FIELD OF THE INVENTION

The invention relates to the field of medicine. More specifically, the invention relates to the identification of a specific genomic and proteomic signature of castration-resistant prostatic cells. In particular, the invention relates to in vitro methods of prognosing the outcome of prostate cancer comprising identifying prostatic cells exhibiting biomarkers from this specific signature. The invention also relates to the uses of said cells in screening methods for identifying new candidate therapies for prostate cancer and more specifically of hormone refractory prostate cancer.

INTRODUCTION

Prostate cancer (PCa) is the most common form of cancer in men and the second leading of cancer related deaths in western countries. At the stage of castration resistance, it remains an incurable disease with more than 300,000 deaths worldwide each year. Treatment options are chosen as a function of the stage of the disease which is determined taking into account notably blood PSA levels, prostate biopsy analyses, and the age of the patient. Management of the disease consequently varies from active surveillance (for low stage PCa) to treatment with curative intention (radical prostatectomy or radiotherapy) in case of clinically localized PCa and ultimately to hormonal therapy and/or chemotherapy.

However, even after treatment with curative intention, about 20-30% of men relapse after five to ten years, biochemical recurrence being detected by increase of PSA level.

Androgen deprivation therapy (ADT) aims to block prostate cancerous cells from getting dihydrotestosterone, hormone which is required for the growth and spread of most PCa cells. ADT has been shown to provide rapid and dramatic beneficial effects in the treatment of metastatic PCa and ADT is also used after PCa recurrence following radical prostatectomy or radiotherapy or as an adjunctive therapy for patients with locally advanced disease undergoing radiation therapy or surgery. Androgen deprivation is reached either through orchiectomy or drug based treatments comprising chemical castration using LHRH agonists or antagonists (aiming at lowering the synthesis of testosterone) or antiandrogen therapy (aiming at preventing androgen receptor signal transduction). While being considered as a gold standard therapy of invasive/metastatic or recurrent prostate cancer, ADT resistance to androgen blockade invariably develops within one to two years of treatment and eventually leads to patient death. There is considerable effort worldwide to understand the molecular mechanisms driving androgen blockade resistance and the emergence of castration-resistant prostate cancer (CRPC). Metastatic CRPC displays a high inter- and intra- patient heterogeneity by implying a mixture of cells displaying various range of Androgen Receptor (AR) expression levels and dependence thereon. Therefore, clonal selection in the course of treatment is a concern for the long-term use of AR inhibitors, though such mechanisms have not been formally proved. Furthermore, the origin of AR independent cancer cells is even more questioned.

WO2013149039 discloses the use of ex vivo organotypic cultures of human prostate cancerous cells and the identification of a group of specific markers of good prognostic regarding the risk of postoperative relapse.

A rare luminal progenitor cell type called LSC ^med (Lin ^" /Sca-l ⁺ /CD49f™ ^ed ) has been recently identified within wild-type (WT) mouse prostate epithelia. These luminal-like cells can generate 2D colonies, 3D spheres (1).

There is still a need for prognostic tool for identifying or predicting a risk of prostate cancer relapse and/or androgen deprivation resistance as well as for new therapeutic tools and targets for the prevention of castration-resistant prostate cancer.

SUMMARY

The present invention provides useful in vitro methods for prognosing the outcome of prostate cancer. The invention stems, inter alia, from the discovery by the Inventors that a particular prostate epithelial cell type is found predominantly in prostate tissue in several in vivo models of prostate tumorigenesis. Furthermore, this cell type population is found notably increased in a premalignant model under androgen deprivation condition thereby highlighting the clinical implication of these cells in castration resistance processes. Last, Inventors also surprisingly identify the tumorigenic properties of these cells in a cancerous context.

Inventors have been able to identify a unique gene signature of the cells, thus providing useful, effective, and reliable tools for identifying or predicting a risk of prostate cancer relapse and/or androgen deprivation resistance.

More particularly, an object of the invention relates to an in vitro method of prognosing the outcome of prostate cancer comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying the overexpression of at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl,

CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt4, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl and Wfdc2.

A particular object of the invention is an in vitro method of prognosing the outcome of prostate cancer comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying an overexpression of Krt4 gene.

A more particular object is the above in vitro method comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying an overexpression of Krt4 gene and an overexpression of at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl and Wfdc2.

Prognostic methods of the invention are particularly suited for identifying or predicting a risk of prostate cancer relapse or progression and/or androgen deprivation resistance from a biological sample of a subject, prostate sample being particularly preferred.

Overexpression of any of the above genes within said sample, can be easily detected either at the protein or RNA level by methods well known in the art. In a particular embodiment said prostatic epithelial cells represent at least 10% of cell exhibiting said overexpression. Alternatively, said methods can be implemented by the detection in said sample of an overexpression of any of said genes of at least 1.5-fold change in comparison with non-cancerous prostatic epithelial cells. Even more particularly said method can be implemented by immunohistochemical labelling of Krt4 encoded protein. In a particular embodiment said methods further comprise a step of detecting, in said prostate sample, loss of basal cells {e.g. through immunohistochemical labelling of a high molecular weight cytokeratin or p63) and/or a step of detecting cancerous cells (e.g. through immunohistochemical labelling of AMACR protein).

In a particular embodiment, prognosing methods of the invention are implemented in at least two samples collected from the same subject at a different time to detect a significant increase of the amount of said prostatic epithelial cells between these at least two samples.

In a particular embodiment, prognosing methods of the invention comprise detecting the risk, presence or progression of hormone refractory prostate cancer.

Furthermore, as shown by the Inventors, the clinical implication of these cells in prostate cancer evolution makes this cell type a valuable tool in the discovery of candidate compounds for the prevention and/or the treatment of androgen blockade resistant prostate cancer.

Another object of the invention thus resides in a method for selecting an active compound in the prevention and/or the treatment of androgen blockade resistant prostate cancer, said method comprising the use of mouse cancerous prostatic epithelial cells resistant to androgen blockade and displaying the overexpression of Krt4 gene and at least one gene selected from 1700016G22Rik, 4930594C411Rik, AA986860, Acsm3, Aldhla3, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, Cbr2, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll5, Cxcll7, Cyp2f2, D17H6S56E-5, Daf2, Defbl4, Defb2, Dusp4, Ednl, F5, Ffar4, Gfpt2, Gml2146, Gml2245, Gml2603, Gml2888, Gml3502, Gml4137, Gml4328, Gm23640, Gm24431, Gm6166, Gprl37b, Gprl37b-ps, Gprc5a, Gsdmc2, Gsdmc3, Gsdmc4, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3b, Reg3g, Rnu73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sprr2a2, Sprr2a3, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2b34, Usp6nl, Wfdcl8 and Wfdc2. A particular object of this invention also resides in a method of selecting an active compound in the prevention and/or the treatment of androgen blockade resistant prostate cancer, said method comprising the use of mouse cancerous prostatic epithelial cells resistant to androgen blockade and being Lin Sca-l ⁺ /CD49r ^ed or CK5 CK8 ⁺ /CK4 ⁺ .

Another object of the invention also resides in a method of selecting an active compound for the prevention and/or treatment of androgen blockade resistant prostate cancer, said method comprising the use of human cancerous prostatic cells resistant to androgen blockade and displaying the overexpression of Krt4 and at least one gene selected from : Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3,

C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl and Wfdc2.

BRIEF DESCRTTPION OF FIGURES Figure 1: LSC ^med are the most abundant epithelial cell type in intact and castrated Pten ^{pc /~} prostate tumors. Representative FACS profiles of ΡΤΕΙΨ^ ^' prostatic epithelial cells gated as Lin-CD49f ⁺ . A) FACS profile of prostate epithelial cells of non-castrated PTE ^{c /} mice. B) FACS profile of prostate epithelial cells of castrated PPEW ^{c /'} mice. Each FACS profile shows gated epithelial populations: basal/stem, LSC ^med (thick black square), and luminal cells. Percentages are noted under each gate name. C) graphical representation of cell type distribution. Bar graphs show mean (±SD) percentages of basal/stem (dark grey), LSC ^med (white bar) and luminal cells (light grey) in n=2-5 independent experiments consisting of 1-10 pooled prostates each.

Figure 2: LSC ^med are present in prostate of Pb-PRL and Wild Type mice, and proliferate after castration. Representative FACS profiles of Pb-PRL and Wild Type prostatic epithelial cells gated as Lin-CD49f ⁺ . A) and B) FACS profile of prostate epithelial cells of non-castrated and castrated Pb-PRL mice, respectively. D) and E) FACS profile of prostate epithelial cells of non- castrated and castrated Wild Type mice, respectively. Each FACS profile shows gated epithelial populations: basal/stem, LSC ^med (thick black square), and luminal cells. Percentages are noted under each gate name. C) and F) graphical representation of cell type distribution. Bar graphs show mean (±SD) percentages of basal/stem (dark grey), LSC ^med (white bar) and luminal cells (light grey) in n=4-5 independent experiments consisting of 1-10 pooled prostates each. ***, significant differences (P<0.001), two-way ANOVA with Sidak's multiple comparisons.

Figure 3: Validation of expression of LSC^-specific genes by qPCR on sorted cell populations from WT, intact and castrated Pb-PRL and intact and castrated Pten ^IK/~ mouse prostates. In the three models, a noticeable upregulation of these genes was observed in LSC ^med compared to basal/stem and luminal cell populations, thereby confirming that LSC ^med signature is maintained across models. Per group, n=2-5 samples of each population isolated from 1-10 pooled mice in independent experiments. Letters denote mouse groups showing significant differences (P<0.05) in a repeated-measures one-way ANOVA with Tukey's post-hoc. B: basal/stem; Lm: LSC ^med ; L: luminal cells.

