SLX4IP Promotes Telomere Maintenance in Androgen Receptor–Independent Castration-Resistant Prostate Cancer through ALT-like Telomeric PML Localization

Cell culture

Cell lines were obtained from the ATCC and maintained at 37°C in 5% CO₂. U2OS, HCT116, WI-38 VA13, and HEK293T cells were cultured in DMEM (Gibco) with 10% [volume/volume (v/v)] FBS (Sigma), 1% of penicillin 10,000 U/mL, streptomycin 10,000 μg/mL, and 25 μg/mL Amphotericin B Mixture (Antibiotic-Antimycotic, Gibco). C4-2B, CWR-R1, 22Rv1, DU145, and PC-3 cells were cultured in RPMI1640 (Gibco) with 10% FBS and 1% antibiotic-antimycotic. LAPC4 cells were cultured in Iscove’s modified Dulbecco’s Medium (Sigma) with 10% (v/v) FBS (Sigma) and 1 nmol/L Dihydrotestosterone (Sigma). For androgen-deprived conditions, C4-2B cells were cultured in RPMI1640 with no Phenol Red (Gibco) with 10% Charcoal-Stripped FBS (Sigma) and 1% antibiotic-antimycotic. Parental and genetically modified cell lines were subjected to Mycoplasma testing via MycoAlert PLUS Detection Kit (July 2018, Lonza) and short tandem repeat validation [April 2019, Case Western Reserve University (CWRU) Genomics Core, Cleveland, OH]. To determine population doubling (PD) times, cells were plated and trypsinized at 72 and 120 hours. Using trypan blue exclusion, viable cell number was determined by using the Countess II (Thermo Fisher Scientific). All experiments, including designated early-passage timepoints, were completed with cells from PD 2 through 8. Late-passage timepoints were completed with cells from the latest PD included in each telomere length experiment or after.

Stable cell line generation

To generate cell lines stably overexpressing 3X FLAG-tagged-SLX4IP, a gBlock was designed to incorporate a 3X FLAG tag. The 3X FLAG-SLX4IP DNA sequence (NCBI nucleotide, NM_001009608.3) was flanked with a SalI endonuclease restriction site on the C-terminus and a BamHI restriction site on the N-terminus. The SLX4IP open reading frame was engineered for bacterial codon optimization using the gBlock Gene Fragment design tool (Integrated DNA Technologies). The gBlock was cloned into pBABE-puro, a gift from Hartmut Land, Jay Morgenstern, and Bob Weinberg (Addgene, #1764; ref. 42). Retrovirus was produced using Lipofectamine 2000 (Thermo Fisher Scientific) to cotransfect pBABE-puro or pBABE-puro.SLX4IP.3XFLAG with pCMV-VSV-G (at a ratio of 6:1, Clontech) into GP2-293T cells (Clontech). Virus was harvested and cleared via centrifugation and/or sterile filtration at 48 hours posttransfection and infections were carried out in the presence of 10 μg/mL of Polybrene (Sigma) followed by selection in 1 (C4-2B) or 1.5 μg/mL (DU145 and PC-3) of Puromycin (Sigma). To generate cell lines stably overexpressing telomerase, an identical strategy was used as outlined above using pBABE-puro.UThTERT+U3-hTR-500, a gift from Kathleen Collins (Addgene#27665; ref. 43).

To generate cell lines with stable knockdown of SLX4IP, two short hairpin RNAs (shRNA) targeting SLX4IP in the pLKO.1-puro lentiviral expression plasmid were purchased from Sigma (clone IDs: NM_001009608.1-426s1c1 and NM_001009608.1-247s1c1). Lentivirus was produced by using Lipofectamine 2000 (Thermo Fisher Scientific) to cotransfect HEK293T cells with pLKO.1-puro Non-Mammalian shRNA control (Sigma, #SCH002), pLKO.1-puro. shSLX4IP.1, or pLKO.1-puro.shSLX4IP.2 with pMD2.G, pRRE, and pRSV-Rev, which were gifts from Didier Trono (at a ratio of 4:1:1:1, Addgene, #12259, #12251, #12253; ref. 44). Viral infections and selection were carried out as described above.

