2.1 Cell lines, cell culture, and androgen deprivation

Androgen-dependent LNCaP (ATCC) were routinely cultured in in RPMI-1640 medium (Corning) supplemented with 10% fetal bovine serum (FBS; Gibco, Invitrogen) and 1% penicillin/streptomycin (PS, Cell grow), and VCaP PC cells were cultured in DMEM medium (Corning) containing 4.5 g/L glucose supplemented with 10% FBS, and 1% PS. Both were grown at 37°C in 5% CO₂. Androgen deprivation was achieved via substitution of 10% charcoal-stripped serum (CSS) for FBS.²⁵ C4-2 and Du145 cells were cultured in RPMI or DMEM medium, respectively, with the same condition as above.

2.2 MS metabolite analysis

Triplicate analysis of cells placed in androgen depleted media was carried out at the West Coast Metabolomics Center at UC Davis. Briefly, untargeted metabolomics profiling screening for all detectable compounds using validated gas chromatography-time-of-flight mass spectrometry (GC-TOF MS) method was used to profile the cells at the two time-points (4 and 8 days after androgen deprivation). Details of the protocols used for metabolite extraction, derivatization, gas chromatography–mass spectrometry (GC-MS) data acquisition, and data processing have been described in detail previously.^{26, 27} Compound identifications were made based on their retention index and comparisons of their mass spectra to the BinBase metabolomics database using the BinBase algorithm.²⁸

2.3 OMI

Fluorescence lifetime images were taken on a custom built inverted multiphoton microscope (Bruker Fluorescence Microscopy), as previously described.²⁰ Briefly, the system consists of an ultrafast laser (Coherent Inc.), an inverted microscope (Nikon; Eclipse Ti), and a 40× water immersion (1.15NA; Nikon) objective. Four independent fields of view were acquired per replicate with two experimental replicates for each condition (8 total). NAD(P)H and FAD images were acquired sequentially for the same field of view using an excitation wavelength of 750 nm and a 440/80 nm emission bandpass filter for NAD(P)H fluorescence, and an excitation wavelength of 890 nm and a 550/100 nm emission bandpass filter for FAD fluorescence. Fluorescence lifetime images were collected using time correlated single photon counting electronics (SPC-150; Becker & Hickl) and a GaAsP photomultiplier tube (H7422P-40; Hamamatsu). A pixel dwell time of 4.8μs was used to acquire 512 × 512 pixel images over 60 s total integration time. The photon count rates were maintained at 2–6 × 10⁵ photons/s to ensure adequate photon observations for lifetime decay fits, and no photobleaching. The instrument response function was measured from second harmonic generation of urea crystals excited at 900 nm, and the full width at half maximum was calculated to be 220 ps. A Fluoresbrite YG microsphere (Polysciences Inc.) was imaged as a daily standard for fluorescence lifetime, fit to a single exponential decay, and the measured lifetime was 2.1 ns (n = 7), which is consistent with published values.²⁹

2.4 Quantification of OMI images

NAD(P)H and FAD fluorescence lifetime images were analyzed using SPCImage software (Becker & Hickl) as previously described.³⁰ Binning of only 3 × 3 pixels was used to preserve spatial resolution for each bin, the fluorescence lifetime decay curve was deconvolved with the instrument response function and fit to a two-component exponential decay model, $$I(t)={\alpha }_{1}\times {\exp }^{\left(\raisebox{1ex}{$ \mbox{-} t$}\!\left/ \!\raisebox{-1ex}{${\tau }_{1}$}\right.\right)}+{\alpha }_{2}\times {\exp }^{\left(\raisebox{1ex}{$ \mbox{-} t$}\!\left/ \!\raisebox{-1ex}{${\tau }_{2}$}\right.\right)}+C$$. In this model I(t) is the fluorescence intensity at time t after the laser excitation pulse, α represents the fractional contribution from each component, C accounts for background light, and τ represents the fluorescence lifetime of each component.³⁰ A two-component model was used because both NAD(P)H and FAD can exist in two conformational states, bound or unbound.^{18, 31} The mean lifetime (τ_m) of both NAD(P)H and FAD were calculated as τ_m = α₁τ₁ + α₂τ₂. The total intensity was calculated from the lifetime decay by summing the photons detected at each pixel in the image. The intensity of NAD(P)H was then divided by the intensity of FAD for each pixel to calculate the ORR.

