BLOC-3, a Protein Complex Containing the Hermansky-Pudlak Syndrome Gene Products HPS1 and HPS4*

The Hermansky-Pudlak syndrome (HPS) is a genetic disorder characterized by defective lysosome-related organelles. HPS results from mutations in either one of six human genes named HPS1 to HPS6 , most of which encode proteins of unknown function. Here we report that the human HPS1 and HPS4 proteins are part of a complex named BLOC-3 (for biogenesis of lysosome-re- lated organelles complex 3). Co-immunoprecipitation experiments demonstrated that epitope-tagged and endogenous HPS1 and HPS4 proteins assemble with each other in vivo . The HPS1 (cid:1) HPS4 complex is predominantly cytosolic, with a small amount being peripherally associated with membranes. Size exclusion chromatography and sedimentation velocity analyses of the cytosolic fraction indicate that HPS1 and HPS4 form a moderately asymmetric protein complex with a molecular mass of (cid:1) 175 kDa. HPS4-deficient fibroblasts from light ear mice display normal distribution and trafficking of the lysosomal membrane protein, Lamp-2, in contrast to fibroblasts from AP-3-deficient pearl mice (HPS2), which exhibit increased trafficking of this lysosomal protein via the plasma membrane. Similarly, light ear fibroblasts display an apparently normal accumulation of Zn 2 (cid:1) in intracellular vesicles, unlike pearl fibroblasts, which exhibit a decreased intracellular Zn 2 (cid:1) storage. Taken together, these observations demonstrate that the HPS1 and HPS4 proteins are components of charomyces cerevisiae AH109 lithium acetate procedure described in instructions for MATCHMAKER two-hybrid kit (Clontech). colony assays, AH109 presence

A group of specialized cytoplasmic organelles including melanosomes and platelet dense bodies are biogenetically related to lysosomes, and are hence referred to as "lysosome-related organelles" (LROs) 1 (1). Both LROs and lysosomes are formed from the trans-Golgi network and endosomes by pathways that share various common steps involving vesicle budding, maturation, translocation, targeting, fusion, and fission. The distinctive properties of LROs, however, require the existence of additional, LRO-specific events for their biogenesis.
Insights into the molecular machinery involved in LRO biogenesis have been recently gained from the study of genetic disorders that affect multiple LROs in both humans and mice (1)(2)(3). In both species, defects in the biogenesis of melanosomes and platelet dense bodies lead to reduced pigmentation and prolonged bleeding, respectively. One of these disorders, known as Chediak-Higashi syndrome in humans and the beige mutation in mice, is caused by mutations in a gene encoding the protein, LYST (4). Another human disorder, Hermansky-Pudlak syndrome (HPS), results from mutations in either one of six genes designated HPS1-HPS6 (5)(6)(7)(8)(9). Mutations in the orthologous genes have been identified in the mouse coat color mutants, pale ear (HPS1 (10,11)), pearl (HPS2 (12)), cocoa (HPS3 (13)), light ear (HPS4 (8)), ruby eye-2 (HPS5 (9)), and ruby eye (HPS6 (9)), which serve as animal models for the human disease. Other studies have led to the identification of additional genes mutated in other mouse models of HPS, including mocha (14), gunmetal (15), buff (16), pallid (17), muted (18), and cappuccino (19). Mutations in the orthologous genes, however, have not yet been reported in HPS patients. Finally, the genes mutated in other mouse HPS models, including subtle gray, reduced pigmentation, and sandy, remain to be identified. HPS2/pearl and mocha encode the ␤3A (6, 12) and ␦ subunits (14), respectively, of the adaptor protein (AP) complex, AP-3. This complex is a known component of the protein trafficking machinery and is involved in the sorting of the transmembrane proteins tyrosinase (21)(22)(23) and the quail neuroretina clone 71 (QNR-71) protein (24) to melanosomes. Gunmetal encodes the ␣ subunit of Rab geranylgeranyl transferase, an enzyme that adds geranylgeranyl groups to Rab GTP-binding proteins (15). Finally, buff codes for VPS33A (16), a component of the HOPS⅐C-Vps complex involved in protein transport to the yeast vacuole and mammalian lysosomes (25,26).