Mouse xenograft studies and IHC

In vivo experiments were performed with approval from the Institutional Animal Care and Use Committee at CWRU (Protocol 2017–0126, Cleveland, OH), which is certified by the American Association of Accreditation for Laboratory Animal Care. Six-week-old male nude mice (NCR nu/nu), bred on-site at The Athymic Animal & Preclinical Therapeutics Facility at CWRU, were inoculated with a 1:1 mix of Matrigel (Gibco, 354234) and 1 × 10⁶ cells subcutaneously. Mouse weights and tumor volumes were determined three times weekly. Tumor volumes were calculated using the hemiellipsoid volume formula following Vernier caliper measurements of tumor diameters in three dimensions. At 30 days postinoculation, tumors were excised and processed for IHC and telomerase activity. IHC and hematoxylin and eosin staining were carried out as published previously (45). Primary antibody dilutions used are as follows: SLX4IP (1:200, Sigma, HPA046372), CDKN1A or p21 (1:500, Cell Signaling Technology, 2947), MKI67 or Ki-67 (1:1,000, Sigma, HPA001164), and active caspase-3 (1:1,000, Millipore, AB3623). Three fields of view for each slide were imaged using an Olympus BX43 Upright Microscope and cells were manually counted and scored using ImageJ software (46).

RNA isolation and analysis

RNA was isolated using TRizol Reagent (Invitrogen), reverse transcribed using the High-Capacity RNA-to-cDNA Kit (Thermo Fisher Scientific), and quantified via qRT-PCR on an Applied Biosystems StepOnePlus Real-time PCR System as described previously (47). Established primer pairs for TERT and GAPDH were used (47).

Western blotting

Western blotting was completed as described previously (47). Briefly, 50 μg of lysate was resolved on a 4% to 20% Mini-PROTEAN TGX Gel (Bio-Rad) and transferred to Nitrocellulose Membrane (Millipore) using the Trans-Blot Turbo (Bio-Rad). Blots were blocked in 5% nonfat milk diluted in TBST and incubated overnight at 4°C with primary antibody. Blots were incubated for 1 hour at room temperature with either horseradish peroxidase–conjugated (Santa Cruz Biotechnology), IRDye 800CW (LI-COR), or IRDye 680RD (LI-COR) secondary antibodies diluted to 1:10,000. Development with ProtoGlow ECL Reagent (National Diagnostics) was used when necessary. Blots were imaged on an Odyssey Fc Imaging System (LI-COR), quantified using Image Studio 5.2 software, and normalized to GAPDH. Primary antibody dilutions used are as follows: 1:20,000 GAPDH (Cell Signaling Technology, 2118), 1:5,000 SLX4IP (Sigma, HPA046372), 1:5,000 FLAG (Sigma, F1804), 1:10,000 AR (Santa Cruz Biotechnology, sc-816), 1:3,000 ENO2 (Cell Signaling Technology, 9536), 1:5,000 p21 (Cell Signaling Technology, 2947), and 1:2,000 ATRX (Cell Signaling Technology, 14820).

Telomere-repeat amplification protocol

The real-time Q-TRAP assay was performed as published previously (48). Tumor lysates were prepared following homogenization with Kontes PELLET PESTLE Grinders (Kimble) under similar lysis conditions. Reactions were performed on an Applied Biosystems StepOnePlus Real-time PCR System. Collected C_t values were then converted to relative telomerase activity units using the telomerase-positive standard curve generated as described previously. Technique validation experiments were carried out with telomerase-positive HEK293T and HCT116 cells and telomerase-negative U2OS and WI-38 VA13 cells, with a standard curve generated using HCT116 cells (Supplementary Fig. S1A and S1B). Additional negative controls were generated by incubating telomerase-positive samples with 1 μg of RNase at 37° C for 20 minutes.

