Comparative evolutionary analysis of VPS33 homologues: genetic and functional insights Paul Gissen 1,2 , Colin A. Johnson 1 , Dean Gentle 1 , Laurence D. Hurst 3 , Aidan J. Doherty 4 , Cahir J. O’Kane 5 , Deirdre A. Kelly 2 and Eamonn R. Maher 1, * 1 Section of Medical and Molecular Genetics, University of Birmingham, Birmingham B15 2TG, 2 The Liver Unit, Birmingham Children’s Hospital, Birmingham B4 6NH, UK, 3 Department of Biology and Biochemistry, University of Bath, Bath BA2 7AY, UK, 4 UK Genome Damage and Stability Centre, University of Sussex, East Sussex BN1 9RQ, UK and 5 Department of Genetics, University of Cambridge, Downing Street CB2 3EH, Cambridge, UK Received December 21, 2004; Revised and Accepted March 21, 2005 VPS33B protein is a homologue of the yeast class C vacuolar protein sorting protein Vps33p that is involved in the biogenesis and function of vacuoles. Vps33p homologues contain a Sec1 domain and belong to the family of Sec1/Munc18 (SM) proteins that regulate fusion of membrane-bound organelles and interact with other vps proteins and also SNARE proteins that execute membrane fusion in all cells. We demonstrated recently that mutations in VPS33B cause ARC syndrome (MIM 208085), a lethal multisystem disease. In con- trast, mutations in other Vps33p homologues result in different phenotypes, e.g. a mutation in Drosophila melanogaster car gene causes the carnation eye colour mutant and inactivation of mouse Vps33a causes buff hypopigmentation phenotype. In mammals two Vps33p homologues (e.g. VPS33A and VPS33B in humans) have been identified. As comparative genome analysis can provide novel insights into gene evol- ution and function, we performed nucleotide and protein sequence comparisons of Vps33 homologues in different species to define their inter-relationships and evolution. In silico analysis (a) identified two homol- ogues of yeast Vps33p in the worm, fly, zebrafish, rodent and human genomes, (b) suggested that Carnation is an orthologue of VPS33A rather than VPS33B and (c) identified conserved candidate functional domains within VPS33B. We have shown previously that wild-type VPS33B induced perinuclear clustering of late endosomes and lysosomes in human renal cells. Consistent with the predictions of comparative analysis: (a) VPS33B induced significantly more clustering than VPS33A in a renal cell line, (b) a putative fly VPS33B homologue but not Carnation protein also induced clustering and (c) the ability to induce clustering in renal cells was linked to two evolutionary conserved domains within VPS33B. One domain was present in VPS33B but not VPS33A homologues and the other was one of three regions predicted to form a t-SNARE binding site in VPS33B. In contrast, VPS33A induced significantly more clustering of melanosomes in melanoma cells than VPS33B. These investigations are consistent with the hypothesis that there are two functional classes of Vps33p homologues in all multicellular organisms and that the two classes reflect the evolution of organelle/tissue-specific functions. INTRODUCTION The functional analysis of novel genes represents a major challenge to human genetics research. Recently, we demon- strated that a rare autosomal recessive disorder, ARC syn- drome, characterized by neurogenic arthrogryposis multiplex congenita, renal tubular dysfunction and neonatal cholestasis with bile duct hypoplasia and low gamma glutamyl trans- peptidase (gGT) activity was caused by mutations in the human VPS33B gene (1). Although the function of the VPS33B protein has not been studied previously in humans, the yeast homologue, Vps33p, a class C vacuolar protein # The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]*To whom correspondence should be addressed. Tel: þ44 1216274434; Fax: þ44 1214141695; Email: [email protected]Human Molecular Genetics, 2005, Vol. 14, No. 10 1261–1270 doi:10.1093/hmg/ddi137 Advance Access published on March 24, 2005 by guest on June 3, 2013 http://hmg.oxfordjournals.org/ Downloaded from
Transcript
Comparative evolutionary analysis of VPS33homologues: genetic and functional insights
Paul Gissen1,2, Colin A. Johnson1, Dean Gentle1, Laurence D. Hurst3, Aidan J. Doherty4,
Cahir J. O’Kane5, Deirdre A. Kelly2 and Eamonn R. Maher1,*
1Section of Medical and Molecular Genetics, University of Birmingham, Birmingham B15 2TG, 2The Liver Unit,
Birmingham Children’s Hospital, Birmingham B4 6NH, UK, 3Department of Biology and Biochemistry,
University of Bath, Bath BA2 7AY, UK, 4UK Genome Damage and Stability Centre, University of Sussex,
East Sussex BN1 9RQ, UK and 5Department of Genetics, University of Cambridge, Downing Street CB2 3EH,
Cambridge, UK
Received December 21, 2004; Revised and Accepted March 21, 2005
VPS33B protein is a homologue of the yeast class C vacuolar protein sorting protein Vps33p that is involvedin the biogenesis and function of vacuoles. Vps33p homologues contain a Sec1 domain and belong to thefamily of Sec1/Munc18 (SM) proteins that regulate fusion of membrane-bound organelles and interact withother vps proteins and also SNARE proteins that execute membrane fusion in all cells. We demonstratedrecently that mutations in VPS33B cause ARC syndrome (MIM 208085), a lethal multisystem disease. In con-trast, mutations in other Vps33p homologues result in different phenotypes, e.g. a mutation in Drosophilamelanogaster car gene causes the carnation eye colour mutant and inactivation of mouse Vps33a causesbuff hypopigmentation phenotype. In mammals two Vps33p homologues (e.g. VPS33A and VPS33B inhumans) have been identified. As comparative genome analysis can provide novel insights into gene evol-ution and function, we performed nucleotide and protein sequence comparisons of Vps33 homologues indifferent species to define their inter-relationships and evolution. In silico analysis (a) identified two homol-ogues of yeast Vps33p in the worm, fly, zebrafish, rodent and human genomes, (b) suggested that Carnationis an orthologue of VPS33A rather than VPS33B and (c) identified conserved candidate functional domainswithin VPS33B. We have shown previously that wild-type VPS33B induced perinuclear clustering of lateendosomes and lysosomes in human renal cells. Consistent with the predictions of comparative analysis:(a) VPS33B induced significantly more clustering than VPS33A in a renal cell line, (b) a putative flyVPS33B homologue but not Carnation protein also induced clustering and (c) the ability to induce clusteringin renal cells was linked to two evolutionary conserved domains within VPS33B. One domain was present inVPS33B but not VPS33A homologues and the other was one of three regions predicted to form a t-SNAREbinding site in VPS33B. In contrast, VPS33A induced significantly more clustering of melanosomes inmelanoma cells than VPS33B. These investigations are consistent with the hypothesis that there are twofunctional classes of Vps33p homologues in all multicellular organisms and that the two classes reflectthe evolution of organelle/tissue-specific functions.
