Sprouty genes function in suppression of prostate tumorigenesis Jennifer L. Schutzmana and Gail R. Martinb,1 Departments of aMedicine and bAnatomy, University of California, San Francisco, CA 94158 Contributed by Gail R. Martin, October 8, 2012 (sent for review August 13, 2012)

P

rostate cancer is the most commonly diagnosed malignancy in men, with more than 240,000 new diagnoses anticipated in 2012. Despite advances in both diagnosis and treatment, ∼28,000 prostate cancer deaths are expected in the United States in 2012 (1). Although androgen deprivation remains a cornerstone of therapy, inevitably the disease becomes resistant to interventions aimed at depleting androgens and treatments are limited for this lethal form of the disease. A better understanding of the genetic and molecular basis for prostate cancer may lead to more effective therapies and prevention strategies. The tumor suppressor gene PTEN, which encodes a key negative regulator of PI3 kinase (PI3K)/AKT signaling, is frequently deleted or mutated in human prostate cancer, resulting in increased AKT activity (2, 3). The importance of PTEN in prostate cancer has been demonstrated in the mouse, where loss of Pten function can recapitulate the progression of the human disease from prostatic intraepithelial neoplasia (PIN) to invasive cancer (4, 5). In addition to increased PI3K/AKT activity, a common feature of aggressive human prostate cancers is hyperactivation of the RAS/ ERK1/2 pathway (6). However, oncogenic mutations in RAS or RAF are rare and the molecular basis for aberrant ERK1/2 activity in human prostate cancer is unknown (7, 8). One possible mechanism for RAS/ERK1/2 pathway hyperactivation is loss-of-function of key negative regulators of the pathway such as the Sprouty genes, which encode intracellular antagonists of receptor tyrosine kinase (RTK) signaling. There are four Sprouty genes in mammals. In the mouse, Sprouty1 (Spry1) and Sprouty2 (Spry2) have been shown to play roles in modulating the spatial and temporal execution of potent signaling cascades that regulate gene expression and cell behaviors

www.pnas.org/cgi/doi/10.1073/pnas.1217204109

such as proliferation, survival, migration, and fate decisions in the embryo (9). The precise mechanism(s) by which Sprouty proteins antagonize RTK signaling remains unclear. Most data indicate that they function to negatively regulate RAS→RAF→MEK→ERK1/2 signaling, but there is limited evidence that they can also affect PI3K/AKT signaling (9, 10). Both of these pathways are commonly deregulated in human prostate cancer (11). Loss of Sprouty gene function has been implicated in human cancers (12). Expression of SPRY1 or SPRY2 is decreased or absent in both primary (15% and 18%, respectively) and metastatic (42% and 74%, respectively) human prostate cancers (11), and both SPRY1 and SPRY2 genes are concomitantly down-regulated in a subset of human prostate cancers (13). These observations suggest that SPRY1 and SPRY2 may function as tumor suppressors in human prostate cancer, consistent with the results of studies showing that Sprouty gene expression can influence tumor development in the mouse lung (14, 15) and liver (16). To investigate their role in prostate tumor suppression, we produced both mice in which Spry1 and Spry2 were inactivated in prostate epithelium that was either wild-type or heterozygous for Pten and mice in which Spry2 was overexpressed in prostate epithelium lacking PTEN. The results of this analysis provide strong evidence that Sprouty genes can function as suppressors of prostate tumorigenesis. Results Loss-of-Function of Spry1 and Spry2 Promotes PIN. Spry1 and Spry2

are expressed in the developing and adult mouse prostate epithelium (Fig. 1 A and B). To explore the possibility that they function as suppressors of tumorigenesis in the prostate, we sought to determine the long-term effects of eliminating them in the mouse prostate epithelium from a very early stage of development. To achieve this, we used the Osr1-cre transgene, which expresses Cre recombinase throughout the embryonic posterior endoderm that gives rise to the prostate epithelium (17), to simultaneously inactivate conditional null alleles of Spry1 and Spry2 (Spry1flox and Spry2flox) (18, 19) and then assessed the effects in adult animals. The cross we performed to produce the animals used in our analysis (Materials and Methods) yielded Osr1-cre;Spry1flox/Δ;Spry2 flox/Δ and Osr1-cre;Spry1flox/+;Spry2flox/+ animals, hereafter referred to as PES1/2DKO (prostate epithelium double knock-out) and PE-S1/ 2Dhet (prostate epithelium double heterozygote) mice, respectively. Importantly, we found that PE-S1/2Dhet prostates were indistinguishable from wild type (FvB/N), and therefore PE-S1/ 2Dhets were used as controls. To confirm that Osr1-cre was indeed functioning as expected to efficiently recombine the Spry1 and Spry2 floxed alleles in the prospective prostate epithelium and determine the extent to which they were recombined in PE-S1/2DKO mutant prostate epithelium, we used laser capture microdissection. We obtained prostate epithelium from hematoxylin and eosin (H&E)–stained paraffinembedded prostate tissue sections at 14 mo of age (Fig. 1 C–E),

Author contributions: J.L.S. and G.R.M. designed research; J.L.S. performed research; J.L.S. and G.R.M. analyzed data; and J.L.S. and G.R.M. wrote the paper. The authors declare no conflict of interest. 1

To whom correspondence should be addressed. E-mail: [email protected].

