Identification of an Interaction between the TPα and TPβ Isoforms of the Human Thromboxane A2 Receptor with Protein Kinase C-related Kinase (PRK) 1

In humans, thromboxane (TX) A2 signals through the TPα and TPβ isoforms of the TXA2 receptor or TP. Here, the RhoA effector protein kinase C-related kinase (PRK) 1 was identified as an interactant of both TPα and ΤPβ involving common and unique sequences within their respective C-terminal (C)-tail domains and the kinase domain of PRK1 (PRK1640–942). Although the interaction with PRK1 is constitutive, agonist activation of TPα/TPβ did not regulate the complex per se but enhanced PRK1 activation leading to phosphorylation of its general substrate histone H1 in vitro. Altered PRK1 and TP expression and signaling are increasingly implicated in certain neoplasms, particularly in androgen-associated prostate carcinomas. Agonist activation of TPα/TPβ led to phosphorylation of histone H3 at Thr11 (H3 Thr11), a previously recognized specific marker of androgen-induced chromatin remodeling, in the prostate LNCaP and PC-3 cell lines but not in primary vascular smooth muscle or endothelial cells. Moreover, this effect was augmented by dihydrotestosterone in androgen-responsive LNCaP but not in nonresponsive PC-3 cells. Furthermore, PRK1 was confirmed to constitutively interact with TPα/TPβ in both LNCaP and PC-3 cells, and targeted disruption of PRK1 impaired TPα/TPβ-mediated H3 Thr11 phosphorylation in, and cell migration of, both prostate cell types. Collectively, considering the role of TXA2 as a potent mediator of RhoA signaling, the identification of PRK1 as a bona fide interactant of TPα/TPβ, and leading to H3 Thr11 phosphorylation to regulate cell migration, has broad functional significance such as within the vasculature and in neoplasms in which both PRK1 and the TPs are increasingly implicated.

The local control of hemostasis and vascular tone is a complex process involving platelets, the endothelium and vascular smooth muscle (VSM), 3 various soluble coagulation factors, vasodilator/vasoconstrictor autocoids, and other diverse mediators (1). Agents such as thromboxane (TX) A 2 that signal through G protein-coupled receptors (GPCRs) to promote platelet aggregation or VSM contraction can typically co-couple to G q -mediated phospholipase C␤ and to G q /G 12 -RhoA activation leading to both Ca 2ϩ -dependent and Ca 2ϩ -independent responses to promote myosin light chain phosphorylation and actin polymerization (2,3). Because of the recognized role of the RhoA effector Rho kinase 1/2 as a therapeutic target in hypertension, many studies have focused on TXA 2 regulation of Rho kinase signaling, but few have investigated its regulation of other RhoA effectors such as protein kinase C-related kinases (PRK) 1 and 2 (4). The PRKs, also referred to as protein kinase novel (PKN), constitute a subfamily of serine/threonine kinases comprising PRK1/PKN␣, PRK2/PKN␥, and PKN␤ (5). They contain three highly conserved regions, including an N-terminal regulatory domain spanning three repeats of charged amino acids (antiparallel coiled-coil fold or ACC finger domains) with leucine zipper-like sequences, a centrally located arachidonic acid-sensitive C2-like auto-inhibitory domain, and a C-terminal catalytic domain (5). Binding of RhoA-GTP within the regulatory domain induces a conformational change in PRK1, for example, leading to release and full activation of its kinase domain by 3-phosphoinositide-dependent protein kinase-1 (PDK1) phosphorylation (5). PRK1 activity has been implicated in numerous cellular processes (5), as well as being widely implicated in androgen-associated prostate cancers and ovarian serous carcinomas (6 -9).
As stated, the prostanoid TXA 2 plays an essential role within the vasculature acting as a potent regulator of platelet activation status, VSM contraction, proliferation, and migration and is widely implicated in a number of cardiovascular disorders, including thrombosis, systemic-and pregnancy-induced hypertension, vessel remodeling, and atherosclerotic progression (10,11). In addition, recent evidence suggests aberrant TXA 2 signaling, and TXA 2 receptor (TP) expression is associated with certain cancers, in particular prostate cancer with direct correlation of tumor Gleason score and pathologic state (12)(13)(14), through as yet unknown mechanisms but potentially through its capacity to regulate RhoA signaling and cell migratory responses. In humans and other primates, TXA 2 actually signals through two distinct TXA 2 receptor isoforms termed TP␣ and TP␤, which differ exclusively in their intracellular C-terminal (C-tail) domains (15). Although the physiologic requirements for two receptors for TXA 2 in humans is not fully understood, they greatly increase the complexity of TXA 2 signaling, and there is increasing evidence that TP␣ and TP␤ have distinct (patho)physiologic roles. For example, TP␣ and TP␤ display distinct patterns of expression, being under the transcriptional control of different promoters (16 -22). Although they show similar coupling to G q /phospholipase C␤ and to G q /G 12 regulation of RhoA activation, their primary effectors, they differentially regulate other secondary effectors, including adenylyl cyclase (15,23). TP␣ and TP␤ also undergo entirely distinct mechanisms of both agonist-induced (homologous) desensitization (24,25) and heterologous desensitization or cross-talk/regulation by other signaling systems such as prostacyclin and nitric oxide (23, 26 -29). Such differences in desensitization of TP␣ and TP␤ signaling occur because of differences in their phosphorylation, occurring mainly within their unique C-tail domains, by the second messenger-regulated protein kinases, including by protein kinase (PK) C, PKA, PKG, and/or by GPCR-regulated kinases 1/2 (23)(24)(25)(26)(27)29).
Hence, exploiting critical differences in their intracellular C-tail domains, the initial aim of this study was to screen for possible novel interacting partners of either TP␣ and/or TP␤ and thereafter to explore the functional relevance of any interaction(s) identified. We report the identification of a novel specific interaction between both TP␣ and TP␤ with the RhoA effector PRK1. This interaction occurs between the kinase domain of PRK1 with common and unique regions within the intracellular C-tail domain of TP␣ and TP␤. Bearing in mind the critical role of TXA 2 as a regulator of RhoA signaling coupled with the identification of its effector PRK1 as a direct interactant of TP␣ and TP␤, the discovery of this novel interaction is likely to have substantial (patho)physiologic implications for processes in which both TXA 2 and RhoA/PRK1 are involved.
