Halofuginone Has Anti-Proliferative Effects in Acute Promyelocytic Leukemia by Modulating the Transforming Growth Factor Beta Signaling Pathway

Promyelocytic leukemia-retinoic acid receptor alpha (PML-RARα) expression in acute promyelocytic leukemia (APL) impairs transforming growth factor beta (TGFβ) signaling, leading to cell growth advantage. Halofuginone (HF), a low-molecular-weight alkaloid that modulates TGFβ signaling, was used to treat APL cell lines and non-obese diabetic/severe combined immunodeficiency (NOD/SCID) mice subjected to transplantation with leukemic cells from human chorionic gonadotrophin-PML-RARα transgenic mice (TG). Cell cycle analysis using incorporated bromodeoxyuridine and 7-amino-actinomycin D showed that, in NB4 and NB4-R2 APL cell lines, HF inhibited cellular proliferation (P<0.001) and induced apoptosis (P = 0.002) after a 24-hour incubation. Addition of TGFβ revealed that NB4 cells were resistant to its growth-suppressive effects and that HF induced these effects in the presence or absence of the cytokine. Cell growth inhibition was associated with up-regulation of TGFβ target genes involved in cell cycle regulation (TGFB, TGFBRI, SMAD3, p15, and p21) and down-regulation of MYC. Additionally, TGFβ protein levels were decreased in leukemic TG animals and HF in vivo could restore TGFβ values to normal. To test the in vivo anti-leukemic activity of HF, we transplanted NOD/SCID mice with TG leukemic cells and treated them with HF for 21 days. HF induced partial hematological remission in the peripheral blood, bone marrow, and spleen. Together, these results suggest that HF has anti-proliferative and anti-leukemic effects by reversing the TGFβ blockade in APL. Since loss of the TGFβ response in leukemic cells may be an important second oncogenic hit, modulation of TGFβ signaling may be of therapeutic interest.


Introduction
Transforming growth factor beta (TGFb) is a cytokine that regulates multiple cellular responses, including inhibition of cell proliferation and induction of differentiation, senescence, and apoptosis [1,2]. Its actions are mediated by binding to the serine/ threonine kinase receptor TbRII that recruits and activates TbRI, which in turn phosphorylates downstream targets. These include the proteins SMAD2 and SMAD3, which translocate to the nucleus in a complex with the common mediator SMAD4 to regulate transcription of target genes [3,4]. The tumor suppressor responses of TGFb are essential for maintaining homeostatic control of normal cell growth and cells in the early phases of tumorigenesis. Among the TGFb-mediated effects in premalignant cells are the suppression of c-Myc expression [5] and the induction of the cell cycle inhibitors p15 and p21. Although these actions imply a tumor suppressor role for TGFb, its effects are both cell-and context-dependent. In that regard, Siegel et al. have shown that activation of TGFb delays the appearance of primary mammary tumors, and mice deficient in TGFb signaling are prone to earlier tumor development, suggesting that the tumor suppressor response of TGFb is important in the early stages of tumorigenesis. In contrast, mice expressing an activated TGFb receptor exhibited increased metastatic lung foci, consistent with a pro-oncogenic effect of this pathway in late-stage disease [6]. In addition, advanced disease is accompanied by increased expression and activation of the ligand but decreased TGFb responsiveness, thus facilitating tumor cell growth [7].
Deregulation of TGFb signaling may alter hematopoiesis, causing a predisposition to leukemia. In contrast to solid tumors, mutations in SMAD genes are rare in leukemia and disruption of TGFb responsiveness is commonly secondary to either (a) altered transcription, as described in acute myeloid leukemia with translocation t(8; 21), in which the AML1/ETO chimeric protein represses transcription of TGFb-responsive genes [8] or (b) disruption of TGFb target gene expression such as the cell cycle regulators c-Myc, p15 and p21, which are commonly associated with leukemogenesis [9].
The role of TGFb in leukemogenesis has been recently studied in acute promyelocytic leukemia (APL), a distinct subtype of acute myeloid leukemia (AML) associated with t (15;17) and expression of the promyelocytic leukemia-retinoic acid receptor alpha (PML-RARa) hybrid protein. A gene expression study using microarrays has revealed that TGFb was downregulated in APL compared with most non-APL samples [10]. In contrast, Raza et al. have described elevated TGFb protein expression by immunohistochemistry in bone marrow biopsies of 23 APL patients [11]. Lin et al. have demonstrated that the cytoplasmic isoform of PML (cPML) is essential for TGFb signaling and Pml-null primary cells are resistant to TGFb-dependent growth arrest, induction of cellular senescence, apoptosis, phosphorylation of Smad2/3, and induction of p15 and p21 expression. Restoration of cPML fully rescued these defects [12]. Since cPML function is impaired in APL blasts, through the formation of cPML/PML-RARa heterodimers, the authors hypothesized that this would be the molecular mechanism of resistance to TGFb anti-proliferative responses [13].
To better characterize the deregulation of the TGFb pathway in APL and to determine its potential as a therapeutic target, we took advantage of the human chorionic gonadotrophin (hCG)-PML/ RARa transgenic model and analyzed the effects of halofuginone (HF; dl-trans-7-bromo-6-chloro-3-[3-(3-hydroxy-2piperidyl)acetonyl]-4(3H)-quinazolinone hydrobromide), which is a low-molecular-weight alkaloid that has been shown to modulate TGFb signaling. In several cultured cell lines, this drug decreased TGFbinduced phosphorylation of SMADs 2 and 3 and induced expression of inhibitory SMAD7 mRNA [14]. HF also reduced tumor growth in in vivo models of pheochromocytoma [15], brain tumors [16], and hepatocellular carcinoma [17]. The effects of HF in hematopoietic malignancies have not been previously described. Our results demonstrate that HF treatment induces anti-proliferative and pro-apoptotic effects, up-regulates TGFb target gene expression, and significantly reduces the leukemic burden in vivo.

