(Nyctalopin on Chromosome X), the Gene Mutated in Congenital Stationary Night Blindness, Encodes a Cell Surface Protein

NYX (Nyctalopin on Chromosome X), the Gene Mutated in Congenital Stationary Night Blindness, Encodes a Cell Surface Protein

Christina Zeitz,1,2,3,4 Harry Scherthan,2,4 Susanne Freier,2 Silke Feil,1,2 Vanessa Suckow,2 Susann Schweiger,2 and Wolfgang Berger1,2

PURPOSE. The complete type of X-linked congenital stationary and are functionally conserved. (Invest Ophthalmol Vis Sci. night blindness (CSNB1) in human and mouse is caused by 2003;44:4184–4191) DOI:10.1167/iovs.03-0251 mutations in the NYX gene. The human NYX protein has been predicted to contain an N-terminal endoplasmic reticulum (ER) -linked congenital stationary night blindness (XLCSNB) is a signaling sequence and a C-terminal glycosylphosphatidylinosi- Xnonprogressive retinal disorder characterized by impaired tol (GPI) anchor. In the current study, these computer predic- night vision and other ocular symptoms such as myopia, hy- tions were verified experimentally by expression of domain- peropia, nystagmus, and reduced visual acuity.1 According to specific cDNA constructs in COS-7 and HeLa cells. Moreover, the clinical phenotype, XLCSNB can be divided into two types: computer-based analysis of the orthologue mouse amino acid complete (CSNB1) and incomplete (CSNB2). These two sub- sequence did not reveal a GPI anchor, which may result in a types can be distinguished on the basis of electroretinogram different protein localization compared with human NYX. (ERG) and by genetic means.2,3 CSNB2 is associated with a Therefore, the cellular localization for the mouse Nyx protein reduced rod b-wave and substantially reduced cone response was also examined. and is due to mutations in CACNA1F, an X chromosomal gene METHODS. A new method was established that differentially encoding the ␣1-subunit of an L-type calcium channel.4,5 visualizes both the protein at the surface of the living cell and CSNB1 is characterized by the complete absence of the rod subsequently in intracellular compartments. The localization of b-wave, but largely normal cone amplitudes. By positional the human and mouse V5-tagged wild-type and mutant NYX cloning we and others have isolated a gene in Xp11.4 (NYX, protein were studied. nyctalopin on chromosome X) that is mutated in patients with 6,7 RESULTS. Human and mouse V5-NYX proteins were dispersed in CSNB1. The gene consists of three exons and codes for a the form of speckles over the entire cell surface. Subsequent novel protein belonging to the leucine-rich repeat (LRR) super- staining of the same cells after detergent extraction revealed family. The predicted polypeptide contains an N-terminal ER that V5-NYX located to the ER and Golgi apparatus. Deletion of signal peptide, a C-terminal glycosylphosphatidylinositol (GPI) the GPI anchor domain of NYX resulted in a time-dependent membrane anchor, and 11 leucine-rich repeats (LRR), flanked loss of V5-NYX from the surface of living cells and accumula- by cysteine-rich LRRs. LRRs are short-sequence motifs present tion of this truncated protein in the ER and Golgi apparatus. in a number of proteins with diverse functions and cellular Deletion of the ER signal sequence in Nyx delocalized the localization. Functionally, these repeats are suggested to medi- 8 intracellular V5-Nyx protein and caused its dispersion in the ate protein–protein interactions. Several membrane proteins cytosol. Furthermore, mutations introduced in the leucine-rich have been shown to be linked to the outer leaflet of the lipid 9,10 repeat (LRR)-region, which has been described as a pathogenic bilayer through a GPI anchor. They are synthesized and variant of NYX, had no effect on subcellular localization of the attached at the endoplasmic reticulum (ER) before modifica- protein. tion and are then transported to the cell surface through the secretory pathway.11 The mammalian Golgi apparatus is the CONCLUSIONS. These data provide evidence that human and central organelle within the secretory machinery. It plays an mouse nyctalopin are membrane-bound extracellular proteins important role in processing, maturation, and sorting of newly synthesized secretory and membrane proteins received from the ER.12,13 GPI-anchored proteins may be localized in mi- From the 1Division of Medical Molecular Genetics and Gene Di- crodomains14 and have been implicated in processes such as agnostics, Institute of Medical Genetics, University of Zurich, Schwer- protein sorting.15 T4-cell activation,16 and cell signaling.17,18 zenbach, Switzerland; the 2Max-Planck Institute for Molecular Genet- Most notably, it has been shown that connectin and chaoptin, 3 ics, Berlin, Germany; and the Institute for Chemistry/Biochemistry, two prominent GPI-anchored LRR proteins, have a pivotal Free University, Berlin, Germany. 4 role in photoreceptor cell adhesion and axon guidance pro- Contributed equally to the work and therefore should be consid- 19,20 ered equivalent first authors. cesses. Supported by Grant T-GE-0602-0222 from The Foundation Fight- The amino acid sequence identity between the mouse or- ing Blindness. thologue, previously identified by screening a genomic mouse 21 Submitted for publication March 11, 2003; revised May 7, 2003; PAC library and human NYX is much higher in the LRR core accepted May 29, 2003. (Ͼ90%) than in the N- and C-terminal sequences (62% and 52%, Disclosure: C. Zeitz, None; H. Scherthan, None; S. Freier, None; respectively). Computer-based protein motif prediction re- S. Feil, None; V. Suckow, None; S. Schweiger, None; W. Berger, vealed the ER signaling peptide, the LRRs, and the GPI anchor None in the human protein sequence. However, only the signal The publication costs of this article were defrayed in part by page sequence and the LRRs were predicted in the mouse sequence. charge payment. This article must therefore be marked “advertise- ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact. To understand the function of NYX, we generated V5- Corresponding author: Christina Zeitz, Division of Medical Molec- tagged human and mouse cDNA expression constructs and ular Genetics and Gene Diagnostics, Institute of Medical Genetics, studied their subcellular localization by immunofluorescence University of Zurich, Schorenstrasse 16, CH-8603 Schwerzenbach, of tagged NYX protein on live and in fixed cells after transfec- Switzerland; [email protected]. tion into COS-7 and HeLa cells. Different protocols were estab-

