RAG2 Is Down-regulated by Cytoplasmic Sequestration and Ubiquitin-dependent Degradation*

Periodic accumulation and degradation of RAG2 (recombination-activating gene 2) protein controls the cell-cycle-dependent V(D)J recombination of lymphocyte antigen receptor genes. Here we show the molecular mechanism of RAG2 degradation. The RAG2 protein is translocated from the nucleus to the cytoplasm and degraded through the ubiquitin/proteasome system. RAG2 translocation is mediated by the Thr-490 phosphorylation of RAG2. Inhibition of this phosphorylation by p27Kip1 stabilizes the RAG2 protein in the nucleus. These results suggest that RAG2 sequestration in the cytoplasm and its subsequent degradation by the ubiquitin/proteasome system upon entering the S phase is an integral part of G0/G1-specific V(D)J recombination.

The ubiquitin/proteasome system plays a major role in target-specific protein degradation and in the regulation of protein expression levels (10,11). The formation of ubiquitin-protein conjugates proceeds via a three-step cascade. First, a ubiquitinactivating enzyme (E1) activates ubiquitin, which is then transferred by a ubiquitin-conjugating enzyme (E2) to a ubiquitin ligase (E3) with which the substrate protein is associated. Finally, E3 catalyzes the conjugation of ubiquitin to the substrate protein. Proteins polyubiquitinated by these enzymes are subjected to degradation by the 26S proteasome. Recent reports have suggested that many proteins, such as p53, IB, ␤-catenin, and p27Kip1, are degraded by the ubiquitin/proteasome pathway (10 -13).
p27Kip1, a cyclin-dependent kinase inhibitor, increases RAG2 stability by inhibiting the cyclin A/CDK2 activity (9). Interestingly, p27Kip1 and RAG2 have similar sequences in the CDK-phosphorylation sites: QT 187 PKKGPL in p27Kip1 and QT 490 PKRNPPL in RAG2. This CDK-phosphorylation site has a critical role in p27Kip1 degradation (13)(14)(15). At the G1-S transition of the cell cycle, CDK2 phosphorylates Thr-187 of p27Kip1. The phosphorylated p27Kip1 is translocated from the nucleus to the cytoplasm, associated with the SCF (Skp1/cullin-1/F-box protein) E3 ubiquitin-ligase complex through this CDK-phosphorylation site, then ubiquitinated and degraded by the 26S proteasome, which promotes the cell-cycle transition from G1 to S. We speculated that a similar scenario may occur for RAG2 degradation. Thus, we examined the ubiquitination of the RAG2 protein and the correlation between RAG2 subcellular localization and its stability.

EXPERIMENTAL PROCEDURES
Plasmids-To generate RAG2 expression vectors (Fig. 1A), mouse RAG2 cDNA fragments were amplified by the reverse transcriptasepolymerase chain reaction (RT-PCR) method from the mouse thymus RNA using specific primers. To amplify wild-type RAG2 cDNA fragments, primers 5Ј-CAACTCGAGATGTCCCTGCAGATGGTAAC-3Ј and 5Ј-GCTCTAGATTAATCAAACAGTCTTCTAAGG-3Ј were used. An amplified fragment was introduced into a pCR-Blunt-TOPO vector (Invitrogen). After digestion of this plasmid with XhoI and EcoRV, the RAG2 cDNA fragment (XhoI-EcoRV) was inserted into the SalI-SmaI site of the pCAT7-neo vector (16) to generate pT7-RAG2 containing the T7-tag sequence upstream of the RAG2 sequence. To generate pEGFP-RAG2, the same RAG2 cDNA fragment was also introduced into the SalI-SmaI site of the pCAEGFP vector in which the T7-tag sequence of pCAT7-neo vector (16) had been replaced with the EGFP sequence of pEGFP-C1 (Clontech). To generate pT7-388 and pEGFP-388, primers 5Ј-CAACTCGAGATGTCCCTGCAGATGGTAAC-3Ј and 5Ј-CCTGGAT-CCTTAAGCACTGAAACAAAATTCCTC-3Ј