Investigation of PC-3 and LNCaP Cell Migration-Boyden chamber assays were used to assess the effects of TP activation on prostate cancer cell migration. Briefly, prior to migration assays, PC-3 and LNCaP cells were plated in 10-cm dishes in growth media containing charcoal-stripped FBS either in the presence or absence of DHT (1 nM) such that they were Ն90% confluent 24 -36 h post-seeding. Alternatively, cells were seeded such that they were Ն50% confluent after 24 -36 h and transfected with siRNA (siRNA PRK1 or siRNA Control ) followed by incubation for a further 48 h. Cells were routinely serumstarved for 24 h before harvesting and QCM TM 24-well colorimetric cell migration assay (Millipore; ECM508) was performed as per the manufacturer's instructions. Briefly, cells were initially rinsed in PBS. While aliquots of cells were retained for immunoblot analysis of endogenous PRK1 and HDJ-2 protein expression, for migration assays, cells (1.0 ϫ 10 6 cells in 300 l of serum-free media either in the presence or absence of 1 nM DHT) were placed in the top chamber and treated with either vehicle (0.01% EtOH) or U46619 (1 M). After the cells settled (ϳ1 h), 500 l of complete growth media in the presence or absence of DHT (1 nM) and containing either vehicle (0.01% DMSO) or U46619 (1 M) was added to the bottom chamber. Cell migration was assessed after 4 h at 37°C. The cells remaining in the top chamber were removed by cotton swabs, and the migrated cells were stained with a crystal violet dye (Millipore, catalog no. 90144) and washed with H 2 O. The cells were then extracted using an extraction buffer (Millipore, catalog no. 90145), and the A 560 nm was measured. Migration was expressed as percentage of basal cell migration. All experiments were performed at least in duplicate, and each experiment was repeated at least two times.
Confocal Microscopy-LNCaP and PC-3 cells were seeded at 2 ϫ 10 5 cells in 2 ml of RPMI 1640 medium, 10% FBS in 6-well plates pre-coated with poly-L-lysine (0.001% for 1 h), and grown at 37°C for 48 h. Thereafter, cells were transiently transfected with pEGFP-C1.PRK1 (31) encoding GFP-tagged PRK1 (GFP: PRK1; 2 g using Lipofectamine LTX (Invitrogen) for LNCaP cells and 1 g using Effectene (Qiagen) for PC-3 cells). Some 48 h post-transfection, cells were stimulated with 1 M U41669 for 0 -240 min prior to fixation in 3.7% paraformaldehyde. Cells were permeabilized with 0.2% Triton X-100 for 10 min on ice, followed by staining with 4Ј,6-diamidino-2-phenylindole (DAPI; 1 g/ml in H 2 O). Images were captured at ϫ63 magnification using Carl Zeiss Laser Scanning System LSM510 and Zeiss LSM imaging software for acquiring multichannel images with filters appropriate for enhanced DAPI and GFP. Images presented in the figures are representative of 10 independent fields.
Data Analyses-Statistical analyses of differences were carried out using the unpaired Student's t test, two-way or one-way ANOVA followed by post hoc Dunnett's multiple comparison t tests, as indicated, throughout employing GraphPad Prism, version 4.00 package. p values of less than or equal to 0.05 were considered to indicate a statistically significant difference. To investigate overall differences between time-dependent, U46619-induced H3 Thr 11 phosphorylation in the absence or presence of DHT, nonlinear regression (R 2 ) and F-test analyses were carried out, where p values of less than or equal to 0.05 were considered to indicate a statistically significant difference. As relevant, single, double, triple, and quadruple symbols signify p Ն 0.05, Ն 0.01, Ն 0.001 and Ն 0.0001, respectively, for post hoc Dunnett's multiple comparison t test analyses.
Thereafter, it was sought to further localize the region(s) within the unique C-tail domain of TP␤ that mediates its interaction with PRK1 640 -942 (Fig. 1B). Although each of the bait and prey strains mated successfully (Fig. 1B, DDO), only diploids expressing either the common proximal region (residues 312-328) or a more distal C-terminal region (residues 366 -392) showed an interaction with PRK1 640 -942 (Fig. 1B, QDO). Taken together, these data identify two regions of importance for PRK1 binding within the C-tail domains of TP␣ and TP␤, namely the proximal common region (residues 312-328) and a more distal C-terminal region unique to TP␤ (residues 366 -392) (Fig. 1, A and B). TP␤ subfragments lacking both of these regions and the TP␣ fragment that lacks the common region (residues 312-328) were unable to interact with PRK1 640 -942 .
The role of individual residues within the common region of TP␣/TP␤ in contributing to their interaction with PRK1 was also investigated (supplemental Fig. 1). This region is of particular interest as much of it (residues 316 -323) is predicted to be organized into the ␣-H8, a structural feature of many GPCRs (32,33). Ala-scanning SDM and Y2H-based screening established that mutation of Leu 316 , Arg 318 , and Leu 323 abolished the interaction of TP␣ 312-343 with PRK1, although mutation of all other residues only impaired it (supplemental Fig. 1). In contrast, whereas mutation of each of the ␣-H8 residues impaired the interaction of TP␤ 312-407 with PRK1 640 -942 , none of those mutations per se completely disrupted the interaction (supplemental Fig. 1).
