Identification and Characterization of Two Novel Isoforms of Pirh2 Ubiquitin Ligase That Negatively Regulate p53 Independent of RING Finger Domains*

Pirh2 is a newly identified E3 ubiquitin ligase known to inhibit tumor suppressor p53 function via ubiquitination and proteasomal degradation. We have identified two novel Pirh2 splice variants that encode different Pirh2 isoforms and named these Pirh2B and Pirh2C. Accordingly, the full-length protein is now classified as isoform Pirh2A. The central region of Pirh2 harbors a RING finger domain that is critical for its ubiquitin ligase function. The Pirh2B isoform lacks amino acids 171–179, whereas Pirh2C is missing C-terminal amino acids 180–261, which for each isoform results in a RING domain deletion and the abrogation of ubiquitin ligase activity. Our findings further indicate that the Pirh2B isoform but not the Pirh2C isoform is capable of binding to Pirh2A, suggesting that the C-terminal region absent in Pirh2C is critical for Pirh2-Pirh2 interactions. Similar to Pirh2A, both Pirh2B and Pirh2C interact with p53; however, interactions between p53 and Pirh2B appear stronger than those between p53 and Pirh2C. Interestingly, although both Pirh2B and Pirh2C are not able to promote in vitro p53 ubiquitination, both are capable of negatively regulating p53 protein stability and promoting the intracellular ubiquitination of p53. Furthermore, like Pirh2A, both isoforms are able to inhibit p53 transcriptional activity. We have also for the first time demonstrated that Pirh2A as well as the novel isoforms also interact directly with MDM2 within a region encompassing MDM2 acidic and zinc finger domains. It is therefore possible that Pirh2A and the novel Pirh2 isoforms identified in this study may also modulate p53 function by engaging MDM2.

The tumor suppressor p53 has been implicated in a growing number of cellular processes, the most recognized of these being the initiation of signaling cascades leading to growth arrest and apoptosis (1). A large body of evidence supports the notion that p53 is capable of regulating these pathways via distinct mechanisms. It is becoming clear that p53 probably exerts its antiproliferative effects through both its well established transcription factor function as well as a transcription-independent role at the mitochondrion (1-3).
Given its potent capacity to control cell fate, p53 in normal cells is held in check at multiple, often interrelated levels, including regulation of p53 protein stability, subcellular localization, and transcriptional activity. It is now well understood that p53 ubiquitination and subsequent proteasomal degradation is one of the principal mechanisms controlling these processes (4 -6). The RING domain containing ubiquitin ligase MDM2 was the first non-viral protein found to be responsible for the ubiquitin-dependent targeting of p53 for proteolysis (7,8). The importance of MDM2 in abrogating p53 function was illustrated in an embryonic lethal MDM2(Ϫ/Ϫ) mouse model in which lethality can be attributed to uncontrolled p53 activity and is rescued by p53 deletion (9,10).
Adding complexity to the once clear role of MDM2 in p53 ubiquitination is the recent discovery of additional RING (Pirh2, COP1, Topors, Synoviolin, and CARPS), HECT (ARF-BP1), and U-Box (CHIP) domain-containing E3 2 ligases, all capable of negatively regulating p53 (4,(11)(12)(13)(14)(15)(16)(17). Notably, Pirh2 and COP1 are both direct transcriptional targets of p53, and thus, both proteins may function redundantly with MDM2 in negative feedback mechanisms (4,11,12). At the present time, the physiological significance of multiple E3 ligases specific for p53 remains to be elucidated. Likewise, whether they function to promote or inhibit tumorigenesis is still unclear and therefore is an issue that certainly warrants further investigation.
By contrast, overexpression of MDM2 in primary tumors is well documented and is believed to be responsible for negatively regulating p53 function in tumors harboring wild-type p53. It is estimated that MDM2 gene amplification occurs in ϳ7% of all cancers (18); however, MDM2 overexpression can also occur independently of gene amplification. For instance, a single nucleotide polymorphism (SNP309) in the MDM2 promoter results in markedly increased MDM2 transcription and an enhanced risk of tumor formation in both germ line and somatic tumors (19). Additionally, at least 40 different alternatively and aberrantly spliced MDM2 transcripts have been characterized, many of which are known to have oncogenic potential or are differentially expressed in certain cancer types (20 -22). MDM2-A, -B, -C, -D, and -E, for example, have been shown to be overexpressed in cancers of the bladder and ovary compared with normal tissues and are capable of transforming NIH3T3 murine fibroblasts (20 -22). Emerging as another important player in p53 inactivation and in tumorigenesis is the E3 ligase Pirh2 (11). Like MDM2, Pirh2 is overexpressed in primary tumors (23,24) and therefore, through p53 inhibition, may very well have oncogenic potential. In contrast to MDM2, however, very little is known about Pirh2 structure and function. Here we report the identification and characterization of two novel isoforms of Pirh2 that we have named Pirh2B and Pirh2C. Both Pirh2B and Pirh2C lack critical regions of the Pirh2 RING finger domain but are able to interfere with p53 function.
Reverse Transcription-PCR and Southern Blotting-Cells were harvested and total RNA was extracted with TRIzol rea-gent (Invitrogen) per the manufacturer's instructions. The cMaster RT Plus PCR System (Eppendorf) or Superscript III RT kit (Invitrogen) was used for first strand full-length Pirh2,  Pirh2B, and Pirh2C cDNA synthesis and cDNA amplification  as per the manufacturer's instructions. Primer sequences for  Pirh2 RT-PCR experiments are found in supplemental Table 1.
Sequencing of Pirh2B and Pirh2C RT-PCR-For Pirh2B, a 2-l aliquot of an RT-PCR from the HCT116 human colon cancer cell line was directly subcloned into the pCR2.1 vector using a TA cloning kit (Invitrogen) as per the manufacturer's instructions. Pirh2C was TA cloned similarly, using 2 l of RT-PCR product amplified from MDA-MB-231 human breast cancer cell RNA. Potential positive bacterial colonies were selected by blue/white screening and grown out, and their plasmids were confirmed by digestion and sequencing.
Transfections-All transfections were performed using Lipo-fectamine2000 reagent (Invitrogen) as per the manufacturer's instructions with the exception of luciferase experiments, which used Mirus TransIt (Mirus, Madison, WI). For GFP-Pirh2 protein expression analysis, HEK293T cells were transiently transfected for 6 h on 60-mm plates. For GFP-Pirh2 localization experiments, HeLa cells were transiently transfected with 2 g of pEGFP-Pirh2 constructs for 6 h on LabTek II chamber slides. Both HEK293T and HeLa cells were then washed with medium following transfection and allowed to express for ϳ16 h.
Immunostaining-HCT116 cells growing on LabTek II slides were transiently transfected with 0.5 ng/well pCEP4-MycPirh2A, -B, or -C or pCEP4-MycMDM2 expression vectors. Approximately 24 h after transfection, cells were fixed with 4% paraformaldehyde for 20 min, followed by methanol at Ϫ20°C for 10 min and 10% goat serum for 3 h at room temperature. Cells were labeled with anti-Myc primary antibody (1:400; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), followed by fluorescein isothiocyanate-conjugated anti-mouse secondary antibody (1:500; Molecular Probes). DNA was counterstained with 4Ј,6-diamidino-2-phenylindole nuclear dye. Cells were analyzed with an Olympus AX 70 fluorescent microscope, and photographs were captured using a digital camera.
