Phosphodiesterase IV Inhibition by Piclamilast Potentiates the Cytodifferentiating Action of Retinoids in Myeloid Leukemia Cells CROSS-TALK BETWEEN THE cAMP AND THE RETINOIC ACID SIGNALING PATHWAYS*

Inhibition of phosphodiesterase IV by N -(3,5-dichloro-pyrid-4-yl)-3-cyclopentyloxy-4-methoxybenzamide(piclamilast)enhancesthemyeloiddifferentiationin-ducedbyall- trans -retinoic acid (ATRA), retinoic acid receptor (cid:1) (RAR (cid:1) ), or retinoic acid receptor X agonists in NB4 and other retinoid-sensitive myeloid leukemia cell types. ATRA-resistant NB4.R2 cells are also partially responsive to the action of piclamilast and retinoic acid receptor X agonists. Treatment of NB4 cells with piclamilast or ATRA results in activation of the cAMP signaling pathway and nuclear translocation of cAMP-dependent protein kinase. This causes a transitory increase in cAMP-responsive element-binding protein phosphorylation, which is followed by down-modulation


Inhibition of phosphodiesterase IV by N-(3,5-dichloropyrid-4-yl) -3 -cyclopentyloxy-4-methoxybenzamide (piclamilast) enhances the myeloid differentiation in-
duced by all-trans-retinoic acid (ATRA), retinoic acid receptor ␣ (RAR␣), or retinoic acid receptor X agonists in NB4 and other retinoid-sensitive myeloid leukemia cell types. ATRA-resistant NB4.R2 cells are also partially responsive to the action of piclamilast and retinoic acid receptor X agonists. Treatment of NB4 cells with piclamilast or ATRA results in activation of the cAMP signaling pathway and nuclear translocation of cAMPdependent protein kinase. This causes a transitory increase in cAMP-responsive element-binding protein phosphorylation, which is followed by down-modulation of the system. ATRA ؉ piclamilast have no additive effects on the modulation of the cAMP pathway, and the combination has complex effects on cAMP-regulated genes. Piclamilast potentiates the ligand-dependent transactivation and degradation of RAR␣ through a cAMP-dependent protein kinase-dependent phosphorylation. Enhanced transactivation is also observed in the case of PML-RAR␣. In NB4 cells, increased transactivation is likely to be at the basis of enhanced myeloid maturation and enhanced expression of many retinoiddependent genes. Piclamilast and/or ATRA exert major effects on the expression of cEBP and STAT1, two types of transcription factors involved in myeloid maturation. Induction and activation of STAT1 correlates directly with enhanced cytodifferentiation. Finally, ERK and the cAMP target protein, Epac, do not participate in the maturation program activated by ATRA ؉ piclamilast. Initial in vivo studies conducted in severe combined immunodeficiency mice transplanted with NB4 leukemia cells indicate that the enhancing effect of piclamilast on ATRA-induced myeloid maturation translates into a therapeutic benefit.
All-trans-retinoic acid (ATRA) 1 and retinoids exert their antileukemic activity through activation of three processes (i.e. cytodifferentiation, growth inhibition, and apoptosis). ATRA induces complete clinical remission in the majority of acute promyelocytic leukemia (APL) patients (1,2) and is part of the standard protocol used for the management of this type of leukemia (3). Paradoxically, APL is characterized by the prototypical t15:17 translocation, which involves the nuclear retinoic acid receptor RAR␣ and leads to the expression of the aberrant fusion protein PML-RAR␣ (1). The success of ATRA in APL represents proof of principle that cytodifferentiation therapy (4) can be applied clinically. However, a more general use of ATRA at the clinical level is limited by toxicity and natural or induced resistance (5)(6)(7).
With this in mind, it would be desirable to devise strategies aimed to increase the efficiency of the intracellular signals regulating the cytodifferentiating, growth-inhibitory, and apoptotic action of retinoids. The identification of agents capable of augmenting the therapeutic index of ATRA may lead to the development of useful pharmacological combinations in the treatment of APL and other acute myeloid leukemias (AMLs). In addition, dissection of the molecular mechanisms underlying the interactions between these agents and retinoids in AML cells is likely to cast light on the process of myeloid maturation.
Analogs of cAMP (8 -12), interferons (13)(14)(15), granulocyte colony-stimulating factor (16), and a novel class of experimental compounds, bis-indols (17), enhance the cytodifferentiating activity of ATRA in AML cell lines. Enhanced cytodifferentia-tion by combinations of bis-indols and ATRA translates into a therapeutically significant effect, as assessed in animal models of APL (17). The cross-talk between the cAMP signaling pathway and ATRA is particularly attractive in terms of the development of pharmacological combinations to be used in the cytodifferentiating treatment of AML. In fact, the cAMP pathway is modulated by various pharmacological agents (18 -20), some of which are already in the clinics for the treatment of asthma (21,22). In this report, we demonstrate that inhibition of phosphodiesterase IV (PDEIV), the enzyme that hydrolyzes cAMP to the corresponding nucleoside monophosphate, by piclamilast, potentiates the retinoid-dependent granulocytic maturation of the APL-derived NB4 and other AML cell lines. The molecular mechanisms underlying this potentiating effect are investigated in the NB4 model. Furthermore, we present initial evidence that the combination of ATRA and piclamilast is more effective than the single components in prolonging the life span of SCID mice transplanted with APL cells.
Intracellular cAMP and PKA activity were determined with the cAMP Biotrak EIA (Amersham Biosciences) and PKA PepTag Assay (Promega, Madison, WI) kits.