Confirmation of the Association between TP␣/TP␤ and PRK1 Using Glutathione S-Transferase Pulldown Assays and Co-immunoprecipitations in Mammalian Cells-To further examine the interaction between PRK1 and TP␣/TP␤, in vitro GST pulldown assays were performed. Consistent with the Y2H studies, PRK1 640 -942 was found to bind GST.TP␤ 312-407 , GST. TP␤ 329 -407 , and GST.TP␣ 312-343 but not to GST.TP␣ 329 -343 or GST, despite near equivalent expression of all proteins ( Fig.  2A). The ability of full-length PRK1 (residues 1-942, herein referred to as PRK1), expressed as a FLAG epitope-tagged form in mammalian HEK 293 cells, to bind GST.TP␣ 312-343 and GST.TP␤ 312-407 was also investigated. PRK1 showed a specific interaction with GST.TP␣ 312-343 and GST.TP␤ 312-407 but not with GST.hIP 299 -386 , a GST protein expressing the C-tail domain of the human prostacyclin receptor (Fig. 2B).
The ability of PRK1 to associate with HA-tagged forms of TP␣ or TP␤ stably expressed in the previously characterized HEK.TP␣ and HEK.TP␤ cell lines (27) was examined through co-immunoprecipitations. PRK1 was detected in the anti-HA immunoprecipitates from HEK.TP␣ and HEK.TP␤ cells but not in corresponding immunoprecipitates from the HEK.␤-Gal cells, encoding HA-tagged ␤-Gal, acting as an additional/alternative control (Fig. 2C). Thereafter, the effect of short term agonist exposure (10 min) on the interaction between TP␣/ TP␤ and PRK1 was investigated in response to stimulation of cells with the selective TXA 2 mimetic U46619. As described previously, in the absence of agonist, PRK1 was specifically detected in the anti-HA immunoprecipitates from HEK.TP␣ and HEK.TP␤ cells and not from HEK.␤-Gal cells. Stimulation of cells with U46619 did not lead to an alteration in the amount of PRK1 associated with TP␣ or TP␤ (Fig. 2D). In all cases, failure to detect PRK1 in the anti-HA immunoprecipitates from the control HEK.␤-Gal cells was not because of lack of PRK1 expression (Fig. 2, C and D, lower panels) or failure of the immunoprecipitation per se in those cells (Fig. 2, C and D, middle panels).
C-terminal Domain of PRK1 Is Important for Association with TP␣/TP␤-The N-terminal region of PRK1, spanning residues 1-511, contains a number of important regulatory domains that influence its C-terminal catalytic kinase domain (Fig. 3A). The three ACC, or HR1, domains are involved in RhoA-GTP binding (34,35), although the C2-like domain participates in AA-induced kinase activation and lies adjacent to the kinase domain itself (36). Hence, here the interaction between TP␣ and TP␤ with PRK1 or with three of its subfragments comprising either the kinase domain (PRK1 561-942 ), the ACC/HR1 domains alone (PRK1 1-357 ), or ACC/HR1 domains along with the AA-binding site (PRK1 1-594 ) were examined through co-immunoprecipitations, where possible interactions with the unrelated HA-tagged ␤ 2 AR served as an additional independent control. As described previously, PRK1 associated strongly and specifically with TP␣ and TP␤ as evidenced by its detection in immunoprecipitates from HEK.TP␣ and HEK.TP␤ but not from control HEK.␤ 2 AR cells (Fig. 3B, upper  panels). The kinase only PRK1 561-942 subfragment was also abundantly expressed in TP␣ and TP␤, but not ␤ 2 AR, immunoprecipitates and at levels comparable with that of PRK1. Conversely, lower levels of PRK1 1-594 were detected in the TP␤ and, to an even lesser extent, in the TP␣ immunoprecipitates. Furthermore, removal of the AA-binding site, as in the PRK 1-357 subfragment, completely abolished binding to TP␣ and substantially reduced binding to TP␤. The observed differences in immunoprecipitation of PRK1 or its subfragments were not due to variations in their expression levels in the cell lines used or in the efficiency of the immunoprecipitations per se (Fig. 3B,  lower and middle panels, respectively). The identity of the faint, nonspecific immunoreactive protein present in the anti-HA precipitates from HEK.␤ 2 AR cells is unknown (Fig. 3B, upper panels). Additional evidence of the involvement of the kinase domain of PRK1 in interacting with the TPs was established whereby ectopic overexpression of PRK1 561-942 , but not PRK1 1-594 or PRK1 1-357 , competed and thereby partially impaired the association of PRK1 with TP␤ and, to a lesser extent, with TP␣ (Fig. 3C). Collectively, these data confirm a constitutive physical interaction between TP␣/TP␤ and PRK1 in mammalian cells and point to a critical role for the C-terminal region of PRK1, comprising its AA-binding C2-like domain and its catalytic kinase domain, in that interaction.
RhoA Association with the TP-PRK1 Complex-PRK1 is a downstream effector of the GTPase RhoA (37). Moreover, both TP␣ and TP␤ modulate RhoA signaling (23). Hence, it was sought to further examine the interaction of TP␣ and TP␤ with PRK1 and to establish whether endogenous RhoA may associate with the complex. Initially, HEK.TP␣, HEK.TP␤, and the control HEK.␤-Gal cell lines, co-transfected with pCMVTag2b: PRK1, were stimulated with U46619 for 0 -120 min, and the presence of PRK1 and endogenous RhoA in the anti-HA immunoprecipitates was examined. In the absence of agonist, PRK1 was detected in the anti-HA TP␣ and TP␤, but not ␤-Gal, immunoprecipitates, although neither short term nor more prolonged U46619 stimulation led to measurable changes in the amount of PRK1 associated with the TPs (supplemental Fig.  2, A-D). Moreover, RhoA was detected in the anti-HA TP␣ and TP␤ immunoprecipitates and at levels that were unaffected by U46619 stimulation (supplemental Fig. 2, A-D). Conversely, RhoA was not detected in HA-␤-Gal immunoprecipitates despite its efficient immunoprecipitation and equivalent expression of endogenous RhoA in all cell lines (supplemental Fig. 2, A and B).
