Cloning and Expression of Glycolipid Transfer Protein from Bovine and Porcine Brain*

(1)

Cloning and Expression of Glycolipid Transfer Protein from Bovine and Porcine Brain*

(Received for publication, November 9, 1999, and in revised form, November 23, 1999)

Xin Lin‡, Peter Mattjus‡, Helen M. Pike‡, Anthony J. Windebank§, and Rhoderick E. Brown‡^¶储 From the‡Hormel Institute, University of Minnesota, Austin, Minnesota 55912 and the Departments of§Neurology and

¶Biochemistry and Molecular Biology, Mayo Clinic and Foundation, Rochester, Minnesota 55905

Glycolipid transfer protein (GLTP) is a small (23–24 kDa), basic protein (pI⬵9.0) that accelerates the inter- membrane transfer of various glycolipids. Here, we re- port the first cloning of cDNAs that encode the bovine and porcine GLTPs. The cDNA open reading frame for bovine GLTP was constructed by bridge-overlapping ex- tension polymerase chain reaction (PCR) after obtain- ing partial coding cDNA clones by hot start, seminested, and rapid amplification of cDNA ends-PCR. The cDNA open reading frame for porcine GLTP was constructed by reverse transcriptase-PCR. The encoded amino acid sequences in the full-length bovine and porcine cDNAs were identical, consisting of 209 amino acid residues, and were nearly the same as the published sequence determined by Edman degradation. The cDNA encoded one additional amino acid at the N terminus (methio- nine), arginine at positions 10 and 200 instead of lysine, and threonine at position 65 instead of alanine. Expres- sion of GLTP-cDNA inEscherichia coliusing pGEX-6P-1 vector resulted in glutathione S-transferase (GST)- GLTP fusion protein. Regulation of growth and induc- tion conditions led to⬃50% of expressed fusion protein being soluble and active. Proteolytic cleavage of GST- GLTP fusion protein (bound to GST-Sepharose) and af- finity purification resulted in fully active GLTP. North- ern blot analyses of bovine tissues showed a single transcript of⬃2.2 kilobases and the following hierarchy of mRNA levels: cerebrum>kidney>spleen⬵lung⬵ cerebellum > liver > heart muscle. Reverse tran- scriptase-PCR analyses of mRNA levels supported the Northern blot results.

Glycosphingolipids have polar carbohydrate head groups of varying complexity that extend into the aqueous milieu as well as nonpolar hydrocarbon chains that embed into the lipid bi- layer of biomembranes. These lipids are present in all eu- karyotes and are localized in cellular membranes, lipoproteins, and other lipid-rich structures. Several lines of evidence impli- cate glycosphingolipids and related metabolites in important

cellular processes such as differentiation, adhesion, proliferation, and cell-cell recognition (1– 4). More recently, the possi- bility of glycosphingolipid-enriched clusters,i.e.“rafts” existing within membranes has generated much interest because of their co-localization with signaling kinases and their putative role in the localization and sorting of proteins carrying glyco- sylphosphatidylinositol anchors and other lipid modifications (5–7). Because of these roles in key cellular processes, it is clear that the trafficking and expression of glycolipids from their site of synthesis in the endoplasmic reticulum-Golgi complex must be effectively coordinated and controlled. Proteins that play roles in these processes warrant further investigation.

Soluble proteins with the ability to selectively accelerate the intermembrane transfer of glycosphingolipids were initially discovered in the membrane-free cytosolic extract of bovine spleen (8, 9). Subsequently, similar activities were found in a wide variety of tissues, including mammalian brain, liver, and kidney as well as spinach chloroplasts (10 –12). Purification of the spleen and brain glycolipid transfer proteins (GLTPs)¹by various means revealed single polypeptides with molecular masses of 22–24 kDa, basic isoelectric points, and an absolute specificity for glycolipids (e.g.see Refs. 9, 13, and 14). GLTPs appear to be cytosolic, and their specificity is directed to glycolipids with a␤-glucosyl or␤-galactosyl sugar linked to either a ceramide or diglyceride hydrophobic backbone (15–17). Edman degradation of the porcine brain protein and other studies revealed 208 amino acid residues with one disulfide bridge (18 –20). Together, these characteristics show GLTP to be distinctly different from other soluble proteins that can interact with glycolipids such as lysosomal sphingolipid activator proteins, other types of lipid transfer proteins, and lectins with mannose recognition/binding domains (10, 14).

Although GLTP behavior has been investigatedin vitro(10 – 12), many aspects of GLTP’s basic structure-function relation- ships as well as its mechanism of action are not well under- stood, and little is known about its in vivo function and localization. To begin to address such issues, here we report the first successful cloning and overexpression of the full-length cDNA of GLTP from bovine and porcine brain as well as the mRNA tissue expression pattern of GLTP.

EXPERIMENTAL PROCEDURES

Materials—Oligonucleotides were synthesized using an Oligo 1000 DNA synthesizer (Beckman, Fullerton, CA) or by BioSynthesis (Lewis- ville, TX). The bovine brain (catalog no. BL1027b, adult male) and the porcine cortex␭gt11 cDNA libraries (catalog no. BL1003b, adult male)

* This work was supported by the Academy of Finland, the Åbo Akademi, Magnus Ehrnrooth, and Hormel Foundations, and United States Public Health Service Grant RO1-GM45928. Portions of this work were presented at the 39th American Society for Cell Biology Annual Meeting, Washington, D.C., December 11–15, 1999 (48). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “adver- tisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank^TM/EBI Data Bank with accession number(s) AF209701 and AF209702.

