Activation of Endocannabinoid Biosynthesis

The HPA axis activity is homeostatically controlled by means of a negative feedback mechanism comprised of both rapid (nongenomic) and delayed (genomic) components, resulting in the reduction of CRH-induced secretion of ACTH from the pituitary gland (Dallman and Jones, 1973; Dayanithi and Antoni, 1989; Jingami et al., 1985; Keller-Wood and Dallman, 1984; Keller-Wood et al., 1988). There are conflicting reports on the genomic versus nongenomic mechanisms controlling fast feedback at the pituitary gland (Dayanithi and Antoni, 1989; Hinz and Hirschelmann, 2000). On the other hand, recent works done in acute hypothalamic slices demonstrated that GCs dose-dependently (Di et al., 2003) and within 1 min (Malcher-Lopes et al., 2006) suppressed excitatory neuronal inputs to all kinds of PVN neuroendocrine cells, including those that regulate the HPA axis, the hypothalamic-pituitary-thyroid (HPT) axis, and the circulating levels of oxytocin and vasopressin. The threshold concentration for this GC-induced suppression of excitation (GSE) was between 10 and 100 nM and the EC50 was between 300 and 500 nM (Di et al., 2003), which are physiological concentrations within the range achieved during stress-induced activation of the HPA axis. The suppression of presynaptic glutamate release was prevented by postsynaptic blockade of G protein-coupled signaling and by cannabinoid CB1 receptor antagonists, and was mimicked and occluded by exogenous cannabinoid application (Di et al., 2003, 2005; Malcher-Lopes et al., 2006), indicating that the effect was mediated by retrograde release of endocannabinoids. GSE was not mimicked by intra-cellular application of the GCs, was insensitive to "classical" cytosolic GC and mineralocorticoid receptor-specific antagonists, and was maintained when DEX was conjugated to the membrane-impermeant protein bovine serum albumin, indicating that it was mediated by a G protein-coupled membrane receptor (Di et al, 2003).

Supporting the role ofthis mechanism on the fast GC-negative feedback on the HPA axis, pretreatment of mice with the CB1 antagonist SR141716 strongly potentiated restraint-induced PVN cells activation (Fos expression)

and corticosterone release (Patel et al., 2004). On the other hand, pretreat-ment with CB1 agonists, or endocannabinoid transport inhibitor AM404, or an antagonist of fatty acid amidohydrolase (FAAH) (an enzyme that hydrolyzes endocannabinoids) significantly reduced or eliminated restraint-induced corticosterone release (Patel et al., 2004). These results indicate that a major part of the nongenomic component of GC-negative feedback operated through the CB1-mediated suppression of glutamate release from presynaptic terminals controlling PVN neuroendocrine cell excitation (Cota, 2008; Di et al., 2003, 2005).

Mass spectrometry analyses provided direct biochemical evidence that GCs robustly activate the biosynthetic pathways leading to the release of both AEA and 2-AG in the PVN even when synaptic activity is suppressed (Malcher-Lopes et al., 2006). Both the increase on tissue levels of endocan-nabinoids triggered by GCs and GSE, the synaptic effect mediated by the endocannabinoids, were completely blocked by the cytokine-like adipocyte peptide leptin (Malcher-Lopes et al., 2006). Leptin has been shown to prevent fasting-induced hyperphagia in rats by blocking the rise on hypothalamic endocannabinoid levels that normally takes place during caloric deficit (Di Marzo et al., 2001; Kirkham et al., 2002). Leptin effect on GSE was mediated by reducing cAMP levels via phosphodiesterase 3B (PDE3B) activation, fact that conduced to the subsequent demonstration that GCs trigger the synthesis and release of AEA and 2-AG through a Gas-cAMP-PKA-dependent mechanism (Malcher-Lopes et al., 2006) (Fig. 11.2). Even though the GC-induced activation of PKA may, in principle, favor the N-acyltransferase-catalyzed production of precursors for both AEA and 2-AG biosynthesis (Cadas et al., 1996; Di Marzo and Deutsch, 1998), GSE was also prevented by GF109203X, a PKC blocker (Di et al., 2003). PKC itself has been shown to suppress CB1-mediated signaling in pituitary cell line AtT-20 (Garcia et al., 1998). On the other hand, AEA is released from dorsal root ganglia sensory neurons in response to both forskolin, an adenylate cyclase stimulant, and the PKC activator phorbol-myristyl-ace-tate, even in the presence of the Ca2+ chelator BAPTA (Vellani et al., 2008). Therefore, it seems that, like PKA, PKC mediates GC-induced biosynthesis and release of endocannabinoids from the postsynaptic cells in the PVN.

