1. Human Biology and Medicine
Download icon

Registered report: IDH mutation impairs histone demethylation and results in a block to cell differentiation

  1. Adam D Richarson
  2. David A Scott
  3. Olga Zagnitko
  4. Pedro Aza-Blanc
  5. Chih-Cheng Chang
  6. David A Russler-Germain
  7. Reproducibility Project: Cancer Biology Is a corresponding author
  1. Sanford Burnham Prebys Medical Discovery Institute, United States
  2. Washington University School of Medicine, United States
Registered Report
Cited
1
Views
939
Comments
0
Cite as: eLife 2016;5:e10860 doi: 10.7554/eLife.10860

Abstract

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 Altmetric scores (Errington et al., 2014). This Registered Report describes the proposed replication plan of key experiments from “IDH mutation impairs histone demethylation and results in a block to cell differentiation” by Lu and colleagues, published in Nature in 2012 (Lu et al., 2012). The experiments that will be replicated are those reported in Figures 1B, 2A, 2B, 2D and 4D. Lu and colleagues demonstrated that expression of mutant forms of IDH1 or IDH2 caused global increases in histone methylation and increased levels of 2 hydroxyglutarate (Figure 1B). This was correlated with a block in differentiation (Figures 2A, B and D). This effect appeared to be mediated by the histone demethylase KDM4C (Figure 4D). The Reproducibility Project: Cancer Biology is a collaboration between the Center for Open Scienceand Science Exchange, and the results of the replications will be published by eLife.

https://doi.org/10.7554/eLife.10860.001

Introduction

Mutations in the metabolic proteins IDH1 and IDH2 are associated with gliomas, acute myeloid leukemias, chondrosarcomas, intrahepatic cholangiocarcinomas, lymphomas, melanomas and colon, thyroid and prostate cancers (for review, see Krell et al., 2013). Previous work has shown that these mutations change the specificity of the reaction catalyzed by IDH proteins; instead of producing α-ketoglutarate from isocitrate, they produce 2-hydroxyglutarate (2HG), a metabolite that can have oncogenic effects (Krell et al., 2013; McKenney and Levine, 2013; Ward et al., 2010; Xu et al., 2011; Zhang et al., 2013). Lu and colleagues expand upon this work to identify a potential mechanism for how 2HG can effect major changes in cell behavior. They present evidence that 2HG interferes with global demethylation that is required for progenitor cells to complete terminal differentiation. Transfection of 3T3-L1 cells with the mutant forms of IDH1 and IDH2 that produce 2HG lead to an increase in global methylation levels and prevented normal in vitro differentiation into adipocytes. The 2HG-sensitive histone demethylase KDM4C appeared to be required for this process, as knockdown of KDM4C recapitulated the phenotype of 2HG production. Examination of glioma samples showed a correlation between IDH mutation status and level of overall methylation (Lu et al., 2012). Taken together, Lu and colleagues’ findings help explain how mutations in IDH1 and IDH2 potentially interface with cancer development and progression.

In Figure 1B, Lu and colleagues examined the effects of mutations in IDH1 and IDH2 on global levels of methylation by transfecting mutant and wild type forms of the genes into 293T cells and using Western blot to assess the levels of various methylation markers. They also confirmed that introduction of the mutated forms of IDH1 and IDH2 correlated with increased intracellular levels of the oncometabolite 2HG. Their findings suggest that mutations in IDH1 and IDH2 correlate with increased levels of many methylation markers, and this key finding is replicated in Protocol 1.

In order to understand the effects of hypermethylation more fully, Lu and colleagues turned to an in vitro model of differentiation; when treated with appropriate signals, 3T3-L1 cells undergo epigenetic changes required for them to differentiation into adipocytes. In Figure 2A and B, they transfect undifferentiated 3T3-L1 cells with the wild type and mutant forms of IDH2 and assess the cells’ ability to differentiate into adipocytes, as determined by staining for lipid droplets with Oil-Red-O. While differentiated 3T3-L1 cells transfected with vector only or wild type IDH2 showed robust Oil-Red-O staining, cells transfected with mutant IDH2 did not, indicating a block in differentiation. qRT-PCR confirmed that cells transfected with mutant IDH variants did not express high levels of known adipocyte markers (Figure 2D). This key finding will be replicated in Protocol 2.

Lu and colleagues identified a histone demethylase, KDM4C, expressed as 3T3-L1 differentiation progressed, that appeared to be sensitive to 2HG. In Figure 4D, they use siRNAs to knock down levels of KDM4C in differentiating 3T3-L1 cells. Western blot analysis and Oil-Red-O staining confirmed that loss of KDM4C increased global methylation levels and inhibited differentiation. This key finding will be replicated in Protocol 3.

Several aspects of Lu’s findings have been corroborated by other work. Multiple groups have demonstrated that perturbations in IDH proteins alter methylation levels; overexpression of the IDH1R132H allele in human tumor cells lines increased global histone methylation levels (Duncan et al., 2012), exogenous IDH2R140Q increased methylation levels in erythroleukemia progenitor cells (Kernytsky et al., 2015) and an immortalized astrocyte cell line expressing IDH1R132H also demonstrated increased levels of methylation (Turcan et al., 2012). Members of the Thompson lab (authors of this study) have confirmed that expression of mutant variants of IDH proteins in 3T3-L1 cells blocked differentiation into adipocytes (Londono Gentile et al., 2013; Ward et al., 2013). Sasaki and colleagues have shown that mutant IDH1 expression increased levels of methylation in mice (Sasaki et al., 2012), while Akbay and colleagues published a similar observation for mutant forms of IDH2 (Akbay et al., 2014). This effect may even hold true for human patients, as there is a marked increase in H3K9me3 levels associated with IDH mutations in oligodendromas and high grade astrocytomas (Venneti 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. 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: Assessing the methylation status and 2HG production of 293T cells transfected with mutant forms of IDH1 and IDH2

This protocol describes how to transfect 293T cells with wild-type and mutant forms of IDH1 and IDH2 and assess levels of global methylation and 2HG production, as seen in Figure 1B and Supplemental Figure 1.

Sampling

  • Experiment will be repeated a total of 6 times for a minimum power of 80%. The metabolite data is qualitative, thus to determine an appropriate number of replicates to initially perform, sample sizes based on a range of potential variance was determined.

    • See "power alculations' for details.

    • The metabolite data displayed in the bottom of Figure 1B were derived from Figure 3B of Figueroa and colleagues (Figueroa et al., 2010).

  • Each experiment consists of five cohorts:

    • Cohort 1: 293T cells transfected with vector only

    • Cohort 2: 293T cells transfected with wild-type IDH1

    • Cohort 3: 293T cells transfected with IDH1R132H

    • Cohort 4: 293T cells transfected with wild type IDH2

    • Cohort 5: 293T cells transfected with IDH2R172K

  • Each cohort is then examined for methylation status by Western blot and levels of 2HG production by GC-MS.