EMSAs were conducted as described (26) with an oligodeoxynucleotide representing the DNA sequence for the binding of cAMP-responsive element-binding protein (5Ј-GAGAGATTGCCTGACGTCAGAGA-GGTAG-3Ј).
Nitro blue tetrazolium reductase (NBT-R) activity was measured in cell extracts, and the number of NBT-R ϩ cells was determined as detailed (17). To determine the morphology of NB4 cells quantitatively, cytospins were stained with May-Grunwald-Giemsa. Approximately 100 cells/microscopic field were scored for the presence of morphologically differentiated cells (27). COS-7 cells (ATCC) were grown in Dulbecco's modified Eagle's medium containing 5% fetal calf serum and transfected with RAR␣, PML-RAR␣, or the mutant RAR␣S369A cDNAs (28) in the presence of the CREB-dependent Som-CAT plasmid (containing 1.4 kb of 5Ј-flanking region of the somatostatin gene promoter) (29) or the RARE-containing ␤ 2 RARE-CAT, DR5-tk-CAT, or TRE-tk-CAT reporter constructs and the normalization plasmid pCH110 (containing the bacterial ␤-galactosidase cDNA) (17). CAT and ␤-galactosidase activities were measured in cell extracts as detailed (17).
Protein extracts of NB4 cells were prepared as described (17). Nuclear and cytosolic fractions were isolated with nuclear extract kit (Active Motif Europe, Rixensart, Belgium). The nuclear extracts are essentially free of cytosolic contamination. Similarly, cytosolic extracts (cytosol) are not significantly contaminated with cell nuclei. In fact, the protein band corresponding to the nuclear protein marker histone H3 is present only in the nuclear fraction, whereas the cytosolic marker caspase-3 is evident only in the cytosolic fraction (see Supplemental Fig.  1S).
In Vivo Experiments-For the in vivo experiments, cells were suspended in 199 Hanks' medium, and 0.1 ml (1 ϫ 10 6 cells/mouse) were intraperitoneally inoculated in SCID mice (17). Piclamilast was dissolved in 0.5% carboxyl methyl cellulose, 0.01% Tween 80 solution and injected at a dose of 10 mg/kg. ATRA (Sigma) was dissolved in the same solution and injected at a dose of 15 mg/kg. Drugs were administered intraperitoneally in a volume of 100 l/animal 1 day after the inoculation, and the treatment continued for 13 days (5 daily injections/week). Data on the survival of animals were analyzed considering the following parameters: median survival time (MST) and percentage increase in life span (MST-treated/MST control ϫ 100) Ϫ 100). Statistical treatment of the results was conducted according to the Cox regression model (33). Fig. 1A, upon RT-PCR analysis, the APL-derived NB4 blast synthesizes detectable amounts of the transcripts coding for members of the PDEIV A, B, and D families. PDEIVC substitutes PDEIVB in the monoblastic cell line U937, which is used as a positive control (34) in these experiments. Treatment of NB4 cells with ATRA for 4 days does not affect the levels of the various PDEIV transcripts.

Inhibition of PDEIV Enhances and Accelerates the Differentiation of the ATRA-sensitive NB4 Cell Line-As illustrated in
Challenge with combinations of ATRA and various PDEIV inhibitors, such as rolipram and piclamilast, enhances the ATRAdependent maturation of NB4 cells (data not shown). Since piclamilast belongs to a chemical series of promising PDEIV inhibitors undergoing phase II/III clinical trials (22), all subsequent experiments were conducted with this molecule.
The effect of piclamilast on the morphological differentiation of NB4 cells is illustrated in Fig. 1B. Relative to what was observed with ATRA alone, NB4 cultures treated for 3 days with ATRA (0.1 M) ϩ piclamilast (30 M) showed an increased proportion of cells with lobated nuclei, cytoplasmic granules, and/or elevated cytoplasm/nucleus volume ratios, three param-eters that are associated with granulocytic maturation (27). No detectable myeloid maturation is evident upon treatment with the PDEIV inhibitor alone. Enhanced morphological maturation by ATRA ϩ piclamilast is associated with a parallel stimulation of the synthesis of ctsD, a retinoid-regulated and lysosomal differentiation marker (35). The enhancing effect of piclamilast is not limited to ctsD and extends to the other surface differentiation marker, CD11b, as demonstrated by the FACS analysis shown in . Granulocytic maturation was quantitatively determined following double blind scoring of the cell cytospins. A minimum of three independent fields (at least 150 cells/field) from each cytospin was considered. Each value represents the mean Ϯ S.D. of three independent culture flasks.°, significantly higher than the corresponding vehicle treated control according to Student's t test (p Ͻ 0.01). *, significantly higher than the corresponding ATRA-treated sample according to Student's t test (p Ͻ 0.01). The expression of the two forms of ctsD observed in NB4 cells was determined in cell extracts by Western blot analysis using specific polyclonal antibodies (right). The filter was rechallenged with an anti-␤-actin antibody. Protein molecular weight markers are indicated. C, NB4 cells were treated for 3 days with vehicle (Me 2 SO) or the indicated concentrations of ATRA in the absence or presence of 30 M piclamilast. The surface expression of the myeloid markers, CD11b, CD11c, CD33, and CD38, was determined in NB4 cells by FACS analysis. The graphs on the left indicate the percentage of marker-positive cells, whereas those on the right show the mean associated fluorescence for each marker. The results are representative of two independent experiments. ATRA-dependent induction of NBT-R in NB4 cells. Enhanced NBT-R is evident both when the