Figure 4: Decreased AR signalling in LSC™* ¹ cells. Relative expression level of AR signaling genes as determined by Affymetrix transcriptomic analysis of androgen signaling target genes in the three prostate cell populations enriched from wild type prostate by flow cytometry. (A) representative genes that are up-regulated by androgens, (B) representative genes that are down- regulated by androgens.

Figure 5: Decreased AR signalling in Pb-PRL mice prostate. Gene expression of AR-target genes by qPCR in whole prostate tissue (dorsal lobe) of intact WT and Pb-PRL mice (n=3-4 unsorted prostate samples). Pb-PRL prostates are characterized by an increased expression of AR repressed gene (Igfbp3) and a decreased expression of AR activated genes. Data are expressed as a mean +SD; asterisks denote significant differences (*P<0.05; **P<0.001) in a two-tailed independent samples t-test.

Figure 6: LSC ^med are not enriched in Hi-myc prostates. (A) Representative FACS profiles of Lin-, CD49f ⁺ cells of Hi-Myc mouse prostates, showing gated epithelial populations: basal/stem, LSC ^med (thick black square), and luminal cells. Percentages are noted under each gate name. (B) Bar graph showing percentages of basal/stem (dark grey), LSC ^med (white bar) and luminal cells (light grey) for Hi-Myc prostates in an experiment with 3 pooled prostates.

Figure 7: LSC ^med are detected by CK4 staining and proliferate upon castration. (A) Representative IHC staining for CK4 in prostates of intact and castrated WT, Pb-PRL and Pten ^{pc /~} mice. Small black arrows indicate CK4 ⁺ cells in the four upper panels. A strongly positive region (large black arrow) and a mainly negative region (large white arrow) are shown for intact Pten ^{pc /~} tissue (bottom left panel). Insets show higher magnification images. (B) Representative images of serial tissue slides of prostates from intact (a-d) or castrated (e-n) WT and Pten ^{pc /~} mice (as indicated) stained for CK4 (left panels) and BrdU (right panels). Dotted lines encircle CK4 ⁺ areas and stars (panels g-j) identify CK4 ⁺ /BrdU ^" areas adjacent to encircled CK4 ⁺ /BrdU ⁺ regions (shown at higher magnifications in panels k-n). Scale bars in A and B (a-j): ΙΟΟμιη; in B (k-n): 20μιη.

Figure 8: CK4 staining in proximal vs distal regions of the prostate gland. (A) Schematic representation of proximal and distal regions of mouse prostate lobes depicted on a ½ prostate showing one dorsal, one lateral and one ventral lobe surrounding the urethra. (B) Representative IHC staining for CK4 in proximal and distal areas of WT and Pb-PRL prostate ventral lobe. Arrows show CK4 ⁺ cells. Scale bar in A: 2mm; in B: ΙΟΟμιη. Figure 9: CK4 staining in human prostate cancer samples. (A) Representative images of serial slides from PDXs of treatment-naive prostatic tumors in intact or castrated hosts, stained for AMACR/p63 and CK4 by IHC. AMACR-positive tumor areas are marked with dotted lines, and show positive staining for CK4. (B) Representative image of a tissue slide from a treatment-na ^' ive Gleason 4 primary prostate cancer stained for CK4.

Figure 10: CK7 staining in Pb-PRL and Pten ^pc~ ' mouse prostate. (A) Representative images of slides from Pb-PRL mouse prostatic tissues showing an intense positive staining for CK7 in distal and proximal glands. (B) Representative images of tissue slides of Pten ^{pc /~} prostate showing an extensive labelling of prostate tissue.

DESCRIPTION OF THE INVENTION

The invention relates to methods of prognosing the outcome of prostate cancer said methods comprising identifying epithelial prostatic cells exhibiting biomarkers identified by the inventors as constituting a specific signature of castration-resistant prostate cells that survive and even proliferate upon androgen blockade. The invention also relates to the uses of said cells in screening methods for identifying new candidate therapies for prostate cancer and more specifically of hormone refractory prostate cancer.

Definitions

As used herein prostate cancer refers to malignant prostate cancer and more particularly to adenocarcinomas.

"Subject" refers to any mammalian subject, preferably a human subject.

Within the context of the invention, the term "prognosing" the outcome of prostate cancer means identifying or assessing the risk of presence, recurrence, or progression of prostate cancer, in particular the risk, presence or progression of hormone refractory prostate cancer. More particularly, prognosis methods of the invention can be used to identify disease subtype, to measure the severity, to monitor the progression of the disease or the conversion of a prostate cancer toward a hormone refractory prostate cancer, to qualify the prostate cancer, and/or to assess the responsiveness or likeness of response of a subject to androgen blockade treatments. In a particular embodiment, prognosis methods of the invention can also be used for patients' stratification in the course of clinical trials or to assess the efficacy of a treatment. As explained above "recurrence of prostate cancer" refers to reappearance of the prostate cancer after a first treatment, said treatment being of any form as, for example, chemotherapy, radiotherapy, hormone therapy, orchiectomy, or a combination thereof.

The terms "hormone refractory prostate cancer", "hormone resistant prostate cancer", "castration-resistant prostate cancer" or "androgen blockade resistant prostate cancer" are herein used interchangeably.

The terms "gene expression signature", "cell signature", or "signature" refer to one or more gene product (typically RNA or protein) that is specifically differentially expressed in a cell type, cell population, biological sample or tissue as compared to a control cell, control cell population, control biological sample or control tissue. Therefore, the identification of such a signature within a cell, cell population, biological sample or tissue is useful for prognosing the outcome of a disease. "Biomarker" is also used to design a gene (i.e. any related transcript or protein) from said signature. More specifically, as used herein "LSC ^med (cell) signature" refers to one or more gene product that have been shown by the inventors as specifically differently expressed in LSC ^med cells when compared to prostate basal/stem cells and/or luminal cells.

"Detection of an overexpression of a gene" as used herein refers to the significant detection of the presence, and/or of an increase in quantity of a product from said gene, i.e. any transcript and/or corresponding protein (i.e. biomarker) in a sample of the subject to be tested as compared to a control. It can also refer to the detection, in a sample of said subject, of cells, or of an increased frequency thereof within a cell population, said cells being characterized by the presence and/or an increase in quantity of product(s) from these genes. In an embodiment, an increase is of at least 5% or even more in comparison with a control sample or reference value or mean value. In a preferred embodiment, increases may be of about 10%, 15%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% or 100% (or even more). In a particularly preferred embodiment said increase is of at least 50%. In another embodiment said presence or increase is characterized by the detectability of a biomarker in a sample of the subject whereas said biomarker is no to barely detectable in control sample. Said tested sample will be then ranked as "positive" for said biomarker, as for example in immunofluorescence, immunohistochemical or FACS assays. In another particular embodiment, a positive sample for any biomarker of the invention can also be characterized by a change of labelling features for said biomarker as for example, a different localization in (a) particular subcellular region(s), and/or alteration in the quality of the labelling (e.g. from a focused to a diffuse labelling or conversely).

The terms "subject suffering or suspected of suffering from prostate cancer" refers to any mammalian subject for which diagnostic of prostate cancer have been made, and/or presenting symptoms which are considered as making probable such a diagnostic. In a first instance, prostate cancer can be suspected from screening with a prostate-specific antigen (PSA) blood test and/or a digital rectal exam. A PSA serum level above the normal values being considered as an indication of probable prostate cancer, in conjunction with other possible symptoms or analyses. A continuous rise of PSA level in the same subject is also indicative of the progression of the disease. Also, rising of PSA levels signs the recurrence of the disease after a cancer therapy, when occurring after a previous treatment of the cancer. Anyway, only an analysis of prostatic tissue can lead to an actual diagnosis of prostate cancer. Such analysis is usually based on the assessment of cell morphology within a prostatic tissue and the assignment of a Gleason score to the tissue based on how much the cells looks like normal, tissue being ranking from 1 (normal tissue) to 5 (very abnormal tissue). Morphological assessment can be associated to an immunohistochemical labelling, in order to identify a loss of basal cells which signs cancer. Basal cells are usually labelled with antibodies directed to high molecular weight cytokeratines (CK 903) or CK5/6 or to p63, a loss of labelling thus allowing to diagnose cancer. Nevertheless, being based on absence of labelling, false positives could arise from technical biases or atypical lesions. Cytokeratin and/or p63 labelling can therefore be coupled to the labelling of AMACR (P504S) which has been found specific for tumour tissues (though not specific for prostatic tumour tissue); detection of CK903 or CK5/6 or p63 negative and AMACR positive foci allows to diagnose prostate cancer with certainty.

As used herein "biological sample" comprises any biopsy sample as incisional biopsy, excisional biopsy, or needle biopsy. "Biological sample" comprises also any autopsy samples, frozen samples dedicated to histologic analyses, fixed or wax embedded sample. As used herein the terms "biological sample" are preferably a prostate biopsy sample, but it can also be metastasis biopsy, or lymph node biopsy from a subject suffering or suspected to suffer from prostate cancer or of prostate cancer relapse. "Biological sample" can also comprise a prostate resection or a sample thereof, or prostatectomy samples. "Biological sample" also includes blood and blood fractions or blood-derived products (tumor cells, circulating tumor DNA/RNA, exosomes), urine and urine fractions or urine-derived products, as well as cells or cell lines or organoids or patient- derived xenografts (PDX) derived from patient prostate samples.

