Biochemical Journal

Research article

SUMO2 and SUMO3 transcription is differentially regulated by oxidative stress in an Sp1-dependent manner

Jing Sang, Kai Yang, Yueping Sun, Yan Han, Hui Cang, Yuying Chen, Guiying Shi, Kangmin Wang, Jie Zhou, Xiangrui Wang, Jing Yi


Protein SUMOylation (SUMO is small ubiquitin-related modifier) is a dynamic process that is strictly regulated under physiological and pathological conditions. However, little is known about how various intra- or extra-cellular stimuli regulate expression levels of components in the SUMO system. SUMO isoforms SUMO2 and SUMO3 can rapidly convert to be conjugated in response to a variety of cellular stresses. Owing to the limitations of sequence homology, SUMO2 and SUMO3 cannot be differentiated between and are thus referred to as SUMO2/3. Whether these two isoforms are regulated in distinct manners has never been addressed. In the present paper we report that the expression of SUMO3, but not SUMO2, can be down-regulated at the transcription level by cellular oxidative stress. In the present study, we checked SUMO2 and SUMO3 mRNA levels in cells exposed to various doses of H2O2 and in cells bearing different levels of ROS (reactive oxygen species). We found an inverse relationship between SUMO3 transcription and ROS levels. We characterized a promoter region specific for the mouse Sumo3 gene that is bound by the redox-sensitive transcription factor Sp1 (specificity protein 1) and demonstrated oxidation of Sp1, as well as suppression of Sp1–DNA binding upon oxidative stress. This revealed for the first time that the expression of SUMO2 and SUMO3 is regulated differently by ROS. These findings may enhance our understanding about the regulation of SUMOylation and also shed light on the functions of Sp1.

  • promoter analysis
  • reactive oxygen species (ROS)
  • small ubiquitin-related modifier 2/3 (SUMO2/3)
  • specificity protein 1 (Sp1)
  • transcription


SUMOylation (SUMO is small ubiquitin-like modifier) is a post-translational modification in which SUMOs are covalently conjugated to target proteins [1]. The mammalian SUMO family mainly includes three isoforms: SUMO1, SUMO2 and SUMO3. The SUMO2 and SUMO3 genes share only approximately 46% sequence similarity to SUMO1, but they share 96% similarity to each other. Thus SUMO2 and SUMO3 cannot be distinguished from each other using antibodies and are referred to as SUMO2/3 in the nomenclature [2,3]. In contrast with SUMO1, SUMO2 and SUMO3 harbour a consensus SUMOylation site at their N-terminal region and distribute to similar subcellular locations. In addition, SUMO1 is commonly conjugated to substrates, whereas SUMO2 and SUMO3 appear to be unconjugated and rapidly convert to be conjugated in response to a variety of cellular stresses [2,49]. Collectively, owing to the similarity in sequence, conjugation dynamics and other characteristics, SUMO2 and SUMO3 have been considered to be more or less functionally redundant. In recent years, by using strategies of overexpression, silencing and rescue with SUMO2 or SUMO3 respectively, several research groups have reported that these two protein isoforms are not mutually replaceable [1013]; this implies that they might be different in terms of substrate and function. However, SUMO2 and SUMO3 are still viewed indiscriminately in most SUMOylation reports to date. It remains unknown whether these two isoforms have independent modes of regulation.

Although a growing number of studies have demonstrated new SUMOylated proteins and related biological events [1417], the regulation of SUMOylation is emerging as a challenging topic [4,6,18]. SUMOs, especially SUMO2/3, after being expressed and processed, can remain free or can be conjugated to targets [2,9,19,20]. Conjugation of SUMOs occurs through a cascade of reactions that are performed by an activating enzyme (E1), a conjugating enzyme (E2) and a SUMO ligase (E3). Many studies have revealed how the components of the enzyme systems necessary for SUMO conjugation are activated under certain physiological or pathological circumstances [6,16,21,22]. Other studies have demonstrated how the reverse process, de-SUMOylation, is regulated by cellular events [15,17,2326]. Conjugation/deconjugation balance is apparently the focus of regulation. Nevertheless, it has seldom been addressed how the expression of SUMO proteins can be subjected to regulation in response to various environmental conditions [4,27,28].

Oxidative stress is one of the most common environmental insults, because ROS (reactive oxygen species), the by-products of cellular metabolism, are excessively generated or inadequately scavenged upon a variety of intra- or extra-cellular stressors' actions. The stressors may include thermal and nutritional variations, hypoxia and reoxygenation in tissues, oncogene activation, protein overload in the endoplasmic reticulum, DNA damage, and drug or radiation exposure, among others [29,30]. Previous research, including our own, has shown that SUMOylation could be regulated by ROS at the level of conjugation or deconjugation. In the present paper we report a previously unknown phenomenon in which ROS can down-regulate the expression of SUMO3, but not SUMO2, at the transcription level [15,17,22,25,31]. In the present study, we characterized a promoter region specific for the SUMO3 gene that is associated by the redox-sensitive transcription factor Sp1 (specificity protein 1) and demonstrated oxidation of Sp1 and suppression of Sp1–DNA binding under cellular oxidative stress. The results of the present study thereby explain why expression of SUMO2 and SUMO3 are distinctively regulated by ROS.


