Emerin Dysregulation Drives the Very-Small-Nuclear Phenotype and Lineage Plasticity That Associate with a Clinically Aggressive Subtype of Prostate Cancer

Patient selection and study design

Human investigations were performed after approval by the Cedars-Sinai Medical Center Institutional Review Board. All patients provided written informed consent prior to any research collection. Under Institutional Review Board# Pro00042197, blood samples from consenting patients with mCRPC were collected for CTC analysis. This study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. For this analysis, samples were considered usable if the following criteria were met: (i) a verified diagnosis of prostate cancer with metastasis, (ii) disease progression with serum testosterone concentration <50 ng/dL (i.e., castration), and (iii) clinical follow-up data on progression/death available. Samples were only collected from patients receiving care at the Cedars-Sinai Cancer Center for Uro-Oncology Research Excellence clinics to ensure access to clinical outcome data. Samples collected between September 2014 and June 2018 were considered for inclusion. Samples were included in the overall survival (OS) analysis if CTC assessment by NanoVelcro assay was performed and confirmed by an experienced, clinical pathologist. Moreover, the CTCs need to be suitable for characterization as vsnCTC positive (+) or negative (−). A subgroup of samples was utilized for an analysis of disease progression while on treatment (i.e., progression-free survival, PFS) if the patient started treatment within 4 weeks of blood collection. All treatments and examinations (laboratory and radiographic) were performed as part of standard care. Only one sample was collected per patient for all analyses in this study. The demographics and clinical characteristics of these patients at the time of blood draw have been summarized in Table 1.

CTC enrichment, imaging, and nuclear size measurements

The workflow and detailed methods for CTC enrichment and vsnCTC enumeration have been summarized in Supplementary Fig. S1. In short, venous blood of patients was collected in acid-citrate-dextrose–containing vacutainers and processed within 24 hours of phlebotomy. Following density gradient centrifugation using a Ficoll-Paque solution (Sigma-Aldrich) per the manufacturer’s specifications, the peripheral blood mononuclear cell layer was isolated. The peripheral blood mononuclear cells (up to 2 mL of whole blood equivalent) were then loaded into a NanoVelcro CTC assay chip to enrich CTCs using our previously published protocol (6). The captured cells were then fixed with 2% paraformaldehyde and subjected to immunocytochemical staining. Candidate CTCs were characterized as 4′,6-diamidino-2-phenylindole (DAPI)–positive/cytokeratin (CK)-positive/CD45-negative cells with round or oval nuclei using an upright fluorescence microscope (Eclipse 90i, Nikon Instruments Inc.). As apoptotic CTCs have been shown to be associated with patient outcomes in cancer (21, 22), each candidate was reviewed and certified by an experienced pathologist for morphology to specifically confirm that the observed event was a CTC (rather than a hematologic cell) and lacked features of apoptosis. Nuclear size was defined as the square root of the product of the long axis and the short axis. The calibration studies for ensuring the accuracy and reproducibility of nuclear size measurements on NanoVelcro CTC assay were conducted in our previous study using prostate cancer cell lines and patient samples (6). A vsnCTC was defined as a CTC with nuclear size less than 8.5 μm based on our previous work (5). In this study, we explored the relationship of OS to minimum, median, and mean CTC nuclear size, as defined by the smallest, median, and average nuclear size among the CTCs obtained in each enumeration study/blood specimen, respectively.

Cell lines

C4-2B cells resistant to enzalutamide (Enza-R) or abiraterone (Abi-R) as well as control (parental) cells were kindly provided by Dr. Allen Gao (University of California Davis, CA). C4-2B/Enza-R cells were maintained in 20 μmol/L Enza-containing medium as previously described (23). C4-2B/Abi-R cells were maintained in 10 μmol/L Abi. C4-2B (RRID: CVCL_4784), DU145 (RRID: CVCL_0105), and 22Rv1 (RRID: CVCL_1045) cells were purchased from ATCC. C4-2B and 22Rv1 were maintained in RPMI-1640 medium with 10% FBS and 1% penicillin/streptomycin. DU145 were cultured in DMEM (Thermo Fisher Scientific) with 10% FBS and 1% penicillin/streptomycin. EMD knockdown DU145 cells, generated by short hairpin RNA (shRNA), were kindly provided by Dr. Michael R. Freeman (20). All cultures were grown in a humidified 5% CO₂ environment at 37°C. Cell line authentication was conducted by ATCC and periodically tested using the Mycoplasma Detection Kit (Lonza), and cells were clear of contamination.

