Deficiency of Sorting Nexin 27 (SNX27) Leads to Growth Retardation and Elevated Levels of N-Methyl-d-Aspartate Receptor 2C (NR2C)

Abstract

Phox (PX) domain-containing sorting nexins (SNXs) are emerging as important regulators of endocytic trafficking. Sorting nexin 27 (SNX27) is unique, as it contains a PDZ (Psd-95/Dlg/ZO1) domain. We show here that SNX27 is primarily targeted to the early endosome by interaction of its PX domain with PtdIns(3)P. Although targeted ablation of the SNX27 gene in mice did not significantly affect growth and survival during embryonic development, SNX27 plays an essential role in postnatal growth and survival. N-Methyl-d-aspartate (NMDA) receptor 2C (NR2C) was identified as a novel SNX27-interacting protein, and this interaction is mediated by the PDZ domain of SNX27 and the C-terminal PDZ-binding motif of NR2C. Increased NR2C expression levels, together with impaired NR2C endocytosis in SNX27^−/− neurons, indicate that SNX27 may function to regulate endocytosis and/or endosomal sorting of NR2C. This is consistent with a role of SNX27 as a general regulator for sorting of membrane proteins containing a PDZ-binding motif, and its absence may alter the trafficking of these proteins, leading to growth and survival defects.

INTRODUCTION

The endocytic pathway is important for diverse cellular and physiological functions, such as nutrient uptake, cell migration, signaling, and development. Altered trafficking in the endocytic pathway is associated with diverse human diseases (2, 5, 24, 30, 36, 42). Proteins on the cell surface are internalized in the early endosome via various routes of endocytosis (10, 25). The early endosome represents the major sorting station of the endocytic pathway, and proteins in the early endosome can be sorted into various post-Golgi structures, such as the recycling endosome for returning to the cell surface, the late endosome (also called the multivesicular body [MVB]) for delivery to the lysosome for degradation, and the trans-Golgi network (TGN) of the Golgi apparatus for retrograde transport to the biosynthetic pathway (14, 15, 26, 39). Other sorting routes, such as direct recycling from the early endosome back to the surface or the direct route from the early endosome to the TGN, also exist (8, 27, 44). Protein trafficking among these compartments is mediated by vesicles/carriers/tubules as shuttling intermediates or by direct fusion of homotypic and/or heterotypic compartments (for example, homotypic fusion of early endosomes and heterotypic fusion of the late endosome with the lysosome) (1, 32). Proteins, such as coat proteins, adaptor proteins, Rab small GTPases and their effectors, and SNAREs, represent core machineries for intercompartmental trafficking (6, 12, 19, 31, 46).

Phosphoinositides are important for compartment-specific recruitment of various cytosolic proteins to distinct post-Golgi membranes, such as the plasma membrane, the early endosome, and the TGN, due to their regulated generation and turnover, as well as their abilities to interact with various protein structural domains (3, 45, 51). PtdIns(3)P is primarily enriched at the early endosome and regulates endosomal recruitment of a large number of proteins harboring the FYVE or phox (PX) domain (7, 9, 40). These FYVE and PX domain proteins function as regulators for endosomal sorting/trafficking. More than 47 PX domain proteins exist in humans, and most of them are also known as sorting nexins (SNXs) (4, 41, 49). Although the roles of some members of the SNX family in endocytic sorting/trafficking have been established, the roles and mechanisms for many SNXs remain to be defined. Among all PX domain proteins, sorting nexin 27 (SNX27) is unique, as it contains a PDZ domain, which is not present in other PX domain proteins. This domain is able to bind a 3-residue PDZ-binding motif usually present at the very C terminus of transmembrane proteins or cytosolic proteins (13, 22, 23, 34). SNX27 has been shown to interact with PDZ-binding motifs of three transmembrane proteins (5-HT4a receptor, Kir3 potassium channels, and β₂-adrenoreceptor) (18) and two cytosolic proteins (diacylglycerol kinase zeta and the cytokine-inducible protein [CASP]) via its PDZ domain. The interaction with membrane proteins seems to regulate intracellular trafficking (for example, endosomal sorting to the lysosome for Kir3.2 and recycling of β₂-adrenoreceptor, while interaction with cytosolic proteins seems to mediate endosomal recruitment (for example, diacylglycerol kinase zeta).

In this study, we show that (i) interaction of the PX domain of SNX27 with PtdIns(3)P mediates its early endosomal association, (ii) endogenous SNX27 is widely expressed and also enriched in the early endosome, (iii) knockout of the SNX27 gene in mice leads to growth retardation and lethality, and (iv) N-methyl-d-aspartate receptor (NMDAR) 2C (NR2C) is a SNX27-interacting protein and absence of SNX27 caused elevated levels and impaired endocytic trafficking of NR2C.

MATERIALS AND METHODS

Vectors and DNA constructs.

The following vectors are used in this study: pCI-neo (Promega), pGEX-4T-1 (Amersham), pGBKT7 (BD Clontech), and pDMyc-Neo and pDHA-Neo, which are modified version of the pCI-neo vector made in our laboratory (11, 40). pDMyc-Neo has two tandem Myc epitope sequences inserted upstream of the multiple-cloning site, while pDHA-Neo has double hemagglutinin (HA) tag sequences inserted upstream of the multiple-cloning site. The coding regions of both isoforms of SNX27 were amplified by PCR from a human Universal Quick Clone cDNA library (Clontech) using Advantage 2 polymerase (Clontech) and specific primers for the common 5′ end (ATGGCGGACGAGGACGGGGAAGGGATT) and the specific 3′ end of SNX27a (CTAGGTGGCCACATCTCTCTGCTGTGACCT) or SNX27b (CTAATATTCCTCTTTTCTCCACTTGAGCTC). A 5′ XhoI site and a 3′ NotI site were introduced during the PCR amplification. Both SNX27a and SNX27b were subcloned into pDMyc-Neo and pDHA-Neo vectors using XhoI and NotI sites. Deletion mutants were generated using PCR and cloned into the vectors. All clones were confirmed by DNA sequencing.