PNAS | December 4, 2012 | vol. 109 | no. 49 | 20023–20028

CELL BIOLOGY

Expression of Sprouty genes is frequently decreased or absent in human prostate cancer, implicating them as suppressors of tumorigenesis. Here we show they function in prostate tumor suppression in the mouse. Concomitant inactivation of Spry1 and Spry2 in prostate epithelium causes ductal hyperplasia and low-grade prostatic intraepithelial neoplasia (PIN). However, when Spry1 and Spry2 loss-of-function occurs in the context of heterozygosity for a null allele of the tumor suppressor gene Pten, there is a striking increase in PIN and evidence of neoplastic invasion. Conversely, expression of a Spry2 gain-of-function transgene in Pten null prostatic epithelium suppresses the tumorigenic effects of loss of Pten function. We show that Sprouty gene loss-of-function results in hyperactive RAS/ERK1/ 2 signaling throughout the prostate epithelium and cooperates with heterozygosity for a Pten null allele to promote hyperactive PI3K/AKT signaling. Furthermore, Spry2 gain-of-function can suppress hyperactivation of AKT caused by the absence of PTEN. Together, these results point to a key genetic interaction between Sprouty genes and Pten in prostate tumorigenesis and provide strong evidence that Sprouty genes can function to modulate signaling via the RAS/ERK1/2 and PI3K/AKT pathways. The finding that Sprouty genes suppress tumorigenesis caused by Pten loss-of-function suggests that therapeutic approaches aimed at restoring normal feedback mechanisms triggered by receptor tyrosine kinase signaling, including Sprouty gene expression, may provide an effective strategy to delay or prevent high-grade PIN and invasive prostate cancer.

PE-S1/2

wild-type Dhet

Spry2

Spry1

14 mo

DKO

sm low grade PIN

ep normal

A

B G

PE-S1/2DKO

% unrecombined relative to control

H

after tissue catapult

D

C

F

after laser cut

E

100

qPCR Spry1 Spry2

80 60

normal

pMEK

ECAD, DAPI

before

I

J pAKT sm

40 sm

20 0 control

(+ / )

/

Osr1-cre; Spry1 flox/ ; Spry2 flox/

K

L

Fig. 1. Loss of Spry1 and Spry2 function causes ductal hyperplasia and lowgrade PIN. (A–B) Paraffin sections of wild-type adult prostates hybridized with probes for Spry1 (A) and Spry2 (B). (C–E) A ventral prostate section from a PE-S1/ 2DKO animal stained with H&E before laser capture microdissection (C), after laser cutting to separate the epithelium from the underlying smooth muscle layer (D; cut region indicated by dashed oval), and after the epithelium was catapulted and collected (E; the region where the tissue has been removed is indicated by red dots). (F) Quantitative PCR analysis of the extent of Spry1 and Spry2 flox allele recombination in PE-S1/2DKO prostate epithelium. Bar graph shows the amount of unrecombined Spry1 or Spry2 flox allele remaining after Cre-mediated recombination in PE-S1/2DKO DNA (which contained only one unrecombined copy of each flox allele before Cre-mediated recombination) as a percent of the amount of unrecombined allele in Spry1 or Spry2 heterozygote (+/Δ) DNA. If there were no recombination, the expected result would be 100% unrecombined, whereas if every flox allele were recombined, the expected result would be 0%; the results obtained with genomic DNA from Spry1 or Spry2 null (Δ/Δ) animals is shown for comparison. (G–L) Ventral prostate sections from animals of the genotypes indicated at 14 mo were stained with H&E (G–H) or immunostained (green) for phospho-MEK1/2 (pMEK) (I–J) or phospho-AKT at Ser473 (pAKT) (K–L); all sections were also immunostained (red) for ECAD and stained with DAPI (blue) (I–L). Note that pAKT is not observed in the prostate epithelium, although it is detected in the smooth muscle surrounding the ducts (K–L). ep, epithelium; sm, smooth muscle.

isolated genomic DNA, and performed quantitative PCR assays (n = 6 samples). We found that, compared with a heterozygous control that did not carry Osr1-cre, the percent of Spry1flox and Spry2flox alleles that remained unrecombined in the DKO prostate epithelium was 10.5–12.5% and 17.5–25.5%, respectively (Fig. 1F). Thus, even if the unrecombined Spry1 and Spry2 floxed alleles were present in entirely nonoverlapping cell populations, the majority of the prostate epithelium assayed was null for both Spry1 and Spry2. Interestingly, despite nearly complete inactivation of both Spry1 and Spry2 throughout the prostate epithelium in PE-S1/2DKO mutants, the overall ductal architecture was preserved and few abnormalities were observed histologically at 8 mo of age. Multiple small focal areas of ductal hyperplasia with increased epithelial cell number and stratification without dysplastic features were identified (n = 3). In a detailed histopathological analysis of prostates at 14 mo of age, ductal hyperplasia was present in all DKO mutant prostates examined (n = 9) and was more frequent and more extensive than at 8 mo of age. In addition, ducts displaying multilayered epithelium with atypical cells and nuclear pleomorphism typical of low-grade 20024 | www.pnas.org/cgi/doi/10.1073/pnas.1217204109

PIN were occasionally observed. In contrast, Dhet control prostates at both 8 and 14 mo of age (n = 2 and n = 9, respectively) showed little or no evidence of ductal hyperplasia and no low-grade PIN (Fig. 1 G and H). In many different cellular contexts, loss of Sprouty gene function results in hyperactive RAS/ERK1/2 signaling in response to growth factors (9). Consistent with those observations, we detected a high level of the phosphorylated (activated) form of MEK1/2 (pMEK) throughout the prostate epithelium in PE-S1/2DKO mutants compared with a low level of pMEK in control PE-S1/2Dhet prostate epithelium (Fig. 1 I and J). The level of pMEK detected in histologically normal prostate epithelium in DKO mutants was ∼fivefold higher than in Dhet controls. In addition to modulating the RAS/ERK1/2 pathway, Sprouty genes can negatively regulate the PI3K/AKT pathway in specific cellular contexts (20). To investigate whether Spry1 and Spry2 loss-of-function led to deregulated AKT signaling, prostate sections from PE-S1/2DKO and Dhet animals were immunostained for phosphorylated (activated) AKT (pAKT) at threonine 308 (Thr308) or serine 473 (Ser473). We did not detect an increase in pAKT in PE-S1/2DKO mutants compared with Dhet or FvBN control prostates, in which there is a low level or no pAKT detectable in the epithelium (Fig. 1 K and L). Thus, loss of Spry1 and Spry2 function resulted in a marked increase in pMEK throughout the adult mouse prostate epithelium but had no effect on pAKT. Together, these data show that Spry1 and Spry2 are necessary in the prostate epithelium to suppress ductal hyperplasia and low-grade PIN, likely by preventing hyperactivation of the RAS/ ERK1/2 signaling pathway. However, loss of Sprouty function alone is not sufficient to drive significant tumorigenesis. Loss of Spry1 and Spry2 Function Promotes PIN in Prostate Epithelium Heterozygous for the Tumor Suppressor Gene Pten. PTEN is a critical