As used herein the terms, "prostatic epithelial cells", "prostate epithelial cells", "prostate epithelium" refers to the two main epithelial cell types, basal and luminal, as well as a minor population of neuroendocrine cells found in prostate. Of course, prostatic epithelial cells also comprise LSC ^med cells that are shown herein to share some common features with both basal and luminal cells, though being a very specific cell population. More particularly "prostatic epithelial cells" refers to cells that express cytokeratins and can be separated by cell sorting using cell-surface markers such as CD49f, Sca-1, CD24, CD44 or other cell-specific markers validated in the art.

As shown in the experimental section, Inventors have identified LSC ^med cells as an important therapeutic target for CRPC because they have progenitor properties, they increase proportionally upon disrupted AR-signalling, they withstand androgen blockade in vivo and they massively proliferate after castration in a relevant model of CRPC (Pten ^{pc /~} ). Inventors have surprisingly found that LSC ^med exhibit a unique gene expression signature that distinguishes them from prostate cell populations designated as luminal and basal/stem cells (Table 1) and defines these cells as a third distinct epithelial prostatic cell entity.

Table 1

LSC ^med signature

(genes overexpressed (>2-fold, p<0.05) compared to expression Enrichement

Enrichement levels found in basal/stem and luminal cells) in unsorted

in unsorted

ENTREZ Pten ^{pc /} -

Mouse Human Hi-Myc

Gene full name Gene ID: prostates* Gene Gene prostates ^†

(mouse) mouse/

symbol symbol

human

WAP four-disulfide 14038/

Wfdcl8 ANOS 1 YES NO core domain 18 3730

Cbr2 carbonyl reductase 2 nd 12409/nd YES NO brain expressed X- 19716/558

Bexl BEX1 NO NO linked 1 59

small proline-rich 100303744

Sppr2a2 nd YES NO protein 2A2 /nd

small proline-rich 100042514

Sprr2a3 nd YES NO protein 2A3 /nd

DNA segment, Chr

D17H6S56

17, human nd 110956/nd YES NO E-5

D6S56E/G7e

G protein-coupled

Gprl37b- receptor 137B, nd 664862/nd YES NO ps

pseudogene

101056162

Gm6166 predicted gene 6166 nd nd nd

/nd

Ids- predicted gene,

Gm24431 nd MGL54542 nd nd

24431/snoRNA gene

08/nd

nd Ids- predicted gene,

Gm23640 MGL54534 nd nd

23640/snRNA gene

17/nd

predicted gene nd

Gml4328 628298/nd nd nd

14328/pseudogene

Gml3502 predicted gene 13502 nd 207933/nd nd nd

1, subfamily A3 NO

NO

sulfurtransferase 100131187 Ly6/Plaur domain 68311/

Lypd2 LYPD2 NO NO

containing 2 137797

scavenger receptor 219151/

Scara3 SCARA3 YES NO

class A, member 3 51435

17536/

Meis2 Meis homeobox 2 MEIS2 NO NO

4212

complement 109828/

C7 C7 YES YES

component 7 730

immunoglobulin

207683/

Igsfll superfamily, member IGSF11 YES NO

152404

11

12223/

Btc betacellulin BTC YES NO

685

syndecan binding 228765/

Sdcbp2 SDCBP2 YES NO

protein (syntenin) 2 27111

dual specificity 319520/

Dusp4 DUSP4 YES NO

phosphatase 4 1846

U73B small nuclear SNORD73 19871/

Rnu73b nd nd

RNA B 114655

expressed sequence 212439/

AA986860 Clorfl l6 YES NO

AA986860 79098

623781/

Gml4137 predicted gene 14137 C15orf62 YES NO

643338

^' compared to WT (yes = found enriched in GSE46799 and/or GSE46473, no : found not enriched in any of GSE46799 or GSE46473) ; fcompared to WT (GO:GSE53202) ; nd : not determined ; Capital or lower case letters are used indifferently in gene names.

As shown in the experimental section LSC specific signature is conserved across mouse models and independent of the prostate pathological status. Strikingly, Inventors have found that LSC ^med cells represents the most abundant epithelial cell type in a model of invasive, castration- resistant, adenocarcinoma (Pten ^{pc /} ), and that this prevalence is not affected by castration. Furthermore, the LSC ^med part within prostate epithelium is significantly increased in WT mice and in Pb-PRL premalignant mouse model, under androgen deprivation condition. Furthermore, LSC ^med from Pten ^{pc /"} mice were found by the Inventors to exhibit tumorigenic properties. Altogether these results highlight the clinical implication of these cells in castration resistance and prostate cancer relapse.

Belonging to the LSC ^med specific signature, the above genes (i.e. any of their transcript or related protein) therefore represent valuable biomarkers, alone or any combination thereof, in the field of prostate cancer, and more specifically to detect the risk, presence or progression of hormone refractory prostate cancer. Indeed, the detection of these cells and/or the detection of a significant increase of frequency of these cells through their specific expression of at least one of these genes within the prostate cell population of a subject is efficient in prognosing the risk for, the presence of, or the evolution of a prostate cancer toward a hormone resistant prostate cancer. In a preferred embodiment said subject is a human subject.

An object of the invention thus relates to an in vitro method of prognosing the outcome of prostate cancer, said method comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying the overexpression of at least one gene selected from Clorfl 16, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2f 1, Defb4A, Defdl 14, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl 1, Illrn, Kcnk5, Krt4, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In an embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of at least one gene selected from Aldhla3, Anxal, Anosl, Apod, Areg, Arll4, Aspa, AtplOb, B4galnt2, Bace2, Btc, C15orf62, Clorfl 16, C3, C7, Cadps, Capnl3, Capsl, Cd55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Dusp4, Ednl, F5, Ffar4, Gprl37b, Gprc5a, Gsdmc, Gsta4, Igsfl 1, Illrn, Kcnk5, Krt23, Krt4, Krt7, Lgals3, Ltf, Mgat4a, Mgat5, Myof, Nbea, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Snord73b, Spns2, Sptlc3, Sultlc2, Tspan8, Ugt2bl0, Usp6nl and Wfdc2. In a more particular embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of at least one gene selected from, Cyp2fl, Krt4, Saal, CxcL17, Ednl, Gsdmc, Pglyrpl, Reg3a, Reg3g and Psca. In another embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of at least one gene selected from Saal, Pparg, Aldhla3, Tspan8, Illrn, Ctse, Cdk6, Clcal, Ltf, Areg, Ednl, Krt4, Krt7, Clu, Ngf, Lgals3, F5, Psca, C3, Scnnla and Anxal.

In another embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of at least one gene selected from Pglyrpl, Tspan8, Kcnk5, Ffar4, Gprl37b, Smiml5, Clcal, Psca, Gprc5a, Scnnla, Anxal, Bace2, Usp6nl, Seel, Sultlc2, Gsta3, Illrn, Apod, Capsl, Ltf, ATPIOB, Myof, Scara3, C7, Igsfl l, Sdcbp2, C3.

In yet another preferred embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of at least one gene selected from Krt7, Spns2, Gfpt2. In a particular embodiment, a combination of at least two biomarkers of the above lists is preferred.

As shown in the experimental section, cytokeratin 4 (CK4 protein, encoded by Krt4 gene) represent a reliable biomarker for LSC ^med cells being specific for these cells and providing results consistent with FACS profiles obtained in the different models (including human tumour xenografts) and conditions used in the experimental section.

Therefore, another object of the invention relates to an in vitro method of prognosing the outcome of prostate cancer comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying an overexpression of Krt4 gene. In a particular embodiment said in vitro method comprises identifying prostatic epithelial cells displaying the overexpression of Krt4 gene and of at least one gene selected from ClorfH6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATPIOB, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In another particular embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of Krt4 gene and at least one gene selected from Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, AtplOb, B4galnt2, Bace2, Btc, C15orf62, Clorfl l6, C3, C7, Cadps, Capnl3, Capsl, Cd55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Dusp4, Ednl, F5, Ffar4, Gprl37b, Gprc5a, Gsdmc, Gsdmc, Gsdmc, Gsta4, Igsfl l, Dim,

Kcnk5, Krt23, Krt4, Krt7, Lgals3, Ltf, Mgat4a, Mgat5, Myof, Nbea, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Snord73b, Spns2, Sptlc3, Sultlc2, Tspan8, Ugt2bl0, Usp6nl and Wfdc2. In another particular embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of Krt4 gene and at least one gene selected from, Cyp2f 1, Saal, CxcL17, Ednl, Gsdmc, Pglyrpl, Reg3a, Reg3g and Psca.

In a more particular embodiment, the prostatic epithelial cells detected in the above in vitro method display an overexpression of Krt4 gene and of at least one gene selected from Saal, Pparg, Aldhla3, Tspan8, Dim, Ctse, Cdk6, Clcal, Ltf, Areg, Ednl, Krt7, Clu, Ngf, Lgals3, F5, Psca, C3, Scnnla and Anxal.

Also, data presented in the experimental section show that Krt7 (aka CK7), Spns2 or Gfpt2 labelling of Pb-PRL and Pten ^{pc /"} mouse prostate display the same features as CK4 labelling. Consequently, Krt7, Spns2, Gfpt2 are particularly suitable for the detection and identification of LSC ^med .