Ethics Statement
This study was approved by the Research Ethics Committee of the University Hospital of the Medical School of Ribeirão Preto, University of São Paulo, process number 3865/2005). Experiments using mice were conducted according to national guidelines for the care and use of laboratory animals (Brazilian College of Animal Experimentation) and was approved by the institutional Animal Experimentation Ethics Committee (protocol number 088/2007).

Cell culture
NB4, a permanent cell line harboring t(15;17) [18], and its derivative NB4-R2, in which all-trans retinoic acid (ATRA)unresponsiveness is associated with a point mutation in the retinoid-binding domain of PML-RARa [19], were used for in vitro assays. Cells were cultured in RPMI 1640 with 10% heatinactivated fetal calf serum (FCS; Gibco BRL, UK) and maintained at 37uC in a CO 2 -humidified incubator.

Treatment of APL cell lines with HF
HF was kindly provided by Prof. Arnon Nagler (Chaim Sheba Medical Center, Tel Hashomer, Israel). Stock solutions of 1 mg/ mL were kept at 280uC until use. Subsequently, working solutions of 10 ng/mL were freshly prepared by diluting the stock solution with autoclaved water (for cell culture assays) or 0.9% NaCl (for in vivo studies).
Cell suspensions containing 5610 5 cells/mL of culture were treated with increasing doses of HF (6.25-200 ng/mL), which was directly added to the medium, and then cells were harvested after 24, 48, or 72 hours of incubation as indicated. Cell viability measurements were recorded with an initial minimum viability of at least 95% as determined by the Trypan blue assay. For cell cycle analysis and gene expression studies, NB4 cells were also subjected to concurrent treatment with TGFb (0.5 ng/mL; Sigma-Aldrich, St. Louis, MO, EUA) as indicated.

