Abstract

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 ‘Tumour micro-environment elicits innate resistance to RAF inhibitors through HGF secretion’ by Straussman and colleagues, published in Nature in 2012 (Straussman et al., 2012). The key experiments being replicated in this study are from Figure 2A, C, and D (and Supplemental Figure 11) and Figure 4C (and Supplemental Figure 19) (Straussman et al., 2012). Figure 2 demonstrates resistance to drug sensitivity conferred by co-culture with some stromal cell lines and identifies the secreted factor responsible as HGF. In Figure 4, Straussman and colleagues show that blocking the HGF receptor MET abrogates HGF’s rescue of drug sensitivity. 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.04034.001

Original article

  1. Tumour micro-environment elicits innate resistance to RAF inhibitors through HGF secretion

    1. R Straussman
    2. T Morikawa
    3. K Shee
    4. M Barzily-Rokni
    5. ZR Qian
    6. J Du
    7. A Davis
    8. MM Mongare
    9. J Gould
    10. DT Frederick
    11. et al.
    Nature 2012;487:500-504

Main text

Introduction

Resistance to oncoprotein-targeted chemotherapy is a common occurrence during cancer treatment and identifying the mechanisms of resistance is important in improving treatment options. Specifically, BRAF-mutant melanomas, which show an initial response to RAF inhibitors, usually become resistant to the therapy (Nickoloff and Vande Woude, 2012). The identification of stroma-mediated resistance in BRAF-mutant melanomas, through the secretion of hepatocyte growth factor (HGF), therefore indicates a potential therapeutic strategy through combination treatment of RAF inhibitors and inhibition of the HGF activated pathway (Straussman et al., 2012). This report is the first to identify paracrine HGF as a potential mechanism for the development of drug resistance (Ghiso and Giordano, 2013; Glaire et al., 2012).

In Figure 2A of their paper, Straussman and colleagues tested the effect of fibroblast-conditioned medium on the proliferation of BRAF-mutant melanoma cells grown in the presence of the BRAF inhibitor PLX4720. Using a cell proliferation assay, they reported that fibroblast-conditioned medium rescued BRAF-mutant melanoma cells from PLX4720 sensitivity, which indicated that a secreted factor was involved. This was a key finding demonstrating that the stromal environment of the tumor cells could mediate their response to drug treatment. This experiment will be replicated in Protocol 3.

Straussman and colleagues went on to identify the secreted factor responsible for acquired drug resistance as HGF. In Figure 2C, they demonstrated that treating melanoma cell lines with PLX4720 in combination with increasing concentrations of exogenous HGF increased proliferation as compared to treatment with drug alone. This finding showed a similar effect to treatment with conditioned media from stromal cells that secrete HGF (see Figure 2A) and supported the hypothesis that HGF is the growth factor responsible for rescuing melanoma cells from drug sensitivity. This experiment will be replicated in Protocol 4.

Straussman and colleagues demonstrated that the HGF-mediated rescue of melanoma cells from drug sensitivity was mediated through HGF's cognate receptor tyrosine kinase MET by treating melanoma cell lines co-cultured with stromal cell lines in the presence of PLX4720 with the MET inhibitor crizotinib, as shown in Figure 2D and Supplemental Figure 11. Treatment with crizotinib reduced the increase in proliferation due to co-culture with an HGF-secreting stromal cell line. This experiment provided further support for the hypothesis that HGF was responsible for rescue from drug sensitivity and also provided evidence that that rescue was MET dependent. This experiment is replicated in Protocol 5.

Lastly, Straussman and colleagues reported sustained activation of both ERK and AKT in HGF-treated melanoma cells during BRAF inhibition and to a lesser extent MEK inhibition, as shown in Figure 4C and Supplemental Figure 19 by Western blot. This confirmed activation of pro-survival pathways in response to HGF treatment even in the presence of PLX4720. These experiments are replicated in Protocol 6.

To date, a direct replication has been reported; Lezcano and colleagues (Lezcano et al., 2014) published a replication of Figure 3 of Straussman et al. Nature 2013, wherein Straussman and colleagues evaluated HGF expression in patient-derived primary melanoma samples and observed a negative correlation between expression of HGF and response to therapy (Straussman et al., 2012). While Lezcano and colleagues' replication also detected the presence of HGF in human melanoma tumor cells and stromal cells with increased expression at disease progression, they did not identify a statistically significant correlation between HGF expression and clinical outcome (Lezcano et al., 2014). While both of the studies come to different conclusions about the association of stromal HGF and clinical outcome, the 95% confidence intervals of the standardized measure of the effect (Cohen's d) for each study substantially overlap. A study published around the same time as the work of Straussman and colleagues supports the negative association between HGF and clinical response to RAF inhibitor treatments through an analysis of HGF levels in patient plasma samples (Wilson et al., 2012).

In other systems, additional labs have observed a similar role for HGF in acquired drug resistance. Caenepeel and colleagues reported that HGF rescued melanoma cell lines, notably SK-MEL-5, from BRAF or MEK inhibition using vemurafenib (an analogue of PLX4720) or PD0325901, respectively, and the rescue was attenuated by MET inhibition (Caenepeel et al., 2013). Nakagawa and colleagues observed that tumor-secreted (not stromal secreted) HGF could induce resistance to the VEGFR inhibitor lenvatinib, and that this resistance could be overcome by co-treatment with golvatinib, a MET inhibitor (Nakagawa et al., 2014). Etnyre and colleagues reported that c-MET and BRAF inhibitors had synergistic inhibitory effects when exposed in combination to melanoma cell lines (Etnyre et al., 2013). Casbas-Hernandez and colleagues co-cultured MCF10 cells with immortalized mammoplasty derived fibroblasts and observed a correlation between the levels of fibroblast-secreted HGF and the differentiation of the MCF10 cells towards a ductal carcinoma phenotype. They also observed a correlation between HGF expression and the more invasive basal-like tumors as opposed to the less invasive luminal tumors (Casbas-Hernandez et al., 2013). HGF is also being evaluated as a potential biomarker to indicate potential treatment choices (Penuel et al., 2013; Xie et al., 2013).

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: determining the range of detection of the replicating lab's plate reader

This is a general protocol that determines the range of detection of the plate reader. Because the plate reader in use by the replicating lab is different than the plate reader used in the original study, we are determining what the range of detection is for the replicating lab's plate reader.

Sampling

  • SK-MEL-5

    1. 8000 cells/well x 4 replicates

    2. 4000 cells/well x 4 replicates

    3. 2000 cells/well x 4 replicates

    4. 1000 cells/well x 4 replicates

    5. 500 cells/well x 4 replicates

    6. 250 cells/well x 4 replicates

    7. 125 cells/well x 4 replicates

    8. 62.5 cells/well x 4 replicates

    9. 31.25 cells/well x 4 replicates

  • The experiment is done a total of once.

Materials and reagents

  • Reagents that are different from ones originally used are noted with an asterisk (*).

ReagentTypeManufacturerCatalog #Comments
pLEX-TRC206 SK-MEL-5CellsOriginal authorsn/aEngineered to express GFP
Synergy HT Microplate Reader*EquipmentBio-TekOriginal equipment used: Molecular Devices SpectraMax M5e Microplate Reader
384-well clear-bottomed platesMaterialCorning3712
Phenol red free DMEM*MediumSigma-AldrichD1145Original unspecified.
Sodium pyruvate solution*ReagentSigma-AldrichS8636This formulation of DMEM does not contain L-glutamine or sodium pyruvate, so these will be supplemented to the medium.
FBS*ReagentSigma-AldrichF4135Original unspecified
100X Pen–Strep–Glut*ReagentSigma-AldrichG1146Original from Invitrogen (15,140-122)
Puromycin dihydrochlorideReagentSigma-AldrichP9620Original unspecified
Biomek FX*EquipmentBeckman CoulterCommunicated by authors. Original from Thermo Scientific (Combi reagent dispenser)
Procedure

  • 1. Seed 4 wells of a 384-well clear-bottom plate with 8000 cells/well all the way to 31.25 cells/well (serial 1:2 dilutions) with pLex-TRC206 SK-MEL-5 cells in 60 µl per well using phenol red free medium using an automated workstation.

  • Note: all cells will be sent for mycoplasma testing and STR profiling.

  • Note: ensure at least 85% of SK-MEL-5 cells are GFP-positive before start of the experiment. Cells can be enriched using FACS or puromycin (0.5–2 µg/ml), however do not grow cells under antibiotic selection on a regular basis.

    • A. Total wells seeded = 36

    • B. Medium for assay: phenol red free DMEM supplemented with 1 mM sodium pyruvate, 10% FBS, and 1X Pen–Strep–Glut.

    • C. Fill wells with 60 µl/well of clear media in at least 2 rows and 2 columns around wells that are being included in the experiment.

  • 2. The next day after seeding, read GFP fluorescence (Synergy HT Microplate Reader).

    • A. Subtract the average reading from media-only wells from the wells with cells.

Deliverables

  • Data to be collected:

    1. Raw GFP fluorescence readings.

    2. Graph of GFP fluorescence readings vs cell number.

Confirmatory analysis plan

  • Statistical Analysis:

    1. Coefficient of determination of data values.

Known differences from the original study

  • Synergy HT Microplate Reader used instead of Molecular Devices SpectraMax M5e Microplate Reader—both can detect GFP fluorescence and the Synergy HT Microplate Reader will be evaluated for sensitivity of detection (Protocol 1) and to determine if the gradient is similar to the original study (≤5%) (Protocol 2).

Provisions for quality control

This protocol will ensure that the replicating lab's plate reader is comparable to the original lab's plate reader.

  • A lab from the Science Exchange network with extensive experience in conducting cell viability assays will perform these experiments.

  • All cells will be sent for STR profiling to confirm identity and mycoplasma testing to confirm the lack of mycoplasma contamination.

