Differential regulation of sentrinized proteins by a novel sentrin-specific protease.

Sentrin-1, also called SUMO-1, is a protein of 101 residues that is distantly related to ubiquitin and another ubiquitin-like protein, NEDD8. Here we report the cloning of a novel sentrin-specific protease, SENP1, which has no homology to the known de-ubiquitinating enzymes or ubiquitin C-terminal hydrolases. However, SENP1 is distantly related to the yeast Smt3-specific protease, Ulp1. A COS cell expression system was used to demonstrate the activity of SENP1 in vivo. When HA-tagged sentrin-1 was co-expressed with SENP1, the higher molecular weight sentrin-1 conjugates were completely removed. Surprisingly, the major sentrinized band at 90 kDa remained intact. The disappearance of the high molecular weight sentrin-1 conjugates also coincided with an increase in free sentrin-1 monomers. SENP1 is also active against proteins modified by sentrin-2, but not those modified by ubiquitin or NEDD8. In addition, sentrinized PML, a tumor suppressor protein that resides in the nucleus, was selectively affected by SENP1, whereas sentrinized RanGAP1, which is associated with the cytoplasmic fibrils of the nuclear pore complex, remained intact. The inability of SENP1 to process sentrinized RanGAP1 in vivo is most likely due to its nuclear localization because SENP1 is active against sentrinized RanGAP1 in vitro. The identification of a nuclear-localized, sentrin-specific protease will provide a unique tool to study the role of sentrinization in the biological function of PML and in the pathogenesis of acute promyelocytic leukemia.

Sentrin-1 (also called SUMO-1) is a small polypeptide that can covalently modify specific proteins in a manner analogous to ubiquitination (1)(2)(3)(4)(5)(6). In mammalian cells, there are three known sentrin family proteins that are expressed in all tissues and appear to have overlapping function (7)(8)(9). It is now clear that the sentrinization pathway in mammalian cells utilizes a unique activating enzyme complex (UBA2/AOS1) and conjugation enzyme (UBC9), to catalyze the modification of a subset of mammalian proteins, such as PML, Sp100, IB␣, RanGAP1, and RanBP2 (3, 4, 9 -16). The biological function of sentrinization in the mammalian system has not been completely eluci-dated for each target substrate. It has been shown, however, that sentrinization of RanGAP1 is required for its translocation from the cytosol to the nuclear pore complex (3,4), whereas sentrinization of PML may regulate its subnuclear localization (9,14) and the assembly of the PML-containing nuclear body (17). Sentrin-1 is closely related to Smt-3, an essential protein in Saccharomyces cerevisiae (18). The enzymatic machinery for Smt-3 conjugation (19,20) is similar to that of the sentrinization pathway, suggesting that the sentrin-1/Smt-3 pathway of ubiquitin-like modification is conserved throughout evolution.
Demonstration of sentrin modification in early studies was complicated by the presence of enzymes that cleaved the isopeptide linkage between sentrin and various target proteins in the cell lysate (2,3). Thus, in analogy to the ubiquitin pathway (21), enzyme(s) capable of removing sentrin from sentrinized proteins must also exist. Recently, Li et al. (22) reported a S. cerevisiae enzyme, Ulp1, which can remove Smt-3 (the yeast homologue of sentrin-1) from its conjugates. Ulp1 also cleaves sentrin-1/SUMO-1, but not ubiquitin, from modified proteins in vitro. Interestingly, Ulp1 is not related to any known de-ubiquitinating enzyme. Here, we report the cloning of a novel protease, called SENP1, which is distantly related to Ulp1 and is active against sentrin but not ubiquitin or NEDD8modified proteins in vivo. We also elucidate the genomic organization of the SENP1 gene and show that SENP1 differentially regulates sentrin-modified proteins in vivo.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture Conditions-COS-M6 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics.
Antibodies-16B12 (Babco, Richmond, CA) is a mouse monoclonal antibody to the peptide sequence YPYDVPDYA of influenza hemagglutinin (HA). 1 Mouse anti-RH (specific for the amino acid sequence, RG-SHHHH) monoclonal antibody was purchased from Qiagen (Santa Clara, CA).