Strikingly, all the other HPS genes cloned to date encode novel proteins with no recognizable homology or structural motifs. Some of these novel proteins have been recently shown to be components of oligomeric complexes. For example, the products of the pallid, muted, and cappuccino genes are subunits of a cytosolic complex named BLOC-1 (for biogenesis of lysosome-related organelles complex 1) (19,27,28), whereas HPS5 and HPS6 form another cytosolic complex named BLOC-2 (9).
Several lines of evidence support the possibility that other HPS gene products might also be components of multiprotein complexes. HPS1/pale ear encodes an 80-kDa cytosolic protein (5,29,30), which is absent from light ear (HPS4-deficient) mouse platelets (8). This is reminiscent of cells deficient in AP-3 (6,31) or BLOC-1 (19,27,28), in which the absence of one subunit leads to the loss of other subunits of the corresponding complex. This is the result of quality control mechanisms that dispose of unassembled subunits of multiprotein complexes. Moreover, the pale ear and light ear mutants (8), or the double-homozygote mutant mice (32), show virtually identical phenotypes of reduced pigmentation limited to the ears and tail (8), indicating that HPS1 and HPS4 may interact physically and/or functionally.
Herein we demonstrate that HPS1 and HPS4 form a complex that we name BLOC-3. Both the endogenous and exogenously expressed, epitope-tagged forms of human HPS1 and HPS4 co-precipitate and co-migrate with each other on sedimentation velocity and gel filtration analyses. The sedimentation coefficient and size exclusion chromatographic properties of BLOC-3 differ from those of AP-3 and BLOC-1, consistent with the existence of multiple complexes involved in the biogenesis of LROs. BLOC-3 appears to be mostly cytosolic on subcellular fractionation, although a small fraction is peripherally associated with membranes. In contrast to AP-3-deficient cells (6,22,29,33), light ear (HPS4-deficient) cells exhibit normal trafficking of lysosomal membrane proteins and normal storage of Zn 2ϩ in intracellular vesicles. These observations indicate that AP-3 and BLOC-3 mediate distinct steps in the biogenesis of LROs.

EXPERIMENTAL PROCEDURES
DNA Constructs-The constructs pCI-HA 3 and pCI-Myc 3 were generated by cloning of annealed primers containing three copies of the HA and Myc epitope tags into XhoI-HindIII and XhoI-EcoRI sites of the mammalian expression vector pCI-neo (Promega, Madison, WI), respectively. The epitope tagging at the amino terminus of HPS1 and HPS4 was performed through PCR amplification of the full-length human HPS1 (GenBank TM /EMBL/DDBJ accession number NM_000195) and HPS4 (GenBank TM /EMBL/DDBJ accession number NM_022081) cDNAs, followed by in-frame cloning into HindIII-EcoRI sites of the pCI-HA 3 (pCI-HA 3 -HPS1) and EcoRI site of the pCI-Myc 3 (pCI-Myc 3 -HPS4) vectors, respectively. The inserts of these two plasmids were also subcloned into XhoI-NotI sites of the vector pCDNA3.1-hygromycin (Invitogen). Yeast two-hybrid constructs in the pGBKT7 (TRP1) and pGADT7 (LEU2) vectors (Clontech) were prepared by cloning of both HPS1 and HPS4 full-length cDNAs.
