Nucleobindin Co-Localizes and Associates with Cyclooxygenase (COX)-2 in Human Neutrophils

The inducible cyclooxygenase isoform (COX-2) is associated with inflammation, tumorigenesis, as well as with physiological events. Despite efforts deployed in order to understand the biology of this multi-faceted enzyme, much remains to be understood. Nucleobindin (Nuc), a ubiquitous Ca2+-binding protein, possesses a putative COX-binding domain. In this study, we investigated its expression and subcellular localization in human neutrophils, its affinity for COX-2 as well as its possible impact on PGE2 biosynthesis. Complementary subcellular localization approaches including nitrogen cavitation coupled to Percoll fractionation, immunofluorescence, confocal and electron microscopy collectively placed Nuc, COX-2, and all of the main enzymes involved in prostanoid synthesis, in the Golgi apparatus and endoplasmic reticulum of human neutrophils. Immunoprecipitation experiments indicated a high affinity between Nuc and COX-2. Addition of human recombinant (hr) Nuc to purified hrCOX-2 dose-dependently caused an increase in PGE2 biosynthesis in response to arachidonic acid. Co-incubation of Nuc with COX-2-expressing neutrophil lysates also increased their capacity to produce PGE2. Moreover, neutrophil transfection with hrNuc specifically enhanced PGE2 biosynthesis. Together, these results identify a COX-2-associated protein which may have an impact in prostanoid biosynthesis.

The COX-2 enzyme has generated particular interest for its implication in inflammation, cellular proliferation, differentiation and tumorigenesis, [15,16], and has recently emerged as a therapeutic target in the treatment and prevention of human cancers [17][18][19][20]. Also, COX-2 mediates physiological events such as kidney functions, post-natal development and female reproductive processes [21][22][23]. In spite of the pivotal roles of COX-2 in many aspects of biology, much remains to be discovered around the regulation of its activity in inflammatory cells. In particular, proteins that associate with COX-2 have yet to be identified.
Nucleobindin (Nuc) is a ubiquitous protein featuring multiple putative functional domains, indicating its potential implication in a number of cellular processes [24][25][26][27][28]. As such, Nuc has been the focus of reports originating from diverse fields including autoimmunity [29], intracellular signaling [30], osteogenesis [26], cancer [31] and inflammation [27]. At the protein level, Nuc is constituted of 460 amino acids, including an N-terminal 25 amino acid signal peptide responsible for its initial localization to the endoplasmic reticulum (ER) [32]. In addition, Nuc contains several classical interaction domains: a DNA binding site, a heterodimerization domain, two EF-hand Ca 2+ -binding sites, a nuclear localization signal [24][25][26] as well as non-classical proteinprotein interaction domains including a G-protein-binding region and an high affinity COX-binding domain, as evidenced by a yeast two-hybrid assay [27,28]. Depending on the model at hand, Nuc has been detected in various subcellular structures such as the nucleus [31,33], mitochondria [34] the cytoplasm [34][35][36][37], the endoplasmic reticulum (ER) [33,34,36] and the Golgi apparatus [35]. The Golgi, like the ER, plays a role as an intracellular Ca 2+ reservoir, which can be released in the cytosol in response to various stimuli, in turn activating a number of intracellular signaling cascades [38]. As such, Nuc may be involved in establishment of the agonist-mobilizable Golgi Ca 2+ store [30].
However, notwithstanding a putative COX-binding site [27] and a relatively well-characterized capacity to bind Ca 2+ [36], the biological functions of Nuc remain elusive.
In the present study, we investigated the expression of Nuc, its subcellular localization, its expression and affinity for COX-2, as well as its impact on COX-2-dependent PGE 2 biosynthesis in human neutrophils. Results obtained identify Nuc as a COX-2associated protein which may have a role in the biosynthesis of prostanoids.
