0 1994 by The American Society for Biochemistry and Molecular Biology, Inc. Tm JOURNAL OF BIOLWICAL CHEMISTRY Vol. 269, No. 32, Isaue of August 12, pp. 20528-20532, 1994

Printed in U.S.A.

Modulation of Human Saposin B Sphingolipid-binding Specificity by Alternative Splicing A STUDY WITH SAPOSIN B-DERIVED SYNTHETIC PEPTIDES*

(Received for publication, February 18, 1994, and in revised form, April 25, 1994)

Sonia LamontagneS and Michel PotierO From the Service de Genetique Mddicale, Hdpital Sainte-Justine et Ddpartements de Pediatrie et de Biochimie, Universitk de Montreal, Montreal, Quebec, Canada H3T IC5

The saposins A, B, C, and D, produced by proteolytic maturation of the same precursor protein, prosaposin, are sphingolipid-binding proteins which function as ac- tivators for lysosomal enzymes involved in sphingolipid hydrolysis. The alternative splicing of the prosaposin gene results in the inclusion or exclusion of exon 8 into transcribed prosaposin mRNA through the use of alter- native acceptor sites. The relative abundance of each alternatively spliced mRNA was determined by reverse transcription-polymerase chain reaction in various hu- man tissues and cell lines. Exon 8 codes for only three amino acid residues, Gln-Asp-Gln, in the saposin B do- main of prosaposin. The prosaposin mRNA containing exon 8 is the major species in cultured skin fibroblasts, brain, and pituitary glands together with a smaller amount of mRNA devoid of exon 8, whereas the prosa- posin mRNA detected in liver and lymphoblasts was de- void of exon 8 insertion. Previous structural modeling studies on saposin B have suggested that the Gln-Asp-Gln insertion occurs in an amphipathic a-helix region of the protein which is implicated in the binding of G,,-gan- glioside. We report that synthetic peptides containing the a-helix, with and without the Gln-Asp-Gln insertion, have different binding affinities for G,,-ganglioside, sul- fatide, and sphingomyelin. The insertion of the Gln-Asp- Gln sequence completely abolishes the capacity of the peptide to bind G,,-ganglioside, whereas its affinity for sulfatide and sphingomyelin is increased about 4-fold and almost 2-fold, respectively. No significant binding of glucosylceramide was observed with both peptides. These results suggest that alternative splicing of prosa- posin mRNAmay change binding specificity of saposin B presumably to adapt to the variable sphingolipid com- position of tissues.

The saposins and their precursor, prosaposin, are sphingo- lipid-binding proteins. The saposins activate hydrolysis of sphingolipid substrates by lysosomal enzymes and may also been involved in sphingolipid transport in the cell. The prosa- posin precursor is processed into four similar saposins named A, B, C, and D. Saposin B has the capacity to bind several sphingolipids such as G,,-ganglioside (Wenger and Inui, 1984)

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Supported by Medical Research Council of Canada Studentship. 5 Supported by Medical Research Council of Canada Operating

tique Medicale, HBpital Sainte-Justine, 3175 Chemin Ste-Catherine, Grant. To whom correspondence should be addressed: Service de GBn6-

MontrBal, QuBbec, Canada H3T 1C5. Tel.: 514-345-4885; Fax: 514-345- 4801.

and globotriaosylceramide (Li et al., 1985) as inferred from stimulation of enzyme activity, as well as galactosyl and glu- cosylceramide (Soeda et al., 1993) and sulfatide (Fischer and Jatzkewitz, 1977) as shown by direct binding assay, We have previously suggested that a predicted amphipathic a-helix be- tween residues Glu5'j and Met65 of saposin B plays a major role in its lipid-binding function (Potier, 1988). A synthetic peptide composed of residues Ser5' to Glu6' of saposin B contained about 44% of residues in an a-helix conformation, as suggested by circular dichroism spectroscopy, and had the capacity to bind G,,-ganglioside in vitro.'

