Aquaporin 5 expression is frequent in prostate cancer and shows a dichotomous correlation with tumor phenotype and PSA recurrence

    Aquaporin 5 expression is frequent in prostate cancer and shows a dichotomous correlation with tumor phenotype and PSA recurrence Alexandra Pust MD, Dominik Kylies, Claudia Hube-Magg PhD, Martina Kluth, Sarah Minner MD, Christina Koop, Tobias Grob PhD, MD, Markus Graefen MD, Georg Salomon MD, Maria Christina Tsourlakis MD, Jakob Izbicki MD, Corinna Wittmer MD, Hartwig Huland MD, Ronald Simon PhD, Waldemar Wilczak MD, Guido Sauter MD, Stefan Steurer MD, Till Krech MD, Thorsten Schlomm MD, Nathaniel Melling MD PII: DOI: Reference:

S0046-8177(15)00388-3 doi: 10.1016/j.humpath.2015.09.026 YHUPA 3712

To appear in:

Human Pathology

Received date: Revised date: Accepted date:

24 April 2015 2 September 2015 19 September 2015

Please cite this article as: Pust Alexandra, Kylies Dominik, Hube-Magg Claudia, Kluth Martina, Minner Sarah, Koop Christina, Grob Tobias, Graefen Markus, Salomon Georg, Tsourlakis Maria Christina, Izbicki Jakob, Wittmer Corinna, Huland Hartwig, Simon Ronald, Wilczak Waldemar, Sauter Guido, Steurer Stefan, Krech Till, Schlomm Thorsten, Melling Nathaniel, Aquaporin 5 expression is frequent in prostate cancer and shows a dichotomous correlation with tumor phenotype and PSA recurrence, Human Pathology (2015), doi: 10.1016/j.humpath.2015.09.026

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Aquaporin 5 expression is frequent in prostate cancer and shows a dichotomous correlation with tumor phenotype and PSA recurrence Alexandra Pust MD a*, Dominik Kyliesa*, Claudia Hube-Magg PhD a, Martina Klutha, Sarah

T

Minner MD a, Christina Koopa, Tobias Grob PhD, MD a, Markus Graefen MD c, Georg

RI P

Salomon MD c, Maria Christina Tsourlakis MD a, Jakob Izbicki MD b, Corinna Wittmer MD a, Hartwig Huland MD c, Ronald Simon PhD a#, Waldemar Wilczak MD a, Guido Sauter MD a,

SC

Stefan Steurer MD a, Till Krech MD a, Thorsten Schlomm MD c,d, Nathaniel Melling MD a,b

Institute of Pathology, University Medical Center Hamburg-Eppendorf, Germany

b

General, Visceral and Thoracic Surgery Department and Clinic, University Medical Center Hamburg-Eppendorf, Germany

c

Martini-Clinic, Prostate Cancer Center, University Medical Center Hamburg-Eppendorf, Germany

d

MA NU

a

Department of Urology, Section for translational Prostate Cancer Research, University

ED

Medical Center Hamburg-Eppendorf, Germany

Corresponding author: Dr. Ronald Simon, Institute of Pathology, University Medical

CE

#

PT

* These authors contributed equally to this work.

Center Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany, Tel: +49 40 7410

AC

57214, FAX +49 40 7410 55997, E-mail: [email protected]

1

ACCEPTED MANUSCRIPT

Summary Aquaporin 5 (AQP5) is an androgen-regulated member of a family of small hydrophobic

T

integral trans membrane water channel proteins regulating cellular water homeostasis and

RI P

growth signaling. To evaluate its clinical impact and relationship with key genomic alterations in prostate cancer, AQP5 expression was analyzed by immunohistochemistry on a tissue microarray containing 12,427 prostate cancers. The analysis revealed weak to moderate

SC

immunostaining in normal prostate epithelium. In prostate cancers AQP5 staining levels were more variable and also included completely negative and highly overexpressing cases.

MA NU

Negative, weak, moderate, and strong AQP5 staining was found in 25.0%, 32.5%, 32.5% and 10.0% of 10,239 interpretable tumors. Comparison of AQP5 expression levels with tumor characteristics showed a dichotomous pattern with both high and low staining levels being linked to unfavorable tumor phenotype. AQP5 was negative in 28%, 23%, 24% and 35% of tumors with Gleason score ≤3+3, 3+4, 4+3 and ≥4+4, while the rate of strongly positive cases continuously increased from 7.0% over 10.0% and 12.0% to 13.0% in cancers

ED

with Gleason score ≤3+3, 3+4, 4+3 and ≥4+4. AQP5 expression was also related to ERG positivity and PTEN deletion (p<0.0001 each). Strong AQP5 positivity was seen in 15.5% of