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, October 2003, Vol. 44, No. 10 NYX, a Cell Surface Protein 4185

lished to identify the localization of V5-NYX. Although V5-NYX GAGCTAGAATTCTCACCGAAAGCTGTCATTCAACG-3Ј (MWG). KpnI/ showed a speckled pattern at the cell surface in living cells, it EcoRI– digested PCR products were isolated from agarose gel clearly colocalized with markers for ER and Golgi in detergent- (Qiaquick; Qiagen) and cloned in KpnI/EcoRI-digested PCR products treated cells. Identical results were obtained with mouse cDNA were isolated from agarose gels (Qiaquick; Qiagen) and cloned in a expression constructs. similarly treated pcDNA4/V5-His expression vector (Invitrogen). In addition, a decisive role of the signal sequence and the GPI anchor for the proper subcellular synthesis and localiza- Cell Culture and Transfection tion of the V5-NYX protein was demonstrated by using deletion constructs. COS-7 and HeLa cells were spread at 1 ϫ 105 cells per well of a six-well Our results show that the ER signaling peptide and GPI plate or at 3 ϫ 106 cells per flask in DMEM/F12 (Invitrogen), contain- anchor are of functional relevance, not only in humans but also ing3mML-glutamine, streptomycin (100 U/mL), penicillin (100 in the mouse, and that V5-NYX is a membrane bound, extra- U/mL), and 10% FCS, 1 day before transfection. Transfection was cellular protein. performed according to a commercial protocol (PolyFect; Qiagen), by using 1 or 15 ␮g plasmid DNA, 10 or 50 ␮L transfection reagent (PolyFect; Qiagen), and 100 or 600 ␮L serum-free medium (Opti-MEM; MATERIALS AND METHODS Invitrogen). The transfection efficiency for all experiments was 20%. Expression Constructs Immunofluorescence Expression constructs were generated by using a mammalian expression system (Invitrogen, Karlsruhe, Germany), in which pcDNA4/V5- COS-7 and HeLa cells were transfected as just described. After 24 His is used for gene fusion. Inserts were amplified with DNA Polymer- hours, transfected cells were analyzed using different staining meth- ase in a kit (HotStarTaq with Q-solution; Qiagen, Hilden, Germany) on ods. To observe proteins that localize exclusively on the plasma mem- genomic DNA and with a second DNA polymerase (ProofStart with brane, we performed live cell staining as follows. The anti-V5 antibody Q-solution; Qiagen) on purified amplified PCR products. To avoid (Invitrogen) was diluted 1:300 in an appropriate volume of cell culture overlong primers, wild-type and mutant inserts were amplified by medium containing 2 ␮L of a crude, heat-denatured lysate prepared using two forward primers. They contained the NYX signal sequence, from nontransfected HeLa or COS-7 cells. After this 10-minute absorp- a V5 recognition site (GGTAAGCCTATCCCTAACCCTCTCCTCG- tion step at 37°C, which removes nonspecific binding sites among the GTCTCGATTCTACG), and 20 nucleotides of the corresponding cDNA. anti-V5 antibodies, the antibody solution was added to the living The reverse primer, used in both PCRs, contained the NYX stop codon transfected cells in the culture dish. After 30 minutes of incubation at to avoid the C-terminal V5- and His-tags positioned downstream in the 37°C, cells were washed three times with medium to remove unbound vector. antibody. The live cells were then incubated for a further 10 minutes After the first PCR reaction, amplified products were purified by gel at 37°C with FITC-conjugated rabbit anti-mouse antibody (Dako, extraction (Qiaquick; Qiagen) and used as a template for the second Glostrup, Denmark) serum diluted 1:500 in culture medium. After PCR. unbound antibodies were removed by three washes with culture The forward primer used for the first PCR amplifying the insert of medium, cells were either instantly fixed in 2% formaldehyde, 1ϫ PBS the human wild-type and mutant expression construct was 5Ј- for 5 minutes, or incubated for a further 30 minutes