| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Centre for Protein Science and Crystallography, Department of Biochemistry and Molecular Biotechnology Program, Faculty of Science, The Chinese University of Hong Kong, Hong Kong
1 Correspondence: Centre of Protein Science and Crystallography, Department of Biochemistry, Faculty of Science, The Chinese University of Hong Kong, Hong Kong. E-mail: shannon-au{at}cuhk.edu.hk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: deconjugation • cysteine protease • oxidative stress
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. Its ability to induce rapid changes in protein functions has reflected its importance in cellular stress response. Recent evidence has further tightened the association between sumoylation and various stress conditions. For instance, global sumoylation has been shown to be up-regulated under oxidative stress from which more new SUMO substrates are identified (3
, 4)
.
Oxidative stress is an intriguing aspect due to the diversified nature of its production and cellular effects. Reactive oxygen species (ROS), such as H2O2, generated intracellularly by the action of phagocytic ROS-generating enzymes, assume an essential beneficial role in the host immune system (5)
. ROS also serve as second messengers in various receptor signaling pathways, like those of insulin (6)
and vascular endothelial growth factor (VEGF) (7)
. Some recent studies suggest that ROS participate in normal aging and regulate longevity in worms (8)
and mammals (9
, 10)
. However, excess ROS, regarded as unwanted byproducts of oxidative phosphorylation, lipid metabolism, and ionizing radiations, can over-oxidize proteins resulting in cell damage. A variety of diseases such as cancer, atherosclerosis, and neurodegeneration can be induced by abnormal cellular redox status (11)
.
For cell signaling, redox-sensitive cysteine residues serve as transducers at different cellular redox states to allow proteins to respond to ROS. The SH group of cysteine residue can be converted to reversible -SOH, disulfide and cyclic sulfenamide, irreversible SO2 and SO3, or even Cys-S-NO, which is formed by further modification of reactive nitrogen species (RNS) (12
, 13)
. The pKa of the sulfhydryl group of most cysteine residues is 
8.5, and therefore is less readily oxidized. However, certain protein cysteine residues that possess a lower pKa by interaction with the nearby positively charged residues, exist as thiolate anions at neutral pH. Several lines of investigation have implicated that cysteine residues play ubiquitous roles in both detecting changes in redox status and mediating a change in protein structure and function. To date, proteins with low pKa cysteine residues include bacterial OxyR (14)
and Hsp33 (15)
, yeast Yap1 (16
, 17)
and peroxiredoxins PrxI and PrxII (18)
, and mammalian cysteine-based phosphatases (19)
.
In sumoylation, cysteine residues are crucial for the formation of E1-SUMO and E2-SUMO thioester intermediates in SUMO conjugation, and for the catalysis of peptide and isopeptide bonds in SUMO deconjugation (20
21
22
23)
. Recent findings demonstrate that a low concentration of H2O2 (1 mM) inhibits SUMO conjugation in vivo by inducing disulfide cross-linkage of the active cysteine residues of Uba2 (the large subunit of E1) and E2 (24)
. Interestingly, changes of the sumoylation pattern cannot be correlated in a manner proportional to the concentration of H2O2. In vitro FRET assay has demonstrated that highly oxidizing conditions (1 and 10 mM H2O2) inhibit the deconjugation activity of SENP1. It appears that increased SUMO conjugation under high oxidative stress could be due to an influence on SUMO proteases.