A previously described an automated cell segmentation pipeline was created in Cell Profiler and applied to NAD(P)H intensity images.³² Briefly, pixels belonging to nuclear regions were manually identified and the resulting round objects were stored as a mask. Cells were identified by propagating out from each nucleus, an Otsu Global threshold was used to improve the propagation and prevent it from continuing into background pixels. Cell cytoplasm was defined as the cell border minus the nuclei. Values for τ_m, τ₁, τ₂, α₁, and intensity of NAD(P)H and FAD as well as the ORR were measured for each cell cytoplasm. For each variable, values for all cells within a treatment group were pooled together, 115–850 cells were analyzed per condition.

2.5 Metabolic inhibitor assay

Cells were seeded in 2 ml of either control or ADT media on 35 mm glass bottom imaging dishes at a density of 20,000 cells for LNCaP and VCaP, and 10,000 cells for C4-2 and Du145 and grown for 4–8 days. Media was refreshed every 2 days with the final replacement occurring 24 h before imaging experiments. Treatments at the corresponding doses were started at the appropriate time before imaging.³³ Two experimental replicates were repeated for each treatment.

2.6 Seahorse metabolic profiling

Oxygen consumption rate (OCR) and ECAR were measured with a Seahorse Extracellular Flux Analyzer XFe96 (Agilent), with three–four replicates per condition. LNCaP and VCaP cells were plated 1 and 2 days, respectively, before assay. Cell culture microplates were coated with Cell-Tak (22.4 μg/ml; Corning) before plating 1 × 10⁴ LNCaP or 3 × 10⁴ VCap cells per well. The assay medium used for LNCaP included pH-adjusted RPMI (Agilent) supplemented with glucose (10 mM), glutamine (2 mM), and pyruvate (1 mM), with VCaP medium using DMEM (Agilent) in place of RPMI. On the day of the assay, cells were washed twice with assay media, always leaving 30 μl of media in the wells, and incubated for 1 h in a 37°C non-CO₂ incubator before assay start.

Basal OCR and ECAR were normalized by DNA per well as quantified by Hoechst fluorescence, a marker for relative cell number. After completion of assay, plates were washed with 25% PBS, after which 100 μl of sterilized nuclease-free water was added and allowed to incubate for 2 h before being frozen at −80°C. Subsequently plates were thawed, incubated with Hoechst 33258 dye at final concentration of 6.7 µg/ml in high salt TNE buffer, and scanned on a plate reader using 360 nm excitation/460 nm emission.

2.7 Western blot

Cells were collected following 4 and 8 days of CSS exposure and western performed using previously described protocols.³³ Blots were then probed with antibodies overnight at 4°C against the following proteins: Fatty acid synthase (FASN) (Santa Cruz, 55580), acetyl-CoA carboxylase alpha (ACCα) (Santa Cruz, 137104), acetyl-CoA carboxylase beta (ACCβ) (Santa Cruz, 377313), carnitine palmitoyltransferase 1 (CPT1) (Santa Cruz, 393070), glutaminase (GLS) (Proteintech, 66265-1-lg), alanine/serine/cysteine-transporter 2 (ASCT2) (MyBioSource, 8292367), androgen receptor (AR) (Santa Cruz, 7305), and α-tubulin (Calbiochem, CP06). Bands were quantified by measuring the signal density, subtracting the background for each lane, and dividing by the tubulin band density before normalization to the cellular control.

2.8 Cell growth assays

DNA levels of control and treated cells were quantitated with a fluorescent DNA assay as described. LNCaP and VCaP, respectively, were plated in 96-well culture plates containing FBS. The following day, the media was switched to either ADT (CSS medium) or regular media. For simultaneous treatment, 2-DG was applied the same time with ADT. For sequential treatment, after 4 days with ADT the cells were treated with 2-DG. Conditions for ADT included: LNCaP, Control: 10% FBS; Low: 8% CSS + 2% FBS; High: 10% CSS; VCaP, Control: 10% FBS; Low: 6% CSS + 4% FBS; High: 10% CSS. Conditions for 2-DG included: 2-DG Low (mM): LNCaP, 1; VCaP 5; High, LNCaP, 5; VCaP 10. Plates were washed and incubated with Hoechst 33258 dye at final concentration of 6.7 µg/ml in high salt TNE buffer and scanned on a plate reader using 360 nm excitation/460 nm emission at Days 6 and 8 for 2 and 4 days 2-DG treatment, respectively.