The orexigenic hormone ghrelin was shown to cause retrograde release of 2-AG, but not AEA, leading to suppression of excitation in the PVN, an effect that was prevented by blockade of DAG lipase (Kola et al., 2008). Even though PKC involvement was not tested by this study, ghrelin is known to be an agonist of the growth hormone secretagogue receptor type 1a, which is a Gaq/11-PKC pathway-coupled receptor. In this case, however, intracellular application of BAPTA did block the synaptic effects of ghrelin (Kola et al., 2008). Hence, it was proposed that ghrelin causes an increase on the intracellular levels of Ca2+, leading downstream to Ca2+-and/or PKC-mediated activation of membrane DAG lipase and subsequent

Figure 11.2 A model for the nongenomic, glucocorticoid-induced biosynthesis and retrograde release of endocannabinoids from PVN neurons. Glucocorticoids act at a putative G protein-coupled membrane glucocorticoid receptor to activate both Gas-cAMP-PKA- and G^g-PLC^-PKC-DAG lipase-mediated intracellular signaling pathways. Mobilization of Ca2+ from endogenous stores, and the stimulatory crosstalk between the two pathways, makes it possible for the activity-independent synthesis and release of both AEA and 2-AG. Endocannabinoids retrogradely released from the postsynaptic neuron act at CB1 receptors to inhibit its presynaptic glutamatergic inputs, thereby suppressing neuronal activation.

Figure 11.2 A model for the nongenomic, glucocorticoid-induced biosynthesis and retrograde release of endocannabinoids from PVN neurons. Glucocorticoids act at a putative G protein-coupled membrane glucocorticoid receptor to activate both Gas-cAMP-PKA- and G^g-PLC^-PKC-DAG lipase-mediated intracellular signaling pathways. Mobilization of Ca2+ from endogenous stores, and the stimulatory crosstalk between the two pathways, makes it possible for the activity-independent synthesis and release of both AEA and 2-AG. Endocannabinoids retrogradely released from the postsynaptic neuron act at CB1 receptors to inhibit its presynaptic glutamatergic inputs, thereby suppressing neuronal activation.

2-AG release (Kola et al., 2008). The combination of results obtained for GCs and ghrelin suggests, therefore, that hormone-stimulated release of endocannabinoids (at least 2-AG) from PVN neurons may depend on both PKC and DAG lipase. In agreement with this postulation, GC-induced activation of the PI-PLC-initiated pathway, leading downstream to DAG lipase activation, has been shown in different cells (Cifone et al, 1999; Marchetti et al., 2003; Tong et al., 2004). In thymocytes (T-cell precursors), for instance, DEX rapid stimulation of DAG generation is mediated by G protein-dependent PI-PLC and PKC activation (Cifone et al., 1999). Moreover, GC-induced activation of this pathway is likely to favor the N-acyltransferase-mediated endocannabinoid biosynthesis, by stimulating the release of intracellular Ca2+ via IP3. This is especially relevant if we consider that both 2-AG and AEA were released in response to GCs by PVN neurons even in the presence of tetrodotoxin (Malcher-Lopes et al., 2006), which completely suppresses action potentials, thereby virtually abolishing activity-dependent Ca2+ influx. On the other hand, simultaneous PKA activation triggered by GCs may further enhance sn-1-DAG lipase activity in neurons in the same way it does for DAG lipase isozymes found in rod membranes (Bisogno et al., 2003; Perez Roque et al., 1998).