To address the deficit in current knowledge about the redox regulation of desumoylation, we studied the oxidative consequences on three different SUMO proteases: yeast Ulp1 and human SENP1 and SENP2. Herein, we report that all these proteases were inhibited under oxidizing conditions and with varying degrees of recovery abilities. The discrepancies were due to their abilities to form intermolecular disulfide dimer and oligomer. In vivo SENP1 dimerization was also detected when cultured cells were incubated with H2O2. The distinct role of the disulfide linkage observed in the SENP1 and Ulp1 was demonstrated by its protection against further irreversible oxidation and enzyme inhibition. The structural basis of the modification was further revealed by the three high-resolution crystal structures of Ulp1 with catalytic cysteine oxidized to sulfenic, sulfinic, and sulfonic acids.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. SENP1C(C535S), SENP1C(C560S), and SENP1C(C613S) were generated by site-directed mutagenesis with oligonucleotides carrying the targeted Cys to Ser mutation. Constructs for SENP2C and Ulp1C were designed as described previously (22
, 25)
, amplified from a human control cDNA library of human MTCTM panel I (BD Biosciences, Franklin Lakes, NJ, USA) or self-prepared genome DNA library from Saccharomyces cerevisiae, and cloned into pGEX-6p-1 expression vector. Yeast E1 was cloned into pGEX-6p-1 to coexpress Uba2 and Aos1 as the human E1. Yeast E2 was cloned into the pGEX-6p-1 vector. Full-length SENP1 was amplified from a human control cDNA library of human MTCTM panel I (BD Biosciences) by PCR and cloned into pHM6 (Roche) and pcDNA4.1 (Invitrogen, Carlsbad, CA, USA) expression vectors for the expression of HA-tagged SENP1 and His-tagged SENP1, respectively.
Protein expression and purification
Recombinant proteins SENP1C, SENP2C, and Ulp1C were expressed in E. coli BL 21 and purified by affinity and gel-filtration chromatography according to standard protocols. GST tag was removed from human E1, E2; yeast E1, E2; and various SUMO proteases prior to gel-filtration chromatography. The purified proteins were stored at –80°C before use.
In vitro hydrolytic activity assay
TDG sumoylation reaction was performed by incubating TDG with E1, E2 and mature SUMO at 37°C, for human SUMO conjugation, or 30°C, for yeast conjugation reaction. The reaction mixtures were incubated for 30 min in the presence of 10 mM ATP. Oxidative treatment was performed by incubation of various SUMO proteases with different concentrations of H2O2, as indicated, for 30 min at 4°C. The oxidized proteins were further incubated with different concentrations of DTT, as indicated, for another 30 min in the recovery assay. The hydrolytic activity was investigated by incubating various wild-type or mutant SUMO proteases with sumoylated TDG, premature SUMO, or thioester-linked E1-SUMO in reaction buffer containing 20 mM Tris, 150 mM NaCl, pH 7.5. In all the above assays, reactions were terminated by adding 5x protein loading dye [10% (w/v) SDS] with or without DTT, as indicated, and the reaction products were subjected to SDS-PAGE analysis with Coomassie stain.
Western blot analysis of SENP1 in CHO cells
Cultured CHO cells were maintained at 37°C in 5% CO2 in Dulbecco’s modified Eagles’s medium (DMEM) (Invitrogen) supplemented with penicillin, streptomycin, and 10% fetal bovine serum (Invitrogen). Cells (1x106) were seeded in a 60 mm dish for 2 days before H2O2 treatment. Cells treated with 0, 0.1, 1, and 10 mM H2O2 for 3 h were then lysed in DTT-containing or DTT-free SDS loading dye and subjected to 15% SDS-PAGE analysis. Immunoblotting was performed with 1:500 goat anti-SENP1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and 1:5000 anti-goat secondary HRP antibody (Santa Cruz). Detection was made by ECL chemiluminescent kit (GE Healthcare, Uppsala, Sweden).
Dimer detection by immunoprecipitation
Transient cotransfection of HA-tagged SENP1 and His-tagged SENP1 in CHO cells was performed by lipofectamine 2000 (Invitrogen). Cells after 24 h cotransfection were treated with or without 30 mM H2O2 for 90 min and lysed in cell lysis buffer (20 mM Tris; 150 mM NaCl, pH 7.4; 1% Triton X-100; 0.5% Nonidet P-40 and 0.2 mM PMSF) for 20 min on ice. The lysate was centrifuged at 14,000 g for 15 min before preclearing with Protein G-Sepharose beads for 2 h at 4°C. The precleared supernatant was incubated with 1.5 µg of anti-HA antibody for 2 h followed by incubation with Protein G-Sepharose beads for another 1.5 h. A negative control experiment was set up in which no anti-HA antibody was added during immunoprecipitation. The beads were spun down and washed three times with the lysis buffer. The samples were boiled in DTT-free SDS loading dye and separated on 10% SDS-PAGE. Immunoblotting was performed with 1:20000 mouse anti-His antibody (GE Healthcare) and 1:5000 anti-mouse secondary HRP antibody (Santa Cruz). Detection was made by ECL chemiluminescent kit (GE Healthcare).