  • SK-MEL-5 cells will be confirmed to have at least 85% of the cells GFP-positive before the start of the experiment.

Protocol 2: determining the detection variability of the replicating lab's plate reader

This is a general protocol that determines the variability in detection of the plate reader. Because the plate reader in use by the replicating lab is different than the plate reader used in the original study, we are determining what the variability of detection is for the replicating lab's plate reader.

Sampling

  • SK-MEL-5:

    1. 2000 cells/well x 384 replicates

  • Experiment will be done a total of once.

Materials and reagents

• Reagents that are different from ones originally used are noted with an asterisk (*).

ReagentTypeManufacturerCatalog #Comments
pLEX-TRC206 SK-MEL-5CellsOriginal authorsn/aEngineered to express GFP
Synergy HT Microplate Reader*EquipmentBio-TekOriginal equipment used: Molecular Devices SpectraMax M5e Microplate Reader
384-well clear-bottomed platesMaterialCorning3712
Phenol red free DMEM*MediumSigma-AldrichD1145Original unspecified. This formulation of DMEM does not contain L-glutamine or sodium pyruvate, so these will be supplemented to the medium.
Sodium pyruvate solution*ReagentSigma-AldrichS8636
FBS*ReagentSigma-AldrichF4135Original unspecified
100X Pen–Strep–Glut*ReagentSigma-AldrichG1146Original from Invitrogen (15,140-122)
Puromycin dihydrochlorideReagentSigma-AldrichP9620Original unspecified
Biomek FXEquipmentBeckman CoulterCommunicated by authors. Original from Thermo Scientific (Combi reagent dispenser)
Procedure

  • 1. Seed all wells of a 384 -well clear-bottom plate with 2000 pLex-TRC206 SK-MEL-5 cells (provided by authors) in 60 µl per well using phenol red free medium using an automated workstation.

  • Note: all cells will be sent for mycoplasma testing and STR profiling.

  • Note: ensure at least 85% of SK-MEL-5 cells are GFP-positive before start of the experiment. Cells can be enriched using FACS or antibiotics, however do not grow cells under antibiotic selection on a regular basis.

    • A. Medium for assay: phenol red free DMEM supplemented with 1 mM sodium pyruvate, 10% FBS, and 1× Pen–Strep–Glut.

    • B. Fill wells with 60 µl/well of clear media in at least 2 rows and 2 columns around wells that are being included in the experiment.

  • 2. The next day after seeding, read GFP fluorescence (Synergy HT Microplate Reader).

    • A. Subtract the average reading from media only wells from the wells with cells.

Deliverables

  • Data to be collected:

    1. Raw GFP fluorescence readings.

    2. Difference of each individual well and the average reading across the plate.

Confirmatory analysis plan

  • Statistical Analysis:

    1. Standard deviation of data values.

Known differences from the original study

  • Synergy HT Microplate Reader used instead of Molecular Devices SpectraMax M5e Microplate Reader—both can detect GFP fluorescence and the Synergy HT Microplate Reader will be evaluated for sensitivity of detection (Protocol 1) and to determine if the gradient is similar to the original study (≤5%) (Protocol 2).

Provisions for quality control

This protocol will ensure that the replicating lab's plate reader is comparable to the original lab's plate reader.

  • A lab from the Science Exchange network with extensive experience in conducting cell viability assays will perform these experiments.

  • All cells will be sent for STR profiling to confirm identity and mycoplasma testing to confirm the lack of mycoplasma contamination.

  • SK-MEL-5 cells will be confirmed to have at least 85% of the cells GFP-positive before the start of the experiment.

Protocol 3: co-culture proliferation assay

This protocol outlines how to culture melanoma cell lines with conditioned medium from three stromal cell lines with or without the RAF inhibitor PLX4720 to analyze cell proliferation rates, as is described in Figure 2A.

Sampling

  • Experiment to be repeated a total of 4 times for a minimum power of 81%.

    1. See Power calculations section for details

  • Each experiment has six conditions to be run in quadruplicate per experiment:

    1. SK-MEL-5 untreated control [additional control]

    2. SK-MEL-5 vehicle (DMSO) control

    3. SK-MEL-5 treated with 2 µM PLX4720 and with unconditioned medium

    4. SK-MEL-5 treated with 2 µM PLX4720 and with conditioned medium from CCD-1090Sk cells that do not secrete HGF

    5. SK-MEL-5 treated with 2 µM PLX4720 and with conditioned medium from PC60163A1 cells that do secrete HGF

    6. SK-MEL-5 treated with 2 µM PLX4720 and with conditioned medium from LL 86 cells that do secrete HGF

Materials and reagents

• Reagents that are different from ones originally used are noted with an asterisk (*).

ReagentTypeManufacturerCatalog #Comments
LL 86 cellsCellsOriginal authorsn/aStromal cell line that secretes HGF
PC60163A1CellsOriginal authorsn/aStromal cell line that secretes HGF
CCD-1090Sk cellsCellsOriginal authorsn/aStromal cell line that does not secrete HGF
pLEX-TRC206 SK-MEL-5CellsOriginal authorsn/aEngineered to express GFP
Synergy HT Microplate Reader*EquipmentBio-TekOriginal equipment used: Molecular Devices SpectraMax M5e Microplate Reader
Pathway 435 BioimagerEquipmentBD BiosciencesOriginal equipment used: Zeiss Axio Observer.Z1
384-well clear-bottomed platesMaterialCorning3712
10 cm tissue culture plates*MaterialsCorning430167Original unspecified
0.45 µm syringe filterMaterialsSigma-AldrichZ355518Original unspecified
10 ml syringeMaterialsSigma-AldrichZ116874Original unspecified
Phenol red free DMEM*MediumSigma-AldrichD1145Original unspecified. This formulation of DMEM does not contain L-glutamine or sodium pyruvate, so these will be supplemented to the medium.
Sodium pyruvate solution*ReagentSigma-AldrichS8636
FBS*ReagentSigma-AldrichF4135Original unspecified
100X Pen–Strep–Glut*ReagentSigma-AldrichG1146Original from Invitrogen (15,140-122)
PLX4720DrugChemietekCT-P4720
DMSO*ReagentSigma-AldrichD8418Original unspecified
Biomek FXEquipmentBeckman CoulterCommunicated by authors. Original from Thermo Scientific (Combi reagent dispenser) and CyBio robotic liquid handler.
Procedure

  1. Prepare Pre-Conditioned Medium (PCM); fresh PCM must be prepared the same day it is used in the treatment of SK-MEL-5 cells; this step is repeated three times to ensure fresh PCM is available on the needed day:

    • A. Three days before the PCM is needed, seed 3 × 10 cm tissue culture plates with 0.5x106 LL 86 cells each, 3 × 10 cm tissue culture plates with 1x106 PC60163A1 cells each, and 3 × 10 cm tissue culture plates with 2 × 106 CCD-1090Sk cells each (9 plates total) in 10 ml of phenol red free medium each and grow for 3 days.

    • B. 3 days after seeding, collect the medium from each cell line using the plate closest to 80–90% confluent.

      • i. 75–95% confluency can be used.

    • C. Filter through 0.45 µm syringe filter with a 10 ml syringe and dilute filtered PCM 1:1 in fresh phenol red free medium. Total volume = 20 ml.

      • i. Use the same day.

      • ii. Do not dilute for day 0 of treatment (these wells will already have 20 µl of media in them).

  2. On day 0, seed 120 wells of a 384-well clear-bottom plate with 1900 pLex-TRC206 SK-MEL-5 cells in 20 µl per well using phenol red free medium using an automated workstation.

    • Note:

      1. All cells will be sent for mycoplasma testing and STR profiling.

      2. Ensure at least 85% of SK-MEL-5 cells are GFP-positive before start of the experiment. Cells can be enriched using FACS or antibiotics, however do not grow cells under antibiotic selection on a regular basis.

      3. Do not exceed a rate of 5–10 µl/s and do not let the tip end closer than 1 mm to the well bottom.

    • A. Fill wells with 50 µl/well of media in at least 2 rows and 2 columns around wells that are being included in the experiment.

      • i. Medium for assay: phenol red free DMEM supplemented with 1 mM sodium pyruvate, 10% FBS, and 1X Pen–Strep–Glut.

    • B. To wells in step A, add 20 µl of fresh undiluted PCM from appropriate stromal cells generated as described in step 1 (see Sampling section for Cohorts) or phenol red free medium alone (Cohort 1).

  3. On day 1 after seeding, read GFP fluorescence (Synergy HT Microplate Reader).

    • A. Subtract the average reading from media-only wells from the wells with cells.

  4. After reading GFP fluorescence, refresh media and add drug using an automated workstation.

    • A. Change the medium for each cohort to 40 µl fresh diluted PCM from appropriate stromal cell lines generated as described in step 1 or phenol red free medium alone.

    • B. Within each cohort, add 10 µl of 5X PLX4720, DMSO dilution, or 10 µl phenol red free medium to each appropriate well to bring the final volume per well up to 50 µl.

      • i. 5X PLX4720: make up stocks of 10 mM PLX4720 in DMSO, then dilute 1:1000 in media to make up 10 µM PLX4720. This is a 5× stock.

      • ii. DMSO dilution: dilute 1 µl DMSO with 999 µl media. Add 10 µl of this mix to DMSO wells.

        1. These dilutions in media prevent toxicity from excess DMSO.

  5. On day 4 after seeding, read GFP fluorescence.

    • A. Subtract the average reading from media-only wells from the wells with cells.

  6. After reading GFP fluorescence, change the medium in appropriate wells to 40 µl fresh diluted PCM from appropriate stromal cell lines generated as described in step 1 or phenol red free medium alone using an automated workstation.

    • A. Add 10 µl of 5X PLX4720, DMSO dilution, or 10 µl phenol red free medium to each appropriate well to bring the final volume per well up to 50 µl.