in medium to CTCGGGAGATCTGGTAAGCCTATCCCTAACCCTCTCCTCGGTCT- determine whether there nyctalopin would be released from the CGATTCTACGTGCGCCCGCGCTTGTCCC-3Ј (polyacrylamide gel elec- plasma membrane. After the detection and fixation of surface proteins trophoresis [PAGE] purified; Invitrogen) and for the second PCR, on the live cell (as described earlier), the intracellular pool of this 5Ј-GCACTTGGTACCATGAAAGGCCGAGGGATGTTGGTCCTGCTT- protein was revealed by extracting the membranes of the cell with CTGCATGCGGTGGTCCTCGGCCTGCCCAGCGCCTGGGCCGTGGGG- 0.2% Triton X-100, 1ϫ PBS for 15 minutes. After three washes in PBS, GCCTGCGCCGGTAAGCCTATCCCTAACCC-3Ј (HPLC purified; Meta- the cells were incubated with preabsorbed anti-V5 in PBGT (0.1% BSA, bion, Martinsried, Germany). The reverse primers for both reactions 0.1% gelatin, 0.05% Tween 20, and 1ϫ PBS22) for 1 hour at RT. After were identical in both reactions: 5Ј-GATCGAGAATTCTCAGTCCATCT- cells were washed three times with PBGT, the primary antibody was Ј Ј GCAGGCCAAA-3 (MWG, Ebersberg, Germany). Amplification of the detected by incubation with Cy3-conjugated F(ab )2 fragment donkey insert of the mouse wild-type Nyx expression construct was performed anti-mouse antibody (AffiniPure; Dianova GmbH, Hamburg, Germany) with the first forward primer, 5Ј-GGTAAGCCTATCCCTAACCCTCTC- diluted 1:500 in PBGT for 30 minutes at RT. The cells were then CTCGGTCTCGATTCTACGTGTCTGCGGGCCTGCCCTG-3Ј (PAGE pu- washed three times with PBGT and postfixed for 5 minutes in 1% rified; Invitrogen); the second forward primer, 5Ј-GCACTTGGTACCAT- formaldehyde PBS to avoid color shifts between epitopes at different GCTGATCCTGCTTCTTCATGCGGTGGTCTTCAGTCTGCCCTAC- subcellular localization. ACCAGGGCCACCGAGGCCGGTAAGCCTATCCCTAACCC-3Ј (PAGE In some experiments, the ER was visualized in live cells by incu- purified, Invitrogen); and the reverse primer for both reactions, 5Ј- bating them with a dye (ER-Tracker Blue-white DPX; Molecular Probes, GATCGAGAATTCTCACTCCCTCTGCAGGCCCC-3Ј (Invitrogen). For Leiden, The Netherlands) diluted 1:1000 in medium. Detection of the expression constructs without the C-terminal or the signaling se- Golgi apparatus was performed with anti-human golgin-97 mouse quence of NYX and Nyx, wild-type plasmids described earlier were monoclonal CDf4 antibody (Molecular Probes) as follows. Cells that used as templates. To amplify the human and mouse insert without the had been stained for proteins on the surface and in the cytosol were GPI anchor, the following primer combination was used: 5Ј-GCAC- again fixed with 2% formaldehyde, 0.015 Triton X-100, and 1ϫ PBS for TTGGTACCATGAAAGGCCGAGGGATGTTG-3Ј (MWG) and 5Ј-GAGCT- 5 minutes. After a triple wash with 0.1% glycine and 1ϫ PBS, cells were AGAATTCTCAGGAGGAGAGGCTGTCGGA-3Ј (MWG); and 5Ј-GATCT- incubated with anti-human golgin-97, mouse monoclonal antibody AGGTACCATGCTGATCCTGCTTCTTCATG-3Ј (MWG) 5Ј-GAGCTA- (Molecular Probes) diluted 1:1000 in PBGT. After three further washes GAATTCTCACCGAAAGCTGTCATTCAACG-3Ј (MWG), respectively. in PBGT, cells were incubated with Cy5-conjugated goat anti-mouse Primers used for amplifying the mouse insert without the signal se- antibody (Dianova GmbH) diluted 1:1000 in PBGT. Finally, cells were quence were 5Ј-GATCTAGGTACCATGCTGATCCTGCTTGGTA- embedded in antifade mounting medium (Vectashield; Vector Labora- AGCCTATCCCTAACCC-3Ј (HPLC purified, Metabion) and 5Ј- tories, Peterborough, UK) containing 0.5 ␮g/mL 4Ј,6-diamidine-2Ј-phe- GAGCTAGAATTCTCAGGAGGAGAGGCTGTCGGA-3Ј (MWG). Primers nylindoledihydrochloride (DAPI) and inspected in a fluorescence mi- used for amplifying the mouse insert without the signal sequence and croscope (Axioscope 1; Carl Zeiss Meditec, Oberkochen, Germany) the GPI anchor were 5Ј-GATCTAGGTACCATGCTGATCCTGCTTG- equipped with a fluorescence imaging system (ISIS; MetaSystems, Alt- GTAAGCCTATCCCTAACCC-3Ј (HPLC purified, Metabion) and 5Ј- lussheim, Germany).

