Analysis of the growth and transformation suppressor domains of promyelocytic leukemia gene, PML.

The promyelocytic leukemia gene (PML) involved in the t(15;17) (q22;q12) translocation in acute promyelocytic leukemia is a growth suppressor. To elucidate the functional domains of PML, several mutants lacking the nuclear localization signal (PMLnls-), the dimerization domain (PMLdim-), the proline-rich domain at the N-terminal (PMLpro-), the proline-rich RING finger motif (PMLpr-), the proline-rich RING finger B-box-1 (PML-prb-), the serine-proline-rich domain at the C-terminal (PMLsp-), and the double mutant (PMLprb-nls-) have been constructed. Immunofluorescence staining of transiently transfected NIH3T3 cells demonstrated that the RING finger motif, dimerization domain, and nuclear localization signal are all required for the formation of PML oncogenic domains (PODs). Immunofluorescence staining of transiently transfected GM637D human fibroblasts indicated that expression of PMLprb-, PM-Lnls-, and PMLprb-nls- led to a significant reduction or, in some cases, complete elimination of PODs. PMLdim-, PMLnls-, PMLpr-, PMLprb-, and PMLprb-nls- mutants were found to lose their ability to suppress transformation of NIH3T3 cells by activated neu, while PMLpro- and PMLsp- mutants did not. These results suggest that the ability of PML to form a POD is essential for suppression of growth and transformation. Furthermore, since PMLprb-, PMLnls-, and PMLprb-nls- mutants could block the suppression effect of wild-type PML on transformation of NIH3T3 cells by the neu oncogene, these PML mutants are potential dominant negative inhibitors of PML. Our study also suggests that the RING finger motif may interact with other nuclear proteins.

The nonrandom chromosomal translocation t(15;17) is a cytogenetic hallmark of acute promyelocytic leukemia (APL) 1 (1), and the genes involved at the breakpoint site have been cloned and characterized (2)(3)(4)(5)(6)(7)(8)(9)(10)(11). It is now clear that the t (15;17) translocation fuses the retinoic acid receptor-␣ (RAR␣) gene on chromosome 17 and the PML gene on chromosome 15. Fusion genes PML-RAR␣ and RAR␣-PML transcribe fusion transcripts and encode potentially oncogenic fusion proteins. Since most of the functional domains of both PML and RAR␣ are retained in the fusion protein PML-RAR␣, it is speculated that PML-RAR␣ is critical for the pathogenesis of APL. Indeed, in cotransfection assays, it was found that PML-RAR␣ represses RA-responsive promoters in the absence of RA (5)(6)(7)9). The stable expression of PML-RAR␣ inhibits apoptosis in U937 cells and makes these cells unable to respond to differentiation induction by 1,25-dihydroxyvitamin D 3 (vitamin D 3 ) and transforming growth factor ␤ 1 (12). In addition, stable expression of PML-RAR␣ in K562 cells interferes with erythroid differentiation induced by hemin (13,14). Thus, expression of this fusion protein in APL cells may be responsible for blocking differentiation and prolonging survival of promyelocytes.
PML-RAR␣ can effectively sequester PML and the retinoic X receptor (RXR) (7,(15)(16)(17). In APL cells, double-color immunofluorescence staining of PML, RXR, and PML-RAR␣ demonstrated that both PML and RXR colocalize with PML-RAR␣ in vivo (16). However, after all-trans-retinoic acid (ATRA) induced differentiation of the APL cells, PML and RXR no longer colocalized with PML-RAR␣, indicating that sequestration of PML and RXR by the PML-RAR␣ fusion protein is critical for the development of APL. This notion is strongly supported by our recent finding that PML is a growth suppressor (18). In brief, we found that (i) increased expression of PML in the APL cells results in a loss of clonogenicity in soft agar assay and tumorigenicity in nude mice (18); (ii) PML suppresses the transformation of rat embryo fibroblasts by cooperative oncogenes and of NIH3T3 cells by activated neu (18,19); (iii) PML is a promoter-specific transcription suppressor, suppressing the promoter activity of the epidermal growth factor receptor and multidrug-resistance genes and having little effect on the promoters of ␤-actin, SV-40, and Rous sarcoma virus (18). From this study and others, we hypothesized that PML-RAR␣ plays a central role in APL as a dominant negative inhibitor of the normal function of PML (growth suppressor) and RXR (coregulator of RAR, thyroid hormone, vitamin D receptor, etc.) and consequently leads to the development of APL by growth stimulation and loss of differentiation induction by hormone or growth factors.