储To whom correspondence should be addressed: The Hormel Insti- tute, University of Minnesota, 801 16th Ave. N.E., Austin, MN 55912.

Fax: 507-437-9606; E-mail: reb@maroon.tc.umn.edu.

1The abbreviations used are: GLTP, glycolipid transfer protein; PCR, polymerase chain reaction; GST, glutathione S-transferase; GalCer, galactosylceramide; GlcCer, glucosylceramide; RACE, rapid amplification of cDNA ends; bp, base pair(s); RT, reverse transcriptase; nsLTP, nonspecific lipid transfer protein; SAP, sphingolipid activator proteins;

GM1, II³-␣-N-acetylneuraminosylgangliotetraosylceramide; GM2, II³-

␣-N-acetylneuraminosylgangliotetriosylceramide.

This paper is available on line at http://www.jbc.org

5104

(2)

were both obtained from CLONTECH (Palo Alto, CA). PCR amplifica- tions were performed with the Expand High Fidelity PCR System (Roche Molecular Biochemicals) on a GeneAmp 2400 thermocycler (Per- kin-Elmer). PCR or RACE products were cloned into pGEM-T Easy (TA) vector (Promega, Madison, WI) and then sequenced using T7 or SP6 promoter primers. DNA sequencing was performed at the Mayo Molec- ular Biology Core Facility with an Applied Biosystems 377 sequencer using thermocycler protocols and fluorescent dye terminators.

GLTP Amino Acid Sequencing—Protein sequencing was performed at the Mayo Protein Core Facility. Briefly, limited proteolysis of native bovine GLTP was achieved by treatment with cyanogen bromide or by trypsinization. Prior to cyanogen bromide treatment, GLTP was re- duced and alkylated. Alternatively, trypsinization was carried out byin situdigestion of gel slices as described by Rosenfeldet al.(21) following SDS-polyacrylamide gel electrophoresis of GLTP. The resulting peptides were separated using an ABI Separation System (Perkin-Elmer) and a Vydac C18 column (2.1 ⫻ 250 mm; The Separations Group, Hesperia, CA). Separation was achieved using a linear gradient of 5%

buffer B to 90% buffer B in 70 min, where buffer B was 60% acetonitrile, 40% water, and 0.9% trifluoroacetic acid, and buffer A was 0.1% trifluoroacetic acid and 5% acetonitrile. The peptide peaks were spotted onto pieces of high density polyvinylidene difluoride (ABI ProBlott) and treated with Biobrene/methanol (1:1; 5␮l) prior to sequencing on an ABI 492 Procise Protein Sequencing System using the pulsed liquid mode. Data were analyzed with the ABI model 610A data analysis software.

Isolation of cDNA Partially Encoding GLTP—A cDNA fragment (379 bp) coding for a GLTP interior region was initially amplified from a porcine brain␭gt11 library by PCR using degenerate oligonucleotides (sense dpF and antisense dpR; Table I). The primers were designed using Oligo 4 from the published interior GLTP amino acid sequences EKEMYG⁸²and MYTKMN²⁰²depicted in Fig. 1. Sequencing of the 379-bp cDNA fragment, after cloning to TA vector, led to the design of gene-specific forward and reverse primers (gspF1 and gspR1; Table I and Fig. 1), which were used to amplify a 345-bp cDNA fragment from a bovine brain␭gt11 cDNA library. The PCR temperature profile program consisted of one cycle of 5 min at 94 °C; 35 cycles of 1 min at 94 °C, 36 s at 63 °C, 90 s at 72 °C; and final extension for 7 min at 72 °C.

Sequencing of the 345-bp bovine cDNA fragment confirmed that it encoded an interior region of GLTP.

To obtain the 5⬘-end encoding the N termini of bovine and porcine GLTPs, cDNA was synthesized from total bovine or porcine brain RNA (1 ␮g) using a first strand synthesis kit (CLONTECH) and a gene- specific antisense primer (gspR2), designed from the known partial nucleotide sequence (Table I). The reaction was performed using a SMART RACE PCR kit (CLONTECH). The 5⬘-end of the cDNA was amplified by PCR using the universal primer mix (5⬘-RACE-UPM;

Table I) along with the antisense gspR2 oligonucleotide using the following temperature profile program: one cycle of 5 min at 95 °C; 35 cycles of 1 min at 94 °C, 36 s at 63 °C, 90 s at 72 °C; and final extension for 7 min at 72 °C. The resulting single 433-bp DNA band was cloned into TA vector and sequenced.

To obtain the 3⬘-end of the bovine GLTP cDNA, a first round PCR was performed using a ␭gt11 reverse universal primer and a gene- specific primer (gspF1) with bovine cDNA-␭gt11 phage library as template (Table I). For the second round of PCR, 5␮l of undiluted product was then used in a reaction mixture in which nested primer (gspF2) replaced primer gspF1 (Table I). The DNA was then denatured for 5 min at 95 °C followed by 20 cycles of 36 s at 55 °C, 90 s at 72 °C, 1 min at 94 °C and by final extension at 72 °C for 7 min.