siRNA transfection

C4-2B, 22Rv1, and DU145 cells were transfected with EMD siRNA (sc-35296, Santa Cruz Biotechnology; RRID: SCR_026442) and control scramble siRNA (sc-37007, Santa Cruz; RRID: SCR_026443) using Lipofectamine 3000 (Thermo Fisher Scientific), as described by the manufacturer. Then, 24 hours after the addition of the transfection mix, liposomes were removed, and fresh medium was added.

Immunofluorescence

Cells were seeded on poly-L-lysin (Sigma-Aldrich) coated Nunc Lab-Tek plates (Thermo Fisher Scientific), then fixed with 2% paraformaldehyde and subjected to immunocytochemical staining as previously described (20). Images were captured using a Leica DMi8 fluorescence microscope and ImageJ (RRID: SCR_003070) was used to measure nuclear size.

Transwell invasion assays

Cells were fasted in 0.1% BSA and then plated (0.1 × 10⁶) on transwell inserts (8 μm) coated with Matrigel matrix, phenol red free (BD Biosciences). Inserts were then placed in 24‐well plates containing 5% FBS and incubated at 37°C for 24 to 48 hours. After incubation, media were aspirated, and cells which did not invade through the chamber on the upper side of the membrane were removed with a swab. Cells attached to the bottom side of the membrane were fixed with 4% paraformaldehyde and stained with 0.1% (v/v) crystal violet. Inserts were washed and photographed at 10× using an inverted microscope (Leica) and MagnaFire SP software. Transwell assays were performed independently at least 3 times.

RNA preparation, cDNA synthesis, and qPCR

Cell lines were lysed in TRIzol (Life Technologies, Invitrogen), and total RNA was prepared using Direct-zol RNA MicroPrep (Zymo Research) according to manufacturer's instructions. 500 ng of total RNA was reverse‐transcribed, and cDNA was used for qPCR analysis on a QuantStudio 5 Real‐Time PCR system (Applied Biosystems; RRID: SCR_021096) per the manufacturer’s instructions. In all cases, target gene expression was normalized to the expression of housekeeping genes such as GAPDH. Relative gene expression was calculated using standard 2−δδCt. Results were generated from at least three independent experiments and are shown as the mean ± SD. Primers used are listed in Supplementary Table S1.

RNA sequencing and gene set enrichment analysis

Total RNA was extracted with the RNeasy Plus Mini Kit (Qiagen; RRID: SCR_008539) according to the manufacturer’s instructions. All the RNA sequencing (RNA-seq) experiments were performed in triplicate. RNA library and transcriptome sequencing were performed by Novogene. Gene set enrichment analysis (GSEA; RRID: SCR_003199) of differentially expressed genes was performed by GSEA 4.0.3. Code and relevant data have been deposited in Code Ocean at: /https://codeocean.com/capsule/7045784/tree/v1.

https://codeocean.com/widget.js?slug=7045784

MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay

C4-2B cells were transfected with either siRNA targeting EMDor control siRNA. Following transfection, the cells were cultured in the presence of Abi acetate (10 μmol/L; 99%, MedKoo Biosciences) or Enza (1 μmol/L). Cell viability was assessed on day 5 of culture using the MTT assay, following the manufacturer's instructions (Thermo Fisher Scientific).

Lentivirus infection

Lentiviral shRNA transduction particles targeting human EMD and lentiviral nontargeting control transduction particles (SHC002V) were purchased from Sigma. The EMD shRNA was a pool of five different siRNA target sequences:

EMD shRNA #1 (oligo ID: TRCN0000083010) hairpin sequence: 5′-AAA​CCC​AGG​GCT​GCC​TTG​GAA​AAG-3′; EMD shRNA #2 (oligo ID: TRCN0000083011) hairpin sequence: 5′- AAA​CCC​AGG​GCT​GCC​TTG​GAA​AAG-3′;EMDshRNA #3 (oligo ID: TRCN0000083008) hairpin sequence: 5′-GTT​GAC​ATG​CAT​TCC​TCC​TTT-3′; EMD shRNA #4 (oligo ID: TRCN0000083009) hairpin sequence: 5′-CCC​AAG​AAA​GAG​GAC​GCT​TTA-3′; and EMDshRNA #5 (oligo ID: TRCN0000083012) hairpin sequence: 5′-AGG​TGC​ATG​ATG​ACG​ATC​TTT-3′. Prostate cancer cells transducted with lentiviruses were supplemented with 8 μg/mL polybrene and then selected by puromycin for 3 to 5 days.