Saposin B is polymorphic in the region of the putative a-helix due to alternative splicing of exon 8 of the prosaposin gene. Splicing generates two major mRNAs, one containing an inser- tion of 9 bp2 coding for the amino acid sequence Gln-Asp-Gln and a second devoid of this insertion (Nakano et al., 1989; Holtschmidt et al. 1991). Because this insertion occurs in a putative a-helix previously implicated in G,,-ganglioside bind- ing,' alternative splicing of the prosaposin gene may be a mech- anism to modulate sphingolipid-binding specificity of saposin B in different tissues. In this study we have first evaluated the relative proportions of prosaposin mRNAs, resulting from al- ternative splicing of exon 8, in various human tissues and cell lines using the polymerase chain reaction on reverse tran- scribed mRNA. We have also studied the sphingolipid binding specificity of saposin B-derived synthetic peptides with and without the Gln-Asp-Gln insertion.

EXPERIMENTAL PROCEDURES Materials-The peptides SapB-18, NH,-Ser-Gln-Tyr-Ser-Glu-Ile-Ala-

Ile-Gln-Met-Met-Met-His-Met-Gln-Pro-Lys-Glu-NH,, and SapB-21, NH,-Ser-Gln-Tyr-Ser-Glu-Ile-Ala-Ile-Gln-Met-Met-Met-His-Met-Gln- Asp-Gln-Gln-Pro-Lys-Glu-NH,, were purchased from Neosystem Labo- ratory (Strasbourg, France). The SapB-21 peptide contains the Gln- Asp-Gln insertion encoded by exon 8 of the prosaposin gene. They were purified by reverse-phase high performance liquid chromatography (pu- rity higher than 97%) and analyzed by fast atom bombardment mass spectrometry. The oligonucleotide primers for PCR amplification were purchased from the Departement de Biochimie, Universit6 de Montreal or l'Institut de Recherche en Biotechnologie (Montreal, Quebec). Before use, the oligonucleotides were purified by Sephadex G-25 chromatog- raphy. Supplies for cell culture, the enzymes Taq DNA polymerase and Superscript reverse transcriptase RNase H- were obtained from Life Technologies, Inc. Sequenase and sequencing kit were obtained from U. S. Biochemical Corp. The RNase inhibitor, RNasine, was purchased from Promega Biotech Inc. ssS-dATP was obtained from DuPont, the G,,-ganglioside was purchased from Calbiochem, whereas the sul- fatide, glucosylceramide, and sphingomyelin were all obtained from Sigma. Manufacturers reported that the purity of the G,,-ganglioside was better than 99% by TLC. The sulfatide gave only one spot by TLC

Champagne, M. J., Lamontagne, S., and Potier, M. (1994) FEES

* The abbreviations used are: bp, base pair; RT, reverse transcription; Lett., in press.

PCR, polymerase chain reaction.

20528

Human Saposin B Sphingolipid Binding 20529

and the preparation was neither contaminated by sphingomyelin nor cerebrosides. The glucosylceramide gave two spots, the major spot which represented about 90% of total was glucosylceramide and the other corresponded to a hydroxy fatty acid derivative of glucosylceram- ide. The sphingomyelin gave few spots, one major representing more than 95% of total corresponded to sphingomyelin and other contami- nating substances were identified as sphingosin and phosphatidylcho- line (less than 1% each).

Isolation of Nucleic Acids from Tissues and Cultured Cells-Human liver, brain, and pituitary glands were obtained at autopsy and were kept frozen a t -80 "C. The cultured lymphoblastoid and skin fibroblast cell lines were obtained from our own cell bank. Total RNA and genomic DNA were prepared according to the method of Sambrook et al. (1989). Human tissues (about 1 g, wet weight) were cut into small pieces and homogenized for 30 s in a Polytron model PT3000 homogenizer in 4 M guanidinium thiocyanate, 0.1 M Tris, pH 7.5, 0.5% (w/v) laurylsarcosi- nate, and 1% (v/v) p-mercaptoethanol. With cultured cells, about 2 x lo7 cells were washed with phosphate-buffered saline, and the cell pellet was directly treated with the denaturing solution. RNA was purified by CsCl gradient centrifugation. With this method, we obtained 100400 pg of total RNA from cultured cells, about 250 pg of total RNNg of tissue, and 100-250 pg of genomic DNA.