Lentiviral shRNA Production and Transduction-pLKO.1 lentiviral vectors expressing shRNAs against Pirh2 and MDM2 were purchased from Open Biosystems (Huntsville, AL). Five constructs were obtained for each and analyzed by Western blotting for their knockdown efficiency. Lentivirus was produced in 293T (2 ϫ 10 6 in 100-mm plates) cells by transient transfection of shRNA construct together with psPAX2 packaging and pMD2.G envelope constructs (Addgene, Cambridge, MA), using Lipofectamine2000. The supernatants were harvested at 24 and 48 h and centrifuged to remove any non-adherent 293T cells. Next, HCT116 cells were transduced with virus carrying Scramble, Pirh2, or MDM2 target in medium supplemented with 8 g of Polybrene at a multiplicity of infection of ϳ0.75. The medium was changed 36 h postinfection and every subsequent 48 h with fresh medium supplemented with 2 g/ml puromycin.
Western Blot Analysis-Cells were harvested and lysed on ice for 20 min in buffer containing 20 mM HEPES (pH 7.4), 2 mM EGTA, 50 mM ␤-glycerophosphate, 1 mM dithiothreitol, FIGURE 1. Amino acid sequence of Pirh2 isoforms. A, an amino acid sequence alignment of full-length Pirh2 and isoforms using the MAFFT version 6.5 protein sequence alignment program (57) (available on the World Wide Web). The Pirh2 RING-H2 domain (residues 145-186) is boxed, and critical Zn 2ϩ -coordinating residues are indicated by asterisks. B, genomic organization of Pirh2 isoforms. Shown is a schematic illustration of the Pirh2 gene structure and the intron-exon makeup of full-length Pirh2 and its isoforms. C, DNA sequences illustrating the splicing of the Pirh2B and -C isoforms. The introduced in-frame stop codon is boxed, and splice junctions are indicated with arrows. 1 mM Na 3 VO 4 , 5 mM NaF, 1% Triton X-100, 10% glycerol, 10 g/ml aprotinin, 2 g/ml pepstatin A, 2 g/ml chymostatin, 2 g/ml leupeptin, 0.1 g/ml okadaic acid, and 400 M phenylmethylsulfonyl fluoride. Extracted protein samples were resolved on 6 -12% SDSpolyacrylamide gels and transferred to nitrocellulose membrane (Bio-Rad). Anti-Pirh2 (BL588) antibodies were purchased from Bethyl Laboratories (Montgomery, TX); anti-GFP was from Roche Applied Science; anti-Myc (9E10), anti-p53 (DO-1, Pab1801, FL-393), anti-MDM2 (SMP14), anti-ubiquitin, and normal mouse and rabbit IgGs were from Santa Cruz Biotechnology; anti-␤-actin was from Sigma; anti-HA was from Covance (Princeton, NJ); and anti-GST was from GE Healthcare. Horseradish peroxidase-conjugated secondary antibodies against mouse and rabbit IgG were obtained from Vector Laboratories (Burlingame, CA). True-blot horseradish peroxidase-conjugated secondary antimouse IgG antibodies used for Western blotting in immunoprecipitation experiments were purchased from eBioscience (San Diego, CA).
Immunoprecipitations-5 ϫ 10 5 H1299 cells were plated on 100-mm plates for ϳ48 h before being transfected with 10 g of total DNA of either pSR␣-Pirh2A and pCEP4-Myc-F.L.Pirh2/Pirh2 isoforms (in a 1:1 molar ratio) or with pCMV-p53 and pCEP4-Myc-Pirh2 isoforms (in a 1:1 molar ratio) for 6 h. After 16 h, harvested cell pellets were lysed for 25 min on ice in RIPA buffer (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.5% Nonidet P-40 plus inhibitors). Approximately 750 g or 1 mg of protein was incubated overnight on a rotator with either 5 g of anti-HA, 5 g of anti-p53 (Pab1801), or 5 g of anti-Myc antibodies, as indicated. Protein complexes were pulled down using 50 l of magnetic protein G Dynabeads (Invitrogen) according to the manufacturer's instructions and analyzed by Western blotting.
GST Pull-down Assay-Purified GST fusion proteins were added to 700 -1500 g of HCT116 p53 (ϩ/ϩ) or p53 (Ϫ/Ϫ) cell lysates in RIPA buffer and incubated overnight on a rotator with 30 l of glutathione-Sepharose beads at 4°C. For direct interaction experiments, 20 g of purified GST-MDM2 was incubated with 10 l of in vitro translated p53, Pirh2A, -B, or -C (Promega, Madison, WI). Bead-bound protein complexes were pulled down at 1000 ϫ g for 5 min at 4°C. The bound beads were then washed three times with the RIPA lysis buffer, denatured at 95°C in SDS-loading buffer, and analyzed by SDS-PAGE.
In Vivo Ubiquitination Assay-3 ϫ 10 5 H1299 cells were plated on 100-mm plates for ϳ48 h before being transfected with 1 g of pCMV-p53 and 2 g of pcDNA3.1-FLAGubiquitin in combination with 2 g of either empty pCEP4, pCEP4-Myc-MDM2, or pCEP4-Myc-Pirh2 constructs for 6 h. After 20 h, cells were treated with 25 M MG-132 for an additional 4 h. Cells were harvested in 200 l of 1% SDS-Tris-buffered saline and lysed by boiling twice for 5 min with vigorous vortexing. 400 l of 1.5% Triton X-Tris-buffered saline was added, and the lysate was centrifuged for 20 min at 4°C. 50 g of lysate was resolved on 8% SDS-polyacrylamide gels and analyzed by Western blotting as described above.
For p53 ubiquitination assays, 5 g of either GST-bound GST or GST-Pirh2 isoforms were incubated for 120 min at 30°C in 50-l reactions as described above with either 1 g of GST-p53 or 2.5 l of in vitro translated p53 (Promega). As indicated, fractions of each reaction were either analyzed directly by 8% SDS-PAGE or immunoprecipitated with 1 g of anti-p53 (FL-393) and 20 l of Protein G beads (Santa Cruz Biotechnology) for 4 h at 4°C in RIPA buffer before being resolved on an 8% SDS-polyacrylamide gel and analyzed by Western blotting.
Luciferase Reporter Assay-Luciferase reporter assays were performed as previously described (25,26). Briefly, HCT116 cells were plated in 6-well plates at ϳ60% confluence and transfected for 12-16 h with 1 g of total DNA using TransIT reagent (Mirus). Cells were harvested following an additional 24-h incubation, and luciferase activity was measured with a luminometer (LUMAT LB 9507; Berthold Technologies).
Pirh2 Stability Assay-1.5 ϫ 10 5 H1299 cells were plated on 60-mm plates for ϳ48 h before being transfected with 1 g of pCEP4-Myc-Pirh2 construct for 6 h. After ϳ12 h, cells were treated with 25 g/ml cycloheximide for the indicated times and then harvested on ice and analyzed using standard Western blot procedures.

