|
|||||||||
|
|
|
|
|
|
|
|
|
|
|
Prepublished online as a Blood First Edition Paper on June 28, 2002; DOI 10.1182/blood-2002-03-0720.
NEOPLASIA
From the Istituto di Ricerche Farmacologiche
"Mario Negri," Laboratory of Molecular Biology, Centro Catullo e
Daniela Borgomainerio, Milano, Italy; the University of Istanbul,
Faculty of Science, Department of Biology, Vezneciler, Istanbul,
Turkey; the Department of Oncology, Sigma-Tau Industrie Farmaceutiche
Riunite S.p.A., Pomezia, Italy; and the Division of Hematology,
Ospedali Riuniti di Bergamo, Largo Barozzi, Bergamo,
Italy.
Enhancing the pharmacologic activity of all-trans
retinoic acid (ATRA) is potentially useful in the management of acute
promyelocytic leukemia (APL) and other types of myeloid leukemia. In
this report, we identify a novel class of experimental agents
selectively potentiating the cytodifferentiating activity of ATRA
and synthetic retinoic acid receptor Acute promyelocytic leukemia (APL) is a form of
acute myelogenous leukemia with typical chromosomal translocations
leading to the expression of abnormal fusion proteins involving the
nuclear retinoic acid receptor (RAR) Although the success obtained with APL patients has raised enthusiasm
for the clinical use of ATRA in the treatment of leukemia and other
neoplastic diseases, the therapeutic efficacy of this compound is still
burdened by problems such as resistance5 and
toxicity.6,7 One possible strategy to increase the
therapeutic index of ATRA is the development of ATRA-based
pharmacologic combinations that are more powerful and easily tolerated
than the individual components.
Relevant aspects of the differentiation program set in motion by
ATRA in APL cells can be reproduced in primary cultures of leukemic blasts and in the derived NB4 cell line,8,9 which is a unique model for the study of the pharmacologic activity of
ATRA and derivatives.10-12 Pharmacologic concentrations of
ATRA arrest the growth of NB4 blasts and differentiate them into cells that resemble mature neutrophils.13,14 This is followed by a slow process of apoptosis.15 We and others used NB4
cells to demonstrate that such diverse compounds as cytokines, cyclic adenosine monophosphate (cAMP) analogs, and STI57116-20
enhance the pharmacologic activity of ATRA and other retinoids.
In this report, we identify a novel class of compounds capable of
potentiating the cytodifferentiating activity of ATRA in NB4 cells.
These molecules have a bis-indol structure and enhance the efficacy of
ATRA not only in vitro but also in vivo in the APL model of the severe
combined immunodeficiency (SCID) mouse receiving transplants of
NB4.21,22
Chemicals
Cell culture, treatments, and transfections
COS-7 cells were transfected with cationic liposomes carrying plasmids
for RAR Cellular proliferation, viability, and cytodifferentiation Cell number and viability were determined following staining with erythrosin (Sigma).27 CD11b, CD38, CD14, ICAM-1, and integrin -5 were measured by flow cytometry using
specific monoclonal antibodies (Becton Dickinson, Franklin Lakes,
NJ).28 The nitrobluetetrazolium (NBT) reduction assay was
performed on extracts of phorbol myristate acetate-stimulated cells,20 and the percentage
of NBT+ cells was determined as
described.20
Western blot analysis, JNK kinase activity, and determination of cytokine secretion and gel retardation assays Extracts of NB4 cells (from 3 × 106 cells) were subjected to Western blot analysis as reported.20 The experiments were performed with antibodies against RAR (RPF'),29 RXR,
4RX1F6,30 CEBP (C-22; Santa Cruz
Biotechnology, Santa Cruz, CA), signal transducers and activators of
transcription-1 (STAT1) p91 (C-24; Santa Cruz Biotechnology), STAT1
p91 and p84 (C-136; Santa Cruz Biotechnology), Tyr701
phosphorylated STAT1 p91 and p84 (Upstate Biotechnology, Lake Placid,
NY), cathepsin D (C-20; Santa Cruz Biotechnology), tyk-2 (C-20;
Santa Cruz Biotechnology), Mlh1 (C-20; Santa Cruz Biotechnology), or
 actin (C-11; Santa Cruz Biotechnology). The antibodies directed
against the phosphorylated forms of extracellular signal-regulated
kinase-1 (ERK-1), ERK-2, Jun N-terminal kinase (JNK), and p38,
and against histone H3 and its acetylated form were from Cell
Signaling Technology (Beverly, MA). Protein bands were visualized with
the enhanced chemiluminescence detection kit (Amersham, Little
Chalfont, United Kingdom). JNK kinase activity was determined on JNK
immunoprecipitates with c-JUN as a substrate, using a commercial kit
(Cell Signaling Technology).