Crystallization of Ulp1C
Purified Ulp1C was incubated with 100 mM H2O2 for 30 min prior to crystallization. Initial crystallization screens were performed at 2.5 mg/ml with a sitting drop method using Hampton Index Screen (Hampton Research, Aliso Viejo, CA, USA). Crystals grew after 3 days in an environment containing 30%v/v pentarythritol ethoxylate (15/4 EO/OH) (pH 6.5) and 0.05 M ammonium sulfate at 16°C. This condition was optimized by adding 10% glycerol as additive and using the hanging drop vapor diffusion procedure.
Crystallization screens were also performed for Ulp1C without H2O2 treatment using Wizard screens I and II (Emerald Biosystems, Bainbridge Island, WA, USA). Crystals grew after 10 days in a buffer containing 30% (w/v) PEG 3000 (pH 7.0) and 0.2 M NaCl at 16°C. After optimization, a data set from a crystal grown in 20% (w/v) PEG 3350 (pH 6.5) and 0.2 M NaCl for 4 wk was collected. To reduce the oxidized form, a crystal grown in the same condition for 2 wk was postsoaked in 50 mM DTT for 48 h before data collection.
Data collection and structure determination
1.9 Å X-ray data set for H2O2-treated Ulp1C crystal, 2.0 Å X-ray data set for Ulp1C crystal, and 2.1 Å X-ray data set for Ulp1C crystal with DTT soaking were collected at 110 K using a Rigaku MicroMax 007 X-ray generator at the Centre for Protein Science and Crystallography, The Chinese University of Hong Kong, and recorded on a RAXIS IV++ image plate. For H2O2-treated Ulp1C crystal, crystallization buffer was used as a cryoprotectant. For the non-H2O2-treated Ulp1C crystals, 20% glycerol containing crystallization buffer was used. DTT (50 mM) was added to the cryoprotectant for the DTT soaking crystal. Images were processed using Mosflm (26)
and scaled and reduced with SCALA from the CCP4 suite (27)
. All the three protein crystals were in spacegroup C2, and unit cell parameters and statistics for the data collected are summarized in Table 1
. Coordinates have been deposited in the Protein Data Bank (PDB ID 2HKP, 2HL8, 2HL9).
|
The three structures were solved by molecular replacement using a molecule of Ulp1-SMT3 (PDB ID code 1EUV) as a search model. Molecular replacement program MolRep (28)
in CCP4 suite was performed with data in the resolution range 15–3.0 Å. The randomly selected 5% of data was reserved for the Rfree calculation for all the three structures. Rigid-body, simulated annealing, and positional refinements of the three oxidized forms were carried out by using the program CNS (29)
. Rounds of refinements were performed with manual rebuilding by using the program O (30)
. The electron density maps from 2Fo–Fc and Fo–Fc calculations were used for model building, and for all the three structures, strong electron density was found close to the S
atoms of the active cysteines. We modeled them as cysteinesulfonic acid, cysteinesulfinic acid, and cysteinesulfenic acid, respectively, The Ramachandran plots drawn by the program PROCHECK (31)
shows that over 88% of all residues in the three oxidized Ulp1 structures fall within the most favored region.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 4)
, and the regulatory mechanism by reversible inhibition of SUMO conjugating enzymes via disulfide linkage of E1 and E2 ligases has recently been reported (24)
. We sought to uncover whether SUMO proteases were affected on oxidation and the underlying molecular basis of the redox regulation of desumoylation. To begin, the oxidative effect on peptidase activity in maturation, isopeptidase activities in desumoylation, and thioesterase activity of the catalytic domain of human SENP1 (SENP1C) (23
, 32)
were examined. Using both SUMO-1 and SUMO-3, SUMO-conjugated thymine DNA glycosylase (TDG), thioester-linked Uba2-SUMO, and C-terminal GST-tagged SUMO precursor were prepared as substrate model (23
, 33)
(Lanes 1 and 2 of Fig. 1
). SENP1C with or without 10 mM H2O2 preincubation was added to the above assays (Lanes 3–6 of Fig. 1
). The results showed that the hydrolytic activities of SENP1C were abolished on H2O2 treatment. To further investigate whether the mechanism of inhibition was reversible or irreversible, recovery assays were performed by incubating H2O2-treated SENP1C with the same concentration of reducing agent DTT (Lanes 7 and 8 of Fig. 1
). In the presence of DTT, over 80% of the substrate turnover was observed, suggesting that H2O2-treated SENP1 had undergone partially, if not totally, reversible modification. As a control experiment, the activities of 10 mM DTT-treated SENP1C were comparable with those of untreated SENP1C (Lanes 9 and 10 of Fig. 1
), indicating that DTT only acted as a reducing agent to neutralize the oxidized effects on the protease.
