Identification of benzodiazepine Ro5-3335 as an
inhibitor of CBF leukemia through quantitative high
throughput screen against RUNX1–CBFβ interaction
Lea Cunninghama, Steven Finckbeinera, R. Katherine Hydea, Noel Southallb, Juan Maruganb, Venkat R. K. Yedavallic,
Seameen Jean Dehdashtib, William C. Reinholdd, Lemlem Alemua, Ling Zhaoa, Jing-Ruey Joanna Yehe, Raman Sooda,f,
Yves Pommierd, Christopher P. Austinb, Kuan-Teh Jeangc, Wei Zhengb,1, and Paul Liua,1
aOncogenesis and Development Section,fZebrafish Core, National Human Genome Research Institute,bNational Center for Advancing Translational Sciences,
cMolecular Virology Section, National Institute of Allergy and Infectious Diseases, anddLaboratory of Molecular Pharmacology, Center for Cancer Research,
National Cancer Institute, National Institutes of Health, Bethesda, MD 20892; andeCardiovascular Research Center, Massachusetts General Hospital,
Charlestown, MA 02129
Edited by Dennis A. Carson, University of California at San Diego, La Jolla, CA, and approved July 26, 2012 (received for review January 3, 2012)
Core binding factor (CBF) leukemias, those with translocations or
inversions that affect transcription factor genes RUNX1 or CBFB, ac-
count for ∼24% of adult acute myeloid leukemia (AML) and 25% of
pediatric acute lymphocytic leukemia (ALL). Current treatments for
CBF leukemias are associated with significant morbidity and mortal-
ity, with a 5-y survival rate of ∼50%. We hypothesize that the in-
teraction between RUNX1 and CBFβ is critical for CBF leukemia and
canbetargeted for drugdevelopment. We developedhigh-through-
put AlphaScreen and time-resolved fluorescence resonance energy
transfer (TR-FRET) methods to quantify the RUNX1–CBFβ interaction
and screen a library collection of 243,398 compounds. Ro5-3335, a
benzodiazepine identified from the screen, wasable to interact with
RUNX1 and CBFβ directly, repress RUNX1/CBFB-dependent transacti-
vation in reporter assays, and repress runx1-dependent hematopoi-
esis in zebrafish embryos. Ro5-3335 preferentially killed human CBF
leukemia cell lines, rescued preleukemic phenotype in a RUNX1–ETO
MYH11 leukemia model. Our data thus confirmed that RUNX1–CBFβ
interaction can be targeted for leukemia treatment and we have
identified a promising lead compound for this purpose.
encoding both proteins play key roles in hematopoiesis (1) and
are involved in leukemogenesis through recurrent chromosome
abnormalities (2), such as a chromosome 16 inversion [(inv)16]
that generates a fusion gene between CBFB and MYH11
(encoding the smooth muscle myosin heavy chain, SMMHC) in
acute myeloid leukemia (AML) subtype M4Eo (3, 4), a trans-
location between chromosomes 8 and 21 that generates a fusion
gene between RUNX1 and ETO in AML subtype M2 (5), and
a translocation between chromosomes 12 and 21 that generates
a fusion gene called TEL–RUNX1 in pediatric precursor B-cell
acute lymphocytic leukemia (ALL) (6). All together, the CBF
leukemias, which contain translocations involving RUNX1 or
CBFB, account for 24% of adult AML cases (7) and 25% of
pediatric ALL cases (8). Although core binding factor (CBF)
leukemias are generally associated with relatively favorable
prognoses, long-term survival for adult patients with CBF AML
is only about 50% (9). Although children with CBF leukemias
have survival rates of >80% (8, 10), standard therapy takes years
to complete. Moreover, the current standard of care for all
patients is frequently associated with significant morbidity and
mortality. Therefore, targeted treatments for CBF leukemia with
high efficacy and low toxicity are clearly desirable.
Previous studies suggest that the physical interactions between
RUNX1 fusion proteins (RUNX1–ETO and TEL–RUNX1) and
CBFβ, and between the CBFβ fusion protein (CBFβ–SMMHC)
and RUNX1 are critical for the pathogenesis of CBF leukemias
(11–13). We therefore hypothesize that inhibitors of the RUNX1–
ranscription factors RUNX1 and CBFβ form a heterodimer
for DNA binding and regulation of gene expression. Genes
CBFβ interaction will be therapeutic for all CBF leukemias.
To identify such inhibitors, we conducted a quantitative high-
throughput screen (qHTS) of a diverse chemical library using
a unique RUNX1–CBFβ interaction assay.
Development of High-Throughput Screen Assays for RUNX1–CBFβ
Interaction. We developed an AlphaScreen assay to measure
the interaction between RUNX1 and CBFβ (Fig. S1A). In this
assay, biotinylated RUNX1 (biotin-RUNX1) is captured by a
streptavidin-coated donor bead and His-tagged CBFβ [His(6×)-
CBFβ] is bound by a nickel chelated acceptor bead. The binding
of RUNX1 to CBFβ brings the donor and acceptor beads in
close proximity, enabling energy transfer between them, resulting
in a chemiluminescence signal (14). An inhibitor of the RUNX1–
CBFβ interaction will disrupt the energy transfer between the
donor and acceptor beads, resulting in a reduction of chem-
iluminescence signal. The signal amplification properties of the
AlphaScreen method allow for the homogenous detection of
protein–protein interactions with high sensitivity, making this
assay amenable to high-throughput screening.
