The von Hippel-Lindau Tumor Suppressor Stabilizes Novel Plant Homeodomain Protein Jade-1*

The von Hippel-Lindau disease gene (VHL) is the causative gene for most adult renal cancers. However, the mechanism by which VHL protein functions as a renal tumor suppressor remains largely unknown. To identify low occupancy VHL protein partners with potential relevance to renal cancer, we screened a human kidney library against human VHL p30 using a yeast two-hybrid approach. Jade-1 (gene forApoptosis and Differentiation inEpithelia) encodes a previously uncharacterized 64-kDa protein that interacts strongly with VHL protein and is most highly expressed in kidney. Jade-1 protein is short-lived and contains a candidate destabilizing (PEST) motif and plant homeodomains that are not required for the VHL interaction. Jade-1 is abundant in proximal tubule cells, which are clear-cell renal cancer precursors, and expression increases with differentiation. Jade-1 is expressed in cytoplasm and the nucleus diffusely and in speckles, where it partly colocalizes with VHL. VHL reintroduction into renal cancer cells increases endogenous Jade-1 protein abundance up to 10-fold. Furthermore, VHL increases Jade-1 protein half-life up to 3-fold. Thus, direct protein stabilization is identified as a new VHL function. Moreover, Jade-1 protein represents a novel candidate regulatory factor in VHL-mediated renal tumor suppression.

The von Hippel-Lindau disease gene (VHL) is the causative gene for most adult renal cancers. However, the mechanism by which VHL protein functions as a renal tumor suppressor remains largely unknown. To identify low occupancy VHL protein partners with potential relevance to renal cancer, we screened a human kidney library against human VHL p30 using a yeast two-hybrid approach. Jade-1 (gene for Apoptosis and Differentiation in Epithelia) encodes a previously uncharacterized 64-kDa protein that interacts strongly with VHL protein and is most highly expressed in kidney. Jade-1 protein is short-lived and contains a candidate destabilizing (PEST) motif and plant homeodomains that are not required for the VHL interaction. Jade-1 is abundant in proximal tubule cells, which are clear-cell renal cancer precursors, and expression increases with differentiation. Jade-1 is expressed in cytoplasm and the nucleus diffusely and in speckles, where it partly colocalizes with VHL. VHL reintroduction into renal cancer cells increases endogenous Jade-1 protein abundance up to 10-fold. Furthermore, VHL increases Jade-1 protein half-life up to 3-fold. Thus, direct protein stabilization is identified as a new VHL function. Moreover, Jade-1 protein represents a novel candidate regulatory factor in VHL-mediated renal tumor suppression.
VHL gene defects are responsible for both von Hippel-Lindau disease (1) and most sporadic clear-cell renal cancers (2)(3)(4)(5). VHL is therefore the most commonly affected renal cancer gene. Clear-cell renal cancer is also the most malignant VHL 1 disease lesion, which suggests VHL protein exerts strongest tumor suppressor activity in renal proximal tubules, which are the precursor cells of this common malignancy.
VHL functions most important for renal tumor suppression remain unclear. Based on the well-recognized association of specific VHL mutations with partial VHL disease phenotypes, a likely hypothesis is that VHL missense mutations may disrupt some, but not necessarily all, VHL functional pathways. For example, VHL mutations that prevent HIF ubiquitination and therefore promote HIF and vascular endothelial growth factor overexpression are precisely those that correlate with hemangioblastoma development (35)(36)(37)(38). Although HIF overexpression contributes importantly to renal cancer pathogenesis (39,40), no VHL biochemical function or protein interaction has been found that correlates with renal cancer risk, leaving unresolved the full role of VHL in renal tumor suppression (41).
To identify molecules potentially important in the pathogenesis of clear-cell renal cancer, we screened an adult human kidney library with human VHL p30 using a yeast two-hybrid approach. We have named the gene encoding a novel, strong VHL-interactor as Jade-1 (gene for Apoptosis and Differentiation in Epithelia). Jade-1 protein is short-lived and most highly expressed in kidney. It contains a candidate PEST degradation domain and plant homeodomain (PHD) motifs, which are not required for the VHL interaction. Jade-1 protein is directly stabilized by VHL protein, which is a new VHL function. Preliminary results suggest Jade-1 is also growth suppressive. Jade-1 may therefore participate in VHL-mediated renal tumor suppression.
Yeast Two-hybrid Analysis-An adult human kidney cDNA library in pB42AD (CLONTECH) was screened against human VHL aa 2-213 in the LexA-expressing, inducible yeast expression vector pGilda (CLON-TECH), according to the manufacturer's instructions (19). pB42AD library clones initially positive by growth on deficient medium and by X-gal staining were rescued and individually screened against both human and mouse VHL in pGilda in yeast before sequencing. Interaction strength in yeast was categorized by the amount of time postcotransformation required for yeast colonies to appear blue on X-gal plates. Clones positive at 24 h were designated as the strongest (3ϩ) interactors. Those positive at 48 or 72 h were designated as 2ϩ or 1ϩ interactors, respectively.
