The Beige/Chediak-Higashi Syndrome Gene Encodes a Widely Expressed Cytosolic Protein*

The human autosomal recessive disorder Chediak-Higashi syndrome and its murine homologue beigeare associated with the formation of giant lysosomes that cluster near the perinuclear region of cells. We prepared a polyclonal antiserum against a glutathione S-transferase-Beige fusion protein and demonstrated by Western analysis that the beige gene encodes a protein of 400 kDa that is expressed in cultured murine fibroblasts as well as most mouse tissues. The protein was not detected in either cultured fibroblasts or mouse tissues from two differentbeige mutants. Cultured fibroblasts transformed with multiple copies of yeast artificial chromosomes that contain the full-length beige gene showed much higher levels of Beige protein than either wild type fibroblasts or mouse tissues. Subcellular fractionation experiments demonstrated that the Beige protein was cytosolic and, under the conditions of isolation, had no measurable membrane association. Cultured mouse fibroblasts in which the Beige protein was overexpressed had smaller than normal lysosomes that were more peripherally distributed than in control cells. These findings, coupled with earlier published results, suggest that the Beige protein regulates lysosomal fission.

homologue of CHS designated beige.
Cells from patients with CHS and from beige mice contain giant intracellular vesicles that cluster around the nucleus. Affected vesicles include lysosomes, melanosomes, platelet dense granules, and cytolytic granules. It is not known whether giant vesicles form because of a defect in vesicle fusion or fission (2,3). The genes responsible for CHS and beige have been identified and are orthologous (4 -6), with the predicted sequence of the Beige/CHS protein yielding little insight into the protein's function.
We demonstrate here that the murine Beige protein is a cytosolic protein expressed in most tissues. A deficiency of Beige protein results a perinuclear clustering of giant lysosomes. In contrast, overexpression of Beige protein results in abnormally small lysosomes, which localize to the periphery of cells. We suggest that the Beige protein is involved in regulating vesicle fission and not fusion.

Generation of Polyclonal Antibodies against the Murine Beige Protein-
The pGEX-2T (Pharmacia Biotech Inc.) vector was used to make a recombinant fusion protein that contained glutathione S-transferase (GST) linked to the carboxyl-terminal 89 kDa of the murine Beige protein (GenBank accession number U52461). This portion of the Beige protein contains the highly conserved BEACH domain and seven WD40 repeats (4,5). This construct was made using PCR to synthesize the beige gene portion using the F primer 5Ј-ACAGGATCCGCTG-CAAGTGAATCCATCAGA-3Ј containing an EcoRI site at its 5Ј end and the R primer 5Ј-AGCGAATTCTCATCCAGCTGCGTAGCTGCT-3Ј that contained a BamHI site at its 5Ј end. These two primers were used with the high fidelity PCR polymerase Expand Long Template PCR System (Boehringer Mannheim) and a template cDNA plasmid named 22B (4), which contains the last 5.6 kb of the murine beige gene. This mixture was amplified using a Perkin-Elmer PCR machine and the following PCR parameters: 94°C for 2 min hot start, followed by 94°C for 30 s, 53°C for 30 s, and 72°C for 4 min for 35 cycles. This reaction yielded the expected band of 2.37 kb, which was then cut with EcoRI and BamHI and gel purified. This EcoRI-BamHI fragment was cloned into the EcoRI-BamHI site of pGEX-2T using standard cloning techniques.
Recombinant plasmids were recovered and assayed for production of the GST-Beige fusion protein (BP55). One clone was used for all of the large scale preparations of the GST-Beige fusion protein as per the manufacturer's instructions. The glutathione-Sepharose purified fusion protein was digested with thrombin to separate GST from the Beige protein portion. This digested protein mixture was run out on a 6% SDS-PAGE gel, stained with 0.05% Coomassie Brilliant Blue-R250 in distilled H 2 0, and the 89-kDa recombinant Beige protein band excised. Some of this band was transferred to PVDF membranes, and the amino terminus was sequenced. The sequence obtained precisely matched the expected Beige protein sequence. This gel purified band was mixed 1:1 with complete Freund's adjuvant and injected into New Zealand White rabbits. All following immunizations were similar except that incomplete Freund's adjuvant was used.
