TALEN-engineered AR gene rearrangements revealendocrine uncoupling of androgen receptorin prostate cancerMichael D. Nyquista,b, Yingming Lia, Tae Hyun Hwangc,1, Luke S. Manlovea,b, Robert L. Vessellad,e,Kevin A. T. Silversteina,c,f, Daniel F. Voytasg,h,2, and Scott M. Dehma,i,2

aMasonic Cancer Center, bGraduate Program in Molecular, Cellular, and Developmental Biology and Genetics, and cBiostatistics and Bioinformatics Core,Masonic Cancer Center, University of Minnesota, Minneapolis, MN 55455; dDepartment of Urology, University of Washington Medical Center, Seattle, WA98195; ePuget Sound VA Health Care System, Seattle, WA 98108; and fSupercomputing Institute for Advanced Computational Research, gDepartment ofGenetics, Cell Biology, and Development, hCenter for Genome Engineering, and iDepartment of Laboratory Medicine and Pathology, University of Minnesota,Minneapolis, MN 55455

Edited by Michael G. Rosenfeld, University of California, San Diego, La Jolla, CA, and approved September 10, 2013 (received for review May 6, 2013)

Androgen receptor (AR) target genes direct development and sur-vival of the prostate epithelial lineage, including prostate cancer(PCa). Thus, endocrine therapies that inhibit the AR ligand-bindingdomain (LBD) are effective in treating PCa. AR transcriptional reac-tivation is central to resistance, as evidenced by the efficacy of ARretargeting in castration-resistant PCa (CRPC) with next-genera-tion endocrine therapies abiraterone and enzalutamide. However,resistance to abiraterone and enzalutamide limits this efficacy inmost men, and PCa remains the second-leading cause of malecancer deaths. Here we show that AR gene rearrangements inCRPC tissues underlie a completely androgen-independent, yetAR-dependent, resistance mechanism. We discovered intragenicAR gene rearrangements in CRPC tissues, which we modeled usingtranscription activator-like effector nuclease (TALEN)-mediatedgenome engineering. This modeling revealed that these AR generearrangements blocked full-length AR synthesis, but promotedexpression of truncated AR variant proteins lacking the AR li-gand-binding domain. Furthermore, these AR variant proteinsmaintained the constitutive activity of the AR transcriptional pro-gram and a CRPC growth phenotype independent of full-length ARor androgens. These findings demonstrate that AR gene rearrange-ments are a unique resistance mechanism by which AR transcrip-tional activity can be uncoupled from endocrine regulation in CRPC.

Prostate cancer (PCa) is the most commonly diagnosed cancerin men and is the second- leading cause of male cancer

mortality (1). The androgen receptor (AR) is a steroid receptortranscription factor that drives prostate development and ho-meostasis and is crucial for PCa growth and survival (2). The ARgene is located on the X chromosome and encodes a modularprotein consisting of three major domains. Exon 1 encodes theNH2-terminal domain (NTD), exons 2 and 3 encode a centralDNA-binding domain (DBD), and exons 4–8 encode theCOOH-terminal ligand-binding domain (LBD). The NTD isresponsible for the majority of AR transcriptional activity, butthis activity is suppressed by the LBD unless AR is bound totestosterone or dihydrotestosterone (DHT) (3–7). Currently, ad-vanced or metastatic PCa is treated by systemic inhibition of an-drogen synthesis and antiandrogens that bind to the AR LBD (8);however, despite robust responses to these endocrine therapies,castration-resistant PCa (CRPC) inevitably develops concurrentwith AR transcriptional reactivation (9, 10).Identification of AR overexpression and high tissue androgen

levels as mechanisms driving AR reactivation in a subset ofCRPC tumors led to the clinical development and recent ap-proval of abiraterone and enzalutamide as new endocrine tar-geting therapies for treatment of CRPC (11–17). However,despite the success of these drugs at improving overall survival,primary and secondary resistance remains a major limitationfor most patients. Point mutations in the AR LBD have been

implicated in resistance to enzalutamide in CRPC cell linemodels (18). Similarly, increased intratumoral steroidogenesishas been observed in CRPC xenograft models that have de-veloped resistance to abiraterone (19). These findings supportcontinued efforts to block the ligand–LBD interaction to achievedurable AR inhibition.An additional occurrence in CRPC is the production of

COOH-terminally truncated AR splice variants (AR-Vs). Di-verse AR-Vs have been reported, all of which contain the ARNTD and DBD but lack the AR LBD (20). Functional studieshave demonstrated that AR-Vs can function as constitutivelynuclear, constitutively active transcription factors (21). AR-Vsare known to be enriched in CRPC and track with poor clinicaloutcomes (22–25). Recently, AR gene rearrangements wereidentified in CRPC cell lines that display high-level expression ofAR-Vs (26, 27). However, AR-Vs are coexpressed with full-length AR in these cell lines, making pinpointing the precisecontributions of AR-Vs to endocrine therapy resistance and aCRPC phenotype challenging. Here we report intragenic ARgene rearrangements in tissues from clinical CRPC metastasesthat completely block full-length AR synthesis. We modeled

Significance

The androgen receptor (AR) is a master regulator in cells ofprostatic origin, including prostate cancer. How AR activity canpersist in tumors that are resistant to second-generation AR-targeted therapies remains unknown. This study describes thediscovery of AR gene rearrangements in clinical prostate cancertissues, and the use of genome engineering in prostate cancercells with transcription activator-like effector nucleases tofunctionally classify these gene rearrangements as drivers ofresistance. This knowledge is expected to lead to better patientmanagement and enable the development of more effectivetherapies for advanced prostate cancer.

