Alpha-centractin is a novel substrate of SETD3 methyltransferase in vitro

Materials

High-purity reagents (analytical grade whenever possible) were purchased from various suppliers: Sigma-Aldrich (USA), Roche Diagnostics (Switzerland), or Merck (Germany). Radiolabeled S-[methyl-³H]adenosyl-L-methionine ([³H]SAM) and scintillation cocktail Ultima Gold LSC were obtained from PerkinElmer (USA). The deuterated form, S-[methyl-²H]adenosyl-L-methionine ([²H]SAM), was purchased from C/D/N Isotopes Inc. (Canada). Synthetic peptides were procured from Caslo Laboratory (Denmark). HisTrap FF crude (pre-charged with Ni²⁺ ions) and PD-10 desalting columns were obtained from Cytiva (USA). Dowex 50W-X4 (200 mesh) ion-exchange resin was purchased from Sigma-Aldrich. Vivaspin-500 concentrator tubes were obtained from Sartorius Stedim (Germany). All other enzymes, DNA modifying enzymes, and the TurboFect transfection reagent were purchased from various suppliers: Thermo Fisher Scientific (formerly Thermo-Fermentas, USA), A&A Biotechnology (Poland), or BioShop (Canada).

SETD3 methyltransferase activity assay

Protein purification

Generation of SETD3 knockout cell lines using CRISPR/Cas9

Isolation and genotyping of SETD3 knockout clones

Single puromycin-resistant colonies were isolated and expanded. After trypsinization, cell suspensions were seeded in 96-well plates. Once confluent, cells were passaged and used for either genotypic analysis or frozen for future use.

Genomic DNA was isolated from puromycin-resistant clones using the Extractme Genomic DNA kit (Blirt). The targeted regions were amplified by PCR, and the products were sequenced to identify mutations. Three clones from each cell line with frameshift mutations leading to premature stop codons were selected for further studies. Mutation analysis was performed using the DSDecode tool (Liu et al., 2015) for automated decoding of superimposed chromatograms and verified using the Indigo tool from the GEAR website (Rausch et al., 2020) (https://www.gear-genomics.com).

Fluorographic detection of SETD3 activity

TurboID-based proximity labeling of proteins

Protein labeling for gel filtration chromatography

Identification of proteins by tandem mass spectrometry

Protein bands visualized on fluorograms (10% polyacrylamide gel) were excised and subjected to in-gel digestion with Trypsin Gold (Promega, Germany) according to the method described by Shevchenko et al. (2006).

For protein identification from fractions obtained in TurboID pulldown experiments and gel filtration of radiolabeled cell lysates (Superdex 200), in-solution digestion was performed. Briefly, 17–34 µg of protein from fraction 15 or streptavidin magnetic beads were reduced with 5 mM DTT in a reaction mixture containing 100 mM ammonium bicarbonate (NH₄HCO₃) and 0.1% RapiGest (Waters, USA) at 50 °C for 30 min. The reaction mixture was then alkylated with 9 mM iodoacetamide for 40 min at room temperature. Subsequently, 2 µg of Trypsin Gold was added, and the digestion proceeded overnight at 30 °C with continuous shaking for on-bead digestion.

Peptides resulting from tryptic digestion were analyzed by nano-UPLC tandem mass spectrometry using an Acquity nano-UPLC system coupled with a Synapt G2 HDMS Q-TOF mass spectrometer (Waters, Milford, USA) equipped with a nanospray source. The instrument was operated in MS^E; mode under default parameters. Briefly, 1−2.5 µg of digested peptides were loaded onto a Waters Symmetry C18 trapping column (20 mm × 180 µm) connected to a Waters BEH130 C18 UPLC column (250 mm × 75 µm). Peptides were eluted with a gradient of 1–85% acetonitrile in water (both containing 0.1% formic acid) at a flow rate of 0.3 µL/min. Eluted peptides were directly introduced into the mass spectrometer.

Data acquisition and analysis were performed using MassLynx 4.1 software (Waters, USA) and ProteinLynx Global Server 2.4 software (PLGS, Waters, USA) with a False Discovery Rate (FDR) set to ≤ 4%. Protein identification was achieved by searching a database containing the complete human (Homo sapiens) reference proteome downloaded from the NCBI Protein database, which was then randomized for the MS/MS software.

