Identification of morphine-6-glucuronide in chromaffin cell secretory granules

We report for the first time that morphine-6-glucuronide, a highly analgesic morphine-derived molecule, is present in adrenal chromaffin granules and secreted from chromaffin cells upon stimulation. We also demonstrate that phosphatidylethanolamine-binding protein (alternatively named Raf-1 kinase inhibitor protein or RKIP) acts as an endogenous morphine-6-glucuronide-binding protein. An UDP-glucuronosyltransferase 2B-like enzyme, described to transform morphine into morphine-6-glucuronide, has been immunodetected in the chromaffin granule matrix, and morphine-6-glucuronide de novo synthesis has been characterized, demonstrating the possible involvement of intragranular UDP-glucuronosyltransferase 2B-like enzyme in morphine-6-glucuronide metabolism. Once secreted into the circulation, morphine-6-glucuronide may mediate several systemic actions (e.g. on immune cells) based on its affinity for mu-opioid receptors. These activities could be facilitated by phosphatidylethanolamine-binding protein (PEBP), acting as a molecular shield and preventing morphine-6-glucuronide from rapid clearance. Taken together, our data represent an important observation on the role of morphine-6-glucuronide as a new endocrine factor.

We report for the first time that morphine-6-glucuronide, a highly analgesic morphinederived molecule, is present in adrenal chromaffin granules and secreted from chromaffin cells upon stimulation. We also demonstrate that PhosphatidylEthanolamine-Binding Protein (alternatively named Raf-1 Kinase Inhibitor Protein) acts as an endogenous morphine-6-glucuronide-binding protein. An UDP-glucuronosyltransferase 2Blike enzyme, described to transform morphine into morphine-6-glucuronide, has been immunodetected in chromaffin granule matrix and morphine-6-glucuronide de novo synthesis has been characterized demonstrating the possible involvement of intragranular UDPglucuronosyltransferase 2B-like enzyme in morphine-6-glucuronide metabolism. Once secreted into circulation, morphine-6glucuronide may mediate several systemic actions (e.g., on immune cells) based on its affinity for mu opioid receptors. These activities could be facilitated by PhosphatidylEthanolamine-Binding Protein, acting as a molecular shield and preventing morphine-6-glucuronide from rapid clearance. Taken together, our data represent an important observation on the role of morphine-6-glucuronide as a new endocrine factor.
For 20 years, cerebral endogenous morphine has been isolated and characterized and its structure shown to be identical to morphine from poppy (1,2). In the 1990s, few groups have succeeded in characterizing endogenous morphine from organs and fluids of vertebrates, including brain and adrenal gland from bovine, heart and adrenal gland from rat, human heart and urine (for review (1)(2)(3)), invertebrates (4-6) and parasites (7,8). In mammals, endogenous morphine synthesis pathway is associated with the biosynthesis pathway of dopamine and catecholamines (for review (1,3,9)). Then, very recently, several crucial steps were reached since Poeaknapo et al. demonstrated that morphine can be formed by a multi-step biosynthetic route (10,11), and Zhu et al. have shown that human primary polymorphonuclear cells are able to synthesize morphine (12). These authors have showed morphine de novo synthesis in human and animal primary and cancer cell cultures.
Chromaffin cells are neuroendocrine cells originating from the neural crest and are the predominant cell type in the adrenal medulla (see review (13)). These cells possess the catecholamine biosynthetic pathway leading to dopamine and adrenaline/noradrenaline synthesis (13). Chromaffin secretory granules contain a complex mixture of peptides and proteins that are co-released with catecholamines into circulation in response to splanchnic nerve stimulation (13). Based on the morphine biosynthetic pathway and on the presence of morphine in bovine and rat adrenal gland (14,15), rat adrenal pheochromocytoma PC-12 cells (10,16)) and eel chromaffin cells (17), we hypothesized that mammalian chromaffin cells might have the capacity to synthesize morphine and that their secretory granules could potentially release this alkaloid into the blood.
Several molecular blood transporters of clinically administrated morphine have been identified (e.g., serum albumin, gamma globulin (18) or alpha 1-acid glycoprotein (19)). In addition, it has been reported that the PhosphatidylEthanolamine-Binding Protein (PEBP) (20), alternatively named Raf-1 kinase inhibitor protein (RKIP) (21), is also able to bind morphine (22). We have recently reported that PEBP is present in secretory granules, as well as in the exocytotic medium of stimulated bovine primary cultured chromaffin cells and in bovine serum (23).
Exogenously administered morphine is catabolized in the liver by the UDPglucuronosyltransferase 2B enzyme family (UGT2B) (24), leading to the formation of morphine 3-glucuronide (M3G) and morphine 6glucuronide (M6G). M3G is totally inactive whereas M6G appears to display stronger analgesic activity than morphine (50 to 600 times depending on animal model) (25) as well as exhibiting much longer half-life (26).
