A Cyanobacteria-derived RNA aptamer resensitizes prostate cancer to hormone therapy

Abstract

Prostate adenocarcinoma (PCa) resistance to androgen receptor (AR) signaling inhibitor therapy is associated with elevated glutamine (L-Gln). Glutamine sensors, present in conserved riboswitches (glnA), control nitrogen metabolism in many organisms, like cyanobacteria. Iterative in silico modifications of glnA found in Synechococcus elongatus and thermodynamic analysis of a 56mer aptamer resulted in high L-Gln specificity and affinity. The optimized aptamer depleted L-Gln from PCa cells by both L-Gln sequestration and extracellular glutaminase activation, serving as an allosteric activator. Glutamine depletion reduced FOXM1 transcriptional occupancy on the promoter of fibroblast growth factor 8 (FGF8), a known mediator of PCa castration resistance. A point mutation in the binding pocket of the 56mer rendered the aptamer ineffective in L-Gln binding and FGF8 regulation. Accordingly, the L-Gln-depleting aptamer, with demonstrated serum stability, limited the proliferation and promoted cell death of castration-resistant PCa alone and in combination therapy with AR antagonists, enzalutamide and apalutamide, in subcutaneous and orthotopic mouse models. Further selective tumor targeting was achieved by functionalizing gold nanoparticles with either the optimized L-Gln aptamer or the point-mutant aptamer. Castration sensitivity was restored by the L-Gln-depleting aptamer but not by the point-mutant. The functionalized nanoparticle demonstrated superior anti-tumor efficacy in an orthotopic PCa model over the untargeted aptamer. The anti-tumor activity of the aptamer helped support L-Gln as an oncometabolite in PCa that can be targeted to sensitize tumors to hormone therapy.

INTRODUCTION

Prostate cancer therapy resistance is a significant factor leading to cancer mortality in American men (1). Androgen signaling plays a critical role in the development and progression of prostate cancer (PCa). Accordingly, it is the primary target for metastatic PCa and can complement first line therapies for localized disease (2). AR inhibitors, such as enzalutamide and apalutamide, are currently employed with great efficacy. However, inherent to the disease, there are multiple mechanisms can lead to PCa castrate resistance (CRPC), inclusive of androgen receptor (AR) amplification or loss, expression of AR splice variants, as well as paracrine factors that bypass the need for AR-dependent proliferative signals can involve lineage plasticity (3–5). We reported that AR signaling inhibition can elevate glutamine (L-Gln) secretion by cancer associated fibroblasts (CAF), associated with elevated circulating L-Gln in PCa patients exhibiting resistance to such therapies (6). L-Gln is a conditionally essential amino acid for cancer cells, as they have a higher demand for protein, nucleotide biosynthesis, and energy production, compared to benign tissues. Inhibiting L-Gln in prostate cancer cells has shown a significant impact on cell growth arrest and cell death with a dramatic reduction of intracellular nucleotides and their derivatives (7). However, antagonizing L-Gln metabolism does not take into account alternative mechanisms of its synthesis or breakdown (8,9). Similarly, due to the redundancies in L-Gln transporters, blocking an individual transporter provides limited benefit (10). Conversely, a non-metabolizable L-Gln analogue, like 6-Diazo-5-oxo-l-norleucine (DON) and DRP104 has showed promising antitumor effects (11). We reasoned that interrupting L-Gln uptake by its extracellular depletion in a tumor-specific manner can be a therapeutic strategy bypassing metabolic adaptation of a specific enzyme or transporter, particularly if combined with hormone therapy.

To target L-Gln, we looked to cyanobacteria structured non-coding RNA riboswitches normally associated with metabolite-binding as a means of gene regulation (12). The riboswitches found in cyanobacteria that sense L-Gln through an aptamer motif, glnA, were chosen as a starting point due to its great stability and high specificity for this amino acid (13–15). Aptamers are single-stranded DNA or RNA molecules that can be part of riboswitches found in both prokaryotic and eukaryotic cells, binding diverse metabolites and vitamins, and metal ions with high specificity. Since L-Gln concentrations in the bacteria can be greater than concentrations observed in the mammalian tumor microenvironment, we wanted to improve aptamer binding affinity through an iterative process of structural analysis, computer modeling, and in vitro testing. This process yielded an optimized aptamer found to effectively limit L-Gln uptake by PCa cells. Aptamer-mediated depletion helped reveal a role for L-Gln in the regulation of FGF8 expression, a known mediator of castrate resistance (16). Restricting L-Gln from prostate tumors was sufficient to restore sensitivity to androgen receptor antagonists through the regulation of the newly found oncogenic signaling axis.

MATERIALS AND METHODS

Ancestry analysis

GlnA sequences from RNAcentral and Rfam (for a total of 1,410 unique species) were retrieved and aligned in parallel to the sequences from the modified aptamers using Clustal Omega (17,18). The values shown in the phylogenetic tree are represented by the length of the branches for the neighbor aptamers, indicating the evolutionary distance between the sequences. Then, an interactive sunburst was created in R with the package plotly. The script was used to build a chart to represent the taxonomic classification of the cyanobacteria species having the glnA aptamer. We used the sunburst accessed in Rfam with the number RF01739 as a reference for building our script. The script consisted of labels and parents, where labels stand for the names of cyanobacteria species and parents represent the taxonomic group for each species. The base script for this chart was: library(plotly); fig <- plot_ly(; labels = c(“Cyanobacteria”,; parents = c(““, “Cyanobacteria”, values = c(82; type = ‘sunburst’,; branch values = ‘total’;); fig. The interactive sunburst can be downloaded in the link indicated at the supplementary data.

Reagents

L-Glutamine (Gibco) was diluted in media for cell culture experiments. Apalutamide (ARN-509, Selleckchem) and FOXM1 inhibitor (FDI-6) (MedChemExpress) were diluted in DMSO (ThermoFisher). The cDNA 56mer template was synthesized by IDT and diluted in nucleases free water to 100 μM concentration for T7 transcription synthesis. The non-labeled RNA 56 merA¹⁵ aptamer was synthesized by Oligo Factory and diluted in PBS 1X for cell culture experiments. In binding experiments, the 56 merA¹⁵ aptamer was diluted in aptamer buffer (50 mM KCl, 50 mM K-acetate and 10 mM MgCl₂, pH 6.8). In animal studies, aptamers were folded in aptamer buffer, then diluted in 0.9% sodium chloride saline solution.