Cloning and purification. Full-length human Nuc was amplified using Expand HIFI+ DNA polymerase (Roche, Laval, Qc, Canada) from pOTB7/Nucleobindin clone (ID 2821805 produced by Invitrogen Life Technologies, Carlsbad, CA, USA). PCR reactions (35 cycles; annealing temp. 60uC) were performed using the following primers: 59-GGA ATT TCA TAT GCC TCC CTC TGG-39 (forward) and 59-CCT AGC TCA TAT GTC ACA GAT GCT GG-39 (reverse); yielding a PCR product of 1409 bp in length. pET/NucDCBD was amplified from pET/Nuc using same DNA polymerase. PCR reaction (30 cycles; annealing temp. 60uC) was performed using the following primers: 59-GGA ATT ACA TAT GAG TCC CGA CAC AGG-39 (forward) and 59-CCT AGC TCA TAT GTC ACA GAT GCT GG-39 (reverse); yielding a PCR product of 1283 bp in length. Sequences of amplified fragments were confirmed by DNA sequencing. The cDNA products were cloned into the NdeI restriction site of pET-15b (Novagen, San Diego, CA, USA). BL21 cells were transformed with pET-15b/Nuc or pET/NucDCBD expression vectors and induced with 1 mM isopropyl-1-thio-b-D-galactopyranoside (Tekniscience, Terrebonne, Qc, Canada). Bacterial extracts were processed for protein purification using a His-Bind Resin column and buffer kit (Novagen).
Production of polyclonal antibodies against nucleobindin. Rabbits were injected with an emulsion of 50-100 mg hrNuc and complete Freund's adjuvant. Total IgGs were purified using a Protein G-coupled sepharoseH 4 fast flow column (GE Healthcare, Waukesha, WI, USA).
Human leukocyte isolation. Neutrophils were isolated as originally described [39] with modifications [4]. Viability was greater than 98%, as determined by trypan blue dye exclusion.
In cell-based assays, neutrophils stimulated with GM/TNF for 2 h were pelleted and resuspended in 700 ml ice-cold HBSS containing the anti-protease cocktail. Suspensions were sonicated on ice and centrifuged (30006 g). Cellular extract aliquots (5 ml) were incubated with hrNuc in a total of 200 ml for 15 min at 37uC, before stimulation with AA (10 mM final) for 30 min at 37uC. Reactions were stopped on ice-cold water; samples were briefly Figure 1. Expression of Nuc in human neutrophils. A) Real-time PCR determination of Nuc, COX-1 and COX-2 messenger RNA expression in neutrophils. Cells were stimulated for 60 min with lipopolysaccharide (LPS ; 100 ng/ml), a mixture of granulocyte/ monocyte colony stimulation factor and tumor necrosis factor-a (GM/ TNF; 1.4 nM and 100 ng/ml respectively), formyl-methionyl-leucylphenylalanine (fMLP; 100 nM) or with PMA (10 nM). Samples were processed for the determination of GAPDH, COX-1, COX-2 and Nuc mRNA expression by real-time-PCR. Shown are integrated results from n = 4 (6SEM) separate experiments performed in identical conditions with different donors. B) Nuc protein expression in neutrophils, as determined by western immunoblotting. Cells were incubated for 2 h with diluent (saline), or with GM/TNF. Samples were processed for the determination of Nuc expression by western immunblotting. Nuc is constitutively present in unstimulated neutrophils. Note that hrNuc migrated slightly slower than neutrophil native Nuc, due to the presence of the signal peptide and of the additional His-Tag sequence. Shown is one immunoblot, representative of four identical experiments performed with different donors. doi:10.1371/journal.pone.0002229.g001 centrifuged and supernatants were assayed for their contents in PGE 2 by ELISA.
Nitrogen cavitation. The procedure was conducted essentially as described [40], with modifications. For each cell preparation, 18 fractions were generated (1 ml each), starting from the bottom of the tube. This procedure allows the distinct separation of azurophil granules, specific granules, gelatinase granules, secretory vesicles, a plasma membrane-enriched fraction, and cytosol [40]. Each fraction was re-centrifuged (100 0006 g, 90 min) in a Beckman TL 100 ultracentrifuge, using a TL 100.2 rotor, in order to pellet Percoll. Fractions (50 ml) were carefully aspirated with a pipet and processed for western immunoblot analysis.