Western blotting

Extracted proteins were separated by 4% to 20% SDS-PAGE gels. The blots were probed with antibodies against EMD (6080531, Leica; RRID: AB_3676183) and β-actin (47778, Santa Cruz Biotechnology; RRID: AB_626632), followed by incubation with alkaline phosphatase-conjugated secondary antibodies (A3687, Sigma-Aldrich; RRID: AB_258103), or HRP anti-mouse (7076, Cell Signaling Technology; RRID: AB_330924) for 1 hour at room temperature. Blots were developed with an ECL substrate (Thermo Fisher Scientific, A38554) and chemiluminescence or by 1-step NBT/BCIP (34042, Thermo Fisher Scientific).

Cisplatin sensitivity assay

To exclude the presence of necrotic or apoptotic cells, DU145 EMD knockdown and control cells by shRNA were stained using the APC Annexin V Apoptosis Detection Kit with PI (BioLegend) and immediately sorted by FACS based on their negative expression of Annexin V/PI. These sorted cells were then used for the cisplatin sensitivity assay.

A total of 1 × 10⁴ DU145 EMDknockdown and control cells by shRNA were seeded in a 96-well plate 16 hours before the addition of Cytotox Green (250 nmol/L) and different concentrations of cisplatin (0, 2, 10, and 50 μmol/L). Dying cells are labeled green, in real time, by the mix-and-read Incucyte Cytotox Green Dye. Images were collected every 2 hours, and the analysis was performed using Incucyte Cell-by-Cell Analysis Software Module, thus enabling individual cell segmentation and classification based on fluorescence. The increase of cell death is directly related to the increase of fluorescence area calculated by Incucyte software analysis.

Statistical analyses

Patients were designated as vsnCTC present/vsnCTC+ or vsnCTC absent/vsnCTC−. If a patient had ≥1 vsnCTC, this patient would be designated as vsnCTC+ regardless of the quantity of vsnCTCs or the presence of non-vsnCTCs. The Kaplan–Meier plot and log-rank test were used for OS and PFS analyses. OS was defined as the time from blood draw to time of death. Progression was defined as clinical progression (per the treating physician) or biochemical/radiographic progression according to Prostate Cancer Working Group 3 (ref. 24) and the RECIST1.1 (ref. 25). Multivariable Cox regression was used primarily to investigate the association of vsnCTC status with time to death and progression. To minimize bias from confounding factors, a model adjusting for the vsnCTC status, age, PSA, lines of CRPC treatment prior to the blood draw, and presence of VM at the time of blood draw was chosen. The results were shown as an HR with 95% confidence intervals (CI). A p-spline plot (26) was used to depict the association between the HR of OS and the CTC nuclear size. Continuous variables were summarized as medians and IQRs and compared using the Mann–Whitney U test. A two-tailed Student t test was used for comparisons between two groups, whereas one-way ANOVA was used for comparisons among three or more groups. Results were presented as the mean ± SD, with P < 0.05 considered statistically significant. All data were analyzed from at least three independent experiments. Graphs were created using GraphPad Prism 9 software (RRID: SCR_002798).

Data availability

RNA-seq data are available at Gene Expression Omnibus with accession ID GSE288991. Additional data generated in this study are available upon request to the corresponding author.

Patients with vsnCTCs have poor survival and exhibit increased resistance to AR-targeted therapy

Our initial study pointed to a relationship between the presence of vsnCTCs and VM (5). In this follow-up series, 93 clinically annotated, blood samples from patients with mCRPC with survival follow-up to death (if applicable) within the Translational Oncology Program Blood & Biospecimen Bank were selected for analysis. CTCs were isolated and analyzed using the NanoVelcro assay. Patient characteristics are shown in Table 1.