Although PKC blockade prevented GSE, a Gaq/11-mediated pathway does not seem to be required by GC-stimulated endocannabinoid biosynthesis, since intracellular application ofantiserum against this subunit did not block GSE (Malcher-Lopes et al., 2006). Transgenic mice lacking Gaq/11 proteins do exhibit normal basal brain levels of 2-AG (Wettschureck et al., 2006), but they show impaired activity-dependent 2-AG synthesis (Wettschureck et al., 2006). Somewhat surprisingly, these results indicate that Gaq/11 is critical for activity-dependent 2-AG biosynthesis. On the other hand, they demonstrate that there are Gaq/11-independent pathways for 2-AG biosynthesis under physiological conditions, which seems to be the case when both AEA and 2-AG are released by PVN neurons in response to GCs by an activity-independent mechanism. Many different hormone receptors activate PKC through its ''classical'' pathway in which Gaq/11 stimulates the hydrolysis ofPIP2 by PI-PLC, yielding the second messengers IP3 and DAG. IP3 acts on specific endoplasmic reticular receptors, thus evoking Ca2+ release into the cytoplasm, which along with DAG leads to the activation and translocation of cytosolic PKC to the cell membrane (Smrcka et al., 1991). Nevertheless, PI-PLCb isozymes can also be activated by Gbg-subunit, albeit somewhat less effectively than by Gaq/11-subunit in some cases (Camps et al., 1992; Gudermann et al., 1996; Hepler et al., 1993; Jin et al., 1998; Katz et al., 1992; Lee et al., 1993; Meldrum et al., 1991; Schnabel et al., 1993). The PI-PLCb isozymes have distinct sites of interaction with Gaq and Gbg, hence Gaq- and Gbg-subunits may concomitantly modulate a single PI-PLCb molecule (Rhee and Bae, 1997), which probably enhances the efficacy of Gaq-coupled receptors in activating PI-PLCb. However, there are also some Gas-coupled receptors connected to signaling cascades via Gbg-subunits. This is the case, for instance, of the b2-adrenergic receptors, which potentiates cytosolic GC receptor transactivation in a hippocampal cell line (HT22) through a Gbg-dependent pathway (Schmidt et al., 2001). Therefore, in principle, the same putative Gascoupled membrane GC receptor may activate both Gas-cAMP-PKA-mediated and Gbg-PLCb-PKC-DAG lipase-mediated pathways, leading to concomitant, activity-independent release ofboth AEA and 2-AG.

In fact, there are some indications that GCs activate PLCb. The non-genomic DEX-mediated stimulation of PI-PLC both on lymphoblastoid and in T cells is biphasic, starts within 15 s, decreases to the control levels after about 2 min and peaks again at 5 min (Bamberger et al., 1999; Graber and Losa, 1995; Tong et al., 2004). This biphasic kinetics is strikingly similar to that of different G-coupled, PLCb-linked receptors, such as a-thrombin and angiotensin II receptors (Hwang et al., 2005; Schelling et al., 1997; Wright et al., 1988). In thymocytes, the nongenomic DEX-induced activation of PI-PLC was strongly reduced by RU486, but DEX could still increase PI-PLC activity by about two fold (Cifone et al., 1999). Therefore, it seems that GCs can nongenomically stimulate PI-PLCb activity through at least two distinct signaling pathways, and one of them does not depend on the cytosolic GC receptor. Evidences also indicate that PI-PLCb is the PI-PLC isozyme that mediates 2-AG production (Hashimotodani et al., 2005; Jung et al., 2005). This is the case, for instance, of 2-AG release triggered by the activation of Gq/11 protein-coupled receptors belonging to group I metabotropic glutamate receptors (mGluR) in primary cultures of corticostriatal and hippocampal slices (Jung et al., 2005).

