Nature | Vol 578 | 13 February 2020 | 309Notably, AR acetylation at the K 630 LKK 633 motif that resembles BET
bromodomain-binding sites in histones has been shown to be important
for AR activity^29. Considering that the N-terminal domain of AR has been
shown to bind to BD1 directly^20 , we speculated that acetylated AR may
interact cooperatively with both BD1 and BD2 of BRD4 at AR–BRD4
co-occupied super-enhancers to regulate a subset of AR-dependent
genes that are therefore sensitive to BD2 inhibition (Extended Data
Fig. 6d, e). In ABBV-744-resistant 22RV1 cells, in which AR-dependent
transcription is driven by AR-V7 (which lacks the K 630 LKK 633 motif^30 ),
ABBV-744 failed to inhibit the AR gene signature, induced limited BRD4
displacement from super-enhancers, and produced weak effects on
proliferation and senescence, collectively supporting the putative
interaction of acetylated AR with BD2 to induce sensitivity to ABBV-744
(Extended Data Fig. 7a–f ). More mechanistic studies will be required
to confirm this hypothesis.
The drug-like properties of ABBV-744 enabled the investigation of
its antitumour efficacy and tolerability. In a mouse xenograft model
using LNCaP cells, treatment with 4.7 mg kg−1 ABBV-744 (1/16 of the
maximum tolerated dose (MTD)) caused a delay in tumour growth
that was equivalent to ABBV-075 treatment at the MTD dose of 1 mg kg−1
(Fig. 4a). Comparing efficacious exposure levels of ABBV-744 in LNCaP
tumour-bearing mice (4.7 mg kg−1; area under the curve, 1.1 μg h ml−1)
and MTD (75 mg kg−1; area under the curve, 13.1 μg h ml−1) demonstrated
that ABBV-744 was able to produce significant antitumour activity at
1/12 of the highest tolerable exposure of ABBV-744 (Extended Data
Fig. 8a). The activity exhibited by ABBV-744 at 1/16 of the MTD of ABBV-
744 was superior to the activities achieved using JQ1 and iBET at their
respective MTDs or, in the case of RVX-208, at the highest feasible dose
in this model (Extended Data Fig. 8b, c). Similarly, ABBV-744 at 1/16 MTD
also displayed equivalent or better antitumour activity compared with
ABBV-075 at MTD in the enzalutamide-resistant MDA-PCa-2b xenograft
model (Fig. 4b). As a control, lowering the dose of ABBV-075 to 1/2 of the
MTD resulted in a significant reduction in antitumour activity to 42%
tumour growth inhibition in the LNCaP xenograft model. Even in the
xenograft model using OPM2 cells, one of the most sensitive models to
DbBi, ABBV-075 at 1/4 of the MTD of ABBV-075 (0.25 mg kg−1) had only
marginal antitumour efficacy (Extended Data Fig. 8d, e).
In toxicity studies in rats, ABBV-075 at 3 mg kg−1 (3× the efficacious
exposure in the LNCaP mouse xenograft model), caused a 59% reduction
in platelets, a decrease in Alcian blue staining of the mucosa and the
loss of goblet cells. By contrast, ABBV-744 at 30 mg kg−1 (25× the effica-
cious exposure) triggered a reduction in platelets of only 20%, and atde
fgBRD4 ChIP–qPCR KLK2 RT–qPCR20 kb hg19+DHT+DHT/ENZ+DHT/ABBV-075+DHT/ABBV-744+DHT+DHT ARH3K27AcBRD4KLK1526627369696969KLK3 KLK2KLK2 KLKP1 KLK4H3KBKBRD4BRD4BRD4AR, non-SESE OtherRegion 1 Region 2(^20) Region 1Region 2
15
10
5
0
10
8
6
4
0
2
15
10
5
+–++++^0 +–++++ –++++
Fold change (IgG) Relative expression
IgGDMSOENZ
ABBV-07
5
ABBV
-744 IgGDMSOENZ
ABBV-07
5
ABBV
-744 DMSOENZ
ABBV-07
5
ABBV
-744
DHT
DMSO
ABBV-075
ABBV-744
Non-AR SE
0.5 0.8
AR-bound SE
0.8
BRD4
–10 kb 10 –10 kb 10 –10 kb 10
SE Other
log
FC in expression 2
log
FC in expression 2
2
1
0
–1
–2
2
1
0
–1
–2
P = 1 × 10 –6 P = 1 × 10 –9
c ABBV-744 ABBV-075
0404030303
a
Vehicle Vehicle VehicleABBV-075ABBV-744
H3K27Ac AR BRD4 H3K27Ac AR BRD4
Transcription start sites
Vehicle Vehicle VehicleABBV-075ABBV-744
b Enhancers
0303030303
4.0 2.0 2.0
–10kb
kb kb kb kb
10 –10kb 10 –10kb 10 –10kb 10 –10kb 10
–10kb 10 –10 10 –10 10 –10 10 –10 10
3.5 2.0 2.0
Fig. 3 | ABBV-744 displaces BRD4 from AR-containing super-enhancers.
LNCaP cells were incubated with 5 nM DHT and vehicle (DMSO), 60 nM
ABBV-075 or 90 nM ABBV-744 for 6 h, and cells were collected for ChIP–seq to
determine H3K27Ac, AR and BRD4 chromatin association. a, b, Rank-ordered
heat maps of H3K27Ac, AR and BRD4 peaks at transcription start sites or
enhancers after the indicated treatment. Rows are ordered according to the
vehicle-treated BRD4 maximum for each region and centred ±10 kb of the
BRD4 peak after treatment with vehicle. Colour scales depict reads per million
(RPM) intensities. Bottom profile plots display log 2 -transformed fold change in
RPM/bp compared with control. BRD4 ChIP experiments were normalized to
spike-in controls. c, Quantification of log 2 -transformed fold change in
expression after ABBV-075 or ABBV-744 treatment for genes associated with
super-enhancers (SE) or non-super-enhancers (other). For all box plots, centre
line indicates the median; box limits are the first and third quartiles; whiskers
range from the first quartile minus 1.5× the interquartile range to the third
quartile plus 1.5× the interquartile range. Unpaired two-tailed Student’s t-test
was used to determine significance for super-enhancers versus other; n = 2.
d, BRD4 profile plots at AR-bound regions that are not located in super-
enhancers (AR, non-SE), AR-bound super-enhancers (AR, SE), or super-
enhancers without AR binding (non-AR SE). e, Gene track of H3K27Ac, AR, and
BRD4 ChIP–seq signals for the indicated treatment conditions at a super-
enhancer that is associated with several AR-dependent genes. f, LNCaP cells
that underwent the indicated treatments for 24 h were collected for ChIP–qPCR
to determine the binding of BRD4 to the indicated regions in the gene track.
g, KLK2 expression in LNCaP cells that underwent the indicated treatments for
24 h was determined by qPCR. f, g, Data are mean ± s.d. (n = 3 biologically
independent samples) and are representative of n > 2 independent
experiments.