cDNA Cloning of the Human SENP1-A tBlastn search of the data bases from a human expressed sequence tag (EST) was performed using the 177 amino acids of HsUlp1, which is homologous to yeast Ulp1 (22). 70 EST sequences were identified in the initial screen. After further analysis, one EST clone (AI261629) was found to have a sequence identical to HsUlp1 (22). Two other EST sequences (AI148063 and N36589) were found to overlap partially with AI261629. The combined amino acid sequence from these three EST clones was extended to a 1113-bp PCR fragment by RACE of both 5Ј and 3Ј ends. When the cloned 1113-bp cDNA was used to search GenBank TM , one genomic DNA fragment located in 12q13.1 (accession number AC004801; 193 kb) was found to have a 209-bp region (nt position 89832-90041) which contains a conserved His site (PI-L-VHW-L), homologous to HsUlp1. Further analysis of the translated sequence from nt position 90041 to * This work was supported in part by National Institutes of Health Grant GM-57502, the DREAM project, and an American Heart Association Established Investigator Award. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  93000 on the 193-kb human genomic DNA led to the identification of another region (nt position 92516 -92611) containing a conserved Cys site. The Cys surrounding sequence (PQQMNGSDCG) is 80% identical to the Cys site region of yeast Ulp1 (PQQPNGYDCG). Two specific oligonucleotides (one from nt position 89832-90041 and another from nt position 92516 -92611) were used as primers for PCR amplification from a human placenta cDNA library. A 450-bp cDNA fragment was cloned, suggesting that the 193-kb DNA fragment contains a novel gene, which was later named SENP1. RACE of the 3Ј end of SENP1 identified a stop codon at nt position 93643-93645 of the 193-kb DNA. RACE of 5Ј end of the 450-bp cDNA led to the identification of the start codon for SENP1 at nt position 37672-37674. A 2511-bp cDNA fragment was cloned by PCR using primers from the RACE fragments. The full length of SENP1 (1.93 kb) was subcloned to pcDNA3 vector for further study. Direct sequencing of PCR fragments and comparison with genomic DNA confirmed the sequence of SENP1 cDNA.
PCR, 5Ј-RACE, and Sequence Analysis-Nested primers were synthesized on the basis of the information obtained from either the positive EST clones or from the genomic DNA. These primers were used to amplify the novel protease gene fragments by PCR from a human placenta cDNA library. Both PCR and RACE were performed as described previously (23). The nucleotide sequences were determined using dye terminator sequencing and an automated sequencer from Applied Biosystems Inc. (Foster City, CA).
Western Blotting-Protein samples were treated at 45°C for 1 h in 300 l of 2% SDS treating solution containing 5%-mercaptoethanol. Western blotting was performed using the protocol provided by ECL FIG. 1. Sequence alignment of SENP1 and Ulp1. Identical amino acids are shaded blocks. Accession numbers for SENP1 and Ulp1are AF149770 and AAB68167, respectively. The underline represents the sequence of the "core domains" of yeast Ulp1 and its related enzymes. The asterisks mark likely residues of the catalytic triad (histidine, aspartate, and cysteine) and a glutamine residue predicted to form the oxyanion hole in the active site. detection system (Amersham Pharmacia Biotech). Horseradish peroxidase (HRP)-conjugated antibodies against mouse IgG or rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA) were used as secondary antibodies.
GST Fusion Proteins Expression and Purification in Escherichia coli-Full-length cDNA of SENP1 was amplified by PCR from a human placenta cDNA library. The PCR product was subcloned into pGEX-5 ϫ1 (Amersham Pharmacia Biotech Inc.) using BamHI and XhoI restriction sites to generate pGEX-SENP1. E. coli BL21 cells carrying pGEX-SENP1 or pGEX-5 ϫ 1 were grown to saturation in 10 ml of LB containing 50 g/ml ampicillin and transferred to 500 ml of LB broth for expansion to an absorbance (600 nm) of 0.8. After the addition of isopropyl-␤-D-thiogalactopyranoside (final concentration: 0.1 mM), the culture was incubated at room temperature for 3 h. The cells were sonicated in lysis buffer (20 mM Tris, pH 8.0, 1 mM EDTA, 100 mM NaCl, 1% IGEPAL CA-630, 100 g/ml egg white lysozyme), and bacterial debris was removed by centrifugation. The sonicated lysates were incubated at 4°C for 1 h with 500 l of a 50% slurry of glutathione-Sepharose 4B beads. After centrifugation, GST-fusion protein bound beads were washed three times and eluted from the beads. Immunohistochemical Localization of Proteins-HEp-2 carcinoma were grown at 37°C in a humidified 5% CO 2 atmosphere. For immunohistochemical staining, cells were grown on round coverslips in 24well plates (Corning Glass, Inc.) until approximately 80% confluent before fixation. Two days after plating on round glass coverslips, cells were fixed at room temperature for 15 min with freshly prepared 1% paraformaldehyde in phosphate-buffered saline and treated as described previously (24). Polyclonal rabbit antiserum against PML was used to visualize PML (25). Cells were analyzed using a Leica confocal laser scanning microscope. Leica image enhancement software was used to balance signal strength, and 8-fold scanning was used to separate signal from noise. Because of the variability among cells in any given culture, the most prevalent cells were photographed and are presented as small groups of nuclei or single nuclear images at high magnification.

