The zebrafish mutant lbk/vam6 resembles human multisystemic disorders caused by aberrant trafficking of endosomal vesicles

Vesicle trafficking is essential for a wide range of cellular processes and defects in vesicle trafficking have been shown to underlie various human diseases. In the arthrogryposis-renal dysfunction-cholestasis (ARC) syndrome,for example, mutations in the VPS33B gene lead to a dysregulation of vesicle fusion, including lysosomes and late stage endosomes(Gissen et al., 2004) (OMIM#208085). Similarly, the Chediak-Higashi syndrome (CHS), Hermansky-Pudlak syndrome (HPS) and Griscelli syndrome (GS) have been linked to a number of proteins involved in the formation, trafficking and fusion of lysosome and lysosome-related organelles [OMIM #214500 (CHS), #203300 (HPS), #214450 (GS1),#607624 (GS2), #609227 (GS3)]. These inherited human diseases are associated with hypopigmentation, prolonged bleeding and immunological defects. It has also been reported that affected individuals display visual defects, including photophobia, strabismus and nystagmus. Although mutations in several genes have been linked to these diseases, it has not yet been possible to identify the molecular defects in all cases.

Lysosome-related organelles are cell type-specific organelles that share several physiological properties with lysosomes(Raposo and Marks, 2002). They include organelles, such as melanosomes in skin and RPE melanocytes, lamellar bodies in type II lung epithelial cells, dense granules in platelets and MHC class II compartments in antigen-presenting cells, and cytotoxic granules of lymphocytes. Tethering and docking of lysosomes and lysosome-related organelles depends on the HOPS complex (homotypic fusion and vacuole protein sorting complex). This complex interacts with both SNAREs and Ypt7p/Rab7(Collins et al., 2005; Price et al., 2000a; Price et al., 2000b; Sato et al., 2000; Wurmser et al., 2000) and is required for SNARE complex assembly(Stroupe et al., 2006). Furthermore, the HOPS complex seems to be important for early-late endosome transition in the Rab-conversion model in mammalian cells(Rink et al., 2005). The HOPS complex comprises the class C Vps protein Vam5p/Vps33p, the protein defective in ARC syndrome; the class C proteins Vam1p/Vps11p, Vam9p/Vps16p and Vam8p/Vps18p; and two class B Vps proteins, Vam6p/Vps39p and Vam2p/Vps41p(Rieder and Emr, 1997; Seals et al., 2000; Wurmser et al., 2000)(reviewed by Bowers and Stevens,2005).

Vam6p/Vps39p was shown to be required for vacuolar protein sorting in yeast. Yeast Vps39p-null mutants exhibit highly fragmented vacuolar morphogenesis and mutant cells accumulate numerous vesicular structures scattered throughout the cytoplasm(Nakamura et al., 1997). Similarly, overexpression studies in cultured HEK cells have implicated Vam6p in the clustering and fusion of lysosomes and late endosomes(Caplan et al., 2001) and blocking Vam6p function with antibodies suggests a possible role in sperm cells during their interactions with the zona pellucida in mice(Brahmaraju et al., 2004).

We have identified a zebrafish mutant (leberknödel; lbk) that carries a mutation in the zebrafish homologue of Vam6p/Vps39p and displays phenotypes similar to those observed in humans suffering from ARC syndrome, CHS, HPS and GS. Our studies reveal multisystemic defects in lbk, including a hypopigmentation of skin melanocytes and the retinal pigment epithelium (RPE). Moreover, lbk displays defects in internal organs (liver, intestine) and the innate immune system. Affected cells display increased numbers and enlarged intracellular vesicles. Physiological and behavioural analyses of visual function in lbkuncovered a reduced visual ability. These analyses suggest that Vam6p/Vps39p has essential functions in a range of tissues during zebrafish development. Notably, no animal model to study the function of Vam6p/Vps39p in vivo has been reported to date.

Zebrafish (Danio rerio) were kept under standard conditions. Strains used in this study were: Tü(Haffter et al., 1996), WIK(Dahm et al., 2005) and lbk^t20290. Staging of embryos in hours (hpf) and days post-fertilization (dpf) was carried out as previously described(Kimmel et al., 1995).

Transplantations were performed as described(Ho and Kane, 1990). Mosaic animals were generated by transplanting 30-40 cells from 3-4 hpf wild-type Tü embryos into the animal pole region, including prospective ectodermal domains, such as the eye and neural crest domains(Kimmel et al., 1995), of age-matched embryos obtained by crossing two lbk^+/-carriers. Larvae displaying a clear lbk^-/- phenotype were scored for the presence of skin melanocytes and RPE cells displaying wild-type levels of melanin.

For histological analyses by light microscopy (LM), larvae were fixed in 4%paraformaldehyde (PFA) in PBS (pH 7.2) at 4°C overnight and washed three times in PBS. The embryos were dehydrated in a standard ethanol series,infiltrated and embedded in Technovit 7100 (Heraeus Kulzer, Germany) for sectioning. Sections (3 μm) of different developmental stages of mutant and wild-type larvae were cut with a glass knife and mounted on Superfrost Plus slides (Microm International, Switzerland). The sections were subsequently stained with 0.5% Toluidine Blue in a 1% sodium tetraborate buffer (pH 9.2)and analysed using an Axioscope 2Mot (Zeiss, Jena, Germany) connected to an AxioCamHR colour camera using the AxioVision 3.0 software (Zeiss).