PML is a member of the family of RING finger proteins capable of forming a homodimer through the dimerization domain (20). As shown by immunofluorescence staining, PML accumulates in the nucleus in a speckled pattern composed of nuclear bodies called PML oncogenic domains (PODs) (17). PODs are also known as ND10 (21) or Kr bodies (16). PML is believed to colocalize with the autoantigens in primary biliary cirrhosis, but the two proteins are not directly known to interact with each other (16,17,22). It is speculated that PML first interacts with other unknown protein factors and then is indirectly associated with the autoantigens. This type of PMLassociated protein factor has yet to be identified. In APL cells induced to differentiated by ATRA, the abnormal microspeckled PML pattern can be reorganized to the normal PODs (16,17). In RA-resistant, APL-derived NB4 cells, POD formation cannot be induced by ATRA treatment (17). Together, these results indicate that the ability of PML to form PODs is required for the induction of differentiation in APL cells.
Taking advantage of the fact that PML can suppress the formation of NIH3T3 cell foci by activated neu (18,19), we have constructed several PML mutants and examined their ability to form PODs in vivo and suppress transformation. Here, we demonstrate that PML mutants lacking the nuclear localization signal, the dimerization domain, and the RING finger domain are unable to form PODs and that their ability to suppress foci formation is significantly reduced. However, mutants lacking the proline-rich domain at the N-terminal and the serine-proline-rich domain on the C-terminal can form PODs in vivo and can suppress foci formation, much like the wild-type PML. Our results corroborate the previous finding that the ability of APL cells to reassemble PODs after ATRA treatment is an important event in differentiation induction (17). Furthermore, we show that PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ are potential dominant negative inhibitors of PML and that the RING finger motif may interact with other nuclear proteins.

MATERIALS AND METHODS
Plasmid Construction-The PML expression plasmid pSG5PML, as described in our previous report (18), was used to construct PML deletion mutants. Mutant PMLnlsϪ (see Fig. 1) was created by digesting pSG5PML with MluI and SmaI, blunt ended with a Klenow fragment, and circularized with T4-DNA ligase. Mutant PMLspϪ was constructed by digesting pSG5PML with MluI and BamHI and then blunt ended and circularized with T4-DNA ligase. Mutant PMLdimϪ was constructed by completely digesting pSG5PML with KpnI, then ethanol precipitated, resuspended in high stringency S1 buffer (200 mM NaCl, 50 mM sodium acetate, pH 4.5, 1 mM ZnSO 4 , 0.5% glycerol), and treated with 3000 units of S1 nuclease (Life Technologies, Inc.)/ml at 37°C for 30 min. The linearized plasmid was further treated with the Klenow fragment and circularized with T4-DNA ligase. Clones with an in-frame deletion were identified by direct DNA sequencing and retained. A clone lacking PML nucleotides 684 -888 (8), a sequence that encodes the entire putative leucine zipper, was designated PMLdimϪ. Deletion mutants PMLproϪ, PMLprϪ, and PMLprbϪ were constructed by polymerase chain reaction amplification of pFM211 (the original cDNA clone used to construct pSG5PML (18)) using the following primers: MU1, 5Ј-AAGGATCCACCATGGAGGAGGAGTTCCAGTT-3Ј; MU2, 5Ј-GAGGATCCACCATGGATAACGTCTTTTTCG-3Ј; MU3, 5Ј-AAG-GATCCACCATGGTGCGTGAGTTCCTGGA-3Ј; PMD, 5Ј-CGATG-CAATTGAGGGTGCAA-3Ј. The underlined regions indicate the position of the perfect Kozak sequence (23) being incorporated into the upstream primer to generate the translation initiation site. To minimize the possibility of mutation, polymerase chain reaction amplifications were carried out by KlenTaq 1 and Pfu polymerases in a 15:1 ratio as described by Barnes (24). Primer sets MU1/PMD, MU2/PMD, and MU3/PMD were used to generate mutant cDNAs of PMLproϪ, PM-LprϪ, and PMLprbϪ, respectively. The cDNA was digested with the restriction endonuclease BamHI, purified on a Centricon-100 spin column (Amicon Inc., Beverly, MA), and inserted into a pSG5 vector (Stratagene) at the BglII site. The correct orientation of the inserted cDNA was verified by brief restriction mapping. A double mutant (PMLprbϪnlsϪ) was created by deleting the SmaI-MluI fragment of PMLprbϪ (this fragment harbors the nuclear localization signal domain). Correct joining of each clone was confirmed by DNA sequencing.
