1 Introduction

The concept that arrestin association with G-protein-coupled receptors (GPCRs) does more than cause homologous receptor desensitization grew out of the realization that arrestin recruitment by the β2-adrenergic receptor (β2AR) resulted in Src tyrosine kinase activation and phosphorylation of extracellular signal-regulated kinase (ERK1/2) (Daaka et al. 1997; Luttrell et al. 1999). Subsequently, Miller et al. (2003) found that the cytokine receptor US28 (a GPCR encoded by the human cytomegalovirus) activates p38 mitogen-activated protein kinase (MAPK) through a G-protein receptor kinase (GRK) and arrestin-dependent mechanism. In addition, Sun et al. (2002) found that the chemotaxic response of HeLa and HEK cells to cytokines mediated by the CXCR4 receptor also required p38 MAPK activation through receptor phosphorylation and arrestin-3 recruitment. Parallel studies showed that the GPCR-arrestin signaling complex activates c-Jun N-terminal kinase (JNK), also through a physical scaffolding mechanism (McDonald et al. 2000; Breitman et al. 2012). These observations lead to the concept that the arrestins can form a scaffold that physically links the GPCR to the three different MAPK signaling cascades: ERK1/2, p38 MAPK, and JNK (Burack and Shaw 2000; Pearson et al. 2001; DeWire et al. 2007).

The steps linking arrestin activation to p38 MAPK phosphorylation have not been fully visualized, but a requirement for apoptosis signal-regulating kinase 1 (ASK1), also known as mitogen-activated protein kinase kinase kinase 5 (MAP3K5), was suggested by the ability of dominant-negative mutant of ASK1 to block p38 MAPK activation (Sun et al. 2002). A plausible model for p38 activation suggests that activated arrestin forms a scaffold containing the required sequential cascade of the three kinases typically involved in a MAPK activation: a MAPKKK (possibly ASK1) activating a MAPKK (possibly MEK3 or MEK6), which in turn activates p38 MAPK (Burack and Shaw 2000; Pearson et al. 2001; Dewire et al. 2007). Presumably, arrestin association with the GRK-phosphorylated GPCR induces a structural rearrangement within arrestin–kinase complex, thereby facilitating the sequential phosphorylation reactions. However, the details of this cascade and differences between the cascades in different cell types and subcellular compartments have not yet been resolved.

Arrestin-dependent p38 MAPK activation results in a range of cellular and behavioral responses. In addition to mediating chemotaxic responses to cytokines (Sun et al. 2002), activation of p38 MAPK via arrestin association regulates apoptosis in mouse embryonic fibroblasts (Yang et al. 2012) and mediates endothelin-induced cell migration of mouse aortic smooth muscle cells (Morris et al. 2012). Arrestin-mediated p38 activation also induces hypertrophy and proliferation of GFAP-immunoreactive astrocytes in the spinal cord and brain (Bruchas et al. 2006; Xu et al. 2007). In addition, we recently found that arrestin-dependent p38 activation plays a key role in the behavioral stress response, and it is the cellular details of this signaling cascade that we would like to summarize in this chapter.

In essence, our studies in mice have shown that:

  1. 1.

    Corticotropin-releasing factor (CRF) is released in the brain and hypothalamus in response to stress exposure.

  2. 2.

    CRF acts broadly in brain to coordinate the physiological, adaptive response to stress necessary for survival.

  3. 3.

    One of the cellular responses to CRF is the stimulated release of the endogenous dynorphin opioid peptides (Land et al. 2008; Bruchas et al., 2009).

  4. 4.

    Dynorphins selectively activate the kappa opioid receptors (KOR), which are Gi/o-coupled GPCRs (Chavkin et al. 1982; Bruchas et al. 2007a, b; Land et al. 2008; Bruchas et al. 2011; Lemos et al. 2012).

  5. 5.

    Sustained KOR activation results in GRK3-mediated phosphorylation of Ser369 in rodent KOR and subsequent arrestin-3 recruitment (Bruchas et al. 2006).