Antibodies-The peptide sequence GKAKQKLLKHGVNLL, corre-  (35). The extracts were then subjected to a first immunoprecipitation (1 st IP) with affinity purified rabbit antibody to HPS1, rabbit antibody to HPS4, mouse monoclonal anti-HA or anti-Myc, as indicated. Washed immunoprecipitates were subsequently denatured by heating at 95°C for 5 min in the presence of SDS and dithiothreitol, diluted with lysis buffer, and subjected to a recapture immunoprecipitation step (2 nd IP) using either anti-HA or anti-Myc. The resulting immunoprecipitates were analyzed by SDS-PAGE on 4 -20% gradients followed by fluorography. The signals shown in lanes 4, 6, and 14 correspond to 2% of the immunoprecipitated material; 5% of the immunoprecipitated material was loaded in lane 13, whereas 100% of the immunoprecipitated was loaded in lanes 3, 5, 11, and 12. B, whole cell extracts from metabolically labeled MNT-1 cells were subjected to immunoprecipitation-recapture analysis, as described in A, using either antibodies to the HPS1 or HPS4 proteins for the 1 st IP and antibodies to an irrelevant protein (c), to the HPS1 or HPS4 proteins for the 2 nd IP step. The ϳ175-kDa band present in lanes 2 and 5 seems to represent a nonspecific interaction since it was immunoprecipitated by the two different antibodies that were used. The identity of the ϳ70-kDa band in lane 2 is uncertain at this time, but may represent a minor degradation product of HPS1. The positions of molecular mass markers are indicated on the left. sponding to residues 694 -708 of human HPS4 (8) was used to generate a rabbit polyclonal antibody (AnaSpec Inc., San Jose, CA). The generation and purification of rabbit antibodies to AP3 ␤3A (34), HPS1 (29), and Pallidin (28) have been described previously. The following mouse monoclonal antibodies were also used: anti-HA (clone HA.11), anti-Myc (clone 9E10) (Covance, Richmond, CA), anti-␥1-adaptin AP1 (clone 100/3) (Sigma), anti-GGA3 (BD Transduction Laboratories, San Jose, CA). The monoclonal rat antibody to mouse Lamp-2 (clone ABL-93) was obtained from the Developmental Studies Hybridoma Bank (Iowa City, IA). Horseradish peroxidase-conjugated secondary antibodies for immunoblotting were from Amersham Biosciences. Alexa 488-conjugated goat anti-rat IgG, Alexa 488-conjugated goat anti-rabbit IgG, and Alexa 568-conjugated goat anti-mouse IgG were purchased from Molecular Probes (Eugene, OR).
Sedimentation Velocity Analysis and Size Exclusion Chromatography-Sedimentation velocity analysis was performed by ultracentrifugation on linear 2-15% (w/v) sucrose gradients (12 ml) prepared in buffer B (25 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EGTA, 1 mM dithiothreitol, 0.5 mM MgCl 2 , and protease inhibitor mixture). Cytosol (350 l) from cells either stably (clone number 42) or transiently expressing HA 3 -HPS1 and Myc 3 -HPS4, prepared in buffer B as described (36), was layered on top of a linear sucrose gradient and centrifuged in a SW41 rotor (Beckman Coulter Inc., Fullerton, CA) at 39,000 rpm for 17 h at 4°C. Eighteen 0.65-ml fractions were collected from the bottom of the tube and analyzed by immunoprecipitation and immunoblotting. The following protein standards from Amersham Biosciences were used (Svedberg coefficients in parentheses): chicken ovalbumin (3.6 S), bovine serum albumin (4.6 S), rabbit aldolase (7.3 S), and bovine catalase (11.3 S).
For size exclusion chromatography analysis, 100 l of cytosol from cells stably expressing HA 3 -HPS1 and Myc 3 -HPS4 (clone number 42) prepared in buffer B was applied to a calibrated Superdex 200 HR 10/30 column (Amersham Biosciences) equilibrated and eluted at 4°C with buffer B at a flow rate of 0.3 ml/min. Fractions (0.4 ml) were collected and analyzed by immunoprecipitation and immunoblotting. Column calibration was performed using the following protein standards from Amersham Biosciences (Stokes radii in parentheses): bovine thyroglob-ulin (85 Å), horse ferritin (61 Å), rabbit aldolase (48.1 Å), bovine serum albumin (35.5 Å), and chicken ovalbumin (30.5 Å). Molecular mass and frictional ratio (f/f 0 ) values were calculated from Stokes radii and sedimentation coefficients assuming a partial specific volume of 0.72-0.75 cm 3 /g, as described (37).