INTRODUCTION
The functional analysis of novel genes represents a majorchallenge to human genetics research. Recently, we demon-strated that a rare autosomal recessive disorder, ARC syn-drome, characterized by neurogenic arthrogryposis multiplex
congenita, renal tubular dysfunction and neonatal cholestasiswith bile duct hypoplasia and low gamma glutamyl trans-peptidase (gGT) activity was caused by mutations in thehuman VPS33B gene (1). Although the function of theVPS33B protein has not been studied previously in humans,the yeast homologue, Vps33p, a class C vacuolar protein
# The Author 2005. Published by Oxford University Press. All rights reserved.For Permissions, please email: [email protected]
*To whom correspondence should be addressed. Tel: þ44 1216274434; Fax: þ44 1214141695; Email: [email protected]
Human Molecular Genetics, 2005, Vol. 14, No. 10 1261–1270doi:10.1093/hmg/ddi137Advance Access published on March 24, 2005
sorting protein, is required for vacuolar biogenesis and has akey role in protein trafficking from Golgi to vacuole (2). Fur-thermore, yeast mutants of the class C vps proteins (consistingof Vps11p, Vps16p, Vps18p and Vps33p) display severe intra-cellular acid–base imbalance, amino acid pool deficiency andtemperature-sensitive growth failure (3). These findingssuggested that the ARC phenotype might be explained byabnormal organelle biogenesis. Consistent with this hypoth-esis, we demonstrated that (a) ectopically expressed VPS33Blocalized to LAMP1-positive late endosomes and lysosomesand (b) overexpression of VPS33B in human renal cellscaused perinuclear clustering of late endosomes and lyso-somes (1). Late endosomes or multivesicular bodies are mem-brane bound organelles which share their biogenesis pathwaywith lysosomes and like melanosomes belong to the class oflysosome-related organelles (LRO) (4,5). Late endosomesconstitute the penultimate stage in the endocytic transport ofproteins to lysosomes and are also involved in recyclingof membrane proteins. Perinuclear clustering of late endosomeshas been shown to occur in the presence of over-expressed classC vps protein homologues and their interacting partners (6).
Inactivation of yeast Vps33 homologues in other species isassociated with phenotypes significantly different from ARC.Thus in Drosophila, a hypomorphic allele of the Vps33 homo-logue carnation (car ) causes the carnation eye colour pheno-type. The car gene product (Carnation) localizes toendosomal compartments and is a homolog of SM regulatorsof membrane fusion (7). In mice, a mutation in themVps33a gene caused the buff (bf ) mouse phenotype that ischaracterized by hypopigmentation and a mild plateletstorage pool deficiency and has been proposed as an animalmodel for Hermansky–Pudlak syndrome (8). Apart from theplatelet dysfunction, the bf mouse does not have any otherphenotypic features in common with ARC syndrome patients.VPS33A and VPS33B share 31% identity and 51% similarity,and are much more closely related to each other than to theyeast Vps33p (24% identity and 41% similarity for VPS33B,and 27% identity and 46% similarity for VPS33A).However, as both genes are ubiquitously expressed, it wouldappear that VPS33A and VPS33B have evolved differentfunctions.
In yeast, the class C vps proteins, together with Vps39p andVps41p form a homotypic vacuolar protein sorting (HOPS)complex which has a role in the recruitment of the mem-brane-bound organelles for the fusion events required forendo/exocytosis and secretion (9,10). Recent findingssuggest that the HOPS complex and the SM-like family ofproteins are essential for vesicular trafficking and may becrucial for ensuring the specificity of SNARE-mediated mem-brane fusion (11–15). SNAREs (soluble N-ethylmaleimide-sensitive factor attachment protein receptor or SNAP receptorproteins) are membrane bound proteins present on bothvesicular (v-SNARE) and target (t-SNARE) membranes. Thev- and t-SNAREs form a complex that pulls opposing mem-branes together (16,17). SNARE function is mediated byupstream regulators and tethering factors such as the HOPScomplex, which also interacts with members of the family ofRab GTPases and SM proteins. These tethering factors bringthe two vesicles into close proximity thus determining thespecificity of the interaction.
To gain insights into the evolution of Vps33p homologuesin multicellular organisms, we undertook nucleotide and pro-tein sequence comparisons in a variety of species andassembled a phylogenetic tree. This analysis suggested thatVPS33A and VPS33B diverged at an early stage of metazoanevolution and that Carnation is more closely related toVPS33A than VPS33B. This finding prompted us to searchfor a Drosophila orthologue of human VPS33B. We investi-gated the function of the VPS33B candidate orthologue bydetermining its ability to induce perinuclear clustering oflate endosomes and lysosomes in human cells. We also inves-tigated the functional importance of potential protein bindingmotifs in VPS33B predicted by comparative sequence analysisand compared the effects of VPS33A and VPS33B over-expression on clustering of melanosomes.
