CRF receptor 1 regulates anxiety behavior via sensitization of 5-HT2 receptor signaling.

Stress and anxiety disorders are risk factors for depression and these behaviors are modulated by corticotrophin-releasing factor receptor 1 (CRFR1) and serotonin receptor (5-HT(2)R). However, the potential behavioral and cellular interaction between these two receptors is unclear. We found that pre-administration of corticotrophin-releasing factor (CRF) into the prefrontal cortex of mice enhanced 5-HT(2)R-mediated anxiety behaviors in response to 2,5-dimethoxy-4-iodoamphetamine. In both heterologous cell cultures and mouse cortical neurons, activation of CRFR1 also enhanced 5-HT(2) receptor-mediated inositol phosphate formation. CRFR1-mediated increases in 5-HT(2)R signaling were dependent on receptor internalization and receptor recycling via rapid recycling endosomes, resulting in increased expression of 5-HT(2)R on the cell surface. Sensitization of 5-HT(2)R signaling by CRFR1 required intact PDZ domain-binding motifs at the end of the C-terminal tails of both receptor types. These data suggest a mechanism by which CRF, a peptide known to be released by stress, enhances anxiety-related behavior via sensitization of 5-HT(2)R signaling.

INTRODUCTION

Anxiety and major depressive disorder often present as co-morbid disorders and the expression and severity of these disorders is commonly associated with stressful experiences1. In response to stress, corticotropin releasing factor (CRF) regulates the activity of hypothalamic-pituitary-adrenal (HPA) axis and triggers changes in other neurotransmitters systems, such as serotonin (5-HT)2-6. CRF is also known to influence anxiety responses and CRF receptor 1 (CRFR1) may be particularly important in this regard7-9. 5-HT also has diverse functional effects in the central nervous system, as well as in the periphery and plays an important role in modulating depression and anxiety-related behaviours in humans and rodents10,11. In particular, pharmacological studies and knockout mice have demonstrated that 5-HT2A and 5-HT2C receptors contribute to anxiety and are pharmacological targets for the treatment of anxiety2,12-17. The targeted deletion of either the 5-HT2AR, 5-HT2CR or CRFR1 in mice is also associated with a reduction in anxiety-related behaviour12,13,18. However, little is known about the molecular mechanisms underlying the cross talk between these two important neurotransmitter systems at the cellular level.

CRF is a 41 amino acid peptide that activates the HPA axis to regulate adrenocorticotropin secretion by the pituitary gland in response to acute and chronic stress19,20. CRF peptide acts through two subtypes of Gs-coupled G protein-coupled receptors (GPCRs) resulting in increased intracellular cAMP formation21,22. Besides its endocrine function in the pituitary, CRF is also involved in a wide variety of effects not related to its pituitary activity indicating it also functions as either a neurotransmitter or neuromodulator in the brain. Consistent with its role as a neurotransmitter, CRF immunoreactive terminals, CRF binding sites and CRF receptor mRNA are widely distributed in areas of the brain that are unrelated to endocrine function23-25. There are also fifteen genes encoding functional serotonin receptors (5-HTR) in the mammalian brain that are classified into 7 families (5-HT1 to 5-HT7), all of which are GPCRs except for 5-HT3Rs which are ionotropic receptors26.

The 5-HT2 and CRF receptors each contribute to the regulation of anxiety behaviors and stress responses and CRF treatment is demonstrated to prolong 5-HT regulation of GABAergic inhibitory transmission27. The molecular and cellular basis for the action of CRF on 5-HT signaling remains unknown, as agents that increase cAMP accumulation do not mimic the effect of CRFR activation27. Therefore, in the present study we tested the hypothesis that CRFR1 activation would increase 5-HT2R-mediated signal transduction. In addition to the well characterized mechanism by which CRF can stimulate 5-HT release from serotonergic neurons to modulate anxiety6,7, we report that CRFR1 activation sensitized 5-HT2 receptor signaling by promoting the recruitment of constitutively internalized 5-HT2 receptor to the cell surface. This new mechanism of 5-HT2R regulation was physiologically relevant as the pre-administration of CRF into the prefrontal cortex of mice significantly enhanced subsequent 5-HT2 receptor-stimulated anxiety-related behaviour. This effect was blocked by a 5-HT2A receptor-selective antagonist. Taken together, our data provide a novel mechanism by which CRFR1 endocytosis and recycling can sensitize 5-HT2R-mediated signaling and anxiety-related behaviours.

