Chromatin regulation through ubiquitin and ubiquitin-like histone modifications

Abstract

Chromatin functions are influenced by the addition, removal, and recognition of histone post-translational modifications (PTMs). Ubiquitin and ubiquitin-like (UBL) PTMs on histone proteins can function as signaling molecules by mediating protein-protein interactions. Fueled by the identification of novel ubiquitin and UBL sites and the characterization of the writers, erasers, and readers, the breadth of chromatin functions associated with ubiquitin signaling is emerging. Here, we highlight recently appreciated roles for histone ubiquitination in DNA methylation control, PTM crosstalk, nucleosome structure, and phase separation. We also discuss the expanding diversity and functions associated with histone UBL modifications. We conclude with a look toward the future and pose key questions that will drive continued discovery at the interface of epigenetics and ubiquitin signaling.

The study of ubiquitin is rooted in chromatin

The early history of ubiquitin discovery is described firsthand [1,2] by those who made three important discoveries. First, ATP-dependent protein degradation was reliant on ubiquitin conjugation by three proteins: E1, E2, and E3. Second, ubiquitin conjugation was selective by way of degron signals. Third, ubiquitin-dependent mechanisms were involved in many cellular processes. Ubiquitin was originally named ubiquitous immunopoietic polypeptide (UBIP) by Goldstein and colleagues who showed that recombinant ubiquitin from bovine thymus could induce mouse T- and B-cell differentiation [3] through immune response mechanisms that are now appreciated as epigenetic in nature [4]. The first protein identified to be ubiquitinated, protein A24, was characterized as a “histone-like” chromosomal protein [5] that had two N-termini and one C-terminus [6]. It is now appreciated that protein A24 is ubiquitinated histone H2A (H2Aub). The two N-termini came from H2A and ubiquitin, and the C-terminus was from H2A, as the ubiquitin C-terminus formed an isopeptide linkage with a lysine side chain on H2A. Since this initial discovery, many physiological processes have been associated with signaling through H2Aub [7]. While H2Aub and H2Bub are well studied for their functions in DNA repair and transcriptional regulation, as reviewed elsewhere [8], we focus this review on recently identified histone ubiquitin and ubiquitin-like modifications and emerging concepts involving ubiquitin signaling as an epigenetic regulator.

Core and linker histones are terminal substrates in the multistep process of ubiquitin attachment

The first step in ATP-dependent ubiquitin transfer is the E1-dependent adenylation of the C-terminus of ubiquitin. A study of the two known human ubiquitin E1 enzymes, UBA1 and UBA6, found a host of nuclear targets for both enzymes [9]. UBA1 was found to associate with DNA breaks via interaction with poly ADP-ribosylated proteins [10]. While UBA1 was found in nuclear and cytosolic fractions, UBA6 was only found in the cytosol [11]. Consistent with cellular locations of the E1 enzymes, UBA1, but not UBA6, was necessary for the DNA damage response [12], suggesting that UBA1 may be the preferred nuclear E1.

The E2 enzyme receives a ubiquitin from the E1 enzyme and carries out the catalytic transfer of ubiquitin to the target protein. Notably, many E2 enzymes appear capable of E3-independent histone ubiquitination in vitro (Figure 1 and Table S1). This is highlighted by the promiscuous UBE2D family, where inclusion of an E2-only control revealed that UBE2B-dependent (E3-independent) ubiquitination of H2A and H2B was as active as the combination of the E3 MSL2-UBE2B, and much more active than MSL2-UBE2D3 or MSL2-UBE2N [13]. Importantly, UBE2B, UBE2H, and UBE2R2 were capable of ubiquitinating all core histones and linker histone H1, independent of an E3 enzyme [14]. It is intriguing to speculate that the most abundant ubiquitinated histone forms (1–1.5% of H2B and 11% of H2A is ubiquitinated [15]) are, in part, a result of E3-independent ubiquitin ligation.

Figure 1 – Regulators of histone ubiquitination.

