Differential Requirement for Src-family Tyrosine Kinases in the Initiation, Progression and Metastasis of Prostate Cancer

Abstract

Prostate cancer (CaP) recurrence after androgen ablation therapy (ADT) remains a significant cause of mortality in aging men. Malignant progression and metastasis are typically driven by genetic and epigenetic changes controlled by the androgen receptor (AR). However, evidence suggests that activated non-receptor tyrosine kinases, including those of the Src family (SFK), directly phosphorylate AR, thereby activating its transcriptional activity in the absence of serum androgen levels. To ascertain whether CaP progression and metastasis require SFK members an autochthonous transgenic adenocarcinoma (AD) of the mouse prostate (TRAMP) model was crossed into Src-, Lyn- or Fyn-null backgrounds. Primary-site CaP formation was dependent on Src, to a lesser extent, Lyn, but not Fyn. Only Src^−/−;TRAMP prostate tumors were marked by reactive stroma. SFK deficiency did not affect progression to neuroendocrine (NE) disease, although there were fewer new cancer cases initiating after 34 weeks in the SFK^−/−;TRAMP mice compared to TRAMP controls. Fifteen to 21% of older (>33 weeks) Lyn- or Fyn-null TRAMP mice lacking primary-site tumors suffered from aggressive metastatic AD growths, compared with 3% of TRAMP mice. Taken with the data that TRAMP mice lacking Src or Lyn exhibited fewer macroscopic metastases compared to Fyn^−/−;TRAMP and TRAMP controls, this suggests that SFK can either promote or suppress specific parameters of metastatic growth, possibly depending on cross-talk with primary tumors. These data identify critical, yet potentially opposing roles played by various SFKs in the initiation and metastatic potential of CaP using the TRAMP model. Implications: Genetically defined mouse models indicate a critical role for Src tyrosine kinase in prostate cancer initiation and metastatic progression.

INTRODUCTION

Prostate Cancer (CaP) is an endocrine malignancy whose strict growth requirement for androgens such as dihydrotestosterone (DHT) gave rise to so-called androgen deprivation therapy (ADT), in which anti-androgens such as bicalutamide induce tumor regression by starving the androgen receptor (AR) of serum ligand levels. In the naïve, androgen-stimulated (AS) setting, AR functions as an androgen-dependent transcription factor that controls the expression of proliferation and survival genes in concert with scores of co-activator or –repressor proteins. ADT failure is marked by castration-recurrent (CR) CaP in the prostate and/or metastatic sites, especially draining lymph nodes and bones. Interestingly, CR-CaP lesions typically express wt-AR that is active in the absence of serum androgen levels, yet still responsive to very low, intracrine levels of androgens thought to be produced at the tumor site (1, 2).

Several mechanisms for AR activation in the CR setting have been described (2), including i) overexpression of wt-AR, ii) expression of AR mutants or splice variants (3, 4), iii) alterations in AR co-regulator expression (5), and iv) AR modification such as by activated serine/threonine kinases (6, 7), and tyrosine kinases such as Ack1 (8) and Src-family kinases (SFK) (9). Progression to CR-CaP is marked by increased total protein phosphotyrosine (10) and relative Ack1 and SFK autophosphorylation/kinase levels (8, 9, 11), likely activated downstream of several receptor tyrosine kinases and G-protein coupled receptors known to be activated during CaP progression. Src potentiates AR function in human CR-CaP cell lines (12, 13) and activated Src is sufficient to induce CaP formation in tissue recombination models involving AR overexpression (10, 14, 15). Indeed, androgen-independent AR transactivation function can be induced following phosphorylation on Y267 by Ack1 (8) or on Y534 by Src (9, 16). The growing appreciation for the role of SFK and Ack1 in AR-dependent CaP progression, as well the role played by SFK in metastatic invasiveness to and growth in the bone environment (14, 16–20), have served as the rationale for a growing number of clinical trials incorporating SFK and broad tyrosine kinase inhibitors in CR-CaP (21–23).