Transient and stable expression.

Transient transfection was done in 6-well plates for indirect immunofluorescence microscopy or in 10-cm plates for immunoprecipitation and immunoblotting experiments. The cells were transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. The cells were tested for transgene expression after 24 h. Stable pooled cells were selected by culturing the cells in the presence of Geneticin (Invitrogen) for at least 2 weeks.

Expression and purification of GST fusion proteins and antibody generation.

The SNX27NT coding region was cloned into the pGEX4T1 vector, and recombinant glutathione S-transferase (GST)-SNX27NT was purified as described previously (40). Anti-SNX27 polyclonal antibodies were raised by injecting purified GST-SNX27NT into rabbits with Freund's complete adjuvant. Booster injections with the antigens in Freund's incomplete adjuvant were administered every 20 days, and 50 ml antiserum bleed was recovered after the 4th injection and thereafter. Anti-SNX27 antibodies were purified using an affinity matrix prepared by chemical coupling of GST-SNX27NT to glutathione-Sepharose using dimethyl pimelimidate dihydrochloride (Sigma). Antibody bound to the Sepharose was eluted with ImmunoPure IgG elution buffer (Pierce) and neutralized with 1 M Tris-HCl, pH 8. The antibodies were then dialyzed against phosphate-buffered saline (PBS) (pH 7.0) and concentrated using Centricon (Millipore) at 5,000 × g.

Protein/lipid overlay assay.

The method used for protein/lipid overlay assays was described previously (11, 20).

Generation of SNX27^−/− mice.

Generation of SNX27^−/− mice was performed as described previously (46). Briefly, the third coding exon, which encodes part of the PX domain (residues 179 to 243) that is crucial for SNX27 function, was chosen for insertion of the Neo gene. A SalI site was introduced after GGTATACAATGTTTACATGGCTGGGAGGCAGCTGTGTTCTAAG, which codes for 179VYNVYMAGRQLCSK193 of exon 3, by site-directed mutagenesis, and a targeting vector was constructed by inserting the Neo cassette into this site with an 8-kb right arm and a 4-kb left arm flanking the insertion. The insertion caused deletion of residue 194 to the C terminus of SNX27.

The targeting construct was transferred into embryonic stem (ES) cells (embryonic day 14 [E14]) via electroporation. Selection of transformants was done using G418. PCR genotyping for SNX27 knockout ES cells was performed using primers RT-01 (5′ TGAGTGCTGGAATTAAAGGCATGC) and RT-neo (5′ CCATCTTGTTCAATGGCCGATCC). PCR genotyping of mouse tail DNAs was performed using primers KO01 (5′ AGAAATGAAGCATCGCTTACCC), KO02 (5′ GTTCCTGTCTCACCAGGTATAC), and KOneo (5′ GGCCGCTTT TCTGGATTCATCG).

Yeast two-hybrid interaction.

The yeast two-hybrid screen and assay were done according to protocols provided by BD Clontech and as described previously (21). The Gal4-DNA binding domain (BD)-fused SNX27FL in pGBKT7 was used to transform the Saccharomyces cerevisiae strain AH109 (BD Clontech). The resulting transformed AH109 yeast cells (MATa) were mated with a Gal4-DNA activating domain (AD)-fused human brain pretransformed Matchmaker cDNA library in Y187 yeast cells (MATα) (BD Clontech). The interaction-positive diploid yeast cells were selected at the highest stringency using synthetic defined (SD) medium (Clontech) lacking tryptophan, leucine, histidine, and adenine (SD/−Trp/−Leu/−His/−Ade) (Quadruple Drop Out medium [QDO]) plates. The Gal4-AD-fused cDNA was extracted from those positive clones and subsequently sequenced. The partial sequences were subjected to a BLAST (Basic Local Alignment Search Tool) search at the NCBI (National Center for Biotechnology Information) website to identify the corresponding cDNAs. The SNX27 bait vector in pGBKT7 and isolated prey vectors were retransformed into AH109 and Y187 yeast cells, respectively. The transformed AH109 and Y187 yeast cells were mated, and the resulting diploid yeast cells containing DNA-BD and DNA-AD fusion constructs were selected on SD medium lacking leucine and tryptophan (SD/−Trp/−Leu) for 2 days. Then, the cells were plated on the SD/−Trp/−Leu/−His/−Ade (QDO) plate to confirm protein-protein interaction.

Neuron culture and pulse-chase.

Cortices from postnatal day 1 to 3 mice were trypsinized (10 min; 37°C), washed, and gently triturated by passing the tissue through a Pasteur pipette with a fire-polished tip. The cells were plated at a density of 100,000/well in poly-d-lysine-coated 24-well polystyrene plates or at 60,000 cells/poly-d-lysine-coated glass coverslip and maintained at 37°C in a humidified incubator with 5% CO₂, 95% O₂ in growth medium. The cells were plated in Neurobasal medium containing 10% fetal bovine serum, B27 serum-free supplement, and glutamax I (all from Gibco). Twenty-four hours after being plated, the culture medium was completely replaced with Neurobasal medium without serum. The cultures were fed three times a week by adding Neurobasal medium until they were used. For pulse-chase, neurons at 4 days in vitro were transfected with myc-NR2c using Lipofectamine 2000 (Invitrogen). Two days after transfection, surface myc-NR2c was live labeled with anti-Myc antibody on ice for 30 min and pulse-chased at 37°C for various times. Surface anti-Myc was then labeled with AF488 secondary antibody without permeabilization, while internalized anti-Myc was labeled with AF555 secondary antibody with permeabilization.