After five boosts, immune serum was obtained from one rabbit that recognized the starting Beige antigen as determined by Western blotting. This serum was further purified by column chromatography in which the antigen was coupled to Sepharose beads using the manufacturers instructions (Pierce).

Preparation of Fibroblast and Mouse Tissue Extracts, Subcellular
Fractionation, and Western Blotting-Mouse fibroblasts (described in Refs. 4 and 7) were maintained in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum or in some instances 1% fetal bovine serum with penicillin and streptomycin (P/S). All cells used in this study, except the bg/bg cell line SB/Le, contained a G418 resistance gene provided by the plasmid pRV1 (8) and were grown in 1 mg/ml G418 (Life Technologies, Inc.) without P/S. For production of the fibroblast protein extracts, confluent monolayers were washed once in phosphate-buffered saline, trypsinized, centrifuged at 200 ϫ g for 5 min, and resuspended in ice cold FB buffer (250 mM sucrose, 20 mM HEPES, pH 7.2, 50 mM KCl, 1 mM MgCl 2 , 0.5 mM EGTA, 20 mM leupeptin (Sigma), 10 mM pepstatin A (Sigma), and 1 mM [4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride] (ICN)). Some experiments were done in a FB buffer that lacked sucrose but that contained 0.5% Triton X-100. Cells were placed in a Dounce homogenizer and centrifuged at 400 ϫ g for 5 min to obtain a postnuclear supernatant (PNS). Mouse tissue homogenates were prepared as described above. Protein determinations were performed using the BCA method (Pierce) and bovine serum albumin fraction V (Sigma) as a standard.
Postnuclear supernatants were fractionated over 10 -60% (w/w) linear sucrose gradients for 3 h at 100,000 ϫ g. Fractions were collected from the bottom of the gradient and pelleted at 200,000 ϫ g for 90 min. The pellets were resuspended in 100 l of a 1% Triton X-100, phosphate-buffered saline, which contained the previously mentioned protease inhibitors, and then analyzed using SDS-PAGE and Western blotting. Hexoseaminidase activity was assayed as described previously (9), and refractive indices were determined on each gradient. All experiments were performed at least three times.
Precast 4 -20% gradient gels (Bio-Rad) or 5% SDS-PAGE gels that did not have a "stacker" were used to separate the protein extracts described above. The SDS-PAGE gels were transferred to PVDF membranes (Gelman Sciences) for 1-2 h using electrophoretic transfer buffer without methanol (25 mM Tris, pH 8, 200 mM glycine). The filters were blocked by incubation in TBST (20 mM Tris, pH 8, 0.9% NaCl, 0.1% Tween 20) with 10% dry milk (Carnation) for 30 min at 37°C or 5% milk overnight at 4°C. Membranes were probed using a 1:500 dilution of affinity purified rabbit anti-Beige antibodies diluted in TBST. For some experiments, the PVDF membranes were stripped in 2% SDS, 62.5 mM Tris, pH 6.7, and 100 mM ␤-mercaptoethanol (stripping buffer) for 30 min at 50°C and then reprobed. For antigen blocking experiments, either the starting GST-Beige fusion protein or GST was mixed with the diluted antibodies in TBST to a final concentration of 20 ng/ml and incubated overnight before addition to the PVDF membranes. The anti-G-Protein ␤ subunit polyclonal antibody was obtained from Santa Cruz Biotechnology, Inc., and was used at a 1:1000 dilution, and an anti-␤ COP antibody was obtained from Sigma and used at a 1:500 dilution.