  6. 6.

    The KOR–arrestin complex initiates the phosphorylation and activation of p38α MAPK at multiple sites within the brain (Bruchas et al 2011; Lemos et al. 2012; Schindler et al. 2012).

  7. 7.

    p38α activation at one of these sites (the nerve terminals of the serotonergic neurons projecting from the dorsal raphe nucleus to the ventral striatum) causes the translocation of the serotonin transporter (SERT; SLC6A4) from an endosomal compartment to the nerve terminal surface (Bruchas et al. 2011; Schindler et al. 2012).

  8. 8.

    Increase in surface expression of SERT pumps serotonin (5HT) back into the nerve terminal more efficiently and thereby produces a transient hyposerotonergic state in the ventral striatum (Bruchas et al. 2011; Schindler et al. 2012).

  9. 9.

    The reduction in 5HT tone in the ventral striatum contributes to the stress-induced dysphoria evident as behavioral aversion in the mice exposed to the stressful experience (Land et al. 2008; Bruchas et al. 2011; Schindler et al. 2012).

Evidence supporting this proposed cascade is summarized below.

2 Kappa Opioid Receptors

Kappa opioid receptors are members of the Gi/o-coupled superfamily of G-protein-coupled receptors (Bruchas and Chavkin 2010). Consistent with the data concerning other members of this class, kappa receptor activation results in a broad range of signaling events including membrane-delimited Gβγ-mediated regulation of calcium and potassium conductances. Gβγ released by kappa receptor activation increases G-protein-gated inwardly rectifying potassium channel (Kir3) activation and positively shifts the activation threshold of voltage-sensitive calcium channels (VSCC), thereby inhibiting calcium conductance (Werz and Macdonald 1984; Cherubini and North 1985; Herlitze et al. 1996). The net effect of these membrane-delimited effects on ion conductance is to reduce somatic excitability and calcium influx at the nerve terminal; kappa receptor activation has been shown to presynaptically inhibit the release of a broad range of neurotransmitters through these ionic mechanisms (Grudt and Williams 1995; Simmons and Chavkin 1996).

G-protein stimulation by kappa receptors also activates a variety of kinases including ERK1/2, JNKs, PKC, and p38 MAPKs in receptor-transfected cells, primary cultures of neurons and astrocytes, and in kappa opioid receptor expressing neurons in the brain (see Bruchas and Chavkin 2010). These activated kinases phosphorylate a variety of substrates, including transcription factors to regulate gene expression and various cytoplasmic proteins to affect neuronal physiology (to be described further below).

Again, like other G-protein-coupled receptors, agonist-activated kappa receptors are substrates for GRKs, which phosphorylate specific serine residues in the carboxy-terminal domain of the receptor [i.e., serine-369 in the rodent kappa receptor (rKOR) and the homologous residue serine-358 in the human kappa receptor (hKOR) sequence] (Appleyard et al. 1999; Li et al. 2002; McLaughlin et al. 2003; Schattauer et al. 2012). Arrestin binding to the GRK-phosphorylated kappa receptor sterically inhibits further G-protein activation and results in homologous desensitization of membrane-delimited signaling, but arrestin association is required for the late phase of ERK1/2 activation (Bruchas et al. 2008; McLennan et al. 2008) and for p38 MAPK activation by kappa receptors (Bruchas et al. 2006; Xu et al. 2007). Thus, arrestin binding to kappa receptors shifts agonist signaling from membrane-delimited pathways to alternative effector pathways.

