Thapsigargin resistance in human prostate cancer cells


  • John P. O'Neill BS,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    Search for more papers by this author
    • The first 2 authors contributed equally to this work.

  • Chidambaram Natesa Velalar PhD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    Search for more papers by this author
    • The first 2 authors contributed equally to this work.

  • Dong Ik Lee PhD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Bin Zhang MD, PhD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Takeo Nakanishi PhD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    2. Division of Medical Oncology, University of Maryland School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Yao Tang MD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Florin Selaru MD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    Search for more papers by this author
  • Douglas Ross MD, PhD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    2. Division of Medical Oncology, University of Maryland School of Medicine, Baltimore, Maryland
    3. Baltimore Veterans Affairs Medical Center, Baltimore, Maryland
    Search for more papers by this author
  • Stephen J. Meltzer MD,

    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    2. Division of Gastroenterology, University of Maryland School of Medicine, Baltimore, Maryland
    3. Baltimore Veterans Affairs Medical Center, Baltimore, Maryland
    Search for more papers by this author
  • Arif Hussain MD

    Corresponding author
    1. Greenebaum Cancer Center, University of Maryland School of Medicine, Baltimore, Maryland
    2. Division of Medical Oncology, University of Maryland School of Medicine, Baltimore, Maryland
    3. Baltimore Veterans Affairs Medical Center, Baltimore, Maryland
    • Division of Hematology and Oncology, University of Maryland Cancer Center, BRB RM: 9-041, 655 W. Baltimore St., Baltimore, MD 21201
    Search for more papers by this author
    • Fax: (410) 328-0805



Thapsigargin (TG) is a potent inhibitor of sarcoplasmic/endoplasmic reticulum Ca2+ ATPases (SERCAs). TG-based prodrugs are being developed for the treatment of prostate cancer (PC). To develop optimal TG-based therapeutics it is important to understand the mechanisms of resistance to TG that may potentially occur in cancer cells.


DU145/TG and PC3/TG cells were derived from human PC DU145 and PC3 cells, respectively, by incremental exposure to TG. Growth assays, Western blot analyses, cDNA microarrays, semiquantitative and real-time polymerase chain reaction (PCR), Northern blot analyses, and immunohistochemistry were used to study these cells.


DU145/TG cells are 1100-fold and PC3/TG cells are 1350-fold resistant to TG. Although expression of both SERCA and p-glycoprotein can mediate TG resistance in hamster cells, neither is modulated in DU145/TG cells. In contrast, in PC3/TG cells, SERCA, and not p-glycoprotein, is significantly overexpressed but cannot by itself account for the 1350-fold resistance to TG in these cells. Several genes not previously identified to be altered by TG selection are modulated in DU145/TG and PC3/TG cells. Furthermore, the spectrum of genes modulated in DU145/TG cells are distinct from that in PC3/TG cells, even though both cells are of prostate origin and share the same TG-resistant phenotype.


PC cells can adapt to SERCA inhibition by TG. However, they demonstrate cell type-specific plasticity with respect to gene expression upon TG selection. Further, previously not described mechanisms of resistance appear to be recruited in the TG-resistant PC cells, which provide a novel model to study mechanisms of resistance and adaptation in PC on TG-mediated dysregulation of Ca2+ homeostasis. Cancer 2006. © 2006 American Cancer Society.

Approximately 30,000 men die from advanced prostate cancer (PC) each year in the U.S.1 Chemotherapy is primarily palliative in metastatic hormone-resistant PC and, at best, has had only a modest effect on overall survival.2, 3 Thus, there is a dire need to identify new targets and treatment strategies in PC. A potentially interesting target in PC is the sarcoplasmic/endoplasmic reticulum Ca2+ATPase (SERCA).4 This enzyme plays a key role in regulating Ca2+ homeostasis within cells.5 Inhibition of SERCA disrupts cellular Ca2+ metabolism and triggers cell death by several pathways in a number of cancer cell lines, including PC.6–11

To our knowledge, the most potent and specific inhibitor of SERCA identified to date is thapsigargin (TG).12 Because SERCA is a housekeeping enzyme, its inhibition by TG may lead to significant toxicities. By limiting delivery of TG-based drugs to the tumor sites of interest, such toxicities might be minimized. The ability of PC cells to secrete the serine protease prostate-specific antigen (PSA) is being exploited to selectively target SERCA in PC cells in preclinical models.4 For instance, PSA-mediated processing and activation of TG-based inactive prodrugs can allow relatively selective targeting of SERCA in the PSA- producing PC cells, resulting in their toxic death.4, 13, 14

The inability to effectively treat advanced PC is due, in part, to its significant resistance to chemotherapy drugs. Hence, it becomes relevant to develop and study models of resistance to drugs being evaluated as potential therapeutics in this disease. Consequently, using DU145 and PC3 cells as 2 examples of androgen-independent PC (AIPC), cell lines highly resistant to TG, designated DU145/TG and PC3/TG, were established to study some of these drug resistance mechanisms in PC.