RESULTS
Homology searching and phylogenetic analysis
Basic Local Alignment Search Tool (BLAST) homologysearch using human VPS33B as a query identified the follow-ing known, annotated proteins: Saccharomyces cerevisiaeVps33p (NP-013500), Drosophila melanogaster Carnationprotein (NP-523410), Homo sapiens VPS33A and VPS33B(NP-061138 and NP-075067), Mus musculus mVps33A andmVps33b (NP-084205 and NP-835171) and Rattus norvegicusr-Vps33a and r-Vps33b (NP-075250 and NP-071622) proteins.In addition, BLAST searches identified two unannotatedhomologues from Danio rerio (ENSDARP00000013152 andENSDARP00000006820), a novel fly Vps33p homologueCG5127-PA (NP-651395) and two Caenorhabditis eleganshomologues known as C56C10.1 (NP-495342) and SLP-1(P34260). Protein sequence alignments demonstrated thatamong the mammalian VPS33B and VPS33A homologues,VPS33B orthologues were more conserved than VPS33Aorthologues. Thus human VPS33B has 96% pairwise identityand 97% similarity (and no gaps) with rat r-Vps33b, butVPS33A has 78% identity and 82% similarity (and 4%gaps) to rat r-Vps33a orthologues (Fig. 1A and B). To studythe evolutionary inter-relationships of VPS33A and VPS33Bhomologues further, we performed both a Bayesian(Fig. 1C) and a quarter puzzling maximum likelihood(Fig. 1D) estimation of phylogeny. Thus we created two phylo-genetic trees rooted with a Vps33p from multiple proteinsequence alignment (Fig. 1C and D). In both cases, thetree divided into two branches containing orthologues ofVPS33A and VPS33B in all multicellular organisms. Thissuggests that either (a) a duplication of the gene occurredbecause the divergence of animals from fungi relativelyclose to the base of the animal lineages or (b) the duplicationwas ancestral to both fungi and animals but lost in the lineageleading to yeast. Carnation was predicted under both methodsto be the orthologue of VPS33A, whereas Drosophila CG5127is orthologous to VPS33B.
Evolutionary conservation of VPS33A and VPS33B
We used overlapping sliding window analysis with a block-size of 90 nucleotides, with jumps of 30 nucleotides, and for
1262 Human Molecular Genetics, 2005, Vol. 14, No. 10
each window calculated the non-synonymous/synonymoussubstitution ratio (Ka/Ks) among human, mouse and ratVPS33B orthologues. This analysis detected strong stabilizingselection across the whole of the gene, with particularly stableregions in the N-terminal 60 codons (nucleotides �0–180)and between codons 250–500 (nucleotides �750–1500)(Fig. 2). Similarly high conservation was found betweenhuman and mouse VPS33A orthologues (data not shown). Sig-nificant sequence differences between codons 100 and 250were present between the human and rat VPS33A orthologues,and this region was also the least conserved in VPS33B.Sequence alignment of VPS33B and VPS33A proteinsrevealed an overall pairwise protein similarity and identityof 51 and 31%, respectively (48 and 27%, respectively, inthe N-terminal 15–60 codons and 52 and 28%, respectively,between codons 240–450) and an insertion of 31 aminoacids (450–480) in VPS33B that was absent from VPS33A(Fig. 3A).
VPS33B structure–function predictions
Using the previously reported crystal structure of the inter-action between neuronal-Sec1 and Syntaxin1 proteins in rat(18) as a model, we searched VPS33B for putative bindingsites for a Syntaxin1 homologue. This revealed three putativebinding motifs (A, B and C) predicted to interact with a Syn-taxin1 homologue (Fig. 3B). These predictions were consistentwith the results of the Ka/Ks ratio analysis as each of the pre-dicted binding motifs were contained within regions understrong stabilizing selection (A between codons 35 and 60, Bbetween codons 260 and 275 and C between codons 310and 340). In addition, although most VPS33B mutations inARC syndrome are predicted to be null mutations, we havepreviously identified a missense substitution at L30P in apatient with severe ARC syndrome. Structural modelling ofthis mutation predicted that the substitution would distort theputative N-terminal binding site A, consistent with the evol-utionary conservation analysis (Fig. 3C).
Function of human Vps33p homologues and vesicularclustering
In order to corroborate the results of the evolutionary conser-vation analysis and structure predictions, we investigated theability of the putative Drosophila VPS33B orthologueCG5127, Carnation and human VPS33A, to induce clusteringof vesicles in a renal cell carcinoma (RCC4) cell line. Therewas clear evidence of vesicular clustering (65%, 26 of 40cells) when compared with mock-transfected cells (0%, zeroof 40 cells), 48 h after transfection of RCC4 cells with wild-type VPS33B (WT). Transfection of RCC4 cells with wildtype VPS33A induced clustering in a smaller number of
Figure 1. Alignment of (A) VPS33A and (B) VPS33B orthologues in S. cer-evisiae, C. elegans, D. melanogaster, D. rerio, M. musculus, R. norvegicus andH. sapiens. Phylogenetic trees for the homologues of the yeast Vps33p protein(C) using Bayesian analysis (50 changes represent an absolute number ofchanges) and (D) using quartet puzzling maximum likelihood analysis. Thehorizontal bar represents a 0.1 evolutionary distance (slightly ,10% diver-gence to allow for multihit correction).
