Modification of small ubiquitin‐related modifier 2 (SUMO2) by phosphoubiquitin in HEK293T cells

Additional complexity in the post‐translational modification of proteins by ubiquitin is achieved by ubiquitin phosphorylation, for example within PINK1‐parkin mediated mitophagy. We performed a preliminary proteomic analysis to identify proteins differentially modified by ubiquitin in HEK293T, compared to phosphomimetic ubiquitin (Ser65Asp), and identified small ubiquitin‐related modifier 2 (SUMO2) as a candidate. By transfecting SUMO2 and its C‐terminal–GG deletion mutant, along with phosphomimetic ubiquitin, we confirm that ubiquitin modifies SUMO2, rather than vice versa. Further investigations revealed that transfected SUMO2 can also be conjugated by endogenous phospho‐Ser65‐(poly)ubiquitin in HEK293T cells, pointing to a previously unappreciated level of complexity in SUMO2 modification, and that unanchored (substrate‐free) polyubiquitin chains may also be subject to phosphorylation.

More recently, the demonstration that ubiquitin itself can be modified through phosphorylation by kinases including PINK1 provided a major breakthrough linking two fundamental signalling pathways in cells; phosphorylation and ubiquitylation [13][14][15]. Ubiquitin can be phosphorylated on Ser, Thr and Tyr residues, which may further modulate its regulatory functions; however, the full physiological and indeed pathological significance of these additional modifications is unclear. In addition to activation of parkin, the consequences of Ser65 phosphorylation of ubiquitin on the structure, polyubiquitin chain assembly, recognition, hydrolysis and downstream mitochondrial homeostasis have been reported [16, 13-15, 17, 18]. Studies have suggested that phospho-Ser65-ubiquitin is the parkin receptor on damaged mitochondria [19], but the pathophysiological relevance of this post-translational modification in the PINK1-parkin mitophagy pathway in neurons and brain tissue is not fully understood. While phosphorylation of the modifier protein, ubiquitin, further increases complexity, it also provides more selectivity and specificity for an apparently universal ubiquitylation process [20,21]. Although the function of Lys63-linked phospho-Ser65-(poly)ubiquitin in parkin activation and directing damaged mitochondria to the autophagy pathway has been established, the broader molecular consequences of ubiquitin phosphorylation, including targets of phosphoubiquitin, have not been fully explored [20,14,22,23,18,24].
Further, comparatively little is known about the regulation of unanchored polyubiquitin chains. In the same way that conjugated polyubiquitin chains can be considered to represent an 'ubiquitin code' , distinct sub-populations of unanchored polyubiquitin chains are likely to underlie different biological activities [25]. An unanchored polyubiquitin chain is a free ubiquitin chain that is not conjugated to a substrate protein. Knowledge of the range of fundamental processes that may be regulated by unanchored polyubiquitin and unanchored phosphorylated-polyubiquitin chains, as well as the precise molecular composition of these ubiquitin polymers (i.e., presence of different isopeptide linkages, etc.) is also lacking. Whether, like conjugated polyubiquitin, such substrate-free ubiquitin assemblies are also regulated by modifications such as phosphorylation, is unclear, but previously established robust purification protocols mean that such questions can now be addressed [26].

Significance Statement
Eukaryotes have an extensive repertoire of PTMs such as phosphorylation and ubiquitylation. Phosphorylation of Serine residues of ubiquitin modifies its structure, polyubiquitin chain assembly and deubiquitylase-dependent chain hydrolysis, providing an additional layer of regulation in ubiquitin signal functions. Site-specific ubiquitin phosphorylation generating pSer65-ubiquitin in the PINK1-parkin pathway is relevant in diseases associated with mitochondrial dysfunction including Parkinson's disease, cancer, cardiovascular and metabolic disorders.
In the same way that conjugated polyubiquitin chains can be considered to represent an 'ubiquitin code' , distinct subpopulations of unanchored (substrate-free) polyubiquitin chains are likely to exert different biological activities. Knowledge of the range of fundamental processes that may be regulated by unanchored polyubiquitin as well as the precise molecular composition of these ubiquitin polymers is also lacking.
This project used proteomic approaches to identify and catalogue mammalian target proteins modified by covalent attachment of pSer65-(poly) ubiquitin from cultured human embryonic kidney cells and identified SUMO2 as a candidate, and indicated that unanchored polyubiquitin chains may also be subject to phosphorylation.
We hypothesised that there are many other protein targets beyond mitochondrial outer membrane proteins [18] that are covalently modified by phosphorylated ubiquitin. This investigation therefore attempted to identify novel mammalian proteins that are modified by covalent modification with phospho-Ser65-(poly)ubiquitin in HEK293T cells.