Author contributions: M.D.N., Y.L., D.F.V., and S.M.D. designed research; M.D.N., Y.L., andL.S.M. performed research; R.L.V. contributed new reagents/analytic tools; M.D.N.,Y.L., T.H.H., K.A.T.S., D.F.V., and S.M.D. analyzed data; and M.D.N., Y.L., D.F.V., andS.M.D. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The data reported in this paper have been deposited in the Gene Ex-pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE49196).1Present address: Quantitative Biomedical Research Center, University of Texas South-western Medical Center, Dallas, TX.

2To whom correspondence may be addressed. E-mail: voytas@umn.edu or dehm@umn.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1308587110/-/DCSupplemental.

17492–17497 | PNAS | October 22, 2013 | vol. 110 | no. 43 www.pnas.org/cgi/doi/10.1073/pnas.1308587110

Dow

nloa

ded

by g

uest

on

Aug

ust 2

2, 2

020

these intragenic rearrangements using transcription activator-like effector nuclease (TALEN) genome engineering (28), andfound that these rearrangements underlie exclusive expression oftruncated AR-Vs and uncoupling of the AR transcriptionalprogram from endocrine regulation. Overall, this study charac-terizes a completely androgen-independent, AR-V–dependentresistance mechanism, and establishes the clinical relevance anda functional context of AR-Vs in CRPC.

ResultsAR-V–Associated Intragenic AR Rearrangements in Clinical CRPC Tissue.LuCaP 86.2 is a PCa xenograft established from a CRPC bladdermetastasis resected from a 79-y-old patient (29). This xenograftexpresses full-length AR encoded by exons 1–8, and also expressesan AR-V protein thought to be an alternative splicing productarising from skipping of AR exons 5–7 (ARv567es) (30). However,because AR is on the X chromosome, discovery of an 8.5-kb de-letion of AR exons 5–7 in LuCaP 86.2 tissue indicates thatARv567es and full-length AR expression may be mutually exclu-sive and restricted to distinct cell populations (27). In line with thisidea, FISH and immunofluorescence staining demonstrated thatall LuCaP 86.2 cells harbored a single AR gene copy and stainedpositive with an antibody specific for the AR NTD (Fig. S1 A andB). However, high-resolution AR copy number analysis usinga multiplex ligation-dependent probe amplification (MLPA) assayrevealed deletion of the AR exon 5–7 segment encoding the ARLBD in approximately half of these cells (Fig. S1C). The relativeratios of deletion-positive and deletion-negative cells remainedstable during long-term propagation in castrated mice, as did ex-pression ratios of full-length AR and ARv567es (Fig. S1 C and D).Taken together, these findings demonstrate the existence of atleast two stable CRPC subclones in heterogeneous LuCaP 86.2tissue, one of which harbors an intragenic deletion of AR exons5–7. To establish the clinical relevance of this finding, we useda PCR-based method and verified heterogeneity for cells with the8.5-kb deletion breakpoint in both the LuCaP 86.2 xenograft andin CRPC cells procured by laser capture microdissection of ar-chival patient tissue (Fig. 1A and Fig. S2 A–C).Interestingly, a different PCa xenograft established from CRPC

abdominal ascites, LuCaP 136, also expresses ARv567es mRNAand protein (29, 30). Whole-exome resequencing of LuCaP 136genomic DNA (29) has not provided an obvious basis forARv567es expression, and we did not detect intragenic deletionof AR exons 5–7 by PCR analysis (Fig. S2D). Thus, we rese-quenced the 183-kb AR gene in LuCaP 136 genomic DNA viahybrid capture, followed by Illumina-based massively parallelpaired-end sequencing. This analysis revealed a copy-neutral8.7-kb inversion encompassing AR exons 5–7 (Fig. 1B, Fig. S3 A–C,and Dataset S1, Tables S1-3). In contrast to heterogeneous ARexpression in LuCaP 86.2 (Fig. 1 C and D and Fig. S4), early-passage LuCaP 136 tissue displayed exclusive expression ofARv567es mRNA (Fig. 1C) and protein (Fig. 1D), consistent withvery few cells harboring a normal AR allele (Fig. S3D). No archivalpatient material corresponding to LuCaP 136 was available, andthis intragenic AR inversion was discovered in tissue that had beenpropagated for only two passages in noncastrate male mice. Laterpassages of LuCaP 136 serially propagated in noncastrate malemice displayed coordinate loss of cells with this AR exon 5–7 in-version allele and AR v567es protein expression (Fig. S5).