Immunofluorescence-FISH analysis

A previously published immunofluorescence-FISH (IF-FISH) protocol was carried out on cells seeded on sterile glass coverslips and tumor sections (47). Coverslips were incubated with PML antibody (Santa Cruz Biotechnology, sc-5621) diluted 1:200 for 1 hour at room temperature. Alexa-488–conjugated anti-rabbit secondary antibody (The Jackson Laboratory) was diluted 1:100 and added to coverslips for overnight incubation at 4°C. Hybridization with 333 ng/mL of a TelC-Cy5–labeled peptide nucleic acid (PNA) oligonucleotide telomere probe (N-CCTAACCTAACCTAA-C, PNA BIO) in PNA buffer (10 mmol/L Tris, pH 7.5 and 70% formamide) was carried out at 95°C for 5 minutes followed by overnight incubation at room temperature. Coverslips were stained with 2 μg/mL of DAPI for 10 minutes and mounted with Fluoromount-G (Thermo Fisher Scientific). Stained slides were imaged following blinding using the Leica HyVolution SP8 gated STED Microscope and LAS X software. Cell count and regions of interest (ROI) representing potential colocalization events were identified in ImageJ (46). These ROIs were analyzed via Coloc 2 using the Manders’ correlation algorithm (49), where a correlation coefficient of ≥0.25 was further analyzed. This coefficient represents the total PML signal relative to the intersection of the PML and telomeric signal in a given ROI. Following this analysis, ROIs were categorized as true APBs (coefficients of >0.85) versus ALT-like PML localization events (coefficients of 0.25–0.85). Each cell was categorized on the basis of highest ROI correlation coefficient found within the cell. Colocalization events within U2OS control cells were included for comparison.

C-circle isolation and analysis

A previously published C-circle dot blot assay was performed with minor modifications (50). Genomic DNA was harvested using the Genelute Mammalian Genomic DNA Miniprep Kit (Sigma) according to the manufacturer’s instructions. DNA (160 ng) was incubated under conditions described in the protocol published previously. Reactions were blotted onto Hybond Membranes (Thermo Fisher Scientific) and UV cross-linked (Spectrolinker XL-1000, Spectronics). Membranes were hybridized with 30 μg/mL of Digoxigenin (DIG) conjugated telomere probe (CCCTAACCCTAACCCTAACCCTAA-DIG, Integrated DNA Technologies) in Digoxigenin Easy Hyb Buffer (Roche) overnight at 37°C. Saline-sodium citrate (SSC, 0.5×) buffer with 0.1% SDS was used to remove excess probe. Membranes were blocked in Digoxigenin Blocking Buffer (Roche) for 30 minutes followed by incubation with 1:3,000 dilution of Digoxigenin-AP, Fab Fragments (Roche) for 3 hours at room temperature. Digoxigenin Wash Buffer (Roche) was used to remove excess antibody and amplified C-circles were detected with CDP-Star Kit (Roche). Membranes were imaged with the Odyssey Fc Imager (LI-COR) and signal intensity ratios were calculated as described previously using Image Studio 5.2 Software. Technique validation experiments were carried out with 160 ng of DNA from telomerase-positive HEK293T, HCT116, and A549 cells and a standard curve of DNA from ALT-positive U2OS and WI-38 VA13 cells (Supplementary Fig. S1C).

Telomere restriction fragment analysis

Genomic DNA was harvested using the GenElute Mammalian Genomic DNA Miniprep Kit (Sigma) according to the manufacturer’s instructions and telomere restriction fragment (TRF) analysis was carried out as published previously (47). Membranes were hybridized as described in the above C-circle analysis protocol at 42°C. SSC (2×) buffer with 0.1% SDS was used to remove excess probe. Membranes were blocked in Digoxigenin Blocking Buffer (Roche) for 30 minutes followed by incubation with 1:10,000 dilution of Digoxigenin-AP, Fab Fragments (Roche) for 3 hours at room temperature. Additional membrane processing and imaging steps were carried out as described above in the C-circle analysis protocol.

Senescence-associated β-galactosidase activity staining

Cells were seeded on sterile glass coverslips and grown to 80% confluency. Senescence β-galactosidase Staining Kit was used following the manufacturer’s protocol (Cell Signaling Technology). Slides were imaged following blinding using an Olympus BX43 Upright Microscope and cells were manually counted and scored using ImageJ software (46).