Materials and reagents

ReagentTypeManufacturerCatalog #Comments
10 cm tissue culture dishesLabwareThermo Scientific130182Original unspecified
Bradford Assay KitReporter assayBio-Rad500-0201EDUOriginal unspecified
DMEMCell cultureCorning15013 CVReplaces original from Invitrogen
Endo-free plasmid maxiprep kitKitQiagen12362Original unspecified
Fetal bovine serum (FBS)Cell cultureCellGro10437-028Original cat # unspecified
293T cellsCell lineATCCCRL-3216Original source unspecified
HRP-conjugated donkey
anti-rabbit secondary
AntibodyGE HealthcareNA934V
HRP-conjugated sheep
anti-mouse secondary
AntibodyGE HealthcareNA931V
IDH1 ORF cloneNucleic acidOriGeneRC210582Replaces ATCC plasmid
in pCMV-Sport6
IDH1R132H ORF cloneNucleic acidOriGeneRC400096Original generated
by authors
IDH2 ORF cloneNucleic acidOriGeneRC201152Replaces Invitrogen
plasmid in pOTB7
IDH2R172K ORF cloneNucleic acidOriGeneRC400103Original generated
by authors
Mouse IgG1 monoclonal
anti-IDH2
AntibodyAbcamAb55271
Nitrocellulose
membrane
Western blot reagentLife TechnologiesLC2006Original source unspecified
Nonfat milkWestern blot reagentCarnationOriginal source unspecified
NuPAGE 4-12% precast
gradient gel
Western blot reagentInvitrogenWG1401BOXOriginal source unspecified
Pierce™ ECL Plus
Western Blotting Substrate
Western blot reagentLife Technologies32132Original unspecified
pLPC vector plasmid
(pLPC H-Ras V12)
Nucleic acidAddgene18741Original source unspecified
Ponceau stainChemicalSIGMAP7170-1LOriginal unspecified
Protease Inhibitor
Cocktail,
InhibitorSigma-AldrichP8340Original unspecified
Rabbit IgG monoclonal
anti-H3
AntibodyCell Signaling Technology4499
Rabbit monoclonal
anti-H3K4me3
AntibodyMillipore17-614
Rabbit polyclonal
anti-H3K36me3
AntibodyAbcamAb9050
Rabbit polyclonal
anti-H3K79me2
AntibodyCell Signaling Technology9757
Rabbit polyclonal
anti-H3K9me2
AntibodyCell Signaling Technology9753
Rabbit polyclonal
anti-H3K9me3
AntibodyAbcamAb8898
Rabbit polyclonal
anti-IDH1
AntibodyProteinTech12332-1-AP
TBS + Tween 20BufferFisher ScientificBP-2471-1Original source unspecified
XCell II blot moduleInstrumentLife TechnologiesEI9051Original unspecified
Acetonitrile, HPLC
grade
ChemicalSpectrumHP412Original source unspecified
ChloroformChemicalFisherC606-4Original unspecified
D-alpha-hydroxyglutaric
acid disodium salt (2HG)
ChemicalSanta Cruz BiotechnologySc-227739Replaces original from Sigma
Methanol, HPLC gradeChemicalMP300141Original source unspecified
N-methyl-N-tert-butyldimethylsilyl
trifluoroacetamide (MTBSTFA)
ChemicalSoltec VenturesGC102Replaces original
from Regis
NorvalineChemicalSigmaN7627Original unspecified
Protein Concentration
Assay; Quick Start
Bradford Assay
Reporter assayBio-Rad500-0205Original unspecified
Lipofectamine 2000Cell cultureLife Technologies11668027Original cat # unspecified

Procedure

Note: 293T cells are maintained in DMEM with 10% FBS at 37°C/5% COAll cells will be sent for STR profiling and mycoplasma testing.

  • Using the pLPC backbone and the OriGene ORF clones, clone in the sequences for wild-type IDH1, wild-type IDH2, IDH1R132H or IDH2R172K to generate the following vectors:

    1. pLPC-IDH1

    2. pLPC-IDH2

    3. pLPC-IDH1R132H

    4. pLPC-IDH2R172K

  • Prep each vector using an endo-free maxiprep kit according to the manufacturer’s instructions.

  • Confirm plasmid identity by sequencing insert and vector integrity by agarose gel electrophoresis.

    1. Note; OriGene ORF clones are shipped with sequencing primers.

  • Plate 293T cells in #10 cm tissue culture dishes and let adhere overnight.

    1. Plate two plates; one will be harvested for Western blot (Step 3), the other for metabolite analysis (Step 4).

  • Transfect 293T cells with appropriate plasmids with Lipofectamine 2000 according to the manufacturer’s instructions.

    1. Note: Prepare separate transfection mixtures for each replicate, then use the same mixture for all plates within that replicate; do not use a single large volume for transfection mixture for all replicates.

      1. pLPC (empty vector)

      2. pLPC-IDH1

      3. pLPC-IDH2

      4. pLPC-IDH1R132H

      5. pLPC-IDH2R172K

    2. Incubate for 3 days.

      1. At this point, the matched plates for each replicate will be harvested; one plate for Western blot analysis (Step 6), the matched plate for GC-MS analysis (Step 7).

    3. Note; from this point forward, the analysis of each replicate must be conducted separately and independently from the other replicates. For example, each replicate should be run on its own gels.

  • Western blot analysis of methylation status:

    1. Acid extraction of histones:

      1. Lyse cells in hypotonic lysis buffer for 1.

        1. Hypotonic lysis buffer: 10mM HEPES, 10mM KCl, 1.5mM MgCl2, 0.5mM DTT, #protease inhibitor cocktail

      2. Add H2SO4 to 0.2N and incubate at 4°C overnight with rotation.

      3. #Centrifuge samples at 6,500xg for 10min at 4°C to pellet debris.

      4. Precipitate proteins with 33% TCA.

      5. Wash with acetone.

      6. Resuspend in de-ionized water.

    2. *#Quantify protein concentration using a Bradford Assay.

    3. #Load ~ 50 µg of protein per well and separate proteins on a 10% NuPAGE 4-12% gradient gel.

    4. #Transfer to nitrocellulose membrane using an XCell II blot module at 25 V for 1-2 hr (start at 100 mA per gel).

    5. *Perform a Ponceau stain and image to confirm transfer of proteins.

      1. Wash out Ponceau.

    6. Block membrane for #1 hr in 5% non-fat milk in PBS with 0.5% Tween-20.

    7. Incubate with primary antibodies #diluted in TBST + 1% nonfat milk at 4°C overnight. *Use the manufacturer’s recommended dilution.

      1. Anti-IDH1

      2. Anti-IDH2

      3. Anti-H3K4me3

      4. Anti-H3K9me2

      5. Anti-H3K9me3

      6. Anti-H3K36me3

      7. Anti-H3K79me2

      8. Anti-H3 (loading control)

        1. #Each antibody will have its own gel run. Membranes will not be stripped and reprobed.

    8. Wash membrane #twice with TBST for a total of 20 min.

      1. Incubate with HRP-conjugated secondary antibodies #diluted in TBST for 1 hr at room temperature. #*Use manufacturer recommended dilutions.

    9. Wash three times with TBST.

    10. Detect signal #using ECL plus according to the manufacturer’s instructions.

    11. Quantify band intensities with ImageJ.

      1. Normalize methylation band intensities to total H3.

      2. Divide normalized band intensities by the vector control band intensity.

  • Gas chromatography-mass spectrometry analysis of 2HG levels. Note; the data in the original paper and the methodology are derived from Figueroa and colleagues (Figueroa et al., 2010).

    1. Gently remove culture medium from cells 3 days after transfection, #wash cells quickly three times with 2 ml ice-cold PBS, and add #0.5 ml ice-cold 80% methanol containing 20 µM L-norvaline per well of a 6-well plate to the cells.

      1. #Quantify protein concentration using the #Bio-Rad Quick Start Bradford Assay.

    2. Incubate 20 min at -80°C.

    3. Centrifuge at 14000xg for 20 min at 4°C.

      1. #Counter-extract samples with chloroform to remove nonpolar metabolites.

    4. Collect supernatant and dry using a #MiVac.

    5. Redissolve dried extracts in #60 µL of a 1:1 mixture of acetonitrile and N-methyl-N-tert-butyldimethylsilyltrifluoroacetamide (MTBSTFA).

    6. Heat the samples for 75 min at 70°C.

    7. GC-MS analysis:

      1. #A Shimadzu QP2010 Plus GC-MS is programmed with an injection temperature of 250°C, injection split ratio 1/10, with injection volume 0.3-1 µl. GC oven temperature starts at 130°C for 4 min, rising to 243°C at 6°C/min and to 280°C at 60°C/min with a final hold at this temperature for 2 min. GC flow rate with helium carrier gas was 50 cm/s. The GC column used is a 15 m x 0.25 mm x 0.25 µm Rxi-5ms (Restek). GC-MS interface temperature is 300°C and (electron impact) ion source temperature is 200°C, with 70 V/ 70 µA ionization voltage/ current. The mass spectrometer is set to scan m/z range 150-600, with ~1 kV detector sensitivity (modified as necessary).

    8. *#In parallel to the sample, run a standard curve of known amounts of 2HG.

    9. Confirm and *#quantify 2HG metabolite peak using standard curve.

    10. Analyze and *#quantify 2HG and glutamate signal (identified by elution time and mass fragment pattern) intensities by integration of peak areas.

  • Repeat independently from Step 4 onwards an additional five times.

Deliverables

  • Data to be collected:

    • Sequence files and agarose gel images confirming vector identity

    • Full gel images of western blots with ladder (as seen in Figure 1B)

      • Images of Ponceau stained membranes

    • Quantification of band intensities (as seen in Supplemental Figure 1A)

    • GC-MS data

    • Quantification of signal intensities of 2HG and glutamate (as seen in Figure 1B)

Confirmatory analysis plan

  • Statistical Analysis of the Replication Data: Note: At the time of analysis we will calculate Pearson’s r to check for correlation between the six dependent variables, normalized intensities measured for each of the histone lysine methylations, for the Western blot data. We will also perform the Shapiro-Wilk test and generate a quantile-quantile plot to assess the normality of the Western blot data and 2HG/glutamate ratios. We will also perform Levene’s test to assess homoscedasticity. If the data appear skewed we will perform a log transformation in order to proceed with the proposed statistical analysis. If the log transformation does not result in similar variance across groups, we will perform the equivalent non-parametric test listed in Power Calculations for this protocol.