were used. DNA fragments encoding deletion mutant 388 were amplified with these primers with RT-PCR, and the subsequent procedure was same as that of the wildtype RAG2 vectors. To generate pT7-T490A and pEGFP-T490A vectors, a point mutation at codon 490 (ACT 3 GCT) of the RAG2 cDNA of the wild-type RAG2 vectors (above) was introduced by the oligonucleotidedirected mutagenesis using the QuickChange site-directed mutagenesis kit (Stratagene). Primers 5Ј-GCAAGAGCATTGCAAGCTCCCAAA-AGAAACC-3Јand 5Ј-GGTTTCTTTTGGGAGCTTGCAATGCTCTTG-* This work was supported in part by a grant from the Japan Atomic Energy Research Institute, by contract with the Nuclear Safety Research Association. 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  C-3Ј were used for this purpose. Human XRCC4 cDNA fragment was prepared from pMycXR (1-334) (Myc-tagged full length human XRCC4 expression vector, Ref. 17) by digestion with XhoI and NotI. The 3Ј-NotI site was converted to a blunt end by Klenow treatment, and this fragment (XhoI-blunt) was introduced into the SalI-SmaI site of the pCAT7-neo vector to generate pT7-XRCC4. The histidine-tagged ubiquitin expression plasmid (pMT107) and HA-tagged p27Kip1 expression plasmid (pHA-p27Kip1) were generous gift from D. Bohmann (18) and J. Kato (15), respectively. Nuclear localization signal (NLS) sequences or nuclear exporting signal (NES) sequences (19) were introduced into pCAT7-based RAG2 expression vectors. pT7-NLS-RAG2 was constructed to express RAG2-NLS which had the NLS of nucleoplasmin (KRPAATKKAGQAKKKK,Ref. 19) at the N terminus of wild-type RAG2. First, pT7-RAG2 plasmid was digested with XhoI and HindIII, and the 3Ј-HindIII site was converted to a blunt end by Klenow treatment to generate the XhoI-blunt site of pT7-RAG2 vector. Second, an annealed DNA fragment of two synthesized oligonucleotides 5Ј-TCGA-GCTAAGAGGCCTGCGGCTACCAAAAAAGCAGGCCAGGCAAAGAA-GAAGAAAGGATCCCC-3Ј and 5Ј-GGGGATCCTTTCTTCTTCTTTGC-CTGGCCTGCTTTTTTGGTAGCCGCAGG CCTCTTAGC-3Ј was introduced into the XhoI-blunt site of pT7-RAG2 to generate pT7-NLS-RAG2. pT7-NES-T490A was constructed to express T490A-NES, which had the NES of PKI (LALKLAGLDI, Ref. 19) at the N terminus of T490A. An annealed DNA fragment of two synthesized oligonucleotides 5Ј-TCGAGCTTTAGCCTTAAAACTAGCAGGCCTTGATATCCA-3Ј and 5Ј-AGCTTGGATATCAAGGCCTGCTAGTTTTAAGGCTAAAGC-3Ј was introduced into the XhoI-HindIII site of pT7-T490A to generate pT7-NES-T490A, In Vivo Ubiquitination Assay-T7-tagged mouse RAG2 or XRCC4 expression vectors (1 g, Fig.1A) and the histidine-tagged ubiquitin expression vector (pMT107, 3 g) were transfected into COS7 cells (2ϫ10 6 cells) using Trans IT transfection reagent (Pan Vera) and 36 h later, half of the cells were treated with lactacystin (10 M), a specific inhibitor for the 26S proteasome (20), for 12 h. Forty-eight hours after transfection, cells were lysed, and ubiquitinated proteins were purified with Ni 2ϩ -charged resin (Fig. 1B). Briefly, cells were solved in 1 ml of guanidine HCl solution (0.5 M NaCl, 6 M guanidine HCl, 50 mM imidazole, 20 mM Tris, pH 8.0) and, after sonication, protein concentration was determined by protein assay (Bio-Rad). Polyubiquitinated proteins were purified from 700 g of crude cell lysates with 20 l of slurry of the Ni 2ϩ -charged resin of HISTAGcatcher (CytoSignal) according to the instructions of the manufacturer. Each 10 g of the crude lysates was precipitated with trichloroacetic acid and used for "Input." Proteins were subjected to Western blot analysis as indicated below.