Analysis of Lysosome Size by Flow Cytometry-Cultured fibroblasts were incubated in 1 mg/ml FITC-dextran (10,000 molecular weight, Molecular Probes) in the standard culture medium described above for 18 h. The cells were washed twice with culture medium and chased for 2 h in culture medium lacking the fluorescent ligand. Culture medium was then replaced with 4°C Hanks' minimal essential medium, and the samples were kept on ice throughout the remainder of the experiment. The cells were scraped and washed once in FB buffer followed by Dounce homogenization in 0.5 ml of FB plus protease inhibitors. Homogenates were centrifuged at 400 ϫ g for 5 min, and the PNS was collected and mixed with 2.5 ml of FACS loading buffer (275 mM sucrose, 20 mM HEPES, pH 7.2, 1 mM EGTA). Flow cytometric analysis was performed on a Becton Dickinson FACSVantage operating with CellQuest software and equipped with a Coherent Enterprise argon laser (Coherent Laser Group) producing 225 mW at 488 nm. For this application the basic instrument configuration was altered by removing the beam expanding lens in the laser focusing optics, removing the ND1 filter in front of the forward light scatter (FSC) detector and setting the acquisition threshold at 150 on log side scatter (log SSC). All other optical filters for light scatter and fluorescence remained in the basic configuration recommended by the manufacturer for FITC fluorescence. Optical alignment was optimized using DNA QC 2-m beads (Becton Dickinson) for log FSC, log SSC and log FL1 (FITC) fluorescence. Consistency of the alignment process was confirmed on a daily basis in two ways: For light scatter, a panel of four polystyrene beads ranging in size from 0.48 to 1.43 m (Polysciences, Inc.) was analyzed, and peak channels were compared with previous results. For log FSC the largest bead appeared just above the 10 3 division on a four log scale (linear channels 1200 -1300) and the smallest beads at or below the 10 1 division (linear channels 5-20). The log SSC peaks ranged from the 10 2 (linear channels 70 -130) down to 10 1 (linear channels [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. In a similar manner, FITC sensitivity was standardized with a FITC Quantum fluorescence kit (Flow Cytometry Standards) that consisted of a blank bead and four beads ranging in fluorescent intensity from 10,000 -256,000 mean equivalent fluorescent units.
Before data acquisition, sample buffer was analyzed to assess the noise level due to scattered laser light, electronic noise, and particles in the buffer. For a 2-min data acquisition period, the noise level typically ranged from 1000 to 5000 events. All background events were nonfluorescent (i.e. they were relatively low in the first decade of the FL1 distribution), so a data acquisition gate was set to include only those events positive for FITC fluorescence (Ͼ10 1 for log FL1). Data on 50,000 fluorescent events were collected and stored in list mode files for off-line analysis using CellQuest software (Becton-Dickenson). Lysosome size was measured as forward scatter height (FSC), and different samples were compared by computing the mean and median forward scatter heights on all 50,000 fluorescent events. The data presented in Fig. 3 are representative of four different experiments, all of which gave similar results.
Visualization of Fibroblast Lysosomes and Immunofluorescence-The visualization of mouse fibroblast lysosomes was performed as described previously in Refs. 11 and 12. Briefly, fibroblasts on coverslips were incubated in the above mentioned culture medium that contained either 0.5 mg/ml Lucifer Yellow-CH (Aldrich) or 1 mg/ml FITC-dextran (Molecular Probes) for 18 h. The medium was removed, and the cells were washed with fresh medium without the fluorescent probe, followed by a chase in culture medium for at least 2 h. The cells were visualized and photographed using fluorescent microscopy as described in Ref. 12.

RESULTS
Western Analysis of the Beige Protein-Western blot analysis of YAC-complemented bg J /bg J fibroblasts or mouse lung extracts detected two proteins, one of approximately 400 kDa and a smaller band of 40 kDa (Fig. 1, A and B). The molecular mass of the large band is similar to the Beige protein's predicted molecular mass of 425 kDa (13). These bands, which were unaffected by preincubation of the antibodies with GST, were not observed when the antibodies were preincubated with the intact GST-Beige fusion protein (Fig. 1). The smaller band may represent the heterotrimeric G-protein ␤ subunit, because an identically sized band was recognized by a commercially available antiserum raised against a carboxyl-terminal peptide of the mouse G-protein ␤ subunit (data not shown). Cross-reactivity between these two proteins might be expected due to the presence of conserved WD40 repeats in the G-protein ␤ subunit (14) and in the Beige protein (4).