    • Western blot:

      • MANOVA (six dependent variables are the normalized intensities for each of the histone lysine methylations; four independent variables are the IDH1 and IDH2 variants (all normalized to vector) with the following planned comparisons using Bonferroni’s correction:

        • Wild-type IDH1 compared to IDH1R132H, for H3K9me2.

        • Wild-type IDH2 compared to IDH2R172K, for H3K9me2.

    • 2HG/glutamate ratios:

      • One-way ANOVA (one dependent variable is the 2HG/glutamate ratio; four independent variables are the IDH1 and IDH2 variants) with the following planned comparisons using Bonferroni’s correction:

        • IDH1WT compared to IDH1R132H

        • IDH2WT compared to IDH2R172K

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

    • For Western blot:

      • The 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 meta-analytic approach to combine the original and replication effects, which will be presented as a forest plot.

    • For 2HG/glutamate ratios:

      • The replication data will be presented as a mean with 95% confidence intervals and will include the original data point, calculated directly from the representative image, as a single point on the same plot for comparison.

  • Additional exploratory analysis:

    • Correlation analysis (Pearson’s r) of each of the six relative histone methylation levels to 2HG/glutamate levels using Bonferroni ‘s correction (as seen in Supplemental Figure 1B).

Known differences from the original study

  • The replication attempt will quantify total amounts of 2HG in addition to the ratio of 2HG to glutamate.

  • Aspects of the Western blot protocol are provided by the replicating lab; complete details of the original protocol were unavailable.

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/vfsbo/).

  • Sequence files and agarose gel images confirming vector identity and integrity

  • Ponceau stains confirming protein transfer to membranes

  • STR profiling and mycoplasma testing results

Protocol 2: Examining the effects of mutations in IDH2 on differentiation of 3T3-L1 cells

This protocol describes how to induce the differentiation of 3T3-L1 cells into adipocytes, which involves extensive chromatin remodeling, after transfection with wild type and mutant forms of IDH2 and assess the level of differentiation by Oil-Red-O staining, as seen in Figure 2A and B, and adipocyte marker expression, as seen in Figure 2D.

Sampling

  • This experiment will use 5 biological replicates for a minimum power of 80%. The metabolite data is qualitative, thus to determine an appropriate number of replicates to initially perform, sample sizes based on a range of potential variance was determined.

    • See Power Calculations for details.

  • Each experiment will consist of three cohorts:

    • Cohort 1: 3T3-L1 cells transduced with vector

    • Cohort 2: 3T3-L1 cells transduced with wild-type IDH2

    • Cohort 3: 3T3-L1 cells transduced with IDH2R172K

  • Each cohort will have 5 plates per biological replicate:

    • One plate will be used to assess IDH2 expression by Western blot.

    • The second plate will be used to assess intracellular levels of 2HG.

    • The third plate will be assessed for adipogenesis by Oil-Red-O staining.

    • The fourth and fifth plates will have mRNA harvested for qRT-PCR analysis.

Materials and reagents

ReagentTypeManufacturerCatalog #Comments
IsobutylmethylxanthineInhibitorSigmaI5879Original cat# unspecified
DexamethasoneChemicalSigmaD4902Original cat# unspecified
InsulinGrowth factorSigmaI3536Original cat# unspecified
TroglitazoneChemicalSigmaT2573Original cat# unspecified
pCL-Eco helper plasmidNucleic acidAddgene12371Original source unspecified
293T cellsCell lineATCCCRL-3216Original source unspecified
3T3-L1 cellsCell lineATCCCL-173Original source unspecified
PuromycinChemicalLife TechnologiesA11138-02Original unspecified
RIPA bufferCell cultureMillipore20188Original source unspecified
Nitrocellulose membraneWestern blot reagentLife TechnologiesLC2006Original source unspecified
Ponceau stainChemicalSigmaP7170Original source unspecified
Rabbit polyclonal anti-IDH1AntibodyProteinTech12332-1-AP
Mouse IgG1 monoclonal anti-IDH2AntibodyAbcamAb55271
HRP-conjugated donkey anti-rabbit secondaryAntibodyGE HealthcareNA934V
HRP-conjugated sheep anti-mouse secondaryAntibodyGE HealthcareNA931V
Oil-Red-OChemicalSigmaO1391Original source unspecified
paraformaldehydeChemicalTousimis1008AOriginal source unspecified
6-well tissue culture platesLabwareSarstedt83.1839Original unspecified
XtremeGene HP reagentCell cultureRoche06366244001Original unspecified
DMEMCell cultureCorning15013 CVReplaces original from Invitrogen
FBSCell cultureCellGro10437-028Original cat # unspecified
OPTI-MEMCell cultureLife Technologies31986070Original unspecified
0.45 µm low binding syringe filterLabwareMilliporeSLHV013SLOriginal unspecified
Endo-free maxiprep kitKitQiagen12362Original unspecified
Protein Concentration Assay; Quick Start Bradford AssayReporter assayBio-Rad500-0205Original unspecified
SpectrophotometerInstrumentBeckman CoulterDU800Original unspecified
TRIzolChemicalInvitrogen15596-018Original cat# unspecified
SuperScript II reverse transcriptaseKitInvitrogen18064-014Original cat# unspecified
7900HT Sequence Detection SystemInstrumentApplied Biosystems
Pparg Taqman assays; Hs00234592_m1Nucleic acidApplied BiosystemsCat. # 4351372Original assay unspecified
Cebpa Taqman assays; Hs00269972_s1Nucleic acidApplied BiosystemsCat. # 4331182Original assay unspecified
Adipoq Taqman assays; Hs00605917_m1Nucleic acidApplied BiosystemsCat. # 4331182Original assay unspecified
18S rRNA Taqman assays; Hs99999901_s1Nucleic acidApplied BiosystemsCat. # 4331182Original assay unspecified

Procedure

Note: 3T3-L1 and 293T cells are maintained in DMEM with 10% FBS at 37°C/5% CO2 .All cells will be sent for STR profiling and mycoplasma testing. pLPC (empty vector), pLPC-IDH2, and pLPC-IDH2R172K are generated in Protocol 1.

  • Generate vector and IDH2 wild type and mutant expressing retroviruses.

    1. #Transfect 293T cells with pCL-Eco helper plasmid and vector or IDH2 vectors.

      1. Spot virus construct in 6 well plates at 1000 ng/well.

      2. Perform X-tremeGeneHP reverse transfection as follows:

        1. Make the helper plasmid mix; 700 ng/well.

        2. Add 4 µL of XtremeGene HP to 400 µL OPTI-MEM.

          • Mix by light tapping.

        3. Mix together the helper plasmids with the XtremeGeneHP reagent and OPTI-MEM.

        4. Add 400 µL of the mix to each well and incubate for at least 30 min at room temperature.

        5. Meanwhile, resuspend 293T cells in DMEM + 10% FBS at 1.2x106 cells/ml.

        6. Add 1600 µL of cells to each well.

    2. #24 hr later, replace media (2 ml total).

    3. #48 hr post transfection, collect supernatant from each well.

      1. Centrifuge at 500xg for 10 min at room temperature to pellet debris.

      2. Filter supernatant through a 0.45 µm syringe filter, aliquot and store at -80°C.

  • Transduce 3T3-L1 cells with viral supernatant.

    1. #Seed cells in 6-well plates and incubate overnight.

      1. Cells should be 50-60% confluent the next day.

    2. #Add viral supernatant to medium.

      1. Supernatant will be added at varying concentrations to determine optimal transduction efficiency; 1:5 to 1:10 – 150 to 300 µL per well.

    3. #Adjust media volume to 1.4 ml per well.

    4. #Add polybrene in 100 µL of media into each well for a final concentration of 8 µg/ml.

    5. #Spinoculate by spinning at 1000xg for 60 min at room temperature.

      1. Incubate overnight.

    6. #Change media to remove viral transduction media.