At this time, there is no direct evidence showing the involvement of Gbg or PI-PLCb isoform in GSE. Nevertheless, the current direct and indirect evidences discussed above support a model in which GCs act at a putative membrane G protein-coupled receptor to activate both Gas-cAMP-PKA- and Gjby-PLCjb-PKC-DAG lipase-mediated pathways, leading downstream to a mutual activation of AEA and 2-AG biosynthesis (Fig. 11.2). The activation of PI-PLCb by Gbg would allow for Gaq/11-independent intracellular raise in Ca2+, which is probably required by the N-acyltransferase-mediated formation of NArPE and sn-1-lyso-2-arachi-donoyl-PC, by the subsequent NAPE-PLD-mediated formation of AEA and PA from NArPE, and by DAG lipase-mediated 2-AG formation from DAG (Fig. 11.1). Due to the relatively low efficacy of the Gbg-mediated pathway, it is possible that PKA-mediated phosphorylation will be necessary to fully activate N-acyltransferase and put on motion downstream biosyn-thetic steps for both AEA and 2-AG. Likewise, PKC activity may be required for a complete DAG lipase activation. The PA formed by NAPE-PLD may be then converted to 2-AG by the sequential actions of PA phosphatase and DAG lipase (Sugiura et al., 2004). The removal of PA by this pathway is likely to increase the rate of NArPE and sn-1-lyso-2-arachidonoyl-PC formation, which in turn may facilitate 2-AG formation by the EGTA-insensitive PLC-mediated pathway (Di Marzo et al., 1996). Since GCs stimulate the release of both AEA and 2-AG, this helps to explain why PKC blockade prevents GSE. In partial agreement with this model, recent data from Jeffrey Tasker's laboratory demonstrated that intracellular application of an antibody against Gbg subunit reduced but did not block GSE (Di et al., 2009). Indicating that an additional mechanism is likely to be involved in PKC activation by GCs. Additionally, PKC activity has been shown to potentiate neurotransmitter-stimulated cAMP accumulation in the brain, as well as the Gas-coupled, b2-adrenoceptor-stimulated cAMP formation in rat pinealocytes (Ho et al., 1988; Karbon et al., 1986).

VI. ENDOCANNAblNOIDS MEtAbOLizAtlON

Several studies suggest that both AEA and 2-AG are internalized via a common carrier-mediated, facilitated diffusion process (Beltramo and Piomelli, 2000; Hillard and Jarrahian, 2000; Piomelli et al., 1999), which in many instances is the rate-limiting step on the inactivation of endocan-nabinoid signaling. This carrier is expressed in several cell types, including glial cells, neurons, endothelial cells, and macrophages (Hillard and Jarrahian, 2000). It has been proposed that such a transporter may also be involved on the release of endocannabinoids to the extracellular space (Hillard and Jarrahian, 2000). The transporter activity is positively modulated by nitric oxide donors (Bisogno et al., 2001; Maccarrone et al., 2000). The expression of neuronal-type nitric oxide synthase (NOS) is upregulated in neuroblastoma cells by DEX (Schwarz et al., 1998), indicating that GCs may indirectly influence endocannabinoid levels by increasing NO-mediated activation of endocannabinoid reuptake. Once internalized, both AEA and 2-AG can be hydrolyzed by the endoplasmic reticular integral membrane-bound FAAH to form AA and ethanolamine from AEA, or AA and glycerol from 2-AG (Bisogno et al., 1999a, 2002; Deutsch et al., 2001; Di Marzo et al., 1999; Goparaju et al., 1998). For AEA, this seems to be the main degrading pathway in most of the cells, including brain neurons (Cravatt et al., 2001), and it seems to be an important metabolic pathway for 2-AG in macrophages (Goparaju et al., 1998). However, in the brain, it is the cytosolic monoacyl-glycerol (MAG) lipase that responds for most (~85%) of 2-AG hydrolysis (Blankman et al., 2007; Dinh et al., 2002; Nomura et al., 2008; Saario et al., 2005).

The cytochrome p4502D6 is one of the major brain cytochrome p450 isoforms and has been implicated in neurodegeneration and psychosis (Snider et al., 2008), and has been shown to convert AEA to 20-HETE-EA and to 5, 6-, 8, 9-, 11, 12-, and 14, 15-EET-EAs in vitro with low Km values (Snider et al., 2008). The cytochrome P4504x1, which is expressed in several human tissues, including heart, brain, and cerebellum, but has an yet unknown biological function, was shown to convert AEA into 14, 15-EET-EA in vitro, whereas AA and 2-AG were not efficiently metabolized by this cytochrome, indicating its specificity (Stark et al., 2008). In the liver, P4504F2 is thought to be the isoform responsible for 20-HETE-EA formation, whereas P4503A4 was identified as the primary enzyme responsible for AEA conversion into EET-EAs in the liver (Snider et al., 2007) and, perhaps in the brain (Bornheim et al., 1995). GCs induce the expression of hepatic cytochromes from the P4503A group (Hoen et al., 2000), suggesting that GCs may also regulate endocannabinoid metabolism via cytochrome p450 in the brain.