PCR Amplification of Reverse Dunscribed mRNA-To estimate the relative amounts of the two alternatively spliced major prosaposin mRNAs, with and without exon 8, in various tissues and cell lines, the region of mRNA corresponding to exon 8 was amplified by RT-PCR using the sense primer 1 (5'-GCATCTC""l'CACCTCATC-3', T,,, = 58.3 "C) and antisense primer 2 (5'-CAGCCAGTATTCTGAAATTG-3', T, = 58.7 "C). Total RNA (1 pg), denatured at 70 "C for 10 min, was incubated with Superscript reverse transcriptase in the presence of 20 pmol of primer 1,19 units of RNasine, 0.5 m~ dNTP, 1 mM dithiothreitol in 20 mM Tris buffer, pH 8.4, 50 mM KCI, 2.5 mM MgC1, and 0.1 mg/ml of nuclease-free bovine serum albumin (buffer A) in a total volume of 20 pl. The reaction was started by incubation at 23 "C for 10 min, and the cDNA was synthesized a t 42 "C for 60 min. The heteroduplex was then denatured at 95 "C for 5 min, and the amplification reaction was started by adding 2.5 units of Taq DNA polymerase, 20 pmol of primer 2, and buffer A in a total volume of 50 pl. After denaturation a t 95 "C for 2 min, the reaction was cycled (22 cycles) a t 95 "C for 45 s, 55 "C for 30 s, and 72 "C for 45 s. The cycles were terminated by a polymerization step a t 72 "C for 7 min. The DNA fragments of 98,104, and 107 bp correspond- ing to the different products of alternative exon 8 splicing were sepa- rated by non-denaturing gel electrophoresis on 15% (w/v) polyacrylam- ide gel in 90 mM Tris buffer, pH 8.0,90 mM borate, 2 mM EDTA. DNAwas stained with ethidium bromide.

To determine the relative proportions of amplified fragments, 35S- dATP was added to the amplification reaction medium and the bands were cut out from the gel after polyacrylamide gel electrophoresis, and the radioactivity was counted by liquid scintillation. The fragments stained with ethidium bromide were also quantitated by micro-densi- tometric scanning of the gel in an Ultroscan XL apparatus (LKE3-Phar- macia). A calibration curve was made using between 0 and 2 pg of total RNA purified from human skin fibroblasts for the reverse transcription reaction followed by PCR amplification to demonstrate linearity of the response up to 1 pg of RNA. The relative proportions of the amplified fragments of 98,104, and 107 bp were unchanged between 0.1 and 2 pg of RNA (standard deviation was less than 2%).

Genomic DNA Amplification-A 84-bp fragment of genomic DNA including exon 8 of the prosaposin gene was amplified by PCR using primers 3 (5'CCTCCGATITCCTG""I'-3', T, = 61.6 "C) and 4 (5'AT- CACCAGACGACGAAGG-3', T, = 61.4 "C). The amplification was car- ried out in a total volume of 50 pl with 0.5 pg of purified genomic DNA from cultured skin fibroblasts or lymphoblasts in the presence of 50 pmol of primers 3 and 4, 0.2 mM dNTP in 10 mM Tris buffer, pH 8.4, 50 mM KC1,2 mM MgCl,, 0.01% (w/v) gelatin. Genomic DNA was denatured at 94 "C for 6 min before adding 2.5 units of Tag DNA polymerase at 80 "C. A 84-bp fragment was amplified by cycling (35 cycles) a t 94 "C for 20 s and 56 "C for 20 s. The final polymerization step a t 72 "C was prolonged for 6 min. The amplification products were analyzed by poly- acrylamide gel electrophoresis as described above and stained with ethidium bromide.

