The cDNAs encoding SENP1 (NP_001254524.1), Lck (NP_001036236.1), ZAP70 (NP_001365523.1), SUMO1 (NP_003343.1), and SUMO3 (NP_008867.2) were amplified by PCR from Jurkat E6.1 T cell cDNA and cloned into the vectors pcDNA3.1-c-Myc (Invitrogen), YFP-C1 (BD), and pGEX-4T-2 (Sigma-Aldrich), respectively. Specific point mutations of SENP1 were introduced by site-directed mutagenesis with a QuikChange Site-Directed Mutagenesis Kit (Stratagene). Primers are shown in Supplementary Table S1.

Antibodies to SENP1 (C-12), Lck (3A5), RanGAP1 (C-5), c-Myc (9E10), β-actin (C4), and GAPDH (FL-335) were from Santa Cruz Biotechnology. Antibodies to SENP1 (ab108981), SUMO1 (ab32058), and SUMO2/3 (ab3742) were from Abcam. Antibodies to HA (C29F4) and glutathione S-transferase (GST) were from Cell Signaling Technology. M2 antibody to Flag (F1804) was from Sigma-Aldrich, and 4G10 (05-1050) was from Millipore. Anti-human CD3 (OKT3) and anti-human CD28 (CD28.2) were from eBioscience. Goat anti-mouse IgG (31160) was from Thermo Fisher Scientific. Horseradish peroxidase-conjugated secondary antibodies were from Jackson ImmunoResearch.

Human leukemia Jurkat T cell line E6.1 cells, TAg cells, JCaM1.6 cells, and SENP1 ^−/− cells were cultured in RPMI 1640 medium (HyClone Logan, UT, United States) supplemented with 10% (vol/vol) fetal bovine serum (FBS, Gibco) and 100 U/ml each of penicillin and streptomycin (Life Technologies) at 37°C and 5% CO₂. Cells in a logarithmic growth phase were transfected by nucleofection (Lonza 4D Nucleofector™ system). After transfection, cells were cultured for 48 h before harvest. HEK293T cells were cultured in DMEM (HyClone) containing penicillin, streptomycin, and 10% FBS. Transfections were carried out with PEI (Sigma-Aldrich). For TCR stimulation with antibodies, cells were washed with serum-free RPMI 1640 medium and stimulated with 5 μg/ml anti-CD3 and 2 μg/ml anti-CD28, which were crosslinked with goat anti-mouse IgG (5 μg/ml). For pervanadate treatment, cells were incubated with tyrosine phosphatase inhibitor pervanadate (10 mM) for 20 min before harvest.

After washing with ice-cold PBS, cells were lysed in lysis buffer (1% Nonidet P40, 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, and 5 mM EDTA), supplemented with protease inhibitors (10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mM PMSF) and phosphatase inhibitors (5 mM sodium pyrophosphate and 1 mM Na₃VO₄). After clearance by centrifugation, whole-cell lysates were incubated overnight with 0.6 µg of antibodies, and proteins were immunoprecipitated for an additional 4 h at 4°C with protein G Sepharose beads (GE healthcare) with gentle shaking. Immunoprecipitates were extensively washed five times with lysis buffer and were separated by SDS-PAGE. The immunoprecipitated proteins were transferred onto a PVDF membrane and probed overnight at 4°C with primary antibodies, followed by incubation for 1 h at RT with HRP-conjugated secondary antibodies. Signals were visualized by enhanced chemiluminescence (ECL; GE Healthcare) and were exposed to X-ray film or on the ChemiDoc XRS+ system (Bio-Rad). Densitometry analysis was performed with ImageJ software.

After immunoprecipitation with SENP1 antibody in Jurkat E6.1 T cells, the immunoprecipitates were resolved by SDS-PAGE. Gel bands of interest were excised and subjected to tryptic digestion. After the desalting procedure, the peptides were analyzed by liquid chromatography-tandem MS. The peptides were analyzed by the LTQ Orbitrap Elite system and EASY-nLC 1000 system (Thermo Fisher Scientific), and then proteins in SENP1 immunoprecipitates were identified with the Protein Discovery software. Mass spectrometry proteomic data from this study are deposited to the ProteomeXchange-PRIDE repository (https://www.ebi.ac.uk/pride/archive/) and assigned the dataset number PXD029720.

LentiCRISPRv2 (one vector system) coexpressing a mammalian codon-optimized Cas9 nuclease along with sgRNA to facilitate editing a SENP1 genomic fragment was transfected in the Jurkat E6.1 T cell line (Shalem et al., 2014; Sanjana et al., 2014). Then, 48 h after transfection, green fluorescent protein (GFP)-positive cells were sorted on a flow cytometer (BD FACSAria II) and seeded individually into 96-well plates. After extended culture, candidate monoclonal cell lines were selected by screening using western blot. The identifications of SENP1 genome-edited cell line are shown in Figure 5 and Supplementary Figure S3.