RESULTS

SNX27 is evolutionarily conserved and targeted to the early endosome by the PX domain.

Among 47 human PX domain-containing proteins, SNX27 is unique in that it is the only PX domain protein containing a PDZ (Psd-95/Dlg/ZO1) domain (4, 41, 49). Human SNX27 has two isoforms: SNX27a (541 amino acids in length) and SNX27b (528 amino acids in length). The only difference is that the C-terminal 15-amino-acid sequence (NIFQMARSQQRDVAT) of SNX27a is replaced by two other residues (EY) in SNX27b. In addition to the N-terminal PDZ domain (residues 40 to 136), SNX27 contains a PX domain (residues 160 to 265) and an RA (Ras association) domain (residues 271 to 362). A BLAST search revealed that SNX27 is evolutionarily conserved and that it also exists in model organisms, such as the fly, worm, and zebrafish. The domain organization is similarly conserved among these different species. However, lower eukaryotic organisms, such as S. cerevisiae or Schizosaccharomyces pombe, do not have SNX27, indicating that SNX27 may play an important role in cellular processes that are unique to multicellular organisms.

The coding regions of both isoforms of SNX27 were amplified by PCR from a human Universal Quick Clone cDNA library (Clontech) and subcloned into pDMyc-Neo and pDHA-Neo vectors. Since similar results of subcellular localization and biochemical characterization were obtained for both SNX27a and SNX27b, we describe our studies using the SNX27a isoform below. To investigate the cellular distribution of SNX27, Myc-SNX27 was expressed in stably transfected A431 cells. Analysis of A431 cells expressing myc-tagged SNX27 by indirect immunofluorescence microscopy revealed that SNX27 is present in vesicular structures characteristic of endocytic compartments (data not shown). Since the SNX27-containing vesicular structures are also marked by the early endosomal marker EEA1, SNX27 is thus targeted to the early endosome. Since SNXs are generally targeted to the endosomal compartments by the PX domain (4, 41, 49), we investigated the role of the PX domain and other domains of SNX27 in endosomal targeting. Four deletion mutants of SNX27 were created: myc-SNX27ΔPDZ (with the N-terminal 159 residues deleted), myc-SNX27ΔPX (with residues 160 to 265 deleted), myc-SNX27NT (with the C-terminal residues 266 to 541 deleted), and myc-SNX27CT (with the N-terminal residues 1 to 265 deleted). Deletion of the PDZ domain- or RA domain-containing C-terminal half did not affect targeting of the mutants to the vesicular structures. However, deletion of the PX domain alone or together with the PDZ domain resulted in cytoplasmic distribution of the mutants (unpublished data). These results suggest that the PX domain of SNX27 is essential for endosomal targeting.

SNX27 is widely expressed in cell lines and tissues.

To study the biochemical properties and function of endogenous SNX27, we generated rabbit antibodies against SNX27. The recombinant N-terminal domain (residues 1 to 265) of SNX27 was expressed as a fusion to GST (GST-SNX27NT) and used to immunize rabbits. Antibodies specific for SNX27 were affinity purified using immobilized GST-SNX27NT. When total lysate of A431 cells was analyzed by immunoblotting analysis, a single polypeptide of the expected size (60 kDa) was detected by the antibodies (Fig. 1 A, lane 1). However, when the antibodies were first incubated with GST-SNX27NT, this 60-kDa band was no longer detected (lane 2). Prior incubation of the antibodies with GST-SNX16FL did not affect the detection of the band (lane 3), suggesting that the 60-kDa band represents the endogenous SNX27 and that the antibodies are specific for SNX27.

Fig. 1.

SNX27 antibodies revealed that SNX27 is widely expressed. (A) Affinity-purified rabbit antibodies against SNX27 were used to probe the membrane blot containing 50 μg A431 cell lysate resolved by SDS-PAGE. The antibodies were used without blocking protein (lane 1) or preincubated with GST-SNX27NT (lane 2) or with GST-SNX16FL (lane 3). (B) Forty micrograms of lysate derived from the indicated cultured cells of human (lanes 1 to 4), monkey (lane 5), rat (lane 6), and mouse (lanes 7 to 9) origin was loaded into each lane and resolved by SDS-PAGE, followed by immunoblotting analysis using anti-SNX27 antibodies. SNX27 (60 kDa) was detected in all the cell lines. The immunoblot was also probed with an antibody against β-actin as a loading control. (C) Forty micrograms of lysates derived from the indicated rat tissues was resolved by SDS-PAGE, followed by immunoblotting analysis using SNX27 antibodies. The immunoblot was also probed with an antibody against β-actin as a loading control.

With the SNX27-specific antibodies, we next examined the expression of SNX27 in various cell lines by immunoblotting analysis (Fig. 1B). Various cell lines were harvested and lysed in RIPA buffer, and 40 μg of cell lysates from each cell line was analyzed. The 60-kDa band of SNX27 was detected in all the cell lines examined, including human A431, HeLa, and HEK293T cells; monkey COS7 cells; rat NRK cells; and mouse PC12, ATT20, and MIN6 cells. These results suggest that the antibodies cross-react with SNX27 in various species and that SNX27 is widely expressed. We next examined SNX27 expression in several rat tissues (Fig. 1C). Proteins of postnuclear supernatants from the indicated tissues were resolved on 10% SDS-PAGE, followed by immunoblotting using the antibodies. Expression of SNX27 was detected in all tissues at various levels. The ubiquitous expression of SNX27 in different cell lines and various tissues indicates that it may play a role in fundamental cellular events.

Endogenous SNX27 is enriched in the early endosome in a phosphatidylinositol 3-kinase (PI3K)-dependent manner.