TRAMP (transgenic adenocarcinoma of mouse prostate) is an autochthonous model in which CaP adenocarcinoma (AD) is induced by the prostate epithelial cell-specific inactivation of p53 and Rb via the expression of SV40 T-antigen driven by a truncated probasin gene (Arr2bp) promoter (24). In this model, an AR-negative neuroendocrine (NE) cancer eventually takes over, seeding both primary prostate sites as well as multiple metastatic sites. We addressed whether CaP formation, progression to NE disease, and metastatic formation in this model would be affected by the loss of Src, Lyn or Fyn, SFK members known to be expressed in human prostate epithelial cells (25). Our data indicate that CaP formation is suppressed by Src and to a lesser extent, Lyn, but not by Fyn, and that in mice forming CaP tumors, loss of the SFK did not prevent the onset of primary-site NE disease. Interestingly, whereas Src and Lyn seem to be required for NE metastasis, a small percentage of older Fyn- and Lyn-deficient mice lacking primary-site tumors exhibited highly aggressive metastatic AD. These data validate the concept of therapeutically targeting SFK to prevent initiation and progression of CaP disease.

MATERIAL AND METHODS

Animals and Genotyping

B6;129S7-Fyn^tm1Sor/J mice (stock number 002385), B6;129S4-Lyn^tm1Sor/J mice (stock number 003515), and B6;129S7-Src^tm1Sor/J mice (stock number 002277) were purchased from Jackson Laboratories. TRAMP mice were kindly obtained through the Mouse Tumor Model Resource, Roswell Park Cancer Institute (Buffalo, NY). Male and female Fyn^−/− and Lyn^+/− mice were bred to homozygous TRAMP^+/+ positive males and females on an FVB background to produce Fyn^−/−;TRAMP^+/wt and Lyn^−/−;TRAMP^+/wt males and females. Src^+/− males and females were mated to homozygous TRAMP^+/+ male and female mice on an FVB background to produce Src^−/−;TRAMP^+/wt male and Src^+/−;TRAMP^+/wt female mice. TRAMP^wt/wt littermates were used for negative controls. Src^−/− female mice have impaired fertility (26) and thus, Src^+/− were used as breeders. Mice that are Src^−/− also exhibit decreased length of long bones leading to a noticeably smaller body size compared to that of heterozygous littermates, slower weight gain, delayed eye lid opening and incisors that fail to erupt (http://jaxmice.jax.org/strain/002277.html), requiring a soft food diet (catalog number S4798-TRAY, Bio Serv) and delayed weaning until 5 weeks of age. Src^−/− mice exhibit an increased mortality rate in the first 3–5 weeks compared to Src^+/− littermates.

Genotyping was performed using DNA extracted as described previously (27) from tails snips taken at weaning. Genotyping was performed by PCR (1 cycle 94°C for 3 min, 35 cycles: 94°C for 30 sec, 54°C for 45 sec, 72°C for 45 sec); primers sets and products sizes for WT and mutant genomes are as follows: Fyn- WT, 415 bp; mutant, 190 bp. Lyn- WT, 340 bp; mutant, 280. Src- WT, 200 bp; mutant, 450 bp. Casein- 500 bp. TRAMP- 600 bp. PCR products were separated by gel electrophoresis on 1.5% agarose gels and visualized following staining with ethidium bromide.

Tumor Palpation and Tissue Collection

Beginning at 8 weeks of age, all male mice where weighed, checked weekly for tumor or metastasis formation by palpation, and tumor volumes, measured by caliper, were calculated using the formula, V=0.5(L×W²), where L is the longest dimension and W is the width. Tumor volume was used as the endpoint unless health or distress issues required earlier euthanasia as recommended by veterinary staff. Animals were anesthetized with isoflurane and blood was collected as a terminal bleed via cardiac puncture. Tumors, metastases and organs were removed post-mortem, and were either used to produce lysates for immunoblotting or fixed in formalin for H&E and IHC analysis. All procedures were in accordance with approval by the RPCI -IACUC.