Accordingly, a particular embodiment of the invention relates to an in vitro method of prognosing the outcome of prostate cancer comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying an overexpression of Krt7 gene. In a more particular embodiment said in vitro method comprises identifying prostatic epithelial cells displaying the overexpression of Krt7 gene and of at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2f 1, Defb4A, Defdl 14, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Dim, Kcnk5, Krt23, Krt4, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In another particular embodiment, the invention relates to an in vitro method of prognosing the outcome of prostate cancer comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying an overexpression of Spns2 gene. In a more particular embodiment said in vitro method comprises identifying prostatic epithelial cells displaying the overexpression of Spns2 gene and of at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa,

ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt4, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Krt7, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In a further particular embodiment, the invention relates to an in vitro method of prognosing the outcome of prostate cancer comprising identifying, in a biological sample of a subject suffering or suspected of suffering from prostate cancer, prostatic epithelial cells displaying an overexpression of Gfpt2 gene. In a more particular embodiment said in vitro method comprises identifying prostatic epithelial cells displaying the overexpression of Gfpt2 gene and of at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2f 1, Defb4A, Defdl 14, Dusp4, Ednl, F5, Ffar4, Krt7, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt4, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In any of the above methods, product of the gene biomarker of the LSC signature as mentioned in Table 1 can be detected by means well known in the art such as nucleic acid probe, RNA or protein aptamer, antibody and the like. In an embodiment, the transcript level can be determined by quantitative Reverse Transcription PCR, northern blotting assays, nucleic acid microarray, in situ hybridization, fluorescent labelling and the like. In another embodiment, the level of any protein or peptide encoded by said gene biomarker of the invention can be determined using western blotting assays, 2-dimensional electrophoresis, immunofluorescence, immunohistochemistry and the like. In an embodiment, high pressure chromatography either coupled or not with mass spectrometry methods (e.g. MS/MS, Maldi/TOF, Maldi TOF/MS...) can be used to detect said gene product. Said means and methods can be used alone or in combination in the methods of the invention. The overexpression of a gene biomarker of the LSC ^med signature in the biological sample of the subject is determined by comparison with the expression level of said gene in a control sample. In an embodiment said control sample is from a subject not suffering from prostate cancer. In another embodiment, said control sample is from a subject not suffering from castration-resistant prostate cancer. In a further embodiment, said control sample has been previously collected from the same subject. Also in a particular embodiment, said control sample has been previously collected from the same subject before triggering an androgen deprivation or an hormonal therapy on this subject.

As shown in the experimental section, a significant presence of LSC ^med cells in a biological sample of a subject can be detected in an unsorted preparation of prostatic cells on the basis of their gene signature as described above, by detecting an increase quantity of the corresponding transcripts in said sample. In an embodiment, the methods of the invention comprise determining an increased quantity of RNA level of at least one gene of LCS ^med signature. In a preferred embodiment said increased RNA level of at least one gene is of at least 1.5 fold change or even more in comparison with non-cancerous prostatic cells or non-cancerous biological sample. In a more preferred embodiment said increase corresponds to 2, 3, 4 times the RNA level found in noncancerous prostatic cells or non-cancerous biological sample.

Experimental data show also that immunohistochemistry methods are effective in detecting presence of LSC ^med cells in a biological sample by the detection of cells positive for one or more protein of the LSC ^med cell signature discovered by the Inventors. Then, in a particular embodiment, the detection, in prostatic epithelium, of more than 10% of cells positive for one or more protein of the LSC ^med cell signature is considered as the sign of a castration-resistant prostate cancer, of prostate cancer progression and/or recurrence. In another particular embodiment said cells positive for one or more protein of the LSC ^med cell signature represent more than 20%, 30%, 40%, 50%, 60%, 70%, 80% of prostatic epithelium in the biological sample of the subject. Any sample suitable for immunohistochemistry (e.g. formalin fixed and/or paraffin embedded) can be used.

In an embodiment, the methods of the invention comprise the detection of epithelial cell positive, as described above, for at least one of the protein encoded by Pglyrpl, Tspan8, Kcnk5, Ffar4, Gprl37b, Smiml5, Clcal, CK4, Psca, Gprc5a, Scnnla, Anxal, Bace2, Usp6nl, Seel, Sultlc2, Gsta3, Dim, Apod, Capsl, Ltf, AtplOB, Myof, Scara3, C7, Igsf 11, Sdcbp2, and C3.

In another embodiment, the methods of the invention comprise the detection of epithelial cell positive, as described above, for at least one of the protein encoded by a gene selected from Krt7, Spns2, Gfpt2. In a particular embodiment said detection is performed either by a FACS analysis or by an immunohistochemical staining for said at least one protein.

In a preferred embodiment, the methods of the invention comprise the detection of cytokeratin 4 positive (CK4 ⁺ ) cells in the biological sample of the subject, Cytokeratin 4 (CK4) protein being known as encoded by the Krt4 gene. In an even more preferred embodiment said detection comprises an immunohistochemical staining for CK4 protein.

In a preferred embodiment, the methods of the invention comprise the detection of cytokeratin 7 positive (CK7 ⁺ ) cells in the biological sample of the subject, Cytokeratin 7 (CK7) protein being known as encoded by the Krt7 gene. In an even more preferred embodiment said detection comprises an immunohistochemical staining for CK7 protein.

In a preferred embodiment, the methods of the invention comprise the detection of Glutamine- fructose-6-phosphate transaminase 2 positive (GFPT2 ⁺ ) cells in the biological sample of the subject, GFPT2 protein being known as encoded by the Gfpt2 gene. In an even more preferred embodiment said detection comprises an immunohistochemical staining for GFPT2 protein. In a preferred embodiment, the methods of the invention comprise the detection of spinster homolog 2 positive (SPNS2 ⁺ ) cells in the biological sample of the subject, SPNS2 protein being known as encoded by the Spns2 gene. In an even more preferred embodiment said detection comprises an immunohistochemical staining for SPNS2 protein.

In another preferred embodiment, the methods of the invention comprise the detection of Scnnla positive (SCNN1A ⁺ ) cells in the biological sample of the subject. In an even more preferred embodiment said detection comprises an immunohistochemical staining for SCNNla protein.

In a further preferred embodiment, the methods of the invention comprise the detection of Psca positive (PSCA ⁺ ) cells in the biological sample of the subject. In an even more preferred embodiment said detection comprises an immunohistochemical staining for PSCA protein.

In another preferred embodiment, the methods of the invention comprise the detection of Cxcll7 positive (CXCL17 ⁺ ) cells in the biological sample of the subject. In an even more preferred embodiment said detection comprises an immunohistochemical staining for CXCL17 protein.

As explained above, diagnostic of prostate cancer is currently made based on an in vitro morphological analysis of prostate cells of the subject. Anyway, immunohistochemical assays can be less subjective and are more often used to diagnose prostate cancer. Thus, in a preferred embodiment, methods according to the invention further comprise immunohistochemical diagnostic test for cancer prostate comprising detecting in vitro in a biological sample from said subject:

- a loss of basal cells in said sample, for example using at least one high molecular weight cytokeratin and/or p63 immunohistochemical labelling, and/or

- AMACR positive cells in said sample.

In an even more preferred embodiment, in vitro methods according to the invention comprise:

- a step of detecting, in the biological sample of the subject, cells positive for at least one of the protein encoded by the gene of the LSC ^med signature as described above, - a step of detecting, in the biological sample of the subject a loss of basal cells in said sample using at least one high molecular weight cytokeratin and/or p63 immunohistochemical labelling, and/or

- a step of detecting AMACR positive cells in said sample, wherein the presence of more than 10% of cells positive for at least one of the protein encoded by the genes of the LSC ^med signature, and/or a loss of basal cells and/or the presence of AMACR positive cells sign(s) a castration-resistant prostate cancer, a progression of prostate cancer and/or recurrence.

In another preferred embodiment, in vitro methods according to the invention comprise:

- a step of detecting, in the biological sample of the subject, CK4 ⁺ cells, - a step of detecting, in the biological sample of the subject a loss of basal cells, in said sample and/or p63 immunohistochemical labelling, and/or

- a step of detecting AMACR positive cells in said sample, wherein the presence of more than 10% of the cells being found CK4 ⁺ , a loss of basal cell and/or the presence of AMACR positive cells sign(s) a castration-resistant prostate cancer, a prostate cancer progression and/or recurrence.

In an embodiment, in vitro immunohistochemical diagnostic test for cancer prostate is performed together or separately, simultaneously or at different time interval, with the detection of cells overexpressing at least one gene of the LSC ^med signature, on the same sample or on a different sample of the subject. In a particularly preferred embodiment said detections are made simultaneously on the same biological sample of the subject. In a more particularly preferred embodiment said detection is made by using an immunohistochemical labelling cocktail, including a basal cell marker (e.g. p63) and a cancer cell marker (e.g. AMACR). In an embodiment, an object of the invention is a kit comprising a mean for detecting the overexpression of at least one gene product of LSC ^med signature as exposed above, selected from Clorfl 16, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl 1, Illrn, Kcnk5, Krt4, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl and Wfdc2. In a very particular embodiment, an object of this invention is a kit comprising a mean for detecting the overexpression of Krt4 gene and of at least one gene selected from Clorfl 16, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl 1, Illrn, Kcnk5, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl and Wfdc2.

In another particular embodiment, an object of this invention is a kit comprising a mean for detecting the overexpression of Psca gene and of at least one gene selected from Clorfl 16, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl 1, Illrn, Kcnk5, Krt23, Krt4, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl and Wfdc2. Another particular object of this invention is a kit comprising a mean for detecting the overexpression of Scnnla gene and of at least one gene selected from Clorf 116, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, nirn, Kcnk5, Krt23, Krt4, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2. Another particular object of this invention is a kit comprising a mean for detecting the overexpression of Cxcll7 gene and of at least one gene selected from Clorfl 16, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsf 11, Illrn, Kcnk5, Krt23, Krt4, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In a particular embodiment, an object of the invention resides in the use of a kit as above for prognosing the outcome of prostate cancer in a mammalian subject, preferably a human subject.

