SNX14 is a bifunctional negative regulator for neuronal 5‐HT₆ receptor signaling

G‐protein‐coupled receptors (GPCRs) are seven‐transmembrane domain proteins that serve to transduce the intracellular effects of a large variety of hormones or neurotransmitters subserving an equally extensive list of physiological responses (Hill, 2006). Most GPCRs couple to heterotrimeric G proteins comprised of α, β and γ subunits at the plasma membrane. In humans, there are over 20 Gα proteins encoded by 16 genes, which are subdivided into four subfamilies based on sequence identity and shared effector systems (Milligan and Kostenis, 2006). The duration of G protein signaling is controlled by the lifetime of the Gα subunit in its GTP‐bound state. Although Gα subunits have intrinsic GTPase activity, auto‐hydrolysis of GTP is often too slow to account for the dynamic switching of GPCR signaling (De Vries et al., 2000; Neubig and Siderovski, 2002). Proteins containing a regulator of G‐protein signaling (RGS) domain bind activated Gα subunits and act as GTP‐activating proteins (GAPs), accelerating GTP hydrolysis (De Vries et al., 2000). So far, RGS proteins for Gα_i, Gα_q and Gα_12/13 have been well‐characterized, but little is known about proteins that might regulate Gαs. Previously, SNX13 has identified been as a RGS for Gαs that is highly expressed in heart and skeletal muscle and has a role in EGFR degradation (Zheng et al., 2001).

The role of 5‐hydroxytryptamine (5‐HT, also known as serotonin) in cognitive processing and performance has been known for many years (Dawson, 2011). Among various 5‐HT receptors, the 5‐HT subtype 6 receptor (5‐HT₆R, also known as HTR6) is one of the most recently discovered of the serotonin receptors. It is positively coupled to adenylyl cyclase and increases cAMP production upon activation. It is located almost exclusively within the central nervous system (CNS), including the striatum, olfactory tubercle, nucleus accumbens, hippocampus, cortex, cerebellum, hypothalamus and amygdala (Monsma et al., 1993; Ruat et al., 1993; Ward et al., 1995).

5‐HT₆R is known to modulate the transmission of several neurotransmitters including acetylcholine, glutamate, dopamine, γ‐aminobutyric acid, epinephrine and norepinephrine (Dawson, 2011; Mitchell and Neumaier, 2005). It is involved in various higher brain functions such as memory and cognition (Marcos et al., 2008; Mitchell et al., 2007) as well as in pathophysiological conditions such as anxiety, depression and obsessive compulsive disorder (OCD) (Svenningsson et al., 2007), Alzheimer's disease (Geldenhuys and Van der Schyf, 2008), drug addiction (Lecca et al., 2004) and schizophrenia (Masellis et al., 2001). Recent evidence also suggests it has a role in obesity (Heal et al., 2011). Over the past decade, 5‐HT₆R has gained increasing attention and its antagonism has become a promising target for the treatment of various neuropsychological diseases. Nevertheless, very little is known about its regulation and function in the brain.

SNX14 is a member of the sorting nexin family, predicted to have for roles in protein sorting and vesicular trafficking (Carlton et al., 2005). It contains a putative RGS domain and a phox homology (PX) domain, an N‐terminal hydrophobic region and a PX‐associated (PXA) domain (Worby and Dixon, 2002). SNX14 has been isolated from mouse embryonic stem cell (ESC)‐derived neurons and was found to be expressed at high levels in the nervous system (Carroll et al., 2001). A recent study has found that SNX14 protein levels are progressively increased during neuronal development, whereas its knockdown severely impairs both excitatory and inhibitory synaptic transmission (Huang et al., 2014). Another study has found that SNX14 mutations cause a distinctive autosomal‐recessive cerebellar ataxia and intellectual disability syndrome (Thomas et al., 2014). Therefore, these results suggest that there is an essential role for SNX14 in neuronal development and function, but the underlying mechanism(s) for this remain poorly understood.

Here, we provide evidence that SNX14 is highly expressed in the brain, and as a sorting nexin, it interacts with 5‐HT₆R and accelerates internalization and degradation of 5‐HT₆R. Knockdown of endogenous SNX14 prolonged the cell surface expression of 5‐HT₆R. Although we found that the SNX14 RGS domain does not retain GAP activity for Gαs, it specifically bound to and sequestered Gαs, thus inhibiting the 5‐HT₆R‐mediated signaling pathway. We further reveal that the binding affinity of SNX14 for Gαs was almost abolished by phosphorylation of RGS domain by protein kinase A (PKA) and that phosphorylation of SNX14 is also required for the endocytic degradation of 5‐HT₆R. Taken together, our results suggest that SNX14 is a dual negative regulator in 5‐HT₆R‐mediated signaling pathways that acts by both sequestering Gαs and inducing endocytosis of 5‐HT₆R.