Identification and Characterization of Pirh2
Variants-Data base searches revealed the presence of several ESTs corresponding to human Pirh2 but with sequences deviating from those encoding the full-length protein (11,27). Using a computer-based approach, we categorized these ESTs relative to their sequence alignments with full-length Pirh2. The first group contained ESTs corresponding directly to full-length Pirh2 (hereafter referred to Pirh2A; GenBank TM accession numbers BG481510, BU166379, and BC047393), the second group of ESTs (BF102905, BI913695, BC015464, BP262622, and W01424) encoded a Pirh2 with a potential deletion in the center (this group was designated Pirh2B), and the third group potential C-terminal truncation (designated as Pirh2C). We sequenced one representative EST from each group, and the deduced amino acid sequence analysis revealed that the Pirh2B (BI913695) variant lacked 27 base pairs encoding 9 amino acids (positions 171-179) within the Pirh2 RING-H2 finger ubiquitin ligase domain (Fig. 1A). The resultant 252-amino acid protein has a predicted molecular mass of 29 kDa. The third Pirh2 iso- The sites of exon splice junctions for each primer are denoted by arrows. C, RT-PCR analysis of Pirh2B expression and Pirh2A-specific controls in HCT116 human colon cancer cells using reverse primers described in B. RT-PCR was performed as described under "Experimental Procedures." D, an amino acid sequence alignment of human (Homo sapiens) Pirh2B with Pirh2B from the Rhesus monkey (Macaca mulatta), chimpanzee (Pan troglodytes), and dog (Canis familiaris) using the MAFFT version 6.5 protein sequence alignment program (57) (available on the World Wide Web). The Pirh2 RING-H2 domain (residues 145-186), including Pirh2B deleted amino acid residues, is boxed, and critical Zn 2ϩ -coordinating residues are indicated by asterisks. E and F, Pirh2C mRNA expression in various human cancer cell lines. E, a schematic illustration of RT-PCR primer design. The RT-PCR primer pair illustrated generated a product of 560 bp in length. F, RT-PCR and analysis of prostate and lung cancer cell lines. JCA-1 cells were previously considered as prostate cancer cells; now they are believed to represent bladder cancer. RT-PCR was performed as described under "Experimental Procedures." form, Pirh2C (BQ429858), encodes a 179-amino acid protein with a predicted molecular mass of ϳ21 kDa. This isoform is substantially smaller than the full-length protein (isoform A) due to lack of amino acids 180 -261, which, like the Pirh2B deletion, includes several residues within the Pirh2 RING finger domain. Particularly, for Pirh2B, this includes the deletion of the Zn 2ϩ -coordinating residue Cys 172 and, for Pirh2C, Cys 183 and Cys 186 (Fig. 1A).
To gain further insight into the molecular basis for the generation of these isoforms, we compared the nucleotide sequences of Pirh2B and Pirh2C with the genomic sequence and organization of the Pirh2 gene (27). Pirh2 is composed of nine coding exons, four of which contribute to amino acids of the Pirh2 RING domain (27). Nucleotide sequences analysis revealed that the 9-amino acid RING finger deletion within Pirh2B corresponds exactly to the 27-base pair Pirh2 exon 7, and thus, Pirh2B could arise as the result of alternative splicing (Fig. 1, B and C). By contrast, Pirh2C appears to involve a second potential splice donor site, which we have named donor site 2 (Fig. 1C). Similar to the first donor site (donor site 1) identified by Beitel et al. (27), donor site 2 also has a nearly perfect 5Ј intron splicing motif (A (64%) G (73%) ͉ G (100%) T (100%) -A (62%) A (68%) G (84%) ). Donor site 2 usage results in the introduction of 28 "intronic" base pairs added between coding exons 7 and 8 and gives rise to Pirh2C by the formation of a premature in-frame stop codon (Fig. 1C). Fig. 1, B and C, shows the genomic organization of the various Pirh2 isoforms as well as alternative donor sites.
Pirh2B and Pirh2C Expression in Various Cell Types-Data base information on ESTs corresponding to Pirh2B and -C showed that the cDNAs of these novel isoforms were present in different cDNA libraries independently generated from multiple different human tissue sources. These findings suggested that the cDNAs corresponding to Pirh2B and -C isoforms appeared to be authentic and not the result of cloning or sequencing artifacts. Next, we sought to experimentally validate the existence of these isoforms, and to that end, their expression at mRNA and protein levels was evaluated in human cancer cell lines. To detect their expression at mRNA levels, RT-PCR analyses were performed. In the case of the Pirh2B isoform, a primer was designed to anneal specifically to the Pirh2 exon 6 and Pirh2 exon 8 splice junction unique to the Pirh2B transcript. The forward primer was designed to correspond to a region common to all isoforms. This RT-PCR strategy was planned to ensure amplification of only Pirh2B transcripts and not those of other isoforms (Fig. 2, A and B). RNA extracted from HCT116 human colon cancer cells was subjected to RT-PCR using this strategy, and as shown in Fig. 2C, Pirh2B is indeed expressed in HCT116 cells. As positive controls, Pirh2A-specific reverse primers encompassing the Pirh2 exon 6-exon 7 or exon 7-exon 8 splice junctions were utilized. As expected, these primers also generated products of predicted size (Fig. 2C). The RT-PCR-amplified Pirh2B-specific products were also subcloned into a cloning vector and sequenced to confirm the Pirh2B sequence. We also performed additional data base searches and noted the existence of sequences derived from other species predicted to encode the Pirh2B isoform. For example, Pirh2B-specific sequences were noted in tissues derived from the Rhesus monkey (GenBank TM accession number XP_001100858), chimpanzee (XP_001154449), and dog (XP_855704) (Fig. 2D), further indicating that Pirh2B is a bona fide isoform.
We noted that, unlike for Pirh2B isoform, there were fewer ESTs corresponding to the Pirh2C isoform. Therefore, we sought to verify the existence of Pirh2C by investigating its expression in a larger pool of human cell lines. A Pirh2C-specific reverse primer was designed to anneal to the 28 "intronic" base pairs we have found to be unique to Pirh2C (Fig. 2E). The forward primer was for a region common to all isoforms. RNAs extracted from prostate (ALVA31, DU145, and PC-3), bladder (JCA-1), lung (A549 and H1299), colon (DLD1, HCT15, HCT116, HT29, RKO, and SW480), and breast (HS578T, MCF-7, MDA-MB-231, MDA-MB-468, and T47D) cancer cell lines were subjected to RT-PCR. Our results indicated that Pirh2C was expressed in all cell lines tested. Representative results indicating Pirh2C expression in prostate, bladder, and lung cancer cell lines are shown in Fig. 2F (top). For controls, the same RNA samples were used to also detect Pirh2A as well as ␤-actin as (Fig. 2F, bottom). , and -C protein species in human cancer cell lines. Immunoprecipitations were performed with 2 g of anti-Pirh2 antibody or normal rabbit IgG as a negative control. The same blot was probed simultaneously for all three isoforms; however, shown above is a shorter exposure for Pirh2A and -B, and a longer exposure for Pirh2C. To show comparable protein input, equal amounts of supernatant from immunoprecipitations were subjected to Western analysis (WB) and probed for ␤-actin, as shown.
It is of note that a slightly larger product of ϳ650 bp was also amplified under these conditions, suggesting the existence of yet another novel isoform containing a portion of intron 7 (Fig.  2F, top). Efforts are currently under way in our laboratory to identify and characterize this potential variant. For additional controls, HEK293T cells were transiently transfected with either empty vector or vectors expressing Pirh2A or Pirh2C, and their RNA was extracted and subjected to RT-PCR. As expected, Pirh2C-transfected cells generated an abundant amount of Pirh2C, whereas in vector-and Pirh2A-transfected cells, only endogenous HEK293T Pirh2C was amplified (Fig.  2F). These results strongly suggested the amplified products to be the Pirh2C isoform. Subcloning and sequencing of these amplified products confirmed their identity to indeed be Pirh2C.