Among 93 patients, 77 (83%) had detectable CTCs, 56 (60%) were vsnCTC+, and 37 (40%) were vsnCTC−. 33 (59%) patients who were vsnCTC+ and 12 (32%) patients who were vsnCTC− had VM at the time of blood draw (P = 0.0192). Interestingly, we also found that patients with mCRPC with greater than two lines of CRPC treatments prior to blood draw had greater vsnCTC positivity (P = 0.0334). Given the established relationship between VM and survival, we hypothesized that there was an association between vsnCTCs, OS, and PFS in mCRPC. As shown in Fig. 1A, vsnCTC+ had significantly shorter OS when compared with patients who were vsnCTC− (median 7.5 vs. 35.8 months, HR = 2.26, 95% CI, 1.46–3.50, log-rank P = 0.0002). After adjusting for potential confounding factors using a multivariate Cox regression model (i.e., PSA, presence of VM at blood draw, prior CRPC treatment line, age, and total CTC number per 1 mL of blood), vsnCTC+ status was independently associated with worse OS (Supplementary Table S2, HR = 1.70, 95% CI, 1.03–2.81, P = 0.0396). Treatments provided after vsnCTC assessment are provided in Table 1 and Supplementary Table S3. In general, there were no patterns of treatment variation between the vsnCTC+ and vsnCTC− group over the natural history of their disease course. Patients with vsnCTC+ also had significantly shorter PFS compared with patients who were vsnCTC− (Fig. 1B, median 3.0 vs. 6.5 months, HR = 1.93, 95% CI, 1.13–3.28, log-rank P = 0.0117). To address potential overlap from our pervious vsnCTC study (5), we performed OS and PFS analyses excluding the 18 patients from the original study (Supplementary Fig. S2). Consistently, patients who were vsnCTC+ had significantly poorer OS and PFS. When specifically considering ARSI treatment, patients who were vsnCTC+ had significantly worse PFS compared with patients who were vsnCTC− (Fig. 1C, median 3.2 vs. 11.3 months, HR = 2.17, 95% CI, 1.09–4.34, log-rank P = 0.0147). Moreover, the CTC nuclear size was significantly smaller in patients with prior ARSI use compared with those with no prior ARSI use (median 7.4 vs. 8.1 μm, rank-sum test P = 0.0314, Fig. 1D). The patients with vsnCTC+ had a lower PSA response rate to ARSI treatment, compared with patients who were vsnCTC− (Fig. 1E). Accordingly, the best PSA change from baseline in each ARSI patient, as shown in the waterfall plot in Fig. 1F, revealed that among the 20 patients who were vsnCTC+, 2 (10%), 4 (20%), and 6 (30%) achieved a best PSA percentage decrease of 90% or more, 50% or more, and 30% or more, respectively. In contrast, of the 17 patients who were vsnCTC−, 4 (24%), 7 (41%), and 8 (47%) achieved a best PSA percentage decrease of 90% or more, 50% or more, and 30% or more, respectively.

In contrast to this, we considered the association between vsnCTC presence and taxane sensitivity. Here, there were fewer patients and in the available dataset, there was no statistically significant association between vsnCTC status and OS of patients receiving taxane therapy (Fig. 1G, median 2.5 vs. 6.0 months, HR = 1.44, 95% CI, 0.58–3.55, log-rank P = 0.4259). To test the hypothesis that a smaller minimum CTC nuclear size is associated with poorer prognosis, a p-spline plot was used to illustrate the relationship between OS and minimum CTC nuclear size (Fig. 1H). Generally, the HR of OS increased as the minimum CTC nuclear size decreased. Mean or median CTC nuclear size did not show a significant association with the HR for OS. Additionally, we performed a p-spline analysis to explore the relationship between CTC number and OS. The p-spline analysis revealed a trend toward worse OS with increasing CTC numbers; however, the 95% CI frequently crossed an HR of 1.0, indicating limited statistical significance (Supplementary Fig. S3). Multivariable Cox regression adjusting for the minimum CTC nuclear size, PSA, baseline presence of VM, and CRPC treatments prior to the blood draw showed that the minimum CTC nuclear size was independently associated with OS (HR = 0.81, 95% CI, 0.70–0.95; P = 0.0099; Supplementary Table S4). This finding suggested that an increase of 1 μm in the minimum CTC nuclear size was linked to a 19% decrease in mortality risk.