Rational in silico modifications

The 67mer aptamer reported by Ames et al. from the Synechococcus elongatus glnA riboswitch was used as a starting point for in silico rational modifications (13). The modified aptamers were designed based on alteration in four principal sections: 1) stem P1, 2) stem-loop P2, 3) E-loop and 4) stem-loop P3, common to glnA of all species. These sections were further divided into 16 smaller subsections that were subjected to changes in length, nucleotide replacements, randomizations to reach the greatest thermodynamic stability. In addition, we truncated both the 5’ and 3’ ends as well as extending the P1 stem with an A-T zipper. The RNAinverse v2.4.18 tool in the ViennaRNA v2.5.1software package was used to calculate one hundred randomization sequences for the Stem P1 section (using the command line: “RNAinverse -Fmp -f 0.1 -R 100 -d2 < seqstruct.txt > inverse.out”). Predictions of secondary structures with lowest MFE were conducted, MFE (fold) structure prediction, MaxExpect structure prediction, Probknot pseudoknot structure prediction and partition function (base-pair probabilities) were calculated for each sequence using RNA structure v6.4 (19,20). Tertiary structure modeling of the candidate aptamers was performed using the RNAComposer v1.0 server (21–23). Then, two thousand docking iterations were simulated in Hex v8.0 for each aptamer using the three-dimensional *.pdb files generated in the presence or absence of Mg²⁺. Prior to docking simulations, molecular geometry optimization of the interaction of Mg²⁺ with Gln was performed in Avogadro v1.2.0. After docking, the estimated equilibrium constant (K_eq) was calculated for the modified aptamers using the formula: ΔG⁰=−RTln(K_eq), where the E_total (K_cal/mol) was considered as the free energy obtained from the docking predictions. Finally, some of the structures were visualized and edited in PyMOL v2.5.4.

Fluorescent intercalator displacement (FID)

The Gln-structural related amino acids binding test (also referred as “amino acid rejection”) was calculated incubating the aptamers (25 μl, 5 μM) in aptamer buffer with 25 μl, 0.5 mM MgCl₂, and a curve (25 μl, 0–50 μM) of the amino following acids: L-Gln, D-Gln, Alanine-Gln, Asparagine, Threonine, Glycine and Cysteine (Sigma) at 18 °C for 15 minutes. After that time, 25 μl of 30 μM thiazole orange (TO) were injected per well in a white 96-well plate using the high-performance plate reader ClarioStar Plus (BMG Labtech) and collecting the fluorescence data every 2 minutes for 30 minutes.

Estimated K_D by nanogold particles

Nanogold aggregation studies were adapted from that previously described for the aptamer-kanamycin binding (24). A standard Gln binding curve was developed by incubating nanogold particles (10 nm diameter, Sigma) with a Gln curve (0–7.4 μM) and 1.2 μM folded aptamers in aptamer buffer at 18°C for 5 minutes. A fresh aliquot of 1/20,000 dilution of 20% poly diallyl dimethylammonium chloride (PDDA) solution (Sigma) was added to the aptamer/nanoparticle mix at 18°C and incubated for 5 min. After incubation, 10 μl of the reaction was placed in a 96 well plate and read it at 655/495 nm wavelength.

Cell culture

CWR22RV1 (22RV1) and C4–2B cells were acquired from American Type Culture Collection (ATCC, Manassas, VA, USA), and ARCaP_M cells were donated by Dr. Leland Chung (Cedars-Sinai Medical Center). All these three cell lines were grown in RPMI-1640 (Gibco) media supplemented with 10% fetal bovine serum (FBS; Atlanta Biologicals). No further testing of authentication was performed following cell line acquisition. Primary cancer-associated fibroblasts (CAF) were generated in the lab from human tumors propagated in DMEM/F12 media (Gibco), supplemented with 5% FBS, 5% Nu Serum (Corning), 10 nM testosterone and 0.68 μM insulin, according to previous reports (25). Cell counts were determined by Bio-Rad TC20 cell counter (Hercules, CA, USA) using a 1:1 ratio of suspended cells and trypan blue. Mycoplasm testing was routinely performed monthly using the MycoAlert^® Mycoplasma Detection Kit (Lonza, Cambridge, MA; cat # LT07–218).

L-Gln uptake

C4–2B, 22Rv1, and ARCaP_M cells were incubated in Gln-free RPMI, 5% FBS overnight prior to Gln-uptake studies. After that time, media was replaced with Gln-free RPMI media (Gibco) supplemented with (R&D systems). Resuspended cells (1×10⁶ per replicate) were incubated in 100 μl Gln-free media with folded aptamers (at indicated concentrations) and ³H-Gln (Perkin Elmer) for 2 hours at 37 °C. Cells were centrifuged at 2,000 rpm for 1 minute and supernatant was discarded, then, pellets were washed twice with PBS 1X. Cells were lysed in 100 μl of 0.2 N NaOH and counted in 5 ml versa scintillation fluid cocktail (Perkin Elmer) by the TRI-CARB 3110TR scintillation counter (Perkin Elmer).

Oxygen consumption rate

Seahorse Bioscience XF96 was used to measure oxygen consumption rate (OCR) in 22Rv1 cells incubated in XF base medium for 1 hr at 37 °C in a CO₂-free incubator for temperature and pH equilibration. Mitochondrial regulators, oligomycin was used to inhibit complex V, carbonyl cyanide p-trifluoro methoxyphenylhydrazone (FCCP) was added to disrupt the mitochondrial proton gradient, and rotenone/antimycin was used to inhibit Complex I and III respectively. Enzalutamide (10 μM) and 56merA¹⁵ (0.5 μM) was used to measure acute treatment effects. Results were normalized to protein content.

RNA sequencing and bioinformatic analysis

The raw data obtained from RNA sequencing was uploaded and processed in the platform BIOBOX. Pipelines comprising STAR + DESeq2 alignments were created to output two excel files: 1) normalized gene counts file and 2) differential expressed genes file. The significant differentially expressed genes common to 22Rv1 and C4–2B cells were used for scatter plots and further pathway enrichment analysis using the platform EnrichR. Gene expression profiling of normal prostates and PCa metastasis were performed in the platform SEEK with data accession: GSE38241 (26). Correlation analysis between FOXM1 and FGF8 expression was investigated using the Cancer tool platform using a cohort of 183 PCa samples, data source number: GSE21032 (27).