Cell lines, cell culture and treatments

The OSC (oral squamous carcinoma) cell line stably expressing antisense MnSOD [manganese SOD (superoxide dismutase)], i.e. OSC-AS-SOD cells, and its parental cell line OSC were gifts from Dr T. Yamamoto (Kochi University, Kochi, Japan) [32,33]. These were cultured in DMEM (Dulbecco's modified Eagle's medium; Gibco). As overexpression of Nox1, a homologue of the catalytic subunit of the NADPH oxidase of phagocytes gp91phox, increased superoxide generation in NIH 3T3 cells [34], an NIH 3T3-Nox1 cell line was established using the following procedures. The pcDNA-Nox1 plasmid was constructed according to methods described previously in the literature [34] and introduced into NIH 3T3 cells by electroporation (Bio-Rad Laboratories). The positive clones were selected by G418 (0.4 mg/ml) and amplified for 10 days, and then maintained in G418 (0.3 mg/ml) in DMEM. Nox1 expression was validated by RT (reverse transcription)–PCR, and the level of ROS was determined. NIH 3T3 cells were cultured in DMEM. L-02 cells and a non-tumour human liver cell line [35,36] was cultured in RPMI 1640 medium (Gibco). HepG2 cells, a perpetual cell line derived from a human hepatocarcinoma [37], were cultured in a 1:1 mixture of DMEM and RPMI 1640 medium. All media were supplemented with 10% newborn calf serum (Biochrom). Cells were grown at 37 °C in a humidified atmosphere with 5% CO2.

To examine the association with ROS, cells were treated with H2O2 and NAC (N-acetylcysteine; Sigma–Aldrich), when needed.

In situ hybridization for SUMO mRNA

Liver specimens (ten) and hepatocarcinoma tissues (20) were derived from autopsic and pathological tissues archived in Ren Ji Hospital and were processed as paraformaldehyde-fixed and paraffin-embedded specimens. The specimens were collected following an Institute-approved protocol. The sequences of the digoxigenin-labelled single-stranded RNA probes for SUMO2 and SUMO3 were 5′-TTGATTGGTTGCCCGTCAAATCGG-3′ and 5′-AATGGAGCCTCAGGGAAGGACAAT-3′ respectively. After hybridization for 24 h, the sections were incubated with an alkaline-phosphatase-conjugated antibody against digoxigenin at room temperature (15 °C) for 3 h and washed with 0.5 M TBS (Tris-buffered saline; 0.5 M Tris base, 150 mM NaCl and 2 mM KCl, pH 7.4) and 0.01 M TBS (pH 9.5). The hybridization signal was visualized by BCIP (5-bromo-4-chloroindol-3-yl phosphate)/NBT (Nitro Blue Tetrazolium) (blue/purple signal). Sections were counterstained with Nuclear Fast Red before examination under an Axioplan 2 fluorescence microscope. The signal intensity in four fields per section was quantified by Zeiss KS400 Version 2.2 software.

Measurement of superoxide in tissues

Normal liver (two) and hepatocarcinoma specimens (two) were freshly collected from surgically resected tissues in Ren Ji Hospital following an Institute-approved protocol and prepared for cryosectioning. Sections (30 μm) were cut. Production of superoxide anions in situ was measured using the oxidative fluorescent dye DHE (dihydroethidium; Beyotime, Shanghai, China) according to a previously described method [38]. Briefly, sections were equilibrated in PBS and mounted on the slides. Fresh buffer containing DHE (2 mM) was then applied on to each section and incubated for 30 min in a light-protected humidified chamber at 37 °C. Sections were evaluated for DHE intensity under fluorescence microscopy. The signal intensity in three fields per section was quantified by Zeiss KS400 Version 2.2 software.

RT–PCR and real-time Q-PCR (quantitative PCR)

RT–PCR and Q-PCR were carried out using standard procedures. After denaturation at 95 °C for 5 min, 28 cycles were performed using the following PCR programmes: 95 °C for 30 s, 56 °C for 1 min and 72 °C for 1.5 min (for Nox1); 95 °C for 30 s, 56 °C for 30 s and 72 °C for 30 s (for SUMO2); or 95 °C for 30 s, 57 °C for 30 s and 72 °C for 30 s (for SUMO3); all were followed by extension at 72 °C for 10 min. Q-PCR was performed on the ABI Prism 7300 system (Applied Biosystems) using SYBR Green and following the manufacturer's instructions. The primers used are indicated in Table 1.

