Interferon Regulatory Factor-8 Is Indispensable for the Expression of Promyelocytic Leukemia and the Formation of Nuclear Bodies in Myeloid Cells*

Interferon (IFN) regulatory factor-8 (IRF-8), previously known as ICSBP, is a myeloid cell essential transcription factor. Mice with null mutation in IRF-8 are defective in the ability of myeloid progenitor cells to mature toward macrophage lineage. Accordingly, these mice develop chronic myelogenous leukemia (CML). We demonstrate here that IRF-8 is an obligatory regulator of the promyelocytic leukemia (PML) gene in activated macrophages, leading to the expression of the PML-I isoform. This regulation is most effective together with two other transcription factors, IRF-1 and PU.1. PML is a tumor suppressor gene that serves as a scaffold protein for nuclear bodies. IRF-8 is not only essential for the IFN-γ-induced expression of PML in activated macrophages but also for the formation of nuclear bodies. Reduced IRF-8 transcript levels were reported in CML patients, and a recovery to normal levels was observed in patients in remission following treatment with IFN-α. We demonstrate a significant correlation between the levels of IRF-8 and PML in these CML patients. Together, our results indicate that some of the myeloleukemia suppressor activities of IRF-8 are mediated through the regulation of PML. When IRF-8 levels are compromised, the reduced PML expression may lead to genome instability and eventually to the leukemic phenotype.

Interferon (IFN) 2 regulatory factor-8 (IRF-8), previously known as ICSBP (1), is a myeloid cell essential transcription factor that belongs to the IRF family (for review see Refs. 2 and 3). This factor is expressed exclusively in immune cells and accordingly plays a major role in the differentiation, development, and function of macrophages (4 -6), dendritic cells (7)(8)(9)(10), and B-cells (11). Mice with IRF-8 null mutation are defective in the ability of myeloid progenitor cells to mature toward macrophage lineage and eventually develop chronic myelogenous leukemia (CML)-like syndrome (6,12). Furthermore, IRF-8 blocks the development of murine Bcr-Ablinduced CML (13) through the inhibition of Bcl-2 (14) and induction of anti-leukemic immunity (15). In humans, down-regulation of IRF-8 expression was reported in CML and acute myelocytic leukemia (AML) patients. Accordingly, a significant increase in IRF-8 expression was correlated with remission following treatment with IFN-␣ (14,16,17). Together, these observations point toward a role for IRF-8 as a tumor suppressor gene.
To characterize the IRF-8 regulatory network in activated macrophages, a differential expression profile of peritoneal macrophages extracted from wild type (WT) mice versus IRF-8 Ϫ/Ϫ mice, before and following 4 h of stimulation with IFN-␥ and LPS, was studied using DNA microarray assay (18). Among the many putative IRF-8-target genes, the promyelocytic leukemia (PML) gene was identified. PML is a tumor suppressor gene that serves as a scaffold protein for nuclear bodies (NBs). PML-NBs associate with numerous proteins and therefore were implicated in a variety of cellular processes such as cell cycle regulation, apoptosis, proteolysis, tumor suppression, DNA repair, and transcription (for review see Refs. 19 -22). These NBs are disrupted during oncogenesis (23) and viral infection (24). PML is involved in acute promyelocytic leukemia (APL) because of a reciprocal translocation between chromosome 15 (PML locus) and chromosome 17 (retinoic acid receptor-␣ (RAR␣) locus) leading to in-frame fusion peptides between these two loci (25)(26)(27). In this study we show that PML is regulated by IRF-8 in myeloid cells, and we provide evidence that the IRF-8-mediated CML tumor suppressor activity may be delegated via PML.

EXPERIMENTAL PROCEDURES
Patient Samples-The cDNAs of peripheral blood from 13 healthy donors, 13 patients in chronic phase of CML at diagnosis, and 7 patients with CML in major or complete cytogenetic remission under an IFN␣-based therapy were studied (14). All patients and healthy donors gave written informed consent for the use of their samples.
Animals-The mouse strains, C57BL/6J (Harlan Biotech, Israel), IRF-1 Ϫ/Ϫ (kindly obtained from Dr. Rubinstein, The Weizmann Institute, Israel, originally from The Jackson Laboratory), and IRF-8 Ϫ/Ϫ (12), were maintained in microisolator cages in a viral pathogen-free facility. All animal work conformed to the guidelines of the animal care and use committee of the Technion.