A time-resolved fluorescence resonance energy transfer (TR-
a different pair of capture reagents for time-resolved fluorescence
(15) as the detection method (Fig. S1B). Similar to AlphaScreen,
the TR-FRET signal is decreased in the presence of an inhibitor of
The AlphaScreen assay was optimized for several parameters
(Fig. S2 A–E) and validated with an untagged CBFβ that dis-
placed the binding between tagged RUNX1 and CBFβ in a con-
centration-dependent manner (Fig. S2F). The AlphaScreen assay
was then miniaturized to 1,536-well plate format for qHTS. The
signal-to-basal ratio was 6.5-fold and the Z′ factor was 0.77, in-
dicating a robust assay for qHTS (Fig. S2G). The TR-FRET assay
Author contributions: L.C., S.F., R.K.H., N.S., J.M., Y.P., C.P.A., K.-T.J., W.Z., and P.L. designed
research; L.C., S.F., R.K.H., V.R.K.Y., S.J.D., W.C.R., L.A., L.Z., R.S., and W.Z. performed re-
search; J.-R.J.Y. contributed new reagents/analytic tools; L.C., S.F., R.K.H., N.S., J.M., V.R.K.Y.,
S.J.D., W.C.R., L.Z., R.S., Y.P., C.P.A., K.-T.J., W.Z., and P.L. analyzed data; and L.C., W.Z., and
P.L. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The primary and secondary chemical screen data have been deposited
in the PubChem database, http://pubchem.ncbi.nlm.nih.gov (accession nos. AID 1484 and
1To whom correspondence may be addressed. E-mail: email@example.com or wzheng@mail.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| September 4, 2012
| vol. 109
| no. 36www.pnas.org/cgi/doi/10.1073/pnas.1200037109
was similarly optimized (Fig. S2H); the signal-to-basal ratio was
fourfold and the Z′ factor was 0.63 (Fig. S2I).
qHTS and Hit Confirmation. The National Institutes of Health
(NIH) Molecular Library Collection of 243,398 compounds was
screened at a seven-concentration titration using the RUNX1–
CBFβ AlphaScreen assay. We identified 148 compounds as the
initial primary hits, which were selected on the basis of their
structural diversity, potency, efficacy, as well as previously re-
ported activities in the literature. Some of these initial hits were
found to be active in multiple other screens performed at the
NIH Chemical Genomics Center (NCGC), suggesting they are
nonspecific inhibitors and were therefore eliminated from fur-
ther consideration. As a result, 131 compounds were selected as
true primary hits for further confirmatory experiments.
For confirmation, these compounds were first rescreened in the
using a biotin-(6×) histidine linker to replace the biotin-RUNX1
and His(6×)-CBFβ proteins in the AlphaScreen assay. The com-
pounds that were active in the RUNX1–CBFβ AlphaScreen assay
but inactive in the counterscreen assay were the confirmed active
compounds. Among the 131 primary hits, 32 were confirmed this
way. To eliminate additional AlphaScreen-related false positive
compounds, such as singlet oxygen or luminescence signal absor-
bent compounds, the same set of compounds was tested in the
RUNX1–CBFβ TR-FRET assay. Ultimately, three compounds,
confirmed as inhibitors of the RUNX1–CBFβ interaction. Results
from the primary and secondary screens have been deposited in
the PubChem database (http://pubchem.ncbi.nlm.nih.gov/, ac-
cession nos. AID 1484 and AID 504373).
Ro5-3335 Was Identified as an Inhibitor of RUNX1–CBFβ Function in
Zebrafish Models. RUNX1 and CBFβ proteins are required for
definitive hematopoiesis during embryogenesis (16–18). Because
interaction with CBFβ is required for RUNX1 activity (17, 19),
inhibitors of the RUNX1–CBFβ complex are expected to block
definitive hematopoiesis in zebrafish embryos.
We treated transgenic cd41-GFP and lck-GFP zebrafish em-
bryos [with GFP expression in thrombocytes/platelets (20) and T
cells (21), respectively] with individual candidate compounds, and
the numbers of GFP+cells in the treated embryos were scored. As
shown in Fig. 1 A and B, NSC140873 significantly reduced the
numbers of GFP+cells, suggesting that it can reduce definitive
hematopoiesis in the embryos. This reduction was not due to
generalized toxicity because the embryos developed normally in
general. Notably, NSC140873 was the only positive compound in
this assay among the 49 compounds tested, which included the
other two lead compounds(MLS000548294 and MLS001048862).
NSC140873 [also known as 2-glycineamide-5-chlorophenyl 2-
pyrryl ketone (GCPK) and MLS000766105] had been previously
reported to inhibit Tat-dependent replication of HIV-1 (22).
NSC140873 has an unstable structure and can be converted
spontaneously in solution to a benzodiazepine, Ro5-3335 (Fig.
1C). Interestingly, Ro5-3335 and its analog, Ro24-7429 (Fig. 1C),
were also reported previously as inhibitors of HIV-1 replication
(23, 24). Therefore, we applied Ro5-3335 and Ro24-7429 to the
cd41-GFP transgenic zebrafish embryos and found that both
compounds similarly inhibited definitive hematopoiesis (Fig. 1D).
In subsequent experiments, we concentrated our efforts on Ro5-
3335, because NSC140873 spontaneously cyclizes to Ro5-3335
and it showed higher potency than Ro24-7429.