View this table:
Table 1 Primers used for constructs, PCR and ChIP

The nucleotides in italic were mutated, and the underlined bold sequences represented the deleted nucleotides in the Sumo3 promoter luciferase (−1448/+31) construct. The underlined sequences (not in bold) are BamHI sites.

ROS detection

DCFH-DA (2′,7′-dichlorodihydrofluorescein diacetate; Sigma) was used as a ROS-capturing reagent using a method described previously [39].

Plasmid constructs and transfection

The mouse Sumo3 promoter (−1448/+31) luciferase reporter was generated by insertion of nucleotides −1448 to +31 of the mouse Sumo3 promoter into the KpnI and SacI sites of the pGL3-basic vector; this is designated SU3 promoter luc (SUMO3 promoter luciferase reporter). Plasmids of the pN3 vector and pN3-Sp1FL-complete were gifts from Professor G. Suske (Philipps-University of Marburg, Marburg, Germany) [40,41]. The plasmid constructs were transiently transfected or co-transfected into cells using Lipofectamine™ 2000 (Invitrogen) according to the manufacturer's instructions.

Deletion and site-directed mutagenesis

Deletion and site-directed mutagenesis in the SU3 promoter luc were performed using the QuikChange® II mutagenesis kit (Stratagene) to generate the constructs of Sp1-I mutation, Sp1-II mutation, USF (upstream stimulatory factor) mutation, Sp1-I deletion and Sp1-II deletion. The Sumo3 promoter (−1448/+31) was used as a template and PCR was conducted using Pfu DNA polymerase. DpnI was applied to remove the DNA template in the PCR product.

Luciferase reporter assay

Cells transfected with dual-luciferase reporters were incubated with 100 μl of 1× passive lysis buffer (from the Promega luciferase assay kit) for 15 min and harvested. Firefly and sea pansy luciferase activities were determined according to the instructions in the Promega Luciferase Assay kit. All assays were repeated at least three times

ChIP (chromatin immunoprecipitation) assays

Protein–DNA was cross-linked in 1% formaldehyde for 10 min at 37 °C before the reaction was quenched using glycine (0.125 M) for 5 min at room temperature. Cells were then harvested and ChIP was performed using a kit (Upstate Biotechnology) according to the manufacturer's instructions. The sonicated chromatin samples were precipitated with anti-Sp1 (sc-59x, Santa Cruz Biotechnology). ChIP using the normal rabbit IgG in place of the primary antibody or no antibody served as negative controls. PCR was then undertaken to detect Sp1 CpG island DNA. After denaturation at 95 °C for 5 min, 30 cycles were performed using the following PCR programme: 95 °C for 30 s, 56 °C for 30 s, 72 °C for 30 s, followed by extension at 72 °C for 10 min. Inputs were used as the loading control and were obtained by eluting DNA from 5 μl of cell lysates prior to the immunoprecipitation step.

In order to distinguish the SP1-bound ectopically expressed Sumo3 promoter elements from endogenous ones, a single BamHI restriction site (GGATCC) was introduced into the plasmids for wild-type as well as Sp1-II-mutated (Sp1-II mutation) or -deleted (Sp1-II deletion) Sumo3 promoters at −322/−321 via mutagenesis. Consequently, after protein–DNA co-immunoprecipitation and PCR, only exogenous Sumo3 promoters could be digested by BamHI and detected as two products of 170 bp and 250 bp. In contrast, endogenous Sumo3 promoter would appear as one 420 bp band. Cells were then transfected with these BamHI site-including plasmids using FuGENE™ HD transfection reagent (Roche). After 48 h, ChIP was performed as described above, and PCR was conducted using a new forward primer to include the BamHI restriction site in both input and ChIP products. BamHI enzyme and NEB buffer 3 were directly added to the PCR products and incubated at 37 °C for 30 min. Finally, the products that were composed of endogenous and exogenous (BamHI cut) elements were loaded on to the agarose gel.


Immunoblotting was performed using routine methods. Rabbit polyclonal anti-Sp1 antibody (Santa Cruz Biotechnology) and mouse monoclonal anti-β-actin antibody (Sigma–Aldrich) were used.


Cell monolayers were fixed with 4% paraformaldehyde, permeablized with 0.2% Triton X-100, and blocked with 5% BSA before incubation with anti-Sp1 antibody at 4 °C overnight. The FITC-conjugated second antibody (Dako) was then used. Cells were examined under an LSM 510 fluorescence microscope (Zeiss). Sp1 fluorescence intensity was quantitatively analysed in 50 cells using Zeiss KS400 Version 2.2 software.