Cell Lines and Transfection-293T17 human embryonic kidney cells and HT1080 fibrosarcoma cells were generously provided by Drs. Z. Luo (Boston University School of Medicine) and R. Widom (Boston University School of Medicine), respectively. 786-O renal cancer cells and HeLa cells were obtained from American Type Culture Collection (Manassas, VA). These lines were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, glutamine, and penicillin-streptomycin (Invitrogen). HA-VHL 786-O (42) and A498 renal cancer lines (44) were generously provided by Drs. W. Kaelin (Dana-Farber Cancer Institute) and W. Krek (Friedrich Miescher Institut, Basel), respectively. 786-O (18,19) and UMRC6 (45) stably-transfected renal cancer lines were described previously. Stable renal cancer lines were maintained as above in 0.2-0.4 mg/ml G418 (Invitrogen). SV40 T antigen-transformed mouse proximal tubule (MPT) cells, generously provided by Dr. M. Loghman-Adham (University of Utah) (46), were grown in the above medium supplemented with 5 units/ml interferon ␥ at 33°C. MPT cells were differentiated without interferon at 37°C for 10 days. Primary culture mouse proximal tubule cells were generously provided by Dr. W. Lieberthal (Boston University School of Medicine) (47). Cells were transfected using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions or by calcium phosphate precipitation (48).
Antibodies-To generate Jade-1 antisera, 2 New Zealand White rabbits were injected with the same carboxyl-terminal 20 residue Jade-1 peptide amino terminally linked to keyhole limpet hemocyanin. Antiserum 1 was affinity-purified using the same peptide coupled to Sepharose (Alpha Diagnostics International, San Antonio, TX). Human VHL antiserum was generously provided by Dr. R. Burk (Albert Einstein College of Medicine, Bronx, NY) (34). Human VHL monoclonal antibody and FLAG M5 monoclonal antibody were from Pharmingen and Sigma, respectively. Sp1, glutathione S-transferase (GST), and HA antisera were from Santa Cruz Biotechnology, as were fluorescein isothiocyanate (FITC)-and cyanine-3 (cy3)-tagged anti-rabbit and anti-mouse secondary antibodies. Horseradish peroxidase-linked anti-rabbit and anti-mouse IgGs were from Bio-Rad.
Immunoprecipitation and Western Blotting-Cultured cells were lysed in lysis buffer (Tris 50 mM, pH 7.6, NaCl 150 mM, EDTA 30 mM, Triton X-100 0.5%,) with Complete protease inhibitor (Roche Molecular Biochemicals), precleared with protein A-agarose bead (Santa Cruz Biotechnology), and mixed with antibody at excess (18,19). Complexes were pelleted with protein A-agarose, washed with lysis buffer, eluted with sample buffer, and analyzed by SDS-PAGE. Immunoblotting was performed as previously described using the antibodies above. For multitissue Western analysis, tissues were minced and homogenized with a Teflon pestle in lysis or KETN buffers (100 mM KCl, 1 mM EDTA, 10 mM Tris, pH 7.5, 0.1% Nonidet P-40) containing protease inhibitor. Human tissue lysates were obtained from Genotech (St. Louis, MO). Relative intensities of positive bands were assessed by densitometry using Image 1.62 (National Institutes of Health). Backgrounds were subtracted to assign densitometry values.
Northern Analysis-Total RNA was prepared using RNAzol (Tel-Test) according to the manufacturer's instructions. A multitissue Northern blot was probed as recommended (CLONTECH) using a 1-kb 5Ј Jade-1 coding sequence fragment. Northern analysis was otherwise performed as described earlier (19).
Immunocytochemistry-Cells were grown on Chamber slides (Lab-Tek) and fixed with 1:1 methanol and acetone for 2 min at room temperature. Immunodetection was performed with affinity-purified anti-Jade-1 serum or VHL monoclonal antibody followed by FITC-or cy3-conjugated anti-rabbit or anti-mouse antibodies. Images were obtained by fluorescence microscopy (Nikon Optiphot) and digital imaging (RT Color camera, Diagnostic Instruments).
Metabolic Labeling-Cells were starved in cysteine (Cys)-and methionine (Met)-free medium for 1-2 h, then fed deficient medium containing 100 Ci/ml 35 S-Met and 35 S-Cys (EasyTag express, PerkinElmer Life Sciences), followed by medium containing 100-fold excess unlabeled Met and Cys as chase for times indicated. Labeling times were 15 min or 1 h for Jade-1 in transiently transfected cells or 2.5 h for endogenous Jade-1 in renal cancer cells. Immunoprecipitations were carried out as above, and proteins were separated by SDS-PAGE. Correct identification of labeled endogenous Jade-1 was aided by control lanes of labeled untagged Jade-1 from transfected 293T17 cells immunoprecipitated with Jade-1 antiserum, with and without competitor peptide. Dried gels were subjected to autoradiography, and bands were quantitated by densitometry using Image 1.62. Detection of labeled endogenous Jade-1 required 3 weeks of autoradiography. Protein half-life was determined by log densitometry plotting. Alternatively, protein half-life was estimated based on densitometry results at the 0and 1-h chase times, using linear regression to identify the time point at which the initial signal strength would be halved. Errors are reported as Ϯ1 S.D.