PT

ERG positive and 5.8% of ERG negative cancers (p<0.0001) as well as in 14.7% of cancers with PTEN deletion and 9.4% of cancers without PTEN deletion. Remarkably, both negativity and strong positivity of AQP5 were linked to unfavorable disease outcome. This was

CE

however only seen in subgroups defined by TMPRSS2:ERG fusion and/or PTEN deletion. In summary, AQP5 can be both overexpressed and lost in subgroups of prostate cancers. Both

AC

alterations are linked to unfavorable outcome in molecularly defined cancer subgroups. It is hypothesized that this dichotomous role of AQP5 is due to two highly different mechanisms as to how the protein can influence cancer cells, i.e. hydraulic motility regulation and Ras/MAPK pathway activation. Keywords: AQP5; prostate cancer; tissue microarray; biochemical recurrence

2

ACCEPTED MANUSCRIPT

T

1. Introduction

RI P

Prostate cancer is the most prevalent cancer in men in Western societies (1). Although the majority of prostate cancers behave in an indolent manner, a small subset is highly aggressive and requires extensive treatment (2, 3). Established prognostic parameters are

SC

limited to Gleason grade and tumor extent on biopsies, preoperative prostate-specific antigen (PSA), and clinical stage. Although these data are statistically powerful, they are often

MA NU

insufficient for optimal treatment decisions in individuals. It is hoped that a better understanding of disease biology will eventually lead to the identification of clinically applicable molecular markers that enable a more reliable prediction of prostate cancer aggressiveness.

Aquaporins (AQPs), a family of 13 small hydrophobic integral trans membrane water channel

ED

proteins (4, 5), have recently gained interest as potential targets for novel antitumor therapy (6-10). AQPs are expressed in basically all normal tissues (http://www.genecards.org/cgiwhere they contribute to regulation

PT

bin/carddisp.pl?gene=AQP5&search=c2d4c0972925c4d4b566bb9ca45e3c28),

of cellular water homeostasis (11). It has been suggested that cancer cells take advantage of this mechanism to form hydraulic motile protrusions, which facilitate cell migration and

CE

invasive tumor growth (12-16). In addition, activated AQPs have been shown to induce the Ras/MAPK growth signaling pathway in vitro (14). In support of a tumor relevant role of

AC

AQPs, altered expression of some of them has been linked to adverse phenotype in many cancers, such as breast cancer (16-18), hepatocellular carcinoma (19), colorectal carcinoma (13, 20), chronic myelogenous leukemia (21), lung cancer (22, 23), ovarian cancer (15, 24), cervical cancer (25) and squamous cell carcinoma (10). In prostate cancer AQP5 is of particular interest, because it is regulated in an androgen receptor (AR) dependent manner (26). A tumor relevant role of AQP5 in prostate cancer is also supported by one study on 60 prostate cancers suggesting a link between AQP5 expression and patient outcome (27).

Based on these promising findings, we sought to clarify the value of AQP5 as a prognostic marker in prostate cancer using our preexisting large prostate cancer tissue microarray (TMA) built from tumor samples of more than 12,000 individual prostate cancer patients. The database attached to this TMA contains comprehensive molecular, pathological and clinical follow up data. Our findings suggest complex interactions with other androgen and nonandrogen dependent pathways with impact on clinical outcome depending on the specific molecular background of prostate cancers. 3

ACCEPTED MANUSCRIPT

2. Materials and Methods

2.1 Patients

T

Radical prostatectomy specimens were available from 12,427 patients, undergoing

RI P

surgery between 1992 and 2012 at the Department of Urology and the Martini Clinics at the University Medical Center Hamburg-Eppendorf. Follow-up data were available for a total of 12,344 patients with a median follow-up of 36 months (range: 1 to 241

SC

months; Table 1). Prostate specific antigen (PSA) values were measured following surgery and PSA recurrence was defined as the time point when postoperative PSA

MA NU

was at least 0.2ng/ml and increasing at subsequent measurements. All prostate specimens were analyzed according to a standard procedure, including a complete embedding of the entire prostate for histological analysis (28). The TMA manufacturing process was described earlier in detail (29). In short, one 0.6mm core was taken from a representative tissue block from each patient. The tissues were distributed among 27 TMA blocks, each containing 144 to 522 tumor samples. For

ED

internal controls, each TMA block also contained various control tissues, including normal prostate tissue. The molecular database attached to this TMA contained

PT

results on ERG expression in 10,711 ((30)), ERG break apart FISH analysis in 7,122 (expanded from (31)) and deletion status of 5q21 (CHD1) in 7,932 (expanded from