To corroborate the conclusion that SNX27 is targeted to the early endosome in a PX domain-dependent manner based on analysis of epitope-tagged SNX27 and its mutants, we next examined the subcellular localization of endogenous SNX27 by indirect immunofluorescence microscopy using SNX27-specific antibodies. Since vesicular endosome-like labeling of endogenous SNX27 was observed in essentially every cell line examined (examples are shown in Fig. 2A), we performed double labeling of SNX27 in A431 cells with organelle markers: early endosomal EEA1 and Golgi SNARE GS28 (Fig. 2B). Endogenous SNX27 colocalized well with EEA1, but not with GS28, confirming that endogenous SNX27 is primarily associated with the early endosome.

Fig. 2.

Endogenous SNX27 is enriched in the early endosome in a PtdIns(3)P-dependent manner. (A) HeLa, A431, and NIH 3T3 cells were fixed and processed for indirect immunofluorescence assay using a rabbit polyclonal antibody against SNX27 (revealed by fluorescein isothiocyanate [FITC]-conjugated goat polyclonal antibody against rabbit IgG). Bar, 10 μm. (B) A431 cells were fixed and processed for indirect immunofluorescence using a rabbit polyclonal antibody against SNX27 (revealed by FITC-conjugated goat polyclonal antibody against rabbit IgG) (a and e), a mouse monoclonal antibody against EEA1 (b), Golgi protein GS28 (f), or LAMP2 (g) (revealed by Alexa Fluor 555-conjugated goat polyclonal antibody against mouse IgG). The presence of yellow labeling indicates overlap in the merged images (d and h). Magnified images of the boxed regions in panels c and g are shown in panels d and h, respectively. (C) A431 cells were fixed and processed for indirect immunofluorescence assay using rabbit polyclonal antibody against SNX27 and mouse monoclonal antibody against α-adaptin and β-adaptin. SNX27 demonstrated colocalization with α-adaptin and β-adaptin of AP2. Magnified images of the boxed regions are shown in the right panels. (D) A431 cells were cultured in the absence of serum for 3 h, followed by treatment with (c and d) or without (a and b) wortmannin (25 nM) for 15 min. The cells were then fixed and processed for indirect immunofluorescence microscopy to detect SNX27 (a and c) and EEA1 (b and d). Like EEA1, endosomal association of SNX27 was abrogated upon treatment with wortmannin. Bars, 10 μm.

Because SNX27 colocalizes with EEA1, we investigated whether SNX27 might be involved in the endocytic trafficking pathway by performing double-labeling immunofluorescence with α-adaptin and β-adaptin, subunits of clathrin adaptor AP-2 involved in endocytosis from the plasma membrane. We found that some SNX27 also colocalized with both α-adaptin and β-adaptin (Fig. 2C), indicating that SNX27 is involved in the endocytic/early endosome trafficking pathway.

Since the PX domain of SNX27 is capable of interacting with PtdIns(3)P (data available upon request) and is essential for endosomal targeting, it is most likely that PX domain-mediated interaction with PtdIns(3)P is the mechanism responsible for SNX27-early endosome association. PI3K inhibitors, such as wortmannin, can inactivate several types of PI3-kinases and thus reduce the level of endosomal PtdIns(3)P (33, 43). Wortmannin was used to validate the PtdIns3P-dependent endosomal association of FYVE or PX domain proteins, such as EEA1, endofin, SNX3, and SNX16 (11, 29, 40, 50). Without wortmannin treatment, SNX27 and EEA1 were colocalized in the vesicular early endosome (Fig. 2D, top). When cells were treated with wortmannin (Fig. 2D, bottom), endosomal localization of SNX27 and EEA1 was significantly reduced compared to untreated cells (Fig. 2D, top). These results suggest that the endosomal association of SNX27 is dependent on the continuous generation of PtdIns(3)P by PI3K. Taken together, these findings indicate that the PX domain of SNX27 mediates its early endosomal association by interacting with endosomal PtdIns(3)P.

Generation of SNX27-deficient mice.

To begin to study the cellular and physiological functions of SNX27, we generated SNX27 knockout mice. The coding region of mouse SNX27 is derived from 11 exons, with the third coding exon encoding amino acids 179 to 243, which form the major part of the PX domain (residues 160 to 265) that is crucial for SNX27 to be targeted to the early endosome. A SNX27 gene-targeting construct was created by inserting the Neo cassette into the third coding exon (Fig. 3A). A SalI site was first introduced into this exon by site-directed mutagenesis, and a targeting vector was generated by inserting the Neo cassette into this site, with an 8-kb right arm and a 4-kb left arm flanking the insertion. The targeting vector was transfected into embryonic stem cells (E14) via electroporation, followed by G418 selection. The correctly targeted ES cells were used for blastocyst microinjection. Chimeras were used for crossing with S129 females to obtain the F1 generation. Crosses between F1 mice gave rise to wild-type (SNX27^+/+), heterozygous knockout (SNX27^+/−), and homozygous knockout (SNX27^−/−) mice, as assessed by a PCR-based method (Fig. 3B). To validate the absence of SNX27 protein in the knockout mice, lysates derived from the brains and kidneys of newborn pups were analyzed by immunoblotting analysis (Fig. 3C). As shown, the level of 60-kDa SNX27 polypeptide was reduced in the heterozygous knockout tissues and completely absent from the homozygous knockout tissues. The reduced levels of SNX27 protein in heterozygous mice is easily explained by a gene dose effect due to the presence of only one functional copy of the SNX27 gene. These results established that the knockout mice have lost the expression of SNX27. In addition, these results also confirmed that the antibodies are specific for SNX27.

Fig. 3.

Generation of SNX27-deficient mice. (A) Schematic drawing of the SNX27 genomic fragment and the targeting vector. (B) Genotyping of SNX27 knockouts. Shown is PCR analysis of genomic DNA isolated from mouse tails for detection of SNX27^+/+, SNX27^+/−, and SNX27^−/− mice. (C) Immunoblotting analysis of tissue lysates derived from the indicated mice using SNX27 antibodies, showing complete absence of SNX27 in the SNX27 knockout mice.