Plasmids and Transfections-Mammalian expression vectors encoding for IRF-8, IRF-1, and PU.1 were all described previously (29). The pGL3-PML containing the 1.44-kb promoter region conjugated to luciferase (30) was a kind gift from Dr. de Thé (Hopital St. Louis, Paris, France). The PGL3-mut-PML promoter harboring 2-bp mutations (AA bases at positions 1422 and 1423 were changed to CG) that destroyed its IFN-simulated response element (ISRE, the DNA binding motif for IRFs) is described hereafter.
Site-directed Mutagenesis-Mutagenesis of the ISRE site within the human PML promoter was performed using the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) using the following primers: 5Ј-GAT CTA AAC CGA GAA TCG CGA CTA AGC TGG GGT CC-3Ј and the complementary oligonucleotide (positions 1404 -1435 in the Gen-Bank TM accession number X91752). The AA bases (1422 and 1423) were changed to CG and created an NruI site.
Isolation of Peritoneal Macrophages-Peritoneal macrophages were harvested as described previously (31). 2.5 ϫ 10 7 cells were plated in tissue culture Petri dishes (100 mm) and kept at 37°C and 5% CO 2 . After 4 h, nonadherent cells were removed, and 24 h later, adherent cells were either not treated or treated for 4 h with 100 units/ml IFN-␥ (CytoLab, Rehovot, Israel) and 10 ng/ml LPS (Sigma).
Isolation of Dendritic Cells (DCs)-The Flt3L method was used with minor modifications. Briefly, BM mononuclear cells were cultured in RPMI 1640 supplemented with 10% fetal bovine serum, and recombinant human Flt3L (100 ng/ml, PeproTech, Rocky Hill, NJ) for 9 days. During the final 24 h of culture, cells were either not stimulated or stimulated with IFN-␥ (100 units/ ml) and LPS (10 ng/ml). Nonadherent DCs were harvested by gentle pipetting leaving adherent cells behind.
RNA Extraction and Northern Blot Analysis-Total RNA from cells and tissues was extracted with Tri-Reagent (Sigma) according to the manufacturer's protocol, separated (15 g) on a 1.6% agarose/formaldehyde gel, and blotted as described before (31). Radiolabeled 600-bp murine PML cDNA fragment, generated by RT-PCR from total RNA of RAW264.7 cells by using the oligonucleotide primers 5Ј-AACTGTGGCTCCGTCCATAC-3Ј (sense) and 5Ј-CCACAGGGAACAACTGACCT-3Ј (antisense), was used as a probe.
The estimated amount of transcripts was normalized to GAPDH mRNA expression to compensate for variations in quantity or quality of starting mRNA and for differences in reverse transcriptase efficiency. The gene of interest/GAPDH ratio was expressed in percentage. The data are presented as the fold of induction for the gene of interest, e.g. the ratio between the relative mRNA levels in treated versus untreated cells.
Retroviral Transduction-Retroviral pMSCV-puro and pMSCV-ICSBP-puro vectors and preparation of retroviruses were described previously (2). Tot2 cells or DCs were transduced with retroviruses by spinoculation (2500 rpm for 1 h at 33°C, twice) in the presence of 4 g/ml Polybrene (Sigma) on days 1 and 2. On day 4, puromycin (2 g/ml) was added to select for transduced cells. Transduced cells were analyzed on day 6 (Tot2) or day 9 (DCs).
Western Blot Analysis-Protein extracts (12 g of nuclear/ cytoplasmic) were separated on a 10% SDS-PAGE and blotted to polyvinylidene difluoride membrane that was subjected to Ponceau S staining (Sigma). Subsequently, Western blot was performed with rabbit polyclonal anti-PML (Santa Cruz Biotechnology) as described previously (34). Immunoreactive proteins were visualized by the SuperSignal West Pico chemiluminescent ECL detection kit (Pierce).
Immunofluorescent Staining-Immunofluorescent staining was performed as described by Boulware and Weber (35). Cells were incubated for 2 h at room temperature with rabbit polyclonal anti-PML antibody (Santa Cruz Biotechnology) and subsequently for 1 h with the secondary antibody (donkey anti-rabbit Rhodamine Red X-conjugate, Jackson ImmunoResearch). Cells were stained with Hoechst 33342 (0.3 g/ml) in PBS, at room tem-perature for 5 min. Following mounting, fluorescence was visualized under Leica TCS-SP2 confocal laser scanning unit.