Ubiquitin writers, erasers, and readers are organized by their associated histone targets and specific lysine residues when known. Unk; histone association is known but target lysine is not defined. Colored boxes correspond to the color of each histone in the nucleosome (PDB ID 5NL0). See also Online Supplemental Information Table S1 for references, data on E2 usage by E3 ligases, and E2 enzymes capable of E3-independent histone ubiquitination.

Finally, an E3 ubiquitin ligase functions to position the E2-ub complex in proximity to the target lysine. The target lysine is nucleophilic toward the C-terminus of ubiquitin and a new attachment is made. Importantly, E3s are scaffolding- or positioning-type proteins, not enzymes in the typical sense of maintaining a catalytic active site. The Really Interesting New Gene (RING), homologous to E6-AP carboxyl terminus (HECT), and RING-between-RING (RBR) are the three types of E3 ligases, with RING being the most common in humans [16]. It is worth noting that free histones often receive more ubiquitin transfer than nucleosomal histones [13,17], suggesting chromatin serves as a barrier or director of some ubiquitin machinery.

Crosstalk between histone and ubiquitin “codes”

Specific attributes of ubiquitin signaling, like substrate and chain type, create a combinatorial signaling “language” referred to as the “ubiquitin code” [18]. Analogous to the “histone code” [19], the ubiquitin code is investigated by studying regulatory proteins referred to as writers (E1, E2, E3), erasers (deubiquitinating enzymes (DUBs)), and readers (ubiquitin binding domains (UBDs)) (Figure 1). In this section, we highlight proteins from each class related to histone targeting, with a focus on the machinery supporting ubiquitination of recently identified sites on core and variant histone proteins.

Studies of the ubiquitin code have primarily focused on how polyubiquitination (or the assembly of various branched ubiquitin forms) is recognized and functions canonically to traffic proteins to the proteasome for degradation. We note that regulation of histone protein stability by the proteasome may be both dependent and independent of polyubiquitination, as reviewed elsewhere [20]. Mono-ubiquitin on histone proteins has roles as a signaling molecule beyond tagging for protein destruction. Both the ubiquitin and histone code hypotheses postulate that post-translational modifications (PTMs) may influence other PTMs to elicit functional outcomes. Consistent with these hypotheses, recent studies have identified connections between small chemical histone PTMs and ubiquitination of chromatin-associated proteins.

Lysine and arginine methylation are prevalent histone PTMs with reported roles in ubiquitin signaling. The ankyrin repeat domains (ARD) of BARD1 recognize the unmethylated state of lysine 20 on histone H4 (H4K20me0), and this domain function was shown to be required for recognition of double strand breaks, homologous recombination, and resistance to PARP inhibitors [21]. As H4K20me0 is enriched at newly synthesized DNA [22], BARD1 ARD recognition of unmodified H4 provides a potential link between newly replicated chromatin and BRCA1-BARD1-dependent histone ubiquitination on all four core histones and the histone variant H2A.X (Table S1). The DNA damage repair protein 53BP1 is a multivalent reader of both H4K20me2 and lysine 15 ubiquitination on histone H2A (H2AK15ub) through its tandem Tudor domain and UBDs, respectively [23,24]. However, there is no consensus on which domain is the primary driver of 53BP1 recruitment to chromatin. Another reader of lysine and arginine methylation with links to ubiquitin is the Polycomb Repressive Complex 2 (PRC2)-associated protein PHF1. PHF1 reads H3K36me3 through its Tudor domain, symmetric di-methylation of arginine 3 on histone H4 (H4R3me2s) through its N-terminal PHD, and DDB1 through its C-terminal PHD [25,26]. As DDB1 is a member of a promiscuous histone CUL4 E3 ligase complex [27], it is intriguing to speculate that the methyl-reader domains of PHF1 are involved in a ubiquitin-dependent interaction between PRC2 and chromatin.

