Defining biological and biochemical functions of noncanonical SET domain proteins

Abstract

Within the SET domain superfamily of lysine methyltransferases, there is a well-conserved subfamily, frequently referred to as the Set3 SET domain subfamily, which contain noncanonical SET domains carrying divergent amino acid sequences. These proteins are implicated in diverse biological processes including stress responses, cell differentiation, and development, and their disruption is linked to diseases including cancer and neurodevelopmental disorders. Interestingly, biochemical and structural analysis indicates that they do not possess catalytic methyltransferase activity. At the molecular level, Set3 SET domain proteins appear to play critical roles in the regulation of gene expression, particularly repression and heterochromatin maintenance, and in some cases, via scaffolding other histone modifying activities at chromatin. Here, we explore the common and unique functions among Set3 SET domain subfamily proteins and analyze what is known about the specific contribution of the conserved SET domain to functional roles of these proteins, as well as propose areas of investigation to improve understanding of this important, noncanonical subfamily of proteins.

The SET (Su(var)3–9, Enhancer-of-zeste, and Trithorax) domain superfamily of proteins is well-characterized as a group of histone-modifying enzymes that regulate critical gene expression programs, among other chromatin functions, and its dysregulation is frequently linked to disease processes including cancer, developmental disorders, and pathologies associated with aging [1]. The SET domain is a catalytic methyltransferase domain which has predominantly been shown to transfer a methyl group from S-adenosylmethionine (SAM) to the ε-amino group of lysine, though other amino acid substrates have also been identified [2]. Furthermore, while histones are a primary substrate for SET domain enzymes, there is substantial evidence for non-histone protein methylation catalyzed by these enzymes and an expanding list of known substrates [3, 4], highlighting the diverse roles of the SET domain superfamily and of protein methylation in regulating the proteome.

There are multiple subfamilies within the SET domain superfamily, each of which are characterized by unique sequence or structural attributes and, frequently, functional similarities[4, 5]. One such SET domain subfamily has been defined based on similarity of domain organization, primary sequence, and biological function to the Saccharomyces cerevisiae protein Set3 [6], and is frequently referred to as the Set3 subfamily. Set3 has a paralog in budding yeast known as Set4, which is 50.7% identical across the SET domain and 31.51% identical across the entire protein (Figure 1; Table 1), although their functions appear to diverge under different environmental conditions [7, 8]. The Set3 subfamily proteins are highly conserved across species (Figure 1) and are generally characterized by the presence of two major protein domains: the PHD finger, which is a chromatin reader domain that frequently recognizes methylated H3 lysine 4 (H3K4) [9, 10], and the SET domain, which diverges from canonical SET domains at the sequence level, and likely in enzymatic activity [11]. Over the last several years, there have been new investigations into the biological functions and biochemical activities of Set3 subfamily members, particularly in yeast, worms (C. elegans), mouse, and humans [12–17]. In this Review, we discuss our current understanding of the contributions of these proteins to diverse biological pathways and evaluate the potential functions of the noncanonical SET domain in Set3 subfamily proteins.

Figure 1. Sequence alignment of SET domains of Set3 SET subdomain family members.

Representative domain structure of Set3 SET subdomain family protein containing PHD and </p/> SET domains, except for SETD5 in zebrafish, mouse, rat, and human which lack PHD domains. Alignment of residues compared to motifs important for cofactor or substrate binding or active site in the catalytically active SET protein MLL1 (indicated in bold) [5, 11, 24]. Human MLL1 sequence was aligned to human MLL5 and SETD5 sequences and S. cerevisiae Set3 and Set4 sequences, then MLL1 sequences were mapped to the alignment in Figure 1. Sequence alignment performed and percent identity calculated with Clustal Omega and Jalview [54, 55]. SMART domain annotations of SET domains were used when possible. The following amino acid residues were used for each organism: S. pombe Set3 215–343, S. cerevisiae Set3 321–462, S. cerevisiae Set4 346–481, C. elegans set-9 930–1062, C. elegans set-26 938–1072, D. rerio Mll5 331–448, D. rerio Setd5 329–450, D. melanogaster UpSET 1118–1282, M. musculus Mll5 328–453, M. musculus Setd5 272–396, R. norvegicus Mll5 328–453, R. norvegicus Setd5 276–397, H. sapiens MLL5 328–453, and H. sapiens SETD5 272–396.