Analysis of Sp1 oxidation by F5M (fluorescein-5-maleimide)

Thiol reduction of Sp1 was assessed using a method described previously [39]. NIH 3T3 cells were incubated in fresh medium containing 10 mM F5M for 30 min at 37 °C after H2O2 treatment. The cells were lysed with ice-cold 1× immunoprecipitation buffer (20 mM Tris/HCl, 150 mM NaCl, 1% Triton X-100, and sodium pyrophosphate, 2-glycerophosphate, EDTA, sodium orthovanadate and leupeptin; Beyotime) containing 1 mM F5M at 4 °C for 30 min, and then lysates were centrifuged (16000 g) at 4 °C for 15 min. The supernatant was incubated with the antibody against Sp1 overnight at 4 °C and the Protein A/G beads (Merck) for another 4 h. Sp1 was then separated from the antibody–Protein A/G bead complex using 1×SDS sample buffer with 0.5 mM F5M. After the denatured proteins were subjected to SDS/PAGE, the F5M fluorescence of the Sp1 bands on the fresh gel were visualized and photographed using the FLA-5000 Imaging System (Fujifilm) at an excitation wavelength of 472 nm. An immunoblot was performed to assess total Sp1 in the identical positions blotted from the same gel.

Statistical analysis

Results presented were derived from at least three independent experiments. Means±S.E.M. were calculated, and groups of data were compared using Student's t test.


SUMO3 transcriptional levels are lower in cells that have higher ROS levels, which is distinct from SUMO2

As expression differences between SUMO2 and SUMO3 have been alluded to [by SAGE (serial analysis of gene expression) at the Genecards website (] in cancer tissues, we first examined the transcriptional levels of these two genes in 20 hepatoma and ten normal liver autopsy samples using in situ hybridization. In general, the hybridization signal for SUMO2 mRNA was similar in normal liver tissues as compared with the cancerous tissues. By contrast, the positive signal for SUMO3 mRNA was remarkably lower in cancerous tissues than in the normal liver (Figure 1A, upper panels). Quantitative image analysis demonstrated that the average intensities of SUMO3 were significantly weaker in the hepatoma samples (Figure 1A, lower panel).

Figure 1 SUMO3 transcriptional levels are lower in cells that have higher ROS levels, which is distinct from SUMO2

(A) Detection of SUMO2 and SUMO3 transcriptional levels in human hepatoma and normal liver tissues by in situ hybridization. Scale bar=30 μm (upper panels). The relative signal intensity in four fields per section from each specimen was analysed (lower panel). (BD) Detection of SUMO2 and SUMO3 transcriptional levels by RT–PCR in HepG2 compared with L-02 cells, OSC-SA-SOD compared with OSC cells, and NIH3T3-Nox1 compared with NIH 3T3 cells (upper panels). Detection of basal ROS levels in cells by flow cytometry after cells were incubated with DCFH-DA. Relative fluorescence intensity represented the relative levels of ROS (lower panels). (E) Detection of superoxide in human hepatoma and normal liver tissues by DHE staining. Scale bar=100 μm (upper panels). The relative signal intensity in three fields per section from each specimen was analysed (lower panel).

We then attempted to verify this phenomenon in a pair of cell lines: hepatoma-derived HepG2 and non-tumour hepatocyte-derived L-02 cells. The RT–PCR results showed that SUMO3 mRNA was markedly reduced in HepG2 cells as compared with that in L0–2 cells, whereas SUMO2 mRNA levels remained similar between the two cell lines (Figure 1B). As we knew that HepG2 cells possessed a higher basal level of ROS than L-02 cells based on our previous work [41a], we suspected that a reduced SUMO3 transcription level might be correlated with an increased ROS level in the cells.

Next, two pairs of cell lines with a similar genetic background but different ROS levels were utilized. The first pair of cell lines were OSC-AS-SOD, which represented a human oral carcinoma cell line with knocked-down SOD and increased ROS, and the parental OSC cells. The second pair of cell lines consisted of NIH 3T3 cells and NIH 3T3-Nox1 cells that stably expressed Nox1, the major subunit of NADPH oxidase, and thus exhibited increased ROS levels. In agreement with the results in the hepatocyte pair of cell lines, in both of the OSC and NIH 3T3 pairs, SUMO3 mRNA levels were markedly lower in cells possessing higher ROS levels. Notably, SUMO2 mRNA levels remained similar, regardless of the differences in ROS levels (Figures 1B–1D). To ascertain this inverse relationship between SUMO3 transcription and ROS levels, we went back to assess the redox states of the hepatoma and normal liver autopsy samples by measurement of DHE staining intensity, which reflects the levels of superoxide in tissues. The results showed that the hepatoma samples with lower SUMO3 expression levels were strongly prone to an oxidized state as compared with normal liver tissue (Figure 1E). Collectively, these data indicate that the transcription of SUMO3, but not SUMO2, may be inversely affected by ROS.

