A synthetic biology approach for identifying de‐SUMOylation enzymes of substrates

A synthetic biology approach using a robust reconstitution system in Escherichia coli enables the identification of plant ubiquitin-like proteases responsible for removing the small ubiquitin-like modifier (SUMO) post-translational modifications from specific protein substrates. SUMOylation is a reversible posttranslational modification that plays a crucial role in various essential biological processes within eukaryotic cells. This reaction involves the transfer of small ubiquitin-like modifier (SUMO) molecules to lysine residues on target proteins, thereby regulating the stability, localization, and activity of these substrates. Despite the low amino acid sequence similarity between SUMO and ubiquitin, they share similar three-dimensional structures and are conjugated onto protein substrates through a comparable three-step catalytic reaction. This includes the activation of SUMO molecules by a SUMO-activating enzyme (E1), followed by the conjugation of SUMO to the cysteine residue of a SUMO-conjugating enzyme (E2), and ultimately culminating in the covalent attachment of lysine residues on substrates mediated by a SUMO ligase (E3) (Park and Yun, 2013). In the SUMOylation cycle, ubiquitin-like proteases (ULPs) play multiple critical functions. First, ULPs are responsible for removing C-terminal fragments from SUMO precursors in order to generate mature SUMO molecules for conjugation. Second, ULPs mediate the de-SUMOylation process by removing SUMO from protein substrates. Interestingly, de-SUMOylation is regulated by ULPs through their isopeptidase activity, which differs from the catalytic mechanism involved in SUMO maturation. As SUMOylation is a conserved strategic response to stresses that allows eukaryotic cells to quickly adapt to environmental changes, de-SUMOylation is crucial for maintaining the appropriate level of SUMO conjugations during transitions between different conditions (Morrell and Sadanandom, 2019). While thousands of SUMOylation substrates have been identified using proteomic approaches, only a small number of enzymes actually plays roles in this process; for example, an E2 (SCE1) and two E3 (SIZ1 and MMS21) have been characterized in Arabidopsis (Augustine and Vierstra, 2018). Therefore, precise control of SUMOylation levels on different substrates may be attributed to de-SUMOylation mediated by ULPs. Multiple members of the ULPs have been well characterized in humans and Arabidopsis, supporting the potential substrate specificity of these de-SUMOylating enzymes (Yates et al., 2016). Thus it is important to identify specific ULPs for substrates in order to understand their dynamic regulation mechanisms. SUMOylation can be detected in vivo by immunoprecipitation using an anti-SUMO antibody or an exogenously expressed tagged-SUMO isoform. However, due to the dynamic changes in SUMOylation statuses within eukaryotic cells, it is challenging to accurately assess the impact of ULPs on the modification level of specific substrates based solely on immunoprecipitation. Furthermore, different ULPs may have redundant functions in de-SUMOylating the same substrates, adding to the challenge of characterizing specific SUMOylation substrates and their corresponding ULPs. To address the existing technological challenges, we have developed a robust system for identifying pairs of protein substrates and ULPs based on a synthetic biology approach, using the well studied Arabidopsis ULPs (Figure 1A). In a previous reconstitution system in Escherichia coli cells expressing the AtSAE1a/AtSAE2 (E1) complex and AtSUMO1GG (a mature form of AtSUMO1 with a C terminus di-glycine motif), with or without SCE1 (E2), SUMOylation of potential substrates was successfully determined (Okada et al., 2009). Due to the absence of SUMOylation and de-SUMOylation enzymes in bacteria, the attachment of SUMO on targeting proteins is stable and easily detectable. Therefore, we introduced ULPs into this system to assess their impact on the levels of substrate SUMOylation. In Arabidopsis, eight functional ULPs have been extensively characterized in previous studies, including ESD4 (EARLY IN SHORT DAYS 4), FUG1 (FOURTH ULP GENE CLASS 1), ELS1/2 (ESD4-LIKE SUMO PROTEASE 1/2), OTS1/2 (OVERLY TOLERANT TO SALT 1/2), and SPF1/2 (SUMO PROTEASE RELATED TO FERTILITY 1/2) (Castro et al., 2018). We cloned these ULP genes fused with a HA tag at their N-termini into the pRSF-Duet-1 vector, in which the original kanamycin resistance gene was replaced by an ampicillin resistance gene (Figure S1A), for expression of ULP proteins in E. coli. Consequently, introducing different