In this report we demonstrate that, in contrast to TG-resistant hamster lung fibroblasts (DC-3F cells) and smooth muscle cells (DDT1-MF2), in which overexpression of the multidrug transporter p-glycoprotein (PGP) and/or SERCA occurs and contributes to the resistant phenotype, neither gene is modulated in the DU145/TG cells.15–17 In PC3/TG cells, SERCA protein is increased approximately 30-fold compared with parental PC3 cells, whereas no change in PGP content occurs. However, because TG inhibits SERCA stoichiometrically, the 30-fold increase in SERCA alone cannot account for the 1350-fold resistance of PC3/TG cells to TG. Altered expression of several other classes of genes occurs as the DU145 and PC3 cells adapt to TG inhibition. Although DU145/TG and PC3/TG cells show the same common phenotype of resistance to TG, distinct sets of genes appear to be modulated in the 2 cell lines, at least among the genes represented on the custom microarrays used in our experiments. The modulated genes are involved in a number of processes ranging from Ca2+ metabolism to growth control; the genes may potentially contribute to the survival of the TG-resistant cells but do not necessarily mediate resistance to TG. Taken together, these studies demonstrate the complex and relatively global effects of TG selection on gene expression in human PC cells, including identification of genes not previously described to be modulated by TG treatment and selection. Furthermore, recruitment of previously uncharacterized, novel pathway(s) leading to TG resistance occur in the PC cells, providing a model to further study mechanisms of adaptation and resistance when the cancer cells are subjected to TG-mediated cellular stress.


Cell Lines

DU145 and PC3 cells were obtained from the American TypeCulture Collection (Rockville, MD) and grown in D-MEM/F-12 medium supplemented with 5% fetal bovine serum in 5% carbon dioxide at 37 °C. These cells were exposed to sequentially increasing concentrations of TG ranging from 0.01 to 1 or 2 μM over the course of several months. The final cell lines were designated DU145/TG (maintained in 1 μM TG) and PC3/TG (maintained in 2 μM TG). In addition, other cells, designated DU145/TGR, were derived from DU145/TG cells by culturing the latter continuously in the absence of TG for 16 weeks.

MTT Assays

MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazo lium bromide) assays to assess cell viability were performed as previously described.18 Cells were plated at 1.5 × 103 cells/well in 96-well plates and incubated overnight in 5% carbon dioxide at 37°C. Twenty-four hours later different concentrations of TG were added to the wells except for controls (where no TG was added). Forty-eight hours after the addition of TG, cell viability was determined by adding MTT and measuring absorbance at 595 nM in a multiwell plate reader. Data were analyzed using SigmaPlot (Systat Software, Point Richmond, CA).

Microarray Experiments

Microarrays were constructed by the University of Maryland Greenebaum Cancer Center core facility (Baltimore, MD) using 8064 cDNAs.19 Total RNA was reverse-transcribed (RT) into cDNA. The cDNA was transcribed in vitro, resulting in amplified RNA (aRNA). The aRNA was subjected to another RT reaction in the presence of Cyanine 3 (Cy 3) or Cyanine 5 (Cy 5) label. Cy3-labeled cDNA probes were made from 3 μg of DU145 aRNA and Cy5-labeled cDNA probes were made from 6 μg of DU145/TG aRNA (to compensate for the reduced binding affinity of Cy5 relative to Cy3), as described.19 Ten independently prepared Cy3-labeled and Cy5-labeled probes were hybridized to separate microarrays. Fluorescent array images were simultaneously collected for both Cy3 and Cy5 at 532 and 635 nanometer (nm) wavelengths, respectively, via a GenePix 4000B fluorescent scanner (Axon Instruments, Foster City, CA). Data acquisition and analysis were performed using GenePix Pro 3.0 microarray software. To normalize for the effect of nonspecific fluorescence, the background intensity was subtracted from the feature intensity before ratio calculations were performed. To quantify changes in gene expression accurately, the mean, median, and standard deviation of the pixel intensities for each raw image was computed with the GenePix Pro 3.0 software. The median of ratios is the median of the pixel-by-pixel ratios of pixel intensities that have had the median background intensity subtracted. The average of median ratios from the 10 independent hybridizations of cDNA was used for comparative analyses.