Immunohistochemistry (IHC) and tissue imaging

IHC was performed as described previously (28) using a purified mouse anti-SV40 Large T Antigen (T_ag)(BD, Franklin Lakes, NJ), polyclonal rabbit anti-synaptophysin (Invitrogen), and polyclonal rabbit anti-cytokeratin, wide spectrum screening (DAKO, Glostrup, Denmark) Tissues were fixed in 10% neutral buffered formalin, placed in paraffin and sectioned at 5 μm. Tissue sections were de-paraffinized rehydrated with a graded alcohol series, and rinsed in distilled H₂O. Antigen retrieval was performed with 1X citrate buffer (pH 6.0, Invitrogen). After washing, endogenous peroxidase activity was blocked with a 3% H₂O₂ in methanol solution and equilibrated with 1X Tris-PO₄ buffer (10x solution 8.4 mM Na₂HPO₄, 3.5 mM KH₂PO₄, 10 mM Tris, 120 mM NaCl, 0.5% Tween-20, pH 7.8) followed by blocking in 1% BSA/Tris-PO₄. Primary antibodies (Ab) were diluted in 1% BSA/Tris-PO₄, placed on tissue sections and incubated overnight at room temperature in a humidified chamber. Slides were washed and the secondary Ab, rabbit anti-mouse or goat anti-rabbit (DAKO) was applied to each tissue section for 2–4 hours. Ab staining was visualized using a 1 mg/ml 3′3-diaminobenzidine chromogen (Sigma, St. Louis, MO) and 1 μl/ml H₂O₂ solution. Slides were counterstained with Meyer’s hematoxylin, dehydrated and coverslipped. Standard staining methods were used to produce H&E staining of all tissues. IHC slides were scanned into an Aperio Spectrum Digital Slide system, and images were viewed and captured using Aperio ImageScope software (Aperio Technologies, Inc, Vista, CA). Microscopic metastases were identified by a combination of analysis of H&E and SV40 T_ag staining.

Immunoblotting (IB) and co-immunoprecipitation (co-IP)

IB was performed as previously described (29) on dorso-lateral lobe RIPA buffer lysates using Ab specific for Fyn (FYN3, Santa Cruz Biotechnology, Santa Cruz, CA), Lyn (#44, Santa Cruz), Src (Cell Signaling, Danvers, MA), phosphoY416-Src (Cell Signaling), and Gapdh (Santa Cruz). Co-IP analyses were performed on DLP RIPA lysates from 8 week-old mice with early palpable CaP tumors as follows: 500 mg total protein/sample was incubated for 8 hours at 4°C with 1 μg AR Ab (Santa Cruz, catalog #sc-816), then incubated overnight with A/G agarose beads (Santa Cruz). After three washes with PBS, the IPs were subjected to IB analysis with either AR Ab (Santa Cruz, Cat. #sc-819) or anti-poY MAb-4G10 (Millipore, cat. #05–1050X). Digital imaging and signal quantification were performed on a Chemi-Genius2 Bio-Imager (Syngene, Frederick, MD) using GeneTools software.

Statistical analyses

Statistical analyses were performed by using the GraphPad Prism v5 (GraphPad Software, San Diego, CA). Differences in tumor incidence were assessed using the Fisher exact test.

RESULTS

In order to assess the role of specific SFK members in CaP initiation, progression and metastatic potential, TRAMP mice, carrying the SV40 T_ag driven by the probasin Arr2bp promoter (30), were crossed onto Src-, Lyn- or Fyn-null backgrounds. Src protein and activation levels increase throughout the initiation and progression phases of CaP-AD and -NE in the TRAMP model (12), suggesting that rising SFK levels might contribute to TRAMP oncogenesis. Deficiency of each of the SFK members was confirmed by immunoblots of prostate dorso-lateral lysates (vs. those from TRAMP mice) using Src, Lyn and Fyn-specific monoclonal antibodies (Fig. 1A). There was no evidence that the loss of one specific SFK led to the compensatory overexpression of other SFKs. Similarly, relative levels of SFK^poY416, which recognizes a shared SFK autophosphorylation site and which correlates with SFK activity (31), were not grossly affected by the loss of Src, Lyn or Fyn (Fig. 1A). Although there was some variation in the relative SFK^poY416 levels between individual SFK^−/−;TRAMP mice, overall relative SFK^poY416 levels increased in all genotypes as the mice aged. This may reflect to increasing activation of upstream receptor tyrosine kinases during CaP progression. Thus, any suppression of tumor initiation in the SFK^−/−;TRAMP mice relative to TRAMP controls would likely be due to the loss of Src, Lyn or Fyn rather than the upregulation of activity by other SFK members.

Figure 1. Tumor formation in the TRAMP model in the absence of SFK.