RESULTS
The Jade-1 Gene and Protein-The human Jade-1 gene was identified as a particularly strong VHL-interacting clone in a yeast two-hybrid screen of human VHL p30. VHL is a low abundance intracellular protein that resides in multiple subcellular compartments (22,(31)(32)(33)(34). Thus, the yeast screen was performed to identify potentially rare or short-lived VHL protein interactions. An adult human kidney library was chosen because half of the mass of an adult human kidney is composed of proximal tubule cells. One million library clones in pB42AD yeast expression vector were screened against human VHL aa 2-213 in LexA-expressing, inducible yeast expression vector pGilda. Positive yeast colonies were confirmed by restreaking under double-selection conditions. Library clones rescued from confirmed positive colonies were tested individually for interaction with human VHL aa 2-213 and 55-143 as well as full-length mouse VHL in pGilda. VHL aa 55-143 were chosen as a minimum substrate-binding VHL beta domain. Jade-1 was one of 14 strongest interactors (3ϩ) in yeast of 40 different genes identified, because VHL-Jade-1-cotransformed colonies appeared blue at 24 h of incubation. Jade-1 also interacted convincingly with human VHL aa 55-143 (3ϩ) as well as mouse VHL (2ϩ) in yeast. VDU1 (10) was also recovered in the screen, supporting the validity of this approach for identifying bona fide VHL interactors. Additional confirmed positive interactors in yeast include four other novel genes, 17 known genes encoding a wide range of proteins, and 14 known genes encoding chaperones or chaperone-like proteins, which are common false-positives. As reported in another screen, VHL-binding protein-1 (3ϩ) and filamin (2ϩ) were also recovered (49), whereas other known VHL-interacting proteins were not.
GenBank TM BLAST searches indicate that the Jade-1 library clone contains the complete coding sequence of a novel, single gene (Fig. 1A). The 3570-nt clone was found in-frame with the B42 activation domain and has 178-nt 5Ј untranslated, 1527-nt coding, and 1831-nt 3Ј untranslated sequence, as well as a polyadenylation signal and poly(A)ϩ tail. The coding sequence is followed by several stop codons (Fig. 1A).
The deduced Jade-1 509 aa sequence (Fig. 1A) has a predicted 58.4-kDa mass and 5.25 isoelectric pH. It has two consensus mid-molecule PHDs (50), also known as leukemia-associated protein (51) or trithorax consensus domains (52), which are 50-to 70-aa C4HC3 zinc-binding motifs (Fig. 1, A and B). Alternatively, the second PHD may represent an extended PHD (53) and include aa 257-371. Jade-1 residues 5-28 comprise a strong candidate PEST domain (PESTfind score of ϩ11) (Fig. 1, A and B), which is a charged, unstructured region that promotes susceptibility to degradation (54). Jade-1 has no signal or transmembrane sequences. Candidate sites for N-glycosylation, myristoylation, and serine or threonine phosphorylation are shown, based on Prosite analysis, although the patterns found are short and not highly specific. No closely homologous proteins have been characterized. However, a transcript called E9 was identified in a differential screen of genes induced with apoptosis in a breast cancer line and is predicted to encode a closely related PEST-and PHD-containing protein (55).
To determine whether transfected Jade-1 encodes a protein of the anticipated size, the protein coding sequence was subcloned into CMV promoter-driven tagged and untagged expression vectors. A 64-kDa expressed protein is consistently seen in Western blots of Jade-1 transfected but not untransfected or control transfected cells (see Figs. 2A, 3A, and 3E), indicating production of Jade-1 protein that is just larger than the predicted 58.4-kDa molecular mass. Jade-1 may therefore be posttranslationally modified.
To confirm the identify of endogenous Jade-1 protein, immunoprecipitability and electrophoretic mobility of vector-expressed and endogenous human Jade-1 proteins were compared. Antisera were generated against a 20-aa peptide corresponding to the Jade-1 carboxyl terminus in two rabbits. By enzyme-linked immunosorbent assay, both antisera detect the immunizing peptide to 1:100,000 dilution. Both Jade-1 immune sera immunoprecipitate from 2 mg of cell lysates a similar prominent 64-kDa band, whereas neither preimmune serum does so, as assessed by SDS-PAGE and Jade-1 Western analysis ( Fig. 2A). 293T17 cells were also transiently transfected with an untagged human Jade-1 expression vector. An band identical in appearance was immunoprecipitable from only 250 g of transfected cell lysate ( Fig. 2A, far right lane), consistent with vector-expressed Jade-1. Similarly, tagged, transfected Jade-1 was identical in immunoprecipitability and appearance, although it was of higher molecular weight (data not shown). Endogenous Jade-1 (Fig. 2B) and transfected Jade-1 immunoprecipitation (data not shown) could also be completely blocked with the immunizing peptide. Thus, both endogenous and transfected Jade-1 protein can be immunodetected and immunoprecipitated in a highly specific manner. Moreover, these observations support the notion that the endogenous and transfected Jade-1 aa sequences are identical. Jade-1 antiserum 1 was used for subsequent experiments.
To examine Jade-1 protein distribution, Western analysis was performed on several tissues (Fig. 2C). The human and mouse Jade-1 proteins, 64 and 61 kDa, respectively, were by far most highly expressed in the kidney. Lower Jade-1 expression was also seen in human pancreas, liver, and heart and in mouse liver. Jade-1 was not readily detectable in human brain. In human cell lines, Jade-1 has been observed in HeLa, 293, and in multiple renal cancer cell lines, although at low levels (data not shown). Thus, although some difference in tissue distribution was found between human and mouse, the kidney was the major site of Jade-1 protein expression.