tumor suppressor gene in the human prostate. Complete Pten loss-of-function in the mouse prostate epithelium invariably causes extensive high-grade PIN that progresses to invasive cancer over a period of several months (4, 5). In contrast, heterozygosity for a Pten null allele causes, at most, low-grade PIN with incomplete penetrance beginning as early as ∼6 mo of age, although this varies greatly with genetic background (21, 22). Pten heterozygosity has therefore been used as a sensitized genetic background in which to look for interactions with other putative tumor suppressor genes (23). Here, we sought to determine whether loss of Spry1 and Spry2 function would exacerbate the effects of Pten heterozygosity. The cross we performed to produce the animals used in our analysis (Materials and Methods) yielded Osr1-cre;Ptenflox/+ ;Spry1flox/Δ; Spry2flox/Δ and Osr1-cre;Ptenflox/+;Spry1flox/+;Spry2flox/+ male mice, hereafter referred to as PE-Pten het;S1/2DKO mutants and PEPten het;S1/2Dhet controls, respectively. We examined prostates beginning at 6 mo of age from littermates on a mixed genetic background (Fig. 2 A and B). All mutant prostates examined (n = 4) displayed multiple, widely spaced areas with low-grade PIN, and in two out of four prostates there were occasional ducts with high-grade PIN. In the control prostates examined (n = 3), two out of three displayed no evidence of PIN at this age, but in one out of three there was occasional low-grade PIN. This difference indicates that the emergence of the low-grade PIN phenotype associated with Pten heterozygosity is accelerated by loss of Spry1 and Spry2 function. A detailed histopathological analysis of prostates collected from mutant and control animals at 12–14 mo was performed (n = 10 per genotype). In PE-Pten het;S1/2DKO mutant prostates, ∼50% of ducts displayed low-grade PIN and ∼15% of ducts displayed highgrade PIN (Fig. 2 C–E). In fact, the extent of high-grade PIN in the mutant prostates was almost certainly much higher, as these ducts were extremely large and likely represent fusions of multiple ducts. Ductal lumens were filled with atypical and dysplastic cells that distorted the overall ductal architecture and were associated with areas of necrosis and abnormal intraepithelial vessels typical of high-grade PIN (Fig. 2D). In addition, evidence of invasion, inSchutzman and Martin

PE-Pten het; S1/2 6 mo

Dhet

DKO

low grade PIN normal

A

B

normal

14 mo

Dhet

DKO

high grade PIN

invasion

hyperplasia

normal

D

100 PE-Pten het; S1/2

% ducts

80 60

total # ducts

Dhet

2068

DKO

3540

40

Overexpression of Spry2 Suppresses PIN Caused by Pten Loss-ofFunction. To further investigate the genetic interaction between

20 0 normal

E

n = 1922 1429

low grade PIN 137 1632

high grade PIN 9 479

Fig. 2. Loss of Spry1 and Spry2 function promotes development of highgrade PIN and invasion in prostate epithelium heterozygous for Pten. (A–D) H&E-stained sections of ventral prostate from 6- and 14-mo-old animals of the genotypes indicated showing representative areas of normal prostate epithelium (A–C), ductal hyperplasia (C), and low-grade (B) and high-grade PIN and invasion (D). An intraepithelial blood vessel and a region of epithelium containing necrotic cells, features of high-grade PIN, are indicated by a dashed line and white arrowhead, respectively (D, Left). The presence of epithelial cells in the mesenchyme demonstrates invasion (arrow in D, Right). (E) Bar graph showing the percent normal ducts and ducts displaying lowgrade or high-grade PIN in PE-Pten het;S1/2DKO (n = 3) and Dhet (n = 2) prostates at 14 mo. The distribution of ducts differed significantly between PE-Pten het;S1/2DKO and Dhet prostates (P < 0.0001, χ2 test).

cluding clusters of epithelial cells in the mesenchyme (Fig. 2D) and vasculature and loss of the smooth muscle around ducts, was found associated with high-grade PIN in mutant prostates. In contrast, in PE-Pten het;S1/2Dhet control prostates, the majority of ducts were histologically normal and only 5–10% of ducts displayed PIN. The extensive high-grade PIN found in PE-Pten het;S1/2DKO prostates is in striking contrast to the absence of high-grade PIN in the oldest PE-S1/2DKO prostates examined (Fig. 1H). Together, these data demonstrate that in the context of Pten heterozygosity, loss of Spry1 and Spry2 function accelerates the development of PIN and promotes more extensive high-grade PIN along with the transition to invasive cancer, and thus, Pten, Spry1, and Spry2 cooperate to suppress tumorigenesis in the prostate epithelium. Loss of Spry1 and Spry2 Function Cooperates with Heterozygosity for Pten to Promote AKT Activation. Given these effects on tumorigen-

esis, we next examined how loss of Sprouty function in epithelium heterozygous for Pten affected RAS/ERK1/2 and PI3K/AKT Schutzman and Martin