Importantly, these findings are consistent with the phenotype
of a zebrafish runx1-truncation mutant (18, 25). To determine
whether Ro5-3335 reduces definitive hematopoiesis through
blocking RUNX1–CBFβ interaction, we treated cd41-GFP zebra-
fish embryos carrying one allele of a runx1 truncation mutation
(runx1+/−) with Ro5-3335. We found that these runx1-deficient
embryos were more sensitive to Ro5-3335 inhibition than the
wild-type cd41-GFP embryos (Fig. 1E). This finding supports the
conclusion that these compounds block RUNX1/CBFβ-dependent
Ro5-3335 Interacts with both RUNX1 and CBFβ but Does Not
Completely Disrupt RUNX1–CBFβ Complex. To demonstrate that
Ro5-3335 physically interacts with RUNX1 and/or CBFβ, we
conducted a UV absorption depletion assay, which is based on the
fact that Ro5-3335 has UV absorption peaks at 330 and 374 nm
(Fig. S3). We used this assay to measure the disappearance or
depletion of Ro5-3335 from solution following exposure to nickel
resin-bound His-tagged RUNX1 or CBFβ. As shown in Fig. 2A,
both bound His-tagged RUNX1 and CBFβ (with the former
showing stronger activity) removed Ro5-3335 from the solution,
indicating they physically bind Ro5-3335. Interestingly, however,
Ro5-3335 did not affect the formation of RUNX1–CBFβ complex
in gel-shift or pull-down assays (Fig. S4). The data suggest that
Ro5-3335 does not completely break apart RUNX1–CBFβ
hibit definitive hematopoiesis in zebrafish em-
bryos. (A) NSC140873 reduced circulating cd41-
(Upper) and lck-GFP+
transgenic zebrafish embryos compared with
DMSO-treated embryos. Red arrows point to cir-
culating cd41-GFP+cells in the heart and lck-GFP+
cells in the thymus, respectively. (B) Dose-de-
pendent reduction of circulating cd41-GFP cells by
NSC140873. (C) Chemical structures of the three
related compounds. (D) Dose-dependent reduction
of circulating cd41-GFP cells by Ro5-3335 and Ro24-
7429. (E) The runx1 W84× truncation mutation
sensitizes fish embryos to the inhibition by Ro5-
3335. *Numbers of untreated runx1+/−embryos or
DMSO-treated runx1+/−embryos with intermediate
or no visible circulating CD41-EGFP+cells were
compared.#Numbers of runx1+/−embryos or wild-
type embryos with intermediate or no visible cir-
culating CD41-EGFP+cells after Ro5-3335 treat-
ments were compared. For all comparisons in B, D,
and E, the P values were calculated with 2 × 2
contingency tables and Fisher’s exact test.
NSC140873, Ro5-3335, and Ro24-7429 in-
(Lower) cells in
Cunningham et al.PNAS
| September 4, 2012
| vol. 109
| no. 36
interaction, but changes the conformation of their complex or
Ro5-3335 Inhibits Transcriptional Regulation by RUNX1 and CBFβ. The
macrophage colony-stimulating factor receptor (MCSFR) pro-
moter is a known target of RUNX1 and CBFβ. Consistent with
previous reports (26, 27), we found that RUNX1 and CBFβ
activated the MCSFR promoter, both individually and syner-
gistically (Fig. 2B). In the presence of Ro5-3335, the activation of
the MCSFR promoter by either protein alone or together was
significantly reduced. The inhibition by Ro5-3335 was most
dramatic on the synergistic activation by CBFβ and RUNX1,
with statistically significant decreases at all concentrations of
Ro5-3335 tested (0.5, 5, and 25 μM) (Fig. 2B).
We also explored the potential function of RUNX1 and CBFβ
in HIV Tat-dependent transcription because Ro5-3335 and
Ro24-7429 were initially characterized as inhibitors of HIV rep-
lication. We found that Tat could competitively inhibit RUNX1–
CBFβ interaction in the TR-FRET assay (Fig. S5A). In addition,
using a surface plasmon resonance (SPR) assay we found that Tat
bound directly to RUNX1 with a Kdvalue of 3.5 μM (Fig. S5B).
Moreover, we found that RUNX1 and CBFβ synergistically
inhibited HIV-1 LTR transactivation in a reporter assay (Fig.
S5C). As expected, the inhibition by RUNX1 and CBFβ could be
reversed by the addition of Ro5-3335.
Ro5-3335 Selectively Reduces Viability of Human CBF Leukemia Cells
in Culture. We then hypothesized that the RUNX1–CBFβ in-
teraction was more critical for the leukemic fusion proteins than
for the normal RUNX1 and CBFβ proteins, and therefore leu-
kemia cells with the CBF fusion proteins would be more sensitive
to inhibitors of the RUNX1–CBFβ interaction than those without
such fusion proteins. To test this hypothesis, we applied Ro5-
3335 to human leukemia cells with or without CBF fusion pro-
teins. As shown in Fig. 2 C and D, cell-killing activities of Ro5-
3335 for human CBF leukemia cell lines were 6- to 50-fold more
potent than for human leukemia cell lines without CBF fusion
proteins. Therefore, the results demonstrated that Ro5-3335
preferentially killed leukemia cells with CBF fusion proteins.
NSC140873 and Ro5-3335 have been tested for their anti-
proliferative activities against the NCI-60 cancer cell lines (28).