CE

(32)), 6q15 (MAP3K7) in 6,069 (expanded from (33)), PTEN (10q23) in 6,704 (expanded from (34)) and 3p13 (FOXP1) in 7,081 (expanded from (35)) cancers. Analysis of patient and corresponding histopathological data for research purposes,

AC

as well as construction of tissue microarrays from archived diagnostic left-over tissues, was approved by local laws (HmbKHG, §12,1) and by the local ethics committee (Ethics commission Hamburg, WF-049/09 and PV3652). All work was carried out in compliance with the Helsinki Declaration. 2.2 Immunohistochemistry Freshly cut TMA sections were immunostained on one day and in one experiment. Slides were deparaffinized and exposed to heat-induced antigen retrieval for 5 minutes in an autoclave at 121°C in pH 7,8 Tris-EDTA-Citrate buffer. Primary antibody specific for AQP5 (rabbit monoclonal antibody (clone EPR3747), Abcam, Cambridge, UK; cat#92320; dilution 1:450) was applied at 37°C for 60 minutes. Bound antibody was then visualized using the EnVision Kit (Dako, Glostrup, Denmark) according to the manufacturer’s directions. The staining intensity and the fraction of stained tumor cells were recorded in each tissue spot. A final IHC score was built from these two parameters according to the following criteria as previously

4

ACCEPTED MANUSCRIPT described (36, 37): negative (no staining at all), weak (1+ staining in ≤70% of tumor cells, or 2+ staining in ≤30% of tumor cells), moderate (1+ staining in >70% tumor cells, or 2+ staining in >30% but ≤70% of tumor cells, or 3+ staining in ≤30% tumor cells), and strong (2+ staining in >70% tumor cells or 3+ staining in >30% tumor Statistics

RI P

2.3

T

cells).

Statistical calculations were performed with JMP® 10.0.2 software (SAS Institute Inc., NC, USA). Contingency tables and the chi²-test were performed to search for

SC

associations between molecular parameters and tumor phenotype. Survival curves were calculated according to Kaplan-Meier. The Log-Rank test was applied to detect

MA NU

significant differences between groups. Analysis of variance (ANOVA) test was applied to search for associations between cell proliferation and AQP5 staining. 3. Results

3.1 Technical issues

ED

A total of 10,239 (82%) of tumor samples were interpretable in our TMA analysis. Reasons for non-informative cases (2,188; 18%) included lack of tissue samples or

PT

absence of unequivocal cancer tissue in the TMA spot. 3.2 AQP5 expression in normal and prostate cancer cells

CE

Normal prostate luminal cells typically showed weak or moderate AQP5 staining, while basal cells were negative. AQP5 immunostaining was localized in the cytoplasm of invasive prostate cancers and was deemed positive in 7,677 of our 10,239 (75%)

AC

interpretable prostate cancers (weak 32.5%, moderate 32.5%, strong 9.9% of cancers). Representative images of negative and positive AQP5 immunostainings are given in Fig. 1. 3.3 Association with TMPRSS2:ERG fusion status and ERG protein expression To evaluate whether AQP5 expression is associated with ERG status in prostate cancers, we used data from previous studies (expanded from (30, 31). Data on TMPRSS2:ERG fusion status obtained by FISH were available from 6,022 and by immunohistochemistry from 8,947 tumors with evaluable AQP5 immunostaining. Data on both ERG FISH and IHC were available from 5,796 cancers, and an identical result (ERG IHC positive and break by FISH or ERG IHC negative and missing break by FISH) was found in 5,532 of 5,796 (95.4%) cancers. AQP5 immunostaining was significantly more frequent in TMPRSS2:ERG rearranged and ERG-positive prostate cancers than in tumors without TMPRSS2:ERG fusion and ERG-negative tumors. Positive AQP5 immunostaining was seen in 88.5% (ERG IHC) and 89.0% (ERG

5

ACCEPTED MANUSCRIPT FISH) of ERG-positive cancers but in only 65.8% and 69.1% of cancers without ERG staining or ERG rearrangement (p<0.0001 each; Fig. 2). 3.4 Associations with tumor phenotype Interestingly, strong AQP5 staining intensity but also lack of staining were both linked

T

to adverse clinico-pathological parameters such as high Gleason grade and

RI P

advanced pathological tumor stage (p<0.0001 each, Table 2), lymph node positivity (p=0.0487) and preoperative PSA levels (p=0.0002) when all prostate cancers were analyzed. No link was found to the surgical resection margin status (p=0.16). Similar