Transcriptional levels of SUMO3, but not SUMO2, are down-regulated by ROS

To mimic the long-term effects of oxidative stress on SUMO gene expression occurring in cell lines and tissues, we exposed NIH 3T3 cells to H2O2 at a concentration of 100 μM every day for 3 days and analysed their SUMO2 and SUMO3 mRNA levels using Q-PCR. We found that SUMO3 transcripts were reduced according to the time of H2O2 treatment, showing more dramatic changes later in time, but that SUMO2 transcription was virtually unchanged (Figure 2A). To further confirm this finding, we examined the transcriptional levels of SUMO3 after exposure to H2O2 at low (100 μM) and higher (500 μM) doses for a short time (6 h), or after pre-treatment with the antioxidant NAC for 4 h. The results showed that SUMO3 transcription was down-regulated by H2O2 treatment in a dose-dependent manner, and that this effect was eliminated in the presence of the antioxidant. These results indicate that ROS selectively down-regulates the transcription of SUMO3 on both long and short time scales. We then applied a high dose (500 μM) of H2O2 for a short time (6 h) treatment as the condition of oxidative stress in the following biochemical experiments using mouse fibroblast NIH 3T3 cells, in which the SUMO3 expression level had reproducibly reduced significantly by approximately 40%.

Figure 2 Transcriptional levels of SUMO3, but not SUMO2, are down-regulated by ROS

(A) Detection of SUMO2 and SUMO3 transcriptional levels by Q-PCR in NIH 3T3 cells exposed to H2O2 for 1–3 days. Mouse Gapdh (glyceraldehyde-3-phosphate dehydrogenase) was used as an internal control. (B) Detection of SUMO2 and SUMO3 transcriptional levels by Q-PCR in NIH 3T3 cells exposed to H2O2 at 100 μM and 500 μM for 6 h. Cells were pre-treated with NAC for 4 h when indicated. Mouse Gapdh was used as an internal control.

The Sp1-binding site is responsible for basal and ROS-regulated Sumo3 promoter activity

To understand the mechanism underlying specific SUMO3 regulation by ROS, we sought to determine the characteristics of the promoters in the two genes. Promoter sequences of SUMO2 and SUMO3 were obtained from the NCBI (National Center for Biotechnology Information) and were compared using Promoter Scan software. We found that, although these two genes had a high homology in their encoding sequences, their promoter sequences differed greatly. SUMO3 had a TATA-less promoter and, according to bioinformatics predictions, the putative binding sites for two transcriptional factors known to be sensitive to oxidative stress, i.e. Sp1 and USF [42], were present (Figure 3A); SUMO2 lacked these features (see Supplementary Figure S1 at We then began to construct a promoter element (−1448/+31) fused with a luciferase reporter for the mouse Sumo3 gene (Figure 3A). The promoter activity of the Sumo3 of this construct was up to 15-fold higher than that of the parental promoter-less construct (pGL3-Basic) in the NIH 3T3 cells (Figures 3B and 3C, compare the upper two bars). This promoter reporter construct was therefore usable for testing activity changes of the Sumo3 promoter under basal and oxidative stress conditions.

Figure 3 The Sp1-binding site is responsible for basal and ROS-regulated Sumo3 promoter activity

(A) The sequence of the mouse Sumo3 promoter. Numbers are relative to the start site ATG (+1). Specific protein-binding sites are underlined. (B) Assays for relative activity of Sumo3 promoter luciferase constructs containing a series of mutations of the putative transcription-factor-binding sites. (C) Assays for relative activity of Sumo3 promoter luciferase constructs containing a series of deletions of the putative transcription-factor-binding sites. (D) Assays for relative activity of Sumo3 promoter luciferase constructs after cells were overexpressed with Sp1 for 48 h. (E) ChIP assay to test Sp1 binding with endogenous and ectopically overexpressed Sumo3 promoters. NIH 3T3 cells were transfected with wild-type or Sp1-II-mutated or -deleted Sumo3 promoter constructs that were introduced with a BamHI restriction enzyme site for 48 h. The band of 420 bp represented endogenous Sumo3 promoters, whereas the bands of 170 bp and 250 bp represented exogenous Sumo3 promoter constructs. Samples precipitated using non-immune IgG or no antibody (NA) were used as negative controls. (F) Assays for relative activity of Sumo3 promoter luciferase constructs after cells were treated with or without 100 μM H2O2 for the indicated time (left-hand panel), or treated with or without the indicated doses of H2O2 for 6 h. Cells were pre-treated with NAC for 4 h when indicated (right-hand panel). (G) Assays for relative activity of wild-type and Sp1-II-mutated or -deleted Sumo3 promoter luciferase constructs after cells were treated with the indicated doses of H2O2 for 6 h. WT, wild-type.

To identify the regions and the corresponding transcription factors of the promoter mediating responses to oxidative stress, site-directed mutation analysis of these sites was performed before the various mutated promoter reporters were transfected into cells. The reporter with a mutation at the second putative binding site for Sp1 (Sp1-II) showed robustly decreased (10-fold) activity; another putative binding site for Sp1 (Sp1-I), as well as the USF site, were irrelevant (Figure 3B).