Table 1.

Functions of Set3 SET domain subfamily proteins and similarity of their SET domains to S. cerevisiae Set3 and Set4 SET domains

Organism	Protein Name	Accession #	% Identity (rel. Set4 SET)	% Identity (rel. Set3 SET)	Function	References
S. cerevisiae	Set3	NP_012954	50.70	100	Transcription repression of sporulation genes, transcription repression from cryptic promoters	[6, 9, 43]
S. cerevisiae	Set4	NP_012430	100	50.70	Azole resistance, hypoxic stress response, oxidative stress response, gene repression	[7, 25, 26]
S. pombe	Set3	NP_594837	28.68	25.00	Heterochromatin silencing/integrity, cytokinesis	[48, 56]
C. elegans	Set-9	NP_001379880	24.82	19.73	Lifespan, germline development	[57]
C. elegans	Set-26	NP_001370101	24.82	19.73	Lifespan, germline development, heat stress response	[12, 49, 57]
D. rerio	Mll5	XP_009298714	27.82	22.79	Neurodevelopment	[58]
D. rerio	Setd5	ABI34483.1	25.56	21.58	Larval development	[50]
D. melanogaster	UpSET	NP_001261819	19.85	19.53	Oogenesis, development, cell specification	[18]
M. musculus	Mll5	NP_081260	28.06	20.57	Hematopoiesis, cell proliferation, spermatogenesis, myeloid differentiation, innate immune response, retinal development	[14, 23, 59, 60]
M. musculus	Setd5	NP_082661	25.18	20.83	Cell cycle progression, embryonic development, neurodevelopment, somitogenesis, cardiac development, vasculogenesis, primordial germline specification, adipogenesis	[17, 24, 29, 51, 61–63]
R. norvegicus	Mll5	NP_001094321	27.34	20.57	-	-
R. norvegicus	Setd5	NP_001100084	25.56	21.58	-	-
H. sapiens	MLL5	NP_001397837	28.78	19.86	cell cycle progression, large intestine cancer, ovary cancer, central nervous system cancer, stomach cancer, pancreatic cancer, thyroid cancer, breast cancer, neurodevelopment, autism spectrum disorder	[15, 35, 64]
H. sapiens	SETD5	NP_001073986	25.18	20.83	neurodevelopment, intellectual disability, prostate cancer, non-small cell lung cancer, esophageal squamous cell carcinoma, pancreatic ductal adenocarcinoma, breast cancer, glycolysis, hepatocellular carcinoma, gene repression	[16, 31, 36, 37, 41,42, 65]