Moreover, gene expression patterns of the NCI-60 cancer cell lines
have been profiled through multiple microarray platforms (29, 30).
We compared the sensitivities of the cell lines to NSC140873 and
Ro5-3335 with expression levels of RUNX1 and CBFB in these
cells. As shown in Fig. S6, we found that the RUNX1 and CBFB
expression patterns in these cell lines correlated with their sensi-
tivities to NSC140873 and Ro5-3335. Those with higher RUNX1
and CBFB expressions are likely to be more sensitive to these two
compounds, and vice versa. This is especially true for CBFB—the
overall expression profile of CBFB in all 60 cell lines correlated
strongly with overall NSC140873 antiproliferative profile (P <
0.05). As expected, there is a robust correlation between the
antiproliferative profiles of the two compounds (P < 0.001). These
findings provide additional support to the hypothesis that these
two compounds target the RUNX1/CBFβ pathway.
Ro5-3335 Rescues the Hematopoietic Defects in the AML1–ETO
(RUNX1–ETO) Transgenic Zebrafish. A RUNX1–ETO transgenic
zebrafish line was recently described (31) that recapitulates some
aspects of human AML when the expression of the RUNX1–
ETO gene is induced by heat shock during embryogenesis. The
DMSO 0.5 µm 5 µm 25 µ µm
Relative luciferase activity
Cell Line IC50 (µ µM)
*ME-1 (CBFB-MYH11) 1.1
HL60 (AML FAB M3) 54.6
Kasumi-6 (AML FAB M2) >200
*REH (TEL-RUNX1) 17.3
RS4 (B-ALL) 108.9
* = CBF leukemia lines
concentration in the supernatants after incubating the indicated components followed by centrifugation. (B) Ro5-3335 inhibits RUNX1/CBFβ transactivation.
Expression of luciferase driven by the MCSFR promoter was assayed in the presence of no transfected RUNX1 or CBFβ (gray bars), RUNX1 alone (white bars),
CBFβ alone (black bars), and RUNX1 and CBFβ together (cross-hatched bars) in the presence of DMSO or the indicated concentrations of Ro5-3335. Luciferase
expression was normalized to Renilla expression driven by the thymidine kinase promoter and displayed as fold change over the MSCFR promoter alone in
DMSO. *P < 0.05, **P < 0.01, ***P < 0.00001 compared with DMSO. (C and D) Antiproliferative activity of Ro5-3335 against CBF leukemia. (C) Cell viability as
measured by ATP activity in the indicated cell lines that were incubated with Ro5-3335 for 48 h at multiple concentrations. (D) IC50values for Ro5-3335 against
the indicated cell lines.
Ro5-3335 inhibits RUNX1/CBFβ functions in vitro. (A) Ro5-3335 UV absorption depletion assay. Absorbance at 374 nm was used to assay Ro5-3335
| www.pnas.org/cgi/doi/10.1073/pnas.1200037109Cunningham et al.
induced phenotype includes accumulation of immature hema-
topoietic cells in the posterior blood island and the concomitant
loss of circulating blood cells during primitive hematopoiesis
(31). The phenotype is likely due to a hematopoietic differenti-
ation defect caused by RUNX1–ETO, which is different from the
definitive hematopoietic stem cell defect in the runx1 mutant
fish. We treated the RUNX1–ETO transgenic embryos with Ro5-
3335 and found that Ro5-3335 rescued the hematopoietic
defects in those embryos in a dose-dependent fashion (Fig. 3).
These data indicate that Ro5-3335 is able to inhibit RUNX1–
ETO function in vivo, which contributes to leukemogenesis.
Pharmacokinetic and Toxicity Studies of Ro5-3335 in Mice. On the
basis of the above results, we then asked whether Ro5-3335 is
capable of inhibiting CBF leukemia progression in mice. We first
determined the pharmacokinetic responses of Ro5-3335 using
oral administration (mixed in the feeding dough or gavage) or i.p.
injection. Oral administration within the feeding dough main-
tained relatively high serum concentrations for a much longer time
than those achieved by oral gavage or i.p. injection, even though
the peak concentrations were not as high (Fig. S7). The peak
serum level of Ro5-3335 (1.2 μM), achieved with 300 mg/kg/d for
5 d in feeding dough, is within the lower range of effective
dosages for suppressing hematopoiesis in zebrafish (Fig. 1E) and
inhibiting human CBF leukemia cells (Fig. 2 C and D).
To evaluate the long-term effects of Ro5-3335, we treated wild-
type mice with Ro5-3335 orally at 300 mg/kg/d for 1 mo. The
treated mice were healthy with no obvious illness. However, they
had reduced total red blood cell count, increased platelets, and
changes in white blood cell differentials in the peripheral blood
(PB) (Fig. S8). Overall bone marrow (BM) cellularity was not af-
granulocyte–monocyte (GM) colonies (Fig. S9). Flow cytometry
both PB and BM (Fig. S10). The data showed that long-term ad-
ministration of Ro5-3335 is well tolerated, although close hema-
tological monitoring is required.
Lin−Cbfb–MYH11 Bone Marrow Cells Are Sensitive to Ro5 Treatment.