Aliquots (10␮l) of the second round PCR mixture were analyzed by electrophoresis on a TAE (40 mMTris acetate, 1 mMEDTA, pH 8.0), 1.5% agarose gel and visualized by ethidium bromide staining to iden- tify the PCR products. Three PCR bands were excised and purified by QIA quick columns (Qiagen, Inc., Chatsworth, CA) and cloned into TA vector prior to sequencing.

Cloning of Full-length cDNA Encoding Bovine Brain GLTP—The full-length bovine GLTP cDNA was constructed by the bridge-overlapping extension PCR method. The PCR mixture (48␮l) contained 20 ng of the 5⬘-end of bovine GLTP cDNA fragment (433 bp) and of the 3⬘-end of bovine GLTP cDNA fragment (428 bp) along with 0.2 mMdNTP, 5%

Me2SO, 1.5 mMMgCl2, and 2.5 units of Advantage polymerase 2 in Advantage DNA polymerase 2 buffer (50 mMKCl, 10 mMTris/HCl, pH 9.0, at 25 °C, 0.1% (w/v) gelatin, and 1% Triton X-100). PCR was carried out using the following temperature profile program: 5 min at 94 °C; 10 cycles of 1 min at 94 °C, 36 s at 60 °C, and 90 s at 72 °C; the addition of primers gspFN and gspRC (1␮l; 20 pMeach) (see Table I), followed by another 30 cycles of 36 s at 57 °C, 90 s at 72 °C, 1 min at 94 °C, and final

elongation for 7 min at 72 °C. After digestion byEcoRI andBamHI, a 627-bp PCR product was cloned into glutathioneS-transferase (GST) fusion protein expression vector pGEX-6P-1 (Amersham Pharmacia Biotech), which also had been digested withEcoRI andBamHI.

Cloning of Full-length cDNA Encoding Porcine Brain GLTP—First- strand cDNA was synthesized (10-␮l reaction) from total pig brain RNA (1␮g) using Superscript^TMII reverse transcriptase (Life Technologies, Inc.) with antisense gene-specific primer gspRC (Table I) according to the SMART RACE PCR kit protocol. A 5-␮l sample aliquot of the reverse transcriptase reaction mixture was used as template for PCR, and the same gene-specific primers (gspFN and gspRC) were used to amplify the porcine full-length GLTP cDNA (627 bp) under the same reaction conditions as for bovine GLTP cDNA.

GLTP Heterologous Expression as GST-GLTP Fusion Protein—The 627-bp cDNA fragment coding for the 209 amino acids was cloned in frame with GST in the pGEX-6P-1 vector at theBamHI andEcoRI restriction sites. Following transformation, cells (Escherichia coliBL21) were grown in 2⫻YT medium at 30 °C until cell density reached anA600

of 1.7. Then expression of GST-GLTP fusion protein was induced with isopropyl-1-thio-␤-^D-galactopyranoside (final concentration 0.1 mM), and cell incubation continued for an additional 2 h. After harvesting, cells were lysed by sonication after treatment with lysozyme. The resulting supernatant, which contained about 50% of the GST-GLTP fusion protein, was incubated with the GST affinity matrix and washed.

GLTP then was released by cleaving with PreScission protease according to the manufacturer’s specifications.

SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting—Pro- teins were analyzed on 16% polyacrylamide gels (0.75 mm) containing 0.1% SDS and transferred to a polyvinylidene difluoride membrane for Western blotting (Millipore Corp., Bedford, MA). To detect GLTP, a rabbit polyclonal antibody was raised against a 12-amino acid peptide (¹⁹⁵YEMYTKMNAELNYKV²⁰⁹; see Fig. 2) that is identical to the C- terminal sequence of bovine and porcine GLTP (BioSynthesis). Prior to use, the serum was treated using a Protein A purification kit (Pierce), and the resulting IgG fraction was immunosorbed against GLTP im- mobilized on polyvinylidene difluoride membranes (22). GST-GLTP fusion protein also was detected with the preceding antibody or with a goat anti-GST antibody (Amersham Pharmacia Biotech). Visualization was achieved with secondary antibodies conjugated to alkaline phos- phatase (Sigma and Bio-Rad).

GLTP Transfer Activity Assays—Two different assays were used to monitor the GLTP-mediated glycolipid transfer between membranes. A fluorescence resonance energy transfer assay utilizing anthrylvinyl- labeled glycolipid (1 mol %) and nontransferable perylenoyl-labeled triglyceride (1.5 mol %) permitted continuous real time monitoring of GLTP activity (23, 24). Fluorescence measurements were carried out using a SPEX Fluoromax spectrofluorimeter (Instruments S.A., Inc.

Edina, NJ). Also used for determination of GLTP activity was a [³H]Gal- Cer radioactive assay in which donor vesicles contained negatively charged phospholipid allowing separation from neutral acceptor vesicles by rapid elution over DEAE-Sephacel minicolumns (14, 24).