RESULTS

CRFR1 activation enhances 5-HT2R signaling

The signaling of both 5-HT2A/C and CRF receptors is linked to the regulation of anxiety behaviors and CRFR activation has previously been shown to modulate 5-HT2R signaling by an unknown mechanism27. Therefore, we examined the mechanism by which CRFR1, a receptor coupled Gαs-stimulated cAMP accumulation, might alter the signaling of Gαq/11-coupled receptors (5-HT2AR and 5-HT2CR) that stimulate increases in inositol phosphate formation. In our initial studies, we utilized human embryonic kidney (HEK 293) cells that do not express endogenous CRFR1 or 5-HT2Rs to examine whether CRFR1 activation altered 5-HT2R signaling. In HEK 293 cells transfected to express either 5-HT2AR or 5-HT2CR in the absence of CRFR1, the treatment of cells with increasing concentrations of 5-HT resulted in a dose-dependent increase in inositol phosphate formation and pretreatment with CRF had no effect on the dose-response curves for inositol phosphate formation for either receptor (Fig. 1a, Supplementary Table 1). However, in cells expressing either 5-HT2AR or 5-HT2CR along with CRFR1, CRF pretreatment (500 nM) for 30 min increased the maximum efficacy (EMAX) for both 5-HT2AR- and 5-HT2CR-stimulated inositol phosphate formation by 40 ± 4.7% and 47 ± 5.5%, respectively (Fig. 1b,c, Supplementary Table 1). The increase in 5-HT2R-mediated inositol phosphate formation observed following CRF pretreatment was not attributable to CRFR1-mediated inositol phosphate formation, as CRF treatment of HEK 293 cells for 30 min did not result in inositol phosphate formation in cells expressing the 5-HT2CR alone, CRFR1 alone or expressing both receptors (Fig. 1d, Supplementary Table 1). To determine whether the observed enhancement in 5-HT2R signaling was specific to CRFR1, we examined whether the coexpression and activation of another Gαs-coupled GPCR also increased 5-HT2R signaling. However, in HEK 293 cells expressing both the β2-adrenergic receptor (β2AR) and 5-HT2AR, isoproterenol (100 μM) pretreatment had no effect on the magnitude of 5-HT2AR-stimulated inositol phosphate responses (Fig. 1e, Supplementary Table 1). Similarly, in cells co-expressing CRFR2 and 5-HT2AR, CRF pretreatment did not increase 5-HT2AR-stimulated inositol phosphate responses (Fig. 1f, Supplementary Table 1). When we examined whether the activation of the 5-HT2AR might increase CRFR1-mediated cAMP formation, we found that 5-HT (10 μM) pretreatment had no effect on CRFR1 responsiveness (Fig. 1g). In addition, we examined the effect of inhibiting either cAMP-dependent protein kinase (PKA) or protein kinase C (PKC) that are activated by CRFR1 and 5-HT2R, respectively and found that inhibition of either kinase had no effect on CRFR1-mediated increases in 5-HT2CR signaling (Supplementary Fig. 1). Thus, CRFR1 activation lead to increased 5-HT2R signaling and this increased 5-HT2R signaling was unique to CRFR1 and could not be mimicked by another Gαs-coupled GPCR.

Figure 1

Effect of CRFR1 activation on 5-HT2R signaling.

Dose response curves for 5-HT-stimulated inositol phosphate (IP) formation in HEK 293 cells pretreated with and without CRF (500 nM) for 30 min in cells transfected with (a) either FLAG-5-HT2AR and FLAG - 5-HT2cR alone, (b) FLAG-5-HT2AR and HA-CRFR1, or (c) FLAG-5-HT2cR and HA-CRFR1. (d) Basal and agonist-stimulated inositol formation in cells expressing FLAG-5-HT2cR alone, HA-CRFR1 alone, or expressing both FLAG-5-HT2cR and HA-CRFR1. Cells were treated with 500 nM CRF with or without a subsequent exposure to 10 μM 5-HT for 30 min. (e) Dose response curves for 5-HT stimulated inositol phosphate formation in HEK 293 cells transfected with FLAG-5-HT2AR and β2AR and pretreated with and without 100 μM isoproterenol (Iso) for 30 min. (f) Dose response curves for 5-HT stimulated inositol phosphate formation in HEK 293 cells transfected with FLAG-5-HT2AR and CRFR2 and pretreated with and without 500 nM CRF for 30 min. (g) Dose response curves for CRF-stimulated cAMP formation in HEK 293 cells transfected with FLAG-5-HT2AR and HA-CRFR1 and pretreated with and without 10 μM 5-HT for 30 min. The data represent the mean ± S.E.M. for 3-6 individual experiments.

It was essential to establish whether the augmented 5-HT2R signaling in response to CRF was observed in prefrontal cortical neurons. Therefore, we first examined whether both receptors were expressed in neurons from the prefrontal cortex of mice. Mouse prefrontal cortical slices were stained with polyclonal antibodies that recognized either endogenous 5-HT2AR or CRFR1 (Fig. 2a,b) and Hoechst (Fig. 2c) to mark cell nuclei. We found that a subpopulation of neurons in the prefrontal cortex stained positive for both 5-HT2AR and CRFR1 protein (Fig. 2d,e). The specificity of the 5-HT2AR antibody was confirmed in parallel Western blot and immunofluorescent studies of prefrontal cortex from 5-HT2AR knockout mice. CRFR1 antibody specificity was confirmed in HEK293 cells expressing HA-CRFR1 (Supplementary Data Fig. 2).