Immunofluorescence. Neutrophils were fixed in 4% paraformaldehyde for 20 min, then washed twice with PBS. Cell suspensions were laid onto poly-L-lysine-coated glass slides and allowed to air-dry. Slides were incubated with a permeabilization buffer (0.5% NP-40, 5% heat-inactivated fetal bovine serum (FBS) and 5% heat-inactivated donkey serum in PBS) for 5 min, washed three times in washing buffer (5% FBS, 0.05% NP-40 in PBS), incubated with a blocking buffer (10% FBS and 10% donkey serum in PBS) for 30 min and washed three times prior to a 1 h incubation with an anti-Nuc antibody (diluted 1/150 in: PBS with 5% FBS, 5% donkey serum and 0.05% NP-40) in a humid environment. After washes, slides were incubated with goat anti- were processed for cavitation and subcellular fractionation, as described in Experimental procedures. Nuc and COX-2 co-localized in ER/Golgicontaining fractions, as determined by western immunoblotting. B) In GM/TNF-stimulated neutrophils, samples were processed as in A) for the determination of the indicated proteins. C) In GM/TNF, and GM/TNF+fMLP (100 nM) stimulated neutrophils, samples were processed as in A) for the determination of the indicated proteins. In each panel, all immunoblots originate from the same membrane. Shown is one immunoblot, representative of four identical experiments performed with different donors. GRP-78: ER/Golgi marker; lactoferrin: marker of specific granules; albumin; marker of secretory vesicles; 58k: 58k Golgi protein (Golgi marker); mPGES-1: microsomal prostaglandin E 2 synthase-1; TXA 2 -Synthase: thromboxane A 2 synthase; cPLA 2 : Type IV cytosolic phospholipase A 2 . D) Schematized protein structure of human Nuc and putative functional domains. The main characterized domains found in Nuc are: a signal peptide directing the protein to the ER; a COX-binding site; a putative nuclear localization signal embedded in to a DNA-binding site; two EF-hand Ca2+-bindins sites; a leucine zipper region. doi:10.1371/journal.pone.0002229.g002 rabbit AlexaFluorH 488 (Molecular Probes, Carlsbad, CA, USA; diluted 1/200) for 30 min in the dark, in a humid environment. Slides were washed once and then incubated with DAPI 0.7 mmoles/ml (Molecular Probes) or with 250 ng/ml propidium Iodine (PI; Sigma) for 5 min in the dark. Slides were washed in PBS and prepared for microscopy with Gel/Mount TM (Biomeda, Foster City, CA, USA). Images were captured by a CoolSNAP HQ camera mounted on an Olympus BX-51 upright microscope using a 606 UPlan Apo objective, and processed with ImagePro 4.5.1 software (Media Cybernetics, Silver Spring, MD, USA). Confocal microscopy was performed on an Olympus BX-61 microscope using a UPlan Apo 1006objective with immersion oil. Data was collected with the FluoView software (Olympus).
Immunoprecipitations with magnetic beads. Anti-Nuc antibodies generated in-house, or irrelevant IgGs, were coupled to Dynabeads M-500 subcellular (Dynal, Norway). Immunomagnetic beads were incubated with selected subcellular fractions at a ratio of 10 ml fraction/5 million beads/ml in PBS pH 7.4, 2 mM EDTA, 5% BSA and the anti-protease cocktail, for 12 h at 4uC. Beads were magnetically immobilized and supernatants were discarded. The beads were washed three times in PBS pH 7.4, 2 mM EDTA, anti-protease cocktail, with decreasing concentrations of BSA (5%, 0.1%, 0%), then resuspended in sample buffer 16 and boiled for 2-3 min.
Pull-down assay. hrNuc was coupled to CNBr-activated SepharoseH 4B beads (GE Healthcare). PMA-stimulated cells were pelleted (microfuge) and resuspended in 0.1% NP-40 lysis buffer [41] for 10 min at 4uC. The lysates were centrifuged at 10006 g for 10 min at 4uC. In this procedure, COX-2 is mainly located in the non-nucleus fraction, which was used for the present assay. Supernatants were incubated with 50 ml of hrNuc-coupled sepharose beads, or with inactivated sepharose beads for 2 h, RT. After incubation, beads were centrifuged (30 sec, 15006 g), washed twice with HBSS + CaCl 2 + antiprotease cocktail, resuspended in sample buffer 16 and heated 2-3 min at 95uC.
Immunoprecipitations under native conditions.