Detection of in vitro methylated α-centractin by tandem mass spectrometry

To obtain a sufficient quantity of methylated α-centractin, formed through the reaction catalyzed by recombinant human SETD3 protein, for subsequent mass spectrometry analysis, the reaction mixture was scaled up. In brief, 20 µg of recombinant human α-centractin was incubated for 2 h at 37 °C in 30 µL of a reaction mixture. This mixture contained 25 mM Tris–HCl, pH 8.0, 10 mM KCl, one mM DTT, 20 µM MnCl₂, 30 µM [²H]SAM, 1.5 µg SAH nucleosidase, and either 1.5 µg SETD3 or 1.5 µg R253A protein. The incubation was terminated by snap-freezing in liquid nitrogen, and the reaction mixtures were stored at −70 °C.

For MS/MS analysis of deuterated protein, 19 µl of the flash-frozen sample (12.6 µg protein) was trypsin-digested ‘in solution’ for 16 h at 30 °C in 30 µl of the reaction mixture containing 100 mM NH₄HCO₃, 0.05% RapiGest (Waters), 1.7 mM TCEP, and 2 µg trypsin (MS grade, Promega). The digestion reaction was stopped by adding trifluoroacetic acid to a final concentration of 1%. The resulting peptides were then analyzed by nano-UPLC-tandem mass spectrometry, following the procedure detailed in the “Identification of proteins by tandem mass spectrometry” section. Detection of the trideuterium-methylated peptides was carried out in fully automated mode using PLGS 2.4 software (Waters). This software had been updated to include trideuterium-methylation (+17.03448 Da) of Cys, Asp, Asn, His, Lys, Arg, Glu, and Gln residues as a potential peptide modification.

Analytical methods

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed using 10% polyacrylamide gels and a Tris-glycine running buffer. Following electrophoresis, the gels were stained with Coomassie Brilliant Blue Protein Staining Solution (Thermo) to visualize proteins. Protein concentration was determined spectrophotometrically using the Bradford assay (1976). Bovine serum albumin (BSA) served as a standard for the assay.

SETD3 methylates targets other than β-actin

Previous autoradiography experiments using total cytoplasmic extracts from HT1080 cells expressing endogenous SETD3 revealed β-actin as the sole methylated protein (Wilkinson et al., 2019). However, this approach might have only detected highly abundant substrates.

To identify potential minor substrates for SETD3, we performed in vitro methylation assays using cell-free lysates from human HAP1 cells lacking SETD3 (SETD3 KO). Lysates were first preincubated with [methyl-¹H]SAM to allow endogenous methyltransferases to modify their substrates. Subsequently, [methyl-³H]SAM was added along with recombinant SETD3 or BSA (control), and protein radiolabeling was performed. Fluorography of proteins resolved by SDS-PAGE revealed several radiolabeled bands with molecular weights of approximately 240, 160, 120, 43, 34, and 29 kDa, present predominantly in lysates containing recombinant SETD3 (Fig. 2). The prominent 43 kDa band likely corresponded to β-actin, while the remaining bands potentially represented novel, less abundant SETD3 substrates. To further substantiate these findings and exclude potential contamination of the radiolabeled bands by proteins from the SETD3 preparation, we conducted a modified fluorographic experiment. In this iteration, BSA was substituted with R253A SETD3, a catalytically inactive mutant of SETD3. Concurrently, radiolabeling of cell-free lysates from wild-type HAP1 cells with SETD3 was performed to compare the degree of protein methylation between SETD3 wild-type and knockout cells. Subsequent fluorography of SDS-PAGE-resolved proteins confirmed the presence of several radiolabeled bands specifically methylated by SETD3, albeit with signals that were more diffuse and less pronounced compared to the initial experiments (Fig. S4). In conclusion, these results suggested the presence of additional SETD3 substrates beyond β-actin in human cells.

To identify proteins modified by SETD3, deuterated SAM ([methyl-²H]) was used instead of tritiated SAM as the methyl donor to distinguish SETD3-catalyzed methylation from other sources. In vitro labelled bands were cut out of the SDS-PAGE gel and analyzed by tandem mass spectrometry (Q-TOF). This analysis identified numerous proteins listed in File S1. Notably, it included condensin complex subunit 1 (NP_055680.3, Band 2), isoleucine tRNA ligase (NP_001361230.1, Band 2), and actin (NP_001186883.1, Band 4), all with specific histidine residues methylated.