Biological and molecular functions of Set3 SET domain subfamily proteins

Sequence analysis and molecular investigation has identified Set3 subfamily members in most eukaryotic species, with a subset listed in Table 1 and sequence alignment of their SET domains shown in Figure 1. In addition to the S. cerevisiae proteins Set3 and Set4, set-9 and set-26 have been identified in C. elegans [12], UpSET inD. melanogaster [18–20], and MLL5/KMT2E (referred to as MLL5 hereafter) and SETD5 in vertebrate and mammalian species [16, 21–23]. As discussed further below, sequence and structural analysis, as well as biochemical activity assays, indicate that these proteins are not likely to have catalytic methyltransferase activity [7, 11, 16, 22]. However their conservation, including retaining a likely non-catalytic SET domain, suggest that they play important functional roles. Indeed, as outlined in Table 1, genetic characterization of mutated or null alleles of Set3 subfamily proteins in different systems has revealed critical links to cell fate specification and differentiation, numerous developmental processes and particularly germline development, stem cell maintenance, cell stress response pathways, cell cycle progression and meiosis, and lifespan, among other processes. While deletion of Set3 or Set4 in yeast species is not lethal, clear defects emerge in processes including sporulation, stress response pathways, and, particularly relevant for pathogenic fungi, a role in azole resistance, biofilm formation, and virulence [6, 7, 11, 13, 16, 24–28]. In mammals, Setd5^−/− mice show embryonic lethality [29] and loss of Mll5 leads to partial postnatal lethality [23], indicating a critical requirement for these proteins in development. Furthermore, altered expression or mutation of mammalian MLL5 and SETD5 is linked to genetically-defined neurodevelopmental disorders. For example, haploinsufficiency of SETD5 in mice causes behavioral and cognitive abnormalities [24], and SETD5 loss-of-function variants are linked to 3p25 microdeletion syndrome [30, 31], a neurodevelopmental disorder. Genome-wide association studies have identified MLL5 variants linked to autism spectrum disorders (ASD) and intellectual disability (ID) [21, 32], and the neurodevelopmental disorder O’Donnell-Luria-Rodan syndrome has recently been identified and is associated with pathogenic variants of MLL5 [15, 33, 34]. Aberrant expression or mutation of both MLL5 and SETD5 has also been implicated in multiple types of cancer [35–37], including leukemias and solid tumors for MLL5 [38–40], and numerous solid tumors and roles in chemotherapeutic resistance for SETD5 [16, 41, 42]. Altogether, genetic and phenotypic data obtained on all organisms studied to date carrying deletions, mutations, or improperly expressed alleles of Set3 subfamily proteins demonstrate a clear contribution of these proteins to physiological pathways and indicate that proper functioning of these proteins is required to limit disease.

At the molecular level, the predominant, known function of these SET domain proteins is the regulation of gene expression, with numerous subfamily members linked to the maintenance of heterochromatin and gene repression in different contexts (Table 1). In cases where it has been investigated, the mechanism through which these proteins regulate gene expression, particularly in the absence of apparent catalytic activity, is predominantly through the recruitment and scaffolding of other chromatin modifiers, particularly histone deacetylases (HDACs) at specific genomic regions [8]. As an example, S. cerevisiae Set3, the first molecularly characterized and hence defining member of this subfamily, is a component of the Set3 complex (Set3C), which is comprised of the two HDACs Hos2 and Hst1, as well as Snt1, Hos4, Sif2, and Cpr1 [6]. Set3C is recruited to chromatin at least in part by the PHD finger of Set3 interacting with the H3K4me2 mark and it represses cryptic transcription through the activity of its associated HDACs [6, 9, 43]. A similar complex exists in S. pombe, though composed of Set3, Hif2, Snt1, and the HDAC Hos2. The D. melanogaster protein UpSET also interacts with the Rpd3/Sin3 HDAC complex, in which it represses aberrant transcription at regions of high gene expression activity and regulates heterochromatin [18–20]. In mammalian species, a diverse set of protein partners has been identified to interact with human MLL5 [44–46], although its chromatin-specific interacting partners that may drive gene expression changes are not yet fully defined. SETD5 has been reported to interact with the NCoR1/HDAC3 complex and the PAF1 complex in multiple cell types [16, 24, 29], as well as with the histone methyltransferase G9a to repress gene expression [16]. These findings from diverse systems suggest that a key role for these noncanonical SET domain proteins in gene expression regulation is likely the recruitment and stabilization of other factors, particularly histone modifying enzymes such as HDACs, at particular regions of the chromatin to control developmental and stress response genes, as well as aberrant or cryptic transcription.

Biochemical assessment of SET domain function in Set3 subfamily proteins

In evaluating the role for Set3 SET domain proteins in diverse biological pathways, there have been both genetic and biochemical investigations of the function of the PHD finger and the SET domains within these proteins. In yeast Set3, worm set-9 and set-26, fly UpSET, and mammalian MLL5, the PHD finger has been shown to recognize H3K4 methyl species and stabilize protein binding at chromatin. Furthermore, deletion or mutation of the PHD finger impairs protein function, indicating its binding activity is a critical component to the activity of these proteins [12, 20, 47, 48]. However, the specific role for the noncanonical SET domain within these proteins has been less clear and its contribution to protein function has not been fully evaluated in many cases.