Human Molecular Genetics, 2005, Vol. 14, No. 10 1263
cells (12.5%, five of 40 cells) (P , 1026 versus VPS33B). Thenovel putative fly VPS33B orthologue (CG5127) inducedclustering in 25% of the cells (10 of 40 cells) whereasempty vector control and Carnation had no effect (each 0/40cells, P ¼ 0.02 for CG5127 versus carnation, and forCG5127 versus empty vector) (Fig. 4). No clustering wasseen in cells transfected with a truncated VPS33B(1–438),and reduced clustering effect (12.5%, P , 1026 versus WT)was seen in the cells transfected with a VPS33B(D46–52)with a deleted predicted syntaxin-binding region A. We alsoinvestigated the effect of the deletion of 31 amino acidspresent in VPS33B and absent in VPS33A. Transfection ofthe cells with pCMVHAþVPS33B(D450–480) had a signifi-cantly reduced clustering effect (10%, P , 0.001 versus WT).
To investigate whether the clustering effect is organelle/tissue-specific, we overexpressed VPS33B and VPS33A inmouse malignant melanoma cells (Fig. 5). Overexpression ofVPS33A caused clustering of melanosomes in 34% (86 outof 250 cells, P , 10213 versus empty vector where clusteringwas observed in 8%, 20 of 250 transfected cells). Over-expression of VPS33B caused clustering of melanosomes in15% of cells (37 of 250 cells, P , 0.05 versus empty vectorand P , 1027 versus VPS33A).
DISCUSSION
We identified two homologues of the yeast Vps33p protein inall multicellular organisms studied. Each of these homologuescontains a Sec1-like domain and corresponds to one of the twodistinct classes of the SM-like protein family, which is highlyconserved among species. Yeast, worm, zebrafish and mam-malian genomes contain between four and seven SM genes,
whose products display significant homology along thewhole sequence. Although it has been suggested that over-expression of a homologue from the same species cannotsubstitute for a null mutation in another SM protein codinggene, it is clear that an orthologue from another species canpartially restore its function (19–22). For example, mousemunc18-1 rescues the unc-18 null mutant phenotype inC. elegans, but munc18-2 cannot. Most of the loss-of-functionmutations described in SM proteins lead to lethal phenotypes,underlining the importance of these proteins in vesicular traf-ficking (11). The function of the SM proteins is not clear butthey are known to associate with target SNAREs (principally,syntaxin proteins) and also tethering factors such as the HOPScomplex proteins (23–25). Recent findings suggest thatdespite high sequence homology between SM proteins theyform complexes with specific SNAREs and play crucial rolein determining the specificity of the SNARE-complex assem-bly (18,26). In yeast, Vps33p has been shown to interactdirectly with vacuolar SNARE Vam3p and its homologuelate endosomal SNARE Pep12p (2,27). The latter also inter-acts with another SM protein Vps45, which is exclusivelyinvolved in the traffic of biosynthetic vesicles from Golgi.Although this finding is surprising, this dual SM action onPep12p may be explained if the presence of Vps33p is a dis-tinguishing feature of the endocytic vesicles originating inthe cell membrane. This suggestion is confirmed by the factthat the two SM proteins cannot be used interchangeably(27). In parallel, functional and genetic studies in yeast impli-cated HOPS complex in protein trafficking along the recyclingpathway to and from late endosomes (28). Furthermore, Vps33pmutants displayed abnormal recycling of the plasma membraneproteins. These findingsmay explain the nature of themolecular
Figure 2. Comparison of non-synonymous/synonymous substitution ratio(Ka/Ks) of human VPS33B versus R. norvegicus (red line) and M. musculus(blue line) orthologues. An overlapping sliding window of 90 nucleotideswas used with jumps of 30 nucleotides. A, B and C are predicted t-SNAREbinding motifs. D is an insertion of 29 codons not present in VPS33A.Mutations that are predicted to be pathogenic in ARC syndrome are indicatedas follows: missense substitution (M), frameshift (F), splice-junction (S) andnonsense (N). Ka/Ks values are shown on the y-axis, and the x-axis displaysthe window midpoint in nucleotides starting from the first nucleotide of thestarting codon.
Figure 3. The predicted structure of the VPS33B protein complexed with thet-SNARE syntaxin1. VPS33B is shown in blue and syntaxin1 in green. (A)The region deleted in the construct VPS33B1-438 is shown in red and the29 amino acid insertion absent in VPS33A in orange. (B) The t-SNAREbinding motifs are shown in pink and the L30P mutation site in red.(C) Close-up of the L30P mutation which is predicted to disrupt alpha helix1 (arrow) and hence the N-terminal binding site for the t-SNARE.
Human Molecular Genetics, 2005, Vol. 14, No. 10 1265
pathophysiology of patients with ARC syndrome. Thusmutations in human VPS33B gene are associated with abnormallocalization of plasma proteins in polarized cells. This mayoccur due to loss of interaction with a SNARE protein at thelate endosomal stage and may lead to their accumulation inthe cytoplasm or mislocalization on the plasma membrane.This hypothesis requires further investigation.
The presence of two VPS33 homologues suggests that theyhave distinct roles in normal cellular functions. The phenotypeof the bf mouse suggests abnormal biogenesis and transport of
the melanosomes and platelet granules. In contrast, there is noevidence of albinism or abnormal organellar biogenesis inpatients with ARC syndrome. However, ARC syndromepatients develop hepatocyte accumulation of lipofuscin gran-ules and abnormal localization of the membrane proteins inpolarized epithelial cells of the liver and kidneys (1). Lateendosomes, melanosomes, lipofuscin and platelet storagegranules, all belong to a class of the LRO that share theirbiogenesis pathways. Thus it appears that VPS33B andVPS33A orthologues function in the LRO-specific pathways
Figure 4. Clustering of LAMP1-positive organelles. RCC4 cells were transfected with a pCMVHAþ Carnation (A–C) or pCMVHAþ CG5127 (D–F) con-structs showing LAMP1 staining (A and D), HA staining (B and E) and merge (C and F, LAMP1 red, HA green and DAPI blue). Scale bar 5 mm. Thearrow in D indicates a cell with clustered LAMP1-positive organelles. (G) A bar chart, showing the percentage of transfected RCC4 cells displaying clustering.�P ¼ 0.02 for CG5127 versus both empty vector and Carnation. ��P, 1026 versus VPS33B.