Enzyme-linked immunosorbent assays (ELISAs) (Endogen, Woburn, MA31) were used for the determination of interleukin-1 (IL-1), tumor necrosis factor (TNF), and monocyte chemoattractant protein 1 (MCP-1). Gel retardation assays were performed on nuclear extracts of NB4 cells, as already described.27 The following double-strand oligonucleotide probes (Santa Cruz) were used: CREB = 5'GAGAGATTGCCTGACGTCAGAGAGCTAG 3'; and Ets-1 = 5'GGATCTCGAGCAGGAAGTTCGA 3'. RNA extraction and gene profiling RNA was extracted with the TRIzol reagent (Invitrogen SRL, S. Giuliano Milanese, Italy) and the polyadenylated fraction selected with the Atlas Pure kit (Clontech, Palo Alto, CA). The human 1.2 ATLAS nylon filters (Clontech) were used for the microarray experiments. Single-strand 32P-labeled cDNA probes were synthesized from poly(A+) RNA as suggested by the manufacturer. Video imaging was performed with the Storm 860 Phosphorimager (Molecular Dynamics, Sunnyvale, CA). Microarray data were quantitated with the ATLAS IMAGE 2.0 analysis software (Clontech).In vivo experiments The in vivo experiments were carried out as described.22 Cells were suspended in 199 Hanks Medium, and 0.1 mL (1 × 106 cells per mouse) were intraperitoneally inoculated in SCID mice. ST1346 was dissolved in Cremophor/ethanol 1:1 solution, diluted 1:4 in phosphate-buffered saline at the concentrations of 15 and 30 mg/kg, respectively. ATRA (Sigma) was dissolved in carboxy methyl cellulose (CMC) 0.5% + 0.01% Tween 80 solution at the concentration of 5 mg/kg. Drugs were administered intraperitoneally in a volume of 10 mL/kg one day after the inoculation, and the treatment continued for 3 weeks (5 daily injections per week.
BISINDs as a novel class of retinoid sensitizers A screening process based on the APL-derived NB4 cell line and aimed to select compounds that potentiate the cytodifferentiating activity of ATRA resulted in the identification of a novel class of agents with bis-indolic structures (BISINDs). As exemplified for the prototypical compounds ST1346 and ST1707 in Figure 1, when NB4 cells are exposed to micromolar concentrations of BISINDs for 3 days, no significant effect on the growth, viability, or differentiation of the leukemic blasts is observed. However, when cells are simultaneously exposed to suboptimal concentrations of ATRA (0.1 µM) and ST1346 or ST1707, NBT-reducing activity (NBT-R), a myeloid differentiation marker,32 is enhanced in a dose-dependent manner. Combinations of ATRA and ST1346 or ST1707 are no more effective than the retinoid alone in terms of growth inhibition. In our experimental conditions, cell viability is always above 80% and is not affected by any treatment.
To get insight into the structural determinants responsible for the
pharmacologic activity of BISINDs, we evaluated 12 congeners (Table
1). No molecule in the series increases
NBT-R, regardless of the concentration used. However, ST1707, ST1346,
ST1422, and ST1860 significantly enhance the induction of NBT-R
afforded by ATRA (0.1µM). ST1381 causes a much lower effect, whereas
ST1730, ST1350, ST1866, and ST1371 are devoid of any potentiating
action. Significantly, ST1385, ST1345, and ST1750 inhibit
ATRA-dependent induction of NBT-R. In the presence of a fixed
concentration of ATRA (0.1µM), the AC50 values for the
potentiating effect vary from 0.3 µM in the case of ST1381 to 8.2 µM in the case of ST1422. The relative potency of each BISIND was
calculated at 10 µM (except for ST1730) and results in the rank
order: ST1707 > ST1346 > ST1422 > ST1860 > ST1381. In the absence of ATRA, all the compounds are devoid of
significant antiproliferative effects. With the exception of ST1730,
all the BISINDs are cytotoxic only at concentrations that are
significantly above their corresponding AC50. A first analysis of structure/activity relationships indicates that the bis-indol is strictly required for the observed pharmacologic activity
(Table 1; data not shown). Attempts to identify other determinants of
the potentiating activity of BISINDs suggest that the alkyl chain bound
to the carbon bridging the 2 indol rings and the terminal carboxylic
group are important.