For ultrastructural analysis by transmission electron microscopy (TEM),larvae were fixed in 2% PFA and 2.5% glutaraldehyde in 0.1 M HEPES buffer (pH 7.2) at 4°C overnight and subsequently washed three times in HEPES buffer. Larvae were then postfixed with 1% OsO₄ in 100 mM PO₄buffer (pH 7.2) for 1 hour on ice, washed with double distilled H₂0, treated with 1% aqueous uranyl acetate for 1 hour at 4°C,dehydrated through a graded series of ethanol, infiltrated with ethanol/resin mixtures and embedded in Epon (using glycidether 100 from Roth, Karlsruhe). Ultra-thin sections were collected on coated slot grids, stained with uranyl acetate and lead citrate and viewed in a Philips CM 10 electron microscope.

For immunohistochemistry, larvae were fixed in 4% PFA in PBS (pH 7.2) for 45 minutes at room temperature, cryoprotected in 30% sucrose for at least 2 hours, embedded in OCT TissueTek (Jung-Leica; Tissue Freezing Medium) and frozen in liquid nitrogen (N₂). Sections (20 μm) were cut at-20°C, mounted on SuperFrost Plus slides, fixed in ice-cold acetone for 1 minute and stored at -20°C. For further use, slides were thawed,air-dried, washed three times in PBS (0.5×, pH 7.4) and blocked with a solution of 20% normal goat serum and 2% bovine serum albumin in PBS containing 0.3% Triton X-100 (PBST) for 1 hour. Sections were then incubated for 2 hours at 4°C in the primary antibody. Primary antibodies used in this study were anti-glutamine synthetase (1:700; Chemicon, Temecula, USA) and anti-rhodopsine (1:200; MilanAnalytica, LaRoche, Switzerland). Antibodies were diluted in PBST. After three washes in PBST, sections were incubated in Alexa546-coupled anti-mouse secondary antibody (1:500; Molecular Probes/Invitrogen, Eugene, OR) for 1 hour, washed three times in PBST, mounted in glycerol and analysed with an Axioscope fluorescence microscope as described above. Apoptosis was detected using the TUNEL (TdT-mediated dUTP nick-end labelling) method on 35 μm cryosections. TUNEL staining was performed using the Cell Death Kit (Roche Molecular Biochemicals) according to the manufacturer's protocol. Oil red O in dextrin was used to visualise lipid-containing vesicles in cryosections (20 μm). Confocal laser scanning microscopy (CLSM) was performed with an LSM510 (Zeiss) and 25×PlanNeofluar and 63× PlanApochromat oil immersion objectives(Zeiss).

Map crosses were set up between heterozygous lbk^+/-(Tü background) and wild-type WIK zebrafish. The offspring from these crosses were inbred and homozygous lbk^-/- and sibling F2 progeny were collected and their DNA extracted. Bulked segregant analysis on 48 lbk^-/- larvae and 48 siblings, respectively, was carried out using 192 simple sequence length polymorphisms (SSLP markers)distributed over the entire genome(Geisler, 2002). Further fine mapping was performed using the total DNA of single homozygous mutant larvae. DNA extraction and PCR were performed as described(Geisler, 2002). The zebrafish vam6 gene was radiation hybrid (RH) mapped on the Goodfellow T51 RH panel as described (Geisler,2002) using six independent primer pairs(Table 1). PCRs for RH mapping were carried out independently in triplicate.

A putative zebrafish vam6 gene was identified via a homology search of the Zv5 Ensemble database(http://www.ensembl.org/Danio_rerio)using the human Vam6p protein sequence as query. Total RNA was isolated from adult zebrafish eyes using the RNAeasy kit (QIAGEN, Switzerland) according to the manufacturer's instructions. cDNA was synthesised using the SuperScript III Reverse Transcriptase Kit (Invitrogen, Switzerland) as described in the manual. The zebrafish vam6 cds was cloned using primers designed against the start and stop regions of the vam6 cDNA(Table 2), respectively, using homology with the human VAM6 gene (VPS39 - Human Gene Nomenclature Database). To obtain 5′ sequences of the vam6cDNA, RACE-PCR was carried out with the SMART RACE cDNA Amplification Kit(Clontech, Switzerland). For RACE primers used to amplify 5′ regions of the vam6 cDNA, see Table 2. PCR products were cloned into the pCRII-Vector (Clonetech),sequenced with T7 and SP6 primers and submitted to GenBank (Accession Number,EF446162).

Heterozygous lbk^+/- carriers were crossed to a collection of ENU-mutated Tü zebrafish and the offspring screened for phenotypes equivalent to that of lbk^-/-, indicating non-complementation of the novel allele with the lbk mutation. Total genomic DNA from 5 dpf larvae showing an lbk-like phenotype was extracted, all 24 exons of the vam6 gene were amplified and the PCR products sequenced (Table 3). All PCR and sequencing reactions were performed independently in triplicate.

