Prostate cancer cells survive anti‐androgen and
mitochondrial metabolic inhibitors by modulating glycolysis
and mitochondrial metabolic activities
Hirak S. Basu PhD1 | Nathaniel Wilganowski BS1 | Samantha Robertson BS1 |
James M. Reuben PhD2 | Evan N. Cohen PhD2 | Amado Zurita MD1 |
Sumankalai Ramachandran PhD1 | Lian‐Chun Xiao MS3 | Mark Titus PhD1 |
George Wilding MD1
Department of Genitourinary Oncology, MD
Anderson Cancer Center, University of Texas,
Houston, Texas, USA
Department of Hematopathology, MD
Anderson Cancer Center, University of Texas,
Houston, Texas, USA
Department of Biostatistics, University of
Texas, MD Anderson Cancer Center, Houston,
Texas, USA
Hirak S. Basu, Department of Genitourinary
Oncology, MD Anderson Cancer Center,
University of Texas, Box 0018‐5, 1515
Holcombe Boulevard, Houston, TX 70030,
Email: [email protected]
Funding information
Congressionally Directed Medical Research
Programs, Grant/Award Numbers: W81XWH‐
15‐1‐0509, W81XWH2010306; National
Cancer Institute, Grant/Award Number: 5
R01 CA185251‐02
Background: Most cancer cells are more glycolytic even under aerobic conditions
compared with their normal counterparts. Recent evidence of tumor cell metabo￾lism, however, shows that some tumors also increase mitochondrial oxidative
phosphorylation (ox‐phos) at some disease states during progression and/or de￾velopment of drug resistance. Our data show that anti‐androgen enzalutamide
(ENZA) resistant prostate cancer (PCa) cells use more mitochondrial metabolism
leading to higher ox‐phos as compared to the ENZA‐sensitive cells and can become
vulnerable to mitochondrial metabolism targeted therapies.
Methods: Seahorse assay, mass spectrometry and high resolution fluorescence
confocal microscopy coupled with image analysis has been used to compare mi￾tochondrial metabolism in ENZA‐treated and ‐untreated anti‐androgen‐sensitive
LNCaP and ‐resistant C4‐2, CWR22ν1, and PCa2b cells. Ex vivo fluorescence mi￾croscopy and image analysis has been standardized to monitor mitochondrial
electron transport (ETS) activity that likely increases ox‐phos in circulating tumor
cells (CTCs) isolated fom patients undergoing AR‐targeted therapies.
Results: Our data show that PCa cells that are resistant to anti‐androgen ENZA
switch from glycolysis to ox‐phos leading to an increased ETS activity. ENZA pre￾treated cells are more vulnerable to ETS component complex I inhibitor IACS‐
010759 (IACS) and mitochondrial glutaminase inhibitor CB‐839 that reduces glu￾tamate supply to tricarboxylic acid cycle. CTCs isolated from 6 of 20 patient blood
samples showed relatively higher ETS activity than the rest of the patients. All six
patients have developed ENZA resistance within less than 6 months of the sample
Conclusion: The enhanced growth inhibitory effects of mitochondrial metabolic
inhibitors IACS and CB‐839 in ENZA pretreated PCa cells provides a rationale for
designing a drug combination trial. Patients can be selected for such trials by
monitoring the mitochondrial ETS activities in their CTCs to maximize success.
Prostate cancer (PCa) is the second leading cause of cancer deaths
among US men. In the last year, over 186,000 men were newly di￾agnosed with PCa and more than 30,000 of them died of the disease.
Clinical management of PCa is challenging due to diverse clinical
outcomes.1 While a majority of the patients, who are diagnosed at
the early stage of the disease are cured by the first line of treatment,
a substantial portion of them progress to advanced, metastatic dis￾ease, which responds poorly to most cancer therapies. Despite in￾tense research during the last decade, the mechanism of developing
anti‐androgen‐resistance remains mostly unknown.
Activation of the androgen receptor (AR) remains central to
most PCa growth, recurrence and progression.1,2 Androgen‐
activated AR translocates to the nucleus and initiates expression of
genes that are involved in PCa proliferation. Targeting the androgen‐
signaling axis using androgen deprivation therapies (ADT) is widely
used in patients with progressing PCas.1–3 Most patients respond to
such therapies for a while but eventually return to the clinic with
castrate‐resistant PCa (CRPC) that grows in low serum androgen.
The mechanism(s) that turns androgen dependent PCa (ADPC) to
CRPC remains mostly unknown. This is a major hurdle to early
identification of patients likely to develop CRPC as well as devel￾oping new therapies for patients with potential to progress.