Cell proliferation and apoptosis assays
For the analysis of proliferation and the cell cycle, NB4 and NB4-R2 cells were treated with HF as described above for 24 hours and then were subjected to immunofluorescent staining of incorporated bromodeoxyuridine (BrdU) and 7-amino-actinomycin (7-AAD), followed by flow cytometric analysis using the BrdU Flow Kit (BD Biosciences, San Jose, CA, USA). In this method, BrdU, an analog of the DNA precursor thymidine, is incorporated into newly synthesized DNA by cells entering and progressing through the S (DNA synthesis) phase of the cell cycle. The incorporated BrdU is stained with a specific BrdU allophycocyanin (APC)-conjugated antibody and 7-amino-actinomycin (7-AAD), a dye that binds to total DNA. With this combination, a two-color flow cytometric analysis permits the enumeration and characterization of cells that are actively synthesizing DNA (BrdU incorporation) according to their cell cycle position (i.e., G0/1, S, or G2/M phases defined by 7-AAD staining intensities). Cells were incubated with 10 mM of BrdU during the last 30 minutes of culture, and then processed according to the manufacturer's recommendations. For the NB4 cell line only, TGFb was added or not to the culture medium to evaluate the tumor suppressive effects of TGFb.
For the analysis of apoptosis after 24 and 48 hours of HF treatment, NB4 cells that were treated with different concentrations of HF were evaluated using the Trypan blue exclusion assay. Concomitantly, apoptosis was determined using the Annexin V and propidium iodide (PI) binding assay (BD Biosciences), and then analyzed by flow cytometry. Each sample was washed in 16 phosphate-buffered saline (PBS), and then incubated with 5 mL of Annexin V, 5 mL of PI, or both at 4uC for 15 min. Subsequently, 400 mL of binding buffer was added to the samples.
All experiments were performed in triplicate and in each sample, 10,000 events were acquired in a FACSCalibur flow cytometer (BD Biosciences) and analysis was performed using Cell Quest software.
The effective dose at 50% (ED 50 ) of HF was calculated based on the inhibition of proliferation in the NB4 cell line as determined by the BrdU incorporation assay. This analysis was performed using CalcuSyn software (Biosoft, Great Shelford, UK).

Analysis of TGFb target gene expression by real-time polymerase chain reaction (PCR)
NB4 and NB4-R2 cells were treated as above for 24 or 72 hours, and wherever indicated, TGFb (0.5 ng/mL) was added 1 hour before HF to the culture. Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Reverse transcription of 500 ng of RNA was performed using the cDNA High Capacity Archive kit (Applied Biosystems, Foster City, CA, USA). Subsequently, the mRNA expression of TGFb target genes that are involved in cell cycle regulation (TGFB, TGFBRI, p15, p21, SMAD3, and MYC) was evaluated by real-time PCR using the Taqman method. All the cDNA samples were diluted five times and were processed in duplicate. The PCR amplification was performed in 40 cycles, using the Taqman PCR master mix, in an SDS (Sequence Detection System) 5700 plataform connected to a 7300 Real-Time PCR System (Applied Biosystems). The probes used for amplification were synthesized using the Assay-on-Demand System (Applied Biosystems) with the following GeneBank sequences: TGFb The expression of the glyceraldehyde-3phosphate dehydrogenase (GAPDH) housekeeping gene was determined using the PDAR reagent (Pre-Developed Assay Reagent; Applied Biosystems) and was used to normalize the data. The 2 2DDCT method was used in the analysis of the PCR data and the relative gene expression in a particular sample was determined as follows: relative amount of target = 2 2DDCT value.