ULP members into this bacteria-based SUMOylation reconstitution system may offer a powerful approach for rapidly identifying de-SUMOylation enzymes for substrates. A synthetic biology approach for identifying de-SUMOylation enzymes of substrates (A) Schematic overview of the reconstitution system. Eight Arabidopsis ULPs (HA-tagged) were individually introduced into bacterial cells containing SUMO1GG (the mature form of SUMO1), SAE1/SAE2 (the SUMO E1 complex), and substrate (Flag-tagged), with or without SCE1 (the SUMO E2), in order to determine the de-SUMOylation enzyme of substrates. (B–E) Identification of the ULPs of FHY1 and NF-YC5 in the established reconstituted system. The Flag-tagged FHY1 or NF-YC5 was expressed with or without each ULP in the reconstituted bacterial cells, and the proteins were then subjected to immunoblotting using anti-Flag and anti-HA antibodies. Representative immunoblots for FHY1-Flag are shown in (B) with quantification data included in (C), while representative immunoblots for NF-YC5 are shown in (D) with quantification data included in (E). Unmodified protein bands are indicated by black triangles, SUMOylated protein bands are indicated by asterisks, and ULP proteins are indicated by black spots. The signals of protein bands were quantified using ImageJ, and the levels of SUMOylation were calculated from relative signals ([SUMOylation/non-SUMOylation] of ULP group/[SUMOylation/non-SUMOylation] of the control group), where the relative SUMOylation level of the control group (without ULP) was set to 1. (F, G) Verification of the impact of ESD4 and ELS1 on the SUMOylation of NF-YC5 in plant cells. HA-ESD4, HA-ELS1, or HA-SPF1 was co-expressed with NF-YC5-GFP and His6-Flag-SUMO1GG in Arabidopsis protoplasts for SUMOylation detection. The GFP-tagged protein was precipitated using anti-GFP agarose beads. Signals of GFP, Flag, and HA were detected in immunoblots with anti-GFP, anti-Flag, and anti-HA antibodies, respectively. Representative immunoblots are presented in (F) and quantification data are included in (G). The data in (C), (E), and (G) are means ± SD from three biologically independent experiments. Significance was analyzed via Tukey's test (one-way ANOVA), with different letters representing significant differences (P < 0.05). To validate the feasibility and specificity of this assay, we initially chose Arabidopsis FAR-RED ELONGATED HYPOCOTYL 1 (FHY1) as a substrate, given its crucial role in regulating plant light responses. FHY1 is known to undergo SUMOylation, with SPF1 (also named ARABIDOPSIS SUMO PROTEASE 1, ASP1) identified as its de-SUMOylation enzyme (Qu et al., 2020). To this end, FHY1 was cloned into the pCDF-Duet-1 vector (Figure S1B) with a C-terminal Flag tag, resulting in the generation of the pCDF-Duet-1-FHY1-Flag construct. This construct was then co-transformed with a pRSF-Duet-1 vector containing one of eight Arabidopsis ULPs (the empty pRSF-Duet-1 vector serving as the negative control) into E. coli competent cells expressing AtSAE1/AtSAE2 (the SUMO E1 complex), and AtSUMO1GG (the mature SUMO1), along with or without AtSCE1 (the SUMO E2). Immunoblotting analysis revealed that, in the absence of ULPs, multiple high-molecular-weight shifts of FHY1-Flag were observed in the presence of SCE1; however, these shifts were absent in samples lacking SCE1. This pattern of FHY1 SUMOylation closely resembled previous results obtained from an in vitro assay using purified enzyme components (Qu et al., 2020), thereby confirming the successful reconstitution of FHY1 SUMOylation within the E. coli system. Most members of ULPs did not exhibit significant suppression on FHY1 SUMOylation compared with the vector control sample. Conversely, expression of SPF1 led to a marked decrease in FHY1's level of SUMOylation (Figure 1B, C), thus supporting its unique role as the de-SUMOylation enzyme for FHY1. These findings align with prior reports and provide compelling evidence for both specificity and reliability within our reconstitution system. Next, it is crucial to conduct a test to determine whether the system was capable of identifying the de-SUMOylation enzyme for uncharacterized SUMOylation substrates. Our recent study showed that the SUMOylation of NUCLEAR FACTOR Y SUBUNIT C10 (NF-YC10) is involved in plant thermotolerance (Huang et al., 2023). According to bioinformatics analysis via GPS-SUMO (Figure S1C), NF-YC5, another member of the Arabidopsis NF-Y family, was selected as a potential SUMOylation substrate for identification of its ULP in our established system. Consequently, we cloned Flag-fused NF-YC5 in the pCDF-Duet-1 vector for protein expression and SUMOylation analysis. The