Direct Sequencing of Amplification Products-To verify the nucle- otide sequence of amplified fragments, the bands stained with ethidium bromide were cut from the gel and the DNA was eluted overnight at 37 "C in 500 pl of 10 m~ Tris, 1 mM EDTA buffer, pH 8. The gel debris were removed by centrifugation through glass wool and DNA was pu- rified by phenol-chloroform extraction. The DNA was recovered by &ha- no1 precipitation in presence of 40 pg of glycogedpl as a carrier. Purified

DNA was resuspended in water and reamplified by PCR using the primers and the method described above. Direct sequencing of purified PCR fragments was carried out according to the method of Winship (1989).

Vrosine Fluorescence Measurements-The interaction of the pep- tides SapB-18 and SapB-21 with sphingolipids was studied by measur- ing intrinsic T y r 3 fluorescence (Konev, 1967) in the presence of increas- ing amounts of sphingolipids above critical micellar concentration which is around lo-' M in aqueous solution (Formisano, 1979). The sphingolipids were solubilized in a chloroform-methanol 2:l (v/v) mix- ture, and the solvent was evaporated under a stream of nitrogen. The sphingolipids were then resuspended in 0.2 M sodium acetate buffer, pH 4.4, containing 5 p SapB-18 or SapB-21, sonicated for 6 to 10 s, and incubated for 10 min a t 20 "C. The fluorescence spectra was measured between 280 and 400 nm under excitation at 280 nm in a Perkin-Elmer model LS 30 spectrophotometer. The peptide-sphingolipid interaction was analyzed using a Scatchard plot (Scatchard, 1949) to determine the number of sphingolipid binding sites per molecule of peptide, n, and the equilibrium dissociation constant, Kd, of the complex:

P + L * P L (Eq. 1)

r -r n

where

r = - = - [L] bound Io - IL [PI total I,,

(Eq. 2)

(Eq. 3)

I, is the fluorescence intensity of the free peptide (PI, and IL is the fluorescence intensity of the peptide in the presence of sphingolipid ligand (L).

RESULTS

Alternative Splicing of Exon 8 of Prosaposin Gene- Holtschmidt et al. (1991) characterized three different mRNAs as a result of alternative splicing of exon 8 of the prosaposin gene. The complete or partial exclusion of exon 8 is due to an alternative internal acceptor splice site in the exon sequence. Since alternative splicing is regulated in a tissue specific man- ner, we used RT-PCR to evaluate the different mRNAs pro- duced in various tissues and cell lines. Fig. 1 shows that the prosaposin mRNA of lymphoblasts and liver excluded almost completely the exon 8 as shown by the amplification of a 98-bp fragment whereas pituitary glands, brain, and cultured skin fibroblasts contained a majority of mRNA specie including the exon 8 sequence as revealed by the amplification of a 107-bp fragment. The 104-bp fragment corresponded to partial exclu- sion of exon 8 is a minor component of pituitary glands, brain (gray matter), and cultured skin fibroblasts prosaposin mRNA. Table I gives an estimation of the relative proportions of vari- ous prosaposin mRNAs in tissues and cell lines. We also tested both gray and white matters from autopsy brain tissue, the former which is rich in GMi-ganglioside and the latter in sulfatide (Sastry, 1985). Although the pattern of bands were similar, the white matter contained about twice the amount of prosaposin mRNA without the insertion of the gray matter (Table I).

To ensure that the PCR fragments were not the result of amplification artefacts, we isolated the amplified fragments of 98 and 107 bp from the gel and verified their nucleotide se- quences by a direct sequencing method (Winship, 1989). In each case, we obtained the expected sequence which confirmed that the fragments resulted from the amplification of prosapo- sin mRNA (data not shown).