Downloaded from iovs.arvojournals.org on 09/24/2021 4186 Zeitz et al. IOVS, October 2003, Vol. 44, No. 10

Cell Fractionation and Western Blot Analysis mouse and the human wild-type V5-tagged constructs. Live cell staining revealed that both proteins showed a similar localiza- For immunochemical detection after 48 hours, transfected COS-7 or tion pattern at the extraplasmatic face of the plasma membrane HeLa cells were collected by centrifugation and homogenized in in both cell types (data not shown). These results suggest that HEPES buffer (0.32 M sucrose, 4 mM HEPES [pH 7.4], including despite low sequence conservation, the N- and C-terminal protease inhibitor cocktail; Complete Mini; Roche, Mannheim, Ger- sequences in human and mouse have a similar role. many) and 2% mercaptoethanol (Merck, Dietikon, Switzerland) with In a second step, live stained V5-NYX–overexpressing cells five to seven strokes with an ice-cold homogenizer with a loose pestle. were fixed, detergent-extracted and differentially stained for After cell disruption, cell debris was removed twice by centrifugation the intracellular V5-NYX pool with anti-V5 antibody. In cells at 1000g for 20 minutes. The supernatant was centrifuged at 100,000g expressing human and mouse V5-NYX protein at high levels, for 1 hour. The remaining supernatant was precipitated with a fourfold strong labeling throughout the cells was observed that con- excess of acetone at 20,000g,at4°C for 15 minutes. The remaining ␮ spicuously surrounded the nuclei (Fig 2D). This was the case in pellet was dissolved in 20 L yeast lysis buffer (2% Triton X-100, 100 both COS-7 and HeLa cells (Fig. 2C). To investigate whether mM NaCl, 10 mM Tris [pH 8.0], and 1 mM EDTA) and the detectable NYX is synthesized at the ER and passed trough the Golgi amount of expressed protein was loaded on 10% SDS-polyacrylamide apparatus, we incubated live cells with an ER tracer (ER- gel. Protein was transferred to polyvinylidene difluoride (PVDF) mem- Tracker; Molecular Probes) and imaged the cells under the brane (Roche Diagnostics, Mannheim, Germany) by electroblotting ϫ microscope (Fig. 2H). After image recording, the cells were and detected in PBST (0.1% Tween 20, 1 PBS) with mouse anti-V5 live stained for surface V5-NYX protein (Fig. 2H), fixed, and antibody (Invitrogen) diluted 1:3000 after blocking the membrane permeabilized and the intracellular pool of V5-NYX was de- overnight with 5% powdered milk (Protifar; Zoetermeer, The Nether- tected in a different color (Fig. 2G). Again anti-V5 live cell lands) and the anti-V5 antibody at 37°C for 15 minutes with a crude, staining showed that V5-NYX is attached to the membrane in heat denatured lysate prepared from nontransfected HeLa or COS-7 ϩ living COS-7 cells, when using either construct. Within the cells. Primary antibody was detected with goat anti-mouse IgG (H L) same cells, V5-NYX and the ER marker displayed a similar horseradish peroxidase–conjugated secondary antibody and chemilu- intracellular localization pattern with V5-NYX localizing to the minescence reaction (NEN Life Science Products-Perkin Elmer, perinuclear ER and a reticular network in the cell body— Rodgau-Ju¨gesheim, Germany). consistent with the NYX synthesis at the ER (Fig. 2E). In addition, colocalization of V5-NYX (Fig. 3D) and the Golgi- specific marker golgin-97 (Fig. 3C) was observed (Fig. 3A), RESULTS suggesting that NYX protein is transported to the Golgi. In Subcellular Localization of Human and control experiments with the empty vector or secondary antibodies alone, no specific fluorescence signals were observed Mouse V5-NYX (data not shown). Hence, it appears that NYX undergoes mem- To examine the subcellular localization of the NYX protein, we brane-bound synthesis at the ER, is transported through the expressed V5-tagged23 human NYX in COS-7 and HeLa cells Golgi apparatus, and finally localizes to the cell surface in both (Fig. 1A). Expression of the fusion protein was confirmed by cell culture systems investigated. This was consistent for both Western blot analysis of the pellet fraction from transfected the mouse and human V5-NYX proteins. cells with the anti-V5 antibody (Fig. 1B). Standard immunofluorescence staining is limited in its abil- Internal Versus External Protein Pools by ity to provide simultaneous visualization of proteins at the cell Retarded Live Cell Staining surface and in intracellular compartments. Therefore, we de- veloped a sequential staining method that first detects the To study the functional relevance of the GPI anchor domain of protein at the extraplasmatic face of the plasma membrane of NYX in mouse and human, expression constructs without the living cells and, after cross-linking fixation, reveals the distri- human anchor motif and mouse C-terminal sequence were bution of the same protein within the cell. generated and transfected in COS-7 and HeLa cells. Subse- Using live cell staining in V5-NYX-expressing COS-7 cells, quently, living cells were stained for V5-NYX and fixed directly fluorescence microscopic analysis of transfected cells revealed (0 minutes) or after additional 30 minutes of culturing in wild-type V5-NYX localized at the cell surface in a punctuate antibody-free medium. If the GPI anchor motif in the two pattern (Figs. 2A, 2B). To address whether the distribution of proteins is functional, the protein would be expected to be the fusion protein is cell-type–dependent, we expressed the retained at the extraplasmatic face of the plasma membrane. same construct in HeLa cells and obtained identical results. Protein with a nonfunctional GPI anchor was expected to be Control experiments with the empty vector plasmid did not seen at the surface at time 0, but should have been reduced or show fluorescence staining (data not shown). lost from the surface at the later time point. In agreement with Comparison of the sequences of the NYX proteins in human this reasoning, both wild-type and mutant cells immediately and mouse revealed that the degree of homology is variable fixed after V5-NYX staining on live cells displayed a similar throughout the protein. Although amino acid identity is very staining pattern at the surface. In cells transfected with wild- high in the LRR core (Ͼ90%), the N-terminal and, even more type V5-tagged DNA, the majority (75.5%) of the cells exhib- striking, the C-terminal sequences showed only average inter- ited V5-NYX at the surface and in the cytoplasm (Fig. 4), species homology (62% and 52%, respectively)21,24 (Fig. 1C). whereas only a few cells showed exclusively intracellular stain- Moreover, computer analysis (www.expasy.org/tools/ pro- ing. Only 50% of cells transfected with V5-NYX constructs with vided in the public domain by the Swiss Institute of Bioinfor- the GPI anchor (NYX⌬GPI) showed mutant V5-NYX at the cell matics, Geneva, Switzerland) does not predict a GPI anchor in surface at time 0. mouse Nyx. Although a hydrophobic amino acid stretch at the A more dramatic effect was seen when live cells were C terminus of Nyx protein could function as a GPI anchor stained for V5-NYX by culturing in antibody-free medium for an motif, the respective hydrophilic motif seems to be too short in additional 30 minutes before fixation. Whereas most of the comparison to known GPI anchors. To determine whether the cells expressing wild-type V5-NYX still showed membrane lo- less-conserved C terminus and signal sequence in mouse and calization of the protein after 30 minutes of cultivation, nearly human NYX influence the subcellular distribution of the two all of the cells expressing the GPI-deleted variant had not proteins, COS-7 and HeLa cells were transfected with the retained the protein at the surface of the plasma membrane

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, October 2003, Vol. 44, No. 10 NYX, a Cell Surface Protein 4187