SNX27 deficiency results in growth retardation and lethality.

Genotyping of more than 150 newborn pups revealed that from a SNX27 double-heterozygous cross, only 16% (compared to the 25% expected) of pups born were homozygous knockout mice (Fig. 4A, a), suggesting that some homozygotes died in the uterus during embryonic development. In addition, the body weights of newborn SNX27^−/− mice (average, ∼1.03 g) were noticeably less than those of SNX27^+/+ and SNX27^+/− mice (average, ∼1.29 g) (Fig. 4A, b and c). These observations indicate that SNX27 plays an important, although not absolutely necessary, role in early embryo development, growth, and survival. Although most SNX27^−/− mice were viable at birth, they failed to thrive, and all died at different times within the first 3 weeks (Fig. 4B), whereas the heterozygotes behaved just like the wild type and liveD up to 24 months. The postnatal growth of SNX27^−/− mice was severely retarded (Fig. 4C), with a clear retardation of body weight gain. The growth retardation was not only reflected in the body weight, but also in multiple organs. The sizes of multiple organs, such as the spleen, kidney, liver, heart, and intestine, were significantly reduced in the SNX27^−/− mice (Fig. 4D). Taken together, these results suggest that SNX27 plays an essential role in postnatal growth and viability, and although its role in embryonic development and survival is obvious, it is not absolutely essential.

Fig. 4.

SNX27 deficiency caused growth retardation and reduced survival during embryonic development and severe postnatal growth retardation and lethality of newborn pups. (A) (a) Among 152 newborn pups, the ratio of SNX27^−/− mice is only 16%, lower than the expected 25%, indicating the loss of some SNX27^−/− embryos during development. Among 152 newborn pups, 25 are SNX27^−/−, 44 are SNX27^+/+, and 83 are SNX27^+/−. (b) Newborn SNX27^−/− mice are smaller than the wild-type pups in the same litter. (c) Body weight comparison among newborn SNX27^+/+, SNX27^+/−, and SNX27^−/− mice, showing that SNX27^−/− mice are significantly lighter. n = 10. Shown are means ± standard deviations. (B) The body weights during postnatal growth of SNX27^−/− and SNX27^+/+ pups were plotted as a function of time, showing severely delayed growth in body weight of SNX27^−/− pups. All SNX27^−/− mice died before being weaned. (C) (Left) Images of SNX27^−/− and wild-type mice from the same litter at 10 days after birth (P10). (Right) Images of SNX27^−/− and wild-type mice from the same litter at 20 days after birth, showing the dramatically reduced size of the former. (D) Images of major organs from SNX27^−/− and wild-type micee at 10 days after birth, showing reduced sizes of multiple organs of the SNX27^−/− mouse.

SNX27 interacts with NR2C.

Since SNX27 is an early endosomal protein with a PDZ domain, one possible cellular function of SNX27 is to regulate endosomal trafficking of membrane proteins containing type I PDZ-binding motifs. This hypothesis is supported by earlier works showing that PDZ-binding motif-containing surface proteins, such as Kir3 potassium channels, 5-HT4a receptor, and β₂-adrenoreceptor, interact with SNX27 (13, 22, 18). Alternatively, SNX27 may mediate endosomal recruitment of cytosolic proteins containing PDZ-binding motifs. This possibility is supported by the reported interaction of SNX27 with diacylglycerol kinase zeta and the cytokine-inducible protein CASP (23, 34). These two modes of action may both work for SNX27, depending on the interacting proteins. In order to study the molecular mechanism responsible for the observed growth retardation and lethality of SNX27^−/− mice, we used yeast two-hybrid screens to identify novel candidate SNX27-interacting proteins. Full-length SNX27 was used as the bait by inserting the coding region into the pGBKT7 vector. AH109 yeast cells transformed with pGBKT7-SNX27FL were mated with yeast Y187 pretransformed with human brain cDNA libraries (Clontech) and subsequently plated on the selective medium with the highest stringency (SD/−Leu/−Trp/−His/−Ade [quadruple dropout supplement {QDO}]). Sixty-eight colonies appeared after 2 weeks of growth selection. After further selection on QDO-X-α-Gal plates, 30 positive colonies were picked up, and the plasmids carried by these yeast cells were extracted and sequenced, revealing 15 independent candidate interacting proteins for SNX27 (Table 1), among which is NR2C.

Table 1.

SNX27-interacting candidates from yeast two-hybrid screening^a

Name	Accession no.	No. of coloniesb	PDZ-binding motif presentc
N-Methyl-d-aspartate receptor 2C subunit precursor (NMDAR2C)	NM_000835	1	Yes (SEV)
Homo sapiens connector enhancer of kinase suppressor of Ras 2 (CNKSR2)	NM_014927	3	Yes (THV)
Diacylglycerol kinase zeta; 104 kDa	NM_001105540	1	Yes (TAV)
Homo sapiens ribosomal protein S6 kinase, 90 kDa, polypeptide3 (RPS6KA3)	NM_004586	1	Yes (TAL)
Liver mitochondrial glutaminase (GA)	NM_013267	1	Yes (SMV)
Very large G protein-coupled receptor	NM_032119	1	Yes (THL)
E2a-Pbx1-associated protein (EB-1)	NM_181670	13	No
Homo sapiens dynactin 1 (p150, glued homolog, Drosophila) (DCTN1)	NM_004082	1	No
ATPase, Na+/K+ transporting, beta 2 polypeptide (ATP1B2)	NM_001678	2	No
Component 1, q subcomponent binding protein (C1QBP)	NM_001212	1	No
Homo sapiens trichoplein, keratin filament binding protein (TCHP)	NM_032300	1	No
RAN binding protein 9	NM_005493	1	No
Protein tyrosine phosphatase, non-receptor type (PTPN2)	NM_080423	1	No
Microtubule-associated protein 1A (MAP1A)	NM_002373	1	No
Rho GDP dissociation inhibitor (GDI) alpha	NM_004309	1	No

^aCandidate interacting proteins for SNX27 as revealed by a yeast two-hybrid screen.

