Discovery of ChAHP2 complex

Eukaryotic transcription factors (TFs) are the key that lock or unlock the information encoded in the DNA sequence. In humans, multiple TFs help to turn our estimated twenty to twenty-five thousand genes "on" or "off" in a space and time dependent manner resulting in organismal development and homeostasis. Research over the last 50 years has led to the discovery of thousands of eukaryotic TFs and their structural motifs that bind DNA in a sequence specific manner. Based on the similarity with the structural motif of known TFs many other proteins are predicted to be TFs, however their genomic targets and role in regulating gene expression still remain to be experimentally verified. ADNP2 is one such protein which, based on the similarity of its sequence with known DNA binding motifs in other TFs, is predicted to be a sequence specific TF. ADNP2 has nine zinc-finger motifs, known sequence specific DNA binding motifs, arranged in two clusters, four at the N-terminal region and five at the C-terminal region. In addition, it has a homeobox domain, also a known DNA-binding domain, and a PxVxL motif, a motif known to be recognized by heterochromatin binding proteins, HP1 proteins. Further, it shows overall domain conservation with its homolog, ADNP, a recently described TF. ADNP was found to stably interact with the chromatin remodeling enzyme CHD4 via its N-terminal region and HP1 via the PxVxL motif to form a chromatin regulatory complex termed ChAHP. Therefore, based on the similarity with ADNP, I hypothesized that ADNP2 also associates with CHD4 and HP1 proteins to form a similar chromatin regulatory complex potentially involved in gene regulation. To test my hypothesis, I used mouse embryonic stem cells (mESCs) as a model system and made use of recent advances in genome editing/CRISPR-Cas9 to engineer cell lines that allowed me to make in vivo observations. By using a strategy of ‘observe, perturb and re-observe’ I attempted to systematically dissect the function of ADNP2 in genome regulation and its mechanism of action.
Proteomics data from my experiments showed that ADNP2 does indeed interact with CHD4 and HP1 proteins to form a stable trimeric complex, which I dubbed ChAHP2. In vitro reconstitution of ChAHP2 in insect cells, done in collaboration with the group of Nicolas Thoma, further corroborated the existence of the ChAHP2 complex. Genome-wide studies of ADNP2 localization in its wild- type (WT) state, upon loss of its interaction with HP1 proteins and upon reduction/loss of H3K9me3 showed that ADNP2, in the context of ChAHP2 complex, binds heterochromatic regions in the genome via HP1 mediated recognition of H3K9me3 mark at these regions. Comparison of ChAHP2 binding sites with ChAHP binding sites showed that both bind retrotransposons, but ChAHP binds SINE retrotransposons whereas ChAHP2 binds endogenous retroviruses (ERVs), and LINE 1 (L1) retrotransposons.
Loss of ADNP has been previously shown to result in increased genomic accessibility and expression of SINE elements. Chromosomal accessibility analysis of cells upon genetic ablation of ADPN2 using assay for transposase-accessible chromatin with sequencing (ATAC-seq) revealed that ADPN2 restricts access to its bound sites suggesting a repressive role for ADNP2 at its bound sites as well. However, loss of ADNP2 alone did not have any effect on repression of ERVs or L1s in mESCs, indicating that ADNP2 is not sufficient to maintain ERV repression. ADNP2 bound ERVs are known to be suppressed by SETDB1 mediated H3K9me3 and loss of SETDB1 results in robust upregulation of these retrotransposons in different cell types. I therefore assessed the effect of ADNP2 loss in combination with SETDB1 depletion (double depletion) and compared it to SETDB1 depletion alone to figure out if both protein act synergistically to repress the deleterious activity of transposons. My experimental data showed that, in mESCs, upregulation of ERVs brought about upon loss of SETDB1 is exacerbated by loss of ADNP2. Mass spectrometry (MS) assessment of proteins originating from ERV RNAs showed that ADNP2 does not affect the translation of ERV RNAs. Assessment of production of virus like particles (VLPs) in cells expressing the ERV RNAs by transmission electron microscopy showed that ADNP2 does not affect the production of VLPs either. Further assessment of the effect of double depletion in a temporal manner showed that ADNP2 affects the kinetics of ERV RNA accumulation in mESCs. Taken together, these results suggest that ADNP2 functions downstream of H3K9me3, either at the level of co-transcriptional or post-transcriptional regulation of ERV RNAs prior to translation.
H3K9me3 level is not a constant mark, and it is known to fluctuated during early embryonic development and differentiation. Therefore, as a final step, to assess the impact of loss of ADNP2 in differentiated cells that might have altered H3K9me3 levels as compared to mESCs, I assessed the expression of retrotransposons in neurons lacking ADNP2. My data showed that loss of ADNP2 alone already results in upregulation specific members of ERVs and a member of L1s in neurons. This indeed supports the idea that ADNP2/ChAHP2 becomes important for retrotransposon repression under certain conditions.
In summary, my findings have incontrovertibly established that ADNP2 nucleates ChAHP2, a trimeric protein complex by associating with HP1s and CHD4 that is distinct from the ChAHP complex. These two complexes have distinct modes of recruitment to the genome and have different targets with different chromatin signatures, yet both appear to be involved in transposon silencing or recognition. My data suggests a repressive role for ADNP2, which might function at the level of co-transcriptional or post-transcriptional regulation of RNA, and it will be exciting to test these possibilities with further experiments.