Fig. 1. AR gene rearrangements linked to AR-V expression in CRPC. (A) PCRanalysis of an 8.5-kb deletion of AR exons 5–7 in genomic DNA isolatedfrom the LuCaP 86.2 xenograft model, CRPC bladder metastasis used toestablish LuCaP 86.2 (Met), or adjacent normal bladder (NB). H2O, water asa no template control. (B) An 8.7-kb inversion of AR exons 5–7 in passage2 of the LuCaP 136 xenograft, which was established from CRPC cells inabdominal ascites fluid. mh, microhomology. (C ) RT-PCR analysis of AR

mRNA in LNCaP cells, LuCaP 86.2 tissue, and LuCaP 136 tissue is shown in theLower panel. Exon organization and relationship with functional proteindomains for full-length AR (AR-FL) and the ARv567es splice variant is shownin the Upper panel. Heteroduplex formation in LuCaP 86.2 PCR products wasconfirmed (Fig. S3). (D) Western blots of the AR NTD and ERK-2 (loadingcontrol) in LNCaP cells, LuCaP 86.2 tissue, and LuCaP 136 tissue.

Nyquist et al. PNAS | October 22, 2013 | vol. 110 | no. 43 | 17493

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 2

2, 2

020

Targeting Nucleases Recreate ARv567es-Associated Genome Re-arrangements. To study the functional significance of theseunique AR gene rearrangements in CRPC tissues, we pursuedgenome engineering in a monoclonal subline isolated from theCWR-R1 PCa cell line (31), termed R1-AD1 (32). R1-AD1cells harbor one intact AR gene copy, express full-length AR,and exhibit growth stimulation in response to androgens andgrowth suppression in response to AR antagonists, includingenzalutamide (32). We constructed two TALEN pairs (28, 33,34), one pair targeting AR intron 4 (AR-int4) and the otherpair targeting AR intron 7 (AR-int7) (Fig. 2A and Fig. S6A).Site-specific dsDNA cleavage by AR-int4 and AR-int7 TAL-ENs was confirmed by assays measuring mutations introducedby TALEN cleavage and imprecise DNA repair (Fig. S6 B andC). When both AR-int4 and AR-int7 TALENs were coexpressed,deletion or inversion events involving the AR exon 5–7 segmentwere observed (Fig. S6 D–F).Using limiting dilution plating and PCR screening, we isolated

and expanded two clonal cell lines derived from R1-AD1, onewith a deletion (R1-D567) and the other with an inversion(R1-I567) of AR exons 5–7 (Fig. 2B). AR exon copy numberanalysis by MLPA demonstrated that R1-AD1 and R1-I567cells harbored one copy of each AR exon, whereas exons 5–7had been deleted in R1-D567 cells (Fig. 2C). Furthermore, thegenome-engineered R1-I567 and R1-D567 cell lines displayedexclusive expression of ARv567es variant mRNA (Figs. 2 D andE and Fig. S7) and protein (Fig. 2F). These findings confirmthat the rearrangements involving the AR exon 5–7 segmentdiscovered in CRPC tissues are causative events underlying theexpression of ARv567es.

ARv567es Drives Androgen Independence in Genome-Engineered CellLines. Because deletion or inversion of AR exons 5–7 eliminatesthe AR LBD, cells harboring these rearrangements should havea growth advantage over androgen-dependent PCa cells underconditions in which full-length AR activity is inhibited. To testthis idea, we performed competitive growth assays in which R1-AD1 cells were transfected with AR-int4 and AR-int7 TALENsand then cultured for 20 d in the presence of androgens [in fetalbovine serum (FBS) or charcoal-stripped serum (CSS) + 1 nMdihydrotestosterone (DHT)] or under conditions of full-lengthAR inhibition (in CSS or CSS + 1 uM enzalutamide). We usedquantitative PCR (qPCR) to measure the enrichment of rare cellpopulations harboring alleles with targeted inversions ordeletions relative to all cells. Maintenance under conditions offull-length AR inhibition resulted in the relative enrichmentof deletion- and inversion-positive cells. In contrast, underandrogen-rich conditions, deletion- and inversion-positive cellsdisplayed relative de-enrichment (Fig. 3A).A possible explanation for the foregoing findings is that de-

letion or inversion of AR exons 5–7 functionally inactivates AR,causing an AR bypass mechanism in CRPC (35). This would beconsistent with the notion that full-length AR is required forAR-dependent CRPC progression, even whenAR-Vs are expressed(36). An alternative explanation could be that ARv567es, whichretains the transcriptionally active AR NTD/DBD core (21) (Fig.1C), may be able to drive a completely ligand-independent, yet AR-dependent, CRPC phenotype. To test this idea, we performed re-porter assays with an AR-responsive mouse mammary tumor virus(MMTV) promoter. In parental R1-AD1 cells, MMTV displayedrobust androgen induction, which was blocked by knockdown ofFig. 2. Engineered inversion or deletion of AR exons 5–7 using TALENs

recapitulates tissue-associated AR splicing events. (A) Genome engineeringstrategy for creating isogenic cell lines harboring inversion or deletion ofAR exons 5–7. TALEN pairs targeted to AR intron 4 (AR-int4) or AR intron 7(AR-int7) are depicted as red arrowheads. (B) Representative Sanger se-quencing trace of the deletion junction from the genome-engineered R1-D567cell line and the 5′ inversion junction from the genome-engineered R1-I567cell line. (C) MLPA genomic copy number analysis in parental R1-AD1 cells andgenome-engineered R1-I567 and R1-D567 cells. (D) RT-PCR analysis of AR

mRNA in parental R1-AD1 cells and genome-engineered R1-I567 and R1-D567 cells. (E) Representative Sanger sequencing trace of the AR exon 4/8splice junction in ARv567es RT-PCR products from D. (F) Western blots of theAR NTD and ERK-2 (loading control) in parental R1-AD1 cells and genome-engineered R1-I567 and R1-D567 cells.