We focused our attention on ST1346. Figure
2 demonstrates that treatment of NB4
cells with ST1346 enhances ATRA-induced NBT-R time dependently (panel
A). Potentiation is observed also in terms of the number of
NBT-R+ cells, which is significantly higher following 3 days of treatment with ATRA + ST1346 than with ATRA alone (panel
B). This demonstrates that the BISIND accelerates the process of
granulocytic maturation activated by ATRA. The growth curve of NB4
blasts observed in control conditions is not altered by ST1346 (panel
C). Similarly, the growth arrest of the leukemic cells triggered by
ATRA is not significantly affected by the BISIND. The level of cell
viability is higher than 85% in all the experimental conditions
considered.
Figure 3A shows that the enhancing effect
of ST1346 on ATRA-induced stimulation of NBT-R is already visible at
the retinoid concentration of 0.01µM and is maintained at higher
concentrations. Potentiation is not the result of a priming effect of
ATRA or ST1346 on NB4 cells because continuous treatment with both
compounds is necessary. As shown in Figure 3B, pretreatment of NB4
cells with ST1346 for one day followed by washing with fresh medium and
incubation with the retinoid for 3 days causes an induction of NBT-R
that is significantly lower than that observed following continuous
treatment for 4 days with the combination of ST1346 + ATRA.
Changing the order of addition of ST1346 and ATRA or extending the
pretreatment period to 3 days results in similar effects. Figure 3C
shows that ST1346-dependent potentiation is observed in the
retinoid-sensitive NB4 parental cell line, but not in the retinoid-resistant NB4.306 subline.24 Figure 3D
demonstrates that the cytodifferentiating effect of ST1346 + ATRA
is evident also in the myeloid leukemia cell line, PR9, which is
engineered to allow conditional expression of the APL oncogene
PML-RAR
In NB4 cells, ST1346 enhances the cytodifferentiating activity not only
of ATRA but also of AM580, a synthetic and selective RAR
Influence of ST1346 on ATRA-dependent myeloid differentiation markers and the gene expression profile of undifferentiated and retinoid-differentiated NB4 cells We established whether the potentiating activity of ST1346 affects the levels of the myeloid CD11b and the monocytic CD14 markers. A typical fluorescence-activated cell-sorter profile of CD11b in NB4 cells treated for 3 days with ATRA (0.01 µM), ST1346 (5 µM), or the combination of the 2 compounds is shown in Figure 5A. A small proportion of NB4 cells is positive for CD11b in basal conditions and following challenge with the BISIND alone. ATRA increases both the number of CD11b+ cells and CD11b cell-associated fluorescence, and these effects are enhanced by ATRA + ST1346. As illustrated in Figure 5B, an increase in the number of CD11b+ cells is evident at ATRA concentrations of 0.001 µM and 0.01 µM and is lost at 0.1 µM (left panel). This is expected, as 0.1 µM ATRA induces expression of CD11b practically on almost all the NB4 cell population. Significantly, however, even at the highest concentration of the retinoid, the amount of CD11b cell-associated fluorescence is higher following challenge with ATRA + ST1346 than with ATRA alone (right panel). None of these effects is evident in the case of CD38, which is maximally induced by ATRA and not further enhanced by ST1346. Treatment of NB4 cells with ATRA or ATRA + ST1346 does not result in the appearance of CD14+ cells.