1266 Human Molecular Genetics, 2005, Vol. 14, No. 10
determined either by a tissue-specific or other, yet unknownfactors.
We identified a putative novel homologue of yeast Vps33p(CG5127) in Drosophila and predicted that it is functionallycloser to the VPS33B. To confirm this we overexpressedCarnation and CG5127 in the human RCC4 cell line. Wecompared the percentage of cells with the distinct morphologyof clustered LAMP1-positive organelles with that achieved byoverexpression of VPS33B and VPS33A. These experimentsidentified that in RCC4 cells VPS33A has a much lowercapacity to produce clustering than VPS33B (12.5 versus65%, P � 0.05). CG5127 retained the ability to induce clus-tering, unlike Carnation protein (25 versus 0%). Thus thesefindings are entirely consistent with the in silico predictionthat CG5127 is a VPS33B orthologue.
In contrast, analysis of clustering of melanosomes in F1P43mouse melanoma cell line found that VPS33A had signi-ficantly more effect than VPS33B (34 versus 15%,P , 1027). This observation demonstrates that the ability ofa Vps33p homologue to induce clustering is organelle and/or tissue-specific. Our results emphasize the conservationand specificity of the pathway for each Vps33 homologue. Itis interesting to note that there is more evolutionaryconservation between human and rodent VPS33B orthologuesthan the VPS33A orthologues, and that the phenotypic conse-quences of VPS33B inactivation are more severe than forVPS33A.
Ka/Ks analysis of VPS33B identified five regions of strongstabilizing evolutionary influence. To gain insights into thepossible relevance of these to VPS33B function, we undertookstructural predictions based on the crystallographic study ofthe nSec1–Syntaxin1 complex (18). Three predicted bindingsites for syntaxin corresponded to three of the five regionswith strong stabilizing selection identified by the Ka/Ks analy-sis (labelled A, B and C). Furthermore, the one known patho-genic VPS33B missense mutation (L30P, which causes severeARC phenotype) is predicted to disrupt the putativeN-terminal binding site A. The other two evolutionary stable
regions were not predicted to be implicated in syntaxinbinding. However, region D corresponded to a 31 aminoacid insertion present in VPS33B and not VPS33A, suggestinga key role in VPS33B function. Accordingly both VPS33Aand a VPS33B protein-lacking region D demonstratedimpaired or absent ability to induce vesicular clusteringwhen compared with wild-type VPS33B.
Clustering and fusion of LAMP1-positive organelles suchas late endosomes, lysosomes or pigment granules in mamma-lian cells may be induced by a variety of mechanisms. Thusoverexpression of Rab7, Rab7-interacting lysosomal protein,Vps18, Vps39 and Vps33b, and a dominant-negative mutantof Rab27a can all induce clustering. The ability to induce clus-tering has been used to implicate candidate proteins in vesicu-lar trafficking processes (6,29–32). Although the precisemechanisms by which these interventions induce clusteringhave not been defined they are known to involve cell motorsand components of the actin cytoskeleton such as actin,ezrin and specific unconventional myosins, all of whichsurround clusters (33). Although we have demonstrated pre-viously that VPS33B could induce clustering, the function ofVPS33A has not been investigated before. To our knowledgethis is the first demonstration of the organelle/tissue-specificnature of clustering. Here we suggest that the interactionbetween VPS33B and syntaxin may be important in clustering,as deletion of the predicted N-terminal syntaxin binding sitediminished the ability to cause clustering by overexpression.However, deletion of the C-terminal portion of the gene alsoimpaired VPS33B-induced clustering. This region ofVPS33B is not predicted to be involved in syntaxin bindingbut does contain the stretch of 31 amino acids that is absentin VPS33A. Thus, the presence of this motif correlated withthe ability to induce clustering and may relate to interactionswith other proteins such as Rab 7 or VPS18. Further studiesto define the functional significance of the VPS33B C-terminalmotif will provide insights into the organelle/tissue-specificeffects of VPS33A and VPS33B and their differing roles inmetazoans.
Figure 5. Clustering of melanosomes in mouse malignant melanoma cells transfected with pCMVHAþVPS33A construct. (A) Blue DAPI staining of nuclei,black melanosomes on phase-contrast microscopy. (B) The same image showing co-localization of the construct stained with anti-HA antibody (red) and themelanosome cluster (black). Arrows point to the cluster of melanosomes.
Human Molecular Genetics, 2005, Vol. 14, No. 10 1267
Our findings demonstrate how a combination of in silicoevolutionary analysis and functional studies can identifycandidate regions critical for normal protein function.Recent discoveries of the role of vesicular trafficking genesin human disease emphasize the importance of these pathwaysfor cell homeostasis and the importance of research in modelorganisms to human genetics.
MATERIALS AND METHODS
Homology search
A homology search was performed using standard BLASTalgorithms on the NCBI BLAST server (http://www.ncbi.nlm.nih.gov/BLAST/) against the non-redundantdatabase, FlyBase (http://flybase.bio.indiana.edu/), Wormbasehttp://www.wormbase.org/), the Zebrafish genome resourceon the NCBI server (http://www.ncbi.nlm.nih.gov/genome/seq/DrBlast.html), Saccharomyces Genome Database (http://www.yeastgenome.org/), UCSC genome browser http://genome.ucsc.edu/cgibin/hgGateway) and Ensembl genomebrowser (http://www.ensembl.org/).
Multiple sequence alignment
Multiple sequence alignment for the Vps33p protein homo-logues was performed using ClustalW and Boxshade softwareon the Baylor College of Medicine website (http://searchlauncher.bcm.tmc.edu/multi-align/multi-align.html). Forthe phylogenetic analysis the alignment was further purgedusing Gblocks to extract regions of high quality informativealignment (34). Default settings were employed.