SNX14 consists of a putative RGS domain, a PX domain, an N‐terminal hydrophobic region and a PXA domain (Fig. 1A; supplementary material Fig. S1A). In situ hybridization showed that SNX14 was highly expressed in the hippocampus, nucleus accumbens and cerebellum (Fig. 1B), where 5‐HT₆R is also abundantly expressed. Immunoblotting also showed that SNX14 was also highly expressed in lung, testis and brain regions, such as the hippocampus, cerebellum and cerebral cortex, but that virtually no expression was observed in heart and muscle where SNX13 is known to be abundant (Zheng et al., 2001), suggesting a mutually exclusive tissue distribution (Fig. 1C). High levels of SNX14 expression were also detected in primary hippocampal neurons, glial cells and in the HT‐22 hippocampal cell line but not in HEK293T, HeLa or COS‐7 cells (Fig. 1D; supplementary material Fig. S1B).

Previous studies have shown that SNX3 regulates endosomal trafficking through its PX‐domain‐mediated interaction with PtdIns(3)P (Xu et al., 2001). Examination of the phosphoinositide‐binding behavior of SNX14 showed that the PX domain bound both PtdIns(3)P and PtdIns(5)P, whereas the PXA domain bound PtdIns(3,4)P₂, PtdIns(3,4,5)P₃ and PtdIns(4,5)P₂. Full‐length SNX14 displayed a combination of the lipid binding patterns of the PX and PXA domains. Accordingly, the heterogeneously expressed SNX14 domain mutants localize in endosome‐like structures in the cytosol or at the plasma membrane, suggesting that SNX14 might play a role in membrane trafficking (supplementary material Fig. S1C,D).

5‐HT₆R is a Gαs‐linked serotonin receptor and SNX14 contains a putative RGS domain. The closely related protein SNX13 has been reported to act as an RGS for Gαs. Given that the expression profiles in the CNS are similar for SNX14 and 5‐HT₆R, we wondered whether SNX14 interacts with 5‐HT₆R. The third intracellular loop (iL3) has a fairly long stretch of amino acids and a previous study has shown that iL3 of 5‐HT₆R interacts with Gαs (Kang et al., 2005). Micro‐liquid chromatography tandem mass spectrometry (LC‐MS/MS) analysis using immunoprecipitated brain lysates with GST‐conjugated iL3 of 5‐HT₆R showed that besides various endocytic proteins, such as dynamin, AP‐2, amphiphysin and epsin, SNX14 was also identified as one of its endogenous binding partners (Fig. 1E).

When heterogeneously expressed in COS‐7 cells, 5‐HT₆R interacted with the full‐length SNX14 and RGS‐PX domain of SNX14 but not with SNX9 and SNX18 (Fig. 1F). We determined that the interaction between 5‐HT₆R and SNX14 was mediated by the iL3 of 5‐HT₆R and the RGS domain of SNX14 (Fig. 1G). The K262 and K265 residues of iL3 appear to be the binding sites for SNX14 because a K262A and K265A double mutant showed little interaction with SNX14 (Fig. 1H). The RGS of SNX14 did not interact with β₂ adrenergic receptor (β₂AR, also known as ADRB2) or dopamine D₁ receptor (D₁DR, also known as DRD1), which are related Gαs‐linked GPCRs (Fig. 1I).

Next, we tested whether the interaction between SNX14 and 5‐HT₆R plays a role in endocytic trafficking of 5‐HT₆R. The RGS‐PX domain of SNX14 was coexpressed with 5‐HT₆R in HEK293T cells and cells were treated with 5‐HT to induce internalization of 5‐HT₆R. Curiously, the expression of the RGS‐PX domain decreased total and surface 5‐HT₆R to the same extent seen in control cells after 30 min exposure to 5‐HT. Treatment of these cells with 5‐HT evoked no further decrement of total or surface 5‐HT₆R (Fig. 2A–D). This endocytic degradation by the RGS‐PX domain was not detected for β₂AR (Fig. 2B–D).