With this wide range of possible cellular signaling responses, it should not be surprising that the ability of a kappa agonist to activate one pathway (its efficacy) does not need to be the same as for all the different pathways. Ligand-directed signaling differences have been documented in other GPCR systems (Urban et al. 2007) (see Chap. 3). Based on these insights, kappa ligands can be conceptually divided into (1) strong agonists (able to activate all of the Gβγ- and arrestin-dependent signaling events), (2) weak agonists (able to activate Gβγ-, but not arrestin-dependent signaling events), (3) neutral antagonists (that bind receptor but do not evoke any signaling responses), and (4) collateral agonists (that bind to kappa receptors to activate one of the alternative signaling pathways without activating Gβγ-dependent responses) (Fig. 1). Arrestin-biased agonists (see Chap. 3) at the kappa receptor that activate arrestin-dependent signaling without efficiently activating Gβγ signaling could be postulated by analogy to the parathyroid hormone and angiotensin II receptors (Gesty-Palmer et al. 2009; Violin et al. 2010); however, examples of this type of ligand have not yet been characterized.

Fig. 1
figure 00141

Ligand-directed signaling differences between kappa opioids. Strong agonists, like dynorphin, activate kappa receptors to stimulate Gβγ-dependent responses including presynaptic inhibition of transmitter release (calcium channel inhibition) and somatic membrane hyperpolarization (potassium channel activation). Sustained kappa receptor activation results in the phosphorylation of specific serine residues in the carboxy-terminal domain of the receptor and subsequent arrestin (green symbol) recruitment. The resulting arrestin activation enables p38 MAPK activation (phosphorylation), and the cellular consequences include astrocyte activation, Kir3 potassium channel phosphorylation (and deactivation), and SERT translocation. At the behavioral level, presynaptic inhibition of transmitter release underlies the analgesic responses, and SERT translocation mediates dysphoria and proaddictive responses. In contrast, long-acting kappa antagonists, like norBNI, cause receptor inactivation through a c-Jun N-terminal kinase (JNK)-dependent mechanism without stimulating Gβγ-dependent responses

Full size image

Dynorphin peptides, salvinorin A, U50,488, U69,593, and enadoline are prominent members of the first category. Buprenorphine, naloxone, and naltrexone are examples of neutral antagonists (although they lack kappa receptor selectivity). A ligand that activated Gβγ signaling but did not efficiently stimulate GRK would be expected to activate the membrane-delimited signaling but not the arrestin-dependent responses (Chavkin 2011). 6′GNTI has been suggested as an example of a G-protein-biased kappa receptor agonist that does not recruit arrestin (Rives et al. 2012). An example of this type of pathway-selective ligand in a different receptor system is morphine, which is a strong opioid analgesic acting through mu opioid receptors, but does not efficiently activate arrestin-dependent responses (Dang and Christie 2012). norBNI and JDTic are examples of the latter category; the selective kappa ligands norBNI and JDTic do not activate Gβγ- or arrestin-dependent pathways, but do effectively activate JNK pathways upon kappa receptor binding (Bruchas et al. 2007a, b; Melief et al. 2010).

3 Stress-Induced Release of Dynorphin Increases Phospho-p38 MAPK in a GRK3- and Arrestin-3-Dependent Manner

Efforts to understand opioid receptor tolerance mechanisms entered a new molecular biology phase after the delta opioid receptors were cloned by Kieffer and Evans in 1994, and the mu and kappa sequences were deduced shortly afterwards (Akil et al 1996). In a series of site-directed mutagenesis studies using Xenopus oocyte expression, the serine-369 residue in the carboxy-terminal domain was found to be the critical GRK phosphorylation site required for homologous rKOR desensitization (Appleyard et al. 1999). To determine if this phosphorylation event also regulated kappa opioid signaling in vivo, we generated a phospho-selective antibody, KOR-p, that could distinguish phosphorylated KOR-pSer369 from the unphosphorylated receptor (McLaughlin et al. 2003). Importantly, the increase in KOR-p immunoreactivity induced by the kappa agonist U50,488 was not evident in GRK3−/− mice. The selective role of GRK3 (without compensation by other GRK isoforms) was a surprise. Mice lacking GRK3 showed reduced analgesic tolerance to U50,488 (McLaughlin et al. 2003).