Real-Time RT Polymerase Chain Reaction

A fluorescein PCR detection system (LightCycler; Roche Applied Science, Indianapolis, IN) was used for analysis via real-time quantitative RT-polymerase chain reaction (PCR).20 Two hundred nanograms (ng) of total RNA from a given sample was reverse-transcribed and amplified using the SYBR Green I RNA Amplification Kit (Roche Applied Science) with gene- specific primers. β-Actin was used as an internal standard for each sample tested. The LightCycler software plots relative fluorescence (measured at 530 nm) during exponential amplification (after correction for background fluorescence) as a function of PCR cycle number and determines the crossing points (CPs) for each RT-PCR reaction. Relative gene expression in the sample was determined using the comparative delta-delta CP method (ΔΔCP) with β-actin as an internal control.21 Optimal and identical real-time amplification efficiencies of the target genes and reference gene (β-actin) were presumed as 2.21 The experiments were repeated at least 2 independent times for each gene analyzed by real-time PCR. The primer sequences were: SERCA2b 5′→3′ CGCGGATCCAATGCCCTCAACAGCTTG, 3′→5′ CGCGGATCCTYANGACCAARAACATRTC; PGP 5′→3′ ATGAAGTTGAATTAGAAAATGCAG, 3′→5′ GGAAAC TGGAGGTATACTTTCATC. The primer sequences for follistatin (FS), TACST1, STC2, and β-actin are listed below.


Total RNA was isolated using Trizol Reagent (Life Technologies, Carlsbad, CA) according to the manufacturer's instructions. The primer sequence and annealing temperature (TA) for some of the up-regulated genes that were identified through microarray analysis were: FS (616 basepairs [bp]) 5′→3′ GTGGAT GATTTTCAACGGG (56 °C), 3′→5′ CAGGCTCATCCGACT TAC (56°C); tumor-associated Ca2+-stimulated transducer 1 (TACST1) (408 bp) 5′→3′ CTTTGTGAATAATAATCGTC (52 °C), 3′→5′ GATCCAGTTGATAACGCG (54 °C); stanniocalcin 2 (STC2) (673 bp) 5′→3′ CTTGCTGCTGCACGAACC (58 °C), 3′→5′ GGTACGAGGATAACGCGG (56 °C); and β-actin (225 bp) 5′→3′ GCTATCCAGGCTGTGCTATC (62 °C), 3′→5′ TGTCACGCACGATTTCC (52 °C). The PCR reactions were carried out in a programmable thermal controller (PTC-100, MJ Research, Waltham, MA).

Northern Blot Analysis

Total RNA (10 μg) was electrophoresed in 1.2% agarose gels containing 0.66 M formaldehyde and transferred to nitrocellulose membranes. The probes for the genes of interest were labeled with [32P]dCTP using the Random Primers DNA labeling kit (Promega, Madison, WI). Hybridization and washing were as described previously.17

Western Blot Analysis

Western blot analyses were performed as described previously.15 Uterine sarcoma cells were used as positive control for PGP. Membranes were probed with the following: SERCA2 IID8F6 primary antibody (Affinity Bioreagents, Golden, CO), PGP-specific primary antibody (Santa Cruz Biotechnology, Santa Cruz, CA), ArpC3 primary antibody (Imgenex, San Diego, CA), and STC2 (K-17) and STC2 (N-18) primary antibodies (Santa Cruz Biotechnology).


Cell suspensions were cytospun onto slides (1 × 104 cells/slide) and fixed in acetone/methanol (1:1) for 15 minutes. Endogenous peroxidases were quenched with 3% hydrogen peroxide (10 minutes at room temperature). Blocking was performed in 2% bovine serum albumin (BSA), 0.2% Tween-20 in phosphate- buffered saline (PBST) for 30 minutes. Mouse antihuman FS primary antibody (R&D Systems, Minneapolis, MN) was applied at 1:100 dilution for 1 hour. The slides were washed in PBST and incubated in EnVision+ dual link-HRP solution (DakoCytomation, Carpinteria, CA) for 20 minutes. Color was developed with DAB (DakoCytomation) and counterstaining was achieved with hematoxylin (Sigma, St. Louis, MO).