GST fusion proteins were expressed in BL21 E. coli cells after induction with IPTG (Sangon Biotech). Bacteria were resuspended in lysis buffer (1× PBS pH 7.4, protease inhibitors, and 1% Triton X-100). Bacterial extracts were sonicated for 10 min and centrifuged. GST fusion proteins were purified from bacterial lysates by incubating with glutathione Sepharose beads (GE Healthcare). The precipitates were washed three times with lysis buffer, then incubated at 4°C overnight with cell lysates from HEK293T cells. Coomassie Brilliant Blue staining was used as loading control.

In vitro endopeptidase activity assay was achieved as previously described (Madu et al., 2013; Xu and Au, 2005). Briefly, GST-SUMO-GFP (Gs-S-Gf) and SENP1 proteins [wild type (WT) or mutants] were expressed in E. coli BL21 cells and purified. To assay the hydrolysis activity, eluted Gs-S-Gf proteins were incubated in the absence (control) or presence of SENP1 WT or its mutants (immobilized on glutathione Sepharose beads), respectively, and then incubated in assay buffer (150 mM NaCl, 10 mM Tris-HCI, and 10 mM DTT) at 37°C for 20 min and boiled in SDS loading buffer, followed by SDS-PAGE and Coomassie staining to analyze.

SENP1 ^−/− T cells generated by CRISPR/Cas9 system were lysed in lysis buffer supplemented with 20 mM N-ethylmaleimide (Sigma-Aldrich). After clearance by centrifugation, 1% SDS (vol/vol) was added to the supernatants. Proteins were dissociated by incubation at 90°C for 10 min, followed by dilution (1:10) with ice-cold lysis buffer immediately. Endogenous total RanGAP1 (including RanGAP1 and RanGAP1-SUMO1) was immunoprecipitated from the above lysis buffer at 4°C overnight with anti-RanGAP1 antibody and immobilized on protein G Sepharose beads. After extensive washing with lysis buffer, immunoprecipitated proteins were equally divided into aliquots as reaction substrate, followed by incubation with GST-SENP1 WT or its mutants (immobilized on glutathione Sepharose beads) in reaction buffer (150 mM NaCl, 10 mM Tris-HCI, and 2 mM DTT) at 37°C for 20 min, respectively. The reaction was stopped by adding SDS loading buffer, followed by SDS-PAGE and immunoblotting to analyze.

The 3D structure of SENP1 (full length, 1–644 aa) was predicted using I-TASSER (https://zhanglab.ccmb.med.umich.edu/I-TASSER/) (Roy et al., 2010; Yang et al., 2015, Yang and Zhang, 2015). The predicted structures of SENP1 Y270F and Y270E mutation were acquired using Missense 3D (http://missense3d.bc.ic.ac.uk/missense3d/) (Ittisoponpisan et al., 2019). The display of structures and measurement of distances were performed by PyMOL software.

To examine the expression level of SENP1, Jurkat E6.1 and SENP1 ^−/− T cells were collected, washed with 1% FBS in PBS buffer, and then permeabilized (Cytofix/Cytoperm Plus; BD) at 4°C for 20 min, followed by washing with 1% FBS/0.5% saponin in PBS buffer. After being sequentially stained with anti-SENP1 antibody on ice overnight and with cross-adsorbed secondary antibody (Alexa Fluor 647) at 4°C for 1 h, the cells were washed three times and resuspended in 1% FBS in PBS buffer, and then examined by using the CytoFLEX flow cytometer (Beckman).

Total RNA was extracted from Jurkat E6.1 and SENP1 ^−/− T cells using TRIzol reagent (Magen) according to the manufacturer’s protocol. RNA was eluted with RNase-free water and quantified using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific). First-strand cDNA was synthesized by reverse transcription with Hifair III 1st Strand cDNA Synthesis Kit (gDNA digester plus, YEASEN, CHINA). Real-time PCR was run on a LightCycler 480 system (Roche) with Hieff qPCR SYBR Green Master Mix (YEASEN, CHINA) in triplicate. Real-time PCR primers sequences are shown in Supplementary Table S1. The relative quantity of mRNA was normalized with GADPH, using the comparative 2^–ΔΔCt method.