Previously we developed a conditional Cbfb–MYH11 knock-in
mouse model (32) that faithfully recapitulates human leukemia
with this fusion gene (33). The expression of Cbfb–MYH11 can
be controlled by Mx1–Cre, which is induced by injecting the mice
with pI:pC (34). We cultured the BM cells from the Cbfb–
MYH11 knockin mice 2 wk after pIpC injection (before leukemia
developed) to determine the effect of Ro5-3335 treatment on
We found that the total cell number after culturing Cbfb–
MYH11 cells with Ro5-3335 for 48 h at 0.1 and 1 μM was sig-
nificantly reduced compared with untreated cells (Fig. S11A). On
the other hand, control BM cells did not show statistically sig-
nificant cell loss after Ro5-3335 treatment (Fig. S11B). More-
over, the Lin−progenitor and stem cells from the Cbfb–MYH11
knockin mice were more sensitive to Ro5-3335 than Lin+cells
(Fig. S11C). In addition, leukemic cells from Cbfb–MYH11 knockin
mice plated in methylcellulose and treated with 1 μM Ro5-3335 for
48 h demonstrated significantly fewer CFU-GM and CFU-
granulocyte–erythroid–monocyte–megakaryocyte (GEMM) col-
onies (Fig. S11 D and E).
These data demonstrate that Cbfb–MYH11 BM cells, espe-
cially the Lin−cells, are more sensitive to the cytotoxic effect of
Ro5-3335 than control BM cells.
Ro5-3335 Reduces Leukemia Burden in Cbfb–MYH11 Knockin Mice.
The Cbfb–MYH11 knockin mice develop leukemia with high
penetrance but variable latency. However, when the leukemia
cells were transplanted, the recipient mice developed leukemia
with short, consistent latency and 100% penetrance. We there-
fore used this leukemia transplantation model to test Ro5-3335.
The Cbfb–MYH11 leukemia cells were transplanted into sub-
lethally irradiated wild-type mice. Ten days after transplantation,
at 300 mg/kg/d orally for 30 d. Other groups were treated with
a standard chemotherapy drug, cytarabine, alone or in combina-
tion with Ro5-3335, to test their potential synergy. We monitored
the disease progression with weekly PB analysis (27, 33).
Ro5-3335 was able to reduce the number of c-kit+cells in the
transplanted mice (Fig. 4A) more effectively than cytarabine. In
addition, the combined treatment with Ro5-3335 and cytarabine
exhibited synergistic effects (Fig. 4B). Morphologically, PB from
mice treated with Ro5-3335, cytarabine, or their combination,
contained mostly differentiated cells, whereas the blood from
untreated mice contained mostly leukemic cells (Fig. 4C). Nec-
ropsy examination revealed that the Ro5-3335–treated mice had
the least amount of leukemic infiltrations among the treated
groups (Fig. 4D). Especially striking was the reduction of leu-
kemic cell infiltration in the livers of Ro5-3335–treated mice
(Fig. 4D and Fig. S12A). Bone marrow (Fig. 4D and Fig. S12B)
and spleen (Fig. 4E and Fig. S12C) infiltrations in the Ro5-3335–
treated mice were also lower than mice in other groups.
Mice receiving cytarabine or both cytarabine and Ro5-3335 had
significantly increased life expectancy compared with untreated
mice (Fig. 4 F and G; P < 0.002 and P < 0.00001, respectively).
Importantly, there was a synergy between Ro5-3335 and cytar-
abine for extending the survival of the treated mice (P < 0.04
compared with cytarabine alone; P < 0.008 compared with Ro5-
3335 alone), suggesting the potential of cytarabine/Ro5-3335
combination therapy clinically. Ro5-3335 alone also prolonged
the survival of the treated mice but the difference with control
mice did not reach statistical significance, because some mice died
early (Fig. 4F), likely not from leukemia.
We have developed two robust screening assays using AlphaScreen
By applying these two assays in a library screen of ∼250,000
Percentage of total embryos with phenotype
DMSO 25 M10 M
embryos. (A) RUNX1–ETO transgenic embryos at 48 h postfertilization (hpf).
blood cells in the heat-shocked embryo. (B) Heat-shocked RUNX1–ETO trans-
or DMSO (Lower). (C) Bar graphs showing the effect of incubating heat-
shocked RUNX1–ETO transgenic embryos with Ro5-3335 at the indicated
concentrations. Black bars, embryos with blood pooling and no circulation;
gray bars, embryos without blood pooling but also no circulation; white bars,
embryos without blood pooling and with circulating cells. Means of three
independent experiments are shown with SDs. *P < 0.03; **P < 0.003.
Cunningham et al.PNAS
| September 4, 2012
| vol. 109
| no. 36
compounds, we have identified and confirmed three compounds as
inhibitors of the RUNX1–CBFβ interaction. Through a series of
that the compound Ro5-3335 binds RUNX1 and CBFβ, inhibits
their functions, and is a potent inhibitor of CBF leukemia. Our
studies, therefore, confirmed the hypothesis that the RUNX1–
CBFβ interaction is indeed critical for the RUNX1 and CBFβ fu-
sion proteinsinleukemiaand isa targetfornew drug development.
Our overall data also suggest that Ro5-3335 is a promising com-
pound that will lead to a unique, targeted treatment for CBF
Our study demonstrates the utility and power of the zebrafish
as a convenient and effective in vivo model to screen compounds
for their specificity, efficacy, and toxicity. Due to the small size
and high fecundity of the zebrafish, it is relatively easy to test
multiple compounds quickly. The availability of genetic tools
such as transgenic fish with tissue-specific expression of fluores-
cent proteins significantly simplified the screening process. Using
the zebrafish model, we were able to promptly focus our efforts
on one compound soon after the primary screen.