results revealed that NF-YC5-Flag exhibited multiple high-molecular-weight bands in the presence of SCE1, while these shifts were not observed in samples without SCE1, confirming that NF-YC5 is indeed a SUMOylation substrate. Further site mutagenesis data revealed that the lysine residues K19 and K33 are the primary SUMOylation sites on NF-YC5 (Figure S1D, E). Subsequently, we investigated the impact of ULPs on the SUMOylation of NF-YC5. Interestingly, it was found that the expression of either ESD4 or ELS1 led to a decrease in the level of SUMOylation of NF-YC5, whereas other ULPs did not have this effect (Figure 1D, E), suggesting that ELS1 and ESD4 are responsible for de-SUMOylating NF-YC5. To validate this conclusion, we conducted an immunoprecipitation assay to investigate the impact of ESD4 and ELS1 on the SUMOylation of NF-YC5 in plant cells. The findings revealed that the overexpression of ESD4 or ELS1 inhibited the SUMOylation of NF-YC5; however, the overexpression of SPF1 (a negative control) did not significantly impact NF-YC5 SUMOylation (Figure 1F, G), which is consistent with our reconstitution system data. These results provide further support for the efficacy and specificity of our established assay in identifying de-SUMOylation enzymes for uncharacterized SUMOylation substrates. In conclusion, we have developed a robust reconstitution system in E. coli for screening the ULPs of SUMOylation substrates. (All the vectors in this system will be shared by Chengwei Yang upon request.) The efficiency and specificity of the established assay were confirmed using the known substrate FHY1 and the unknown substrate NF-YC5. The specificity of ULPs may arise from their interactions with substrates but, given that the interaction between de-SUMOylation enzymes and substrates may be transient and weak, identifying enzyme–substrate pairs based on interactions can be challenging. Therefore, our current approach is well suited for offering insights into the investigation of functional associations between de-SUMOylation enzymes and their substrates in plants, even for processes regulated by redundant enzymes. However, as the assay was conducted in bacterial cells, it is necessary to verify the identified enzyme–substrate pairs in plant cells. In this work, we utilized eight classic ULPs in Arabidopsis; however, certain studies suggested that additional proteases, such as Desi3a, also function as de-SUMOylation enzymes in Arabidopsis (Orosa et al., 2018). Therefore, new de-SUMOylation enzymes can be included in our toolbox. Furthermore, given that detection of SUMOylation in plant cells typically relies on SUMO antibodies or transgenic plants, conducting screenings in bacterial cells offers a faster and simpler alternative, especially for plant species in which transgene expression is difficult to achieve. As endogenous SUMO proteases are not present in bacterial cells, it is easier and more reliable to detect SUMOylation and assess the effects of ULPs using our current system. Although our system was developed using components from Arabidopsis, due to the conservation of SUMOylation mechanisms across eukaryotes, this strategy can also be applied to identify ULPs for substrates from other plant species as well as mammals. We would like to thank Dr. Zhenkun Zhang (Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences) for the pRSF-Duet-1 vector and Professor Andreas Bachmair (University of Vienna) for the SUMO E1 plasmid. This work was supported by the Major Program of Guangdong Basic and Applied Research (2019B030302006), the National Natural Science Foundation of China (32270292, 32270752), the Natural Science Foundation of Guangdong (2019A1515110330, 2018B030308002, 2024A1515011071), Guangdong Modern Agro-industry Technology Research System (2023KJ114), the Program for Changjiang Scholars, the Scientific Research Start-up Fund for PhD of Zhaoqing University (240013), and Youth Foundation of Zhaoqing University (QN202439). The authors declare no conflicts of interest. C.Y. and J.L. supervised the project; Junwen H., Junjie H., J.W., M.Z., and S.L. performed experiments; J.J., T.C., and L.S. provided technological support; Junwen H., Junjie H., J.L., and C.Y. analyzed the data and wrote the manuscript. All authors read and approved the contents of this paper. Additional Supporting Information may be found online in the supporting information tab for this article: http://onlinelibrary.wiley.com/doi/10.1111/jipb.13838/suppinfo Figure S1. Plasmid information and identification of NUCLEAR FACTOR Y SUBUNIT C5 (NF-YC5) SUMOylation sites Table S1. Primers used in this study Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.