      1. Replace with fresh media.

    7. Grow cells with 2.5 µg/ml puromycin for 7 days to select for stable expression of either wild-type or mutant IDH2.

      1. Maintain cells in puromycin.

      2. Also treat a non-transduced well of 3T3-L1 cells as a control showing susceptibility to puromycin.

      3. Split each biological replicate into 5 plates for the four assays being performed.

        1. Plate 1 is for Western blot

        2. Plate 2 is for GC-MS

        3. Plate 3 is for Oil-Red-O staining (harvested 7 days after differentiation)

        4. Plate 4 and 5 are for qRT-PCR (harvested 0 and 4 days after differentiation)

  • Generate whole cell lysates from the first plate of each cohort:

    1. Lyse cells and sonicate in RIPA buffer.

      1. RIPA buffer: 1% sodium deoxycholate, 0.1% SDS, 1% Triton X-100, 0.01 M Tris pH 8.0 and 0.14 M NaCl

      2. #Sonicate for 1 min, at 180 watts with rounds of 10 sec on/10 sec off. Keep sample on ice during sonication.

    2. Centrifuge lysates at 14000xg for 10 min at 4°C.

    3. Collect supernatant and measure total protein concentration #using a Bradford assay.

    4. Perform Western blot as outlined in Protocol 1 Step 6 using the following primary antibodies *at the manufacturer’s recommended dilution:

      1. Anti-IDH1

      2. Anti-IDH2

  • Harvest the second plate for metabolite analysis by mass spectrometry as described in Protocol 1 Step 4.

  • Induce 3T3-L1 cells to differentiate into adipocytes.

    1. Incubate cells for 2 days with a differentiation cocktail composed of 0.5 mM isobutylmethylxanthine, 1 mM dexamethasone, 5 mg/ml insulin and 5 mM troglitazone supplementing the standard media.

    2. After 3 days, maintain cells with 5 mg/ml insulin until harvested.

  • After 7 days of differentiation, assess adipogenesis by Oil-Red-O staining in the third plate from each cohort.

    1. Wash cells in PBS and fix in 3% paraformaldehyde for 20 min at room temperature.

    2. Wash cells with de-ionized water.

    3. Stain with Oil-Red-O solution according to the manufacturer’s protocol.

    4. Image stained wells by brightfield microscopy and *quantify Oil-Red-O staining by extracting oil-red-o in isopropanol and reading absorbance at 500 nm.

  • Harvest fourth plate for RNA extraction at Day 0 of differentiation and the fifth plate at Day 4 of differentiation and perform qRT-PCR to assess expression levels of adipocyte markers at each time point.

    1. Harvest cells and extract RNA using TRIzol according to the manufacturer’s instructions.

    2. Reverse transcribe RNA and synthesize cDNA using SuperScript II reverse transcriptase according to the manufacturer’s protocol.

      1. Assess purity and concentration of RNA and cDNA spectrophotometrically; record A260/A280 and A260/A230 ratios.

    3. Perform qPCR on a 7900HT Sequence Detection system using Taqman gene expression assays for the following genes:

      1. Pparg

      2. Cebpa

      3. Adipoq

      4. 18S rRNA for normalization.

        1. *Primers sequences and PCR cycling conditions will be optimized.

  • Repeat independently from Step 2 onwards an additional four times.

Deliverables

  • Data to be collected:

    • Whole gel images of Western blots with ladder (as seen in Figure 2A)

    • *Densitometric quantification of bands

      • Also normalized to the loading control.

    • Images of wells stained with Oil-Red-O (as seen in Figure 2B)

    • *Quantification of Oil-Red-O levels for each cohort

    • All raw qRT-PCR data

    • Graph of gene expression over time for each of the three adipocyte markers (as seen in Figure 2D)

Confirmatory analysis plan

  • Statistical Analysis of the Replication Data: Note: At the time of analysis we will calculate Pearson’s r to check for correlation between the three dependent variables, normalized gene expression for each of the adipocyte markers, for the qRT-PCR data. We will also perform the Shapiro-Wilk test and generate a quantile-quantile plot to assess the normality of the qRT-PCR data and 2HG/glutamate ratios. We will also perform Levene’s test to assess homoscedasticity. If the data appears skewed we will perform a log transformation in order to proceed with the proposed statistical analysis. If this is not possible we will perform the equivalent non-parametric test listed in Power Calculations for this protocol.

    • Western Blot:

      • Confirmatory; no analysis performed

    • 2HG/glutamate ratio:

      • One-way ANOVA (one dependent variable is the 2HG/glutamate ratio; three independent variables are the vector and IDH2 variants) with the following planned comparison using Fisher’s LSD correction:

        • IDH2R172Kcompared to IDH2WT

    • qRT-PCR:

      • One-way MANOVA (three dependent variables are the normalized gene expression of each of the adipocyte markers on day 4; three independent variables are the vector and IDH2 variants) with the following planned comparisons using Bonferroni’s correction:

        • IDH2R172compared to vector for each gene (three comparisons total)

        • IDH2R172Kcompared to IDH2WT for each gene (three comparisons total)

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

    • For qRT-PCR:

      • The 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 meta-analytic approach to combine the original and replication effects, which will be presented as a forest plot.

    • For 2HG/glutamate ratios:

      • The replication data will be presented as a mean with 95% confidence intervals and will include the original data point, calculated directly from the representative image, as a single point on the same plot for comparison.

  • Additional exploratory analysis:

    • Oil-Red-O staining:

      • One-way ANOVA (one dependent variable is the A500 readings; three independent variables are the vector and IDH2 variants) with the following planned comparison using Fisher’s LSD correction:

        • IDH2R172K compared to IDH2WT

Known differences from the original study

  • Aspects of the Western blot protocol are provided by the replicating lab; complete details of the original protocol were unavailable.

  • Aspects of the viral production protocol are adapted from the replicating lab’s in-house protocol.

    • Viral supernatant will be collected only at 48 hr post-transection and will not be combined with viral supernatant collected at 72 hr.

  • In addition to imaging Oil-Red-O stained plates, the replication attempt will quantify the amount of Oil-Red-O staining spectrophotometrically.

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/vfsbo/).

  • Sequence files and agarose gel images confirming vector identity and integrity

  • Ponceau stains confirming protein transfer to membranes

  • STR profiling and mycoplasma testing results

  • Absorbance data for RNA and cDNA

Protocol 3: Assessing the role of KDM4C on differentiation of 3T3-L1 cells

This protocol describes how to treat 3T3-L1 cells with an siRNA against the histone demethylase KDM4C, whose activity is inhibited by 2HG, and assess the effect of loss of KDM4C activity on methylation and differentiation, as seen in Figure 4D and Supplemental Figure 8.

Sampling

  • This experiment will be repeated 3 times for a minimum power of 80%. The Western blot data is qualitative, thus to determine an appropriate number of replicates to initially perform, sample sizes based on a range of potential variance was determined.

    • See Power calculations for details.

  • Each experiment consists of five cohorts:

    • Cohort 1: 3T3-L1 cells treated with scramble control siRNAs

    • Cohort 2: 3T3-L1 cells treated with siRNAs #1 against KDM4C

    • Cohort 3: 3T3-L1 cells treated with siRNAs #2 against KDM4C

    • Cohort 4: 3T3-L1 cells treated with siRNAs #3 against KDM4C

    • Cohort 5: untreated 3T3-L1 cells [additional control]

  • Each cohort is induced to differentiate, followed by:

    • Assessment of methylation by Western blot for:

      • Anti-KDM4C

      • Anti-H3K9me3

      • Anti-H3

      • Anti-β-actin

    • Assessment of differentiation by Oil-Red-O staining

Materials and reagents

ReagentTypeManufacturerCatalog #Comments
3T3-L1 cellsCell lineATCCCL-173Original source unspecified
DMEMCell cultureCorning15013 CVReplaces original from Invitrogen
FBSCell cultureCellGro10437-028Original cat # unspecified
KDM4C siRNA #1Nucleic acidSynthesis left to the discretion of the replicating lab
KDM4C siRNA #2Nucleic acidSynthesis left to the discretion of the replicating lab
KDM4C siRNA #3Nucleic acidSynthesis left to the discretion of the replicating lab
Scrambled control siRNANucleic acidDharmaconD-001810-01-20
Lipofectamine RNAiMAXCell cultureInvitrogen13778-030Original cat# unspecified
isobutylmethylxanthineInhibitorSigmaI5879Original cat# unspecified
dexamethasoneChemicalSigmaD4902Original cat# unspecified
insulinGrowth factorSigmaI3536Original cat# unspecified
TroglitazoneChemicalSigmaT2573Original cat# unspecified
RIPA bufferCell cultureMillipore20188Original source unspecified
Nitrocellulose membraneWestern blot reagentLife TechnologiesLC2006Original source unspecified
Ponceau stainChemicalSigmaP7170Original unspecified
Mouse IgG2a monoclonal anti-β-actinAntibodySigmaA5316
Rabbit IgG monoclonal anti-H3AntibodyCell Signaling Technology4499
Rabbit polyclonal anti-H3K9me3AntibodyAbcamAb8898
Rabbit polyclonal anti-KDM4CAntibodyAbcamAb85454
Oil-Red-OChemicalSigmaO1391Original source unspecified
paraformaldehydeChemicalTousimis1008AOriginal source unspecified

Procedure

Note: 3T3-L1 cells are maintained in DMEM with 10% FBS at 37°C/5% CO2. All cells will be sent for STR profiling and mycoplasma testing.