enzymatic activity is measured in cell extracts and when the number of NBT-R ϩ cells is counted (data not shown). The enhancing effect of piclamilast is similar to that observed with the cAMP analogue, 8-CPT-cAMP (data not shown). Fig. 2A demonstrates that treatment with piclamilast (30 M) does not result in a significant increase in the number of NBT-R ϩ NB4 at any of the time points considered. However, when the PDEIV inhibitor is added to ATRA, the retinoid-dependent increase of NB4 positive cells is enhanced and accelerated. Enhanced NBT-R activity is not only time-dependent but also dose-dependent relative to ATRA ( Piclamilast Interacts with ATRA, RAR, and RXR Agonists-Piclamilast enhances the ATRA-dependent induction of NBT-R activity in the retinoid-sensitive HL-60 myeloid leukemia cell line (Fig. 3A) and in primary cultures of blasts isolated from an AML patient (Fig. 3B). In both cell types, piclamilast ϩ ATRA induces a significantly greater antiproliferative effect than ATRA or piclamilast alone (data not shown). In AML blasts, the enhancing action of the PDEIV inhibitor on NBT-R induction and growth inhibition requires long exposure times (7 days) and is associated with a reduction in cell viability. ATRA ϩ piclamilast reduces the total number of cells observed in control conditions by 69 Ϯ 5% (mean Ϯ S.D., n ϭ 3), and this compares to 37 Ϯ 2 and 10 Ϯ 1% in the case of ATRA and piclamilast, respectively. Exposure of the blasts to piclamilast for 7 days does not result in a significant cytotoxic effect relative to what observed in basal conditions (viability: 64 Ϯ 5 versus 61 Ϯ 3%), whereas the combination of ATRA ϩ piclamilast is more cytotoxic than the retinoid alone (viability 37 Ϯ 4 versus 50 Ϯ 4%). In U937 myelomonocytic cells, piclamilast and ATRA induce NBT-R activity on their own, although the combination is no more effective than the single components ( Fig.  3C). In NB4 cells, the RAR/RXR heterodimeric pathway is involved in the cross-talk between piclamilast and ATRA. In fact, the RAR␣ agonist, AM580 (36,37), can substitute for ATRA (Fig. 3D).
Granulocytic maturation of certain ATRA-sensitive and -resistant AML cells is induced by combinations of cAMP analogs and RXR agonists (38). Treatment of NB4 cells for 3 days with 0.01 M CD2915, a selective RXR agonist, does not result in a significant increase in the percentage of NBT-R ϩ cells, whereas a higher concentration (0.1 M) augments the proportion of cells expressing the differentiation marker (Fig. 3D). With both concentrations of CD2915, the addition of piclamilast to the PKA Mediates the Cytodifferentiation Induced by ATRA and Piclamilast-In NB4 cells, a dose-dependent suppression of the enhancing effect of piclamilast on retinoid-induced NBT-R activity is evident with Rp-8Br-cAMP, a competitive inhibitor of cAMP (Fig. 4A). At all of the concentrations tested, the cAMP analog does not alter the growth and viability of NB4 cells treated with vehicle, piclamilast, ATRA, or piclamilast ϩ ATRA. Similar suppressive effects on the ATRA-dependent induction of NBT-R activity are caused by substitution of Rp-8Br-cAMP with the PKA inhibitor H89 (10 M) (39) (Fig. 4B). Like Rp-8Br-cAMP, H89 does not revert the growth inhibition afforded by treatment of NB4 cells with ATRA or the combination of ATRA and piclamilast (data not shown). The concentration of H89 used is effective in inhibiting PKA activity as assessed in undifferentiated NB4 cells treated with the kinase inhibitor for 16 h (Fig. 4B, inset).
The Rap-exchanging factor, Epac, mediates some of the bio-logical effects of cAMP (40). Treatment of NB4 cells for 3 days with O-Me cAMP, a selective Epac activator (41), does not affect the percentage of NBT-R ϩ NB4 cells regardless of the presence/absence of ATRA and/or piclamilast in the growth medium (Fig. 4C). Furthermore, the expression of NBT-R activity in NB4 cells treated with ATRA ϩ piclamilast is not modified by the addition of O-Me cAMP. Finally, in the experimental conditions considered, the activator of Epac has no significant effect on the number and viability of cells (data not shown).
Piclamilast and ATRA Modulate the cAMP Signaling Pathway-Proliferating NB4 cells contain detectable amounts of intracellular cAMP (Fig. 5A). Treatment with piclamilast for 6 h leads to an almost 3-fold elevation in the cyclic nucleotide levels, which are back to base line by 18 h. ATRA has no major influence on the constitutive or piclamilast-induced levels of cAMP.
Detectable and relatively constant amounts of PKA are determined in unstimulated NB4 cells (Fig. 5B). A significant elevation of PKA activity is observed upon treatment with piclamilast for 6 h, whereas an over 4-fold increase (similar to that afforded by the PKA agonist dibutyryl-cAMP) is evident by 18 h. PKA activity is back to basal levels by 96 h. Unexpectedly, the addition of ATRA to the growth medium for 18 h results in a level of PKA activation that is similar to that observed with piclamilast. However, treatment of NB4 cells with ATRA ϩ piclamilast for the same amount of time is no more effective than the PDEIV inhibitor or the retinoid alone. Undifferentiated NB4 cells synthesize significant amounts of the PKA regulatory subunits, PKA-rI␣ and PKA-rII␣ (Fig. 5C). Treatment of cells with piclamilast does not alter the intracellular levels of either PKA-rI␣ or PKA-rII␣. By contrast, downmodulation of PKA-rII␣ is evident following incubation with ATRA for 24 and 72 h. At both time points, the addition of piclamilast to ATRA results in suppression of the retinoid-dependent down-modulation of PKA-rII␣.