These results suggest that SNX14 dramatically increases the rate of constitutive endocytosis and degradation of 5‐HT₆R. This is confirmed by experiments using Tet‐On inducible expression of the RGS‐PX domain in cells transiently or stably expressing 5‐HT₆R (Fig. 2E). Using liposome‐mediated delivery of purified the RGS‐PX domain of SNX14, we show that 5‐HT₆R gradually decreased from the cell surface after delivery of purified the RGS‐PX domain (Fig. 2F). We found that not only the RGS‐PX domain but also full‐length SNX14 had a similar effect and further found that phosphoinositide‐binding property of the PX domain is required given that the ΔPX mutant of SNX14 failed to induce 5‐HT₆R internalization (Fig. 3A–E). Endogenous SNX14 was recruited to the plasma membrane by 5‐HT treatment in HT‐22 cells (supplementary material Fig. S2A). Total internal reflection fluorescence (TIRF) imaging also demonstrated that full‐length SNX14 was recruited to the plasma membrane by 5‐HT treatment and that some of SNX14 spots were re‐internalized (Fig. 3F–J).

We next investigated whether knockdown of SNX14 alters the expression of 5‐HT₆R in HT‐22 cells that endogenously express high levels of SNX14 and 5‐HT₆R. Three independent short hairpin RNA (shRNA) constructs targeting SNX14 were made (Fig. 4A) and suppression of SNX14 expression by shRNA1 and shRNA2 in HT‐22 cells was confirmed, whereas shRNA3 had no effect (Fig. 4B). When endogenous expression of SNX14 was suppressed by shRNA1, shRNA2 or shRNA1+2, the levels of 5‐HT₆R were significantly higher than those in the control or shRNA3‐transfected cells (Fig. 4B). Furthermore, 5‐HT exposure did not further downregulate 5‐HT₆R in shRNA1+2‐transfected cells, supporting the conclusion that SNX14 accelerates endocytosis of 5‐HT₆R (Fig. 4C–E). The levels of SNX14 stayed relatively constant over time after 5‐HT treatment even in the presence of phosphatase (Fig. 4F,G).

SNX14 contains a putative RGS domain (Fig. 1A; supplementary material Fig. S1A). Given that the closely related protein, SNX13 has been reported to act as a RGS for Gαs (Zheng et al., 2001), we evaluated whether SNX14 also acts as a RGS‐GAP for Gαs. We found that SNX14 bound to the activated form of Gαs [activation was achieved by addition of aluminium magnesium fluoride (AMF), GTPγs or the Q227L mutation, each of which promotes conformational changes that resemble the transition state of Gαs (Tesmer et al., 1997)]. SNX14 did not bind to wild‐type Gαi1 or the active form of Gαi1 (Q204L) even in the presence of AMF. The isolated RGS domain of SNX14 interacted with constitutively active Gαs (Q227L) and with wild‐type Gαs but not with wild‐type Gαi1 or the active form of Gαi1 (Q204L) (Fig. 5A). SNX14 or activated Gαs was, respectively, pulled down with GST–Gαs or the GST–RGS domain from mouse brain lysates (Fig. 5B). In addition, purified His–Gαs (AMF‐activated) interacted with the purified GST–RGS domain but not with GST–RGS4 (Fig. 5C). Mutationally activated Gαs (Q227L) also strongly bound to the purified GST–RGS domain but not to GST–RGS4 (Fig. 5C), suggesting that there is a specific interaction between the RGS domain of SNX14 and Gαs. We next investigated whether SNX14 could potentiate the GTPase activity of Gαs. GTP single‐turnover assays showed that SNX14 did not promote the GTPase activity of Gαs (Fig. 5D).

Given that SNX14 does not possess GAP activity for Gαs, we queried the physiological meaning of the SNX14–Gαs interaction by performing cAMP production assays. 5‐HT increased cellular cAMP in control cells but not in cells expressing the RGS domain of SNX14. cAMP elevation in the presence of forskolin, which activates Gαs in the absence of adenylyl cyclase, was equivalent in the presence or absence of SNX14 (Fig. 5E). These results suggest that SNX14 acts as a negative regulator that inhibits Gαs‐mediated cAMP production perhaps by binding and sequestering Gαs. We found that endogenous SNX14 was located in endosome‐like punctate structure, but upon addition of 5‐HT, SNX14 translocated to the plasma membrane (supplementary material Fig. S2B,C). TIRF imaging also showed that SNX14 appeared to be recruited to the plasma membrane by 5‐HT treatment coincidently with Gαs, and they disappeared together (supplementary material Fig. S2C).

The iL3 region of 5‐HT₆R is known to directly bind to Gαs (Kang et al., 2005). Given that we found that the RGS of SNX14 also interacts with iL3 of 5‐HT₆R (Fig. 1G,H), we tested whether the RGS of SNX14 and Gαs competitively bind to iL3 of 5‐HT₆R. The interaction between RGS of SNX14 and iL3 of 5‐HT₆R gradually decreased, with a concomitant increase in the interaction between Gαs and iL3 of 5‐HT₆R upon increasing the amount of Gαs protein in the reaction (Fig. 6A–D), suggesting that there is a competition between RGS of SNX14 and Gαs in their binding to iL3 of 5‐HT₆R.