Statistical analysis

Statistical analyses were performed using two-tailed Student t test (in vitro and in vivo data comparing two groups) or one-way ANOVA with multiple comparisons (in vitro data comparing at least three groups) in GraphPad Prism 8 software. Patient expression data were analyzed through the Prostate Cancer Transcriptome Atlas (PCTA) on the basis of disease course using rank-sums test between metastatic CRPC (mCRPC) and primary subsets (www.thepcta.org; ref. 51). P values less than 0.05 were considered statistically significant. All in vitro data are represented as mean values from three independent experiments performed in triplicate, unless otherwise noted.

AR-independent CRPC in vitro models exhibit an atypical ALT-like phenotype

Because of the understudied nature of ALT’s contribution to telomere maintenance in CRPC, we sought to determine whether TMM phenotype correlated with AR dependence. Five in vitro CRPC models were interrogated for telomerase expression, activity, ALT hallmarks, and telomere length distribution. C4-2B, CWR-R1, and 22Rv1 cell lines retained AR expression (AR-dependent; refs. ^{52, 53}), whereas DU145 and PC-3 cells exhibited AR loss (AR-independent; refs. ^{54, 55}). Despite being castration sensitive, LAPC4 cells represent a characterized telomerase-positive AR-dependent model capable of acquiring ALT hallmarks (56), and were included as controls. Relative mRNA expression of TERT and telomerase activity across cell lines were analyzed alongside telomerase-positive HCT116 and ALT-positive U2OS controls (Fig. 1A and B; Supplementary Fig. S1D). Interestingly, significant variability was noted among CRPC cell lines regardless of AR status.

With the exception of C4-2B cells, CRPC cell lines exhibited limited telomerase activity. Based on this result, we reasoned ALT-like processes might be engaged to assist with telomere maintenance in CRPC cells with limited telomerase activity. As ALT activity cannot be directly measured, we evaluated surrogate measures of ALT activity, including APB abundance, C-circles, and telomere length heterogeneity (14, 21, 22). APBs were visualized via IF-FISH for ALT-associated PML protein localizing at the telomere, which is highly indicative of ALT-mediated HR events (21). Traditionally, APBs are instances of nearly complete overlap of PML foci with ultra-bright telomeric foci illustrated in the ALT-positive U2OS control (Fig. 1C and D; ref. 21). However, we identified atypical telomeric PML localization events exhibiting only partial overlap, representing an ALT-like hallmark (Fig. 1C and D). AR-independent DU145 and PC-3 cell lines had a modest proportion of APB-positive cells at approximately 7% and 9%, respectively, compared with ALT-positive control, with approximately 39%; however, AR-dependent cell lines had significantly fewer APB-positive cells ranging from approximately 0.3% to 1.8% (Fig. 1C and D). In addition, DU145 and PC-3 cell lines had a significantly greater proportion of cells with ALT-like PML localization events at approximately 22% and 34%, respectively, whereas AR-dependent cell lines exhibited these events infrequently, ranging from approximately 1.2% to 8.7%. To complement this finding, telomeric C-circles, which are byproducts of ALT-related events, were evaluated via C-circle amplification (CCA; ref. 50). A signal ratio greater than or equal to two when comparing the CCA reaction with control is highly indicative of ALT (50), and only AR-independent cell lines reached this threshold, albeit at a significantly lower level than U2OS control (Fig. 1E). Finally, telomere length was evaluated to determine whether CRPC cell lines contain long, heterogeneous telomeres, a known characteristic of traditional ALT-positive cells (14). Interestingly, DU145 and PC-3 telomeres did not mimic the telomere length distribution present in the U2OS control, and only exhibited an average telomere length of approximately 4.9 and 6 kb, respectively (Fig. 1F). Moreover, they did not differ dramatically from the telomere length distribution patterns of their AR-dependent counterparts.

Taken together, these data indicate CRPC cell lines exhibit variable telomerase expression and activity. Those that were AR-dependent did not contain an appreciable amount of ALT-like hallmarks, supporting telomerase as the primary TMM. Conversely, in DU145 and PC-3 cells lacking AR expression, relatively low telomerase activity was accompanied by an atypical ALT-like phenotype, including ALT-like telomeric PML localization events and C-circles, but lacking long, heterogeneous telomeres.