Immunoprecipitations were performed as described earlier [42], with modifications. Briefly, neutrophils stimulated with GM/TNF were centrifuged, and the cell pellets were lysed by adding cold lysis buffer (10 mM Tris-HCl, pH 7.4, 137.2 mM Nacl, 1 mM EDTA, 0.6% CHAPS, 2 mM orthovanadate, and the protease inhibitor cocktail) for 5 min on ice. The insoluble material was discarded after centrifugation at 13 0006g at 4 uC during 5 min. The supernatant was harvested, then precleared with protein A-Sepharose at 4 uC for 30 min. Resulting supernatants were incubated at 4 uC either with 8 mg of anti-COX2 (mouse) antibodies or 8 mg of anti-Nuc (rabbit) antibodies for 1 h followed by 2 h incubation with protein A-Sepharose beads. The beads were collected and washed three times with cold lysis buffer. Laemmli sample buffer (26) was added to the beads, which were boiled for 7 min. Transfection of recombinant nucleobindin. We used the Pro-Ject Protein Transfection Reagent (Pierce, Rockford, IL), according to the manufacturer's instructions. Briefly, neutrophils were stimulated with GM-CSF/TNF-a at 37uC. After 15 min, a mixture of 7.5 ml of the reagent with 2 mg of the protein (Nuc or NucDCBD) were added and incubated for 4 h at 37uC. AA (10 mM) was added and samples were incubated for an additional 30 min at 37uC. Samples were centrifuged and supernatants were assayed for PGE 2 content by ELISA. Cell pellets were processed for SDS-PAGE.
Statistical analysis. Where applicable, statistical analysis was performed by Student's non-paired t-test (two-tailed), and significance (*, **) was considered attained when p was ,0.05.

Nucleobindin is constitutively expressed in human phagocytes
Expression of Nuc in neutrophils was first assessed. To this end, cells were incubated with agonists known to stimulate inflammatory gene expression in these cells: lipopolysaccharide (LPS), the formylated synthetic peptide fMLP, the phorbol ester PMA, or a mixture of granulocyte-macrophage colony-stimulating factor and tumor necrosis factor-a (GM/TNF). Following stimulations, mRNA levels of COX-1, COX-2 and Nuc were determined by real-time PCR. While each of the agonists elicited an increase in expression of COX-2 mRNA, as previously reported [3,43], that of Nuc only varied in a modest fashion and, in most conditions, variations were comparable to that of COX-1, a constitutively-expressed gene in neutrophils (Fig. 1A). Similar results were obtained in human monocytes stimulated with LPS (data not shown). These results are consistent with the reported structure for the promoter region of the Nuc gene, featuring typical elements of house-keeping genes [32]. Data obtained at the protein level also indicate that resting neutrophils constitutively express Nuc (Fig. 1B).

Nuc and COX-2 are localized in the Golgi and ER of neutrophils
The subcellular localization of Nuc, and of the enzymatic machinery responsible for prostaglandin biosynthesis was investigated in human neutrophils. To this end, we used the nitrogen cavitation technique coupled to fractionation on a PercollH density Figure 4. Demonstration of co-localization for Nuc and COX-2 in neutrophils, by electron microscopy. Neutrophils stimulated with GM/ TNF were processed for the detection of Nuc (A), COX-2 (B) and Golgi (C), by indirect immunostaining and electron microscopy. For each protein, labeling was mainly found in a single cluster situated between nuclear lobes, in the center of the cell. D) Samples were processed for the double detection of Nuc and of COX-2 by electron microscopy. To this end, a polyclonal chicken anti-Nuc antibody and a polyclonal rabbit anti-COX-2 antibody, were used in sequence, as described in the Experimental procedures. Co-localization of Nuc (18 nm beads, indicated by thick arrows) and COX-2 (6 nm beads, thin arrows) is clearly seen in a cluster between nuclear lobes. In each panel, a single neutrophil is shown in the upper-left corner; the magnified region of interest is represented by the respective white square. doi:10.1371/journal.pone.0002229.g004 gradient, a well-recognized procedure which allows for the separation of intracellular compartments such as: four distinct populations of granules, secretory vesicles, plasma membranes and the cytosol [44]. Neutrophils were stimulated for 2 h with GM/ TNF, a condition which efficiently up-regulates their expression of COX-2 [3]. Following stimulation, cells were processed for nitrogen cavitation, subcellular fractionation, and detection of proteins of interest by western immunoblotting. The fractionation pattern was validated with the use of specific cellular compartment markers. Lactoferrin, a marker of specific granules, was predominantly found in fractions 6 to 8, whereas albumin, the marker for secretory vesicles, was predominantly in fractions 9 to 11 ( Fig. 2A), in accordance with previous reports [44,45]. Nuc was mainly found in fraction 11, both in diluent-and in GM/TNF-stimulated cells. When COX-2 was up-regulated, it co-localized with Nuc, in the fractions 10 to 12 of GM/TNF-stimulated cells ( Fig. 2A, right panel) and also with the glucose-related protein (GRP)-78, marker for Golgi and ER structures. Localization patterns of Nuc, COX-2 and GRP-78 systematically matched with each other, suggesting that Nuc and COX-2 both reside in the Golgi and ER structures [44].