Next, the expression of Pirh2B and Pirh2C at the protein level was investigated using RKO and HCT116 human colon cancer cells and H1299 human lung cancer cells. Western blotting revealed the existence of two closely migrating bands, and because Pirh2A and Pirh2B differ only by 9 amino acids or ϳ1 kDa in molecular mass, the upper and lower bands are consistent with their molecular masses and could potentially represent Pirh2A and Pirh2B, respectively (Fig. 3A). As shown in Fig. 3A, longer exposure of the same blot revealed the presence of a ϳ21 kDa band migrating at the calculated molecular mass for Pirh2C and identical in mass to exogenously expressed Pirh2C (data not shown). To further demonstrate that each isoform is expressed at the protein level, endogenous Pirh2 was immunoprecipitated from RKO, HCT116, and H1299 cell lines and analyzed by Western blot. These results, shown in Fig. 3B, clearly show bands consistent in size with those of Pirh2A and Pirh2B. Similar to direct Western blot results, upon longer film exposure, a band migrating at the predicted molecular mass of Pirh2C was also present (Fig. 3B). Collectively, these results indicate that both Pirh2B and Pirh2C are bona fide isoforms that are expressed at both the mRNA and protein levels in various cell lines representing multiple human tissues.
Subcellular Localization of Pirh2 Isoforms-The nuclear and nucleolar subcellular localization of Pirh2A (full-length) is an important factor influencing Pirh2 function and stability (27,28). Despite studies describing Pirh2 localization, however, the exact location of a putative nuclear localization signal has not been determined. Therefore, as a first step to gain some insight into the function(s) of Pirh2B and Pirh2C and how they may differ from the full-length Pirh2, we generated expression con- FIGURE 4. Subcellular distribution of GFP-tagged Pirh2 isoforms. A, a schematic illustration of full-length Pirh2 (Pirh2A) and Pirh2 isoform GFPtagged expression constructs. Numbers represent the amino acid residues corresponding to the Pirh2 RING domain. B, Western blot (WB) was performed using an anti-Pirh2 antibody to detect exogenous GFP-Pirh2 and endogenous Pirh2 expression. C, HeLa cells were transfected with expression vectors carrying GFP or GFP-Pirh2 using Lipofectamine2000. Cells were washed with PBS, fixed with 4% paraformaldehyde, and incubated with 4Ј,6-diamidino-2phenylindole (DAPI) to stain nuclei. D, HCT116 cells seeded on Lab-Tek II slides were fixed and processed for immunostaining with anti-Myc specific antibodies or normal mouse IgG and fluorescein isothiocyanate-conjugated secondary antibodies, as described under "Experimental Procedures." Nuclei were stained with 4Ј,6-diamidino-2-phenylindole nuclear dye. C and D, cells were analyzed with an Olympus AX 70 fluorescent microscope, and photographs were captured using a digital camera. structs in which GFP is tagged to the N terminus of Pirh2A, Pirh2B, and Pirh2C (Fig. 4A). Western blot analysis using anti-Pirh2 antibody was performed on lysates from HeLa cells trans-fected with the constructs to confirm the expression and correct size of the GFP-tagged Pirh2 (Fig. 4B). As shown in Fig. 4C, GFP-Pirh2A displays diffuse nuclear and cytoplasmic localiza-  tion, which is in agreement with previous reports (24,28). Two distinct patterns are seen for GFP-Pirh2A, including a diffuse localization similar to GFP alone (Fig. 4C, top) and a more commonly observed diffuse localization with a speckled or punctate pattern (bottom). Subcellular distribution of Pirh2B and Pirh2C was similar to that of Pirh2A, since both novel isoforms also compartmentalized to the nucleus and the cytosol (Fig. 4C). To verify that the GFP tag did not interfere with the localization of Pirh2A, -B, or -C, immunofluorescence staining was performed on HCT116 cells expressing Myc-tagged Pirh2 isoforms. As seen in Fig. 4D, similar results were obtained. Hence, these results suggest that the region most proximal to the Pirh2 C terminus, including a region of its RING domain, may not determine its localization.
Pirh2B and Pirh2C Differ in Their Ability to Interact with Pirh2A-It has previously been reported that the RING E3 ligase MDM2 is capable of homodimerizing with itself through RING-RING domain interactions (29 -31) as well as with alternatively spliced MDM2 isoforms (32,33). Pirh2 is a RING finger ligase, and whether it can also oligomerize via its RING domain has not been investigated. We have identified two novel Pirh2 isoforms that have alterations in their RING finger region; therefore, we sought to investigate whether full-length Pirh2 (Pirh2A) is capable of oligomerization with itself or with other novel isoforms. To do this, expression constructs containing HA-tagged Pirh2A were co-transfected with vectors expressing Myc-Pirh2A, -B, or -C into H1299 human lung cancer cells. The representative results, shown in Fig. 5A, demonstrate that HAtagged Pirh2A does indeed interact with Myc-Pirh2A. Likewise, the Pirh2B isoform also interacts with Pirh2A. Interestingly, the C terminus-deficient Pirh2C was incapable of interacting with Pirh2A (Fig. 5A). To further validate these findings, the reciprocal experiment was performed whereby the same lysate was immunoprecipitated with antibodies against the Myc epitope. As shown in Fig. 5B, identical results were obtained. These findings suggest that unlike MDM2-MDM2 interactions, an intact RING domain is not required for Pirh2 self-association and that the region responsible for Pirh2-Pirh2 interactions probably lies within the C terminus of this protein.
Pirh2B and Pirh2C Cannot Autoubiquitinate and Have Altered Protein Stability-In addition to facilitating substrate ubiquitination, many E3 RING domain-containing proteins, such as MDM2, can also perform autoubiquitination, thereby targeting themselves for proteasomal degradation (8,34). It has been demonstrated that even a single mutation at the critical residues within the RING domains of either MDM2 or Pirh2 abolishes intrinsic ubiquitin ligase activity (8,24,27). Therefore, we found it relevant to investigate the ligase activity of Pirh2B and Pirh2C, which lack Cys 172 or Cys 183 and Cys 186 Zn 2ϩ -coordinating residues, respectively. To assess this possibility, we performed in vitro ubiquitination assays using recombinant GST or GST-Pirh2 proteins (Fig. 6, A and B). As shown in Fig. 6, C and D, Pirh2A was capable of autoubiquitination, whereas Pirh2B and Pirh2C were not, and that is probably due to the absence of essential residues in the their RING domains.
These results demonstrate that Pirh2B and Pirh2C lack intrinsic ubiquitination activity and cannot autoubiquitinate. Next, we sought to investigate stability of both isoforms, and to FIGURE 7. p53 physically interacts with Pirh2A, -B, and -C isoforms. A, immunoprecipitation (IP) of Myc-Pirh2 proteins. Wild-type p53 and Myc-Pirh2 constructs were co-transfected in a 1:1 molar ratio into H1299 cells and immunoprecipitated with anti-Myc antibody or normal mouse IgG as a negative control to detect p53-Pirh2 interactions. B, immunoprecipitation of p53. Lysate was immunoprecipitated with anti-p53 (Pab1801) antibody or normal mouse IgG as a negative control to detect p53-Pirh2 interactions. A and B, the relative expression of p53 and Myc-tagged Pirh2 proteins for each immunoprecipitation was confirmed by Western blot (WB) analysis (Input, right). C, GST pull-down of endogenous p53. Lysate from p53 (ϩ/ϩ) HCT116 cells was incubated with GST, GST-Pirh2A, GST-Pirh2B, or GST-Pirh2C, and Pirh2-p53 binding was assessed as described under "Experimental Procedures." Lysate from p53 (Ϫ/Ϫ) HCT116 cells was incubated with GST-Pirh2A as a negative control for p53. p53 expression was confirmed by Western blot analysis (Input, right). A-C, to confirm comparable protein input, equal amounts of supernatant from each experiment were subjected to Western analysis and probed for ␤-actin, as shown. Blots were sequentially probed with p53 (DO-1), Myc, GST, and ␤-actin antibodies, as indicated. that end, protein stability assays were performed. Myc-Pirh2 proteins were expressed, and cells were treated with cycloheximide to inhibit protein synthesis. A representative of multiple experiments shown in Fig. 6, E and F, reveals that both Pirh2B and Pirh2C have enhanced protein half-life compared with the full-length Pirh2 protein (Pirh2A). These findings therefore show that the stability of Pirh2 isoforms Pirh2B and Pirh2C was enhanced compared with full-length Pirh2, and that differences in protein stability are due, in part, to their inability to target themselves for proteasomal degradation.