(A) Loss of Src, Lyn or Fyn was confirmed by immunoblotting of dorso-lateral CaP lysates from either TRAMP control mice (left panel) or in SFK^−/−;TRAMP mice (right panels), in comparison to relative levels of Src^poY416 (correlating with overall SFK activation levels) or Gapdh (as a protein loading control). N.S., non-specific. (B) Tumor formation vs. age (in weeks) of the SFK^−/−;TRAMP vs. TRAMP controls based on the following numbers of mice studied and statistical power relative to TRAMP controls: Src^−/−;TRAMP- 36 (p=0.001), Lyn^−/−;TRAMP- 60 (p= 0.143), Fyn^−/−;TRAMP- 72 (p=0.735), TRAMP- 10. +, censored. Note that while every TRAMP mouse (10/10) developed a tumor by 36 weeks of age (median 23 weeks), the Src^−/−;TRAMP mice developed tumors more slowly (median 36 weeks) and less frequently (10 of 26 mice total). Additionally, there is no significant difference in mice that develop tumors or live to 36 weeks without tumors: overall Log-rank p=0.09; Fyn^−/−;TRAMP vs. TRAMP, p=0.578); Lyn^−/−;TRAMP vs. TRAMP, p=0.38; Src^−/−;TRAMP vs. TRAMP, p=0.172. (C) Smoothed hazards for TRAMP vs. Lyn^−/−;TRAMP mice showing decreased hazard for tumor incidence in older (>29 wks) Lyn^−/−; TRAMP mice. Shaded region reflects the difference incidence between genotype groups (blue = TRAMP, red = Lyn^−/−; TRAMP) prior to 24 wks and after 29 wks. Permutation p-values (calculated by log-rank test) are based on the area shaded in blue if the group identifiers are randomly re-sorted. (D) Left panel- DLP lobe lysates from 8 week-old SFK-null;TRAMP or TRAMP mice (3 mice/group) were immunoprecipitation using AR-specific Ab and then analyzed by IB for AR or total phosphotyrosine (poY). Right panel- The relative AR-poY from each group was quantified by densitometric analysis as described in Materials and Methods, using the mean value for parental TRAMP =1. Error bars, S.E. of 3 mice/group. *, p<0.01 using student’s one-tailed T-test.

The oncogenic progression of CaP in TRAMP mice is marked by the onset of AR-positive AD from 8–12 weeks after birth followed by the onset and eventual predominance of NE disease. In contrast, although limited developmental defects are detected in select SFK-deficient mice, such as osteopetrosis in Src-null mice (32), the loss of Src, Lyn or Fyn alone is not sufficient to induce developmental or pro-oncogenic effects in the prostate, suggesting that the redundant expression of other SFK is compensatory for the lost function. In order to assess how the loss of specific SFK affected TRAMP initiation and progression, we identified cancer cases based on the appearance of palpable DLP tumors, which occurred in 48% of parental TRAMP mice by 20 weeks of age (Fig. 1B), in agreement with previous assessments (24). The kinetics of tumor initiation in Fyn^−/−;TRAMP and Lyn^−/−;TRAMP mice during weeks 10–28 was indistinguishable from that of parental TRAMP mice. The Fyn^−/−;TRAMP mice continued to exhibit similar tumor initiation kinetics to that of TRAMP mice through 34 weeks, after which, there was a slight slowdown in tumor initiation rate in Fyn^−/−;TRAMP mice, whereas there was a statistically significant slowing of the tumor initiation rate in Lyn^−/−;TRAMP mice after week 30 compared to the parental TRAMP mice (Fig. 1C). In contrast, Src^−/−;TRAMP mice showed much less tumor initiation starting at week 18 (p=0.001). These data indicate that TRAMP-associated tumor initiation is dependent on Src, somewhat on Lyn but less so on Fyn. It should be noted that SFK losses did not alter the propensity of the CaP lesions to form in DLP lobes.

To address possible mechanisms for the relatively suppressed oncogenesis in Src^−/−;TRAMP mice, we immunoprecipitated androgen receptor (AR) from DLP lysates of SFK-null;TRAMP or TRAMP mice (3 mice each) during early AD formation (8 week-olds), immunoblotted these proteins using anti-phosphotyrosine antibody, and then quantified the relative AR-poY levels by densitometry. Early-onset lesions were chosen because, compared to NE lesion- which are typically AR-negative, they would more likely be driven by AR-mediated signaling. Fig. 1D shows that the relative AR-poY levels in Src^−/−;TRAMP AD lesions were significantly lower than those in parental TRAMP, Fyn^−/−;TRAMP or Lyn^−/−;TRAMP AD lesions. This correlates with the decreased early tumor formation Src^−/−;TRAMP mice (Fig. 1B), and suggests that Src is the most critical of the three SFK in activating AR function by direct phosphorylation.