Jade-1 message and expression pattern were characterized with a human multitissue Northern blot (Fig. 2D). A 3.6-kb Jade-1 transcript, identical to the expected size, was most prominent in kidney. This message was also highly expressed in pancreas and skeletal muscle, but was found in all tissues tested with longer exposure (data not shown). The absence of Jade-1 protein in tissues, such as brain and skeletal muscle, where message was clearly expressed suggests that the primary level of Jade-1 control may not be at the RNA level. Another major 6-kb transcript was also seen with a distribution similar to Jade-1 (Fig. 2D). Thus, as observed for Jade-1 protein, Jade-1 message was most highly expressed in kidney, and a major band corresponds to the library clone.
To confirm that Jade-1 is expressed in renal proximal tubule cells, Western blots of kidney cortex and cultured mouse proximal tubule cells were performed. Kidney cortex is roughly 80% proximal tubule cells by mass. Jade-1 was prominently expressed in renal cortex (Fig. 2E, right lane) and had lower expression in renal medulla (data not shown). However, when mouse proximal tubule cells proliferated in primary culture, Jade-1 expression was greatly reduced, and the protein exhibited a slight reduction in molecular mass (Fig. 2E, left lane). These cells are at least 95% mouse renal proximal tubule cells (47). Jade-1 expression was also high in a temperature-sensitive SV40 T antigen transformed mouse proximal tubule (MPT) line and increases with differentiation (Fig. 2F). Together these observations indicate that Jade-1 is expressed in renal cancer precursors and that differentiation or perhaps quiescence increases Jade-1 expression.
Jade-1 and VHL Proteins Interact-To verify the strong VHL-Jade-1 interaction observed in yeast, binding was tested in transiently transfected mammalian cells. Coding sequence for Jade-1 and the other positive library clones was inserted into mammalian expression vector pFLAG-CMV2. HA-tagged VHL was cotransfected into 293T17 cells with FLAG-Jade-1 or other FLAG-tagged library clones, several of which are included for comparison. In these experiments, immunoprecipitation of HA-VHL coimmunoprecipitated much FLAG-Jade-1, less FLAG-clone 10, little FLAG-clone 4, and no FLAG-clone 5, as shown by anti-FLAG Western blotting (Fig. 3A, HA-IPs). Jade-1, clone 10, and clone 4 expression levels in whole cell lysates were comparable (Fig. 3A, lysates), which suggests the VHL interaction with Jade-1 is stronger than with clones 10 or 4. As shown, more than 30% of Jade-1 coimmunoprecipitated with VHL, based on comparison of input and detected amounts of protein. FLAG-clone 5 expression levels in whole cell lysates were undetectable in this experiment. Conversely, FLAG immunoprecipitation of the Jade-1 library clone and to a lesser extent library clone 10 robustly coimmunoprecipitated HA-VHL (Fig. 3B). In contrast, no interaction was observed with clones 4 or 5. Thus, Jade-1 exhibited a strong bidirectional interaction with VHL, whereas clone 10 had a modest bidirectional interaction. Clone 4 had a weak unidirectional interaction. A strong VHL-Jade-1 interaction was also observed in HeLa and HT1080 cells (data not shown). The Jade-1 interaction with VHL in mammalian cells was stronger than that of any other library clone tested. VHL therefore binds Jade-1 avidly and selectively. Because Jade-1 did not bind p53 (data not shown), its interaction with VHL is also selective.
To detect an interaction between VHL and endogenous Jade-1 protein, immunoprecipitations were performed in VHL transiently transfected 293T17 cells and in VHL stably transfected renal cancer cells. Following transient transfection of VHL and immunoprecipitation using a VHL antiserum, endog-enous Jade-1 was detectable by Western blotting, whereas Jade-1 was not visible following control immunoprecipitation (Fig. 3C). Conversely, immunoprecipitation with Jade-1 serum, but not preimmune serum, allowed detection of transiently transfected VHL by Western analysis (Fig. 3D). VHL stably transfected renal cancer cell lines provide similar results (data  -1 gene and protein. A, confirmation that the Jade-1 library clone encodes the same protein as endogenous (endog.) Jade-1. Immunoprecipitations (IPs) were performed on whole cell lysates (2 mg of protein total) from VHL stably transfected 786-O renal cancer cells or untransfected 293T17 cells, or on lysates (250 g of protein total) from 293T17 cells transfected with untagged Jade-1 (trfd. Jade). Preimmune (pre) or corresponding immune serum (post) from two different rabbits was used for immunoprecipitations, followed by Western analysis for Jade-1 with antiserum 1. B, the Jade-1 antiserum is specific for Jade-1 protein. Endogenous Jade-1 protein was immunoprecipitated (IP) from untransfected 293T17 cell lysates using Jade-1 antiserum in the absence or presence of the immunizing peptide (ϩpep). Western analysis was performed with Jade-1 antiserum. C, human (above) and mouse (below) multitissue Western analysis was performed using Jade-1 antiserum. Human (Hs) Jade-1 is 64 kDa, whereas the mouse (Mm) protein is 61 kDa (arrows). Skeletal muscle (sk. mm.), placenta (placnt.), pancreas (pancr.). D, Jade-1 human multitissue Northern analysis reveals the Jade-1 transcript at 3.6 kb, and possibly an alternatively spliced form at 6 kb. Abbreviations are as in C. E and F, high Jade-1 expression is associated with proximal tubule cell differentiation. E, mouse (Mm) kidney cortical tissue was finely