Western Blot Analysis-Unless otherwise noted, transfected cells were lysed in 60 mM Tris (pH 7.6) and 1% SDS and boiled for 5 min. Protein concentration was determined by protein assay (Bio-Rad). Equal amounts of crude lysate (10 g) were subjected to SDS-PAGE, then transferred to a nitrocellulose filter. The filter was probed with the anti-T7 tag antibody (Novagen), the anti-GFP antibody (Clontech) or the anti-HA tag antibody (12CA5, Boehringer Mannheim).
Confocal Microscopy-COS7 cells plated on glass-bottomed plates (Mak Tek Corporation) were transiently transfected with EGFP-fused constructs with or without pHA-p27Kip1. Twenty-four hours after transfection, living cells were examined under the confocal microscope (Leica TCS-NT SP).

RESULTS
Ubiquitination of RAG2 Protein-To show that the RAG2 protein is a substrate for ubiquitination, we adopted a highly sensitive in vivo ubiquitination assay system originally developed by Treier et al. (Fig. 1B) (18). We generated three T7 epitope-tagged RAG2 vectors to express wild-type RAG2 (pT7-RAG2), a point mutant where Thr-490 was replaced by alanine (pT7-T490A) and a deletion mutant that contained the N-terminal 388-amino acid region (pT7-388) (Fig. 1A). The core functional domain of RAG2 for V(D)J recombination has been defined as an N-terminal 1ϳ383-amino acid region (21,22). We confirmed the recombination activity of the three constructs (data not shown). Each construct, together with the histidine-tagged ubiquitin expression vector (pMT107), was transfected into COS7 cells (18). The T7-tagged XRCC4 expression construct (pT7-XRCC4) was used as a control. Half of the cells were treated with lactacystin, a specific inhibitor of the 26S proteasome (20). Polyubiquitinated proteins were precipitated using Ni 2ϩ -charged resin and detected by Western blot analysis using the anti-T7 antibody (␣-T7).
As shown in Fig. 1C, RAG2 proteins were clearly polyubiquitinated, as evidenced by the blockade of proteasome-mediated degradation with lactacystin (lanes 1 and 2). Ubiquitinated XRCC4 was not detected even after lactacystin treatment (lanes 3 and 4). In crude cell lysates, the expression level of RAG2, but not that of XRCC4, was increased by blocking the proteasome-mediated degradation with lactacystin treatment (Fig. 1C, INPUT, lanes 1-4). These results clearly indicate that RAG2 is ubiquitinated and degraded by the 26S proteasome. 388, a mutant RAG2 whose C-terminal 140 amino acids are deleted. Each has a T7 or EGFP tag at the N terminus. B, in vivo ubiquitination assay system (18). For explanation, see text. C, visualization of polyubiquitinated RAG2 proteins. Expression vectors for the His-tagged ubiquitin and for the RAG2 proteins or XRCC4 protein were transfected into COS7 cells, and the cells were then treated (Lac (ϩ)) or untreated (Lac (Ϫ)) with lactacystin. Polyubiquitinated proteins were precipitated by Ni 2ϩ -charged resin and subjected to Western blot analysis. Polyubiquitinated proteins were visualized with the anti-T7 antibody (␣-T7, bracket). Equal aliquots of whole-cell lysate were subjected to SDS-PAGE and Western blot analysis using the ␣-T7 ( Lee and Desiderio reported that Thr-490 phosphorylation targets the RAG2 protein for rapid degradation, relieving the T490A mutant of this destruction (9). Therefore, we next examined the ubiquitination and proteasome-mediated degradation of mutant RAG2 proteins, T490A and 388 (Fig. 1C,  lanes 5-10). As expected, the T490A protein was more abundant compared with the RAG2 protein in the absence of lactacystin (INPUT, lanes 8 and 6). After lactacystin treat-ment, the expression level of T490A increased and the ubiquitination became visible (lane 7) but was weaker than that of wild-type RAG2 (compare lanes 5 and 7). This result suggests that the phosphorylation of Thr-490 is not essential for but facilitates the ubiquitination and subsequent proteasome-mediated degradation of the RAG2 protein. Unexpectedly, a deletion of the C-terminal 140-amino acid region including the Thr-490 rendered the RAG2 protein more sensitive to ubiquitination and proteasome-mediated degradation (lanes 9 and 10; compare with lanes 5 and 6). These data suggest that the following. 1) Sites for the ubiquitination and interaction with a ubiquitin ligase are located in the N-terminal 1ϳ388-amino acid region of RAG2.
