Oncoprotein GT198 is a direct target of taxol

Taxol (paclitaxel) is one of the most successful chemotherapeutic drugs in the treatment of human cancer. It has recently been questioned whether the mechanism of action in mitotic arrest, which is ubiquitously present in all cells, is sufficient to explain the tumor specificity, clinical efficacy, and side effects of taxol. In this report, we have identified a new protein target of taxol as GT198 (gene symbol PSMC3IP, also known as Hop2). GT198 is an oncoprotein and a DNA repair factor involved in human common solid tumors. The GT198 gene carries germline mutations in breast and ovarian cancer families and recurrent somatic mutations in tumor microenvironment. Mutant GT198 was identified in pericyte stem cells on capillary blood vessels inducing tumor angiogenesis. GT198 is a DNA-binding protein dimer, also stimulates DNA repair, regulates meiosis, participates in homologous DNA recombination, and activates nuclear receptor-mediated gene expression. Here we show that taxol directly binds to the DNA-binding domain of GT198 in vitro. Taxol serves as an allosteric inhibitor to block DNA binding to GT198 with an IC50 of 8.6 nM. Labeled taxol colocalizes with GT198 in interphase nuclei of cultured cells. Decreased GT198 expression desensitizes taxolinduced cell death, and taxol inhibits GT198 nuclear foci formation during DNA repair. Together, these results demonstrate that GT198 is a previously unrecognized direct protein target of taxol. The finding of taxol target as an oncoprotein GT198 in common solid tumors provides a rationale for the clinical efficacy of taxol. We anticipate that GT198 may serve as a clinical predictive marker of taxol efficacy as well as a new drug target for future anti-cancer therapy.


INTRODUCTION
Taxol is a natural compound originally isolated from the Pacific yew tree Taxus brevifolia, and is one of the most successful chemotherapeutic drugs in the treatment of human cancer (1,2). Taxol improves overall survival rates in patients with breast, ovarian, lung cancer, and other types of solid tumor (3,4). The mechanism of action of taxol has previously been linked to mitotic arrest (5,6), apoptosis, and more recently to defects in mitotic spindle assembly and chromosome segregation in primary breast tumors (7). However, unanswered questions remain that the mechanism of action in mitotic arrest, which is ubiquitously present in all cells, may not be sufficient to explain the tumor specificity, clinical efficacy, and side effects of taxol (8)(9)(10)(11).
We and others have previously identified a breast and ovarian cancer initiating gene GT198 (gene symbol PSMC3IP, also known as Hop2 or TBPIP) (12)(13)(14)(15). Here, we report that taxol directly interacts with GT198 and inhibits its DNA-binding activity. GT198 is a DNA-binding protein that functions as a transcriptional coactivator stimulating nuclear receptor-mediated gene activation (16,17). GT198 also acts as a crucial DNA repair factor by participating in homologous DNA recombination (12,13,18). Germline mutations in GT198 have previously been identified in familial and early-onset breast and ovarian cancer (14,19), as well as in familial ovarian disease (20). Somatic mutations in GT198 are prevalent and recurrent in sporadic breast, ovarian, prostate, uterine, and fallopian tube cancers, where they lead to constitutive activation of GT198 in transcription (14,15). In sporadic breast and ovarian cancers, GT198 is mutated in tumor microenvironment. The GT198 mutant cells include pericyte stem cells and the decedent vascular smooth muscle cell lineage. These include myoepithelial cells and adipocytes in breast cancer (21), hormone-producing luteinized theca cells in ovarian cancer (22), myofibroblasts in prostate and bladder cancers (23), pericytes in skin and brain tumors as well as in multiple common solid tumors (24).
The expression of GT198 in normal tissues (16,22), is intriguingly correlated with the spectrum of taxol-induced clinical side effects (4,25). We then did a preliminary screen of commonly used chemotherapy drugs using GT198 as a protein target, and found that GT198 is inhibited by a number of oncology drugs including doxorubicin, etoposide, and taxol (23). In this study, we sought to detailed analysis of GT198 as a direct target of taxol.