minced and either lysed and tested immediately for Jade-1 expression by Western analysis (fresh) or subjected to 10 days primary culture (prim. cult.) (47) and tested. The upper band present in both lanes but more prominent in primary culture cells is nonspecific. F, Jade-1 Western analysis of a temperature-sensitive T antigen-transformed MPT cell line (46) grown under permissive, proliferating conditions (prolif.) or with partial differentiation (diff.) at 37°C for 10 days. Other FLAG-tagged library clones, FL-4, FL-5, and FL-10, or empty vector (FL-vec), are shown for comparison. B, cotransfections were performed as in A but were followed by FLAG immunoprecipitation and VHL Western blotting. C and D, endogenous Jade-1 binds transfected VHL. C, 293T17 cells were transiently transfected with VHL, and whole cell lysates were immunoprecipitated using a VHL antiserum or control (ctrl) GST antiserum and assayed for the presence of Jade-1 by Western blotting. D, lysates from VHL-transfected 293T17 cells were immunoprecipitated using preimmune serum (pre) or Jade-1 immune serum (post). VHL was detected using a VHL monoclonal antibody. E-G, the Jade-1 PHD regions are not required for interaction with VHL. E, expression levels of cotransfected Jade-1 (J) (upper panels) or VHL protein (V) (lower panels) as measured by Western blotting of the same whole cell lysates used for immunoprecipitations in F and G. In 293T17 cells, VHL or empty pFLAG-CMV2 (ev) was cotransfected with FLAG-or HA-tagged Jade-1 or truncations (see Fig. 1B for construct schematics), or with empty pCR3.1 uni HA (ev). Lower panels are independent immunodetections of the same blots probed in the upper panels. F, Jade-1 immunoprecipitations of cell lysates from E were followed by VHL immunoblotting. G, VHL immunoprecipitations of the same cell lysates were followed by Jade-1 immunodetection. not shown). These observations indicate that endogenous Jade-1 binds VHL.
To determine the Jade-1 regions responsible for the VHL interaction, HA-tagged Jade-1 truncations were generated lacking the amino terminus (and candidate PEST domain) (del1), the amino terminus and PHD regions (del2), or both PHDs alone as a double internal deletion (dd), as diagrammed in Fig. 1B. In whole cell lysates, transient expression of fulllength Jade-1 and the dd and del2 proteins was robust, whereas del1 expression was lower (Fig. 3E, upper panels). VHL expression was comparable in these cotransfections (Fig.  3E, lower panels). Jade-1 immunoprecipitation of these same cell lysates permitted coimmunoprecipitation of VHL with fulllength Jade-1, dd, and del1, but not with del2 (Fig. 3F). Likewise, VHL immunoprecipitation allowed coimmunoprecipitation of full-length Jade-1, dd, and del1, but not del2 (Fig. 3G). The VHL interaction with del1 did appear reduced, even taking into account lower del1 expression. As expected, immunoprecipitation in the absence of either partner did not show the interaction (Fig. 3, F and G, end lanes). Thus, these experiments demonstrate that the Jade-1 carboxyl terminus and PHD regions themselves are not absolutely required and that the amino terminus and the inter-PHD region in particular may be most important for interaction with VHL. These findings also support the specificity of the VHL-Jade-1 interaction.
Localization of Jade-1 and VHL-To determine subcellular compartments where Jade-1 might reside, cells were transiently transfected with HA-tagged Jade-1. In 293T17 cells, transiently transfected HA-Jade-1 gives a strong diffuse and speckled cytoplasmic signal (Fig. 4A). Occasionally, intense perinuclear Jade-1 fluorescence was seen, as shown. Cotransfected FLAG-VHL exhibited a nearly identical immunofluorescence pattern (Fig. 4B). Moreover, the Jade-1 and VHL proteins were almost completely colocalized (Fig. 4C). The HA antiserum and FITC-linked anti-rabbit secondary antibodies gave no detectable fluorescence with untransfected or vector transfected cells (data not shown), which indicates signal is highly specific for HA-tagged Jade-1 protein. VHL was detected with a monoclonal antibody and cy3-conjugated anti-mouse secondary antibody, a combination that also showed negligible background or bleedthrough fluorescence to the other fluorophore wavelength (data not shown). In MPT cells, transiently transfected HA-Jade-1 was seen in prominent nuclear speckles, in addition to the diffuse and speckled cytoplasmic pattern (Fig.  4D). Nuclear speckle localization has also been described for other PHD proteins (56,57).
Endogenous Jade-1 was localized using Jade-1 antiserum 1 that had been affinity-purified against the immunizing peptide. Several lines of evidence suggest this antibody is specific for human Jade-1 in immunofluorescence studies despite its recognition of several bands by Western blotting. First, the antiserum specificity for Jade-1 in immunoprecipitations was high, because no cross-reacting bands detectable by Western blot were immunoprecipitable (see Fig. 2, A and B). Second, although antiserum 1 readily detected 61-kDa mouse Jade-1 by Western blot as well as nonspecific bands, it could not immunoprecipitate any of these proteins (data not shown). Moreover, the mouse Jade-1 immunofluorescence signal was negligible at the antibody dilution used for human cells. These observations suggest strongly that the antiserum 1 Jade-1 immunofluorescence signal corresponds only to the highest affinity target, human Jade-1, and not any lower affinity, non-immunoprecipitable protein, such as mouse Jade-1.