Using immunocytochemical, biochemical and proteomic strategies, the present study reveals that bovine chromaffin cell secretory granules store M6G-PEBP complex. We were able to show that M6G is secreted from bovine chromaffin cells upon nicotinic stimulation. Coimmunoprecipitation and affinity chromatography experiments also revealed the interaction between the alkaloid carrier (i.e., PEBP) and an UGT2Blike enzyme necessary for the conversion of morphine into M6G. Finally, M6G de novo synthesis was demonstrated by using the UGT2Blike enzyme present in chromaffin granules.
All glass cover slips were mounted on a glass slide with a drop of Mowiol 4-88.
Isolation of chromaffin cell subcellular fractions-Chromaffin cell plasma membrane, cytoplasm, intragranular matrix and granule membrane were prepared as described by Smith and Winkler (27) and modified by us (23,28). After protein quantification using Bradford technique, the purity of subcellular fractions was assayed as previously described by using subcellular markers (23). Total microsomal fraction was purified according to Levesque et al. (29).
Isolation of proteins exocytotically released from stimulated chromaffin cells-Chromaffin cells (2.5x10 6 ) cultured for 3 days were washed 4 times 5min with 10 ml of Locke's solution at 37°C in order to get rid of the culture medium and floating dead cells. Cells were subsequently stimulated for 10 min with 10 µM nicotine in Locke's solution as previously described (28).
The extracellular medium of unstimulated cells and Locke's solution were treated as controls. Secreted medium and controls were centrifuged at 800 g for 10 min at 4°C and filtered on a 0.22 µm syringe filter to remove cells that might be present in secretions. Samples were acidified up to 0.1% trifluoroacetic acid in order to prevent degradation (23).
Mass spectra analysis-MS and MS-MS analyses were performed using electrospray mass spectrometry (ES-MS) on a Q-TOF II (Bio-Tech, Manchester, UK) in positive mode. Scanning was done from 50 to 700 m/z in 1s and calibration was performed by using phosphoric acid 0.1% in water/acetonitrile 50/50. MS and MS-MS analysis were done by nanospray of line using "NanoES spray capillaries" from Protana (Odense, Denmark). For MS-MS experiments, argon 5.5 gas was used for the collision gas and the collision energy was set to 10-30 eV. MS-MS spectra were acquired using the selection of the precursor ion by the quadrupole and fragments were analyzed by the time of flight. The absence of M6G prior to sample application was systematically controlled using MS and MS-MS mode.
Affinity chromatography-Purified anti-bovine PEBP 1-11 antibody (3 mg prepared as described in (30)) or commercial anti-human UGT2B antibody (200 µg) were coupled to a HiTrap NHS-activated HP 1ml column (Amersham Pharmacia Biotech) according to manufacturer's instructions. Antibody-coupled column was first washed with 5 volumes of buffer A (75 mM Tris-HCl pH 8.0). Samples (i.e., cytoplasm, microsome, lysosome, mitochondria, intragranular matrix) resuspended in buffer A (2 mg in 1 ml final volume) were loaded on the column and let stand for 10 min at room temperature. The column was then washed with 10 volumes of buffer A prior to elution performed with 1 ml of buffer B (150 mM NaCl, 100 mM glycine, pH 3.0). The eluate was immediately neutralized with 20 µl of buffer C (1.5M Tris-HCl, pH 8.0) prior to RP-HPLC analysis or desalting.
Control experiments on the antibodies were performed as described above.
Tryptic digestion and mass spectrometry. Silver stained bands were excised and washed with 100 µl of 25 mM NH 4 HCO 3 , dehydrated twice with 100 µl of acetonitrile and dried with a Speed Vac evaporator before reduction (10 mM DTT in 25 mM NH 4 HCO 3 ) and alkylation (55 mM iodoacetamide in 25 mM NH 4 HCO 3 ) (33). For tryptic digestion, gel pieces were resuspended in three gel volumes of trypsin (12.5 ng/µl) freshly diluted in 25 mM NH 4 HCO 3 and incubated overnight at 35°C. The digested peptides were then extracted from gel in a buffer containing 25% H 2 O, 70% acetonitrile, 5% formic acid and analyzed by matrix assisted laser desorption/ionization time-offlight (MALDI TOF) and LC/MS/MS. For nano-HPLC, a CapLC system (Micromass Ltd., Manchester, UK) was used. Samples were concentrated on a precolumn and peptides were separated on a 15 cm x 75 µm i.d. column packed with 3 µm 100 Å C18 PepMap (LC-Packings). The MS and MS/MS analyses were performed with a Q-TOF 2 hybrid quadrupole/time-of-flight mass spectrometer (Micromass Ltd., Manchester, UK). Data analysis was performed with Global Server (MicroMass, Ltd., Manchester, UK) software and Mascote (Matrix Science Ltd., London, UK) against NCBI (The National Center for Biotechnology Information) database.
M6G de novo synthesis-UGT2B was purified by affinity chromatography on an Äkta Purifier HPLC system. The eluted material was desalted on Sep Pak cartridge and analyzed by Western blot using an anti-UGT2B antibody in order to confirm the presence of UGT2B-like material. Eluted fraction was also submitted to gel migration followed by silver staining. Silver stained bands were excised and treated as described above.
Affinity-purified UGT2B-like enzyme was dried and incubated 3 h at 37°C in a reaction volume of 100 µl (100 µM UDP-glucuronic acid, 2 mM saccharolactone, 0.9 µM morphine, 50 mM Tris HCl, pH 5.5). The resulting sample was deproteinized and desalted prior to Q-TOF MS-MS analysis to examine the presence of M6G. Controls experiments were performed with the boiled affinity column eluate, the reaction buffer alone and morphine used for the assay.

