(A) Overexpression of Cerox1 in mouse wildtype and DicerΔ/Δembryonic stem (ES) cells (inset graph). The overexpression of Cerox1 in wildtype mouse embryonic stem cells results in an increase in complex I subunit transcripts, with no observed change in expression of two control subunits (Ndufs2, Ndufv1) that were also unaffected in N2A cells. Overexpression of Cerox1 in DicerΔ/Δembryonic stem cells results in no increase in the expression of any complex I subunit. 2-sided t-test; **p<0.01, *p<0.05, ns = not significant. Error bars s.e.m. n = 3 biological replicates. (B) Predicted MREs whose presence is conserved in both the mouse and human Cerox1. Coloured MREs indicate those MREs whose presence is conserved between mouse and human and whose miRNAs are expressed in N2A cells. miRNA site types are as follows: miR-28–5p, 8mer-A1; miR-138–5p, 6mer; miR-370–3p, 7mer-m8; miR-488–3p, 7mer-m8; miR-708–5p, 7mer-A1. The grey predicted MREs represent those that are conserved, but whose miRNAs are not expressed in N2A cells (miR-125a-3p, miR-199/199–5p, miR-302ac/520f, miR-485/485–5p, miR-486/486–5p, miR-501/501–5p, miR-654–3p, miR-675/675–5p). (C) Overexpression of the 5xMRE mutant failed to alter expression of complex I subunit transcripts that otherwise all increase in abundance following wild-type Cerox1 overexpression. Fold changes of wildtype Cerox1 or the 5xMRE Cerox1 mutant compared to the control are indicated in the inset panel. The numbers of MREs predicted by TargetScan v7.0 (Agarwal et al., 2015) in these transcripts’ 3’UTRs for the four conserved, N2A expressed miRNA families are indicated (see also Supplementary file 3). Due to known widespread noncanonical miRNA binding (Helwak et al., 2013), predictions were also extended across the gene body (bracketed MREs). 2-sided t-test; **p<0.01, *p<0.05, ns = not significant. Error bars s.e.m. n = 3 biological replicates. (D) Overexpression of the 5xMRE mutant failed to alter OXPHOS enzymatic activity compared to the control for any of the complexes measured. A one-way ANOVA was applied to test for differences in activities of the mitochondrial complexes between a control and overexpression of wildtype Cerox1 and the 5xMRE mutant. A post-hoc Dunnett’s test indicated that the overexpression of wildtype Cerox1 resulted, as expected, in significantly increased complex I and IV activities of 30% and 17% respectively (F [2, 21]=4.9, p=0.017; F[2, 20]=4.6, p=0.033), while comparisons for the 5xMRE mutant with the control were not significant. There was no significant difference in the activities of complex II (F[2,19]=3.5, p=0.26), complex III (F[2,19]=0.08, p=0.5) or citrate synthase (F[2,20]=2.6, p=0.42). n = 6 biological replicates. Significance levels, one-way ANOVA, Dunnett’s post hoc test *p<0.05. (E) Four to six fold overexpression of each of four miRNAs with predicted MREs whose presence is conserved in both mouse and human Cerox1 resulted in a decrease in Cerox1 transcript level, with overexpression of miR-488–3p resulting in >90% knock down of Cerox1. This was not observed when the miRNA miR-137–3p, which has no predicted MREs in Cerox1, was similarly overexpressed. Error bars s.e.m. n = 3 biological replicates. (F) Fluorescent in situ hybridisation detection of miR-488–3p (magenta) and Cerox1 (green) in N2A cells. Scale bar = 5 μm. A no probe control (Figure 5—figure supplement 1D) indicated some background Fast Red signal (miRNA detection) localised to the nucleus, but no background for Alexa Fluor-488 (lncRNA detection).