Taxol binds to the DNA-binding domain of GT198
To test for a direct interaction in vitro, recombinant His-tagged GT198 proteins were analyzed for binding kinetics on 96-well plates using a commercially available Oregon greenlabeled taxol (Figure 1a). A previously reported biotinylated taxol (26), is also analyzed for labeled taxol interaction with GT198 in cells. We first verified the biotin moiety on biotinylated taxol by streptavidin binding (Figure 1b). The integrity of both Oregon green-and biotin-labeled taxols were further evaluated by cell survival dose-response curves, which showed that both the green-and biotin-labeled taxols retain cytotoxicity albeit with much reduced efficacy (EC50=67.1 nM and 70.7 nM, respectively) compared to unlabeled taxol (EC50=3.3 nM) (Figure 1c). The addition of carbon C7 side chains reduced but did not abolish the binding of labeled taxol to cellular targets, allowing for subsequent testing of the labeled taxols. The true in vitro binding affinity of taxol can only be measured through competition of DNA binding to GT198 by unlabeled taxol.
First, we measured the binding affinity of green-taxol to His-tagged GT198 and its fragment proteins (Figure 1d), and obtained an EC50 of 245 nM for the wildtype GT198 (Figure 1e). The binding site was mapped using truncated GT198 protein fragments (Figure 1d). GT198 is a small homodimeric DNA-binding protein. Each monomer contains 217 amino acids including an N terminal domain, a leucine zipper (LZ) dimerization domain (16), a DNA-binding domain (DBD) capable of binding to either single-or double-stranded DNA (12,15), and a C terminal auto-regulatory domain (15). Only the GT198 DBD domain was required for the interaction of greentaxol (Figure 1e), whereas the N and C terminal domains were dispensable (Figure 1e and Supplementary Figure S1a). We also measured the DNA binding affinity of GT198 using a biotinlabeled oligonucleotide and obtained an EC50 value of 52 nM (Figure 1e). This high binding affinity of DNA to GT198 permitted the sensitive detection in subsequent competition assays (Figure 2).
Furthermore, we confirmed that the GT198 DBD was required for binding to biotin-labeled taxol using GST-GT198 fusion proteins (Supplementary Figure S2a-b), as well as endogenous GT198 protein immunoprecipitated from HeLa nuclear extracts (EC50=176 nM) (Supplementary Figure S1b). Although the binding affinities of biotin-and green-labeled taxols to GT198 may be comparable (Figure 1c), biotin detection is less sensitive than fluorescence and we were not able to assay using small quantity of His-tagged GT198 proteins, but only able to measure its interaction using GST fusions or immuneprecipitated cellular GT198. Together from the data above, we conclude that labeled taxol binds to the DBD of GT198.

Taxol inhibits the DNA binding of GT198
Binding of unlabeled taxol to GT198 was then assessed indirectly using competition binding assays, in which increasing concentrations of unlabeled taxol compete for binding of labeled taxol. The results suggest that unlabeled taxol binds to GT198 at a much higher binding affinity since it competes for green-taxol with an IC50 of 8.8 nM using full-length GT198 (Figure 2a). This folddifference in binding affinity between unlabeled and labeled taxol is comparable to the observed fold-difference in their cytotoxicities (Figure 1c). Consistently, DNA binding to GT198 was competed off using unlabeled taxol with an IC50 of 8.6 nM (Figure 2b), suggesting potent inhibition of GT198 DNA binding activity by unlabeled taxol. Double-reciprocal plots revealed competitive binding of GT198 between labeled and unlabeled taxols (Figure 2c). The constant Vmax with increased Km in the presence of taxol suggests that labeled green-taxol binds to the same site on GT198 as does unlabeled taxol. In contrast, inhibition of DNA binding by taxol was allosteric or noncompetitive (Figure 2d). The reduced Vmax with constant Km for DNA binding in the presence of taxol suggests that taxol and DNA bind to different sites on the GT198 DBD (Figure 2d). The inhibition of DNA binding by taxol may result from a protein conformational change (modeled in Figure 2e). Of note, a rather incomplete inhibition of DNA binding was observed even using high concentrations of taxol (Figure 2b), further suggesting an allosteric mechanism of inhibition. Potentially, even incomplete inhibition could have sufficient impact at the functional level, since the binding affinity is rather high. Taken together, these results suggest that taxol directly inhibits the DNA binding of GT198 in vitro at single digit nanomolar concentrations, within a concentration range effective for cytotoxicity (Figure 1c), and is also consistent with observed drug efficacy in clinical chemotherapy.