The loss of Fyn, Lyn or Src had different effects on the overall pathological progression of AD or NE CaP disease compared to TRAMP controls based on, respectively, cytokeratin or synaptophysin expression (Table 1). For example, most AD-only tumors in Fyn^−/−;TRAMP or Lyn^−/−;TRAMP mice occurred early (weeks 21–28), although this was a little delayed compared to TRAMP controls, which had their highest incidence in 16–20 week-old mice. Src^−/−;TRAMP mice displayed AD-only tumors even later, from weeks 25 to >32. Similar to TRAMP controls, AD/NE-combination tumors appeared in Fyn^−/−;TRAMP throughout the 16–32 week period. In contrast, Lyn^−/−;TRAMP and Src^−/−;TRAMP showed much later development of AD/NE-combination tumors. Whereas more NE-only tumors were found in older TRAMP mice (25–32 weeks), there was a paradoxical early appearance of these tumors in 16–20 week-old SFK^−/−;TRAMP mice, and these lesions were marked by reactive stroma, based on increased mesenchymal layers and cellular infiltrates (Fig. 2A). Later-forming NE CaP tumors, based on a dearth of cytokeratin staining and an abundance of synaptophysin staining, could be found in all SFK^−/−;TRAMP and TRAMP groups (Fig. 2B). As a control, we showed that the unaffected VP lobes of control C57BL/6 mice had abundant cytokeratin staining throughout the luminal cell layer, but only few synaptophysin-positive cells in the basal layer (inset: arrows). Taken together, these data indicate that the loss of specific SFK affected the initiation and progression of AD and NE disease in the TRAMP model.

Table 1.

Incidence of adenocarcinoma (AD) vs. neuroendocrine (NE) tumors in SFK^−/−;TRAMP vs. TRAMP mice

Age (wk)	AD				AD/NE				NE
	Fyn−/−; TRAMP+	Lyn−/−; TRAMP+	Src−/−; TRAMP+	TRAMP	Fyn−/−; TRAMP+	Lyn−/−; TRAMP+	Src−; TRAMP+	TRAMP	Fyn−/−; TRAMP+	Lyn−/−; TRAMP+	Src−/−; TRAMP+	TRAMP
16–20	0/15 (0)*	0/11 (0)	0/3 (0)	20/29 (69)	2/15 (13.3)	0/11 (0)	0/3 (0)	6/29 (20.7)	13/15 (86.7)	11/11 (100)	3/3 (100)	3/29 (10.3)
21–24	0/7 (0)	2/11 (18.2)	0/1 (0)	N.D.	0/7 (0)	0/11 (0)	1/1 (100)	N.D.	7/7 (100)	9/11 (81.8)	0/1 (0)	N.D.
25–28	0/8 (0)	1/12 (8.3)	1/1 (100)	12/19 (63.2)	2/8 (25.0)	0/12 (0)	0/1 (0)	4/19 (21.5)	6/8 (75)	11/12 (91.7)	0/1 (0)	3/19 (15.8)
29–32	5/10 (50)	0/1 (0)	1/1 (100)	0/16 (0)	3/10 (30.0)	0/1 (0)	0/1 (0)	4/16 (25.0)	2/10 (20)	1/1 (100)	0/1 (0)	12/16 (75)
>32	3/4 (75)	3/4 (75)	1/3 (33.3)	N.D.	0/4 (0)	1/4 (25.0)	2/3 (66.7)	N.D.	1/4 (25)	0/3 (0)	0/3 (0)	N.D.
All ages	7/44			32/59‡

^*# with CaP/total # per genotype (% with CaP per genotype)

AD: >80% cytokeratin-positive IHC.

NE: >80% staining synaptophysin-positive.

AD/NE: individual VP lobes containing cytokeratin- and synaptophysin-positive lesions.

N.D., not done

^‡χ² p=0.00017 for the difference in proportions for the across-age summary of TRAMP vs. Fyn−/−;TRAMP AD tumors.

Figure 2. Effect of SFK loss on TRAMP pathogenic progression.