SC

associations were also found for high Gleason grade and advanced pathological tumor stage (p<0.0001, p=0.01 respectively, Supplementary Table 1) in the subgroup

MA NU

of ERG-negative cancers. In ERG-positive cancers, negative and strong AQP5 staining was linked to all the evaluated parameters (high Gleason grade, advanced tumor stage, lymph node metastasis (p<0.0001 each), preoperative PSA levels (p=0.0003), surgical resection margin (p=0.02; Supplementary Table 2)). 3.5 Association with other key genomic deletions

Earlier studies had provided evidence for distinct molecular subgroups of prostate

ED

cancers defined by TMPRSS2:ERG fusions and several genomic deletions. Previous studies by others and us described a strong link of PTEN and 3p13 deletions to ERG

PT

positivity and of 5q21 and 6q15 deletions to ERG negativity (32-35). To study, whether AQP5 expression might be particularly associated with one of these genomic

CE

deletions, AQP5 data were compared to preexisting findings on PTEN (10q23), 3p13 (FOXP1), 6q15 (MAP3K7) and 5q21 (CHD1) deletions. In the analysis of all tumors, AQP5 expression proved to be significantly linked to all four genomic deletions

AC

mentioned above (PTEN, 6q15 and 5q21 p<0.0001 each, 3p13 p=0.0056; Fig. 3A). For PTEN and 3p13 deletions the link was established with positive AQP5 staining and for 6q15 and 5q21 deletions a correlation with negative staining was found. In the subgroup of ERG-negative cancers the associations were reduced to deletions of PTEN and 6q15 (p=0.0003 and p=0.0319 respectively; Fig 3B), but in the same manner as described for all cancers. The subgroup of ERG-positive cancers revealed that negative AQP5 staining was associated with deletions of 3p13 (p<0.0001; Fig. 3C), which represents inverse behavior compared to the link found for all cancers. 3.6 Association with tumor cell proliferation (Ki67 labeIing index) High-level AQP5 staining was significantly linked to increased cell proliferation as measured by Ki67 labeling index in all cancers (p<0.0001) as well as in cancers of identical Gleason score (p<0.001 each; Table 3) except for the smallest subgroup of cancers with Gleason score ≥4+4 (p=0.71). 3.7 Association with PSA recurrence

6

ACCEPTED MANUSCRIPT Follow-up data were available for 9,631 patients with interpretable AQP5 immunostaining on the TMA. A convincing association between AQP5 expression and early PSA recurrence was not seen in all tumors despite statistical significance (p=0.03; Fig. 4A). In the subset of ERG- negative tumors strong AQP5 staining was

T

linked to early biochemical recurrence (p=0.009; Fig. 4B), while, surprisingly, PSA

RI P

recurrence in ERG- positive cancers was linked to lack of AQP5 staining (p=0.004; Fig. 4C). Cancers without PTEN deletion showed no correlation of staining intensity to PSA recurrence-free survival (p=0.15; Fig. 4D). Furthermore, cancers with PTEN

SC

deletion showed reduced PSA recurrence-free survival for both negative and strong staining (p=0.002; Fig. 4E) as opposed to weak and moderate staining. The most

MA NU

striking correlation between AQP5 staining and early PSA recurrence was found for the subset of ERG-negative cancers, in which negative and strong AQP5 expression proved to be prognostic factors even in cancers harboring deletions of PTEN, which itself is one of the strongest known prognosticators in prostate cancer (p<0.0001; Fig.

AC

CE

PT

ED

4F).

7

ACCEPTED MANUSCRIPT

4. Discussion Our results show that prostate cancers can display highly increased or markedly reduced AQP5 expression, both of which are linked to PSA recurrence in molecularly defined tumor

T

subsets, particularly in PTEN deleted cancers.

RI P

AQP5 staining was typically weak to moderate in normal prostatic epithelium and also in most cancers except for a fraction of tumors with obviously reduced expression (25%) or clear-cut overexpression (10.0%). That an overall frequency of 81.7% AQP5 positive

SC

prostate cancers was observed in an earlier study (27), coming close to the 75.0% AQP5 expressing tumors in our study, suggests that the experimental conditions were comparable.