Growth Inhibition Assays and Expression of SERCA and PGP

We used short-term MTT assays to determine the effects of TG on wildtype (wt) and TG-selected PC cells. In this assay, the tetrazolium salt (MTT) distinguishes living cells from dead cells; only live cells take up and metabolize MTT to a formazan product that can be measured spectrophotometrically.18 Although MTT-based short-term assays have proven to be very useful in assessing both cell proliferation and cell toxicity, such short-term analysis is, to an extent, dependent on the rate of cell death. By contrast, clonogenic assays are not affected by rate of cell death and take into account total cell kill, including delayed cell killing (apoptosis or otherwise). Differences in survival between 2 cell lines noted in a short-term assay may therefore not be apparent in a clonogenic assay; this may be particularly true for nonhematopoietic cells.22

Survival of wt and TG-selected PC cells in increasing concentrations of TG, as assessed by MTT assays, is shown in Figure 1. Within the context of the limitations imposed by the MTT assay, it is apparent that survival of wt cells in TG is different than for TG-selected cells 48 hours after treatment with TG. Comparison of EC50 values at this timepoint demonstrates that DU145/TG cells are 1100-fold resistant to TG relative to DU145 cells (Fig. 1A). Interestingly, DU145/TG cells that have been taken off TG for 16 weeks (designated DU145/TGR) continue to be highly resistant to TG (Fig. 1A), suggesting a stable change with respect to the TG-resistant phenotype has occurred so that the DU145/TGR cells maintain high levels of resistance despite being off TG selection for months. The other AIPC cell line, PC3/TG, selected for resistance to TG is approximately 1350-fold resistant to the drug relative to PC3 cells after 48 hours of exposure to TG (Fig. 1B). Thus, the wt and TG-selected cells have distinct phenotypes with respect to their relative sensitivity to TG, as determined by the MTT assay.

Figure 1.

Relative sensitivity of wild-type and thapsigargin (TG)-selected cells to TG. Percentages of surviving cells after incubation in different concentrations of TG for 48 hours are shown relative to controls (which were not treated with TG). Data depict the mean with standard deviation of 3 independent experiments.

Previously, we showed that TG-resistant hamster lung fibroblasts and smooth muscle cells overexpress SERCA2 and PGP, which may account, in part, for their resistance to TG.15–17 Interestingly, neither SERCA2 (Fig. 2A and 2B) nor PGP (Fig. 3A and 3C) are overexpressed in DU145/TG cells, suggesting that other mechanisms of TG resistance are likely to be operative in these cells. In PC3 cells, in contrast, SERCA2 is highly overexpressed (Fig. 2C and 2D). However, as with DU145/TG cells, expression of PGP is not altered in PC3/TG cells when compared with the parental PC3 cells (Fig. 3B), suggesting that PGP does not contribute to TG resistance in the PC3/TG cells. Although a 30-fold increase in SERCA protein occurs in PC3/TG cells (Fig. 2D), given that TG inhibits SERCA stoichiometrically, the enhanced expression of SERCA by itself cannot account for the 1350-fold resistance to TG observed in these cells. Taken together, these data suggest that other, previously not identified mechanisms of TG-resistance are also recruited in PC3/TG cells.

Figure 2.

Sarcoplasmic/endoplasmic reticulum Ca2+ ATPases (SERCA2) expression in DU145- and PC3-derived cells. (A) Real-time polymerase chain reaction (PCR): DU145, DU145/TG (thapsigargin). (B) Western blot analysis: DU145, DU145/TG. (C) Northern blot analysis: PC3, PC3/TG. (D) Western blot analysis: PC3, PC3/TG. kDa: kilodaltons.

Figure 3.

P-glycoprotein (PGP) expression in prostate cancer cells. (A) Real-time polymerase chain reaction (PCR): DU145, DU145/TG (thapsigargin). (B) Real-time PCR: PC3, PC3/TG. (C) Western blot analysis: DU145, DU145/TG. kDa: kilodaltons.