|
Oxidative induction of Cys 603-Cys 613 intermolecular disulfide linkage of SENP1
Disulfide bridge formation is a common mediator for reversible protein modifications in the redox regulation of various signaling pathways (14
, 35)
. To investigate whether similar modification occurs in SENP1 on oxidation, we analyzed the molecular size of the enzyme after H2O2 treatment by nonreducing SDS-PAGE. Loss of the hydrolytic activities on H2O2 treatment suggests that the active-site Cys 603 is affected. Therefore, mutant SENP1C(C603S), in which Cys 603 is substituted by serine (23)
, was included in the assay. A concentration of 0.5 µg/µl protease was added in the reaction. In Lanes 5–8 of Fig. 2
A, a disulfide-linked dimer and oligomer of SENP1C were observed. However, SENP1(C603S) mutant exhibited a much weaker dimerization pattern, indicating that the active-site Cys 603 is one of the major thiol donors for the disulfide linkage and some other cysteine residues are also involved. On the other hand, the oxidative effect on SUMO was investigated. SUMO-1 containing one single cysteine residue also formed an intermolecular dimer (Lanes 9–10 of Fig. 2A
). Incubation of SENP1C and SUMO-1, together under oxidation conditions, induced formation of the higher-ordered complexes SENP1C-SUMO-1 and SUMO-1-SENP1-SUMO-1, indicating that more than one cysteine residue in SENP1C is responsible for the molecular association (Lanes 1–4 of Fig. 2A
). To verify the dimer formation of SENP1 is not due to nonspecific protein-protein interactions, analysis was repeated but with a titration of lower concentrations of SENP1C (Fig. 2B
). The dimeric form was detected when the protein concentration was as low as 0.01 µg/µl.
|
Further identification of the cysteine residues involved in the H2O2-induced intermolecular disulfide bond formation was based on the positions of Cys 535, Cys 560, Cys 603 (active site cysteine), Cys 513, and Cys 608 in the crystal structure of SENP1C (23
, 36)
. Residue Cys 608 is buried in the structure core and is less likely to be modified. Hence, mutants SENP1C(C535S), SENP1C(C560S), and SENP1C(C613S) were constructed. Under oxidation, SENP1C(C535S) and SENP1C(C560S) exhibited a very similar dimerization pattern as that of the wild-type SENP1C, while the pattern of SENP1C(C613S) resembles that of SENP1C(C603S) (Fig. 2C
). The results demonstrate that Cys 603 and Cys613 are the two main residues for dimer and oligomer formation.
Multiple sequence alignment of all three human SUMOs, (SUMO-1, -2, and -3) and yeast SMT3 revealed one conserved cysteine in human but not in yeast SUMO members (37)
. In this study, we examined whether H2O2 induced dimer formation of SUMO-3 and SMT3. As expected, SUMO-3, but not SMT3, can form a dimer (Fig. 2D
).
In vivo dimerization of full-length SENP1
Prior to any detailed characterization of the importance of SENP1C dimerization, in vivo monomer-dimer transformation of SENP1 under oxidation was studied. Cultured CHO cells were exposed to 0, 0.1 mM, 1 mM, and 10 mM H2O2 for 3 h, and the cell pellets were lysed in DTT-containing and DTT-free SDS loading dyes for Western blot analysis with an anti-SENP1 antibody. From Fig. 3
, a SENP1 dimer-sized band was clearly detected when the cells were incubated with 1 mM and 10 mM H2O2.