Detection of TGFb protein expression
After treatment with HF as described above, total protein extracts were obtained according to Schreiber et al. [20]. Briefly, NB4 cells were washed twice in cold PBS, lysed with lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM Na 2 EDTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM bglycerophosphate, 1 mM Na 3 VO 4 , 1 mg/mL leupeptin) containing a protease inhibitor mixture P8340 (Sigma, St. Louis, MO, USA) and homogenized in a Dounce system (model D-130, Biosystems, Brazil) for one minute on ice. Lysates were centrifuged at 20,0006 g for 30 minutes at 4uC and the supernatants were removed. Protein concentration was determined by the Bradford's method [21]. Proteins were submitted to SDS-PAGE and a total of 30 mg of protein from each sample was transferred to polyvinylidene fluoride (PVDF) membranes (GE Lifesciences, Pittsburgh, PA, USA) [22]. Membranes were blocked with 5% non-fat dry milk in 0.1% Tween-TBS and incubated with the specific antibodies. Mouse anti-b-actin was purchased from Santa Cruz Biotechnology (California, USA). Rabbit anti-TGFb, anti-TGFb receptor II, anti-Smad3 and horseradish peroxidase-conjugated goat antirabbit IgG secondary antibody were purchased from Cell Signaling (Beverly, MA, USA), and goat anti-mouse IgG secondary antibody from GE Lifesciences (Pittsburgh, PA, USA). The antibody-protein complex was detected using the ECL Western Blotting Detection Reagents (GE Lifesciences).

HF treatment in an APL transplant model
To analyze in vivo effects of HF, irradiated immunodeficient non-obese diabetic/severe combined immunodeficiency (NOD/ SCID) mice were injected with leukemic cells from hCG-PML-RARa transgenic mice (TM). The TM were kindly provided by Prof. Pier Paolo Pandolfi (Beth Israel Deaconess Medical Center, Harvard Stem Cell Institute, Boston, MA, USA) and their generation has been described elsewhere [23]. Notably, in this transgenic model, a lethal form of leukemia that closely resembles human APL occurs after a long pre-leukemic phase (12-15 months) and affects only 10-15% of the TM [24]. We have established a transplant model, in which all animals develop leukemia after 14 days from the transplant. Briefly, leukemic cells, previously maintained at 280uC, were thawed and suspended in RPMI 1640 with 10% FCS. After Trypan blue exclusion testing and 12 hours after sublethal cobalt irradiation with 250 cGy, 2610 6 viable cells were intravenously injected into the ocular plexus of CB17-Prkdc scid /J 10-to 12-week-old NOD/SCID mice (The Jackson Laboratory, Bar Harbor, Maine, USA). Animals were maintained under pathogen-free conditions and received autoclaved food and water ad libitum. Experiments were conducted according to institutional and national guidelines for the care and use of laboratory animals The definition of the dose and mode of administration of HF was based on previous reports of the in vivo use of the drug in solid tumors and fibrosis models [14,15,17,25,26,27,28], and pilot experiments were performed to confirm leukemic infiltration in transplanted animals and to test the efficacy and toxicity of HF. Twenty-four hours after the transplant procedure, NOD/SCID mice received treatment with vehicle only (0.9% NaCl; n = 5) or 150 mg/kg/day HF (n = 5) as an intraperitoneal injection for 21 consecutive days. At the end of the experiment (day 21), mice were sacrificed under ketamine anesthesia after being subjected to a cardiac puncture to obtain peripheral blood (PB) samples. Nonleukemic, age-matched NOD/SCID were used as controls (n = 3). Animals were maintained under pathogen-free conditions and received autoclaved food and water ad libitum.

Analysis of the hematological and molecular responses to HF in vivo
For monitoring PB counts, mice were bled from the tail before transplantation and 10 days after the beginning of HF injections. Automated counts (hemoglobin, white blood cells, and platelets) were performed using a T-890 Coulter cell counter (Coulter Corporation, Hialeah, FL, USA), and differential counts were obtained from Leishman-Wright-Giemsa-stained smears. In addition, after euthanasia, bone marrow (BM) cells were obtained by flushing the bone cavities of femurs and tibiae with RPMI 1640 containing 10% FCS. Cells were washed once, and then the pellet was resuspended in PBS at a concentration of 10 6 /mL. Approximately 10 5 cells were used for cytospin slide preparation and staining with Leishman-Wright-Giemsa, and then 10 6 cells were used for DNA extraction and PCR analysis to detect PML-RARa, as previously described by van Dongen [29].
For morphological analysis of PB and bone marrow slides, a minimum of 100 PB and 200 BM cells were counted, and then myeloid cells were classified as immature, intermediate, or mature, according to the Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice [30]. Because extensive spleen infiltration was observed in leukemic animals, the relation between the spleen and body weight was also assessed to further quantify the hematological response to HF.