Activation of PKA is accompanied by relocalization of the catalytic subunit to the nuclear compartment. The relocalization of the catalytic subunit was studied in NB4 cells treated with piclamilast, ATRA, and the combination of the two compounds following separation of nuclear and cytosolic fractions by differential centrifugation. Detectable amounts of the PKA catalytic subunit, PKA-cI␣ (42), are present in the cytosol of undifferentiated NB4 cells (Fig. 5D). Piclamilast, ATRA, and We determined the levels of the transcription factor, CREB (43), and its phosphorylation in the nucleus of NB4 cells (Fig.  6A) by Western blot as well as the amounts of CREB interacting with the corresponding DNA consensus sequence by EMSA (Fig. 6B). In EMSA, the retarded bands are specific for the CREB complexes, as demonstrated by competition experiments using wild type and mutated oligonucleotides as well as supershift with an anti-CREB antibody (data not shown). Fig. 6A demonstrates that undifferentiated NB4 cells synthesize large amounts of CREB constitutively. A proportion of the transcription factor is in its active state, as demonstrated by the level of phosphorylation. A rapid and transient increase in the phosphorylation and activity of CREB is observed after 2 h of treatment with piclamilast, ATRA, or ATRA ϩ piclamilast. In all cases, activation of CREB is significant but minor (ϳ2-fold induction). By 6 h, the stimulating effect of the PDEIV inhibitor is over. Stimulation by piclamilast is followed by progressive suppression of CREB phosphorylation (see 24-h time point) consequent to down-regulation of the CREB protein.
Whereas the association of ATRA and piclamilast is no more efficient than the single components in activating CREB, it accelerates the down-regulation of the transcription factor afforded by the PDEIV inhibitor (see the 6-h panel). Regardless of the stimulus applied, CREB and the relative state of phosphorylation are back to control levels by 72 h. As illustrated in Fig.  6B, the constitutive activation of CREB in undifferentiated NB4 cells and the effects of piclamilast and/or ATRA on the levels and state of phosphorylation of the transcription factor are confirmed by EMSA at 24 and 72 h. Using the same assay (Fig. 6B, bottom panel), we demonstrate that treatment of NB4 cells with H89 for 24 h suppresses the activation of CREB observed in control and piclamilast-treated cells, suggesting PKA involvement. The PKA inhibitor does not affect ATRA or ATRA ϩ piclamilast down-regulation of CREB activity.
The effect of piclamilast and/or ATRA on the expression of some CREB-dependent genes in NB4 cells (vinculin, THBS, EGR-1, and BTG) is complex, time-dependent, and not strictly related to the CREB activation pattern described above (Fig.  6C). Vinculin and BTG mRNAs are never up-regulated by piclamilast, whereas THBS and EGR1 mRNAs are responsive to the PDEIV inhibitor. In the case of THBS, this effect is transient, being evident only at 6 h, whereas piclamilast-dependent induction of the EGR1 transcript is delayed (72 h). Interestingly, ATRA induces the four CREB-dependent mRNAs, albeit with different kinetics. However, this induction is enhanced by piclamilast only in the case of vinculin, BTG, and THBS, and the phenomenon is evident only when the retinoid is used at a concentration of 0.01 M.
Given the complexity of the effects on CREB in NB4 cells, we evaluated the direct action of piclamilast and/or ATRA on the transcriptional activity of the CREB-containing promoter of the somatostatin gene. These experiments were conducted in COS-7 cells, since NB4 blasts are refractory to transient and stable transfection. As shown in Fig. 6D, the somatostatin reporter construct is activated by piclamilast. In contrast, de-FIG. 6. Effect of piclamilast and ATRA on CREB and CREB-responsive genes. A, NB4 cells seeded at an initial concentration of 150,000/ml were treated with vehicle, ATRA (0.1 M), piclamilast (30 M), or the combination of the two compounds for the indicated amount of time. Equivalent amounts (20 g) of nuclear cell extracts were subjected to Western blot analysis with antibodies recognizing CREB and the corresponding phosphorylated form (CREBp) as well as ␤-actin. B, NB4 cells seeded as in A were treated for the indicated amount of time. In the case of the experiments involving H89, the PKA inhibitor was present in the medium at a concentration of 10 M. Nuclear extracts were prepared and subjected to EMSA with a radiolabeled double-stranded oligodeoxynucleotide corresponding to the DNA consensus sequence for the binding of CREB. Competitor, the presence of a 100-fold molar excess of the cold oligodeoxynucleotide in the binding reaction. The results are representative of at least three independent experiments. C, total RNA was extracted from NB4 cells treated with the indicated stimuli. Identical amounts of RNA (20 g) were subjected to Northern blot analysis using specific THBS, EGR1, vinculin, and BTG cDNA probes radiolabeled with [ 32 P]dCTP. Ethidium bromide staining of representative RNA gels, illustrating the 28 S ribosomal RNA band, is shown at the bottom. These gels demonstrate that equal amounts of intact RNA were loaded in each lane of the gels used for Northern blot analysis. D, COS-7 cells were co-transfected with RAR␣ (0.1 g) and the cAMP-dependent gene reporter construct containing the somatostatin promoter as well as the normalization construct pCH110 coding for bacterial ␤-galactosidase. Eighteen hours following transfection, cells were treated with the indicated concentrations of ATRA and/or piclamilast. Thirty-six hours following the addition of the compounds, cell extracts were prepared and assayed for CAT and ␤-galactosidase activities. The data represent CAT activity in arbitrary units following normalization for the level of transfection using ␤-galactosidase. Each value is mean Ϯ S.D. of three replicate culture dishes.°, significantly higher than the relative control sample according to Student's t test (p Ͻ 0.01). The results are representative of two independent experiments. spite co-transfection of the RAR␣ cDNA, the CREB-regulated somatostatin promoter is not stimulated by ATRA. Furthermore, the addition of ATRA to piclamilast does not modulate the transcriptional activation of the PDEIV inhibitor. These data indicate that the activation of PKA and CREB induced by ATRA in NB4 cells may be limited to the myeloid cellular context. More importantly, what was observed in COS-7 cells indicates that the cross-talk between the retinoid and cAMPdependent signaling pathways is unlikely to involve the activation of CREB.