3) The phosphorylation of Thr-490 abrogates such inhibitory function of the C-terminal 140-amino acid region.
Regulation of Subcellular Localization of RAG2-As mentioned before, upon phosphorylation, p27Kip1 is translocated from the nucleus to the cytoplasm and degraded through the ubiquitin/proteasome system (13,15). To investigate whether RAG2 degradation correlates with its subcellular localization similar to p27Kip1, we examined the localization of three GFP fusion proteins, EGFP-RAG2, EGFP-T490A, and EGFP-388, transiently expressed in COS7 cells (Fig. 1D). Although it is believed that RAG2 resides in the nucleus, EGFP-RAG2 is located in both the cytoplasm and the nucleus at various ratios (Fig. 1D, upper two panels, and see Fig. 3B, upper left panel). On the other hand, the T490A mutant is located exclusively in the nucleus (Fig. 1D, lower left panel), whereas the 388 mutant is predominantly located in the cytoplasm (Fig. 1D, lower right  panel). RAG2 is physiologically co-expressed with RAG1 in lymphocyte precursors (1)(2)(3). Thus, its localization may be affected by the RAG1 protein. However, the co-expression of the RAG1 protein did not alter the localization pattern for the RAG2 protein in our system (data not shown). Thus, it is Expression vectors for each of RAG2 proteins together with the EGFP vector were transfected into COS7 cells, and the cells were either untreated (Lac (Ϫ)) or treated (Lac (ϩ)) with lactacystin. Same amounts of the crude lysates were subjected to Western blot analysis using the anti-T7 (top) and anti-GFP (bottom) antibodies.

FIG. 3. p27Kip1 expression increases the nuclear localization and stability of the RAG2 protein.
A, increased stability of RAG2 by co-expression of p27Kip1. Expression vectors for EGFP-RAG2 and EGFP were introduced into COS7 cells together with (ϩ) or without (Ϫ) the p27Kip1 expression vector. Whole-cell lysates were subjected to Western blot analysis using the anti-GFP antibody to detect EGFP-RAG2 and EGFP (top) or the anti-HA antibody to detect p27Kip1 (bottom). B, RAG2 is localized exclusively in the nucleus when p27Kip1 is co-expressed. EGFP-RAG2 (upper panels) or EGFP (lower panels) was expressed with (right panels) or without (left panels) p27Kip1 in COS7 cells. presumable that RAG2 shuttles between the nucleus and the cytoplasm independent of RAG1, that the C-terminal 140amino acid region is inhibitory to the cytoplasmic localization of RAG2, and that the phosphorylation of the Thr-490 in this region abrogates this inhibition. Together with the result shown in Fig. 1C, the cytoplasmic localization of RAG2 by Thr-490 phosphorylation or the C-terminal deletion accelerates its degradation through the ubiquitin/proteasome system.
Correlation of Subcellular Localization and Stability of RAG2 Protein-To confirm the correlation between the subcellular localization and the stability of the RAG2 protein (Fig. 2), we first examined the stability of the RAG2 protein that is restricted to the nucleus or the cytoplasm by fusing a NLS or a NES to the RAG2 protein, respectively. The expression constructs encoding the wild-type RAG2, NLS-fused RAG2, T490A mutant, and NES-fused T490A mutant were transfected into COS7 cells, and their subcellular localization was confirmed by immunofluorescent detection using the anti-RAG2 antibody ( Fig. 2A). We next examined the amount of RAG2 protein in whole-cell lysates by Western blot analysis. The NLS-fused RAG2 protein was expressed in the cells at a higher level than the wild-type RAG2 protein (Fig. 2B, lanes 1 and 2). By contrast, the expression level of the T490A protein exclusively localized in the nucleus was markedly diminished by the fusion to NES (lanes 3 and 4). The expression level of these proteins was equivalent to that of the wild-type RAG2 protein in the cells treated with lactacystin ( lanes 5-8, top). Lactacystin treatment did not affect the expression level of co-expressed EGFP ( lanes 5-8, bottom). Thus, the cytoplasmic RAG2 is less stable than the nuclear RAG2 due to its proteasome-dependent degradation.