Bioinformatics Analysis of Promoter Region-Analysis of the promoter sequence was carried out with GEMS Launcher (Genomatix Software GmbH, Munich, Germany) using the FrameWorker tool which is based on the algorithms described.

IRF-8 Is Essential for the Induced Expression of PML in Macrophage
Cells and for the Constitutive Expression in Hematopoietic Organs-To identify genes that are regulated by IRF-8 and IRF-1 in macrophages, DNA microarray analysis was employed (18). For this purpose, we compared the expression profile of genes in peritoneal macrophages from WT and from IRF-8 Ϫ/Ϫ mice, before and following 4 h of stimulation with IFN-␥ and LPS. This analysis led to the identification of PML as a putative gene regulated by IRF-8 in activated macrophages (18). To further validate these results, Northern blot analyses were performed using nonstimulated as well as stimulated peritoneal macrophages and the macrophage cell lines RAW264.7 (expressing IRF-8) and CL-2 (originating from IRF-8 KO mice (28)). As seen in Fig.  1A, a strong induction of PML mRNA was observed following 4 h of treatment with IFN-␥ and LPS in peritoneal macrophages and in RAW264.7 cells (Fig. 1A, lanes 2 and 6, respectively). This PML up-regulation was not observed in peritoneal macrophages and CL-2 cells originating from IRF-8 KO mice (Fig. 1A, lanes 4 and 8, respectively). By using real time PCR, it was clear that the mRNA level of PML was induced by 5-fold in peritoneal macrophages from WT mouse strain and by over 7-fold in the macrophage cell line RAW264.7 but not in cells from IRF-8 and IRF-1 null mice (Fig. 1B). RAW264.7 cells and peritoneal macrophages were also treated with either IFN-␥ or LPS alone in 8-h intervals up to 24 h (data not shown). PML transcript levels were not affected by LPS at all times. However, IFN-␥ induced both IRF-8 and PML mRNA to similar levels shown in Fig. 1B for the combined treatment (IFN-␥ and LPS, for 4 h), and therefore the latter was used in subsequent experiments. Taken together, PML induction was compromised only in cells from the IRF-8 Ϫ/Ϫ and IRF-1 Ϫ/Ϫ mouse strains. We next examined the expression level of PML in hematopoietic tissues (thymus, liver, spleen, and lung), and in a nonhematopoietic tissue (heart). mRNA was extracted from these tissues from the WT as well as from the IRF-8 Ϫ/Ϫ mouse strains and subjected to Northern blot analysis. A strong mRNA signal corresponding to PML was observed in samples from the WT tissues except for the heart (Fig. 1C, compare lanes 1, 3, 5, and 7 . Relative expression of mRNA corresponding to PML was calculated as the ratio between the level of the amplified PML and GAPDH. The bar graph represents the fold of induction for PML, e.g. the ratio between the relative mRNA levels in treated versus untreated cells. mRNA was also extracted from various tissues of WT and KO mice (as indicated) and Northern blot was performed (C). The PML protein level was analyzed by Western blot (D) in cytoplasmic (cyt) and nuclear extracts (nuc) of RAW264.7 or CL-2 cells before (Ϫ) and after treatment for 18 h with IFN-␥ and LPS (ϩ). Ponceau staining was performed to ensure an equal amount of loading of proteins. FEBRUARY 23, 2007 • VOLUME 282 • NUMBER 8 to lane 9). We attribute this strong signal to the hematopoietic nature of these tissues, including the lung, which is rich in alveolar macrophages. Interestingly, PML mRNA levels in all samples extracted from the KO mice were significantly lower than in the WT counterparts. In fact, they were similar to the level of PML mRNA observed in the heart of the WT mice (Fig. 1C,  lanes 2, 4, 6, 8, and 10). Taken together, the data indicated that IRF-8 is essential for the expression of PML in hematopoietic tissues.

IRF-8 Regulates PML
To test the effect of IRF-8 on PML protein expression, cytoplasmic and nuclear extracts were prepared from RAW264.7 and CL-2 cells before and following activation with IFN-␥ and LPS. Western blot analysis was performed using antibody against PML, and as seen in Fig. 1D, induced expression of PML is observed only in nuclear extract from the WT activated cells (e.g. RAW264.7 cells) in perfect alignment with the mRNA data.