      • i. 5X PLX4720: make up stocks of 10 mM PLX4720 in DMSO, then dilute 1:1000 in media to make up 10 µM PLX4720. This is a 5× stock.

      • ii. DMSO dilution: dilute 1 µl DMSO with 999 µl media. Add 10 µl of this mix to DMSO wells.

  7. On day 7 after seeding, read GFP fluorescence and document bright-field and GFP images (BD, Pathway 435 Bioimager).

    • A. Subtract the average reading from media-only wells from the wells with cells.

  8. Data analysis:

    • A. Remove background fluorescence by subtracting the average reading from media-only wells from the wells with cells for each plate reading.

    • B. Subtract the readings of day 1 from the other plates (day 4 and day 7) for the same wells.

    • C. Average the quadruplicates.

    • D. Calculate the effect of PLX4720 in the presence or absence of conditioned media by normalizing the number of cells after 7 days of treatment (as measured by GFP fluorescence) to the number of cells present in the SK-MEL-5 vehicle control condition.

  9. Repeat experiment independently three additional times.

Deliverables

  • Data to be collected:

    1. Raw GFP fluorescence readings from days 1, 4, and 7.

    2. Normalized fluorescence proliferation data.

    3. Fluorescent and bright-field micrographs of cells from day 7.

    4. Bar chart of relative proliferation as a % of untreated control for all conditions. (Use data from Day 7–Day 1 background) (Compare to Figure 2A)

    5. A semi-logarithmic graph of proliferation (log) vs time (linear) over three time points after seeding.

Confirmatory analysis Plan

  • Statistical analysis of the replication data:

    • A. One-way ANOVA comparing the proliferation of PLX4720-treated cells cultured with unconditioned medium, CCD-1090Sk conditioned medium, LL 86 conditioned medium, or PC60163A1 conditioned medium.

      1. Planned comparisons with the Bonferroni correction:

        • Unconditioned medium to PC60163A1 conditioned medium

        • Unconditioned medium to LL 86 conditioned medium

        • CCD-1090Sk to PC60163A1 conditioned medium

        • CCD-1090Sk to LL 86 conditioned medium

  • Meta-analysis of original and replication attempt effect sizes:

    • A. Compare the effect sizes of the original data to the replication data and use a meta-analytic approach to combine the original and replication effects, which will be presented as a forest plot.

Known differences from the original study

  • The replication will only use one of the three melanoma cell lines used by the original authors, the SK-MEL-5 cell line. The replication will exclude SK-MEL-28 and G-361 cells.

  • The replication will include an additional control, untreated SK-MEL-5 cells in addition to the vehicle (DMSO) treated SK-MEL-5 cells used in the original study.

  • A Synergy HT Microplate Reader will be used instead of a Molecular Devices SpectraMax M5e Microplate Reader—both can detect GFP fluorescence and the Synergy HT Microplate Reader will be evaluated for range of detection (Protocol 1) and detection variability (Protocol 2)

  • A BD Pathway 435 Bioimager used instead of a Zeiss Axio Observer.Z1—both are fluorescence microscopes with high-throughput screening capabilities.

  • The replicating lab does not have a ViCell XR cell viability counter, and thus will seed a larger number of cells per well (1900 instead of 1700 cells/well).

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/p4lzc/).

  • A lab from the Science Exchange network with extensive experience in conducting cell viability assays will perform these experiments.

  • All cells will be sent for STR profiling to confirm identity and mycoplasma testing to confirm the lack of mycoplasma contamination.

  • SK-MEL-5 cells will be confirmed to have at least 85% of the cells GFP-positive before the start of the experiment.

Protocol 4: recombinant HGF proliferation assay

This protocol assesses changes in proliferation when melanoma cells are treated with the RAF inhibitor PLK4720 with or without HGF, as is described in Figure 2C. The cells are also treated with a MEK inhibitor, PD184352.

Sampling

  • Experiment to be repeated a total of three times for a final power of 99%.

    1. See Power calculations section for details

  • Each experiment has 12 conditions to be done in quadruplicate per experiment:

    1. SK-MEL-5 untreated control [additional control]

    2. SK-MEL-5 vehicle (DMSO) control

    3. SK-MEL-5 2 µM PLX4720 + 0 ng/ml HGF

    4. SK-MEL-5 2 µM PLX4720 + 6.25 ng/ml HGF

    5. SK-MEL-5 2 µM PLX4720 + 12.5 ng/ml HGF

    6. SK-MEL-5 2 µM PLX4720 + 25 ng/ml HGF

    7. SK-MEL-5 2 µM PLX4720 + 50 ng/ml HGF

    8. SK-MEL-5 1 µM PD184352 + 0 ng/ml HGF

    9. SK-MEL-5 1 µM PD184352 + 6.25 ng/ml HGF

    10. SK-MEL-5 1 µM PD184352 + 12.5 ng/ml HGF

    11. SK-MEL-5 1 µM PD184352 + 25 ng/ml HGF

    12. SK-MEL-5 1 µM PD184352 + 50 ng/ml HGF

Materials and reagents

• Reagents that are different from the ones originally used are noted with an asterisk (*).

ReagentTypeManufacturerCatalog #Comments
pLEX-TRC206 SK-MEL-5CellsOriginal authorsn/aEngineered to express GFP
PLX4720DrugChemietekCT-P4720
PD184352DrugSanta Cruzsc-202759AMEK inhibitor
384-well clear-bottomed platesMaterialCorning3712
0.45 µm syringe filterMaterialsSigma-AldrichZ355518Original unspecified
10 ml syringeMaterialsSigma-AldrichZ116874Original unspecified
Phenol red free DMEM*MediumSigma-AldrichD1145Original unspecified. This formulation of DMEM does not contain L-glutamine or sodium pyruvate, so these will be supplemented to the medium.
Sodium pyruvate solution*ReagentSigma-AldrichS8636
FBS*ReagentSigma-AldrichF4135Original unspecified
100X Pen–Strep–Glut*ReagentSigma-AldrichG1146Original from Invitrogen (15,140-122)
HGFReagentSigma-AldrichH5791Original from RayBiotech (228-10,702-2)
DMSO*ReagentSigma-AldrichD8418Original unspecified
Synergy HT Microplate Reader*EquipmentBio-TekOriginal equipment used: Molecular Devices SpectraMax
M5e Microplate Reader
Biomek FXEquipmentBeckman CoulterCommunicated by authors. Original from Thermo Scientific (Combi reagent dispenser) and CyBio robotic liquid handler.
Procedure

  1. On day 0, seed 48 wells of a 384-well clear-bottom plate with 2800 pLex-TRC206 SK-MEL-5 cells in 40 µl of phenol red free medium each using an automated workstation.

    • Note:

      1. All cells will be sent for mycoplasma testing and STR profiling.

      2. Ensure at least 85% of SK-MEL-5 cells are GFP-positive before start of the experiment. Cells can be enriched using FACS or antibiotics, however do not grow cells under antibiotic selection on a regular basis.

      3. Do not exceed a rate of 5–10 µl/s and do not let the tip end closer than 1 mm to the well bottom.

        • A. Fill wells with 60 µl/well of clear media in at least 2 rows and 2 columns around wells that are being included in the experiment.

        • B. Medium of all cell lines for assay: phenol red free DMEM supplemented with 1 mM sodium pyruvate, 10% FBS, and 1× Pen–Strep–Glut.

  2. On day 1 after seeding, read GFP fluorescence (Synergy HT Microplate Reader).

    • A. Subtract the average reading from media-only wells from the wells with cells.

  3. After reading GFP fluorescence, add to the appropriate wells 10 µl 6X HGF or phenol red free medium alone. Then add to the appropriate wells the following: 10 µl 6X PLX4720, 10 µl 6X PD184352, 10 µl DMSO dilution, or 10 µl phenol red free medium alone.

    • A. 6X HGF: make up stocks of 100 μg/ml HGF, then dilute accordingly to make 6X working concentrations of each required HGF dilution.

    • B. 6X PLX4720: make up stocks of 12 mM PLX4720 in DMSO, then dilute 1:1000 in media to make up 12 μM PLX4720 for use at 6X for the assay to avoid excessive DMSO toxicity.

    • C. 6X PD184352: make up stocks of 6 mM PD184352 in DMSO, then dilute 1:1000 in media to make up 6 μM PD184352 for use at 6X for the assay to avoid excessive DMSO toxicity.

    • D. DMSO dilution: dilute 1 µl DMSO with 999 µl media. Add 10 µl of this mix to DMSO dilution wells.

      • A. These media dilutions are to avoid toxicity from excessive DMSO.

  4. On day 4 after seeding, read GFP fluorescence.

    • A. Subtract the average reading from media-only wells from the wells with cells.

  5. After reading GFP fluorescence, change the medium in all wells to 40 µl fresh phenol red free medium using an automated workstation. Then add to the appropriate wells 10 µl 6X HGF or phenol red free medium alone. Then add to the appropriate wells the following: 10 µl 6X PLX4720, 10 µl 6X PD184352, 10 µl DMSO dilution, or 10 µl phenol red free medium alone.

    • A. 6X HGF: make up stocks of 100 μg/ml HGF, then dilute accordingly to make 6X working concentrations of each required HGF dilution.

    • B. 6X PLX4720: make up stocks of 12 mM PLX4720 in DMSO, then dilute 1:1000 in media to make up 12 μM PLX4720 for use at 6X for the assay to avoid excessive DMSO toxicity.

    • C. 6X PD184352: make up stocks of 6 mM PD184352 in DMSO, then dilute 1:1000 in media to make up 6 μM PD184352 for use at 6X for the assay to avoid excessive DMSO toxicity.

    • D. DMSO dilution: dilute 1 µl DMSO with 999 µl media. Add 10 µl of this mix to DMSO dilution wells.