Chromatin Immune Precipitation

22RV1 cells were incubated with 56merA¹⁵ (500 nM) or vehicle for 24 h in 2 mM Gln having RMPI, 5% FBS. After incubation, the cells were placed on ice and standard ChIP analysis procedures were followed (4). The antibodies used for immunoprecipitation were IgG (Millipore), RNA-pol II (Millipore) and FOXM1 (Invitrogen). Then, quantitative PCR was performed to determine FOXM1 enrichment on the FGF8 promoter.

ELISA and enzyme activity of GLS1

100 ×10³ cells were plated into a 12-well plate, after 48 hours, condition media was collected and centrifuged at 14,000 rpm. Aliquots were prepared in PBS 1X for ELISA and the protocol was performed following the manufacturer’s instructions (My BioSource). The enzyme activity was performed using the GLO-glutamate assay (Promega), then, Km was calculated according to Michaelis-Menten equation. The initial velocity of the product (mM/min glutamate) versus the substrate (mM L-Gln) was plotted.

Quantitative real-time PCR and RNA sequencing

Total RNA was extracted using the mini kit RNeasy (Qiagen) or Trizol (Sigma) according to the manufacturer’s indications. Reverse transcription was performed for quantitative real-time PCR, using oligonucleotide primers (Supp Table S1), then the data was analyzed by the 2-ΔΔCt method and represented as the normalized expression relative to the expression of the constitutive ribosomal gene, 36B4 mRNA. RNA sequencing was performed using paired end reads per run on the Illumina NovaSeq 6000 platform of Novogene.

Cell viability assays

Cell viability was measured after 48 hours incubation with vehicle or corresponding treatments by cell counting using trypan blue exclusion in the cell counter (Bio Rad, TC120), or by crystal violet staining as showed in the corresponding figure legend. L-Gln was used at 2 mM concentration for cell viability experiments unless a different concentration is indicated in the figure legends.

Immunoblotting and Immunohistochemistry

Protein from tumor lysates were quantified and western blotting was performed for FOXM1 and GAPDH using alkaline phosphatase-conjugated secondary antibodies (Sigma-Aldrich) on nitrocellulose membranes (Millipore) following separation on acrylamide gels. Harvested tumors were fixed, paraffin embedded and 5 μm sections subjected to immunohistochemical analysis. Sections were subjected to immunohistochemical staining for phosphorylated histone H3 (1:1000, Cell Signaling, Danvers, MA), survivin and TUNEL (Thermo Fisher Scientific Inc). Gold nanoparticles were visualized in tissues sections by silver enhancement staining following the manufacturer’s instructions (Sigma Aldrich).

Plasma stability

Stability of the aptamers in human plasma was tested before in vivo studies. The aptamers (1 μl, 10 μM) were incubated in 10 μl of plasma at 37 °C for 0–24 hours. After incubation, the plasma was diluted in 10 μl of PBS 1X, then, a qPCR reaction was performed. A standard curve of each aptamer was performed to correlate Ct values with aptamer concentration.

Animal studies

All animal procedures were performed according to the protocol approved by the Institutional Animal Care and Use Committee at Cedars-Sinai Medical Center. Male NSG mice (Jackson Labs, Bar Harbor, ME, USA), 6–8 weeks old, were used for subcutaneous or prostatic orthotopic grafting, as previously described (4,28). For the subcutaneously grafted tumors, 22RV1 cells were combined with prostatic CAF in a ratio of 1×10⁶ / 3×10⁶ respectively. After the tumors reached 100 mm³, mice were randomly selected and treated daily for 4 days with either vehicle (saline) or 56merA¹⁵ (4 mg/kg). In the gold nanoparticles experiments, the subcutaneous engraftments were treated three times per week with either enzalutamide (1 mg/kg) or 56merA¹⁵ (0.72 mg/kg) functionalizing 10 nm golds nanoparticles (171.1 ng/mouse). Tumors of the orthotopically grafted mice 22RV1 (2×10⁵) and CAF (6×10⁵) were expanded for 8 weeks prior to treatment with apalutamide (50 mg/kg) or apalutamide in combination with 56merA¹⁵. In the experiments involving golds nanoparticles, C42B-ER (2×10⁵) and CAF (6×10⁵) recombinant tumors orthotopically grafted were expanded in mice for 5-weeks under enzalutamide prior treatment with either 56merA¹⁵ (daily) or 56merA¹⁵-Au^np (3-times per week) interperitoneally. Apalutamide and enzalutamide were administered 3-times per week by oral gavage.

Statistical analysis

The student’s T-test was used to compare vehicle to each treatment. Two-way ANOVA was used to compare the effect of multiple treatments in multiple groups. The results were expressed as individual data points, or as the mean ± S.D. The p values < 0.05 were considered statistically significant using GraphPad Prism software (GraphPad, San Diego, CA).

Data Availability

RNA sequencing data is deposited at GEO repository with accession number GSE232596. An interactive sunburst plot of cyanobacteria GlnA riobswitches: https://doi.org/10.7910/DVN/EWB0OR. All other raw data are available upon request from the corresponding author.

RESULTS

Synechococcus elongatus glnA aptamer optimization

The glnA aptamer is a highly conserved L-Gln sensor element of a riboswitch found in many organisms (13–15). The 1,410 glnA phylogenetic neighboring aptamers of bacteria and viruses including 82 of the cyanobacteria order were aligned to develop an interactive sunburst chart (Fig. 1A) (29–31). The corresponding phylogenetic tree represented energy-minimized modeling of the unbound aptamer (minimum free energy, MFE) and relative binding free energy (RBFE) of aptamer/L-Gln binding for each species in two parallel heatmaps. The secondary structures were obtained by partition function (Fig. 1B)(19,20). RNA tertiary structures of the modified aptamers were predicted and docking simulations conducted by running two thousand iterations for each aptamer (21,32). Although the glnA aptamer from Phormidium Tenue, a cyanobacteria of the Oscillatoriaceae family, had a favorable calculated MFE (−21.9 kcal/mol) and RBFE (−171.66 kcal/mol), it was not considered further due to its limited predicted L-Gln specificity compared to other amino acids by in silico predictions (Supp Table S1). Synechococcus elongatus glnA aptamer of 67 ribonucleotides (67mer) was predicted to have high specificity but low affinity for L-Gln (33). We chose to improve the affinity of the 67mer through rational in silico modifications guided by constraints dictated by 16 evolutionarily conserved motif elements among the glnA aptamers (Supp Fig. S1A). The strength of ligand binding affinity of the aptamers based on their calculated RBFE values was predicted to reflect its binding affinity. The K_d and K_int values were obtained by nanogolds aggregation and fluorescent intercalator displacement (FID) assays as complementary approaches to confirm our modeling results with in vitro transcribed aptamers (Supp Table S2)(34,35). Sections of the 67mer with the least conservation were randomized for iterative in silico testing cycles of MFE and equilibrium constant (K_eq) in the Gln-bound state. These modifications comprised changes in sequence length, nucleotide replacements, and A-T zipper rearrangement. Accordingly, aptamers of 60 and 56 ribonucleotides (60mer and 56mer, respectively) were chosen for further analysis based on their lower L-Gln docking RBFE values. The 60mer was a result of shortening of the 3’ tail from the original 67mer resulting in a significantly greater L-Gln binding K_eq to that of the 67mer (Supp Table S3, Supp Fig. S1B). The 56mer had an additional 5’ deletion of the P1 stem, resulting in elevated L-Gln binding. These modifications in sections one, two, fourteen, fifteen and sixteen significantly affected the overall tertiary structure. The in silico L-Gln docking estimations of the native 67mer were congruent to the reported S. elongatus glnA crystal structure by Ren et al., having the L-Gln binding site between the P1 stem and E loop (Supp Fig. S1C) (33). For the 56mer, a predicted L-shaped conformation was modeled with the L-Gln binding pocket of unpaired purines within the P2 (G²³ to G²⁴) and the E loop (A⁴⁷ to A⁴⁸) sections (Fig. 1C, 1D). Common to all three aptamers, the secondary and tertiary structures had a long-range ‘linchpin’ Watson-Crick G-C pair through P1 and E Loop (13,33).