Tissue and Cell Preparations and Total RNA Isolation—All tissue samples were obtained from healthy animals within minutes after slaughter and were immediately frozen in liquid nitrogen. Total RNA from bovine tissues was isolated by an RNA STAT-60 acidic guani- dinium isothiocyanate/chloroform procedure as described briefly below (Tel-Test Inc., Friendswood, TX). Tissue samples (50 –100 mg) that had been kept frozen in liquid nitrogen were homogenized directly in 1 ml of the RNA STAT-60 reagent. After the addition of chloroform (200␮l) and shaking for 15 s, the sample was centrifuged at 1200 rpm for 15 min at 4 °C. The aqueous phase was collected in a tube containing 2-propanol (500␮l), was incubated at room temperature for 10 min, and then was centrifuged at 12,000 rpm for 10 min at 4 °C. The supernatant was discarded, and the pellet was washed in 70% ethanol, dried, and redis- solved in RNase-free water. Prior to use, RNA concentrations were determined spectrophotometrically, and RNA integrity was verified by electrophoresis.

RT-PCR Analysis—GLTP expression levels in the different tissues were initially analyzed by RT-PCR using total RNA extracted from different tissues by the RNA STAT-60 procedure described above. The RT reaction was performed according to protocol (CLONTECH) but with 2␮g total RNA in each transcription reaction. After cDNA synthesis using gene-specific primer gspRC and a Superscript II RT^TMkit (Life Technologies, Inc.), cDNA was amplified by PCR using gene- specific primers (gspFN and gspRC) under the same conditions as described for the production of the full-length GLTP cDNA from bovine brain. Aliquots of 10␮l of the RT-PCR reaction mixture were analyzed by electrophoresis on a TAE (40 mMTris acetate, 1 mMEDTA, pH 8.0)

(3)

1.5% agarose gel and visualized by ethidium bromide staining to iden- tify the RT-PCR products.

Northern Blot Analysis—Total RNA (20␮g) was electrophoresed on 1% agarose gels containing 2.2Mformaldehyde and transblotted onto nylon membranes (Roche Molecular Biochemicals) using the TurboBlot system (Schleicher & Schuell). Hybridization was carried out at 68 °C for 24 h in 50% formamide hybridization solution containing 20 ng/ml RNA probe. The mRNA hybridization probe for GLTP consisted of antisense RNA labeled with DIG-CTP using T7 RNA polymerase (Roche Molecular Biochemicals) along with a 345-bp cDNA partially encoding GLTP, inserted into TA cloning vector but linearized using SalI. Following hydridization, the membrane was washed twice at room temperature with 2⫻SSC buffer (30 mMsodium citrate, pH 7.0, 300 mM

NaCl) containing 0.1% SDS and twice at 68 °C with 0.5⫻SSC buffer containing 0.1% SDS. For detection, anti-DIG-AP antibody was used in conjunction with CDP-star chemiluminescent reagent (Roche Molecular Biochemicals).

RESULTS

Cloning of cDNA Encoding GLTP—Because the primary amino acid sequence of porcine brain GLTP had been determined earlier by Edman degradation (18), we initially set out to establish the similarity to a GLTP purified to homogeneity from bovine brain (14). This was necessary because Ledeen and co-workers (25) had reported two distinct glycolipid transfer protein activities in bovine brain. Because the purified bovine and porcine GLTPs appeared to be N-terminally blocked (14, 18), we fragmented the beef brain protein with cyanogen bromide or by in situ trypsinization of SDS gel slices, purified three of the resulting peptides by reverse-phase HPLC, and sequenced these fragments. The three internal peptides matched identically with amino acid sequences 57– 65, 80 –91, and 197–208 (carboxyl terminus) of porcine brain GLTP except for a single amino acid substitution in residues 57– 65 as well as in residues 197–208.

As a result of the high degree of similarity, we relied on the porcine brain GLTP primary amino acid sequence to design and produce degenerate oligonucleotide primers (sense primer DF and antisense primer DR; Table I) for the initial amplification of cDNA encoding GLTP from a␭gt11 cDNA library. The subsequent PCR experiments resulted in the successful amplification a 379-bp cDNA fragment encoding an interior region of GLTP. From the sequence of this 379-bp cDNA fragment, we then designed gene-specific oligonucleotide primers (gspF1 and gspR2; Table I and Fig. 1) to use in the amplification of the 3⬘- and 5⬘-cDNA ends encoding the C and N termini of GLTP. Hot start, seminested PCR approaches were used to obtain the 3⬘ region and resulted in a 428-bp fragment encoding the GLTP C terminus. Downstream and flanking the TAG stop codon was a

33-bp untranslated region. However, the canonical AATAAA polyadenylation site or poly(A)^⫹ tail was absent, suggesting that part of the 3⬘-end of the mRNA was missing in the original bovine␭gt11 cDNA library.