There is a growing body of literature detailing the crosstalk between lysine acetylation and ubiquitination in the DNA damage response. Depletion of lysine acetyltransferases KAT5 (TIP60) or PCAF resulted in decreased acetylation of lysine 120 on H2B (H2BK120ac) and increased H2BK120ub following DNA damage [28]. This ‘acetyl-ubiquitin switch’ suggests that acetylation functions at certain residues to block ubiquitin signaling. Another example of ubiquitin regulation by acetylation is found between EP300 (p300) and KDM3A (JMJD1A, JHDM2A). EP300 acetylates JMJD1A, a H3K9me2 demethylase [29], on lysine 421 which blocks polyubiquitination by STUB1, thus maintaining JMJD1A levels [30]. This study reveals a potential ubiquitin-dependent link between acetylation and methylation on both histone and non-histone proteins.

Ubiquitination and phosphorylation also have reported connections in chromatin regulation. EGFR stimulates lysine 48 linked polyubiquitination of H3K4 by RNF8-UBE2L6 to induce H3 degradation [31]. Both phosphorylation and ubiquitination were blocked however, when threonine 11 on H3.3 (H3.3T11ph) was mutated to alanine [31]. These findings implicate signaling between growth factors and kinase-dependent phosphorylation of H3T11 in the regulation of histone abundance.

Numerous E3 ligases have the capability for histone PTM recognition. PHRF1 and TRIM33 are two E3 ligases that contain histone reader domains; PHRF1 has a PHD finger [32,33], and TRIM33 has a PHD-Bromodomain [34]. The UHRF1 and UHRF2 E3 ligases also have histone and DNA reading domains, the former will be discussed below for its role in DNA methylation maintenance (Figure 2).

Figure 2 – Ubiquitin-dependent regulation of DNA methylation maintenance.

(left) The E3 ligase UHRF1 writes multiple mono-ubiquitin sites on histone H3 and PAF15, (middle) both of which can bind to the ubiquitin interacting motif (UIM) in the replication foci targeting sequence (RFTS) of the DNMT1 maintenance DNA methyltransferase. Engagement of ubiquitinated proteins by the DNMT1 UIM enhances DNMT1’s catalytic activity to facilitate maintenance methylation. While longer H3 peptides used for crystallization (as indicated in the Figure), only residues 2–20 and 2–19 were resolved for H3K18ubK23ub and H3K9me3K18K23ub, respectively. (right) The DUB USP7 is potentially involved in DNA methylation regulation by controlling stability of other DNA methylation regulators and by directly removing H3 and other ubiquitin marks read by DNMT1. The indicated USP7 inhibitors are in various stages of preclinical evaluation.

As the breadth of histone PTM writer, eraser, and reader activities directly and indirectly associated with ubiquitin regulators continues to grow, the parallel concepts put forward in the ubiquitin and histone code hypotheses are beginning to intersect. It is becoming clear that both small and large chemical attachments to chromatin-associated proteins are part of a dynamic and collective signaling response that functions to fine-tune the regulation of DNA-templated cellular processes.

Methyl degrons connect lysine methylation to the ubiquitin-proteasome system

A methyl degron is a methylation-dependent ubiquitination event that leads to protein degradation. So far, methyl degrons only exist for lysine methylation. This process involves a methyltransferase, methyllysine reader, and a Cullin-RING ubiquitin ligase. The first study to describe this mechanism reported that DCAF1, a CUL4 substrate-recognition subunit (SRS), bound to RORα at methylated lysine 38 and led to its proteasomal degradation [27]. Methyl degron mechanisms have also been reported for DNA methyltransferase 1 (DNMT1) and the transcription factor E2F1, where the methyllysine-reading L3MBTL3 and DCAF5 serve as the CUL4 SRS [35]. These studies demonstrated roles for lysine methylation in the regulation of protein stability, in a way that was previously unrecognized.