Additional fractionation experiments were performed with GM/TNF-stimulated neutrophils in order to localize additional enzymes implicated in prostanoid synthesis. Microsomal PGE 2synthase-1, thromboxane-synthase, as well as 58k, an additional marker for the Golgi, also co-localized with COX-2 and Nucpositive fractions (Fig. 2B). For the specific case of type IV cytosolic phospholipase A 2 (cPLA 2 ), GM/TNF-treated cells were stimulated with fMLP prior to the fractionation process in order to induce its phosphorylation and translocation to membranes [41]. In this situation, cPLA 2 was readily detected in COX-2-and Nuccontaining fractions (Fig. 2C).

Nuc and COX-2 are localized in proximity of each other in neutrophils
Results obtained so far suggest a Golgi and ER localization for Nuc and for the enzymatic machinery mediating prostanoid synthesis. We sought further confirmation for this co-localization in intact cells, by immunofluorescence, confocal micrscopy, and electron microscopy. For immunofluorescence experiments, resting neutrophils were fixed and permeabilized, then processed for the detection of Nuc by indirect labeling. As can be appreciated in Fig. 3A (left panel), immunoreactive Nuc (green labeling) appeared predominantly embedded near the center of the cells and between nuclear lobes, typically with one or two main spots per cell and a small number of secondary spots, a pattern consistent with a Golgi and ER localization [46]. Analysis of the samples by confocal microscopy further precised the central location of Nuc within the cell (Fig. 3B). A 3D representation of the confocal data (Movie S1) nicely illustrates the central localization of Nuc within the neutrophil, consistent with its Golgi and ER localization. In electron microscopy experiments, intact neutrophils stimulated with GM/TNF were fixed and embedded; ultrathin slices were incubated with specific polyclonal antibodies for the detection of Nuc (Fig. 4A), COX-2 (Fig. 4B) or the Golgi marker GRP-94 (Fig. 4C). For each of the three proteins, labeling was found mainly clustered in a central area localized in the vicinity of nuclear lobes, and in a small number of secondary sites, largely confirming the immunofluorescence and confocal microscopy data and supporting the idea that Nuc and COX-2 both localize in the Golgi and ER. In a separate set of experiments, samples were subjected to a double-labeling and prepared for electron microscopy showed proximity between COX-2 and Nuc (Fig. 4D). The pattern of labeling demonstrated that both proteins can be located very close to each other, clustered on the luminal side of vesicular structures (Nuc, thick arrows; COX-2, thin Immunomagnetic beads coated with anti-Nuc IgGs or with irrelevant IgGs were incubated with an aliquot from the positive fractions (10 to 12) showed in Fig. 1. Anti-Nuc-coated beads immunoprecipitated a structure that was positive for COX-2 and GRP-78. Results are from one experiment, representative of n = 3 distinct experiments performed in identical conditions. D) Left panel: Nuc was immunoprecipitated from COX-2-expressing neutrophils, using anti-Nuc or irrelevant anti-IgG antibodies, as described in the Experimental procedures; samples were processed for the detection of COX-2 by western immunoblot. Right panel: COX-2 was immunoprecipitated using anti-COX-2 or irrelevant anti-IgG antibodies and samples were processed for the detection of Nuc by western immunoblot (IP: Immunoprecipitation; WB: western immunoblot). Results are from one experiment, representative of n = 2 distinct experiments performed in identical conditions. doi:10.1371/journal.pone.0002229.g005 arrows). Experiments performed with resting neutrophils yielded virtually no COX-2 labeling (data not shown).