Electrophoresis and Immunoblotting-SDS-PAGE and immunoblotting analysis were performed as described (36). Horseradish peroxidase-labeled antibodies were detected by using the Western lightning, chemiluminescence reagent plus (PerkinElmer Life Sciences).
Antibody Internalization and Fluorescence Microscopy-Fibroblasts grown on glass coverslips were incubated for 15 min at 37°C in the presence of primary antibodies diluted in Dulbecco's modified Eagle's medium, 0.5% (w/v) bovine serum albumin, and 25 mM HEPES, pH 7.4. Subsequently, cells were washed for 5 min in ice-cold phosphate-buffered saline, fixed in 4% formaldehyde for 15 min, and then processed for immunofluorescence (20). Images were acquired on a Zeiss LSM 410 confocal microscope (Carl Zeiss Inc., Thornwood, NY). To detect intracellular zinc store sites, either fixed or unfixed cells were incubated for 1 h with 25 M zinquin ethyl ester (Toronto Research Chemicals, Ontario, Canada) in Dulbecco's modified Eagle' medium containing 25 mM HEPES, pH 7.4, and 0.1% (w/v) bovine serum albumin. Stained samples were washed extensively with phosphate-buffered saline and mounted on glass slides using Fluoromount G (Southern Biotech, Birmingham, AL). Images were acquired on a Zeiss LSM 410 confocal microscope using a 413-nm laser.
General Biochemical Procedures-Metabolic labeling of cultured cells with [ 35 S]methionine, preparation of Triton X-100 extracts, preparation and ultracentrifugation of detergent-free extracts, salt extraction of microsomal membranes, and immunoprecipitation-recapture were performed as described (35,36).
Yeast Culture, Transformation, and Two-hybrid Assays-The Saccharomyces cerevisiae strain AH109 (Clontech) was transformed by the lithium acetate procedure as described in the instructions for the MATCHMAKER two-hybrid kit (Clontech). For colony growth assays, AH109 transformants were resuspended in water to 0.1 A 600 /ml, then 5 l were spotted on plates lacking leucine and tryptophan, in the presence or absence of histidine, and incubated at 30°C for 4 -5 days.

RESULTS
Co-immunoprecipitation of HPS1 and HPS4-The possible association of HPS1 with HPS4 was investigated in M1 human fibroblasts stably transfected with HA 3 -tagged HPS1 and/or Myc 3 -tagged HPS4. The association of these proteins was analyzed by immunoprecipitation-recapture (38) using extracts of M1 clones metabolically labeled with [ 35 S]methionine (Fig. 1). In cells expressing only HA 3 -HPS1, this protein could be de- tected as an ϳ88-kDa protein by the use of antibody to the HA epitope in both the immunoprecipitation and recapture steps (Fig. 1A, lane 4). This molecular mass is consistent with that previously reported for the endogenous protein (29,30), accounting for the molecular mass of a triple HA epitope. Interestingly, use of an antibody to HPS4 in the immunoprecipitation step and an antibody to HA in the recapture step also resulted in the detection of the ϳ88-kDa HA 3 -HPS1 (Fig. 1A,  lane 3), consistent with the association of the endogenous HPS4 and the recombinant HA 3 -HPS1 in these cells. In cells expressing only Myc 3 -HPS4, immunoprecipitation and recapture with anti-Myc identified this protein as an ϳ98-kDa species (Fig.  1A, lane 6). This apparent molecular mass was larger than the ϳ78 kDa predicted from its amino acid sequence (8), even accounting for the contribution of a triple Myc epitope, indicating that HPS4 migrates anomalously on SDS-PAGE. Importantly, immunoprecipitation with anti-HPS1 followed by recapture with anti-Myc also yielded the ϳ98-kDa Myc 3 -HPS4 (Fig.  1A, lane 5). This demonstrated that the endogenous HPS1 assembles with the transgenic Myc 3 -HPS4. Similar experiments performed with cells expressing both HA 3 -HPS1 and Myc 3 -HPS4 demonstrated that these two epitope-tagged proteins also associate in vivo (Fig. 1, lanes 11-14).