CTCs from patients who are vsnCTC+ exhibit signs of nuclear shape instability by decreased EMD expression

To examine whether the vsn phenotype is a result of nuclear shape instability, we compared EMD expression from CTCs between patients with vsnCTC+ mCRPC and vsnCTC− mCRPC. Expression of EMD in white blood cells (WBC; CK⁻, CD45⁺, and DAPI⁺) from patients with prostate cancer was utilized as a comparator. Immunofluorescence analysis showed that patients who were vsnCTC+ exhibited a loss of EMD along with mislocalization of EMD into puncta as described previously (Fig. 2A; ref. 20). Additionally, EMD expression was significantly decreased in CTCs from patients with vsnCTC+ mCRPC when compared with patients with vsnCTC− mCRPC (Fig. 2B, P < 0.05) and WBCs (P < 0.0001) at the single-cell level. There was no significant difference in EMD expression in CTCs from vsnCTC− patients and WBCs (P = 0.32). A significant positive correlation was observed between the EMD expression and the nuclear size of CTCs from patients with mCRPC (Fig. 2C, Spearman r = 0.52, P < 0.0001). In contrast, no significant correlation was found between CK and nuclear size (Fig. 2D, Spearman r = 0.17, P = 0.20), suggesting that the reduction in nuclear size was associated with the loss of EMD expression in CTCs from patients with advanced prostate cancer.

Depletion of EMD evokes the emergence of the vsn phenotype and promotes invasion and lineage plasticity in vitro

To examine the causal relationship between EMD loss and the vsn phenotype, we developed EMD-depleted cell lines by siRNA. DU145/EMD− cells exhibited nuclei that were 22% smaller than controls (Fig. 3A and B; P < 0.001). The EMD knockdown in 22Rv1 and C4-2B cells by siRNA similarly exhibited a 16% and 19% reduction in nuclear size respective to their isogenic control lines (P < 0.0001, P < 0.0001; Fig. 3C and D; Supplementary Fig. S4). The C4-2B/EMD− cells also had elevated invasiveness, as determined by transwell migration assay when compared with C4-2B/control (Ctrl) cells (P < 0.01; Fig. 3E).

As prostate cancers progress in the face of AR suppression, more poorly differentiated histologic phenotypes emerge that can exhibit small cell morphologic features or neuroendocrine features such as the expression of chromogranin A, synaptophysin, or other IHC markers of this phenotype (27). It is known that after multiple lines of antiandrogen therapies, 20% to 25% of mCRPCs will have a shift in gene expression consistent with the development of a neuroendocrine phenotype marked by expression of genes associated with cellular plasticity and stemness (28, 29). Given the increased invasiveness following EMD loss, along with the clinical associations of vsnCTCs with ARSI resistance, VM, and multiple lines of CRPC treatments, we next investigated whether EMD loss is associated with lineage plasticity. We identified markers associated with lineage plasticity based on published studies (30–32). In addition to increased invasiveness, C4-2B/EMD− cells showed elevated expression of genes associated with lineage plasticity (Fig. 3F), with similar findings observed in DU145/EMD− cells (Fig. 3G).

Transcriptomic changes in EMD knockdown cells

We next sought to uncover the molecular mechanisms underlying the role of EMD in prostate cancer biology. To do this, we performed GSEA on RNA-Seq data following EMD knockdown (Supplementary Fig. S5) to identify pathway enrichment in prostate cancer cells. Given that EMD knockdown is associated with the vsn phenotype and promotes lineage plasticity, metastasis, epithelial–mesenchymal transition, and dedifferentiation, these gene sets were found to be the most significantly upregulated pathways following the loss of EMD (Fig. 4A–C).

In addition, KRAS signaling, which is known to be associated with cancer stemness and metastatic progression in prostate cancer (33, 34), was enriched in EMD knockdown prostate cancer cells. Interestingly, we also observed an enrichment of RB1-loss–related pathways in EMD knockdown cells. As RB1 downregulation has been implicated in the transition from CRPC to a neuronally differentiated form (35, 36), this finding suggested that EMD loss could potentially contribute to this transition. These results are consistent with our data showing that EMD loss is associated with lineage plasticity.