Gene Expression Profiles in the TG-resistant Cells

We studied the gene expression profile of DU145, DU145/TG, PC3, and PC3/TG cells using custom cDNA microarrays. The majority of the genes (approximately 75%) show similar expression between the wt and resistant cells. An example of the distribution profile of up-regulated and down-regulated genes for DU145/TG cells relative to DU145 cells is shown as a scatterplot in Figure 4, which demonstrates that a majority of the genes are located close to the “line of identity,” (i.e., demonstrate similar expression between the 2 cell lines). Of the 8064 genes on the array, approximately 560 (i.e., 7% of the total DNA elements) are expressed sequencing tags (ESTs), whereas <2% are of unknown function. Greater than 16% of the 8064 genes show differences of ≥2-fold between the wt and TG-resistant cells. Representative examples of genes whose expression has increased by ≥2.5-fold are listed in Table 1 (PC3/TG) and Table 2 (DU145/TG); these genes were grouped into several functional classes using the OntoExpress software.23

Figure 4.

Scatterplot of expressed genes in DU145 and DU145/TG (thapsigargin) cells. Expression of 8064 genes is depicted. The mean fluorescent intensities for each cDNA on the microarray upon hybridization with Cy3-labeled DU145 cDNA and Cy5-labeled DU145/TG cDNA probes are shown on the X and Y axes, respectively. The “line of identity” reflects essentially equal expression in DU145 and DU145/TG cells, with points lying above this line reflecting overexpressed genes and points below this line reflecting down-regulated genes in the resistant cells relative to the DU145 cells. Examples of some of the up-regulated genes: 1-FS, 2-TACST1, 3-STC2.

Table 1. Genes Up-Regulated in PC3/TG Cells
Functional CategoryAccession No.Fold Difference
Calcium-binding proteins
 ATPase, Ca2+ transporting (SERCA2)H8535510.5
 S-100 calcium-binding protein A1H499834.6
Cytoskeletal structure/mobility
 Actin related protein (Arp), 2/3 complexH7396110
Immune function
 Microseminoprotein betaAI6587277
Kinase and transferase activities
 Mevalonate kinaseH082058
 Matrix metalloproteinase 15AA4433005.2
 Aldehyde dehydrogenase 6AA4552352.7
 ATPase, Na+/K+ transporting, beta 3AA4892752.6
 DNA polymerase deltaAA5042042.5
Signal transduction
 Phosphodiesterase 2AAA8762194.7
 Protein phosphatase 1BC0409744.6
 Death-associated proteinAA4590512.9
 Interleukin 13 receptor, alpha 2R527962.6
 Cytochrome c-1AA8652652.5
 ADP-ribosylation factor 4T713162.5
Cell growth
 Insulin-like growth factor binding proteinBC0578063.8
 Cell division cycle 2BC0145633.5
 Proliferating cell nuclear antigenAA4502653.1
 Cyclin E2AA5209992.6
 Cyclin-dependent kinase 2AI6530172.5
 Endothelin 1H110032.5
DNA-binding proteins
 Uracil-DNA glycosylaseH151112.5
Table 2. Genes Up-Regulated in DU145/TG Cells
Functional CategoryAccession No.Fold Difference
Adhesion/Ca2+ binding molecules
 Natural killer cell transcription 4AA4589654.2
 Mucin 1, transmembraneAA4880733.7
 Down syndrome candidate region 1AA6297073.2
 Regulator of G-protein signaling 2 (24 kDa)AI6756702.9
 Suppression of tumorigenicity 14AA4892463.8
 2′,5′-Oligoadenylate synthetase 1AA1467733.3
 Cell division cycle 25AR090632.8
 Diacylglycerol kinase, alpha (80 kDa)AA4569003
 Histidine triad nucleotide-binding proteinT575562.9
 Plasminogen activator, urokinaseAA2846682.6
Signaling/transcription factors
 Tumor-associated calcium signal transducer 1AI3408836
 Myxovirus (influenza) resistance 1AA4568864.4
 Nuclear receptor subfamily 2H688383.6
 p53-responsive gene 2AI3567093.5
 Bone marrow stromal cell antigen 2AA4853713.3
 Major histocompatibility complexAA4642463.3
 Interferon-stimulated protein (15 kDa)AA4060203
 Forkhead box D1AA0693722.8
 T-box 19AI6309802.6
Structural proteins
 Ladinin 1T977103
 Keratin 19AA4642502.7
 Ribosomal protein L44AA6693592.6
Transport proteins
 Lipocalin 2AA4011374.3
 Metallothionein 1EAA8723833.5
 Stanniocalcin 2AA6764083.2