Figure 1. Rational in silico modifications of the glnA aptamer.

A. Sunburst plot illustrates the top-level cyanobacteria taxonomic group in the center and each subsequent level of the hierarchy in concentric rings (left). Phylogenetic association among the cyanobacteria orders Stigonematales (S), Chroococcales (C), Pleurocapsales (P), Chroococcidiopsidales (H), Nostocales (N), Oscillatoriales (O), and undefined order (–). The numbers on the sunburst plot correspond to the order of the highest equilibrium constants in order (K_eq, 1–10) with those names in the heatmap depicting calculated MFE and RBFE of unbound and L-Gln-bound aptamer, respectively (right). B. Secondary structure showing the long-range ‘linchpin’ Watson-Crick G-C pair through P1 and E-Loop (blue dashes line). The colors of the nucleotides suggest the probability of the predicted position. C. 3D structure of the 56mer bound to L-Gln within the E loop pocket. D. The L-Gln binding pocket of the 56mer indicates the interacting nucleotides and magnesium (Mg⁺⁺) in the transverse plane. E. A heatmap representation of relative binding affinity of aptamers to indicated amino acids based on fluorescent intercalator displacement (FID) assay. Data analyzed using tow-tailed student t test, * P < 0.05, ** P < 0.01, *** P < 0.001.

The optimized aptamers having lowest RBFE and highest K_eq were transcribed for kinetic studies. To measure the L-Gln specificity of the aptamers with structurally related-amino acids, a fluorescent intercalator displacement (FID) assay was performed with thiazole orange. The lower fluorescence intensity indicated higher amino acid binding based on greater thiazole orange displacement (36). As predicted by in silico modeling, the 60mer affinity for L-Gln was greater than that observed for the 67mer, and the 56mer was significantly greater than either of the other aptamers (ANOVA P < 0.0001; Fig. 1E). The FID approach supported the 56mer to have significantly greater affinity to L-Gln compared to structurally similar asparagine, glutamate, threonine, glycine, alanine-glutamine, and even its enantiomer, D-Gln, not observed with the 67mer and 60mer (Supp Table S4) (13,33).

56merA¹⁵-mediated L-Gln depletion

Aptamer binding was further characterized and modifications evaluated. To better determine the binding kinetics of the aptamers we chose a label-free colorimetric gold nanoparticle-based method, commonly used in biosensor studies (24). We engineered 15, 20 and 25 nucleotide polyA-tails located on the 3’ ends of the aptamers to immobilize the aptamers to gold nanoparticles (Au^NP). Unexpectedly, the addition of this 15 base-3’ polyA-tail on the 56mer (56merA¹⁵) resulted in a lower calculated RBFE and higher K_eq compared to the native 56mer (Supp Tables S5). Due to the importance of cations being crucial for riboswitches activities, the previously reported importance of Mg²⁺ by glnA was confirmed for L-Gln binding by the 56merA¹⁵ in our model, and nanoparticle-based kinetics (Supp Table S6, Supp Fig. S2A) (33,37). Although in silico studies demonstrated limited L-glutamate (L-Glu) binding by the 56merA¹⁵, this was further validated by FID studies over a broad concentration gradient (Supp Fig. S2B). The L-Gln binding pocket of the 56merA¹⁵ identified by in silico docking studies was tested by incubating the aptamer with short oligonucleotides complementary to the P2, P3 and E loops sections of the aptamer (Fig. 2A). The spectrophotometric red-shift of the nanoparticle-associated aptamer helped to determine which oligonucleotide interrupted L-Gln binding. Although blocking the P3 loop had negligible effect, the two oligos complementary to the E loop sections, significantly limited the L-Gln binding (P < 0.01); moreover, blocking the P2 loop further suppressed L-Gln binding (P < 0.001). To pinpoint the critical nucleotides for L-Gln interaction, in silico mutation analysis supported the P2 section to be important. The replacement of guanine at position 24 (G²⁴) with adenine (A) or uracil (U) resulted in the poorest predicted binding with higher free energy (Supp Table S7). The 56merA¹⁵ binding affinity for L-Gln (K_D = 0.261 μM) was found to be more than 10-fold greater than that of the G²⁴→A²⁴ substitution (mut56merA¹⁵, K_D = 7.6 μM) and 100-fold greater than that of the 60mer (K_D = 25.3 μM, Fig. 2B). The 56merA¹⁵ binding affinity was more than 1000-fold of the native 67mer (K_D = 323.3 μM) (13). As a point of reference, 67mer K_D determined by nanogold-based assay was comparable to the K_D range of 100–600 μM reported for S. elongatus glnA, previously determined by in-line probing and isothermal titration calorimetry experiments (13,33). The FID assay showed the critical role of G²⁴ in the 56merA¹⁵ binding pocket based on a significant fluorescence decrease in in the presence of L-Gln, compared to the mut56merA¹⁵ in a 30 minute-timescale (Fig. 2C). The characterization of the 56merA¹⁵ through multiple biochemical measurements supported its high specificity and affinity for L-Gln.