Endogenous Jade-1 appeared in prominent nuclear speckles in 786-O (Fig. 4E) and A498 renal cancer cells (Fig. 4H). In cytoplasm, diffuse, filamentous, and speckled endogenous Jade-1 fluorescence was seen as well. Stably transfected VHL colocalized with endogenous Jade-1 in a subset of Jade-1-positive nuclear, perinuclear, and cytoplasmic speckles (Fig. 4, F-I). Endogenous Jade-1 and stably transfected VHL also colocalized diffusely in the nucleus of this cell. Jade-1-and VHLpositive cytoplasmic speckles were present in roughly equal abundance, however, only about 10% of each protein colocalized with the other (Fig. 4I). Some Jade-1-and VHL-positive cytoplasmic speckles appeared in ring-like groups, as indicated by the circles. Thus, VHL and Jade-1 colocalized in several compartments, although the colocalizing protein fractions were small, which suggests that the protein interactions may be dynamic.
VHL Stabilizes Jade-1 Protein Expression-To establish a biological relationship between VHL and Jade-1 protein, Jade-1 protein expression was analyzed in VHL-deficient renal cancer cell lines and compared with VHL-expressing stable derivatives by Western analysis and immunoprecipitation. In 786-O, UMRC6, and A498 renal cancer cells, endogenous Jade-1 protein expression is 3-to 10-fold higher in VHL stably transfected derivatives than in parental or empty vector lines (Fig. 5A, upper panels). Protein loading for each sample pair is comparable based on Ponceau S membrane staining (Fig. 5A,  lower panel). The UMRC6 cells have lower Jade-1 expression than 786-O and A498 cells; consequently, UMRC6 Jade-1 signal detection here required longer exposure. These results have been confirmed by Jade-1 immunoprecipitation (data not shown). Despite expressing even less wild-type VHL protein than many non-cancer renal lines, including 293T17 cells (data not shown), UMRC6 VHL cells still exhibited 3-fold increased Jade-1 expression. This observation supports the notion that elevated Jade-1 protein levels are not merely the result of VHL overexpression. Furthermore, increased Jade-1 expression was readily seen in other VHL-transfected 786-O stable lines in comparison with additional 786-O empty vector lines and was evident regardless of the degree of cell confluence (data not shown). These results establish a consistent biological relationship whereby the presence of VHL increases Jade-1 expression.
To initially assess how VHL might increase Jade-1 expression, Jade-1 was transiently transfected into 293T17 cells with and without VHL. One day following transfection the amount of transfected Jade-1 protein was largely unaffected by VHL cotransfection (Fig. 5B, far left panel). In contrast, VHL cotransfection substantially increased Jade-1 abundance 3 days post-transfection (Fig. 5B, right panels). In HeLa and HT1080 cancer cell lines in particular VHL dramatically increased Jade-1 protein. As controls, cotransfected empty vector or beta galactosidase did not increase Jade-1. In addition, VHL cotransfection did not increase the expression of p53 or VHLbinding proteins PKC and the C2H2 zinc-finger transcription factor Sp1 (Fig. 5C), supporting the notion that VHL specifically increases Jade-1 abundance. VHL-dependent increases in transfected Jade-1 suggest that the VHL effect on Jade-1 is not at the transcriptional or mRNA level, because Jade-1 gene control elements were not present and the regulatory elements in the different expression vectors were similar. Because increased Jade-1 expression was most notable by late post-transfection, we examined whether VHL affects Jade-1 protein stability.
To explore the mechanism whereby VHL increases Jade-1 abundance, pulse-chase metabolic labeling experiments were performed in renal cancer cell lines. 786-O cells stably transfected with wild-type VHL or an empty expression vector were pulsed with radiolabeled 35 S-Met and 35 S-Cys. Labeled Jade-1 was immunoprecipitated from 786-O cell lysates and subjected to SDS-PAGE, followed by autoradiography and densitometry.  Fig. 6A, in which the Jade-1 protein half-life is 41 min without VHL and 62 min with VHL. Similar results were obtained with A498 stable lines. The amount of labeled Jade-1 at the end of the 2.5-h labeling period was higher in VHL-expressing cells than VHL-deficient cells, most likely due to increased degradation of Jade-1 during labeling without VHL. This hypothesis is supported by the fact that the Jade-1 half-life was shorter than the labeling period itself, particularly in VHL-null cells. Furthermore, a similar discrepancy in Jade-1 abundance with and without VHL was seen at the 0-chase time point in transient transfections with 1-h labeling (Fig. 6B, right panel) but not with 15-min labeling (Fig. 6B, left  panel). Because relatively small fractions of the proteins interact (Fig. 4, F-I), VHL most likely stabilizes Jade-1 through a "hit-and-run" mechanism.