Immunolabelling of morphine-like components in chromaffin cells
We extend our previous observations related to the presence of morphine and its derivatives in adrenal medulla and PC-12 (10,14-16) by investigating the subcellular localization of morphine and its derivatives in chromaffin cells by laser confocal microscopy. Morphine localization was compared with CGA immunoreactivity, a specific marker of the intragranular matrix of secretory granules (23). Morphine-like immunolabelling (Fig. 1A, green) was observed as bright dots in the cytoplasm, similar to that obtained with CGA (Fig. 1A, red).
The superimposition of the two labelled materials may suggest a partial intragranular colocalization in chromaffin granules ( Fig. 1A and B, Merge and arrows). Control experiments using anti-morphine antibody blocked with M6G prior to immunocytochemistry experiments or secondary antibody alone (Fig. 1C) were carried out to determine the specificity of labelling.

Characterization of M6G in chromaffin intragranular matrix
In order to clearly demonstrate the presence of M6G in chromaffin granules, the intragranular matrix of bovine (33 mg) was isolated and loaded on a RP-HPLC column to purify and characterize the alkaloids present. The HPLC gradient was specifically designed to separate M3G, morphine, M6G, COD and MA (500 pmoles ; Fig. 2A). Comparison of the chromatography profile of the deproteinized intragranular material with the elution profile of standards indicated the presence of a peak with the same retention time as M6G. Q-TOF MS-MS analysis of this peak allowed its unambiguous identification as M6G (462.15 Da) according to its fragmentation profile generating morphine (i.e., 286.13 Da ; Fig. 2B).