FIGURE 1. Motif analysis of human and mouse nyctalopin. (A) Diagrams of cDNA constructs for NYX expression in COS-7 and HeLa cells. (a) The human (NYX) cDNA coding for a signal sequence, an LRR region and a GPI anchor and the same construct with its mouse orthologue. Deletion constructs without either the C-terminal (b) or the N-terminal sequence (c) or both of them (d) were cloned to analyze the function of these motifs. A V5 tag was inserted downstream of the signal sequence for detection of the protein on ﬂuorescence microscopy and Western blot analysis. (B) The high-speed pellet fraction was probed with an anti-V5 antibody and revealed proteins of the expected size. The calculated molecular masses of V5-tagged fusion proteins are as follows: V5-NYX (human construct), 55 kDa; V5-Nyx (mouse construct), 55kDa; V5-NYX⌬GPI (human construct without GPI anchor), 51 kDa; V5-Nyx⌬GPI (mouse construct without C-terminal sequence), 50 kDa; V5-Nyx⌬SS (mouse construct without signal sequence), 52 kDa; and V5-Nyx⌬GPISS (mouse construct without C-terminal and signal sequence), 49kDa. (C) Shaded box- es: amino acid identities. The degree of the amino acid sequence conservation between human and mouse in the signal sequence is 62% and in the GPI anchor region 52%.23,24

(Fig. 4). Only approximately 20% of these cells showed surface attach to the membrane. The V5-NYX staining pattern obtained staining, but this pattern appeared to be very weak. That the in COS-7 was identical with that in HeLa cells (data not shown). remaining 20% of the transfected cells still showed surface To determine whether the localization of mouse Nyx differs staining, even though weak, can be explained by the fact that from that of human NYX, the same series of experiments were 30 minutes of additional incubation with antibody-free medium performed with the wild-type mouse construct and a construct is not sufficient for the complete dissociation of mutant V5- without the C-terminal sequence. Again, in cells transfected NYX from the surface in cells strongly expressing this protein. with the mouse wild-type construct, most of the cells exhibited However, the weak surface staining and the significant reduc- V5-Nyx at the surface and in their interior, whereas only a few tion of the surface pattern suggest that V5-NYX without the cells showed exclusively intracellular staining (data not membrane anchor is synthesized at the ER and transported by shown). This was seen at time 0 of live cell staining and after the Golgi complex to the plasma membrane, where it fails to 30 minutes of further antibody-free cultivation and fixation. In

Downloaded from iovs.arvojournals.org on 09/24/2021 4188 Zeitz et al. IOVS, October 2003, Vol. 44, No. 10

FIGURE 3. Colocalization studies of expressed V5-tagged NYX and golgin-97 in COS-7 cells. (A) Merged image (magenta)oftheﬂuores- cent signal of V5-NYX (red) and of the anti-golgin antibody signal (blue). (B) Anti-V5 NYX FITC (green) revealed only a few signals at the cell surface, indicating that these cells have not yet implemented the protein in the extraplasmatic face of the plasma membrane. (C) Local- ization of the Golgi marker protein golgin-97 by Cy5-labeled secondary antibodies (blue). (D) V5-NYX protein signals (red) in overexpressing COS-7 cells. These cells were chosen for display because the moderate expression allowed for unequivocal identiﬁcation of the Golgi compartment.

alone or without the signal sequence and the GPI motif were transfected in HeLa cells. Sequential immunofluorescence staining against surface and intracellular V5-Nyx protein without the signal sequence showed that the plasma membrane FIGURE 2. Colocalization studies of intracellular V5-NYX and an ER staining was abolished in HeLa cells (Figs. 6B, 6F, respectively). marker in COS-7 cells. (A, E) Composite images of consecutive immu- For V5-Nyx proteins without the signaling sequence alone or nofluorescence staining showing V5-NYX at the extracellular membrane in green (FITC), the intracellular V5-NYX protein pool in red the signaling sequence and membrane anchor, a diffuse gran- (Cy3), the ER in white (ER tracer dye) and the nucleus stained in blue ular cytosolic distribution was detected (Figs. 6A, 6C, 6E, 6G). (DAPI). (B, F) Same cells as in (A) and (E), showing the surface staining No fluorescence staining was found in control experiments only (green channel). (C, G) Red channel of the cell shown in (B) and where plasmid without insert was used for the transfection (F) revealing the intracellular V5-NYX pool only. The protein is ex- (data not shown). pressed at high levels and concentrated around the nuclei. (D) Nuclei (counterstaining, DAPI, blue) of the cells of the left column.(H) Same cell as in (E) in vivo stained for the endoplasmic reticulum by an ER tracer (white) before fixation. The ER marker exhibits a reticulate distribution, which partially colocalizes with the endogenous V5- tagged NYX shown in (G).

cells transfected with V5-Nyx constructs without the C-terminal sequence (V5-Nyx⌬GPI) that contained the hydrophilic microdomain that was predicted to be too short to function as a GPI anchor, a reduced number of V5-NYX⌬GPI–expressing cells showed the protein at the surface and in their interior only when ﬁxed immediately after protein staining (t ϭ 0; Fig. 5A–C). After 30 minutes of further cultivation nearly all of the FIGURE 4. Measurement of internal versus cell surface protein pools transfected cells showed exclusive intracellular staining (Fig. by retarded live cell staining in HeLa cells. Live cells were incubated 5E–G). In summary, our results show that despite low se- with ﬁrst and second antibody in medium. (u) Internal and surface stained protein; (f) exclusively internal protein. At 0 min 75.5% of the quence conservation in the N- and C-terminal region of human transfected cells expressing wild-type V5-NYX showed internal stain- NYX and mouse Nyx and although no GPI anchor motif was ing, as well as surface staining. 24.5% of the transfected cells showed predicted to occur in the mouse protein, the corresponding exclusively internal staining. In contrast, 55.5% of the transfected cells protein motifs are functionally homologous. expressing mutant V5-NYX⌬GPI showed internal staining, as well as surface staining. 54.5% showed exclusively internal distribution. Thirty Subcellular Localization of V5-Nyx without the ER minutes after incubation in antibody-free medium 76.5% of the trans- Signaling Peptide fected cells expressing wild-type V5-NYX still showed surface and internal staining. Of the V5-NYX–transfected cells, 23.5% showed ex- To investigate whether the predicted signal sequence is re- clusively internal staining. In contrast, only 21.5% of the V5-NYX⌬GPI- quired for the intracellular transport of the mRNA to the ER transfected cells showed surface staining and internal staining at 30 membrane, constructs either without the signal sequence minutes, whereas 78.5% showed exclusively internal staining.