To model the in vivo protein stabilization mechanism, pulsechase metabolic labeling experiments were performed in 293T17 cells transiently transfected with Jade-1, with and without wild-type VHL, using 1-h labeling periods. Wild-type VHL cotransfection increased immunoprecipitable, radiolabeled Jade-1 at the 0-, 2-, and 4-h chase time points (Fig. 6B,  right panel). Labeled, coimmunoprecipitated VHL appeared in abundance at the bottom of the blot. Using 0-and 1-h chase time points from four experiments and estimating by linear regression, VHL increased cotransfected Jade-1 half-life from 39 Ϯ 3 to 106 Ϯ 14 (S.D.) min, similar to the VHL effect on endogenous Jade-1. A multipoint decay log plot provided similar transfected Jade-1 half-life results (Fig. 6C). The plotted line slopes differ by more than 3-fold with and without VHL (slope Ϫ0.326 versus Ϫ1.03, respectively), indicating VHL substantially increases the Jade-1 protein half-life. Shortening the labeling period to 15 min equalized immunoprecipitable labeled Jade-1 with or without VHL at chase time 0 (Fig. 6B,  left panel), supporting the notion that VHL does not increase Jade-1 protein synthesis. In fact, Jade-1 production appeared slightly decreased with cotransfected VHL, a reproducible finding consistent with VHL-mediated repression of a cotransfected promoter (18,19). Thus, these results support the endogenous Jade-1 data that VHL increases Jade-1 abundance primarily by prolonging Jade-1 protein half-life. DISCUSSION Jade-1 is a novel protein that interacts physically and functionally with the VHL tumor suppressor. No closely homologous protein has been characterized previously. Jade-1 is shortlived and contains a candidate PEST degradation domain and PHD motifs as major structural features. Jade-1 is most highly expressed in kidney and resides in cytoplasm and the nucleus, both diffusely and in prominent speckles. Jade-1 expression is high in differentiated proximal tubule cells, which are renal cancer precursors, and expression falls with proliferation. The strong VHL-Jade-1 interaction does not depend on intact PHD fingers, but may require the inter-PHD region and the PESTcontaining amino terminus. Most importantly, VHL increases both transfected and endogenous Jade-1 levels by directly stabilizing the protein, which represents a new VHL function. Preliminary results not presented here also suggest that Jade-1 is growth suppressive and may promote apoptosis. Thus, Jade-1 protein is a candidate regulatory molecule that may participate in VHL-mediated renal tumor suppression.
Jade-1 is a novel protein-coding gene. Initially, the Jade-1 nt secondary antibodies were used to localize Jade-1 in this single cell, shown in a background of untransfected cells. E, endogenous Jade-1 is found in nuclear speckles in renal cancer cells. Untransfected 786-O renal cancer cells were probed with affinity-purified anti-Jade-1 serum and FITC-labeled anti-rabbit secondary antibody. A single 786-O cell of typical appearance is shown. As in D, nuclear speckles are in the same visual plane as the nucleus. F-I, colocalization of endogenous Jade-1 with stably transfected VHL. Wild-type VHL stably transfected A498 cells were probed for both VHL and Jade-1. F, VHL was detected using anti-VHL monoclonal and cy3-labeled anti-mouse secondary antibodies. G, a merged view of the cell from F and H is shown. H, Jade-1 was detected using affinity-purified anti-Jade-1 serum and an FITC-labeled anti-rabbit secondary antibody. I, colocalization of endogenous Jade-1 and stably transfected VHL to cytoplasmic speckles. Higher magnification view of perinuclear region from G, with a 90°rotation counterclockwise. Nucleus is to the right. Circled regions highlight groups of speckles with Jade-1-VHL colocalization. B, VHL increases Jade-1 expression 3 days, but not 1 day, post-cotransfection. Cultured cell lines as indicated were transiently cotransfected with a Jade-1 expression vector and a second expression vector, either empty (Ϫ), or containing wild-type VHL (V) or ␤-galactosidase (Bgal). 1 or 3 days following transfection Jade-1 protein levels were determined by Western analysis. C, the VHL-mediated increase in expression is specific for Jade-1. FLAG-tagged p53 (FL-p53), FLAG-tagged protein kinase C (FL-PKCz), or transcription factor Sp1 were cotransfected with either vector alone (Ϫ) or wild-type VHL (V). FLAG or Sp1 Western blotting was performed on cell lysates harvested 3 days after transfection. and aa sequences were not found in GenBank TM . Subsequently, cDNA clones were deposited with predicted protein coding sequences identical to Jade-1, including "hypothetical protein" FLJ22479, which is represented by protein accession numbers NP_079176 (New Energy Development Organization (Japan) human cDNA sequencing project, Japan) and XP_033946 (NCBI annotation project). Protein FLJ14714 is identical to FLJ22479 and has been assigned accession number AK027620 (New Energy Development Organization (Japan) project, Japan). Another cDNA encodes a putative "unnamed protein product" with accession number BAB55239 (New Energy Development Organization (Japan) project, Japan). No protein expression or functional information accompanies these GenBank TM reports. Characterization of the human Jade-1 cDNA clone identified revealed the presence of a com-plete 3Ј-end. Expressed sequence tag sequences were reviewed to determine the Jade-1 initiation codon, and the MKF site was chosen because no clone with substantially more 5Ј sequence was found. Correct identification of the initiation codon is supported by the identical sizes of untagged, transfected Jade-1 protein and the endogenous, immunoprecipitable Jade-1 protein. Other Jade-1 transcripts exist as well, as suggested by the prominent 6-kb message on Northern analysis and a 95-kDa Jade-1 immunoreactive band on Western blotting (data not shown). Although such bands may reflect homologs or alternative transcripts, longer Jade-1 coding sequence clones have been deposited in public databases. For example, an 846-aa Jade-1 protein would be predicted based on combined overlapping Celera Genomics cDNA clone CT8385 and clone KIAA1807 (accession number AB058710) (Kazusa DNA Re- search Institute) (58), although their non-coding sequences are incomplete. The predicted 846-aa Jade-1 protein would have a mass near 95 kDa and would contain the two PHD regions but would lack the candidate PEST sequence or any additional recognizable domains. This larger protein also contains internally 13 of the 20 aa in the immunizing peptide, which still might be sufficient for immunoreactivity. Although alternative splicing may change interaction with and regulation by VHL or other functions, 509-aa Jade-1 may be the first member of a group of proteins stabilized by the VHL tumor suppressor.