IMAC purification of ubiquitin-modified proteins
Cells from each transfection were scraped in 6 mL pre-chilled PBS after which 10% of each cell suspension was pelleted at 3000 × g for 5 min at 4 • C. After discarding the PBS by carefully pipetting, each pellet was re-suspended in 50 μL of 1 × Laemmli sample loading buffer (60 mM Tris-Cl pH 6.8, 2% SDS, 10% glycerol, 5% β-mercaptoethanol, 0.01%

Identification of proteins modified by covalent attachment of phosphoubiquitin by LC-MS/MS
The shotgun proteomic approach was employed to identify proteins in IMAC-purified mixtures as detailed in literature [29][30][31]. Protein on

2.4
Confirmation of the covalent modification of transfected SUMO2 by ubiquitin in vivo SUMO2 and SUMO2∆GG (a C-terminal-GG deleted SUMO2 variant incapable of substrate conjugation but able to accept posttranslational modifications) cDNAs engineered in pEGFP-C1 vector to generate GFP-tagged SUMO2-GG and GFP-tagged SUMO2∆GG constructs were co-transfected with equal amounts of HF-UbWT construct created earlier into HEK293T cells as follows; GFP-SUMO2-GG and HF-UbWT, and GFP-SUMO2∆GG and HF-UbWT. Following harvesting, cells were lysed in DLB and proteins were purified by IMAC as described previously. After washes, 25 μL of 1 × Laemmli loading buffer containing 200 mM imidazole was added to elute bound proteins.

Modification of transfected SUMO2 by endogenous phospho-Ser65-ubiquitin
HEK293T cells seeded at 1.5 × 10 6 cells per 10 cm dish were transfected at ∼80% confluence with SUMO2 cloned into pcDNA3.1Zeo+ to express a His-FLAG-tagged SUMO2 (HF-SUMO2) by the PEI protocol outlined above. 40 h after incubation in fresh complete culture media, the media was replaced with complete culture media containing 10 μM CCCP, an inhibitor of oxidative phosphorylation and returned to the tissue culture incubator (5% CO 2 , 100% humidity) for 8 h prior to harvesting in PBS. CCCP is known to increase total or conjugated phospho-Ser65-(poly)ubiquitin and has been used to demonstrate the subsequent recovery of cells from oxidative stress and in mitophagy [16,13]. The cells were then lysed in DLB and proteins bound to HF-SUMO2 were purified by IMAC as previously described. Bound pro-

2.6
Detection of Ser65 phosphorylation in purified unanchored (poly)ubiquitin chains in HEK293T cells

IMAC-purification of ubiquitin-modified proteins from HEK293T cells
We performed a pilot screen of cellular targets of (phospho-

Identification of ubiquitin-modified proteins from HEK293T cells by mass spectrometry
Denaturing IMAC-purified samples were subjected to LC-MS/MS analysis in order to catalogue endogenous proteins modified by transfected ubiquitin. The initial analysis identified a total of 32 proteins in purified fractions from transfections with HF-UbWT, HF-UbS65D, or both (but not HF-EV controls), from a minimum of two peptide sequences, including 19 associated with HF-UbS65D only (Table   S1). Data presented on Table 1 represent data from which controls F I G U R E 1 Immunoblot of purified protein targets of ubiquitylation by recombinant wild-type and Ser65Asp-ubiquitin in HEK293T cells. His FLAG-tagged UbWT and UbS65D mutant proteins were expressed in HEK293T cells. Following cell lysis in DLB buffer and sonication at 3 x 30 s on ice, cell fragments were pelleted at 13,000 g for 10 min. Proteins bound to UbWT and UbS65D were purified by IMAC, resolved by SDS-PAGE and immunoblotted against VU-1 ubiquitin (left panel) or anti-His-tag antibodies (right panel). Blots were stripped and reprobed with anti-β-actin antibodies (lower panels) (proteins identified in transfects with HF-EV controls) had been manually excluded. One of these HF-UbS65D-modified proteins was SUMO2 ( Figure 2). A subsequent more preparative scaled-up repeat using four times the amount (by protein) of transfected cell lysates following the same protocol described above, identified a total of 316 different proteins in corresponding IMAC-purified samples were detected, again including SUMO2 (Table S1) (Table 1 and Table S1). Across both experiments, a total of 22 out of 93 residues of the endogenous SUMO2 sequence were detected in the purified samples from HF-UbS65D-transfected cells, representing sequence coverage of 24% (Figure 2). These analyses therefore suggested (preferential) modification of SUMO2 by transfected phosphomimetic HF-UbS65D compared to HF-UbWT in HEK293T cells. SUMO2 has not previously been described as target of phospho-Ser65-ubiquitin, although mixed/hybrid chains with ubiquitin have been reported [32,33], and ubiquitylation prediction software programmes (i.e., UbiSite, RUBI and MuBiSiDa) analyses conducted in this project indicate the presence of ubiquitylation sites. SUMO2 was therefore considered a potentially novel covalent target of phospho-Ser65-ubiquitin in HEK293T cells. As SUMO2 itself is an ubiquitin-like modifier, we first determined whether it was indeed undergoing modification with ubiquitin, or itself was the modifier of ubiquitin, in subsequent experiments. To this end, we co-transfected HEK293T cells with HF-UbWT and GFP-SUMO2 with or without its critical Cterminal diglycine residues that are required to mediate conjugation to targets, whilst not affecting the ability to 'accept' conjugation (i.e., GFP-SUMO2∆GG). After denaturing IMAC-purifications, anti-ubiquitin immunoblotting indicated that deletion of the diglycine residues of transfected SUMO2 had no effect on its ability to be modified by HF-UbWT, with high molecular weight ubiquitin conjugates present in purified fractions (Figure 3). GFP-tagged protein reactive bands TA B L E 1 LC-MS/MS analyses of HEK293T proteins captured on denaturing IMAC after expression of HF-EV control, UbWT and UbS65D