Identification of M6G secreted from nicotine-stimulated cultured chromaffin cells
In order to characterize M6G as a secretory product from stimulated chromaffin cells, alkaloid standards (390 pmoles ; Fig. 3A1) and extracellular medium were analysed by RP-HPLC. After stimulation with 10 µM nicotine, the secreted material was submitted to 1N HCl and chloroform/isopropanol extractions. Alkaloids present in the aqueous phase were separated by HPLC on a RP C18 column (Fig. 3A2). A major peak corresponding to the M6G standard was observed.
In extracellular media corresponding to basal secretions, M6G could not be detected (Fig. 3A3) demonstrating the genuine release of M6G upon nicotine stimulation.
Q-TOF MS-MS analysis of M6Gcorresponding peak unambiguously demonstrated the presence of M6G (462.15 Da) according to its fragmentation profile whereby morphine is generated (i.e., 286.13 Da; Fig. 3B). Similar analysis revealed the absence of M6G in secretions from unstimulated cells (Fig. 3C).

Evidence for interaction between PEBP and M6G
We recently reported that PEBP is present in chromaffin granules and able to translocate from cytoplasm to the intragranular space probably via a granule membrane raft-binding mechanism (23). Based on this property, we then examined the capacity of PEBP to bind intragranular morphinelike components using immunoaffinity experiments carried out on intragranular matrix, cytoplasm and microsome, mitochondria, as well as on lysosome fractions. Immunoaffinity experiments using the purified antibody directed against PEBP [1][2][3][4][5][6][7][8][9][10][11] were carried out to isolate the putative intragranular PEBP-morphine-like complex(es). Subcellular fraction extracts (20 mg each) were loaded on a HighTrap-NHS column coupled with the anti-PEBP 1-11 antibody. The eluted PEBP binding-molecules were concentrated and then purified by RP-HPLC. The chromatograms on figure 4A represent the elution profile of a mixture of 500 pmol of standards ( Fig.  4A1 ; M3G, morphine, M6G, COD and MA), as well as the purification of immunoaffinity eluate from 20 mg of intragranular matrix (Fig. 4A2), cytoplasm and microsome (Fig.  4A3), mitochondria and lysosomal fractions ( Fig. 4A4  and 4A5). Peaks marked with arrows on chromatograms representing the affinity elution of granule matrix, cytoplasm and microsomes ( Fig.  4A2 and 4A3), correspond to the elution time of the M6G standard. Only the intragranular matrix and, at a lower level, the fraction corresponding to the cytoplasm and microsomes, contain M6G-like molecules, but not morphine. MS-MS Q-TOF mass spectrometry analysis of the M6Gcorresponding peak of the elution from the granule matrix (Fig. 4A2) detected M6G (462.15 Da) and its fragmentation-derived fragment (i.e., morphine 286.13 Da ; Fig. 4B). Control experiments using MS and MS-MS mode on the elution buffer confirmed that the eluted M6G did not result from contamination (Fig. 4C).