In recent years, much effort has been devoted to identifying and
understanding genomic and epigenomic changes that drive PCa
progression. Despite some progress in our understanding of genomic
and epigenomic alterations linking to CRPC and/or metastatic CRPC
such as aggressive variant (AVPC) and neuroendocrine (NEPC) var￾iant diseases, only a few genomic markers made significant impact on
clinical management of PCa, thus far. This is because many ADT and
anti‐androgen therapy resistant, lethal, metastatic PCas showed little
genomic difference from their organ confined counterparts.4 Trans￾lational studies to better interpret the genomic data to predict dis￾ease progression and therapy resistance are only currently being
Nearly a century ago Otto Warbürg and his coworkers demon￾strated that most cancer cells use more glycolysis for their energy
need compared with their normal counterparts incubated under
identical condition (Warbürg effect).5 Based on this principle, 18F‐
labeled 2‐deoxy‐2‐fluoro‐D‐glucose (18F‐FDG) that uses cellular glu￾cose uptake pathway was approved by FDA for cancer imaging in
patients by positron emission tomography (PET). PCa, however,
generally exhibits weak FDG‐PET intensity except at some ag￾gressive, therapy resistant state and/or specific organ site(s) of me￾tastasis.6,7 This led to our current assumption that certain metabolic
adaptation happens during PCa progression and/or during the de￾velopment of therapy resistance, albeit the exact role of metabolic
changes leading to such effect remains elusive.
In addition to glycolysis, tumor cells also engage mitochondrial
ox‐phos to support growth. Enhanced mitochondrial ETS activity is
responsible for cell proliferation and invasion of a variety of cancer
cell lines.8 In the last 5 years, the concept of ATP production by fatty
acids and/or amino acid oxidation under certain conditions has been
realized in cancer cells.9 Proper maintenance of mitochondrial
membrane potential by the ETS is also essential to prevent apoptosis
and to support proliferation.10 It has been reported that ovarian and
esophageal cancer cells switch from glycolysis to ox‐phos, while
transitioning from proliferative to invasive state.11,12 Depending on
the disease state, some tumors, including PCa, can also exhibit an
intermediate metabolic phenotype performing a truncated or re￾versed tri‐carboxylic acid (TCA) cycle in the mitochondria that export
certain TCA cycle intermediates such as citrate, α‐keto‐glutarate
(α‐KG), pyruvate, and so forth to the cytoplasm13 and related re￾ferences therein. These could then be utilized in fatty acid, amino
acid and nucleotide biosynthesis in the rapidly dividing cells. Tar￾geting mitochondria in these cells could be an effective strategy to
block cell invasion and proliferation.
The last 5 years has seen an increased effort in developing more
specific mitochondria targeting agents for anticancer drug develop￾ment. Among these, targeting mutant IDH1 inhibitor, AG‐120, and a
glutaminase inhibitor, CB‐839, are currently undergoing Phase II
clinical trials either as a single agent or in combination with other
approved therapies in acute myeloid leukemia (AML)14 and renal cell
carcinoma,15 respectively. Several other agents targeting glucose
transporter 1 (Glut1)******, hexokinase, pyruvate kinase M2 (PKM2)
**** and a mitochondrial pyruvate transporter are currently under￾going preclinical development and/or entered early Phase clinical
trials.16 A new agent IACS‐010759 (IACS) specifically targeting mi￾tochondrial Complex I17 in the ETS has been developed in our In￾stitution and is undergoing a Phase I trial against AML and other
solid tumors including PCa. To establish an effective mitochondrial
metabolism targeting agent against PCa, it is imperative to under￾stand the disease state, when tumors are mostly dependent on this
pathway for survival and therefore, will be most vulnerable to these
agents. Other therapies that can modify the tumor metabolism that
is more akin to that of this disease state should also enhance the
efficacies of these novel agents. Monitoring the mitochondrial ETS
activity that relates to ox‐phos status using minimally invasive cir￾culating tumor cell (CTC) analysis will help identify the disease state
as well as could be used as a pharmacodynamic marker for clinical
trial/use of these agents.
The data presented here show that PCas adapt to anti‐androgen
therapies by switching from glycolysis to ox‐phos for their bioener￾getic need and such adaptation can be monitored in patients by
analyzing mitochondrial ETS activity in CTCs in patient blood sam￾ples; we also propose that PCa tumors at this state will be most
vulnerable to the mitochondrial metabolism targeted therapies.
2.1 | Cell lines
LNCaP and CWR22ν1 cells are obtained from American Type
Culture Collection (ATCC). C4‐2 cells are a kind gift from Professor
800 | BASU ET AL.
Leelund Chung, University of California, Los Angeles (UCLA) and
PCa2b cells are a kind gift from Nora Navone, Genitourinary
Oncology, University of Texas MD Anderson Cancer Center.
2.2 | Seahorse analysis
Cells were cultured (20,000 cells per well) on a 96‐well Seahorse
plate in Dulbecco’s modified Eagle’s medium (DMEM) media sup￾plemented with 1% FBS and 4% charcoal stripped FBS
(DMEMF1C4)18 at 37°C, 5% CO2 for 48 h. Test compound at desired
concentrations was then added and incubated for 24 h. Cells were
then assayed following manufacturer supplied glycolysis and oxygen
consumption protocols (Agilent). The extracellular acidificationa rate
(ECAR) data are then corrected to measure the acidification rate
exclusively due to glycolysis by subtracting the pH change due to
mitochondrial metabolic activity (as measured from cellular oxygen
consumption rate [OCR]) and expressed as glycolytic proton excre￾tion rate (GlycoPER) using the manufacturer supplied algorithm.