      • A. These media dilutions are to avoid toxicity from excessive DMSO.

  6. On day 7 after seeding, read GFP fluorescence and document bright-field and GFP images (BD, Pathway 435 Bioimager).

    • A. Subtract the average reading from media-only wells from the wells with cells.

  7. Data analysis:

    • A. Remove background fluorescence by subtracting the average reading from media-only wells from the wells with cells for each plate reading.

    • B. Subtract the readings of day 1 from the other plates (day 4 and day 7) for the same wells.

    • C. Average the quadruplicates.

    • D. Calculate the effect of PLX4720 and PD184352 in the presence or absence of HGF by normalizing the number of cells after 7 days of treatment (as measured by GFP fluorescence) to the number of cells present in the SK-MEL-5 vehicle control condition.

  8. Repeat the experiment independently two additional times.

Deliverables

  • Data to be collected:

    1. Raw GFP fluorescence readings from days 1, 4, and 7.

    2. Normalized fluorescence proliferation data.

    3. Fluorescent and bright-field micrographs of cells from day 7.

    4. Bar chart of relative proliferation as a % of untreated control for all conditions. (Use data from Day 7 - Day 1 background) (Compare to Figure 2C)

    5. A semi-logarithmic graph of proliferation (log) vs time (linear) over 3 time points after seeding.

Confirmatory analysis Plan

  • Statistical Analysis:

    1. Compare the proliferation rate of PLX4720-treated cells treated with 0, 6.25, 12.5, 25, or 50 ng/ml HGF. Also compare each HGF cohort to the proliferation rate of vehicle-treated and untreated cells.

      • A. One-way ANOVA

    2. Compare the proliferation rate of PD184352-treated cells treated with 0, 6.25, 12.5, 25, or 50 ng/ml HGF. Also compare each HGF cohort to the proliferation rate of vehicle-treated and untreated cells.

      • A. One-way ANOVA

  • Meta-analysis of original and replication attempt effect sizes:

    1. Compare the effect sizes of the original data to the replication data, using a meta-analytic approach to combine the original and replication effects which will be presented as a forest plot.

Known differences from the original study

  • The replication will only use one of the three melanoma cell lines used by the original authors, the SK-MEL-5 cell line. The replication will exclude SK-MEL-28 and G-361 cells.

  • The replication will include an additional control, untreated SK-MEL-5 cells in addition to the vehicle (DMSO) treated SK-MEL-5 cells used in the original study.

  • A Synergy HT Microplate Reader used instead of a Molecular Devices SpectraMax M5e Microplate Reader—both can detect GFP fluorescence and the Synergy HT Microplate Reader will be evaluated for range of detection (Protocol 1) and detection variability (Protocol 2)

  • A BD Pathway 435 Bioimager used instead of a Zeiss Axio Observer.Z1—both are fluorescence microscopes with high-throughput screening capabilities.

  • The replicating lab does not have a ViCell XR cell viability counter and thus will seed a larger number of cells per well (2800 instead of 2500 cells/well).

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/p4lzc/).

  • A lab from the Science Exchange network with extensive experience in conducting cell viability assays will perform these experiments.

  • All cells will be sent for STR profiling to confirm identity and mycoplasma testing to confirm the lack of mycoplasma contamination.

  • SK-MEL-5 cells will be confirmed to have at least 85% of the cells GFP-positive before the start of the experiment.

Protocol 5: inhibitor proliferation assay

This experiment confirms that the rescue from drug sensitivity is due to HGF signaling by co-treating cells with crizotinib, an inhibitor of MET, the receptor tyrosine kinase for HGF, as seen in Figure 2D and Supplemental Figure 11.

Sampling

  • Run the experiment six times in total for a minimum power of 80%.

    1. See Power calculations section for details

  • Each experiment has 10 cohorts:

    1. Each cohort consists of

      • • SK-MEL-5 cells alone

      • • SK-MEL-5 co-cultured with LL86 cells

      • • SK-MEL-5 co-cultured with CCD-1090Sk cells

        • Each condition will be run in quadruplicate.

    2. The cohorts are treated with the following drugs:

      • • Cohort 1: no drug treatment [additional control]

      • • Cohort 2: treated with vehicle (DMSO) control

      • • Cohort 3: treated with 0.2 µM crizotinib and vehicle

      • • Cohort 4: treated with 0.2 µM PHA-665752 and vehicle [additional control]

      • • Cohort 5: treated with 2 µM PLX4720 and vehicle

      • • Cohort 6: treated with 2 µM PLX4720 and 0.2 µM crizotinib

      • • Cohort 7: treated with 2 µM PLX4720 and 0.2 µM PHA-665752 [additional]

      • • Cohort 8: treated with 1 µM PD184352 and vehicle

      • • Cohort 9: treated with 1 µM PD184352 and 0.2 µM crizotinib

      • • Cohort 10: treated with 1 µM PD184352 and 0.2 µM PHA-665752 [additional control]

Materials and reagents

• Reagents that are different from ones originally used are noted with an asterisk (*).

ReagentTypeManufacturerCatalog #Comments
pLEX-TRC206 SK-MEL-5CellsOriginal authorsn/aEngineered to express GFP
LL 86 cellsCellsOriginal authorsn/aStromal cell line that secretes HGF
CCD-1090Sk cellsCellsOriginal authorsn/aStromal cell line that does not secrete HGF
PLX4720DrugChemietekCT-P4720BRAF inhibitor
PD184352DrugSanta Cruzsc-202759AMEK inhibitor
crizotinibDrugActive BiochemA-1031MET inhibitor
PHA-665752DrugSigma-AldrichPZ0147MET inhibitor [additional control]
384-well clear-bottomed platesMaterialCorning3712
Phenol red free DMEM*MediumSigma-AldrichD1145Original unspecified. This formulation of DMEM does not contain L-glutamine or sodium pyruvate, so these will be supplemented to the medium.
Sodium pyruvate solution*ReagentSigma-AldrichS8636
FBS*ReagentSigma-AldrichF4135Original unspecified
100X Pen–Strep–Glut*ReagentSigma-AldrichG1146Original from Invitrogen (15,140-122)
DMSO*ReagentSigma-AldrichD8418Original unspecified
Synergy HT Microplate Reader*EquipmentBio-TekOriginal equipment used: Molecular Devices SpectraMax M5e Microplate Reader
Biomek FXEquipmentBeckman CoulterCommunicated by authors. Original from Thermo Scientific (Combi reagent dispenser) and CyBio robotic liquid handler.
Procedure

  1. On day 0, seed 40 wells of a 384-well clear-bottom plate with 1900 LL86 stromal cells in 20 µl phenol red free media, seed 40 wells with 1900 CCD-1090Sk stromal cells in 20 µl media, and seed 40 wells with phenol red free medium alone using an automated workstation.

    • Note:

      1. All cells will be sent for mycoplasma testing and STR profiling.

      2. Ensure at least 85% of SK-MEL-5 cells are GFP-positive before the start of the experiment. Cells can be enriched using FACS or antibiotics, however do not grow cells under antibiotic selection on a regular basis.

      3. Do not exceed a rate of 5–10 µl/s and do not let the tip end closer than 1 mm to the well bottom

        • A. Total wells seeded: 120

        • B. Fill wells with 60 µl/well of clear media in at least 2 rows and 2 columns around wells that are being included in the experiment.

        • C. Medium of all cell lines for assay: phenol red free DMEM supplemented with 1 mM sodium pyruvate, 10% FBS, and 1X Pen–Strep–Glut.

  2. In wells from Step 1, seed 1900 pLex-TRC206 SK-MEL-5 cells in 20 µl phenol red free medium per well using an automated workstation.

  3. On day 1 after seeding, read GFP fluorescence (Synergy HT Microplate Reader).

    • A. Subtract the average reading from media-only wells from the wells with cells.

  4. Add appropriate drugs to each well (final volume = 60 µl).

    • A. Formulation of drug stock solutions:

      • i. 6X PLX4720: make up stocks of 12 mM PLX4720 in DMSO, then dilute 1:1000 in media to make up 12 µM PLX4720 for use at 6× for the assay to avoid excessive DMSO toxicity.

      • ii. 6X PD184352: make up stocks of 6 mM PD184352 in DMSO, then dilute 1:1000 in media to make up 6 µM PD184352 for use at 6× for the assay to avoid excessive DMSO toxicity.

      • iii. 6X crizotinib: make up stocks of 1.2 mM crizotinib in DMSO, then dilute 1:1000 in media to make up 1.2 µM PD184352 for use at 6× for the assay to avoid excessive DMSO toxicity.

      • iv. 6X PHA-665752: make up stocks of 1.2 mM PHA-665752 in DMSO, then dilute 1:1000 in media to make up 1.2 µM PD184352 for use at 6× for the assay to avoid excessive DMSO toxicity.

      • V. DMSO dilution: dilute DMSO 1:1000 in medium to avoid excessive DMSO toxicity.

    • B. Cohort 1: add 20 µl phenol red free medium

    • C. Cohort 2: add 10 µl DMSO dilution and 10 µl medium

    • D. Cohort 3: add 10 µl 6X crizotinib and 10 µl medium

    • E. Cohort 4: add 10 µl 6X PHA-665752 and 10 µl medium

    • F. Cohort 5: add 10 µl 6X PLX4720 and 10 µl medium

    • G. Cohort 6: add 10 µl 6X PLX4720 and 10 µl 6X crizotinib

    • H. Cohort 7: add 10 µl 6X PLX4720 and 10 µl 6X PHA-665752

    • I. Cohort 8: add 10 µl 6X PD184352 and 10 µl medium

    • J. Cohort 9: add 10 µl 6X PD184352 and 10 µl 6X crizotinib

    • K. Cohort 10: add 10 µl 6X PD184352 and 10 µl 6X PHA-665752

  5. On day 4 after seeding, read GFP fluorescence.

    • A. Subtract the average reading from media-only wells from the wells with cells.