Optokinetic response (OKR) measurements were performed as described previously (Rinner et al.,2005b). Briefly, single larvae were placed in the centre of a Petri dish (35 mm diameter) filled with 3% pre-warmed (28°C)methylcellulose facing with their right eye a screen at an apparent distance of 4.65 cm within their visual field. Moving sine-wave gratings were projected by a HP vp6111 projector onto the screen. The projection size on the screen was 8×6 cm, subtending a visual angle of 65.6° horizontally and 53.1° vertically. Eye movements were recorded by an infrared-sensitive CCD camera triggered by the visual stimulation. A custom developed programme based on the LabView IMAQ software (National Instruments, version 5.1) was used to control the stimulation and the camera as well as to analyse the resulting images. Contrast sensitivity functions for mutant and wild-type larvae were measured by the eye velocity as a function of the contrast of the moving grating. Statistics were derived as described previously(Rinner et al., 2005a) by bootstrap sampling of differences between randomly drawn samples(n=10,000) from a pooled data set of wild-type and mutant larvae. The significance level chosen was at 5% for all experiments.

For recordings of electroretinograms (ERGs)(Makhankov et al., 2004), 5 dpf larvae were dark-adapted for at least 30 minutes prior to positioning them in the recording chamber. The recording chamber was shielded from interference from external sources of electromagnetic radiation by placing it inside a tight Faraday cage. Each larva was placed on its side on the surface of a moist sponge with E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂and 0.33 mM MgSO₄) and paralyzed by directly adding a drop of the muscle relaxant Esmeron (0.8 mg/ml; Organon Teknika, Eppelheim, Germany) onto the larva. An Ag/AgCl electrode system was used to record the ERG responses. The recording electrode was positioned on the centre of the cornea. The reference Ag/AgCl pellet was placed under the body of the larva. All pre-recording steps were carried out under dim red light illumination. An additional 5-minute period in complete darkness was chosen to adapt the larva to dark prior to measurements. White light stimuli (100 mseconds) were used to elicit ERG responses with interstimulus intervals of 5 seconds. Light stimuli were fixed at five relatively different light intensities ranging from 2 lux(OD 5) to 20,000 lux (OD 0). Unattenuated light stimulus intensities were measured over the head of the larvae using a light meter (Tektronix J17, Texas Instruments, USA) and found to account for 3100 lux (optical density (OD)equal to 0 log units). Different light intensities were adjusted using neutral density filter wheels. A virtual instrument (VI) under NI LabVIEW 5.1 was developed for use in all experiments. Sampling was carried out in buffered acquisition mode with a sampling rate of 1000 Hz. The resulting ERGs, as the corneal sum field potentials of the retina in response to light, are shaped like in other vertebrates: a small negative deflection, termed the a-wave,which reflects photoreceptor activation, is followed by a stronger positive deflection, the b-wave, which reflects second order neuron activation. The small a-wave is often masked by the much larger b-wave because of interference. Therefore, in contrast to the b-wave, the a-wave is not a robust measure in larval zebrafish. As a consequence, the b-wave was taken as an indirect measure of outer retina activation. For rescue experiments, a light intensity of 6000 lux was chosen to elicit ERG responses in sibling and lbk^-/- larvae. ERG responses obtained from one larva were averaged three to seven times depending on signal-to-noise ratio. Statistical analysis was performed using GraphPad Prism4 (GraphPad Software, San Diego,CA) software and graphs were generated using Origin v.7 (OriginLab,Northampton, MA).

Salmonella enterica serovar Typhimurium expressing DsRed(van der Sar et al., 2003)were injected with a FemtoJet injection apparatus (Eppendorf) at a constant pressure of 80 hPa for 0.5 seconds into the common cardinal vein of 48 hpf larvae embedded in low-melting point agarose as previously described(van der Sar et al., 2003). To ensure delivery of equal amounts of bacteria, mutant and sibling larvae were injected with the same injection capillary and the injection of the bacteria was monitored under a fluorescence stereo-microscope (Zeiss). For bacterial counts, larvae were macerated in PBS and plated onto ampicillin-containing bacterial plates as described (van der Sar et al., 2003; van der Sar et al., 2006). Bacterial growth was assessed after 24 hours of incubation at 37°C with a fluorescence stereo-microscope.

The rescue was performed by injecting a pSGH2 vector containing the vam6 cds under the control of a heat-shock promoter(Bajoghli et al., 2004) into one-cell stage embryos derived from a cross of two lbk^+/-individuals. At 3 dpf, injected larvae were heat shocked at 38°C for 2 hours and subsequently transferred back to 28°C. The extent of the rescue was determined at 5 dpf by assessing the level of pigmentation in skin melanocytes and RPE and in ERG measurements.

A 25 nucleotide morpholino antisense oligonucleotide was designed against the ATG region of the vam6 mRNA: GAACTGGTTCGTATGCGTCGTGCAT. The morpholino was injected using a pressure of 30 hPa for 4 mseconds into one-cell stage wild-type embryos at different oligonucleotide concentrations:100 μM, 500 μM and 700 μM, corresponding to approximately 1.5 ng, 7.5 ng and 10.5 ng of morpholino. As a control, the generic control morpholino provided by GeneTools (Philomath, OR) was used.

For PCR analysis, total RNA was isolated from 100 wild-type zygotes and 24 hpf embryos, respectively, using the RNAeasy kit and cDNA synthesised using the SuperScriptIII Reverse Transcriptase kit. PCR was performed at an annealing temperature of 60°C for 35 cycles using primers listed in Table 2.