Jurkat E6.1 T cells transfected with empty vector or YFP-SENP1 WT, Y270F, Y270E, and C603S were stimulated for 24 h with anti-CD3 plus anti-CD28, and the determination of the concentration of IL-2 in cultured supernatants by ELISA was achieved as described (Wang et al., 2019,2015). Briefly, a 96-well plate was coated with capture antibody to IL-2 at 4°C overnight. After pre-blocking at RT for 1 h, triplicates of standards and samples were then incubated at RT for 2 h in the plate. Incubation with detection antibody to IL-2 at RT for 1 h was done, followed by incubation with avidin-HRP at RT for 30 min. The amount of bound avidin was then assessed with TMB peroxidase that was acidified by 2N H₂SO₄. The absorbance of each well at 450 and 570 nm was then measured with a spectrophotometric plate reader (BioTek).

Data is representative of three independent experiments. GraphPad Prism 5.0 software was used for graphs and statistical analysis, performed with a two-tailed, unpaired Student’s t-test. For all tests, statistical significance was quantified by p-values. When p-values were below 0.05 (*), 0.01 (**), 0.001 (***), or 0.0001 (****), differences were considered statistically significant. In the figures, p-values are indicated as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, and ns, not significant. Graphs represent mean ± standard error of the mean (SEM).

To investigate whether SENP1 is regulated by TCR signaling, we first analyzed SENP1-associated proteins in T cells. In mass spectrometric analysis, nine peptides of Lck were shown in SENP1-immunoprecipitated complex from Jurkat E6.1 T cells (Figure 1A), and two of them had higher frequencies (Figure 1B), suggesting Lck may interact with SENP1. Coimmunoprecipitation analysis verified the association between overexpressed SENP1 and Lck in HEK293T cells (Figure 1C) and between endogenous SENP1 and Lck in Jurkat E6.1 T cells (Figure 1D).

Since Lck is a crucial tyrosine kinase in TCR proximal signaling (Shah et al., 2016), we next tested if SENP1 could be phosphorylated by Lck as a substrate. In HEK293T cells, tyrosine phosphorylation of exogenous SENP1 was detected when SENP1 was coexpressed with Lck, whereas it was undetectable when SENP1 was coexpressed with ZAP70, an important downstream kinase of Lck (Figure 2A). In Jurkat E6.1 T cells, anti-CD3 plus anti-CD28 costimulation induced tyrosine phosphorylation of endogenous SENP1; however, this induced phosphorylation disappeared in Lck-deficient JCaM1.6 cells (Figure 2B). We further cotransfected SENP1 with Lck or its kinase dead mutant (K273R) in HEK293T cells and observed that SENP1 was tyrosine phosphorylated only when cotransfected with Lck WT but not with its K273R mutant (Figure 2C), indicating that the kinase activity of Lck was essential for SENP1 phosphorylation. Together, these results demonstrated that SENP1 could be phosphorylated by tyrosine kinase Lck but not ZAP70.

Regarding phosphorylation sites, Lck has a strong preference for Ile, Leu, or Val at the −1 position, whereas it cannot tolerate Asp at this position, and Lck has a moderate preference for bulky hydrophobic residues at the +3 position (Shah et al., 2018; Shah et al., 2016). To identify the crucial phosphorylation site on SENP1 by Lck, we checked all tyrosine residues and their −1 and +3 position residues on SENP1 and noticed that four tyrosine residues Tyr119, Tyr194, Tyr270, and Tyr349 are in accordance with the rule of Lck phosphosite preference (Figure 3A). Interestingly, all of them are located in the N-terminal non-catalytic domain of SENP1 (Figure 3B). By mutating tyrosine to phenylalanine, we constructed Y119F, Y194F, Y270F, and Y349F mutants of SENP1, respectively. Through cotransfection of SENP1 YF mutants separately with Lck in HEK293T cells, we examined the effect of each YF mutation on phosphorylation of SENP1 by Lck. We found that Lck-mediated tyrosine phosphorylation of SENP1 remarkably disappeared when mutating Tyr270 to Phe (Figure 3C). Similarly, Y270F mutation notably decreased tyrosine phosphorylation of SENP1 induced by pervanadate (tyrosine phosphatase inhibitor) treatment in Jurkat TAg cells (Figure 3D). However, TCR-induced phosphorylation of SENP1 only reduced modestly with anti-CD3 stimulation (Figure 3E), suggesting that other tyrosine sites of SENP1 may also be phosphorylated upon TCR stimulation. SENP1 Tyr270 was a middle conserved site among various species, as its conservative index is 0.7 in alignment of 58 species in the database of VarSite; meanwhile, SENP1 Tyr270 is completely conserved in primates (Supplementary Figure S1), suggesting Y270 phosphorylation might be related to primate-specific immune responses.