To exclude the possibility that the observed associations of PRK1 or RhoA with TP␣ or TP␤ may be an artifact of overexpression of PRK1, the ability of endogenous PRK1 and endogenous RhoA to associate with the TP isoforms was examined. As with the overexpressed PRK1, similar levels of association between TP␣/TP␤ and PRK1 were observed in both nonstimulated and U46619-stimulated cells (Fig. 4, A and B; supplemental Fig. 2, E and F). Moreover, consistent with previous findings, RhoA was associated with the TP␣/TP␤-PRK1 immunoprecipitates in ternary complexes that were not affected by U46619 stimulation (Fig. 4, A and B; supplemental Fig. 2, E and F). Collectively, these data confirm a physical interaction between TP␣/TP␤ and PRK1 in a constitutive ternary complex with the PRK1 effector RhoA in mammalian cells that is independent of TP agonist activation.

TP␣ and TP␤ Are Not Phosphorylation Targets of PRK1-
Several functional targets of PRK1 have been identified, ranging from its ligand-dependent interaction and activation of the nuclear androgen receptor (8) to its interaction and phosphor-ylation of vimentin and glial fibrillary acidic protein, to inhibit filament formation (38). Moreover, both TP␣ and TP␤ are recognized targets of PKC phosphorylation (24 -26, 28). Hence, in view of its many regulatory functions, it was sought to investigate whether PRK1 might target TP␣ and/or TP␤ by direct phosphorylation. To this end, the ability of purified preparations of PRK1 to phosphorylate GST fusion proteins encoding the intracellular loop domains (IC 1 -IC 3 ) and the C-tail domains of TP␣ or TP␤ were examined in vitro. The known PRK1 substrate histone H1 served as a control in the in vitro kinase assays (39). Although histone H1 was readily phosphorylated, none of the purified recombinant GST proteins encoding the respective intracellular subdomains of TP␣/TP␤ were phosphorylated in vitro by PRK1 (supplemental Fig. 3). Moreover, despite repeated attempts, whole cell phosphorylations in HEK.TP␣ or HEK.TP␤ cells established that PRK1 did not lead to phosphorylation of TP␣ or TP␤ either in the absence or presence of U46619 stimulation or following overexpression of PRK1 (data not shown).
PRK1 Activity Is Increased in Response to U46619 Stimulation-As stated, it has been previously established that agonist activation of TP␣ and TP␤ leads to robust activation of RhoA (23). Moreover, it has been established here that PRK1 constitutively interacts with both TP␣ and TP␤ in a complex that also contains RhoA. Hence, we sought to investigate here whether endogenous PRK1 associated with the immune complexes is functionally active and whether its activity may respond to agonist stimulation of TP␣ and/or TP␤ either expressed in the respective clonal HEK.TP␣ and HEK.TP␤ cell lines or endogenously in vascular endothelial EA.hy926 cells or primary (1°) HUVECs and in 1°h.CoASM cells. Initially, to confirm the specificity of PRK1 activation, HEK.TP␣, HEK.TP␤, and control HEK.␤-Gal cells were stimulated with U46619 for 10 min, and following immunoprecipitation with anti-PRK1 antibody, precipitates were used as source of kinase for the in vitro phosphorylation assays using histone H1 as the specific PRK1 substrate. Results show substantial phosphorylation of histone H1 when PRK1 was immunoprecipitated from agonist-stimulated HEK.TP␣ and HEK.TP␤ cells (Fig. 5, A and B). Conversely, relative levels of histone H1 phosphorylation were significantly lower when PRK1 was precipitated from the control HEK.␤-Gal cells despite near equivalent immunoprecipitation of PRK1 in all cell types (Fig. 5, A and B). Moreover, time course assays established that, in the absence of agonist, PRK1 resulted in histone H1 phosphorylation, which was transiently increased in the presence of U46619, with maximal responses occurring at 10 -30 min for both TP␣ and TP␤ (Fig. 5, C and D). The precise physiologic impact of the observed increase in PRK1 activity in response to TP agonist activation, as determined by analysis of histone H1 phosphorylation, is unclear.
PRK1 immunoprecipitated from EA.hy926, 1°h.CoASM cells, or 1°HUVECs also led to increases in histone H1 phosphorylation in vitro in the response to agonist simulation (supplemental Fig. 4, A and B; and data not shown) and were lower than those observed in HEK.TP␣ and HEK.TP␤ cells. Such differences are most likely reflective of the levels of endogenous TP␣ and TP␤ expressed in the former cell types (10 -20 fmol/mg cell protein (40)) relative to those levels in the clonal HEK.TP␣ or HEK.TP␤ cell lines (ϳ2 pmol/mg of cell protein). Furthermore, PRK1 was not detected in immunoprecipitates from either 1°HUVECs or 1°hCoASMCs using an affinitypurified antibody (No. 168) directed to both TP␣ and TP␤ (supplemental Fig. 4, C-E).
As a means of verifying that the increased PRK1-induced phosphorylation of histone H1 observed in the presence of U46619 was due to activation of TP␣ and/or TP␤, PRK1 present in the anti-HA TP␣ and anti-HA TP␤ immune complexes from HEK.TP␣ and HEK.TP␤ cells was also used as a source of kinase in the in vitro assays. Consistent with previous data, PRK1 was efficiently co-immunoprecipitated with both TP␣ and TP␤ and not with the control ␤-Gal (Fig. 5, E and F, lower panels). Endogenous PRK1 present in the anti-TP␣ and anti-TP␤ precipitates resulted in efficient in vitro phosphorylation of histone H1, although, as expected, no phosphorylation was detected using the anti-HA immune complexes from HEK.␤-Gal cells (Fig. 5, E and F, upper left panels). Moreover, whereas PRK1 present in the anti-TP␣ and anti-TP␤ precipitates phosphorylated histone H1 in vitro in the absence of agonist, phosphorylation was increased following U46619 stimulation in both cases (Fig. 5, E and F, upper right panels). Hence, collectively, these data confirm that PRK1 associated in immune complexes with TP␣ and TP␤ is functionally active and that it undergoes enhanced activation in response to receptor stimulation.