However, mass spectrometry analysis only successfully identified labeled proteins in two of the six analyzed bands. This was likely due to the limited amount of protein present in the excised bands. To overcome this limitation and improve protein yield for mass spectrometry, we fractionated cell-free lysates labeled with both [methyl-²H] and [methyl-³H]SAM on a Superdex-200 gel filtration column. This aimed to partially purify SETD3 substrates before mass spectrometry analysis. Surprisingly, nearly all radiolabeled proteins eluted in the void volume of the column. This suggests that these proteins might exist within the lysates as protein complexes exceeding 600 kDa in molecular mass (Figs. 3A and 3B). The labeled fractions were then resolved by SDS-PAGE, and the methylated polypeptides were visualized using fluorography. Notably, the 43 kDa band corresponding to actin displayed the strongest signal, indicating it as the most prominently methylated protein (Fig. 3C). The remaining SETD3 substrates were only clearly detectable in the most radioactive fraction (fraction 15) after a 40-fold concentration step (Fig. 3D). Mass spectrometry analysis of fraction 15 revealed the presence of numerous proteins, including SETD3, actin, and several deuterated polypeptides (File S2). Interestingly, the presence of SETD3, with a native molecular weight of approximately 130 kDa, in fraction 15 suggests its potential involvement in the high-molecular-weight protein complexes eluting in the void volume of the Superdex 200 column.

SETD3 interacts with several proteins in human cells

Protein methyltransferase activity and substrate preference can be influenced by interactions with non-substrate partner proteins. Some methyltransferases exhibit context-dependent substrate methylation based on their interacting partners (Herz, Garruss & Shilatifard, 2013). Therefore, we investigated potential in vivo interactions between SETD3 and other proteins using a TurboID proximity labeling method (Branon et al., 2018). This technique employs a TurboID biotin ligase that rapidly conjugates biotin to nearby proteins. Subsequently, these biotinylated proteins can be isolated and identified by mass spectrometry (MS/MS).

We constructed two plasmids encoding SETD3 fused to the TurboID ligase at either the N or C-terminus. HEK293T cells lacking endogenous SETD3 (SETD3 KO) were transfected with both constructs for 48 h. Following a 30-minute TurboID labeling step, cells were lysed, and biotinylated proteins were enriched using streptavidin beads. Half of the beads were used for protein elution with biotin, and the eluted proteins were analyzed by Coomassie blue staining (Fig. 4A) and western blotting with an anti-biotin antibody (Fig. 4B). The remaining beads underwent tryptic digestion to generate peptides for LC-MS/MS analysis for protein identification.

As shown in Fig. 4B, minimal endogenous biotinylated proteins were isolated from control cells transfected with an empty plasmid (File S3). Biotin addition to the culture medium resulted in non-specific overlabeling of endogenous proteins, potentially overloading the streptavidin beads (Fig. 4B). Notably, efficient protein enrichment occurred even without exogenous biotin, facilitating the identification of biotinylated proteins in that cell lysate. In total, LC-MS/MS identified 23 unique human proteins in the streptavidin-purified samples (Table 2, File S3), including SETD3 and actin. Interestingly, compared to cells incubated with biotin, a more diverse set of proteins were identified from the sample using biotin-free cell lysates. This difference likely arose from: (i) excessive protein labeling with added biotin, (ii) overloading of beads with a few highly labeled proteins, and (iii) insufficient protein digestion due to excessive biotinylation of lysine residues (Li et al., 2021).

Similar experiments were conducted in HeLa and HAP1 cells lacking SETD3. However, protein biotinylation was significantly lower in these cell lines compared to HEK293T cells, possibly due to reduced expression of the recombinant SETD3-TurboID fusion proteins.

Several proteins may be the substrates of SETD3

The SET domain, a common feature of protein methyltransferases that modify lysine residues, often exhibits high substrate specificity due to recognition of flanking sequences surrounding the targeted lysine (Dai et al., 2019). Similarly, SETD3 appears to exhibit this characteristic.

Previous studies demonstrated that the β-actin peptide (residues 66–88) fits into a specific groove on SETD3, with residues Ile71, His73, and Trp79 occupying three hydrophobic pockets (Guo et al., 2019; Dai et al., 2019). Substitution of Ile71 and Trp79 in the β-actin peptide with shorter or charged residues significantly reduced its suitability as a substrate (Bilgin et al., 2022; Al-Fakhar et al., 2023). These findings suggest that the mechanisms governing β-actin peptide binding likely apply to other SETD3 substrates as well.

