INSIGHTS | PERSPECTIVES
with preexisting phenotypic differences.
Mu et al. used RNA-sequencing data sets
of clinical samples to identify transcription
factors whose expression correlated with
co-occurring alterations in the RB1 and
TP53 signaling pathways. Among these,
SOX2 expression was induced by silencing
RB1 and TP53 in LNCaP-AR cells, and SOX2
expression in these cells was necessary and
sufficient for the expression of basal epithelial and neuroendocrine markers as well as
for enzalutamide resistance. RB1 loss has
been associated with lineage plasticity in
other cancer cell models, and in induced
pluripotent stem cells, RB1 inhibited the
basal transcription of SOX2 and other pluripotency genes ( 8). These data suggest a
model whereby the induction of SOX2 expression subsequent to RB1 and TP53 loss
contributes to neuroendocrine differentiation and AR-pathway independence.
Ku et al. examined the effects of Rb1 and
Tp53 deletion in a mouse model of meta-
static prostate cancer that is based on the
lack of phosphatase and tensin homolog
(Pten), another tumor suppressor. Prostate-
specific deletion of Pten led to adenocarci-
noma (expressing AR) as expected, whereas
the simultaneous deletion of Rb1 produced
an aggressive metastatic disease that was
initially an adenocarcinoma but progressed
to heterogeneous histology with markers
of adenocarcinoma and neuroendocrine
disease, including variable AR expression.
Most tumors responded transiently to cas-
tration (a reduction in androgen produc-
tion), and two of five recurrent tumors
that were analyzed contained Tp53 muta-
tions, suggesting a role for Tp53 loss in cas-
trate resistance. Gene expression analysis
showed marked overlap between the dif-
ferentially expressed genes in mice lacking
Pten and Rb1 and mice lacking Pten, Rb1,
E2F (including Sox2 and Ezh2), neuroen-
docrine lineage genes, and genes related to
stem cells and epigenetic reprogramming.
Ezh2 inhibition or silencing increased AR
and decreased Syp expression in cell lines
derived from mouse prostate tumors lack-
ing Pten and Rb1. These cell lines also
became sensitive to inhibition by enzalu-
tamide. Similarly, Ezh2 inhibition restored
enzalutamide responsiveness in the RB1-
and TP53-depleted LNCaP-AR cells in the
study by Mu et al.
The studies by Mu et al. and Ku et al.
indicate that RB1 loss not only facilitates
the outgrowth of neuroendocrine-like
cells that have decreased AR signaling but
also facilitates lineage reprogramming
by stimulating expression of SOX2. The
studies also suggest a role for TP53 loss,
although RB1 loss alone appears sufficient
to drive neuroendocrine differentiation in
the mouse model, perhaps related to the
presence of less-differentiated luminal epi-
thelial cells. Further studies in additional
metastatic prostate cancer models will be
important to assess the generalizability of
these results. Furthermore, RB1 loss may
contribute other functional effects. RB1
loss alone can confer resistance to AR-
targeted therapies ( 9). The extent to which
TP53 or RB1 loss directly contributes to
regulating AR expression and/or activity
observed in neuroendocrine prostate can-
cer is also unclear. Indeed, RB1 loss may
directly enhance AR expression and signal-
ing ( 10). Nonetheless, results in the mouse
model indicate that Rb1 loss does not di-
rectly suppress AR and instead suggest
that there may be a selective advantage to
decreasing AR expression once cell growth
becomes AR independent.
SOX2, an E2F-regulated gene, was identified as a downstream mediator of RB1-
dependent lineage plasticity. SOX2 plays
context-dependent roles in both maintaining pluripotency and driving neuronal progenitor differentiation ( 11). It will
be important to determine the various
roles for SOX2 and to what extent in vitro modeling can be generalized to disease
progression in vivo. Moreover, the effects
of EZH2 inhibition are a key finding and
parallel other studies showing the importance of EZH2-dependent mechanisms in
supporting CRPC-NE growth, including
models initiated by other drivers ( 12). Notably, EZH2 is also a direct AR coactivator
( 13). This suggests that EZH2 inhibition in
combination with other therapeutics, including enzalutamide, has the potential to
increase responses in patients with CRPC-NE disease. j
1. P. Mu et al ., Science355, 84 (2017).
2. S. Y. Ku et al., Science 355, 78 (2017).
3. P.A. Watson, V.K.Arora, C.L.Sawyers, Nat. Rev. Cancer15,
4. A.Aparicio,C.J.Logothetis, S.N.Maity, Cancer Discov. 1,
5. J. H. E. Small et al ., J. Clin. Oncol. 33 (suppl.), abstr. 5003
6. H. Beltran et al. , Nat. Med. 22, 298 (2016).
7. J. L. Bishop et al. , Cancer Discov. 7, 1 (2016).
8. M. S. Kareta et al., Cell Stem Cell 16, 39 (2015).
9. A.Kumaretal.,Nat.Med. 22,369(2016).
10. A.Sharma et al., J. Clin. Invest.120,4478(2010).
11. A. Sarkar, K. Hochedlinger, Cell Stem Cell 12, 15 (2013).
12. E. Dardenne et al., Cancer Cell 30, 563 (2016).
13. K. Xu et al. , Science 338, 1465 (2012).
After treatment with an AR antagonist, cells
with altered RB1 and TP53 are selected.
Factors including SOX2 and EZH2 contribute
to dediferentiation and plasticity.
Cells established are most often
reprogrammed to the neuroendocrine
lineage that is resistant to
In castrate-resistant prostate cancer
(luminal epithelial adenocarcinoma),
cells express and depend upon
androgen receptor (AR+) for growth.
and TP53 cell
Efective AR blockade
A model of progressive reprogramming