  • Transfect with 3T3-L1 cells with siRNAs against KDM4C:

    1. Plate out equal densities of single cell suspensions of 3T3-L1 cells in #6-well plates.

      1. *Optimize the number of cells to plate per well.

      2. Plate out two plates per siRNA pool (control vs. siKDM4C).

        1. One will be harvested on Day 3 of differentiation for Western blot analysis.

        2. One will be used on Day 7 of differentiation for Oil-Red-O analysis.

    2. Transfect with the following siRNAs at a final concentration of 40 nM using Lipofectamine RNAiMAX according to the manufacturer’s instructions.

      1Sense5’-GCUUGAAUCUCCCAAGAUATT-3’
      Antisense5’-UAUCUU GGGAGAUUCAAGCTT-3’
      2Sense5’-CAAAGUAUCUUGGAUCAAATT-3’
      Antisense5’-UUUGAUCCAAGAUACUUUGCC-3’
      3Sense5’-GAGGAGUU UCGGGAGUUCAACAAAU-3’
      Antisense5’-AUUUGUUGAACUCCCGAA ACUCCUC-3’
      1. Transfect control wells with a scrambled control siRNA.

      2. Also plate control wells with no transfection.

    3. Incubate for 3 days.

  • Induce differentiation of control siRNA and antisense siRNA transduced 3T3-L1 cells as specified in Protocol 2 Step 6.

  • 3 days after differentiation, harvest one plate from each treatment and prepare whole cell lysates as specified in Protocol 2 Step 7.

  • Perform Western blot analysis on all whole cell lysates from Day 3 as described in Protocol 2 Step 7.

    1. Probe with:

      1. Anti-KDM4C

      2. Anti-H3K9me3

      3. Anti-H3

      4. Anti-β-actin

    2. Quantify band intensities with ImageJ.

      1. Normalize H3K9me3 band intensities to total H3.

      2. Normalize KDM4C band intensities to ß-actin.

  • At Day 7 of differentiation, assess level of differentiation by Oil-Red-O staining as specified in Protocol 2 Step 8.

    1. Image wells and quantify Oil-Red-O expression.

  • Repeat experiment an additional two times.

Deliverables

  • Data to be collected:

    • Whole gel images of all Western blots with ladder (as seen in the top of Figure 4D)

    • Images of Oil-Red-O stained wells (as seen in the bottom half of Figure 4D)

  • Quantification of Oil-Red-O staining at Day 7 of differentiation (compare to Supplemental Figure 8B)

Confirmatory analysis plan

  • Statistical Analysis of the Replication Data: Note: At the time of analysis we will calculate Pearson’s r to check for correlation between the two dependent variables, normalized intensities measured for KDM4C and H3K9me3, for the Western blot data. We will also perform the Shapiro-Wilk test and generate a quantile-quantile plot to assess the normality of the Western blot and Oil-Red-O data. We will also perform Levene’s test to assess homoscedasticity. If the data appears skewed we will perform a log transformation in order to proceed with the proposed statistical analysis. If this is not possible we will perform the equivalent non-parametric test listed in Power Calculations for this protocol.

    • Quantification of Oil-Red-O staining:

      • One way ANOVA (one dependent variable is the A500 readings; four independent variables are the control and three KDM4C siRNA sequences) with the following planned comparisons using Bonferroni’s correction:

        • Control siRNA compared to KDM4C siRNA #1

        • Control siRNA compared to KDM4C siRNA #2

        • Control siRNA compared to KDM4C siRNA #3 [additional exploratory analysis]

    • Western blot:

      • One-way MANOVA (two dependent variables are the normalized intensities measured for KDM4C and H3K9me3; four independent variables are the control and three KDM4C siRNA sequences) with the following planned comparisons using Bonferroni’s correction:

        • H3K9me3 levels:

          • Control siRNA compared to KDM4C siRNA #1

          • Control siRNA compared to KDM4C siRNA #2

          • Control siRNA compared to KDM4C siRNA #3

        • KDM4C levels (QC):

          • Control siRNA compared to KDM4C siRNA #1

          • Control siRNA compared to KDM4C siRNA #2

          • Control siRNA compared to KDM4C siRNA #3

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

    • Oil-Red-O staining for siRNA #1 and #2:

      • 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 meta-analytic approach to combine the original and replication effects, which will be presented as a forest plot.

        • There is no originally reported data from siRNA #3, therefore it will not be included.

    • Western Blot:

      • The replication data will be presented as a mean with 95% confidence intervals and will include the original data point, calculated directly from the representative image, as a single point on the same plot for comparison.

Known differences from the original study

  • The replication will perform the Oil-Red-O quantification for all three siRNAs, not just #1 and #2 as presented in Supplemental Figure 8.

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/vfsbo/).

  • Ponceau stains confirming protein transfer to membranes

  • STR profiling and mycoplasma testing results

Power calculations

Note: details of all power calculations can be found at https://osf.io/rb32p/

Protocol 1

Summary of original data

Note: data estimated from published figures.

Supplemental Figure 1: normalized WB band intensity (normalized to Vector)MeanSDN
IDH1WTH3K9me21.70.83
H3K9me310.23
K3K4me31.20.63
H3K27me30.40.33
H3K36me31.20.43
H3K27me20.80.43
IDH1R132HH3K9me27.92.53
H3K9me34.11.23
K3K4me33.40.83
H3K27me32.50.53
H3K36me31.70.83
H3K27me24.72.53
IDH2WTH3K9me23.21.13
H3K9me32.11.23
K3K4me31.90.33
H3K27me31.90.83
H3K36me31.40.43
H3K27me21.50.93
IDH2R172KH3K9me211.43.83
H3K9me34.91.63
K3K4me341.43
H3K27me33.61.63
H3K36me31.80.73
H3K27me25.43.73
Figure 1B: 2HG/glutamate ratios
MeanAssumed N
IDH1WT0.0053
IDH1R132H0.0523
IDH2WT0.0233
IDH2R172K1.563

Test family

  • Western blot; Figure 1B/Supplemental Figure 1A: Note: Since we do not have the raw data, we were unable to perform power calculations using a MANOVA. We are approximating sample sizes with corrected one-way ANOVAs for each DV (normalized histone methylations).

    • Bonferroni-corrected one-way ANOVAs (one per DV) followed by Bonferroni corrected planned comparisons:

      • Wild-type IDH1 compared to IDH1R132H, collapsed across all histone lysine methylations.

      • Wild-type IDH2 compared to IDH2R172K, collapsed across all histone lysine methylations.

        • Note: Only H3K9me2 is being included since this is the histone modification with the largest effect size reported. A correlation among all the histone methylations will also be performed prior to performing the proposed analysis plan.

  • 2HG/glutamate ratios; Figure 1B:

    • One-way ANOVA followed by Bonferroni-corrected pairwise comparisons for the following:

      • IDH1WT compared to IDH1R132H, for H3K9me2

      • IDH2WT compared to IDH2R172K, for H3K9me2

Power calculations

  • Power calculations were performed using R software (version 3.2.2) (R Core Team, 2015) and G*Power (version 3.1.7) (Faul et al., 2007)

  • Partial η2 calculated as in Lakens (2013)

  • Western blot calculations:

    • Note: Due to the large variance, these parametric tests are only used for comparison purposes. The sample size is based on the non-parametric tests also listed. For the ANOVA a Kruskal-Wallis would be performed as the non-parametric alternative, which would require an ~15% increase in sample size calculated for the parametric test listed.