Phylogenetic tree generation
A phylogenetic tree was created using MrBayes softwaredownloaded from http://morphbank.ebc.uu.se/mrbayes/.MrBayes is a program for the Bayesian estimation of phylo-geny, which is based upon the posterior probability distri-bution of the trees, conditioned on the observations (35).The posterior probability distribution of trees is calculatedusing a simulation technique, Markov chain Monte Carlo(MCMC). The analysis was performed on the protein align-ment using a mixed model for amino acid evolution. Theanalysis was run for a million generations, resulting in10 000 trees. A conservative burn in of 7000 trees wasemployed. The 50% consensus rule tree was then constructedon the remaining 3000 trees with the support values givenbeing the proportion of the trees supporting the consensustree. A second tree was created from the same alignmentusing Tree Puzzle, a quartet puzzling maximum likelihoodmethod (36). The Mueller Vingron model of amino acid evol-ution was employed with one fixed and eight gamma rates.The nodal values are the support values estimated by themethod.
Ka/Ks sliding window analysis
Analysis of the molecular evolution of the VPS33A andVPS33B genes was undertaken using a sliding window
approach for the human, mouse and rat genes. The variablesmeasured in each pairwise comparison were: (a) Ka, the rateof non-synonymous substitution per non-synonymous site (asubstitution being a mutation that has gone to fixation) and(b) Ks, the rate of synonymous substitution per synonymoussite. The ratio Ka/Ks provides a measure of the form of selec-tion occurring. The meaning of particular values of Ka/Ks
ratios is: Ka/Ks ¼ 1, sequence is evolving neutrally; Ka/Ks , 1, the sequence is under stabilizing selection; Ka/Ks � 1, the sequence is under directional selection. Mostanalyses result in ratios of between 0.1 and 0.2. We used over-lapping sliding window with the size of the block of 90nucleotides and jumps of 30 nucleotides. The protocol of Li(37) was employed to estimate Ka and Ks in each window.
Structural modelling
Structural representation of the VPS33B and VPS33A proteinswas performed using RasMol, a program for moleculargraphics visualization (www.openrasmol.org).
Molecular cloning of the Vps33p homologues
We obtained human cDNA by RT–PCR amplification of thetotal RNA from a healthy subject. Forward primers forVPS33B and VPS33A genes included the start codon and thereverse primers contained the native stop codon (all primersequences available on request). A VPS33B construct with adeleted putative N-terminal binding site (deleted aminoacids 46–52 MSPLDRI) was made using Bgl II restrictionendonuclease and PCR of the replacement insert. A VPS33Bdeletion construct VPS33B(D450–480) was made to removethe amino acids absent in VPS33A by PCR of the flankingDNA with the primers to create a BamHI restriction site.The two flanking fragments were then ligated and insertedinto the pCMVHA vector. A VPS33B(1–438) construct wasmade by PCR including start codon and the stop codonmimicking the truncating mutation R438X. cDNAs werecloned into the pCMVHA expression vector (Invitrogen).
We cloned cg5127 and carnation cDNA by RT–PCRamplification of total RNA from D. melanogaster embryos(a kind gift from Dr Badenhorst, University of Birmingham).To allow subcloning of the cDNA fragments into thepCMVHA expression vector, we designed the forwardprimers with EcoRI (VPS33B, Car, cg5127) or Sal1(VPS33A ) restriction sites and reverse primers with a Kpn1restriction sites. All constructs were verified by sequencingusing an ABI 3730 DNA Sequencer. Expression of epitope-tagged protein following transfection of the constructs wasverified by western blotting with a monoclonal anti-HA anti-body (Sigma Aldrich).
Cells and antibodies
We grew human adult RCC4 and mouse malignant melanomaF1P43 cells in Dulbecco’s minimal essential mediumsupplemented with 5% fetal calf serum (Sigma Aldrich).F1P43 cell line was a generous gift from Dr Elena Sviderskaya(St George’s Hospital Medical School, University of London,UK). We obtained monoclonal antibodies against LAMP1
1268 Human Molecular Genetics, 2005, Vol. 14, No. 10
(Developmental Studies Hybridoma Bank, University ofIowa). Other antibodies were obtained commercially.
Transient transfection and immunofluorescence
For transient transfection experiments, we grew RCC4 cells andmalignant melanoma F1P43 cells on glass coverslips to 80%confluence and transfected them with 500 ng of plasmidDNA. RCC4 cells were transfected with pCMVHA emptyvector control, pCMVHAþVPS33B, pCMVHAþVPS33B(D46–52), pCMVHAþVPS33B(1–438), pCMVHAþ
VPS33B(D450–480), pCMVHAþVPS33A, pCMVHAþ
Car, pCMVHAþ CG5127) constructs using Effectene reagent(Qiagen, GmbH) for 24 h. We then removed transfectioncomplexes and allowed cells to grow for another 24 h innormal medium. We washed RCC4 cells in phosphate bufferedsaline (PBS), fixed them in 4% paraformaldehyde in PBS for30 min and permeabilized them in PBST buffer [PBS, 10%(v/v) normal swine serum, 0.1% (v/v) Tween 20]. We detectedendogenous LAMP1 using mouse monoclonal antibodiesdiluted 1:100 in PBST. We detected HA antigen using rabbitpolyclonal anti-HA antibody (Sigma Aldrich) at 1:100 inPBST. We detected primary antibodies using TRITC-conjugated anti-mouse IgG and FITC-conjugated anti-rabbitIgG antibodies diluted 1:100 in PBST. All antibodies wereapplied at room temperature in as moist chamber. Finally weapplied antifade (Vectashield, Vector Laboratories) containingDAPI (2 mg/ml). All experiments were performed in triplicate.