The possibility exists that the 9-bp insertion detected at the mRNA level could be due to a polymorphism at the genomic DNA level instead of alternative splicing since we used tissues and cells from different individuals to detect the alternatively spliced mRNAs. To test for this possibility, we amplified with primers 3 and 4, a 84-bp fragment of prosaposin genomic DNA

20530 Human Saposin B Sphingolipid Binding

FIG. 1. Alternative splicingof exon 8 of the prosaposin gene in tissues and cell lines determined by RT-PCR The fragments of 98, 104, and 107 bp ampli- fied by RT-PCR from total RNA were separated by electrophoresis in 15% (w/v) polyacrylamide non-denaturing gel. 107- These fragments correspond to mRNA without exon 8 insertion, with a 6-bp in- 04- sertion and a 9-bp insertion, respectively. 98- Brain tissue is gray matter.

- 104

TABLE I Percentage of fragments amplified by RT-PCR of prosaposin mRNAs

isolated from various human tissues and cultured cells

Brain RT-PCR Skin frasyrt fibroblasts Gray matter gland Liver Lymphoblasts

White Pituitary

1 2 1 2 1 2 matter -

bP B B 70 8 % 107 72-c 6" 78-c 6 62 65 76 6 5 2 3 104 4 ? 1 11 k 3 9 8 9 NDb ND ND ND 98 24 -c 5 11 t4 29 27 15 94 95 98 97

n = 6 n = 4

- 89 - 80

A

Mean f SD. ND, not detected.

containing exon 8 and used it for direct sequencing from cul- tured lymphoblasts and skin fibroblasts (data not shown). No polymorphism of fragment length was observed in this region of DNA in 10 unrelated individuals. Thus, we conclude that the variable 9-bp insertion observed in prosaposin mRNA is not the result of a genomic DNA polymorphism.

Sphingolipid Binding Specificity of SapB-18 and SapB-21- Fig. 2 shows helical wheel representations of the putative am- phipathic a-helices predicted without and with the three-amino acid insertion (Potier, 1988). The SapB-18 peptide has a mean hydrophobicity per residue of 0.40, and the Gln-Asp-Gln inser- tion decreases this value to 0.10 as computed according to the method of Eisenberg et al. (1984). However, the mean hydro- phobic moment per residue which is an indicator of the asym- metric distribution of hydrophobic and hydrophilic residues around the helices was only slightly affected by the insertion from 0.28 to 0.26. The Gln-Asp-Gln insertion increased the negative charge on the hydrophilic side of the a-helix because of the Asp residue insertion on the same side as the G ~ u ~ ~ residue (Fig. 2).

The fluorescence spectra of synthetic peptides SapB-18 and SapB-21 in the presence of increasing sphingolipid concentra- tions above the CMC were determined. The sphingolipid-pep- tide interaction decreased the fluorescence intensity maximum of the T y r 3 residue of SapB-18 and SapB-21. We also observed a red-shift of fluorescence maximum of about 15 nm with in- creasing sphingolipid concentration.

Fig. 3 compares relative fluorescence intensity as a function of the molar ratio of sphingolipid versus peptide concentration in the binding assay. SapB-18 binds the G,,-ganglioside but we report that the insertion of the sequence Gln-Asp-Gln in SapB-21 abolishes G,,-ganglioside binding (Fig. 3A). However, Fig. 3B shows that the three-amino acid insertion increased the binding capacity of the peptide for the sulfatide, whereas Fig. 3C shows little change in sphingomyelin binding. Both SapB-18 and SapB-21 did not bind glucosylceramide (data not

FIG. 2. Helical wheel diagrams of the putative a-helix of sapo- sin B without (A) and with ( B ) the sequence Gln-Asp-Gln inser- tion corresponding to exon 8 coding sequence. The hydrophilic amino acid residues are indicated in gray.

shown) suggesting that the quenching of intrinsic Tyr fluores- cence intensity is due to specific binding and not to the presence of a sphingolipid in the solution.