The cloning of the murine beige gene was confirmed by the identification of a unique mutation within the bg allele (15), which is predicted to result in a truncated Beige protein that is missing the region of Beige that was used to create the GST-Beige fusion antigen (4,13). Fibroblasts from bg/bg mice lacked the 400-kDa band present in control mouse fibroblasts (Fig.  1B). The bg J allele has been shown to be associated with reduced levels of beige gene mRNA (6). The 400-kDa band was absent from extracts of bg J /bg J mouse fibroblasts (Fig. 1B). In all tissue extracts, both control and mutant, a band larger than 400 kDa was also detected. This band could represent crosshybridization to the CDC4L/BGL gene product (16), which is a large gene of unknown function that is expressed in a wide variety of cell types and which is highly related (i.e. 50% identical over a 300-amino acid stretch) to the Beige/CHS protein (4).
We cloned the beige gene using a YAC complementation strategy in which YACs were introduced into bg J /bg J mouse fibroblasts (7). During the course of these studies we isolated two complemented cell lines (195-3 and 195-4), in which the beige gene was overexpressed as determined by Northern analysis (4). Overexpression in these cells results from the amplification of the YACs as episomal DNA elements (7). Western analysis of extracts from these two cell lines showed a very strong 400-kDa band (Fig. 1B). This band was increased 13-fold in clone 195-3 and 6-fold in clone 195-4 compared with wild type cells. A third complemented cell line (113-1) that contained a different YAC also showed the expected 400-kDa band but at levels that were equal to those of control fibroblasts. Three cell lines (137-1, 151-1, and 151-2) that contained noncomplementing YACs did not show the 400-kDa band but still showed the unknown larger band (Fig. 1B). These results indicate that the 400-kDa band is the Beige protein due to its presence in all YAC-complemented cell lines, its absence in all noncomplemented cell lines, and its absence in two beige mutants.
Cytosolic Localization of the Beige Protein-Although the beige gene mRNA is expressed at very low levels in vivo, it can be detected by reverse transcription-PCR in most tissues tested (4). The 400-kDa Beige protein was detected in almost all mouse tissues tested, with the highest levels of expression seen in brain, spleen, and lung (Fig. 1C), and undetectable levels seen in heart and small intestine. Most tissues examined showed expression levels similar to that seen in the control mouse fibroblasts. The 400-kDa band seen in control mouse tissues was absent in all bg/bg mouse tissues. Subcellular fractionation of homogenates from YAC-complemented cell lines or wild type mouse lung indicated that the Beige protein was cytosolic and was not associated with membranes. Subcellular fractionation using sucrose or Percoll gradients indicated that the Beige protein showed no association with lysosomes or any other definable organelle. Using a variety of buffer conditions and nucleotides (i.e. ATP, GTP, or GTP␥S), we did not observe any association of the Beige protein with membranes (data not shown).
Examination of the distribution of Beige protein in wild type and uncomplemented cells by indirect immunofluorescence showed a weak diffuse cytoplasmic staining. Examination of the YAC-complemented cell lines also showed a diffuse cytoplasmic staining, but the signal seen was significantly brighter than in wild type or mutant cells (data not shown). The immunofluorescence data are consistent with the biochemical data and indicate that the Beige protein is cytosolic.