We found that Ro5-3335 interacts with both RUNX1 and
CBFβ, with stronger affinity for the former (Fig. 2A). However,
Ro5-3335 did not disrupt RUNX1–CBFβ interaction completely
(Fig. S4). Ro5-3335 may be sandwiched inside the RUNX1/CBFβ
interaction surface and increase the distance between the two
proteins (hence the reduced signals in the AlphaScreen and TR-
FRET assays). It is also possible that Ro5-3335 binding may in-
duce conformational changes in RUNX1 or CBFβ without com-
pletely disrupting the heterodimer. More extensive structural
biological studies are needed to discern these possibilities. Such
changes may result in changes of affinity for target DNA as well as
alterations in interaction with other partner proteins.
Ro5-3335 and its analog Ro24-7429 were previously developed
as anti-HIV drugs that block HIV gene expression by inhibiting
Tat-mediated transactivation (22, 35). Ro24-7429 entered phase
II clinic trial but failed to show clinic efficacy in HIV patients,
although the drug was tolerated well by patients (24). To our
knowledge no one has reported activity of these compounds
against leukemia, nor has anyone reported that they target
RUNX1 and CBFβ. Furthermore, we have found no publication
linking RUNX1 or CBFβ to HIV Tat. Our data, however, sug-
gest that the RUNX1 and CBFβ proteins are potentially involved
in HIV infection through Tat-mediated transactivation and are
potentially important for the pathogenesis of AIDS. Inter-
estingly, two recent reports demonstrated that CBFβ also inter-
acts with the HIV protein Vif (36, 37). This interaction was
shown to be required for the degradation of APOBEC3G, a po-
tent inhibitor of viruses including HIV. It will be interesting to
determine whether the effect of RUNX1 and CBFβ on Tat-
mediated transactivation is related to the interaction between
CBFβ and Vif. Equally interesting is to determine whether Ro5-
3335 can disrupt this CBFβ–Vif interaction.
cells on day 21 after transplantation of Cbfb–MYH11 leukemia cells, as measured by FACS. *P < 0.003; **P < 0.0002; ***P < 0.00007, compared with saline-
injected mice. (B) Percentages of c-kit+peripheral blood cells on day 33 after transplantation of Cbfb–MYH11 leukemia cells, as measured by FACS. *P =
0.026, **P = 0.005, compared with the combined treatment group. (C) Wright–Giemsa stained peripheral blood cells in the leukemic mice at 21 d after
transplantation. Ro5, Ro5-3335; Cyt, cytarabine. (D) Graded levels of leukemic infiltration into liver, spleen, and bone marrow. Hepatic leukemic infiltration
was decreased in mice treated with cytarabine (P < 0.02), Ro5-3335 (P < 0.001), and both cytarabine and Ro5-3335 (P < 0.03) compared with animals treated
with saline alone. Ro5-3335–treated mice had less hepatic infiltration than did cytarabine-treated mice (P < 0.05). Splenic infiltration was decreased in Ro5-
3335–treated animals compared with saline alone (P < 0.007). Combined Ro5-3335 and cytarabine treatment significantly reduced leukemic infiltration in the
bone marrow (P < 0.04). (E) Average spleen weight of transplanted mice by treatment group. *Ro5-3335–treated animals had significantly smaller spleens
than saline-treated mice (P < 0.02), whereas the other two treated groups did not. (F) Kaplan–Meier survival curves of the transplanted mice, grouped by their
treatments. (G) Average survival days of the transplanted mice by treatment group. Mice receiving cytarabine (*P < 0.002) or both cytarabine and Ro5-3335
(**P < 0.00001) had significantly increased life expectancy compared with mice treated with saline alone. In addition, mice receiving both cytarabine and Ro5-
3335 significantly increased life span compared with cytarabine alone (#P < 0.04) or Ro5-3335 alone (##P < 0.008).
Ro5-3335 decreases leukemia burden and slows leukemia progression in a murine transplantation model. (A) Percentages of c-kit+peripheral blood
| www.pnas.org/cgi/doi/10.1073/pnas.1200037109Cunningham et al.
Lead compound discovery for a new drug target in industry Download full-text
usually involves a screen of 1–3 million compounds, whereas
we screened only ∼243,000 compounds to identify inhibitors of
the RUNX1–CBFβ interaction. Nevertheless, we were able to
identify Ro5-3335 and Ro24-7429 from the original lead com-
pound, NSC140873, through subsequent modeling and virtual
search of commercially available compounds. Therefore, our ap-
proach demonstrated an alternative path for rapid drug discov-
ery that combines initial screen of a limited compound collection
with extensive informatics analysis and lead search. This ap-
proach may be applied to other rare and neglected diseases, for
which adequate funds are not always available for the routine,
expensive drug discovery process.