^bNumber of colonies recovered in the screen.

^cEncoded protein contains a PDZ-binding motif.

Sequencing of the NR2C construct recovered from the yeast two-hybrid screen revealed that it contains the region encoding the C-terminal 171-residue region (residues 1066 to 1236) of NR2C, including the C-terminal PDZ-binding motif (SEV) (we termed this C-terminal region NR2C-CT). To validate the interaction of SNX27 with NR2C and to define the domain of SNX27 that is responsible for its interaction with NR2C, truncated mutations of SNX27 were made in the pGBKT7 vector and include SNX27NT, SNX27ΔPDZ, SNX27ΔPX, SNX27ΔRA, and SNX27CT (Fig. 5 A). These constructs, together with SNX27FL, were each transformed into yeast AH109, whereas the pGADT7/NR2cCT plasmid retrieved from the positive clone was retransformed into yeast Y187. The AH109 cells carrying various SNX27 mutant constructs were mated with Y187 cells carrying pGADT7/NR2C-CT. The resulting diploid yeast cells were grown on SD/−Leu/−Trp for 2 days. Then, the cells were streaked on highly stringent QDO medium lacking leucine, tryptophan, adenine, and histidine (Fig. 5B). The two SNX27 constructs lacking the PDZ domain (SNX27ΔPDZ and SNX27CT) cannot support growth under selection for interaction, while other SNX27 constructs harboring the PDZ domain (SNX27/NT, SNX27ΔPX, and SNX27ΔRA) had an ability similar to that of SNX27FL to support growth under selection. These results not only confirm the interaction of SNX27 with NR2C, but also suggest that the PDZ domain of SNX27 is essential for interaction with NR2C.

Fig. 5.

The PDZ domain of SNX27 is critical for interaction with NR2C. (A) Six SNX27 constructs (schematically depicted) were cloned in the pGBKT7 vector fused to GAL4BD and transformed into yeast AH109. The purified NR2C-CT in pGADT7 was retransformed into yeast Y187. (B) The indicated AH109 and Y187 yeast cells were mated and plated on leucine- and tryptophan-free SD plates (SD/−Leu/−Trp). To confirm interaction, mated yeast cells were grown on the highest-stringency medium lacking leucine, tryptophan, histidine, and adenine (SD/−Leu/−Trp/−His/−Ade).

We next tested whether the PDZ-binding motif of NR2C-CT is important for interaction with SNX27, as assessed by a coimmunoprecipitation assay. NR2C-CT and NR2C-CTΔSEV (amino acids 1066 to 1233) were subcloned into pDMyc-Neo for expression as N-terminal Myc-tagged fragments. HEK-293 cells were cotransfected with HA-SNX27 (or its mutants) and myc-NR2C-CT or myc-NR2C-CTΔSEV, and cell lysates were immunoprecipitated with anti-Myc antibodies (Fig. 6 B). The immunoprecipitates were analyzed by immunoblotting analysis with anti-Myc antibodies to determine the efficiency of the immunoprecipitation (Fig. 6C, middle), as well as anti-HA antibodies to detect the amounts of coimmunoprecipitated HA-SNX27 (Fig. 6C, top). Myc-NR2C-CT efficiently corecovered HA-SNX27FL (lanes 1) AND SNX27NT (lanes 3), but not SNX27ΔPDZ (lanes 2). Deletion of the SEV motif of NR2C-CT abolished its ability to corecover SNX27FL (lanes 4). These results further demonstrate that SNX27 interacts with NR2C-CT and that this interaction depends on the PDZ domain of SNX27 and the PDZ-binding motif of NR2C-CT.

Fig. 6.

The PDZ domain of SNX27 mediates interaction with the C-terminal PDZ-binding motif of NR2C. (A) Alignment of the C-terminal amino acid sequences of human, rat, and mouse NR2C showing the conserved class I PDZ-binding motif (SEV). (B) Diagrammatic representation of the NR2C and SNX27a deletion mutants. (C) HEK-293 cells were cotransfected with HA-SNX27 (or its indicated mutants) and Myc-NR2C-CT or Myc-NR2C-CTΔSEV. Total cell lysates were prepared and incubated with monoclonal antibody against Myc cross-linked to protein G-Sepharose (BD). The immunoprecipitates were resolved by SDS-PAGE and then processed for immunoblotting using rabbit polyclonal antibody against the HA tag (top) to assess the efficiency of coimmunoprecipitation as a measure of interaction. Following stripping, the immunoblots were reprobed with rabbit polyclonal antibody against the Myc tag (middle) to ensure the efficiency of immunoprecipitation of the Myc-tagged proteins. Five percent of the lysates used in the immunoprecipitations were also analyzed for the expression of HA-tagged SNX27 or its mutants (bottom). IP, immunoprecipitation; WB, Western blotting.

SNX27 deficiency leads to increased levels and impaired endocytosis of NR2C.

To establish the physiological relevance of the interaction of SNX27 and NR2C, we first validated the interaction of endogenous NR2C and SNX27 (Fig. 7 A). Mouse brain extracts were immunoprecipitated with antibodies against SNX27, and the resultant immunoprecipitate was analyzed by immunoblotting analysis. As shown, a noticeable amount of endogenous NR2C was corecovered with SNX27. To detect colocalization of NR2C with SNX27, a double immunofluorescence assay was subsequently performed on primary cultured neurons from the cerebellum, known to express significant levels of NR2C. As shown in Fig. 7B, there was significant colocalization between NR2C and SNX27 in cultured cerebellar neurons.