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, October 2003, Vol. 44, No. 10 NYX, a Cell Surface Protein 4189

function.2 This information suggests a defect in synaptic transmission between photoreceptors and second-order neurons—in particular, bipolar cells. The role in this process of NYX, the gene defective in patients with CSNB1, is not clear. However, because of specific domains of the predicted protein, it has been speculated that NYX encodes an extracellular membrane-anchored protein with leucine-rich motifs, presum- ably responsible for protein–protein interactions.7 Several adhesive proteins belonging to the leucine-rich repeat family (e.g., connectin and chaoptin) have been shown to be attached to the plasma membrane through a GPI anchor.8 To study the subcellular synthesis and localization of NYX, we expressed V5-tagged fusion constructs in COS-7 cells. Using live cell staining we localized V5-NYX to the extraplasmatic face of the plasma membrane where it displayed a speckled distribution. This specific localization of V5-NYX is consistent with previous findings which showed that GPI-anchored proteins are localized in microdomains (200–300 nm) or “rafts” on the plasma membrane.14,26 These rafts could dictate the sorting of associated proteins and/or provide sites for assembly of cytoplasmic signaling molecules.27 Next, we were interested in the intracellular distribution of NYX. Therefore, we differentially

FIGURE 5. Immunoﬂuorescence staining of HeLa cells expressing V5- Nyx without the C-terminal sequence. (A–C) A cell that expresses Nyx without the C-terminal sequence showed intracellular localization (red) and a reduced speckled surface staining pattern (green) directly after in vivo immunostaining. (E–G) Cell that lost surface staining after 30 minutes of cultivation in antibody-free medium. Most of the expressing cells showed intracellular protein localization exclusively. (C, G) Intracellular V5-Nyx⌬GPI was present in a clear reticular staining (most obvious in the periphery) which indicates accumulation of the protein without the C-terminal sequence in the ER. (D, H) Counter- staining of DNA (DAPI) reveals the nuclei of transfected cells above.

Subcellular Localization of Mutant V5-NYX To study the consequences of mutations previously identiﬁed in CSNB1, we expressed V5-tagged mutated variants in COS-7 and HeLa cells. Mutations were introduced by direct ampliﬁ- cation of the open reading frame on the DNA of patients. In this way, a deletion of the amino acid sequence AELP at position 243-246, a missense mutation leading to an L3P exchange at position 347 in the LRR core and another missense mutation leading to a G3V at position 370 were studied. None of them altered the subcellular localization of the protein (data not shown). FIGURE 6. Functional analysis of the signal sequence and the GPI anchor. (A) Diffuse cytosolic distribution of V5-Nyx without the puta- DISCUSSION tive ER signal sequence and (E) the signal sequence and the last 26 Rod-mediated vision depends on the transmission of visual amino acids of Nyx in HeLa cells. (B, F) Deletion of the signal sequence of Nyx causes loss of protein at the plasma membrane in HeLa cells. (C, information from the photoreceptor outer segment through G) Intracellular protein shows a granular cytoplasmic staining, with the the retina. Visual information is transmitted postreceptorally to 25 typical ER “honeycomb” structure and the perinuclear Golgi staining a depolarizing class of bipolar cells. being absent (compare Figs. 2 and 3 with Fig. 6). (D, H) Cells were The complete type of congenital stationary night blindness embedded in mounting medium containing DAPI to stain the nuclei of is characterized by the absence of postreceptoral rod-mediated transfected cells.

Downloaded from iovs.arvojournals.org on 09/24/2021 4190 Zeitz et al. IOVS, October 2003, Vol. 44, No. 10