The endogenous HPS1 and HPS4 were also found to interact with each other on immunoprecipitation-recapture analyses of untransfected MNT-1 (human melanoma) cells (Fig. 1B). This could be shown by immunoprecipitation with anti-HPS1 and recapture with anti-HPS4 (Fig. 1B, lane 3) or vice versa (Fig.  1B, lane 6). Taken together, these observations indicate that  both the endogenous and epitope-tagged HPS1 and HPS4 associate with each other in vivo. We did not detect co-immunoprecipitation of HPS1 or HPS4 with Pallidin (data not shown), indicating that the HPS1⅐HPS4 complex is distinct from BLOC-1.
The HPS1⅐HPS4 Complex Is Mostly Cytosolic-To determine the extent to which the HPS1⅐HPS4 complex is associated with membranes, M1 cells stably expressing HA 3 -HPS1 and Myc 3 -HPS4 were homogenized in buffer A (20 mM HEPES, pH 7.4, 0.25 M sucrose, 1 mM EGTA, 0.5 mM MgCl 2 , 1 mM dithiothreitol, and protease inhibitor mixture) and centrifuged for 10 min at 800 ϫ g. The supernatant of this centrifugation (i.e. postnuclear supernatant) was further centrifuged for 2 h at 120,000 ϫ g, yielding cytosol (C) and membrane (M) fractions (Fig. 2, upper panel). Analysis by immunoblotting using anti-HA and anti-Myc showed that most of these proteins were in the cytosol with only a small fraction (less than 10%) being associated with membranes. These observations were consistent with similar analyses previously performed for endogenous HPS1 (29,30), and also with immunofluorescence microscopy analyses that showed a predominantly cytosolic pattern for HA 3 -HPS1 and Myc 3 -HPS4 stably expressed in M1 cells (Fig.  3). The small fraction of membrane-associated HA 3 -HPS1 and Myc 3 -HPS4 could be extracted to various degrees with 1 M Tris-HCl (pH 7.4), 1 M NaCl, 0.2 M Na 2 CO 3 (pH 11.3) or 3 M urea (Fig. 2, upper panel). These observations indicated that the association of HA 3 -HPS1 and Myc 3 -HPS4 with membranes is mostly peripheral. These properties of the HPS1⅐HPS4 complex are similar to those described before for the BLOC-1 complex (27,28).
Hydrodynamic Properties of the HPS1⅐HPS4 Complex-The interaction of HPS1 with HPS4 was further characterized by conducting analyses of the sedimentation behavior of these proteins. To this end, extracts of M1 cells stably expressing both HA 3 -HPS1 and Myc 3 -HPS4 were subjected to sedimentation velocity analysis on sucrose gradients. Gradient fractions were assayed by immunoblotting with anti-HA and anti-Myc (Fig. 4). The distribution of Myc 3 -HPS4 on the gradients exhibited two peaks corresponding to sedimentation coefficients of 3.8 S (fraction number 5) and 6.3 S (fraction number 8) (Fig.  4A). The second peak coincided with the HA 3 -HPS1 peak (Fig.  4A). Immunoprecipitation with antibodies against one epitope followed by immunoblotting with antibodies to the other revealed a single peak at 6.3 S (Fig. 4, B and C). Therefore, the HPS1⅐HPS4 complex sediments as a 6.3 S species whereas the 3.8 S peak of Myc 3 -HPS4 likely represents excess, unassembled protein. The sedimentation coefficient of the HPS1⅐HPS4 complex (6.3 S) was distinct from that of BLOC-1 as shown by immunoblotting with anti-Pallidin antibody (5.2 S, Fig. 4F and Refs. 27 and 28). This further confirmed that HPS1⅐HPS4 and BLOC-1 are distinct complexes.
