|Home | About | Journals | Submit | Contact Us | Français|
Small ubiquitin-like modifiers (SUMO) conjugation to cellular proteins is a reversible posttranslational modification that mediates the protein’s function, subcellular localization, and/or expression. The SUMO proteases (SENP) deconjugate modified proteins and thus are critical for maintaining the level of SUMOylated and un-SUMOylated substrates required for normal physiology. Altered expression of SENPs is observed in several carcinomas. This review focuses on how the change in SENP levels disturbs SUMO homeostasis and contributes to cancer development and progression. We reported that one member of the SENP family, SENP1, can transform normal prostate epithelia to a dysplasic state and directly modulate several oncogenic pathways in prostate cells, including AR, c-Jun, and Cyclin D1. Assessment of tissue from human prostate cancer patients indicates elevated mRNA levels of SENP1 and the SUMO2/3 deconjugating enzyme, SENP3. The induction of SENP3 in cancer cells initiates the angiogenic pathway; specifically, SENP3 regulates the transcriptional activity of hypoxia-inducible factor 1α via deSUMOylation of the coregulatory protein p300. Unlike prostate cancer, enhanced SUMOylation is favored with onset of breast cancer and correlated with the reduced SENP6 mRNA levels found in several breast cancer tissue arrays. Preventing enhanced SUMO conjugation of cellular substrates in breast cancer cells reduces tumorigenesis. Hence, distortion of SUMO equilibrium contributes to the initiation and progression of cancer, specifically in prostate and breast cancers. The deSUMOylation machinery may be key to restoring balance to the SUMO system and thus serve as ideal targets for therapeutic agents.
SUMO (small ubiquitin-like modifier) is readily conjugated to numerous cellular protein substrates in a sequence of events termed SUMOylation. SUMOylation is a dynamic, reversible covalent modification. Mammals express 4 SUMO isoforms: SUMO1, SUMO2, SUMO3, and SUMO4. However, posttranslational modification of cellular substrates by SUMO4 is questionable, as it remains in the inactive form in vivo.1,2 SUMO2 and SUMO3 share 95% sequence homology and hence are collectively referred to as SUMO2/3. Once conjugated to their cellular substrate, SUMO2 and SUMO3 efficiently form polymeric chains, while in contrast, SUMO1 polychains are rarely observed.3
The initial step for SUMOylation requires the inactive SUMO to be cleaved at the carboxy terminus by the hydrolase activity of the SUMO-specific cysteine proteases (SENPs). This enzymatic activity of the SENPs exposes 2 glycine residues and generates an active SUMO. The SENPs exhibit differences in this endopeptidase activity. SENP1 and SENP2 process all 3 isoforms efficiently; SENP3 and SENP5 express activity for only SUMO2 and SUMO3; and SENP6 and SENP7 do not demonstrate hydrolase activity.
Conjugation of SUMO is mediated via a sequential process that closely mimics ubiquitylation. SUMO is activated by the formation of a thioester bond between the carboxy-G residue and the C residue on the heterodimeric E1 activating enzyme, SAE1/SAE2.4 This activated SUMO is then transferred from the E1 enzyme to the E2 conjugating enzyme, Ubc9.4,5 The SUMO protein can then be ligated to its cellular substrate directly by E2 or may require the ligase activity of an E3 enzyme. SUMOylation occurs at a particular sequence, ψKXE, on the target protein; ψ is a hydrophobic residue and X is any amino acid.6,7 Hence SUMO conjugation of the target protein requires recognition of this sequence by the E2-SUMO (or E3-SUMO) complex (Figure 1).