2.3 | Lactate/pyruvate assay
A high pH HILIC method was developed on the Agilent 6560 LC‐IM‐
QTOF‐MS. A Waters® BEH Amide column (2.1 × 100 mm, 3.5 μm)
was used with a gradient of Buffer A: 95:5 H2O/ACN + 20 mM
NH4AcO + 20 mM NHOH (adjusted to pH 9) and Buffer B: 5:95 H2O/
2.4 | DNA assay
Cell growth was assayed using our published DNA assay used rou￾tinely in our laboratory.18
2.5 | MitoSOX Red (MSR) dye fluorescence with
MitoTracker Green (MTG) dye
The PCa cells plated in each well of a 96‐well plate were treated with
desired concentrations of test agents for 96 h. Cells were then in￾cubated with MSR for 3.5 h to stain superoxide producing mi￾tochondria followed by MTG dye for 30 min to stain all
mitochondria.19 The nonoverlapping fluorescence emission in￾tensities of MSR and MTG were determined in a fluorescence plate
2.6 | Microscope estimation for MSR fluorescence
At the completion of the treatment, cells were treated with MSR dye
that specifically binds to the mitochondrial membrane and fluoresces
upon oxidation only by the superoxide generated by the
mitochondrial ox‐phos activity19 following manufacturer supplied
protocol (Invitrogen). All microscopic images of mitochondrial
fluorescence were collected at ×500 maginification using a ×100
stimulated emission depletion (STED) objective in a high resolution
Leica SP8 STED fluorescence laser confocal microscope equipped
with UV and white light laser using sequential scanning and ob￾servation protocol at desired excitation and emission wavelengths at
50–100 nm resolution and analyzed using Image J image analysis
software. Mitochondrial intensities in cells in each focal plane at 0.5‐
μM‐thick optical slices and mean pixel intensities of all mitochondria
in the field were calculated after appropriate background subtraction
to reduce the nuclear fluorescence to zero. Images of cells in 5–10
random areas (depending on cell density) on each slide containing
about 600–1200 mitochondria were calculated. Each experiment
was repeated twice and all data were normalized to the fluorescence
of the control cells.
2.7 | Patient samples
Patients with CRPC undergoing AR‐targeted therapies such as en￾zalutamide (ENZA) or abiraterone in combination ADT therapy were
recruited and provided written informed consent under IRB ap￾proved protocol PA15‐0956 in accordance with the provisions of the
Declaration of Helsinki.
2.8 | Enrichment and immunocytochemistry (ICC)
staining of CTCs
Patient blood sample (10 ml) was collected by venipuncture,
stored in a chilled collection tube under an IRB approved pro￾tocol and CTCs were isolated using a Parsortix microfluidic sys￾tem Angle Europe Limited. The system enriches tumor cells based
on size and deformability without biasing the selection with any
pre‐determined surface antigen.20 After filtering blood through a
6.5‐µm critical gap, reversing the flow of buffer allows for en￾richment of live CTCs. CTCs were confirmed by ICC staining of
nucleated cells for AR, EpCam and pan‐cytokeratine.21 We used
the same method to identify AR‐positive nucleated PCa CTCs
from the patient‐derived xenograft (PDX) MDA‐PCa‐203 and
also from a patient blood sample. The conditions for blood col￾lection and storing in ice for less than 30 min before CTC isola￾tion remain identical for all samples to control for any metabolic
alterations during collection, transport and isolation.
2.9 | Quantification of mitochondrial superoxide
production in CTCs
CTCs were plated in a chamber slide and were blinded for patient
information using a pathologist supplied coding. Cells in some
chambers were incubated with MSR dye for 4 h in a 37°C CO2/air
BASU ET AL. | 801
incubator. Cells are then fixed and stained with Hoechst 33342
(Molecular Probes, Inc.), primary anti‐human AR (Santa Cruz
Biotechnology, Inc.) with secondary AlexaFluor 488 (Invitrogen) and
primary EpCAM (Abcam) with secondary AlexaFluor 680 (Invitrogen)
following our standardized immunocytochemistry protocol (see
2.10 | Statistical analysis
Patient disease outcome information was unblinded and subjected to
rigorous statistical analysis. Mitochondrial data for CTCs were
summarized using descriptive statistics. Linear mixed model was
fitted to compare data between responding (samples 1–6 and 8–13)
and non‐responding (samples 14–20) patients, including patients as
random effects to account for the intra‐cluster correlation for clus￾tered data. Wilcoxon rank sum test was also used for two group
comparisons. SAS9.4 software (SAS Institute) was used for the
3.1 | ADPC cells are more glycolytic than are
CRPC cells
Cellular OCR and pH changes in the media due to cellular acid
efflux (ECAR; mostly lactic acid from glycolysis) by cultured cells
can be measured by Seahorse assay at a high precision.22 This
assay has now been accepted as a gold standard for measuring
changes in glycolysis and cellular oxygen consumption (mostly
due to mitochondrial ETS activity) in cultured cells.22,23 The
ECAR data from ADPC LNCaP, LNCaP‐derived CRPC C4‐2 and
PDX CRPC PCa2b cells isolated from a patient’s bone lesion were
incubated in androgen reduced media (see Section 2). The results
are shown in Figure 1A. The corresponding basal OCR data are
shown in Figure 1B. All Seahorse data are normalized to viable
cell counts (~85%–90% viable under these conditions) as de￾termined by a propidium iodide dye fluorescence assay per￾formed after the completion of the Seahorse experiment.24 The
data show that C4‐2 and PCa2b cells are significantly less gly￾colytic and use more mitochondrial metabolism than do the
LNCaP cells.