Binding properties of SapB-18 and SapB-21 for the various sphingolipids were also analyzed by a Scatchard plot according to Equation 2. Table I1 summarizes Kd values and number of binding sites, n, for each sphingolipid studied. Sphingolipids bind to the SapB-18 and SapB-21 peptides with Kd values be- tween 7 and 17.8 VM. When binding was observed, there was a single sphingolipid-binding site per peptide molecule. As com- pared to SapB-18, the peptide with the Gln-Asp-Gln insertion lost completely its capacity to bind the G,,-ganglioside but affinities for sulfatide and sphingomyelin were increased about 4- and almost 2-fold, respectively.

DISCUSSION

Alternative splicing mechanisms increase coding capacity of the genome by multiplying biological functions that can be obtained from a single gene. There are several examples in the literature where alternative splicing results in a polymorphism at the protein level which alter the binding specificity or func- tion of expressed proteins. These include hormone precursors,

Human Saposin B Sphingolipid Binding 20531

A I

0.0 0.5 1.0 1.5 2.0

[GM-1 ganglioside] / [SapB] Molar ratio

B

0 1 2 3 4

[sulfatide] / [sap61

Molar ratio

C 1

0.0- 0 1 2 3 4 5

[sphingomyelin] I [SapB] Molar ratio

SapB-21 (0) as a function of sphingolipid concentration. The FIG. 3. Relative fluorescence intensity of SapB-18 (m) and

points represent means and standard deviations of at least three dif- ferent experiments.

DNA-binding proteins and structural proteins where post-tran- scriptional modifications allows tissue specific expression of proteins with variable functions. The nuclear transcriptional factor gene responsible for the synthesis of the CAMP-respon- sive element modulator generates four mRNAs by cell-specific alternative splicing. Three mRNAs code for antagonists of the CAMP transcriptional response (Foulkes et al. 19911, whereas the fourth mRNA, expressed in adult testis, contains an inser- tion of two glutamine-rich domains encoded by two exons re- sponsible for the transcriptional activation function (Foulkes et al. 1992). In other examples, a specific binding site is either abolished or created by alternative splicing. In the adrenodoxin reductase, a NADPH binding site is lost because alternative splicing causes the insertion of 6 amino acid residues in a Pap NADPH binding domain (Lin et al., 1990) whereas with myosin

TABLE I1 Scatchard analysis of sphingolipid-binding capacity of the synthetic

peptides SapB-18 and SapB-21

n Glycolipid K d

SapB-18 SapB-81 SapB-18 SapB-81

P M

G,,-ganglioside 7.0 2 2.1" NBb 0.82 2 0.25 NB Sulfatide 17.8k 5.6 4.62 1.1 1.15 20.36 1.21 2 0.29 Sphingomyelin 11.8 2 3.4 7.9 2 1.1 1.47 5 0.42 1.06 2 0.14

Mean 2 S.D. of at least three determinations. * NB, no detectable binding.

I of intestinal brush border, a calmodulin binding site is created by the insertion of 29 amino acid residues (Halsall and Ham- mer, 1990). For the wtl gene product, a transcription factor implicated in Wilms' tumorogenesis, alternative splicing modi- fies its binding specificity toward a different cognate nucleotide sequence. The two proteins, produced with and without a three- amino acid insertion between two zinc finger domains, have completely different physiological roles (Bickmore et al., 1992).

