ceRNA depletion on PTEN expression
We replicated an experiment to test the microRNA dependency of the putative PTEN ceRNAs. This experiment used the same isogenic wild-type and DICER mutant (DicerEx5) HCT116 colon carcinoma cells as the original study. The DicerEx5 cell line, which was engineered to disrupt a well-conserved segment of the N-terminal helicase domain in exon 5 of DICER, while leaving the RNase III domains intact, displays a hypomorphic phenotype in the processing of mature microRNAs (Cummins et al., 2006). This experiment is similar to what was reported in Figure 3G–H and Supplemental Figure S3B of Tay et al., 2011 and described in Protocol 3 in the Registered Report (Phelps et al., 2016). Wild-type and DicerEx5 HCT116 cells were transfected with siRNAs targeting the same putative PTEN ceRNAs as the original study. Knockdown efficiency, measured by RT-qPCR, revealed the average reduction in gene expression relative to control siRNA was 81% in both cell lines for all putative PTEN ceRNAs, with the greatest biological variability in DicerEx5 HCT116 cells when targeting CNOT6L (Figure 3—figure supplement 1B).
Depletion of the putative PTEN ceRNAs resulted in downregulation of PTEN protein in wild-type HCT116 cells to an average of 80%, 43%, or 61% for siRNA-mediated depletion of SERINC1, VAPA, or CNOT6L, respectively, relative to control siRNA (average PTEN expression = 100%) (Figure 3A–B, Figure 3—figure supplement 1A). As a control, siRNAs targeting PTEN reduced PTEN protein levels to an average of 1.6%. To compare the relative PTEN expression among the various conditions, we planned to conduct four comparisons using the Bonferroni correction to adjust for multiple comparisons. The comparison of PTEN protein levels between control siRNA and siRNA targeting VAPA, CNOT6L, or PTEN were statistically significant, while the comparison of control siRNA and siRNA targeting SERINC1 were not (see Figure 3 legend). The original study reported statistically significant decreases in PTEN protein levels with siRNA-mediated depletion of SERINC1 (average PTEN expression = 53%), VAPA (average PTEN expression = 52%), CNOT6L (average PTEN expression = 59%), or PTEN (average PTEN expression = 1.9%) compared to control siRNA (average PTEN expression = 100%) in wild-type HCT116 cells (Tay et al., 2011).
Figure 3.  PTEN protein expression in wild-type and DICER mutant HCT116 cells depleted of PTEN ceRNAs.
Wild-type (WT) and DICER mutant (DicerEx5) HCT116 cells were transfected with non-targeting control siRNA (siNC) or siRNA plasmids targeting SERINC1 (siSER), VAPA (siVAPA), CNOT6L (siCNO), or PTEN (siPTEN). Cells were harvested 72 hr later for Western blot analysis. (A) Relative protein expression (PTEN/HSP90) are presented for each condition. Western blot bands were quantified, PTEN levels were normalized to HSP90, with protein expression presented relative to siNC. Means reported and error bars represent SD from three independent biological repeats for wild-type HCT116 cells and four repeats for DicerEx5 HCT116 cells. Analysis of wild-type HCT116 cells: one-way ANOVA (equal variance) on PTEN/HSP90 expression: F(4,10) = 25.4, I=3.18×10−5. Planned contrasts between siNC and siSER: t(10) = 1.94, uncorrected I=0.082 with a priori Bonferroni adjusted significance threshold of 0.0125, Bonferroni corrected p=0.326; siNC and siVAPA: t(10) = 5.44, uncorrected p=2.85×10−4, Bonferroni corrected p=0.0011; siNC and siCNOT: t(10) = 3.69, uncorrected p=0.0042, Bonferroni corrected p=0.017; siNC and siPTEN: t(10) = 9.34, uncorrected p=2.97×10−6, Bonferroni corrected p=1.19×10−5. Analysis of DicerEx5 HCT116 cells: one-way ANOVA (unequal variance) on PTEN/HSP90 expression: F(4,6.0) = 19.3, p=0.0014. Planned comparisons: siNC and siSER: two-sample t-test, t(6) = 3.96, uncorrected p=0.0074 with a priori Bonferroni adjusted significance threshold of 0.0125, Bonferroni corrected p=0.030; siNC and siVAPA: two-sample t-test, t(6) = 0.896, uncorrected p=0.405, Bonferroni corrected p>0.99; siNC and siCNOT: Welch’s t-test, t(4.36) = 2.92, uncorrected p=0.039, Bonferroni corrected p=0.156; siNC and siPTEN: two-sample t-test, t(6) = 4.15, uncorrected p=0.0060, Bonferroni corrected p=0.024. (B) Representative Western blots probed with an anti-PTEN antibody and anti-HSP90 antibody. Additional details for this experiment can be found at https://osf.io/drcbw/.