Pirh2B and Pirh2C Differ in Their Ability to Interact with the Tumor Suppressor p53-Pirh2 has been shown to be an important negative regulator of the tumor suppressor p53; it interacts with p53 and promotes p53 ubiquitination and degradation (11). We next sought to investigate what effects, if any, Pirh2B and Pirh2C might have on p53 function. To do this, we first evaluated the ability of p53 to bind to these Pirh2 isoforms. p53-null H1299 cells were co-transfected with vectors expressing wild-type p53 and the Pirh2 isoforms, and their lysates were subjected to immunoprecipitation. The representative results, shown in Fig. 7A, demonstrate that the Pirh2B isoform clearly interacted with wild type p53, and consistent with previous reports (11), Pirh2A and p53 also displayed mutual interac-tions. Pirh2C and p53 also reproducibly displayed mutual interactions; however, such interactions appeared to be rather weak (Fig. 7A) (data not shown). The opposite experiment was also performed whereby immunoprecipitations were conducted with antibodies against Myc-tagged Pirh2 rather than against p53. As shown in Fig.  7B, these experiments yielded similar results; however, interactions between p53 and Pirh2C were not seen under these conditions, further suggesting that p53-Pirh2C interactions were weak. As an additional means to confirm the results of coimmunoprecipitation experiments, lysates from isogenic p53 (ϩ/ϩ) or p53 (Ϫ/Ϫ) HCT116 human colon cancer cells were incubated with purified recombinant GST-Pirh2A, -B, and -C proteins or GST alone and analyzed by GST pull-down assays. Fig. 7C shows the results of these experiments, demonstrating that the endogenous p53 from the wild-type p53-expressing HCT116 cells interacts with recombinant GST-Pirh2A and the Pirh2B isoforms but not with Pirh2C. While this manuscript was in revision, a structural study by Sheng et al. (35) found that Pirh2 associated with p53 via contacts with the Pirh2 N and C termini. Using various synthetic deletions, one of which corresponded to the region deleted in Pirh2C, the authors concluded that primary binding is dependent upon a C-terminal region of full-length Pirh2 and that the Pirh2 N terminus harbors only weak p53 binding potential (35). Their results appear to correlate well with our findings and provide an explanation for the weak p53 binding that we have noted for C-terminally deleted Pirh2C.
Pirh2B and Pirh2C Can Down-regulate p53 Protein Levels and Promote Intracellular p53 Ubiquitination-Our results so far indicate that neither Pirh2B nor Pirh2C exhibits intrinsic ubiquitin ligase activity (Fig. 6, C and D), and Pirh2C appears to only weakly interact with endogenous or exogenously expressed p53 (Fig. 7). Interestingly, previous studies have indicated that MDM2 can still negatively regulate p53 stability, localization, and transcriptional activity independently of its ubiquitination function (36 -38). Therefore, we investigated the effect of Pirh2B or Pirh2C on p53 protein levels. To that end, wild-type p53-expressing HCT116 cells were transfected with Pirh2A, Pirh2B, or Pirh2C, and endogenous p53 levels were analyzed by Western blotting. A representative experiment shown in Fig. 8A clearly indicates that, as expected, Pirh2A negatively regulates endogenous p53 protein levels. FIGURE 8. Pirh2A and Pirh2B and C isoforms negatively regulate p53 protein levels and promote intracellular p53 ubiquitination. A, endogenous, wild-type p53-expressing HCT116 colon cancer cells were transiently transfected with either 2.5 or 5.0 g of empty pCEP4 vector or pCEP4-Myc Pirh2A, -B, or -C vector, as indicated. Western blotting for p53 was performed using anti-p53 (DO-1) antibody. B, a graphic illustration of one representative experiment showing that Pirh2 isoforms down-regulate p53 protein levels. Relative p53 protein levels for 2.5 and 5.0 g of Pirh2-transfected cells were determined by normalizing p53 to ␤-actin and plotted relative to 2.5 and 5.0 g of vector-transfected p53 levels, respectively. Densitometry analysis was performed using Amersham Biosciences ImageQuant TL v2003.03. C, H1299 cells were transiently transfected with Myc-tagged Pirh2 or MDM2, p53, and FLAG-tagged ubiquitin, as indicated. Cells were harvested, and cell extracts were prepared as described under "Experimental Procedures." p53 ubiquitination was detected using anti-p53 (DO-1) antibody. A nonspecific band detected by the DO-1 antibody is denoted by an asterisk. Exogenous Myc-Pirh2 and endogenous MDM2 expression was confirmed by Western blot (WB) analysis (Input, bottom right).
Interestingly, both Pirh2B and, to a lesser extent, Pirh2C also down-regulated p53 to levels comparable with that mediated by full-length Pirh2 (Fig. 8A). Fig. 8B shows quantitative results of the representative experiment shown in Fig. 8A. The same experiment was performed one more time, and similar results were obtained.
Because Pirh2B and Pirh2C, like Pirh2A, are both capable of negatively regulating p53 protein stability, we next sought to investigate whether Pirh2B and Pirh2C could affect the intracellular ubiquitination of p53 despite their demonstrated lack of intrinsic ligase activity. H1299 cells, which do not harbor endogenous p53, were transfected with constructs expressing Myc-tagged Pirh2 or MDM2, p53, and FLAG-tagged ubiquitin and, after 20 h of expression, were treated with the proteasome inhibitor, MG-132. As expected, cells transfected with either Pirh2A or MDM2 showed markedly increased levels of higher molecular weight p53, indicative of p53 ubiquitination (Fig.  8C). Interestingly, cells transfected with either Pirh2B or Pirh2C also demonstrated notably increased levels of p53 ubiquitination (Fig. 8C), albeit Pirh2A and Pirh2B appeared to be more efficient at promoting p53 ubiquitination than did Pirh2C. Although isoform C is more stable and abundant, it is still less efficient in promoting intracellular ubiquitination of p53. These findings therefore indicate that "ubiquitinationdead" isoforms of Pirh2 are still capable of promoting the covalent attachment of ubiquitin to p53.
To investigate the possibility that Pirh2B and Pirh2C may engage endogenous Pirh2A to promote p53 ubiquitination, we utilized an RNA interference approach to study the effect of the various Pirh2 isoforms on p53 in cells where endogenous Pirh2 expression had been efficiently knocked down. Several Pirh2specific shRNA vectors were tested for their ability to knock down Pirh2 expression in HCT116 cells (data not shown). Fig.  9A shows Pirh2 shRNA 10849, which was the most efficient and was therefore used for subsequent experiments. Of note, Pirh2 shRNA 10849 is specific to nucleotides 168 -188 of the 786-bp Pirh2 open reading frame and targets Pirh2A, -B, and -C for knockdown. Therefore, it was necessary to mutate Pirh2 expression vectors in order to successfully reintroduce them into Pirh2 shRNA 10849-infected HCT116 cells. The mutated constructs were tested in the Pirh2 10849 knockdown cells, and as shown in Fig. 9B, Pirh2A, -B, and -C levels were comparable with those seen in Scramble negative control shRNA-infected cells after 24 h of expression. Exogenously introduced Pirh2A, -B, and -C were expressed at physiologically relevant levels in that they were comparable with the endogenous Pirh2A levels in HCT116 Scramble cells (data not shown).