In a very particular embodiment, an object of this invention is the use of a kit comprising at least a mean for detecting CK4 ⁺ cells for prognosing the outcome of prostate cancer in mammalian subject, preferably a human subject.

Another particular object of this invention is the use of a kit comprising at least a mean for detecting PSCA ⁺ cells for prognosing the outcome of prostate cancer in mammalian subject, preferably a human subject.

A particular object of this invention is the use of a kit comprising at least a mean for detecting SCNNAl ^"1" cells for prognosing the outcome of prostate cancer in mammalian subject, preferably a human subject. A further object of this invention is the use of a kit comprising at least a mean for detecting

CXCL17 ^"1" cells for prognosing the outcome of prostate cancer in mammalian subject, preferably a human subject. A more particular object of this invention is any of the use of a kit as above in prognosing the risk of relapse or of developing an androgen blockade resistant prostate cancer in a mammalian subject, preferably a human subject.

Advantageously, any of the above-mentioned methods or kits can be used in a method for determining, adapting or modifying the treatment of prostate cancer in a subject suffering or suspected of suffering from said cancer. Indeed, experimental data show the clinical implication of LSC ^med cells in castration resistance and prostate cancer relapse, these cells having been found by the Inventors as increased upon androgen deprivation conditions and as having tumorigenic properties. Hence in instances where a patient is diagnosed as bearing prostatic epithelial cells as LSC ^med cells, androgen deprivation therapy (orchiectomy, chemical castration or antiandrogen therapy) should be considered as being not a suitable therapeutic intervention and alternative therapeutic strategies should be considered instead.

Consequently, an object of the invention is a method for determining, adapting or modifying the treatment of prostate cancer in a subject suffering or suspected of suffering from said cancer, said method comprising not applying androgen deprivation therapy in a patient shown as having prostatic epithelial cells exhibiting an overexpression of any of the gene of the LSC ^med signature, or combination thereof, as described above elsewhere in this specification.

In a particular embodiment, the method for determining, adapting or modifying the treatment of prostate cancer comprises not applying androgen deprivation therapy when said patient is shown as having prostatic epithelial cells exhibiting an overexpression of at least Krt4 gene or of Krt4 gene with at least one gene selected from Clorf 116, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2

In another particular embodiment, the method for determining, adapting or modifying the treatment of prostate cancer comprises not applying androgen deprivation therapy when said patient is shown as having prostatic epithelial cells exhibiting an overexpression of at least Krt7 gene or of Krt7 gene with at least one gene selected from Clorf 116, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsf 11, nirn, Kcnk5, Krt23, Krt4, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pgl rpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In further particular embodiment, the method for determining, adapting or modifying the treatment of prostate cancer comprises not applying androgen deprivation therapy when said patient is shown as having prostatic epithelial cells exhibiting an overexpression of at least Spns2 gene or of Spns2 gene with at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Gfpt2, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsf 11, Illrn, Kcnk5, Krt23, Krt4, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Krt7, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

In another particular embodiment, the method for determining, adapting or modifying the treatment of prostate cancer comprises not applying androgen deprivation therapy when said patient is shown as having prostatic epithelial cells exhibiting an overexpression of at least Gfpt2 gene or of Gfpt2 gene with at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Defb4A, Defdl l4, Dusp4, Ednl, F5, Ffar4, Krt7, C15orf62, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsf 11, Illrn, Kcnk5, Krt23, Krt4, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, OsbpB, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl, and Wfdc2.

Results shown herein identify LSC ^med cells as the source of prostate cancer relapse notably after androgen deprivation, and therefore as a new therapeutic target for the prevention and/or treatment of castration-resistant prostate cancer. Consequently, an object of the invention relates to a method of selecting an active compound in the prevention and/or the treatment of androgen blockade resistant prostate cancer, said method comprising the use of mouse prostatic epithelial cells resistant to androgen blockade and displaying the overexpression of at least one gene selected from 1700016G22Rik, 4930594C411Rik, AA986860, Acsm3, Aldhla3, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, Cbr2, CD55, Cdc42ep5, Cdk6, Clcal, Clca3, Clu, Ctse, Ctsh, Cxcll3, Cxcll5, Cxcll7, Cyp2f2, D17H6S56E-5, Daf2, Defbl4, Defb2, Dusp4, Ednl, F5, Ffar4, Gfpt2, Gml2146, Gml2245, Gml2603, Gml2888, Gml3502, Gml4137, Gml4328, Gm23640, Gm24431, Gm6166, Gprl37b, Gprl37b-ps, Gprc5a, Gsdmc2, Gsdmc3, Gsdmc4, Gsta3, Gsta4, Igsf 11, Illrn, Kcnk5, Krt23, Krt4, Krt7, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3b, Reg3g, Rnu73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sprr2a2, Sprr2a3, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2b34, Usp6nl, Wfdcl8 and Wfdc2. In a more particular embodiment the above method of selecting an active compound in the prevention and/or the treatment of androgen blockade resistant prostate cancer comprises the use of mouse prostatic epithelial cells resistant to androgen blockade and displaying the overexpression of at least one gene selected from AA986860, Aldhla3, Anxal, Apod, Areg, Arll4, Aspa, AtplOb, B4galnt2, Bace2, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, CD55b, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2f2, Daf2/, Dusp4, Ednl, F5, Ffar4, Gml4137, Gprl37b, Gprc5a, Gsdmc2, Gsdmc3, Gsdmc4, Gsta4, Igsf 11, Illrn, Kcnk5, Krt23, Krt4, Krt7, Lgals3, Ltf, Mgat4a, Mgat5, Myof, Nbea, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3b, Reg3g, Rnu73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Tspan8, Ugt2b34, Usp6nl and Wfdc2. In a particular embodiment said method comprises the use of mouse prostatic cells resistant to androgen blockade and displaying the overexpression of at least Krt4 gene.

FACS analysis can be used to select LSC ^med based on the selection of prostatic epithelial cells displaying Lin Sca-l ⁺ /CD49r ^ed or CK5 CK8 ⁺ /CK4 ⁺ features. Consequently, another object of the invention is a method of selecting an active compound in the prevention and/or the treatment of androgen blockade resistant prostate cancer, said method comprising the use of mouse prostatic epithelial cells resistant to androgen blockade, said mouse prostatic epithelial cells being found Lin Sca-l ⁺ /CD49r ^ed and/or CK57CK87CK4 ⁺ in FACS analysis. Another object of the invention is a method of selecting an active compound in the prevention and/or treatment of androgen blockade resistant prostate cancer, said method comprising the use of human prostatic cells resistant to androgen blockade and displaying the overexpression of at least one gene selected from Clorfl l6, Acsm3, Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, ATP10B, B4galnt2, Bace2, Bexl, Btc, C3, C7, Cadps, Capnl3, Capsl, CD55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2f 1, Defb4A, Defdl 14, Dusp4, Ednl, F5, Ffar4, Gfpt2, Gml4137, Gprl37b, Gprc5a, Gsdmc, Gsta3, Gsta4, Igsfl l, Illrn, Kcnk5, Krt23, Krt7, Krt4, Leng9, Lgals3, Ltf, Lypd2, Meis2, Mgat4a, Mgat5, Myof, Nbea, Ngf, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, Snord73b, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Spns2, Sptlc3, Sultlc2, Sultlel, Tcf21, Tnfaip6, Tspan8, Tstdl, Ugt2bl0, Usp6nl and Wfdc2.

In a particular embodiment said method comprises the use of human prostatic cells resistant to androgen blockade and displaying the overexpression of least one gene selected from Aldhla3, Anosl, Anxal, Apod, Areg, Arll4, Aspa, AtplOb, B4galnt2, Bace2, Btc, C15orf62, Clorfl l6, C3, C7, Cadps, Capnl3, Capsl, Cd55, Cdc42ep5, Cdk6, Clcal, Clu, Ctse, Ctsh, Cxcll3, Cxcll7, Cyp2fl, Dusp4, Ednl, F5, Ffar4, Gprl37b, Gprc5a, Gsdmc, Gsta4, Igsfl l, fllrn, Kcnk5, Krt23, Krt4, Krt7, Lgals3, Ltf, Mgat4a, Mgat5, Myof, Nbea, Nt5c2, Osbpl3, Pglyrpl, Pparg, Prss22, Psca, Reg3a, Reg3g, SlOOal, Saal, Sbsn, Scara3, Seel, Scnnla, Sdcbp2, Slc25a24, Slc25a48, Smiml5, Snord73b, Spns2, Sptlc3, Sultlc2, Tspan8, Ugt2bl0, Usp6nl and Wfdc2. In a more particular embodiment said method comprises the use of human prostatic cells resistant to androgen blockade and displaying the overexpression of at least Krt4 gene.

Compounds for the prevention and/or the treatment against androgen blockade resistant prostate cancer can be selected using in vitro assays well known in the art. Such assays comprise, for example, MTT cytotoxicity assays whereby cell proliferation can be measured as a function of compound concentration in the culture medium. Assays can be done on culture plates or, for example, in soft agar to follow the formations of 2D and 3D colonies respectively. Such in vitro cell toxicity/proliferation assays are particularly convenient for high throughput screening of compounds libraries. Xenografts of LSC ^med cells containing tumours and the subsequent follow up of tumour growth in the animals subjected to a treatment with candidate compound(s) are further tests that can be implemented in the methods of the invention.