Quantification of TGFb in the serum of leukemic mice treated with HF
Serum and BM samples of transplanted and control NOD/SCID mice were obtained after euthanasia, as previously described, and used to quantify TGFb using the enzyme-linked immunosorbent assay (ELISA). Serum samples were stocked at 280uC until use. BM cells were washed twice with 16PBS and total protein extracts were prepared by suspending cell pellets in a lysis buffer (20 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, and 1 mM PMSF) supplemented with a cocktail of protease and phosphatase inhibitors (Sigma Aldrich), and then subjecting the samples to three cycles of sonication. The resulting lysates were centrifuged at 13000 rpm for 10 minutes at 4uC, and then supernatants (a minimum total protein of 1 mg) were stocked at 220uC until use. Subsequently, serum and BM protein samples were subjected to activation of latent TGFb1 to immunoreactive TGFb by acidification, and then used in a sandwich ELISA assay using the Quantikine TGFb1 Immunoassay kit (R&D Systems, Minneapolis, MN, USA), according to the manufacturer's recommendations. The optical density was determined using a microplate reader set at an absorbance of 450 nm and a standard curve was generated to report the results of TGFb quantification in pg/mL.

Statistical analysis
The effects of HF on APL cell lines regarding cell proliferation, apoptosis, BCL-2 expression, and transcription of TGFb target genes were analyzed by multivariate analysis using the mixed linear model. Comparisons simultaneously included the following variables: treatment with variable doses of HF, cell type (either NB4 or NB4-R2), and addition of TGFb. For in vivo experiments, differences in blood cell counts, percentage of immature cells, comparisons of the spleen/body weight ratio, and TGFb quantification in leukemic and control mice were evaluated by analysis of the variance (ANOVA) between groups followed by the Bonferroni's correction post-test. In all comparisons, a significance level of P,0.05 was considered to be significant. Statistical analyses were performed using SPSS 13.0 software.