Figure 2. Characterization of L-Gln-binding aptamers.

A. Short complementary oligonucleotides were used to identify the L-Gln binding pocket through nanogolds binding assay. Colorimetric changes of the nanogolds due to oligonucleotide interruption of the 56merA¹⁵/L-Gln interaction were measured. B. 67mer-A¹⁵, 60mer-A¹⁵, 56merA¹⁵ and mut56merA¹⁵ L-Gln binding kinetics (K_D) were calculated by gold nanoparticle-based colorimetric assay. C. Fluorescent intercalator displacement (FID) assay was used to determine the kinetics of L-Gln binding to 56merA¹⁵ and mut56merA¹⁵ in the indicated timescale. D. Internalization kinetics (K_int) of [³H]-labeled L-Gln was determined in the context of increasing concentrations of 56merA¹⁵. Data analyzed using two-tailed t test and standard deviation with statistical significance, ** P < 0.01, *** P < 0.001.

We determined the ability of the aptamer to interrupt L-Gln uptake by PCa cells. This was achieved by measuring the uptake of L-Gln ([³H]-L-Gln) in the presence or absence of aptamers in cell lines with differential L-Gln sensitivities. The 56mer significantly decreased the uptake of extracellular L-Gln in both the more Gln-sensitive PCa line (C4–2B) and the less Gln-sensitive PCa line (ARCaP_M), lower uptake than either the 67mer or 60mer aptamers (Supp Fig. S2C). Accordingly, the 56merA¹⁵ reduced 22Rv1 cells uptake of [³H]-L-Gln in a dose-dependent manner (P < 0.0001; Supp Fig. S2D). At the highest dose of [³H]-L-Gln we achieved a 30% uptake inhibition by the aptamer. In a standard competition assay with [³H]-L-Gln and increasing 56merA¹⁵ concentrations revealed an internalization constant (K_int) of 24.31 nM by 22Rv1 cells (Fig. 2D).

A single L-Gln molecule binding capacity of the aptamer has been reported, however, there was a 10-fold difference in binding affinity and K_int, suggesting a secondary mechanism of aptamer-mediated L-Gln depletion. Considering that secretion of mitochondrial glutaminase enzyme (GLS1) has been reported by injured neuronal cells and its role in processing L-Gln (38,39), we performed a GLS1 ELISA on conditioned media from multiple cell lines. There was an apparent differential expression of GLS1 found on the conditioned media, with no distinguishing expression status among benign and malignant human prostate lines: 22Rv1, RWPE-1, BPH1, PC3, ARCaP_M, and C42B (Fig. 3A). However extracellular GLS activity measured from conditioned media from the same cell lines suggested similar activity, except for ARCaP_M cells, having the greatest activity. We reasoned that 56merA¹⁵ and GLS1 could interact in the extracellular space in the context of elevated L-Gln. The GLS1 expression, and to a much lesser extent GLS2 expression by 22Rv1 cells was a result of extracellular L-Gln (2 mM), regardless of 56merA¹⁵ administration (Fig. 3B). Subsequently, we found that the addition of 56merA¹⁵ to cultured 22Rv1 cells resulted in a significant elevation in extracellular GLS1 activity through the measurement of the conversion to glutamate (P < 0.0001; Fig. 3C). The addition of the specific GLS1 antagonist, CB839, inhibited GLS1 activity and to a lesser extent, in the presence of the aptamer (Supp Fig. S3A). Further incubation of L-Gln with recombinant GLS1, with increasing concentrations of 56merA¹⁵, resulted in dose-dependent elevation glutamate generation (Fig. 3D). Importantly, the direct measure of enzymatic activity showed 56merA¹⁵ significantly elevated GLS1 V_max (maximum enzymatic velocity) with a minimal change in the K_M (Michaelis constant [P < 0.0001]; Fig. 3E). Although 56merA¹⁵ was designed for L-Gln sequestration, the kinetic analysis indicated that the aptamer could serve as an allosteric enhancer of the GLS1 activity helping efficient extracellular L-Gln depletion. Modeling the interaction of the L-Gln/56merA¹⁵/GLS1 complex had a lower RBFE (−875.76 kcal/mol), compared to either L-Gln with either 56merA¹⁵ or GLS1 independently (Fig. 3F). The incubation 56merA¹⁵ with either 22Rv1 or C42B cells resulted in a significant depletion of extracellular L-Gln over time (P < 0.01; Fig 3G). L-Gln levels were significantly lowered by 56merA¹⁵, in both intracellular and conditioned media, with L-Glu concentrations lowered only intracellularly (Supp Fig. S3B). The capacity for L-Gln depletion by 56merA¹⁵ seemed to indicate a mechanism involving L-Gln sequestration and catalyzing extracellular GLS1 activity.

Figure 3. Characterization of L-Gln-binding aptamers.

A. Conditioned media from indicated benign and PCa cell lines were measured for GLS1 content by ELISA and extracellular GLS (ecGLS) activity normalized to cell number. B. A docking model was generated of L-Gln, 56merA¹⁵ and GLS1 to calculate RBFE of the complex. C. GLS1 activity was measured in 22Rv1 conditioned media following 8 hours of incubation with 200 nM 56merA¹⁵ in the presence and absence of GLS1 inhibitor, CB839. D. Michaelis-Menten kinetics of GLS1 was calculated in the presence and absence of 56merA¹⁵, where the V_max of GLS1 was found to be 1.70 mM/min and the addition of 56merA¹⁵ increased it to 1.85 mM/min. The K_M was calculated to 0.93 mM and 0.96 mM, respectively. E. GLS1 and GLS2 expression in 22Rv1 cells under L-Gln and 56merA¹⁵ treatment. F. GLS1 activity in conditioned media was measured as a factor of glutamate generated in 5 min by addition of 56merA¹⁵ in a concentration-dependent manner. G. Glutamine was measured in conditioned media following vehicle or 56merA¹⁵ incubation with 22Rv1 or C42B cells for the indicated times and analyzed by multiple comparison ANOVA. Data were analyzed using two-tailed T test indicate standard deviation with statistical significance unless otherwise stated. NS indicates not significant, * P < 0.05, ** P < 0.01, and **** P < 0.0001.