Pten and Sprouty genes in prostate tumor suppression, we used a conditional Spry2-GOF (gain-of-function) transgene (24) in a mouse model of human prostate cancer based on homozygosity for a Pten null allele in the prostate epithelium (5). After Cremediated recombination, the Spry2-GOF transgene expresses a low level of Spry2 RNA (24). Importantly, this level of Spry2 overexpression does not result in any gross prostate abnormalities and Osr1-cre; Spry2-GOF males survive for at least 6 mo. For this analysis, we produced Osr1-cre; Pten flox/flox male mice without and with Spry2-GOF, hereafter referred to as PE-Pten KO and PEPten KO;S2GOF mice. Prostates from these animals were used to determine the frequency at which PIN occurred following Osr1cre-mediated loss of Pten function in the mixed genetic background we used and the extent to which Spry2 overexpression can suppress prostate tumorigenesis. Prostates were examined at 2 and 4 wk and 9 mo of age, and the number of ducts histologically normal versus those displaying PIN was quantified (Fig. 4K). In PE-Pten KO prostates, PIN was detected in 31% of ducts as early as 2 wk (Fig. 4A). This is earlier than has previously been reported in prostate-specific Pten KO mice (4, 5), most likely because the Osr1-cre transgene we used functions in the embryo, whereas the PB-Cre and PB-Cre4 transgenes used by others function only after birth (17, 25, 26). In mutants carrying the Spry2-GOF transgene at this age, twofold fewer ducts (15%) displayed PIN (Fig. 4B). By 4 wk, the frequency of affected ducts was 66% in PE-Pten KO and 31% in PE-Pten KO; S2GOF prostates (Fig. 4 C and D). At 9 mo, 79% of ducts in PEPten KO mice displayed extensive high-grade PIN, but the frequency of PIN was only 42% in the PE-Pten KO;S2GOF prostates, and those ducts that did display PIN were less severely affected than in PE-Pten KO prostates (Fig. 4 E and F). Together, these findings indicate that Spry2 overexpression has an early suppressive effect on prostate tumorigenesis caused by Pten inactivation and that this effect persists for many months. It is well established that loss of Pten function invariably results in hyperactive PI3K/AKT signaling and in turn this causes prostate tumorigenesis (4, 5). As expected, immunostaining of serial sections of PE-Pten KO prostates demonstrated widespread absence PNAS | December 4, 2012 | vol. 109 | no. 49 | 20025

CELL BIOLOGY

C

signaling. Similar to what was observed in PE-S1/2DKO prostates (Fig. 1 I and J), we found that pMEK was increased throughout the PE-Pten het;S1/2DKO mutant prostate epithelium (Fig. 3 A–C). To assess PI3K/AKT signaling, we immunostained for pAKT-Thr308 and pAKT-Ser473. In PE-Pten het;S1/2DKO prostates, pAKT was detected at a high level in ducts affected by PIN (Fig. 3 D–F). However, in contrast to what was observed for pMEK, pAKT was not detected in the normal epithelium (Fig. 3F, compare with Fig. 3C), despite the reduction in Pten gene dosage. These data demonstrate that loss of Spry1 and Spry2 function can result in aberrant activation of AKT specifically in the context of Pten heterozygosity. Moreover, the observation that a high level of pAKT was detected only in ducts with PIN suggests that it is the activation of PI3K/AKT signaling, either alone or in combination with hyperactive RAS/ ERK1/2 signaling, that causes the PIN. However, PIN is known to result from hyperactive PI3K/AKT signaling caused solely by a total loss of Pten function (4, 5). To rule out the possibility that the high level of pAKT we observed in ducts affected by PIN in PE-Pten het;S1/2DKO prostates was due to a loss of the wild-type allele of Pten, we immunostained prostate sections for PTEN. In both PE-Pten het;S1/2DKO and Dhet prostates, robust PTEN expression was detected throughout the epithelium (Fig. 3 G and H). Thus, loss of heterozygosity of Pten is not the explanation for the high-grade PIN found in PE-Pten het; S1/2DKO prostates, and the high level of pAKT we detected in ducts displaying high-grade PIN occurred despite the presence of one copy of Pten, which encodes its negative regulator.

of PTEN in the epithelium, which was correlated with a high level of pAKT and high-grade PIN (Fig. 4 G and I). In contrast, in PEPten KO;S2GOF prostates, we found that in epithelium lacking PTEN, pAKT was not detected, and moreover the ducts that were PTEN- and pAKT-negative appeared histologically normal (Fig. 4 H and J). These data provide strong evidence that Sprouty overexpression inhibits AKT hyperactivation caused by loss of Pten, thereby suppressing tumorigenesis. Discussion Spry1 and Spry2 gene expression is frequently absent or significantly decreased in human prostate cancers (11, 13, 27, 28), raising the possibility that Sprouty genes normally function as tumor suppressors by negatively regulating signal transduction that must be kept in check to prevent tumorigenesis. To rigorously test this hypothesis, we used a genetic approach using the Osr1-cre transgene to conditionally delete both Spry1 and Spry2 in the mouse prostate epithelium. Despite extensive Cre-mediated inactivation of both Sprouty genes throughout the prostate epithelium and a consequent widespread increase in RAS/ERK1/2 activation, only modest effects, including frequent ductal hyperplasia and occasional low-grade PIN (Fig. 1), were detected. Thus, loss of both Spry1 and Spry2 function in the prostate epithelium is not sufficient to cause significant tumorigenesis. Similar findings have been reported for mouse models based on expression of an oncogenic form of Kras in the mouse prostate epithelium, which, like loss of 20026 | www.pnas.org/cgi/doi/10.1073/pnas.1217204109

PE-Pten KO

PE-Pten KO

PE-Pten KO; S2GOF

2 wks

normal

9 mo

PTEN

a

w

gr

PE-Pten KO; S2GOF

normal

IN

P de

lo

A

ECAD, DAPI

Fig. 3. Hyperactivation of RTK signaling pathways is caused by loss of Spry1 and Spry2 function in prostate epithelium heterozygous for Pten. (A–H) For each genotype indicated, sections of the ventral prostate from animals at 14 mo of age were immunostained (green) for pMEK (A–C), pAKT-Ser473 (D–F), or PTEN (G–H); all sections were also immunostained (red) for ECAD and stained with DAPI (blue). Note that a high level of pMEK, indicating hyperactive RAS/ERK1/2 signaling, is observed throughout the prostate epithelium when Spry1 and Spry2 have been deleted (B–C); an area of normal prostate epithelium is indicated with a white arrow adjacent to an area of PIN (C). A high level of pAKT is observed in the PE-Pten het;S1/2DKO prostate where PIN is found (E and F) but not in normal epithelium (F, white arrow). In panel H, which shows a serial section adjacent to the one shown in panel E, strong immunostaining for PTEN is observed throughout the epithelium, demonstrating that the pAKT and high-grade PIN found in PE-Pten het;S1/2DKO prostates is not due to loss of Pten heterozygosity.