Materials and Methods
The tagged proteins (Table S1) were purified in-house or obtained from
Genecopoeia and SAIC-Frederick. The RUNX1–CBFβ AlphaScreen and TR-FRET
assays followed the standard protocols with minor modifications. A total of
243,398 compounds from the NIH Molecular Library Collection (http://mli.nih.
gov/mli/compound-repository/mlsmr-compounds) were used in the primary
screen, and all compounds were serially diluted and screened at seven con-
centrations. Ro5-3335 and Ro24-7429 were provided by the Developmental
Therapeutics Program, National Cancer Institute. All cell lines except for ME-1
(38) were obtained from ATCC. The HIV-1 LTR luciferase and the MCSFR re-
porter assays were performed as described (27, 39). The SPR assay followed
manufacturer’s instructions (BioRad). The cytotoxicity assay with ATPLite was
performed following manufacturer’s instructions (PerkinElmer). The NCI-60
cell line analyses were performed as described (40). Zebrafish embryos at
6–24 h postfertilization (hpf) were treated with compounds in DMSO [<1%
(vol/vol)] for up to 5–6 d when the GFP cells were scored. Transgenic RUNX1–
ETO zebrafish embryos were heat shocked as described (31) and then exposed
to Ro5-3335 for 24 h before evaluation. Leukemic cells from transgenic mice
that carry the CBFβ–SMMHC fusion protein (33) were transplanted by retro-
orbital injection. Ten days after transplantation, the mice were divided into
groups of 5–6 andreceived saline 1 mL (i.p. injection) daily on days10–15 after
transplantation; cytarabine (Bedford Labs) 100 mg/kg/d (i.p. injection) daily
dough on days 10–40 after transplantation, or both cytarabine and Ro5-3335.
Please see SI Materials and Methods for a more complete description.
ACKNOWLEDGMENTS. The authors thank Pingjun Zhu for assistance in the
assay development; Sam Michael for the robotic screen; Paul Shinn for the
compound management; Jingjun Hong and Yawen Bai for advice on
protein purification; Joel Morris and Jerry Collins for compounds; Kevin
Bishop, Trang Jennifer Nguyen, the National Human Genome Research
Institute (NHGRI) mouse, zebrafish, flow cytometry, cytogenetics, and mi-
croscopy cores for technical assistance; Shelly Hoogstraten-Miller and
Irene Ginty for help with mouse experiments and nursing care; Xin Xu
for PK analysis; Urraca L. Tavarez for demonstrating how to knead dough;
Jerrold Ward and Dave Bodine for evaluating histology sections; members
of the P.L. laboratory, Alan Wayne, and David Loeb, for helpful dis-
cussions; and Francis Collins for critical reading of the manuscript. This
work was supported by National Institutes of Health (NIH) Grant X01
MH083259-01 (to P.L.), the Molecular Libraries Initiative of the NIH
Roadmap for Medical Research, the Intramural Research Program of
NHGRI, NIH, and NIH grants CA140188 and AG031300 (to J.-R.J.Y.). L.C.
was a fellow in the joint Johns Hopkins–National Cancer Institute Pediatric
Hematology/Oncology Fellowship program and a recipient of the compet-
itive NIH intramural loan repayment program.
1. Ito Y (2008) RUNX genes in development and cancer: Regulation of viral gene ex-
pression and the discovery of RUNX family genes. Adv Cancer Res 99:33–76.
2. Speck NA, Gilliland DG (2002) Core-binding factors in haematopoiesis and leukaemia.
Nat Rev Cancer 2:502–513.
3. Liu P, et al. (1993) Fusion between transcription factor CBF beta/PEBP2 beta and
a myosin heavy chain in acute myeloid leukemia. Science 261:1041–1044.
4. Liu PP, Hajra A, Wijmenga C, Collins FS (1995) Molecular pathogenesis of the chromo-
5. Erickson P, et al. (1992) Identification of breakpoints in t(8;21) acute myelogenous
leukemia and isolation of a fusion transcript, AML1/ETO, with similarity to Drosophila
segmentation gene, runt. Blood 80:1825–1831.
6. Golub TR, et al. (1995) Fusion of the TEL gene on 12p13 to the AML1 gene on 21q22 in
acute lymphoblastic leukemia. Proc Natl Acad Sci USA 92:4917–4921.
7. Look AT (1997) Oncogenic transcription factors in the human acute leukemias. Sci-
8. Loh ML, et al. (2006) Prospective analysis of TEL/AML1-positive patients treated on
Dana-Farber Cancer Institute Consortium Protocol 95-01. Blood 107:4508–4513.
9. Appelbaum FR, et al. (2006) The clinical spectrum of adult acute myeloid leukaemia
associated with core binding factor translocations. Br J Haematol 135:165–173.
10. von Neuhoff C, et al. (2010) Prognostic impact of specific chromosomal aberrations in
a large group of pediatric patients with acute myeloid leukemia treated uniformly
according to trial AML-BFM 98. J Clin Oncol 28:2682–2689.
11. Roudaia L, et al. (2009) CBFbeta is critical for AML1-ETO and TEL-AML1 activity. Blood
12. Shigesada K, van de Sluis B, Liu PP (2004) Mechanism of leukemogenesis by the inv
(16) chimeric gene CBFB/PEBP2B-MHY11. Oncogene 23:4297–4307.
13. Hyde RK, Liu PP (2010) RUNX1 repression-independent mechanisms of leukemogen-
esis by fusion genes CBFB-MYH11 and AML1-ETO (RUNX1-RUNX1T1). J Cell Biochem
14. Eglen RM, et al. (2008) The use of AlphaScreen technology in HTS: Current status. Curr
Chem Genomics 1:2–10.
15. Degorce F, et al. (2009) HTRF: A technology tailored for drug discovery - a review of
theoretical aspects and recent applications. Curr Chem Genomics 3:22–32.