|
To verify whether the observed dimer-sized band resulted from the formation of intermolecular disulfide bond as shown in the in vitro studies, we cotransfected CHO cells with both HA- and His-tagged SENP1 for the immunoprecipitation assay. Different concentrations of H2O2 (1, 10 mM, 30 mM, and 100 mM) were tested and changes of endogenous SENP1 were examined (Fig. 3B
). Though much intense dimer-sized band was found in 100 mM H2O2 treatment, a lower concentration of 30 mM H2O2 was used for the immunoprecipitation assay, which would represent a physiologically strong oxidative stress. Cotransfected cell lysates with or without 30 mM H2O2 treatment were immunoprecipitated by anti-HA antibody and detected by Western blotting using anti-His antibody. From Fig. 3C
, a SENP1 dimer-size band corresponding to (HA-tagged SENP1)-(His-tagged SENP1) dimer was detected in cells stimulated with H2O2 (Lane 2), indicating that in vivo SENP1 homodimerization occurs in response to oxidative stress.
Interaction of SENP1 and SUMO-1 on oxidation
The crystal structures of SENP2/SUMO and SENP1/SUMO complexes reveal that SUMO proteases interact strongly with SUMO to facilitate the hydrolytic reaction (22
, 23
, 36)
. We questioned whether the noncovalent interaction of SUMO and SENP1C is affected by their dimerization under oxidation as shown in Fig. 2A
. Active-site mutant SENP1C(C603S) was also studied as it bound to SUMO-1 tightly (23)
but did not display strong dimerization following H2O2 treatment. In the interaction assay, N-terminal His-tagged SUMO-1 was firstly immobilized onto nickel resin and then incubated with SENP1C or SENP1C(C603S) in a buffer with or without 10 mM H2O2 for 30 min. To examine both the total and multimeric bound SENP1C, the pull-down complexes were analyzed by nonreducing (Fig. 4
A) or reducing (Fig. 4B
) SDS-PAGE. By comparing the nonspecific binding of SENP1C or SENP1C(C603S) with nickel resin (Lanes 3–4 and 7–8 of Fig. 4
), both monomeric SENP1C and SENP1(C603S) were pull-downed, but less SENP1C was bound to the SUMO under oxidation (Lane 1 of Fig. 4
). Moreover, no dimeric SENP1C was detected. Taken together with observation of the small amount of pull-downed dimeric SENP1C(603S) (Lane 6 of Fig. 4
), the result indicates that formation of disulfide linkage at catalytic Cys 603 hinders the SUMO binding surface.
|
Dimer formation of other SUMO proteases
To further understand if disulfide-linked dimer formation is ubiquitous to other SUMO proteases, the oxidative effects on the catalytic domains of three well-characterized SUMO proteases, human SENP1 and SENP2, and yeast Ulp1 were examined. Different SUMO proteases at a concentration of 0.5 µg/µl were incubated with various concentrations of H2O2 from 0.02 mM to 20 mM for 30 min, and the reactions were terminated by addition of DTT-free SDS loading dye prior to SDS-PAGE analysis.
From Fig. 5
A, dimeric SENP1C was detected at all concentrations of H2O2, even under the H2O2-free condition. The extent of dimerization and oligomerization was found to be maximal in response to treatment with 0.8 mM or higher H2O2. Though the human SUMO protease SENP2 shares the highest sequence identity with SENP1 (22)
, neither SENP2C dimer nor oligomer was observed (Fig. 5B
) while yeast Ulp1C, which does not contain any corresponding Cys 613 for disulfide linkage in SENP1C, dimerized under oxidizing condition (Fig. 5C
). However, it is noted that less Ulp1C dimer was detected, in particular on exposure to 20 mM H2O2. Notably, 20 mM H2O2 treatment induced a band shift of the monomeric proteases. This phenomenon is consistent with previous studies that the thiol group of cysteine can be irreversibly modified to SO2 or SO3 on high oxidative stress (23
, 38)
.
|
To study the recovery ability of these modified proteases, DTT was added to the incubation mixtures to reverse the oxidation effect. As both SENP1C and Ulp1C dimerized in 4 mM H2O2, the three SUMO proteases treated with 4 mM H2O2 were further incubated with 4 mM DTT. The results showed that all dimer or oligomer dissociated to the monomeric form on reduction (Lane 7 in Fig. 5
).