F1P43 cells were transfected with pCMVHA empty vectorcontrol, pCMVHAþVPS33B and pCMVHAþVPS33A asdescribed previously. We detected HA antigen using mono-clonal anti-HA antibody (Sigma Aldrich) at 1:100 in PBST.We detected primary antibodies using TRITC-conjugatedanti-mouse IgG.
We visualized the images with a Photometrics SenSys KAF1400-G2 CCD fitted to a Zeiss Axioplan epifluorescencemicroscope. Melanosomes were visualized using phasecontrast microscopy on the same system using an oil immer-sion objective. We captured images using SmartCapture 2software (Digital Scientific) running on a Macintosh G4computer.
Detection of clustering and statistical analysis
The captured images were analysed using Adobe Photoshop6.0 software. For the assessment of clustering in RCC4cells, the image of a single cell was isolated and the totalnumber of pixels and the median level of red luminosity(the intensity of the red signal) were detected. The numberof pixels occupied by the red signal with the intensityof above the median level of luminosity was recorded(median.pxls.). Clustering (Q) was measured as a fraction ofthe cell occupied by the red signal of above median intensity(Q) ¼ median.pxls/total.pxls). We defined clustered cells asthose with a Q-value that was less than two standard devi-ations compared with the average value of Q in controlcells. To standardize the level of expression of transfectedconstruct in each individual cell, we only selected cells withthe level of green luminosity within one standard deviationfrom the mean of the positive control (VPS33B wild-type
construct). There was no significant difference betweenexpression levels in the selected cells between differenttransfections.For the assessment of melanosome clustering in F1P43 cells,we observed the number of cells with pigmented clusters inthe field of view. The transfection efficiency was estimatedby measuring the red signal and no significant differencewas found between VPS33A and VPS33B. The significanceof the result was assessed by Fisher’s exact test.
ACKNOWLEDGEMENTS
The authors wish to thank Professor J. Paul Luzio for helpfuldiscussion and reading of the manuscript. This work wassupported by grants from Children Living with InheritedMetaBolic conditions (CLIMB) charity and BirminghamChildren’s Hospital Research Foundation. P.G. is a WellChildand RCPCH Research Fellow.
REFERENCES
1. Gissen, P., Johnson, C.A., Morgan, N.V., Stapelbroek, J.M., Forshew, T.,Cooper, W., McKiernan, P.J., Klomp, L.W.J., Morris, A.A.M.,Wraith, J.E. et al. (2004) Mutations in VPS33B, encoding a regulator ofSNARE-dependent membrane fusion, cause arthrogryposis–renaldysfunction–cholestasis (ARC) syndrome. Nat. Genet., 36, 400–404.
2. Sato, T.K., Rehling, P., Peterson, M.R. and Emr, S.D. (2000) Class C Vpsprotein complex regulates vacuolar SNARE pairing and is required forvesicle docking/fusion. Mol. Cell, 6, 661–671.
3. Banta, L.M., Robinson, J.S., Klionsky, D.J. and Emr, S.D. (1988)Organelle assembly in yeast. Characterization of yeast mutants defectivein vacuolar biogenesis and protein sorting. J. Cell Biol., 107, 1369–1383.
4. Oh, J., Ho, L., Ala-Mello, S., Amato, D., Armstrong, L., Bellucci, S.,Carakushansky, G., Ellis, J.P., Fong, C.T., Green, J.S. et al. (1998)Mutation analysis of patients with Hermansky–Pudlak syndrome: aframeshift hot spot in the HPS gene and apparent locus heterogeneity.Am. J. Hum. Genet., 62, 593–598.
5. Chiang, P.W., Oiso, N., Gautam, R., Suzuki, T., Swank, R.T. andSpritz, R.A. (2003) The Hermansky–Pudlak syndrome 1 (HPS1) andHPS4 proteins are components of two complexes, BLOC-3 and BLOC-4,involved in the biogenesis of lysosome-related organelles. J. Biol. Chem.,278, 20332–20337.
6. Poupon, V., Stewart, A., Gray, S.R., Piper R.C. and Luzio, J.P. (2003)The role of mVps18p in clustering, fusion, and intracellular localizationof late endocytic organelles. Mol. Biol. Cell, 14, 4057–4027.
7. Sevrioukov, E.A., He, J-P., Moghabi, N., Sunio, A. and Kramer, H. (1999)A role for the Deep Orange and Carnation eye color genes in lysosomaldelivery in Drosophila. Mol. Cell, 4, 479–486.
8. Suzuki, T., Oiso, N., Gautam, R., Novak, E.K., Panthier, J-J.,Suprabha, P.G., Vida, T., Swank, R.T. and Spritz R.A. (2003) The mouseorganellar biogenesis mutant buff results from a mutation in Vps33a, ahomologue of yeast vps33 and Drosophila carnation. Proc. Natl Acad. Sci.USA, 100, 1146–1150.
9. Rieder, S.E. and Emr S.D. (1997) A novel RING finger protein complexessential for a late step in protein transport to the yeast vacuole.Mol. Biol.Cell, 8, 2307–2327.
10. Price, A., Seals, D., Wickner, W. and Ungerman, C. (2000) The dockingstages of yeast vacuole fusion requires the transfer of proteins from acis-SNARE complex to a Rab/Ypt protein. J. Cell. Biol., 148, 1231–1238.
11. Toonen, R.F. and Verhage, M. (2003) Vesicle trafficking: pleasure andpain from SM genes. Trends Cell Biol., 13, 177–186.
12. Liu, J., Ernst, S.A., Gladycheva, S.E., Lee, Y.Y., Lentz, S.I., Ho, C.S.,Li, Q. and Stuenkel E.L. (2004) Fluorescence resonance energy transferreports properties of syntaxin1A interaction with Munc18-1in vivo. J. Biol. Chem., 279, 55924–55936.