Sense and antisense probes for in situ hybridisation were in vitro transcribed from a linearized pCRII-Vector containing the full-length vam6 cds using SP6 and T7 polymerases, respectively, in the presence of digoxigenin-coupled nucleotides (DigRNA labelling kit; Roche Molecular Biochemicals, Switzerland). Transcripts were hydrolysed to obtain fragments of∼300-500 nucleotides in length. To prevent melanisation of skin melanocytes and the RPE, embryos used for in situ hybridisation were treated with 0.2 mM PTU (1-phenyl-2-thiourea; Sigma). PTU-treated embryos were staged,fixed in 4% PFA in PBS (pH 7.25) at 4°C overnight, dehydrated in a standard methanol series and stored in 100% methanol at -20°C. In situ hybridisation was performed using an automated in situ hybridisation apparatus(Hölle&Hüttner, Tübingen, Germany). Hybridisation was carried out overnight at 58°C. Probe detection was performed with an alkaline phosphatase-coupled anti-Dig Fab-fragment (Roche) and NBT/BCIP staining buffer. Stained embryos were cleared with 50% methanol/50% glycerol and transferred into 100% glycerol for imaging.

The zebrafish mutant leberknödel(lbk^t20290) was identified in a large-scale ENU mutagenesis screen (Glass and Dahm,2004). External analyses revealed that, starting at 3 dpf, lbk^-/- larvae show a pronounced hypopigmentation of skin melanocytes and the RPE. Instead of the evenly black melanocytes characteristic of wild-type larvae, lbk melanocytes show only scarce and patchy pigmentation (Fig. 1, see Fig. S1 in the supplementary material). It is interesting that in addition to melanosome-containing cells the iridophores are affected in lbk. Iridophores derive their reflective properties from vesicular organelles containing guanine crystals(Morrison and Frost-Mason,1991). These light reflections are absent in lbk, both in the eye and body, at all stages of development(Fig. 1E-H), indicating that the biogenesis of these organelles is disrupted. We cannot, however, exclude the possibility that iridophores are missing altogether, although, in view of the overall lbk phenotype, this possibility appears unlikely. Further to the pigment cell phenotypes, starting at 4 dpf, the intestinal tract of lbk develops a brownish discolouration(Fig. 1A-F,I,J) and the liver becomes enlarged (Fig. 1I,J). lbk^-/- larvae fail to develop swim bladders and die between 7 and 8 dpf.

To identify the mutation underlying the lbk phenotype, we mapped the mutation on the zebrafish genome. Pooled DNA of 48 lbk^-/- and 48 sibling 5 dpf larvae, respectively, was tested with 192 SSLP markers (Knapik et al., 1998) resulting in a linkage of the mutation to chromosome 17 between markers z22083 and z4053 (Fig. 1K). To narrow down the critical interval of the mutated locus, we tested additional SSLP markers located in this genomic region. Analysis of 282 meioses identified z9692 and z4053 as the closest SSLP markers(Fig. 1K).

As the phenotype observed in lbk suggests a defect in endosomal vesicle trafficking (see below), we performed RH mapping of several genes involved in this process. This approach resulted in a linkage of the zebrafish vam6 gene to the EST fc27c07.x1 on chromosome 17. This EST is located in close proximity to z9692 on the T51 RH map(Fig. 1K)(Geisler et al., 1999). Given this close linkage, we cloned and sequenced the zebrafish vam6-coding sequence (cds), which has a length of 2628 bp corresponding to a protein of 875 amino acids (see Fig. S2 in the supplementary material).

Sequencing of pooled vam6 cDNA from lbk^-/-larvae and phenotypically wild-type siblings, respectively, revealed a nonsense mutation at position 1066 of the vam6 cds(Fig. 1L), which segregated with the mutant phenotype. This point mutation was confirmed by sequencing exon 11 from genomic DNA of the larvae identified as recombinant in the fine mapping approach. All 15 individuals displaying the lbk phenotype were genotypically homozygous for the point mutation, whereas all phenotypically wild-type individuals were either heterozygous or homozygous for the wild-type allele (not shown). The putative truncated protein product lacks the C-terminal 520 amino acids, including the central clathrin homology(CLH) repeat domain, which has been demonstrated to play an essential role in the clustering and fusion of lysosomes(Caplan et al., 2001), and Ypt7p-interacting sequences (Wurmser et al., 2000) (Fig. 1M,N).