B 4 wks

G

IN

eP

ad

w lo

gr

H

sm

pAKT

H

normal

I 100

C

D

J PE-Pten KO no S2GOF

80

9 mo high grade PIN

% ducts with PIN

G

N

F

PI

E

e

D

ad

C

Sprouty gene function, causes an increase in RAS/ERK1/2 activation (29, 30). In striking contrast, when Spry1 and Spry2 were concomitantly deleted along with one copy of the potent tumor suppressor gene Pten, we observed widespread PIN, frequently of high grade, as well as evidence of invasion. Tumorigenesis was not due solely to the decrease in Pten dosage, as it was not observed in PE-Pten het; S1/2Dhet control animals (Fig. 2). Tumorigenesis was also not due to loss of heterozygosity or silencing of Pten, as we detected a high level of PTEN throughout both PE-Pten het;S1/2DKO and Dhet prostates (Fig. 3 G and H). Interestingly, our mice, in which Pten was heterozygous in the prostate epithelium, survived to at least 12–14 mo of age, whereas mice with complete Pten loss-of-function combined with oncogenic Kras (29) or Braf (31) expression in the prostate died at a much earlier age. This decreased viability is likely due to the development of aggressive disease caused by lack of PTEN. However, distinct effects caused by oncogenic Kras or Braf expression compared with loss of Sprouty function, such as potential differences in the level of RAS/ERK1/2 signaling or the activation of different downstream effectors, cannot be ruled out as contributing factors. A key question raised by our data is why Spry1 and Spry2 loss-of-function causes significant tumorigenesis only when it occurs in the context of Pten heterozygosity. The explanation we favor is based on our observations that AKT is aberrantly activated

gr

B

pAKT

A

PTEN

ECAD, DAPI

pMEK

DKO

w

DKO

lo

PE-Pten het; S1/2 Dhet

normal

S2GOF

60 40 20 0

E

F

K

2 wks total # 258 248 ducts

4 wks 256 222

9 mo 4345 2936

Fig. 4. Spry2 overexpression suppresses PIN caused by Pten loss-of-function. (A–F) H&E-stained sections of ventral prostate from animals of the genotypes indicated. At both 2 and 4 wk of age, the PE-Pten KO prostate displayed both normal ducts and ducts with low-grade PIN (A and C), whereas the PE-Pten KO;S2GOF prostate displayed approximately half as many ducts with PIN and the majority of ducts were normal (B and D). By 9 mo of age, almost all ducts in the PE-Pten KO prostate displayed high-grade PIN (E). In contrast, fewer than half of ducts displayed PIN in the PE-Pten KO;S2GOF prostate, and overall these ducts were less severely affected than those in the PE-Pten KO prostate (F compare with E ). (G–J) Serial sections immunostained (green) for PTEN (G and H) or pAKT (I and J); all sections were also immunostained (red) for ECAD and stained with DAPI (blue). Note that PTEN is absent from the epithelium, as expected, in both genotypes (G and H) and pAKT is detected at a high level in the epithelium of the PE-Pten KO prostate (I) but not the PE-Pten KO;S2GOF prostate (J). (K) Bar graph depicting the frequency of ducts displaying PIN of any grade in PE-Pten KO and PE-Pten KO;S2GOF animals at 2 and 4 wk and 9 mo of age. The distribution of normal ducts and ducts with PIN differed significantly between PE-Pten KO and PEPten KO;S2GOF prostates at each age (P < 0.0001, Fisher’s exact test).

Schutzman and Martin

(Fig. 3 E and F) thus provides strong evidence that Sprouty genes can negatively regulate the PI3K/AKT pathway in vivo. Further evidence for a genetic interaction between Sprouty genes and Pten was provided by our Spry2 GOF studies (Fig. 4) showing that overexpression of Spry2 can suppress PIN formation caused by Pten loss-of-function over many months. Thus, in prostates lacking PTEN but with a modest level of Spry2 overexpression, there were many more normal ducts than in the Pten null prostate. Significantly, in these normal ducts, we detected little or no pAKT despite the absence of PTEN (Fig. 4 G–J). PTEN loss-of-function, which is one of the most common and important oncogenic changes in human prostate cancer (2, 3), can be accompanied by silencing of Sprouty genes potentially as a result of multiple mechanisms including hypermethylation (27, 34) and aberrant microRNA expression (35). Because we have shown here that Sprouty genes are modulators of tumorigenesis caused by loss of Pten function, future studies aimed at obtaining a deeper understanding of the molecular consequences of Sprouty loss-of-function should provide insight into the rewiring of key oncogenic signaling pathways that occurs in the cancer cell. Moreover, Sprouty genes now represent a potential target for novel therapeutics. In the clinic, a large proportion of men diagnosed with prostate cancer will have clinicopathological features, including Gleason score, PSA value, and clinical staging, that suggest indolent disease. Under the current guidelines, these patients are often encouraged to pursue a course of active surveillance. However, some of these men will ultimately develop more aggressive invasive or disseminated lethal prostate cancer. Knowledge of the PTEN and Sprouty gene expression status in prostate biopsies from men at risk for prostate cancer could potentially help to risk-stratify patients with PIN, enabling clinicians to select those patients with disease that is more likely to progress to aggressive invasive cancer and who may benefit from more frequent surveillance or earlier intervention. Moreover, in high-risk patients with PIN that lacks PTEN and with low or absent Sprouty expression, novel treatment strategies, such as RNA activation (36), aimed at targeted reactivation of Sprouty gene expression, even at modest levels, might prove to be an effective intervention to delay or prevent the development of high-grade PIN and invasive prostate cancer by restoring a normal homeostatic feedback mechanism to modulate potent signal transduction pathways. Materials and Methods Mice. Mice carrying Tg(Osr1-cre)4Mrt (Osr1-cre) (17), the Spry1tm1Jdli (Spry1flox) and Spry1tm1.1Jdli (Spry1Δ) alleles (18), the Spry2tm1Mrt (Spry2flox) and Spry2tm1.1Mrt (Spry2Δ) alleles (19), the Ptentm2.1Ppp (Ptenflox) allele (5), and Tg(CAG-Bgeo,Spry2,-ALPP)1Mrt (Spry2-GOF) (24) were maintained and genotyped as previously described. FvB/N mice were used for in situ hybridization analysis of Spry1 and Spry2 expression. The breeding schemes used to obtain the mutant and control animals that were studied are illustrated in Table 1. Mice were maintained and all animal experiments were performed in accordance with