  6. Change the medium in relevant wells to 40 µl fresh media, then add appropriate drugs as per Step 4 using an automated workstation.

  7. On day 7 after seeding, read GFP fluorescence and document bright-field and GFP images (BD, Pathway 435 Bioimager).

    • A. Subtract the average reading from media-only wells from the wells with cells.

  8. Data analysis:

    • A. Remove background fluorescence by subtracting the average reading from media-only wells from the wells with cells for each plate reading.

    • B. Subtract the readings of day 1 from the other plates (day 4 and day 7) for the same wells.

    • C. Average the quadruplicates.

    • D. Calculate the effect of PLX4720, PD184352, crizotinib, PHA-665752, PLX4720 + crizotinib, PLX4720 + PHA-665752, PD184352 + crizotinib, PD184352 + PHA-665752, DMSO, or untreated in the presence or absence of stromal cells by normalizing the number of cells after 7 days of treatment (as measured by GFP fluorescence) to the number of cells present in the vehicle control treated SK-MEL-5 cells alone condition.

  9. Repeat experiment independently five additional times.

Deliverables

  • Data to be collected:

    1. Raw GFP fluorescence readings from days 1, 4, and 7.

    2. Normalized fluorescence proliferation data.

    3. Fluorescent and bright-field micrographs of cells from day 7.

    4. Bar chart of relative proliferation as a % of untreated control for all conditions. (Use data from Day 7 - Day 1 background) (compare to Figure F11)

    5. A. semi-logarithmic graph of proliferation (log) vs time (linear) over three time points after seeding.

Confirmatory analysis plan

  • Statistical analysis of replication data:

    1. Three-way ANOVA comparing the proliferation of vehicle-treated, PLX4720-treated, or PD184352-treated cells also treated with vehicle, crizotinib, or PHA-665752 cultured with or without stromal cells followed by:

    2. Two-way ANOVA comparing the proliferation of vehicle-treated cells treated with vehicle, crizotinib, or PHA-665752 cultured with or without stromal cells.

    3. Two-way ANOVA comparing the proliferation of PLX4720-treated cells treated with vehicle, crizotinib, or PHA-665752 cultured with or without stromal cells.

      1. Planned comparisons with the Bonferroni correction:

        • Vehicle-treated LL 86 cells compared to vehicle-treated no stromal cells

        • Vehicle-treated LL 86 cells compared to vehicle-treated CCD-1090Sk cells

        • Vehicle-treated LL 86 cells compared to crizotinib-treated LL 86 cells

        • Vehicle-treated LL 86 cells compared to PHA-665752-treated LL 86 cells

    4. Two-way ANOVA comparing the proliferation of PD184352-treated cells treated with vehicle, crizotinib, or PHA-665752 cultured with or without stromal cells.

      1. Planned comparisons with the Bonferroni correction:

        • Vehicle-treated LL 86 cells compared to crizotinib-treated LL 86 cells

        • Vehicle-treated LL 86 cells compared to PHA-665752-treated LL 86 cells

  • Meta-analysis of original and replication attempt effect sizes:

    1. Compare the effect sizes of the original data to the replication data, using a meta-analytic approach to combine the original and replication effects which will be presented as a forest plot.

Known differences from the original study

  • Supplemental Figure 11 tests co-culture of SK-MEL-5 cells with 9 stromal cell lines. We have chosen LL86 cells, which showed the largest rescue of proliferation, and CCD-1090Sk cells, which showed the least rescue.

  • Additional controls added by the replication team:

    1. Treatment with PHA-665752

      • • In addition to inhibiting MET, crizotinib also targets ALK, ROS1, and RON. In order to confirm that the effects of crizotinib are due to targeting of MET, we will also use a more selective MET inhibitor, PHA-665752 (Cui, 2014; Parikh et al., 2014).

    2. The replication will include an additional control, untreated SK-MEL-5 cells in addition to the vehicle (DMSO) treated SK-MEL-5 cells used in the original study.

  • A Synergy HT Microplate Reader used instead of a Molecular Devices SpectraMax M5e Microplate Reader

    1. Both can detect GFP fluorescence and the Synergy HT Microplate Reader will be evaluated for range of detection (Protocol 1) and detection variability (Protocol 2)

  • A BD Pathway 435 Bioimager used instead of a Zeiss Axio Observer.Z1

    1. Both are fluorescence microscopes with high-throughput screening capabilities.

  • The replicating lab does not have a ViCell XR cell viability counter and thus will seed a larger number of cells per well (1900 instead of 1700 cells/well).

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/p4lzc/).

  • A lab from the Science Exchange network with extensive experience in conducting cell viability assays will perform these experiments.

  • All cells will be sent for STR profiling to confirm identity and mycoplasma testing to confirm the lack of mycoplasma contamination.

  • SK-MEL-5 cells will be confirmed to have at least 85% of the cells GFP-positive before the start of the experiment.

Protocol 6: inhibitor Western blot assay of ERK and AKT signaling

This experiment assesses the protein levels of various activated downstream pathway signaling component proteins in the presence or absence of HGF and drugs, as seen in Figure 4C and Supplemental Figure 19.

Sampling

  • Repeat the experiment six times in total for a minimum power of 85%.

    1. See Power calculations section for details

  • Each experiment contains seven conditions:

    1. SK-MEL-5 cells treated with:

      • • Untreated [additional control]

      • • Vehicle (DMSO) control

      • • 2 µM PD184352

      • • 2 µM PLX4720

      • • 25 ng/ml HGF + vehicle

      • • 25 ng/ml HGF + 2 µM PD184352

      • • 25 ng/ml HGF + 2 µM PLX4720

Materials and reagents

• Reagents that are different from ones originally used are noted with an asterisk (*).

ReagentTypeManufacturerCatalog #Comments
Mouse anti-c-MetAntibodyCell Signaling31481:1000 dilution; 145 kDa
Rabbit anti-pMet Tyr 1349AntibodyCell Signaling31331:1000 dilution; 145 kDa
Mouse anti-AKTAntibodyCell Signaling29201:2000 dilution; 60 kDa
Rabbit anti-pAKTAntibodyCell Signaling40601:2000 dilution; 60 kDa
Mouse anti-MEKAntibodyCell Signaling46941:1000 dilution; 45 kDa
Rabbit anti-pMEKAntibodyCell Signaling91541:1000 dilution; 45 kDa
Mouse anti-ERKAntibodySanta Cruz1359001:200 dilution; 44,42 kDa
Rabbit anti-pERKAntibodyCell Signaling43701:2000 dilution; 44,42 kDa
Rabbit anti-GAPDHAntibodyCell Signaling21181:1000 dilution; 37 kDa Loading control
pLEX-TRC206 SK-MEL-5CellsOriginal authorsn/aEngineered to express GFP
PLX4720DrugChemietekCT-P4720
PD184352DrugSanta Cruzsc-202759AMET inhibitor
Odyssey Infrared Imaging SystemEquipmentLi-COR
6-well tissue culture plates*MaterialsCorning3516Original unspecified
DMEM*MediumSigma-AldrichD6429Original from Invitrogen (10,569-010).
FBS*ReagentSigma-AldrichF4135Original unspecified
100X Pen–Strep*ReagentSigma-AldrichP4333Original from Invitrogen (15,140-122)
DMSO*ReagentSigma-AldrichD8418Original unspecified
HGFReagentSigma-AldrichH5791Original from RayBiotech (228-10,702-2)
PhosSTOP phosphatase inhibitorReagentRoche04906837001
Complete mini protease inhibitorReagentRoche
NuPAGE sample reducing agentReagentInvitrogenNP0009
TruPAGE 4–12% TEA-tricine gels*ReagentSigma-AldrichPCG2003Original: NuPage (WG1402BOX)
TruPAGE TEA-Tricine SDS Running Buffer (20X)ReagentSigma-AldrichPCG3001Original unspecified
TruPAGE LDS Sample Buffer (4X)ReagentSigma-AldrichPCG3009Original unspecified
TruPAGE Transfer Buffer (20X)ReagentSigma-AldrichPCG3011Original unspecified
Odyssey blocking bufferReagentLI-COR927-40,000
Chameleon Kit Pre-stained Protein LadderReagentLI-COR928-90000Original unspecified
IRDye® 800CW Goat anti-RMouse IgG (H + L)AntibodyLi-COR926-32210Original unspecified
IRDye 680RD Goat anti-Rabbit IgG (H + L)AntibodyLi-COR926-68071Original unspecified
PBS, without MgCl2 and CaCl2ReagentSigma-AldrichD8537Original unspecified
IGEPAL CA-630 (NP-40 substitute)ReagentSigma-AldrichI8896Original unspecified
Tween 20ReagentSigma-AldrichP1379Original unspecified
DC Protein Assay Kit IIReagentsBio-Rad500-0112
Odyssey Application SoftwareSoftwareLi-COR
Ponceau stain*ReagentSigma-AldrichP3504Not included in the original study
Immobilon-FL PVDF membraneReagentEMD MilliporeIPFL00010Original unspecified
Procedure

  1. On day 0, plate 5 × 105 pLex-TRC206 SK-MEL-5 cells in 2 ml media per well for a total of 7 wells across 2 × 6-well plates.

    • Note:

      1. All cells will be sent for mycoplasma testing and STR profiling.

      2. Ensure at least 85% of SK-MEL-5 cells are GFP-positive before start of the experiment. Cells can be enriched using FACS or antibiotics, however do not grow cells under antibiotic selection on a regular basis.

        • A. Medium of all cell lines for assay: DMEM supplemented with 10% FBS and 1× Pen–Strep.

  2. On day 1 add the appropriate additives to each well.

    • A. Formulation of stock solutions:

      • Note: these dilutions are to avoid toxicity from excessive DMSO.