Additionally, the stability of UHRF1, a DNMT1 cofactor and E3 ubiquitin ligase, is regulated by SET8 lysine methylation [36]. In this study, SET8 overexpression increased UHRF1 ubiquitination. Furthermore, when lysine demethylase 1 (LSD1) was knocked out, a decrease in UHRF1 protein was noted. It was however inconclusive whether changes in UHRF1 protein stability, through antagonistic activities of SET8 and LSD1, affected global DNA methylation. While an increasing number of proteins are reported to contain methyl degrons, the broader functions of this regulation have yet to be demonstrated with certainty.

Nucleic acid-binding writers of histone ubiquitination

Histone-targeting E3 ligases are likely in close proximity to nucleic acids, assuming their action is directed toward chromatin and not free histones. Consequently, it is unsurprising that a handful of ubiquitin writers bind directly to DNA or RNA. SHPRH, an E3 ligase, bound to 147 bp and 227 bp DNA; with an apparent reduction in affinity for DNA when it was assembled into nucleosomes [37]. However, the DNA binding function of SHPRH has not been explicitly linked to its E3 ligase activity. DZIP3 is an RNA-binding E3 ligase [38] known to write H2AK119ub [39] and TP53ub [40], but again the exact link between RNA binding and E3 ligase activity is unclear. RNF4 is a DNA binding E3 ligase whose activity toward H3 is enhanced on poly-nucleosome substrates, possibly due to binding to linker DNA [41]. This increased activity may be due to allosteric activation, as has been reported for the DNA-binding E3 ligases UHRF1 and UHRF2 [17,42]. The E3 ligase activities of UHRF1 and UHRF2 are stimulated by binding to DNA through their SET and RING associated (SRA) domains. For UHRF1, the addition of linker DNA to recombinant mononucleosomes stimulated histone ubiquitination in a manner dependent on its interaction with histones and its UBL (a binding site for the E2 enzyme UBE2D1) and RING domains [43–45]. The UHRF1 SRA domain binds DNA with increased affinity for hemi-methylated forms relative to unmodified DNA. While binding to hemi-methylated DNA induces a conformational change in UHRF1 [17,46], it is unknown how the addition of a single methyl group to a cytosine nucleotide is able to regulate the E3 ligase activity of UHRF1, especially toward itself [43]. Future work in this area will determine whether E3 ligases binding to DNA/RNA enhance ubiquitination of targets through allosteric regulation, increasing binding affinity to substrate, or a mechanism yet to be described.

The maintenance of DNA methylation patterns is emerging as a ubiquitin-dependent process

UHRF1 writes multiple mono-ubiquitin marks on both histone H3 [47] and PCNA-associated factor (PAF15/PCLAF) [48]. The multiple mono-ubiquitin marks are read by the DNMT1 ubiquitin interacting motif (UIM) [49,50] and serve as a binding site on chromatin for DNMT1. When bound to DNA, DNMT1 catalyzes methylation of cytosines on the newly synthesized daughter strand of DNA (Figure 2). There is no consensus however among published studies on how much of the genome requires UHRF1 for its maintenance methylation. UHRF1-independent DNA methylation maintenance may be facilitated by the affinity of DNMT1 for DNA or by other E3 ligases capable of writing H3 multi-mono-ubiquitin, like NEDD4 or CBL [51]. Ubiquitin marks involved in the regulation of DNA methylation (at least on histone H3 and H2B) are erased by USP7 [52,53]. USP7 also controls the stability of DNMT1 [54] and UHRF1 [55] and is now under clinical investigation as a cancer therapeutic with several drugs recently developed (Figure 2) [56–58].

Ubiquitin-like modifiers are found on many histone proteins

In addition to modification by ubiquitin, there are several ubiquitin-like proteins (UBLs) whose functions are still emerging. UBLs share similar structural features and enzymatic cascades for attachment to target proteins. Despite similar secondary and tertiary structures, the electrostatic surface potential varies greatly between UBLs (Figure 3a). Functional consequences of UBL attachment may be recognized through UBL-interacting domains, harnessing specific charges on the surface of each UBL. High-throughput studies have identified many histones as targets for UBL modification. However, there are relatively few functions attributed to histone-UBL attachment. We focus here on studies that identified UBL-histone attachments and some associated biological consequences.