ARSI-resistant cells exhibit a vsn phenotype and EMD loss

Given that patients who are vsnCTC+ demonstrate higher ARSI resistance compared with patients who are vsnCTC−, we next examined nuclear size in a well-established model of C4-2B ARSI-resistant cell lines developed by Liu and colleagues (23, 37). We observed that the C4-2B/Abi-R and C4-2B/Enza-R sublines exhibited a significant reduction in nuclear size (18.7% and 21.3%, respectively) compared with parental control cells (C4-2B/parent), as measured by the NanoVelcro assessment method (Fig. 5A), confirmed by confocal imaging (Fig. 5B). No statistically significant difference in nuclear size was observed between C4-2B/Abi-R and C4-2B/Enza-R cells. The EMD expression was found to be significantly reduced in both C4-2B/Abi-R and C4-2B/Enza-R cells compared with C4-2B/parent cells (Fig. 5C). In line with our observation that EMD loss is associated with the vsn phenotype, we found that the invasion capacity was higher in C4-2B/Abi-R and C4-2B/Enza-R cells compared with C4-2B/parent cells (Fig. 5D). These findings prompted us to explore whether lineage plasticity, a known pathway for therapeutic resistance, is present in ARSI-resistant C4-2B cell lines, based on previously established studies (30–32). Interestingly, our results indicated that neuroendocrine differentiation markers were upregulated in the C4-2B ARSI-resistant sublines, whereas basal markers were downregulated (Fig. 5E). These results are consistent with our findings that EMD loss is linked to lineage plasticity in prostate cancer cell lines.

To determine whether cells with EMD loss, which exhibit the vsn phenotype, have differential sensitivity to ARSI therapy, we compared the sensitivity of AR+ prostate cancer cells with EMD knockdown to their parental controls when treated with ARSIs. We selected C4-2B cells as they are castration-resistant cells but sensitive to ARSIs. Treatment of C4-2B/Ctrl cells with 10 μmol/L Abi acetate for 5 days significantly reduced cell viability (Fig. 5F, P < 0.001). In contrast, Abi had no effect on C4-2B/EMD− cells, indicating that C4-2B/EMD− cells were more resistant to Abi compared with the control cells (Fig. 5F). We also found 22Rv1/EMD− cells were relatively resistant to Abi when compared with control cells (Supplementary Fig. S6A). This resistance maybe due to EMD’s close association with the endoplasmic reticulum (ER), in which Abi acts by inhibiting CYP17 (38). As EMD is located on the nuclear envelope, which is continuous with the ER, its disruption could impair ER function, contributing to Abi resistance.

We also treated the C4-2B sublines with Enza. Surprisingly, despite the loss of EMD, these cells remained sensitive to Enza. Enza significantly inhibited the viability of C4-2B/EMD− cells (Fig. 5F, P < 0.01). When comparing impact of Enza on the C4-2B/EMD− and C4-2B/Ctrl cells, the viability of C4-2B/EMD− cells remained higher than that of C4-2B/Ctrl cells though the difference was not statistically significant. In addition, we found 22Rv1/EMD− cells exhibited a similar response to Enza as the control cells (Supplementary Fig. S6B). This finding suggested that the preserved sensitivity of the EMD knockdown in C4-2B and 22Rv1 cells to Enza may be due to nuclear instability caused by EMD loss, which would not affect the ability of Enza to block AR nuclear translocation (39), ultimately reducing cell viability.

Platinum-based chemotherapy demonstrates greater efficacy in castration-resistant tumor cells with the vsn phenotype

As the emergence of vsnCTCs and the vsn phenotype precedes the onset of VM, we hypothesized that the biological changes driving the vsn phenotype would increase the sensitivity of the underlying disease to platinum-based chemotherapy. To test this hypothesis, we compared the platinum sensitivity of vsn phenotype cells (DU145/EMD−) with non-vsn phenotype cells (DU145/Ctrl; Fig. 6). Our data demonstrated that DU145/EMD− cells exhibited a higher degree of cell death when exposed to cisplatin compared with DU145/Ctrl cells across various doses, suggesting the potential benefit of platinum-based chemotherapy in this context.