stained V5-NYX—first, on the surface of live cells and then the assumption that the absence of the GPI anchor results in after fixation within the same cell. V5-NYX is first localized to the failure of the folded polypeptide to attach the glycolipid the perinuclear ER and is subsequently accumulated in the anchor that is essential for binding to the inner ER membrane Golgi apparatus, as verified by colocalization experiments with and to the extraplasmatic face of the plasma membrane. The ER- and Golgi-specific markers. These findings are in accor- remaining protein at the surface of expressing cells is presum- dance with the ER-bound synthesis of membrane-bound pro- ably due to the high amount of protein expressed in single teins that are attached to the inner ER membranes before cells. It is still trafficked through the ER (targeted by the modification and transport to the cell surface through the N-terminal ER signal sequence) and Golgi to the cell surface but secretory pathway.11 The Golgi apparatus, which serves as the it fails to attach. central organelle within the secretory pathway, plays an im- Deletion of the N-terminal amino acids alone or in addition portant role in processing, maturation, and sorting of newly to the removal of the GPI anchor resulted in a failure of synthesized proteins received from the ER.12,13 Colocalization membrane-bound V5-NYX synthesis, leading to an enrichment of V5-NYX with the Golgi suggests that the protein may be of this protein in the cytosol. The typical perinuclear, reticu- modified (e.g., glycosylated), an assumption that requires fur- late, and surface staining pattern obtained with the full-length ther analysis. constructs was completely lost. This is consistent with the To exclude cell-type–specific differences in subcellular dis- N-terminal amino acids containing an essential part of the ER tribution of fusion proteins that have been observed for other signal sequence for the secretory pathway through an ER– proteins,28,29 we investigated the distribution of nyctalopin, Golgi-like compartment. not only in COS-7 but also in HeLa cells. V5-NYX was found to Our results agree with the expectation that correct process- localize to the extracellular surface in both cell types. Thus, our ing of the GPI anchor in the ER-Golgi network is essential for results indicate that the localization of V5-NYX at the plasma the successful secretion of nyctalopin, which is additionally membrane is probably cell-type independent. dependent on an N-terminal sequence. To study the effect on Amino acid sequence alignment of NYX to its mouse ortho- the localization of mutations, we synthesized three different logue showed that the LRR core is most highly conserved mutant constructs (Leu347Pro, ⌬AELP, and Gly370Val) and (Ͼ90%), whereas the signal sequence and the GPI anchor transfected them in the same way as wild-type constructs. region showed lower sequence identity (62% and 52%, respec- These mutations were chosen because they are all found in the tively).21,24 Previously, the gene and the mutation responsible LRR region, which has been described to contain most of the for the mouse nob (no b-wave) phenotype was identified.30 disease-associated mutations in humans.7 However, we ob- The investigators confirmed that the nob mouse is a model for served no obvious change in the distribution of the mutated human CSNB1 and showed that the similar phenotype is proteins (data not shown). Because these mutations did not caused by an 85-bp deletion in the mouse Nyx gene. Based on show any effect on the localization, our data suggest that the computer prediction, Gregg et al.30 also suggested that the location per se of NYX is necessary but not sufficient for putative GPI anchor site in the human has no counterpart in function. Further studies will have to show which proteins the mouse Nyx protein. However, these conclusions hitherto may interact with nyctalopin and determine whether muta- remained untested. Our data now provide firm evidence that tions disrupt these interactions leading to the complete type of the subcellular locations of both the human and mouse wild- XLCNSB and a similar phenotype in the nob mouse. However, type NYX proteins are identical. V5-NYX in both species local- our findings add NYX to a number of other GPI-anchored LRR izes at the extraplasmatic face of the plasma membrane and to proteins such as chaoptin and connectin, which have a pivotal the ER and Golgi apparatus inside the cells. Therefore, it is role as cell adhesion molecules in neuronal development in possible that the computer-based predictions are not of rele- Drosophila.20,32 Whereas chaoptin is located on photorecep- vance for the mouse Nyx protein. tor cells,32 connectin is distributed on the surface of a subset of In vivo localization analysis of a protein deleted for the embryonic muscles, on the growth cones and axons of the C-terminal sequence of the mouse V5-Nyx protein showed that motor neurons that innervate theses muscles, and on several this domain is required for membrane binding, clearly indicat- associated glial cells.20 Biochemical analysis revealed that cha- ing functional conservation of the respective protein motifs in optin is linked to the extracellular surface of the plasma mem- human and mouse NYX. A similar situation has been reported brane by covalent attachment to glycosylphosphatidylinositol. in which a GPI anchor motif had been predicted for only one It has been suggested that chaoptin plays an important role in of the functionally conserved homologous cell surface pro- photoreceptor cell morphogenesis to organize closely opposed teins, whereas both proteins were capable of mediating GPI membranes of photoreceptor cells by direct adhesive interac- anchor attachments.31 tions.19 Connectin has been shown to reveal a repulsive func- Time-course cell staining is a powerful method for analysis tion during motor neuron growth cone guidance and synapse of functionality of GPI anchors. The typical punctate staining formation.20 for wild-type V5-NYX on the cell surface was retained after in In the current study, we showed that NYX is a GPI-an- vivo staining and subsequent 30-minute culturing in antibody- chored, LRR-cell surface protein that localizes in vivo in foci free medium. In contrast, removal of the GPI anchor resulted in over the cell surface. One could thus hypothesize that NYX a reduction of surface-bound V5-NYX protein: only 50% of plays an important role in the human retina by establishing transfected cells showed staining at the cell surface when and/or specifying contacts between rod photoreceptors and stained in vivo for V5-NYX. When cells were cultured for an postsynaptic neurons, including bipolar, amacrine, and gan- additional 30 minutes in antibody-free medium before detec- glion cells. Alternatively, this protein could act as a ligand tion of V5-NYX, they displayed a defect in protein retention. and/or a receptor molecule in a cell-signaling pathway in the Only 20% of expressing cells showed a few punctate V5- human retina. Future studies are needed to show whether NYX NYX⌬GPI signals at the plasma membrane, suggesting that the is able to mediate homophilic adhesion of retinal cells. V5-NYX⌬GPI protein seen directly after immunostaining was unable to bind tightly at the cell surface. This is probably due Acknowledgments to a constant flow of overexpressed protein through the secretory pathway, which cannot be anchored to the cell surface. CZ thanks Thomas Haaf for introduction to fluorescence microscopy. Intracellular, anchorless V5-NYX protein was associated with The authors thank Barbara Kloeckener for helpful discussions and the reticulate ER network. These findings are consistent with critical reading of the manuscript, and Hans-Hilger Roper for support.