17494 | www.pnas.org/cgi/doi/10.1073/pnas.1308587110 Nyquist et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

2, 2

020

full-length AR (Fig. 3B). Conversely, in R1-D567 and R1-I567cells, MMTV displayed constitutively high basal activity, butno androgen induction. However, high basal MMTV activity inR1-I567 and R1-D567 cells was blocked by knockdown ofARv567es (Fig. 3B). These findings demonstrate that ARv567esis a constitutively active transcription factor in R1-I567 and R1-D567 cells.We next tested whether constitutive ARv567es transcrip-

tional activity could drive androgen-independent growth inR1-I567 and R1-D567 cells. DHT enhanced the growth ofparental R1-AD1 cells, which was inhibited by knockdown offull-length AR (Fig. 3 C and D). Conversely, R1-I567 and R1-D567 displayed robust androgen-independent growth, whichwas inhibited by knockdown of ARv567es (Fig. 3 C and D).These data demonstrate that these AR gene rearrangementsunderlie a shift from a growth profile that is dependent onandrogens and full-length AR to a growth profile driven byconstitutive ARv567es activity.

ARv567es Effects a Constitutive Form of the Androgen/AR TranscriptionalProgram. To determine whether ARv567es functions globally as aconstitutive transcriptional regulator, we performed gene expressionmicroarray analysis of parental R1-AD1 cells and R1-D567 cells.We first identified genes that were differentially expressed betweenfull-length AR “on” (1 nM DHT) and “off” (vehicle control) statesin R1-AD1 cells, then identified genes that were differentially ex-pressed between ARv567es on (control siRNA) and off (siRNAtargeting AR exon 1) states in R1-D567 cells. Overall, the majorityof genes that displayed differential expression between ARv567eson and off states in R1-D567 also displayed a similar directionalchange between the AR on and off states in parental R1-AD1 cells(Fig. 4A). This finding was confirmed for several shared target genes(FKBP5, FASN, and LIMA1) by direct qRT-PCR analysis (Fig.S8). In line with this, knowledge-based Ingenuity Pathway Analysis(IPA) revealed that the only significant multigene networks dis-playing differential expression in parental R1-AD1 cells or R1-D567 cells had AR as the prominent central hub (Fig. S9). Takentogether, these findings indicate that ARv567es supports constitu-tive activation of a transcriptional program very similar to theprogram regulated by full-length AR.

We tested this idea more rigorously using gene set enrichmentanalysis (GSEA) (37). Genes that were induced or repressed 1.2-fold by DHT in parental R1-AD1 cells were positively or nega-tively enriched, respectively, in the R1-D567 gene expressiondataset (Fig. 4B). Testing the reciprocal relationship yieldedidentical results; genes that were induced or repressed 1.2-foldby ARv567es in R1-D567 were positively or negatively enriched,respectively, in the R1-AD1 gene expression dataset (Fig. 4C andDataset S1, Tables S4 and S5). Based on these findings, weconclude that AR gene rearrangements are a CRPC resistancemechanism through which the AR transcriptional program canbe uncoupled from endocrine regulation.

DiscussionARv567es and other AR-Vs are expressed in CRPC, but themechanisms underlying this expression, and their functionalsignificance, are poorly understood (20). In previous studies, weshowed that intragenic AR rearrangements occur in CRPC celllines in which truncated AR-Vs were first discovered, explaininghigh-level AR-V expression concurrent with full-length AR inthese models (26, 27). For example, the 22Rv1 cell line harborsa 35-kb tandem duplication encompassing AR exon 3, and thisintragenic duplication is associated with high-level mRNA andprotein expression of the truncated AR-V7 (also referred to asAR3 and AR 1/2/3/CE3) and AR 1/2/3/2b (also referred to asAR-V4 and AR5) variants (21, 22, 24, 26, 38). The CRPC CWR-R1 cell line harbors a population of cells with a 48-kb intragenicdeletion of AR intron 1, and these deletion-positive cells displayhigh-level AR-V7 expression and resistance to enzalutamide (22,24, 27, 32). These findings are associative, however, and have notestablished a clear cause-and-effect relationship between AR generearrangements and functional AR-V expression. The data pre-sented here demonstrate that CRPC LuCaP 86.2 and LuCaP 136tissues also harbor intragenic rearrangements, associated with ex-pression of theARv567es variant. This study provides an importantbreakthrough, establishing a causal relationship between specificgenome rearrangements discovered in CPRC tissues and func-tionally significant AR-V expression. By modeling these specificintragenic AR gene rearrangements using a unique TALEN ge-nome engineering approach, we have demonstrated that thesegenomic events underlie a true androgen-independent phenotype