Subcellular localization of UGT2B-like enzyme in chromaffin cells
Given the presence of M6G in secretory granules, the presence of an enzyme able to convert morphine to M6G was investigated. It has been reported that UGT2B family enzymes which act for this specific enzymatic reaction are present in the microsomal fraction of several organs including in liver (34). Immunocytochemical experiments were performed on primary cultured bovine chromaffin cells using antibodies specific to UGT2B enzymes. Confocal laser microscopy revealed UGT2B immunoreactivity as dispersed dotted pattern in cytoplasm (Fig. 5A, green labelling), that was found to colocalize partially with CGA immunoreactivity (Fig. 5A, red). At higher magnification, the colocalization of UGT2B with the granular marker CGA was further suggested particularly in cell extensions (Fig. 5B).
UGT2B labelling specificity was assessed by absorbing antibody with the commercial UGT2Bderived peptide or by incubating the second fluorescent antibody alone (Fig. 5C).
In order to confirm UGT2B-like enzyme localization, SDS-PAGE followed by Western blot analysis was performed on the intragranular matrix and microsomal fractions. In this experiment, 10 µg of each fraction was loaded on gel and immunodetection was carried on with the anti-UGT2B antibody. Analysis of the chromaffin granule matrix extract showed a strong labelling as a unique band with an apparent molecular mass of 55-60 kDa, as expected for UGT2B enzymes (34) (Fig. 6A, lane 1). One band with same molecular mass was also detected in the microsomal fraction (Fig. 6A, lane 2). Control experiments using adsorbed anti-UGT2B antibody and commercial blocking peptide confirmed antibody specificity (Fig. 6A, lane 5).
Immunoprecipitation experiments using anti-UGT2B antibodies were performed to investigate if a putative complex involving UGT2B and PEBP exist in the granule matrix. These data have indicated that a complex between UGT2B-like enzyme and PEBP exist in chromaffin granules since PEBP is co-precipitated by the antibody directed against UGT2B family (data not shown).
The physico-chemical characteristics of the UGT2B-like enzyme were examined on 2D-gel electrophoresis using 300 µg of intragranular protein material. A single board spot at 55-60 kDa and pI of 5 was observed (data not shown) suggesting the structural heterogeneity of the protein.
Evidence for M6G de novo synthesis by UGT2B-like enzyme De novo synthesis experiment was designed in order to determine whether the UGT2B-like enzyme present in intragranular matrix is able to transform morphine into M6G. The UGT2B-like enzyme was first purified from intragranular matrix using an anti-UGT2B affinity column. The resulting eluate was then separated by SDS-PAGE.
To complete this work, a duplicate gel was silver stained. A band at 55 kDa corresponding to the putative UGT2B-like enzyme, as well as other bands were observed (Fig. 7B). Thus, four additional bands were observed at 24, 26, 28 and 60 kDa. Proteomic analysis indicated the presence in the 24, 26, and 28 kDa bands of PEA 1-165 , PEA  and PEA 1-208 fragments, respectively (36) (data not shown). The proteomic analysis of the 55 kDa silver stained band did not show a similarity to a known protein. UGT2B enzyme was not identified because of the absence of bovine UGT2B in data protein data banks and the poor conservation UGT2B proteins that was described to be a barrier to identify orthologues across species (for review: (37)).
M6G de novo synthesis experiments were carried out at the intragranular pH of 5.5 (13) using the affinity eluate. After incubation, the reaction medium was deproteinized and desalted in order to analyse the presence of newly synthesized M6G. to incubation indicated the absence of M6G (data not shown). An additional control was performed by incubating with boiled affinity-eluted UGT2Blike enzyme (Fig. 7D). A 464.16 Da molecular mass component was visible that corresponded to an unidentified molecule ; it did not display the specific M6G fragmentation pattern at 286.13Da corresponding to morphine. This last control demonstrated that the formation of M6G was totally abolished upon enzyme heat inactivation.

Presence of M6G into chromaffin granules
In the present work we demonstrate for the first time that M6G (and not morphine) is present in secretory chromaffin granules of bovine adrenal medulla. Since P450 enzymes are localized in the ER membrane of these cells, we surmise that morphine is synthesized in the ER-Golgi apparatus by salutradine synthase (i.e., cytochrome P450 reductase) (38). Using together immunocytochemical experiments and biochemical techniques, M6G was identified within chromaffin granules. Some diffuse nongranular labelling can be attributed to morphinelike components present in the ER.

Presence of a PEBP-M6G complex into chromaffin granules
Recently the cytoplasmic PEBP, also named RKIP (20) has been identified in bovine adrenal gland and chromaffin cells (23,39). In addition, its presence in chromaffin intragranular matrix, secretion materials and serum has been shown by our group (23). PEBP has been described to bind various molecules, including indocyanine, phosphatidylethanolamine, bromosulfophtaleine and hormones, such as oestradiol-17β and dehydroepiandrosterone (40). It was thus postulated that PEBP might represent an organic anion transporter (OAT) with the same affinity as bovine serum albumin (40). In the present study we further demonstrate that in the chromaffin intragranular matrix, M6G is a novel endogenous ligand to PEBP. This binding data is in agreement with previous reports that described PEBP as a morphine-binding protein (22).
In our experimental conditions, it is important to point out that the quantity of M6G could not be determined precisely because a part of the M6G bound to PEBP has been lost probably during the deproteinisation step (i.e., precipitation of a part of the PEBP-M6G complex).

