Role of PML in Cell Growth and the Retinoic Acid Pathway

See allHide authors and affiliations

Science  06 Mar 1998:
Vol. 279, Issue 5356, pp. 1547-1551
DOI: 10.1126/science.279.5356.1547


The PML gene is fused to the retinoic acid receptor α (RARα) gene in chromosomal translocations associated with acute promyelocytic leukemia (APL). Ablation of murine PML protein by homologous recombination revealed that PML regulates hemopoietic differentiation and controls cell growth and tumorigenesis. PML function was essential for the tumor-growth–suppressive activity of retinoic acid (RA) and for its ability to induce terminal myeloid differentiation of precursor cells. PML was needed for the RA-dependent transactivation of the p21 WAF1/CIP1 gene, which regulates cell cycle progression and cellular differentiation. These results indicate that PML is a critical component of the RA pathway and that disruption of its activity by the PML-RARα fusion protein may be important in APL pathogenesis.

Acute promyelocytic leukemia is a distinct subtype of myeloid leukemia that is invariably associated with chromosomal translocations involving the RARα locus (1). In 99% of APL cases, RARα is fused to thePML gene, leading to the production of a PML-RARα chimeric protein (2).

Retinoic acid receptors are nuclear hormone receptors that act as RA-inducible transcriptional activators, in their heterodimeric form, with retinoid-X receptors (RXRs), a second class of nuclear retinoid receptors (3). Retinoic acid controls fundamental developmental processes, induces terminal differentiation of myeloid hemopoietic progenitors, and has tumor- and cell-growth–suppressive activities (4). PML is an interferon (IFN)-inducible gene (5) that encodes a RING-finger protein typically concentrated within discrete speckled nuclear structures called PML nuclear bodies (PML NBs) or PML oncogenic domains (2, 5). Through its ability to heterodimerize with PML and RXR, PML-RARα is thought to interfere with both PML and RAR/RXR-RA pathways, thus acting as a double dominant negative oncogenic product (5, 6). However, the normal function of PML and its contribution to APL pathogenesis are unknown.

To investigate these aspects, we disrupted the PML gene in the mouse germ line (7, 8). By homologous recombination in murine embryonic stem (ES) cells, we substituted part of exon 2 of the PML gene, which encodes the RING-finger domain, with a neomycin resistance gene cassette (Fig.1, A to C). Mice homozygous for the PML mutation (PML−/−) were born with the expected Mendelian frequency, were indistinguishable at the gross phenotypic level from PML+/+ and PML+/− littermates, and were fertile; however, the PML−/− mice were extremely susceptible to spontaneous Botryomycotic infections (9). Successful disruption of the PML gene was inferred from the lack of PML mRNA and PML NBs in mouse primary embryonic fibroblasts (MEFs) from PML−/− embryos (Fig. 1, D and E) (7).

Figure 1

Targeted disruption of the PMLgene. (A) Map of the murine 5′ PML genomic region determined by restriction mapping, Southern blot hybridization, and DNA sequencing (15). The targeting vector is derived from a 6.8-kb Eco RI fragment of the PML gene. The TK and Neo selectable markers are shown as hatched boxes (7). Also shown is the endogenous PMLgenomic region after correct integration of the targeting construct by homologous recombination and the three probes used for Southern blot analysis (solid lines) (7). E, Eco RI; K, Kpn I; N, Nar I; No, Not I; A, Apa I; B, Bam HI; and Ec, Eco RV. (B) Southern blot analysis with A, B, and Neo probes of Bam HI–digested DNA from recombined ES cell clones and AB1 untransfected ES cells confirms proper recombination. (C) Southern blot analysis with the probe A of Eco RI–digested tail DNA from littermates obtained from intercrossing two PML+/− mice. DNAs from two mice show the disappearance of the wild-type (WT) band. R, recombinant bands. (D) Northern blot showing that the homozygous PML mutation abolishesPML mRNA expression. For up-regulation of PMLexpression, PML−/− and PML+/+ MEFs were treated with murine IFNα+β. The integrity and amount of RNA as well as stimulation by IFNs were shown by rehybridizing the same blot with β-actin and IFI 204 probes; h, hours; 28 S and 18 S, ribosomal RNA. (E) PML−/− and PML+/+ MEFs were studied by immunofluorescence in basal conditions or upon IFN treatment for 24 hours (7). PML−/− MEFs do not show any PML nuclear staining. The nuclei of the PML−/−cells were visualized by costaining with 4′6′-diamidino-2-phenylindole (DAPI) (right). PML-Ab, PML rabbit antiserum.