HF exerts anti-proliferative actions on APL
Cell cycle analysis, evaluated by the immunofluorescent staining of incorporated BrdU and 7-AAD, showed that, in both NB4 and NB4-R2 cell lines, HF inhibited cellular proliferation (P,0.001) and induced apoptosis (P = 0.002) in a dose-dependent manner. Increasing doses of HF blocked cell cycle progression at the G1/S transition (Fig. 1A-C and Figure  S1). In addition, doses greater than 35 ng/mL of halofuginone also induced an accumulation of cells in G2+M. Notably, NB4-R2 cells were more susceptible to HF than NB4 cells (P,0.001) (Fig. 1A-C and Figure S1). Similar results were detected using the Trypan blue exclusion test. Compared to control samples, treatment with 50 ng/mL HF reduced NB4 cell viability by 44.3% (40-46.77%) and 70.5% (62.3-72.8%) after 24 and 48 hours of treatment, respectively. Indeed, no increase in viable cells numbers was observed after 24 hours of treatment with more than 50 ng/mL HF (Fig. 1D-E). Simultaneously, doses of HF greater than 50 ng/mL induced initial and latestage apoptosis after 24 hours of treatment as determined by Annexin V and PI binding assays and flow cytometric analysis (P,0.001). In addition, after 48 hours of incubation, HF induced a 1.6-and 2.5-fold increase in apoptosis at 100 and 200 ng/mL, respectively ( Fig. 2A-B).
When TGFb was added to NB4 cell cultures with HF at concentrations of 25-50 ng/mL, a significant increase (P,0.001) in growth inhibition was observed, suggesting an additive effect between the two molecules. At doses $100 ng/mL of HF, less than 10% of the cells were in S phase, suggesting that a maximum effect was reached (Fig. 3A). Regarding apoptosis, the effect of HF was similar in the presence or absence of TGFb (Fig. 3B), except for the treatment with 12.5 ng/mL of HF. Although the multivariate analysis by the mixed linear model revealed that apoptosis is identified by green or blue lines, respectively. Results confirmed that, in both cell lines, HF inhibited cell proliferation (P,0.001) and caused apoptosis (P = 0.002), although the pro-apoptotic effect in NB4 cells was visually less evident than the one observed in NB4-R2. (D-E) Analysis of the number of viable NB4 cells by Trypan blue exclusion according to the incubation time (P,0.001) with various doses of HF (P,0.001), indicated by the right-sided legend. Forty-eight hours after incubation with doses greater than 50 ng/mL of halofuginone, the number of viable NB4 cells significantly reduced. doi:10.1371/journal.pone.0026713.g001 addition of TGFb is statistically relevant for apoptosis (P = 0.013), this was due to the results detected with 12.5 ng/mL of HF.
The ED 50 of HF for the inhibition of proliferation after 24 hours of incubation was calculated based on the percentage of cells in S phase. Without concomitant treatment with TGFb, the ED 50 of HF was 18 ng/mL (95% confidence interval (95% CI): 15.5-20.8 ng/ mL; R 2 : 0.99). In the presence of TGFb, the ED 50 was 10 ng/mL (95% CI: 7.45-15.7 ng/mL; R 2 : 0.98), thus corroborating the potential additive anti-proliferative action of TGFb and HF.
To verify that these effects were caused by HF treatment and not a late cellular response to cell growth inhibition, some genes were selected for a 24-hour HF incubation and real time PCR analysis, including the addition of exogenous TGFb to NB4 cultures. Consistent with the 72-hour treatment assay, after 24hour treatment, HF significantly induced the transcription of TGFb target genes (TGFbRI and p21) and repressed MYC expression in NB4, regardless of the addition of exogenous TGFb (Fig. 6). Indeed, as analyzed using the mixed linear model, the results were similar in the presence or absence of TGFb for TGFbRI (P = 0.699), SMAD3 (P = 0.963), p21 (P = 0.295), and MYC (P = 0.768).
HF demonstrates anti-leukemic effects in the in vivo model of APL Table 1 shows that HF treatment for 21 days resulted in an increase to normal hemoglobin levels and platelet counts in leukemic mice. The WBC was significantly lower in leukemic mice treated with HF compared with untreated controls. However, the cytomorphological analysis of Leishman-stained PB smears demonstrated that HF treatment reduced the percentage of immature cells in the bone marrow (66.3617.9% versus 2769.3%; P,0.01) and peripheral blood samples (36.5621.04% versus 1565.09%; P = 0.06), although blasts were still detectable (Table 1). PML-RARa detection by PCR confirmed the engraftment of leukemic cells in all transplanted animals. As expected by the morphological analysis of the bone marrow, samples obtained after treatment tested positive for PML-RARa.
Since extensive spleen infiltration has been observed in the hCG-PML-RARa mouse model [23], the relation between the spleen and body weight was assessed to further quantify the hematological response to HF. Compared with NOD/SCID wildtype (WT) control mice, as expected, leukemic animals showed a higher spleen/body weight ratio (P,0.001). Treatment with HF resulted in a significant reduction of the spleen/body weight ratio (P,0.05; Table 1). In addition, morphological analysis of Leishmanstained slides of spleen imprints confirmed the infiltration of immature cells resembling promyelocytes in leukemic animals and demonstrated normal cellular distribution in the HF-treated mice.

HF reverses TGFb inhibition in PML-RARa leukemic mice
TGFb quantification in serum samples showed lower levels of the cytokine in leukemic mice compared with WT controls (P,0.0001). In animals that received HF treatment, TGFb levels were similar to those in controls, suggesting that the drug could reverse TGFb inhibition in PML-RARa leukemic mice. When TGFb protein expression was verified in the bone marrow lysate, a similar tendency of TGFb down-regulation in leukemic animals was observed, although the data were not statistically significant (Fig. 7).