Although alternative splicing of the prosaposin gene by in- clusion or exclusion of a 9-bp exon was described a few years ago, its biological function was still unknown. Here, we present evidence that the insertion of the Gln-Asp-Gln sequence in a putative a-helix previously implicated in sphingolipid binding may alter the binding specificity of saposin B. We have used a synthetic peptide containing the 52 to 69 sequence of saposin B, which we have previously found to adopt an a-helical confor- mation in solution by circular dichroism spectroscopy and which has the capacity to bind G,,-ganglioside,' to show that sphingolipid binding specificity toward the G,,-ganglioside, the sulfatide and the sphingomyelin is altered by inserting the Gln-Asp-Gln sequence in the synthetic peptide. The insertion of the Gln-Asp-Gln residues coded by exon 8 occurs near the C- terminal end of the putative a-helix (Fig. 2). Alternative splic- ing of prosaposin gene is tissue-specific, and the prosaposin mRNA isolated from cultured lymphoblasts and liver tissue did not contain the exon 8 insertion, whereas skin fibroblasts, the pituitary glands, and brain contained more than 62% of mRNA with the 9-bp insertion (Table I). However, because there is no available antibody that is now able to discriminate between the saposin B with and without the Gln-Asp-Gln insertion, the re- spective protein concentrations in tissues are unknown and are only estimated from the relative quantities of the two mRNAs.

It is unlikely that the putative a-helix contained in the syn- thetic peptides SapB-18 and SapB-21 are the only structural elements implicated in sphingolipid binding and recognition by saposin B.' Other structural elements of saposin B may also contribute to sphingolipid binding or define its binding speci- ficity. Therefore, prosaposin and saposin B will not necessarily have the same sphingolipid binding specificity as SapB-18 and SapB-21 synthetic peptides. However, the a-helix of the SapB-18 peptide is certainly a major element in defining sphin- golipid-binding because Kd values determined for the isolated prosaposin and saposin B were similar to those determined with the SapB-18 and SapB-21 peptides (between 7 and 17 JIM) (Hiraiwa et al., 1992). However, these affinity constants for sphingolipid binding by prosaposin and saposin B should be taken with caution because it is not clear whether the prosa- posin and saposin B preparations used in these binding assays contained or not the Gln-Asp-Gln insertion.

Changing the sphingolipid-binding specificity of saposin B by alternative splicing may reflect a general mechanism of gene regulation at the tissue level. Most examples in the literature on the role of alternative splicing as a basic regulatory system that would affect binding specificity of proteins concern the

20532 Human Saposin B Sphingolipid Binding

nuclear transcription factors or more generally DNA-binding proteins. We provide for the first time evidence of altered lipid- binding specificity of a protein due to tissue specific alternative splicing. Considering that the sphingolipid content vary con- siderably from tissue to tissue, the prosaposin and saposin B may need to adapt to different requirements for binding prop- erties of sphingolipids. For instance, the brain contains a greater quantity and variety of most sphingolipids; the gray matter being rich in GM,-ganglioside and the white matter in sulfatide (Sastry, 1985). Thus, in this tissue there are both prosaposin and saposins B, with and without the Gln-Asp-Gln insertion, in order to offer a greater spectrum of sphingolipid- binding functions as compared to the liver and lymphoblasts where the only saposin B produced is devoid of the insertion. Relative proportion of prosaposin precursor and mature sa- posins also vary with tissue. The saposins dominate in liver whereas prosaposin is abundant in brain (Sano et al., 1989; Kondoh et al., 1993) However, it seems difficult at this stage of the investigation to directly correlate the presence of prosapo- sin and saposin B, with or without the Gln-Asp-Gln insertion in a given tissue, with the sphingolipid content of this tissue since many other proteins are involved in lipid degradation such as the other saposins but also the lysosomal enzymes themselves. Several authors have proposed that prosaposin (Hineno et al., 1991) and its rat seminal fluid homolog, the sulfated glycopro- tein 1 (Collard et al., 19881, may be involved in the transloca- tion of membrane sphingolipids. Supporting this idea, Hiraiwa et al. (1992) provided evidence that prosaposin is able to trans- locate radiolabeled GM,-gangliosides from donor liposomes to acceptor erythrocyte membranes in vitro. In this context, the lipid-binding specificity of prosaposin and saposin B may be of great physiological importance.