To identify the genes modulated in NB4 cells by ST1346 alone or in
combination with ATRA, we analyzed the expression of approximately 1200 genes with cDNA microarrays. We focused on genes expressed either early
(8 hours) or relatively late (3 days) along the granulocytic maturation
process, and the results are summarized in Table
2. We observed 38 genes
induced by ATRA. The expression of 20 of these genes is enhanced by
cotreatment with ST1346 for either 8 or 72 hours (group A). Among the
products of these genes, the myeloid nuclear cell differentiation
antigen, intercellular adhesion molecule-1 (ICAM-1), MCP-1,
migration inhibitory factor-related protein (MRP)-14, MRP-8,
and granulocyte colony-stimulating factor (G-CSF) receptor are known
myeloid differentiation markers. For some genes in group A (PRK1, G-CSF
receptor, LIM-kinase, LD-78, c-fgr, and Tyk-2), the combination of
ATRA + ST1346 maintains increased expression for a longer time
than ATRA alone. For other genes (MAPKAP-K2 and cathepsin D), treatment
of NB4 cells with the retinoid and the BISIND accelerates the
up-regulation observed in the presence of ATRA. The remainder of genes
whose expression is induced by ATRA is not further up-regulated by the
addition of ST1346 to the medium (group B). Indeed, in some cases, the BISIND inhibits the induction of gene transcription afforded by the
retinoid (P19INK, TNF receptor, RALB, etc). Group C consists of 10 genes that are activated only following treatment of NB4 cells with
ATRA + ST1346. The majority of these genes are turned on at 8 hours, and increased activity is either maintained or turned off at 72 hours. Paradoxically, however, the steady-state levels of the
transcripts encoding human DNA replication licensing factor (hRlf) beta
subunit, vascular endothelial growth factor (VEGF), and
classical late-infantile neuronal ceroid lipofuscinosis (CLN2) show a
biphasic pattern of expression upon treatment with ATRA + ST1346,
that is, an increase at 8 hours and a decrease at 72 hours.
Group D is composed of the genes coding for cathepsin C, heat shock factor 1, and replication factor C, which are activated by ST1346 alone. Group E consists of 19 genes whose expression is significantly down-regulated by ATRA and includes those coding for the 2 cytokines IL-1 and TNF. Finally, group F includes 3 genes, ARF1, DNA polymerase Delta, and M-CSF, whose level of expression is significantly down-regulated only by ATRA + ST1346. For a number of genes in Table 2, we measured the relative protein or
RNA products. Flow cytometric analysis of ICAM-1 demonstrates that more
than 90% of the undifferentiated NB4 blasts express the adhesion
molecule. As shown in Figure 6A, ST1346
(10 µM) does not alter the levels of cell-associated ICAM-1, whereas
ATRA augments it in a dose-dependent manner. Cotreatment of NB4 cells
for 3 days with the BISIND and ATRA enhances the surface expression of
ICAM-1 by approximately 2-fold at both the concentrations of the
retinoid considered. By contrast, ATRA and ATRA + ST1346 are equally efficient in up-regulating integrin
Effects of ST1346 on the levels, stability, and activation of retinoic acid receptors and histone acetylation NB4 blasts express significant amounts of both RAR and
PML-RAR, and transfection of this cell line is technically
challenging. Thus, the effect of ST1346 on the trans-activating
properties of RAR, PML-RAR, or RXR was investigated in COS-7
cells, which are endowed with relatively low levels of nuclear retinoic
acid receptors23 and easily transfected. As illustrated in
Figure 7A, following cotransfection with
a void plasmid (pSG5) and TRE-TK-Luc (containing a luciferase
reporter gene controlled by a retinoid-dependent artificial promoter),
treatment of cells with ATRA results in an approximately 5-fold
increase in the levels of luciferase activity relative to what is
observed in basal conditions. This effect is explained by the presence
of endogenous retinoic acid receptors in COS-7 cells. ST1346 has no
significant effect on either the basal or the ATRA-induced levels of
luciferase activity. Forced expression of RAR RXR RAR RXR,
and PML-RAR (data not shown) increases the ATRA-dependent induction
of luciferase activity to varying degrees. Again, ST1346 has no
significant effect on the ligand-dependent or independent
trans-activating properties of the various receptors. Varying the
concentrations of ATRA and ST1346 or changing the type of recipient
cell lines and reporter constructs does not alter the results (data not
shown). Although our data do not rule out possible specific effects
related to the myeloid cell context, they suggest that ST1346 does not
modulate the transcriptional activity of RAR or PML-RAR
complexes directly.