Unlike endopeptidase activity, all SENP family members exhibit isopeptidase activity to cleave the isopeptide bond between the glycine residue of SUMO and the lysine side chain of a substrate.7 The catalytic activity is maintained within a highly conserved 200 amino acid–region in the C-terminus of the proteases. The 6 mammalian SENP enzymes can deconjugate monoSUMOylated proteins or dissemble polymeric SUMO side chains. SENP1 and SENP2 can deSUMOylate cellular substrates modified by any of the 3 SUMO isoforms, while the remaining 4 SENPs are more efficient at deconjugating SUMO2 and SUMO3 than SUMO1. Differences in the subcellular localization of the SENPs contribute to the selectivity for deSUMOylation of specific cellular proteins. SENP1, SENP6, and SENP7 are distributed in the nucleoplasm. SENP2, SENP3, and SENP5 are more compartmentalized with SENP2 in the nuclear pore complex and SENP3 and SENP5 in the nucleolus. In addition, SENP1 and SENP2 express nuclear export sequences that allow these 2 SENPs to shuttle in and out of the nucleus. Based on these similarities and differences, the SENP can be categorized into independent subfamilies: Family 1: SENP1 and SENP2; Family 2: SENP3 and SENP5; Family 3: SENP6 and SENP7.
The posttranslational modification of SUMO to cellular substrates is vital for normal cell physiology. Studies in knockout mice define a few of these cellular events and also demonstrate the fine balance in SUMOylation/deSUMOylation required for normal embryonic development. Loss of SUMO conjugation with knockout of Ubc9 results in severe embryonic lethality just after development into blastocysts.8 The Ubc9-deficient cells exhibit chromosomal defeats and abnormal nuclear morphology, suggesting that SUMOylation of specific targets are critical for maintaining both these processes. On the other hand, excessive SUMO conjugation with knockout of either SENP1 or SENP2 is also embryonic lethal.9–11 Our recent studies in mouse embryonic fibroblasts (MEF) from SENP1 knockout mice illustrate how adequate SUMOylation/deSUMOylation dictates protein expression in the cell.9 In these cells, the ablation of SENP1 increases SUMOylation of the hypoxia-inducible factor 1α (HIF1α), leading to HIF1α ubiquitylation, and its subsequent proteosomal-mediated degradation under hypoxic conditions. Chiu et al. report that loss of SENP2 in MEF cells inhibits cell cycle progression, specifically G1- to S-phase transition.10 In our SENP2 knockout model, we demonstrate that SUMO homeostasis contributes to epigenetic regulation of gene transcription.11 Ablation of SENP2 leads to excessive SUMOylation of a Polycomb group (PcG) protein, which in turn facilitates binding of PcG to the histone trimethylation marker H3K27me3 and, finally, repressing transcription of several PcG-regulated genes concurrently. Collectively, these knockout mice models demonstrate that the individual components of the SUMO machinery are critical for a plethora of cellular events, including but not limited to chromosomal integrity, nuclear architecture, protein stability, cell cycle progression, epigenetic regulation, and gene transcription.
Under normal physiological conditions, only a fraction of the total number of proteins expressed in the cell is SUMO modified. Hence, the isopeptidase activity is essential to maintain a stable level of SUMO conjugated and unconjugated substrates. As defined above, the loss of a specific SENP gravely alters cell biology. The expression of several SENPs is altered in multiple cancers. The remainder of this review will discuss (1) what dictates changes in SENP expression with the onset of cancer, (2) how the balance of SUMO modification is lost, and (3) how the altered SENP expression contributes to carcinogenesis.
Elevated SENP1 levels are observed in thyroid oncocytic adenocarcinoma and prostate cancer.12–14 However, the upregulation of SENP1 is a relatively early event in the carcinogenesis of the prostate; using in situ hybridization, we observe an increase of SENP1 mRNA in the precancerous lesions called prostatic intraepithelial neoplasia as compared to normal adjacent prostate epithelia in human samples.13 A recent study suggests that the SENP1 induction may initially be required to counter the higher levels of free unconjugated SUMO1 observed in the normal prostate gland as compared to other organs.15 However, persistent elevation of SENP1 directly facilitates the transformation of the normal prostate gland to dysplasic state as observed in our transgenic mice model. A SENP1 transgene was directed to the mouse prostate gland with an androgen-driven promoter. SENP1 levels were significantly elevated in the prostate epithelia with the expression of the transgene at 4 months of age. This induction of the SENP1 produced distinct hyperplasia with enhanced expression of the secretary epithelial cells crowding into the lumen of the prostate from 3 out of 4 founders. This dysplasic growth was not observed in prostate samples from age-matched wild-type mice.13 Hence upregulation of SENP1 is sufficient to illicit transformation of the prostate gland.