To identify additional lbk alleles and thus confirm that a mutation in vam6 underlies the lbk phenotype, we performed complementation analyses with described mutants displaying similar external phenotypes (Glass and Dahm,2004; Kelsh et al.,1996). These analyses identified no additional lbkalleles. We therefore performed an allele screen by crossing heterozygous lbk^+/- individuals with ENU-mutated Tü zebrafish. Over 1000 genomes were screened for external phenotypes equivalent to that observed in lbk, indicating non-complementation of the novel allele with the lbk mutation. This approach identified a compound heterozygous larva (lbk^*; see Fig. S3 in the supplementary material) harbouring in one allele of the vam6 gene the original C→T exchange found in lbk and in the second allele a T→A exchange at position 374 of the vam6 cds(Fig. 1O). The latter mutation results in the exchange of a conserved methionine residue for a lysine at amino acid position 125 in the N-terminal citron homology (CNH) domain of the Vam6p protein (see Figs S2, S3 in the supplementary material). This domain has been shown to be required for lysosome clustering and fusion(Caplan et al., 2001). Importantly, siblings from the same cross carrying the lbk nonsense allele, did not carry the T→A exchange identified in lbk^* compound heterozygotes(Fig. 1O), confirming that this mutation disrupts Vam6p in the lbk^* larva. Sequencing of 142 vam6 alleles confirmed that this T→A exchange does not occur in the genetic pool from which the founder fish for the ENU-mutagenesis were derived (see Fig. S3 in the supplementary material). The compound heterozygous mutant was not viable and the allele was not recovered to generate a permanent line. Therefore, the information that can be derived from the analysis of the lbk^* mutant is limited to the one experiment reported here.

To investigate if the defect caused by the mutation in lbk is cell-autonomous, we performed blastomere transplantations. Wild-type cells from 1000-cell stage embryos were transplanted into age-matched embryos obtained from a cross of two heterozygous lbk carriers. At 5 dpf, the resulting larvae were screened and larvae with a predominantly lbk^-/- mutant skin phenotype (as judged by the presence of strongly hypopigmented melanocytes) were investigated for the presence of melanocytes of wild-type appearance. As the lbk^-/-phenotype is fully penetrant such that mutants never display melanocytes with normal levels of pigmentation, melanocytes of wild-type appearance had to be derived from transplanted cells. In four independent experiments, a total of 26 lbk^-/- background larvae displaying melanocytes with wild-type pigmentation in their skin were identified (see Fig. S4 in the supplementary material). The development of transplanted wild-type cells into normally pigmented melanocytes in an lbk^-/- background indicates that the lbk mutation acts cell autonomously.

Hypopigmented retinas often have additional defects and affected zebrafish were shown to display decreased visual ability(Schonthaler et al., 2005). Similarly, in humans several pathological conditions with RPE hypopigmentation are associated with more generalised ocular symptoms. This prompted us to examine the eye phenotype in lbk more closely. Histological analyses revealed that the overall patterning of the neural retina is normal(Fig. 2A-F). However, sections taken at different developmental stages revealed not only a severe hypopigmentation of the RPE, but also a progressive thickening(Fig. 2A-H) up to three times the thickness of the wild-type RPE (Fig. 2R). This thickening results from an increase of RPE cell size,rather than increased numbers of RPE cells(Fig. 2S). TEM analyses of lbk revealed substantially fewer and mostly immature and aberrantly shaped melanosomes (Fig. 2I-N,Q). As melanin is, however, present in the lbk^-/- RPE and skin melanocytes (see Fig. S5 in the supplementary material), the mutation affects the biogenesis of melanosomes rather than melanin biosynthesis. This is supported by the observation that the number of mature melanosomes is substantially reduced in lbk(Fig. 2V).

In addition to optically shielding photoreceptor cells (PRCs), the RPE is essential for PRC function by phagocytosing shed PRC outer segments. Outer segments harbour the photopigments that allow the detection of light and the shedding of parts of these segments is an essential process in the maintenance of normal PRC function. The RPE cytoplasm in lbk is filled with numerous vesicles whose number increases between 5 and 7 dpf(Fig. 2I-L). Some of these vesicles contain thick membrane stacks reminiscent of the membrane discs in PRC outer segments (Fig. 2M,N),suggesting a defect in the fusion of endocytic vesicles with lysosomes.

To test this hypothesis, we used rhodopsin as a molecular marker for rod outer segments (Fig. 2O,P). lbk larvae (5 dpf) show an increase in rhodopsin-positive vesicles inside the RPE, while in the wild-type, rhodopsin antibodies only label continuous rod outer segments. Quantification confirmed that the lbkRPE contains increased numbers of outer segment-containing vesicles(Fig. 2T). Staining of eye cryosections with Oil Red O confirmed the presence of numerous lipid-containing vesicles in the lbk RPE(Fig. 4J). These data further suggest a failure in the mutant RPE to metabolise phagocytosed PRC membrane stacks.

At 5 dpf, the PRC outer segments are fully developed in the zebrafish. In lbk mutants, however, they are reduced in length and their regular palisade arrangement is disrupted (Fig. 2I-L,U). Moreover, the microvilli from the RPE that normally interdigitate with the outer segments, cannot be detected. As development proceeds, these phenotypes get stronger resulting in PRCs with very short or even no outer segments (Fig. 2L).

Although we found the outer retina to be severely affected, the inner retina showed no evidence of morphological alterations in lbk(Fig. 2A-F). Similarly,staining of retinal Müller glia cells, which span the entire inner retina and are thus a good marker for inner retinal architecture, revealed no differences in their number, arrangement and morphology between wild-type and lbk^-/- larvae (see Fig. S6 in the supplementary material). Starting at 7dpf, however, apoptotic cells can be detected in the neural retina (Fig. 2W).