We next tested whether SNX14 is phosphorylated by cAMP‐dependent PKA, a downstream effector of Gαs‐mediated signaling pathways. Indeed, SNX14 was phosphorylated at serine and threonine, but not tyrosine residues, upon forskolin treatment (Fig. 7A–C). We found that recombinant RGS was strongly phosphorylated in vitro by the catalytic subunit of PKA (PKAnes) (Fig. 7D) and LC‐MS/MS showed that the RGS domain was phosphorylated on S382 and S388 (supplementary material Fig. S3). Although these serine residues are not canonical sites for PKA phosphorylation, PKA‐mediated phosphorylation was markedly reduced in an S382A/S388A double mutant of RGS (supplementary material Fig. S3; Fig. 7D). This was further confirmed in cells transfected with RGS or various S382/S388 mutants and PKAnes. The phosphorylation of RGS increased by PKAnes was substantially reduced in S382A/S388A single or double mutants (Fig. 7E).

We next tested whether phosphorylation of SNX14 affects its binding to Gαs. Firstly, when transfected with PKAnes, there was dramatic reduction in the interaction between the RGS‐PX domain of SNX14 and Gαs in the presence of AMF or a constitutively active Gαs (Q227L) (Fig. 7F). Secondly, the single or double phospho‐mimetic mutants (S382D/S388D) of RGS either markedly reduced or completely abolished the binding between RGS and Gαs (Fig. 7G). These results suggest that the phosphorylation of RGS domain by PKA dramatically decreases its affinity for Gαs.

We next tested whether phosphorylation affects the SNX14‐induced endocytic trafficking of surface 5‐HT₆R. Interestingly, the S382A/S388A double phospho‐deficient mutant did not promote endocytosis of 5‐HT₆R whereas the S382D/S388D double phospho‐mimetic mutant had an effect on 5‐HT₆R levels similar to the RGS‐PX domain (Fig. 8A,B). These results suggest that PKA‐mediated phosphorylation of SNX14 is required for its effect on internalization of 5‐HT₆R. We tested further this possibility using the PKA inhibitor, H‐89. After H‐89 treatment, co‐expression of the RGS‐PX domain did not decrease the initial levels of either total or surface levels of 5‐HT₆R (Fig. 8C–F). Given that we showed that phosphorylated the RGS‐PX domain no longer binds Gαs (Fig. 7F,G), our results imply that phosphorylated SNX14 might be ‘re‐routed’ for preferential binding to 5‐HT₆R, thus enhancing endocytic internalization and degradation of that receptor.

In this study, we found that SNX14, as might be expected of a sorting nexin, accelerates internalization and degradation of 5‐HT₆R. We also found that SNX14 specifically binds and sequesters Gαs, thus inhibiting downstream cAMP production. We, thus, provide strong evidence that SNX14 plays roles as an endogenous negative regulator of 5‐HT₆R trafficking and signaling.

Besides Gαs, 5‐HT₆R is known to interact with various proteins such as Fyn tyrosine kinase (Yun et al., 2007), Jun activation domain‐binding protein‐1 (Jab1, also known as COPS5) (Yun et al., 2010), mammalian target of rapamycin (mTOR) (Meffre et al., 2012), microtubule‐associated protein 1B light chain (MAP1B‐LC) (Kim et al., 2014) and cyclin‐dependent kinase 5 (Cdk5) (Duhr et al., 2014). The C‐terminus of 5‐HT₆R interacts with the Fyn and activation of 5‐HT₆R stimulates extracellular‐regulated kinase 1/2 through the Fyn‐dependent pathway (Yun et al., 2007). 5‐HT₆R also interacts with mTORC1 and activates the phosphoinositide 3‐kinase (PI3K), Akt and mTOR signaling pathway in prefrontal cortex, and recruitment of mTOR by 5‐HT₆R contributes to the perturbed cognition in schizophrenia (Meffre et al., 2012). In addition, the C‐terminus and iL3 of 5‐HT₆R selectively interact with the N‐terminus and C‐terminus of Jab1, respectively. Interestingly, hypoxia increases the expression level of 5‐HT₆R and Jab1, indicating that 5‐HT₆R and the Jab1 complex and its downstream signaling pathway protect the cells against hypoxia (Yun et al., 2010). A recent study has further shown that direct recruitment of CDK5 and p35 (also known as CDK5R1) to the C‐terminus of 5‐HT₆R leads to the phosphorylation of this region and promotes neurite outgrowth by activating Cdc42 (Duhr et al., 2014). Such diverse 5‐HT₆R‐mediated downstream signaling pathways and second messengers might explain why agonists and antagonists of 5‐HT₆R result in conflicting effects in different studies (Marazziti et al., 2011).