Figure 3 -. Structural insights into ubiquitin-like proteins and nucleosomal ubiquitin.

(a) Cartoon representations of ubiquitin and ubiquitin-like domain secondary structure (upper), and electrostatic potential (lower). Alignment was performed by command cealign in PyMOL and electrostatic potential was calculated using the APBS module in PyMOL. Units are given in energy per charge (k, Boltzmann constant; T, temperature in Kelvin; e, charge of an electron). Only single UBL units are shown for ISG15 and FAT10. PDB IDs used in order of appearance are 1UBQ, 5IA7, 1NDD, 6GF1, 1Z2M, 1P0R, 4WJQ, 1WM2, 1U4A. (b) Structures of nucleosomes containing ubiquitinated histones were aligned based on the histone and DNA; H2Aub nucleosomes (upper), H2Bub nucleosomes (lower). PDB IDs for each ubiquitin are indicated by color and the nucleosome core particle is from PDB 6PX1.

The ubiquitin-like protein FAT10 (ubiquitin d) is essentially a di-ubiquitin molecule, whose flexible linker between the two ubiquitin-like domains modulates its activation and conjugation [59]. The histone variant macro H2A.1 was immunoprecipitated with an anti-FAT10 antibody from HEK293 cells after stimulation by IFNγ and TNFα [60]. Linker histone H1.2, H2B, and variant histones H2A.Z, H2A.X, and H2A.V immunoprecipitated with protein G-tagged FAT10 overexpressed in HeLa cells [61]. However, there are no functions attributed to histone FAT10ylation.

Interferon-stimulated gene 15 (ISG15) is a UBL that also resembles two linked ubiquitin molecules. In order to identify sites of ISGylation in vivo, Zhang et al. compared wild type mice, Isg15^−/−, and Usp18^C61A/C61A; the latter is a catalytically dead ISG15 isopeptidase mutant that results in more conjugated ISG15 [62]. The authors then compared KGG enriched peptides (a remnant from trypsin digestion of ubiquitin, ISG15, and NEDD8 attachments to target lysines) from the liver of each mouse genotype after infection with Listeria monocytogenes and found H3.3K23isg, H2A.XK119isg, and H2BK5isg. As DNMT1 is a reader of multi-mono-ubiquitinated proteins [50,63], it is interesting to speculate that the DNMT1 UIM may interact with FAT10 and/or ISG15.

Neural precursor cell expressed developmentally down-regulated protein 8 (NEDD8) is structurally similar to ubiquitin. Canonical histones H2A, H2B, and H4 were pulled down by His-NEDD8 expressed in HEK293T cells [64]. The same study found that RNF168 was responsible for H2A neddylation and that neddylation and ubiquitination of H2A were competitive. RNF111-UBE2M neddylates several lysines on the N-terminus of H4, which recruits the DNA damage response factor RNF168 to chromatin through its UIM [65]. Interestingly, the de novo DNA methyltransferases DNMT3A/B interact directly with NEDD8 through a region mapped between their ADD and catalytic domains [66], adding another possible link between DNA methylation and UBLs.

While small ubiquitin-like modifiers (SUMO1/2/3) have been identified on all core histones, some H2A variants, and histone H1, the most prominent site of SUMO1/3 appears to be on lysine 12 of histone H4 (H4K12sumo) [67]. H4K12sumo prevented magnesium-dependent chromatin compaction, albeit to a lesser extent than H4K16ac. It is also noteworthy that SUMO-modified proteins accumulate in PML bodies after inhibition of ubiquitination [68], suggesting a function for maintenance of genomic integrity.

Ubiquitin-like protein 5 (UBL5, HUB1 in yeast) is a UBL with a C-terminal di-tyrosine motif. It remains unclear whether UBL5 makes covalent attachments like other UBLs. However, UBL5 pulled down various histones and was implicated in mRNA splicing and sister chromatid cohesion [69]. Additionally, UBL5 has been found near Cajal bodies, suggesting involvement in phase separation of splicing components [70].