Piclamilast Modulates the ATRA Signaling Pathway-PDEIV inhibition exerts modulatory effects on the nuclear retinoic acid receptors, and this is likely to be at the basis of the cross-talk between ATRA and piclamilast in AML cells. The effects of piclamilast on the transactivation properties of RAR␣ and PML-RAR␣ in COS-7 cells are shown in Fig. 7A. Piclamilast enhances the ligand-dependent activation of RAR␣ regardless of the reporter construct, ␤2RAR-CAT or DR5-CAT, considered. The former consists of the chloramphenicol acetyltransferase (CAT) reporter gene controlled by the retinoic acidresponsive element (RARE) of the ␤2RAR promoter. In the latter, an artificial RARE controls the thymidine kinase promoter. In the case of DR5-CAT, the enhancing effect of piclamilast on the ligand-dependent transactivation of RAR␣ is similar to that caused by 8-CPT-cAMP. Enhanced transactivation by piclamilast ϩ ATRA is accompanied by decreased steady-state levels of RAR␣, as demonstrated by the Western blots shown. This is due to an increase in the proteasome-dependent degra-  ; DR5) as well as the normalization construct pCH110. Eighteen hours following transfection, cells were treated with the indicated concentrations of ATRA, piclamilast, or 8-CPT-cAMP and combinations thereof. Thirty-six hours after the addition of the compounds, cell extracts were prepared and assayed for CAT and ␤-galactosidase activities. The data represent CAT activity in arbitrary units following normalization for the level of transfection using ␤-galactosidase. Each value is the mean Ϯ S.D. of three replicate culture dishes. The results are representative of at least three independent experiments. Parallel Western blot analyses with anti-RAR␣ and anti-␤-actin antibodies are represented above the bar graphs. These experiments were performed on pooled extracts obtained from the same samples used for the determination of CAT activity and are normalized for the content of ␤-galactosidase. B, COS-7 cells were co-transfected with RAR␣ (0.1 g) or PML-RAR␣ (0.1 g) and the ATRA-dependent gene reporter constructs DR5-CAT (1 g; DR5) or TRE-CAT (1 g; TRE) as well as the normalization construct pCH110. Enzymatic measurements and Western blot analyses for PML-RAR␣ and RAR␣ were conducted as in A. The results are representative of at least three independent experiments. C, COS-7 cells were co-transfected with the wild type form of RAR␣ (0.1 g; RAR␣) or the mutant RAR␣S369A (0.1 g), in which the PKA phosphorylation site has been inactivated by site-directed mutagenesis (Ser 369 to Ala 369 ), and the ATRA-dependent gene reporter construct DR5-CAT (1 g; DR5) as well as the normalization construct pCH110. Enzymatic measurements and Western blot analysis were conducted as in A or B.°, significantly higher than the relative control sample according to Student's t test (p Ͻ 0.01). *, significantly higher than the relative ATRA or piclamilast sample according to Student's t test (p Ͻ 0.01). D, NB4 cells were treated for 72 h with the indicated concentrations of ATRA, piclamilast, and ATRA ϩ piclamilast. Cell extracts from pools of three separate culture dishes were subjected to Western blot analysis with anti-RAR␣ and ␤-actin antibodies. The results are representative of at least two independent experiments. The arrow indicates the position of the band corresponding to PML-RAR␣. dation of retinoid receptors that accompanies ligand-dependent transactivation (44 -46). Indeed, the phenomenon is blocked by the proteasome inhibitor MG132 (data not shown).
Piclamilast enhances the ligand-dependent transactivation not only of RAR␣ but also of PML-RAR␣ (Fig. 7B), although PML-RAR␣ is a less sensitive target than RAR␣. These effects are observed with ␤2RAR-CAT (data not shown), DR5-CAT, and TRE-CAT, another thymidine kinase-based reporter construct containing a palindromic artificial RARE. As in the case of RAR␣, ligand-dependent activation of PML-RAR␣ is associated with degradation of the receptor. However, this phenomenon is not modulated significantly by piclamilast.
The above results suggest a direct action of piclamilast on RAR␣ and PML-RAR␣ and demonstrate a direct correlation between the rate of degradation and the state of activation of RAR␣. To test this, we transfected RAR␣S369A, a receptor mutant whose PKA phosphorylation site has been inactivated (Fig. 7C). RAR␣S369A maintains responsiveness to ATRA-dependent transactivation but is refractory to piclamilast. Significantly, the PDEIV inhibitor has no detectable effect on the ligand-dependent degradation of RAR␣S369A. Consistent with this, inhibition of PKA by H89 suppresses the enhanced transactivation and degradation of RAR␣ afforded by ATRA ϩ piclamilast (data not shown).