Crystal structures of reversible and irreversible oxidized forms of Ulp1C
The in vitro assays described have demonstrated that inhibition of SUMO proteases induced by H2O2 treatment is via reversible or irreversible modification of the active-site cysteine residue. To obtain more insight into the structural basis of the redox regulation, we attempted to crystallize the proteases with or without H2O2. Crystallization of SENP1 was unsuccessful, instead Ulp1 crystals were obtained after 3 days in the presence of 100 mM H2O2 in crystallization buffer [30%v/v pentaerythritol ethoxylate (15/4 EO/OH) (pH 6.5) and 0.05 M ammonium sulfate]. In the absence of H2O2, crystals were obtained after 10 days in another crystallization condition [20% (w/v) PEG 3350 (pH 6.5) and 0.2 M NaCl], and a data set was collected after 4 wk of crystallization. The phase was determined by molecular replacement using Ulp1 (31)
(PDB ID: 1EUV) as a search model. During model building and refinement, extra electron densities within 1.8 Å connecting to the active site Cys 580 S atoms in the Fo–Fc maps were clearly noted in the two structures. Modified cysteinesulfonic or cysteinsulfinic acid was then modeled into the structures from crystals grown with or without H2O2, respectively (Fig. 6
A, B). The oxidized Cys 580-SO2 found in the latter structure was unexpected, and we hypothesized that oxidation by atmospheric oxygen might take place when the crystal was left for 4 wk before data collection. Another set of data was, therefore, collected 2 wk after crystallization, and the crystal was postsoaked in 50 mM DTT for 48 h before data collection. Surprisingly, a positive density peak with contour level >5
in proximity to the Cys 580 S atom was observed in the Fo–Fc map, and the cysteine was modified to cysteinesulfenic acid (Fig. 6C
). The statistics of the X-ray data and refinement are summarized in Table 1
.
|
The crystal packing of the three structures is very similar. The N
atom of Lys 602 (Lys 602sym) of a molecular symmetry is 2.70 and 3.84 Å apart from the O
1 of the Cys 580-SOH and O2 of Cys 580-SO2 modified structures, respectively (Fig. 6B, C
). The side chain of Lys 602sym in the -SO3 structure is disordered, and it is not clear whether there is any close intermolecular interaction with Cys 580-SO3. From the structural alignment using DALI (39)
, all three crystal structures share the same secondary structural elements as described previously in Ulp1. In the latter two structures, the oxygen atoms of the modified cysteines are stabilized by the surrounding residues, the O2 and O3 of the Cys580-SO3 are hydrogen-bonded to the N
of His 514 and of Trp 448, respectively (Fig. 6A
), whereas the O1 of the Cys580-SO2 is hydrogen-bonded to the N of His 514 (Fig. 6B
). The S atoms of both modified cysteines are far apart from the N of His 514 in the catalytic triad (Cys 580, His 514, and Asp 531). Comparison with the modified cysteinsulfenic acid, the O1 atom is not in close proximity to any atom, instead the S of Cys-SOH is 3.61 Å away from the N of His 514 (Fig. 6C
). The difference in hydrogen bonding and stabilization, however, does not alter the C
of the modified cysteine residues (Fig. 6D
). The main chain atoms in the hydrolytic pocket align well among the three structures. Superimposition of the structures with cysteinesulfenic and cysteinesulfonic acids further reveals the shift of Trp 448 and His 514 as described. Inspection of the electron density maps of two other cysteine residues, Cys 571 positioned on a surface loop and Cys 585 on an interior helix (Fig. 6E, F
), reveals that Cys 580 was the only oxidized residue.