Ubiquitin-fold modifier 1 (UFM1) maintains structural similarity to ubiquitin, despite minimal conservation in their amino acid sequences. In vitro UFMylation did not deposit UFM1 on H2A, H2B, or H3 and only wrote mono-ufmylation on H4K31 [71]. The previous study found that H4K31ufm was involved in AKT activation and the DNA damage response. In another study, H2A and H2A.V were identified after pulldown by 6xHis-UFM1 genetically complemented in UFM1 knockout cells [72], although spectral counts were low for each histone in this mass spectrometry dataset.

The flexibility of ubiquitin and UBL attachment presents a challenge for structural studies

Resolved structures of ubiquitin molecules attached to nucleosomes have also included ubiquitin reader proteins, presumably restricting the flexibility of ubiquitin attachment. Of the published histone-ubiquitin structures (Table 1), the majority of nucleosomal structures have been solved with C-terminal H2Bub, and two structures have been solved with N-terminal H2Aub, a surrogate for C-terminal H2Bub (Figure 3b). As far as H3ub is concerned, only two studies have resolved H3ub and they have been only in the context of short peptides [50,73]. The general consensus from structural work is that reader proteins make contacts with both ubiquitin and the acidic patch on the nucleosome, perhaps offering an explanation as to the general consistency in the location of ubiquitin relative to the nucleosome (Figure 3b). While H4K31ub-containing nucleosomes have been used for structural studies [74,75], both groups resolved the nucleosome core particle while failing to resolve ubiquitin. These data suggested flexibility of H4K31ub. Models of the combination H2Bub/H4K31ub nucleosomes [74] posit both ubiquitin molecules can occupy a similar space. There are no structures of H3 or H4 ubiquitination in a nucleosomal context. Perhaps larger nucleosomal arrays will help reveal how ubiquitin (and UBL) attachment to nucleosomes modulates the large-scale chromatin architecture.

Table 1 –

Structural studies of ubiquitinated histones

PDB	Histoneub	Chemistry of Ubiquitin Linkage	Other Histone PTMs	Bound Proteins	Reference
6PX1	H2A	Ubiquitin fused to N-terminus of H2A via Ser-Gly linker	H3K36M	-	[90]
6NZO	H2A	Ubiquitin fused to N-terminus of H2A via Ser-Gly linker	H3K36M	Set2 (C. thermophilum)	[90]
6PX3	H2A	Ubiquitin fused to N-terminus of H2A via Ser-Gly linker	H3K36M	Set2 (C. thermophilum)	[90]
5KGF	H2A	RNF168 ubiquitination of H2A(K13R/K36R)/H2B dimers	H4kc20me2	53BP1 (H. sapiens)	[23]
6JMA	H2B	DCA-linkage between ub G76C and H2BK120C	-	DOT1L (H. sapiens)	[91]
4ZUX	H2B	DCA-linkage between ub G76C and H2BK120C	-	SAGA DUB module (S. cerevisiae)	[92]
6T9L	H2B	DCA-linkage between ub G76C and H2BK120C	H3Kc4me3	SAGA DUB module (S. cerevisiae)	[93]
6UH5	H2B	Disulfide bond between ub G76C and H2BK120C	-	COMPASS (K. lactis)	[94]
6NN6	H2B	DCA-linkage between ub G76C and H2BK120C	-	DOT1L (H. sapiens)	[95]
6NQA	H2B	DCA-linkage between ub G76C and H2BK120C	H3K79Nle, H3M90Nle, H3M120Nle	DOT1L (H. sapiens)	[96]
6NJ9	H2B	DCA-linkage between ub G76C and H2BK120C	H3k79Nle, H3M90Nle, H3M120Nle	DOT1L (H. sapiens)	[96]
6NOG	H2B	DCA-linkage between ub G76C and H2BK120C	-	DOT1L (H. sapiens)	[96]
6FTX	H2B	DCA-linkage between ub G76C and H2BK120C	H3Kc36me3	Chd1 (S. cerevesiae)	[97]
6KIW	H2B	DCA-linkage between ub G76C and H2BK120C	-	MLL3, WDR5, RBBP5, ASH2L (H. sapiens)	[98]
6KIU	H2B	DCA-linkage between ub G76C and H2BK120C	-	MLL1, WDR5, RBBP5, ASH2L (H. sapiens)	[98]
6VEN	H2B	DCA-linkage between ub G76C and H2BK120C	H3K4Nle, H3M90Nle, H3M120Nle	COMPASS (S. cerevesiae)	[99]
6J99	H2B	DCA-linkage between ub G76C and H2BK120C	-	DOT1L (H. sapiens)	[100]
6PZV	H3(1–25)	Disulfide bonds between ub G76C and H3K18C/K23C	H3K9me3, H3R26W	DNMT1 RFTS/UIM (B. taurus)	[73]
5WVO	H3(1–37)	Disulfide bonds between ub G76C and H3K18C/K23C	H3K37W	DNMT1 RFTS/UIM (H. sapiens)	[50]