IRF-8 Is Essential for the Expression of Specific PML Isoform(s)-
In humans, a large number of alternatively spliced PML transcripts generate a variety of PML isoforms that differ in their C-terminal sequences (see illustration in Fig. 2A). Recent studies assign specific activities to specific isoforms (36 -39). Although the murine PML genomic locus exhibit similar architecture, the murine orthologues were not described yet.
To characterize the hematopoietic-specific PML isoform(s), we used a human promyelocytic cell line, U937, that was either activated or not with IFN-␥ and LPS for 4 h. The relative mRNA levels for PML isoforms (I-VI) as well as for IRF-8 was determined. We found that in addition to the expected induction in IRF-8 mRNA level (ϳ18-fold; Fig. 2B), the level of PML-I was significantly induced (10-fold; Fig. 2B). The levels of the other isoforms were marginally elevated as compared with that of IRF-8 and PML-I. This induction of PML-I was not observed in HeLa cells, and the expression levels were rather low before and following activation (data not shown). These results suggest that PML-I isoform is the major splice variant that functions in IFN-␥ and LPS-activated U937 cells.
The IRF-8-regulated Expression of PML Is Mediated by ISRE Element(s) Located in the PML Promoter-PML expression is induced by IFNs through identified ISRE and GAS elements in its promoter (30). However, the transcription factors involved in this induced expression were not determined. To determine the role of IRF-8 in the regulated expression of PML, reporter gene assays were performed. The characterized human PML promoter was transfected into RAW264.7 and CL-2 cells. Because IRF-8 is commonly engaged in heterocomplexes with PU.1 (3), the role of the latter in this regulated expression was studied as well. PU.1 is a transcription factor obligatory for hematopoiesis (40). In addition, because PML was selected in our DNA microarray analysis of macrophages from both IRF-8 Ϫ/Ϫ and IRF-1 Ϫ/Ϫ mice, we hypothesized that IRF-1 may also be an important regulator of PML, and therefore its effect on PML expression was investigated. Treatment of RAW264.7 cells with IFN-␥ and LPS for 18 h led to the induced expression of the reporter gene (data not shown) that was equivalent to that observed following co-transfection of IRF-8-expressing vector (Fig. 3A, black columns). In addition, transfection of either IRF-1 or PU.1 expression vectors led to a significant increase in the reporter gene levels similar to that of IRF-8. Combined transfection of IRF-8 and IRF-1 led to a further increase in reporter activity. A strong additive activation of the reporter gene was observed when all three factors were cotransfected.
Similar results were obtained with the CL-2 cells, lacking IRF-8. Here the reporter gene induced by IRF-8 was more profound (4-fold higher) than that of IRF-1 or PU.1 alone (Fig. 3A, white columns). This finding highlights the major role of IRF-8 in the expression of PML.
To test the importance of the characterized ISRE (41), the binding site for IRFs in the PML promoter, 2-bp substitutions at positions 1422 and 1423 (AA to GC) were introduced to the PML reporter construct (mut ISRE). As seen in Fig. 3A (gray bars), only PU.1 could still affect this mutated reporter construct in RAW264.7 cells. The effects of IRF-8 and IRF-1 were reduced by more than 50%. This demonstrates that transcriptional activation of the PML promoter by both IRF-8 and IRF-1 is indeed mediated by this ISRE.
To further characterize the binding activities of these three transcription factors, EMSA was performed using nuclear extracts from resting RAW264.7 cells. A DNA segment from the human PML promoter, encompassing the previously characterized ISRE element, was used as a probe (30). A major slow migrating band was observed (Fig. 3B, indicated by arrowhead) that was competed effectively by an excess of unlabeled PML ISRE and not by an excess of mutated ISRE oligonucleotide (Fig.  3B, lanes 2 and 3, respectively). This indicates that the competed band results from specific binding to the ISRE. To identify the proteins engaged in this binding, specific antibodies directed against ISRE-bound transcription factors were included in the binding reaction. It is clear that antibodies directed against IRF-1, IRF-8, and PU.1 led to the disappearance of this band (Fig. 3B, lanes 5-7, respectively). This was not observed with antibodies directed against IRF-2 or IRF-4 (Fig.  3B, lanes 8 and 9, respectively) or with preimmune serum or control IgG (Fig. 3B, lanes 4 and 10, respectively). This suggests this band represents a complex composed of IRF-1, IRF-8, and PU.1, which is characteristic of macrophage-specific promoters (for review see Ref. 3).