Treatment of APL blasts with ATRA is associated with degradation
of both PML-RAR Influence of ST1346 on nuclear transcriptional factors and complexes involved in ATRA-dependent granulocytic maturation of APL cells Various types of transcriptional factors, including the STAT1 and CEBP proteins as well as the ets and
cAMP-dependent CREB binding complexes, are implicated in the
process of APL blast differentiation activated by
ATRA.37-40 As expected37 and as illustrated
in Figure 8A (left panel), treatment of
NB4 cells for 3 days with ATRA causes a significant increase in the
amounts of STAT1 (91 kDa) and STAT1(84 kDa). ST1346 enhances the
ATRA-dependent induction of both forms of STAT1 (left panel). A
quantitative assessment of the enhancing effect of ST1346 on the
ATRA-dependent induction of STAT1 demonstrates that the BISIND
causes a 3-fold induction of the protein over what is observed in the
presence of the retinoid alone (right panel). At least a proportion of the ATRA- and ST1346 + ATRA-induced STAT proteins must be in the transcriptionally active state, as indicated by the level of
phosphorylation of Tyr701 (left panel). As shown in Figure 8B,
CEBP is not synthesized by undifferentiated NB4 cells. Following
ATRA treatment, significant amounts of CEBP are already evident at
12 hours and increase up to 72 hours. ST1346 treatment does not result
in the synthesis of the CEBP protein and has only marginal effects
on the level of induction afforded by the retinoid. Figure 8C
demonstrates that nuclear extracts of undifferentiated NB4 cells
contain detectable amounts of CREB and Ets-1
complexes, as assessed by gel retardation assays. Treatment of cells
for 4 hours with ST1346, but not ATRA, results in the disappearance of
the Ets-1 complex. This inhibitory effect is completely
suppressed if cells are simultaneously exposed to ST1346 and
ATRA. ST1346-dependent inhibition disappears following 16 hours
of culturing. By this time, NB4-associated Ets-1 binding activity is similar in all the experimental conditions considered. In
the case of CREB, ATRA reduces the amounts of the specifically retarded
band observed in control conditions at both 4 and 16 hours. This effect
is reversed by addition of ST1346 to the culture medium.
Role of JNK in the differentiation of NB4 cells triggered by ATRA and enhanced by ST1346 MAP kinases, ERK, p38, and JNK are involved in the process of differentiation of myeloid cells.41-44 As illustrated in Figure 9A, undifferentiated NB4 cells show high levels of constitutive ERK phosphorylation. The predominant form of phosphorylated ERK is p42. Treatment of cells with ATRA, ST1346, and the combination of the 2 compounds for 8 or 72 hours does not affect the steady-state levels or the phosphorylation of ERK p42 and p44 proteins. As shown in Figure 9B, in conditions of complete suppression of ERK activation (data not shown), the MEK inhibitor U0126 does not affect the proportion of NBT+ NB4 cells nor the level of cell-associated NBT-R activity observed following treatment with vehicle, ATRA, ST1346, or ATRA + ST1346. This indicates that constitutive ERK phosphorylation and activation do not play any significant role in the process of granulocytic maturation activated by ATRA and enhanced by ST1346.