Approaches toward disrupting ubiquitin signaling on chromatin for cancer therapy

Owing to the multi-step process of ubiquitin attachment, there are many potential places to develop inhibitors. We note that inhibitors are under development for nearly every step in ubiquitin regulation, as comprehensively reviewed elsewhere [76]. We focus here on antagonists of ubiquitin signaling for potential cancer therapies. Several inhibitors targeting ubiquitin signaling are in clinical trials for cancer therapy. PTC596 was found to reduce BMI1 protein load, a subunit of the PRC1 complex required for writing H2AK119ub, and has shown efficacy as an anticancer agent in xenograft rodent models for leukemia [77]. This molecule has completed a phase 1 clinical trial for advanced stage solid tumors (NCT02404480) and was deemed safe for human use with ‘manageable gastrointestinal side effects’ [78]. Currently, PTC596 is in a phase 1b trial for ovarian cancer (NCT03206645), leiomyosarcoma (NCT03761095), and childhood gliomas (NCT03605550). Significant efforts to therapeutically target USP7 (Figure 2), an eraser of H2AK119ub and H3ub (Figure 1) are underway [56,57]. Inhibition of USP7 by XL177A upregulates TP53 target genes [58] and additional effects may be elicited through H3ub with potential links to DNA methylation regulation. Modulation of ubiquitin signaling at the upstream level of ubiquitin adenylation via inhibition of the E1 enzyme is another therapeutic strategy that has progressed to human trials. Inhibition of UBE1 by the small molecule TAK-243 (MLN7243) holds promise as an anticancer treatment, as it induced cell cycle arrest and ER stress, impaired DNA damage repair in human cancer cell lines, and demonstrated antitumor activity in mice with human xenografts [79]. TAK-243 had a phase 1 clinical trial (NCT02045095) in advanced solid tumors terminated (for sponsorship reasons) and another phase 1 trial with this molecule is being planned for several hematologic malignancies (NCT03816319).

Phase separation and ubiquitination of chromatin

The separation of cellular components into membrane-less structures has long been described with varying terminology [80]. Functional consequences associated with ‘phase separation’ of chromatin and chromatin regulators are emerging, and ubiquitination appears to be a regulator of this phenomenon. A key observation was an increase in ubiquitin-containing cellular aggregates after inhibition of histone deacetylases (HDACs) by treatment with SAHA (vorinostat) [81]. In addition, treatment of Jurkat or primary human peripheral blood mononuclear cells by doxorubicin resulted in aggregation of H2A and H2B, with the cytoplasmic accumulation of H2B being reversible by treatment with the E1 inhibitor PYR-41 [82]. In Drosophila melanogaster, lipid droplets are a known nuclear reservoir for the histone variant H2A.V [83], raising the possibility that ubiquitin and histone sequestration in phase-separated bodies are linked.