3.2 | Anti‐androgen ENZA reduces glycolysis in
viable PCa cells
The ECAR and OCR in ENZA‐sensitive LNCaP and ENZA‐resistant
C4‐2 cells treated for 24 h with graded concentrations of ENZA
within achievable human serum level (22 μM)25 are shown in
Figure 1C,D. ENZA significantly lowers the ECAR with small changes
in OCR in both cell lines. These data have been further confirmed by
determining the changes in total (cellular+media) lactate/pyruvate
ratio using mass spectroscopy (Figure 1E). LNCaP cells produce more
lactate and less pyruvate than do C4‐2 cells confirming the Seahorse
data. ENZA significantly lowers lactate levels in the ENZA‐sensitive
LNCaP, but not in the ‐resistant C4‐2 cells at or near their respective
IC50 doses.
3.3 | Prolonged treatment with ENZA increases
mitochondrial superoxide production
ENZA treatment for 72–96 h at respective IC50 doses did not pro￾duce enough viable PCa cells for reliable Seahorse assay results. To
circumvent this problem, we have standardized a confocal micro￾scopy method coupled with an image analysis assay to estimate
mitochondrial ETS activity based on superoxide production25 (see
Section 2). The representative fluorescence images of LNCaP and
C4‐2 cells after 96 h treatment with ENZA (at or near the IC50 doses,
see above) were obtained using a fluorescence STED confocal mi￾croscope (Figure 2A). The fluorescence intensities of MSR dye are
quantitated (see Methods for detail) by Image J segmentation
(Figure 2B). The data are shown in Figure 2C–E. ENZA‐treatment
significantly increases the MSR fluorescence intensity in the ENZA‐
resistant CRPC C4‐2 and CWR22ν1 cells as compared to the control
untreated cells (Figure 2D,E). In ENZA‐sensitive LNCaP cells the
increase observed in not statistically significant (Figure 2C). The ef￾fect of ENZA is almost completely reversed by androgen treatment
suggesting the changes are mainly due to an on‐target effect of
ENZA interacting with AR.
To confirm that the microscopy image analysis data are not
due to any selection bias, a macroscale assay to quantitate mi￾tochondrial activities has also been standardized. The MSR/MTG
ratio represent the mitochondrial ox‐phos normalized per mi￾tochondrion. The increase in this ratio for LNCaP and C4‐2 cells
treated with ENZA compared with the ratio in untreated control
cells is small but significant (Figure 2F). This confirms the mi￾croscopic observation of relatively higher mitochondrial ETS
activity in the cells that survive in the presence of ENZA. The
relative increases in the MSR/MTG ratio is less in C4‐2 cells than
that observed in their microscopic images. This could be because
the macroscale assay may include some mitochondrial staining by
MTG in the dead cells that have reduced or stopped ETS activity
and thus, do not cause MSR fluorescence, but still retain some
membrane poteintial for MTG staining. These cells are not in￾cluded in microscopic analysis. On the other hand, the increasing
trend of ox‐phos in LNCaP cells observed in the microscopy as￾say, but did not attain statistical significance did reach statistical
significance in the 96‐well plate‐based assay due to
the ability to analyze a lot more cells as compared with the
limited number of fields analyzed using the high resolution
802 | BASU ET AL.
3.4 | Cells that switch to ox‐phos are more
vulnerable to mitochondria targeted agents
It is hypothesized that cells depending more on ox‐phos for their
energy need should be more vulnerable to mitochondrial
metabolism targeted agents. To establish this, we used two
agents – IACS‐010759 (IACS) and CB‐839 that are currently
undergoing clinical trials as anticancer drugs. IACS specifically
targets complex I of the mitochondrial ETS17 and CB‐839 inhibits
glutaminase26 that produces glutamate from glutamine, which is
FIGURE 1 (A) Glycolysis (ECAR) and (B) oxygen consumption rate (OCR) in LNCaP, C4‐2 and PCa2b cells grown in androgen reduced media
for 48 h; (C) ECAR and OCR in LNCaP and (D) C4‐2 cells grown in androgen depleted media for 48 h and then treated with ENZA for 24 h. All
ECAR and OCR data were normalized to viable cells as determined by propidium iodide staining (see text) at the end of each assay. Each data
point and error bar represent the mean and standard deviation of readings from 15 separate wells repeated three times. (E) Mass spectroscopic
analysis of lactate/pyruvate ratio in C4‐2 and LNCaP cells incubated in androgen depleted medium for 48 h and then treated for 24 h with
ENZA at their respective IC50 doses (20 and 10 µM, respectively). Each data point and the error bar are the mean and standard deviation of
2 parallel samples run at least twice. *p < .01, **p < .001. ECAR, extracellular acidificationa rate; ENZA, enzalutamide; OCR, oxygen
consumption rate
BASU ET AL. | 803
then converted to α‐ketoglutarate α‐KG and used as one of the
fuels for the TCA cycle.