The effect of Pirh2A, -B, and -C on p53 protein levels in these cells where endogenous Pirh2 had been considerably knocked down was then investigated. A representative experiment shown in Fig. 9C clearly demonstrates that, as expected, Pirh2A remained capable of negatively regulating p53 protein stability in Pirh2 shRNA cells and at levels comparable with those seen in Scramble cells. Interestingly, both Pirh2B and, to a lesser degree, Pirh2C also down-regulated p53 in Pirh2 shRNA cells at levels similar to those seen in both Scramble cells (Fig. 9C) and in previous experiments (Fig. 8A). Our results indicated that expression of exogenous Pirh2 isoforms consistently decreased endogenous p53 levels, although the extent of p53 down-regulation was modest in certain instances, which could be due to altered transfection efficiencies in Pirh2 knocked down cells. Nevertheless, these findings indicate that Pirh2B probably does not utilize full-length Pirh2 (Pirh2A) to facilitate p53 ubiquitination. Our earlier finding that Pirh2C is incapable of interacting with full-length Pirh2 (Fig. 5, A and B), coupled with the maintained ability of Pirh2C to negatively regulate p53 in the absence of Pirh2A, strongly suggests that Pirh2C also does not appear to require Pirh2A for p53 ubiquitination.
Pirh2 Isoforms Enhance MDM2 Levels and Physically Associate with MDM2-Western blot analysis of MDM2 expression revealed that endogenous MDM2 levels are increased in cells FIGURE 9. Pirh2A, -B, and -C isoforms negatively regulate p53 protein levels in endogenous Pirh2 knockdown background. A, HCT116 human colon cancer cells were infected with either Scramble shRNA or Pirh2 shRNA (Pirh2 10849) expressing lentivirus. Reduction of endogenous Pirh2 levels was shown by Western blot analysis using anti-Pirh2 antibody, as described under "Experimental Procedures." Note that Pirh2A and Pirh2B were not resolved on this 8% SDS-polyacrylamide gel. B, Western blot analyses were performed to demonstrate Pirh2A, -B, and -C expression of mutant pCEP4-Myc-Pirh2 expression constructs (Pirh2mut3) in Pirh2 shRNA-infected HCT116 cells. C, Scramble or Pirh2 shRNAinfected HCT116 colon cancer cells were transiently transfected with 5.0 g of empty pCEP4; pCEP4-Myc Pirh2A, -B, or -C mutant vector; or pCEP4-MycMDM2, as indicated, and endogenous p53 levels were evaluated. Western blotting for p53 was performed using anti-p53 (DO-1) antibody.
transfected with Pirh2A, -B, or -C, indicating that Pirh2 might act as a positive regulator of MDM2 protein levels (Fig. 8C). To further elucidate whether a potential relationship between MDM2 and Pirh2 exists, immunoprecipitation experiments were carried out to investigate possible interactions between MDM2 and the various Pirh2 isoforms. Results of a represent-ative experiment in Fig. 10A demonstrate that exogenously expressed Pirh2A, -B, and -C each interact with endogenous MDM2. The same blot was also probed for p53, and as seen in Fig. 10A, each Pirh2 isoform also interacted with endogenous wild-type p53. As an additional control, cells were transfected with Myc-tagged p53, and as is seen in Fig. 10A, endogenous   FIGURE 10. A, immunoprecipitation (IP) of endogenous MDM2 by Myc-Pirh2 isoforms. HCT116 cells, which express endogenous MDM2, were transfected with Myc-tagged Pirh2A, -B, or -C or Myc-p53, and their lysates were subject to immunoprecipitation with anti-Myc-conjugated agarose beads as described under "Experimental Procedures." In the case of Myc-p53-transfected cells, less input protein was used for the immunoprecipitation to minimize the strength of the MDM2 signal during Western blot (WB). The relative expression of Myc-tagged Pirh2 and p53 proteins was confirmed by Western blot analysis (Input, bottom panel). Detection of endogenous p53 required more sensitive enhanced chemiluminescent reagent; therefore, the portion of the membrane containing Myc-p53 was removed to avoid overexposure of endogenous p53 bands. B, GST-MDM2 pull-down of Pirh2 isoforms. Lysate from p53 (Ϫ/Ϫ) or p53 (ϩ/ϩ) HCT116 cells was incubated with GST or GST-MDM2, and Pirh2-MDM2 binding was assessed as described under "Experimental Procedures." Exogenous Pirh2 and endogenous p53 and MDM2 expression was confirmed by Western blot analysis of input protein (Input, bottom panel). A and B, to confirm comparable protein input, equal amounts of supernatant from immunoprecipitations were subjected to Western analysis and probed for ␤-actin, as shown. Blots were sequentially probed with MDM2, p53, Myc, and ␤-actin antibodies, as indicated. C, Pirh2A, -B, and -C interact with MDM2 directly. Equal amounts of in vitro translated p53, Pirh2A, -B, or -C were incubated with either GST or GST-MDM2 and Pirh2-MDM2 binding was assessed as described under "Experimental Procedures." pcDNA3.1-MycPirh2 was used for in vitro translation reactions; therefore, doublets probably represent translation from start site upstream from the Myc tag as well as the native Pirh2 start site. D, schematic representation of GST-MDM2 constructs used for GST-MDM2 Pirh2 pull-downs. Regions corresponding to p53 binding as well as the MDM2 acidic, zinc finger, and RING finger domains are illustrated. Numbers shown below correspond to the amino acid residues of each MDM2 domain. E, mapping the region(s) or MDM2 responsible for MDM2-Pirh2 interactions. Lysate from HCT116 cells was incubated with GST or the various GST-MDM2 constructs, and the binding of endogenous Pirh2 was assessed as described under "Experimental Procedures." Note that Pirh2A and Pirh2B were not resolved on this 8% SDS-polyacrylamide gel. The same blot was probed for p53 (DO-1) as a positive control.
MDM2 very strongly interacted with its well established target, p53.
Pirh2 and MDM2 interactions have not been reported previously. As an additional means to confirm the novel finding that MDM2 and each Pirh2 isoform are capable of mutual interactions, GST pull-down assays were performed. As demonstrated in Fig. 10B, Pirh2A, -B, and -C transiently expressed in HCT116 cells each clearly interacted with GST-MDM2. Moreover, interactions between MDM2 and Pirh2 isoforms appear to be p53-independent, since the two proteins associate in p53 (Ϫ/Ϫ) cells in addition to wild-type p53-expressing HCT116 cells. As a positive control, the same blot was also probed for p53, and as expected, MDM2 bound p53 very strongly (Fig.  10B). To elucidate whether MDM2-Pirh2 interactions are direct, we performed GST pull-down assays by incubating in vitro translated Pirh2A, -B, or -C with purified GST-MDM2. As seen in Fig. 10C, Pirh2A, -B, and -C each interacted with GST-MDM2, as did p53, a well established direct binding partner of MDM2.