Fig. 3. ARv567es expression induced by AR generearrangements drives an androgen-independentPCa phenotype. (A) Relative enrichment of cellswith AR intragenic deletion or inversion events in-duced by transfecting R1-AD1 cells with AR-int4 andAR-int7 TALENs. Cells were grown to confluenceafter transfection and then replated on day 0 underandrogen-rich (FBS or CSS + 1 nM DHT) or castrate[CSS or CSS + 1 μM enzalutamide (Enz)] conditions.On day 20, plates were tested for enrichment ofcells with AR deletion or inversion alleles relative toall AR alleles (intron 2) by qPCR (n = 6). P valueswere calculated using the two-tailed t test. Errorbars represent SD. (B) Promoter-reporter assaysfollowing cotransfection with an AR-responsiveMMTV-luciferase reporter and siRNAs targeted toAR exon 7 or AR exon 1. Transfected cells weretreated for 24 h with 1 nM DHT as indicated. Resultsare presented as means (n = 3). P values were cal-culated using the two-tailed t test. Data are shownfor one triplicate experiment representative ofthree independent biological replicates, each ofwhich reached statistical significance (P < 0.05). Er-ror bars represent SD. (C) Western blots for the AR NTD or ERK-2 (loading control) in R1-D567 and parental R1-AD1 cells transfected with siRNAs targeting ARexon 1. (D) Growth assays of parental R1-AD1 cells and genome-engineered R1-I567 and R1-D567 cells transfected as in C (n = 4). P values shown are thelargest of any comparison between bracketed groups and were calculated using the two-tailed t test. Data are shown for one quadruplicate experimentrepresentative of two independent biological replicates, each of which reached statistical significance (P < 0.05). Error bars represent SD.

Nyquist et al. PNAS | October 22, 2013 | vol. 110 | no. 43 | 17495

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 2

2, 2

020

through a mechanism of switching AR expression from full-lengthAR to ARv567es. Furthermore, we have established ARv567es asthe active protein driving this androgen-independent phenotypeby effecting a broad androgen/AR transcriptional program.A rearrangement-dependent mechanism of AR-V expression

in CRPC tissues contrasts with the previous report of acuteincreases in AR-V7 expression in VCaP and late-passage LNCaPcells in response to endocrine targeting therapy (39), supportinggrowth of VCaP cells when full-length AR is inhibited (40). Thisplasticity in AR-V expression might be attributable to a negativefeedback loop in which transcription of the AR gene is activelyrepressed by androgen-bound full-length AR via recruitment oflysine-specific demethylase 1 (41). Based on this mechanism, ARinhibition would lead to transcriptional derepression and in-creased expression of both full-length AR and AR-V7. However,unlike AR gene rearrangements, this mechanism does not ap-pear to affect relative expression ratios, and AR-V7 remains aminor constituent of overall AR expression (39, 40).Tumor heterogeneity and subclonal architecture are not ap-

parent when lysates from cell lines or tissues are assessed formRNA or protein expression. A frequent finding in CRPC celllines and tissues, including LuCaP 86.2 and LuCaP 136, is theconcurrent expression of AR-Vs and full-length AR (22–25, 30).Modeling this coexpression by transfecting AR-Vs in LNCaPcells led to the conclusions that full-length AR and ARv567escan physically interact (30), that ARv567es facilitates nuclearlocalization and transcriptional activation of full-length AR un-der castrate conditions (30), that full-length AR is required forARv567es to drive features of the CRPC phenotype (30, 36), andthat antiandrogens such as enzalutamide can inhibit full-lengthAR and thereby inhibit any effects of AR-Vs (36). Our findingsunderscore the importance of understanding the subclonal ar-chitecture of CRPC cell lines and tumor tissue when assessingAR-V expression and function. Indeed, there have been conflictingreports about the association of AR-V expression levels with clinicaldisease progression (22–25, 42), and our findings suggest that this

could be related to the variable degree to which heterogeneousCRPC tumors are enriched for AR-V–expressing cells.The majority of men who progress on abiraterone and enzalu-

tamide display rising serum levels of the AR-regulated prostate-specific antigen gene, indicating that these tumors remain AR-driven (13, 15, 43). This finding has spurred ongoing development ofadditional endocrine therapies that act on the AR LBD (18, 44, 45).Importantly, the AR gene rearrangements discovered and modeledin the present study provide a complete genetic block of full-lengthAR and represent a previously unanticipated endocrine uncou-pling escape mechanism. Identification of AR-Vs as the ultimatefunctional outcome of this mechanism demonstrates the need fordevelopment of new therapies targeted to the AR-V core, whichis composed of the transcriptionally active AR NTD and ARDBD (46, 47). In addition, this work highlights an opportunity todevelop AR gene rearrangements as stable biomarkers of re-sistance to endocrine-targeting therapies for PCa.

Materials and MethodsTissues. Genomic DNA samples from the LuCaP series of PCa xenograftsand deidentified clinical CRPCa tissue were obtained from the University ofWashington’s Prostate Cancer Biorepository, which was developed andmanagedby one of the coauthors (R.L.V.) and has been described previously (25, 29, 30).