Co-expression of p27Kip1 Increases RAG2 Localization and Stability in the Nucleus-We next examined the localization of RAG2 when it was stabilized by co-expression with p27Kip1. Co-expression of p27Kip1 resulted in the increase of the EGFP-RAG2 expression level as reported previously (9) (Fig. 3A) and, strikingly, in the exclusive localization of the EGFP-RAG2 protein in the nucleus (Fig. 3B, upper panels) without affecting the localization of EGFP (lower panels). Thus, p27Kip1 induces the localization of RAG2 in the nucleus, leading to the stabilization of the RAG2 protein. Taken together, we conclude that the nuclear localization of the RAG2 protein increases its stability and that the cytoplasmic sequestration accelerates its degradation through the ubiquitin/proteasome system. RAG2 translocation from the nucleus to the cytoplasm appears to be mediated by Thr-490 phosphorylation. Thus, the p27Kip1-inhibition of this cyclin A/CDK2-mediated Thr-490 phosphorylation stabilizes the RAG2 protein in the nucleus. DISCUSSION V(D)J recombination, the process by which lymphocyte antigen receptor genes are assembled, is initiated by doublestranded DNA cleavage with RAG1/RAG2 proteins and is completed by repairing the cleaved DNA with the general repair machinery. This entire process is restricted to the G0/G1 phase of the cell cycle. The onset of V(D)J recombination correlates well with the RAG2 expression level, namely, the RAG2 protein is accumulated at G0/G1 and its expression level decreases rapidly at the G1-S transition of the cell cycle. Thus, the disappearance of the RAG2 protein turns off V(D)J recombination. The RAG2 expression level is determined by the post-transcriptional mechanism because the expression level of RAG2 mRNA is constant through the entire cell cycle (8), but the precise molecular mechanism of RAG2 degradation has not been fully clarified. Here we showed that two novel mechanisms are in-volved in RAG2 degradation: RAG2 protein translocation from the nucleus to the cytoplasm and the ubiquitination of the RAG2 protein.
Based on previous studies (8,9) and our own, we propose a model for the RAG2 degradation system (Fig. 4). After completion of V(D)J recombination, the cell cycle goes into the S phase. At the G1-S transition, p27Kip1 is degraded, and subsequently cyclin A/CDK2 becomes active and phosphorylates the Thr-490 of RAG2. This phosphorylation abates the C-terminal regulatory domain function, which is inhibitory for the cytoplasmic localization of RAG2, and promotes RAG2 translocation from the nucleus to the cytoplasm. The ubiquitination of RAG2 may occur in both the nucleus and the cytoplasm, but the active degradation is mainly executed in the cytoplasm by the 26S proteasome. The reason for the difference in RAG2 degradation between the nucleus and the cytoplasm is presently unclear. Exactly how Thr-490 phosphorylation promotes the cytoplasmic localization of RAG2 and what ubiquitin ligase could mediate the ubiquitination of RAG2 remain to be determined.
RAG2 and p27Kip1 are regulated in a similar manner, that is, by phosphorylation-dependent translocation and degradation through the ubiquitin proteasome system. In the case of p27Kip1, the Jab1 protein associates with the phosphorylated p27Kip1 and transports p27kip1 from the nucleus to the cytoplasm (15). RAG2 localization, however, was not affected by the co-expression of the Jab1 protein (data not shown). Thus, Rag2 translocation may be induced by other proteins that bind to various sites including the phosphorylated Thr-490. V(D)J recombination, the process in which lymphocytes acquire immune diversity, is potentially deleterious, because the process involves DNA cleavage. This rapid cytoplasmic sequestration/degradation of RAG2 at the G1-S transition before the onset of genomic DNA replication would ensure that the S phase does not initiate with RAG-mediated DSBs being present in the DNA. Moreover, G0/G1-specific V(D)J recombination is also consistent with the fact that the nonhomologous end-joining pathway that is responsible for the repair of DSB in V(D)J recombination is most active during G0/G1 (23). Thus, the cell cycle regulation of V(D)J recombination serves to protect cells against deleterious effects of DNA cleavage.
Here we show the molecular mechanism of the rapid sequestration/degradation of the RAG2 protein. A similar mechanism might be employed in other proteins whose nuclear localization may be harmful.