Figure 2

Effect of CRFR1 activation on 5-HT2R signaling in neurons.

Shown are representative laser scanning confocal micrographs demonstrating the coexpression of endogenous (a) 5-HT2AR (green) and (b) CRFR1 (red) in 30 μm neuronal slice derived from prefrontal cortex of C57/BL6 mice. Neurons are also stained for (c) neuclei (Hoechst). (d) Shown is the colocalization of the 5-HT2AR and CRFR1 in a subpopulation of neurons (dashed circles). (e) Shown is a magnified view of 5-HT2AR and CRFR1 coloclization in a subpopulation of neurons (dashed circles) in dashed box in Fig. 2d. Cortical layers are identified with roman numerals. Bar = 50 μm.

We next found that CRF (500 nM) pretreatment of mouse neuronal cultures for 30 min significantly increased 5-HT (50 μM)-stimulated [3H]-myo-inositol conversion to inositol phosphate. Importantly, in slices prepared from prefrontal cortex CRF pretreatment increased 5-HT-stimulated inositol phosphate formation by 2.3 ± 0.2 fold and when the 5-HT2A/CR selective agonist 2,5-dimethoxy-4-iodoamphetamine (DOI; 10 μM) was used, CRF pretreatment increased inositol phosphate formation by 1.5 ± 0.2 fold. Thus, consistent with what we observed in an overexpression system the pretreatment of endogenous CRF receptor increased 5-HT/DOI-stimulated inositol phosphate formation in prefrontal neuronal cultures and tissue.

Mechanism underlying CRF-mediated increases in 5-HT2R signaling

The sensitization in 5-HT2R signaling was unique to CRFR1 and was independent of the activity of second messenger-dependent protein kinases activated by either receptor (Supplementary Fig. 1). Therefore, we examined whether agonist-stimulated CRFR1 internalization contributed to the sensitization of 5-HT2R signaling. First, we tested whether the expression of a dominant-negative inhibitor of clathrin-mediated endocytosis (dynamin I-K44A) altered CRFR1-mediated increases in 5-HT2AR signaling in HEK 293 cells. We found that dynamin I-K44A expression completely eliminated CRFR1-dependent increases in 5-HT2AR-stimulated inositol phosphate formation following CRF pretreatment (Fig. 3a). Previous studies have demonstrated that CRFR1, 5-HT2AR and 5-HT2CR are internalized28,29. Therefore, we examined the localization of HA-epitope tagged CRFR1 and FLAG-epitope tagged 5-HT2R that were immunofluorescently labeled at the cell surface at 4°C and then allowed to warm to 37°C in both HEK 293 cells and rat cortical neurons. We found that both FLAG-5-HT2AR (Fig. 3b) and FLAG-5-HT2CR (Fig. 3c) were internalized from the cell surface in the absence of agonist, whereas no constitutive endocytosis was observed for the HA-CRFR1 (Fig. 3b,c). Similarly, in transfected neurons FLAG-5-HT2AR, but not CRFR1 was observed to internalize from the cell surface in the absence of agonist treatment (Fig. 3d). In contrast, when rat cortical neurons were warmed to 37°C and treated with 100 nM CRF both HA-CRFR1 and FLAG-5-HT2AR (untreated) were endocytosed and were colocalized within the same intracellular vesicles (Fig. 3e). Similar to what was observed for the HA-CRFR1, agonist-stimulated HA-β2AR also colocalized with FLAG-5-HT2AR in vesicles after isoproterenol treatment (Fig. 3f), but this does not translate into an alteration in 5-HT2AR signaling (Fig. 1e). We also found that HA-CRFR1 and FLAG-5-HT2AR were colocalized to both Rab5- and Rab4-positive endocytic organelles (Supplemental Data Fig. 3). Thus, not only was the localization of the 5-HT2R between the cell surface and intracellular compartments of cell dynamically regulated, CRFR1 endocytosis was required for the sensitization of 5-HT2R responses to agonist.

Figure 3

Role of endocytosis in CRFR1-dependent augmentation of 5-HT2R signaling.