Effect of Beige Protein on Lysosome Size and Distribution-We examined the effects of Beige protein overexpression on lysosomal morphology using the fluorescent fluid phase marker Lucifer Yellow to visualize fibroblast lysosomes. All YAC-containing cell lines were derived from a C57BL/6J-bg J / bg J mouse fibroblast cell line (MCHSF2) that showed the classic beige/CHS phenotype of a reduced number of giant lysosomes clustered near the nucleus ( Fig. 2A and Refs. 4, 7, and  17). The control mouse fibroblast cell line, established from a C57BL/6J mouse, showed the typical wild type morphology of numerous small lysosomes that were scattered throughout the cell but that were predominantly located within the cell body (Fig. 2C). The YAC-complemented clone 113-1, which contained wild type levels of Beige protein, showed a lysosomal  (4,7,21). The boxes represent different genes that we had isolated as part of our positional cloning studies (with Beige and Nidogen being named). These genes are presented as markers to demonstrate the degree of overlap between the YACs shown. For a much more detailed YAC map please see Ref. 21. morphology similar to control fibroblasts (Fig. 2D). The two YAC-complemented overexpressing cell lines (195-3 and 195-4), however, showed a phenotype characterized by the presence of more peripheral lysosomes (Fig. 2, E and F). Approximately 50% of the cells in these two cell lines showed this altered phenotype, with another 45% displaying a wild type morphology and 5% displaying a beige phenotype. This phenotypic variability can be explained by the variability in the number of YAC copies per cell. Fluorescent in situ hybridization studies, as well as cloning studies, indicated that these episomal YACs were unstable and were readily lost from cells (7). Finally, the YAC-containing but noncomplemented clones 137-1 (Fig. 2B), 151-1, and 151-2 (data not shown) displayed the parental beige phenotype. All cell lines tested were maintained in 1 mg/ml G418, which was necessary to maintain the neomycin resistance gene that was contained on the YACs (7). Cell line 137-1 showed YAC amplification equal to cell line 113-1, which represented 2-4-fold amplification of the YACs, whereas cell lines 195-3 and 195-4 had 12-and 6-fold amplifications, respectively.
Visual examination of lysosomes in the YAC-complemented cell lines 195-3 and 195-4 suggested that they had smaller lysosomes than control cells. This observation was confirmed by quantifying lysosome size using a flow cytometric assay. As initially described by Murphy and colleagues, flow cytometric analysis can quantitatively determine the size and fluorescence intensity of individual intracellular endocytic vesicles within a crude cell homogenate (18,19). Mouse fibroblast lysosomes were labeled with FITC-dextran using protocols that specifically label lysosomes as determined by the co-localization of the fluorescent ligand with the lysosomal enzyme hexoseaminidase on Percoll gradients (data not shown). The labeled cells were homogenized, and the PNS was analyzed on a Becton Dickinson FACSVantage. As expected, the bg J /bg J fibroblasts (MCHSF2) had a greater percentage of larger lysosomes when compared with either control (C57-2CF) or Beige overexpressing fibroblasts (195-4) (Fig. 3A). The Beige overexpressing cell line 195-4 had a significantly larger number of smaller lysosomes than the wild type cell line. Similar results were obtained when the median or mean for the forward scatter height on all fluorescence positive events was determined (Table I). These measurements yield a relative number for the average size of lysosomes from one cell sample and allows for a direct comparison of the average vesicle size between different cell types. These data confirm the observation that the cells with the lowest amount of Beige protein or no Beige protein (bg J /bg J MCHSF2 cells or bg/bg SB/Le cells) have the largest lysosomes. Lysosome size was reduced upon expression of normal levels of Beige protein (C57-2CF and 113-1) and was further decreased in cells expressing the highest levels of Beige protein (195-3 and 195-4). DISCUSSION We have begun to define the function of the Beige/CHS protein by developing an affinity purified polyclonal antibody directed against the Beige protein and by observing the effects of overproduction of the Beige protein on lysosome morphology. The specific nature of our antibody preparation was established in five ways. First, a band of approximately 400 kDa was identified on Western blots in multiple mouse tissues and cell lines. The molecular mass of the Beige protein predicted by the gene sequence is 425 kDa. Second, this band was absent on Western blots from bg/bg and bg J /bg J mice. Third, the signal from the 400-kDa band was blocked by the addition of the GST-Beige fusion protein into the Western blotting mix but was not blocked by GST alone. Fourth, the 400-kDa band was present in all YAC-complemented cell lines but was missing in all noncomplemented lines. Finally, the 400-kDa band, seen on Western blots from YAC-complemented cell lines, was identical in size to the band seen in control mouse tissues.