Although mRNAs encoding HPS1 and HPS4 are expressed in most cells and tissues, the HPS1 or HPS4 mutant phenotype is manifested primarily in specialized cell types such as melanocytes and the platelet-precursor megakaryocytes. It is thus conceivable that the composition, and consequently the size, of the HPS1⅐HPS4 complex might be different in these specialized cell types. To address this issue, we co-expressed HA 3  ior of the complex as described above. The complex was found to migrate with an identical sedimentation coefficient of 6.3 S in both M1 and MNT-1 cells (Fig. 5, compare B and E). Hence, the distinct requirement for HPS1 and HPS4 in melanocytes is not likely because of a difference in the size, and presumably the composition, of the HPS1⅐HPS4 complex. However, we cannot rule out that endogenous levels of a melanocyte-specific protein putatively interacting with the HPS1⅐HPS4 complex could only affect the sedimentation properties of a small fraction of the recombinant complex.
The hydrodynamic properties of the HPS1⅐HPS4 complex were further characterized by size exclusion chromatography. An extract of M1 cells stably expressing HA 3 -HPS1 and Myc 3 -HPS4 was run on a Superdex 200 HR column and the eluted fractions analyzed by either immunoblotting or immunoprecipitation followed by immunoblotting (Fig. 6). Again, the distribution of total Myc 3 -HPS4 showed two apparent peaks corresponding to Stokes radii of 65.2 and 55 Å (Fig. 6A). The first peak co-eluted with total HA 3 -HPS1 (Fig. 6A) and thus corresponded to the HPS1⅐HPS4 complex (Fig. 6B). This was further confirmed by the elution of HA 3 -HPS1 as identified by immunoprecipitation and immunoblotting (Fig. 6C). This gel filtration analysis also showed that the elution profile of the HPS1⅐HPS4 complex was distinct from those of the AP-3 complex detected by immunoblotting for ␤3A-adaptin (Fig. 6E), and BLOC-1 detected by immunoblotting for pallidin (Fig. 6F).
The sedimentation coefficient and Stokes radius of the HPS1⅐HPS4 complex were used to calculate its molecular mass and frictional ratio, as described (37). The calculated molecular mass of the complex is ϳ175 kDa, which approximates the sum of the molecular masses of the HPS1 and HPS4 polypeptides. The frictional ratio of ϳ1.8 indicates that it is a moderately asymmetric complex.
Yeast Two-hybrid Analysis of the HPS1-HPS4 Interaction-To examine whether the interaction between HPS1 and HPS4 proteins is direct, we performed yeast two-hybrid analyses (Fig. 7). GAL4bd constructs encoding the full-length HPS1 or HPS4 proteins were co-expressed with GAL4ad constructs encoding the same proteins. There was no evidence of interaction of HPS1 with HPS4, or of self-association of these two proteins (Fig. 7A), despite expression of all of these proteins, as demonstrated by immunoblotting of the co-transformed yeast cells (Fig. 7B).