Fig. 7.

Interaction of endogenous SNX27 and NR2C; SNX27 deficiency leads to enhanced levels of NR2C. (A) Mouse brain extracts were immunoprecipitated with antibodies against SNX27, and the immunoprecipitates were resolved by SDS-PAGE, followed by immunoblot analysis using antibodies against SNX27 and NR2C. A noticeable amount of NR2C was coprecipitated with SNX27. (B) SNX27 colocalized with NR2C in cultured neurons. Primary cerebellar neurons from wild-type mice were cultured for 7 days and stained with SNX27 and NR2C. Colocalization of NR2C and SNX27 was observed in the merged image.

Newborn SNX27^−/− and wild-type pups from the same litter were sacrificed, and the brain lysates were analyzed by immunoblotting analysis. If the interaction of SNX27 with NR2C has a role in regulating endocytosis and/or endosomal sorting of NR2C to the lysosome, then we should expect a comparable increase of NR2C levels in vivo. The same brain lysates derived from newborn SNX27^−/− and wild-type pups were analyzed by immunoblotting (Fig. 8 A). As shown, the level of NR2C was increased by about 35 to 40% in the SNX27^−/− brain compared to the wild-type brain (Fig. 8A, right). To determine whether this mechanism is common to other NMDA subunits, we analyzed brain lysates for NR1 expression in SNX27^−/− and wild-type mice and found that there was no significant difference in protein expression levels for the NR1 subunit (Fig. 8A). To determine whether NR2C expression is also enhanced at the transcriptional level in SNX27 knockout mice, we performed real-time PCR to compare the mRNA levels of NR2C transcripts in brain tissues from wild-type and knockout mice. No significant difference in mRNA transcript levels was observed in wild-type and knockout mice (Fig. 8B).

Fig. 8.

Elevated NR2C expression and impaired NR2C endocytosis in SNX27-deficient cerebellar neurons. (A) Brain lysates from newborn wild-type or SNX27^−/− pups were resolved by SDS-PAGE, followed by immunoblotting using antibodies against SNX27, NR1, NR2C, or β-actin. The level of NR2C was increased in SNX27-negative brain lysate. The quantitative results of NR2C levels normalized to β-actin in four independent experiments are presented as means and standard deviations; the levels in the wild type were arbitrarily set as 100%. (B) NR2C mRNA transcript levels in wild-type or SNX27^−/− brains were quantified by real-time PCR. The mRNA transcript levels in wild-type and SNX27^−/− brains were not significantly different. The experiment was repeated three times, and similar results were obtained. (C) (Left) Cryosections of cerebellum from SNX27-deficient and wild-type mice stained with NR2C antibody. Significantly more NR2C was expressed in SNX27-deficient cerebellum than in wild-type cerebellum. (Middle) Quantitation of immunofluorescence intensity shown in the cryosections. (Right) Primary cerebellar neurons from SNX27-deficient and wild-type mice were cultured for 7 days and stained with NR2C antibody. Similar to panel A, significantly more NR2C was expressed in SNX27-deficient cerebellar neurons than in wild-type cerebellar neurons. (D) Primary cortical neurons from SNX27^+/+ and SNX27^−/− mice were transfected with Myc-NR2C on day 4 in vitro (4 DIV) and pulse-chased with anti-Myc antibody on 6 DIV. Surface NR2C receptors were labeled with AF-488 secondary antibody without permeabilization, while internalized NR2C receptors were labeled with AF-555 secondary antibody following permeabilization. Reduced NR2C endocytosis was observed in SNX27-deficient neurons compared to wild-type neurons.

To qualitatively examine the inverse relationship between SNX27 and NR2C expression, cryosections of cerebellum from SNX27-deficient and wild-type mice were prepared and stained with NR2C antibody (Fig. 8C, left). Significantly more NR2C protein was expressed in the SNX27-deficient cerebellum than in the wild-type cerebellum in both the molecular and granular layers. Moreover, significantly more NR2C was expressed in primary cerebellar neurons from SNX27-deficient mice than in those from wild-type mice (Fig. 8C, right).

Since SNX27 colocalized with clathrin adaptor AP2, in addition to its localization in the early endosome, we monitored the internalization of Myc-NR2C in cortical neurons via pulse-chase to determine whether the rate of endocytosis was affected. Primary cortical neurons from SNX27^+/+ and SNX27^−/− mice were transfected with Myc-NR2C and pulse-chased with anti-Myc antibody for various periods (Fig. 8D). Eighty-five percent more NR2C was endocytosed in SNX27^+/+ than in SNX27^−/− neurons (Fig. 8D, right), as evidenced by the greater number of internalized NR2C (red puncta) in SNX27^+/+ neurons at 5 and 15 min after pulse-chase than in wild-type neurons. In a separate experiment, leupeptin, a protease inhibitor, was added to the media 30 min before the pulse-chase and throughout the incubation at 37°C, and no significant effect was observed (data not shown). These results indicate that SNX27 may act to regulate endocytic trafficking of NR2C.

DISCUSSION

PX domain-containing proteins, such as SNXs, are emerging as important regulators of endocytic trafficking. Although the functions of many SNXs remain to be investigated, the available results suggest that different SNXs may regulate the sorting/trafficking of different proteins. For example, SNX1 and SNX2 have distinct and overlapping functions in vivo (38), and they both interact with Vps26, Vps29, and Vps35 to form mammalian retromers to regulate endosomal trafficking to the Golgi apparatus for proteins such as cation-independent mannose-6-phosphate receptor (37). In our present study, we first generated SNX27-deficient mice, which were found to have growth defects and increased lethality. Second, we established that interaction of the PX domain of SNX27 with PtdIns(3)P mediates targeting of SNX27 to the early endosome, which is consistent with several recent reports (22, 23, 34). Third, using SNX27-specific antibodies, we showed that SNX27 is widely expressed in cell lines and various rat tissues and that it interacts with NR2C via its PDZ domain. Finally, NR2C levels are elevated in SNX27-deficient mice, and its surface internalization is compromised in SNX27-deficient neurons.