Analysis of peripheral blood (PB) from PML−/− mice (10) revealed a marked reduction of circulating granulocytes (neutrophils: PML+/+, 1518 ± 220 cells/μl; PML−/−, 795 ± 243 cells/μl; P< 0.02; basophils: PML+/+, 247 ± 169 cells/μl; PML−/−, 69 ± 19 cells/μl; P < 0.01; eosinophils: PML+/+, 478 ± 142 cells/μl; PML−/−, 136 ± 74 cells/μl; P < 0.03) and an overall reduction of circulating myeloid cells (monocytes: PML+/+, 365 ± 196 cells/μl; PML−/−, 203 ± 77 cells/μl; P < 0.05), which caused leukopenia. Flow-cytometric analysis of cells from the spleen, lymph nodes, thymus, and bone marrow (BM) and differential counts of BM cells demonstrated a reduction of both granulocytes and monocytes in the BM of PML−/− mice (11). Thus, PML−/− mice have an impaired capacity for terminal maturation of their myeloid cells.

PML overexpression in cultured cells is accompanied by growth inhibition (12). To assess the effects of PML inactivation on cell proliferation, we studied the growth of early passages of MEFs (13), which normally express PML (Fig. 1E) and whose proliferative properties are well characterized (14). At low density, PML+/+, PML+/−, and PML−/− cultures were morphologically indistinguishable. However, PML−/− MEFs grew faster than PML+/+MEFs (Fig. 2A), as confirmed by3H-labeled thymidine incorporation (Fig. 2B). PML+/− MEFs showed an intermediate growth rate (Fig. 2, A and B). In situ terminal deoxynucleotidyl tranferase labeling experiments revealed comparable numbers of apoptotic cells (15). Furthermore, PML inactivation markedly enhanced the ability of MEFs to form colonies (Fig. 2C). The S phase population of PML−/− MEFs was increased with a concomitant decrease in the G0/G1 population (Fig. 2D), a change analogous to that observed in retinoblastoma−/− MEFs (14). PML−/− MEF monolayers achieved higher cellular densities and formed foci (Fig. 2, E and F) but, unlike fully transformed cells, were unable to grow in a semisolid medium.

Figure 2

Growth properties of MEFs of different PML genotypes. (A) Growth curves. Each time point is the average of triplicate measurements. Doubling time of MEFs: PML−/−, 26.4 ± 3.1 hours; PML+/−, 29.8 ± 3.3 hours; PML+/+, 32.9 ± 3.9 hours (13). (B) [3H]thymidine incorporation. Cells (5 x 103 per well) were distributed in 96-well plates and cultured in the presence of [3H]thymidine. Each time point is the average of triplicate measurements (13). PML+/+, open bars; PML+/−, solid bars; PML−/−, hatched bars. (C) Clonogenic efficiency (13). The colonies (>10 cells) were scored under the microscope 8 days after plating. The scoring of colonies at day 16, when the colonies were bigger in size and detectable by the eye, gave superimposable results. (D) Analysis of cell cycle stages in PML+/+(left) and PML−/− (right) MEFs. Cultures were pulsed with BrdU, labeled with an antibody to BrdU to detect DNA synthesis (vertical axis) and propidium iodide to detect total DNA (horizontal axis), and analyzed by two-dimensional flow cytometry (13). (E) Loss of contact inhibition in PML−/−MEFs. Cells were seeded at 3 × 105 cells in 60-mm dishes and, after 2 weeks, fixed and stained. PML−/− MEFs grew to a higher density than PML+/+ MEFs of the same passage. Scale bar, 60 μm. (F) [3H]thymidine incorporation of PML−/− and PML+/+ MEFs cultured at confluence. The conditions were as in (B), except that 4 × 104 cells per well were seeded.

These findings suggest that PML is a negative growth regulator and therefore may function as a tumor suppressor. Although the incidence of spontaneous tumors in the PML−/− cohort was not increased during the first year of life, mutant mice succumbed to infections, severely compromising the long-term assessment of tumor incidence (9). We therefore studied tumorigenesis in two experimental models designed to accelerate tumor formation (16, 17). In the first, we exposed the skin of mice [because PML is highly expressed in keratinocytes (Fig.3B)] to a single application of the tumor initiator dimethybenzanthracene (DMBA) followed by treatment for several weeks with the tumor promoter 12-0-tetradecanoylphorbol-13-acetate (TPA), a protocol that gives rise to papillomas that occasionally progress to carcinomas after several months (16). PML−/− mice developed more papillomas (Fig. 3A), although the frequency of tumors undergoing malignant transformation was similar in the two groups (PML+/+, 1.8%; PML−/−, 2.3%). In the second model, DMBA was injected into the salivary gland of PML−/− and PML+/+ mice, a procedure that normally produces sarcomas and fibrosarcomas (16,17). PML−/− mice developed more tumors than control mice (greater than twofold; P < 0.04) (Fig. 3, D to I) (16). Unexpectedly, 50% of the tumors observed in the PML−/− group were T and B cell lymphomas (only one B cell lymphoma arose in the wild-type cohort; P < 0.02), and 21% were fibrohistiocytoma-like lesions (rare tumors with a histiocytic-macrophagic cellular component) (Fig. 3E). Lymphomas in PML−/− mice were aggressive metastatic malignancies (Fig.3D). They appeared to be of clonal origin because the infiltrating lymphoid population homogeneously expressed either CD4 or CD8 markers (T lymphomas) and either κ or λ chains (B lymphomas) (Fig. 3, H and I). Macrophage tumoricidal activity, natural killer cell, and cytotoxic T lymphocyte activities, which are required for efficient surveillance against tumors, were normal in PML−/− mutants (9); however, upon concanavalin A activation, splenic lymphocytes in PML−/− mutants showed a proliferative advantage despite normal production of interleukin-10 (IL-10), IL-4, IL-6, and IFN-γ (15). PML can, therefore, antagonize the initiation, promotion, and progression of tumors of different histological origins.