To confirm this Sp1-binding site, we performed binding-site-deletion analysis. The results revealed that loss of the regions spanning the USF-binding site and Sp1-I had no substantial influence on Sumo3 promoter activity. In contrast, removal of six nucleotides from Sp1-II eliminated basal Sumo3 promoter activity (Figure 3C). This suggests that the sequence of −91/−86 is essential for regulatory elements on the Sumo3 promoter. Next, we overexpressed Sp1 in cells co-transfected with Sumo3 reporters. The results showed that overexpressed Sp1 boosted the transcriptional activity of the wild-type Sumo3 promoter, but slightly enhanced that of the Sp1-II-mutated version. It also showed that there was no effect on the Sp1-II-deleted versions (Figure 3D), indicating that Sp1, by binding to the Sp1-II site, is the transcription factor that promotes basal Sumo3 transcription.

To demonstrate binding of Sp1 with wild-type or Sp-II-mutated/deleted promoters, three plasmids with an inserted BamHI restriction enzyme site were constructed to allow us to distinguish ectopically expressed Sumo3 promoters from endogenous promoters (see Supplementary Figure S2 at ChIP was then performed, and positive PCR products spanning the Sp1-II site and the BamHI site were amplified. The PCR results showed that exogenous Sumo3 promoters were effectively expressed since the BamHI-digested 170 bp and 250 bp bands were predominant, as compared with the endogenous 420 bp band (Figure 3E, lower panel). After ChIP, precipitation with normal rabbit serum or without antibody did not yield any positive DNA bands. As expected, Sp1 bound with either the endogenous or exogenous wild-type Sumo3 promoters. However, Sp1 binding with the Sp1-II-mutated version was greatly decreased, and binding with the Sp1-II-deleted version was entirely diminished; binding with the endogenous Sumo3 promoter remained unchanged in these samples (Figure 3E, upper panel). These results confirmed that the Sp1-II site at the −91/−86 sequence is an Sp1-binding site for maintenance of basal Sumo3 transcription.

We then sought to verify that the above-proven element was responsive to ROS. The cells expressing this reporter were exposed to H2O2. As expected, the results of the reporter assay upon H2O2 treatment showed that ROS potently repressed the Sumo3 promoter activity in a time-dependent manner (Figure 3F, upper panel). The repression appeared significant (approximately 40%) at 6 h post-treatment, which was consistent with the Q-PCR data (Figure 2). A recovery was seen at 24 h post-H2O2 exposure, which could be attributed to disappearance of H2O2. The cells were then treated with H2O2 in various doses. Maximum repression was observed at 500 μM H2O2 for 6 h. These responses were redox-specific, as the repression was reversed by application of the antioxidant NAC (Figure 3F, lower panel). After cells transfected with wild-type and Sp-II-mutated/deleted promoters were exposed to 500 μM H2O2 for 6 h, activity of the wild-type Sumo3 promoter was markedly decreased, but activity of the Sp1-mutated or -deleted versions remained unchanged (Figure 3G). These results suggest that the Sp1-binding site on the Sumo3 promoter is responsive to ROS and is required for repressed transcription of Sumo3 under oxidative stress.

Down-regulation of SUMO3 transcription by ROS is caused by attenuation of Sp1–DNA binding ability that might be related to oxidative modification of Sp1

Next, we performed ChIP to determine whether the Sp1 association with the Sumo3 gene promoter at the −91/−86 element was changed under oxidative stress in vivo. Remarkably, H2O2 treatment led to an obvious decrease in Sp1 binding to Sumo promoters and this effect was reversed in the presence of an antioxidant (Figure 4A). These results clearly indicate that the attenuated interaction between Sp1 and its target DNA sequences accounts for the decrease in Sumo3 promoter activity under oxidative stress, while simultaneously indicating that the attenuated Sp1–DNA binding is caused by ROS.

Figure 4 Down-regulation of SUMO3 transcription by ROS is caused by attenuation of Sp1–DNA binding ability that might be related to oxidative modification of Sp1

(A) ChIP assay for the DNA-binding activity of Sp1 on the promoter of Sumo3 in vivo. Samples precipitated using non-immune IgG or no antibody (NA) were used as negative controls. NIH 3T3 cells were treated with H2O2 for 6 h and pre-treated with NAC when indicated before harvest. The ratio of immunoprecipitated promoter elements over total promoter elements was quantified. (B) Detection of the Sp1 mRNA level by RT–PCR after NIH 3T3 cells were treated with H2O2 for 6 h and pre-treated with NAC when indicated. The relative mRNA level was represented by the ratio of grey levels of the Sp1 bands over control bands. (C) Detection of Sp1 protein levels by immunoblotting after NIH 3T3 cells were treated with H2O2 for 6 h and pre-treated with NAC when indicated. The relative protein level was represented by the ratio of grey levels of the Sp1 bands over control bands. (D) Detection of Sp1 nuclear localization by immunofluorescence. Sp1 staining intensity in 50 cells was analysed quantitatively. (E) Assessment of oxidation of Sp1 by F5M. NIH 3T3 cells were treated with H2O2 at various concentrations for 6 h before incubation with reduced thiol-specific-conjugating fluorescence from F5M. Sp1 was immunoprecipitated by anti-Sp1. Sp1 in the reduced form was shown by F5M fluorescent bands (upper panel), and total Sp1 was shown by immunoblotting (lower panel). The relative protein reduction level was represented by the ratio of relative fluorescence intensity of the F5M-stained Sp1 bands over total Sp1 bands.