To analyse the physiological effect of the lbk mutation on visual performance, we tested larvae by optokinetic response (OKR) measurements. In this test, eye tracking movements of immobilised larvae exposed to moving visual stimuli are used to assess visual ability. In contrast to wild-type larvae, moving gratings consistently failed to evoke eye movements in all lbk^-/- larvae tested, indicating that mutants are behaviourally blind (Fig. 3A). However, this effect is not due to the inability of the mutants to move their eyes, as spontaneous eye movements were regularly observed. We further performed electroretinogram (ERG) analyses of 5 dpf wild-type and lbklarvae (Fig. 3B). In wild-type larvae the b-wave amplitude increased with increasing light intensities. In lbk, however, it remained small even at high light intensities. At 7dpf, the ERG of lbk^-/- is flat, indicating that at this stage the mutant retina no longer responds to light stimuli (not shown).

Subsequent analyses revealed additional phenotypes in internal organs,including the liver and intestine. Starting at 6dpf, the liver of lbklarvae becomes progressively enlarged and darkly discoloured(Fig. 1I,J). Histological analyses confirmed this enlargement and showed a significant swelling of liver cells (Fig. 4A,B). This phenotype was confirmed when lbk^-/- mutants were crossed to a transgenic line, Tg (ef1:GFP), that expresses GFP under the control of the intestinal promoter ef1 (Field et al.,2003) and sections imaged by confocal microscopy(Fig. 4C). To quantify the increase in liver size observed in lbk, we related the total liver area to the number of hepatocyte nuclei in a given section. By this measure, 5 dpf lbk larvae show a 30% increase of liver cell area compared to their siblings (Fig. 4D).

TEM analyses revealed that the cytoplasm of hepatocytes in lbk is filled with numerous, sometimes very large vesicles (up to 30 μm in diameter; Fig. 4E,F),indicative of a defect in vesicle trafficking in this cell type. At 7 dpf, the hepatocytes of lbk larvae show substantial necrotic changes(Fig. 4G,H). These changes might also explain the discoloration of the liver observable in external views(Fig. 1J). Oil Red O staining further showed an accumulation of vesicles containing large amounts of lipids in the lbk liver (Fig. 4I).

We further examined the intestines of lbk and age-matched sibling larvae. TEM analyses of 7 dpf larvae revealed a vesicle phenotype also in the cells of the mutant intestinal bulb. Although wild-type cells contain few large vesicles, the number of vesicles is significantly increased in lbk with a concomitant reduction in vesicle size(Fig. 4K,L; see Fig. S7 in the supplementary material). This further indicates a possible defect in the fusion of intracellular vesicles in lbk.

Similar to RPE cells, macrophages are highly phagocytotically active and contain numerous lysosomes and lysosome-related organelles, such as major histocompatibility complex class II compartments(Raposo and Marks, 2002). This prompted us to examine the morphology of vesicles in this cell type by incubating 7 dpf lbk and sibling larvae in Neutral Red, which is selectively retained in macrophage lysosomes(Herbomel et al., 2001). This staining suggests that intracellular vesicles in lbk macrophages may be enlarged, seem to display an amorphous shape and are more heterogeneous in size than in wild-type larvae (Fig. 5A,B; inserts).

To obtain functional information on the innate immune system in lbk, we injected DsRed-expressing bacteria into mutant and sibling larvae at 48 hpf (Fig. 5C). Phagocytosis of the injected bacteria was complete within 10 minutes of the injection. Clearance of the fluorescent bacteria was monitored by in vivo fluorescence microscopy as well as by plating dissociated larvae onto selective media plates and counting of the resulting bacterial colonies. We found that while siblings efficiently cleared injected bacteria within 32 hours, mutants still contained numerous bacteria(Fig. 5D-M). Importantly,although phagocytosis appeared unaffected in lbk, the subsequent failure of the mutant phagocytes to degrade the bacteria indicates a defect in the fusion of phagosomes with lysosomes.

To confirm that a mutation in vam6 underlies the lbkphenotype, we performed a rescue experiment. The zebrafish vam6 cds was cloned into an expression vector under the control of a heatshock-inducible promoter (Bajoghli et al., 2004). This vector was injected at the one-cell stage and expression induced by heat shock at 3 dpf. In total, three independent experiments were performed and evaluated. To assess the rescue of the mutant phenotype, the larvae were screened at 5 dpf for their external morphology. vam6-injected lbk larvae developed darker skin melanocytes and RPE (Fig. 6A-C; insets)than uninjected and control vector-injected (no vam6 cds) lbk larvae. The larvae were further tested by ERG analysis to assess the rescue of the visual ability. vam6-injected lbk larvae showed an increase in the b-wave amplitude to 20-60% of wild-type levels(Fig. 6A-D). By contrast,uninjected control lbk larvae did not show a b-wave(Fig. 3B).

To provide further evidence that a mutation in vam6 underlies the lbk phenotype, we designed an antisense morpholino against a region containing the start codon (ATG) of the vam6 mRNA. Injection of the ATG-morpholino into wild-type embryos at the one-cell stage resulted in a similar, but stronger phenotype than that observed in lbk(Fig. 6E). Injection of a control morpholino did not result in larvae displaying any phenotypes(Fig. 6E). The weaker phenotype of lbk larvae compared with the vam6 knock-down could be explained by the presence of maternally supplied wild-type vam6 mRNA,which could serve as a template for the synthesis of functional Vam6p during early embryonic development. To test this hypothesis, we isolated mRNA from wild-type zygotes, reverse transcribed it into cDNA and performed a PCR analysis. This showed that vam6 mRNA is maternally supplied in zebrafish (Fig. 6F) and its expression persists through later stages of development(Fig. 6F,G; see Fig. S8 in the supplementary material). The Vam6p protein is part of the HOPS complex, a large multi-subunit protein complex. In this context it is interesting that the mRNA for Vps18, another member of the HOPS complex, has been found to be maternally supplied in zebrafish zygotes in a microarray experiment(http://zf-espresso.tuebingen.mpg.de/).