Cell Lines. The LNCaP (CRL-1740) cell line was obtained from American TypeCulture Collection, and CWR-R1 cells were provided by Dr. Elizabeth Wilson,University of North Carolina, Chapel Hill, NC. The R1-AD1 cell line (CWR-R1,androgen-dependent 1, referred to as “deletion-negative clone 1” in theoriginal publication) (27) is a subline derived from single-cell cloning of theCWR-R1 cell line. R1-AD1 cells are androgen-responsive and contain a structurallynormal copy of the AR gene (27, 32). Cells were maintained in in RPMI 1640medium supplemented with 10% (vol/vol) FBS in a 5% CO2 incubator at 37 °C.

MMTV-Luciferase Reporter Assays. Promoter-reporter assays with an MMTV-luciferase reporter were performed as described previously (38).

Cell Growth Assays. Cell growth was monitored by crystal violet staining asdescribed previously (26). For 20-d growth-enrichment assays, cells were elec-troporated with either a CMV-GFP expression vector and pBluescript stuffervector or equal masses of all four TALENs. Transfected cells were plated on six-well tissue culture plates and grown to confluence. Confluent cells were trypsi-nized and reseeded on six-well tissue culture plates in RPMI 1640 medium sup-plemented with 10% (vol/vol) FBS, 10% (vol/vol) CSS, 10% (vol/vol) CSS + 1 μMenzalutamide (Selleck Chemicals), or 10% (vol/vol) CSS + 1 nM DHT (Sigma-Aldrich). Cells were cultured for 20 d under these conditions with trypsinizationand reseeding when the cells became confluent. After 20 d, genomic DNA washarvested for analysis by qPCR.

Western Blot Analysis. Western blot analysis was performed as describedpreviously (26) using antibodies specific for the AR NTD (N-20; Santa CruzBiotechnology) and ERK2 (D-2, Santa Cruz Biotechnology).

Generation of R1-D567 and R1-I567 Cell Lines. For isolation of clonal cell lineswith targeted intragenic deletion or inversion of the AR exon 5–7 segment,cells were electroporated with TALENs, plated, and allowed to recover for3 d. Transfected cells were transferred to tissue culture dishes at limitingdilution. When colonies became visible (∼3 wk to 1 mo), cells were picked,trypsinized, and plated on 96-well plates. Clones were expanded and splitinto two separate plates, one of which was used for PCR screening. GenomicDNA was extracted using the QuickExtract Kit (catalog no. QE0905T; Epi-centre-Illumina) according to the manufacturer’s protocol. Genomic DNAwas then used for PCR screening with rearrangement-specific primers(Dataset S1). Positive clones were expanded. The clonal purity of R1-D567and R1-I567 cell lines was assessed by secondary PCR screening, and met thecriteria of positive PCR signals with rearrangement-specific primers andnegative PCR signals with WT-specific primers (Dataset S1).

MLPA. MLPA assays were performed as described previously (27).

Gene Expression Microarray Analysis. Gene expression microarray analysisusing Illumina Human HT-12 V4 BeadChips was performed as describedpreviously (32). Triplicate biological samples were used for analysis. R1-D567

Fig. 4. ARv567es induced by AR gene rearrangements drives constitutive,androgen-independent expression of the AR transcriptional program. (A)Heat map of microarray data showing genes differentially expressed in R1-D567 cells transfected with siCNTL vs. siAR targeting AR exon1 (ARv567estranscriptome), along with the responses of these same genes to 1 nM DHTin parental R1-AD1 cells. Data represent mean centered expression changesin log2 scale for three independent biological replicates. (B) GSEA interrogatingR1-D567 gene expression data for enrichment of genes that were either up-regulated by 1.2-fold (Upper, red) or down-regulated by 1.2-fold (Lower, blue)in R1-AD1 cells in response to treatment with 1 nM DHT. False discovery rate(FDR) q-values are shown for each plot. (C) GSEA as in B interrogating R1-AD1gene expression data for enrichment of genes that were either up-regulatedby 1.2-fold (Upper, red) or down-regulated by 1.2-fold (Lower, blue) in R1-D567cells transfected with siCNTL vs. siAR targeting AR exon 1.

17496 | www.pnas.org/cgi/doi/10.1073/pnas.1308587110 Nyquist et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

2, 2

020

cells were electroporated with either control siRNA (siCNTL) or siAR-exon1(Dataset S1). R1-AD1 cells were electroporated with siCNTL. Electroporatedcells were plated in RPMI 1640 medium supplemented with 10% CSS. After48 h, cells were switched to serum-free RPMI 1640 medium supplementedwith 1 nM DHT or 0.1% (vol/vol) ethanol (vehicle control) for 24 h. Gene listswere derived from genes displaying an ≥1.2-fold expression increase ordecrease (P = ≤ 0.05 with Benjamini–Hochberg false discovery rate correc-tion) in R1-AD1 1 nM DHT vs. EtOH and in R1-D567 siCNTL vs. siAR-exon1.Heat maps were generated using Cluster 3.0 software (http://bonsai.hgc.jp/∼mdehoon/software/cluster/software.htm). Data are available at the Na-tional Center for Biotechnology Information’s Gene Expression Omnibus(accession no. GSE49196).

GSEA. Differentially expressed genes were analyzed for enrichment in nor-malized gene expression datasets using GSEA analysis software v2.0.10(http://www.broadinstitute.org/gsea/index.jsp). Normalized gene expressiondata were ranked using the Signal2Noise metric, and GSEA was performedagainst 1,000 random gene set permutations. All other analysis options werekept at the default setting.