To gain additional insight into the interactions between MDM2 and the Pirh2 variants, next, we sought to identify the regions of MDM2 involved in interactions with Pirh2. To that end first, we generated several deletion variants of MDM2 fused with GST and purified each one. Fig. 10D shows a schematic illustration of the GST-MDM2 deletion variants. Next, we carried out GST pull-down assays to identify the regions of MDM2 involved in interactions with Pirh2. As shown in Fig. 10E, endogenous Pirh2 from HCT116 cells interacted most strongly with the MDM2 variant 151-491, which contains the MDM2 central acidic domain as well as the N-terminal zinc and RING finger domains. Endogenous Pirh2 also bound MDM2 variant 1-275 and MDM2 variant 276 -491, which retain the MDM2 acidic and zinc finger domains, respectively. However, as seen in Fig. 10E, loss of either the acidic or zinc finger domain resulted in a diminished ability to bind Pirh2 versus either full-length MDM2 or MDM2 variant 151-351. Similar results were seen for exogenously expressed Pirh2A, -B, or -C; however, in each case, deletion of the zinc finger domain appeared to have a more dramatic negative effect on Pirh2 binding than did the acidic domain deletion (supplemental Figs. 1-3). Moreover, neither MDM2 variant 1-150 nor MDM2 variant 351-491, is capable of interacting with Pirh2.
Pirh2B and Pirh2C Cannot Directly Ubiquitinate p53-To examine whether Pirh2B and Pirh2C are acting intracellularly as E3 ubiquitin ligases despite disruptions in their RING domains, in vitro ubiquitination experiments were done with in vitro translated p53 as a substrate. The results of these experiments, as shown in Fig. 11A, confirm that Pirh2A, but not Pirh2B and Pirh2C, can ubiquitinate p53 in vitro. To verify that the ubiquitination seen was, in fact, p53 and not Pirh2A autoubiquitination, in vitro ubiquitination assays were performed on GST-p53, followed by denaturation and immunoprecipitation of GST-p53 by a polyclonal anti-p53 antibody (FL-393; Santa Cruz Biotechnology). As shown in Fig. 11B, a similar pattern of p53 ubiquitination was seen when p53 was isolated from the reaction by immunoprecipitation.

Pirh2B and Pirh2C Negatively Regulate p53 Transcriptional
Activity-Full-length Pirh2 (Pirh2A) has been shown to negatively regulate the transcriptional activation function of p53 (11). Given that Pirh2A, Pirh2B, and Pirh2C isoforms promote p53 ubiquitination intracellularly, we also tested their effects on p53 transcriptional activity to investigate what effects Pirh2B and Pirh2C expression will have on p53 function. To examine the effect of Pirh2B and Pirh2C on the transcriptional function of endogenous p53, we introduced into wild-type p53 expressing HCT116 cells the PG13-luciferase reporter construct that harbors 13 copies of the wild-type p53 binding site (39,40). As shown in Fig. 12, Pirh2A interferes with p53 induction of the PG13 promoter, as does MDM2, the well established negative regulator of p53. Interestingly, p53-mediated PG13 activity is also inhibited by Pirh2B and Pirh2C (Fig. 12). Consistent with their effects on PG13 promoter activity, Pirh2B and Pirh2C as well as Pirh2A also interfere with p53-mediated transcriptional activation of the PUMA promoter luciferase construct, a key FIGURE 11. A, GST-immobilized Pirh2 was tested for its ability to ubiquitinate in vitro translated p53 in the presence of E1, E2 (UbcH5b), and His-tagged ubiquitin, as indicated. Ubiquitinated GST-p53 was detected using anti-ubiquitin antibody. B, GST-immobilized Pirh2 was tested for its ability to ubiquitinate purified GST-p53 in the presence of E1, E2 (UbcH5b), and ubiquitin, as indicated. Following the in vitro ubiquitination, a fraction of each reaction was denatured in 1% SDS and subjected to immunoprecipitation using anti-p53 antibody (FL-393). Ubiquitinated p53 was detected using anti-ubiquitin antibody. WB, Western blot.

DISCUSSION
A large body of evidence indicates that the deregulation of p53 by proteins and pathways implicated in p53 ubiquitination appears be a principal means by which p53 is inactivated in human cancers. Pirh2 is one candidate protein that is known to negatively regulate p53 through ubiquitination and is overexpressed in primary tumors (23,24). Despite initial information implicating it in negative regulation of p53, much remains to be elucidated as to how Pirh2 functions to inhibit p53 and how it may contribute to tumorigenesis.
To this end, we report here the identification of two novel Pirh2 isoforms, Pirh2B and Pirh2C, that each lack a portion of their central RING domain as a result of differential mRNA splicing. Expression of Pirh2B and Pirh2C isoforms is not restricted to a single cell or tissue type, since our results indicate that the Pirh2B and Pirh2C isoforms are expressed in various cells representing multiple tissues. Furthermore, ESTs corresponding to these novel isoforms are present in several cDNA libraries generated from various tissues. Moreover, we have found that Pirh2B expression appears to be conserved across species, since sequences highly homologous and predicted to encode Pirh2B are expressed in tissues derived from the Rhesus monkey, chimpanzee, and dog. Pirh2C may also be similarly conserved; however, we did not find homologous sequences encoding Pirh2C in other species. Because only two ESTs representing the human Pirh2C homologue were identified (BQ429858 and BX459245), it may be possible that sequences representing Pirh2C from other species have not yet been reported in the data bases.
The splicing mechanisms generating these isoforms remain unclear; however, the alternate splicing of an E3 ligase targeting p53 is not without precedence. For example, MDM2 variants are known to arise from the alternate usage of either a 5Ј-untranslated region or a p53-responsive region within MDM2 intron 1 (44 -47). Pirh2B is generated by the alternative splicing of exon 7. Although the consensus p53 response element (p53RE) for Pirh2 has been mapped to Pirh2 intron 3 (11), very little is known regarding regulation of Pirh2 gene expression. Further studies are therefore needed to elucidate the exact mechanisms controlling the splicing events responsible for the generation of this isoform. The splicing of Pirh2C, unlike Pirh2B, appears to occur at a second, previously undefined 5Ј donor site (donor site 2) 28 base pairs into Pirh2 intron 7. By Northern blot analysis, Pirh2A (full-length Pirh2) has been demonstrated to be the predominantly expressed isoform (11), and therefore donor site 1 is likely to be the stronger donor site. However, our results clearly demonstrate that Pirh2C exists naturally, perhaps through "leaky" utilization of donor site 2. A number of factors could influence the differential splicing of Pirh2C, such as lack of definition between the two canonical donor sites, or it could be due to comparatively weak elements within Pirh2 exons or introns that provide information to the mRNA splicing machinery (48). Further in-depth studies are also needed to elucidate the molecular mechanisms underlying the generation of this isoform. Future studies are also needed to explore the relative expression of these isoforms in matched normal and cancerous tissue specimens and would prove helpful in further elucidating their function in human malignancies.
