p97/VCP drives turnover of SUMOylated centromeric CCAN proteins and CENP-A

Abstract

The centromere is a unique chromatin domain that links sister chromatids and forms the attachment site for spindle microtubules in mitosis. Centromere inheritance is largely DNA sequence-independent but strongly reliant on a self-propagating chromatin domain featuring nucleosomes containing the H3 variant CENP-A. Unlike other histones, CENP-A is maintained with unusually high stability in chromatin. Previously, we have shown that mitotic maintenance of CENP-A and other CCAN proteins is controlled by a dynamic SUMO cycle and that the deSUMOylase SENP6 is necessary for stable maintenance of CENP-A at the centromere. Here, we discover that the removal of SENP6 leads to a rapid loss of the constitutive centromere-associated network (CCAN), followed by a delayed loss of centromeric CENP-A, indicating that the CCAN is the primary SUMO target. We found the ATP-dependent segregase p97/VCP removes centromeric CENP-A in a SUMO-dependent manner and physically interacts with the CCAN and CENP-A chromatin. Our data suggest a direct role of p97 in removing centromeric CENP-A via SUMOylated CCAN proteins thereby ensuring centromere homeostasis and potentially preventing ectopic CENP-A accumulation.

SUMO control of centromere homeostasis

“…There has been some evidence that RNF4 is involved, as depletion of RNF4 can rescue centromeric CENP-A levels that are lost in a SENP6 mutant ( Fu et al, 2019 ). However, a recent study could not corroborate this observation ( van den Berg et al, 2023 ). Instead, depletion of RNF4 resulted in a loss of CENP-A rather than suppress excessive SUMOylation, suggesting that RNF4 may play a more direct positive role in maintaining CENP-A.…”

“…However, in contrast to the CCAN components mentioned above, CENP-A itself does not appear to be SUMOylated indicating its levels are controlled via SUMOylation of the downstream CCAN components ( Liebelt et al, 2019 ; Mitra et al, 2020a ). The dynamics of centromere protein loss upon SENP6 depletion showed that while the CCAN proteins CENP-T, CENP-I, CENP-H and CENP-C are rapidly delocalized from the centromere, CENP-A is removed with a delay, indicating its loss is secondary to CCAN protein removal ( van den Berg et al, 2023 ) ( Figure 1C ).…”

“…Recent work revealed that the SUMO-dependent turnover of the human CCAN is dependent on p97 ( van den Berg et al, 2023 ). Interestingly, while CENP-A itself does not appear to be SUMOylated ( Liebelt et al, 2019 ; Mitra et al, 2020a ), it binds to p97 and its turnover is mediated by p97 in a SUMO-dependent manner, indicating that p97 is recruited to extract CENP-A via other CCAN members.…”

“…How p97 is targeted to CENP-A remains to be answered. A recent attempt to identify adaptors found no evidence for the involvement of the canonical factors NPL4 and UFD1, suggesting p97 is targeted by more specialized adaptors ( van den Berg et al, 2023 ).…”

“…Mislocalized CENP-A can also become a substrate for the E3 ubiquitin ligase Psh1 for polyubiquitination ( Hewawasam et al, 2010 ; 2014 ; Ranjitkar et al, 2010 ). Psh1-mediated ubiquitylation does not appears to be SUMO-dependent and in this case Cse4 modification renders it a target for the p97 homolog Cdc48 thereby removing it from the chromatin, revealing a potential analogous mechanism in yeast and humans, although through different signals ( Ohkuni et al, 2022 ; van den Berg et al, 2023 ) ( Figure 2 ). In human cells, non-centromeric CENP-A has been shown to turn over more rapidly than the centromeric pool ( Falk et al, 2015 ).…”