Ligand binding directs RAR␣ and PML-RAR␣ along the proteasome degradation pathway in the natural context of the NB4 blast (29,(47)(48)(49)(50). As illustrated in Fig. 7D, treatment for 24 h with 1 M ATRA results in a significant diminution of the levels of both RAR␣ and PML-RAR␣, whereas similar phenomena are not observed with a lower concentration of the retinoid (0.1 M). Whereas piclamilast has no effect on its own, the addition of the PDEIV inhibitor to 0.1 M ATRA results in a degradation of the two RAR receptors similar to that induced by 1 M ATRA. Fig. 8A shows the effect of piclamilast, ATRA, and ATRA ϩ piclamilast on the expression of ATRA-responsive genes (granulocyte colony-stimulating factor, CD38, sialoadhesin, PRAM, ctsD, and pLym) in NB4 cells. Whereas ctsD and granulocyte colony-stimulating factor are markers of granulocytic maturation, the functional significance of the other genes is unknown. Except for CD38 (see also Fig. 1C) and granulocyte colonystimulating factor, expression of the mRNAs considered is enhanced by co-treatment with ATRA ϩ piclamilast, although the PDEIV inhibitor has no effect on its own. The enhancing effect of piclamilast is more evident when ATRA is present at 0.01 M, a concentration at which the retinoid does not increase the levels of sialoadhesin, PRAM, and ctsD and is marginally effective on pLym. MCP-1 and the cytokine, TNF, are two proteins regulated by ATRA at the gene level in opposite directions (51). MCP-1 is a myeloid differentiation marker (17), whereas TNF is a potential effector molecule in the APL-related ATRA syndrome (52). Fig. 8B demonstrates that piclamilast and ATRA increase the levels of MCP-1 and decrease those of TNF. The combination of piclamilast and ATRA is more effective than the single components in augmenting the production of MCP-1 but not in inhibiting TNF accumulation.
Effect of Piclamilast, ATRA, and the Association on cEBP, STAT1, and ERK-cEBP transcription factors are important mediators of myeloid maturation (53)(54)(55)(56). Fig. 9A demonstrates that detectable amounts of cEBP␤ and cEBP⑀ are not present in undifferentiated NB4 cells. Treatment of NB4 cells with piclamilast or ATRA induces the 36 -38-kDa (LAP) and the 23 kDa (LIP) forms of cEBP␤ as well as cEBP⑀. In all cases, the phenomena are already detectable at 6 h and persist until 72 h. A, NB4 cells were treated for 72 h with the indicated concentrations of ATRA, piclamilast, and ATRA ϩ piclamilast. Cells (pools of three independent culture dishes) were harvested, and total RNA was extracted. Identical amounts of RNA (20 g) were subjected to Northern blot analysis using the indicated cDNA probes labeled with [ 32 P]dCTP. The ethidium bromide staining of a representative RNA gel, illustrating the 28 S ribosomal RNA band, is shown at the bottom. B, NB4 cells were treated for the indicated amounts of time with ATRA, piclamilast, and ATRA ϩ piclamilast. Culture medium was collected and used for the determination of MCP-1 and TNF␣ with specific ELISAs.°, significantly higher or lower than the relative control sample according to Student's t test (p Ͻ 0.01); *, significantly higher than the relative ATRA or piclamilast sample according to Student's t test (p Ͻ 0.01).
Piclamilast exerts a particularly rapid and strong effect on LIP. Except for the enhancing effect of ATRA ϩ piclamilast on the expression of LAP observed at 6 h, the combination is no more effective than the PDEIV inhibitor or the retinoid alone in inducing the various forms of cEBP␤ and cEBP⑀. Indeed, long term treatment (72 h) with high concentrations of ATRA (0.1 M) and piclamilast slightly inhibits the induction of cEBP␤. Induction of cEBP␤ and -⑀ by ATRA is predominantly transcriptional, whereas the effect of piclamilast on these proteins is mainly translational or post-translational. In fact, a rapid and long lasting induction of the two transcripts is afforded by ATRA, whereas treatment with piclamilast is not associated with induction of cEBP␤ or -⑀ mRNAs (Fig. 9B). The addition of piclamilast to ATRA has only marginal effects on the expression of the mRNAs coding for cEBP␤ and -⑀ relative to what is observed with the retinoid alone. Interestingly, ATRA results in a dose-dependent diminution of the levels of the cEBP␣ mRNA, which is not affected by ATRA ϩ piclamilast (see 24 h).
The transcription factor STAT1 is also involved in the process of myeloid maturation of leukemic cells (57)(58)(59). Piclamilast exerts a significant enhancing effect on the induction of STAT1 observed in ATRA-treated NB4 cells (Fig. 9C). This results in a detectable increase in STAT1 tyrosine phosphorylation, which is necessary for the activation of the protein. The PKA inhibitor, H89, blocks the positive effect of piclamilast on the regulation of STAT1 by ATRA.
Activation of the MAP kinase, ERK, has been involved in the process of myeloid maturation triggered by ATRA in the HL-60 myeloid cell line (60,61). Piclamilast, ATRA, and ATRA ϩ piclamilast do not affect the constitutive expression of ERK proteins in NB4 cells (Fig. 10A). However, treatment for 10 min and 1 h with the PDE IV inhibitor is associated with decreased ERK phosphorylation and activation. Inhibition of ERK phosphorylation by piclamilast is protracted if ATRA is added (see 1 day and 3 days). Fig. 10B demonstrates that co-treatment with U0126, a selective ERK inhibitor, does not have any detectable effect on the number of NBT-R ϩ cells observed in NB4 cells challenged with vehicle, ATRA, piclamilast, and the combination. U0126 has a strong inhibitory action on the growth of undifferentiated NB4 cells and potentiates the antiproliferative and cytotoxic actions of ATRA. However, the addition of piclamilast to U0126 or U0126 and ATRA does not add to the magnitude of the observed effects.