Although Pirh2A, Pirh2B, and Pirh2C showed similarity in their subcellular distributions, each protein differed in terms of protein turnover, as demonstrated by protein stability assays and Pirh2 stable cell lines. This is due, in part, to their inability to autoubiquitinate, which we have confirmed in vitro. Ubiquitination of Pirh2C by Pirh2A also probably does not occur due to lack of physical interactions between the two proteins. Of note, as this manuscript was in preparation, a study reported that artificial deletion of the C-terminal end of the Pirh2A isoform abolished Pirh2-Pirh2 interactions (49), a finding that validates our results with the naturally occurring Pirh2C variant. Furthermore, our results for Pirh2 are unlike those established for MDM2, which is understood to form MDM2 dimer complexes through RING-RING interactions (29 -33). An alternative yet unverified possibility could also be that enhanced Pirh2B and Pirh2C stability is the result of lysine deletions to which ubiquitin can be covalently attached. Of the 14 lysines present on Pirh2A, Pirh2B lacks Lys 178 , whereas Pirh2C does not harbor Lys 239 and Lys 241 . Hence, Pirh2B and Pirh2C stability could be enhanced due to the lack of critical lysine residues targeted for ubiquitination.
Unexpectedly, we discovered that Pirh2B and Pirh2C both negatively regulate p53 protein stability possibly by facilitating the intracellular (in vivo) ubiquitination of p53, despite our findings that each lacked intrinsic ubiquitination activity. The means by which this occurs is currently unclear; however, our findings appear to rule out the possibility that either functions through Pirh2A, since knockdown of Pirh2A by RNA interference had no apparent effect on their ability to negatively regu-FIGURE 12. Pirh2A and Pirh2B and -C isoforms inhibit p53 transcriptional activity. Wild-type p53-expressing HCT-116 human colon cancer cells growing in 6-well plates were transiently transfected as described under "Experimental Procedures" with PG13 or PG13 with p53 plus Pirh2 or MDM2. Induction of luciferase activity from the PG13-Luc construct was plotted relative to PG13 and p53 alone. The average of three experiments Ϯ S.E. is shown. late p53 stability. The fact that Pirh2C does not interact with Pirh2A yet still facilitates p53 ubiquitination intracellularly only further supports the notion that these isoforms may mediate p53 ubiquitination inside the cell independent of full-length Pirh2. Still, because Pirh2 knockdown is less than 100%, it remains a possibility that the remaining Pirh2A expressed is sufficient for Pirh2B or Pirh2C-mediated p53 ubiquitination.
Our results indicate that expression of Pirh2B and Pirh2C, as well as Pirh2A (full-length Pirh2), enhance the levels of endogenous MDM2. In addition, our results have, for the first time, demonstrated that endogenous MDM2 and each of the various Pirh2 isoforms are capable of mutual interactions. Our findings show that interactions between MDM2 and full-length Pirh2, Pirh2B, and Pirh2C are direct, which is another novel finding of our study. We have delineated the domains on MDM2 responsible for interactions with Pirh2 to be both the central acidic domain (amino acids 223-274) and the N-terminal zinc finger domain (amino acids 305-322). Pirh2 and its isoforms can be grouped with a growing number of proteins that associate with the MDM2 acidic and zinc finger regions, including YY1 (50), pRb (51), ARF (52), and ribosomal proteins L5, L11, and L23 (53).
Interestingly, loss of the MDM2 RING finger in MDM2 variant 1-350 decreased Pirh2 binding; however, the association remained stronger than binding between Pirh2 and full-length MDM2. MDM2 variant 351-491, which contains only the RING finger domain of MDM2, possessed no Pirh2 binding capability. Reciprocally, the truncated RING domains of Pirh2B and Pirh2C did not have any apparent effect on their association with MDM2. Collectively, these results therefore suggest that unlike interactions between numerous other RING domain-containing proteins, interactions between Pirh2 and MDM2 are not mediated through their RING domains (54 -56). Notably, no association was seen for MDM2 variant 1-150, which contains the MDM2 p53 binding domain, indicating that Pirh2 and p53 bind MDM2 in disparate regions of MDM2.
Importantly, our findings demonstrate that p53 does not facilitate the association between MDM2 and Pirh2, based on the fact that MDM2 and the Pirh2 isoforms exhibit direct interactions in the absence of p53. In addition to their mutual crosstalk in the context of p53, our results therefore also raise the notion that Pirh2 and MDM2 may fulfill coordinate p53-independent functions within the cell, and future investigations are needed to explore this possibility. Collectively, these data strongly suggest that Pirh2A, -B, and -C may promote p53 ubiquitination indirectly through an MDM2-mediated mechanism. Surprisingly, however, in HCT116 MDM2 shRNA-infected knockdown cells, Pirh2A, -B, and -C were each capable of negatively regulating p53 protein stability despite loss of MDM2 expression (data not shown). Whether this is due to the fact that a certain amount of MDM2 is still expressed is an issue that warrants additional study. With the very recent discovery of several additional p53-specific E3 ligases, the potential for Pirh2A, -B, or -C to regulate still other ubiquitin ligases through these mechanisms also requires examination.
In addition to our novel findings characterizing a relationship between Pirh2 and MDM2, the notion that Pirh2 and its isoforms function at least in part through MDM2 is also supported by a report showing that co-expression of full-length Pirh2 and MDM2 has an additive effect on p53 ubiquitination (11). Given that the two proteins bind disparate regions of p53 (11,35), this outcome could be the result of paralleled ubiquitination of p53, and this could be true for Pirh2A that harbors intrinsic E3 ligase activity. However, our results indicate that both Pirh2B and Pirh2C lack ubiquitin ligase activity and Pirh2C has a diminished ability to bind p53. It is therefore possible that these isoforms (as well as Pirh2A) could positively regulate MDM2 ligase activity toward p53 through a mechanism that involves Pirh2-MDM2 interactions. A hypothetical model showing this possibility is illustrated in Fig. 13. Whether Pirh2A, -B, or -C increases MDM2 protein levels or its p53specific E3 ligase function via their mutual interactions with MDM2 could be an intriguing possibility. Although speculative, each Pirh2 isoform could potentially promote MDM2 ligase activity by enhancing MDM2 protein stability post-translationally. Such a possibility would provide a reasonable explanation for increased MDM2 despite Pirh2-mediated inhibition of p53 transcriptional activity. Still further, Pirh2A, -B, or -Cmediated alterations in a signaling pathway upstream to MDM2 represent yet another potential mechanism by which Pirh2 may regulate MDM2 protein levels. Future in-depth studies examining the interactions between these two important p53-specific E3 ligases and the means by which they appear to cooperate to promote p53 ubiquitination will help to shed more light on this intriguing issue. Likewise, exploration into possible interactions between Pirh2 and other more recently discovered E3 ligases is also needed.
Our results also indicate that p53 transcriptional activity is inhibited by both Pirh2B and Pirh2C, in addition to Pirh2A. Of note, our report is the first to demonstrate that two naturally occurring Pirh2 isoforms with disrupted RING domains are capable of promoting p53 ubiquitination and inhibiting p53 transcriptional function. Our results are in agreement with those of a previous study (11) showing that a truncated Pirh2 protein designed to lack its entire RING finger domain also inhibits p53-mediated transcription. Because it is well established that p53 ubiquitination interferes with its transcriptional activity, it is likely that Pirh2B-and Pirh2C-mediated increases in p53 ubiquitination/degradation are at least in part responsible for the observed inhibition. However, in the case of MDM2, in addition to ubiquitination/degradation, it is well accepted that p53 inhibition also involves a physical interaction at the N terminus of p53 to block its association with the basal transcription machinery (36 -38). Because Pirh2 is known to bind to the central DNA binding and tetramerization domains of p53 (11,35), it is possible that the full-length Pirh2A and its novel isoforms may interfere with the ability of p53 to bind to DNA (4) and hence regulate p53 function independent of their ubiquitin ligase activities. Further in depth studies will help to better elucidate how Pirh2A and the two novel isoforms elicit their ubiquitin-dependent and/or independent effects on the tumor suppressor p53.