Attempts to obtain the cDNA encoding the N terminus of GLTP from the␭gt11 library were carried out by PCR using the

␭ forward universal and gspR2 primers (Table I). A 406-bp fragment resulted that lacked the ATG start codon and encoded an N terminus of PVAEH instead of the expected ALLAEH (based on the pig sequence). To verify this difference, RACE- PCR was employed (see “Experimental Procedures”). However, RACE-PCR yielded a 433-bp cDNA fragment that not only coded the N terminus predicted by the pig primary sequence but also included the ATG start codon along with an upstream G⫹C-rich region consisting of 10 base pairs. The cDNA fragments encoding both the N and C termini of GLTP were then used to construct a full-length GLTP cDNA by bridge-overlapping extension PCR. The resulting 670-bp sequence contained a single open reading frame (627 bp) coding for the amino acid residues of GLTP (Fig. 1).

Identical Amino Acid Sequences of Bovine and Porcine GLTP—Application of a similar PCR-based strategy yielded a GLTP cDNA clone from porcine brain with a 627-bp open reading frame encoding a 209-amino acid protein with a calcu- lated mass of 23.9 kDa. Although several nucleotide bases were different (Fig. 1,boxed nucleotides), the encoded amino acids in the bovine and porcine cDNAs were identical. Both cDNAs, however, encoded for one additional amino acid at the N terminus (methionine) as well as for arginine at positions 10 and 200 instead of the lysines originally reported at positions 9 and 199 based on Edman degradation (18). We also found that threonine was encoded at position 65 instead of the alanine originally reported at position 64. These differences, illustrated in Fig. 2 (vertical arrows), were verified in several clones generated by different PCRs.

A hydropathy plot analysis (26) revealed hydrophobic segments near amino acids 40, 100, and 150 as well as near the C terminus (Fig. 3). The hydrophobicity of these regions suggests that certain of these segments could be potential membrane interaction sites. Analysis by the PSORT II program (27) did not reveal any signal peptide sequences but predicted the protein to be cytoplasmic, which agrees with previous suggestions (10, 11).

Heterologous Expression of GLTP as a GST Fusion Pro- tein—In earlier studies, Saupeet al.(28) showed that HET-C2 protein, a fungal protein showing similarity to GLTP, could be TABLE I

Primer nucleotide sequences and corresponding GLTP amino acid sequences

Primer designation^a Primer sequence GLTP amino acid sequence

Degenerate primers

dpF (sense) 5⬘-GAGAAGGAGATGTACGG-3⬘ ⁷⁷^EKEMYG⁸²

dpR (antisense) 5⬘-TTCATCTTGGTGTACAT-3⬘ ¹⁹⁷^MYTKMN²⁰²

Gene-specific Primers

gspF1 5⬘-GCCGAGTGGCCCAAAGTGGGGGCC-3⬘ ⁸³^AEWLKVGA⁹⁰

gspF2 (nested) 5⬘-CTGATGTGGCTGAAAAGAGG-3⬘ ⁹⁴^LMWLKR⁹⁹

gspR1 5⬘-CATCTCATAGATGACATCGATGGT-3⬘ ¹⁹⁰^TIDVIYEM¹⁹⁷

gspR2 5⬘-GACGCGGATGAGATTGGGGTGGTT-3⬘ ¹¹⁹^NHPNLIRV¹²⁶

Vector Primers

3⬘-␭gt11 (Reverse universal primer) 5⬘-TTGACACCAGACCAACTGGTAATG-3⬘ 5⬘-␭gt11 (Forward universal

primer)

5⬘-GGTGGCGACGACTCCTGGAGCCCG-3⬘

5⬘-RACE primer 5⬘-CTAATACGACTCACTATAGGGCAAGCAGTGGTAACAACGCAGAGT-3⬘ Universal primer mix 5⬘-CTAATACGACTCACTATAGGGC-3⬘

Expression vector primers

gspFN 5⬘-GGGGGATCCATGGCGCTGCTGGCGGAGC-3⬘ ¹^MALLAE⁶

gspRC 5⬘-GGGGAATTCTACACCTTGTAGTTGAGCTC-3⬘ ²⁰⁴^ELNYKV²⁰⁹

aPrimers are designated as follows: dp, degenerate primer; F, forward; R, reverse; gsp, gene-specific primer. The underlined nucleotides indicate the restriction enzyme site.

(4)

successfully expressed as a GST fusion protein inE. coliusing the pGEX expression vector. For this reason, we utilized the pGEX expression vector. The 627-bp cDNA open reading frame for GLTP was cloned in-frame with GST into the pGEX-6P-1 vector at theBamHI andEcoRI restriction sites. The primers used in the PCR were gspFN and gspRC with adjacent restriction sites for BamHI and EcoRI, respectively (Table I). The resulting pGEX-GLTP was used to transform competent BL21 cells. The extent to which the fusion protein accumulated in inclusion bodies was critically dependent upon growth and induction conditions. Cells grown at 37 °C to modest densities (OD⬵0.7) and induced for 6 h with isopropyl-1-thio-␤-D-galactopyranoside produced GST-GLTP fusion protein that almost all pelleted following lysis and centrifugation, presumably due

to localization to inclusion bodies. After the pelleted fusion protein was solubilized using urea (8^M), attempts were made to refold into active form. However, refolding by the method of Chen et al. (29) failed to restore GLTP activity (data not shown). As a result, cell growth conditions were adjusted to reduce inclusion body formation. This was achieved by growing the cells at only 30 °C, but to higher densities (OD⬵1.7), and then inducing with isopropyl-1-thio-␤-D-galactopyranoside for only 2 h. These conditions resulted in GST-GLTP fusion protein, which was about equally distributed between the soluble and pelleted fractions after lysis and centrifugation. Analysis of the soluble fraction by SDS-polyacrylamide gel electrophoresis revealed overexpression of the 53-kDa GST-GLTP fusion protein (Fig. 4a,lane 1). Immunoblotting showed that the fusion protein reacts with immunosorbed, protein A-purified polyclonal antibody raised to the C-terminal region of GLTP (Fig.