MA NU

As Li et al analyzed large sections, this concordance indirectly also argues against substantial tissue heterogeneity of AQP5 expression, which would have led to markedly lower rates of positivity in our tissue microarray analysis (27). Comparison of AQP5 expression with tumor recurrence revealed an uncommon, dichotomous relationship, which appears to be dependent of the cellular microenvironment, i.e. presence of TMPRSS2:ERG fusion and/or PTEN deletion. Interpretation of this

ED

observation is difficult. It might be speculated, that this unusual impact on outcome is related to the two distinct cellular functions of AQP5. It has been suggested that AQP5 both induces

PT

hydrostatic pressure changes contributing to localized cell membrane protrusions facilitating cell motility (38) but also triggers cell proliferation via activation of the Ras/MAPK growth

CE

pathway (14). It appears possible that the role of AQP5 in regulating these mechanisms is differently or even inversely influenced by ERG activation and/or PTEN inactivation. Interaction of PTEN and AQP5 is supported by experimental evidence (39, 40) showing that

AC

the small GTPase Ras related protein 1 (Rap1) has a regulatory effect on pathways involving both PTEN and AQP5. A critical role of the cellular microenvironment on the prognostic impact of AQP5 is also supported by earlier studies linking its overexpression to unfavorable prognosis in some tumor types but to favorable clinical outcome in others. AQP5 expression is associated with poor prognosis in breast cancer (16-18), hepatocellular carcinoma (19), colorectal carcinoma (13, 20), chronic myelogenous leukemia (21), lung cancer (22, 23), ovarian cancer (24) (15), cervical cancer (25) and oral squamous cell carcinoma (10), but to better outcome in gallbladder (41) and biliary tract cancer (42). AQP5 is an example of a biomarker that shows variable prognostic value only in specific molecular subsets. This is not uncommon. We have earlier identified other markers (e.g. mTOR (43), SEC14L1 (44), NY-ESO (45) with prognostic impact limited to a particular molecular environment, such as cancers with TMPRSS2:ERG fusion and PTEN deletion in the above mentioned cases. Although AQP5 does not appear to be a clinically suitable

8

ACCEPTED MANUSCRIPT prognostic marker in prostate cancer, it is noteworthy that some AQPs have recently gained interest as potential targets for novel anti-cancer therapies. In vitro models suggest that inhibition of AQP3 and 8, for example by small interfering RNA (8) or growth pathway kinase inhibitors (7, 9, 46), decreases cell proliferation and migration in cell lines obtained from

T

pancreatic, gastric and esophageal cancers. Moreover, knock down of AQP1 and AQP3

RI P

conferred impaired angiogenesis and resistance to skin cancers, respectively, in mouse models (41). The high prevalence of AQP5 expression in prostate cancer suggests that

SC

AQP5 inhibiting molecules could be tested in prostate cancer once they become available.

In summary, our results demonstrate that AQP5 can be both up-regulated and down-

MA NU

regulated in prostate cancers and that both of these conditions can lead to unfavorable clinical outcome in certain molecularly defined prostate cancer subgroups. Acknowledgements

We are grateful to Janett Lütgens, Sünje Seekamp, Inge Brandt, Bianca Kelp and Anne

Hanseatic City of Hamburg.

[1]

PT

References

ED

Meyer for excellent technical assistance. This work was supported by the Free and

Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA: a cancer journal for

[2]

CE

clinicians 2010;60:277-300.

Wilt TJ, Brawer MK, Jones KM, et al. Radical prostatectomy versus observation for

[3]

AC

localized prostate cancer. N Engl J Med 2012;367:203-213. Thompson IM, Jr., Tangen CM. Prostate cancer--uncertainty and a way forward. N

Engl J Med 2012;367:270-271. [4]

Agre P, Kozono D. Aquaporin water channels: molecular mechanisms for human

diseases. FEBS Lett 2003;555:72-78. [5]

Yakata K, Hiroaki Y, Ishibashi K, et al. Aquaporin-11 containing a divergent NPA motif

has normal water channel activity. Biochim Biophys Acta 2007;1768:688-693. [6]

Liu W, Wang K, Gong K, Li X, Luo K. Epidermal growth factor enhances MPC-83

pancreatic cancer cell migration through the upregulation of aquaporin 3. Mol Med Rep 2012;6:607-610. [7]

Chang H, Shi YH, Talaf TK, Lin C. Aquaporin-8 mediates human esophageal cancer

Eca-109 cell migration via the EGFR-Erk1/2 pathway. Int J Clin Exp Pathol 2014;7:76637671. [8]

Kusayama M, Wada K, Nagata M, et al. Critical role of aquaporin 3 on growth of

human esophageal and oral squamous cell carcinoma. Cancer Sci 2011;102:1128-1136. 9

ACCEPTED MANUSCRIPT [9]

Wang J, Gui Z, Deng L, et al. c-Met upregulates aquaporin 3 expression in human

gastric carcinoma cells via the ERK signalling pathway. Cancer Lett 2012;319:109-117. [10]

Ishimoto S, Wada K, Usami Y, et al. Differential expression of aquaporin 5 and

aquaporin 3 in squamous cell carcinoma and adenoid cystic carcinoma. Int J Oncol

Verkman AS. More than just water channels: unexpected cellular roles of aquaporins.