Additional Materials and Methods. Detailed information on FISH, immuno-fluorescence staining, next-generation paired-end AR gene resequencing,transient transfections, RT-PCR, genomic PCR, cloning of PCR products, TALENnuclease construction, Surveyor nuclease assays, and microarray samplepreparation and analysis is available in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank Colm Morrissey and Lisha Brown at theUniversity of Washington for facilitating tissue procurement; the Universityof Minnesota Biomedical Genomics Core facility for performing AR generesequencing and microarray analyses; the Minnesota SupercomputingInstitute for providing computing, bioinformatics, software, and datastorage support for this project; and LeAnn Oseth and Betsy Hirsch forperforming FISH and MLPA analyses in the Cytogenetics Shared ResourcesLaboratory at the Masonic Cancer Center, with support from ComprehensiveCancer Support Grant P30 CA077598. This work was supported by grantsfrom the National Institutes of Health (R01 CA174777, to S.M.D., and R01GM098861, to D.F.V.), an American Cancer Society Research Scholar Grant(RSG-12-031-01, to S.M.D.), a Department of Defense Prostate CancerResearch Program New Investigator Award (W81XWH-10-1-0353, to S.M.D.),and a Novel Methods pilot grant from the University of Minnesota Clinicaland Translational Science Award Award (8UL1TR000114-02, to S.M.D.). S.M.D.is a Masonic Scholar of the Masonic Cancer Center, University of Minnesota.

1. Siegel R, NaishadhamD, Jemal A (2013) Cancer statistics, 2013. CA Cancer J Clin 63(1):11–30.2. Garraway LA, Sellers WR (2006) Lineage dependency and lineage-survival oncogenes

in human cancer. Nat Rev Cancer 6(8):593–602.3. Callewaert L, Van Tilborgh N, Claessens F (2006) Interplay between two hormone-

independent activation domains in the androgen receptor. Cancer Res 66(1):543–553.4. Christiaens V, et al. (2002) Characterization of the two coactivator-interacting surfa-

ces of the androgen receptor and their relative role in transcriptional control. J BiolChem 277(51):49230–49237.

5. Dehm SM, Regan KM, Schmidt LJ, Tindall DJ (2007) Selective role of an NH2-terminalWxxLF motif for aberrant androgen receptor activation in androgen depletion-independent prostate cancer cells. Cancer Res 67(20):10067–10077.

6. He B, et al. (2004) Structural basis for androgen receptor interdomain and coactivatorinteractions suggests a transition in nuclear receptor activation function dominance.Mol Cell 16(3):425–438.

7. Jenster G, van der Korput HA, Trapman J, Brinkmann AO (1995) Identification of twotranscription activation units in the N-terminal domain of the human androgen re-ceptor. J Biol Chem 270(13):7341–7346.

8. Ryan CJ, Tindall DJ (2011) Androgen receptor rediscovered: The new biology andtargeting the androgen receptor therapeutically. J Clin Oncol 29(27):3651–3658.

9. Attard G, Richards J, de Bono JS (2011) New strategies in metastatic prostate cancer:Targeting the androgen receptor signaling pathway. Clin Cancer Res 17(7):1649–1657.

10. Chen Y, Clegg NJ, Scher HI (2009) Anti-androgens and androgen-depleting therapiesin prostate cancer: New agents for an established target. Lancet Oncol 10(10):981–991.

11. Attard G, et al. (2008) Phase I clinical trial of a selective inhibitor of CYP17, abir-aterone acetate, confirms that castration-resistant prostate cancer commonly remainshormone driven. J Clin Oncol 26(28):4563–4571.

12. Chen CD, et al. (2004) Molecular determinants of resistance to antiandrogen therapy.Nat Med 10(1):33–39.

13. de Bono JS, et al.; COU-AA-301 Investigators (2011) Abiraterone and increased sur-vival in metastatic prostate cancer. N Engl J Med 364(21):1995–2005.

14. Montgomery RB, et al. (2008) Maintenance of intratumoral androgens in metastaticprostate cancer: A mechanism for castration-resistant tumor growth. Cancer Res68(11):4447–4454.

15. Scher HI, et al.; AFFIRM Investigators (2012) Increased survival with enzalutamide inprostate cancer after chemotherapy. N Engl J Med 367(13):1187–1197.

16. Titus MA, Schell MJ, Lih FB, Tomer KB, Mohler JL (2005) Testosterone and dihy-drotestosterone tissue levels in recurrent prostate cancer. Clin Cancer Res 11(13):4653–4657.

17. Tran C, et al. (2009) Development of a second-generation antiandrogen for treatmentof advanced prostate cancer. Science 324(5928):787–790.

18. Balbas MD, et al. (2013) Overcoming mutation-based resistance to antiandrogenswith rational drug design. eLife 2:e00499.

19. Mostaghel EA, et al. (2011) Resistance to CYP17A1 inhibition with abiraterone incastration-resistant prostate cancer: Induction of steroidogenesis and androgen re-ceptor splice variants. Clin Cancer Res 17(18):5913–5925.