4b,lane 1). Interestingly, the GST-GLTP fusion protein accel- erated the intervesicular transfer of both fluorescently labeled GalCer (Fig. 5, right panel, trace c) and [³H]GalCer (not shown). No transfer activity was observed in mock-transfected cells that contained pGEX vector without the GLTP coding insert (Fig. 5, right panel,trace d). However, the addition of soluble sonicate containing the GST-GLTP fusion protein (Fig.

5,trace d,arrow) caused an immediate burst in transfer activ- ity. Other control experiments involving [³H]GalCer revealed FIG. 1.Nucleotide sequences for bovine and porcine glycolipid

transfer protein and the deduced amino acid sequences.The single letter code amino acid sequence is numbered on theleft, and the nucleotide sequence is numbered on theright. The termination codon of the bovine glycolipid transfer protein occurs at nucleotide 627, which corresponds to amino acid 209. Thedotsin the porcine cDNA sequence indicate identical nucleotides, and the differences are shown. The positions of the two degenerate primers, dpF (sense) and dpR (antisense) areunderlinedin the single letter amino acid sequence, and theunder- lined cDNA sequences indicate the positions of the gene-specific primers.

FIG. 2.Alignment of the GLTP amino acid sequence and the fungal HET-C2 protein.Protein sequences were deduced from their cDNAs. The SIM alignment algorithm was used to align the amino acid sequences for maximum identity (47). Individual amino acids are shown in single letter code and numbered on theright.Black boxes, sequence identities between the GLTP and HET-C2;dots, conservative differ- ences; hyphens, gaps. The four differences in GLTP amino acid se- quence obtained from the cDNAsversusfrom Edman degradation by Abe (18) are marked withvertical arrows.

FIG. 3.Hydropathy analysis. A hydropathy plot of the deduced amino acid sequence for GLTP was obtained using the Kyte and Doolittle analysis with an average window size of 19 amino acid residues plotted at one-residue intervals (26). Positive values represent increased hydrophobicity.

(5)

that the transferred radioactivity co-migrated with the GalCer standard upon analysis by thin layer chromatography (data not shown).

Affinity purification of GLTP was achieved by adsorbing the soluble GST-GLTP fusion protein to glutathione-Sepharose-4B, washing, and then proteolytically cleaving to release the bound GLTP. Fig. 4 shows the resulting GLTP purity as assessed by SDS-polyacrylamide gel electrophoresis (Fig. 4a,lane 2) and by immunoblotting against immunosorbed, protein A-purified polyclonal antibody to the C-terminal region of GLTP (Fig. 4b, lane 2). By these criteria, the recombinant GLTP appeared pure and behaved identically with native GLTP purified from bovine brain (Fig. 4,aandb,lane 3) as described by Brownet al. (14). The purification yielded ⬃0.5 mg of recombinant GLTP/liter of cultured cells within 3– 4 working days. In contrast, purification of similar amounts of native GLTP from half a dozen bovine brains requires about 6 – 8 weeks (14). The activity of the purified recombinant GLTP was assessed by fluorescently labeled GalCer. As shown in Fig. 5 (left panel), the fluorescence resonance energy transfer assay reveals that the initial rates of anthrylvinyl-GalCer transfer by recombinant GLTP (trace a) are as good or better than those of native GLTP (trace b) that had been stored for several months at⫺70 °C.

These results were confirmed by intervesicular transfer assays utilizing [³H]GalCer (data not shown).

Tissue Expression Levels of GLTP mRNA—Because RT-PCR has proven to be a fast, sensitive, and semiquantitative method for analyzing gene expression, we initially used RT-PCR to analyze the wild-type tissue expression of the 627-bp mRNA open reading frame encoding GLTP. This was accomplished using the gspFN forward and gspRC reverse primers as described under “Experimental Procedures.” The resulting agarose gel (Fig. 6a) shows the following hierarchy of 627-bp mRNA levels: cerebrum ⬎kidney⬎spleen ⬵cerebellum⬵lung ⬎ liver⬎heart muscle.

To confirm the RT-PCR results, mRNA levels also were determined by Northern blot hybridization. Antisense RNA to a 345-bp cDNA fragment partially encoding GLTP served as the hybridization probe (Table I). Northern blot analyses of bovine tissues (Fig. 6b) showed single transcripts of ⬃2.2 kilobases and a similar expression level hierarchy, except that lung, cerebellum, and spleen mRNA levels appeared to be similar.

The relatively high levels of GLTP mRNA in cerebrum and kidney are consistent with the higher levels of glycosphingolipids that are generally found in these tissues.