There are many functional interactions between signaling and membrane trafficking (Ponimaskin et al., 2005; Renner et al., 2012). The best example is β‐arrestin, which binds to GPCRs and evokes their internalization through clathrin‐coated pits but also functions to activate various signaling cascades (DeWire et al., 2007). Likewise, SNX14 modulates G‐protein signaling as well as GPCR trafficking, suggesting that SNX14 bilaterally modulates Gαs‐linked GPCRs signaling as a dual negative regulator. Although not examined in this study, Gαs is known to traffic through lipid rafts and its sequestration attenuates Gαs signaling (Allen et al., 2009). Participation of SNX14 in this process might add another level of regulation relevant to psychiatric disorders (Donati et al., 2008; Donati and Rasenick, 2003).

Although we have not investigated a possible role of SNX14 on receptor insertion or recycling, we believe that the role of SNX14 resides in the endocytic realm for the following reasons. First, when we treated with dynasore to block the endocytosis, we found that the levels of 5‐HT₆R gradually increased even in the presence of SNX14 (supplementary material Fig. S4A). This suggested that SNX14 does not affect the insertion of the receptor. Second, a lysosomal inhibitor, chloroquine also induced a gradual increase of 5‐HT₆R levels in the presence of SNX14 (supplementary material Fig. S4B), suggesting that SNX14 does not affect the recycling of the receptor.

Regulators of Gαs‐linked GPCRs should have crucial roles in CNS, but specific RGS proteins for Gαs in the brain have yet to be found. Although the RGS domain of SNX13 is reported to function as a specific GAP for Gαs (Zheng et al., 2001), its expression is high in heart and skeletal muscles but virtually absent in brain. The RGS domain of axin is known to bind Gαs, but it also interacts Gα₁₂ or Gα_o, thus showing no specificity for Gαs (Castellone et al., 2005). SNX25 also contains an RGS domain and induces TGF‐β receptor degradation or interacts with D₁, D₂ dopamine receptors playing a role in trafficking. Our results show that SNX14 contains a RGS domain specific for Gαs. Despite its inability to act as a RGS‐GAP, we demonstrated that SNX14 successfully inhibits the production of cAMP, thus we proposed that although it is not a classical GAP for Gαs, SNX14 acts as a negative regulator that inhibits Gαs‐mediated downstream signaling pathways by binding and sequestering Gαs.

Regulation of RGS function by feedback phosphorylation is seen in other RGS or RGS‐like proteins. Phosphorylation of RGS2 by protein kinase C (PKC) reduces its GAP activity (Cunningham et al., 2001) and ERK‐mediated phosphorylation of RGS‐GAIP increases its GAP activity (Ogier‐Denis et al., 2000). Through phosphorylation, signaling cascades can be fine‐tuned in either a feedback or feed‐forward manner. The binding affinity of SNX14 for Gαs is dramatically decreased by PKA‐mediated phosphorylation of its RGS domain; thereby relieving Gαs inhibition by SNX14 and facilitating Gαs activation of adenylyl cyclase and the subsequent cAMP signaling pathways. Thus, activation of the 5‐HT₆R‐mediated cAMP–PKA pathway depends on the extent of SNX14 phosphorylation by PKA, suggesting that this might fine‐tune 5‐HT₆R‐mediated signaling in the brain which might contribute to the cellular responses observed in neurological and psychiatric disorders, suggesting SNX14 as a potential therapeutic target.

The full‐length of mouse SNX14 was amplified by PCR from an Image cDNA clone (Invitrogen, Carlsbad, CA) and subcloned into pEGFP‐C1 (Clontech, Palo Alto, CA) and pGEX‐4T‐1 (Amersham Biosciences, Arlington Hts, IL). The PXA‐RGS‐PX domain (amino acids 130–937), the RGS‐PX domain (amino acids 336–937), the PXA domain (amino acids 130–304), the RGS domain (amino acids 336–468) and the PX domain (amino acids 558–937) of SNX14 were subcloned into EGFP or mCherry empty vector. 6×His‐RGSPX was cloned into pcDNA‐3.1A (Invitrogen, San Diego, CA). HA‐tagged β₂AR, Gαs (WT), Gαs (Q227L), Gαi1 (WT) and Gαi1 (Q204L) were cloned into pCMV4‐FLAG vector (Sigma, St. Louis, MO). HA–β₂AR, HA‐RGS4 and 3×HA‐5‐HT₆R were purchased from the Missouri S&T cDNA Resource Center (Rolla, MO). PM‐GFP was constructed by subcloning of the GAP‐43 membrane anchoring domain into pDisplay‐EGFP vector (Invitrogen). The fidelity of all constructs was verified by DNA sequencing.