(a) Dose response curves for 5-HT stimulated inositol phosphate (IP) formation in HEK 293 cells transfected with FLAG-5-HT2AR and HA-CRFR1 and pretreated with and without 500 nM CRF for 30 min in the presence of dominant-negative dynamin I-K44A. The dose response curves represent the mean ± S.E.M. for 4 independent experiments. Shown are representative laser scanning confocal micrographs showing the distribution of (b) FLAG-5-HT2AR and HA-CRFR1 and (c) FLAG-5-HT2CR and HA-CRFR1 in HEK 293 cells labeled with FLAG and HA antibodies at 4°C and then warmed to 37°C for 30 min in the absence of agonist. (d) Shown are representative laser scanning confocal micrographs showing the distribution of FLAG-5-HT2AR and HA-CRFR1 labeled with FLAG and HA antibodies at 4°C and warmed to 37°C for 30 min in the absence of agonist. (e) Shown are representative laser scanning confocal micrographs showing the distribution of FLAG-5-HT2AR and HA-CRFR1 transfected into rat cortical neurons labeled with FLAG and HA antibodies at 4°C and treated with 500 nM CRF and warmed to 37°C for 30 min. (f) Shown are representative laser scanning confocal micrographs showing the distribution of FLAG-5-HT2AR and HA-β2AR transfected into HEK 293 cells labeled with FLAG and HA antibodies at 4°C and treated with 100 μM Iso and warmed to 37°C for 30 min. Micrographs are representative images of multiple cells imaged on three independent occasions. Bar = 10 μm.

To further assess the role of the intracellular trafficking of both the 5-HT2AR and CRFR1 in the CRF-dependent regulation of 5-HT2AR signaling, we examined whether the inhibition of receptor recycling with monensin would block CRF-mediated increases in 5-HT2AR signaling. Treatment of cells with 100 μM monensin did not affect 5-HT2AR signaling in the absence of CRF pretreatment (Fig. 4a). However, monensin treatment attenuated the increase in 5-HT2AR signaling observed following CRF pretreatment (Fig. 4a). To assess whether the effect of monensin treatment was related to the recycling of receptors through endosomes, we utilized dominant negative Rab4-S28N and Rab11-S25N proteins to selectively inhibit receptor recycling via rapid (Rab4 positive) and slow (Rab11 positive) recycling endosomes. We found that the overexpression of Rab4-S28N, but not the overexpression of Rab11-S25N, blocked the increase in 5-HT2AR-mediated inositol phosphate formation induced by CRFR1 pre-activation (Fig. 4b,c). Biotinylation of cell surface FLAG-5-HT2AR also revealed that CRF pretreatment increased the cell surface expression of the 5-HT2AR by 3.7 ±1.8 fold (Fig. 4d). Accordingly, the endocytosis and recycling of CRFR1 was essential for regulating 5-HT2AR signaling via mechanism that resulted in increased 5-HT2AR expression at the cell surface.

Figure 4

Role of receptor recycling in CRF modulated 5-HT2R signaling.

(a) Dose response curves for 5-HT stimulated inositol phosphate (IP) formation in HEK 293 cells transfected with FLAG-5-HT2AR and HA-CRFR1 and pretreated with and without 500 nM CRF for 30 min following the pretreatment of cells with and without 100 μM monensin for 30 min. (b) Dose response curves for 5-HT stimulated inositol phosphate formation in HEK 293 cells with transfected FLAG-5-HT2AR, HA-CRFR1 and Rab4S8N and pretreated with and without 500 nM CRF for 30 min. (c) Dose response curves for 5-HT stimulated inositol phosphate formation in HEK 293 cells transfected with FLAG-5-HT2AR, HA-CRFR1 and Rab11-S25N and pretreated with and without 500 nM CRF for 30 min. (d) Increase in cell surface 5-HT2AR localization following 30 min pretreatment of CRFR1 with 500 nM CRF. The cell surface expression of the 5-HT2AR represents the mean ± S.E.M. for 4 independent experiments. The full length blot is presented in Supplementary Fig. 5. * P<0.05 versus untreated control.

All three receptors encoded class I PDZ domain interacting motifs at the end of their carboxyl-terminal tails and both the 5-HT2AR and 5-HT2cR were previously demonstrated to interact with PDZ domain containing proteins that regulate receptor trafficking30-34. Therefore, we examined whether the deletion of three amino acids from the 5-HT2AR (ΔSCV) and CRFR1 (ΔTAV) carboxyl-terminal tails would affect cell surface recruitment of the 5-HT2AR following CRF treatment. When tested, we found that the deletion of either the 5-HT2AR or CRFR1 PDZ domain binding motifs attenuated the CRF-dependent increases in 5-HT2AR at the cell surface (Fig. 5a). Since a loss of the PDZ binding motifs on either the 5-HT2AR or CRFR1 resulted in a loss of CRFR1-dependent recruitment of 5-HT2AR to the cell surface, we tested whether PDZ domain interactions were required for CRFR1-mediated sensitization of 5-HT2R signaling. Truncation of the final three amino acid residues of the CRFR1 carboxyl terminal tail (ΔTAV) prevented CRFR1-mediated increases in 5-HT2CR signaling following CRF pretreatment (Fig. 5b). Similarly, increased 5-HT2cR inositol phosphate formation in response to CRFR1 activation was not observed following the deletion of either the deletion of either the 5-HT2CR PDZ (ΔSSV) or 5-HT2AR (ΔSCV) domain binding motifs (Fig. 5c,d). We also found that the treatment of HEK293 cells with a peptide that encoded the HIV Tat protein membrane transducing domain fused to the last 10 amino acid residues corresponding to the CRFR1 carboxyl-terminal tail prevented CRFR1-mediated sensitization of 5-HT2AR signaling (Fig. 5e). Thus, intact PDZ domain protein interactions with both receptors were required for CRFR1-dependent sensitization of 5-HT2R responses.