Table 1. Breeding schemes used to obtain mutant and control animals Parental male Osr1-cre

Tg/Tg

;Spry1

Parental female Δ/+

;Spry2

Δ/+

Osr1-creTg/Tg;Spry1Δ/+;Spry2Δ/+

Osr1-creTg/0;Ptenflox/+;Spry2-GOFTg/0

Spry1

flox/flox

;Spry2

flox/flox

Ptenflox/flox;Spry1flox/flox;Spry2flox/flox

Ptenflox/flox

Genotypes of interest Tg/0

flox/Δ

flox/Δ

Exp freq

Osr1-cre ;Spry1 ;Spry2 (PE-S1/2DKO) Osr1-creTg/0;Spry1flox/+;Spry2flox/+ (PE-S1/2Dhet)

1:4 1:4

Osr1-creTg/0;Ptenflox/+;Spry1flox/Δ;Spry2flox/Δ (PE-Pten het;S1/2DKO) Osr1-creTg/0;Ptenflox/+;Spry2flox/+;Spry2flox/+ (PE-Pten het;S1/2Dhet)

1:4

Osr1-creTg/0;Ptenflox/flox;Spry2-GOFTg/0 (PE-Pten KO;S2GOF) Osr1-creTg/0;Ptenflox/flox (PE-Pten KO)

1:8

1:4

1:8

The text in bold shows the abbreviations used in this paper for the genotypes shown. Exp freq, expected frequency.

Schutzman and Martin

PNAS | December 4, 2012 | vol. 109 | no. 49 | 20027

CELL BIOLOGY

in PE-Pten het;S1/2DKO but not in PE-S1/2DKO prostates and that pAKT is localized specifically to areas with PIN but is not detected in the ∼40% of ducts with normal epithelium. These data strongly suggest that the observed aberrant AKT activation is responsible for tumorigenesis, either on its own or in conjunction with the increase in pMEK that is found in prostates in which Spry1 and Spry2 have been inactivated in the epithelium (Figs. 1J and 3 B and C). AKT activation alone in a Pten wildtype background has been shown to be sufficient for the development of PIN in a mouse model (32), and deregulated PI3K/ AKT and RAS/ERK1/2 signaling has been shown to function synergistically to promote prostate tumorigenesis in mouse tissue recombinants and mutant mouse models (29, 31, 33). Thus, the combination of Spry1 and Spry2 loss-of-function and reduction in Pten dosage may promote tumorigenesis either as a consequence of its effect solely on the level of AKT activation or its effect on the level of both MEK and AKT activation. Importantly, our data also indicate that aberrant activation of AKT per se is dependent not only on reduced Pten dosage but also on the high level of pMEK caused by loss of Spry1 and Spry2 function, as pAKT is not detected in PE-Pten het;S1/2Dhet prostates in which there is a low level of pMEK. In keeping with the large body of evidence demonstrating the exquisite sensitivity of the negative feedback mechanisms that serve to regulate the RAS/ ERK1/2 and PI3K/AKT pathways and the cross-talk between them, we propose that there is sufficient negative feedback to prevent aberrant AKT activation either when Sprouty genes are inactivated (PE-S1/2DKO) or when the dosage of the Sprouty genes and Pten are concomitantly reduced (PE-Pten het;S1/2Dhet), but not when Sprouty genes are inactivated in combination with a reduction in Pten dosage (PE-Pten het;S1/2DKO). Under these circumstances, the level of negative feedback inhibition falls below a threshold necessary to limit AKT activation and prevent tumorigenesis. It remains unknown why a high level of pAKT was sporadic and was not detected throughout the PE-Pten het;S1/2DKO prostate epithelium. This may reflect small differences in local growth factor signaling that become amplified when negative regulators like Sprouty proteins are absent and the level of PTEN is reduced. One potential molecular mechanism by which a Sprouty protein could impact both the RAS/ERK1/2 and PI3K/AKT pathways is via its interaction with GRB2, the SH2/SH3 domaincontaining adapter protein that links RTK activation to both pathways (9). By this mechanism, Sprouty loss-of-function could coordinately deregulate these two key oncogenic pathways. Although it is well-established that Sprouty genes negatively regulate RAS/ERK1/2 signaling in multiple contexts (9, 10), and our data show that this is indeed the case in the adult prostate epithelium (Fig. 1 I and J), there are limited in vitro data demonstrating their effect on PI3K/AKT signaling (20). Our finding that Spry1 and Spry2 loss-of-function in the context of Pten heterozygosity can result in increased pAKT in the adult prostate

protocols approved by the Institutional Animal Care and Use Committee of the University of California, San Francisco.

sections were captured using a Nikon Spectral C1si Confocal microscope (Nikon Imaging Center at UCSF) with a 20× objective.