The higher the valency of protein interactions, the higher the propensity to phase separate [84]. As many of the ubiquitin regulators have multiple ubiquitin-like and ubiquitin binding domains, it reasons that ubiquitin may be a regulator of phase separation. Indeed, ubiquitin can both positively and negatively regulate phase transition of proteins. The proteasome shuttling protein UBQLN2 phase separates and is associated with stress granules in cells, a behavior that may be dependent on its ability to self-oligomerize [85]. However, the previous study found that interactions with ubiquitin (and polyubiquitin) transitioned UBQLN2 from condensed to disperse, demonstrating a role for ubiquitin in removing proteins from phase separated bodies.

Hyperosmotic stress caused nuclear accumulation of proteasomes in droplet formations in a process that was both dependent on RAD23B binding to multiple ubiquitin chains and blocked by the E1 inhibitor TAK-243 [86]. The previous study showed that after hyperosmotic stress, histones H1.0, H1.2, H1.4, H1.X, H2B, and H3.1 were among the most ubiquitinated proteins. These data show a positive regulatory function of ubiquitin in nuclear phase separation and implicate involvement with ubiquitinated histones.

Mechanistic connections are being made between ubiquitin transfer and molecular condensation. SPOP is a CUL3 SRS that self-oligomerizes through BACK and BTB domains. Both BACK and BTB domains are necessary to localize to nuclear speckles [87], demonstrating a possible recruitment mechanism for CUL3 complexes to phase separated bodies. A potential connection to cancer was made with SPOP, whereby disease-associated mutations caused SPOP to phase separate in vitro [88]. This data suggests that E3 ligase dysregulation is secondary to inappropriate condensation. In yeast, Bre1-Lge1 made condensates that brought Rad6 (UBE2B) and nucleosomes into contact, increasing H2B ubiquitin [89]. In the previous context, the phase separated bodies are, in some sense, acting like the E3 ligase and increasing the frequency of contacts between the E2 and substrate.

Concluding Remarks

Further connections between chromatin and ubiquitin are bound to be made through continued studies of multidomain ubiquitin regulators. We anticipate that similar connections exist for many of the UBL modifications discussed in this review. Structural studies, enabled largely by single particle cryo-electron microscopy, will likely aid in revealing roles for ubiquitin and UBL histone modifications as modulators of chromatin architecture. We also anticipate techniques that enable high resolution imaging of cellular components, like cryo-electron tomography, may find utility in revealing roles for ubiquitin in the phase separation of nuclear elements and chromatin. Finally, as human clinical trials are now underway with compounds that modulate the writing and erasing of ubiquitin modifications, determining the extent to which these molecules function through chromatin regulation will be important for understanding both the success and failure of clinical ventures targeting the ubiquitin system. The continued study at the interface of epigenetics and ubiquitination (see Outstanding Questions) will surely reveal fundamental biological mechanisms and associated pathologies.

Outstanding Questions.

• What is the breadth of ubiquitin and ubiquitin-like modifications that are associated with chromatin?

• What are the structural and functional consequences associated with histone tail and histone core domain modifications?

• How do chromatin architecture, DNA modifications, and the histone code influence activities of ubiquitin and ubl writers, erasers, and readers?

• To what degree are the functions of ubiquitin-like modifiers (FAT10, ISG15, SUMO, UFM1) conserved or divergent in the regulation of chromatin-related processes?

Highlights.

• Histones are among the most abundant ubiquitinated proteins in cells, perhaps due to their E3-independent ubiquitination by many E2 enzymes.

• The histone and ubiquitin “codes” function together to regulate DNA-templated cellular processes.

• The maintenance of DNA methylation through cell division is emerging as a ubiquitin-dependent process.

• Inhibitors of ubiquitin transfer are an emerging class of targeted therapeutics for cancer management.

• Ubiquitin and ubiquitin-like proteins may play key roles in phase separation of chromatin and other nuclear components.