Trafficking of Lysosomal Membrane Proteins in HPS4-deficient Cells-Defects in the AP-3 complex in fibroblasts and B-lymphoblastoid cells from ␤3A-deficient HPS2 patients (6), and in fibroblasts from the ␤3A-deficient (6,22) and ␦-deficient mice (6) result in enhanced trafficking of lysosomal membrane proteins such as CD63 and Lamp-1 via the cell surface. Nevertheless, these lysosomal proteins still reach the lysosomes following an endocytic route (6,39). To determine whether the HPS1⅐HPS4 complex might be similarly involved in the trafficking of lysosomal membrane proteins, we compared the uptake of antibodies to Lamp-2 and Lamp-1, and the steady-state distribution of these proteins, in primary cultures of wild-type, pearl (␤3A-deficient), pallid (BLOC-1), ruby eye (BLOC-2), and light ear (HPS4-deficient) fibroblasts. We observed that wildtype, pallid, and ruby eye fibroblasts took up barely detectable amounts of anti-Lamp-2 (Fig. 8, A, C, and D) antibody, whereas pearl cells internalized noticeable higher amounts (Fig. 8B). Using this assay, light ear cells exhibited a pattern of internalization similar to that of the wild-type cells (Fig. 8E). The steady state distributions of Lamp-2 in the cells from all five mice were not detectably different (data not shown). Similar results were observed when the internalization of Lamp-1 antibody and the steady state distributions of Lamp-1 were studied (data not shown). Together with previous findings (29), these observations indicate that, unlike AP-3, the HPS1⅐HPS4 complex as well as BLOC-1 and BLOC-2 are not directly involved in the trafficking of lysosomal membrane proteins. Vesicular Storage of Zinc Ions in HPS4-deficient Cells-Another phenotype associated with AP-3 deficiency in mice is a decrease in vesicular pools of Zn 2ϩ (22,27), which is probably caused by mislocalization or degradation of Zn 2ϩ transporters such as ZnT-3 (14). Vesicular Zn 2ϩ can be visualized by staining with the membrane-permeant Zn 2ϩ fluorescent probe, zinquin (40). Zinquin staining of fibroblasts from pearl mice revealed decreased accumulation of Zn 2ϩ (Fig. 9B), as previously reported (27). Fibroblasts from light ear (HPS4-deficient) mice, as well as those from pallid (BLOC-1) and ruby eye (BLOC-2) mice, displayed apparently normal zinquin staining (Fig. 9E). Therefore, vesicular storage of Zn 2ϩ in fibroblasts is also unaffected by the lack of expression of HPS4. This observation adds to the notion that AP-3 on one side and the HPS1⅐HPS4, BLOC-1, and BLOC-2 complexes on the other side act at different steps in the biogenesis of LROs.

DISCUSSION
The evidence presented here indicates that the HPS1 and HPS4 proteins are part of a novel complex that, in keeping with the nomenclature in the field, we propose to name BLOC-3 (for biogenesis of lysosome-related organelles complex 3). The molecular mass of this complex, calculated from sedimentation velocity and gel filtration analyses, is ϳ175 kDa. This corresponds approximately to the sum of the molecular masses of the HPS1 and HPS4 polypeptides, indicating that BLOC-3 has one copy of each of these polypeptides. Yeast two-hybrid assays have failed to demonstrate an interaction of HPS1 and HPS4 expressed as chimeras with the Gal4 DNA binding domain and transcription activation domain, respectively, even though the chimeras are well expressed in the co-transformed yeast cells. A trivial explanation could be that the chimeras expressed in yeast do not behave like the endogenous or epitope-tagged proteins do in mammalian cells. Alternatively, these results could point to the existence of a third component of the BLOC-3 complex, which allows the transgenic HPS1 and HPS4 proteins to come together when expressed by transfection into mammalian cells. This component would have to be very small, however, to fit within the margin of error in the estimation of the size of the BLOC-3 complex.
The physical association of the HPS1 and HPS4 proteins explains the absence of the HPS1 protein from light ear (HPS4deficient) fibroblasts, as unassembled subunits of multiprotein complexes are most often degraded (41). This has also been observed for the unassembled subunits of the AP-3 complex (6, 31) and BLOC-1 (19,27,28) in other forms of HPS. The assembly of HPS1 with HPS4 is also consistent with the similar phenotype observed in pale ear (HPS1-deficient), light ear (HPS4-deficient), and double-homozygote pale ear/light ear mice (8,32).