Knockout of the SNX27 gene in mice revealed that SNX27 plays an essential role in postnatal growth and survival. Although the role of SNX27 during embryonic development is not absolutely necessary, as most SNX27^−/− embryos developed to term and were born, the defect in postnatal growth of SNX27^−/− pups is severe, with much delayed body weight gain, reduced sizes of multiple organs, and lethality before weaning. Interestingly, none of more than 150 SNX27^−/− pups survived to weaning under standard husbandry conditions. Close examination showed that the number of SNX27^−/− pups born is significantly less than predicted by Mendelian ratio (16% rather than the expected 25% of all pups born), and birth weight is significantly less than that of wild-type pups, which together indicate that SNX27 plays a vital role in embryonic development and survival.

The underlying mechanism responsible for SNX27's role in postnatal growth is currently unknown. Since the PDZ domain of SNX27 may potentially interact with multiple proteins containing the PDZ-binding motif in the early endosome, SNX27 may participate in the regulation of the trafficking and functions of many different proteins. The combined consequences of altered trafficking and functionality of multiple proteins in multiple organs may be the basis for the observed growth retardation and postnatal lethality. The availability of SNX27^−/− knockout mice will greatly facilitate the study of the trafficking and functionality of specific SNX27-interacting proteins in various cell types and tissues within the scientific community.

Since SNX27 has been shown to interact with several proteins containing PDZ-binding motifs, such as Kir3 potassium channels, 5-HT4a receptor, diacylglycerol kinase zeta, and the cytokine-inducible protein CASP, SNX27 may prove to be a general regulator at the early endosome for a multitude of proteins containing the PDZ-binding motif. To begin to address this issue, we used yeast two-hybrid interaction screens to identify candidate SNX27 proteins. Our screens have uncovered 15 proteins (Table 1) as candidate interacting proteins for SNX27. Among these 15 proteins is diacylglycerol kinase zeta, which has been shown recently to interact with SNX27 (34). The other 14 proteins are novel candidates and include, very interestingly, the ion channel receptor, NR2C, and the very large G protein-coupled receptor 1 (GPR98), together with insulin-stimulated kinase 1 (ISPK-1). Like diacylglycerol kinase zeta, NR2C, GPR98, and ISPK-1 contain C-terminal PDZ-binding motifs and are likely to interact with the PDZ domain of SNX27. The other 11 proteins do not contain C-terminal PDZ-binding motifs, and they may interact with SNX27 via other mechanisms, if their interactions can be validated.

The NMDA receptors are ligand-gated ion channels, which consist of one core NR1 and one or more NR2A, NR2B, NR2C, and/or NR2D subunits forming the heteromeric complex (17). Newly synthesized NMDARs are delivered to the neuronal surface by the secretory pathway (47). PSD-95 family or membrane-associated guanylate kinases (MAGUKs) bind the extreme C-terminal PDZ-binding motifs of NMDARs via the PDZ domain to anchor NMDARs to the synaptic sites (16, 48). The interaction of NMDARs with the PSD-95 family (including SAP102, SAP97, and PSD-93) stabilizes the number of surface NMDARs, which represents a balance between receptor internalization and insertion (35). Since the recovered NR2C gene clone in our yeast two-hybrid screens encodes the C-terminal region of NR2C with a PDZ-binding motif, and also, surface NMDARs undergo clathrin-mediated endocytosis and are internalized to the early endosome (17, 35), we have characterized the interaction of SNX27 and NR2C in greater detail. Yeast two-hybrid analysis of the interaction of NR2C with various SNX27 mutants showed that the PDZ domain is essential for the interaction, as all mutants lacking a PDZ domain no longer interacted, unlike the PDZ domain-containing mutants, which are capable of such an interaction. Coimmunoprecipitation experiments using cells expressing myc-NR2C-CT and HA-SNX27 showed that SNX27 is corecovered with NR2C-CT using anti-Myc antibodies. Furthermore, the corecovery of SNX27 with NR2C-CT was abolished when the PDZ domain was deleted. In addition, the ability of NR2C to corecover SNX27 was abrogated when the SEV motif was deleted from NR2C. These results suggest that the PDZ domain of SNX27 mediates interaction with the PDZ-binding motif of NR2C. Since endogenous NR2C could be coimmunoprecipitated with endogenous SNX27 when brain lysates were immunoprecipitated with SNX27 antibodies, the interaction of NR2C and SNX27 likely occurs in vivo.

Furthermore, in the absence of SNX27, the level of NR2C was increased in SNX27^−/− mice relative to wild-type mice, similar to Kir3.2c (22). Since the mRNA for NR2C remained unchanged in SNX27^−/− mice, the increase of NR2C is most likely due to altered trafficking in response to the absence of SNX27. In the present study, using antibody pulse-chase of Myc-tagged NR2C, we found that NR2C internalization is compromised in SNX27-deficient neurons. Based on the increased levels of Kir3.2 and NR2C, two validated interacting proteins of SNX27, in SNX27^−/− mice, it can be envisioned that the expression levels of other proteins whose endocytic trafficking is regulated by SNX27 may be similarly increased in SNX27^−/− mice. Systematic analysis of the abundance of various PDZ-binding motif-containing proteins by immunoblotting analysis or SILAC proteomics (28) may reveal proteins whose levels are increased in SNX27^−/− mice or derived cells, and these proteins are likely novel candidates whose endosomal sorting is regulated by interaction with the PDZ domain of SNX27. These future studies will deepen our understanding of SNX27 in regulating endocytic trafficking of an increasing number of proteins and their cellular and physiological roles.