Several observations suggest Jade-1 is growth suppressive or might participate in apoptosis. Stable Jade-1 overexpression in 786-O renal cancer cells so far has not been possible. We typically achieve 30 -40% success rates for stable expression of any transgene and 20% success rates with VHL, which is growth suppressive in these cells. However, 0 of 25 Jade-1-transfected, drug-resistant stable 786-O colonies exhibited increased Jade-1 expression. In addition, transient transfection of Jade-1 into 293T17 cells increases apoptosis by 50%, as measured both by Hoechst fluorescence and direct visualization and by tagged Annexin V binding and fluorescence-activated cell scanning. 2 This latter observation may explain the difficulty generating a Jade-1 stable cell line. In addition, the closely homologous E9 gene is increased 9-fold with induction of apoptosis in breast cancer cells (55). The predicted E9 polypeptide has a candidate PEST domain and two PHD regions, which further suggests such proteins may participate in apoptosis. As shown here, Jade-1 expression also correlates with differentiation of mouse proximal tubule cells and VHL status. The role of Jade-1 in renal cancer is unclear, but the differentiation or quiescence relationship is intriguing. Jade-1 expression may fall in any proliferative renal epithelial cell lesion, such as renal cysts, even in the absence of VHL gene defects. The prominent Jade-1 message in normal kidney and pancreas is also interesting, because both tissues develop cysts and cancers in VHL disease. Jade-1 may therefore be an important negative regulator of epithelial cell growth.
These data associate VHL with a PHD protein, which has not been described previously. The C4HC3 PHD finger binds two zinc ions and is structurally similar to the C3HC4 RING zincbinding domain. The PHD is a critically important motif, because it is lost or mutated in several diseases, including the acute leukemias, reviewed in Ref. 59, X-linked alpha thalassemia with mental retardation (60), and others. Many PHD proteins are transcription factors or cofactors and have chromatin modifying roles (50,(61)(62)(63)(64). Thus, by increasing Jade-1 protein, VHL might alter chromatin, which has not been previously suggested and might help explain its promotion of a renal epithelial differentiation program (26). A subset of PHD proteins is involved in apoptosis, such as requiem (65), ing1 (66), and perhaps E9 (55). Given the well-recognized roles of VHL in transcriptional control of gene expression (7,18) and protection against stress-induced apoptosis (28 -30, 67), PHD protein Jade-1 could be an important participant in these VHL pathways.
Direct protein stabilization by VHL is also a new finding. Although VHL promotes the ubiquitination and destruction of several interacting proteins (7)(8)(9)(10), no protein partner thus far is stabilized by VHL like Jade-1. Jade-1 is also stabilized in a highly specific manner, because this effect is not seen with VHL-interacting proteins Sp1 or PKC nor with p53. VHL might increase Jade-1 stability in several ways. Because the VHL-Jade-1 interaction appears transient, a "hit-and-run" mechanism appears most likely, which may involve a Jade-1 post-translational modification rather than VHL-dependent steric hindrance of Jade-1 degradation. Moreover, the Jade-1 PEST domain is a strong candidate degradation-susceptibility region, or degron. VHL might therefore alter PEST domain phosphorylation, for example, which is well known to affect protein stability. Although Jade-1 protein appears tissue-restricted, Jade-1 message is more widely expressed, suggesting that control of Jade-1 expression may occur predominantly at the protein level. Several proteins that do not interact with VHL are increased with VHL reintroduction, such as cell cycle inhibitors p27 (28) and p21, and Bcl2 (30,67). As expected of proteins increased by a tumor suppressor, they can be growth suppressive (68), which by inference suggests Jade-1 might also inhibit growth. Jade-1 stabilization also supports the notion that VHL may play a global role in determining protein fate. VHL-mediated control of the deubiquitinating enzyme VDU1 (10) supports this assertion, as does the widening role of VHL in defense against cell stress (28 -30, 67).
We pursued a yeast two-hybrid approach to identify low occupancy VHL protein partners that might have importance in renal cancer. This work identifies Jade-1, a novel PEST-and PHD-containing protein, as a strong VHL interactor that may help control gene expression and differentiation in renal cancer precursor cells. Moreover, these studies identify direct protein stabilization as a new VHL function. Further elucidation of the role of Jade-1 in renal tumor suppression and renal epithelial cell growth and development may therefore be of considerable interest.