Figure 5

Receptor determinants of CRF-dependent increases in 5-HT2R signaling.

(a) Shown is the change in cell surface 5-HT2AR and 5-HT2AR-ΔSCV localization following 30 min pretreatment of CRFR1 with 500 nM CRF as well as the change in cell surface 5-HT2AR localization following 30 min pretreatment of CRFR1-ΔTAV with 500 nM CRF. The cell surface expression of the 5-HT2AR represents the mean ± S.E.M. for 4 independent experiments. *P<0.05 versus untreated control. (b) Dose response curves for 5-HT stimulated inositol phosphate (IP) formation in HEK 293 cells transfected with FLAG-5-HT2cR and either HA-CRFR1 or HA-CRFR1 lacking a PDZ domain binding motif (ΔTAV) pretreated with and without 500 nM CRF for 30 min. (c) Dose response curves for 5-HT stimulated inositol phosphate formation in HEK 293 cells transfected with HA-CRFR1 and either FLAG-5-HT2CR or FLAG-5-HT2CR lacking a PDZ domain binding motif (ΔSSV) pretreated with and without 500 nM CRF for 30 min. (d) Dose response curves for 5-HT stimulated inositol phosphate formation in HEK 293 cells transfected with HA-CRFR1 and either FLAG-5-HT2AR or FLAG-5-HT2AR lacking a PDZ domain binding motif (ΔSCV) pretreated with and without 500 nM CRF for 30 min. (e) Dose response curves for 5-HT stimulated inositol phosphate formation in HEK 293 cells transfected with HA-CRFR1 and FLAG-5-HT2AR pretreated for 1 h with a Tat-fusion peptide corresponding to the last 10 amino acid residues of the CRFR1 carboxyl-terminal tail and then treated with and without 500 nM CRF for 30 min. Dose response curves represent the mean ± S.E.M. for 3-5 independent experiments.

CRF treatment enhances 5-HT-mediated anxiety-related behaviours

To assess the role of CRF in the regulation of 5-HT2R-mediated anxiety behaviour, two anxiety-related behaviours were examined in mice: (1) the latency for mice to enter the center of an open field and (2) the latency for mice to enter the open arm of an elevated plus maze. Having established the molecular mechanism by which CRFR1 activation sensitized 5-HT2R responses in vitro, we examined whether the infusion of CRF peptide (1.5 μg) into the medial prefrontal cortex followed by the intraperitoneal administration of the 5-HT2R selective agonist DOI (0.15 mg/kg) would sensitize 5-HT-mediated anxiety-related behavioral responses. The latency of mice to enter the center of an open field varied as a function of the intracerebral infusion (CRF vs vehicle) x systemic challenge (DOI vs vehicle) interaction, F(1,35)=7.01, p < 0.01. Follow-up analysis of the mean latencies for mice to enter the center square in a 5 min open field test revealed that neither the CRF nor the DOI treatments alone influenced performance relative to the vehicle-vehicle condition (Fig. 6a). However, among mice that received both CRF and DOI treatment the latency to enter the central portion of the maze was significantly longer than that of mice that received only a single drug treatment or vehicle (Fig. 6a). In the plus-maze test, the latency to enter an open arm, as well as the number of entries onto the open arms, also varied as a function of the IC infusion (CRF vs vehicle) x systemic challenge (DOI vs vehicle) interaction, F(1,35) = 7.85, 3.89, p < 0.01 and 0.05, respectively. Follow-up comparisons indicated that DOI itself produced a modest reduction in the latency to enter an open arm (p < 0.08) and the number of arm entries emitted (p < 0.10), whereas CRF infusion had no effect (Fig. 6b,c). However, among mice that received both the CRF and DOI treatments a marked increase of the open arm latency and a decreased frequency of open arm entries was evident relative to mice that received either treatment alone (Fig. 6b,c). In contrast to these findings, the number of entries into the closed arms, which approximately doubled the open arm entries, did not vary with either the CRF or DOI treatments, or as a function of their interaction (p > .15) (Fig. 6d). Likewise, the time spent in the closed arms did not vary as a function of the treatments mice received (F < 1) (Fig. 6e).