H3 Thr 11 Phosphorylation in Prostate PC-3 and LNCaP Cells-PRK1 has been established to play a key role in the regulation of transcription by the nuclear androgen receptor (AR) through its specific phosphorylation of histone H3 at a critical Thr 11 residue (H3 Thr 11 ), thus initiating chromatin remodeling and potentiating androgen-induced gene expression, such as within the prostate (8,9). Furthermore, this occurs through androgeninduced association of PRK1 with the AR (8,9). Hence, because of the findings herein demonstrating a direct interaction between TP␣/TP␤ with PRK1, it was sought to explore whether U46619-mediated activation of PRK1 might actually induce phosphorylation of histone H3 at Thr 11 (H3 Thr 11 ) in 1°H UVECs and 1°h.CoASMCs, somewhat similar to that now established for the androgens (8,9). To this end, H3 Thr 11 phosphorylation was analyzed using a specific anti-phospho-H3 Thr 11 , where cells growth-arrested in metaphase with colcemid served as a positive control for H3 Thr 11 phosphorylation, and blots were co-screened with anti-histone H3 antibody to ensure uniform total histone H3 loading (9,41). Although colcemid induced robust phosphorylation, the level of H3 Thr 11 phosphorylation in response to U46619 in either 1°HUVECs and 1°h .CoASMCs was unaltered over the course of the treatment (0 -60 min; supplemental Fig. 4, F and G) suggesting that, in those cell types at least, TP␣/TP␤-induced PRK1 activation is not associated with H3 Thr 11 phosphorylation.
As stated previously, in addition to PRK1, increased TXA 2induced signaling and TP isoform(s) expression have been associated with human prostate cancer (12)(13)(14). Hence, in view of the direct interaction of PRK1 with TP␣ and TP␤ and of the by now established role of androgen-activated PRK1 in inducing H3 Thr 11 phosphorylation (8,9), it was sought to investigate whether U46619-mediated activation of TP/PRK1 signaling might induce H3 Thr 11 phosphorylation in the human prostate carcinoma PC-3 and LNCaP cell lines. Moreover, it was also sought to investigate whether TP-mediated PRK1 activation might affect androgen-induced responses by comparing the effect of U46619 on dihydrotestosterone (DHT)-induced H3 Thr 11 phosphorylation in the androgen-insensitive PC-3 relative to the androgen-sensitive human adenocarcinoma LNCaP cell lines, where colcemid-arrested cells served as a positive control for the assay in both cell lines (Fig. 6, A and B). In contrast to 1°HUVECs and 1°h.CoASMCs, stimulation of both PC-3 and LNCaP cell types with U46619 led to significant increases in H3 Thr 11 phosphorylation with maximal responses occurring at 30 -60 min in both cell lines (Fig. 6, A and B). In all cases, rescreening of the anti-phospho-H3 Thr 11 blots with anti-histone H3 itself confirmed that any differences in H3 Thr 11 phosphorylation in either PC-3 or LNCaP cells, e.g. in response to U46619, were not due to discrepancies in histone H3 levels (Fig. 6, A and B). Although treatment of PC-3 cells with DHT did not induce a significant increase in H3 Thr 11 phosphorylation per se (Fig. 6A), co-stimulation with U46619 led to similar and time-dependent increases in H3 Thr 11 phosphorylation but to levels that were not significantly different from those in the absence of DHT (Fig. 6A, F-test analysis, p ϭ   0.9039). Conversely, pretreatment of LNCaP cells with DHT led to a significant increase in H3 Thr 11 phosphorylation (Fig.  6B). Moreover, co-stimulation of LNCaP cells with U46619 in the presence of DHT led to robust, time-dependent increases in H3 Thr 11 phosphorylation relative to those levels in the presence of U46619 alone (Fig. 6B, F-test analysis, p Ͻ 0.0001), with maximal responses occurring at 30 -60 min post-U46619treatment. In point of fact, at 30 min of post-agonist stimulation, maximal levels of H3 Thr 11 phosphorylation in LNCaP  F) immunoprecipitates were used as a source of PRK1 to examine U46619-induced in vitro phosphorylation of histone H1 (10 g; 30°C for 30 min). Phosphorylated histone H1 was visualized by autoradiography, although all immunoprecipitates were immunoblotted (IB) with anti-PRK1 antibody (A-F, upper and lower panels, respectively). The relative positions of the 30-and 104-kDa molecular size markers are to the left of the panels. The bar charts represent mean percentage changes in phosphorylation relative to PRK1 expression in the anti-PRK1 immunoprecipitates, where basal levels in the absence of U46619 are assigned a value of 100%. The asterisks indicate that levels of phosphorylation were significantly increased in response to U46619 stimulation, relative to vehicle-treated cells, where ** signifies p Ն 0.01 for post hoc Dunnett's multiple comparison t test analysis. Data; n Ն 3.

Thromboxane Receptor Interaction with PRK1
APRIL 29, 2011 • VOLUME 286 • NUMBER 17 JOURNAL OF BIOLOGICAL CHEMISTRY 15449 cells were up to 3-fold greater in the presence of DHT plus U46619 relative to basal levels in the absence of either agent (Fig. 6B, two-way ANOVA, p Ͻ 0.0001). In addition, to investigate whether the increased H3 Thr 11 phosphorylation in the latter cell types is specific to the TPs or a more general phenomenon, both PC-3 and LNCaP cells were stimulated with 17-phenyl trinor PGE 2 or cicaprost, selective agonists for the related G q -coupled EP 1 subtype of the PGE 2 receptor or the G s -coupled prostacyclin receptor, respectively, and their effect on H3 Thr 11 phosphorylation was compared with that of the TP agonist U46619 (supplemental Fig. 5, A and B). Although U46619 resulted in substantial increases in H3 Thr 11 phosphorylation in both PC-3 and LNCaP cells, it was established that neither 17-phenyl trinor PGE 2 nor cicaprost did so in either cell line (supplemental Fig. 5, A and B). The precise physiologic impact of the observed TP-mediated PRK1 activation, as determined herein by analysis of H3 Thr 11 phosphorylation, is not fully evident at this time. However, given the serious nature of enhanced H3 Thr 11 phosphorylation as a key marker of androgen-dependent gene expression, any change in H3 Thr 11 phosphorylation in response to TP activation, possibly leading to enhanced AR-dependent gene expression, is likely to be of substantial significance, such as in the context of the enhanced TP expression associated with prostate cancer (12)(13)(14).