Sequence alignment of the noncanonical SET domains of the Set3 subfamily members show residue substitutions and a number of regions within the active site that are normally conserved in canonical, catalytically active SET domain proteins (Figure 1). One such region is the asparagine-histidine (NH) motif, highlighted in Figure 1, which is crucial for hydrogen bonding with the methyl donor SAM and is altered to arginine-arginine (RR) across all members of the Set3 SET domain subfamily. In MLL5, the first R in the altered NH motif has been shown to inhibit proper positioning of a loop within the SET domain, causing the cofactor docking site to be sterically blocked [11]. Another key residue is the tyrosine (Y) motif, which is involved in target lysine positioning [5] in the catalytic site. This is altered to either phenylalanine (F) or tryptophan (W) in the Set3 SET domain subfamily, with the exception of fly UpSET, which has a histidine (H) in that position. Furthermore, the XGXG motif at the beginning of the SET domain sequence, which plays a role in cofactor binding [11], is absent from Set3 SET subfamily proteins. These noncanonical residues and motifs, as well as additional changes highlighted in Figure 1, are a key indicator that these SET domains are not likely to possess catalytic methyltransferase activity and that they diverge from canonical SET domains in their function.

To experimentally assess the catalytic activity of these SET domains, biochemical assays of recombinant or purified endogenous proteins have been performed extensively to test for methyltransferase activity of Set3-related proteins. In budding yeast, TAP-purified Set3C lacked methylation activity [6], and recombinant Set4 did not show activity on multiple histone substrates in vitro [7]. Recombinant fly UpSET did not methylate purified histones in vitro either on its own, or with the addition of an O-GlcNAc transferase, which has been proposed to stimulate methyltransferase activity for MLL5 [44], or with nuclear extracts [18]. Recombinant set-26 was reported to methylate histones at H3K9 using purified histones or nucleosomes in vitro [49], however a subsequent report showed that loss of set-26 along with its paralog set-9 did not have any genome-wide changes in H3K9me3 levels or distribution [12], suggesting H3K9 is not likely a physiological target of set-26.

For the mammalian proteins, initial reports showed that neither the human, recombinant SET domain of SETD5 nor full-length SETD5 purified from HEK293T cells was able to methylate nucleosomes in vitro [24]. A subsequent report suggested that a truncated mouse Setd5 did methylate nucleosomes in vitro and specifically targeted H3K36 [50]. However, comprehensive investigation of multiple sources of SETD5 recombinant protein, including mouse and human, as well as a SETD5-containing complex from HEK293T cells, showed that it lacked methyltransferase activity on histone, nucleosomes, over 9000 human proteins on a protein array, and a fractionated human cell extract, providing additional evidence that SETD5 does not methylate a non-histone protein [16]. Similarly, in vitro assays using truncated mouse Mll5 containing the SET domain did not detect methylation activity on purified histones or recombinant nucleosomes [23], and similar results were obtained with a recombinant, truncated human MLL5, containing an N-terminal helix, the SET domain, and the post-SET domain [11]. In addition, preincubation of recombinant MLL5 with nuclear extracts was not able to stimulate activity on histone substrates [11], providing evidence that an additional cofactor or protein partner may not be sufficient to promote methyltransferase activity of MLL5. This truncated, human MLL5 was also subject to isothermal calorimetry (ITC) to test binding of the methyl donor SAM, which was not detectable [11]. While there are sequence alterations within the SET domain of MLL5, and related proteins, which suggested it would not bind SAM (Figure 1), nonetheless, this is the only member of the Set3 protein subfamily in which SAM binding has been directly tested and reported. Additional biochemical validation as to whether these noncanonical SET domains bind SAM may provide further insight into their potential functions.

Molecular genetic evaluation of SET domain function in Set3 subfamily proteins

The lack of biochemical activity for these noncanonical SET domains raises the question as to how, or whether, they contribute to protein function. In a number of organisms, targeted deletions of the SET domain have been evaluated for phenotypic consequences and impact on protein stability and protein-protein interactions. In some cases, the SET domain has been reported to be dispensable for some protein functions, while required for others. For example, in S. pombe, deletion of the Set3 SET domain did not affect the role of Set3 in heterochromatin silencing, however it was important for maintaining Set3 recruitment to some heterochromatic loci [48]. In mice, the SET domain of Setd5 is not required for its role in adipogenesis [51], while it is crucial for its function in cell proliferation in the developing retina [17].