In this study, we provide the first phenotypic description of a lack of vam6 gene function in a multicellular organism. Zebrafish lbk mutants display defects in several tissues and cell types (liver,intestinal tract, macrophages, PRCs, RPE and skin melanocytes) as well as compromised vision and innate immunity. The range of phenotypes resembles pathological symptoms observed in individuals suffering from syndromes caused by defects in lysosomes and lysosome-related organelles(Stinchcombe et al., 2004),including the ARC syndrome, CHS, GS and HPS. For example, CHS, GS and HPS are characterised by a hypopigmentation of the skin and RPE, as well as a significant visual impairment (Barak and Nir, 1987; Clark and Griffiths,2003; Griscelli et al.,1978; Hermansky et al.,1975). These symptoms are also observed in lbkmutants.

Melanosomes are part of the large group of lysosome-related organelles and serve to synthesize and store melanin. The melanocytes in lbk mutants display significantly fewer and lighter melanosomes than are observed in wild-type animals. The presence of melanosomes with wild-type levels of melanin pigmentation in the lbk skin and RPE shows that the mutation does not cause a block in melanin synthesis or melanosome biogenesis per se. The increase in immature melanosomes and the substantially decreased number of mature (stage IV) melanosomes in lbk rather suggests a significantly reduced efficiency of melanosome maturation, a process critically dependent on the fusion of endosomal vesicles with premelanosomes(Dell'Angelica, 2003). Alternatively, the stop codon mutation we identified in lbk, UAG, has been suggested to have intermediate read-through fidelity. Thus, the mutant embryos might retain a low level of functional protein, as has been observed in the human syndromes described in this manuscript and other lysosomal storage disorders (Brooks et al.,2006).

The hypopigmentation of the lbk RPE probably contributes to the reduced visual ability. Moreover, the RPE contains numerous vesicles containing undigested PRC outer segments. Although indicating that phagocytosis by the RPE is not compromised, this novel phenotype suggests a defect in lysosome-mediated degradation of phagocytosed outer segments, a central function of the RPE, which may lead to a generalised metabolic dysregulation. As the RPE is essential for maintaining the PRCs, compromising the viability of the RPE will probably also negatively affect PRC function. This notion is supported by our TEM studies, which reveal shortened PRC outer segments. The presence of an a-wave in ERGs, however, suggests that lbk PRCs retain at least some functionality up until 5 dpf. Nonetheless, lbk larvae are severely visually impaired. Visual defects have also been reported for CHS and HPS syndromes (OMIM #214500;#203300). Interestingly, visual problems have not been diagnosed in individuals with GS to date (OMIM #214450; #607624; #609227). This may,however, be due to inconsistent testing of affected individuals or subsymptomatic reductions in visual acuity.

In addition to the RPE, there is also an accumulation of vesicles in hepatocytes and the cells lining the intestinal tract. These cells typically show high endo- and exocytic activities and large numbers of intracellular lysosomes and lysosome-related vesicles. The larger number of intracellular vesicles in the lbk liver and RPE is accompanied by an increase in overall liver and RPE size. As there is no apparent defect in the formation of vesicles but rather in their subsequent fusion (and hence recycling), it could be envisaged that the observed increase is due to an accumulation of vesicles,which in turn leads to an overall increase in cytoplasmic volume. An increase in liver volume (hepatomegaly) has also been described for individuals affected by CHS (OMIM #214500) and ARC syndrome(Gissen et al., 2004). Similarly, the intestinal phenotype observed in lbk is reminiscent of the granulomatous colitis described for individuals with HPS (OMIM#203300).

Defects in lysosome-related organelles might also be responsible for the immunological deficiencies observed in individuals with CHS and GS(Faigle et al., 1998; Griscelli et al., 1978). Macrophages, for instance, are crucial for the body's innate immune response. One of their key functions is the phagocytosis of pathogens. Pathogen-containing endosomes subsequently fuse with lysosomes and major histocompatibility complex class II (MHCII) compartments, a lysosome-related organelle, in which the pathogens are degraded and foreign peptides loaded onto MHCII complexes. Loaded MHCII complexes are subsequently delivered to the surface of macrophages for presentation to cells of the adaptive immune response. We found that lbk macrophages display an increase in intracellular vesicles, which might be indicative of a defect in the endosomal pathway and/or a defect in MHCII vesicle delivery to the plasma membrane. Importantly, our analyses show that the innate immune response to bacterial infection is significantly compromised in lbk.