Examples

A. Protocols 1. Animals

Animal experiments using Pten ^{pc /~} mice were reviewed and approved by the Austrian ministry authorities and conducted according to relevant regulatory standards (BMWF-66.009/0281- I/3b/2012 ). Animal experiments using Pb-PRL mice were approved by the local ethical committee for animal experimentation (authorization CEEA34.VG.095.12). Animal experiments using Hi-Myc were reviewed and approved by Monash Animal Research Platform Animal Ethics Committee (approval number: MARP/2011/161).

Pten ^{pc /~} mice (pten ^loxP/loxP mice crossed with Pb-Cre4 transgenic males) were generated as described previously (2) and maintained on a C57BL/6 and Sv/129 mixed genetic background. Hi- Myc (ARR2/Pb-MYC) mice were maintained on a pure FVB/N background.

Pb-PRL mice (prolactin transgene driven by the short probasin promoter) were generated on the C57BL/6J background (>20 backcrosses), as previously described (3).

Experiments were performed using 6-8 month old mice, i.e. when pre-malignant (Pb-PRL) and aggressive malignant (Pten ^{pc /~} and Hi-Myc) phenotypes were well established. For all genotypes, non-transgenic littermates were used as controls and are referred to as WT animals.

Colonies were housed in controlled conditions, on a 12/12-hour light/dark cycle with normal food and water provided ad libitum. Where indicated, mice were surgically castrated at the age of 2 (WT and Pb-PRL) or 3 weeks (Pten ^pcV~ ) and for the latter, BrdU (1 mg/ml) was administered in the drinking water during 4 days before sacrifice. Prostate samples were obtained by microdissection immediately after sacrifice by cervical dislocation.

2. Prostate subpopulation sorting by FACS

Cell sorting was performed on a BD FACS Aria III (BD Biosciences, San Jose, CA). Prostate cells were isolated from freshly dissected mouse prostates by tissue mincing and collagenase digestion, trypsin and DNase treatment, followed by homogenization with syringe and needle and passing through a 40 μιη nylon mesh. Cell suspensions from all genotypes were subjected to differential centrifugation using Histopaque-1119 (Sigma- Aldrich) to reduce the prostate secretion contents. Isolated cells were then stained on ice for 20 minutes. Antibodies included FITC-coupled lineage "Lin" antibodies (anti-CD31, CD45 and Terl l9, 11-0311-85, 11-0451-85, 11-05921-85, respectively), PE-coupled anti-CD49f (integrin alpha-6; 12-0495-83) and APC-coupled anti-Sca-

1 (lymphocyte antigen 6A-2/6E-1; 17-5981-82) all from eBioscience, San Diego, CA. Dead cells were stained with SYTOX Blue (S34857, Life Technologies, Carlsbad, CA).

Sorted cells were loaded onto cytospin slide chambers of a Shannon Cytospin 2 (Thermo Scientific) and centrifuged at 500 rpm at room temperature for 10 min. They were collected in DMEM medium, supplemented with 50% FBS, glutamine, and penicillin-streptomycin, or in RA1 Lysis Buffer (Macherey-Nagel, Diiren, Germany) to perform RNA extraction.

3. RNA extraction and amplification RNA was extracted from FACS sorted cells with a Nucleospin RNA XS kit (Macherey-Nagel) according to the manufacturer's instructions. RNA quality and concentration measurements were performed using an Agilent RNA 6000 Pico kit on a BioAnalyzer (Agilent Technologies, Santa Clara, CA). RNA samples were amplified using an Ovation PicoSL WTA System V2 (NuGEN Technologies, San Carlos, CA), according to manufacturer's instructions. Resulting cDNA was purified using Agencourt RNAClean XP Beads (Beckman Coulter, Brea, CA).

4. Microarray analysis

Gene expression analysis was performed using GeneChip® Mouse Transcriptome Arrays 1.0 (Affymetrix, Santa Clara, CA). Prior to hybridization, cDNA was fragmented and biotin-labeled using the Encore Biotin Module (NuGEN). Biotinylated DNA fragments were hybridized onto the array chips using the Hybridization Wash Stain kit (Affymetrix). The chips were washed, stained, and scanned using the Affymetrix Model 450 Fluidics Station, the Affymetrix Model 3000 scanner and the Command Console software for piloting the GeneChip systems. For data analysis, raw data CEL files were imported in R/Bioconductor using the Oligo package [http://www.r- project.org/]. Expression levels were normalized using the RMA algorithm from the Affymetrix package. Expression levels and background noise were computed using a custom algorithm within R as follows. Assuming that a maximum of 80% of genes are expressed on any given microarray, 20% of probes were tagged with the lowest intensity for each microarray as background. A threshold was fixed at two standard deviations over the mean of the background. All probes for which normalized intensities were lower than the computed threshold were designated as background for each array. When comparing gene expression levels between two groups, a probe was included in the analysis if its intensity exceeded the background in at least 80% of the samples from at least one group. Group comparisons were done using Student's t-test and lists were filtered at P <0.05 and fold change > 1.5. Cluster analysis was performed by hierarchical clustering using the Spearman correlation similarity measure and average linkage algorithm.

5. Quantitative PCR

For qPCR, iTaq Universal SYBR Green Supermix (Biorad, Hercules, CA) was used, and reactions were run on a Viia7 Real-Time PCR System (Applied Biosystems, Foster City, CA) or a qTower 2.0 real-time thermal cycler (Analytik Jena). Primers are listed Table 2. Results were normalized to Eeflal expression.

Table 2

Gene Sequence Primer Forward (5'-3') Sequence Primer Reverse (5 '-3')

Cbr2 GGGCAGGGAAAGGGATTGG CCACACACACGGGCTCTATTC

Cxcll5 ATGGGTGAAGGCTACTGTTGG AGAGGCTTTTCATGCTCAACAC

Cxcin AGGTGGCTCTTGGAAGGTG GGTGACATCGTTTGAGAAATTGC

Cyp2f2 GGACCCAAACCTCTCCCAATC CCGTGAACACCGACCCATAC

Ednl GCACCGGAGCTGAGAATGG GTGGCAGAAGTAGACACACTC

Gsdmc2 CTGTGGAATGCTTGTCCGATG CCTCCAGGTCCGTTGATTGG

Gsdmc4 TGAGGAGCCTGCCAATCTAAA ATGTGGGGTGCTAGAATCCTT

Krt4 CAATGACAAAGGTCGCCTACA AGGACTCTCGTGAAGTTGATCTC

Psca GGACCAGCACAGTTGCTTTAC GTAGTTCTCCGAGTCATCCTCA

Pglyrpl GCCATCCGAGTGCTCTAGC CTTGTGGTAATGCTGCACATTG Reg3b ACTCCCTGAAGAATATACCCTCC CGCTATTGAGCACAGATACGAG Reg3g CAGACAAGATGCTTCCCCGT GCAACTTCACCTTGCACCTG Saal GAAGGAAGCTAACTGGAAAAACTC TCCTGAAAGGCCTCTCTTCCAT 6. Human prostate cancer samples

Fixed primary prostate tumor samples were obtained from patients with individual patient consent (4). Three localized tumors were used to generate patient-derived xenografts (PDX). They were obtained from radical prostatectomy patients with written consent from Cabrini Hospital (03-14- 04-08) and Monash University (2004/145 and CF11/3577-2011001899) Human Research Ethics Committees. Tissues were cut into pieces and implanted under the kidney capsule of host NOD/SCID or NSG mice as described previously (5,6). Implanted tissues were harvested as a function of xenograft growth. Typically, for grafts in castrated mice, harvest occurred at 4 weeks after castration. Collected grafts were fixed in 10% formalin, embedded in paraffin and treated for immunostaining.

7. Immunohisto/cy to chemistry (IHC/ICC) and immunofluorescence (IF) All samples were fixed in 4% PFA, paraffin embedded, and underwent heat-induced antigen retrieval in citrate buffer at pH 6. IHC/ICC and IF were performed as previously described (1). Vector Elite ABC HRP kit with DAB substrate (Vector Laboratories, Burlingame, CA) was used for detection of IHC slides, with hematoxylin as counterstain. Slides were scanned with a Nanozoomer 2.0 (Hamamatsu, Massy, France) and analyzed using NDP.view 2.5.14 software (Hamamatsu). For IF, nuclei were stained with Hoechst dye, and samples were analyzed with a lOx or 20x objective under an Axio Observer.Zl inverted microscope (Carl Zeiss Microscopy, Germany).

Primary antibodies included mouse anti-CK4 (MA1-35558; ThermoFisher Scientific, Waltham, MA), rabbit anti-Krt7 (HPA007272; Sigma-Aldrich, Saint Louis, USA), rabbit anti-p63 (clone 7JUL, Novocastra, Leica Biosystems), rabbit anti-AMACR (clone 13H4, Dako, Santa Clara, CA), mouse anti-BrdU (ref. 11170376001, Sigma Aldrich), and antibody diluent (MM- France, Francheville, France) was used as negative control. Fluorescent-labelled secondary antibodies used for IF were goat anti-mouse IgG-CFL 594 (sc-362277; Santa Cruz Biotechnology), goat anti-rabbit Alexa Fluor 488 (A-11070; Invitrogen) and goat anti-rat IgG- CFL 647 (sc-362293 Santa Cruz Biotechnology). 8. In vivo prostate regeneration assay

The in vivo prostate regeneration assay was performed as previously described (16). Briefly, prostates from 3 Pten ^{pc /"} mice were digested. 8 10 ⁴ FACS-sorted LSC ^med cells or Basal/stem cells were mixed with 1.5 10 ⁵ urogenital sinus mesenchymal (UGSM) cells mixed with growth factor reduced matrigel (Corning, Corning, NY) (1: 1, v/v) in a final volume of 150 μί, and transplanted subcutaneously in immunodeficient male SCID mice for 10 weeks. Regenerated tissues were collected, fixed in 4% buffered formalin and embedded in paraffin for histological, morphological and/or IHC analysis.