To test whether these three transcription factors also occupy the PML promoter region in vivo, ChIP analysis was performed in RAW264.7 and CL-2 cells (Fig. 3C). Although the binding of IRF-1 was not significant in resting RAW264.7 cells, the binding of IRF-8 and PU.1 was evident even under noninduced conditions. Following activation, strong induction of IRF-1 binding was evident (at least 20-fold), and a moderate increase in the binding of both IRF-8 and PU.1 was observed (2.25-and 3-fold, respectively; Fig. 3C). As expected, no binding of IRF-8 to the PML promoter was evident in CL-2 cells deficient for IRF-8. Yet the binding of both IRF-1 and PU.1 was augmented in activated cells (ϳ2-fold; Fig. 3C). The presence of acetylated histone 3 served as a positive control, whereas nonrelevant IgG served as a negative control in the ChIP assays. Together, our results point to the pivotal role of IRF-8 in the regulation of PML expression in macrophages through a specific ISRE located within the promoter and through a cooperative interaction with IRF-1 and PU.1.

IRF-8 Is Essential for the Formation of PML-NBs in
Macrophages-PML is the major scaffold protein of the NBs, and its disruption was reported in various myeloleukemias (19). To test the necessity of IRF-8 for the formation of PML-NBs, RAW264.7 cells were either treated or not treated with IFN-␥ and LPS for 18 h. The cells were stained with antibodies directed against PML and were visualized by immunofluorescence using confocal microscopy. It is clear from Fig. 4A that PML-NBs are observed only in the activated macrophage cells as described previously (42). Peritoneal macrophages were harvested, activated, and stained. Interestingly, PML-NBs were observed only in the WT cells treated with IFN-␥ and LPS and not in the IRF-8 KO cells (Fig. 4A). Because IRF-8 is also essential for the differentiation and maturation of DCs (7,43), we tested the presence of PML-NBs in conventional DCs. Unlike macrophages, NBs were detected in DCs of WT mice even without activation (Fig. 4B). However, NBs were not observed in DCs extracted from IRF-8 Ϫ/Ϫ mice unless transduced with IRF-8 (Fig. 4C).
To demonstrate that PML expression is dependent on IRF-8, we used the Tot2 myeloid progenitor cell line derived from an IRF-8 Ϫ/Ϫ mouse strain. These cells differentiate to mature macrophages within 6 days only following transduction of IRF-8 (6). As seen in Fig. 5, an increased level of PML mRNA was observed 6 days later only in cells transduced with IRF-8 ( Fig. 5B) and not in Tot2 cells transduced with the empty vector (Fig. 5A). Furthermore, activation with IFN-␥ and LPS caused a sharp increase in PML mRNA level that was accompanied by the appearance of PML-NBs only in IRF-8-transduced cells (Fig. 5B). Together, these results show that IRF-8 is essential for the formation of PML-NBs in activated macrophages.
Aberrant Expression of IRF-8 in CML Patients Is Accompanied by Abnormal PML Expression-As mentioned above, IRF-8 levels are very low in the peripheral blood of primary diagnosed CML patients. These levels are restored to normal values during remission following treatment with IFN-␣ (16,17). To further characterize the role of IRF-8 in the transcriptional regulation of PML in the context of malignancy, we examined whether PML follows the same expression pattern as IRF-8 in CML patients. cDNAs from peripheral blood of 13 healthy donors, 13 chronic phase CML patients at primary diagnosis, and 7 patients in remission following treatment with IFN-␣ were analyzed by real time RT-PCR. As seen in Fig. 6, a significant reduction in IRF-8 mRNA levels was observed in CML patients in comparison with healthy donors (median 0.27 and 1.89, respectively) that was accompanied by a significant  ). 24 h later, cells were harvested, and reporter activity was determined. Similarly, the luciferase reporter gene harboring the same PML promoter region with two point mutations in the ISRE (gray bars) was also transfected to RAW264.7 cells as above (for details see "Experimental Procedures"). This experiment was repeated three times yielding similar results. B, EMSA was performed with nuclear extract from untreated RAW264.7 cells. Competitors and antibodies were incorporated as indicated Pre, preimmune serum. The major complex is indicated by an arrowhead. C, fast ChIP assays were performed with RAW264.7 (upper) and CL-2 cells (lower) before (open bars) and after (black bars) 12 h of activation with IFN-␥ and LPS using antibodies directed against IRF-1, IRF-8, PU.1, and acetylated histone 3 (H3). ChIP with goat IgG served as negative control. Each experiment was repeated four times, and relative binding quantity and statistical significance were determined as described (33). FEBRUARY 23, 2007 • VOLUME 282 • NUMBER 8 reduction in PML mRNA levels (median 0.23 and 0.75, respectively). As expected, an increase in IRF-8 mRNA expression was observed in patients in remission under IFN-␣ (median 1.75). This increase was accompanied by a sharp up-regulation in PML levels in these patients in remission (median 3.71). These results strongly imply a role for IRF-8 as a regulator of PML in human myeloid cells and suggest a link between the disregulated expression of the PML gene and CML.