To investigate whether PRK1 actually interacts with the TP␣/TP␤ endogenously expressed in the PC-3 and LNCaP cells, immunoprecipitations were performed with an affinitypurified antibody (No. 168) directed to both TP isoforms. In the absence of agonist, PRK1 was detected in the anti-TP␣/TP␤ immunoprecipitates from both PC-3 and LNCaP cells but not in corresponding immunoprecipitates using the preimmune IgG (Fig. 7, A and B). Consistent with previous data in HEK.TP␣ and HEK.TP␤ cells (Fig. 4), stimulation of PC-3 or LNCaP cells with U46619, either in the presence or absence of prestimulation with DHT, did not alter levels of PRK1 associated with anti-TP␣-TP␤ immune complexes in either cell type (Fig. 7, A  and B). Noteworthy, because of the relatively low levels of TP␣/ TP␤ endogenously expressed in either PC-3 or LNCaP cells (ϳ120 fmol/10 6 cells (14)), secondary screening of the immunoprecipitates precluded detection of either receptor isoform (data not shown). However, in parallel experiments, the specificity of the affinity-purified antibody to immunoprecipitate both HA-tagged TP␣ and TP␤ from HEK.TP␣ and HEK.TP␤ cells was confirmed, although neither receptor isoform was present in immunoprecipitates using equivalent amounts of the preimmune IgG (supplemental Fig. 4E). Furthermore, through preliminary image analysis using the GFP-tagged form of PRK1 (GFP:PRK1), it was confirmed that PRK1 underwent transient  ; 30 min). In all cases, isolated histones were resolved by SDS-PAGE and immunoblotted (IB) with anti-phospho-H3 Thr 11 (upper panel) or anti-histone H3 (lower panel) antibody followed by chemiluminescence detection. The bar charts represent mean percentage changes in H3 Thr 11 phosphorylation relative to histone H3 levels and are expressed in arbitrary units (Ϯ S.E., n ϭ 3), where basal levels in vehicle-treated cells and in the absence of U46619 are assigned a value of 100%. The asterisks indicate that levels of H3 Thr 11 phosphorylation were significantly increased in response to U46619 stimulation, relative to vehicle-treated cells. # indicates that levels of H3 Thr 11 phosphorylation were significantly increased in response to U46619 stimulation in the presence of DHT. † signifies that levels of H3 Thr 11 phosphorylation were significantly decreased in the presence of RO-31-8221. In these cases, single and double symbols signify p Ն 0.05 and p Ն 0.01, respectively, for post hoc Dunnett's multiple comparison t test analysis. $ signifies that levels of H3 Thr 11 phosphorylation were significantly increased in the presence of DHT, and $$$ indicates p Ն 0.001 for unpaired Student's t test. The insets in E and F confirm overexpression of FLAG-tagged dominant negative PRK1 K644E variant, and Ø signifies the empty vector pCMVTag2b. translocation from the cytosolic to the nuclear fraction following stimulation of LNCaP (Fig. 7C) and PC-3 (supplemental Fig.  6) cells with U46619, where maximal nuclear localization was observed at 30 min post-agonist stimulation.
Further confirmation that the agonist-induced increases in H3 Thr 11 phosphorylation observed in both PC-3 and LNCaP cells are due to PRK1-induced mechanisms was established whereby the broad spectrum PRK1 inhibitor RO-31-8220 partially inhibited colcemid-and U46619 phosphorylation and the DHT responses in LNCaP cells (Fig. 6, C and D). As the PRK1 inhibitor RO-31-8220 can also inhibit other kinases, including GSK␤, S6K, RSK, MSK, and PKC␣ (42), additional approaches were used to examine PRK1 specificity. Although overexpression of the wild type PRK1 led to modest increases in basal H3 Thr 11 phosphorylation, which was further increased on U46619 stimulation (supplemental Fig. 5, C and D), overexpression of a kinase-defective dominant negative PRK1 K644E (9) impaired U46619-induced H3 Thr 11 phosphorylation in both PC-3 and LNCaP cells (Fig. 6, E and F, respectively). Furthermore, siRNA directed to PRK1 (siRNA PRK1 ), but not to the scrambled control sequence (siRNA Control ), substantially reduced PRK1 expression (Fig. 8, A and B) and U46619-induced H3 Thr 11 phosphorylation, both in the absence or presence of DHT, in PC-3 and LNCaP cells (Fig. 8, C and D). Moreover, the DHT-induced H3 Thr 11 phosphorylation, both in the absence and presence of U46619, observed in LNCaP cells was partially, but not completely, impaired by the siRNA PRK1 but not by the siRNA Control . The decreases in PRK1 expression in the presence of siRNA PRK1 were not due to unequal loading of the protein samples per se as evidenced by immunoblotting of membranes for the ubiquitously expressed molecular chaperone HDJ-2/ DNA-J protein, which served as an internal loading control (Fig. 8, A and B).
U46619-induced activation of TP␣/TP␤ endogenously expressed in PC-3 cells has been previously shown to increase cell motility and migration, possibly accounting for the increased correlation between TP expression and signaling in prostate cancers (14). Hence, here it was sought to explore TPmediated cell migration in the androgen-responsive LNCaP and nonresponsive PC-3 cell lines and to investigate whether or siRNA Control or, as a control, nontransfected, followed by pretreatment for a further 24 h with either vehicle (0.01% EtOH) or 1 nM DHT prior to assessment of migration for 4 h in either the vehicle (0.01% EtOH), 1 M U46619, 1 nM DHT, or 1 M U46619 plus 1 nM DHT. In all cases, mean cell migration in vehicletreated cells was assigned a value of 100% and agonist-stimulated migration expressed as a relative percentage. *, #, and $ signify that levels of H3 Thr 11 phosphorylation were significantly increased in response to U46619, U46619 in the presence of DHT or DHT stimulation, relative to vehicle-treated cells. † signifies that levels of H3 Thr 11 phosphorylation were significantly decreased in cells transfected with siRNA PRK1 compared with siRNA Control . In all cases, single and double symbols signify p Ն 0.05 and p Ն 0.01, respectively, for post hoc Dunnett's multiple comparison t test analysis. The insets in E and F represent immunoblot analysis of endogenous PRK1 expression in PC-3 and LNCaP cells, where blots were screened with anti-HDJ2 to confirm uniform protein loading.