Taxol colocalizes with GT198 in interphase nuclei
We further examined the cellular distribution patterns of GT198 and taxol. In GT198 transfected HeLa cells, biotinylated taxol colocalizes with overexpressed GT198 and its DBD fragment but not with its N terminal domain (Figure 3a). As previously reported, GT198 is nuclear and its fragments are cytoplasmic (15). The colocalization is consistent with the observed binding of taxol to the DBD of GT198. The punctate pattern of endogenous GT198 coincided more precisely than overexpressed GT198 to the biotin-labeled taxol staining in interphase nuclei (Figure 3b). In mitotic phase cells, endogenous GT198 and taxol staining were significantly increased relative to the interphase suggesting that the GT198 expression is correlated with the cell cycle (Figure 3b). Similarly, green-taxol and GT198 nuclear staining patterns closely coincided in interphase nuclei, although additional cytoplasmic staining was observed with green-taxol (Figure 3c). This cytoplasmic signal was not due to microtubule binding as detected using anti--tubulin, since tubulin staining did not coincide with green-taxol in either interphase or mitotic phase cells (Figure 3c). Notably, the nuclear but not the cytoplasmic staining of green-taxol was blocked by unlabeled taxol (Supplementary Figure S3), suggesting this cytoplasmic signal is perhaps non-specific. Thus, taxol colocalizes specifically with GT198 but not microtubules in HeLa cells.

Decreased GT198 desensitizes Taxol-induced cell death and taxol inhibits GT198 foci formation
To analyze the functional consequence of taxol binding to GT198, we reduced GT198 expression using transient transfection of siRNA in HeLa cells. GT198 siRNA desensitized, whereas GT198 overexpression sensitized cells to the cytotoxic effects of taxol (Figure 4a). Similarly, in mouse P19 stem cells, the stably expressed DBD fragment of GT198 (aa 126-217) desensitized cytotoxicity of taxol relative to stable expression of GT198 (Figure 4b), consistent with the notion that DBD acts as a dominant negative of wild type GT198 (15). In particular, a significant portion of DBDtransfected stable cells remained viable even after exposure to high concentrations of taxol ( Figure  4b), suggesting that reduced GT198 activity may render taxol less effective. GT198 is also a DNA repair factor, and forms nuclear foci upon irradiation (Figure 4c). Strikingly, taxol treatment results in GT198 cytoplasmic translocation and disrupts GT198 foci formation in irradiated cells (Figure 4c). This finding is consistent with the observation that translocation of DNA repair proteins to the cytoplasm occurs following taxol treatment (10). Since GT198 cytoplasmic translocation is characteristic of wild type GT198 inhibition (15,22), this result suggests that taxol inhibits GT198 activity, including its function in DNA repair.