DISCUSSION
PML and PML-NBs were implicated in a variety of cellular processes, such as cell cycle regulation, apoptosis, proteolysis, tumor suppression, DNA repair, and transcription. Despite this, the function of PML bodies is still unknown. IRF-8 is a key factor in the differentiation, maturation, and function of myeloid cells. Intriguingly, in this study we show that the expression of PML in hematopoietic tissues and the induced expression of PML in response to IFN-␥ in myeloid cells are dependent on IRF-8. The stimulation of PML expression does not occur in myeloid cells from IRF-8 Ϫ/Ϫ mice but can be rescued by transduction of IRF-8 to these cells. This not only restores PML mRNA levels but also the formation of PML-NBs, suggesting that IRF-8 through the regulation of PML expression affects the existence and hence the activities of these NBs. We show that IRF-8, in association with IRF-1 and PU.1, binds to the ISRE motif within the PML promoter (in vitro and in vivo) and induces its transcription in differentiating myeloid cells in response to activation with IFN-␥ and LPS. Because deregulated IRF-8 expression is associated with the molecular pathology of CML (16,17), we looked for a correlation between PML expression level and CML. We show that the reduced expression level of IRF-8 in primary diagnosed CML patients is accompanied by a significant reduction in PML transcription. Like IRF-8, PML levels are restored to the normal range when these patients are in remission because of treatment with IFN-␣.
Interestingly, higher expression of PU.1 is required for macrophages than for neutrophil development (40). In addition, PU.1 mutations have been observed in some AML patients (49). The importance of IRF-1 in myeloid cell differentiation is highlighted because of its functional inactivation in various neoplastic disorders such as accelerated exon skipping in CML (50) and deletions in AML and myelodysplasias (51). IRF-8 is also a myeloid commitment factor. IRF-8 Ϫ/Ϫ mice harbor an increased  Transduced cells were selected with puromycin. Each day following transduction, cells were harvested, and total RNA was prepared to analyze the expression level of PML transcript by real time RT-PCR. Six days later, cells were treated with IFN-␥ and LPS for 6 h, and RNA was prepared and analyzed as above. In addition, to visualize the presence of PML-NBs (intense red spots within the nuclei (blue)), cells were treated with IFN-␥ and LPS for 18 h and subjected to immunofluorescent and Hoechst staining as described under Fig. 4A. number of myeloid progenitor cells defective in the ability to differentiate to macrophages and consequently mature mainly to granulocytes. Accordingly, these IRF-8-deficient mice eventually develop a CML-like disease (6,12). In humans, 66% of AML and 79% of CML patients exhibit reduced levels of IRF-8 or no expression at all. PML modulates key tumor suppressor genes such as p53 and the retinoblastoma gene, and consequently PML KO mice are tumor-prone (52,53). Furthermore, PML-RAR␣ is thought to act as a dominant negative of PML, thereby promoting APL. The ability of PML to modulate key tumor-suppressive pathways whose inactivation has been directly involved in human tumorigenesis suggests that its functional and physical inactivation may participate in the pathogenesis of malignancies other than APL (19). IRF-8 is also a myeloid-specific tumor suppressor gene, and its reduced expression leads to myeloleukemias (16,17). Together with our results, this suggests that the tumor suppression activity of IRF-8 might be mediated, at least in part, through the regulation of its target gene PML. The tumor suppression activity of PML is partially attributed to its role in genome stabilization. PML inactivation, or in this case down-regulation, is thought to cause genomic instability and/or render cells less prone to apoptosis in response to DNA damage allowing the accumulation of additional genetic lesions that cause a neoplastic phenotype (20). Along this line, ablation or down-regulation of IRF-8 will lead to the down-regulation of PML expression in myeloid cells and hence may result in genome instability and reduced apoptosis. Indeed, myeloid cells from IRF-8deficient mice exhibit reduced spontaneous apoptosis and a significant decrease in apoptosis induced by DNA damage (54). Furthermore, IRF-8-deficient mice develop nonrapid leukemia suggesting that additional events lead to the leukemic