Subcellular fractionation analyses revealed that BLOC-3 is mostly cytosolic, with only a very small fraction associated with membranes. This could be because of dissociation from membranes during cell lysis. However, we have also observed a FIG. 7. Yeast two-hybrid analysis of HPS1 and HPS4 interaction. The AH109 yeast strain was co-transformed with full-length HPS1 and HPS4 cDNAs fused to GAL4 DNA-binding domain (GAL4bd) and GAL4 transcription activation domain (GAL4ad), respectively, and vice versa. A, after selection, co-transformants were plated on plates with (ϩ His) or without (Ϫ His) histidine. Interactions were detected based on the ability of the co-transformants to grow in the absence of histidine. B, whole extracts of the co-transformed yeast cells were subjected to SDS-PAGE and immunoblotting analysis using mouse monoclonal anti-Myc and anti-HA antibodies. Notice that the yeast two-hybrid control proteins (p53 and SV40 LT-Ag) are not fused to epitope tags. mostly cytosolic pattern of localization by immunofluorescence microscopy of cells stably expressing both HA 3 -HPS1 and Myc 3 -HPS4 and we were unable to detect any HPS1⅐HPS4 complex associated to membranes in cells permeabilized prior to fixation (data not shown). In this regard, BLOC-3 also resembles BLOC-1, which is mostly localized to the cytosol (27,28). Because both complexes contribute to the biogenesis of membrane-bound organelles, it is likely that the small fraction that is associated with membranes represents the active form of each complex. Studies by Spritz and colleagues (30) have shown that the membrane-associated form of HPS1 localizes to tubulovesicular and pre-melanosomal structures in melanotic cells. Morphological analyses of skin melanocytes from light ear (HPS4-deficient) mice have revealed an increase in the proportion of immature melanosomes, which is probably because of a kinetic block in their maturation (42). A reduced number of mature melanosomes and aberrant melanosomes were also observed in retinal pigmented epithelial cells of light ear (HPS4-deficient) mice (8). Thus, it is likely that BLOC-3 functions at an early stage of melanosome biogenesis.
The exact biochemical pathway in which BLOC-3 is involved is unknown and, unfortunately, the sequences of both the HPS1 and HPS4 proteins are not informative in this regard. Orthologs of these proteins exist in all mammals with sequenced genomes, as well as in Drosophila, but not in the yeasts S. cerevisiae and Schizosaccharomyces pombe. This correlates with the existence of LROs in metazoans but not in yeasts. It is likely that, like BLOC-1 and BLOC-2, BLOC-3 is part of the molecular machinery specifically dedicated to LRO biogenesis. In contrast, AP-3 (35,43,44) and HOPS/Vps-C (25,26) exist in all eukaryotes including yeast, exhibit homologies to other proteins, and have known functions in protein trafficking in the endosomal-lysosomal system. These complexes are probably components of the general endosomal-lysosomal trafficking machinery, and their absence leads to other phenotypic abnormalities in addition to those derived from LRO defects. Indeed, we have observed that AP-3 deficiency in fibroblasts causes enhanced trafficking of both Lamp-1 and Lamp-2 via the plasma membrane, whereas BLOC-3 deficiency does not appreciably affect the trafficking of these proteins. Similarly, AP-3-deficient cells exhibit reduced vesicular accumulation of Zn 2ϩ , whereas BLOC-3-deficient cells display apparently normal Zn 2ϩ accumulation.
The possibility that BLOC-3 might be involved in some aspect of lysosome biogenesis, however, cannot be ruled out. Light ear (HPS4-deficient) and pale ear (HPS1-deficient) mice have enlarged lysosomes with increased amounts of lysosomal hydrolases in the kidney, and decreased secretion of lysosomal hydrolases into the urine (45). In addition, macrophages from pale ear mice exhibit reduced secretion of mature lysosomal hydrolases upon treatment with ammonium chloride (20). It remains to be determined whether the lysosomes that undergo exocytosis in these cell types are true lysosomes or LROs.
The biochemical characterization of proteins that are defective in human and mouse models of HPS is thus beginning to uncover a distinct molecular machinery for LRO biogenesis in which various multiprotein complexes play essential roles. The description of the BLOC-3 complex reported here is another step toward unraveling the nature of this machinery.