The anti‐SNX14 antibody was raised against RAENTERKQNQNY (amino acids 759–771) and anti‐human SNX14 polyclonal antibody was purchased from Sigma. Other antibodies used were: anti‐GFP (EMD Millipore, Billerica, MA), anti‐His (QIAGEN, Valencia, CA), anti‐transferrin receptor, anti‐EEA1 (BD Biosciences, San Jose, CA), anti‐β‐actin, anti‐β‐tubulin, anti‐β‐tubulin‐III, anti‐phosphoserine, anti‐phosphothreonine, anti‐phosphotyrosine (EMD Millipore), anti‐FLAG, anti‐GST (Sigma), anti‐5‐HT₆R (H‐143; Santa Cruz Biotechnology, Santa Cruz, CA), anti‐rat HA (Roche, Basel, Switzerland), anti‐mouse HA (COVANCE, Princeton, CA), anti‐Gαs (Santa Cruz Biotechnology) and secondary antibodies conjugated to Alexa Fluor 405, 488 or 594 (Invitrogen).

Protein–lipid overlay assays were performed using PIP strips (Echelon Research Laboratories, Salt Lake City, UT). Membranes were blocked with 3% (w/v) BSA in Tris‐buffered saline with 0.1% Tween 20 (TBS‐T) for 1 h at room temperature. Membranes were incubated overnight at 4°C in 3% BSA/TBS‐T, together with 0.5 µg/ml of various purified GST‐tagged proteins. Membranes were further incubated with anti‐GST antibody for 1 h at room temperature, probed with horseradish peroxide (HRP)‐conjugated anti‐rabbit IgG antibody (Jackson ImmunoResearch, West Grove, PA) and visualized by enhanced chemiluminescence (GE Healthcare Bio‐Sciences, Uppsala, Sweden).

Experiments were performed in accordance with guidelines set forth by the Seoul National University Council Directive for the proper care and use of laboratory animals. Mice were perfused transcardially with heparinized PBS followed by ice‐cold 4% paraformaldehyde in PBS. Brains were removed, immersed in the same fixative for 4 h, cryoprotected by infiltration with 10–30% sucrose solution, embedded in an Optimal Cutting Temperature compound and rapidly frozen in 2‐methanol. Specimens were cut into 35‐µm coronal sections on a cryostat and immunostained with anti‐SNX14 antibody. Sections were treated with 1% FBS with 0.3% Triton X‐100 in PBS for 30 min, blocked with 1% H₂O₂ in PBS, rinsed with PBS, incubated with the primary antibody for overnight at 4°C and exposed to the HRP‐conjugated anti‐rabbit IgG (Rockland Immunochemicals, Gilbertsville, PA) for 1 h at room temperature. Immunostaining was visualized by staining with diaminobenzidine (Vector, Burlingame, CA) for 3 to 5 min and viewing under a FSX100 microscope (Olympus, Tokyo, Japan).

HEK293T cells were transfected using Lipofectamine‐2000 (Invitrogen). At 48 h after transfection, the cells were starved for 16 h and treated with chemicals followed by 10 µM 5‐HT for 20 min or 25 µM forskolin for 3 min. The cells were lysed and cAMP was measured with a cAMP direct immunoassay kit (Abcam, Cambridge, MA).

BL21 bacterial cells were transformed with GST–RGS, GST–RGS mutants (S382D/S388D and S382A/S388A), 6×His‐Gαs and 6×His‐Gαi1. In the case of Gαs, 90 nM protein was applied with 900 nM of either RGS or RGS mutants (S382D/S388D and S382A/S388A) or 500 nM human RGS proteins whereas 80 nM Gαi1 protein was used for assay. G proteins were loaded with [³²P]GTP and GTPase assays were carried out as described previously except that the assay temperature was 4°C (Berman et al., 1996).