Nuc binds to COX-2 with high affinity
Affinity between Nuc and COX-2 [27] was confirmed in neutrophils by incubating sepharose beads coated with hrNuc with lysates obtained from resting-or PMA-stimulated neutrophils. COX-2 co-immunoprecipitated along with hrNuc, assessed by western immunoblottings (Fig. 5A). Chelation of Ca 2+ with 5 mM EGTA did not prevent immunoprecipitation of COX-2 by hrNuc (Fig. 5B), indicating that the association between the two proteins does not chiefly rely on availability of Ca 2+ . Presence of Nuc and COX-2 within the same organelle was also demonstrated by immunoprecipitation experiments. An aliquot of Nuc-positive fractions resulting from nitrogen cavitation experiments (right panel of Fig. 2A, fraction 11) was incubated with anti-Nuc polyclonal antibodies covalently-linked to magnetic beads. These anti-Nuc-coated beads specifically immunoprecipitated structures which, in addition to containing Nuc, also contained COX-2 and GRP-78 (Fig. 5C), showing that the organelles containing Nuc also harbor COX-2 and Golgi/ER structures. Finally, direct association between Nuc and COX-2 was demonstrated by two ways through immunoprecipitation experiments. First, COX-2-expressing neutrophil lysates were treated with an anti-Nuc antibody and presence of COX-2 in the immunoprecipitates was confirmed by western immunoblots (Fig. 5D, left panel). Second, lysates were conversely treated with an anti-COX-2 antibody, and Nuc could also detected in immunoprecipitates (Fig. 5D, right panel). Together, these results demonstrate co-localization, proximity and direct association between Nuc and COX-2 in human neutrophils.

Nucleobindin increases cyclooxygenase-2-dependent prostaglandin E 2 generation
We addressed the intriguing possibility that Nuc may impact on COX-2 activity, first by using an in vitro enzymatic assay with purified human recombinant (hr)COX-2. In this highly-simplified system, addition of hrNuc increased the generation of PGE 2 in a concentration-dependent fashion, up to 4 fold over basal levels (Fig. 6A). Pre-treatment of Nuc with the anti-Nuc polyclonal antibody prevented the increase in PGE 2 production (Fig. 6B), pointing to a specific implication of Nuc in this process. In addition, the Nuc-enhanced PGE 2 production was entirely prevented by the specific COX-2 inhibitor NS-398, confirming a COX-2 mediated event (Fig. 6B). These results, showing enhancing impact of Nuc on COX-2-dependent PGE 2 production, support the concept of a physiologically significant interaction between Nuc and COX-2.
This point was specifically addressed: lysates from COX-2expressing neutrophils were stimulated with AA, alone or in the presence of increasing quantities of hrNuc, and production of PGE 2 was measured. In these experiments, exogenous hrNuc increased the production of PGE 2 in a concentration-dependent fashion, up to 5fold over basal levels (Fig. 7A). And, once again, the Nuc-enhanced PGE 2 production was prevented by pre-incubation of hrNuc with an anti-Nuc antibody, or by the presence of NS-398 (Fig. 7B), confirming a Nuc and COX-2-mediated event. Further evidence of a functionally relevant interaction between Nuc and COX-2 was obtained by transfecting Nuc into intact COX-2-expressing neutrophils. Cells were transfected either with full-length hrNuc or with hrNuc lacking the COX-binding domain (NucDCBD), then stimulated with AA. As can be appreciated in Fig. 7C, addition of full length hrNuc specifically caused a significant increase in PGE 2 biosynthesis by neutrophils, while NucDCBD was ineffective. Results are expressed as percentage of inhibition of PGE 2 production, when compared to the production obtained in the absence of antibody or inhibitor (mean6s.e.m., n = 4. *: significantly higher than samples incubated without hrNuc; **: significantly higher than samples incubated with 2.0 mg hrNuc or less). doi:10.1371/journal.pone.0002229.g006

Discussion
All approaches taken, be that cavitation coupled to subcellular fractionation, electron microscopy, immunofluorescence, or confocal microscopy, indicated that Nuc mainly clusters in a central place in the vicinity of -but distinct from-nuclear lobes, a region in which the Golgi resides [46], and in a small number of nearby additional spots, consistent with ER localization. Conversely, Nuc does not appear to be present in granules or in the plasma membrane of human neutrophils. It remains possible that lower levels of Nuc could be found elsewhere, but the thorough approach undertaken herein strongly points to a Golgi/ER localization for Nuc.
In different cell types , COX-2 has also been reported to localize in the Golgi and ER. In addition to an N-terminal signal peptide which causes initial ER integration during translation, a Cterminal KDEL-like signal (PTEL) is also present in the structure of COX-2 and thought to be recognized by a membrane-bound receptor that continually retrieves the proteins from later compartments of the secretory pathway and returns them to the ER [47]. Results from the present study further document a Golgi and ER localization for Nuc and COX-2 in neutrophils. In addition, activated cPLA 2 , mPGES-1 and TXA 2 -synthase, in fact all enzymes of the prostanoid biosynthesis machinery, were also found in the same Golgi-and ER-containing fractions, along with Nuc and COX-2. Fractionation procedures and microscopy experiments further showed that, in human neutrophils, Nuc and COX-2 can localize in proximity of each other. Finally, immunoprecipitations and pull down assays each confirmed direct interaction and a high affinity between Nuc and COX-2.