Without characteristic features, the N-terminal regions of SENPs presumably control their enzymatic properties intrinsically (Mikolajczyk et al., 2007). We questioned if SENP1 Y270 phosphorylation could regulate its enzyme activity. Firstly, we assessed the impact of SENP1 Y270 phosphorylation on its endopeptidase activity by an in vitro assay (Figures 4A,B). We fused full-length SUMO1, i.e., precursor with GST at its N-terminus, with GFP at C-terminus to construct the recombinant SUMO1 precursor simultaneously fused with two tags (GST-preSUMO1-GFP or Gs-pS1-Gf, 64 kDa) and used it as a substrate of SENP1 endopeptidase according to the literatures (Xu and Au, 2005; Madu and Chen, 2012). Proteolytic cleavage of Gs-pS1-Gf by SENP1 yields two products, GST-SUMO1 (GST fused with mature SUMO1, approximately 38 kDa) and GFP moieties (HSTV peptide fused to N-terminus of GFP, about 26 kDa). The ratio of product/substrate reflects the reaction efficiency (Xu and Au, 2005; Madu and Chen, 2012). As expected, the capability of SENP1 to remove C-terminal GFP moiety from Gs-pS1-Gf was abolished by the catalytic inactive C603S mutation. However, neither a tyrosine phosphorylation-deficient Y270F mutant nor a Y270E mutant (which mimics the negative charge of tyrosine phosphorylation in the enzyme microenvironment) of SENP1 had any impact on its endopeptidase activity towards pre-SUMO1, indicating that Y270 phosphorylation on SENP1 did not regulate its endopeptidase activity towards pre-SUMO1.

Since SENP1 prefers pre-SUMO1 > pre-SUMO2 > pre-SUMO3, we guessed that SENP1 Y270 phosphorylation might also not affect its endopeptidase activity towards pre-SUMO2 or pre-SUMO3. We further performed an in vitro assay against SUMO3 precursor Gs-pS3-Gf (Xu and Au, 2005; Mikolajczyk et al., 2007). Due to the substrate preference of SENP1 endopeptidase activity, the bands related to GST-SUMO3 were faint. Unlike Gs-pS1-Gf, there was a band near above 25 kDa in all lanes when using Gs-pS3-Gf even in the absence of GST-SENP1 (labelled with GFP/GST in Figure 4B). Western blot showed that the Gs-pS3-Gf protein could be cleaved to generate GST-tag protein for uncertain reason (Supplementary Figure S2); thus, the band near above 25 kDa contained mixed proteins of GFP moiety (VPESSLAGHSF peptide fused to the N-terminus of GFP) and GST-tag. Nevertheless, we performed endopeptidase activity assay towards pre-SUMO3 with Coomassie staining and immunoblotting at the same time (Figures 4B,C). As expected, GST-SUMO3 could not be detected when reacting with SENP1 C603S mutant. Interestingly, GST-SUMO3 could not also be detected when reacting with SENP1 Y270E mutant, suggesting that SENP1 Y270E specifically decreases its hydrolytic ability to SUMO3 precursor.

The above data indicates that tyrosine phosphorylation on Y270 of SENP1 may modulate its endopeptidase activity with selectivity, which has no effects on its endopeptidase capacity toward pre-SUMO1, but decreases its endopeptidase activity against pre-SUMO3.

Next, we examined whether SENP1 Y270 phosphorylation could influence its isopeptidase activity. Considering that the presence of endogenous SENP1 may interfere with the effects of ectopic-expressed SENP1, we set to construct a SENP1 ^−/− T cell line by CRISPR/Cas9 system. The genome sequencing and western blot results showed that the expression of CRISPR/Cas9-gRNA targeting the first exon of SENP1 induced a thymidine nucleotide insertion in SENP1 locus to knockout SENP1 gene (Figures 5A,B; Supplementary Figures S3A,B).