HEK293T cells were co‐transfected with 3×HA‐5‐HT₆R and various SNX14 mutants, lysed in a lysis buffer (20 mM Tris‐HCl pH 8, 10% glycerol, 137 mM NaCl, 2 mM EDTA), and centrifuged for 20 min at 14,000 g at 4°C. 500 µg total proteins from the supernatant were incubated with anti‐GFP, anti‐FLAG, anti‐phosphoserine, anti‐phosphothreonine or anti‐phosphotyrosine antibody overnight at 4°C, incubated with 30 µl of protein‐A–Sepharose for 1 h, pelleted by centrifugation and analyzed by SDS‐PAGE. Proteins on the gels were transferred onto PVDF membranes (Pall Life Sciences, Ann Arbor, MI) and incubated with primary antibodies for 1 h at room temperature. The immunoreactions were detected with SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific, Rockford, IL).

HEK293T, COS‐7 and HT‐22 cells were cultured at 37°C and 5% CO₂ in Dulbecco's Modified Eagle's Medium (DMEM, Invitrogen) supplemented with 10% FBS (Hyclone, Logan, UT) and transfected using Lipofectamine‐2000. Fluorescence images were acquired on an Olympus IX‐71 microscope (Olympus) using a CoolSNAP‐HQ camera (Roper Scientific, Tucson, AZ) driven by MetaMorph software (Molecular Devices, Sunnyvale, CA). The confocal images were acquired on an Olympus FV‐1000 confocal microscope with 100×, 1.4 NA oil‐immersion objective lens. For immunocytochemistry, cells were fixed in 4% formaldehyde and 4% sucrose in PBS for 15 min, permeabilized for 5 min in 0.25% Triton X‐100 in PBS and blocked for 30 min in 10% BSA in PBS at 37°C. The cells were incubated with primary antibodies for 2 h and secondary antibodies for 45 min at 37°C in 3% BSA in PBS. For the colocalization assay, COS‐7 cells were permeabilized with 0.05% saponin for 1 min at 4°C before fixation. Analysis and quantification were performed with MetaMorph and SigmaPlot 8.0 (Systat Software, San Jose, CA).

Cells were imaged using an Olympus IX‐71 microscope fitted with a 60×, 1.45 NA TIRF‐lens and controlled by Cell^M software (Olympus). 488 and 561 nm diode lasers were coupled to the TIRF microscopy condenser through two independent optical fibers. Using software, we changed the angle of the incident laser to get the calculated penetration depth of <120 nm. Cells were typically imaged either in single channel or in two channels by sequential excitation with 0.1‐s exposures and detected with a Andor iXon 897 EMCCD camera (512×512; Andor Technologies, Belfast, Northern Ireland). The Image J program (National Institutes of Health) was used for analysis.

GST‐RGS of SNX14 (0.5 mg/ml) was incubated with 1000 units/ml catalytic subunit of PKA (New England Biolabs, Ipswich, MA) at 30°C for 2 h in 50 mM Tris‐HCl pH 7.5, 10 mM MgCl₂ and 200 µM ATP. For the autoradiography, the reaction mixture was supplemented with γ‐labeled ATP to a final specific activity of 200 µCi/µmol. RGS phosphorylation was assayed in the presence of reducing agents.

3×HA‐5‐HT₆R stable cells were transfected with various SNX14 mutants. 48 h after transfection, cells were starved for 16 h in serum‐free media and treated with 10 µM 5‐HT for the times indicated. The cells were washed with ice‐cold PBS, incubated with 0.25 mg/ml EZ‐Link® Sulfo‐NHS‐SS‐Biotin in PBS for 30 min at 4°C, rinsed in Quenching Solution with TBS to remove free biotin reagents, and lysed. After centrifugation, the supernatants were incubated with 50 µl of 50% slurry of immobilized NeutraAvidin for 2 h at 4°C. To detect internalized 5‐HT₆R (Fig. 3D,E), 3×HA‐5‐HT₆R stable cells were transfected with various SNX14 mutants. After starvation, surface proteins were biotinylated with 0.25 mg/ml EZ‐Link® Sulfo‐NHS‐SS‐Biotin in PBS for 30 min at 4°C. Unreacted biotin was removed by 50 mM glycine in PBS for 10 min at 4°C and washed with cold PBS. The cells were treated with 10 µM 5‐HT for indicated times and washed with cold PBS. Non‐endocytosed biotin was cleaved in 50 mM glutathione, 75 mM NaCl and 75 mM NaOH in FBS solution for 20 min at 4°C. After blocking and washing, cells were lysed and immunoprecipitated with NeutraAvidin for 2 h at 4°C. Biotinylated proteins were eluted in SDS sample buffer followed by immunoblotting with an anti‐HA antibody.

Immunoprecipitated protein samples were separated by one‐dimensional SDS‐PAGE and protein bands of interest were excised from the gel for an in‐gel digestion procedure. Tryptic digest peptides obtained from the in‐gel digestion procedure were dissolved in 0.1% formic acid and 5% acetonitrile solution for mass spectrometry analysis.