Figure 3

Role of PML in tumorigenesis. (A) Rate of appearance of papillomas in PML−/− (n = 14) and PML+/+(n = 14) mice treated with DMBA and TPA. Graph shows the average number of papillomas ± SE, and the arrow indicates the time when the tumor promotion treatment with TPA was terminated. One of two independent experiments is shown (16). (Band C) PML expression in the skin and lymphocytes. (B) Immunohistochemical analysis of the skin was performed on paraffin tissue sections from newborn mice with a PML rabbit antiserum (7). PML is readily detectable in its nuclear speckled configuration in keratinocytes. Scale bar, 50 μm. (C) Splenic lymphocytes were studied by immunofluorescence (7) (top). Nuclei were visualized by DAPI (bottom). PML NBs are detectable in all lymphocytes. PML is also expressed in thymic and BM lymphocytes (15). (D andE) Histopathological analysis of the tumors that developed in PML−/− mice. Twenty-five mice of each group were injected with DMBA (16). Tumors in PML−/−[four T cell lymphomas, three B cell lymphomas, three malignant fibrohistocytomas (MFH), one angiosarcoma, and three fibrosarcomas] and PML+/+ [one B cell lymphoma, one MFH, two fibrosarcomas, two soft tissue sarcomas, and one benign papilloma] mice were identified by external examination after 4.86 ± 0.53 and 5.67 ± 0.69 months (P < 0.02), respectively, upon DMBA injection, after which animals were killed for pathological examination. (D) Marked splenomegaly in a PML−/− mouse that developed a T cell lymphoma (right), as compared with the spleen of a wild-type age-matched control mouse (left). Scale bar, 0.5 cm. These lymphomas were metastasizing tumors that involved the spleen, thymus, lymph nodes, liver, and vertebrae. (E) Hematoxylin and eosin staining of a subcutaneous tumor with large dysplatic histiocytes (arrows) displaying multinucleation and numerous prominent nucleoli. This tumor was diagnosed as an MFH. Scale bar, 50 μm. (F to I) Immunophenotyping of lymphomas from PML−/− mice. (F) The tumor is positive for the T cell surface antigen CD3 and negative for the B cell surface antigen B220 (inset). (G) The tumor is positive for the B cell surface marker B220 and negative for the T cell surface antigen CD3 (inset). These tumors were diagnosed as T and B cell lymphomas, respectively. The homogeneous expression of CD4 or CD8 (T lymphomas) (H) and κ or λ (B lymphomas) (I) surface markers supports the clonal origin of these tumors. Scale bars, 25 μm.

We next investigated whether PML was required for the growth-suppressive activity of RA. Retinoic acid markedly inhibited the growth of PML+/+ MEFs but had little effect on the growth of PML−/− MEFs (Fig. 4, A and B). Treatment with RA did not increase cell death in these experiments (15).