RI P

[11]

T

2012;41:67-75.

J Cell Sci 2005;118:3225-3232.

Verkman AS, Hara-Chikuma M, Papadopoulos MC. Aquaporins--new players in

cancer biology. J Mol Med (Berl) 2008;86:523-529. [13]

Kang SK, Chae YK, Woo J, et al. Role of human aquaporin 5 in colorectal

MA NU

carcinogenesis. Am J Pathol 2008;173:518-525. [14]

SC

[12]

Woo J, Lee J, Kim MS, et al. The effect of aquaporin 5 overexpression on the Ras

signaling pathway. Biochem Biophys Res Commun 2008;367:291-298. [15]

Yan C, Yang J, Shen L, Chen X. Inhibitory effect of Epigallocatechin gallate on

ovarian cancer cell proliferation associated with aquaporin 5 expression. Arch Gynecol Obstet 2012;285:459-467.

Lee SJ, Chae YS, Kim JG, et al. AQP5 expression predicts survival in patients with

ED

[16]

early breast cancer. Ann Surg Oncol 2014;21:375-383. Jung HJ, Park JY, Jeon HS, Kwon TH. Aquaporin-5: a marker protein for proliferation

PT

[17]

and migration of human breast cancer cells. PLoS One 2011;6:e28492. Shi Z, Zhang T, Luo L, et al. Aquaporins in human breast cancer: identification and

CE

[18]

involvement in carcinogenesis of breast cancer. J Surg Oncol 2012;106:267-272. [19]

Guo X, Sun T, Yang M, et al. Prognostic value of combined aquaporin 3 and

[20]

AC

aquaporin 5 overexpression in hepatocellular carcinoma. Biomed Res Int 2013;2013:206525. Wang W, Li Q, Yang T, et al. Expression of AQP5 and AQP8 in human colorectal

carcinoma and their clinical significance. World J Surg Oncol 2012;10:242. [21]

Chae YK, Kang SK, Kim MS, et al. Human AQP5 plays a role in the progression of

chronic myelogenous leukemia (CML). PLoS One 2008;3:e2594. [22]

Zhang Z, Chen Z, Song Y, et al. Expression of aquaporin 5 increases proliferation

and metastasis potential of lung cancer. J Pathol 2010;221:210-220. [23]

Chae YK, Woo J, Kim MJ, et al. Expression of aquaporin 5 (AQP5) promotes tumor

invasion in human non small cell lung cancer. PLoS One 2008;3:e2162. [24]

Yang JH, Shi YF, Cheng Q, Deng L. Expression and localization of aquaporin-5 in the

epithelial ovarian tumors. Gynecol Oncol 2006;100:294-299. [25]

Zhang T, Zhao C, Chen D, Zhou Z. Overexpression of AQP5 in cervical cancer:

correlation with clinicopathological features and prognosis. Med Oncol 2012;29:1998-2004.

10

ACCEPTED MANUSCRIPT [26]

Moehren U, Denayer S, Podvinec M, Verrijdt G, Claessens F. Identification of

androgen-selective androgen-response elements in the human aquaporin-5 and Rad9 genes. Biochem J 2008;411:679-686. [27]

Li J, Wang Z, Chong T, et al. Over-expression of a poor prognostic marker in prostate

Schlomm T, Iwers L, Kirstein P, et al. Clinical significance of p53 alterations in

RI P

[28]

T

cancer: AQP5 promotes cells growth and local invasion. World J Surg Oncol 2014;12:284.

surgically treated prostate cancers. Mod Pathol 2008;21:1371-1379. [29]

Kononen J, Bubendorf L, Kallioniemi A, et al. Tissue microarrays for high-throughput

[30]

SC

molecular profiling of tumor specimens. Nat Med 1998;4:844-847.