Leukocyte 12-lipoxygenase (12-LOX) and 15-lipoxygenase (15-LOX) oxygenate 2-AG, producing the hydroperoxyeicosatetraenoic acid glyceryl esters 12-HETE-G and 15-HETE-G, respectively (Kozak et al., 2002b; Moody et al., 2001). 2-AG is metabolized by 12-LOX 40% as efficiently as AA, whereas 15-LOX oxygenates 2-AG comparably or preferably to AA. 15-HETE-G is an agonist of the peroxisome proliferator-activated receptor a, a ligand-activated transcription factor belonging to the steroid hormone receptor superfamily (Kozak et al., 2002b), which has been implicated in anti-inflammatory and antiatherogenic actions in vascular cells, and in the control of genes involved in systemic lipid metabolism (Marx et al., 2004). The brain metabolite, 12-hydroxy-AEA, produced by direct oxygenation of AEA by 12-LOX has an affinity to CB1 receptor corresponding to twice that of AEA, although less active than the parent compound in inhibiting the murine vas deferens twitch response (Hampson et al., 1995). LOX inhibitors reduce the AEA-induced activation of vanilloid receptors in the guinea-pig bronchus (Craib et al., 2001). Likewise, AEA evoked depolarization of guinea-pig vagus nerve via activation of vanilloid receptors is partially suppressed by blockade of LOX activity (Kagaya et al., 2002). However, it is not clear whether the LOX products involved in vanilloid receptor activation in these cases are produced directly from AEA or from AEA-derived AA. GCs suppress EGF-induced expression of 12-LOX in human epidermoid carcinoma A431 cells through classical GC receptor activation (Chang et al., 1995), indicating that in some tissues GCs may regulate LOX-mediated metabolism of endocannabinoids via 12-LOX.

For more than a century, paracetamol has been among the most popular medicines, being widely used as analgesic, antipyretic, and nonsteroidal antiinflammatory drug. Its mode of action, however, remained a mystery until 2005, when it was demonstrated that it indirectly causes the activation of cannabinoid receptors (Hogestatt et al., 2005). In the brain and spinal cord, paracetamol, following deacetylation to its primary amine (p-aminophenol), is conjugated with AA to form N-arachidonoylphenolamine, also known as AM404, which is a blocker for the endocannabinoid transporter. Therefore, paracetamol causes CB1 and CB2 receptors activation by preventing endo-cannabinoids reuptake and subsequent metabolization. Interestingly, AM404 is also an inhibitor of COX enzymes in the brain (Bertolini which is likely to further contribute to endocannabinoid built-up, since COX is implicated in the nonhydrolytic, metabolization of endocannabi-noids in the CNS and peripheral tissues. For instance, the vasodilator action of AEA and 2-AG in rat small mesenteric arteries is limited by local endothelial activity of FAAH, MAG lipase, or COX2 (Ho and Randall, 2007), so that inhibition of COX2 by nimesulide potentiated AEA-induced vascular relaxation (Ho and Randall, 2007). In some cases, it seems that FAAH removes AA from endocannabinoids, which are then converted to prostaglandins by COX2 (Ho and Randall, 2007). Several cannabinoids are known to elicit systemic vasodilatation, mainly via cannabinoid CB1 and vanilloid TRPV1 receptors. However, both AEA and 2-AG are also able to increase pulmonary arterial pressure by a mechanism that is not mediated by CB1, CB2, or TRPV1 receptors (Wahn et al., 2005). The unspecific COX inhibitor acetylsalicylic acid (100 mM), the specific COX2 inhibitor nimesulide (10 mM), the specific EP1 prostanoid receptor antagonist SC-19220 (100 mM), and FAAH inhibitor methyl arachidonyl fluorophosphonate (0.1 mM) all prevented this effect, indicating that endocannabinoids increase pulmonary arterial pressure via COX2 metabolites of AA (Wahn et al., 2005).