DNA Sequence Analysis of the PML Mutants-Sequencing of the mutant PML plasmids was performed as described previously (8).
In Vitro Transcription and Translation-In vitro transcription of the PML mutant plasmids in the expression vector pSG5 was performed with T7 RNA polymerase using an RNA transcription kit obtained from Stratagene. In vitro translation was done by incorporating [ 35 S]methionine (1,200 Ci/mmol) (DuPont NEN) into proteins using a rabbit reticulocyte lysate in vitro translation system obtained from Life Technologies, Inc. In vitro translated proteins were electrophoresed in an SDS-10% polyacrylamide gel, dried, and autoradiographed.
Gene Transfection-For high efficiency transfection of the PML mutant plasmids into the NIH3T3 cells, lipofectamine-mediated gene transfers were performed according to the manufacturer's instructions (lipofectamine was purchased from Life Technologies, Inc.). To study the effect of PML and its mutants on transformation of NIH3T3 cells by activated neu oncogene, a modified calcium phosphate coprecipitation method was used. In this transfection assay, 2 g of pSV2lacZ was included in all transfection experiments to monitor transfection efficiency.
Briefly, exponentially growing cells were trypsinized, seeded at 5 ϫ 10 5 cells per 10-cm plate, and incubated overnight in 10 ml of complete Dulbecco's modified Eagle's medium. Then, 10 -20 g of plasmid DNA premixed with 0.5 ml of 0.25 M CaCl 2 , 0.5 ml of 2 ϫ Hanks' buffered solution (50 mM Hepes, 280 mM NaCl, 1.5 mM Na 2 HPO 4 , pH 7.04 Ϯ 0.02) was added, and the mixture was incubated for 15-20 min at room temperature. Calcium phosphate-DNA solution (1 ml) was added dropwise to the plate of cells, after which the mixture was swirled gently and incubated for 16 -24 h at 37°C under 5% CO 2 . The medium was removed, and the cells were rinsed twice with growth medium, refed, and incubated for 24 h at 37°C under 5% CO 2 . The cells were split at a 1:10 ratio for focus-forming assay.
Focus-forming Assay-A focus-forming assay of NIH3T3 cells was performed as described in our previous reports (18). Foci were visualized after 12-14 days of incubation by staining with 0.5% crystal violet in 25% methanol.
Immunofluorescence Staining and Western Blotting-The mutant plasmids were transfected into NIH3T3 cells by lipofectamine-mediated gene transfer according to the manufacturer's instructions. Total protein was isolated after 48 h of culture. Protein fractions isolated from the NIH3T3 cells transfected with different mutant PMLs were then subjected to immunofluorescence staining and Western blotting using the affinity-purified antipeptide antibody as described in our previous reports (19,25).