The high degree of cellular resolution of the immunohistochemical KOR-p staining provided a new opportunity to detect sites of dynorphin action in the brain, and we next adopted a partial sciatic nerve ligation method previously shown by Porecca and colleagues to evoke endogenous dynorphin release (Wang et al. 2001). We found that KOR-p immunoreactivity was increased in the spinal cord following partial sciatic nerve ligation in wild type, but not in KOR−/−, prodynorphin−/−, or GRK3−/− mice (Xu et al. 2004). Sustained dynorphin release following nerve ligation produced tolerance to the analgesic effects of U50,488, but nerve ligation did not produce tolerance in GRK3−/− or prodynorphin −/− mice (Xu et al. 2004). These results established that kappa opioid receptor desensitization occurred both in vivo and in vitro through a GRK-/arrestin-dependent mechanism.

4 Astrocyte Activation by Dynorphin Occurs Through an Arrestin/p38 MAPK Mechanism

One of the striking features of nerve ligation is that it causes the robust activation of astrocytes, as documented by the increase in number of GFAP-immunoreactive cells in the spinal cord. However, we were surprised to observe that the increased GFAP immunoreactivity was not evident in prodynorphin−/− or GRK3−/− mice and that the activation of astrocytes by nerve ligation could be blocked by the p38 MAPK inhibitor SB 203580 (Xu et al. 2007). Using KOR-transfected AtT20 cells, we found that kappa receptor stimulation increased phospho-p38 immunoreactivity and that the increase could be blocked by a dominant-negative form of arrestin but not evident if kappa receptor phosphorylation was blocked by alanine substitution for Ser369 in KOR (Bruchas et al. 2006). A GRK3-/arrestin-dependent mechanism of p38 activation in astrocytes stimulated by kappa agonists in vivo and in vitro was also documented by confocal imaging and Western blot analysis (Bruchas et al. 2006; Xu et al. 2007). p38 MAPK activation was not evident in ether striatal astrocytes or neurons isolated from KOR−/− or GRK3−/− mice, and cultured striatal astrocytes pretreated with siRNA for arrestin-3 were also unable to activate p38 in response to U50,488 treatment (Bruchas et al. 2006). McLennan et al. (2008) also found that proliferation of immortalized astrocytes in culture could be stimulated by kappa opioids in a Gβγ- and arrestin-dependent manner. They attributed these effects to pERK activation—not p38; however, in a subsequent study, they reported that both ERK and p38 pathways stimulated oligodendrogenesis in a similar culture system (Hahn et al. 2010). Extending these findings, we found that forced swim stress also activates GFAP-immunoreactive astrocytes in hippocampus and cortex by stimulating this dynorphin-KOR-GRK3-arrestin ⟹ phospho-p38 MAPK cascade (Messinger and Chavkin unpublished observations).

5 Kappa Receptor Activation of Arrestin/p38 MAPK Regulates the Potassium Channel Kir3

Prior studies showed that tyrosine phosphorylation in the N-terminal cytoplasmic domain of the G-protein-gated inwardly rectifying potassium channel, Kir3.1, facilitates channel deactivation by increasing the intrinsic GTPase activity of the channel (Ippolito et al. 2002, 2005). Dynorphin released during forced swim stress or following sciatic nerve ligation also resulted in tyrosine phosphorylation of Kir3.1 at these regulatory residues (Clayton et al. 2009). Channel phosphorylation in the dorsal horn of the spinal cord of nerve-ligated mice required GRK3 phosphorylation of the kappa opioid receptor, arrestin recruitment, and subsequent p38 MAPK activation (Clayton et al. 2009). Whole cell voltage clamp of AtT20 cells expressing kappa receptors demonstrated that p38 activation reduced the potassium current through a Src kinase-dependent mechanism; the enhanced channel deactivation could be blocked by the Src inhibitor PP2. Similar mechanisms also regulate Kir3 current in serotonergic neurons of the dorsal raphe nucleus in the brain (Lemos et al. 2012). Acute activation of kappa receptors in these neurons increases potassium conductance through G-protein-gated inwardly rectifying channel, but sustained kappa receptor activation by repeated stress exposure causes channel phosphorylation and subsequent channel inactivation through the arrestin-dependent p38 MAPK mechanism (Lemos et al. 2012).