Therefore, to identify the most promising potential substrates from the proteins identified in fluorography and TurboID experiments, we employed the following criteria: (i) the candidate protein should be [²H]-methylated at a histidine residue, (ii) the candidate protein’s molecular weight should correspond to one of the bands visualized in the fluorography experiment, and (iii) the sequence flanking the methylated histidine should contain relatively hydrophobic residues at positions −2 and +5 relative to the histidine, analogous to Ile71 and Trp79 in the β-actin peptide. Based on these criteria, fourteen different proteins, including β-actin, were selected as potential SETD3 substrates, along with their sequences containing the putative methylation site (Table 3). Notably, some of these potential substrates, such as actin and fatty acid synthase, were identified in both fluorography and TurboID experiments.

Alpha-centractin is a new substrate of SETD3

To verify the activity of human SETD3 towards its potential substrates in vitro, peptides were synthesized and tested as acceptors of [³H]methyl from [³H]SAM in a radiochemical assay. As shown in Fig. 5, SETD3 catalyzed the methylation of β-actin and α-centractin peptides. Interestingly, the enzyme also displayed activity towards a peptide derived from the isoleucine tRNA ligase, albeit at a significantly lower level.

To further confirm the specific methylation of α-centractin, a radiochemical assay was performed using recombinant human α-centractin as the substrate, alongside β-actin as a positive control. The α-centractin protein was overexpressed and purified from Escherichia coli using the same established protocol previously employed for β-actin production. As shown in Fig. 6, wild-type SETD3 methylated recombinant α-centractin, demonstrating its suitability as a substrate in vitro. Notably, methylation was absent in the presence of R253A SETD3, a catalytically inactive mutant, confirming the enzyme’s specific activity. However, the methylation efficiency of wild-type SETD3 towards α-centractin was considerably lower compared to β-actin, suggesting that the latter serves as a significantly better substrate for the enzyme. This can be attributed to the kinetic properties of SETD3 in the presence of α-centractin (Table 4). The enzyme displayed a considerably lower affinity for α-centractin (3.827 µM) compared to β-actin (0.752 µM) (Kwiatkowski et al., 2018). Furthermore, its k_cat with α-centractin and a saturating concentration of SAM (0.31 s⁻¹) was roughly half that estimated for β-actin (0.65 s⁻¹). The enzyme’s affinity for SAM was also somewhat reduced with α-centractin (0.242 µM) compared to β-actin (0.116 µM) (Kwiatkowski et al., 2018). However, it is important to note that we conducted these kinetic studies using only sub-saturating concentrations of α-centractin, as the substrate tended to precipitate at higher concentrations (Fig. S5).

To confirm the specific methylation of H77 in recombinant human α-centractin, we incubated the protein with either homogeneous recombinant human SETD3 or R253A mutant enzymes. This incubation was performed in the presence of [²H] SAM. Following the reaction, the proteins in the assay mixture were tryptic-digested. The resulting α-centractin peptides were then analyzed using nano-UPLC-Q-TOF. Each MS analysis achieved at least 37% sequence coverage of α-centractin. The peptide YPMEHGIVK (M + H = 1,090.5804 Da), which contains the H77 residue, was the only trideuterium-methylated peptide detected by the PLGS 2.4 software (File S4). Manual inspection of the MS/MS spectra of both the parent ion and its fragments definitively confirmed the presence of trideuterium-methylated H77 (Fig. 7). No trideuterium-methylation of H77 was detected in the YPMEHGIVK peptide (M + H = 1,073.5479 Da) derived from α-centractin protein that had been incubated with the R253A mutant protein (see Fig. 7).

While β-actin is a known substrate for SETD3 in human cells, there may be others

Previous research, including work from our laboratory, identified SETD3 as an enzyme that specifically methylates H73 residue on β-actin (Kwiatkowski et al., 2018; Wilkinson et al., 2019). Currently, β-actin remains the only confirmed substrate for this methyltransferase. However, SETD3 has been implicated in various cellular processes apparently unrelated to actin regulation, such as myocyte differentiation, cell cycle control, and the response to hypoxic conditions (Eom et al., 2011; Kim et al., 2011; Cohn et al., 2016). Additionally, studies have shown SETD3 localization within the nucleus of mammalian cells (Eom et al., 2011; Cohn et al., 2016), hinting at the potential for different substrates in various cellular compartments. This raises a key question: is β-actin the sole substrate for SETD3 methyltransferase activity?