Characterization of PML Deletion Mutants-
The major features of the PML protein ( Fig. 1) include the proline-rich domain at the N-terminal, which resembles the transactivating domain of some transcription factors (7,20,26). A RING finger motif, B-box-1, and B-box-2 lie immediately downstream. Other important domains include the ␣-helical domain responsible for dimerization, a nuclear localization signal, and a prolineserine-rich domain at the C-terminal believed to be involved in PML phosphorylation. In each mutant PML constructed by us as shown in Fig. 1 and as described under "Materials and Methods," the deleted region was confirmed by direct DNA sequencing. For mutants PMLproϪ, PMLprϪ, and PMLprbϪ, the 5Ј-ends of these clones were sequenced, and the presence of the correct ATG translation start site and the perfect Kozak's sequence was confirmed in each clone.
To confirm that these mutant clones would translate their respective mutant PML proteins, each of these plasmids were transfected into NIH3T3 cells by lipofectamine-mediated gene transfer. Western blot analysis of protein fractions isolated from these cells indicated that PMLdimϪ and PMLnlsϪ encoded mutant PML proteins of the expected size (approximately 80 -90 kDa) ( Fig. 2A, lanes 6 and 7). However, mutant PMLproϪ, PMLprϪ, and PMLprbϪ encoded mutant proteins significantly smaller than expected. We have previously reported that although its predicted size is about 72 kDa, the PML protein migrates at about 90 kDa during SDS-polyacryl- Functional Domains of PML amide gel electrophoresis, probably due to its acidic nature (18). As shown in Fig. 2A, lane 2, PMLproϪ with a deletion of the first 48 amino acids migrated at about 70 kDa, indicating that is much smaller than the wild-type PML. To confirm that there is no additional deletion within the cDNA sequence, we completely sequenced one strand of PMLproϪ and found no deletion in any other region of the cDNA. We speculate that the significant difference in size between PMLproϪ and PML is the result of a change in secondary structure. The proline-rich domain may be responsible for the slower running nature of the wild-type PML in SDS-polyacrylamide gel electrophoresis. This notion is supported by the fact that both PMLprϪ and PML-prbϪ migrated at molecular weights agreeable with their predicted sizes ( Fig. 2A, lanes 3-5).
To further confirm these findings, the PML mutant plasmids were transcribed and translated in vitro. Fig. 2B showed that the molecular weights of the in vitro transcribed and translated PMLproϪ, PMLprϪ, PMLprbϪ, and PMLdimϪ proteins agree with those derived from the Western blotting analysis as shown in Fig. 2A.
Cellular Localization of PML Mutant Proteins-Since PML had been shown to concentrate within PODs (16,17,22,25), we intended to define which domains of PML are required for the formation of PODs. How deleting various domains of PML affects its cellular localization was determined by transfecting the PML mutants into NIH3T3 cells. The PML mutant proteins were visualized in the cell by immunofluorescence staining using the affinity-purified PML antibody as described in our previous reports (18,19). As shown in Fig. 3, d and h, PMLproϪ and PMLspϪ did not affect the cellular location of PML. A typical nuclear speckled pattern made up of PODs was found in cells transfected with these plasmids. Deletion of the region harboring the nuclear localization signal resulted in cytoplasmic and perinuclear localization (Fig. 3b), presumably because the PML mutant protein could not enter the nucleus. A similar pattern of cellular distribution of PMLdimϪ, PMLprϪ, and PMLprbϪ was found in NIH3T3 cells transfected with these plasmids. A diffuse instead of speckled nuclear pattern was found (Fig. 3, c, e, and f). The double mutant (PMLprbϪnlsϪ), unlike PMLnlsϪ, was diffused throughout the cytoplasm (Fig.  3g). Since only an additional deletion of the RING finger motif in PMLprbϪnlsϪ, it is suggested that the RING finger motif may be responsible for interaction with other protein factors and that its deletion disrupts the perinuclear pattern. Together, the above results indicate that the dimerization domain, the RING finger domains, and the nuclear localization signal are all required for the organization of PODs, since deletion of these regions resulted in diffuse nuclear distribution and the inability of PMLs to assemble into POD.