13. Voets, T., Toonen, R.F., Brian, E.C., de Wit, H., Moser, T., Rettig, J.,Sudhof, T.C., Neher, E. and Verhage, M. (2001) Munc18-1 promoteslarge dense-core vesicle docking. Neuron, 31 581–591.
Human Molecular Genetics, 2005, Vol. 14, No. 10 1269
14. Rowe, J., Calegari, F., Taverna, E., Longhi, R. and Rosa, P. (2001)Syntaxin 1A is delivered to the apical and basolateral domains ofepithelial cells: the role of munc-18 proteins. J. Cell Sci., 114,3323–3332.
18. Misura, K.M., Scheller, R.H. and Weiss, W.I. (2000) Three-dimensionalstructure of the neuronal-Sec1–syntaxin 1a complex.Nature, 404, 355–362.
19. Gengyo-Ando, K., Kitayama, H., Mukaida, M. and Ikawa, Y. (1996) Amurine neuralspecific homolog corrects cholinergic defects inCaenorhabditis elegans unc-18 mutants. J. Neurosci., 16, 6695–6702.
20. Khan, A.H., Thurmond, D.C., Yang, C., Ceresa, B.P., Sigmund, C.D. andPessin, J.E. (2001) Munc18c regulates insulin-stimulated glut4translocation to the transverse tubules in skeletal muscle. J. Biol. Chem.,276, 4063–4069.
21. Verhage, M., Maia, A.S., Plomp, J.J., Brussaard, A.B., Heeroma, J.H.,Vermeer, H., Toonen, R.F., Hammer, R.E., van den Berg, T.K.,Missler, M., Geuze, H.J. and Sudhof T.C. (2000) Synaptic assembly of thebrain in the absence of neurotransmitter secretion. Science, 287, 864–869.
22. Bhattacharya, S., Stewart, B.A., Niemeyer, B.A., Burgess, R.W.,McCabe, B.D., Lin, P., Boulianne, G., O’Kane, C.J. and Schwarz, T.L.(2002) Members of the synaptobrevin/vesicle-associated membraneprotein (VAMP) family in Drosophila are functionally interchangeable invivo for neurotransmitter release and cell viability. Proc. Natl Acad. Sci.USA, 99, 13867–13872.
23. Gallwitz, D. and Jahn, R. (2003) The riddle of the Sec1/Munc-18proteins – new twists added to their interactions with SNAREs.Trends Biochem. Sci., 28, 113–115.
24. Jahn, R. and Sudhof, T.C. (1999) Membrane fusion and exocytosis. Ann.Rev. Biochem., 68, 863–911.
25. Betz, A., Okamoto, M., Benseler, F. and Brose, N. (1997) Directinteraction of the rat unc-13 homologue Munc13-1 with the N terminusof syntaxin. J. Biol. Chem., 272, 2520–2526.
26. Munson, M., Chen, X., Cocina, A.E., Schultz, S.M. and Hughson, F.M.(2000) Interactions within the yeast t-SNARE Sso1p that control SNAREcomplex assembly. Nat. Struct. Biol., 7, 894–902.
27. Subramanian, S., Woolford, C.A. and Jones, E. (2004) The Sec1/Munc18protein, Vps33p, functions at the endosome and the vacuole ofSaccharomyces cerevisiae. Mol. Biol. Cell, 15, 2593–2605.
28. Bugnicourt, A., Froissard, M., Sereti, K., Ulrich, H.D., Haguenauer-Tsapis, R. and Galan, J-M. (2004) Antagonistic roles of ESCRT and Vpsclass C/HOPS complexes in the recycling of yeast membrane proteins.Mol. Biol. Cell, 15, 4203–4214,
29. Jordens, I., Fernandez-Borja, M., Marsman, M., Dusseljee, S.,Janssen, L., Calafat, J., Janssen, H., Wubbolts, R. and Neefjes, J. (2001)The Rab7 effector protein RILP controls lysosomal transport byinducing the recruitment of dynein-dynactin motors. Curr. Biol., 11,1680–1685.
30. Cantalupo, G., Alifano, P., Roberti, V., Bruni, C.B. and Bucci, C. (2000)Rab-interacting lysosomal protein (RILP): the Rab7 effector required fortransport to lysosomes. EMBO, 20, 683–693.
31. Caplan, S., Hartnell, L.M., Aguilar, R.C., Naslavsky, N. andBonifacino, J.S. (2000) Human Vam6p promotes lysosome clustering andfusion in vivo. EMBO, 20, 683–693.
32. Hume, J.A.N., Collinson, L.N., Rapak, A., Gomes A.Q., Hopkins, C.R.and Seabra, M.C. (2001) Rab27a regulates the peripheral distribution ofmelanosomes in melanocytes. J. Cell. Biol., 152, 795–808.
33. Richardson, S.C.W., Winistorfer, S.C., Poupon, V., Luzio, J.P. andPiper, R.C. (2004) Mammalian late Vps orthologues participate in earlyendosomal fusion and interact with the cytoskeleton. Mol. Biol. Cell, 15,1197–1210.
34. Castresana, J. (2000) Selection of conserved blocks from multiplealignments for their use in phylogenetic analysis. Mol. Biol. Evol., 17,540–552.
35. Huelsenbeck, J.P. and Ronquist, F. (2001) MRBAYES: Bayesianinference of phylogenetic trees. Bioinformatics, 17, 754–755.
36. Strimmer, K. and von Haeseler, A. (1996) Quartet puzzling: a quartetmaximum likelihood method for reconstructing tree topologies. Mol. Biol.
Evol., 13, 964–969.
37. Li, W.-H. (1993) Unbiased estimation of the rates of synonymous andnonsynonymous substitution. J. Mol. Evol., 36, 96–99.
1270 Human Molecular Genetics, 2005, Vol. 14, No. 10