Here, we show that SETD3 methylates several polypeptides in human HAP1 cells that are distinct from β-actin. These novel substrates appear to be present in minute quantities. This is evidenced by their very weak labeling and the faint protein bands observed during SDS-PAGE analysis in the gel regions where they were visualized (see Fig. 2 and Fig. S4). Furthermore, it is highly unlikely that these minor SETD3 substrates are, in fact, major substrates for other histidine-specific methyltransferases. Unlike lysine- or arginine-specific methyltransferases, no cross-reactivity has been observed among histidine-specific methyltransferases to date. Every known enzyme in this class—including SETD3, METTL9, METTL18, and C9orf41—catalyzes the methylation of its own specific substrates. Due to the low abundance of these novel substrates compared to β-actin, their detection was only possible by using SETD3-deficient cells as a substrate source for in vitro labeling experiments dependent on SETD3 activity. This might explain why these alternative substrates have not been observed previously in experiments using lysates from cells expressing endogenous SETD3 (Wilkinson et al., 2019).

Mass spectrometry analysis identified numerous proteins within the labeled bands, but only a few exhibited histidine residue methylation. In vitro methylation experiments using synthetic peptides derived from candidate proteins (see Fig. 5) revealed detectable methylation of the Isoleucyl-tRNA synthetase 1 (IARS1) peptide. This suggests that IARS1 might be a SETD3 substrate present in the labeled bands. However, attempts to produce the recombinant IARS1 protein were unsuccessful, which prevented us from further testing it as a substrate for SETD3 in vitro. Also noteworthy is that the absence of SETD3-dependent methylation of the remaining tested peptides does not definitively exclude the possibility that their parent proteins are still substrates of the methyltransferase. Synthetic peptides have already been shown to be poor substrates for some histidine methyltransferase which may efficiently methylate full-length proteins in vitro or in vivo (Webb et al., 2010; Davydova et al., 2021; Małecki et al., 2021). This likely occurs because some methyltransferases require complex interactions with the entire protein substrate for proper recognition and efficient methylation.

ACTR1A is a new substrate for SETD3 in vitro

We found that all SETD3 substrates visualized in fluorography were involved in the formation of protein complexes that also included the SETD3 enzyme (see Figs. 3B–3D and File S2). This aligns with previous studies demonstrating that: (i) endogenous SETD3 in skeletal muscle is tightly bound to myofibrils, forming a relatively stable complex with myofibrillary proteins (Vijayasarathy & Rao, 1987; Kwiatkowski et al., 2018), and (ii) SETD3 interacts with a large number (around 170) of different intracellular proteins in mammalian cells (Cohn et al., 2016). The exact mechanisms and reasons behind the formation of these complexes remain unclear. However, it is tempting to speculate that these interactions might play a role in how SETD3 recognizes substrates and performs its enzymatic activity.

Our work has established that α-centractin (ACTR1A) physically interacts with SETD3 and serves as a substrate for its methyltransferase activity. This conclusion is supported by two findings: (i) proximity labeling experiments using HEK293T cell lysates identified ACTR1A as a protein interacting with SETD3 (see Fig. 4) and (ii) both human recombinant α-centractin and a synthetic peptide derived from this protein were methylated in enzymatic assays (Figs. 5–7). The observation that α-centractin is a SETD3 substrate is unsurprising, considering its close resemblance to β-actin in both amino acid sequence and spatial structure (Fig. 8). However, more importantly, our data demonstrates that SETD3 methylates β-actin significantly more efficiently than ACTR1A. This aligns with the current understanding of SETD3 substrate specificity, which likely relies on precise recognition of the amino acid sequence surrounding the methylation site (Guo et al., 2019; Dai et al., 2019).

Potential physiological role of ACTR1A methylation

α-Centractin (or actin-related protein Arp 1) is a crucial subunit of dynactin. Dynactin is a key component of the cytoplasmic dynein motor complex, which plays a central role in transporting various cellular cargo along microtubules in eukaryotic cells (Urnavicius et al., 2015; Reck-Peterson et al., 2018). Notably, among all actin-related proteins in eukaryotes, α-centractin exhibits the highest similarity to conventional β-actin. It shares the ability to bind and hydrolyze ATP, and can spontaneously polymerize into short filaments resembling, but not identical to, actin filaments (Bingham & Schroer, 1999). Importantly, an α-centractin filament composed of eight monomers can bind a single β-actin monomer, forming the central core of the 23-subunit (11 distinct proteins) dynactin complex.

Previous research has shown that methylation of a specific histidine residue (H73) in β-actin promotes its polymerization. We hypothesize that methylation of the corresponding histidine residue (H77) in α-centractin by SETD3 might similarly stimulate filament formation, thereby facilitating the assembly of the dynactin complex. Consequently, a loss of SETD3 activity could potentially impact not only actin polymerization but also dynein-dependent processes like intracellular transport.