        • i. 1000X HGF: make a stock of 25 µg/ml HGF.

        • ii. 1000X PLX4720: make a stock of 20 mM PLX4720 in DMSO, then dilute 1:10 in media to make a 2 mM PLX4720 working solution.

        • iii. 1000X PD184352: make a stock of 20 mM PD184352 in DMSO, then dilute 1:10 in media to make a 2 mM working solution.

        • iv. DMSO dilution: dilute DMSO 1:10 in medium.

    • B. For media only: add 2 µl media

    • C. For DMSO: add 2 µl DMSO dilution

    • D. For 2 µM PD184352: add 2 µl 1000X PD184352

    • E. For 2 µM PLX4720: add 2 µl 1000X PLX4720

    • F. For 25 ng/ml HGF + DMSO: add 2 µl 1000X HGF and 2 µl DMSO dilution

    • G. For 25 ng/ml HGF +2 µM PD184352: add 2 µl 1000X HGF and 2 µl 100X PD184352

    • H. For 25 ng/ml HGF +2 µM PLX4720: add 2 µl 1000X HGF and 2 µl 100X PLX4720.

  3. 24 hr after drug treatment, prepare cells for lysis.

    • A. Quickly wash cells with ice-cold PBS and remove excess PBS.

    • B. Add 0.5 ml or less of ice-cold lysis buffer to wells on ice.

      • i. Lysis buffer: 50 mM Tris pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% NP-40, 1 mg/ml NaF, and one pellet per 10 ml each of PhosSTOP phosphatase inhibitor and complete mini protease inhibitor.

    • C. Scrape cells off dish with cell scraper.

    • D. Collect cells in a 1.5 ml centrifuge tube on ice.

    • E. Incubate on ice for 30 min with periodic vortexing.

    • F. Spin down at 4°C and remove supernatant into separate tube.

  4. Determine protein concentration by using the DC Protein Assay Kit II following manufacturer’s instructions.

  5. Mix 50 µg total cell lysate with NuPAGE sample reducing agent and run on two 4–12% TEA-tricine gels with a protein molecular weight marker at 120 V.

  6. Transfer onto membrane using replicating lab's transfer protocol.

  7. After the transfer, stain the membrane with Ponceau to visualize the transferred protein. Image membrane, then wash out the Ponceau stain [additional quality control step].

  8. Wet membrane with PBS for 5 min, then block membranes in Odyssey blocking buffer (LI-COR, 927-40,000) following manufacturer's instructions.

  9. Probe membrane with the following primary antibodies diluted in Odyssey blocking buffer at 4°C with gentle shaking, overnight.

    • A. Mouse anti-c-Met (Cell Signaling, 3148); 1:1000; 145 kDa

    • B. Rabbit anti-pMet Tyr 1349 (Cell Signaling, 3133); 1:1000; 145 kDa

    • C. Mouse anti-AKT (Cell Signaling, 2920); 1:2000; 60 kDa

    • D. Rabbit anti-pAKT (Cell Signaling, 4060); 1:2000; 60 kDa

    • E. Mouse anti-MEK (Cell Signaling, 4694); 1:1000; 45 kDa

    • F. Rabbit anti-pMEK (Cell Signaling, 9154); 1:1000; 45 kDa

    • G. Mouse anti-ERK (Santa Cruz, 135900); 1:200; 44,42 kDa

    • H. Rabbit anti-pERK (Cell Signaling, 4370); 1:2000; 44,42 kDa

    • I. Rabbit anti-GAPDH (Cell Signaling, 2118); 1:1000; 37 kDa

      • i. Loading control

    • J. Note: multiple gels will need to be run to probe for this many proteins. Do not strip between probing with different phospho antibodies, just wash membrane well (4 × 10 min PBS-T) and then add next antibody. Suggest grouping as follows:

      • i. Gel 1: Probe pAKT [rabbit 60 kDa], then pMEK [rabbit 45 kDa], then AKT [mouse 60 kDa], then MEK [mouse 45 kDa], then GAPDH [rabbit 37 kDa].

      • ii. Gel 2: Probe pMet Tyr 1349 [rabbit 145 kDa], then pERK [rabbit 44,42 kDa], then c-Met [mouse 145 kDa], then ERK [mouse 44,42 kDa], then GAPDH [rabbit 37 kDa].

  10. Wash membranes in PBS +0.1% Tween 20 4 × 5 min.

  11. Detect primary antibodies with anti-rabbit or anti-mouse IRDye secondary antibodies (LICOR) diluted in Odyssey blocking buffer for 30–60 min protected from light following manufacturer's instructions.

  12. Wash membranes in PBS +0.1% Tween 20 4 × 5 min.

  13. Rinse membrane with PBS to remove residual Tween 20.

  14. Detect near infrared fluorescence with the Odyssey Infrared Imaging System.

  15. Quantify signal intensity with Odyssey Application Software.

    • A. For each antibody subtract background intensity from values and then divide by the GAPDH loading control.

    • B. Calculate the effect of PLX4720, PD184352, or vehicle in the presence or absence of HGF by normalizing the band intensities (after background and loading correction) to the band intensity of the SK-MEL-5 vehicle control condition.

  16. Repeat experiment independently five additional times.

Deliverables

  • Data to be collected:

    1. Odyssey images of probed membranes (full images with ladder).

    2. Raw and quantified signal intensities normalized for GAPDH loading and total pan-protein levels.

    3. Bar graphs of normalized mean signal intensities (compare to Figure 19).

Confirmatory analysis plan

  • Statistical analysis of replication data:

    1. Two-way ANOVA comparing the relative phopho-AKT band intensities of cells treated with vehicle, PLX4720, or PD184352 in the presence or absence of HGF.

      • A. Planned comparisons with the Bonferroni correction:

        • PLX4720-treated cells in the absence of HGF compared to PLX4720-treated cells in the presence of HGF.

    2. Two-way ANOVA comparing the relative phopho-ERK band intensities of cells treated with vehicle, PLX4720, or PD184352 in the presence or absence of HGF.

      • A. Planned comparisons with the Bonferroni correction:

        • PLX4720-treated cells in the absence of HGF compared to PLX4720-treated cells in the presence of HGF.

    3. Two-way ANOVA comparing the relative phopho-MET (Tyr1349) band intensities of cells treated with vehicle, PLX4720, or PD184352 in the presence or absence of HGF.

      • A. Planned comparisons with the Bonferroni correction:

        • Cells treated in the absence of HGF and treated with vehicle, PLX4720, or PD184352 compared to cells treated in the presence of HGF and treated with vehicle, PLX4720, or PD184352.

  • Meta-analysis of original and replication attempt effect sizes:

    1. Compare the effect sizes of the original data to the replication data and use a meta-analytic approach to combine the original and replication effects, which will be presented as a forest plot.

Known differences from the original study

  • Provider lab transfer protocol used instead of iBlot Gel Transfer Device (Invitrogen, IB1001) using Program 4—both are capable of transferring protein efficiently, and to determine completeness of the transfer the gel may be stained (Step 8).

  • The replication will include an additional control, untreated SK-MEL-5 cells in addition to the vehicle (DMSO) treated SK-MEL-5 cells used in the original study.

  • The replication will not include the pMet Tyr1234/5, RAF1, and pRAF1 antibodies included in 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/p4lzc/).

  • A lab from the Science Exchange network with extensive experience in conducting cell viability assays and performing Western blots will perform these experiments.

  • All cells will be sent for STR profiling to confirm identity and mycoplasma testing to confirm the lack of mycoplasma contamination.

  • SK-MEL-5 cells will be confirmed to have at least 85% of the cells GFP-positive before the start of the experiment.

Power calculations

All calculations are determined in order to reach at least 80% power.

Protocol 1

No power calculations required.

Protocol 2

No power calculations required.

Protocol 3

Summary of original data:

Note: original data values were shared by authors.

SK-MEL-5 cells onlyMeanSEMSDN
Unconditioned medium30.76.0610.53
CCD-1090Sk conditioned medium32.00.731.263
PC60163A1 conditioned medium83.311.6620.33
LL 86 conditioned medium83.67.2612.63

• Standard deviation was calculated using the formula, SD = SEM*(SQRT n)

Test family

  • ANOVA: fixed effects, omnibus, one-way, alpha error = 0.05

    • Power calculations were performed from effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

    • ANOVA F statistic calculated with Graphpad Prism 6.0

    • Partial η2 calculated from Lakens (2013)

Power calculations for replication

GroupsF test statisticPartial η2Effect size fA priori powerTotal sample size
Unconditioned, CCD-1090Sk, PC60163A1, and LL 86 conditioned mediumF(3,8) = 15.90950.85642.44208793.7%a8a (4 groups)
  • A total sample size of 16 will be used based on the planned comparison calculations making the power 99.9%.

Test family

  • Two tailed t-test; difference between two independent means Bonferroni’s correction: alpha error = 0.0125.

    • Calculations were performed from effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

Power calculations for replication

Group 1Group 2Effect size dA Priori powerGroup 1 sample sizeGroup 2 sample size
Unconditioned mediumPC60163A1 conditioned medium3.25479880.9%44
Unconditioned mediumLL 86 conditioned medium4.56127780.7%a3a3a
CCD-1090Sk conditioned mediumPC60163A1 conditioned medium3.56698688.0%44
CCD-1090Sk conditioned mediumLL 86 conditioned medium5.76279994.8%b3b3b
  • 4 per group will be used based on the other comparisons making the power 98.3%.

  • 4 per group will be used based on the other comparisons making the power 99.9%.

Protocol 4

Summary of original data

Note: original data values were shared by authors.