We have determined the effects of both agents on ECAR (see
above) in LNCaP and C4‐2 cells with or without ENZA pretreatment
for 24 h. The results are shown in Figure 3A–F. Treatment with IACS
or CB‐839 decreases ox‐phos with a concomitant increase in glyco￾lysis in the surviving cells (Figure 3A,C,E). Pretreatment with ENZA
at close to the IC50 doses of the respective cell lines (10 μM for
LNCaP and 20 μM for C4‐2) for 24 h before IACS addition markedly
blocks the observed increase in glycolysis in both cell lines
(Figure 3B,D,F).
We anticipated that a decrease in mitochondrial ox‐phos, when
coupled with ENZA‐induced reduction in glycolysis, should have a
major reduction in cellular bioenergetics. This will enhance the
FIGURE 2 Representative confocal microscopy images of an optical section at ×100 and ×500 magnification each of (A) LNCaP and (B) C4‐2
cells treated first 24 h with ENZA at their respective IC50 doses (10 and 20 μM, respectively) and then incubated with MSR dye for 4 h
and examples of corresponding Image J segmentation (bottom section) for fluorescence intensity analysis. Mean pixel fluorescence intensities
of oxidized MSR dye in the mitochondria were calculated from individual mitochondrion MSR fluorescence image quantitation and averaged
over all mitochondria per cell as described in the text. of (C) LNCaP, (D) C4‐2 and (E) CWR22ν1 cells treated with vehicle control, 10 µM ENZA
(LNCaP) and 20 µM ENZA (C4‐2 and CWR22ν1), androgen mimetic 2 nM R1881, and ENZA + R1881 (E + R) were calculated from individual
mitochondrion MSR fluorescence image quantitation and averaged over all mitochondria per cell as described in the text. (F) MSR/MTG in
C4‐2 and LNCaP cells treated for 96 h with vehicle (control) and with IC50 dose of ENZA. Each data point and standard deviation are the mean
of the readings from six wells run in triplicates and repeated at least three times. *p < .05, **p < .005. ENZA, enzalutamide; MSR, MitoSOX Red;
MTG, MitoTracker Green
804 | BASU ET AL.
FIGURE 3 ECAR and OCR in (A) LNCaP, (C) C4‐2 treated with graded concentration of IACS for 24 h and (E) C4‐2 treated with graded
concentration of CB‐839 and (B), (D) and (F) are the corresponding treatment with IACS and CB‐839 after 24 h pretreatment with ENZA. All
ECAR and OCR data were normalized to viable cells as determined by propidium iodide staining (see text) at the end of the assay. Each data
point and error bar represent the mean and standard deviation of readings from 15 separate wells repeated at least three times. *p < .005.
ECAR, extracellular acidificationa rate; ENZA, enzalutamide; OCR, oxygen consumption rate
BASU ET AL. | 805
growth inhibitory activity of the mitochondria targeted agents in
ENZA pretreated cells. We tested the growth inhibitory effects of
72 h treatment with a graded concentration of IACS or CB‐839 with
or without 24 h ENZA pretreatmenton LNCaP and C4‐2 and a single
concentration of each of the agents on PCa2b cells. Concentrations
of CB‐839 were kept within the range of human serum Cmax and that
of IACS is kept much below the Cmax reported in preclinical in vivo
studies.17 ENZA was used at or below its published human serum
24 Cell growth data in terms of cellular total DNA fluores￾cence.18 are shown in Figure 4. The data clearly demonstrate that the
growth inhibitory effects of both IACS and CB‐839 are more pro￾nounced in ENZA pretreated cells. The effect was more apparent in
C4‐2 cells than in LNCaP cells probably due to relatively more mi￾tochondrial metabolism dependency of C4‐2 cells treated with ENZA
FIGURE 4 Effect of graded concentrations of IACS on the growth of (A, C) LNCaP and (B, D) C4‐2 cells after 96 h treatment with IACS or
CB‐839 alone or after 24 h pretreatment with ENZA. (E) Effects of ENZA, IACS, and ENZA + IACS (F) and the effects of ENZA, CB‐839 and
ENZA + CB‐839 on the growth of LNCaP, C4‐2, and PDX‐derived cells. Each data point and error bar represent the mean and standard
deviation of readings from 18 separate wells repeated at least twice. *p < .005. ENZA, enzalutamide; PDX, patient‐derived xenograft
806 | BASU ET AL.
than that of LNCaP cells (see Figure 2). Similar increase in growth
inhibitory effect of CB‐839 was also observed in two other anti‐
androgen resistant cell lines CWR22ν1 and PCa2b pretreated with
ENZA (data not shown).
3.5 | Reduction of mitochondrial ROS is related to
increased growth inhibition
We then used high resolution microscopy to observe changes in
mitochondrial ROS as measured by mean pixel MSR dye fluor￾escnence intensities (see Figure 2) in LNCaP, C4‐2, CWR22ν1
and PCa2b cells treated with CB‐839 with or without ENZA
pretreatment are shown in Figure 5. Mitochondrial superoxide
production remains unchanged in ENZA‐sensitive LNCaP cells,
but significantly increase after ENZA treatment in most
androgen‐resistant cell lines. As observed from the Seahorse
assay (Figure 3), CB‐839 causes a decrease in mitochondrial
metabolism in all cell lines. Significantly more reduction of mi￾tochondrial superoxide production was observed for CB‐839
treatement of ENZA pretreated cells. This effect is not seen in
cells, when CB‐839 is added either before or at the same time of
ENZA addition (data not shown).