20. Dehm SM, Tindall DJ (2011) Alternatively spliced androgen receptor variants. EndocrRelat Cancer 18(5):R183–R196.

21. Chan SC, Li Y, Dehm SM (2012) Androgen receptor splice variants activate AR targetgenes and support aberrant prostate cancer cell growth independent of the canonicalAR nuclear localization signal. J Biol Chem 287(23):19736–19749.

22. Guo Z, et al. (2009) A novel androgen receptor splice variant is up-regulated duringprostate cancer progression and promotes androgen depletion-resistant growth.Cancer Res 69(6):2305–2313.

23. Hörnberg E, et al. (2011) Expression of androgen receptor splice variants in prostatecancer bone metastases is associated with castration-resistance and short survival.PLoS ONE 6(4):e19059.

24. Hu R, et al. (2009) Ligand-independent androgen receptor variants derived from splicingof cryptic exons signify hormone-refractory prostate cancer. Cancer Res 69(1):16–22.

25. Zhang X, et al. (2011) Androgen receptor variants occur frequently in castration-resistant prostate cancer metastases. PLoS ONE 6(11):e27970.

26. Li Y, et al. (2011) Intragenic rearrangement and altered RNA splicing of the androgenreceptor in a cell-basedmodel of prostate cancer progression. Cancer Res 71(6):2108–2117.

27. Li Y, et al. (2012) AR intragenic deletions linked to androgen receptor splice variantexpression and activity in models of prostate cancer progression. Oncogene 31(45):4759–4767.

28. Cermak T, et al. (2011) Efficient design and assembly of custom TALEN and other TALeffector-based constructs for DNA targeting. Nucleic Acids Res 39(12):e82.

29. Kumar A, et al. (2011) Exome sequencing identifies a spectrum of mutation fre-quencies in advanced and lethal prostate cancers. Proc Natl Acad Sci USA 108(41):17087–17092.

30. Sun S, et al. (2010) Castration resistance in human prostate cancer is conferred by a fre-quently occurring androgen receptor splice variant. J Clin Invest 120(8):2715–2730.

31. Gregory CW, Johnson RT, Jr., Mohler JL, French FS, Wilson EM (2001) Androgen re-ceptor stabilization in recurrent prostate cancer is associated with hypersensitivityto low androgen. Cancer Res 61(7):2892–2898.

32. Li Y, et al. (2013) Androgen receptor splice variants mediate enzalutamide resistancein castration-resistant prostate cancer cell lines. Cancer Res 73(2):483–489.

33. Bogdanove AJ, Voytas DF (2011) TAL effectors: Customizable proteins for DNA tar-geting. Science 333(6051):1843–1846.

34. Miller JC, et al. (2011) A TALE nuclease architecture for efficient genome editing. NatBiotechnol 29(2):143–148.

35. Miyamoto DT, et al. (2012) Androgen receptor signaling in circulating tumor cells asa marker of hormonally responsive prostate cancer. Cancer Discov 2(11):995–1003.

36. Watson PA, et al. (2010) Constitutively active androgen receptor splice variants ex-pressed in castration-resistant prostate cancer require full-length androgen receptor.Proc Natl Acad Sci USA 107(39):16759–16765.

37. Subramanian A, et al. (2005) Gene set enrichment analysis: A knowledge-based ap-proach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA102(43):15545–15550.

38. Dehm SM, Schmidt LJ, Heemers HV, Vessella RL, Tindall DJ (2008) Splicing of a novelandrogen receptor exon generates a constitutively active androgen receptor thatmediates prostate cancer therapy resistance. Cancer Res 68(13):5469–5477.

39. Hu R, et al. (2012) Distinct transcriptional programs mediated by the ligand-dependent full-length androgen receptor and its splice variants in castration-resistantprostate cancer. Cancer Res 72(14):3457–3462.

40. Liu LL, et al. (2013) Mechanisms of the androgen receptor splicing in prostate cancercells. Oncogene, in press.

41. Cai C, et al. (2011) Androgen receptor gene expression in prostate cancer is directlysuppressed by the androgen receptor through recruitment of lysine-specific deme-thylase 1. Cancer Cell 20(4):457–471.

42. Zhao H, et al. (2012) Transcript levels of androgen receptor variant AR-V1 or AR-V7 donot predict recurrence in patients with prostate cancer at indeterminate risk forprogression. J Urol 188(6):2158–2164.

43. Ryan CJ, et al. (2013) Abiraterone in metastatic prostate cancer without previouschemotherapy. N Engl J Med 368(2):138–148.

44. Clegg NJ, et al. (2012) ARN-509: A novel antiandrogen for prostate cancer treatment.Cancer Res 72(6):1494–1503.

45. DeVore NM, Scott EE (2012) Structures of cytochrome P450 17A1 with prostate cancerdrugs abiraterone and TOK-001. Nature 482(7383):116–119.

46. Andersen RJ, et al. (2010) Regression of castrate-recurrent prostate cancer by a small-molecule inhibitor of the amino-terminus domain of the androgen receptor. CancerCell 17(6):535–546.

47. Dehm SM, Tindall DJ (2007) Androgen receptor structural and functional elements:Role and regulation in prostate cancer. Mol Endocrinol 21(12):2855–2863.

Nyquist et al. PNAS | October 22, 2013 | vol. 110 | no. 43 | 17497

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 2

2, 2

020