RNA in Situ Hybridization, Histology, and Immunofluorescence. For RNA in situ hybridization, prostates were fixed in 4% (wt/vol) paraformaldehyde overnight at 4 °C and embedded in paraffin. RNA in situ hybridization was performed according to standard protocols. To generate digoxigenin-labeled probes, plasmids containing the mouse Spry1 and Spry2 sequences were used for in vitro transcription (37). For histological analysis, prostates were fixed in Histochoice (Electron Microscopy Sciences) overnight at 4 °C and embedded in paraffin. We stained 5 μm sections, spaced 200 μm apart, with H&E according to standard protocols. Immunostaining was performed on paraffin sections after antigen retrieval by boiling the sections in 0.05% citriconic anhydride, pH7.5. The following primary antibodies were used: anti-phospho-MEK1/2 rabbit antibody (Ser217/ 221) (Cell Signaling #9121, 1:100), anti-phospho-AKT (Ser473) rabbit monoclonal antibody (Cell Signaling #3787, 1:100), anti-phospho-AKT (Thr308) rabbit antibody (Cell Signaling #9275, 1:50), anti-PTEN mouse monoclonal antibody (Cell Signaling #9556, 1:100), and anti-E-Cadherin (ECAD) rat monoclonal antibody (Invitrogen clone ECCD-2, 1:500). For immunofluorescence assays, anti-phospho-MEK1/2, anti-phospho-AKT, and anti-PTEN primary antibodies were detected after amplification using the Vectastain ABC kit (Vector Labs) and Tyramide amplification (Perkins Elmer); anti-ECAD primary antibody was detected using an Alexa-fluor conjugated secondary antibody (Molecular Probes, 1:200). Nuclei were stained with DAPI. The pMEK level in individual cells from normal prostate epithelium in PE-S1/2DKO and Dhet animals at 14 mo of age (n = 3 prostates per genotype, 125 cells per prostate) was quantified using Image J software to determine the integrated density (Mean Gray Value × Area) for each cell; pooled results for each genotype were compared using an unpaired t test (two-tailed P value < 0.0001). Brightfield images of H&E-stained sections were captured using a Leica DM5000B upright microscope with a 20× objective. Images of immunostained

Laser Capture Microdissection and Quantitative PCR. Laser capture microdissection was performed on H&E-stained 5 μm paraffin-embedded prostate sections from PE-S1/2DKO animals using a Zeiss PALM LMD (UCSF Laboratory for Cell Analysis Core). Two samples from different regions of the prostate from three different PE-S1/2DKO mice (n = 6 samples) were collected, and genomic DNA was isolated using the QiaAmp DNA Micro Kit per the manufacturer’s instructions. Control DNA was isolated from tails of Spry1Δ/+ and Spry2Δ/+ animals. For quantitative PCR analysis, all reactions were performed in triplicate, and Braf was used as the endogenous reference gene. The following primer sequences were used: Spry1 sense: CTT TGT GCC TAC CCT GCT TGC TCT GCT ACC; Spry1 anti-sense: AGG GCG GTG GGT CCA GTC GTA ACA GC; Spry2 sense: CCT CTG TCC AGG TCC ATC AGC ACT GTC AGC; Spry2 anti-sense: GCA GCA GCA GGC CCG TGG GAG AAG; Braf sense: TGA GTA TTT TTG TGG CAA CTG C; and Braf anti-sense: CTC TGC TGG GAA AGC GGC (38).

1. Siegel R, Naishadham D, Jemal A (2012) Cancer statistics, 2012. CA Cancer J Clin 62(1): 10–29. 2. Whang YE, et al. (1998) Inactivation of the tumor suppressor PTEN/MMAC1 in advanced human prostate cancer through loss of expression. Proc Natl Acad Sci USA 95(9):5246–5250. 3. Wu X, Senechal K, Neshat MS, Whang YE, Sawyers CL (1998) The PTEN/MMAC1 tumor suppressor phosphatase functions as a negative regulator of the phosphoinositide 3kinase/Akt pathway. Proc Natl Acad Sci USA 95(26):15587–15591. 4. Wang S, et al. (2003) Prostate-specific deletion of the murine Pten tumor suppressor gene leads to metastatic prostate cancer. Cancer Cell 4(3):209–221. 5. Trotman LC, et al. (2003) Pten dose dictates cancer progression in the prostate. PLoS Biol 1(3):E59. 6. Gioeli D, Mandell JW, Petroni GR, Frierson HF, Jr., Weber MJ (1999) Activation of mitogen-activated protein kinase associated with prostate cancer progression. Cancer Res 59(2):279–284. 7. Gioeli D (2005) Signal transduction in prostate cancer progression. Clin Sci (Lond) 108 (4):293–308. 8. Palanisamy N, et al. (2010) Rearrangements of the RAF kinase pathway in prostate cancer, gastric cancer and melanoma. Nat Med 16(7):793–798. 9. Mason JM, Morrison DJ, Basson MA, Licht JD (2006) Sprouty proteins: Multifaceted negative-feedback regulators of receptor tyrosine kinase signaling. Trends Cell Biol 16(1):45–54. 10. Cabrita MA, Christofori G (2008) Sprouty proteins, masterminds of receptor tyrosine kinase signaling. Angiogenesis 11(1):53–62. 11. Taylor BS, et al. (2010) Integrative genomic profiling of human prostate cancer. Cancer Cell 18(1):11–22. 12. Lo TL, et al. (2006) Sprouty and cancer: The first terms report. Cancer Lett 242(2): 141–150. 13. Fritzsche S, et al. (2006) Concomitant down-regulation of SPRY1 and SPRY2 in prostate carcinoma. Endocr Relat Cancer 13(3):839–849. 14. Shaw AT, et al. (2007) Sprouty-2 regulates oncogenic K-ras in lung development and tumorigenesis. Genes Dev 21(6):694–707. 15. Minowada G, Miller YE (2009) Overexpression of Sprouty 2 in mouse lung epithelium inhibits urethane-induced tumorigenesis. Am J Respir Cell Mol Biol 40(1):31–37. 16. Lee SA, et al. (2008) Integration of genomic analysis and in vivo transfection to identify Sprouty 2 as a candidate tumor suppressor in liver cancer. Hepatology 47(4):1200–1210. 17. Grieshammer U, Agarwal P, Martin GR (2008) A Cre transgene active in developing endodermal organs, heart, limb, and extra-ocular muscle. Genesis 46(2):69–73. 18. Basson MA, et al. (2005) Sprouty1 is a critical regulator of GDNF/RET-mediated kidney induction. Dev Cell 8(2):229–239. 19. Shim K, Minowada G, Coling DE, Martin GR (2005) Sprouty2, a mouse deafness gene, regulates cell fate decisions in the auditory sensory epithelium by antagonizing FGF signaling. Dev Cell 8(4):553–564.