We then sought to investigate the changes in Sp1 under oxidative stress. The transcription level of Sp1 itself was first examined by RT–PCR. The results suggested that the mRNA of Sp1 was not altered by H2O2 treatment (Figure 4B). Immunoblotting was then performed to compare the protein levels of Sp1 after cells were treated with H2O2 for 6 h; it was found that the protein levels of Sp1 remained the same as untreated samples (Figure 4C). Next, nuclear localization of Sp1 in NIH 3T3 cell was analysed by immunofluorescence; the image analysis showed no difference in the average Sp1 nuclear intensity before and after H2O2 treatment (Figure 4D). These results excluded the possibility of a decrease in the quantity of Sp1 in the nucleus upon oxidative stress.

Sp1 is known for being redox-sensitive [43]. The cysteine residues in Sp1, especially those that form zinc fingers essential for binding with DNA, can be targets for oxidative modification by ROS [43,44]. We suspected that oxidative modification of Sp1 occurred in cells exposed to the applied doses of H2O2. F5M staining was performed to determine the oxidative status of Sp1. As shown in Figure 4(E), the F5M fluorescence of Sp1 decreased significantly with increased doses of H2O2, whereas the total protein levels of Sp1 remained stable. This confirmed that H2O2 treatment could, in fact, induce oxidation of Sp1.

We have thus revealed the mechanism that causes SUMO2 and SUMO3 to be differentially regulated by cellular responses to oxidative stress. Because the promoter region of the SUMO3 gene, not the SUMO2 gene, has a specific Sp1-binding site, Sp1 binding guarantees basal transcription of SUMO3 (Figure 5, upper panel). However, as Sp1 is a redox-sensitive transcription factor, oxidative modification of Sp1 may impair its binding ability with DNA. Hence, ROS can potently down-regulate SUMO3 transcription (Figure 5, lower panel).

Figure 5 Hypothetical illustration of the Sp1-dependent mechanism underlying the transcriptional down-regulation of SUMO3 by cellular oxidative stress


SUMO2 and SUMO3 are two genes located at distinct chromosomes, but their high homology makes characterization of the functional specificity of the two proteins difficult. Hence information regarding the differences between SUMO2 and SUMO3 has been very limited [1013]. The present study provides evidence for the first time that SUMO2 and SUMO3 may have distinct ways to be regulated, supporting the notion that they are not genetically redundant. Despite the fact that the two isoforms can be conjugated to the same spectrum of target proteins and with the same spatial characteristics, the present study indicates that they may be specifically controlled by different cellular signals.

We demonstrate for the first time that, at the level of transcriptional regulation, SUMO3 has a unique responsiveness to oxidative stress that is entirely different from SUMO2. As SUMO2/3 conjugation is rapidly induced by various stress responses [14,15,17,19,31], the regulation of SUMO2/3 by ROS has emerged as an interesting topic [15,17,18,22,23,25]. We have found previously that global SUMO2/3 conjugations were increased upon H2O2 treatment, whereas SUMO2/3 modification of certain proteins, for instance p300 and PML (promyelocytic leukaemia protein), was decreased simultaneously. This suggests that, under oxidative stress, specific SUMO2/3 deconjugation occurs in parallel with a general conjugation [15,17]. These deconjugations of SUMO2/3 from specific proteins are mediated by SENP3 [SUMO1/sentrin/SMT3 (suppressor of mif two 3 homologue 1)-specific peptidase 3], a SUMO2/3-specific protease that is rapidly stabilized under mild oxidative stress [15,17] (NIH 3T3 cells with 10 μM H2O2, HeLa cells with 50 μM H2O2). Under more intense oxidative stress, SUMOylation may undergo more complex regulation mainly due to inactivation of conjugating enzymes (SUMO1 conjugation in HeLa cells with 1 mM H2O2) or deconjugating proteases [SENP1 and SENP2 in CHO (Chinese-hamster ovary) cells with 1 mM or 10 mM H2O2]. Therefore it has been known that the SUMOylation/de-SUMOylation equilibrium can be regulated by ROS in various ways. In the present study we highlight that ROS can regulate SUMOylation in an additional way: transcriptional down-regulation of SUMO3 under conditions of long-term sublethal oxidative stress. Given that a free pool of SUMO2/3 that becomes conjugated to substrates upon cellular stress is believed to exist [13,6], we postulate that the resources of this pool may be subjected to regulation. Therefore redox-regulation of the expression of SUMOs at the transcription level may serve as a long-term regulatory mechanism. This sort of regulation is likely to play a role in cellular adaptive responses to certain long-term pathological conditions such as chronic inflammation, and, as indicated by the present study, cancer.