Discussion
The present study demonstrates for the first time that HF has anti-leukemic properties, reducing tumor growth and inducing apoptosis in vitro and in vivo. A few studies have reported that HF inhibits angiogenesis in vitro [31] and T cell activation [32], with doses ranging from 5 to 400 ng/mL. Previous studies regarding Cell cycle effects were determined by the BrdU incorporation assay. Continuous and dotted lines represent the effects of HF in the absence or presence of exogenous TGFb in cell cultures, respectively. The multivariate analysis by the mixed linear model revealed that addition of TGFb is statistically relevant (P,0.001) for cell proliferation and apoptosis (P = 0.013). Regarding the latter, the percentage of apoptotic cells was similar in samples treated with 25-200 ng/mL of HF irrespective of TGFb addition. Therefore, the significant difference reflects the results detected with 12.5 ng/mL of halofuginone. The effective dose at 50% (ED 50 ) of HF for the inhibition of proliferation was 18 ng/mL, whereas, in the presence of TGFb, the ED 50 was reduced to 10 ng/mL, thus suggesting a potential additive anti-proliferative action of TGFb and HF. Data represent the results of three independent experiments. doi:10.1371/journal.pone.0026713.g003 HF pharmacokinetics have suggested that intraperitoneal delivery of a 1.5-mg/kg dose in mice produced plasma concentrations between 173 and 209 ng/mL at 10 minutes after administration, with a 100% bioavailability and rapid distribution of the drug to all tissues, except the brain. In addition, HF doses greater than 1.5 mg/kg proved excessively toxic to mice [33]. In vivo studies using HF have reported significant anti-tumoral effects of the drug with intraperitoneal doses varying from 50 to 300 mg/kg with no significant toxicity. Therefore, the doses we have chosen for both in vitro and in vivo assays were within the range previously reported to be effective and non-toxic. Importantly, HF has been previously shown to reduce solid tumor growth, but no published data exist on its actions in hematologic diseases.
HF exerted a potent anti-proliferative effect with an ED 50 of 18 ng/mL (95% CI: 15.5-20.8 ng/mL), and was as effective in NB4 as in NB4-R2 cells. HF-induced cell growth inhibition was associated with dose-dependent up-regulation of TGFb target genes. Treatment resulted in an increase in the number of TGFb  mRNA transcripts and TGFb protein expression, which is consistent with the previously demonstrated positive feedback of the pathway [34]. Moreover, the expression of other genes that encode proteins of the TGFb pathway (including TGFbRI and SMAD3) were modulated by HF treatment. SMAD3, when phosphorylated and released from the receptor complex by TGFbRI, forms a heterodimeric complex with SMAD4, which then translocates to the nucleus to regulate transcription of target genes. Consistent with the activation of the TGFb pathway induced by HF, our results showed up-regulation of the cyclindependent kinase inhibitors p15 and p21 and down-regulation of MYC, which may have contributed to TGFb growth inhibition. MYC appears to be a very specific target of TGFb because its transcription is mediated by SMAD3 binding to a responsive element in the promoter of the gene [35,36]. In addition, our findings agree with the evidence that, upon TGFb stimulation, down-regulation of MYC creates a positive feedback loop that further amplifies p15 and p21 expression [1,2,37].
HF inhibited proliferation and induced apoptosis in the absence of exogenous TGFb, but addition of the latter potentiated the effects of HF and resulted in a reduction of the ED 50 concentration by 45% and further up-regulated the expression of TGFbRI, SMAD3 and p21. Of note, as shown in figure 6, the addition of TGFb alone to NB4 cells was able to up-regulate its target genes. This can be attributed to the TGFb property of activating its own mRNA expression and protein secretion. Therefore, one could hypothesize that the up-regulation of TGFb target genes by HF results from a combination of increased secretion of TGFb and a direct effect of the drug on transcription.