20. Basson MA, et al. (2006) Branching morphogenesis of the ureteric epithelium during kidney development is coordinated by the opposing functions of GDNF and Sprouty1. Dev Biol 299(2):466–477. 21. Di Cristofano A, Pesce B, Cordon-Cardo C, Pandolfi PP (1998) Pten is essential for embryonic development and tumour suppression. Nat Genet 19(4):348–355. 22. Freeman D, et al. (2006) Genetic background controls tumor development in PTENdeficient mice. Cancer Res 66(13):6492–6496. 23. Shen MM, Abate-Shen C (2010) Molecular genetics of prostate cancer: New prospects for old challenges. Genes Dev 24(18):1967–2000. 24. Basson MA, et al. (2008) Specific regions within the embryonic midbrain and cerebellum require different levels of FGF signaling during development. Development 135(5):889–898. 25. Maddison LA, Nahm H, DeMayo F, Greenberg NM (2000) Prostate specific expression of Cre recombinase in transgenic mice. Genesis 26(2):154–156. 26. Wu X, et al. (2001) Generation of a prostate epithelial cell-specific Cre transgenic mouse model for tissue-specific gene ablation. Mech Dev 101(1–2):61–69. 27. McKie AB, et al. (2005) Epigenetic inactivation of the human sprouty2 (hSPRY2) homologue in prostate cancer. Oncogene 24(13):2166–2174. 28. Kwabi-Addo B, et al. (2004) The expression of Sprouty1, an inhibitor of fibroblast growth factor signal transduction, is decreased in human prostate cancer. Cancer Res 64(14):4728–4735. 29. Mulholland DJ, et al. (2012) Pten loss and RAS/MAPK activation cooperate to promote EMT and metastasis initiated from prostate cancer stem/progenitor cells. Cancer Res 72(7):1878–1889. 30. Pearson HB, Phesse TJ, Clarke AR (2009) K-ras and Wnt signaling synergize to accelerate prostate tumorigenesis in the mouse. Cancer Res 69(1):94–101. 31. Wang J, et al. (2012) B-Raf activation cooperates with PTEN loss to drive c-Myc expression in advanced prostate cancer. Cancer Res 72(18):4765–4776. 32. Majumder PK, et al. (2003) Prostate intraepithelial neoplasia induced by prostate restricted Akt activation: The MPAKT model. Proc Natl Acad Sci USA 100(13):7841–7846. 33. Gao H, et al. (2006) Combinatorial activities of Akt and B-Raf/Erk signaling in a mouse model of androgen-independent prostate cancer. Proc Natl Acad Sci USA 103(39):14477–14482. 34. Kwabi-Addo B, Ren C, Ittmann M (2009) DNA methylation and aberrant expression of Sprouty1 in human prostate cancer. Epigenetics 4(1):54–61. 35. Darimipourain M, Wang S, Ittmann M, Kwabi-Addo B (2011) Transcriptional and posttranscriptional regulation of Sprouty1, a receptor tyrosine kinase inhibitor in prostate cancer. Prostate Cancer Prostatic Dis 14(4):279–285. 36. Kang MR, et al. (2012) Intravesical delivery of small activating RNA formulated into lipid nanoparticles inhibits orthotopic bladder tumor growth. Cancer Res 72(19): 5069–5079. 37. Klein OD, et al. (2006) Sprouty genes control diastema tooth development via bidirectional antagonism of epithelial-mesenchymal FGF signaling. Dev Cell 11(2):181–190. 38. Dankort D, et al. (2009) Braf(V600E) cooperates with Pten loss to induce metastatic melanoma. Nat Genet 41(5):544–552.

20028 | www.pnas.org/cgi/doi/10.1073/pnas.1217204109

ACKNOWLEDGMENTS. We thank Andrew Hsieh and Bilge Reischauer for advice on qPCR analysis; Vivian Weinberg for advice on statistical analysis; Greg Hamilton (Laboratory for Cell Analysis at UCSF) for providing technical training in laser capture microdissection; Prajakta Ghatpande for technical assistance; and Jonathan Licht, Martin McMahon, Ross Metzger, Davide Ruggero, and Nan Tang for helpful suggestions and insightful criticism. We also thank the University of California at San Francisco (UCSF) Mouse Pathology Core. This work was supported by National Institutes of Health (NIH) Grant R01 CA078711 (to G.R.M.). J.L.S. was supported by a Prostate Cancer Foundation Young Investigator Award, a UCSF Clinical and Translational Science Institute Pilot Research Award for Junior Investigators in Basic and Clinical/Translational Sciences, an American Cancer Society Postdoctoral Fellowship, an American Society of Clinical Oncology Young Investigator Award, a UCSF Prostate Cancer Center Fellowship, and NIH Training Grant NIH T32DK007636.

Schutzman and Martin