Figure 6

Analysis of CRF pretreatment on 5-HT2R-mediated anxiety-related behaviours.

(a) Mean latencies for mice to enter the center square in a 5 min open field. (b) Mean latency to enter the open arms of the elevated plus maze. (c) The frequency of entries in to the open arms of the elevated plus maze. (d) The frequency of entries in to the closed arms of the elevated plus maze. (e) Time spent in the closed arms of the elevated plus maze. In all experiments, either vehicle or CRF (1.5 μg in 1 μl) was administered to the medial prefrontal cortex via a surgically implanted cannulae for 5 min and 5 min later mice were intraperotineally injected with vehicle or DOI (0.15 mg/kg) prior to behavioral testing. 9-10 mice were used in each test group. P< 0.01 versus vehicle/vehicle treated control. Data represents mean ± SD. *P< 0.01 versus vehicle/vehicle treated control.

In a follow up series of experiments we examined whether the synergistic effects of DOI and CRF treatment could be antagonized by the pretreatment of mice with the 5-HT2AR selective antagonist M100907. We found that the latencies to enter the open arms of the plus maze varied as a function of the DOI x CRF x M100907 interaction, F(1,41) = 6.00, p = 0.018 (Fig. 7a). The tests confirmed that treatment with DOI alone did not influence the latencies to enter the open arms, whereas CRF infusion provoked a moderate, but statistically significant increase in response latencies. In mice that received the combination of systemic DOI following CRF administration to the prefrontal cortex, latencies to enter the open arms were still longer (Fig. 7a). When mice were treated with M100907 alone or with M100907 plus DOI none of the mice entered the open arms of the plus maze. Likewise, when given M100907 in conjunction with CRF, latencies were longer than in mice that received CRF alone, although several mice did enter onto the open arms (Fig. 7a). As predicted, when mice received M100907 in conjunction with DOI and CRF the latencies to enter the open arms of the maze were markedly reduced from that elicited by the combination of DOI plus CRF. Thus despite the fact that M100907-treated mice displayed a significant reluctance to enter the open arms of the maze, M100907 effectively attenuated the effects of the DOI-CRF combination.

Figure 7

Analysis of CRF pretreatment on 5-HT2R-mediated anxiety-related behaviours following M100907 treatment.

(a) Mean latency to enter the open arms of the elevated plus maze in a 5 min test period. (b) The frequency of entries in to the open arms of the elevated plus maze. (c) Time spent in the open arms of the elevated plus maze. (d) The frequency of entries in to the closed arms of the elevated plus maze. (e) Time spent in the closed arms of the elevated plus maze. In all experiments, either vehicle or CRF (1.5 μg in 1 μl) was administered to the medial prefrontal cortex via a surgically implanted cannulae for 5 min and 5 min later mice were intraperotineally injected with vehicle or DOI (0.15 mg/kg) and mice were pretreated i.p. with either vehicle or 0.25 mg/kg of M100907 in a volume of 0.3 ml prior to DOI administration before behavioral testing. 6-8 mice were used in each test group. Data represents mean ± SD. *P< 0.05 versus respective vehicle control. ** P< 0.05 versus respective M100907 treatment. ϕP< 0.05 relative to M100907 and CRF treatment.

The analysis of both the number of open-arm entries and the time spent in the open arms revealed responses which paralleled that of the response latencies (Fig. 7b,c). Specifically, the DOI x CRF x M100907 interaction was highly significant, F(1,41) = 10.78, 15.04, p < 0.001, and the follow up tests confirmed that neither CRF nor DOI alone affected the frequency of open arm entries. By contrast the combination of these treatments significantly reduced open arm entries and reduced the time spent in the open arms, as observed in the preceding studies. The M100907 profoundly influenced the frequency of open arm entries and time spent on the open arms (as described in the analysis of the latencies) in that mice treated with the compound (alone or in combination with DOI) did not make any entries onto the open arm, and most animals treated with M100907 and CRF also failed to make open arm entries (Fig. 7b,c). However, when animals received all three compounds, open arm entries and time on the open arms increased significantly relative to mice that either received DOI and CRF (but not M100907) or those that received CRF and M100907 (but not DOI). However, the number of entries were clearly fewer than that of animals that were either untreated or that had received only DOI (Fig. 7b).