COX2 may also promote bioactive molecules synthesis using endocan-nabinoids as substrate. For instance, AEA-induced inhibition of interleukin-2 (IL-2) in primary splenocytes is not mediated by the cannabinoid CB1 or CB2 receptors, nor by the endocannabinoid membrane transporter or FAAH, but it was reduced by COX inhibitors (Rockwell and Kaminski, 2004), suggesting the involvement of oxygenated endocannabinoid products. In another example, AEA inhibited the growth of colorectal carcinoma cell lines HT29 and HCA7/C29, which express COX2, but had little effect on the SW480 carcinoma cell line, which express very low COX2 (Patsos et al., 2005). The induction of cell death by AEA in the COX2-expressing cell lines was abrogated by the COX2-selective inhibitor NS398, whereas inhibition of FAAH potentiated this effect, indicating that AEA-induced cell death was mediated via direct oxygenation of AEA by COX2 (Patsos et al., 2005). In fact, both endocannabinoids are efficiently oxygenated by COX2, but not COX1, to produce hydroxy endoperoxides analogous to AA-derived PGH2: PG-H2-ethanolamide (PG-H2-EA) and PGH2-glycerol (PGH2-G) from AEA and 2-AG, respectively. 2-AG is oxygenated by COX2 as effectively as AA, whereas AEA oxygenation by COX2 occurs with a relatively high (micromolar) Km, suggesting that this reaction may only occur in tissues in which high amounts of AEA are found (Kozak and Marnett, 2002). After COX2-mediated oxygenation, PG-H2-EA and PGH2-G can be converted into a range of prostaglandin ethanolamides (prostamides) and prostaglandin glyceryl esters (glycerylpros-taglandins), respectively (Kozak et al., 2000; Rouzer and Marnett, 2008; Woodward et al., 2008; Yu et al., 1997). In murine RAW cells, the endocannabinoid-derived prostanoids PG-H2-EA and PGH2-G serve as precursors for glycerol esters and ethanolamides of the prostaglandin E2, prostaglandin D2, and prostaglandin F2a in reactions catalyzed by the respective prostaglandin synthases (Kozak et al., 2002a). Similarly, PGH2-G and PG-H2-EA served as substrates to produce the corresponding endocannabinoid-derived prostacyclin derivatives by prostacyclin synthases. It was also demonstrated that the sequential action of COX2 and throm-boxane synthase on AEA and 2-AG could generate thromboxane A2 ethanolamide and thromboxane A2 glycerol ester, respectively. However, endocannabinoid-derived thromboxanes are produced at much reduced levels when compared with AA-derived thromboxanes, indicating that this pathway may not be as physiologically significant (Kozak et al., 2002a).

The endocannabinoid-derived prostanoids PGH2-G and PG-H2-EA and their metabolites represent a unique class of lipids, and the scope of their biological functions is only beginning to be elucidated. Zymosan, a preparation of protein—carbohydrate complexes from yeast cell wall, which is used to induce experimental sterile inflammation, has been shown to stimulate PGH2-G synthesis by resident peritoneal macrophages (Rouzer and Marnett, 2005), indicating its involvement in innate immune and inflammatory responses. However, prostamides and glycerylprostaglandins do not interact with any known natural or recombinant prostanoid receptors involved in such responses. In most cases, their biological effect can be achieved with very small concentrations, implicating high-affinity interaction between these molecules and unidentified specific receptors. In RAW macrophage cells, for example, picomolar concentrations of PGE2-G dose-dependently induces Ca2+ mobilization from endoplasmic reticulum stores and activation ofPLCb-PI3-PKC pathway, leading to the phosphorylation ofERK1 and ERK2 (Nirodi et al., 2004). PGE2-G and its chemically stable synthetic analog PGE2-G-serinolamide are potent reducers of intraocular pressure in dogs (Woodward et al., 2008), whereas the prostamides PGF2a-EA, PGD2-EA, and PGE2-EA are potent inducers of cat iris sphincter contraction (Matias et al., 2004). Subsequent work has demonstrated that the compound AGN204396 blocked prostamides-induced, but not PGE2-G-induced iris contraction, indicating separate receptors for prostamides and glycerylprostaglandins (Woodward et al., 2007). PGE2-G was detected in vivo and was shown to induce mechanical allodynia and thermal hyperalgesia in rats (Hu et al., 2008). Even though PGE2-G can be quickly metabolized into PGE2, these effects were not blocked by prostanoid receptors antagonists (Hu et al., 2008). Concerning pain sensitivity, these are the opposite effects ofthe parent endocannabinoid 2-AG, indicating that COX2 can modulate pain sensation by oxygenating 2-AG (Hu et al., 2008). Given the regulation of COX2 expression by GCs, this implies that COX2 activity may represent a context-sensitive enzymatic switch that converts the anti-inflammatory and antinociceptive mediator 2-AG into a pronociceptive and proinflammatory prostanoid.

Was this article helpful?

0 0

Post a comment