Our previous reports indicate that SENP1 modulates several pathways that are critical for prostate gland carcinogenesis. This SENP enzyme modulates the transcriptional activity of the androgen receptor (AR) via deSUMOylation of the coregulatory protein HDAC1.16 Interestingly, the AR directly dictates the transcription of the SENP1 gene.17 Androgen-activated AR readily binds the SENP1 promoter at a specific DNA binding site called an androgen response element to initiate transcription of the SENP1 gene and elevate SENP1 mRNA. A positive feedback loop exists between AR and SENP1: SENP1 enhances AR transcriptional activity, which potentiates SENP1 expression. Disruption of this feedback loop with siRNA-targeted knockdown of SENP1 significantly blunts androgen-driven prostate cell proliferation.17 Hence SENP1 exhibits an intriguing relationship with AR, which initiates a prominent signal cascade for the development of prostate cancer. In addition to the transcriptional activity of AR, SENP1 facilitates c-Jun-dependent transcription18 and increases expression of the cell cycle regulator Cyclin D1.13 We are currently deciphering whether these mechanisms and yet-unidentified pathways are responsible for the transformation of the prostate gland in SENP1 transgenic mice.
SENP3 is also elevated in prostate cancer and additional carcinomas, including ovarian, lung, rectum, and colon.19 The tumor suppressor protein p19ARF is known to dictate SENP3 turnover; it initiates SENP3 phosphorylation, ubiquitylation, and subsequent proteosomal-mediated degradation.20 Loss of ARF is observed with the onset of several human cancers,21,22 and hence, deregulation of the ARF-mediated SENP3 turnover could attribute to the elevated SENP3 levels observed in various carcinomas. Alternatively, induction of SENP3 can be mediated via reactive oxygen species (ROS); ROS inhibits the ubiquitin-proteosomal mediated degradation of SENP3 to increase SENP3 protein levels.23 Increasing administration of H2O2 produces a dose-dependent induction of SUMO2/3, but not SUMO1, and conjugates and facilitates the redistribution of SENP3 from the nucleolus to the nucleoplasm. This relocalization changes the set of substrates deconjugated by SENP3, including the SUMO2/3-modified HIF1α. Enhanced expression of SENP3 increases HIF1α transcriptional activity but not through deconjugation of SUMO2/3-modified HIF1α. Instead, SENP3 mediates this induction of HIF1α transcription via deSUMOylation of the coregulatory protein p300. In this manner, overexpression of SENP3 facilitates the expression of HIF1α-regulated vascular endothelial growth factor (VEGF), which is critical for vascular development. When SENP3 was stably overexpressed in HeLa cells and subsequently xenografted into nude mice, the SENP3 overexpression produced more aggressive tumors, as exemplified via the greater tumor volume and angiogenesis in the xenograft animals. Hence the induction of SENP3 directly contributes to cancer progression, possibly most notably in cancers with increased ROS levels (like prostate cancer24).
It is intriguing to speculate that SENP1 may play a similar role in the cancer progression. As discussed above, data from SENP1 knockout mice indicate that the loss of SENP1 potentiates HIF1α degradation and consequently lowers VEGF levels.9 Reduction of VEGF hinders development of new vasculature and contributes to the lethality of SENP1 knockout. Based on these studies from SENP1−/− MEF cells, it is feasible that induction of SENP1 carcinomas facilitates angiogenesis via enhanced stability of HIF1α; this is currently being investigated in our laboratory.