The analysis of the entries to the closed arms indicated that behavior was significantly influenced by the DOI x CRF x M100907 interaction, F(1,41) = 9.29, p < 0.01 (Fig. 7d). The follow up tests indicated that DOI, CRF and the combination of these treatments increased closed arm entries relative to mice that had received only the vehicle treatments. Thus, one cannot ascribe the reduced open arm entries induced by the CRF-DOI combination to reduced motor activity. The M100907 treatment alone reduced the frequency of arm entries, irrespective of the other treatments received, although the magnitude of this effect was less pronounced in mice that had also received DOI + CRF. The time spent in the closed arms was unaffected by either the DOI or CRF or their combination (Fig. 7e). However, time spent in the closed arms was increased by M100907 in those mice that received this treatment alone, or either DOI or CRF. However, time spent in the closed arms among mice that received the combination of the three treatments did not differ from that of mice that received the CRF + DOI or those that received DOI + M100907. However, the time spent in the closed arms among mice that received the combination of DOI, CRF and M100907 was indistinguishable from that of mice that received only vehicle, or either CRF or DOI alone (Fig. 7e). Taken together our data in mice showed that CRFR activation resulted in increased 5-HT2R signaling in vivo and that the activation of both receptors had an important effect on behavioural responses associated with anxiety.

DISCUSSION

We demonstrated here that CRF acted through CRFR1 to sensitize 5-HT2R-mediated signaling and anxiety behaviours thereby linking CRF-mediated stress responses to anxiety and depression. Our findings indicated that enhanced 5-HT2R sensitivity following CRF pretreatment in vivo as evidenced by increased anxiety-related behaviour in mice. This observation showed that CRF could potentiate 5HT2R mediated behaviours and has implications regarding the mechanisms by which stressors may exacerbate the anxiogenic effects of 5HT2R activation. Importantly, our behavioural data, which showed a functional interaction between CRF and 5-HT, were supported at the cellular level. Thus, we demonstrated both that CRFR1 activation positively modulated 5-HT2R signaling in cortical neurons and that these two receptors were co-expressed in the same neuronal populations. The molecular mechanism underlying the sensitization of 5-HT2R signaling by CRFR1 required agonist-stimulated CRFR1 endocytosis and recycling which resulted in increased cell surface expression of 5-HT2Rs and increased second messenger responses to 5-HT treatment (Supplemental Data Fig. 4). These findings provide an additional mechanism by which receptor endocytosis and recycling contribute to the regulation of GPCR responsiveness in general and specifically show how CRFR1 activation can positively modulate 5-HT2R signaling thereby leading to pathophysiological behavioural responses.

We observed that anxiety responses in both an open field emergence and in a plus-maze test were sensitized in mice that were pretreated with CRF administered to the prefrontal cortex, followed by systemic administration of a low dose of DOI. When administered alone, neither of these treatments affected performance in these tests, demonstrating that the CRF and DOI treatments acted synergistically to provoke the anxiety responses. The behavioral change could not be attributed to diminished motoric activity, as entries into the closed arms of the plus-maze were unaffected by the treatments. It should be said that when significantly higher doses of DOI were employed (0.625 and 1.25; data not shown) elevated arm entries were evident (as opposed to reduced open-arm entries), likely reflecting an overall arousal. It has been reported that CRF influences anxiety processes, and that CRFR1 may be especially relevant in this regard7-9. Likewise, pharmacological studies have pointed to the involvement of 5-HT manipulations in attenuating anxiety and that 5-HT2AR and 5-HT2CR may contribute to CRF-mediated anxiety22,11,15,16. Thus, both the CRF and 5-HT systems when sufficiently activated will independently lead to anxiety responses. The 5-HT2AR selective antagonist M100907 itself also provoked marked reductions of open arm entries suggesting that M100907 could independently induce an anxiety-like response. As entries into the closed arm were observed, it was clear that the absolute failure to enter the open arm was not due to motor impairments, and instead it was likely that the reduced activity reflected an overall increase of anxiety. However, of particular significance, was the observation that the anxiety-provoking effects of CRF and DOI cotreatment were antagonized by M100907 pretreatment. Thus our observations indicated that cross-talk between CRF- and 5-HT-mediated signaling processes occured in the prefrontal cortex and that CRF sensitized 5-HT2-processes to promote stressor-like effects, such as anxiety35.

Based on our data, we propose a multistep mechanism whereby CRF peptide activation of CRFR1 enhances 5-HT2R signaling by increasing the availability of 5-HT2R at the surface of cells to be activated by agonist and to couple to the activation of phospholinositol phosphatase Cβ-mediated inositol phosphate formation (Supplemental Data Fig. 4). We found that agonist-activation of CRFR1 promoted the dynamin-dependent internalization of CRFR1 into the intracellular endosomal compartment of the cell and we found that 5-HT2AR and 5-HT2CR were internalized to endosomes in a constitutive manner. Thus, following agonist treatment internalized CRFR1 facilitated the cell surface recycling of 5-HT2R from endosomes resulting in increased 5-HT2R protein at the cell surface. The CRFR-dependent enhancement of 5-HT2R signaling also required the interaction of PDZ domain containing proteins with both receptors, since the deletion of PDZ binding motifs in the carboxyl-terminal tail domains of either CRFR1, 5-HT2AR or 5-HT2CR prevented CRF-mediated sensitization of 5-HT2R signaling. Interestingly, the activation of CRFR2, another CRFR expressed in the brain, did not sensitize 5-HT2AR signaling and consistent with this observation examination of the CRFR2 carboxyl-terminal tail revealed that the canonical PDZ binding motif was disrupted.