(A) Comparison of dorso-lateral CaP tumors in 20 week-old Src−/−; or Fyn−/−;TRAMP mice, showing extensive stromal formation in the absence of Src. (B) Ventral lobes from 16–25 week-old SFK^−/−;TRAMP, TRAMP or C57BL/6 mice were analyzed by H&E staining or by IHC for SV40 T_ag, high molecular weight cytokeratins (CK) or synaptophysin. Inset, 60X magnification showing typical cell-cell CK staining in the luminal epithelium of normal VP lobes with rare peripheral synaptophysin-positive cells (arrows). (C) NE LN-metastases in 19–25 week-old SFK^−/−;TRAMP or TRAMP mice based on synaptophysin-positive, CK-negative staining.

We then analyzed whether the loss of SFK affected the relative proliferation, cell death and T_ag expression rates in 20 week-old mice. Proliferation was measured by Ki67 staining whereas cell death was measured by TUNEL staining (Table 2). Although the Lyn^−/−;TRAMP CaP proliferation rate was less than half that of the Fyn^−/−;TRAMP and Src^−/−;TRAMP tumors, it was comparable to that of TRAMP tumors. Cell death rates were statistically comparable between the genotypes. Lastly, there was a slight decrease in the T_ag positivity in Src^−/−;TRAMP tumors, possibly owing to the aforementioned increase in reactive stroma.

Table 2.

Rates of proliferation, cell death and T-antigen expression^*

	Fyn−/−;TRAMP	Lyn−/−;TRAMP	Src−/−;TRAMP	TRAMP
Ki67	6.6 +/− 1.0	3.0 +/− 0.6	7.4 +/− 0.4	4.8 +/− 1.6
TUNEL	7.0 +/− 0.6	9.4 +/− 1.6	7.2 +/− 0.6	7.6 +/− 0.6
T-ag (nuclear)	95.6 +/− 1.0	95.8 +/− 0.8	89.6 +/− 2.0	94.6 +/− 1.8

^*Percent positive cells based on counting at least 50 cells in each of 6 independent fields at 20x magnification from two 20-week-old mice/genotype.

To assess whether SFK play roles in general or organ-specific metastasis formation, 16–33 week-old TRAMP or SFK^−/−;TRAMP mice bearing roughly equally-sized primary tumors were examined post-mortem for macroscopic metastases. Whereas Fyn^−/−;TRAMP mice exhibited similar rates and organ distribution of metastases compared to parental TRAMP mice, Lyn^−/−;TRAMP and Src^−/−;TRAMP mice were relatively deficient in macroscopic metastases (Table 3). Indeed, most metastasis formation in all mouse crosses was detected in local draining pelvic lymph nodes, yet the Lyn^−/−; and Src^−/−;TRAMP mice exhibited little or no macro-metastasis formation in other peripheral organs. Similar results were found when micrometastases were analyzed (Table 4). The vast majority of these metastases were of NE origin (Fig. 2C), most likely arising from primary-site NE tumors. Taken together, these data indicate that NE macro-metastasis formation is suppressed in the absence of Src and Lyn, but not in the absence of Fyn.

Table 3.

Incidence of macroscopic metastases in tumor-bearing mice^*

	Fyn−/−;TRAMP	Lyn−/−;TRAMP	Src−/−;TRAMP	TRAMP#	TRAMP‡
Pelvic LN	35/49 (71.4)	22/40 (55.0)	4/26 (15.4)	9/25 (36.0)	12/40 (30.0)
Renal LN	14/49 (28.6)	3/40 (7.5)	0/26 (0.0)	N.D.	N.D.
Kidney	6/49 (12.2)	2/40 (5.0)	1/26 (3.9)	0/37 (0)	1/43 (2.3)
Liver	6/49 (12.2)	0/40 (0.0)	0/26 (0.0)	1/37 (2.7)	2/43 (4.6)
Lung	3/49 (6.1)	1/40 (2.5)	0/26 (0.0)	12/37 (32.4)	2/43 (4.6)
other	1/49 (2.0)	1/40 (2.5)	0/26 (0.0)	1/37 (2.7)	N.D.

LN: lymph node; SV: seminal vesicle; N.D.: not determined.

^*tumor-bearing mice with palpable prostate tumors at 15–33 weeks of age

^#published TRAMP data described in (24, 30).

^‡Data using RPCI TRAMP stock used to generate the SFK−/−;TRAMP genotypes in the current study, as described in: “Early growth inhibition is followed by increased metastatic disease with vitamin D (calcitriol) treatment in an aggressive mouse model of prostate cancer.” Adebusola Alagbala Ajibade, Jason S. Kirk, Ellen Karasik, Bryan Gillard, Michael T. Moser, Candace S. Johnson, Donald L. Trump, and Barbara A. Foster (submitted for publication).