SLX4IP expression correlates with AR independence and ALT-like hallmarks

ALT hallmarks, including those in CRPC (27), often correlate with loss of chromatin remodeler, ATRX. While ATRX deletions are potential ALT-initiating events, these genomic alterations are not universally causative of ALT hallmarks across in vitro models (56, 57). Mutational analysis of ATRX (58-60) and protein expression across CRPC cell lines revealed DU145 and PC-3 cells, which display an atypical ALT-like phenotype, retain wild-type ATRX expression (Fig. 2A). Thus, ATRX loss may be responsible for ALT initiation in some cases, but is noncontributory to the ALT-like hallmarks that we observed. As such, additional ALT regulators were investigated.

The relatively uncharacterized protein, SLX4IP, was identified as a promoter of ALT-mediated telomere maintenance in breast cancer models (41); however, its role in CRPC remains ambiguous. Inspection of SLX4IP expression data from the PCTA (51) revealed significantly higher expression in patients with mCRPC, which includes both AR-dependent and -independent cases, when compared with those with primary disease (Supplementary Fig. S2). Interestingly, relative SLX4IP protein expression in DU145 and PC-3 cells, which exhibit ALT-like hallmarks and AR independence, was significantly higher than in AR-dependent cell lines that lack ALT-like hallmarks (Fig. 2A and B). The correlation among elevated SLX4IP expression, ALT-like hallmarks, and known role of SLX4IP in telomere maintenance supports SLX4IP as a potential regulator of the observed atypical ALT phenotype in AR-independent CRPC in vitro (37, 38, 41).

To investigate the ability of SLX4IP to promote ALT-like hallmarks in CRPC cell lines, a 3X FLAG-tagged SLX4IP construct was stably overexpressed via retroviral transduction in C4-2B (C4-2B.SLX4IP) and PC-3 cells (PC-3.SLX4IP; Fig. 2C). In C4-2B.SLX4IP cells, a significant reduction in TERT expression and telomerase activity was concomitant with SLX4IP overexpression (Fig. 2D and E; Supplementary Fig. S2). This effect was diminished in PC-3 cells that have limited telomerase expression and activity at baseline (Fig. 2D and E; Supplementary Fig. S2). As predicted, SLX4IP overexpression coincided with a significant increase in the percentage of cells with ALT-like PML localization events in both C4-2B.SLX4IP and PC-3.SLX4IP cell populations (Fig. 2F and G). Although modest increases in the percentage of APB-positive cells were noted with SLX4IP overexpression, the changes were deemed not statistically significant. Finally, SLX4IP overexpression resulted in a significant elevation in C-circle signal ratio in C4-2B cells (Fig. 2H). A slight, but statistically insignificant, elevation was noted in C-circle abundance in PC-3.SLX4IP cells. Together, these data indicate SLX4IP overexpression is capable of promoting ALT-like hallmarks to varying degrees but is most consistently correlated with elevations in ALT-like PML localization events.

These data support the role of SLX4IP in promoting ALT-like hallmarks, but it is unclear whether SLX4IP expression correlates with telomere length changes. After defining PD time (Supplementary Fig. S2), TRF analysis revealed control C4-2B and PC-3 cells exhibited minimal telomere length changes over 45 days (Fig. 2I and J). If SLX4IP-mediated ALT-like hallmarks are not promoting telomere maintenance it would be expected that, at least in the case of C4-2B. SLX4IP cells, telomeres would shorten because of the observed significant reduction in telomerase expression and activity (Fig. 2D and E). However, slight telomere lengthening was noted in both C4-2B.SLX4IP and PC-3.SLX4IP cells, supporting the role of SLX4IP-mediated ALT-like PML localization events in promoting telomere maintenance (Fig. 2I and J; Supplementary Fig. S2). Despite this observation, changes in telomere length distribution were not present, consistent with the atypical ALT-like presentation described above.