RESULTS AND DISCUSSIONS
cDNA Cloning and Genomic Organization of SENP1-Li et al. (22) recently reported a novel protease, Ulp1, specific for Smt3, the yeast homologue of sentrin-1. In the same report, a human EST sequence homologous to Ulp1 was tentatively termed HsUlp1. Using a combined PCR cloning and data base search technique described previously (12, 23), we identified a partial-length HsUlp1. Using the partial-length HsUlp1 sequence as a query in a BLASTn sequence search, we detected a conserved region of 209 bp from a 193-kb human genomic DNA fragment (accession number AC004801). This 209-bp region contains one conserved histidine residue present in all Ulp1related proteins from different species (22). Further analysis of the translated sequence from the 193-kb genomic DNA by the exon trapping technique (26) resulted in the discovery of another region that contains one conserved cysteine residue. To determine whether this genomic DNA encodes a functional protein, we amplified a 450-bp cDNA fragment from a human placenta cDNA library with PCR using primers based on the conserved histidine and cysteine site sequences. Extension of the cloned cDNA by RACE resulted in the identification of a 2511-bp cDNA clone from a human placenta cDNA library. The 2511-bp cDNA clone contains an open reading frame of 1929 bp, encoding a protein of 643 amino acids (Fig. 1). The predicted coding region of this open reading frame is preceded by an in-frame stop codon and has a consensus sequence commonly associated with initiation methionines (27). The protein encoded by this open reading frame was named SENP1 (sentrinspecific protease-1) because it possesses a protease activity against sentrin-modified proteins but not ubiquitin or NEDD8modified proteins in vivo (see below).
As shown in Fig. 1, SENP1 is 21% identical and 50% similar to yeast Ulp1. The similarity between SENP1 and yeast Ulp1 is confined primarily to the C-terminal region of ϳ200 amino acids, within which an ϳ90-residue segment has been proposed to form a core structure common to a diverse and widespread group of cysteine proteases (22). Similar to yeast Ulp1, SENP1 contains the four conserved catalytic residues of an adenoviral protease (22).
The SENP1 gene is located in 12q13.1 because it is carried by the Homo sapiens 12q13.1 PAC RPCI1-228P16 fragment (Roswell Park Cancer Institute Human PAC Library; accession number AC004801). It spanned about 61 kb of contiguous DNA (Fig. 2). The SENP1 gene is composed of 18 distinct exons ranging between 39 and 487 bp. Both the 5Ј and 3Ј acceptor splice sites in each of the introns followed the GT-AG consensus sequence for eukaryotic genes (Fig. 2). Both exon 1 (112 bp) and exon 2 (47 bp) encoded most of the 5Ј-untranslated region, whereas exon 3 contained the remaining 14 bp of the 5Ј-un-translated region plus the first 36 amino acids. Exons 5 through 17 encoded most of the amino acids, with exon 18 (ϳ487 bp) containing the final 20 codons and an extensive 3Ј-untranslated region of ϳ423 bp.