Table 4.

Incidence of microscopic metastases in tumor-bearing mice^*

	Fyn−/−;TRAMP	Lyn−/−;TRAMP	Src−/−;TRAMP	TRAMP
Pelvic LN	37/45 (82.2%)	33/35 (94.3%)	5/25 (20.0%)	19/39 (48.7%)
Renal LN	11/45 (24.4%)	2/35 (5.7%)	0/25 (0.0%)	N.D.
Kidney	8/45(17.8%)	1/35 (2.9%)	1/25 (4.0%)	1/43 (2.3%)
Liver	6/45 (13.3%)	0/35 (0.0%)	0/25 (0.0%)	2/43 (4.7%)
Lung	18/45 (40.0%)	4/35 (11.4%)	1/25 (4.0%)	3/20 (15.0%)
SV	3/45 (6.7%)	0/35 (0.0%)	0/25 (0.0%)	N.D.

^*identified by microscopic analysis of H&E-stained tissue sections from tumor-bearing mice where tumor was palpated between 15–33 weeks of age.

N.D., not determined.

We noted that between 35–65% of SFK^−/−;TRAMP mice older than 16 weeks remained tumor free, as defined initially by palpation and then by postmortem pathological confirmation, compared to 41% of TRAMP controls. However, whereas only 3% of tumor-free TRAMP mice exhibited metastases, 21% of Fyn^−/−;TRAMP and 15% of Lyn^−/−;TRAMP mice were found to have macrometastases post-mortem; one 36 week-old Src^−/−;TRAMP mice was tumor free yet presented with a large pelvic lymph node mass (5 g), initially thought to be a palpable prostate tumor, plus renal metastases and abdominal ascites (Table 5). Except for the single Src^−/−;TRAMP case, these mice were identified based on a lack of palpable prostate tumors yet rapid onset of cachexia or even sudden death, suggesting an underlying aggressive disease. Unlike the typical NE metastases found in tumored mice (Fig. 2C), which were found mainly in draining lymph nodes, these lesions were distributed throughout multiple organs, and moreover, they displayed AD rather than NE markers (Fig. 3A &B). This suggests that Fyn and Lyn may suppress the early dissemination of AD progenitor cells, which might evolve into metastatic growths independent of primary CaP progression.

Table 5.

Incidence and sites of macroscopic metastases in tumor-free mice^*

	Fyn−/−;TRAMP	Lyn−/−;TRAMP	Src−/−;TRAMP	TRAMP
Total # mice studied	62	52	26	621**
Tumor-free mice	29 (47%)	18 (35%)	17 (65%)	254 (41%)
Tumor-free with metastases	13 (21%)	8 (15%)	1 (100%)‡	18 (3%)
Metastatic sites
Pelvic LN	4/13 (30.8%)	3/8 (37.5%)	1/1 (100.0%)‡	6/18 (33.3%)
Renal LN	2/13 (15.4%)	2/8 (25.0%)	0/0 (0.0%)	2/18 (11.1%)
Kidney	7/13 (53.9%)	3/8 (37.5%)	1/100 (100.0%)‡	6/18 (33.3%)
Liver	6/13 (46.2%)	1/8 (12.5%)	0/0 (0.0%)	2/18 (11.1%)
Lung	6/13 (46.2%)	2/8 (25.0%)	0/0 (0.0%)	2/18 (11.1%)

^*mice with no palpable or post-mortem pathological prostate tumors at $16 weeks of age

^**unpublished data, B.A. Foster lab (RPCI).

^‡36 week-old Src^−/−;TRAMP mouse first palpated at 32 weeks with a lower abdominal mass (assumed to be a prostate tumor), that presented with a 5 g pelvic LN mass, renal metastases and extensive abdominal ascites, yet no prostate tumor upon pathological examination.

Figure 3. AD lung metastases in SFK^−/−;TRAMP or TRAMP mice lacking primary-site tumors.