Local conformational rearrangements of Ulp1
Structure superimposition of the oxidized Ulp structures and the Ulp1/SMT3 complex (25)
revealed that reversible or irreversible modification of Cys 580 did not induce any significant structural changes (Fig. 7
A, B). However, orientation differences of Trp 448 and His 514 are remarkable. Previously, we proposed that the active-site cysteine residue and the histidine of the catalytic triad in human SUMO proteases need a conformational change before the cleavage reaction (23)
. Specifically, rearrangement of Trp 465, His 533, and Trp 534 (equivalent residues Trp 448, His 514, and Trp 515 in Ulp1, respectively), is required. In the present study, the distance between S atom of Cys 580 and N of His 514 in the -SOH and Ulp/SMT3 structures are similar (3.61 and 3.71 Å, respectively), suggesting that His 514 in the catalytic triad of Ulp1 may not undergo local rearrangement as shown in the SENP2/SUMO-1 complex (22)
. Instead, His 514 may facilitate Cys 580 in the apoenzyme to form a strong nucleophile, thiolate anion (Cys-S-), which is readily oxidized. Besides, the three oxidized Ulp1 structures revealed that alteration of Trp 448 blocks the hydrophobic tunnel for the C-terminal polypeptide of SMT-3. It is not clear if the shift of Trp 448 is due to attraction of the O atoms of oxidized Cys 580, or the apoenzyme of Ulp1 does exist in a close form.
|
Redox regulation of SENP1C, SENP2C, and Ulp1C
To further understand the redox regulation of SUMO proteases, an attempt was made to examine the recovery ability to assess how these enzymes tolerate oxidative stress. SUMO proteases were first treated with 4 or 20 mM H2O2 for 30 min and then recovered by treatment with 4 or 20 mM DTT, respectively, for another 30 min. The deconjugation activities of untreated, oxidized, and recovered SUMO proteases toward SUMO-TDG were studied. To distinguish the differences, enzyme dosage and reaction incubation time were adjusted so that the deconjugation reaction reached a near, but not total, completion. As a control experiment, the activities of 4 or 20 mM DTT-treated SENP2C and Ulp1C were studied, their hydrolytic activities were comparable to those of untreated protease (Fig. 8
B).
|
From Fig. 8
, SENP1C displayed the highest recovery ability among the three proteases. Estimation of band intensity indicated that 90% and 50% of substrate turnover by SENP1C after treatment with 4 mM and 20 mM H2O2, compared with, respectively, 80% and 30% substrate turnover by yeast Ulp1C. However, only 20% of substrate turnover by SENP2C was observed following 4 mM H2O2 treatment, and the enzyme was almost completely irreversibly inhibited after exposure to 20 mM H2O2. Notably, the order of the recovery ability of the three proteases is dependent on the ability of dimerization (Fig. 5)
. This strongly indicates that the reversible disulfide linkage of SUMO proteases protects the enzymes from irreversible inhibition under oxidizing conditions. Although the dimerization-defective mutant SENP1C(C613S) possesses normal hydrolytic activity, it exhibits a very low recovery ability comparable to that of SENP2. This result further supports our proposed redox model that intermolecular disulfide bond formation of SUMO proteases protects the enzyme from irreversible oxidization and inhibition.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. A variety of proteins including peroxiredoxin, transcription factors OxR and Fos, and protein tryrosine phosphatase PTP1B have been shown to be thiol-based regulatory switches. Among all known examples, conformational changes on oxidation of the cysteine residues are the common underlying mechanism. Recently, several lines of investigation on sumoylation have implicated oxidation stress induces up-regulation of the modification (3
, 4)
and induces reversible oxidative inhibition of E1 E2 SUMO ligases (24)
. As all SUMO conjugating and deconjugating enzymes possess catalytic cysteines, we raise questions on whether the SUMO modification adopts a similar redox regulatory mechanism. In the present study, we study the molecular basis of the redox regulation of SUMO proteases from human and yeast. Both functional and structural analyses reveal that oxidation leads to modification of the catalytic cysteine residue of SUMO proteases from the reversible form (Cys-SOH) to the irreversible forms (Cys-SO2 and Cys-SO3) and resulting in loss of protease activity. Results of our functional assays agree with those reported by Bossis and Melchior in 2006 (24)
, that a higher H2O2 concentration (10 mM) is required to suppress the activity of SENP1 dramatically. Here, we further elucidate the mechanism behind the inhibition. Our results demonstrate that modified Cys-SOH rapidly undergoes dehydration with Cys-SH of another protease molecule to form an intermolecular disulfide bridge. Such monomer-dimer transformation is shown in both in vitro and in vivo assays. Cross-linkage of the SUMO proteases can maintain the active cysteines and prevent them from irreversible oxidation to Cys-SO2 and Cys-SO3. Identification of catalytic Cys 603 and Cys 613 for the cross-linkage that inactivates SENP1C and SUMO interaction provides a deeper understanding of the observed results. Interestingly, the three SUMO proteases SENP1, SENP2, and Ulp1 exhibited different degrees of dimer/oligomer formation and recovery abilities. It is possible that desumoylation by SUMO proteases is under different redox modifications. Absence of any disulfide linkage in the SENP2 suggests an intricate but specific regulatory mechanism of the availability of SUMOs and desumoylation in response to oxidative stress.