DISCUSSION
In summary, our results provide the first evidence that a DNA repair factor and oncoprotein GT198 is a direct target of taxol. From a clinical perspective, the study provides anticipated rationale for the clinical side effects of taxol. In human cancer patients, taxol-induced lymphocyte toxicity (27), neutropenia (4), testicular shrinkage (28), and ovarian damage (29), are potentially correlated with the endogenous GT198 protein expression specifically in testis, ovary, spleen, and thymus (16,22). The acute taxol toxicity in rodents was also found in ovary (30), testis, hematopoietic, and lymphoid systems (31). In contrast, the low toxicity of taxol in major organs is correlated with undetectable GT198 protein level therein (16). Moreover, GT198 stimulates glucocorticoid receptor-mediated transcription (16). The glucocorticoid receptor ligand dexamethasone is an effective premedication prior to taxol (25), potentially due to a compensation of taxol-induced GT198 loss in normal cells.
In human ovarian cancer, for example, GT198 is overexpressed in the luteinized theca cells and causes steroid hormone overproduction in the tumor microenvironment (22). Mutant luteinized theca cells provide hormone stimuli to recapitulate epithelial tumor cells. GT198 inhibition by taxol would be predicted to suppress steroid hormone for cancer initiation thereby result in high clinical efficacy. Consistently, in human breast cancer, GT198 is expressed in mutant pericyte stem cells, the decedent fibroblasts, adipocytes, and myoepithelial cells (21). Taxol treatment of human breast cancer would inhibit the cancer-inducing cell lineages in tumor microenvironment and lead to observed clinical efficacy in the treatment. In addition, GT198 regulates androgen receptor-mediated transcription (16), and taxane chemotherapy that would suppress GT198 has already been shown to impair androgen receptormediated transcription in prostate cancer (9,32). At the cellular level, taxol inhibits microtubule trafficking of a number of DNA repair proteins (10). Inhibited GT198 within the DNA repair complex could potentially lead to observed defects in microtubule trafficking (33). Finally, GT198 is highly expressed in cultured cancer cell lines (Supplementary Figure S4) (16), and in xenografted mouse tumor (24), which could underlie the historic success in taxol identification using GT198-positive cell lines or xenografted tumors (1,34). Together these observations strongly support that GT198 is a direct protein target of taxol. In the future, GT198 may be a useful prognostic marker in predicting the clinical efficacy of taxol or taxane in human cancer treatment. GT198 can also serve as a therapeutic drug target to identify new drugs.    (Supplementary Fig. 3). Secondary antibodies were Alexa 595 (red) and Alexa 488 (green)-conjugated anti-mouse or anti-rabbit antibodies. Cells were counterstained with DAPI in blue.

Green-taxol binding and competition assays
The binding of Oregon green-labeled taxol (Molecular Probe, Oregon Green ® 488 Taxol, P-22310) to GT198 was detected through analyzing green fluorescence. Black 96-well plates (Thermo Scientific, #7805) were coated to dry overnight at 37 o C with 400 ng/well of recombinant His-tagged GT198 proteins together with 5 g/well of purified BSA (NEB) in a volume of 50 l. BSA alone was included as a control for background. No subtraction of background was needed due to very low background. Duplicate wells were used for each experimental point (n = 2). The GT198-coated plates were blocked with 5% BSA in PBS with 0.1% Triton X-100 (TPBS) for 1 h. The binding was carried out in 100 l/well for 4 hrs to overnight at 4 o C using serial diluted green-taxol (0.91, 2.

DNA binding and competition assays
The binding of biotinylated DNA to GT198 was detected by chemiluminescence. The DNA binding to GT198 is non-sequence specific (15). A single-stranded 25-mer biotin-oligonucleotide [Biotin]-cctggggttgctgaggtcctggcag was used in the assay since it is sufficient to bind one GT198 dimer. White MicroLite TM 2+ 96-well plates (Thermo Scientific, #7572) were similarly coated with Histagged GT198 proteins and blocked with BSA as above. The binding in duplicates (n = 2) was carried out for 4 hrs to overnight at 4 o C using serial

Binding assays using GST-GT198 and immunoprecipitated GT198
In vitro binding assays using GST fusions were performed in duplicates by incubating GST-GT198 resins (10 l, 2 g) and biotinylated taxol (2 M) in the binding buffer containing 1 mM ZnSO4 for overnight at 4 o C. For immunoprecipitated GT198, experiments in duplicates were performed by incubating agarose beads-captured immune complex and biotinylated taxol (2 M) in the binding buffer for overnight at 4 o C. Washed beads were incubated with streptavidin-peroxidase (1 U/ml) for 1 h at 4 o C and further washed three times before transfer to 96-well plates to measure. Bound taxol was detected by chemiluminescence with 50 l/well ECL detection reagents in a Dynex MLX luminometer.