Experiments utilizing subcellular fractionation of fibroblast and tissue extracts localized the Beige protein to the cytoplasm. We were unable to detect any membrane association between the Beige protein and any subcellular fraction. This finding was unexpected for two reasons. First, the beige protein has homology to the VPS15 and Huntington protein that encode proteins localizing to both cytosolic and membrane fractions (5). The VPS15 and Huntington proteins may cycle on and off membranes and are believed to be involved in vesicle dynamics. Second, the main morphologic abnormalities in cells lacking Beige/CHS protein occur in membrane bound organelles. In a previous report, we complemented the beige defect in a cell culture system using a somatic cell fusion approach (20). We observed complementation only after a normal and beige lysosome had fused. We concluded that the Beige protein required a membrane interaction to exert its effect, but when the beige/ CHS genes were cloned, the predicted protein sequence contained no obvious signal sequence or transmembrane domains. The cytosolic localization of the Beige protein may accurately reflect the distribution of the protein, in which case its role may be to interact with a second protein. This second protein may then interact with membranes. The existence of a protein interaction domain is suggested by the known function of WD40 repeats. An alternative that cannot be eliminated by our data is that the lack of membrane association may result from the techniques used for the cell fractionation studies, but no membrane association was observed by immunolocalization.
Overexpression of the Beige protein in YAC-complemented cell lines was the result of a 6 -12-fold amplification of the YACs as demonstrated by Southern and fluorescent in situ  hybridization analysis (7). Amplification resulted in overproduction of beige gene mRNA as determined by Northern analyses (4). Although the beige gene cDNA is quite large (11.6 kb of open reading frame), Northern blotting detected only one band in the YAC-complemented cell clones, 195-3 and 195-4 (4).
In agreement with the results of our Northern analysis, we detected the 400-kDa band only in YAC-complemented cell lines and wild type murine tissues. The 400-kDa band was also the only band that was increased in intensity in our YACcomplemented cell lines (Fig. 1B). These results strongly suggest that the 400-kDa Beige protein band is functionally significant to the beige phenotype.
Overexpression of the Beige protein in bg J /bg J fibroblasts produced a phenotype different from that in the beige mutants, the wild type cells, or the YAC-complemented clone 113-1, which contained wild type levels of Beige protein. Cells lacking the Beige protein (most CHS and beige alleles are believed to be null alleles; Refs. 4 -6) contained large lysosomes clustered near the nucleus. Cells with wild type levels of Beige protein contained smaller lysosomes clustered in the cell body. Overexpressing cells contained even smaller lysosomes located near the cell periphery. There is a tight correlation between the observed phenotype and the level of Beige protein with highest levels producing a phenotype, exactly opposite to that seen in beige/CHS mutants. It remains possible that the complementing YACs overexpress other genes, in addition to beige, and that these genes are responsible for the observed phenotype, but this seems extremely unlikely.
We previously demonstrated that beige/CHS giant lysosomes are as capable of fusing with other lysosomes as are wild type lysosomes (11,20). This finding strongly suggests that the formation of beige/CHS giant lysosomes results from a reduced rate of fission. The hypothesis that Beige protein regulates fission is supported by our finding that cells overexpressing Beige protein contain smaller than normal lysosomes. We have recently developed a lysosome-lysosome fusion assay (22) and have determined that cytosol from cells overexpressing the Beige protein had no effect on lysosome fusion in vitro. This finding, coupled with the smaller than normal lysosomes produced when the Beige protein was overexpressed, suggests that the defect in beige/CHS is one of lysosomal fission and not fusion. A molecular mechanism mediating fission is not obvious, but it is possible that one domain of the Beige protein provides a structural support for lysosomes and that another domain, through a linking protein, pulls at the lysosomal membrane until fission occurs.