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, October 2003, Vol. 44, No. 10 NYX, a Cell Surface Protein 4191

References 17. Stulnig R, Berger M, Sigmund T, Stockinger H, Horejsi V, Waldhaüsl W. Signal transduction via glycosyl phosphatidylinosi- 1. Heon E, Musarella MA. Congenital stationary night blindness: a tol-anchored proteins in T cells is inhibited by lowering cellular critcal review for molecular approaches. Wright AF, Jay B, eds. cholesterol. J Biol Chem. 1997;272:19232–19247. Molecular Genetics of Inherited Eye Disorders. Chur, Switzerland: 18. Rodgers W, Crise B, Rose JK. Signals determining protein tyrosine Harwood Academic; 1994:277–301. kinase and glycosyl-phosphatidylinositol-anchored protein target- 2. Miyake Y, Yagasaki K, Horiguchi M, Kawase Y, Kanada T. Congen- ing to a glycolipid-enriched membrane fraction. Mol Cell Biol. ital stationary night blindness with negative electroretinogram: a 1994;14:5384–5391. new classification. Arch Ophthalmol. 1986;104:1013–1020. 19. Krantz DE, Zipursky SL. Drosophila chaoptin, a member of the 3. Boykott KM, Pearce WG, Musarella MA, et al. Evidence for genetic leucine-rich repeat family, is a photoreceptor cell-specific adhe- heterogeneity in X-linked congenital stationary night-blindness. J sion molecule. EMBO J. 1990;6:1969–1977. Hum Genet. 1998;62:865–875. 20. Nose A, Takeichi M, Goodman CS. Ectopic expression of connec- 4. Bech-Hansen NT, Taylor MJ, Maybaum TA, et al. Loss-of-function tin reveals a repulsive function during growth cone guidance and mutations in a calcium-channel alpha-1-subunit gene in Xp11.23 synapse formation. Neuron. 1994;13:525–539. cause incomplete X-linked congenital stationary night blindness. 21. Pesch K, Zeitz C, Fries JE, et al. Isolation of the mouse nyctalopin Nat Genet. 1998;19:264–267. gene Nyx and expression studies in mouse and rat retina. Invest 5. Strom TM, Nyakatura G, Apfelstedt-Sylla E, et al. An L-type channel Ophthalmol Vis Sci. 2003;44:2260–2266. gene is mutated in incomplete X-linked congenital stationary night 22. Scherthan H, Jerratsch M, Li B, et al. Mammalian meiotic blindness. Nat Genet. 1998;19:260–263. telomeres: protein composition and redistribution in relation to 6. Bech-Hansen NT, Naylor MJ, Maybaum TA, et al. Mutations in NYX, nuclear pores. Mol Biol Cell. 2000;11:4189–4203. encoding the leucine-rich proteoglycan nyctalopin, cause X-linked 23. Southern JA, Young DF, Heaney F, Baumgartner W, Randall RE. complete congenital stationary night blindness. Nat Genet. 2000; Identification of an epitope on the P and V proteins of simian virus 26:319–323. 5 that distinguishes between two isolates with different biological characteristics. J Gen Virol. 1991;72:1551–1557. 7. Pusch CM, Zeitz C, Brandau O, et al. The complete form of X-linked congenital stationary night blindness is caused by muta- 24. Zeitz C, Schweiger S, Scherthan H, Freier S, Berger W. Functional tions in a gene encoding a leucine-rich repeat protein. Nat Genet. characterization of protein domains in NYX (nyctalopin on chro- 2000;26:324–327. mosome X) (Abstract). Am J Hum Genet. 2002;71:326. 8. Kobe B, Deisenhofer J. The leucine-rich repeat: a versatile binding 25. Daw NW, Jensen RJ, Brunken WJ. Rod pathways in mammalian retinae. Trends Neurosci. 1990;1:110–115. motif. Trends Biochem Sci. 1994;19:415–421. 26. Varma R, Mayor S. GPI-anchored proteins are organized in submi- 9. Ferguson MA. Glycosyl-phosphatidyinositol membrane anchors: cron domains at the cell surface. Nature. 1998;394:798–801. the tale of a tail. Biochem Soc Trans. 1992;20:243–256. 27. Simon K, Ikonen E. Functional rafts in membranes. Nature. 1997; 10. Takeda J, Kinoshita T. GPI-anchor biosynthesis. Trends Biochem 387:569–570. Sci. 1995;20:367–371. 28. Schwahn U, Paland N, Techritz S, Lenzner S, Berger W. Mutations 11. Barz WP, Walter P. Two endoplasmatic reticulum (ER) membrane in the X-linked RP2 gene cause intracellular misrouting and loss of proteins that facilitate ER-to-Golgi transport of glycosylphosphati- the protein. Hum Mol Gen. 2001;10:1177–1183. dylinositol-anchored proteins. Mol Cell Biol. 1999;10:1043–1059. 29. Mayor S, Rothberg KG, Maxfield FR. Sequestration of GPI-an- 12. Palade G. Intracellular aspects of the process of protein secretion. chored proteins in caveolae triggered by cross-linking. Science. Science. 1975;189:347–358. 1994;264:1948–1951. 13. Rothman JE. Mechanisms of intracellular transport. Nature. 1994; 30. Gregg RG, Mukhopadhyay S, Candille SP, et al. Identification of the 372:55–63. gene and the mutation responsible for the mouse nob phenotype. 14. Friedrichson T, Kurzchalia TV. Microdomains of GPI-anchored Invest Ophthalmol Vis Sci. 2003;44:378–384. proteins in living cells revealed by crosslinking. Nature. 1998;394: 31. Voth BR, Kell BL, Phalgun BJ, Ivens AC, McMaster WR. Differen- 802–805. tially expressed Leishmania major gp63 genes encode cell surface 15. Brown DA, Rose JK. Sorting of GPI-anchored proteins to glyco- leishmanolysin with distinct signals for glycosylphosphatidylinosi- lipid-enriched membrane subdomains during transport to the api- tol attachment. Mol Biochem Parasitol. 1998;93:31–41. cal cell surface. Cell. 1992;68:533–544. 32. Zipursky S, Venkatesh T, Teplow DB, Benzer S. Neuronal devel- 16. Resta R, Thompson LF. T cell signaling through CD73. Cell Signal- opment in the Drosophila retina: monoclonal antibodies as mo- ling. 1997;9:131–139. lecular probes. Cell. 1984;36:15–26.

Downloaded from iovs.arvojournals.org on 09/24/2021