When focusing on the delineation of a physiologically-relevant function for this high-affinity association, hrNuc specifically increased the COX-2-mediated formation of PGE 2 , in three distinct settings: purified COX-2, cell lysates, and transfected intact cells. Amongst all known prostanoids, purified human neutrophils only release PGE 2 and TXA 2 from COX-2 activity [3]. In turn, only PGE 2 was considered in this study as it can be non-enzymatically produced from COX-generated PGH 2 . The increase in COX-2-dependent PGE 2 generation was concentration-dependent and reached up to five fold increase, relative to basal levels. Also, presence of a putative COX-binding domain [27] was necessary for increasing PGE 2 production. These experiments could not, however, take into consideration crucial factors such as the micro-environment, compartmentalization, or cellular architecture. Moreover, post-translational modifications of native Nuc, absent in hrNuc, may affect interactions in an as of yet unknown fashion. In this regard, Nuc does not appear to be Nor O-linked glycosylated, but possesses up to 10 potential phosphorylation sites including three protein kinase C sites which may well impact on its conformation and propensity to associate with other proteins, including COX-2 [26]. In turn, further studies documenting a thorough determination of the stoichiometry of this association, as well as of involvement of final-step enzymes (e.g., mPGES-1, TXA 2 -S), which will most likely require combinations of cell-free and cell line-based experimental set ups, will be necessary before the architecture of this pivotal enzymatic complex can be progressively unveiled. Nonetheless, results obtained indicate a high affinity between Nuc and COX-2.
In summary, we found that Nuc mainly localizes in the Golgi and ER of human neutrophils, along with COX-2 and other enzymes involved in prostanoid generation. Nuc can associate with COX-2 with high affinity and increase the resulting PGE 2 generation. The potential role of Nuc in the regulation of PGE 2 Figure 7. Nuc increases PGE 2 biosynthesis in a COX-2-dependent manner in human neutrophils. A) COX-2-expressing neutrophil extracts were incubated with AA (10 mM) for 30 min, alone or in the presence of indicated quantities of hrNuc. PGE 2 production was measured by ELISA. Results are expressed as percentages of maximum PGE 2 production (mean6s.e.m., n = 3). B) Left bar; hrNuc (1 mg) was treated with a polyclonal anti-Nuc antibody prior to incubation with the cellular extracts. Right bar; the COX-2 specific inhibitor NS-398 (50 mM) was used to confirm a COX-2-mediated PGE 2 biosynthesis. Results are expressed as inhibition of PGE 2 production, when compared to the production obtained in the absence of antibody or inhibitor (mean %6s.e.m., n = 4). C) Bar graph; GM/TNF-treated neutrophils were transfected with full-length hrNuc, or lacking a COX-binding domain (NucDCBD) using the Pro-Ject procedure as described in Experimental procedures, then stimulated with AA (10 mM). PGE 2 production was measured by ELISA. (mean6s.e.m., n = 3. *: significantly higher than samples incubated without Nuc). Western immunoblots; Cells treated as described above were processed for the determination of cellular COX-2, Nuc, hrNuc and NucDCBD protein levels. Note that hrNuc migrates higher than endogenous Nuc (or NucDCBD), because of the signal peptide sequence still being present. Immunoblots are from one experiment, typical of three independent experiments performed in identical conditions. doi:10.1371/journal.pone.0002229.g007 production is of interest in a large number of physiological settings, and the present report might be a first step in a characterization of this pivotal enzymatic complex.

Supporting Information
Movie S1 3D visualization of Nuc localization. A 45 sec video sequence, based on confocal data obtained on the sample presented in B), illustrates the 3 dimensional localization of Nuc in the center of one human neutrophil. Nuc is clearly observed in the center of the cell and within a limited number of vesicles in the vicinity of the nucleus, typical of the Golgi and endoplasmic reticulum, respectively. The second half of the video is in 'edge' mode, which delineates the contours of localization of Nuc, putting in evidence a central reservoir at the center of the cell as well as close-by endoplasmic reticulum vesicles. Found at: doi:10.1371/journal.pone.0002229.s001 (3.36 MB MOV)