PRK1 expression may influence that migration (Fig. 8, E and F). Stimulation with U46619 led to substantial increases in migration of both LNCaP and PC-3 cells (Fig. 8, E and F; Nontransfected). Furthermore, DHT also increased migration of LNCaP but not of PC-3 cells, and this effect was augmented by co-stimulation of the former cells with U46619 (Fig. 8, E and F, Nontransfected; two-way ANOVA, p ϭ 0.0004). Disruption of PRK1 expression with the siRNA PRK1 , but not the siRNA Control , specifically impaired U46619-induced cell migration in both PC-3 and LNCaP cells (Fig. 8, E and F). In addition, siRNA PRK1 also partially impaired DHT-induced cell migration in LNCaP cells both in the absence and presence of U46619 (Fig. 8F).
Taken together, these data establish that agonist-induced activation of the TPs endogenously expressed in the prostate adenocarcinoma PC-3 and LNCaP cell lines leads to H3 Thr 11 phosphorylation, a previously recognized marker of chromatin remodeling exclusively associated with androgen-induced responses. Moreover, the TXA 2 mimetic U46619 can significantly augment the androgen-induced H3 Thr 11 phosphorylation in LNCaP but not in PC-3 cells. It was established that PRK1 directly interacts with TP␣/TP␤ endogenously expressed in both PC-3 and LNCaP cells and disruption of PRK1, such as through targeted siRNA, substantially impairs TP-mediated H3 Thr 11 phosphorylation and cell migration in response to U46619. Collectively, these data provide a potential molecular basis for the role of TXA 2 and its receptor in prostate cancer and in other conditions in which aberrant TXA 2 /RhoA/PRK1 signaling is implicated.

DISCUSSION
In this study, we report the discovery of a novel interaction between the TP␣ and TP␤ isoforms of the human TXA 2 receptor (TP) with PRK1, an effector of certain members of the Rho subfamily of monomeric GTPases.
TP␣and TP␤-mediated intracellular signaling to G q and G 12 is well characterized (15,23). However, less is known about their signaling to extracellular stimuli that do not involve direct coupling to heterotrimeric G proteins. TP␣/TP␤ can form homo/heterodimers or oligomers (43), raising the possibility of multiple protein associations at a single TP complex. Moreover, the roles of various GPCR-interacting proteins are increasingly recognized in regulating novel intracellular cascades through direct protein-protein interaction, most often involving the intracellular loops and/or C-tail domain(s) of the GPCR, and which do not necessarily involve classic G protein signaling (44). A number of novel associations with either TP␣ and/or TP␤ have been previously identified. The proteasome activator PA28␥ specifically interacts with TP␤, enhancing its degradation by a proteasome-dependent mechanism (45). Interaction of TP␤ with the nucleoside diphosphate kinase Nm23-H2 leads to its Rac1-dependent endocytosis (46), although its interaction with Rab11 participates in its agonist-induced trafficking through the slow endosome pathway (47). More recently, interactions between TP␣/TP␤ with angio-associated migratory cell protein (48) and with KIAA1005 were reported (49). Interestingly, both PRK1 and KIAA1005 contain C2 domains. However, the significance of these conserved functional domains in the two TP interactants was not examined in this study and will be investigated in future studies.
The direct interaction between TP␣/TP␤ and PRK1 identified in this study was found to be constitutive in mammalian cells. Although there was no agonist-dependent alteration in the association of TP␣/TP␤ with either PRK1 or RhoA, TP agonist stimulation enhanced PRK1 activation leading to phosphorylation of its general substrate histone H1. Although the precise nature of the interaction with PRK1 remains to be determined, Y2H studies identified two regions of importance within the intracellular C-tail domains of TP␣/TP␤, namely the common region (residues 312-328), proximal to transmembrane (TM) 7, and a more distal region of TP␤ (residues 366 -392). In the absence of one or both of these regions, binding to PRK1 640 -942 in yeast is severely reduced or completely abolished. Using GST-based in vitro approaches, the specific requirement of the common 312-328 region within the C-tail domains of TP␣ and TP␤ as a critical binding determinant with both the kinase domain and PRK1 was confirmed. Examination of the subdomains of PRK1 reaffirmed an essential role for the C-terminal domain of PRK1, incorporating the AA-binding site and the kinase domain, in the interaction with TP␣/TP␤, although the N-terminal ACC/HR1 domains, involved in Rho/ Rac binding, were not required. Moreover, ectopic expression of PRK1 561-942 , but not PRK1 1-357 or PRK1 1-594 , specifically competed the interaction of PRK1 with TP␤ and, to a lesser extent, with TP␣. Additional experimentation is required to clarify why PRK1 1-594 can bind to TP␤, and to a lesser extent to TP␣, but does not compete with binding of the full-length PRK1 to either receptor isoform.
More precise mapping of the regions within TP␣/TP␤ and PRK1 that contribute to the interaction is beyond the scope of this study and will be the subject of further investigations. Noteworthy, using the Y2H screening approach, we also investigated the role of the three IC loops in the interaction with PRK1 and its subdomains. However, because of their limited sizes for Y2H-type interaction studies, results generated were inconclusive, and therefore, the possibility that any one of all of the IC domains may contribute to the interaction with PRK1 in mammalian cells cannot be excluded.