Kinetic data analysis
The binding EC50 values are the 50% of maximal binding concentrations of green-taxol or DNA. The inhibition IC50 values are the taxol concentrations of 50% inhibition of green-taxol or DNA binding. In cell viability assays, EC50 values are the taxol concentrations that reduce total cell viability by 50%. The EC50 and IC50 values were calculated through Kaleidagraph software (Synergy Software) using nonlinear regression sigmoidal dose-response curve fit. The equation for calculation is y = m1 + (m2 -m1)/(1 + (x /m3) m4 ), where m1 is the minimum, m2 is the maximum, m3 is the EC50 or IC50 value calculated by Kaleidagraph, and m4 is the slope at midpoint of the curve. However, data points were drawn with better fitted slopes using nonlinear regression sigmoidal curve fit provided by Kaleidagraph software: y = m1 + m2/(1 + exp(-(x-m3)/m4)), where exp = е x , although this equation was less tolerant in calculation and was not used to calculate values. The analysis of Vmax and Km in competitive or non-competitive binding assays was carried out in double-reciprocal plots, using means of four high concentration data points, omitting points at lower concentrations with reciprocal values off-scale.

Cell viability assays
HeLa cells (ATCC, CCL-2) were maintained in DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 0.1 µg/µl streptomycin. Cells were incubated in 5% CO2 at 37 o C. For transient transfections, GT198 plasmid (0.4 g/well) or GT198 siRNA (100 nM) was transfected into HeLa cells in 48-well plates for 4 hrs before the cell viability assay. Mouse embryonal carcinoma P19 cells (ATCC, CRL-1825) were maintained in -MEM supplemented with 2.5% fetal bovine and 7.5% bovine calf serum, and antibiotics above. Stably transfected P19 cells expressing GT198 and its DBD-containing fragment 126-217 were selected using 400 g/ml of G418 for three weeks. Positive clones were confirmed by PCR and Western blot analyses. The cell viability assays were carried out in duplicate wells for each condition in 48-well plates after taxol treatment for 72 hrs at various concentrations (0.032, 0.16, 0.8, 4, 20, 100, 500, 2500 nM). The tetrazolium compound MTS reagent can be reduced by cells into a colored formazan product (CellTiter 96® AQueous One Solution Cell Proliferation Assay, Promega, G3580). The MTS reagent was diluted in phenol red-free medium to minimize the background and incubated with cells at 200 µl/well for 1 h at 37 o C. The reacted colored medium was transferred to flat-bottom clear 96-well plates and the absorbance at 490 nm was determined using a Tecan Safire microplate reader. P values were determined by two-tailed paired t test.

Northern dot blot analysis
The Northern dot blots containing cDNAs from primary human tumors (T) with paired adjacent normal tissues (N) and cancer cell lines were obtained from Clontech (Cancer Profiling Array, #7840-1). The blots contained normalized cDNA isolated from tumors and the corresponding adjacent normal tissues from individual cancer patients. However, adjacent normal tissues often possess early molecular changes undetectable by pathology standards. The GT198 probe was prepared by random-primed 32 P-DNA synthesis using human GT198 full-length cDNA as template. Northern hybridization was performed according to manufacturer's protocol before visualized by autoradiography.

Statistics
Statistical analyses were performed using Prism software. In cell viability assays, statistical significance of differences between groups was determined by two-tailed paired t test. In GST binding assays, P values were determined by twotailed unpaired t test. In binding assays with zinc, P values were determined by linear regression. Each binding assay was repeated for at least three times in the laboratory. Data represent mean ± s.e.m of duplicate experiments (n = 2) as indicated. Duplicates rather than triplicates were used in each experimental data point due to highly consistent in vitro experimental conditions that yield variations are sufficiently small.