Confirmation of SUMO2 as a covalent target of endogenous phospho-Ser65-ubiquitin
In order to induce the formation of endogenous phospho-Ser65ubiquitin since steady-state levels in cells are relatively low [34], Nevertheless, replicating these experiments in different cell lines using HEK293T as vehicle will enable conclusive evidence to be drawn.

F I G U R E 5
Detection of Ser65 phosphorylation in endogenous unanchored (poly)ubiquitin in HEK293T cells. 6.0 mg of HEK293T protein lysate in NETN buffer previously treated for 8 h with 10 μM CCCP (+CCCP) or diluent DMSO (-CCCP) were purified by ZnF-UBP domain affinity chromatography using 100 μL of 10 mg/mL ZnF-UBP protein on beads. Captured proteins were resolved by SDS-PAGE and immunoblotted with VU-1 ubiquitin or phospho-Ser65-ubiquitin antibodies. Commercial phospho-Ser65-ubiquitin was used as a positive control of Ser65 phosphorylation of unanchored (poly)ubiquitin chains. Traces of weak phospho-Ser65-ubiquitin reactivity is evident in purified fractions of unanchored polyubiquitin, consistent with the presence of endogenous phospho-Ser65-ubiquitin

3.5
Detection of phospho-Ser65-ubiquitin in affinity purified unanchored (poly)ubiquitin following CCCP treatment of HEK293T cells Unanchored polyubiquitin chains are free ubiquitin chains that are not conjugated to a substrate protein. Given our observations that 'hybrid' chains consisting of phospho-Ser65-ubiquitin and SUMO2 may be physiologically relevant, we finally investigated if unanchored polyubiquitin chains, that is, covalent assemblies of multiple ubiquitins in a substrate-free form that regulate a range of different biological pathways, also represent overlooked substrates of modification by phosphorylation [35-39, 6, 40]. Such chains (as well as monoubiquitin) can be affinity purified using an ubiquitin-binding domain (UBD) with high affinity exclusively for the free C-terminus of ubiquitin (but not SUMO2), the ZnF-UBP domain denoted as FUBE: that is, free ubiquitin-binding entity [25]. Heating of cell lysates prior to purification is a critical step in order to resolve unanchored polyubiquitin chains from heat-sensitive ubiquitin-protein conjugates ( Figure 5, right panel) [41]. Unanchored polyubiquitin chains were successfully purified by ZnF-UBP-Sepharose affinity chromatography [26] both from cells with (+CCCP) and without CCCP (+DMSO) treatment, and not with control ZnF-UBP-Sepharose (no protein)

Concluding remarks
His FLAG-tagged human recombinant UbWT and UbS65D mutant were created and employed to purify protein targets modified by covalent attachment of (poly)ubiquitin or phospho-Ser65-(poly)ubiquitin in HEK293T cells by the denaturing IMAC protocol. Proteomic analysis suggested differential modification of various potentially novel target proteins by ubiquitin compared to UbS65D including endogenous SUMO2, with S65D phosphomimetic mutant modifying proteins more efficiently. By transfecting GFP-SUMO2 and its C-terminal-GG deletion mutant, along with ubiquitin, we confirm that ubiquitin modifies SUMO2 rather than vice versa. Further, we found that transfected HF-SUMO2 is modified by endogenous phospho-Ser65-ubiquitin in