ARC syndrome, CHS, GS and HPS are caused by defects in vesicle trafficking,particularly of lysosomes, lysosome-related organelles and late endosomes, and mutations in genes involved in vesicle transport, sorting, docking and fusion were shown to underlie these diseases. Moreover, ARC syndrome was demonstrated to be caused by mutations in the vps33b gene, a member of the HOPS complex, in humans (Gissen et al.,2004). Similarly, morpholino knock-down experiments targeting the zebrafish vps33b gene result in a phenotype resembling ARC syndrome(Matthews et al., 2005). Interestingly, the buff (bf) mouse, which harbours a mutation in the Vps33a gene, has been proposed as a model for HPS(Suzuki et al., 2003),suggesting a link between ARC syndrome and the CHS, GS and HPS group of diseases. The phenotypes observed in lbk encompass both symptoms observed in individuals with CHS, GS and HPS (liver, macrophage,hypopigmentation and visual phenotypes) and ARC syndrome (liver and intestinal phenotypes), providing further evidence for a link between these disorders.

Moreover, we have identified the gene mutated in lbk as vam6. Like Vps33p, the Vam6p protein is a component of the HOPS complex, which is required for SNARE complex assembly during vesicle docking and fusion. Mutations in the two alleles identified in this study lead to a premature STOP upstream of the CLH repeat domain and Ypt7p-interaction sequences, and the exchange of a highly conserved amino acid in the N-terminal CNH domain, respectively. The latter has been shown to be essential for lysosome clustering and fusion (Caplan et al., 2001). The Ypt7p-interaction domain mediates protein-protein interactions between Vam6p/Vps39p and Ypt7p/Rab7, a Rab-GTPase with essential functions in early to late endosome transport and late endosome-lysosome fusion, as well as in axonal retrograde transport(Deinhardt et al., 2006; Stein et al., 2003). It has been shown in yeast that Vam6p acts as a guanine nucleotide exchange factor for Ypt7p (Wurmser et al.,2000). When this exchange fails, Ypt7p is locked in its GDP-bound state and vesicle docking is blocked. This probably contributes to the vesicle phenotypes observed in lbk. In mammalian cells, both Rab7 and the HOPS complex are essential for the conversion of early into late endosomes(Rink et al., 2005). Interestingly, RNAi-mediated knock-down of Rab7 in HeLa cells leads to a cellular vesicle phenotype reminiscent of that observed in fibroblasts from individuals with CHS (Davies et al.,1997). Similarly, knock-down of RAB-7 in C. elegans leads to enlarged early and late endosomes and knock-down of HOPS complex members yield defects in gut lysosome formation(Poteryaev et al., 2007).

The next step in vesicle fusion, following the initial docking steps, is assembly of SNARE complexes. This process requires physical interactions between the HOPS complex and SNAREs(Collins et al., 2005; Price et al., 2000a; Price et al., 2000b; Sato et al., 2000). Mutations in Vam6p that abolish the functions of the CLH or CNH domains, respectively,might hamper these interactions and additionally contribute to the accumulation of vesicles in lbk.

The mutation in the vam6 gene in lbk acts in a cell-autonomous fashion and affects cells derived from all three germ layers(skin melanocytes, RPE, PRCs in ectoderm; liver and intestine in endoderm; and macrophages in mesoderm). The most severe phenotypes are detected in cells that display high levels of phagocytic (RPE cells, macrophages) and secretory activity (hepatocytes, intestinal cells). Owing to the importance of vesicle trafficking in these cells, they are probably the first and most severely affected. The occurrence of only partially overlapping symptoms in ARC syndrome, CHS, GS and HPS, as well as the zebrafish lbk mutant could be due to the fact that lysosome-related organelles can be cell type-specific(Raposo and Marks, 2002) and thus defects in proteins required for these organelles might only affect certain cell types. The comparatively widespread organ dysfunction observed in lbk suggests that Vam6p is a core component with non-redundant functions in vesicle trafficking in a range of cell types.

It has recently been shown that a viral insertion disrupting the vps18p gene, another member of the HOPS complex, in zebrafish results in similar phenotypes as those observed in lbk, including a hypopigmentation of skin and RPE melanocytes, a lack of iridophore reflections, visual defects, hepatomegaly and an accumulation of vesicles in hepatocytes (Maldonado et al.,2006; Sadler et al.,2005). This provides further evidence for the importance of the HOPS complex in various cell types and organs.

In conclusion, the overlap in symptoms in human syndromes caused by defective lysosome trafficking and phenotypes observed in lbk suggest this mutant as a model for these inherited human diseases and suggests a novel disease gene. Importantly, there is currently no mouse model for loss of function of Vam6p. Studies on lbk may offer important insights into the mechanisms underlying the observed symptoms and make these pathologies amenable to experimental manipulation.

The authors thank Matthew Harris for providing ENU-mutated zebrafish for the allele screen and for critically reading the manuscript; Annemarie Meijer for providing DsRed-expressing Salmonella and advice; Philippe Herbomel, Matthias Gesemann, Sarah Magee and Anne Spang for helpful discussions; Pawel Pasierbek for help with confocal LSM; Ines Gehring,Marianne Leißer and Brigitte Sailer for technical assistance; and John Vessey for critically reading the manuscript. This work was supported by an EMBO Long-term Fellowship (R.D.) and by the European Commission as part of the ZF-MODELS Integrated Project in the 6th Framework Programme, Contract No. LSHG-CT-2003-503496 (S.C.F.N., R.D. and R.G.).