L-Gln regulation of FGF8 and ARSI sensitivity

Next, we chose to use the aptamer as a tool to interrogate the role of L-Gln in PCa progression. Initially, the impact of the 56merA¹⁵ treatment on oxygen consumption rate by 22Rv1 cells was measured. The 56merA¹⁵ reduced the peak respiration rate compared to 2 mM L-Gln containing media, interestingly lower than that achieved by glutamine-free media (Fig. 4A). As such, we measured the capacity of PCa lines to generate L-Gln as a source of anaplerotic activity reflected by the Seahorse assay. We found that the 22Rv1, C4–2B, and ARCaP_M lines, generated significant levels of L-Gln (P < 0.0001), however L-Glu accumulation was uneven among the lines (Supp Fig. S3C). To better understand downstream ramifications of aptamer-mediated L-Gln depletion, RNA sequencing was performed on 22Rv1 and C4–2B cells in the presence or absence of the 56mer aptamer (Fig. 4B). Enrichment analysis in both cell lines showed upregulation of PBX1, EED, CBX2, EZH2, and SUZ12 target gene activity, involved in the maintenance of cell fate and condensed chromatin state, among the top five transcriptional pathways involving polycomb group complex proteins (Fig. 4C). The common pathways downregulated by 56merA¹⁵ in both cell lines included stem-associated master regulators as FOXM1, OCT1 (POU2F1), MYC, and SOX2. It captured our attention that FOXM1 was the most significantly downregulated pathway by the aptamer, in line with previous reports (6). Exploring this pathway, a genome-wide integrated Chromatin immunoprecipitation Enrichment Analysis (ChEA) predicted FOXM1 as a top upstream transcriptional regulator of FGF8 expression (40). In addition, the Taylor cohort of 218 PCa patients suggested a positive correlation between FGF8 and FOXM1 expression (r = 0.5137; Fig. 4D) (27,41). Accordingly, multiple anatomically distinct PCa tissues demonstrated FOXM1 and FGF8 co-expression, in contrast to the normal prostate samples from organ donors. Metastatic PCa tissues had elevated FOXM1 and FGF8 compared to their localized counterparts. Consequently, treating 22Rv1 cells with the FOXM1 inhibitor, FDI-6, suppressed FGF8 expression to a greater extent than that by 56merA¹⁵, analogous to that observed with glutamine-free media (Supp Fig. S3D). The impact of L-Gln on FOXM1 transcriptional activity was tested through chromatin immunoprecipitation (ChIP) analysis, where 56merA¹⁵ treatment significantly downregulates FOXM1 enrichment on the FGF8 promoter (Fig. 4E). Interestingly, AR inhibition by enzalutamide elevated FGF8b expression over 10-fold but the inhibition of glutaminase activity by CB839 was found to promote FGF8b further, to rule out the role of the citric acid cycle in the signaling axis (Supp Fig. S3E). Enzalutamide interestingly reduced maximal respiration of 22Rv1 cells, compared to vehicle or that of Gln-free media; the addition of 56merA¹⁵ further suppressed respiratory capacity (Supp Fig. S3F). The development of an aptamer with high specificity and affinity for L-Gln enabled its extracellular depletion revealing a novel metabolic mechanism in FGF8 expression that potentially impact castrate sensitivity.

Figure 4. Impact of L-Gln depletion.

A. In the Seahorse experiment, the 22Rv1 cells were pre-incubated with 2 mM Gln overnight, and oxygen consumption rate was measured in response to L-Gln-free media (vehicle) or 2 mM L-Gln in the presence or absence of 56merA¹⁵ for the duration of the study. B. RNA sequencing analysis of C4–2B and 22Rv1 treated with vehicle or 56mer for 48 hrs. Common significantly up and down regulated genes in the two cell lines are indicated in the volcano plot. C. Enriched analysis identified transcription factor pathways up and down regulated, common to 22Rv1 and C4–2B cells in response to the 56mer. Combined score enrichment (p-value × z-score) indicated. D. Correlation analysis of FOXM1 and FGF8 expression was performed on the Taylor dataset (27). E. ChIP analysis for FOXM1 enrichment on the FGF8 promoter was tested in the presence and absence of the 56merA¹⁵ in 22Rv1 cells. IgG and RNA polymerase II (Pol II) were used as negative and positive controls, respectively. The percentage of viable cells was normalized versus the treatment control. Data analyzed using two-tailed T test indicate standard deviation with statistical significance, **** P < 0.0001.

Since FOXM1 and FGF8 are important in castrate resistance we hypothesized that antagonizing L-Gln signaling with 56merA¹⁵ might enhance efficacy of the AR antagonists (16,42). We explored the effect of apalutamide on the glutaminase activity in conditioned media from 22Rv1 cells. We found the extracellular GLS1 activity to be significantly greater following apalutamide treatment (P < 0.0001; Fig. 5A). Therefore, we tested the effect of the 56merA¹⁵ with apalutamide on the FOXM1 expression in 22RV1, C4–2B and ARCaP_M cells. FOXM1 was downregulated by 56merA¹⁵ in 22Rv1 and C42B cells, and all three lines in the context of apalutamide (Fig. 5B). In further examination of the FGF8b isoform, previously identified by Bluemn et al. to be associated with bypassing androgen dependence, was downregulated by apalutamide and 56merA¹⁵, as single agents (P < 0.05), with greater suppression when combined (P < 0.001; Fig. 5C) (16). In addition, the mut56merA¹⁵ was not able to suppress the expression of FGF8b. Subsequent survival analysis with castrate resistant 22Rv1 cells showed that both apalutamide and 56merA¹⁵ limited cell viability, alone and in combination, but, the mut56merA¹⁵ had a negligible effect (Fig. 5D). Considering the promising impact that 56merA¹⁵ demonstrated on L-Gln depletion and FOXM1 signaling, animal studies were considered. Initially, we performed a stability test by incubating the 56merA¹⁵ in human plasma at 37 °C to find a half-life of 3.78 hours, as determined by qPCR (Supp Fig. S4A). To best mimic prostate tumors, tissue recombinants of 22Rv1 cells and primary human prostatic CAF in a 1:3 ratio were grafted in mice (25). Subcutaneous tissue recombinant tumors were expanded to 1 cm³ prior to intraperitoneal treatment with 4 mg/kg 56merA¹⁵ or saline, daily administrated for four days. This timepoint was chosen for it has the greatest hormone-dependent apoptotic response in mice (43). L-Gln-depletion by 56merA¹⁵ resulted in significant elevation of tumor cell death and decrease in mitosis, based on immunohistochemical localization of TUNEL and phosphorylated histone H3, respectively (Fig. 5E). In the subsequent orthotopic model, 6 weeks after grafting, the mice were randomized to receive 50 mg/kg apalutamide in the presence or absence of 4 mg/kg 56merA¹⁵ daily for four days. Combined therapy resulted in a significant elevation of cell death and reduced mitosis, compared to those receiving apalutamide alone (Fig. 5F). Western blotting of the orthotopic tumor tissues demonstrated 56merA¹⁵-mediated suppression of apalutamide-induced FOXM1 expression (P < 0.01; Fig. 5G). The ramifications of L-Gln-mediated master regulator activity was determined in terms of its impact on lineage plasticity. The aptamer reduced all but one of the neuroendocrine markers tested as well as elevated all but one of the basal differentiation markers (Fig. 5H). The luminal markers were inconsistently expressed by tumors. However, the extensive cell death with the aptamer treatment in the context of apalutamide would clearly affect the number of assessable luminal cells in the tumors. The data showed that elevated L-Gln to be necessary and sufficient for castrate resistance, as its depletion sensitized prostatic tumors to AR inhibition.