Effect of PML Mutants on Suppression of Transformation of NIH3T3 Cells by Activated neu-In our previous report (19), we demonstrated that PML suppresses neu-transformed NIH3T3 cells by suppressing both the expression of neu and its signaling events. To investigate the ability of these PML mutants to suppress transformation of NIH3T3 cells, each individual mutant was cotransfected with c-Neu104 (genomic neu) by calcium phosphate coprecipitation as described under "Materials and Methods." Results of these studies (Fig. 4) showed that PMLdimϪ, PMLnlsϪ, PMLprϪ, PMLprbϪ, and PMLprbϪnlsϪ lost their ability to suppress transformation of NIH3T3 by activated neu. This suggests that the dimerization domain, nuclear localization signal, and RING finger motif are all required for the PML

FIG. 3. Cellular distribution of PML and its mutants in transfected NIH3T3 cells.
NIH3T3 cells transfected with 2-3 g of the respective plasmids by lipofectamine were cultured for 48 h, and immunofluorescence staining was performed as described previously (19). a, negative control; b, PMLnlsϪ; c, PMLdimϪ; d, PMLproϪ; e, PM-LprϪ; f, PMLprbϪ; g, double mutant (PMLprbϪnlsϪ); h, PMLspϪ. Magnification, ϫ1000. growth suppressor function. As shown in Fig. 3, all of these PML mutants share one common property, i.e. they are unable to assemble into a nuclear speckled pattern or the PODs. This suggests that the ability of PMLs to organize into PODs is required for PML to exert its growth or transformation suppressor function. This notion is supported by the finding that PML mutants PMLproϪ and PMLspϪ, which are capable of forming PODs (Fig. 3, a and d), were both able to suppress neu-induced transformation of NIH3T3 cells, just like the wildtype PML (Fig. 4).
PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ as Potential Dominant Negative Inhibitors of PML-PML-RAR␣ reportedly plays a critical role in APL as a dominant negative inhibitor of the normal PML and RXR function; the fusion protein acts by forming PML-RAR␣⅐PML and PML-RAR␣⅐RXR heterodimers through the dimerization domain of PML and RAR␣ (15)(16)(17).
To investigate the effect of mutant PMLs on human cells that normally express wild-type PML, pPMLprbϪ, pPMLnlsϪ, and pPMLprbϪnlsϪ were transfected into GM637D cells, an SV40transformed human fibroblast cell line, by lipofectamine according to the procedures recommended by the manufacturer (Life Technologies, Inc.). In GM637D cells, PMLprbϪ displayed a diffuse nuclear pattern (Fig. 5b) without the typical POD pattern (Fig. 5a). PMLnls accumulated outside the nucleus, mostly in the perinuclear zone (Fig. 5d). Interestingly, the double mutant PMLprbϪnlsϪ showed a diffuse but fine staining pattern throughout the cytoplasm (Fig. 5c), completely unlike those of PMLprbϪ and PMLnlsϪ. Together, these results indicate that the expression of PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ in GM637D cells led to a significant reduction or, in some cases, complete elimination of the PODs. This in turn suggests that PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ are capable of sequestration PML from its normal cellular localization and are potential dominant negative inhibitors of the wild-type PML.
The dominant negative inhibitory effect of PMLprbϪ, PM-LnlsϪ, and PMLprbϪnlsϪ were further demonstrated by their ability to inhibit the biologic function of the wild-type PML. As shown in Fig. 6, PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ were remarkably capable of knocking out the ability of wild-type PML to suppress transformaton of NIH3T3 cells by activated neu. This inhibition is similar to that caused by the PML-RAR␣ fusion protein encoded from the breakpoint of the t(15;17) translocation in APL (18). Furthermore, PMLprbϪ and PMLprbϪnlsϪ inhibited PML function in a dose-dependent manner (data not shown).

DISCUSSION
To understand the function of PML, a growth suppressor involved in the t(15;17) translocation in APL, we constructed several mutant plasmids and used them to analyze its functional domains. We found that (i) the nuclear localization signal, dimerization domain, and RING finger motif of PML are all essential for the formation of PODs; (ii) the ability of PML to form PODs in vivo is required for its growth suppressor function; and (iii) PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ are potential candidates for dominant negative inhibition of the wild-type PML. Moreover, we have uncovered evidence that the RING finger motif of PML may interact with other proteins.