9. Statistics

Two-way analysis of variance (ANOVA) with Sidak's multiple comparisons were used to compare population percentages among intact and castrated mouse groups. Repeated measures one-way ANOVA tests were used to evaluate gene-expression differences among cell types for each genotype. Post hoc multiple comparisons were performed by Tukey's test. A value of P < 0.05 was used as significance cut-off for all tests. All analyses were performed using GraphPad Prism version 6.00 for Windows, GraphPad Software, San Diego CA.

B. Results 1. LSC ^med cells are present in various in vivo models for prostate cancer cells, are

resistant and proliferate under androgen deprivations conditions. a. Tumours of Prostate-specific Pten-deficient mice exhibit a significant increase in LSC ^med cells. Pro state- specific Pten-deficient mice {Pten ^{pc /~} ) are known to develop invasive prostate adenocarcinomas. This model has been shown to mimic the progression of the disease toward invasive carcinoma and subsequent metastasis, as seen in humans (7).

As shown in Figure 1, Inventors surprisingly found that 82% of the whole Pten ^{pc /~} epithelia of intact prostates was made of LSC ^med (Fig. 1A and C). Of note, as evidenced in Figures IB and C, no change regarding the LSC ^med prevalence was observed in the prostate epithelium of castrated Pten ^{pc /~} mice, within which cell population proportions are almost the same than in non-castrated mice, thereby signing an androgen deprivation resistance of these cells. b. Castration triggers a massive increase in LSC population proportion in a premalignant model and in Wild Type mice.

Wild Type mice exhibit a very low prevalence of LSC ^med cells (< 10%, Fig. 2A and C) in intact prostates and a significant increase of LSC ^med cells population after castration (>20%, Fig. 2B and C).

An even more striking outbreak of LSC ^med cells is observed in castrated Pb-PRL mice, wherein an almost two-fold enrichment in LSC ^med cells population of prostate epithelium is noticed (from 36% to more than 70% of the epithelia before and after castration, respectively Fig. 2D-F).

Altogether these results show that LSC ^med cells are found significantly increased in prostate cancerous epithelium in several in vivo models for prostate cancer and that, furthermore, these cells are resistant and even proliferate under androgen deprivation conditions, thereby suggesting a role for these cells type in prostate tumorigenesis and in the progression to castrate resistant prostate cancer (CRPC): some LSC ^med subsets may be more tolerant to castration than others, leading to clonal expansion.

2. Gene expression analyses

a. LSC ^med cells exhibit a specific gene expression signature Microarray analyses of the above WT FACS-sorted LSC ^med , basal/stem, and luminal cell populations were performed. WT prostate were chosen to circumvent any bias due to prostate pathology. Results shown that the gene expression profiles of the three epithelial cell subpopulations are different, though LSC ^med shares some similarities to basal/stem and to luminal cells. Indeed, 111 genes were differentially expressed (>2-fold, p<0.05) compared to basal/stem and luminal cells. Most of these genes were over-expressed in LSC ^med , while a minor portion was downregulated in LSC ^med compared to basal/stem and luminal cells (Table 1). b. LSC cells signature is conserved across mouse models LSC ^med signature is detected in Pten ^{pc ,~} mouse prostates

As shown above, LSC ^med represent the vast majority of epithelial cells in Pten ^{pc /~} mouse prostates (Fig. 1). LSC ^med signature of Table 1 was used to query two independent transcriptomic datasets of Pten ^{pc /~} mouse prostate whole tissue (i.e. unsorted cells) relative to WT (GSE46799 and GSE46473) (8,9). Both analyses revealed a strong overlap of LSC ^med -specific genes with those differentially expressed in Pten ^{pc /~} mice vs WT, most of the genes being concordantly up- or downregulated in both unsorted Pten ^{pc /~} prostates and sorted WT LSC ^med (Table 1, not shown for down-regulated genes). Hence the gene expression profile of unsorted Pten ^{pc /~} mouse prostates do reflect an enrichment of LSC ^med -specific genes, thereby showing that the signature can be detected using microarrays without any sorting of the cells.

LSC ^med signature is confirmed by qPCR in WT, premalignant Pb-PRL and malignant Pten ^{pc ,~} mouse prostates sorted cells.

Thirteen genes from the LSC ^med signature (see Table 2) were selected and compared for their actual level of expression in sorted cell populations from WT, premalignant Pb-PRL and malignant Piew^-prostates by qPCR (Fig. 3). In the three models, a noticeable upregulation of these genes was observed in LSC ^med compared to basal/stem and luminal cell populations, thereby confirming that LSC ^med signature is maintained across models. Importantly, castration was not found to affect this population-specific expression pattern even when the absolute expression levels varied post- castration.

Altogether these findings show that the LSC ^med signature is intrinsic and specific to the LSC ^med cells, independently of the prostate pathological status.

c. LSC enrichment is linked to the intrinsic level of AR-signaling activation

LSC ^med display low androgen signaling

As shown on Fig. 1 and 2, LSC ^med survive and are proportionally enriched upon castration thereby demonstrating their androgen-independence. Based on the list of 144 genes defined as androgen-regulated in the prostate (10), inventors observed that androgen signalling was markedly lower in LSC ^med , with AR-activated genes significantly downregulated and AR-repressed genes, upregulated or unchanged relative to luminal cells (representative examples are given in Fig. 4, ENTREZ gene ID in Table 3 below). The intrinsically low androgen signalling of LSC ^med thus explains why they tolerate androgen-deprivation.

Table 3

Representative example of AR-activated genes Representative example of AR-repressed genes Gene symbol Gene ID mouse/human Gene symbol Gene ID mouse/human

NKX3-1 ID: 18095/ID: 4824 AHNAK ID: 66395/ID: 79026

FKBP11 ID: 66120/ID: 51303 THBS1 ID: 21825/ID: 7057

RCN1 ID: 19672/ID: 5954 TACSTD2 ID: 56753/ID: 4070

MCOLN2 ID: 68279/ID: 255231 F3 ID: 14066/ID: 2152

PDIA6 ID: 71853/ID: 10130 Cxcl5/CXCL6 ID: 20311/ID: 6372

APOF ID: 103161/ID: 319 GADD45B ID: 17873/ID: 4616

CRELD2 ID: 76737/ID: 79174 IFIT3 ID: 15959/ID: 3437

AZGP1 ID: 12007/ID: 563 IGFBP3 ID: 16009/ID: 3486

CRABP1 ID: 12903/ID: 1381 AW112010 ID: 107350/nd

FAM3B ID: 52793/ID: 54097 Car8/CA8 ID: 12319/ID: 767

LSC are enriched in low androgen signalling contexts

A down-regulation of AR expression and/or AR transcriptional activity has already been reported in intact Pten ^{pc /~} prostates (7, 10, 11). Inventors also found that the expression of several androgen-responsive genes is significantly altered in Pb-PRL prostates compared to WT mice (Fig. 5), showing that AR-signalling is globally decreased in this model. On the opposite, again based on the list of 144 genes defined as androgen-regulated in the prostate (10), AR-signaling was found unaffected in Hi-Myc mice (not shown), another established model of prostate cancer, using an independently generated set of tran scrip tomic data (GO: GSE53202, (12)).

Remarkably, the FACS profiles of Hi-Myc prostates bearing adenocarcinomas showed no LSC ^med enrichment when compared to Pten^ ^"7" or Pb-PRL prostates (Fig. 6). Confirming this result, the expression of LSC ^med signature genes in prostates from Hi-Myc mice was similar to WT mice, and markedly lower than in Pten ^{pc A} mice (Table 1, GO: GSE53202, (12)).

Together, these results show that LSC ^med enrichment prior to castration is specific to mouse models with decreased AR-signaling and not observed in all the model of prostate cancer. The detection of an enrichment, within a prostatic epithelial sample, even without any treatment, in cell exhibiting LSC ^med signature is therefore the sign of a probable resistance to androgen- deprivation therapy.

3. Cytokeratin 4 and 7 are protein markers for LSC

CK4 (Krt4 gene product, table 1), whose expression was validated by qPCR as overexpressed in LSC ^med only (Fig. 3), was tested as a marker to identify LSC ^med on tissue slides. To confirm that CK4 protein expression was specific for LSC ^med , all the three sorted cell populations (LSC ^med , luminal, basal/stem cells) were immunostained for CK4.

Only LSC ^med cells were found positive for CK4 (data not shown, IF experiment). LSC ^med also co-expressed CK8 but not CK5 proteins (not shown), consistent with gene expression studies and their luminal cell type characteristics.

In tissue sections of intact prostates, CK4 ⁺ cells are observed with an increased frequency from WT, to Pb-PRL, to Pten ^{pc /"} tissues (Fig. 7A, left column), consistent with FACS analyses (Fig. 1).

Castration is found to result in an increase of the frequency of CK4 ⁺ cells in WT and Pb-PRL prostates while their prevalence remained high in Pten ^{pc /"} mice (Fig. 7A, right column), which is also consistent with the proliferation of LSC ^med cells observed upon castration (Fig. l).