Expectedly, SENP1 ^−/− T cells displayed much stronger SUMO1 and SUMO2/3 modification (Supplementary Figure S3C). By checking the mRNA expression of SENP2 (another sub-family member of SENP1) in SENP1 ^−/− T cells, we found that SENP1 knockout did not affect the transcription of SENP2 (Supplementary Figure S3D), and western blot showed that the expressions of TCR proximal signaling proteins were not affected in SENP1 ^−/− T cells (Supplementary Figure S3E). Therefore, the increased sumoylation in SENP1 ^−/− T cells should be the direct result of SENP1 knockout. Global sumoylation reflects the isopeptidase activity of SENP1; hence, we performed an in vivo SUMO1-profiling assay in this SENP1 knockout cell line to assess the impact of SENP1 Y270 phosphorylation on its isopeptidase activity (Figure 5C). As expected, SENP1 WT overexpression efficiently reduced the overall sumoylation of cellular proteins, and the capability of SENP1 to remove SUMO1 conjugates from protein substrates was abolished by C603S mutation. Compared with SENP1 WT, the Y270F mutation obviously decreased its desumoylation activity, although this decrease was less than that caused by C603S mutation (Figure 5C), indicating that Y270 tyrosine phosphorylation contributed to the isopeptidase activity of SENP1. Interestingly, the Y270E negative charge mimic mutation did not increase the desumoylation activity of SENP1 but only showed similar isopeptidase activity to SENP1 WT in vivo (Figure 5C). A similar pattern of desumoylation activities of SENP1 WT and its mutants is shown in Figure 5D, when analyzing their activities by the ratio of RanGAP1-SUMO1/RanGAP1 in SENP1 ^−/− T cells with rescue expression of YFP-SENP1 or its mutants (Figure 5D). Considering the similar activities between SENP1 WT and Y270E mutant, we guessed that in T cells, SENP1 Y270E mutant did not show a stronger activity than SENP1 WT that might result from a certain degree of constitutive Y270 phosphorylation of SENP1 WT. To exclude the possible influence of basal phosphorylation on SENP1, we performed an in vitro isopeptidase activity assay, in which E. coli recombinant GST fusion proteins of SENP1 WT (which could not be phosphorylated) and its mutants were incubated with immunoprecipitated endogenous total RanGAP1 (including RanGAP1 and RanGAP1-SUMO1) from SENP1 ^−/− T cells, respectively (Figure 5E). Basing on the ratio of RanGAP1-SUMO1/RanGAP1, we observed that the desumoylation activity of SENP1 was decreased by both Y270F and C603S mutations, but was increased by Y270E mutation (Figure 5E), indicating Y270 phosphorylation could increase SENP1 isopeptidase activity. Similar results from in vivo SUMO2/3-profiling assay also suggested that SENP1 Y270 phosphorylation influenced its isopeptidase activity (Figure 5F). Combining the above in vivo and in vitro results, we conclude that Y270 phosphorylation is required for the optimal isopeptidase activity of SENP1.

In T cells, SENP1 overexpression inhibits TCR-induced IL-2 production (Wang et al., 2015). Thus, we further examined the effects of Y270 phosphorylation on TCR-induced IL-2 production in Jurkat E6.1 T cells by ELISA (Figure 5G). Consistently, TCR-induced IL-2 production was inhibited by overexpression of SENP1 WT and was increased by C603S mutation, confirming that desumoylation is overall a negative factor in T cell activation. Meanwhile, the IL-2 production was enhanced by Y270F mutation and inhibited by Y270E mutation (Figure 5G), aligned with their desumoylation effects (Figures 5C–F).

We next investigated how Y270 phosphorylation regulated SENP1 isopeptidase activity. Firstly, we analyzed whether SENP1 Y270 phosphorylation could directly change its protein conformation, especially the catalytic core domain, to regulate its enzyme activity. The predicted structures of SENP1 WT (1–644 aa), Y270F, and Y270E mutation were obtained, and His533, Asp550, and Cys603 formed the catalytic triad (Figure 6A). We measured the distances between these key residues in these structures and found that neither Y270F nor Y270E mutation brought obvious changes to it (Figure 6B). Since the binding of SUMO1 β-grasp domain (residues 20–92) induces structural changes in the active site of SENP1 and then enhances the enzyme activity of SENP1 (Chen et al., 2014; Guo and Zhou, 2016), the binding of SUMO to SENP1 is hence a prerequisite for SENP1 isopeptidase activity, and we wondered whether Y270 phosphorylation on SENP1 could regulate its isopeptidase activity via modulating its binding to SUMO-conjugated substrates. We performed the GST pull-down assay by using GST fused to the N-terminus of mature SUMO1 (1–97 aa) or SUMO3 (1–92 aa). Y270F mutation reduced SENP1 binding to both GST-SUMO1 and -SUMO3. In contrast, Y270E mutation increased its binding to both (Figures 6C,D). These results indicated that Y270 phosphorylation may increase the binding propensity between SENP1 and sumoylated proteins, which can in turn enhance the isopeptidase activity of SENP1. We supposed that allosteric activation of SENP1 by SUMO1 may bring residue 270 closer to SUMO in space, and Y270 phosphorylation may stabilize and enhance the binding between SENP1 and SUMO, which could lead to more efficient desumoylation of the substrates.