Figure 4

Effect of PML on RA growth-suppressive and differentiating activities. (A and B) Growth of PML+/+ and PML−/− MEFs in the presence of RA. (A) Each time point is the average of triplicates from a 13.5 days post coitum (d.p.c.) early passage preparation of PML+/+ and PML−/− MEFs. Cells were cultured with or without 1 μM RA. (B) The data are expressed as a percentage of growth of RA-treated versus untreated cells (untreated = 100% of growth), at day 6 of culture. The values are averages calculated from triplicates of three (n = 3) experiments performed with 13.5 d.p.c. preparations of PML+/+ (open bars) and PML−/− (hatched bars) MEFs and five (n = 5) independent experiments performed with two different 15.5 d.p.c. preparations of PML+/+ and PML−/− MEFs. (C) Hemopoietic colonies from in vitro BM cultures (19) from two PML−/− and two PML+/+ mice, scored in triplicate, are shown. Bars are as in (B). (D) Aberrant and impaired RA differentiating activity in PML−/− and PML−/− PML/RARα hemopoietic progenitors (19). The data are expressed as a percentage of colony formation of RA-treated (1 μM) versus untreated cells (untreated = 100%). The bars indicate the mean values ± SD. Triplicate measurements from two PML−/− (hatched bars), two PML+/+ (open bars), and two PML−/− PML/RARα (solid bar) mice are shown. (E) PML−/− and PML+/+ MEFs were transiently transfected by calcium phosphate precipitation with the p21 WAF1/CIP1 promoter-luciferase reporter plasmid pGL2 (10 μg per transfection) (20), together with TK–β-galactosidase (2 μg), in the presence or absence of 1 μM RA. Transactivation is expressed as a percentage of luciferase activity, as deduced from arbitrary light units normalized to β-galactosidase activity. The values shown are averages ± SD calculated from triplicate platings, from one representative experiment out of three. Bars are as in (B). (F) Protein immunoblot analysis of p21WAF1/CIP1 expression in PML−/−and PML+/+ BM cells upon treatment with RA at concentrations of 10−5 M, 10−6 M, or 10−7 M for 72 hours.

Because RA induces terminal myeloid and granulocytic differentiation (3, 4, 18), we tested whether the reduction in myeloid cells in the PML−/− mice resulted from an impaired response of PML−/− progenitors to RA (19). In in vitro methylcellulose colony assays of hemopoietic progenitors, BM cells from PML−/− and PML+/+ mice were comparable in their ability to form erythroid and myeloid colonies (Fig. 4C). In the presence of RA, the number of myeloid colonies obtained from the PML+/+progenitors was increased as expected (18, 19), but this effect was completely abrogated in PML−/− cells (Fig.4D). Thus, the presence of PML is crucial for the growth-inhibitory activity of RA, as well as for RA induction of myeloid differentiation.

To determine if PML-RARα could restore RA activity, we evaluated the RA responsiveness of BM cells from PML−/−PML-RARα transgenic mice (19). We obtained these mutants by crossing PML−/− mice with PML-RARα transgenic mice that express the fusion gene only in the myeloid promyelocytic compartment (19). Retinoic acid significantly reduced the number of myeloid colonies derived from PML−/− PML-RARα BM cells, suggesting that PML-RARα can directly mediate RA growth-inhibitory activity in a PML-independent manner (Fig. 4D).

To explore the mechanism by which PML mediates RA function, we tested whether PML is required for the RA-dependent transactivation of the cyclin-dependent kinase inhibitorp21 WAF1/CIP1 gene, which can be activated by nuclear receptors, including RXRα/RARα (20). We transfected PML−/− and PML+/+ MEFs with ap21 WAF1/CIP1 promoter-reporter construct and assessed the response to RA treatment. Consistent with previous results (20), RA stimulated in PML+/+ MEFs the basal activity of the p21 WAF1/CIP1promoter by two to three times (Fig. 4E). In the absence of PML, the RA-dependent transactivation of the promoter was fully abrogated (Fig.4E). Accordingly, concentrations of RA at 10−7 or 10−6 M did not activate the endogenousp21 WAF1/CIP1 gene in PML−/−BM cells (Fig. 4F). Thus, PML is essential for the RA-dependent induction of p21WAF1/CIP1. Because p21WAF1/CIP1up-regulation can result in terminal differentiation of hemopoietic cells (20), the lack of p21WAF1/CIP1 induction in PML−/− cells might partially explain the role of PML in controlling hemopoietic cell differentiation.

Our findings demonstrate that PML controls cell proliferation, tumorigenesis, and the differentiation of hemopoietic precursors. These functions are, at least in part, based on the ability of PML to interact with the RA pathway and in particular its ability to mediate RA growth-suppressive and differentiating activities. Preliminary results indicate that PML can be part of the RXR/RAR transcription complex (21), providing a direct explanation for its effect on RA function. This function is consistent with the role of PML in the RA-dependent transactivation of specific genes such as p21 WAF1/CIP1 and with the development of T and B cell lymphomas in mice deficient in RARα (22). These results provide a framework for understanding the molecular pathogenesis of APL. Whereas APL might result from the functional interference of PML-RARα with two independent pathways, PML and RXR/RAR, we show here that these proteins act, at least in part, in the same pathway. Thus, by simultaneously interacting with RXR and PML, PML-RARα may inactivate this pathway at multiple levels, leading to the proliferative advantage and the block of hemopoietic differentiation that characterize APL.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: p-pandolfi{at}


View Abstract

Stay Connected to Science

Navigate This Article