Here, we establish elevated SLX4IP expression promotes ALT-like PML localization events and C-circles in C4-2B cells and elevates ALT-like hallmarks from baseline in PC-3 cells. Moreover, these phenotypic changes are coupled to telomere maintenance, suggesting SLX4IP-mediated ALT-like PML localization promotes functional telomeric elongation events.

SLX4IP is essential for ALT-like hallmarks and telomere maintenance in AR-independent CRPC

To identify the effects of SLX4IP loss on the atypical ALT phenotype of AR-independent CRPC in vitro, stable cell lines with SLX4IP knockdown (KD.1 and KD.2) and nontargeting control (NS) derived from PC-3 cells were generated using lentiviral transduction (Fig. 3A and B). An approximate 50% knockdown of SLX4IP in PC-3 cells led to a reduction in both percentage of APB-positive cells and percentage of cells with ALT-like telomeric PML localization events (Fig. 3C and D), while having no significant effect on C-circle abundance (Fig. 3E). With this specific loss of ALT-like PML localization events, we reasoned SLX4IP knockdown may trigger a compensatory induction of telomerase to provide adequate telomere maintenance (61). However, TERT expression and telomerase activity were not induced, but marginally diminished, in PC-3.KD.1 and KD.2 cells (Fig. 3F and G; Supplementary Fig. S3). To solidify the role of SLX4IP-mediated ALT-like PML localization events in telomere maintenance, telomere length changes were investigated following SLX4IP knockdown.

If the observed SLX4IP-mediated atypical ALT phenotype is involved in telomere maintenance, then a reduction in SLX4IP expression without compensatory telomerase induction should lead to accelerated telomere shortening (34). To test this hypothesis, telomere length changes were evaluated over 45 days coupled with PD time determination. Calculation of PD time revealed a dramatic reduction in cell number over time in PC-3.KD.1 and KD.2 cells suggesting an impairment in proliferative capacity, a phenotype observed with impaired telomere maintenance (refs. ^{18, 62}; Supplementary Fig. S3). In addition, PC-3.KD.1 and KD.2 cells demonstrated an average telomere loss of approximately 700–800 bp over 24 PDs, whereas control lines exhibited minimal telomere length changes in the equivalent number of PDs (Fig. 3H; Supplementary Fig. S3). This translates to more than triple the rate of telomere loss in control lines (Fig. 3I; Supplementary Fig. S3). Because cells unable to sufficiently maintain telomere length are programmed to senesce to prevent loss of genomic information as they approach a minimum threshold length (18, 62), we reasoned that accelerated telomeric shortening in PC-3.KD.1 and KD.2 would lead to premature senescence. As expected, a considerable increase in the proportion of senescence-associated β-galactosidase (SA β-gal) staining (63) was observed in late-passage cells with SLX4IP knockdown compared with late-passage controls (Fig. 4A and B). Elevations in relative p21 protein expression, a senescence-associated marker (64), further corroborated the senescent phenotype (Fig. 4C and D). Moreover, SLX4IP-mediated senescence is most likely due to inadequate telomere length maintenance, as no significant differences in β-gal–positive cells (Fig. 4A and B) or p21 protein expression (Fig. 4C and D) were observed in early-passage PC-3.NS.1, NS.2, KD.1, and KD.2 cells.

In contrast to PC-3 cells, SLX4IP knockdown in DU145 cells revealed a compensatory increase in telomerase expression and activity coupled with a reduction in cells with ALT-like telomeric PML localization (Supplementary Fig. S4). Although these cells have the ability to induce telomerase expression and promote functional assembly of the enzyme, a significant impairment in PD time and modest telomere shortening were observed (Supplementary Fig. S5). Moreover, SLX4IP knockdown led to a prominent increase in both β-gal positively stained cells and relative p21 protein expression in late-passage DU145-derived cell lines that were not present in early-passage cells (Supplementary Fig. S5). Altogether, these data demonstrate diminished SLX4IP expression is associated with the induction of senescence in late-passage PC-3 and DU145 cells.