DISCUSSION

We have cloned the bovine and porcine brain cDNAs for GLTP and shown that they contain 627-bp open reading frames encoding identical amino acid sequences. This represents the first reported cloning of GLTP. The amino acid sequence encoded by the cDNAs is identical to the sequence previously determined by Edman degradation (18) with the exception of four amino acids. The cDNAs encode one additional amino acid (Met) at the N terminus, arginines at positions 10 and 200 (instead of lysines), and threonine at position 65 (instead of alanine). The absolute conservation of the amino acid sequence encoded by both bovine and porcine cDNA suggests an important role in cellular glycosphingolipid trafficking and/or metabolism. Yet searches of on-line protein and DNA data bases show no similarity between GLTP and other proteins previously reported to have glycolipid intermembrane transfer activity and/or to participate in glycolipid metabolism. Such proteins include two classes of soluble proteins: nonspecific lipid transfer proteins (e.g.Refs. 30 and 31) and sphingolipid activator proteins (e.g.Refs. 32 and 33).

Nonspecific lipid transfer proteins (nsLTPs) are present in both animals and plants and represent a group of soluble proteins that catalyze thein vitrotransfer of a wide range of lipids, including glycolipids. Although able to transfer different neutral glycosphingolipids and ganglioside GM1 between membranes (34), nsLTPs clearly differ from GLTP with regard to lipid selectivity and actively transfer all common phosphoglyc- erides as well as cholesterol. Also, despite having basic isoelectric points (pI 8.6 –9.6) like GLTP (pI 9.0), nsLTPs are much smaller (13.2 kDa) than GLTP (23.9 kDa) and are initially encoded in precursor form (15.3 kDa; pre-nsLTP). A tripeptide targeting sequence (Ala-Lys-Leu) in the C-terminal end directs certain nsLTPs to peroxisomes, where a 20-amino acid presequence is cleaved. No known organelle targeting sequences appear in the primary sequence of GLTP, which is consistent with its putative cytoplasmic localization. Given the differences in selectivity and structure along with the lack of sequence homology revealed by searches of on-line data bases, available evidence indicates that GLTP is distinctly different from known nsLTPs and other lipid transfer proteins.

Another class of small soluble proteins with the ability to interact specifically with glycolipids are the sphingolipid activator proteins (SAPs), which serve as nonenzymatic cofactors in the degradation of glycosphingolipids by the acidic glycosi- dases (35). Although one SAP,i.e.GM2 activator protein, has been shown to displayin vitroglycolipid transfer activity (36), this SAP is clearly selective for ganglioside GM2 in contrast to the broader selectivity displayed by GLTP for many glycolipids (13, 15–17). Also, the GM2 activator protein appears to differ from GLTP in that it is synthesized as a prepropolypeptide that includes a presequence directing synthesis on the rough endoplasmic reticulum followed by extrusion of the nascent polypep- tide into the lumen for trafficking to the lysosome. Other SAPs (four of five) are encoded by a single gene and enter the lysosome as a single 65–73-kDa precursor chain that is processed into four similarly sized (⬃13 kDa), fairly homologous polypeptides (SAPs A–D) that are heavily glycosylated and have acidic pI values (4.6). Prosaposin and all of the saposins bind gangliosides, sulfatides, and cerebrosides to varying degrees and fa- cilitate glycolipid insertion into erythrocyte ghosts or brain microsomes to differing extents (37, 38). Also, the sulfatide activator protein (SAP A) markedly accelerates the transfer of complex gangliosides compared with simpler glycolipids having shorter carbohydrate chains (39). Yet, subtle differences in glycolipid selectivity along with other structural differences and the lack of sequence homology revealed by searches of FIG. 4.SDS-polyacrylamide gel electrophoresis and immuno-

blotting of GLTP.PanelA, total protein staining after electrophoresis using Coomassie Brilliant Blue.Lane 1, sonicatedE. colicells expressing GST-GLTP fusion protein; lane 2, affinity-purified recombinant GLTP obtained after protease cleavage of the fusion protein;lane 3, native GLTP purified from bovine brain. B, immunostaining after transblotting to a polyvinylidene difluoride membrane and detecting with anti-GLTP antibody.Lanes 1–3, same as inA.

(6)

on-line data bases strongly suggest that GLTP also is distinctly different from all known SAPs.

In contrast, it is clear from the on-line multiple sequence analysis routines of the EBML and Poˆle Bio-Informatique data bases that GLTP is homologous to the fungal protein, HET-C2 (Fig. 2). This protein, encoded at thehetloci, plays a decisive role in regulating the compatibility and stability of heterokaryons that form by hyphal fusion between different individual organisms (28). The resulting vegetative heterokaryons contain genetically distinct nuclei within a common cytoplasm. If these nuclear components contain different specificities at any of several hetloci, incompatibility interactions are triggered, resulting in growth cessation and cell destruction by a lytic process. The genetic basis of this vegetative incompatibility has been studied in various ascomycetes including Neurospora, Aspergillus, andPodospora (40, 41). The cloned het-c2allele encodes a 208-amino acid protein with significant homology to GLTP (29% identity and an additional 30% similarity). In the case of the Poˆle Bio-Informatique on-line data base, in addition to the 29% identity, strong similarity was noted for 20% of sequence, whereas weak similarity was noted for 10% of the sequence. The similar amino acids are distributed all along the sequence (Fig. 2), which agrees with earlier findings (28). Pre- liminary experiments indicate that cloned HET-C2 does dis-

playin vitroglycolipid transfer activity.²This finding is espe- cially interesting, because earlier preliminary investigations of Saccharomyces cerevisiae, a unicellular yeast incapable of a mycelial growth stage, revealed no trace of GLTP activity (42).