The PML protein contains several important domains, including a C3HC4 (3 Cys, 1 His, and 4 Cys) zinc-finger motif (RING finger domain); two additional Cys/His regions, referred to as B-box-1 and B-box-2; and an ␣-helical coiled-coil region (21, 26 -28). Immunofluorescence staining demonstrated that PML is a major component of novel nuclear bodies designated PODs (17,20,26). We have conclusively demonstrated that the nuclear localization signal, the dimerization domain, and the RING finger motif of PML, but not the proline-rich domain at the N-terminal and the serine-proline-rich domain at the Cterminal, are required for the formation of PODs. Our results, as shown in Fig. 3, e and f, agree with the observation of Kastner et al. (7) that PML proteins bearing a mutation in the cysteine-rich (RING finger) motif (Glu 59 -Cys 60 3 Glu 59 -Leu 60 ) had a diffuse nuclear staining pattern. During the preparation of this manuscript, Borden et al. (29) also demonstrated by site-directed mutagenesis that the RING finger motif of PML is FIG. 4. The effect of mutant PMLs on suppression of neu-induced transformation of NIH3T3 cells. Cotransfection of plasmids by calcium phosphate coprecipitation and focus-forming assays were performed as described previously (18). In each transfection assay, 10 g of plasmid was transfected. The results shown represent means of three independent experiments with three plates each. In each transfection, 2 g of pSV-␤Gal (␤-galactosidase cDNA under the control of the SV40-early promoter) was included, and the activity of ␤-galactosidase was determined to monitor transfection efficiency. Plasmid c-neu104 containing the activated neu oncogene has been described previously (18); PML and its mutants used in the study are described in Fig. 1.
necessary for POD formation.
In our previous reports, we showed that PML is a growth and transformation suppressor (18,19). In this study, we demonstrated that the RING finger motif, dimerization domain, and nuclear localization signal are responsible for this suppression of growth and transformation. Since these regions of PML are also required for the formation of PODs, we therefore concluded that the ability of PML to assemble into PODs is essential for the growth suppression effect of PML. This finding is corroborated by the observations that ATRA-induced differentiation of NB4 and APL cells is associated with a reorganization of the microspeckled nuclear pattern into PODs and that no such response was found in the ATRA-resistant NB4 cells (16,17,22).
At present, the composition and function of PODs is not completely understood. Our finding that PML is a growth and transformation suppressor sheds new light on the in vivo biologic function of this new form of nuclear body. PODs, which are associated with the nuclear matrix, reportedly do not contain any of the proteins known to be involved in pre-mRNA splicing, transcription, and DNA synthesis (17,30). In addition to PML, the PODs may include the SP100, NDP-55, and Vmw110 (also known as ICP0) proteins and an unidentified 65-kDa protein (16,17,21,22,30). These proteins, however, do not directly interact with PML. In studies of APL and herpes simplex virus immediate-early protein, PODs appeared to be extremely important (both directly and indirectly) for myeloid differentiation as well as regulation of gene expression (e.g. activation of both cellular and viral genes in infection) (16,17,21,22,30,31).
Recently, the growth suppressor function of PML was confirmed by another group, who found that when skin, breast, and colon malignant cells turned invasive, PML expression was lost (32). These data imply that PML may be an antioncogene involved not only in APL but also in non-hematological oncogenesis. Interestingly, Terris et al. (33) reported that expression of PML increased significantly in the inflammatory tissues and during normal or pathological proliferative states, indicat-ing that PML plays a role in the inflammatory process and in cell growth control.