Sp1 belongs to a superfamily of zinc-finger-containing DNA-binding proteins and is ubiquitously expressed [45]. Sp1-binding sites are GC-rich elements and essential for the basal transcription of TATA-less promoters [40,41]. Most cell types constitutively express active Sp1 [41], which is consistent with the expression profile of the SUMO3 gene [3]. In the present study we identified that the mouse Sumo3 promoter is TATA-less, rich in GC and contains a Sp1-binding site at the −91/−86 region that is critical for basal as well as ROS-regulated transcription. In addition, comparing the mouse Sumo3 promoter sequences with the human equivalent, it is convincing that they may have similar patterns of regulation. However, the Sumo2 promoter has none of these features, which explains why transcription of the two SUMO isoforms may respond to ROS regulation differentially.

Sp1 was initially thought to serve mainly as a constitutive activator of housekeeping genes. It is now known that Sp1 can function in transcriptional regulation of a wide range of genes in positive or negative ways [46,47]. Because of the high abundance of Sp1 proteins, modulation of Sp1 functions is not likely to occur through its expression, as suggested by the results of the present study (Figure 3D) that exogenous overexpression of Sp1 cannot enhance its transcriptional activity to a significant extent. Instead, Sp1 activity can be effectively and sensitively controlled through various post-translational modifications including phosphorylation, acetylation, glycosylation, ubiquitination, SUMOylation and oxidation [4850], which allow it to activate or repress transcription in response to physiological and pathological stimuli. Sp1 is known to be a redox-sensitive transcription factor and is characterized by thiol modifications of the cysteine residues forming zinc-finger structures. This in turn dramatically diminishes the Sp1–DNA binding ability and suppresses transcription of Sp1-targeted genes [43]. Oxidation of Sp1 by ROS is a critical and frequently operated mechanism [47], which makes Sp1 a negative regulatory player in gene expression. We have revealed that oxidation of Sp1 by ROS is also an important factor responsible for the distinct transcriptional regulation patterns of SUMO2 and SUMO3 under cellular oxidative stress. Therefore this finding may enhance our understanding of the regulation of SUMOylation and also shed light upon Sp1 functions. Whether this Sp1-dependent transcriptional down-regulation of SUMO3 is involved in oxidative stress-related human diseases such as, for instance, cancer needs to be studied in the future.


Jing Sang designed and performed the experiments, interpreted data and wrote the manuscript. Kai Yang designed and performanced ChIP assays with the BamHI-site-containing plasmids, and performed other experiments. Yueping Sun designed some experiments and was involved in discussion. Yan Han was involved in preliminary findings. Hui Cang carried out some plasmid construction and mutagenesis. Yuying Chen performed immunofluorescence analysis. Guiying Shi carried out ROS detection by FACS. Kangmin Wang performed image analysis. Jie Zhou and Xiangrui Wang collected clinical samples. Jing Yi designed experiments, interpreted data and wrote the manuscript.


This work was supported by grants from the National Natural Science Foundation of China [grant number 30971437 (to J.Y.)]; the Shanghai Municipal Science and Technology Commission [grant number 08JC1413800 (to J.Y.)]; the Shanghai Municipal Education Commission [Leading Academic Discipline Project, grant number J50201]; the Ren Ji Collaboration Project (to X.W. and J.Y.); and a Doctorial Student Innovation Fellowship in Jiao Tong University [grant number BXJ201001 to (J.S.)].


We thank Professor G. Suske for his gifts of plasmids of pN3 vector and pN3-Sp1FL-complete, and also thank Dr T. Yamamoto (Kochi University, Kochi, Japan) for OSC and OSC-AS-SOD cells.

Abbreviations: ChIP, chromatin immunoprecipitation; DCFH-DA, 2′,7′-dichlorodihydrofluorescein diacetate; DHE, dihydroethidium; DMEM, Dulbecco's modified Eagle's medium; F5M, fluorescein-5-maleimide; NAC, N-acetylcysteine; OSC, oral squamous carcinoma; Q-PCR, quantitative PCR; ROS, reactive oxygen species; RT, reverse transcription; SENP, small ubiquitin-related modifier 1/sentrin/SMT3 (suppressor of mif two 3 homologue 1)-specific peptidase; SOD, superoxide dismutase; Sp1, specificity protein 1; SUMO, small ubiquitin-related modifier; TBS, Tris-buffered saline; USF, upstream stimulatory factor


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 41a.
  43. 42.
  44. 43.
  45. 44.
  46. 45.
  47. 46.
  48. 47.
  49. 48.
  50. 49.
  51. 50.
View Abstract