2.9 Statistical analysis

Comparisons for mass spectroscopy altered metabolites and Seahorse metrics were completed as paired two-tail t-test. Differences in OMI variables between treatment and control cells were tested using a one-way analysis of variance of control versus treatment, all p-values <0.05 were defined as significant. Treatment effect size of OMI of inhibitors was calculated using Glass's Delta (GD) [(µ_treatment − µ_control)/σ_control].^{34, 35}

3.1 Acute ADT induces metabolite profile alterations in PC with an abundance of glycolytic intermediates

Our group has phenotypically characterized the response of PC cells (LNCaP and VCaP) to ADT and found that most cells initially respond with decreased proliferation accompanied by apoptosis in a subset of cells within 48 h (2%–4%).^{6, 33} Cells then undergo senescence and quiescence by Day 8, with their associated phenotypic and gene expression changes. To obtain an overview of metabolism during this time frame, MS was used to temporally measure changes in the metabolite profile of the hormone sensitive (LNCaP) cell line at 4 and 8 days after ADT. Hormonal depletion was accomplished in cell culture using CSS removing androgens. MS evaluation of 252 metabolites found 49 and 135 were significantly different (p < 0.05) compared to control after 4 and 8 days of treatment, respectively (Figure 1A). At 4 days, 45% of the significantly altered metabolites were increased relative to control. In the 8-day group, this number increased to 59%, suggesting that some metabolites that were initially depressed relative to control became more abundant with longer exposure to ADT.

Metabolites were then organized into groups based on function (Figure 1A). Notably, there was a large group of metabolites that were significantly increased following ADT that belong in the carbohydrate metabolism category, specifically glycolysis (Figure 1B,C). Also of interest is an accumulation of substrate upstream to alpha-ketoglutarate in the TCA cycle (Figure 1D). These changes in metabolite levels after ADT motivated additional study of functional changes in redox state and oxygen consumption to assist in identifying whether metabolite accumulation might be due to increased flux through these pathways upstream, or decreased flow downstream.

3.2 ADT decreases the ORR in hormone sensitive, but not castration resistant cell lines in the acute phase

We expanded our analysis to both hormone sensitive and resistant cell lines and employed OMI to monitor the oxidation–reduction state of live cells.²⁰ ORR measures the autofluorescence intensity of NAD(P)H relative to FAD and provides label-free images of the oxidation–reduction (redox) state of the cell thus reflecting cellular metabolism.^{36, 37} The fluorescence properties of NADH and NADPH overlap and are jointly referred to as NAD(P)H.

Qualitative ORR images in control and Day 8 ADT conditions confirm that cells are viable (Figure 2A). LNCaP cells exhibit a morphological shift to a more elongated shape after 8 days of ADT consistent with prior studies.^{38, 39} No morphological change after ADT was seen in the castration resistant lines C4-2 and Du145.⁴⁰ The ORR significantly decreases with ADT in LNCaP and VCaP, at both 4 and 8 days compared to hormone replete conditions (Figure 2B,C). The reduction in the ORR with ADT in hormonesensitive lines is driven by decreases in NAD(P)H intensity and increases in FAD intensity (Supporting Information: Figure 1), reflecting a more oxidized cell redox state. As a control, the androgen resistant lines C4-2 and DU-145 were imaged after 8 days of ADT exposure and show an increase in the ORR or no change, respectively (Figure 2D,E).^{29, 41} NADH(P)H and FAD mean lifetimes also change when cells are grown in ADT media, however, the changes are not consistent across hormone sensitive or castration resistant cell lines (Supporting Information: Figure 2).⁴² Thus, OMI demonstrates that hormone sensitive cell lines show significant decreases in ORR following ADT compared to castration resistant lines, confirming that ADT induces an oxidized redox state in hormone sensitive PC cells. These data indicate that the abundance of glycolytic intermediates seen with MS is likely due to a decrease in intracellular glycolytic flux.

3.3 Seahorse and western blot analysis of ADT-treated cells reveals overall depression in energy metabolism resulting in a quiescent/senescent phenotype

To further examine the baseline metabolic responses of PC and the changes that are induced with ADT, we employed the Seahorse extracellular flux assay. The Seahorse platform detects real-time changes in cellular bioenergetics through measurements of OCR and ECAR. Together, OCR and ECAR can characterize the metabolic phenotypes of cells.^{43, 44} OCR and ECAR significantly decreased with 8 days of ADT exposure compared to control in both LNCaP and VCaP cells (Figure 3A,B and Supporting Information: Figure 3). OCR is related to mitochondrial respiration while ECAR is related to anaerobic glycolysis and CO₂ production via the TCA cycle. The reduction of both OCR and ECAR with ADT across both cell lines indicates a decrease in overall energy metabolism. We find that hormone sensitive cell lines have an energetic/glycolytic metabolic phenotype under control conditions but shift to a more quiescent/senescent phenotype after 8 days of ADT (Figure 3C).