The inability of telomerase induction to fully rescue DU145.KD.1 and KD.2 cell lines from telomere shortening and senescent-associated markers raises the question of whether stable telomerase overexpression in PC-3.KD.1 and KD.2 cells, which do not induce telomerase, can promote sufficient telomere maintenance and prevent senescence. To answer this question, PC-3.KD.1 and KD.2 cells with stable telomerase overexpression (PC-3.KD.1+TELO and PC-3.KD.2+TELO) were generated via retroviral transduction and followed over 45 days. Following confirmation of telomerase overexpression (Fig. 4E and F; Supplementary Fig. S5) and PD determination (Supplementary Fig. S5), TRF analysis revealed minimal telomere length changes in PC-3.KD.1+TELO cells over 32 PDs versus approximately 600 bp loss in PC-3.KD.1 cells over 24 PDs (Fig 4G). Telomere length reductions of approximately 500 bp in PC-3.KD.2+TELO cells over 32 PDs versus approximately 800 bp in only 30 PDs in PC-3.KD.2 cells were also noted (Fig 4G). These data translate to a deceleration in the rate of telomere loss to only approximately 4 and approximately 14 bp per PD in PC-3.KD.1+TELO and PC-3. KD.2+TELO cells, respectively (Fig. 4H). Telomere length preservation via telomerase was further reflected by a significant reduction in the proportion of SA β-gal–positive cells and p21 expression in PC-3.KD.1+TELO cells and p21 expression in PC-3.KD.2+TELO cells when compared with cells with SLX4IP knockdown alone (Fig. 4I-L). Together, these data support that the senescent phenotype observed following SLX4IP knockdown is very likely through a telomere-based mechanism.

Loss of SLX4IP-mediated ALT-like PML localization events revealed senescence-associated markers in both PC-3 and DU145 cells demonstrating AR independence. Remarkably, the compensatory induction of telomerase observed in DU145 cells was unable to completely prevent telomere shortening and senescence. This effect was recapitulated following telomerase reconstitution in PC-3 cells with SLX4IP knockdown. These data support ALT-like PML localization is necessary for adequate telomere maintenance in AR-independent CRPC in vitro models. In addition, these data identify SLX4IP as an essential regulator of the atypical ALT phenotype in this AR-independent environment.

SLX4IP-mediated ALT-like hallmark depletion and induction of senescence is maintained in vivo

To ascertain whether loss of ALT-like PML localization events via SLX4IP knockdown is maintained in vivo, male nude mice were inoculated subcutaneously with PC-3.NS.1 or PC-3.KD.1 cells and tumor volumes were subsequently monitored over 30 days. IHC revealed SLX4IP knockdown led to a significant increase in the percentage of p21 positively stained cells (Fig. 5A and B). Consequently, this effect translated to a significant reduction in average tumor volume at time of sacrifice (Fig. 5C). As no considerable difference in Ki-67 or active caspase-3 staining was observed between groups, reductions in tumor volume were likely due to the senescent phenotype associated with SLX4IP knockdown (Fig. 5A). Indeed, SLX4IP knockdown resulted in a reduction of ALT-like PML localization events (Fig. 5D and E). Analogous to our in vitro studies, evaluation of telomerase activity among xenografts did not reveal a compensatory increase in telomerase activity following SLX4IP knockdown (Fig. 5F and G). Together, these in vivo results further support SLX4IP as a mediator of ALT-like PML localization in AR-independent CRPC.

Androgen-deprived conditions promote ALT-like PML localization in an SLX4IP-dependent manner

Androgen deprivation therapy triggers AR independence in patients (2, 65-67). Similarly, AR-dependent cell lines undergo AR loss when placed in growth media lacking androgens (68-70). Because we identified a unique relationship between ALT-like hallmarks and AR-independent CRPC, we reasoned AR-dependent CRPC cells lacking ALT-like hallmarks grown in androgen-deprived conditions may limit telomerase and gain ALT-like hallmarks as they transition to AR independence. Moreover, this promotion of ALT-like hallmarks, specifically ALT-like PML localization, is expected to be reliant on SLX4IP. Therefore, C4-2B cells with stable knockdown of SLX4IP (KD.1 and KD.2) and NS were generated (Fig. 6A and B) and placed in charcoal-stripped media (+CSS) to mimic androgen deprivation and induce AR loss (68-71).