SENP1 Is a Sentrin-specific Protease-A COS cell expression system was used to demonstrate the activity of SENP1 in vivo. Briefly, HA-tagged sentrin-1 was introduced into COS cells by liposome-mediated transfection as described previously (2). Total cell lysates were prepared 16 h after transfection for Western blot analysis using anti-HA antibody. As shown in Fig. 3A, lane 1, a 90-kDa band and higher molecular weight sentrin-1 conjugates were detected. When HA-tagged sentrin-1 was coexpressed with His-tagged SENP1, the higher molecular weight sentrin-1 conjugates were completely removed (lane 2). However, the 90 kDa band, which most likely represents sentrinized RanGAP1, remained intact. The disappearance of the high molecular weight sentrin-1 conjugates also coincided with the accumulation of free sentrin-1 monomers. A similar pattern was observed when Myc-tagged SENP1 was co-expressed with HA-tagged sentrin-1 (lane 3). The activity of SENP1 is restricted to sentrin-1 because it did not appear to have any activity against ubiquitin-modified proteins (lanes 4 -6). Fig.  3B shows that SENP1 is also active against sentrin-2 modified proteins. Again, the 90-kDa band was not affected by SENP1, and the sentrin-2 monomer accumulated in COS cells overexpressing His-tagged SENP1. We also tested the activity of SENP1 against NEDD8-modified proteins. As expected, SENP1 was unable to affect NEDD8 conjugates (Fig. 3C). Taken together, SENP1 is a sentrin-specific protease that selectively removes sentrin from sentrinized proteins. SENP1 Processes Sentrinized PML but Not Sentrinized Ran-GAP1-The inability of SENP1 to reduce the 90-kDa band suggests that SENP1 cannot remove sentrin from all sen- trinized proteins. Thus, we tested the effect of SENP1 on two specific sentrin conjugates, RanGAP1 and PML. RanGAP1 is a 70-kDa cytosolic protein that can be modified by a single molecule of sentrin-1 (28,29). Sentrinized RanGAP1 (90 kDa) is a component of the nuclear pore complex and plays a role in regulating nuclear transport. When HA-tagged RanGAP1 was expressed in COS cells, a 70-kDa unmodified form of RanGAP1 and a 90-kDa band corresponding to sentrinized RanGAP1 were observed (Fig. 4A, lane 1). Co-expression of SENP1 was unable to remove native sentrin from HA-tagged RanGAP1 (lane 2). This is consistent with the results shown in Fig. 3, A and B, which show that the 90-kDa band is resistant to SENP1. Li et al. (22) have shown that SUMO-1 (sentrin-1)-modified RanGAP1 could be cleaved by Ulp1 using an in vitro assay. To further explore the difference between our in vivo results and Li's in vitro observations, we tested GST-tagged SENP1 fusion protein in an in vitro assay (Fig. 4B). As shown, sentrinized RanGAP1 could be effectively processed by GST-SENP1 but not by GST in vitro. The discrepancy between the in vitro and in vivo activity of SENP1 against sentrinized RanGAP1 could be explained by the observation that SENP1 is localized in the nucleus (see below). Thus, the nuclear localized SENP1 does not have access to sentrinized RanGAP1, which is attached to the cytoplasmic fibrils of the nuclear pore complex. This interpretation is also supported by our previous finding that the majority of high molecular weight sentrinized proteins, which are sensitive to SENP1 (Fig. 3, A and B), are localized in the nucleus (2). To test this hypothesis further, we study the in vivo activity of SENP1 against a well studied nuclear protein, PML. As expected, native sentrin was completely removed from sentrinized PML in COS cells expressing His-tagged SENP1 (Fig. 4A, lane 4).
The importance of cellular localization in the activity of SENP1 is also supported by the following co-localization experiment in which Myc-tagged SENP1 was transfected into HEp-2 cells. Fig. 5, B, C, E, and F show that SENP1 is evenly distributed throughout the nucleus. The question of whether SENP1 removes sentrin-1 from sentrinized proteins was tested by colabeling with anti-sentrin-1 antibody. As shown in Fig. 5, D-F, in a field with seven cells, two HEp2 cells have been transfected with SENP1-containing plasmid and express SENP1 in the nucleus (red). In these two transfected cells, sentrin-1 (green) is localized in a nuclear rim pattern consistent with the nuclearpore localization of RanGAP1. In contrast, in the five cells that were not transfected, sentrin-1 is localized not only in the nuclear rim, but also in nuclear dots consistent with the localization of PML-containing nuclear body (also see Fig. 5, A-C).
Thus, SENP1 appears to selectively remove sentrin-1 from the PML-containing nuclear body but not from the nuclear rim where RanGAP1 is localized. Thus, the cellular localization results in HEp2 cells (Fig. 5) are consistent with the Western blot results of the COS cells (Fig. 4). Our results also suggest that removal of sentrin-1 from the PML-containing nuclear body does not destabilize it. PML, a RING finger protein with tumor suppressor activity, has been implicated in the pathogenesis of acute promyelocytic leukemia that arises following a reciprocal chromosomal translocation that fuses the PML gene with the retinoic acid receptor ␣ (RAR␣) gene (30). In acute promyelocytic leukemia, two forms of PML-RAR␣ fusion proteins have been reported. Remarkably, both forms of PML-RAR␣ fusion proteins could not be sentrinized in vivo (10). Further experiments are needed to study the role of SENP1 in regulating the biological function of PML and in the pathogenesis of acute promyelocytic leukemia.
In summary, SENP1 is a sentrin-specific protease that preferentially removes sentrin from sentrinized proteins in the nucleus. Survey of the data base reveals at least four other human EST sequences that have significant homology with SENP1. Thus, it is likely that additional sentrin-specific protease will be identified in the near future that is localized in different cellular compartments and is involved in the regulation of different sentrinized proteins in vivo.