One-way ANOVA: α=0.00833, 4 groupsα
DVF(3,8)Partial η2Effect size fA priori powerTotal Sample Size
H3K9me210.4860.797261.9830292.1%1121
H3K9me37.02740.724921.6233596.4%1161
H3K4me36.61970.712841.5755695.1%1161
H3K27me36.03390.693511.5042392.7%1161
H3K36me30.62760.190511.0612590.2%24
H3K79me23.00330.529691.07984280.0%224
  1. 1 With 6 samples per group (24 total), achieved power is 99.9%.

  2. 2 Since the original effect size will not be detectable with the proposed sample size, this is the effect size that can be detected at 80% power with the given sample size. The original effect size was 0.48512.

Planned contrasts; two-tailed t-test: α=0.004167
Group 1Group 2Effect size dA priori powern/group
IDH1WTIDH1R132H3.3403984.9%151
IDH2WTIDH2R172K2.9313887.3%6
Planned contrasts; two-tailed Wilcoxon-Mann-Whitney: α=0.025
Group 1Group 2Effect size dA priori powern/group
IDH1WTIDH1R132H3.3403980.9%252
IDH2WTIDH2R172K2.9313884.0%6
  1. 1 With 6 samples per group, achieved power is 95.3%.

  2. 2 With 6 samples per group, achieved power is 93.4%.

  • 2HG/glutamate ratio calculations:

    • Note: The original data does not indicate the error associated with multiple biological replicates. To identify a suitable sample size, power calculations were performed using different levels of relative variance.

2%; one-way ANOVA: α=0.05, 4 groups
F(3,8)Partial η2Effect size fA priori powerTotal Sample Size
7240.70.9996352.1190199.9%8
Planned comparisons; 2-tailed t-test: α=0.025
Group 1 versusGroup 2Effect size dA priori powern per group
IDH1WTIDH1R132H63.618299.9%2
IDH2WTIDH2R172H69.660699.9%2
15%; one-way ANOVA: α=0.05, 4 groups
F(3,8)Partial η2Effect size fA priori powerTotal Sample Size
128.720.979706.9477299.9%8
Planned comparisons; 2-tailed t-test: α=0.025
Group 1 versusGroup 2Effect size dA priori powern per group
IDH1WTIDH1R132H8.4824283.5%2
IDH2WTIDH2R172H9.2880888.4%2
28%; one-way ANOVA: α=0.05, 4 groups
F(3,8)Partial η2Effect size fA priori powerTotal Sample Size
36.9420.932683.7220199.9%8
Planned comparisons; 2-tailed t-test: α=0.025
Group 1 versusGroup 2Effect size dA priori powern per group
IDH1WTIDH1R132H4.5441592.4%3
IDH2WTIDH2R172H4.9757695.9%3
40%; one-way ANOVA: α=0.05, 4 groups
F(3,8)Partial η2Effect size fA priori powerTotal Sample Size
18.1020.871602.6054296.0%8
Planned comparisons; 2-tailed t-test: α=0.025
Group 1 versusGroup 2Effect size dA priori powern per group
IDH1WTIDH1R132H3.1809189.6%4
IDH2WTIDH2R172H3.4830394.2%4

In order to produce quantitative replication data, we will run the experiment six times. Each time we will quantify the 2HG/glutamate ratio. We will determine the standard deviation across the biological replicates and combine this with the reported value from the original study to simulate the original effect size. We will use this simulated effect size to determine the number of replicates necessary to reach a power of at least 80%. We will then perform additional replicates, if required, to ensure that the experiment has more than 80% power to detect the original effect.

Protocol 2

Summary of original data

Note: data estimated from published figures.

Figure 2A: 2HG/glutamate ratioAssumed N
Mean
Vector0.13
IDH1R172K5.33
IDH2WT0.13
Figure 2D: Relative expression of adipocyte markers
PpargMeanSDN
VectorDay 01.450.8233
Day 413.9920.8163
IDH2WTDay 02.5211.0763
Day 410.9660.8793
IDH2R172KDay 01.1340.8233
Day 44.2230.9413
CebpaMeanSDN
VectorDay 01.1230.4213
Day 43.0531.1883
IDH2WTDay 01.930.4563
Day 44.8070.5653
IDH2R172KDay 00.6670.4913
Day 40.2460.213
AdipoqMeanSDN
VectorDay 0058.6213
Day 4572.414193.1033
IDH2WTDay 0058.6213
Day 4448.27686.2073
IDH2R172KDay 0058.6213
Day 441.37927.5863

Test family

  • 2HG/glutamate ratios; Figure 2A bottom:

    • One way ANOVA followed by Fisher’s LSD for the following comparison:

      • IDH2WT vs. IDH2R172K

  • qRT-PCR; Figure 2D: Note: Since we did not have the raw data, we were unable to perform power calculations using a MANOVA. We are approximating the sample sizes with corrected one-way ANOVAs for each DV (gene).

    • Bonferroni-corrected one-way ANOVAs (one per gene) followed by Bonferroni corrected comparisons for Day 4 timepoints:

      • IDH2R172Kcompared to vector for each gene (3 comparisons total)

      • IDH2R172Kcompared to IDH2WT for each gene (3 comparisons total)

Power calculations

  • Power calculations were performed using R software (version 3.2.2) (R Core Team, 2015) and G*Power (version 3.1.7) (Faul et al., 2007)

  • Partial η2 calculated as in Lakens (2013)

  • 2HG/glutamate ratios:

    • Note: The original data does not indicate the error associated with multiple biological replicates. To identify a suitable sample size, power calculations were performed using different levels of relative variance.

2%; one-way ANOVA: α=0.05, 3 groups
F(2,6)Partial η2Effect size fA priori powerTotal Sample Size
7214.50.9995849.018899.9%6
Planned comparisons; two-tailed t-test: α=0.05
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HIDH2WT84.938399.9%2
15%; one-way ANOVA: α=0.05, 3 groups
F(2,6)Partial η2Effect size fA priori powerTotal Sample Size
128.260.977156.5386699.9%6
Planned comparisons; two-tailed t-test: α=0.05
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HIDH2WT11.325299.8%2
28%; one-way ANOVA: α=0.05, 3 groups
F(2,6)Partial η2Effect size fA priori powerTotal Sample Size
36.8090.924643.5028598.5%6
Planned comparisons; two-tailed t-test: α=0.05
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HIDH2WT6.0670484.2%2
40%; one-way ANOVA: α=0.05, 3 groups
F(2,6)Partial η2Effect size fA priori powerTotal Sample Size
18.0360.857392.4519485.1%6
Planned comparisons; two-tailed t-test: α=0.05
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HIDH2WT4.2468896.6%3

In order to produce quantitative replication data, we will run the experiment five times. Each time we will quantify the 2HG/glutamate ratio. We will determine the standard deviation across the biological replicates and combine this with the reported value from the original study to simulate the original effect size. We will use this simulated effect size to determine the number of replicates necessary to reach a power of at least 80%. We will then perform additional replicates, if required, to ensure that the experiment has more than 80% power to detect the original effect.

  • qRT-PCR:

    • Note: Due to the large variance, these parametric tests are only used for comparison purposes. The sample size is based on the non-parametric tests also listed. For the ANOVA a Kruskal-Wallis would be performed as the non-parametric alternative, which would require an ~15% increase in sample size calculated for the parametric test listed.

Pparg
One-way ANOVA: α=0.0167, 3 groups
F(2,6)Partial η2Effect size fA priori powerTotal Sample Size
96.8540.969965.6819599.5%161
Planned comparisons; two-tailed t-test: α=0.0083
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HVector11.092199.9%232
IDH2R172HIDH2WT7.4055998.6%232
Planned comparisons; two-tailed Wilcoxon-Mann-Whitney: α=0.0083
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HVector11.092199.9%232
IDH2R172HIDH2WT7.4055996.9%232
  1. 1 With 5 samples per group (15 total), achieved power is 99.9%.

  2. 2 With 5 samples per group, achieved power is 99.9%.

Cebpa
One-way ANOVA: α=0.0167, 3 groups
F(2,6)Partial η2Effect size fA priori powerTotal Sample Size
26.8430.899475.6819590.5%161
Planned comparisons; two-tailed t-test: α=0.0083
Group 1 versusGroup 2Effect sizeA priori powern per group
IDH2R172HVector3.2904891.6%5
IDH2R172HIDH2WT10.701199.9%232
Planned comparisons; two-tailed Wilcoxon-Mann-Whitney: α=0.0083
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HVector3.2904889.0%5
IDH2R172HIDH2WT10.701199.9%232
  1. 1 With 5 samples per group (15 total), achieved power is 99.9%.