3.6 | Increase in ox‐phos in CTCs from patients
developing resistance to AR‐targeted therapies
First, we have standardized a method of isolating CTCs from blood
samples from consented PCa patients. PCa CTCs were identified as
CD45 negative nucleated cells that are positive for AR and epithelial
cell adhesion marker (EpCam). Representative images of the isolated
CTCs stained for DNA (blue), AR (green) and EpCam (red) are shown
in Figure 6A.
To quantitate mitochondrial superoxide production in the CTCs,
we followed ex vivo MSR dye oxidation in the CTCs following the
protocol standardized for cell lines in our laboratory (see Figure 2). A
representative image of an optical section of a CTC isolated from a
PCa patient blood sample and its corresponding segmentation and
analysis are shown in Figures 6B and 6C, respectively. The back￾ground correction and threshold are adjusted to set all fluorescence
pixels within the nuclei below <5 fluorescence units to correct for
the mitochondria that are not in each optical section (0.5 μm) that
are being analyzed. The mean pixel intensities of MSR dye fluores￾cence in 5–6 optical sections per cell (approximately 1500–2000
mitochondria per cell) were determined from each CTC.
Thus far, we have collected 22 blood samples from 20 patients
with CRPC (age range 52–73) undergoing AR‐targeted therapies
FIGURE 5 Mean pixel fluorescence intensities of oxidized MSR dye in the mitochondria were calculated from individual mitochondrion
MSR fluorescence image quantitation and averaged over all mitochondria per cell as described in the text. (A) LNCaP, (B) C4‐2, (C) CWR22ν1,
and (D) PCa2b cells treated with vehicle control, 10 µM ENZA (LNCaP) and 20 µM ENZA (C4‐2, CWR22ν1 and PCa2h) for 96 h, CB‐839 for
72 h, and ENZA (96 h)+CB‐839 (72 h) computed from microscopic image analysis (see text). *p < .005. CTC, circulating tumor cell;
ENZA, enzalutamide; MSR, MitoSOX Red
BASU ET AL. | 807
FIGURE 6 (See caption on next page)
808 | BASU ET AL.
either ENZA or abiraterone in combination with androgen depriva￾tion. The patients are considered responding to therapy during
sample collection, when their prostate specific antigen (PSA) levels
are stable and are under 4 ng/dl for two consecutive serum PSA tests
within 6 months before and after sample collection. Patients with
rising PSA within that period are considered developing resistance to
From each 10 ml blood sample, we detected sufficient (10 or
more) analyzable CTCs from 18 out of 20 patients. Mean MSR dye
fluorescence intensities of a minimum of 10 CTCs per patient sample
are shown in Figure 5D. Samples #14–20 showed significant increase
in MSR fluorescence intensities representing enhanced mitochon￾drial ETS activity as compared to samples #1–6 and #8–13. Patients
contributing samples #15–20 have progressed to therapy resistance
within 3 months after the sample collection. Samples #6 and #7
(from one patient) and #13 and #14 (from another patient) were
collected longitudinally within a time span of 3 months for each
individual. As shown in Figure 6E,F, both samples #7 and #14 showed
significantly higher MSR dye intensity as compared to sample #6 and
#13, respectively. At the time of collection of samples #6 and #13,
both patients were responding to treatment with androgen synthesis
inhibitor abiraterone acetate plus ADT as concluded from their low
serum PSA level. At the time of collection of sample #7, the patient
had a stable PSA measurement, but the patient showed progressive
disease within 3 months after collection of the sample #7 and could
not be followed further for serum PSA or radiographic progression.
At the time of collection of sample #14, the patient's serum PSA
increased and subsequent follow up showed steadily increasing PSA
followed by radiographic progression of the disease.
While glycolysis is ~4 fold less in CRPC C4‐2 cells relative to its
parental ADPC LNCaP cells (Figure 1A), oxygen consumption in C4‐2
cells is only about 1.5‐fold more than that in LNCaP cells (Figure 1B).
As ox‐phos generates 18 times more ATP than does glycolysis, a
relatively smaller increase in oxygen consupmtion can compensate
for a larger decrease in glycolysis. It is noted that ENZA‐treatment
causes a marked decrease in glycolysis (Figure 1C) in ENZA‐sensitive
LNCaP cells relative to that in ENZA‐resistant C4‐2 cells (Figure 1D).
The Seahorse data are supported by metabolite estimation data
related to ENZA‐treatment of LNCaP and C4‐2 cells (Figure 1E). As
C4‐2 is much less glycolytic than is LNCaP (Figure 1A,E), it is difficult
to ascertain if the lack of ENZA‐effect in C4‐2 cells is related to its
ENZA resistance or the lack of sensitivity of the assay in cells at a
very low glycolytic state.