Figure 5. L-Gln depletion complements AR inhibition.

A. Extracellular GLS (ecGLS) activity measured from 22Rv1 conditioned media after 72hrs. B. FOXM1 expression was measured in indicated PCa cell lines by RT-PCR following treatment with vehicle (2 mM Gln), apalutamide in the presence or absence of 56merA¹⁵ or mut56merA¹⁵. C. The expression of FGF8b isoform was measured by rtPCR in 22Rv1 following the indicated treatments. Statistical analysis was performed by one-way ANOVA. D. Surviving 22Rv1 cells were quantified by trypan blue excluded cell counting 48 hours after indicated treatments in the presence or absence of 2 mM L-Gln. The 100% live cells refer to the number of cells originally plated. E. Vehicle (saline) or 56merA¹⁵ treatment of subcutaneously grafted tissue recombinants of 22Rv1 and carcinoma associated fibroblasts had differential TUNEL and phosphorylated histone H3 (p-HH3) immunohistochemical staining (n ≥ 7). F. Orthotopically grafted tissue recombinants of 22Rv1 and CAF were treated with apalutamide alone or in the presence of 56merA¹⁵ resulted in differential TUNEL and p-HH3 immunohistochemical staining (n ≥ 7). G. Orthotopic tumors from mice treated with apalutamide in the presence or absence of 56merA¹⁵ were evaluated for the expression of FOXM1 with protein control, GAPDH, by western blot. The normalized data is summarized in the bar graph. H. The same tissues were evaluated for the indicated PCa lineage plasticity markers using rtPCR in triplicate. The scale bar represents 100 μm. Data analyzed using two-tailed T test indicate standard deviation with statistical significance unless otherwise stated, * P < 0.05, ** P < 0.01, *** P < 0.001, and **** P < 0.0001.

Tumor targeting by aptamer functionalization of gold nanoparticles

Challenges of the aptamer-based depletion of L-Gln was reaching sufficient depletion levels under longer term use. We employed gold nanoparticles (Au^NP) to provide some tumor selectivity, having known safety and increased localization in tumoral tissues due to an effect known as enhanced permeability and retention (EPR) (44,45). Functionalization of 10 nm Au^NP by the aptamer polyA-modification (56merA¹⁵-Au^NP), made earlier for kinetic analysis, was demonstrated to have an absorption peak at 520 nm in an aptamer concentration dependent manner, resulting in a clear color change in the aptamer-Au^NP complex (Fig. 6A). The efficacy of the 56merA¹⁵-Au^NP in enzalutamide sensitization was initially tested in a subcutaneous 22Rv1/CAF tissue recombinant model. After allowing the tumors to expand to approximately 1 cm³, the mice were randomized for treatment with enzalutamide in the presence or absence of 56merA¹⁵-Au^NP or mut-56merA¹⁵-Au^NP under longitudinal monitoring. Single-agent enzalutamide and 56merA¹⁵-Au^NP provided significant reduction in tumor size, compared to vehicle (Fig. 6B). The combination therapy of enzalutamide and 56merA¹⁵-Au^NP further suppressed tumor growth, however mut-56merA¹⁵-Au^NP provided no such improvement in tumor reduction compared to vehicle. To determine the contribution of the CAF in the tumor models, a parallel subcutaneous grafting study was conducted with only 22Rv1 introduction for 14 days of treatment (Supp Fig S4B). Here, enzalutamide had little impact on tumor size, but 56merA¹⁵ administration as a single agent or in combination with enzalutamide had a similarly significant reduction in tumor volume, compared to vehicle. Next, in an orthotopic model using an enzalutamide-resistant C42B line with CAF, enzalutamide treatment with daily 56merA¹⁵ administration was found to be inferior to giving 56merA¹⁵-Au^NP 3-times per week in terms of tumoral growth (Fig. 6C). The functionalized Au^NP were localized in the orthotopic tumors by silver enhancement staining (Fig. 6D). At the time of harvest, 56merA¹⁵-Au^NP treated mice had significantly greater aptamer localized in the tumors compared to the liver and kidney tissues, in support of an EPR effect (Sup Fig. S4C). However, the mice given 56merA¹⁵ had no detectible aptamer by qPCR. Accordingly, 56merA¹⁵-Au^NP treatment was found to be superior to 56merA¹⁵ in decreasing both survivin and phosphorylated histone H3 localization in the tumors. These findings were supported by the observed concomitant reduction in FOXM1 and FGF8 expression by the 56merA¹⁵-Au^NP compared to 56merA¹⁵ treatment. In conclusion, the Au^NP-based delivery of the 56merA¹⁵ aptamer efficiently sensitized castrate resistant PCa models to AR inhibition.

Figure 6. Targeted L-Gln depletion by 56merA¹⁵-Au^NP support enzalutamide sensitivity.