PLX4720-treated SK-MEL-5 cellsMeanSEMSDN
0 ng/ml HGF32.111.019.13
6.25 ng/ml HGF73.53.095.353
12.5 ng/ml HGF84.37.2712.63
25 ng/ml HGF93.713.022.53
50 ng/ml HGF96.98.5614.83

• Standard deviation was calculated using the formula, SD = SEM*(SQRT n)

Test family

  • ANOVA: fixed effects, omnibus, one-way, alpha error = 0.05

    • Power calculations were performed from effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

    • ANOVA F statistic calculated with Graphpad Prism 6.0

    • Partial η2 calculated from Lakens (2013)

Power calculations for replication

GroupsF test statisticPartial η2Effect size fA priori powerTotal sample size
0, 6.25, 12.5, 25, and 50 ng/ml HGFF(4,10) = 8.07960.76371.79775197.7%a10a (5 groups)
  • A total sample size of 15 will be used as a minimum making the power 99.9%.

Summary of original data

Note: original data values were shared by authors.

PD184352-treated SK-MEL-5 cellsMeanSEMSDN
0 ng/ml HGF30.39.79173
6.25 ng/ml HGF58.112.722.03
12.5 ng/ml HGF65.84.527.833
25 ng/ml HGF80.10.661.143
50 ng/ml HGF89.73.195.533

• Standard deviation was calculated using the formula, SD = SEM*(SQRT n)

Test family

  • ANOVA: fixed effects, omnibus, one-way, alpha error = 0.05

  • Power calculations were performed from effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

  • ANOVA F statistic calculated with Graphpad Prism 6.0

  • Partial η2 calculated from Lakens (2013)

Power calculations for replication

GroupsF test statisticPartial η2Effect size fA priori powerTotal sample size
0, 6.25, 12.5, 25, and 50 ng/ml HGFF(4,10) = 9.04930.78351.90235186.8%a10a (5 groups)
  • A total sample size of 15 will be used as a minimum making the power 99.9%.

Protocol 5

Summary of original data

Note: numbers were shared by original authors.

Stromal cellsBRAF/MEK inhibitorMET inhibitorMeanSEMSDN
NoneVehicleVehicle1000.000.003
NoneVehicleCrizotinib97.40.390.703
NoneVehiclePHA-66575297.4a0.39a0.70a3a
NonePLX4720Vehicle32.210.818.73
NonePLX4720Crizotinib28.48.8115.33
NonePLX4720PHA-66575228.4a8.81a15.3a3a
NonePD184352Vehicle24.413.222.93
NonePD184352Crizotinib26.62.734.703
NonePD184352PHA-66575226.6a2.73a4.70a3a
LL 86VehicleVehicle99.22.053.603
LL 86VehicleCrizotinib99.13.636.303
LL 86VehiclePHA-66575299.1a3.63a6.30a3a
LL 86PLX4720Vehicle91.09.3216.13
LL 86PLX4720Crizotinib33.47.2812.63
LL 86PLX4720PHA-66575233.4a7.28a12.6a3a
LL 86PD184352Vehicle56.911.119.23
LL 86PD184352Crizotinib25.43.646.303
LL 86PD184352PHA-66575225.4a3.64a6.30a3a
CCD-1090SkVehicleVehicle99.70.801.403
CCD-1090SkVehicleCrizotinib100.83.405.903
CCD-1090SkVehiclePHA-665752100.8a3.40a5.90a3a
CCD-1090SkPLX4720Vehicle31.18.4014.53
CCD-1090SkPLX4720Crizotinib27.16.1010.63
CCD-1090SkPLX4720PHA-66575227.1a6.10a10.6a3a
CCD-1090SkPD184352Vehicle23.710.217.73
CCD-1090SkPD184352Crizotinib26.910.117.53
CCD-1090SkPD184352PHA-66575226.9a10.1a17.5a3a
  • All PHA-665752 treatment values were made the same as the corresponding crizotinib treatment as these inhibitors are assumed to have the same effect. PHA-665752 is an additional MET inhibitor added to the experimental design.

• Standard deviation was calculated using the formula, SD = SEM*(SQRT n)

Test family

  • 3-way ANOVA between subjects: fixed effects, special, main effects and interactions, alpha error = 0.05

    • Power calculations were performed from effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

    • ANOVA F statistic calculated with R software 3.1.1 (R Core Team, 2014)

    • Partial η2 calculated from Lakens (2013)

Power calculations for replication

GroupsF test statisticPartial η2Effect size fA priori powerTotal sample size
All 27 groupsF(8,54) = 3.6903 (interaction)b0.35350.73945380.4%a43a (27 groups)
  • A total sample size of 162 will be used based on the planned comparison calculations making the power 99.9%.

  • 10,000 simulations were run using the summary data to randomly assign data values and the interaction F statistic was computed for a 3-way ANOVA between subjects design. The average F statistic was calculated and used in the power calculations.

Test family

  • 2-way ANOVA between subjects: fixed effects, special, main effects, and interactions, alpha error = 0.05

    • Power calculations were performed from effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

    • ANOVA F statistic calculated with Graphpad Prism 6.0

    • Partial η2 calculated from Lakens (2013)

Power calculations for replication (BRAF/MEK inhibitor)

GroupsF Test statisticPartial η2Effect size fA Priori powerTotal sample size
All 9 vehicle groupsF(4,18) = 0.9381 (interaction)0.17250.495450a80.0%a54a (9 groups)
All 9 vehicle groupsF(2,18) = 0.5678 (main effect: stromal cells)0.05930.436865a80.0%a54a (9 groups)
All 9 vehicle groupsF(2,18) = 0.9546 (main effect: MET inhibitor)0.09590.436865a80.0%a54a (9 groups)
All 9 PLX4720 groupsF(4,18) = 4.7285 (interaction)0.51241.02511582.1%b19b (9 groups)
All 9 PD184352 groupsF(4,18) = 1.8076 (interaction)0.28660.63382880.8%c36c (9 groups)
  • A sensitivity calculation was performed since the original data showed a non-significant effect with the computed effect size shown that can be detected with 80% power.

  • 54 total will be used based on the planned comparison calculations making the power 99.9%.

  • 54 total will be used based on the planned comparison calculations making the power 96.0%.

Test family

  • • Two tailed t-test; difference between two independent means, Bonferroni’s correction: alpha error = 0.0125.

    • Power calculations were performed for effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

Power calculations for replication (PLX4720 group)

Group 1Group 2Effect size dA Priori powerGroup 1 sample sizeGroup 2 sample size
No stromal cells treated with PLX4720 and vehicleLL 86 stromal cells treated with PLX4720 and vehicle3.36991883.8%a4a4a
LL 86 stromal cells treated with PLX4720 and vehicleLL 86 stromal cells treated with PLX4720 and crizotinib3.98441894.2%b4b4b
LL 86 stromal cells treated with PLX4720 and vehicleCCD-1090Sk stromal cells treated with PLX4720 and vehicle3.90969293.4%c4c4c
LL 86 stromal cells treated with PLX4720 and vehicleLL 86 stromal cells treated with PLX4720 and PHA-6657523.98441894.2%d4d4d
  • 6 per group will be used based on the PD184352 planned comparisons making the power 99.1%.

  • 6 per group will be used based on the PD184352 planned comparisons making the power 99.9%.

  • 6 per group will be used based on the PD184352 planned comparisons making the power 99.9%.

  • 6 per group will be used based on the PD184352 planned comparisons making the power 99.9%.

Test family

  • • Two tailed t-test; difference between two independent means, Bonferroni’s correction: alpha error = 0.025.

    • Power calculations were performed for effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

Power calculations for replication (PD184352 group)

Group 1Group 2Effect size dA Priori powerGroup 1 sample sizeGroup 2 sample size
LL 86 stromal cells treated with PD184352 and vehicleLL 86 stromal cells treated with PD184352 and crizotinib2.20455086.0%66
LL 86 stromal cells treated with PD184352 and vehicleLL 86 stromal cells treated with PD184352 and PHA-6657522.20455086.0%66

Protocol 6

Summary of original data

Note: numbers were estimated from bar chart in Supplemental Figure S19.

pAKT

Growth FactorBRAF/MEK inhibitorMeanSEMSDN
VehicleVehicle1.000.000.003
VehiclePD1843520.840.330.573
VehiclePLX47201.220.581.003
HGFVehicle5.110.560.973
HGFPD18435211.565.229.043
HGFPLX47209.112.113.653
pERK

Growth FactorBRAF/MEK inhibitorMeanSEMSDN
VehicleVehicle1.000.000.003
VehiclePD1843520.000.000.003
VehiclePLX47200.080.070.123
HGFVehicle1.600.120.213
HGFPD1843520.390.210.363
HGFPLX47201.610.651.133
pMET(Tyr1349)

Growth FactorBRAF/MEK inhibitorMeanSEMSDN
VehicleVehicle1.000.000.003
VehiclePD1843522.912.003.463
VehiclePLX47202.872.915.043
HGFVehicle9.445.118.853
HGFPD18435216.446.5811.403
HGFPLX472013.735.679.823

• Standard deviation was calculated using the formula, SD = SEM*(SQRT 3)]

Test family

  • 2-way ANOVA between subjects: fixed effects, special, main effects, and interactions, alpha error = 0.05

    • Power calculations were performed from effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

    • ANOVA F statistic calculated with Graphpad Prism 6.0

    • Partial η2 calculated from Lakens (2013)

Power calculations for replication

GroupsF Test statisticPartial η2Effect size fA Priori powerTotal sample size
pAKTF(1,12) = 15.9141 (main effect: growth factor)0.57011.15157485.9%a11a (6 groups)
pERKF(1,12) = 13.0042 (main effect: growth factor)0.52011.04104285.0%b12b (6 groups)
pERKF(2,12) = 7.5790 (main effect: BRAF/MEK inhibitor)0.55811.12381382.9%c13c (6 groups)
pMET (Tyr1349)F(1,12) = 9.4520 (main effect: growth factor)0.44060.88748582.8%d14d (6 groups)
  • 36 total will be used based on the planned comparison calculations making the power 99.9%.