Whereas the induction of SENP1 and SENP3 in prostate cancer would favor enhanced deSUMOylation of cellular substrates, it is likely that SUMOylation would be prevalent in breast carcinomas (Figure 2). Using bioinformatics analysis of published microarray data, a recent report demonstrated downregulation of SENP6 mRNA in breast tumor tissue as compared to normal tissue.25 Currently, it is unknown how onset of breast cancer elicits a change in SENP6 mRNA expression. In addition to the reduction of SENP6, mRNA of the SUMO conjugation machinery—specifically, SUMO1, Ubc9, and the E3 PIAS3—is elevated in data from this study and an additional report.25,26 Augmentation of SUMO conjugation in breast cancer cells increases tumor formation; stable overexpression of Ubc9 in the breast cancer cell line MCF7 increases SUMOylation and tumor volume when xenografted into mice.27 In contrast, decreasing SUMOylation with the expression of the dominant negative Ubc9 inhibits tumor volume in the xenograft models. It is possible that restoring reduced SENP6 levels in breast cancer cells could produce results similar to the dominant negative Ubc9 and inhibit tumor formation. Hence, in some cancers, enhancing net SUMO conjugation (possibly via downregulation of one or more SENPs) may contribute to the progression of the carcinoma.
Numerous studies indicate that SUMO homeostasis is lost with onset of various carcinomas.25–29 Induction of the SUMO E2 Ubc9 is observed in several carcinomas, and initial studies suggested targeting the molecule for cancer therapy.27 Altering the expression of the sole SUMO E2 would directly dictate SUMO conjugation of all protein substrates within a cell. It is feasible that enhanced SUMOylation of only a select subset of cellular substrates truly contributes to carcinogenesis. Hence, promoting a general change in the overall level of SUMOylation by targeting Ubc9 may not be the most effective method to combat the cancer. The SUMO proteases provide more attractive alternative targets for therapy because 6 SENPs exhibit different (1) affinity to deSUMOylate conjugated substrates, (2) isopeptidase activity against the 3 SUMO isoforms, and (3) expression levels in several carcinomas. In this effort, we and others have attempted to develop specific inhibitors for the family of SENPs30 but none are currently available. A selective inhibitor of either SENP1 or SENP3 may restore equilibrium to the skewed SUMO system in prostate cancer (Fig. 1) and prove an effective therapeutic agent. As prognostic markers, elevated SENP1 levels in the prostate gland could identify individuals with a heightened risk of developing prostate cancer. In addition, assessment of SENP3 levels in prostate cancer samples may serve as diagnostic markers for enhanced angiogenesis and, consequently, more aggressive carcinomas. Additional studies on extensive tissue array to validate the use of either SENP1 or SENP3 as biomarkers for prostate cancer are warranted.
Many questions remain unanswered with respect to SENPs and cancer. Does the expression of one or multiple SENPs fluctuate with the onset of a given carcinoma? Are specific combinations of SENPs altered in the initiation and progression of cancer? Is there a change in function and/or localization of cancer-induced SENPs? Are cancer-induced SENPs subject to more mutation?
The current studies do indicate that deregulation of either SUMO conjugation or deconjugation can contribute to cancer progression. Whether SUMOylation or deSUMOylation facilitates cancer depends on several factors. First, the tissue type greatly influences whether SUMO modification or demodification is required to maintain dysplasic cancer growth. Second, the cancer cell will favor either SUMO deconjugation or conjugation depending on what cellular substrates require to elicit a procancer response. Finally, from the current body of data, it is difficult to tell whether changes in the level of SUMOylation or deSUMOylation persist throughout the different stages of a given cancer. It is possible that both SUMO conjugation and deconjugation are critical for cancer progression—for example, one arm may be favored for the cancer stage that initiates tumor growth, while the other arm is favored for cancer metastasis.
This work was supported in part by RO1 CA239520.
The authors declare no potential conflicts of interest with respect to the publication of this article.