Mitchell, and Julie Tranchemontagne for discussion and help with some Acknowledgments-We are indebted to Damian Labuda, Grant

technical aspects of this work. We also thank Marie-Josee Champagne for advice with the fluorescence measurements and Robert Collu and

Denis Gauvreau (through Projet IMAGE) for frozen human tissues and cultured cell lines. The secretarial help of Micheline Patenaude is also gratefully appreciated.

REFERENCES

Bickmore, W. A,, Oghene, K., Little, M. H., Seawright, A., van Heyningen, V., and

Collard, M. W., Sylvester, S. R., Tsuruta, J. K., and Griswold, M. D. (1988) Bio-

Eisenberg, D., Schwartz, E., Komaromy, M., and Wall, R. (1984) J. Mol. Biol. 179,

Fischer, G., and Jatzkewitz, H. (1977) Biochim. Biophys. Acta 481, 561-572 Formisano, S., Johnson, M. L., Lee, G., Aloj, S. M., and Edelhoch, H. (1979) Bio-

Foulkes, N. S., Borrelli, E., and Sassone-Corsi, P. (1991) Cell 64, 739-749 Foulkes, N. S., Mellstrom, B., Benusiglio, E., and Sassone-Corsi, P. (1992) Nature

Halsall, D. J., and Hammer, J. A,, I11 (1990) FEBS Lett. 267, 126-130 Hineno, T., Sano,A., Kondoh, K., Ueno, S. I., Kakimoto,Y., andYoshida, K. I. (1991)

Hiraiwa, M., Soeda, S., Kishimoto, Y., and OBrien, J. S. (1992) Proc. Natl. Acad.

Holtschmidt, H., Sandhoff, K., Kwon, H. Y., Harzer, K., Nakano, T., and Suzuki, K.

Kondoh, K., Sano, A., Kakimoto, Y., Matsuda, S., and Sakanaka, M. (1993) J.

Konev, S. V. (1967) Fluorescence and Phosphorescence of Proteins and Nucleic

Li, S. C., Kihara, H., Serizawa, S., Li, Y. T., Fluharty, A. L., Mayes, J. S., and

Lin, D., Shi, Y., and Miller, W. L. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 8516-

Nakano, T., Sandhoff, K., Stumper, J., Christomanou, H., and Suzuki, K. (1989) J.

Potier, M. (1988) Biochem. Biophys. Res. Commun. 165,32-37 Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Labo-

ratory Manual, 2nd Ed., Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

Sano, A., Hineno, T., Mizuno, T., Kondoh, K., Ueno, S. I., Kakimoto, Y., and Inui, K. (1989) Biochem. Biophys. Res. Commun. 165,1191-1197

Sastry, P. S. (1985) Prog. Lipid Res. 24, 68-176 Scatchard, G. (1949) Ann. N . E: Acad. Sci. 51, 660-672 Soeda, S., Hiraiwa, M., OBrien, J. S., and Kishimoto, Y. (1993) J. Biol. Chem. 268,

18519-18523 Wenger, D. A,, and Inui, K. (1984) The Molecular Basis of Lysosomal Storage

Disorders (Barranger, J. A., and Brady, R. O., eds) Academic Press, New York Winship, P. R. (1989) Nucleic Acid Res. 17, 1266

Hastie, N. D. (1992) Science 257,235-237

chemistry 27,45574564

125-142

chemistry 18, 1119-1124

355,8044

Biochem. Biophys. Res. Commun. 176, 668474

Sci. U. S. A. 89, 11254-11258

(1991) J. Biol. Chem. 266, 7556-7560

Comp. Neurol. 334,590402

Acids, Plenum Press, New York

Shapiro, L. J. (1985) J. Biol. Chem. 260, 1867-1871

8520

Biochem. (Tokyo) 105, 152-154