Figure 9C demonstrates that NB4 cells express detectable amounts of p38 protein, and the quantity of the polypeptide is not altered by any of the treatments considered. Following 72 hours of challenge, ATRA increases p38 phosphorylation. ST1346 has no detectable effect on the ATRA-induced level of p38 phosphorylation. Figure 9D demonstrates that the specific p38 inhibitor PD169316 enhances the induction of NBT-R activity observed in the presence of ATRA. Interestingly, addition of the inhibitor to ST1346 and ATRA results in a level of NBT-R activity that is slightly but significantly higher than that observed in the presence of the combination of the BISIND and the retinoid. Altogether, our results indicate that activation of p38 plays a negative role in the process of granulocytic maturation of APL cells. However, they suggest that enhancement of NB4 cell differentiation by ST1346 does not involve p38 inhibition. As shown in Figure 9E, 2 forms of JNK, p46 and p54, are synthesized by undifferentiated NB4 cells. Both forms are phosphorylated in basal conditions, although the level of p54 phosphorylation is higher than that of p46. Following 8 or 72 hours of treatment with ATRA, ST1346, or the combination of the 2 compounds, no significant variation in the amounts of p46 and p54 are observed. At 8 hours, the level of p46 or p54 is similar in all the experimental groups considered; at 72 hours, ATRA inhibits JNK phosphorylation. This down-regulation is relieved in cells treated with ATRA + ST1346. As expected and as illustrated in Figure 9F, inhibition of JNK phosphorylation by ATRA is associated with a decrease in the corresponding c-JUN kinase activity. JNK activity is brought back to control levels by the combination of ST1346 + ATRA. Treatment of NB4 cells with the JNK inhibitor SP600125 blocks JNK activity in all the experimental conditions considered. This is accompanied by a decrease in the percentage of NBT-R-positive NB4 cells observed upon challenge with ATRA + ST1346 to a level similar to that observed following treatment with ATRA alone. The effect is evident not only in terms of the number of NBT-R-positive cells but also in terms of cell-associated enzymatic activity. Thus, prolonged JNK activation by ATRA + ST1346 may be crucial for the enhancement of the retinoid cytodifferentiating effect by the BISIND. In vivo efficacy of the combination of ATRA + ST1346 To evaluate whether the potentiating effect of ST1346 on ATRA cytodifferentiating activity has an impact on the therapeutic efficacy of the retinoid, we performed experiments in SCID mice that received transplanted NB4 cells intraperitoneally.22 As shown in Table 3, treatment of animals with a suboptimal dose of ATRA (5 mg/kg) or 2 doses of ST1346 (15 and 30 mg/kg) does not cause a significant increase in the mean survival time (MST). Coadministration of ST1346 and ATRA results in an increase of MST, which is statistically significant at the higher dose of BISIND. The effect on MST and increased lifespan (ILS) obtained with ATRA + ST1346 (30 + 5 mg/kg) is similar to that observed with a maximally tolerated dose of ATRA (15 mg/kg) (data not shown). No dosage regimen is accompanied by significant toxicity as determined by the lack of body weight loss and by the absence of toxic deaths. These results suggest that enhancement of ATRA-induced cytodifferentiation by BISIND can be exploited to increase the in vivo efficacy of the retinoid.
Here, we report on the identification of a novel class of experimental agents, BISINDs, that potentiate the cytodifferentiating activity of ATRA in APL and other acute myeloid leukemia (AML) cells. The BISIND prototypical molecule ST1346 exerts a general stimulating effect on many aspects of the granulocytic maturation program set in motion by ATRA. The potentiating effect of ST1346 is specifically directed toward ATRA-induced cytodifferentiation, as the compound leaves unaltered the antiproliferative and apoptogenic activity of the retinoid. Administration of ST1346 to SCID mice xenografted with NB4 cells dose-dependently stimulates the therapeutic effect of an otherwise ineffective dose of ATRA and significantly increases the lifespan of leukemia-bearing animals. These results are consistent with the fact that enhancement of the cytodifferentiating potential of ATRA is a viable strategy to incr | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
) or ATRA (0.1µM) (
) for 3 days. Aliquots of cells
were used for the determination of NBT-reducing activity (A,D), total
number of cells (B,E), and cell viability (C,F). The results shown are
the means ± SD of 3 separate cultures. *Significantly higher than
the corresponding group treated with vehicle (P < .01
according to the Student t test). °Significantly higher
than the sum of the effects observed in the corresponding ST1346-,
ST1707-, and ATRA-treated groups, following 2-way analysis of variance
and measurement of the F score of the interaction
(P < .01 according to Tukey test).
myeloid leukemia cell
types, that is, the myelocytic HL-60, the myelomonoblastic U937, and
Kazumi, as well as the myeloblastic KG1 cell lines. Treatment of the
ATRA-sensitive HL-60 blasts with increasing concentrations of ST1346
and a suboptimal concentration of ATRA leads to significant induction
of NBT-R. In this case, maximal induction of NBT-R is observed
following addition of 50 µM ST1346 to ATRA. In the experimental
conditions used, U937 and Kazumi cells are refractory to challenge with
ATRA, while the combination of ST1346 + ATRA causes a variable
proportion of both cell populations to become NBT positive. It is worth
noticing that maximal induction of NBT positivity in Kazumi cells by
ST1346 + ATRA is observed at 2 different concentrations of the
BISIND (10 and 50 µM) and takes longer than in other cell types. KG1 is totally unresponsive to both ATRA and the combination of the bis-indol and the retinoid.