In Vivo Activity of the Combination of Piclamilast ϩ ATRA-To evaluate whether the potentiating effect of piclamilast on ATRA cytodifferentiating activity has therapeutic impact, we transplanted NB4 cells intraperitoneally in SCID mice. Animals were treated with vehicle, ATRA, piclamilast, and the combination of the two compounds and evaluated for survival. Fig. 11  vehicle-treated animals of 37% in the case of ATRA and 48% in the case of ATRA ϩ piclamilast. In no experimental group did we observe signs indicative of major systemic toxicity, such as treatment-associated lethality or significant body weight loss. DISCUSSION Inhibition of PDEIV by piclamilast enhances and accelerates the granulocytic maturation program activated in APL and other myeloid leukemia cells by ATRA, RAR␣, and RXR agonists through increases in the intracellular levels of cAMP. Cooperation between piclamilast and ATRA is likely to be mediated by the RAR/RXR pathway, since it is reproduced with the selective RAR␣ agonist, AM580, and is not observed in the retinoid-resistant NB4.R2 and NB4/007 cell lines (data not shown), in which this pathway is disrupted (24,62). Interestingly, not only ATRA-sensitive cell lines, but also the retinoidresistant, NB4.R2 subline, respond to combinations of piclami-last and RXR agonists. This indicates that piclamilast facilitates myeloid maturation not only via retinoid-dependent RAR/RXR activation, but also through the recently described RXR/RXR pathway (8 -11, 38). The data obtained in vitro are encouraging in view of the potential therapeutic application of combinations between selective PDE IV inhibitors and ATRA, RAR, or RXR agonists in the treatment of AML. Indeed, our initial in vivo studies in the APL model of the SCID mouse transplanted with NB4 cells demonstrate that treatment with ATRA and piclamilast is well tolerated and does not result in increased toxicity relative to ATRA and/or piclamilast alone. More importantly, the combination is more effective than ATRA alone in terms of increased survival of the leukemiabearing animals. At present, our data demonstrate that the therapeutic efficacy of the combination of piclamilast and ATRA is only slightly superior to that of ATRA alone; however, the dosage and administration schedule need to be optimized.
The molecular mechanisms underlying the cross-talk between piclamilast and ATRA are summarized in Table I. Epac, is a direct cAMP target (40,41) and is known to antagonize some of the cellular responses to PKA (63). However, Epac does not play any role in our differentiation model. Furthermore, the direct Epac activator, O-Me cAMP, does not disrupt the interaction between ATRA and piclamilast. Piclamilast stimulates PKA and translocates the catalytic subunit of the enzyme to the nucleus of NB4 cells. In the nucleus, the activated PKA subunit increases the constitutive phosphorylation of CREB and its DNA binding activity. CREB activation is transient and limited in magnitude and does not completely reflect PKA activation and export to the nucleus. The activation is followed by down-regulation of the entire CREB system by 24 h. The system is back to pretreatment levels by 72 h. The transient and limited activation of CREB is in line with the weak effects that piclamilast exerts on CREB-dependent genes. In NB4 cells, the early action of ATRA on the various components of the cAMP signal transduction system is similar to that of piclamilast in many respects. The retinoid activates PKA and translocates it to the nucleus. Furthermore and unlike piclamilast, ATRA down-modulates one of the two regulatory subunits of FIG. 10. Effect of piclamilast, ATRA, and the combination on ERK levels and phosphorylation. A, NB4 cells (150,000/ml) were treated with the indicated compounds for various lengths of time. Western blot analysis were performed on cell extracts, using antibodies directed against ERK and the corresponding phosphorylated form of the protein.
The results are representative of at least two independent experiments. B, NB4 cells were treated for 3 days with the indicated compounds. The percentage of NBT-R ϩ cells (left graph) and the total number of cells (right graph) were determined. The viability of cells is indicated in parenthesis.°, significantly higher or lower than the relative control sample according to Student's t test (p Ͻ 0.01). *, significantly higher or lower than the relative ATRA or piclamilast sample according to Student's t test (p Ͻ 0.01).
FIG. 11. In vivo activity of piclamilast alone and in combination with ATRA. SCID animals were transplanted intraperitoneally with NB4 cells (6 ϫ 10 6 cells/animal). Animals were injected with NB4 cells on day 0. On day 3, animals were randomized in four experimental groups receiving one daily dose of vehicle, piclamilast (10 mg/kg), ATRA (15 mg/kg), and the combination of the retinoid and the PDE IV inhibitor. Treatment was continued for 13 days. The Kaplan-Meier's survivial curves of the various experimental groups are illustrated.
PKA. At present, the mechanisms underlying PKA activation by the retinoid are poorly understood. Indeed, at the time points considered, ATRA treatment does not lead to a detectable increase in intracellular cAMP. However, it was recently demonstrated that the retinoid causes a very rapid pulse of cAMP accumulation in NB4 cells (64). Regardless of the mechanism of PKA activation, this phenomenon is likely to be responsible for the transient and early activation of CREB observed upon treatment of NB4 cells with ATRA. Interestingly, activation of CREB by the retinoid seems to be limited to the NB4 and possibly the myeloid cellular context. In fact, in COS-7 cells transfected with RAR␣, ATRA does not activate the CREB-responsive promoter of the somatostatin gene.