Confirmation of Microarray Results

SERCA2 is among the genes represented on the custom arrays used in our experiments. It is highly overexpressed in PC3/TG cells compared with PC3 cells on microarray analysis (Table 1), consistent with the Northern blot analysis data in Figure 2C. Differential expression of SERCA2 is not found between DU145 and DU145/TG cells on microarray (not shown), which is also confirmed by real-time PCR (Fig. 2A). Interestingly, although ARPC3 is also highly up-regulated in PC3/TG cells on microarray (Table 1), this enhanced expression is not associated with an increase in ARPC3 protein (Fig. 5). SERCA2, in contrast, is highly up-regulated, both at the RNA (Fig. 2C) (Table 1) and protein levels (Fig. 2D), in PC3/TG cells.

Figure 5.

Western blot analysis. ARPC3 expression in PC3 and PC3/TG (thapsigargin) cells.

To validate the microarray results in DU145/TG cells, 3 of the most up-regulated genes (i.e., FS, tumor-associated calcium signaling transducer 1 [TACST1/Trop-2], and stanniocalcin 2 [STC2]) (Table 2) were analyzed further. As seen in Figure 6A, Northern blot analyses demonstrated that all 3 genes are overexpressed in the DU145/TG cells compared with DU145 cells. Expression of these genes in DU145 and DU145/TG cells was also determined by real-time PCR, which in general provides more accurate information with respect to gene expression. Relative expression of TACST1, FS, and STC2 are summarized in Figure 6B. Despite some discordance with the microarray and Northern blot analysis data in terms of fold increase, real-time PCR demonstrates that all these genes are overexpressed in DU145/TG cells. Approximately 6-fold (FS), 60-fold (TACST1), and 3-fold (STC2) differences in their expression levels is observed between DU145 and DU145/TG cells (Fig. 6B). The apparent discrepancy between the different assays point to the necessity for validating microarray data.

Figure 6.

Verification of genes up-regulated on microarray in DU145 and DU145/TG (thapsigargin) cells. (A) Northern blot analysis. Probes for the specific genes were generated using the appropriate primers. (B) Relative gene expression for follistatin (FS), TACST1, and STC2 based on real-time polymerase chain reaction (PCR).

Demonstrating altered expression of the above 3 genes at the protein level in DU145/TG cells has been more problematic. Antibodies for TACST1 are not commercially available, whereas the 2 antibodies for STC2 (STC2 K-17 clone and STC2 N-18 clone, Santa Cruz Biotechnology) tested in our model did not give any signal on Western blot analysis or immunohistochemistry (IHC) in either wt or resistant cells. Furthermore, although STC2 is coexpressed in estrogen receptor (ER)-positive breast cancer tissue,24 and is an estrogen target in ER-positive MCF-7 breast cancer cells,25 our attempt to use MCF-7 cells as positive control for STC2 also did not result in a detectable signal with either antibody. Thus, at present it is not clear whether the enhanced levels of TACST1 and STC2 RNA are associated with a corresponding increase in their respective protein levels in the DU145/TG cells. DU145/TG cells do appear to have increased levels of FS protein, as demonstrated on IHC (Fig. 7B and 7C). Although in other cancer cells FS has been immunolocalized to the cytoplasm,26 in DU145/TG cells enhanced immunoreactivity is observed around the cell membrane (Fig. 7B and 7C). A paradigm for this exists in oocytes where FS is immunolocalized not only in the ooplasm but also the zona pellucida.27 No differences in FS expression were found between PC3 and PC3/TG cells on microarray, and FS was also essentially undetectable in either cell line on IHC (Fig. 7E,F).

Figure 7.

Immunohistochemistry. Fixed cells were stained with (A-C, E, and F) mouse antihuman follistatin (FS) primary antibody or (D) mouse immunoglobulin (Ig) G (negative control). (A) DU145; (B-D) DU145/TG (thapsigargin); (E) PC3; (F) PC3/TG.