A. The absorbance of a 56merA¹⁵ functionalized gold nanoparticle (56merA¹⁵-Au^NP) was measured by UV spectrophotometry at indicated concentrations of aptamer. B. Subcutaneous 22Rv1/CAF xenografts were expanded for 2 weeks and treated for an additional 2 weeks with vehicle or enzalutamide, with and without either mut56merA¹⁵-Au^NP or 56merA¹⁵-Au^NP. Tumor volumes were estimated longitudinally from the time of initial treatment and multiple comparison ANOVA was used to analyze the data. C. Orthotopically grafted C42B E-R/CAF tissue recombinants were expanded for 5 weeks prior to intraperitoneal treatment with 56merA¹⁵ or 56merA¹⁵-Au^NP in the presence of enzalutamide for two weeks. Final tumor weights were measured. D. Histochemical analysis on the orthotopic tumors demonstrated differential silver enhanced Au^NP (Ag-enhanced Au^NP), as well survivin and p-HH3 immunohistochemical localization was quantitate (n ≥ 6). The scale bar represents 100 μm. In addition, RNA from frozen tumor tissues were analyzed for the expression of FOXM1 and FGF8. Statistical significance of data analyzed were shown using two-tailed student T test, unless otherwise stated, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

DISCUSSION

PCa resistance to AR signaling inhibitor therapy is a leading cause of mortality from the disease. Multiple approaches can be taken to address the observed elevation in L-Gln associated with hormone therapy resistance. As intracellular glutaminase inhibition and individual transporter strategies have not been effective, we chose to limit L-Gln access to the cancer cells. Choosing aptamers as a means of L-Gln targeting was not initially considered to be easily translatable due to the limited clinical examples of aptamer-based therapeutics. This is in part due to the challenges of cross-reactivity and specificity aptamers generated through Systematic Evolution of Ligands by EXponential (SELEX) enrichment platforms (46). Exploiting the evolutionary specificity advantage of the S. elongatus cyanobacterial glnA riboswitch and optimizations constrained to areas of less sequence conservation enabled a significant increase in L-Gln affinity. However, the depletion of L-Gln in preclinical PCa models was achievable due to the surprising role the aptamer played in catalyzing extracellular glutaminase activity. The added functionalization of Au^NP with the L-Gln targeting aptamer allowed for tumor enrichment and restored sensitivity to AR antagonists, highlighting a potential therapeutic strategy to overcome CRPC.

Methodologic approaches using nanogold/spectroscopy and FID, supported the initial modeling of the L-Gln binding kinetics and binding pocket estimations, corroborating calculated RBFE. The 56merA¹⁵ aptamer may help to overcome an otherwise unfavorable oxidizing environment outside the cell to one more favorable for the reaction by proximal presentation of L-Gln and Mg⁺⁺ to GLS1. Aptamer-facilitated ATP-independent hydrolysis resulted in a significant increase in the GLS1 reaction rate (47). The identification of extracellular GLS1 activity was born from the surprising observation that glutamine depletion capacity was about 10-fold greater than aptamer binding affinity. A previous report cited extracellular GLS1 expression by injured neural cells (39). Although we did not find a pattern of elevated GLS1 secretion by PCa cells compared to their benign counterparts under untreated conditions, extracellular glutaminase was notably elevated by AR signaling inhibition. 56merA¹⁵ was identified as an allosteric catalytic activator, where the aptamer helped bring an enzyme and substrate together to increase the effective concentration of glutamine for optimal hydrolysis to glutaminase. In line with these results, in silico modeling suggested GLS-1 interaction within the aptamer-Gln complex, resulting in a lower free energy compared to the GLS-1-Gln complex in the absence of the aptamer. The release of glutamate by the reaction would free the aptamer to bind other neighboring L-Gln molecules, increasing the number of L-Gln molecules depleted by a single aptamer. Interestingly, the GLS1 allosteric enhancer properties of 56merA¹⁵, seemed to involve non-overlapping sites of the GLS1 interaction that of CB839-mediated suppression (48).

In its capacity to serve as a biologic tool, 56merA¹⁵ revealed a definitive role for L-Gln in tumor progression and AR antagonist sensitivity. Among the top upregulated transcriptional regulators was PBX1, a known tumor suppressor (49). In addition, the activity of multiple polycomb-group proteins was elevated (EZH2, CBX2, PHC1, EED, SUZI12) in response to the glutamine-targeting aptamer. It would suggest changes in the chromatin landscape. Considering the role of EZH2 particularly mediating lineage plasticity and AR stabilization, we found that 56merA¹⁵ had a limited impact on basal versus luminal differentiation of the prostate tumors in the context of apalutamide. However, we found that 56merA¹⁵ treatment resulted in lower FOXM1 expression and decrease in downstream neuroendocrine markers. As a recognized pro-metastatic factor, exploring the effects of FOXM1 on androgen sensitivity led us to identify its direct regulation of FGF8 expression. FGF8 signaling is a key effector in the maintenance of cell proliferation of castrate resistance (16). Based on its impact of FOXM1 and FGF8 on PCa progression, L-Gln can be considered an oncometabolite in the context of CRPC. 56merA¹⁵, alone and in combination with apalutamide limited cell viability. While we contend that L-Gln derived from fibroblasts and produced by PCa cells themselves contribute to observed tumor progression, we cannot rule out off-target effects of the aptamer.

The tumor dependence on L-Gln is amplified by AR inhibition (6). As stromal fibroblasts have consistently demonstrated capacity for potentiating resistance to AR inhibition (6,25,50), we showed that glutamine depletion by the aptamer provide elevated killing and reduced proliferation of the tissue recombinant tumor models compared to AR inhibition alone (apalutamide or enzalutamide). Systemic depletion of L-Gln by asparaginase in PCa mouse models was previously observed to support radiation sensitivity (51). Unfortunately, asparaginase administration is not a clinically viable option for PCa patients. Our work revealed a novel catalytic aptamer for L-Gln depletion in coupling glutaminase activity. 56merA¹⁵ limited tumor progression in both subcutaneous and orthotopic models through FOXM1 and FGF8 signaling downregulation in the context of two independent AR antagonists. The functionalization of gold nanoparticles with the 56merA¹⁵ aptamer helped to target L-Gln with the administration of only a fraction of that required for the aptamer alone. It should be noted that, PCa patients are generally given AR antagonists in the context of castrate conditions. As such, further preclinical studies are required to better realize the therapeutic potential of 56merA¹⁵-Au^NP in the treatment of CRPC patients. While we are not aware of somatic alterations in glutamine transporters, prostate cancer patients identified to have elevated circulating glutamine levels would likely be good candidates for such aptamer treatment. Based on our pervious publication, patients with plasma glutamine levels >2 mM were found to be insensitive to androgen receptor signaling inhibition (6).

Significance.

Depletion of glutamine, which can mediate hormone therapy resistance in prostate cancer patients, with a cyanobacteria-derived catalytic aptamer blocks FGF8 expression and sensitizes hormone refractive prostate tumors to androgen receptor inhibitors.

Abbreviations:

PCa

prostate cancer

AR

androgen receptor

CAF

cancer associated fibroblasts

L-Gln

glutamine

L-Glu

L-glutamate

FID

fluorescent intercalator displacement

FGF8

fibroblast growth factor 8

EPR

enhanced permeability and retention