  • 36 total will be used based on the planned comparison calculations making the power 99.9%.

  • 36 total will be used based on the planned comparison calculations making the power 99.9%.

  • 36 total will be used based on the planned comparison calculations making the power 99.9%.

Test family

  • Two tailed t-test; difference between two independent means, Bonferroni’s correction: alpha error = 0.05.

    • o Note: calculations were performed for effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

Power calculations for replication (pAKT group)

Group 1Group 2Effect size dA priori powerGroup 1 sample sizeGroup 2 sample size
pAKT, vehicle, PLX4720pAKT, HGF, PLX47202.94395893.1%a4a4a
  • 6 per group will be used based on the pERK planned comparisons making the power 99.5%.

  • Note: HGF/PD184352 compared to vehicle/PD184352 is not included as the number of needed samples is too large.

Test family

  • Two tailed t-test; difference between two independent means, Bonferroni’s correction: alpha error = 0.05.

    • o Note: calculations were performed for effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

Power calculations for replication (pERK group)

Group 1Group 2Effect size dA priori powerGroup 1 sample sizeGroup 2 sample size
pERK, vehicle, PLX4720pERK, HGF, PLX47201.91085984.6%66
  • Note: HGF/PD184352 compared to vehicle/PD184352 is not included as the number of needed samples is too large.

Test family

  • Two tailed t-test; difference between two independent means, Bonferroni’s correction: alpha error = 0.05.

    • o Note: calculations were performed for effects reported in the original study using G*Power software (version 3.1.7) (Faul et al., 2007).

Power calculations for replication (pMET(Tyr1349) group)

Group 1Group 2Effect size dA priori powerGroup 1 sample sizeGroup 2 sample size
All 3 vehicle (no HGF) conditionsAll 3 HGF conditions1.58154583.7%a8a (3 conditions)8a (3 conditions)
  • 18 per group (6/condition) will be used based on the pERK planned comparisons making the power 99.6%.

References

Acknowledgements

The Reproducibility Project: Cancer Biology core team would like to thank the original authors, in particular Ravid Straussman and Michal Barzily-Rokni, for generously sharing critical information and reagents to ensure the fidelity and quality of this replication attempt. We would also like to thank the following companies for generously donating reagents to the Reproducbility Project: Cancer Biology: American Type Culture Collection (ATCC), BioLegend, Charles River Laboratories, Corning Incorporated, DDC Medical, EMD Millipore, Harlan Laboratories, LI-COR Biosciences, Mirus Bio, Novus Biologicals, and Sigma–Aldrich.

Decision letter

Charles L Sawyers, Reviewing editor, Memorial Sloan-Kettering Cancer Center, United States

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

Thank you for sending your work entitled “Registered report: Tumour micro-environment elicits innate resistance to RAF inhibitors through HGF secretion” for consideration at eLife. Your Registered report has been reviewed by Charles Sawyers, Ravid Straussman as one of the original authors, and a biostatistician.

Charles Sawyers has assembled the following comments to help you prepare a revised submission.

All the reviewers agree that you have appropriately identified the most salient features of Straussman et al. for replication and that the replication experiments are well designed. Protocols 3, 4 and 5 are the key experiments (stromal conditioned media rescue, recombinant HGF rescue and inhibition of rescue with crizotinib). One reviewer felt that protocol 6 (survey of other signaling pathways activated by HGF) was optional.

1) Two reviewers felt that more attention should be given to the Lezcano et al., 2014 publication that reportedly failed to confirm a correlation between HGF expression and outcome (Figure 3 in Straussman et al.).

Specifically:

a. It should be noted in the text that the same group did replicate some other key findings of Straussman et al. – the presence of HGF in human melanoma tumors (in both melanoma cells and stromal cells) and the finding that HGF is significantly enhanced in disease progression.

b. The testing for a correlation between pre-treatment HGF and clinical outcome was done by Lezcano et al. using a cohort of 23 pre-treatment samples. While we fully support the claim by Lezcano et al. that “rigorous validation studies are thus indicated for approaches that seek to personalize such therapies to maximize therapeutic efficacy,” we wonder if testing of 23 samples can be considered as rigorous. As no power calculations are mentioned in Lezcano et al., we would like to see some discussion of whether Lezcano et al. were sufficiently powered to make positive or negative associations. If not, how large would the sample sizes need to be?

c. Wilson et al. (PMID: 22763448) tested the correlation between plasma HGF and PFS/OS on 126 melanoma patients and did find a statistically significant negative correlation that supports the findings in Straussman et al. As this is the only available big cohort testing HGF and clinical outcome on BRAFi, this should be adding it to the literature summary in the introduction.

2) We are aware of 2 groups that have directly replicated several of the in vitro experiments of the paper and have published some results. These should be added these to the literature summary.

a. A group from Amgen attempted to directly replicate the key findings from Straussman et al. Their findings can be found here: http://cancerres.aacrjournals.org/cgi/content/meeting_abstract/73/8_MeetingAbstracts/3405. They show that HGF can rescue melanoma cell lines from BRAFi and MEKi and that this rescue is attenuated by METi.

b. A group from the University of Illinois was able to demonstrate that c-MET inhibition is synergistic with BRAF inhibition in melanoma cell lines: http://cancerres.aacrjournals.org/cgi/content/meeting_abstract/73/8_MeetingAbstracts/2078

3) Regarding statistical power, we also have the following suggestion:

While it is very useful for you to leverage the previously reported effects to compute minimum power a priori, what you really need is to guarantee a minimum power on your own data. This can be done, a priori, by including some cross-study variation. This will be helpful for you to plan on the number of replicates and so forth. Papers by Giovanni Parmigiani and collaborators at the Dana Farber provide some estimates about cross-study variation that could be used for this purpose. Worst case, you should budget some additional variability because of cross-study reproducibility, and increase the sample size as appropriate. We also want you to compute and report power post-hoc/on-the-fly on your own data. Some minimum power should be guaranteed using summaries of your own data.

Comments on the specific protocols:

Protocols 1 and 2 - We think that the protocols are mixed and experiment detailed in protocol 2 should be protocol 1 and vice versa. This should be corrected. Below we refer to the protocols as they appear in the file that we received.

Protocol 1

• When growing SK-MEL-5-GFP cells make sure that >85% of cells are GFP labeled. If number of GFP positive cells are dropping one can use FACS or antibiotics to enrich again for GFP positive cells. We did not grow the cells under antibiotic selection on a regular basis.

• Microplate reader used is different from original and should be labeled with a *.

• We used Corning #3712 plates and did specify that in the methods section. Please remove the * and remove the comment: “Original unspecified”.

• 1c – as specified in the methods section we maintained cells in DMEM from Invitrogen (#10569-010). While using phenol-red free DMEM for the screens is a good idea (we did the same) I would recommend supplementing it with sodium pyruvate as the DMEM that we used had Sodium pyruvate in it. When using Phenol-red free media we used to add Sodium pyruvate from Cellgro (Cat #25-000-CI) to a final of 1 mM.

• 1d - we have plated cells on 384-well plates using the Combi cell platter (http://www.thermo.com.cn/Resources/201306/21143420640.pdf). This resulted in very accurate plating. I don't know how the replicating lab is planning to plate cells on 384-well plates. If manual plating is planned make sure that no air bubble is present at the bottom of the well after plating as this can frequently occur for those unexperienced with manual plating of 384-well plates.

Protocol 2

• Read GFP only after cells have completely settled down. As indicated in the paper we used to plate cells on day 0 and read GFP for the first time on day 1.

• Read GFP from wells with media and no cells as well. Before analyzing results subtract reading from clear–media wells from wells that have cells. We noticed that reading from media-only wells can change from day from day and thus subtract the reading from media-only wells from wells with cells. To this end we always make sure to have media-only wells on each plate with a total volume that is equivalent to test-wells. This remark is true to all experiments in all protocols.

Protocol 3

• 2b - Seems like 50ul and not 60ul is a better control for the wells that will have 20ul of cancer cells +20ul of PCM +10ul of drug.

• From the protocol it seems like stromal cells are plated once. I have plated stromal cells 3 times (each time 3 days before I needed it) to make sure that I have fresh PCM on days 0,1 & 4.

• This protocol involves a few cycles of media change in 384-well plates. We have done so using a CyBi robotic liquid handler. Do the replicating lab plan to use a robotic liquid handler? From my experience it is not easy to take out the exact same amount of media from 384-wells manually making sure not to touch the bottom and disturb the cells. If a robotic plate handler can be used I would recommend using it, as manual handling of hundreds of 384-wells might be a source of a lot of noise in the experiment. Lastly - both extraction and addition of liquid from the wells should be done gently. Cells are under the treatment of BRAFi and might be displaced more easily that non-treated cells. If using a robotic system please do not exceed a rate of 5-10μl/s and do not let the tip end closer than 1 mm to the well bottom.

• 8a - Subtract the reading from media only wells first and only then subtract reading of day 1 from day 7. This remark is true for all protocols.

Protocol 4

• 1a – please correct number to match your planned 2500 cells/well.

• 5 – I think a step is missing in which all media will be taken out, 40ul of fresh media added and only then HGF and drugs are added again.

Protocol 5

• 4 – PLX4720 must be diluted to 20 mM or less before diluted into media. This remark is true to all other planned experiments.

Protocol 6

• We used media with phenol-red for these experiments.

• 2 – On day 1 we did not change media to fresh. We only added drugs/HGF as indicated.

• Again – Make sure that stocks of PLX4720 are not over 20 mM when added to media.

• 3 – Cells were washed with cold PBS quickly on ice. Lysis buffer was added to the wells on ice. Cells were scraped and cell extract moved into Eppendorf tube.

DOI: http://dx.doi.org/10.7554/eLife.04034.002

Author response