6 Kappa Receptor Activation of Arrestin/p38 MAPK Activates the Serotonin Transporter

Selective kappa agonists produce feelings of dysphoria in humans and aversion responses in experimental animals (Pfeiffer et al. 1986; Shippenberg and Herz 1986). Stress-induced release of the endogenous dynorphin opioid peptides selectively activates kappa opioid receptors and produces dysphoria in experimental animals (McLaughlin et al. 2006; Bruchas et al. 2007a, b; Land et al. 2008). The dysphoria caused by stress-induced activation of the dynorphin-kappa opioid systems results in a potentiation of the rewarding valence of cocaine and reinstatement of extinguished cocaine drug seeking, which may help explain how stress increases the risk of drug addiction.

We used a conditional gene deletion approach to define the molecular events responsible for these behavioral responses. Using mice having lox-p excision sequences flanking the p38α MAPK, we found that selective inactivation of p38 signaling in serotonergic neurons of the dorsal raphe nucleus blocked defeat-induced social aversion and stress-induced reinstatement of cocaine place preference (Bruchas et al. 2011). In addition, selective excision of p38α MAPK in serotonergic neurons blocked stress-induced potentiation of cocaine place preference (Schindler et al. 2012). These behavioral responses were each caused by stress-induced dynorphin release, kappa opioid receptor activation, GRK3-dependent kappa receptor phosphorylation, and subsequent arrestin recruitment and activation. Previous reports had demonstrated a role for p38 MAPK in the modulation of the plasma membrane serotonin transporter (SERT, SLC6A4) function in vitro (Zhu et al. 2004, 2005; Samuvel et al. 2005), and using a cell-surface biotinylation and Michaelis-Menten kinetic analysis of 5HT transport, we found that stress-induced activation of p38α in serotonergic neurons causes SERT translocation from a cytoplasmic endosomal compartment to the cell surface (Bruchas et al. 2011; Schindler et al. 2012). Although the dorsal raphe sends afferent projections broadly throughout the forebrain, dynorphin-dependent SERT translocation was evident only in the serotonergic projection to the ventral striatum (Schindler et al. 2012). These findings suggest that stress-induced dysphoria mediated by arrestin/p38 MAPK activation is caused by a transient hyposerotonergic state in the nucleus accumbens (Fig. 2).

Fig. 2
figure 00142

Stress exposure regulates serotonergic neurotransmission. Dynorphin opioid peptide release from medium spiny neurons (MSN) in the ventral striatum activates kappa opioid receptors (KOR) expressed on the terminals of the serotonergic neurons innervating the nucleus accumbens. Through a G-protein receptor kinase 3 (GRK3)-dependent mechanism, arrestin activates p38α MAPK in the nerve terminals, thereby increasing cell-surface expression of the serotonin transporter. The transient hyposerotonergic state in the nucleus accumbens likely contributes to the dysphoria underlying stress-induced aversion and stress-induced potentiation of the rewarding valence of abused drugs

Full size image

7 Conclusions

These studies show that kappa receptor activation by either endogenous dynorphin release in vivo or pharmacological activation in vitro causes p38 MAPK activation through a GRK3- and arrestin-dependent mechanism. p38 MAPK is likely to have a broad range of substrates and to regulate a diverse group of processes. Several of these have been identified, but others are plausible. Several important questions have not been resolved, including the characterization of specific signaling steps linking kappa receptor activation of arrestin to p38 MAPK activation and differences in signaling in different cell types and subcellular compartments. Since arrestin/p38 signaling is essential for the dysphoric and proaddictive effects of kappa opioids, but not for their analgesic effects, we expect that pathway-selective kappa agonists, which need to be identified, will have therapeutic advantages.