Mitochondrial ETS activity relates to ox‐phos in most cells not
treated with an uncoupling agent such as oligomycin. The ETS ac￾tivity could be measured at a much higher accuracy in single cells
using high resolution fluorescence microscopy of MSR dye fluores￾cence coupled with computer‐aided image analysis as compared to
Seahorse assay. A significant increase in MSR dye fluorescence in
ENZA‐treated cells shows that the relatively higher ETS activity is
indeed related to ENZA‐resistance in two ‐resistant cell lines C4‐2
and CWR22ν1 as compared to ‐sensitive LNCaP cells. The single cell
data are confirmed in a 96‐well plate‐based measurement of changes
in MSR intensity per mitochondria (MSR fluorescence normalized to
MTG fluorescence of total mitochondria per cell) in LNCaP and C4‐2
cells. Unlike the LNCaP cells, the MSR fluorescence in the ‐resistant
C4‐2 cells significantly increases with ENZA treatment. This effect is
more pronounced for IACS than for CB‐839 probably because IACS
directly inhibits ox‐phos by blocking the mitochondrial ETS activity,
while CB‐839 only indirectly affects ox‐phos by blocking the supply
of α‐KG, which is one of several metabolites that feed the TCA cycle
to generate electrons for the ETS.
Thus, we have shown that CRPC cells in general, are less gly￾colytic and use more mitochondrial ETS than their androgen de￾pendent counterparts. All PCa cells surviving in the presence of anti‐
androgens further increase their dependence on mitochondrial ETS
activity. Mitochondrial metabolism targeted inhibitors currently un￾der preclinical and clinical development decreases ox‐phos. Thus, we
hypothesize that CRPC cells surviving in the presence of anti‐
androgens will be more vulnerable to mitochondrial metabolism
targeting agent.
As anticipated, the mitochondria‐targeted metabolic inhibitors
strongly suppress mitochondrial ETS activity. ENZA pretreatment
further decreases mitochondrial activity (Figure 5) and increases
growth inhibitory effect of CB‐839 (Figure 4E,F). Interestingly, this
effect is not observed, when ENZA was added either with or after
CB‐839 treatment (data not shown). We hypothesize that the ENZA
induced increase in mitochondrial ETS activity in androgen‐resistant
cells happens due to changes in cellular metabolism either at or
upstream of the ETS (site of action of IACS) or glutaminase activity
FIGURE 6 (A) CTC isolated from a PCa patient blood sample. The CTCs were identified from CD45 negative nucleated cells that are
positive for androgen receptor and epithelial cell adhesion molecule as shown above the respective images; (B) a representative image of an
optical section of MSR fluorescence due to oxidation by mitochondrial superoxide of a PCa patient CTC and (C) image segmentation analysis of
the image in 5B for pixel fluorescence intensity analysis; (D) mean MSR dye fluorescence intensities of 10 or more CTCs obtained from each
PCa patient undergoing androgen‐signaling axis targeting therapies. Each sample was collected from an individual patient, except samples #6,
#7 and #13, #14 (marked in gray). Samples marked with a line on top are from patients who developed resistance to anti‐androgen therapy. (D)
Mean CTC MSR dye fluorescence intensities in patients with stable and progressive disease, p value for linear mixed model (***p < .0001). (E)
Mean CTC MSR dye fluorescence intensities for two patients while stable and near the time of progression, p value (**p < .005). CTC, circulating
tumor cell; PCa, prostate cancer; MSR, MitoSOX Red [Color figure can be viewed at wileyonlinelibrary.com]
BASU ET AL. | 809
(target of CB‐839). Therefore, ENZA cannot further increase the
mitochondrial ROS in any of the cells that are either pretreated or
co‐treated with the metabolic inhibitors. In fact, ENZA pretreatment
further decreases mitochondrial ETS activity in CB‐839 treated cells.
This suggests that ENZA‐treated cells become more dependent on
mitochondrial metabolism and reduce other cellular metabolic ac￾tivities upstream to mitochondrial metabolism such as glycolytic and/
or lipid biosynthetic pathways that may indirectly contribute to mi￾tochondrial ETS activity.
High resolution microscopic image analysis of CTCs showed a
significantly more mitochondrial ETS activities during or just before
the development of resistance to anti‐androgen therapy in all pa￾tients studied thus far.
ENZA treatment makes PCa cells more vulnerable to metabolic in￾hibitors that are under clinical development. The high resolution
microscopy method may be developed further for clinical application
to identify the disease state in patients, when they will best respond
to mitochondrial metabolism and ETS inhibitors.
We thank MD Anderson Genitourinary Oncology clinical research
group and all patients who participated in this study and their fa￾milies for their help and support. We also thank Joe Marszalek, Inst
Head, CCCT Transl Biology, MDACC for the supply of IACS‐010759.
The study was supported by the National Institute of Health: R01
CA185251‐02, DoD CDMRP: W81XWH‐20‐1‐0306 and W81XWH‐
15‐1‐0509, MD Anderson Cancer Center moonshot program, and
the Department of Genitourinary oncology, MDACC.
All Authors declare no potential conflict of interest with this
The data that support the findings of this study are available from
the corresponding author upon reasonable request.
Hirak S. Basu http://orcid.org/0000-0001-7733-8008
1. Tindall DJ. PCa: Biochemistry. Molecular Biology and Genetics. New
York: Springer; 2013.
2. PCa: diagnosis and management. NICE Guideline Updates Team (UK).
London: National Institute for Health and Care Excellence (UK);
May 2019.