There is also evidence that this noncanonical SET domain may contribute to protein-protein interactions in some contexts, however it is not required in all cases. In budding yeast, the SET domain is crucial for the integrity of Set3C, as no members of Set3C were detected via TAP purification of Set3 carrying a deletion of the SET domain [6]. However, deletion of the SET domain of human MLL5 did not affect its interactions with binding partners OGT or USP7 via co-immunoprecipitation from HEK293T cells [44]. Moreover, there was no disruption to protein interactions of SETD5 lacking the SET domain to HDAC3, G9a, and other interacting partners [16]. While further investigation regarding the potential function of the SET domain in direct protein-protein interactions is required, these findings suggest that there may be differential requirements for the SET domain in mediating some interactions.

It is plausible that the SET domain within Set3 subfamily proteins plays an important structural role to maintain the appropriate three-dimensional structure of the protein, as well as support critical protein-protein interactions. Indeed, there are examples in which deletion of the SET domain decreases protein abundance in cells, likely due to reduced protein stability. In budding yeast, endogenous deletion of the SET domain in Set3 has also been reported to reduce protein expression [6], and the SET domain of Set4 appears important for its stability [7]. Similarly, deletion of the SET domain in endogenous Setd5 in mouse embryonic stem cells led to decreased expression [24] and overexpression of SETD5 with an ASD-associated mutation (R308*) that truncates the SET domain leads to decreased SETD5 expression in HEK293T cells [24], though deletion of the SET domain in human SETD5 did not impact its expression [16]. These observations indicate that the SET domain likely contributes to protein stability and folding in Set3 subfamily proteins, and importantly, that assessment of mutants lacking the SET domain should include evaluation of protein abundance to ensure stable expression of the mutant protein. In some cases, this will therefore present challenges in evaluating the specific contribution of the SET domain to protein function, however further investigation of its potential functions using targeted point mutants, structural, and biochemical analysis will likely shed more insights on its specific contributions to the biological and biochemical functions of Set3-related proteins.

Future Implications and Concluding Remarks

The predominant known activity of the SET domain is to catalyze methylation of lysine residues on histones and other proteins [5]. As discussed here, there is a small, but highly conserved, subfamily within the SET domain superfamily that contain a noncanonical SET domain with sequence substitutions that preclude catalytic activity, and extensive biochemical investigation has not yielded substantial evidence of methyltransferase activity. Nonetheless, the Set3-related proteins have clear biological functions, such as being essential for proper development, and are directly linked to disease states, including inherited neurodevelopmental disorders. Given these roles, an improved understanding of the role of the conserved SET domain in these proteins has the potential to yield important new insights to physiological and pathological processes. In eukaryotes, up to 10% of proteins within enzymatic families are pseudoenzymes [52, 53], or proteins that share a domain that resembles a catalytically competent domain but lack activity. Pseudokinases have been well-documented and extensively studied; however, there is an emerging list of other enzymatic families that contain pseudoenzymes, including many associated with catalyzing post-translational modifications, such as pseudophosphatases and pseudodeubiquitinases, among others [53]. These pseudoenzymes can have a diversity of functions, including scaffolding the assembly of protein complexes and mediating protein-protein interactions, allosterically regulating canonical enzymes, acting as molecular switches in signaling events, regulating the accessibility of substrate for canonical enzymes, or other, unanticipated enzymatic activities [53]. In the Set3 SET domain subfamily, there are a number of examples of proteins which scaffold the assembly of larger complexes, including yeast Set3, fly UpSET, and human SETD5. Whether this is the predominant function of other subfamily members, or if these noncanonical SET domains have additional molecular or biochemical roles, remains to be determined. An increased understanding of the biological functions of these proteins and further biochemical, structural, and genetic investigation of the specific roles for the SET domain will yield new insights into the potential noncatalytic roles for these unique SET domain proteins.

Abbreviations:

SET

Su(var)3–9, Enhancer-of-zeste, and Trithorax

SAM

S-adenosylmethionine

KMT

lysine methyltransferase

PHD

plant homeodomain

HDAC

histone deacetylase