Although significant overexpression of FS, TACST1, and STC2 is observed in DU145 cells selected long-term with TG (i.e., DU145/TG cells), short-term exposure (24 hours) of wt cells to TG also induces these genes (Fig. 8, Lane 2). In contrast, in DU145/TGR cells, which have been taken off TG selection for 16 weeks but continue to be highly resistant to TG, expression of these genes returns to basal levels (Fig. 8, Lane 4). Thus, it is apparent that TG itself, rather than selection for TG resistance, can induce these genes because enhanced expression is observed in DU145/TG cells (which are continuously maintained in TG; Figs. 6 and 8, Lane 3) as well as DU145 cells treated short-term with TG, but not DU145/TGR cells (which are grown in the absence of TG). Further, although FS, TACST1, and STC2 may be involved in the adaptive response to SERCA pump inhibition in the DU145 model, they do not appear necessary in mediating TG resistance because their expression returns to basal levels, whereas resistance is maintained in the DU145/TGR cells.

Figure 8.

Reverse-transcriptase polymerase chain reaction (RT-PCR) analysis of DU145-derived cells. Twenty-five μL of RT-PCR reactions were run on 1% agarose gels. Lane 1, DU145; Lane 2, DU145 cells treated with 1 μM thapsigargin (TG) for 24 hours before RNA harvest; Lane 3, DU145/TG; Lane 4, DU145/TGR.


When subjected to certain forms of nutritional deprivation, or exposed to certain cytotoxic agents, cancer cells can mount a complex response to the cellular stress in which the endoplasmic reticulum (ER) plays a central role.28–31 This so-called “ER stress response,” in which specific genes are expressed, whereas others are repressed, can serve a protective function or lead to apoptosis, depending on the experimental conditions, cellular context, and the specific molecular pathways recruited.28–33 TG, by depleting Ca2+ from the ER via inhibition of SERCA, is a potent inducer of ER stress and apoptosis in a number of cancer cell lines, including PC. Depletion of ER Ca2+ can result in elevation of cytosolic Ca2+ via store-operated channels.34 This, in turn, can activate Ca2+, Mg2+-dependent endonucleases and calcineurin-dependent dephosphorylation of Bad, triggering apoptosis.35, 36 Recent studies suggest that TG can also up-regulate DR5, resulting in caspase 8 activation and recruitment of extrinsic or/and intrinsic pathways of apoptosis (the latter via caspase 8-mediated cleavage of Bid).10, 11, 33 Interestingly, GADD153/CHOP, expression of which is increased as part of the ER stress response to TG, can up-regulate DR5, providing a link between ER stress and apoptosis.33 Further, enhanced toxicity to other agents such as diindolylmethane is observed in cancer cells prestressed with TG.31

Given TG's potent effects on a number of pathways mediating apoptosis, and its ability to induce cell death in cancer cells with a low proliferative index, such as prostate cancer, a concerted effort is being made to develop TG-based drugs as potential therapeutics in this disease.4, 13, 14 Development of resistance to TG would compromise such a treatment strategy, and hence it becomes relevant to identify the cellular pathways that may potentially lead to TG resistance in PC. In this report we demonstrate that the androgen-independent DU145 and PC3 cells can in fact adapt to TG-mediated cellular stress and develop high levels of resistance to this drug (Fig. 1). Although enforced expression of Bcl2 in DU145 cells can alter the balance of other regulators of cell proliferation and cell death, and lead to TG resistance, altered expression of Bcl2 was not observed in either TG-resistant cell line (data not shown).37

Previously, we showed that in hamster lung fibroblasts and smooth muscle cells selected for resistance to TG, both SERCA and PGP become overexpressed.15–17 Because SERCA is the most specific target of TG, it would be advantageous for cells to overexpress this gene in an attempt to overcome TG inhibition, similar to what is often observed in cells selected for resistance to antimetabolites.38 Because TG may be a substrate for PGP,15 enhanced expression of PGP could also render cells resistant to TG by decreasing intracellular accumulation of TG. Neither of these 2 mechanisms, however, appears to be operative in the DU145/TG cells, because expression of both SERCA and PGP are similar between the wt and resistant cells (Figs. 2A and 2B, 3A and 3C), suggesting recruitment of other resistance mechanism(s). Interestingly, the DU145/TG cells maintain resistance despite being taken off TG for several months (Fig. 1A), suggesting that a stable genetic change mediating resistance has occurred. One possibility is that a specific gene (or set of genes) becomes overexpressed in the DU145/TG cells, leading to their decreased sensitivity to TG. Loss of a certain gene(s) that would normally cause sensitivity to TG in wt cells (for instance, gene(s) mediating apoptosis) could have also occurred during the selection process, resulting in TG resistance in DU145/TG cells.