The results obtained with HF treatment in a murine transplant model of APL in NOD/SCID mice reinforced the potential antileukemic effects of the drug. Mice transplanted with PML-RARa cells and treated with HF presented hematological remission in peripheral blood, bone marrow, and spleen as determined by blood cell counts and cytological analysis. HF did not induce differentiation of leukemic blasts, and the improvement of hematopoiesis resulted from the decrease of the leukemic burden.
In addition, consistent with the evidence of the downregulation of TGFb signaling in APL, TGFb protein levels were decreased in leukemic PML-RARa animals and HF increased TGFb levels to values very similar to those of control non-leukemic mice. The partial hematological remission observed in HF-treated animals may be associated with the direct induction of the cytokine TGFb itself, known to play a key role in the regulation of human hematopoietic stem cell quiescence, proliferation and differentiation [38]. Although the effect of TGFb as a potent negative regulator of hematopoieses may contribute to tumor control, its role in leukemogenesis has been more recently associated with the findings of disruption in TGFb signaling either by mutational inactivation or down-regulation of its components [13]. Therefore, we have attributed the TGFb-dependent growth inhibitory effects of HF to the indirect transcription regulation of cell cycle target genes.
In contrast to our findings, previous studies of fibrosis models have shown that HF inhibits TGFb-induced phosphorylation of Table 1. Peripheral blood cell counts, percentage of immature cells, and spleen/body weight ratio of wild-type control mice (WT; n = 3) and NOD/SCID leukemic mice that were treated with halofuginone (HF) (LEUK-HF; n = 5) or left untreated (LEUK; n = 5).  SMAD2 and 3, abrogating subsequent signaling [14,26,39,40]. This may be explained by the fact that TGFb functions are both cell type-and context-dependent. In tumorigenesis, it can switch from tumor suppressor in the premalignant stages to prooncogene at later stages of the disease. We believe that malignant cells of APL have decreased or altered TGFb-responsiveness and that HF was able to restore TGFb signaling activation. Of interest, it has been shown that low doses of HF selectively inhibited the differentiation of the proinflammatory T helper 17 murine and human cells (TH17 cells) by activating a cytoprotective signaling pathway, without directly modulating TGFbinduced SMAD2 phosphorylation or other lymphocytes responses to TGFb [41]. Indeed, Wu et al demonstrated that TH17 cells and interleukin-17 concentrations were significantly increased in peripheral blood samples from AML patients when compared with those from healthy volunteers, suggesting that TH17 cells may play a role in leukemogenesis [42]. Although not dependent on TGFb regulation, this effect strengthens the potential of HF as an anti-leukemic drug.
Recent evidence has suggested that deregulation of the TGFb pathway may play a role in APL pathogenesis. PML-RARa may interrupt the interaction of nuclear PML with the SMAD2/3/4 complex, leading to blockade of TGFb transcriptional activity [13]. In this context, considering the effects on TGFb target gene expression, HF could reverse this blockade and prevent cell proliferation in an in vitro model of APL. Consistent with the hypothesis of TGFb blockade in APL, our results did not show significant differences in cell growth inhibition and apoptosis when NB4 and NB4 cells treated with TGFb were compared, suggesting that APL cells are resistant to the anti-proliferative effects of this cytokine.
We can hypothesize that although the disruption of the TGFb pathway itself is not sufficient to initiate malignant transformation, the loss of the TGFb response may be a critical second step that contributes to leukemia progression. In this context, the modulation of TGFb signaling may have therapeutic interest in APL. Figure S1 Cell cycle status of NB4 and NB4-R2 cells after treatment with increasing doses of halofuginone. Cell cycle status according to BrdU incorporation by NB4 (upper graphic) and NB4-R2 (lower graphic) cells after treatment with increasing doses of halofuginone. The multivariate analysis using the mixed linear model confirmed that, in both cell lines, the drug inhibited cell proliferation (P,0.001) and caused apoptosis (P = 0.002), although the pro-apoptotic effect in NB4 cells was visually less evident than the one observed in NB4-R2. (TIF)