Weischenfeldt J, Simon R, Feuerbach L, et al. Integrative genomic analyses reveal an

MA NU

androgen-driven somatic alteration landscape in early-onset prostate cancer. Cancer Cell 2013;23:159-170. [31]

Minner S, Enodien M, Sirma H, et al. ERG Status Is Unrelated to PSA Recurrence in

Radically Operated Prostate Cancer in the Absence of Antihormonal Therapy. Clinical cancer research 2011;17:5878-5888. [32]

Burkhardt L, Fuchs S, Krohn A, et al. CHD1 Is a 5q21 Tumor Suppressor Required for

[33]

ED

ERG Rearrangement in Prostate Cancer. Cancer Research 2013;73:2795-2805. Kluth M, Hesse J, Heinl A, et al. Genomic deletion of MAP3K7 at 6q12-22 is

PT

associated with early PSA recurrence in prostate cancer and absence of TMPRSS2:ERG fusions. Modern pathology 2013;26:975-983. Krohn A, Diedler T, Burkhardt L, et al. Genomic deletion of PTEN is associated with

CE

[34]

tumor progression and early PSA recurrence in ERG fusion-positive and fusion-negative prostate cancer. Am J Pathol 2012;181:401-412. Krohn A, Seidel A, Burkhardt L, et al. Recurrent deletion of 3p13 targets multiple

AC

[35]

tumour suppressor genes and defines a distinct subgroup of aggressive ERG fusion-positive prostate cancers. J Pathol 2013;231:130-141. [36]

Simon R, Mirlacher M, Sauter G. Immunohistochemical analysis of tissue

microarrays. Methods in molecular biology 2010;664:113-126. [37]

Minner S, Wittmer C, Graefen M, et al. High level PSMA expression is associated with

early PSA recurrence in surgically treated prostate cancer. Prostate 2011;71:281-288. [38]

Papadopoulos MC, Saadoun S, Verkman AS. Aquaporins and cell migration. Pflugers

Arch 2008;456:693-700. [39]

Huang Y, Wernyj RP, Norton DD, et al. Modulation of specific protein expression

levels by PTEN: identification of AKAP121, DHFR, G3BP, Rap1, and RCC1 as potential targets of PTEN. Oncogene 2005;24:3819-3829.

11

ACCEPTED MANUSCRIPT [40]

Ciuffreda L, Di Sanza C, Cesta Incani U, et al. The mitogen-activated protein kinase

(MAPK) cascade controls phosphatase and tensin homolog (PTEN) expression through multiple mechanisms. J Mol Med (Berl) 2012;90:667-679. [41]

Sekine S, Shimada Y, Nagata T, et al. Role of aquaporin-5 in gallbladder carcinoma.

Sekine S, Shimada Y, Nagata T, et al. Prognostic significance of aquaporins in

RI P

[42]

T

Eur Surg Res 2013;51:108-117.

human biliary tract carcinoma. Oncol Rep 2012;27:1741-1747. [43]

Muller J, Ehlers A, Burkhardt L, et al. Loss of pSer2448-mTOR expression is linked to

SC

adverse prognosis and tumor progression in ERG-fusion-positive cancers. Int J Cancer 2013;132:1333-1340.

Burdelski C, Barreau Y, Simon R, et al. Saccharomyces cerevisiae-like 1

MA NU

[44]

overexpression is frequent in prostate cancer and has markedly different effects in Etsrelated gene fusion-positive and fusion-negative cancers. Hum Pathol 2015;46:514-523. [45]

Grupp K, Ospina-Klinck D, Tsourlakis MC, et al. NY-ESO-1 expression is tightly linked

to TMPRSS2-ERG fusion in prostate cancer. Prostate 2014;74:1012-1022. [46]

Zhang L, Zhou F, Li Y, et al. Fas-associated factor 1 is a scaffold protein that

ED

promotes beta-transducin repeat-containing protein (beta-TrCP)-mediated beta-catenin

AC

CE

PT

ubiquitination and degradation. J Biol Chem 2012;287:30701-30710.

12

ACCEPTED MANUSCRIPT

Table 1 Pathological and clinical data of the arrayed prostate cancers No. of patients (%) Study cohort on Biochemical relapse TMA(n=12,427) among categories

T

Parameter

AC

CE

PT

ED

MA NU

SC

RI P

Follow-up (mo) n 11,665 (94) 2,769 (24) Mean 48.9 Median 36.4 Age (y) ≤50 334 (3) 81 (24) 51-59 3,061 (25) 705 (23) 60-69 7,188 (58) 1,610 (22) ≥70 1,761 (14) 370 (21) Pretreatment PSA (ng/ml) <4 1,585 (13) 242 (15) 4-10 7,480 (61) 1,355 (18) >10-20 2,412 (20) 737 (31) >20 812 (7) 397 (49) pT stage pT2 8,187 (66) 1,095 (13) pT3a 2,660 (22) 817 (31) pT3b 1,465 (12) 796 (54) pT4 63 (1) 51 (81) Gleason grade ≤3+3 2,983 (24) 368 (12) 3+4 6,945 (56) 1,289 (19) 4+3 1,848 (15) 788 (43) ≥4+4 584 (5) 311 (53) pN stage pN0 6,970 (91) 1,636 (24) pN+ 693 (9) 393 (57) Surgical margin Negative 9,990 (82) 1,848 (19) Positive 2,211 (18) 853 (39) Percentage in the column “Study cohort on TMA” refers to the fraction of samples across each category. Percentage in column “Biochemical relapse among categories” refers to the fraction of samples with biochemical relapse within each parameter in the different categories. Abbreviation: TMA tissue micro array