16. Okuda T, van Deursen J, Hiebert SW, Grosveld G, Downing JR (1996) AML1, the target
of multiple chromosomal translocations in human leukemia, is essential for normal
fetal liver hematopoiesis. Cell 84:321–330.
17. Wang Q, et al. (1996) The CBFbeta subunit is essential for CBFalpha2 (AML1) function
in vivo. Cell 87:697–708.
18. Sood R, et al. (2010) Development of multilineage adult hematopoiesis in the ze-
brafish with a runx1 truncation mutation. Blood 115:2806–2809.
19. Tang YY, et al. (2000) Energetic and functional contribution of residues in the core
binding factor beta (CBFbeta ) subunit to heterodimerization with CBFalpha. J Biol
20. Lin HF, et al. (2005) Analysis of thrombocyte development in CD41-GFP transgenic
zebrafish. Blood 106:3803–3810.
21. Langenau DM, et al. (2004) In vivo tracking of T cell development, ablation, and
engraftment in transgenic zebrafish. Proc Natl Acad Sci USA 101:7369–7374.
22. Kira T, Hashimoto K, Baba M, Okamoto T, Shigeta S (1996) 2-Glycineamide-5-chlor-
ophenyl 2-pyrryl ketone, a non-benzodiazepin Tat antagonist, is effective against
acute and chronic HIV-1 infections in vitro. Antiviral Res 32:55–62.
23. Hsu MC, et al. (1991) Inhibition of HIV replication in acute and chronic infections
in vitro by a Tat antagonist. Science 254:1799–1802.
24. Haubrich RH, et al. (1995) A randomized trial of the activity and safety of Ro 24-7429
(Tat antagonist) versus nucleoside for human immunodeficiency virus infection. The
AIDS Clinical Trials Group 213 Team. J Infect Dis 172:1246–1252.
25. Jin H, et al. (2009) Definitive hematopoietic stem/progenitor cells manifest dis-
tinct differentiation output in the zebrafish VDA and PBI. Development 136:647–654.
26. Rhoades KL, et al. (1996) Synergistic up-regulation of the myeloid-specific promoter
for the macrophage colony-stimulating factor receptor by AML1 and the t(8;21)
fusion protein may contribute to leukemogenesis. Proc Natl Acad Sci USA 93:
27. Kamikubo Y, et al. (2010) Accelerated leukemogenesis by truncated CBF beta-SMMHC
defective in high-affinity binding with RUNX1. Cancer Cell 17:455–468.
28. Holbeck SL, Collins JM, Doroshow JH (2010) Analysis of Food and Drug Administra-
tion-approved anticancer agents in the NCI60 panel of human tumor cell lines. Mol
Cancer Ther 9:1451–1460.
29. Reinhold WC, et al. (2010) Exon array analyses across the NCI-60 reveal potential
regulation of TOP1 by transcription pausing at guanosine quartets in the first intron.
Cancer Res 70:2191–2203.
30. Zoppoli G, et al. (2011) Coordinated regulation of mitochondrial topoisomerase IB
with mitochondrial nuclear encoded genes and MYC. Nucleic Acids Res 39:6620–6632.
31. Yeh JR, et al. (2008) AML1-ETO reprograms hematopoietic cell fate by down-
regulating scl expression. Development 135:401–410.
32. Castilla LH, et al. (1996) Failure of embryonic hematopoiesis and lethal hemorrhages
in mouse embryos heterozygous for a knocked-in leukemia gene CBFB-MYH11. Cell
33. Castilla LH, et al. (1999) The fusion gene Cbfb-MYH11 blocks myeloid differentiation
and predisposes mice to acute myelomonocytic leukaemia. Nat Genet 23:144–146.
34. Kuo YH, et al. (2006) Cbf beta-SMMHC induces distinct abnormal myeloid progenitors
able to develop acute myeloid leukemia. Cancer Cell 9:57–68.
35. Shahabuddin M, Volsky B, Hsu MC, Volsky DJ (1992) Restoration of cell surface CD4
expression in human immunodeficiency virus type 1-infected cells by treatment with
a Tat antagonist. J Virol 66:6802–6805.
36. Jäger S, et al. (2012) Vif hijacks CBF-β to degrade APOBEC3G and promote HIV-1 in-
fection. Nature 481:371–375.
37. Zhang W, Du J, Evans SL, Yu Y, Yu XF (2012) T-cell differentiation factor CBF-β reg-
ulates HIV-1 Vif-mediated evasion of host restriction. Nature 481:376–379.
38. Yanagisawa K, Horiuchi T, Fujita S (1991) Establishment and characterization of a new
human leukemia cell line derived from M4E0. Blood 78:451–457.
39. Yedavalli VS, Jeang KT (2011) Matrin 3 is a co-factor for HIV-1 Rev in regulating post-
transcriptional viral gene expression. Retrovirology 8:61.
40. Gmeiner WH, Reinhold WC, Pommier Y (2010) Genome-wide mRNA and microRNA
profiling of the NCI 60 cell-line screen and comparison of FdUMP with fluoro-
uracil, floxuridine, and topoisomerase 1 poisons. Mol Cancer Ther 9:3105–3114.
41. Zuber J, et al. (2009) Mouse models of human AML accurately predict chemotherapy
response. Genes Dev 23:877–889.
Cunningham et al. PNAS
| September 4, 2012
| vol. 109
| no. 36