Transcription factors bind negatively selected sites within human mtDNA genes

Abstract

Transcription of mitochondrial DNA (mtDNA)-encoded genes is thought to be regulated by a handful of dedicated transcription factors (TFs), suggesting that mtDNA genes are separately regulated from the nucleus. However, several TFs, with known nuclear activities, were found to bind mtDNA and regulate mitochondrial transcription. Additionally, mtDNA transcriptional regulatory elements, which were proved important in vitro, were harbored by a deletion that normally segregated among healthy individuals. Hence, mtDNA transcriptional regulation is more complex than once thought. Here, by analyzing ENCODE chromatin immunoprecipitation sequencing (ChIP-seq) data, we identified strong binding sites of three bona fide nuclear TFs (c-Jun, Jun-D, and CEBPb) within human mtDNA protein-coding genes. We validated the binding of two TFs by ChIP-quantitative polymerase chain reaction (c-Jun and Jun-D) and showed their mitochondrial localization by electron microscopy and subcellular fractionation. As a step toward investigating the functionality of these TF-binding sites (TFBS), we assessed signatures of selection. By analyzing 9,868 human mtDNA sequences encompassing all major global populations, we recorded genetic variants in tips and nodes of mtDNA phylogeny within the TFBS. We next calculated the effects of variants on binding motif prediction scores. Finally, the mtDNA variation pattern in predicted TFBS, occurring within ChIP-seq negative-binding sites, was compared with ChIP-seq positive-TFBS (CPR). Motifs within CPRs of c-Jun, Jun-D, and CEBPb harbored either only tip variants or their nodal variants retained high motif prediction scores. This reflects negative selection within mtDNA CPRs, thus supporting their functionality. Hence, human mtDNA-coding sequences may have dual roles, namely coding for genes yet possibly also possessing regulatory potential.

Keywords: CEBPb; ChIP-seq; Jun-D; c-Jun; mitochondrial DNA; negative selection; transcription.

Fig. 1.—

(A) A schematic phylogenetic tree representing nodal and tip variants. Tip variants, filled circles; nodal variants, open circles. (B) Decision tree describing the approach to identify negative selection in TFBS.

Fig. 2.—

ChIP-seq peaks correspond to DNase1 hypersensitivity sites in human cells. x axis: mtDNA nucleotide positions. y axis: (A–D) The blue area represents positive strand reads, whereas the green area corresponds to reads from the negative strand. The dark green area designates the overlapping region between reads from the two strands. Black rectangle below the panel—JASPAR predicted TF-binding motif. y axis: (E–G) The F score (see Materials and Methods) of each mtDNA position in DNAse-seq experiments; the lower the score the more protected is the DNA by proteins. (A) Jun-D, (B) c-Jun site 1 (bind1), (C) c-Jun site 2 (bind2), (D) CEBPb, (E, F) DNase-seq of HepG2 cells, (G) DNase-seq of IMR90 cells. (H) A scheme summarizing the identified mtDNA-binding sites of c-Jun, Jun-D, and CEBPb. D-Loop—the main noncoding region. 12S, 16S—rRNA genes. Capital letters—tRNAs genes. ND genes, CO1-3, Cytb, and ATP6/8—protein-coding subunits of OXPHOS complexes 1, 3, 4, and 5. Arrows point to the binding sites of the relevant TFs; asterisk, c-Jun-binding site 2.

Fig. 3.—

ChIP-seq validation experiment by ChIP-qPCR. ChIP experiments performed using HepG2, foreskin (FSF), and H9-hESC cells with anti-c-Jun or anti-Jun-D antibodies were followed by qPCR. (A) Jun-D-binding peaks. (1) The identified mtDNA-binding site. (2) mtDNA site lacking a ChIP-seq peak. (B) c-Jun-binding validation. (1) and (2) correspond to the identified mtDNA-binding sites, with (1) being the most prominent site. (3) mtDNA site lacking a significant ChIP-seq peak. (C) ChIP-qPCR signals for Jun-D. (D) ChIP-qPCR signals for c-Jun. Note that the PCR signal is stronger in the binding sites as compared with the negative controls (nonbinding sites and H9-hESC cells [H9]) that showed no binding.

Fig. 4.—

Mitochondrial localization of c-Jun and Jun-D in human cells. (A) Immunogold labeling of Jun-D and c-Jun in mitochondria of HepG2 cells. Mitochondria-associated signals are indicated by arrows. (A) Jun-D immunogold labeling. (B) c-Jun immunogold labeling. (C) A control experiment was conducted using secondary antibodies alone. (D) Quantification of mitochondrial particles (black) versus cytoplasmic particles (gray) reflecting antibody labeling of c-Jun and Jun-D. ***P value < 0.001. y axis: Number of gold particles, x axis: Analyzed proteins. (E) To assess the mitochondrial localization of c-Jun and Jun-D by subcellular fractionation, histoneH3 and VDAC1 were used as markers for the purity of the nuclear (N) and mitochondrial (M) fractions, respectively. The presented results summarize three independent fractionation experiments.

Fig. 5.—

Number of nodal events in the binding sites of Jun-D, c-Jun, and CEBPb. Number of nodal events in the binding site of the identified TFs represented by filled rhombus; the weaker binding site of c-Jun—bind 2—is represented by an open rhombus. Distribution of nodal events within each of the predicted binding sites (CNRs) represented by box plots (note—the whiskers do not represent average and SD). y axis: Number of mutational events (variants); x axis: The analyzed TFs. We noticed that for Jun-D the median equals to the first quarterly and for c-Jun the median equal to the third quarterly. Filled circle, the average of nodal events in the CNRs of each TF.

Fig. 6.—

Percentage of variant events which retained the predicted binding motif of c-Jun. The strong binding site of c-Jun (bind 1) is represented by a filled rhombus and the weaker binding site of c-Jun (bind-2) is represented by an open rhombus. Nodal events do not change the prediction score in c-Jun bind 1, but change it in c-Jun bind 2.The distribution of the mutational events which retain the binding motif of the CNRs is represented by a box plot. y axis: percentage of mutational events which retain the binding motif prediction motif; x axis: TF.