To identify some of the underlying genetic changes, we employed cDNA microarrays. Although we employed microarrays that represented a relatively small set of expressed cDNAs, several interesting features emerged from the analysis. Using the OntoExpress software,23 the genes were grouped into several functional but not mutually exclusive categories. Several genes fell into more than 1 category but for clarity they are listed under only 1 functional group (Tables 1 and 2). Although microarrays provide useful information with respect to global gene expression patterns, they are limited in scope regarding quantitation, and thus the need for validation studies. Furthermore, enhanced expression of a particular gene may not necessarily lead to a corresponding increase in the respective protein; therefore, the biological significance of altered gene expression also needs to be evaluated in this context.

Among the genes that appeared to be most highly expressed in DU145/TG cells on microarray (Table 2), we further confirmed expression of FS, TACST1, and STC2 by Northern blot analysis and real-time PCR (Fig. 6). TACST1 and STC2 have been implicated in cellular Ca2+ metabolism, and may potentially play a role in the adaptive response to TG-mediated dysregulation of Ca2+ homeostasis in DU145 cells. However, their biological relevance is difficult to ascertain in this model because, due to the limitations already noted, we were unable to determine whether an increase in TACST1 or STC2 protein also occurred in the DU145/TG cells. We did observe increased immunoreactivity with anti-FS antibody in the DU145/TG cells (Fig. 7). By interacting with members of the TGFβ family, particularly activins and bone morphogenetic proteins, FS modulate/inhibit their actions.39 In DU145 cells, others studies have shown that FS blocks the growth inhibitory effects of activin A.40 At present we do not know whether activins are also modulated by TG treatment or TG selection, but an increase in FS may represent 1 adaptive response by the DU145/TG cells that allows them to maintain continued growth in TG. Although these genes may be part of an adaptive response to TG, they are unlikely to play a significant role in mediating TG resistance because DU145/TGR cells continue to be highly resistant in the absence of enhanced FS, TACST1, or STC2 expression.

The significant overexpression of SERCA in PC3/TG cells noted on microarray correlated with increased SERCA protein in these cells, whereas this was not the case with ARPC3, which is highly expressed in PC3/TG cells at the RNA level (Table 1) but not the protein level (Fig. 5). Although an increase in SERCA can lead to relative resistance to TG, the known stoichiometric relation between SERCA content and TG inhibition12 suggests additional factor(s) are also likely involved to account for the high levels of TG resistance in the PC3/TG cells. Of the over 8000 genes analyzed by microarray, those overexpressed by a factor of 2-fold or greater (e.g., SERCA, ARPC3, mevalonate kinase [MK], β-microseminoprotein [β-MSP], and membrane type-2 matrix metalloproteinase [MMP15]) in PC3/TG cells (Table 1) are essentially not up-regulated in DU145/TG cells. Similarly, the highly up-regulated genes in DU145/TG cells (e.g., FS, TACST1, STC2) are not so in PC3/TG cells. The cellular functions modulated by FS, TACST1, and STC2 are quite distinct from that modulated by MK, β-MSP, MMP15, and ARPC3. Even though PC3 and DU145 cells are both of prostate origin and androgen-independent, it is not readily apparent why they have such a diverse response with respect to their gene expression profiles when selected for a common phenotype, i.e., TG resistance. Whereas some of the genes may be necessary for the PC cells to survive in TG, modulation of other genes may simply represent an epiphenomena of the selection process and not required for cell survival in TG.

In summary, the results of the current study demonstrated that highly TG-sensitive PC cells can be selected for resistance to this drug. Genes other than PGP or SERCA are modulated during selection for TG resistance. Further, a cell type-specific pleiotropic response with respect to gene expression occurs in the 2 independently selected PC cell lines. Although other target genes likely exist and could potentially be identified if a whole genome microarray were used, our analysis reveals the multiple and different effects of TG on the genome of PC cells as they adapt to the inhibitory actions of TG. Conceivably, a subset of the altered genes may contribute to the selected cells' ability to maintain growth and division in TG. However, the specific gene(s) mediating TG resistance in this model have yet to be identified and studies are currently in progress to clarify these pathways of resistance in the PC cells.