In WT prostates, CK4 ⁺ cells were found to be restricted to the proximal regions of the gland (Fig. 8), where stem/progenitor cells have been shown to reside (13). In contrast, in Pb-PRL mice, the CK4 ⁺ cells, beside an increased frequency, were found distributed throughout the proximal and distal ductal regions (Fig. 8). Notably, such a distribution is also observed for CK7 ⁺ cells, with an increased labelling of luminal epithelial cells within proximal as well as distal areas of the gland (Fig. 10A, circled areas).

Pten ^{pc /"} mouse tissues showed clusters of CK4 ⁺ cells located adjacent to negative regions, suggesting clonal expansion (Fig. 7A, left column black and white arrows, respectively). Accordingly, staining of serial sections for BrdU and CK4 showed much higher proliferation of CK4 ⁺ cells in intact Pten ^{pc /"} than in intact WT prostates (Fig 7B). Also, an intense CK7 labelling of numerous prostate epithelial cells is observed in these tissues, in accordance with the above observations that LSC ^med constitute the majority cell type within Pten ^{pc /"} mice prostate epithelia

(Fig. 10B). As for CK4 ⁺ labelling, clusters of CK7 ⁺ cells located adjacent to negative regions are also observed (not shown).

These results validate the use of CK4 and CK7 as a specific marker of LSC ^med , and the role of these cells in the emergence of androgen blockade resistance foci.

Hence, genes of the LSCmed signature as identified herein by the inventors provide valuable tools for use for detecting the risk, presence or progression of hormone refractory prostate cancer.

4. LSCmed ofPterf mice are tumor initiating cells. Inventors tested the tumorigenicity of LSC ^med cells in the in vivo prostate regeneration as exposed in the Protocol section A). The evolution of grafts made of LSC ^med cells or basal/stem cells of Pten ^{pc /"} mice mixed with UGSM cells was assayed 10 weeks after the subcutaneous cells engraftment into SCID mice.

As expected, the basal/stem cells grafts were found to form tumoral glandular structures: high grade Prostatic Intraepithelial Neoplasia (PIN) lesions with some foci of carcinoma in situ (not shown). They were characterized by epithelial cell stratification, nuclear atypia, cribriform gland. Noteworthy, immuno staining of CK5, CK4 and CK8 showed presence of basal cells, luminal and LSC ^med cells in the formed tumoral structures.

As expected also, transplantation of UGSM cells alone did not form discernible glandular structure.

More strikingly, Inventors made the observation that grafts from Pten ^{pc /"} LSC ^med cells were able to also form tumoral prostate epithelial structures (not shown). Indeed, up to now only basal/stem cells grafts were found to form tumoral glandular structures in this assay (17). Predominantly the histomorphology of these structures were found comparable to that of carcinoma in situ with focal micro invasion, PIN and extended atrophic glands. The tissue grafts were characterized by glandular structures with multiple atrophic, cystic glands, as well as less differentiated irregular arranged smaller glands. Focal non-atrophic irregular glands with loss of CK5 positive basal cell layer were found. Those glands showed micro-invasion through the basal membrane and cribriform luminal structures and were composed exclusively of CK4/CK8 positives cells as found for LSC ^med cell (not shown).

These results strongly support tumor-initiating features for LSC ^med cells. 5. LSC ^med ofPten ^{pc /} - mice massively proliferate under androgen depletion condition

CK4/BrdU staining was used to address whether Pten ^{pc /"} LSC ^med play a role in cancer relapse. CK4 ⁺ /BrdU ⁺ cells proliferation capacity was assayed three weeks after castration. In contrast to castrated WT prostates used as control, castrated Pten ^{pc /"} prostates exhibited large clusters of CK4 ⁺ /BrdU ⁺ cells located adjacent to negative regions (Fig. 7B), again suggesting clonal expansion. This analysis proved that LSC ^med of Pten ^{pc /"} mice not only survive castration, but massively proliferate under these conditions of androgen depletion.

Then, the identification of an enrichment in LSC ^med cells and/or the detection of an LSC ^med signature from a sample of a subject could help in predicting the responsiveness to androgen therapy and the risk of relapse and therefore in adapting the treatment of the subject.

Indeed, androgen deprivation therapy could be considered of no use in subject whose sample present such a signature and even unbefitting and potentially harmful given the proliferation of LSC ^med cells observed by the Inventors in castrated animals (Fig. 1 and Fig. 7B).

6. CK4-positive cells are present in androgen-deprived patient-derived xenografts

(PDXs)

To show that LSC ^med -like cells were present in human prostate cancer, PDXs of treatment- naive tumors, where castrate-tolerant cells reside (14), were searched for CK4 ⁺ cells in intact as well as castrated hosts. Figure 9 shows dual localization of AMACR ⁺ (Alpha-methylacyl-CoA racemase, a prostate cancer marker, Ref. 15), p63 ^" and CK4 ⁺ cells on serial sections (encircled regions) of both specimens. In addition, CK4 ⁺ cells were found present in primary tumor tissues (a representative section from a Gleason 4 prostate cancer is shown in Fig. 9B).

These results indicate that CK4 ⁺ cells are present in human prostatic tumors and are castrate- tolerant, consistent with a key role in the progression to castration-resistant prostate cancer.

REFERENCES

1. Sackmann-Sala L, Chiche A, Mosquera-Garrote N, et al. Prolactin- Induced Prostate Tumorigenesis Links Sustained Stat5 Signaling with the Amplification of Basal/Stem Cells and Emergence of Putative Luminal Progenitors. Am J Pathol 2014; 184: 3105-3119.

2. Pencik J, Schlederer M, Gruber W, et al. STAT3 regulated ARF expression suppresses prostate cancer metastasis. Nat Commun 2015; 6: 7736.

3. Rouet V, Bogorad RL, Kayser C, et al. Local prolactin is a target to prevent expansion of basal/stem cells in prostate tumors. Proc Natl Acad Sci U SA 2010; 107: 15199-15204.

4. Oudard, S., Banu, E., Beuzeboc, P., Voog, E., Dourthe, L.M., Hardy-Bessard, A.C., Linassier, C, Scotte, F., Banu, A., Coscas, Y., et al. (2005). Multicenter randomized phase II study of two schedules of docetaxel, estramustine, and prednisone versus mitoxantrone plus prednisone in patients with metastatic hormone-refractory prostate cancer. J Clin Oncol 23, 3343-3351.

5. Lawrence, M.G., Taylor, R.A., Toivanen, R., Pedersen, J., Norden, S., Pook, D.W., Frydenberg, M., Papargiris, M.M., Niranjan, B., Richards, M.G., et al. (2013). A preclinical xenograft model of prostate cancer using human tumors. Nat Protoc 8, 836-848.

6. Toivanen, R., Frydenberg, M., Murphy, D., Pedersen, J., Ryan, A., Pook, D., Berman, D.M., Taylor, R.A., and Risbridger, G.P. (2013). A preclinical xenograft model identifies castration- tolerant cancer-repopulating cells in localized prostate tumors. Sci Transl Med 5, 187ral71. 7. Wang S, Gao J, Lei Q, et al. Prostate-specific deletion of the murine Pten tumor suppressor gene leads to metastatic prostate cancer. Cancer Cell 2003; 4: 209-221.

8. Chen Y, Chi P, Rockowitz S, et al. ETS factors reprogram the androgen receptor cistrome and prime prostate tumorigenesis in response to PTEN loss. Nat Med 2013; 19: 1023-1029.

9. Wang G, Lunardi A, Zhang J, et al. Zbtb7a suppresses prostate cancer through repression of a Sox9-dependent pathway for cellular senescence bypass and tumor invasion. Nat Genet

2013; 45: 739-746.

10. Mulholland DJ, Tran LM, Li Y, et al. Cell autonomous role of PTEN in regulating castration- resistant prostate cancer growth. Cancer Cell 2011; 19: 792-804.

11. Carver BS, Chapinski C, Wongvipat J, et al. Reciprocal feedback regulation of PI3K and androgen receptor signaling in PTEN-deficient prostate cancer. Cancer Cell 2011; 19: 575- 586.

12. Ellwood-Yen K, Graeber TG, Wongvipat J, et al. Myc-driven murine prostate cancer shares molecular features with human prostate tumors. Cancer Cell 2003; 4: 223-238.

13. Tsujimura A, Koikawa Y, Salm S, et al. Proximal location of mouse prostate epithelial stem cells: a model of prostatic homeostasis. / Cell Biol 2002; 157: 1257-1265.

14. Toivanen R, Frydenberg M, Murphy D, et al. A preclinical xenograft model identifies castration-tolerant cancer-repopulating cells in localized prostate tumors. Sci Transl Med 2013; 5: 187ral71.

15. Browne TJ, Hirsch MS, Brodsky G, Welch WR, Loda MF, Rubin MA. Prospective evaluation of AMACR (P504S) and basal cell markers in the assessment of routine prostate needle biopsy specimens. Hum Pathol. 2004;35: 1462-8.

16. Xin L, Ide H, Kim Y, Dubey P, Witte ON. In vivo regeneration of murine prostate from dissociated cell populations of postnatal epithelia and urogenital sinus mesenchyme. Proc Natl Acad Sci U S A. 2003;100 Suppl 1: 11896-903.

17. Lawson DA, Zong Y, Memarzadeh S, Xin L, Huang J, Witte ON. Basal epithelial stem cells are efficient targets for prostate cancer initiation. Proc Natl Acad Sci U S A. 2010;6:2610-5.