Analysis of SLX4IP expression revealed a modest, yet significant increase in SLX4IP expression when comparing C4-2B.NS cells cultured in +CSS, hypothesized to adopt an ALT-like phenotype, with their normal growth media (+FBS) counterparts (Fig. 6A and B). Reductions in AR expression and upregulation of ENO2 expression, markers of AR independence, across groups confirmed the transition was initiated under androgen-deprived conditions (Fig. 6A; Supplementary Fig. S6; refs. ^{71, 72}). Notably, a difference in AR and ENO2 expression was not observed between cell lines grown in +CSS, indicating SLX4IP itself is not responsible for inducing AR independence.

As expected, relatively high telomerase expression and activity were noted in cells grown in +FBS, however, both were significantly reduced with +CSS (Fig. 6C and D; Supplementary Fig. S6). Moreover, few APB-positive cells and ALT-like PML localization events were identified in C4-2B.NS, KD.1, and KD.2 cells grown in +FBS. Following +CSS growth, C4-2B.NS cells exhibited a significant increase in the percentage of cells with ALT-like PML localization events, but not APBs (Fig. 6E and F). This significant increase was diminished with SLX4IP knockdown. Together, these data support C4-2B cells shift from high telomerase expression and activity to low telomerase expression and activity with ALT-like PML localization events following androgen deprivation. Without sufficient SLX4IP expression, this phenotypic shift is blunted.

Reductions in telomerase with a blunted atypical ALT phenotype make telomere shortening an expected fate. Following PD time determination (Supplementary Fig. S6), cell lines grown in each condition were followed for 45 days to characterize telomere length changes. As expected, C4-2B cells grown in +FBS regardless of SLX4IP expression exhibited minimal telomere length changes consistent with their telomerase-positive status (Supplementary Fig. S6). C4-2B-derived cell lines grown in +CSS demonstrated telomere shortening; however, this phenotype was markedly amplified in the case of SLX4IP knockdown (Fig 6G; Supplementary Fig. S6). Specifically, C4-2B.KD.1 and KD.2 cells exhibited a dramatic loss of approximately 1,800 and approximately 2,500 bp in 45 PDs compared with the approximately 1,400 bp loss in 54 PDs in C4-2B.NS cells. Insufficient telomere maintenance was unsurprising because C4-2B–derived cell lines demonstrate limited telomerase expression and activity following growth in +CSS, regardless of SLX4IP expression. However, the accelerated loss of telomere length in C4-2B.KD.1 and KD.2 cells supports the role of ALT-like PML localization events in promoting some degree of telomere maintenance in the C4-2B.NS cells following androgen deprivation, albeit not to the same degree as telomerase in +FBS.

After identification of telomere shortening, senescent-associated markers were evaluated. Early-passage comparisons revealed SLX4IP knockdown and growth in +CSS independently trigger immediate β-gal positivity (Fig 7A and B) and elevations in relative p21 protein expression (Fig 7C and D) to some degree. This pattern remained consistent from early- to late-passage cells with the exception of C4-2B. KD.1 and KD.2 cells grown in +CSS, which exhibited dramatic telomere shortening. β-gal staining of these late-passage C4-2B.KD.1 and KD.2 cells revealed a significant increase in β-gal–positive cells compared with C4-2B.NS cells grown in +CSS and compared with C4-2B.KD.1 and KD.2 cells grown in +FBS (Fig. 7A and B). Consequently, these late-passage β-gal–positive populations can likely be attributed to the impaired telomere maintenance highlighted above. Corroborating this pattern, relative p21 protein expression mimicked our previous finding and was significantly elevated in late-passage SLX4IP-knockdown cells cultured in +CSS (Fig. 7C and D). Therefore, with androgen deprivation, AR-dependent CRPC limits telomerase expression and activity and gains ALT-like PML localization events in an SLX4IP-dependent manner. Without sufficient SLX4IP expression and promotion of these ALT-like hallmarks, telomere shortening and senescence are induced in this in vitro model of disease progression following androgen deprivation.