The role of the common 312-328 region, encoding the ␣-H8 domain (32,33) of TP␣/TP␤, in contributing to their interaction with PRK1 was also investigated herein. Located proximal to TM7, perpendicular to the heptahelical TM bundles, the ␣-H8 domain can play an essential role in mediating interactions between certain GPCRs and their GPCR-interacting proteins in addition to influencing receptor expression and/or trafficking (50 -52). Moreover, it may act as a conformational switch between the active and inactive states of certain GPCRs. Hence, our discovery of a role for the putative ␣-H8 in the interaction of TP␣/TP␤ with PRK1 is indeed consistent with its functional importance and in mediating protein:protein interactions. Although mutation of certain residues (Leu 316 , Arg 318 , and Leu 323 ) within the ␣-H8 domain completely disrupted the interaction between TP␣ 312-343 and PRK1, mutation of other residues only impaired it. In contrast, mutation of residues within ␣-H8 impaired the interaction of TP␤ 312-407 with PRK1 in each case, although no single mutant completely disrupted impaired TP-mediated H3 Thr 11 phosphorylation. Collectively, these data establish that TP-mediated PRK1 activation can independently lead to H3 Thr 11 phosphorylation in prostate carcinoma cell lines but that it can also cooperate to augment androgen-induced H3 Thr 11 phosphorylation. These findings are entirely significant in that they are the first to establish that agents other than androgens can induce H3 Thr 11 phosphorylation and that it occurs through a similar PRK1-dependent mechanism. Furthermore, in preliminary experiments, it was established that U46619 induced PRK1 translocation to the nucleus, whereas disruption of PRK1 expression impaired TPmediated cell migration in both cell types and blocked a U46619-induced augmentation of LNCaP cell migration in the presence of DHT. Hence, it is indeed tempting to speculate that the observed cooperativity between TP-and AR-mediated PRK1 activation leading to H3 Thr 11 phosphorylation may account for the documented association between increased TXA 2 signaling, including TP expression, in androgen-driven prostate cancer (12,14). It will be of significant interest to establish whether activation of TP signaling, in particular through the novel pathway identified herein involving PRK1-mediated H3 Thr 11 phosphorylation, occurs in other carcinomas in which androgens are implicated such as in ovarian carcinomas showing elevated serum androgen levels (6). The importance of PRK family members in transcriptional regulation is further endorsed by recent reports that key members of class 11a histone deacetylases are phosphorylated by PRK1 and/or by the related PRK2. More specifically, HDAC-5, -7, and -9 but not -4 are phosphorylated by PRK1/2 within their nuclear localization signal, thereby impairing histone deacetylase nuclear import and promoting transcriptional activation (53). PRK1 has been also established to specifically interact with the TNF receptorassociated factor 2 (TRAF2), involving a direct interaction between residues 580 and 584 of PRK1 located in the linker region between its AA-sensitive C2-like and catalytic domains (54). TRAF2 is one of the major mediators of TNF receptor FIGURE 9. Model of AR-and TP-dependent H3Thr 11 phosphorylation in response to PRK1 activation. The cell-permeable androgen testosterone (T) is converted to its active metabolite DHT by the cellular 5-␣-reductase. DHT, in turn, binds to the hormone binding domain of the AR, leading to its release from an inactive complex with heat shock proteins (HSP), promoting AR dimerization and activation. The ligand-bound AR can activate gene expression by (i) translocating to the nucleus leading to transcriptional activation of target gene(s) containing an androgen response element (ARE) within its promoter. In addition, (ii) the ligand-bound AR can activate PRK1 which, in turn, translocates to the nucleus where it specifically catalyzes the phosphorylation of H3 Thr 11 initiating chromatin remodeling, promoting of AR-dependent transcriptional activation of target genes, such as within the prostate. Herein, it was established that PRK1 is recruited into a complex with TP␣ and TP␤. Agonist (U46619)-dependent activation of TP␣ and/or TP␤ leads to activation of PRK1. It was also established that, similar to that of the AR, TP␣/TP␤-dependent activation of PRK1 can also lead to increased H3 Thr 11 phosphorylation, enhancing transcriptional activation of AR-responsive target genes and promoting cell proliferation and/or migration, such as within the prostate. Additionally, in the androgenresponsive LNCaP cells, co-stimulation with the TP agonist and DHT augments PRK1-dependent H3 Thr 11 phosphorylation.
signaling, transducing TNF␣ signaling to its many functional targets, including activation of NF-Band c-Jun kinase (JNK)mediated inflammatory responses and/or apoptotic cascades. Disruption of PRK1 expression impairs TRAF2-induced NF-B activation linking PRK1 to the regulation of TNF␣-mediated inflammatory responses (54). Interestingly, in a follow-up study, PRK1 was found to specifically phosphorylate the related TRAF1, which lacks the JNK/IKK signaling effector domain but not its interactant TRAF2 or other TRAF members, leading to the recruitment of TRAF1 to the TNF receptor, silencing the receptor complex by PRK1 (55). Moreover, within the vasculature, PRK1 has been implicated in the mediation of VSM-specific gene expression promoting VSM differentiation, such as in response to transforming growth factor-␤1 (56,57). Keeping in mind the central role of TXA 2 within the vasculature, including the mediation of inflammatory disease, coupled with the finding herein of a direct interaction with and activation of PRK1, it will be of significant interest to investigate the possible interplay between those critical pathways, be it at the cellular or (patho)physiologic levels.
In conclusion, as outlined in the model presented in Fig. 9, this study provides evidence of a novel constitutive interaction between TP␣/TP␤ and PRK1, in complex with RhoA. Furthermore, results demonstrate an agonist-dependent increase in PRK1 activity and a significant TP-dependent increase in H3 Thr 11 phosphorylation and associated cell migration in prostate carcinoma cell lines, a modification that was until this study almost exclusively viewed as an androgen-induced marker of chromatin remodeling and transcriptional activation. Although the involvement of RhoA in the TP-mediated H3 Thr 11 phosphorylation by PRK1 is currently unknown, requiring additional experimentation, the discovery herein of a direct interaction between TP␣/TP␤ with PRK1 is significant. For example, such a discovery is likely to impact on the understanding of a range of (patho)physiologic processes in which aberrant TXA 2 / TP, RhoA, and PRK signaling is implicated, not the least within certain neoplasms and in vascular and hypertensive disease. Further investigation will reveal a more complete understanding of the physiologic, and possible (patho)physiologic or clinical, significance of this interaction and whether TP antagonism might offer a therapeutic advantage in such conditions.