Tryptic digest peptide samples obtained from the in‐gel digestion procedure were also resuspended in 1×NEBuffer‐3 and alkaline phosphatase solution (10 unit/µl) was added to remove phosphate moieties. After incubation for 2 h at 37°C, the dephosphorylation was stopped by adding formic acid solution to a final concentration of 5%.

In‐gel digested peptide samples and alkaline phosphatase‐treated peptide samples were subjected to micro‐LC‐MS/MS experiments using an LTQ ion trap mass spectrometer (ThermoElectron, San Jose, CA). Details of the micro LC‐MS/MS analysis and protein database search process are as described elsewhere (Choi et al., 2011). Briefly, peptides were separated by C18 reversed phase capillary column and Agilent HP1100 quaternary LC pump. Solvent A (3% HCOOH, 5% CH₃CN and 92% H₂O) and Solvent B (3% HCOOH, 5% CH₃CN and 92% H₂O) were used to make a 120‐min gradient. MS/MS spectra obtained in the micro‐LC‐MS/MS analysis were searched against an in‐house protein database containing both SNX14 and EGFP using Bioworks Ver.3.2 and Sequest Cluster System. The differential modification search options for phosphorylation modification (+80 on Ser, Thr, Tyr) and oxidation (+16 on Met) were considered in the search and the maximum number of modifications that were allowed per peptide was seven. Manual assignments of fragment ions in each MS/MS spectrum were performed to confirm the database search results.

The small hairpin RNAs (shRNAs) from three nucleotides 42–52 (shRNA‐1), 183–189 (shRNA‐2), and 536–542 (shRNA‐3) of the mouse SNX14 cDNA sequence (NM_172926) were designed. A pair of complementary oligonucleotides was synthesized separately with the addition of an ApaI site at the 5′‐end and an EcoRI site at the 3′‐end. The forward primer sequences were 5′‐CAGGTATCTGCATGTCTTATTCAAGAGATAAGACATGCAGATACCTGTTTT‐3′, 5′‐GGTGGATATTCCATCTATTTTCAAGAGAAATAGATGGAATATCCACCTTTT‐3′ and 5′‐GAGGATGACTCTCCAGTAGTTCAAGAGACTACTGGAGAGTCATCCTCTTTT‐3′ (the underlined letters are the SNX14‐siRNA sequences). The annealed cDNA fragment was cloned into the ApaI‐EcoRI sites of pSilencer‐1.0‐U6‐mRFP vector (Ambion, Austin, TX). The knockdown efficiency was tested in the HT‐22 cells and EGFP–SNX14‐transfected HEK293T cells.

The cloned 6×His‐RGSPX in pET‐23b vector (EMD Millipore) was transformed into BL‐21 and cultured in 2×YT medium supplemented with ampicillin. After overnight induction with 0.5 mM IPTG at 30°C, cells were sonicated in the lysis buffer (50 mM NaH₂PO₄ pH 8.0, 300 mM NaCl, 5 mM Decyl‐β‐D‐Maltopyranoside (Anatrace, Maumee, OH) and centrifuged for 60 min at 13,000 g. The supernatants were incubated with Ni‐NTA agarose (Qiagen) at 4°C for 30 min. The beads were then washed extensively with lysis buffer and the protein was eluted with an elution buffer (50 mM NaH₂PO₄ pH 8.0, 200 mM NaCl, 200 mM imidazole). To further purify 6×His‐RGSPX, whole proteins were filtered with an Ultracel‐30K centrifugal filter (Millipore, Billerica, MA) at 5000 g for 10 min and the supernatant was immediately applied to a Superdex 200 10/300 GL column (GE Healthcare) at a flow rate of 0.5 ml/min at 4°C. Proteins were collected by a FPLC‐NGC chromatography system. The purified 6×His‐RGSPX protein was introduced to 5‐HT₆R stable HEK293 cells using Chariot (Active Motif, Carlsbad, CA). Briefly, 4 µl of Chariot was diluted in 40 µl of 80% DMSO, mixed with 40 µl of PBS containing purified the RGS‐PX domain (1.25 µg) and followed by 30 min incubation at room temperature. The complex was diluted in serum‐free DMEM to a volume of 200 µl and added to 5‐HT₆R stable cells. The cells were incubated with the complexes in serum‐free medium for 1 h, followed by 1 ml serum‐containing medium for 2 h.

We thank Jung‐Ah Kim for critical reading of the manuscript. Confocal and TIRF microscopy data were acquired in the Biomedical Imaging Center.