A key question in the context of the cross-talk between piclamilast and ATRA concerns the point(s) at which the cAMP and the retinoid pathways converge, since this/these may be critical for enhanced myeloid maturation. Combining piclamilast and ATRA has no significant additive or synergistic effects on cAMP levels, PKA activation, and nuclear translocation relative to what is observed with the retinoid and the PDEIV inhibitor alone. In contrast, the association of ATRA and piclamilast accelerates the down-regulation of CREB afforded by piclamilast and prevents the retinoid-dependent down-modulation of PKA-rII␣. Accelerated CREB down-regulation by ATRA ϩ piclamilast may lead to a more rapid arrest of cell growth, which, in turn, may accelerate the retinoid-dependent myeloid maturation program. Indeed, growth arrest and myeloid maturation of NB4 cells are tightly linked (37). The effect on PKA-rII␣ may also be relevant in the context of the myeloid maturation of APL blasts, since this regulatory subunit is believed to be involved in cell differentiation, whereas PKA-rI␣ seems to transduce proliferative signals (65). Hence, tipping the intracellular balance of PKA-rII␣ and PKA-rI␣ toward the former may facilitate the process of granulocytic maturation.
The action of piclamilast on the nuclear retinoid receptors is dramatic. Indeed, modulation of the primary targets of the TABLE I Summary of the biological effects of piclamilast and ATRA The table summarizes the results obtained in NB4 or COS cells treated with piclamilast, ATRA, and a combination of the two compounds on various components of the cAMP and the retinoid signaling pathways. The effects of ATRA and piclamilast, alone or in combination, on the myeloid maturation of NB4 cells are also summarized. The time frame of the various effects is indicated. The observed phenomena can be described as early or late events on the basis of their occurrence within or after the first 24 h. Plus and minus signs indicate an increase and a decrease in the observed parameter, respectively. If the combination of piclamilast and ATRA gives an increase or a decrease significantly higher or lower than that observed with ATRA, a double plus or double minus sign is shown. In the case of cEBP␤, ϩ/Ϫ indicates attenuation of the increase observed with ATRA or piclamilast. ϫ, no variation in the parameter taken into consideration. retinoid signaling pathway by piclamilast is likely to play a prominent role in the potentiating effect of the PDEIV inhibitor. Piclamilast enhances the ligand-dependent activation of both RAR␣ and PML-RAR␣, probably through a direct action of PKA on the state of phosphorylation and activation of the two receptors. Stimulation of the ligand-dependent transactivation of RAR␣ and PML-RAR␣ is observed when two minimal artificial promoters are used, and the effect is suppressed in the case of the mutant RAR␣S369A receptor. Piclamilast enhances not only the retinoid-dependent transactivation but also the degradation of RAR␣. Transactivation and degradation of the RAR␥ receptor are intimately linked and possibly coupled processes (44,45). Blocking degradation suppresses transactivation, and the opposite is also true. Mutation of the PKA phosphorylation site has no significant effect on ATRA-induced degradation, whereas it abrogates the increased proteolysis afforded by piclamilast ϩ ATRA. This suggests that PKA activation and RAR␣ phosphorylation on Ser 369 by piclamilast facilitate the transactivation/degradation coupling. Interestingly, PML-RAR␣ behaves differently from RAR␣. The aberrant fusion protein is degraded to the same extent by ATRA and ATRA ϩ piclamilast. This suggests that the two receptors have different susceptibilities to ligand-dependent proteasomal degradation. Regardless of the mechanistic details, PKA activation results in enhanced expression of many, but not all, ATRA-regulated genes in NB4 cells. Enhanced activity of a large number of retinoid-activated genes may be at the basis of enhanced myeloid maturation. STAT1, the transcription factors of the cEBP family, and the MAP kinase ERK are regulated by retinoids and have been implicated in the process of myeloid maturation (57)(58)(59)(60)(61). Our data are consistent with the idea that STAT1 plays a role in the process of granulocytic maturation set in motion by ATRA and enhanced by PDEIV inhibitors. In fact, the amounts as well as the activation state of STAT1 correlate with enhanced granulocytic maturation of NB4 cells by ATRA ϩ piclamilast. Although piclamilast and ATRA induce cEBP␤ and -⑀ through different molecular mechanisms, no significant interactions between the two compounds on these molecular targets are evident at the majority of the time points considered. The only exception is the enhanced induction of cEBP␤ observed early (6 h) during the differentiation process. This effect may be of some significance for the granulocytic maturation of NB4 cells. In this context, it is relevant that piclamilast and ATRA induce not only the forms of cEBP␤ that act as transcriptional activators (LAP1 and LAP2) but also the purportedly transcriptional inhibitor, LIP. Consistent with the idea that PKA modulates the ERK pathway in a negative fashion (66 -68), phosphorylation and activation of the MAP kinase is reduced by piclamilast. Interestingly, prolonged down-regulation of ERK phosphorylation by the combination of piclamilast and ATRA may have relevance for the inhibition of cell growth, which is more evident upon treatment of NB4 cells with the combination than with piclamilast or ATRA alone. However, the data obtained with the U0126 inhibitor suggest that ERK activation (60,61) is not a necessary event for the myeloid maturation of APL cells.
In conclusion, our results concur in defining the molecular mechanisms underlying the cross-talk between the cAMP and the retinoid signal transduction pathways. Furthermore, they indicate that PDE IV represents a valuable pharmacological target for future efforts aimed at the differentiation therapy of myeloid leukemia.