13

ACCEPTED MANUSCRIPT

Table 2 Association between AQP5 immunostaining and prostate cancer phenotype Evaluable (n)

Negative

All cancers

10239

25.0

32.5

Tumor stage pT2 pT3a pT3b-pT4

6655 2268 1276

26.0 22.8 23.6

33.0 31.5 32.0

Gleason grade ≤3+3 3+4 4+3 ≥4+4

2302 5806 1591 485

Lymph node metastasis N0 N+

5710 587

32.4 33.2 32.4

8.7 12.4 12.0

<0.0001

34.3 32.2 33.8 24.7

30.9 34.4 29.9 27.6

7.0 10.1 12.4 12.8

<0.0001

23.9 27.4

33.0 30.0

32.5 29.6

10.7 12.9

0.049

1267 6130 2033 705

22.4 24.5 26.2 30.1

31.9 32.3 34.4 31.2

36.4 33.0 29.3 30.6

9.3 10.2 10.1 8.1

0.0002

8184 1866

24.9 25.5

32.9 31.4

32.6 32.0

9.6 11.1

0.16

PT

SC

MA NU 27.8 23.3 23.9 34.8

RI P

9.9

ED

AC

Surgical margin negative positive

p value

32.5

CE

Preoperative PSA level (ng/ml) <4 4-10 >10-20 >20

T

AQP5 (%) Weak Moderate Strong

Parameter

14

ACCEPTED MANUSCRIPT

Table 3 Association between AQP5 immunostaining results and Ki67 labeling index.

pGleason ≤3+3 p<0.0001

1472 2024 2032

strong

559

negative weak moderate

347 460 450

strong

94

3.4

0.1

1.5 2.2 2.6

0.1 0.1 0.1

2.6

0.2

785 1168 1256

2.0 2.6 2.9

0.1 0.1 0.1

341

3.2

0.1

227 322 258

3.0 3.2 3.9

0.2 0.2 0.2

92

4.4

0.4

99 63 61

5.1 4.4 4.2

0.5 0.6 0.6

strong negative weak moderate

31 746 1073 1063

4.9 2.7 2.9 3.2

0.9 0.1 0.1 0.1

strong negative weak moderate

284 89 222 284

3.5 3.9 3.2 3.9

0.2 0.3 0.2 0.2

84

4.2

0.3

negative weak moderate

pGleason 3+4 p<0.0001

strong

strong

PT

negative weak moderate

AC

CE

pGleason ≥4+4 p=0.71

PTEN norm p=0.0001

PTEN del p=0.015

ED

negative weak moderate

pGleason 4+3 p=0.0009

T

negative weak moderate

MA NU

all p<0.0001

KI 67 labeling index Mean SD 2.3 0.1 2.7 0.1 3.0 0.1

n

RI P

AQP5

strong Abbreviation: SD standard deviation

SC

Parameter p value

15

ACCEPTED MANUSCRIPT Figure legends

Fig 1 Representative pictures of AQP5 immunostaining in prostate cancer with (A) negative,

T

(B) weak, (C) moderate and (D) strong staining

RI P

Fig 2 Association between AQP5 immunostaining and ERG-status (IHC/FISH) in all cancers Fig 3 Association between AQP5 immunostaining and 10q23 (PTEN), 5q21 (CHD1), 6q15

SC

(MAP3K7), 3p13 (FOXP1) deletions in (A) all cancers, in the (B) ERG-negative and the (C) ERG-positive subset

MA NU

Fig 4 Kaplan-Meier curves for prostate antigen (PSA) recurrence and AQP5 expression in (A) all cancers, (B) the ERG- negative subset, (C) the ERG- positive subset, (D) PTEN

AC

CE

PT

ED

normal cancers (E) PTEN deleted cancers and (F) ERG- positive and PTEN deleted cancers

16

Figure 1

AC

CE

PT

ED

MA NU

SC

RI P

T

ACCEPTED MANUSCRIPT

17

MA NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Figure 2

18

AC

Figure 3

CE

PT

ED

MA NU

SC

RI P

T

ACCEPTED MANUSCRIPT

19

AC

Figure 4

CE

PT

ED

MA NU

SC

RI P

T

ACCEPTED MANUSCRIPT

20