(A) Lung and VP lobe pairs from individual >33 week-old SFK^−/−;TRAMP or TRAMP mice stained as in Fig. 2B, showing either hyperplastic or HG-PIN lesions in the VP lobe simultaneous with AD lung metastases. Note that this Src^−/−;TRAMP case had one small tumor each in the DP (arrow) and VP lobes and thus, was not scored as prostate tumor-free in the Src^−/−;TRAMP column in Table 5. All images are at 5X magnification. (B) Cytokeratin and synaptophysin staining in examples of the metastases cited in Table 5, scored as the number of prostate tumor-free mice over total number of mice studied per genotype, with percentages in parentheses. *, see Src^−/−;TRAMP case above.

DISCUSSION

An increasing number of studies demonstrate that CR-CaP, the lethal phenotype of prostate cancer, is driven by the activation of AR in the absence of serum androgen levels. Data showing that AR can be activated by direct phosphorylation by Ack1 and SFK non-receptor tyrosine kinases, whose activities increase during CaP progression, strongly suggests that this pathway may help drive CR-CaP progression. Indeed, activated versions of these kinases are sufficient to induce androgen-independent growth of CaP cell lines (8, 9), and activated Src plus AR overexpression can induce CaP in tissue recombination models using primary prostate epithelial cells (14, 15). A study by Cai et al. (14) ranked Src as most able to induce CaP initiation using this latter system, with Fyn and then Lyn ranking progressively less active.

The current study is the first to address whether specific SFK members are required for CaP initiation, progression of NE tumors, and metastatic potential. In agreement with the tissue recombination studies, Src was found to be the most potent CaP inducer. This correlated with our finding that relative AR-poY was decreased only in early Src-null;TRAMP CaP lesions (8 weeks), likely to reflect AR-driven AD lesions. Moreover, our study indicates that Src and Lyn are equally critical to the formation of distal macrometastases in the TRAMP model. Importantly, the suppression of AD and NE formation in Src^−/−;TRAMP mice and to a lesser extent, in Lyn^−/−;TRAMP mice, was not due to the upregulation of other SFK members, but rather likely due to the loss of Src or Lyn. Fyn seemed not to be important for CaP initiation during the 16–34 week post-birth phase. In contrast, progression to NE disease, which is AR-independent, was not affected by the loss of Src, Lyn or Fyn. This strengthens the notion that the role of SFK is to promote AR-driven AD disease, and indeed, AD-CaP initiation was delayed in TRAMP mice lacking Src or Lyn.

The loss of Src, and to a lesser extent the loss of Lyn, suppressed formation of NE micro- and macro-metastases in various organs compared to TRAMP controls. Although in the case of Src^−/−;TRAMP mice, this might be attributed to a severe delay in general disease onset, the dearth of metastases in the Lyn^−/−;TRAMP group occurred at a period in which primary tumor formation was statistically comparable to that of TRAMP control. This strongly suggests that Lyn, and possibly Src, play critical roles in metastasis formation in the TRAMP model.

An interesting phenomenon was noted, namely that a minority of late-stage Lyn^−/−;TRAMP or Fyn^−/−;TRAMP mice suffered from aggressive metastatic disease in the absence of primary-site tumors. These mice displayed prostatic hyperplasias or intraepithelial neoplasias whereas their metastases were predominantly AD lesions based on staining for cytokeratins but not for synaptophysin. It is unclear whether Src also suppresses this process: only one Src^−/−;TRAMP mouse, which was shown ultimately to lack a prostate tumor, presented with extensive AD metastases. This suggests that AD metastases arise from cells that disseminated early, and then, at a low frequency, progress oncogenically at distal sites. Moreover, the lack of primary-site tumors in these cases suggests some sort of suppressive cross-talk between the AD metastases and tumor initiating cells in the prostate. Nonetheless, the mechanism by which Lyn and Fyn (and possibly, Src) inhibit formation of these AD metastases requires further analysis. Indeed, metastatic colonization (33), recruitment of endothelial cells to metastatic sites (34) and tumor invasiveness (35) require Src activity, whereas activation of Src pathways suppresses tumor dormancy, likely by favoring activation of ERK vs. P38 MAP kinase pathways (36). This suggests that SFK might suppress the proliferation of already disseminated dormant cells, such that in SFK-null backgrounds, the incidence of AD-metastases in mice lacking primary-site tumors increases. Thus, although the preponderance of our data suggests that the therapeutic targeting of SFK would prevent CaP initiation, progression and metastasis formation based on the TRAMP model, there is worry that this might derepress the dormancy of disseminated AD-initiating tumor cells.