Uhrf2 coordinates repressive epigenetic pathways

Interestingly, a second member of the Uhrf family, Uhrf2, harbours similar domains (Bron- ner et al., 2007). Until now, Uhrf2 has been described to play a role in intranuclear degradation of polyglutamine aggregates (Iwata et al., 2009) and cell cycle control (Mori et al., 2002; Li et al., 2004; Mori et al., 2011). However, a potential function of Uhrf2 in maintenance DNA methylation has not been reported. We systematically in- vestigated the function and interplay of distinct Uhrf2 domains in DNA and histone tail substrate recognition and reported first data on cell-type specific functions of Uhrf1

and Uhrf2 (Chapter 3.4). Firstly, we compared the expression pattern of uhrf1 and

uhrf2 in ESCs and somatic cells, during differentiation and in differentiated mouse tis-

sues and found opposite expression patterns; whileuhrf1is expressed in ESCs and down

regulated during differentiation, which is consistent with previous reports (Fujimori et

al., 1998; Hopfner et al., 2000), uhrf2 is up regulated and highly expressed in differen-

tiated mouse tissues. Using our in vitro binding assays, we investigated the DNA and

binds to unmodified histone H3 and H3K9me3 mediated by an aromatic cage in the tandem Tudor domain. Consistently, acetylation of H3K9, underrepresented in hete- rochromatin, prevented the binding of Uhrf2. Notably, in contrast to recent publica- tions, we could not detect any histone binding for the isolated PHD domains of Uhrf1

and Uhrf2in vitro. It was shown, that the PHD domain, which is located adjacent to the

tandem Tudor domain, binds to histone H3 unmodified arginine 2 (H3R2) (Rajakumara et al., 2011; Xie et al., 2011; Wang et al., 2011; Hu et al., 2011). The binding of the PHD domain to H3 is abolished by methylation of H3R2 but not by H3K4 and H3K9 methy- lation. ChIP studies demonstrated that Uhrf1’s ability to repress its direct target gene expression is dependent on PHD binding to unmodified H3R2, thereby demonstrating the functional importance of this recognition event and supporting the potential for crosstalk between histone arginine methylation and Uhrf1 function (Rajakumara et al., 2011). In general, arginine methylation is catalysed by the PRMT family of proteins and physiological roles for protein arginine methylation have been established in several bi- ological processes like signal transduction, mRNA splicing, transcriptional control, DNA repair and protein translocation (Bedford and Clarke, 2009). However, little is known about how histone arginine residues are recognized and to what extent protein argi- nine methylation and demethylation occurs. Interestingly, we could demonstrate that the combination of the PHD and the tandem Tudor domain of Uhrf2 displayed an increased binding to H3K9me2/me3 in comparison to the single TTD, which was not observed for the corresponding construct of Uhrf1. The increased binding depends on the highly conserved linker region between the tandem Tudor and PHD domain of Uhrf2, suggesting a cooperative interplay of different Uhrf2 domains. A similar func- tional importance of linker sequences has been described for BPTF and histone lysine demethylases (Li et al., 2006; Horton et al., 2010). In case of BPTF, the largest sub- unit of the ATP-dependent chromatin-remodelling complex, NURF, a PHD and a bromo

domain are connected via a a-helical linker (Li et al., 2006). Both investigated his-

tone demethylases harbour a PHD that binds H3K4me3 and a jumonji domain that demethylates either H3K9me2 or H3K27me2 and the substrate specificity is controlled by the linker connecting both functional domains (Horton et al., 2010). These data demonstrate an additional level of complexity involving multivalent binding of histone- tail peptides by mixed two- effector modules (Ruthenburg et al., 2007).

In contrast to Uhrf1, Uhrf2 did not show a DNA binding preference for hemimethylated DNA. However, binding to heterochromatin-specific H3K9me3 peptides induced a sig- nificant preference of Uhrf2 for hemi- over unmethylated DNA. Vice versa, binding to DNA enhanced binding of Uhrf1 and Uhrf2 to the H3K9me3 peptide. Consistently, SILAC-based proteomic analysis identified enrichment of Uhrf1 at nucleosomes con- taining repressive DNA and H3K9 methylation marks (Bartke et al., 2010). A similar ef- fect was reported for MSL3 that specifically binds to H4K20me1 via a chromodomain

only in the presence of DNA (Kim et al., 2010).

Similar to Uhrf1, the subcellular localization and binding dynamics of Uhrf2in vivo are

dependent on histone H3K9 methylation but not on DNA methylation. Consistently, mutant Uhrf1 protein deficient for H3K4me0/K9me3 binding fails to reduce the expres- sion of a target gene, p16INK4A, when overexpressed (Nady et al., 2011). Remarkably,

transient expression of Uhrf2 inuhrf1-/-_{ESCs did not restore the DNA methylation pattern}

at major satellites, pointing to functional differences between Uhrf1 and Uhrf2in vivo.

Notably, a recent publication reveals that Uhrf fails to recruit Dnmt1 to replication foci during S phase of the cell cycle, demonstrating that the cell cycle-dependent inter- action between Uhrf1 and Dnmt1 is a key regulatory mechanism for DNA methylation (Zhang et al., 2011). Also, the opposite expression pattern argues against a functional

redundancy of both genes and explain the drastic loss of DNA methylation inuhrf1-/-

ESCs despite the presence of intactuhrf2alleles.