The three crystal structures of Ulp1 with catalytic Cys 580 oxidized to cysteinesulfenic, cysteinesulfinic and cysteinesulfonic acids not only validate the reversible and irreversible enzyme inhibition on different degrees of oxidation as observed in the in vitro assays but also provide a structural basis for the redox regulation. Due to the difficulty in reproducibly crystallizing the Ulp1 crystals in the absence of an oxidant, attempts to set up a time course H2O2 soaking analysis to reveal the transition from -SH to -SOH, -SO2, and -SO3 states were unsuccessful. However, it is interesting to recognize that Cys 580 of Ulp1 protease grown without H2O2 is oxidized to Cys-SOH and then to Cys-SO2 when incubated longer in the crystallization buffer. It appears that Ulp1 protease is extremely sensitive to mild oxidation, even at the atmospheric oxygen level. We anticipated that soaking the crystal in 50 mM DTT would reduce the sulfenic group to thiol group. However, cysteinesulfenic acid is in a stabilized form, and this contradicts with the DTT recovery assay shown in the in vitro experiments. From the crystal structure, we speculate that intermolecular hydrogen bond between O1 of Cys 580 and N of Lys 602sym might prevent the DTT reduction reaction. From Fig. 5
, interaction with Lys 602sym is a consequence of the oxidation that modification from Cys580-SH to -SOH attracts the side chain of the Lys 602sym to the active-site, while further oxidation to -SO3 would repel Lys 602sym due to orientation of the -SO3 group and its interaction with Trp 448 and His 514, respectively.
As a cysteine protease, the catalytic triad comprises the active residue Cys 580 of Ulp1 coordinated by His 514, which is in turn stabilized by Asp531. The presence of the basic His 514 residue promotes the formation of a strong nucleophile, thiolate anion (Cys-S–), which is readily oxidized. Taken into account the pKa value of 8.5 of cysteine residue and the microenvironment surrounding another two cysteines, Cys 571 and Cys 585, we can then explain why only Cys 580 in the Ulp1 is oxidized in the crystallization buffer (pH 6.5) (Fig. 6E, F
). The three structures determined reveal that the apoenzymes are already in active forms. This is dissimilar from many other cysteine proteases such as HAUSP, a member of the UBP family of ubiquitin-specific proteases (40)
, which requires substrate binding to align the active cysteine to the catalytic histidine for deprontonation. This might explain why only little structural and functional information of oxidized cysteine proteases was reported to date (41)
.
In summary, in addition to the identification of the disulfide-bridge of the active-site cysteines of E1 and E2 that reversibly regulate the sumoylation (24)
, we demonstrate the SUMO proteases may serve as redox sensor and effector that undergo both reversible and irreversible modification on exposure to various degrees of oxidative stress. Unlike most of the redox-regulatory switches identified, the reversible modification of SUMO proteases is at the intermolecular level. Noticeably, dimerization would bury the SUMO binding interface and reversibly inhibit the enzyme activity, whereas oxidation to -SO2 and -SO3 causes an irreversibly inhibitory effect. Moreover, considering the different recovery abilities of SUMO proteases SENP1, SENP2, and Ulp1 and their distinct abilities of dimer formation under oxidative stress, we hypothesize that the reversible disulfide cross-linkage is conducive to the enzyme stability during oxidizing conditions, and the functions of the proteases can be restored rapidly. It is anticipated that the seven human SUMO proteases identified to date, together with their specific tissue distribution and substrate specificities, would generate a diverse but specific intracellular redox response.
|
ACKNOWLEDGMENTS |
|---|
Received for publication December 10, 2006. Accepted for publication July 19, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. T. H. Yeh SUMOylation and De-SUMOylation: Wrestling with Life's Processes J. Biol. Chem., March 27, 2009; 284(13): 8223 - 8227. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