We next examined the protein expression of several key regulatory enzymes in FA and glutamine utilization (Figure 3D). An analysis of protein concentrations can provide insight into flux through pathways, especially with enzymes that act as rate limiting steps, but do not directly reflect enzyme activities.¹⁷ These pathways were specifically chosen as OMI and seahorse indicated a decreased role of glycolysis and PC is known to rely on FA metabolism.^{16, 45} Three enzymes related to FA synthesis, including FASN and two isoforms of ACC showed decreased expression after ADT (Figure 3D,E). Similarly, the expression of enzymes responsible for glutamine uptake and hydrolysis, ASCT2 and the kidney-type glutaminase (KGA) isoform of GLS1, decrease after both 4 and 8 days of ADT (Figure 3D,E). Importantly, AR expression decreases in early ADT, at both 4 and 8 days of ADT consistent with the literature.³⁹ In summary, hormone sensitive PC cells treated with ADT that exhibit overall depression in energy metabolism concurrently appear to downregulate key regulatory enzymes in FA and glutamine utilization in the acute phase.

3.4 ADT reduces PC cell sensitivity to targeted metabolic pathway inhibitors in the early phase

There has been interest in whether inhibitors of metabolic pathways might improve the response to ADT given recent data demonstrating modest improvements in survival with metformin and statins in combination with ADT.^{8, 9} OMI is a new approach to gauge the efficacy of drugs in living cells. Based on our data, we selected several inhibitors of glycolysis, glutaminolysis, FA synthesis, electron transport chain (ETC), and FA oxidation (Figure 4A) to test on hormone sensitive PC cells after ADT. Specifically, 2-DG (2-Deoxy-d-glucose) inhibits glycolysis, Malonate inhibits succinate dehydrogenase subunit A and the ETC, BPTES [Bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide] blocks GLS1 and thus glutamine conversion to glutamate, TOFA [5-(Tetradecyloxy)−2-furoic acid] inhibits FA synthesis via ACC, and etomoxir blocks FA oxidation by inhibiting CPT1. Previous work indicates that T cells and breast cancer cells that are sensitive to 2-DG, Malonate, BPTES, and etomoxir have a decreased ORR, while cells sensitive to TOFA have an increased ORR.^{46, 47} Significant changes with the addition of these agents to ADT may result in either an increased or decreased ORR dependent on the pathway inhibited.

Both hormone sensitive cell lines, LNCaP and VCaP, demonstrate altered ORR responses to metabolic inhibitors after ADT compared to control conditions (Figure 4B,C). Responses demonstrate some variation between lines which is unsurprising due to their different genetic phenotypes.⁴⁸ LNCaP cells in FBS control conditions show altered ORR (normalized to the media control) for all metabolic inhibitors (Figure 4B). However, after 8 days of ADT LNCaP cells only remain sensitive to 2-DG and Malonate treatment, and no longer respond to the other metabolic inhibitors tested (Figure 4B). VCaP cells in FBS alter ORR with 2-DG, Malonate and TOFA treatment. After ADT, VCaP cells remain sensitive to 2-DG and Malonate but are also sensitive to BPTES (Figure 4C). We find that both cell lines maintain consistent sensitivity to glycolysis inhibition (2-DG) both before and after ADT. The effect size, for 2-DG treatment of LNCaP was augmented with ADT (GD = 0.374) compared to control (GD = 0.353) as was VCaP (ADT GD = 0.675; control GD = 0.663). This enhanced sensitivity to 2-DG after ADT is consistent with our mass spectroscopy results (Figure 1). We also saw that both cell lines were consistently sensitive to ETC inhibition (Malonate); however, sensitivity to Malonate was reduced after ADT exposure in both LNCaP (control GD = 0.972; ADT GD = 0.678) and VCaP (control GD = 0.329; ADT GD = 0.284). Additional Mito-fuel flex results indicate changes in preferred fuel sources with ADT that support this resistance to metabolic inhibitors during acute ADT (Supporting Information: Figures 3 and 4).

Given both LNCaP and VCaP cells demonstrate consistent reductions in the ORR to 2-DG after ADT indicating a drug response, we examined whether this translated into changes in cell number. We treated cell lines with both 2-DG and ADT in two different ways, sequentially and simultaneously. First, the cells were treated by ADT for 4 days to induce an alteration in the redox ratio, and then 2-DG was administered for 2 or 4 days. Exposure to either ADT or 2-DG alone caused a reduction in growth in both cell lines (p < 0.05) compared to controls (Figure 5). Combining ADT followed by 2-DG resulted in a moderate calculated synergistic effect (0.5–0.8) in both cell lines. Second, the two cell lines were treated with both 2-DG and ADT simultaneously, however, the combination of the two agents did not show synergistic inhibition (Supporting Information: Figure 5). This emphasized the importance of ADT in inducing a metabolic susceptibility before drug inhibition.