We found that sensitization of 5-HT2R signaling was dependent on receptor endocytosis as dynamin I-K44A expression could block this effect. This suggested that the internalization of either the CRFR1 or the 5-HT2Rs was essential for sensitizing 5-HT2R signaling. Several lines of evidence suggest that it is the internalization of CRFR1 that is essential for this effect. First, both 5-HT2AR and 5-HT2CR are found to be predominantly intracellularly in neurons of the rat prefrontal cortex36,37. Second, in the present study we found that both 5-HT2AR and 5-HT2CR were constitutively internalized in both HEK 293 cells and neurons, although cell surface expression of 5-HT2AR has been reported38-40. However, the mechanism underlying the observed constitutive endocytosis was unclear and may be consequence of the fact the serum used to culture cells may contain 5-HT. Independent of the mechanism by which 5-HT2R were internalized, we propose that it was the internalization and recycling of the CRFR1 that dynamically regulated the subcellular equilibrium of 5-HT2R resulting in the redistribution of 5-HT2R to the cell surface resulting in the sensitization of 5-HT2R signaling.

The CRFR1-mediated increases in 5-HT2AR signaling were also blocked by either the treatment of cells with monensin, which prevents the trafficking of intracellular vesicles or the overexpression of a dominant-negative Rab4-S28N mutant protein that blocked rapid recycling of GPCRs to the cell surface. Thus, CRFR1 sensitization of 5-HT2R signaling required increased 5-HT2R recycling and cell surface expression. The intracellular localization of 5-HT2R may prevent over-stimulation of serotonergic synapses. The regulated recruitment of this intracellular pool of 5-HT2R may function to promote altered post-synaptic signal adaptation to physiological stimuli, such as CRF peptide release in response to stress leading to the activation of CRFR1 in 5-HT2R expressing neurons of the prefrontal cortex. Such plasticity at serotonergic synapses may be akin to the alterations in AMPA receptor trafficking involved in synaptic plasticity associated with long term potentiation41.

We found that CRFR1-dependent alterations in 5-HT2R signaling required intact PDZ binding motifs at the carboxyl-terminal tails of both CRFR1 and 5-HT2Rs. Thus, these receptors may exist as components of a macromolecular protein complex via the recruitment of PDZ domain containing scaffold proteins. Although PDZ protein interactions have not been reported for the CRFR1, several PDZ domain-containing proteins have been demonstrated to interact with both 5-HT2Rs. Examples of PDZ domain containing proteins that interact with both 5-HT2R and 5-HT2CR include MAGI-2, MPP3, MUPP1, PSD-95 and SAP9730-34. Each of these PDZ domain containing proteins are comprised of multiple PDZ domains that would allow them to form complexes with more than one GPCR. PDZ domain containing proteins have also been demonstrated to regulate GPCR signaling, desensitization and trafficking. For example, PSD-95 inhibits β1AR internalization, but facilitates the association of the β1AR with NMDA receptors, whereas SAP97 interactions are involved in β1 AR recycling42. PSD-95 overexpression increases rat 5-HT2CR desensitization and facilitates both constitutive and agonist-induced rat 5-HT2CR internalization38. In contrast, PSD-95 interactions with 5-HT2AR leads to augmented 5-HT2AR signaling without altering the kinetics of 5-HT2A R desensitization30. PSD-95 is also required for proper dendritic targeting and expression of 5-HT2A and 5-HT2C receptors in vivo34. Thus, PDZ domain containing proteins may not only contribute to the formation of CRFR1/5-HT2R protein complexes, they may be involved in the regulation of the co-trafficking of the receptors between cellular compartments.

In summary, the endocytosis and recycling of GPCRs plays an important role in regulating the desensitization and resensitization of GPCRs as well as modulating their signaling via G protein-independent signal transduction pathways43. Here, we identified an additional mechanism by which the endocytosis and recycling of one GPCR influenced the activity of a second GPCR by recruiting constitutively internalized receptors to the cell surface. As a consequence, we found that agonist-stimulated CRFR1 internalization resulted in the sensitization of 5-HT2R signaling by allowing the recruitment of internalized 5-HT2R to the plasma membrane. Our studies provide a novel biochemical mechanism to explain how CRFR1 activation sensitizes 5-HT2R-mediated anxiety behaviours in response to stress that is likely to be applicable to other receptor-mediated signaling pathways and behavioral responses.