As stated above, the PML-RAR␣ fusion protein is reportedly a dominant negative inhibitor of PML and RXR (7,17,34). Previous immunofluorescence staining of PML indicated cytoplasmic and/or nuclear localization with a micropunctuated pattern in APL blasts and in cells stably transfected with PML-RAR␣. Here, we have demonstrated that PMLprbϪ, PM-LnlsϪ, and PMLprbϪnlsϪ are potential dominant negative inhibitors against the normal function of PML and that transient expression of PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ significantly reduces or completely eliminates normal PODs in GM637D cells. Furthermore, we have shown that PMLprbϪ, PMLnlsϪ, and PMLprbϪnlsϪ can block the effect of wild-type PML on transformation of NIH3T3 cells by activated neu. We have noticed that stable transfectants overexpressing PM-LnlsϪ in GM637D cells eventually died (data not shown). Since overexpression of PML in B104 -1-1 (19) and HeLa cells 2 did not result in cell death, we therefore reason that PMLnlsϪ probably induced the observed cell death through its perinuclear localization. It is possible that the RING finger motif interacts with other important unknown proteins essential for cell growth. Sequestration of these proteins by PMLnlsϪ may be responsible for the cell death. Therefore, PMLnlsϪ may not be a suitable dominant negative inhibitor. Conversely, PML-prbϪ and PMLprbϪnlsϪ, both without the RING finger motif, should be more suitable for testing the effect of knocking out wild-type PML in normal cells and for studying whether elimination of cellular PML will result in growth stimulation. We plan to establish in our laboratory stable transfectants of GM637D constitutively expressing the PMLprbϪ and PMLprbϪnlsϪ and to investigate the effects of these dominant negative inhibitors on proliferation and growth of GM637D cells. Such studies will be helpful in further understanding the function of PML in the control of cell growth.
As stated above, PML is characterized by the RING finger motif. This motif has the sequence C 1 -X 2 -C 2 -X (9 -27) -C 3 -X (1-3) -H 1 -X 2 -C 4 -X 2 -C 5 -X (4 -48) -C 6 -X 2 -C 7 , where C represents cysteine, H represents histidine, X can be any amino acid, and the numbers in parentheses refer to the length in amino acids of the sequences (20,27). The RING finger motif may play a important cellular role by acting as a DNA binding or proteinprotein interacting domain (9,20,29). It is necessary then to determine whether DNA or protein is involved in the interaction with the RING finger motif of PML protein. The fact that treatment of the nuclei with RNase and DNase in previous studies did not disrupt the PODs in human cell lines suggests that nucleic acids are not a component of this structure (30). 3 Consequently, this implies that it is the RING finger motif that normally interacts with protein factors. Our results presented in Figs. 3 and 4 show that the immunostaining pattern of mutant PMLnlsϪ is clearly distinct from that of double mutant PMLprbϪnlsϪ and that PMLnlsϪ accumulated in the cytoplasm, mostly in the perinucleus. In contrast, the staining pattern of the double mutant was fine and diffuse throughout the cytoplasm. This suggests that the RING finger motif accounts for these differences. We therefore speculate that the RING finger motif of PML interacts with other nuclear protein factors and is responsible for the differences in cellular distribution between PMLnlsϪ and PMLprb-nlsϪ.
We conclude then that the RING finger motif, ␣-helical region, and nuclear localization signal of PML are all required for the formation of PODs. We also conclude that the ability of 2 X.-F. Le 6. PMLprb؊, PMLnls؊, and PMLprb؊nls؊ as potential dominant negative inhibitors of the wild-type PML. Cotransfection of plasmids by calcium phosphate coprecipitation and focus-forming assays were performed as described previously (18). In each transfection assay, 10 g (unless otherwise specified in parentheses) of plasmid was transfected. The results shown represent the means of three independent experiments with three plates each. Plasmid pSV-␤Gal was included, and the activity of ␤-galactosidase was determined to monitor transfection efficiency. PML to form PODs is essential for its growth and transformation suppressor function. However, studies are needed (i) to define more precisely the specific contribution of each domain in PML and (ii) to define the composition and function of PODs so as to understand in depth their role during human oncogenesis. In this direction, we have been assessing the effects of the two dominant negative inhibitors of wild-type PML described in this report (PMLprbϪ and PMLprbϪnlsϪ) on the proliferation and differentiation of GM637D cells, which stably express mutant PMLs.