Presence of a UGT2B-like enzyme within chromaffin granules
Once synthesized, we surmise that morphine could bind to PEBP in the early granule stage, before its presentation to a conversion enzyme (i.e., UGT2B-like enzyme), producing the final active M6G material. Our data suggest that such a complex, comprising UGT2B-like enzyme and PEBP exists in these experimental conditions. UDP-glucuronosyltransferases (UGTs) represent a superfamily of glycosylated enzymes (35) localized into the endoplasmic reticulum (29,41) which catalyses the glucuronidation reaction of several drugs and steroid hormones. The glucuronidated products are more polar, less toxic and easier to excrete through bile or urine. According to their sequence homologies, UGTs include two classes : UGT1 and UGT2, the later subdivided in 2A and 2B components (41). To date, the UGT2B subfamily catabolize morphine into M3G and M6G (24) and amongst these, the UGT2B7 (42,43) appears to be the only one able to produce M6G from morphine. Using immunochemical techniques, we have shown the presence of UGT2B-like immunoractivity in chromaffin granules. Our data also demonstrate the presence of a glycosylation on the intragranular UGT2B, a posttranslational modification reported to occur on other UGT2B family members (35). The affinity-purified chromaffin granule UGT2B has been shown to transform morphine into M6G, suggesting for the first time its implication in a metabolic process. Up to now, this property of UGT2B has never been reported.
In our experiments, the proteomic analysis of the 55 kDa silver stained band (Fig. 7B) did not show similarity to any protein because of the lack of bovine UGT2B sequences in the protein databases or the presence of a new UGT2B variant. Indeed, 12 different UGT2B genes exist in humans (7 genes and 5 pseudogenes) and many remnant genes are also present (for review: (37)), whereas no UGT2B gene or protein were described in cow. UGT2B proteins display with 70% of similar sequences, the higher conservation being present in their C-ter domain. This low conservation was described to be a barrier to identify orthologues across species (37).

Relevance of PEBP-M6G complex upon stress situation
Since PEBP and M6G are bound together within granule matrix and are cosecreted from bovine chromaffin cells upon stimulation (see Fig.  3) (23), it is likely that the PEBP-M6G complex exists after secretion and is present into blood. A large variety of proteolytic enzymes are present in chromaffin secretory granules, acting to process precursor proteins such as PEA and chromogranins/secretogranins (36). In contrast, intragranular PEBP is highly resistant to proteolytic degradation (23). PEBP may also be unaffected by proteolysis in circulation, thus protecting M6G for bodily clearance (i.e., excretion into urine and/or bile).
Furthermore, our data suggest that the analgesia observed upon acute stress (often related, but never totally explained by enkephalin and corticoid release (44)) might also be due, totally or in part, to M6G secretion, with PEBP acting as a transporter and molecular shield. After secretion, M6G may mediate several systemic actions based on its affinity for mu opioid receptors present on the cell surface of neurons, neuroendocrine cells, endothelium and immune cells (2,(45)(46)(47).
Interestingly, data form litterature support a separate and select mu opiate receptor for M6G. Thus, Rossi et al., using an antisense probe targeting Gi alpha 1 in mammals found that both heroin-and M6G-evoked analgesia but not that induced by morphine were blocked (48,49). These results show that heroin, 6-acetylmorphine, fentanyl and etonitazine can all produce analgesia through a novel mu analgesic system which is similar to that activated by M6G (48). Antisense mapping studies on exons 1, 2 and 3 of MOR-1 (i.e., mu1 opioid receptor) in mice suggested the presence of a novel mu receptor subtype responsible for M6G analgesia that may represent a splice variant of MOR-1 (50,51). In addition we previously demonstrated that in three murine macrophage cell lines (J774.2; RAW 264.7; BAC1.2F5), the mu opiate receptor subtype is mu3 because it binds only opiate alkaloids (i.e., M6G), excluding M3G or any of the opioid peptides tested (52).
Clinical relevance of M6G for pain control It is known that M6G may have a greater affinity to mu1 opioid receptor (responsible for analgesia) than to mu2 (responsible for respiratory depression), suggesting that M6G can offer benefit as systemic analgesic (53,54). M6G is presently under Phase III trial in post-operative pain (CeNeS pharmaceutical, Cambridge, U.K.) highlighting its biological potential.
Interestingly, since 20 years, cellular therapy using adrenal chromaffin cell transplants was tested for pain management in both acute and chronic pain models (55,56). The resulting antinociceptive effect was related to the secretion of opioid peptides and catecholamines from the transplants. Taking into account the present findings and the highly potent analgesic activity of M6G, we suggest that this alkaloid also participates in the analgesia observed in those early experiments.