Figure 3—figure supplement 1.  Knockdown efficiency and individual repeats of PTEN protein expression in wild-type and DICER mutant HCT116 cells transfected with siRNA against PTEN ceRNAs.
This is the same experiment as Figure 1. (A) Independent biological repeats of Western blot assay. PTEN/HSP90 protein expression is presented for each condition relative to the siNC condition. (B) Independent biological repeats of RT-qPCR analysis. Expression of each transcript after transfection of its respective siRNA relative to negative control transfection (siNC) is presented. Transcripts listed on y-axis. Means reported and error bars represent SD from three technical replicates. One-sample t-tests of transcript expression data after transfection of respective siRNA to a constant of 1 (relative value of siNC). Wild-type HCT116 cells: SERINC1: t(2) = 8.41, uncorrected p=0.014, Bonferroni corrected p=0.055; VAPA: t(2) = 120, uncorrected p=6.98×10−5, Bonferroni corrected p=2.79×10−4; CNOT6L: t(2) = 10.4, uncorrected p=0.0091, Bonferroni corrected p=0.036; PTEN: t(2) = 30.6, uncorrected p=0.0011, Bonferroni corrected p=0.0043. DicerEx5 HCT116 cells: SERINC1: t(2) = 28.5, uncorrected p=0.0012, Bonferroni corrected p=0.0049; VAPA: t(2) = 48.3, uncorrected p=4.28×10−4, Bonferroni corrected p=0.0017; CNOT6L: t(2) = 3.23, uncorrected p=0.084, Bonferroni corrected p=0.336; PTEN: t(2) = 242, uncorrected p=1.71×10−5, Bonferroni corrected p=6.83×10−5. Additional details for this experiment can be found at https://osf.io/drcbw/. For DicerEx5 HCT116 cells, we found depletion of the putative PTEN ceRNAs resulted in higher PTEN protein levels (SERINC1: 290%; VAPA: 144%; CNOT6L: 256%) relative to control siRNA (average PTEN expression = 100%), while targeting PTEN reduced PTEN protein levels to an average of 2.6%. (Figure 3A–B, Figure 3—figure supplement 1A). To compare the relative PTEN expression among the various conditions, a similar analysis as described above for wild-type HCT116 cells was performed for DicerEx5 HCT116 cells. We found that PTEN protein levels between control siRNA and siRNA targeting SERINC1 or PTEN were statistically significant, while the comparisons between control siRNA and siRNA targeting VAPA or CNOT6L were not (see Figure 3 legend). The original study reported PTEN downregulation by ceRNA depletion was attenuated in DicerEx5 HCT116 cells with average PTEN expression around the same as control siRNA (control siRNA: 100%; SERINC1: 117%; VAPA: 108%; CNOT6L: 113%), while the average PTEN expression in cells transfected with siRNA-mediated depletion of PTEN was 1.3% (Tay et al., 2011). Similar to the siRNA-mediated depletion of putative PTEN ceRNA in DU145 cells described above, the original study reported a knockdown of greater than 90% for most conditions (Tay et al., 2011). The level of knockdown required to yield a given phenotype varies because it is system-dependent (Bailoo et al., 2014), thus the difference in achieved knockdown between the original study and this replication attempt should be considered when interpreting these results. Further, the original study reported lower RSDs for PTEN protein levels across all the siRNA conditions in the DicerEx5 HCT116 cells compared to the wild-type HCT116 cells (DicerEx5: 0.1–9% vs wild-type: 8–17%), while this replication attempt observed larger RSDs compared to the original study, especially for DicerEx5 HCT116 cells (DicerEx5: 29–61% vs wild-type: 12–46%). Importantly, the individual biological repeats were largely consistent relative to the control siRNA condition (Figure 3—figure supplement 1A). This difference in variance between the original study and this replication attempt could influence if statistical significance is reached. To summarize, for this experiment, we found results that were generally in the same direction as the original study, varied in terms of statistical significance, and in DicerEx5 HCT116 cells effects that were of a larger magnitude than the original study. This absence of an attenuated ceRNA effect in this replication attempt suggests the null hypothesis that there is no difference in PTEN protein expression when the microRNA machinery is disrupted can be rejected.