3. Logothetis CJ, Aparicio A, Koinis F, Corn PG, Efstathiou E. PCa: Quo
Vadis? Eur Urol. 2019. pii: S0302‐2838(19)30521‐4 [Epub ahead of
4. Culig Z. Distinguishing indolent from aggressive PCa. Recent Results
Cancer Res. 2014;202:141‐147.
5. Vander Heiden MG, Cantley LC, Thompson CB. Understanding the
Warburg effect: the metabolic requirements of cell proliferation.
Science. 324(5930):1029‐1033.
6. Fraum TJ, Ludwig DR, Kim EH, et al. tracers: essentials for the ur￾ologist. Can J Urol. 2018;25(4):9371‐9383.
7. Jadvar H. Is there use for FDG‐PET in PCa? Semin Nucl Med. 2016;
8. Zong WX, Rabinowitz JD, White E. Mitochondria and Cancer. Mol
Cell. 2016;61(5):667‐676.
9. DeBerardinis RJ, Chandel NS. Fundamentals of cancer metabolism.
Sci Adv. 2016;2(5):e1600200.
10. Gottlieb E, Armour SM, Harris MH, Thompson CB. Mitochondrial
membrane potential regulates matrix configuration and cytochrome
c release during apoptosis. Cell Death Differ. 2003;10:709‐717.
11. Lin CS, Lee HT, Lee SY, et al. High mitochondrial DNA copy number
and bioenergetic function are associated with tumor invasion of
esophageal squamous cell carcinoma cell lines. Int J Mol Sci. 2012;
12. Caneba CA1, Bellance N, Yang L, Pabst L, Nagrath D. Pyruvate
uptake is increased in highly invasive ovarian cancer cells under
anoikis conditions for anaplerosis, mitochondrial function, and mi￾gration. Am J Physiol Endocrinol Metab. 2012;303(8):E1036‐E1052.
13. Gómez‐Cebrián N, Rojas‐Benedicto A, Albors‐Vaquer A, López‐
Guerrero JA, Pineda‐Lucena A, Puchades‐Carrasco L. Metabolomics
contributions to the discovery of PCa biomarkers. Metabolites. 2019;
9(3) PMID: 30857149.
14. Bewersdorf JP, Stahl M, Zeidan AM. Are we witnessing the start of a
therapeutic revolution in acute myeloid leukemia? Leuk Lymphoma.
15. Hoerner CR, Chen VJ, Fan AC. The 'Achilles Heel' of metabolism in
renal cell carcinoma: glutaminase inhibition as a rational treatment
strategy. Kidney Cancer. 2019;3(1):15‐29.
16. Fortunato S, Bononi G, Granchi C, Minutolo F. An update on patents
covering agents that interfere with the cancer glycolytic cascade.
ChemMedChem. 2018;13(21):2251‐2265.
17. Molina JR, Sun Y, Protopopova M, et al. An inhibitor of oxidative CB-839
phosphorylation exploits cancer vulnerability. Nat Med. 2018;24(7):
18. Basu HS, Thompson TA, Church DR, et al. A small molecule
polyamine oxidase inhibitor blocks androgen‐induced oxidative
stress and delays PCa progression in the transgenic adeno￾carcinoma of the mouse prostate model. Cancer Res. 2009;
19. Mukhopadhyay P, Rajesh M, Yoshihiro K, Haskó G, Pacher P.
Simple quantitative detection of mitochondrial superoxide pro￾duction in live cells. Biochem Biophys Res Commun. 2007;358(1):
20. Miller MC, Robinson PS, Wagner C, O’Shannessy DJ. The Parsortix™
cell separation system—a versatile liquid biopsy platform. Cytometry
A. 2018;93(12):1234‐1239.
21. Fujii T, Reuben JM, Huo L, et al. Androgen receptor expression on
circulating tumor cells in metastatic breast cancer. PLOS One. 2017;
22. TeSlaa T, Teitell MA. Techniques to monitor glycolysis. Methods
Enzymol. 2014;542:91‐114.
23. TeSlaa T, Teitell M. Conceptual background and bioenergetic/mi￾tochondrial aspects of oncometabolism. Methods in Enzymology.
Cambridge: Academic Press; 2014.
24. Wrobel K1, Claudio E, Segade F, Ramos S, Lazo PS. Measurement of
cytotoxicity by propidium iodide staining of target cell DNA. Ap￾plication to the quantification of murine TNF‐alpha. J Immunol
Methods. 1996;189(2):243‐249.
810 | BASU ET AL.
25. Scher HI, Beer TM, Higano CS, et al., PCa Foundation/Department
of Defense PCa Clinical Trials Consortium. Antitumour activity of
MDV3100 in castration‐resistant PCa: a phase 1‐2 study. Lancet.
26. Hoerner CR, Chen VJ, Fan AC. The ‘achilles heel’ of metabolism in
renal cell carcinoma: glutaminase inhibition as a rational treatment
strategy. Kidney Cancer. 2019;3(1):15‐29.
How to cite this article: Basu HS, Wilganowski N, Robertson S,
et al. Prostate cancer cells survive anti‐androgen and
mitochondrial metabolic inhibitors by modulating glycolysis and
mitochondrial metabolic activities. The Prostate. 2021;81:
799‐811. https://doi.org/10.1002/pros.24146
BASU ET AL. | 811