  2. 2 With 5 samples per group, achieved power is 99.9%.

Adipoq
One-way ANOVA: α=0.0167, 3 groups
F(2,6)Partial η2Effect size fA priori powerTotal Sample Size
15.2690.835792.2560396.3%191
Planned comparisons; two-tailed t-test: α=0.0083
Group 1 versusGroup 2Effect sizeA priori powern per group
IDH2R172HVector3.8500187.8%242
IDH2R172HIDH2WT6.3575294.5%333
Planned comparisons; two-tailed Wilcoxon-Mann-Whitney: α=0.0083
Group 1 versusGroup 2Effect size dA priori powern per group
IDH2R172HVector3.8500184.0%444
IDH2R172HIDH2WT6.3575290.8%333
  1. 1 With 5 samples per group (15 total), achieved power is

  2. 99.9%.

  3. 2 With 5 samples per group, achieved power is 97.9%.

  4. 3 With 5 samples per group, achieved power is 99.9%.

  5. 4 With 5 samples per group, achieved power is 96.8%.

Protocol 3

Summary of original data

Note: data estimated from published figures.

Figure 4D and S8A: Western BlotBand intensity (normalized to H3)
KDM4CControl siRNA1
KDM4C siRNA #10.50971
KDM4C siRNA #20.27671
KDM4C siRNA #30.02492
H3K9me3Control siRNA1
KDM4C siRNA0.36952
  1. 1 These values were normalized to ß-Actin as seen in Supplemental Figure 8A.

  2. 2 These values were normalized to total H3 as seen in Figure 4D. Also there is no data for siRNAs #1 and #2. We have assumed similar values for siRNA #3 for the purposes of these calculations.

Supplemental Figure 8B: Oil-Red-O quantificationMeanSDN
Control1.060.033
siRNA #10.420.153
siRNA #20.690.093
siRNA #310.690.093
  1. 1 There is no data for siRNA #3. We have assumed similar values as siRNA #2 for the purposes of these calculations.

Test family

  • Western blot; Figure 4D and S8A: Note: Since we did not have the raw data, we were unable to perform power calculations using a MANOVA. We are approximating the sample sizes with corrected one-way ANOVAs for each DV (normalized protein).

    • One-way ANOVAs followed by Bonferroni corrected comparisons:

      • H3K9me3 levels in control siRNA compared to each KDM4C siRNA (3 comparisons total)

      • KDM4C levels in control siRNA compared to each KDM4C siRNA (3 comparisons total)

  • Quantification of Oil-Red-O staining; Figure S8B:

    • One way ANOVA followed by Bonferroni corrected comparisons:

      • Control siRNA compared to KDM4C siRNA #1

      • Control siRNA compared to KDM4C siRNA #2

      • Control siRNA compared to KDM4C siRNA #3

Power calculations

  • Power calculations were performed using R software (version 3.2.2) (R Core Team, 2015) and G*Power (version 3.1.7) (Faul et al., 2007).

  • Partial η2 calculated as in Lakens (2013).

  • Western Blot

    • Note: The original data does not indicate the error associated with multiple biological replicates. To identify a suitable sample size, power calculations were performed using different levels of relative variance.

2%; One-way ANOVA: α=0.025, 4 groups α
DVF(3,8)Partial η2Effect size fPowerTotal sample size
H3K9me32114.70.9987428.16199.9%8
KDM4C3865.50.9993138.07399.9%8
Planned comparisons; two-tailed t-test: α=0.0083
DVGroup 1 versusGroup 2Effect size dA priori powern per group
H3K9me3ControlKDM4C #153.09999.9%2
ControlKDM4C #253.09999.9%2
ControlKDM4C #353.09999.9%2
KDM4CControlKDM4C #142.40799.9%2
ControlKDM4C #262.55299.9%2
ControlKDM4C #384.33599.9%2
15%; One-way ANOVA: α=0.025, 4 groups
DVF(3,8)Partial η2Effect size fPowerTotal sample size
H3K9me337.5950.933773.754799.2%8
KDM4C68.720.962645.076499.9%8
Planned comparisons; two-tailed t-test: α=0.0083
DVGroup 1 versusGroup 2Effect size dA priori powern per group
H3K9me3ControlKDM4C #17.080097.8%3
ControlKDM4C #27.080097.8%3
ControlKDM4C #37.080097.8%3
KDM4CControlKDM4C #15.654388.5%3
ControlKDM4C #28.340399.6%3
ControlKDM4C #311.24599.9%3
28%; One-way ANOVA: α=0.025, 4 groups
DVF(3,8)Partial η2Effect size fPowerTotal sample size
H3K9me310.7890.801822.011498.7%12
KDM4C19.7220.880892.719589.2%8
Planned comparisons; two-tailed t-test: α=0.0083
DVGroup 1 versusGroup 2Effect size dA priori powern per group
H3K9me3ControlKDM4C #13.792886.8%4
ControlKDM4C #23.792886.8%4
ControlKDM4C #33.792886.8%4
KDM4CControlKDM4C #13.029185.9%5
ControlKDM4C #24.468095.8%4
ControlKDM4C #36.024092.1%3
40%; One-way ANOVA: α=0.025, 4 groups
DVF(3,8)Partial η2Effect size fPowerTotal sample size
H3K9me35.28680.664721.408080.2%12
KDM4C9.66370.783731.903797.6%12
Planned comparisons; two-tailed t-test: α=0.0083
DVGroup 1 versusGroup 2Effect size dA priori powern per group
H3K9me3ControlKDM4C #12.655087.2%6
ControlKDM4C #22.655087.2%6
ControlKDM4C #32.655087.2%6
KDM4CControlKDM4C #12.120485.6%8
ControlKDM4C #23.127688.3%5
ControlKDM4C #34.216893.3%4

In order to produce quantitative replication data, we will run the experiment three times. Each time we will quantify band intensity. We will determine the standard deviation of band intensity across the biological replicates and combine this with the reported value from the original study to simulate the original effect size. We will use this simulated effect size to determine the number of replicates necessary to reach a power of at least 80%. We will then perform additional replicates, if required, to ensure that the experiment has more than 80% power to detect the original effect.

  • Oil-Red-O staining:

    • Note: Due to the large variance, these parametric tests are only used for comparison purposes. The sample size is based on the non-parametric tests also listed. For the ANOVA a Kruskal-Wallis would be performed as the non-parametric alternative, which would require an ~15% increase in sample size calculated for the parametric test listed.

One-way ANOVA: α=0.05, 4 groups
F(3,8)Partial eta2Effect size fPowerTotal Sample Size
20.9390.887032.802297.8%181
  1. 1 With 3 samples per group (12 total), achieved power is 99.9%.

Planned comparisons; two-tailed Wilcoxon-Mann-Whitney: α=0.0167
Power Calculations
Group 1Group 2Effect size dPowern/group
ControlKDM4C #15.916896.9%3
ControlKDM4C #25.515693.3%3
ControlKDM4C #35.515693.3%3
Planned comparisons; two-tailed t-test: α=0.0167
Group 1Group 2Effect size dPowern/group
ControlKDM4C #15.916897.7%3
ControlKDM4C #25.515695.6%3
ControlKDM4C #35.515695.6%3

References

  1. 1
  2. 2
  3. 3
  4. 4
    An open investigation of the reproducibility of cancer biology research
    1. TM Errington
    2. E Iorns
    3. W Gunn
    4. FE Tan
    5. J Lomax
    6. BA Nosek
    (2014)
    eLife, 3, 10.7554/eLife.04333.
  5. 5
    G*power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences
    1. F Faul
    2. E Erdfelder
    3. AG Lang
    4. A Buchner
    (2007)
    Behavior Research Methods 39:175–191.
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19

Decision letter

  1. Irwin Davidson
    Reviewing Editor; Institut de Génétique et de Biologie Moléculaire et Cellulaire, France

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "Registered report: IDH mutation impairs histone demethylation and results in a block to cell differentiation" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Irwin Davidson (Reviewing Editor) and Randy Schekman (Senior Editor).

The reviewers have discussed the reviews with one another and the Reviewing editor has drafted this decision to help you prepare a revised submission.

Summary:

The reviewers found this proposed study to have important impact on the functional consequences of the frequently occurring IDH mutations and their mode of action. The provided protocol appears complete, but several points concerning the statistical methods remain to be clarified.