phenotype (12,55). In addition, as shown here, macrophages and dendritic cells extracted from IRF-8 Ϫ/Ϫ mice exhibit low levels of basal PML expression that could not be induced with IFN-␥. Therefore, this can lead to a long leukemia latency during which additional genetic events are taking place and account for a later aggressive stage. Similar events were reported with APL transgenic models; the expression of the PML-RAR␣ fusion protein was not sufficient for APL development and acquired nonrandom chromosomal abnormalities that may have led to the acquisition of additional changes that provide an advantage to the transformed cells (56). Accordingly, the low level of IRF-8 in peripheral blood of primary diagnosed patients is accompanied by a very low level of PML that is restored upon remission because of IFN-␣ treatment. Together, these data suggest that the multistage leukemia observed upon repression of IRF-8 expression may be mediated by PML. Lack of PML expression may lead to genomic instability, enhanced survival of cells with damaged DNA, and the subsequent development of aggressive leukemia (22). Our data therefore suggest that PML also functions as a tumor suppressor gene in CML.
The PML gene consists of nine exons, and because of alternative splicing a large number of PML proteins that differ in their C terminus were reported in humans (57). Although the murine PML isoforms were not characterized yet, an increasing body of evidence assigns different activities to some of the human isoforms suggesting that PML splice variants could have potentially diverse cellular functions (36 -39, 58). We report here that in U937 cells the expression level of PML-I is mainly elevated in activated cells. This fits with recent a report demonstrating that PML-I protein is much more abundant in macrophages (59). It was recently reported that PML-I forms a complex with AML1, targets it into nuclear bodies, enhances AML1-mediated transcription, and stimulates differentiation of myeloid cells (37). The translocation generating the AML1-ETO fusion protein (8,21) is one of the most frequent chromosomal abnormalities associated with AML. Interestingly, overexpression of AML1-ETO in bone marrow cells derived from IRF-8 Ϫ/Ϫ mice induces myeloblastic transformation that is typical of AML. Thus, the IRF-8 defect synergizes with AML1-ETO to transform and expand a myeloblast population (60). Together, this suggests that IRF-8, through the regulation of PML-I, affects the activity of AML1 and subsequently the capacity of myeloid progenitor cells to differentiate to monocyte/macrophage lineage rather than to granulocytes.
Myeloid progenitor cells can receive two incoming differentiation cues, IFN-␥ and RA. Although RA preferentially stimulates granulopoiesis (61), IFN-␥ stimulates maturation toward monocytes/macrophages (62). This ability of RA and IFN-␥ to modulate myeloid progenitor cell differentiation toward these two lineages is mediated via PML. Thus, our simplified working model assigns a pivotal role for PML as an integrator of incoming signals. RA, through the binding to its receptor complex RAR␣-retinoid X receptor, leads to PML-dependent differentiation of myeloid progenitor cells to granulocytes (52,63). The analogous route of differentiation, mediated by IFN-␥, leads to the induction of IRF-8 through the JAK-STAT signaling (64). On the one hand, IRF-8 leads to the induction of key regulatory components such as the transcription factor EGR-1 (65) and on the other hand to the induction of the PML-I isoform that forms a complex with AML1 resulting in the differentiation of myeloid progenitor cells to monocytes. In conclusion, our study suggests an important role for PML in myeloid cell differentiation to monocytes and extends its tumor suppression activity not only to APL but also to CML. The expression levels of IRF-8 and PML of each healthy donor, patient with CML, and patient in remission following cytogenetic response to IFN-␣ were quantitated from peripheral blood cDNA by real time RT-PCR and is outlined in the percentage of normalized to GAPDH. Healthy donors, n ϭ 13; CML primary diagnosed patients, n ϭ 13; IRF-␣ major cytogenic responders, n ϭ 7. Medians are indicated by lines. ns indicates not significant; * indicates p value Ͻ 0.05; ** indicates p value Ͻ 0.01; *** indicates p value Ͻ 0.001 as determined by Kruskall Wallis test, adjusted for multiple comparisons with Dunn multiple comparisons t test.