Essential revisions:

Normality and homoscedasticity will be assessed using adequate tests and methods. However what kind of transformation do the authors plan to apply in case of skewed data? Do the authors have an idea of how the data could be skewed? Similarly, which test do the authors intend to apply in case of non-normal data?

Power calculations assume normally distributed data. In the opposite case non parametric tests will be used with much less power. Power calculations should be taken with caution, as they are performed according to an optimistic scenario.

Protocol 1:

The authors should clarify if the MANOVA analyses are planned to analyze the results of western blots and that:

a) Response variables are intensities measured for H3K9me2, H3K9me3, etc.;b) The factor of variability is the cohort with 5 different levels: control, IDH1-WT, IDH1 mutant, IDH2-WT, IDH2 mutant;c) Contrasts to be tested are the ones listed under the subsection “Confirmatory analysis plan” (H3K9me2, WT vs. mutant, etc.).

Clarifications should be made for each model used in this protocol, as it is not always clear how the suggested model fits the data. The same comments hold true for the other protocols, for example the 2HG/glutamate ratios where there is only one response variable and a factor of variability with 5 levels with 2 tested contrasts.

Protocol 2:

For the 2HG/glutamate ratio, please describe the model (one factor with three levels, one contrast?).

For the qRT-PCR, is it a per-gene model? What are the response variables? Please clarify.

Protocol 3:

Again please clarify the MANOVA model (response variable, factor levels, contrasts).

https://doi.org/10.7554/eLife.10860.002

Author response

Essential revisions: Normality and homoscedasticity will be assessed using adequate tests and methods. However what kind of transformation do the authors plan to apply in case of skewed data? Do the authors have an idea of how the data could be skewed? Similarly, which test do the authors intend to apply in case of non-normal data?

We have included the potential of a logarithmic transformation in case of skewed data in the revised manuscript. This might be the case with the 2HG/glutamate ratios that are listed in the protocols, which only show representative images, but display them on a log scale, which suggests the data might be skewed.

In the case of non-normal data, the revised manuscript includes the non-parametric tests that would be applied. When there is reason to suspect that a non-parametric test might be applied, such as a large difference in variance between the groups being tested, we have included non-parametric as well as parametric power calculations as described in response to the second question.

Power calculations assume normally distributed data. In the opposite case non parametric tests will be used with much less power. Power calculations should be taken with caution, as they are performed according to an optimistic scenario.

We agree and understand the limitations of using the original data we have been able to obtain, or estimate, as a basis for the sample size for the replication attempt. We also agree that some of the assumptions made are according to an optimistic scenario. However, we hope that by performing power calculations using non-parametric tests, when there is reason to believe the data would violate the assumptions of a parametric test, we will ensure the sample size used for the replication study will be suitable for a parametric test (what we aim to perform if possible) or the non-parametric test (what we quite possible will need to perform). Further, as pointed out, this still stands risk of decreased power depending on the nature of the replication data (such as if there is normal distribution) in order to perform the proposed comparisons. Additionally, we have revised the analysis plans and power calculations to reflect the proposed models and contrasts, including the potential non-parametric tests that might be performed.

Protocol 1: The authors should clarify if the MANOVA analyses are planned to analyze the results of western blots and that:

a) Response variables are intensities measured for H3K9me2, H3K9me3, etc.;b) The factor of variability is the cohort with 5 different levels: control, IDH1-WT, IDH1 mutant, IDH2-WT, IDH2 mutant;c) Contrasts to be tested are the ones listed under the subsection “Confirmatory analysis plan” (H3K9me2, WT vs. mutant, etc.).

Clarifications should be made for each model used in this protocol, as it is not always clear how the suggested model fits the data. The same comments hold true for the other protocols, for example the 2HG/glutamate ratios where there is only one response variable and a factor of variability with 5 levels with 2 tested contrasts. Protocol 2: For the 2HG/glutamate ratio, please describe the model (one factor with three levels, one contrast?).

For the qRT-PCR, is it a per-gene model? What are the response variables? Please clarify. Protocol 3: Again please clarify the MANOVA model (response variable, factor levels, contrasts).

Thank you for this suggestion. We have included the model of each proposed analysis plan, to reflect the DV, IV, and contrasts proposed. Additionally, we have revised the analysis plans and power calculations to reflect the proposed models and contrasts, included the potential non-parametric tests that might be performed.

https://doi.org/10.7554/eLife.10860.003

Article and author information

Author details

  1. Adam D Richarson

    Cancer Metabolism Facility, Sanford Burnham Prebys Medical Discovery Institute, La Jolla, United States
    Contribution
    ADR, Drafting or revising the article
    Competing interests
    ADR: Cancer Metabolism Facility at Sanford Burnham Prebys Medical Discovery Institute is a Science Exchange associated laboratory
  2. David A Scott

    Cancer Metabolism Facility, Sanford Burnham Prebys Medical Discovery Institute, La Jolla, United States
    Contribution
    DAS, Drafting or revising the article
    Competing interests
    DAS: Cancer Metabolism Facility at Sanford Burnham Prebys Medical Discovery Institute is a Science Exchange associated laboratory
  3. Olga Zagnitko

    Cancer Metabolism Facility, Sanford Burnham Prebys Medical Discovery Institute, La Jolla, United States
    Contribution
    OZ, Drafting or revising the article
    Competing interests
    OZ: Cancer Metabolism Facility at Sanford Burnham Prebys Medical Discovery Institute is a Science Exchange associated laboratory
  4. Pedro Aza-Blanc

    Functional Genomics Core, Sanford Burnham Prebys Medical Discovery Institute, La Jolla, United States
    Contribution
    PA-B, Drafting or revising the article
    Competing interests
    PA-B: Functional Genomics Core at Sanford Burnham Prebys Medical Discovery Institute is a Science Exchange associated laboratory
  5. Chih-Cheng Chang

    Functional Genomics Core, Sanford Burnham Prebys Medical Discovery Institute, La Jolla, United States
    Contribution
    C-CC, Drafting or revising the article
    Competing interests
    C-CC: Functional Genomics Core at Sanford Burnham Prebys Medical Discovery Institute is a Science Exchange associated laboratory
  6. David A Russler-Germain

    Washington University School of Medicine, St Louis, United States
    Contribution
    DAR-G, Drafting or revising the article
    Competing interests
    No competing interests declared.
    ORCID icon 0000-0003-1009-2247
  7. Reproducibility Project: Cancer Biology

    Contribution
    RP:CB, Conception and design, Drafting or revising the article
    For correspondence
    fraser@scienceexchange.com
    Competing interests
    RP:CB: EI, FT, JL, NP: Employed by and hold shares in Science Exchange Inc
    1. Elizabeth Iorns, Science Exchange, Palo Alto, United States
    2. William Gunn, Mendeley, London, United Kingdom
    3. Fraser Tan, Science Exchange, Palo Alto, United States
    4. Joelle Lomax, Science Exchange, Palo Alto, United States
    5. Nicole Perfito, Science Exchange, Palo Alto, United States
    6. Timothy Errington, Center for Open Science, Charlottesville, United States

Funding

Laura and John Arnold Foundation

  • The Reproducibility Project: Cancer Biology

The Reproducibility Project: Cancer Biology is funded by the Laura and John Arnold Foundation, provided to the Center for Open Science in collaboration with Science Exchange.

Acknowledgements

We thank Courtney Soderberg at the Center for Open Science for assistance with statistical analyses. We would also like to thanks the following companies for generously donating reagents to the Reproducibility Project: Cancer Biology; American Type and Tissue Collection (ATCC), Applied Biological Materials, BioLegend, Charles River Laboratories, Corning Incorporated, DDC Medical, EMD Millipore, Harlan Laboratories, LI-COR Biosciences, Mirus Bio, Novus Biologicals, Sigma-Aldrich, and System Biosciences (SBI).

Reviewing Editor

  1. Irwin Davidson, Reviewing Editor, Institut de Génétique et de Biologie Moléculaire et Cellulaire, France

Publication history

  1. Received: August 18, 2015
  2. Accepted: February 15, 2016
  3. Version of Record published: March 14, 2016 (version 1)

Copyright

© 2016, Richarson et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 939
    Page views
  • 197
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Comments

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

  1. C Lu et al.
    1
    1. Cancer Biology
    Curated by Roger Davis et al.
    Collection Updated

    Investigating reproducibility in preclinical cancer research.

    1. Human Biology and Medicine
    Andrew S Cowburn et al.
    Research Article