Also, our searches of the Saccharomyces Genome Data base provided no evidence of open reading frames encoding GLTP or homologous proteins.

Together, these results suggest that GLTP might play an important role in cytoplasmic processes triggered by cell-cell interactions and related to cell-cell compatibility such as the control of cell apoptosisversusproliferation. Such a role could be one of sensing or helping to modulate glucosylceramide (GlcCer) levels at key locations within cytoplasmically oriented compartments. The importance of GlcCer levels within cells is well documented. Depletion of endogenous GlcCer by metabolic inhibitors arrests the cell cycle (44). Elevations of GlcCer lead to tissue hyperplasia in mice including hepatocyte proliferation and epidermal mitogenesis (45, 46). The molecular biological approaches described here are likely to help provide the means not only to gain insights into thein vivorole of GLTP but also to accelerate an understanding of basic structure-function re- lationships that govern this protein’s ability to interact with glycolipids in membrane surfaces.

Acknowledgments—We thank Dr. Zigang Dong for use of the oligo- nucleotide synthesizer and for helping with the initial degenerate primer syntheses, Dr. Dan McCormick for performing the partial amino acid sequencing, Dr. Kangjian Wu and Gary Folkert for assistance in designing PCR primers and oligonucleotide probes, the Mayo Sequenc- ing Facility for DNA sequencing, Dr. Jagi Gill and Pia Roos for assistance with the DNA sequencing, Dr. Beatrice Turcq for providing pGEX vector with thehet-c2insert, Dr. David Hoggsett for advice regarding GLTP heterologous expression, Dr. Wei-Ya Ma for initial help with cell transformations, and Dr. J. G. Molotkovsky for synthesizing the fluorescent glycolipids.

REFERENCES

1. Hakomori, S., and Igarashi, Y. (1995)J. Biochem.(Tokyo)118,1091–1103 2. Riboni, L., Viani, P., Bassi, R., Prinettti, A., and Tettamanti, G. (1997)Prog.

Lipid Res.36,153–195

3. Okazaki, T., Kondo, T., Kitano, T., and Tashima, M. (1998)Cell. Signal.10, 685– 692

4. Kolesnick, R. N., and Kro¨nke, M. (1998)Annu. Rev. Physiol.60,643– 665 5. Simons, K., and Ikonen, E. (1997)Nature387,569 –572

6. Brown, R. E. (1998)J. Cell Sci.111,1–9

7. Brown, D. A., and London, E. (1998)Annu. Rev. Cell Dev. Biol.14,111–136

2P. Mattjus, H. M. Pike, E. Espagne, B. Turcq, and R. E. Brown, unpublished observation.

FIG. 5.Transfer activity of the recombinant GLTP.Glycolipid transfer activity was measured using the fluorescence resonance energy transfer approach described under “Experimental Procedures.”Ashows the activity of the purified cloned and native GLTPs (0.3␮g/ml).Trace a depicts the cloned GLTP following affinity purification after expression as a GST-GLTP fusion protein inE. coli;trace bdepicts the native GLTP purified from bovine brain.Bshows the GLTP activity inE. colisonicates transformed with pGEX vector either with or without the cDNA insert encoding GLTP (3␮g/ml).Trace cdepicts the sonicate ofE. coliexpressing GST-GLTP fusion protein;Trace ddepicts the sonicate ofE. coli expressing GST alone. Thearrowindicates the addition of cloned GLTP to verify the fidelity of the assay.

FIG. 6.GLTP RNA expression levels in bovine tissues.A, RT- PCR analysis. The amplified products were electrophoresed using an agarose gel and stained with ethidium bromide as described under

“Experimental Procedures.”Lane A, cerebrum;lane B, kidney;lane C, spleen;lane D, liver;lane E, heart muscle;lane F, lung;lane G, cere- bellum (from animals 1.5–2 years in age).M, a 100-bp DNA ladder.B, Northern blot analysis. Total RNA extracts (20␮g/lane) were electrophoresed on 1% agarose gels containing 2.2 M formaldehyde, were transblotted onto nylon membranes, and were probed by hybridization using a 345-bp antisense RNA probe labeled with DIG-CTP.Lanes A–G, same as forA.

(7)

8. Metz, R. J., and Radin, N. S. (1980)J. Biol. Chem.255,4463– 4467 9. Metz, R. J., and Radin, N. S. (1982)J. Biol. Chem.257,12901–12907 10. Sasaki, T. (1990)Experientia46,611– 616

11. Sasaki, T., Abe, A., and Roerink, F. (1990) inSubcellular Biochemistry: Intra- cellular Transfer of Lipid Molecules(Hilderson, H. J., ed) Vol. 16, pp.

113–127, Plenum Press, New York

12. Bankaitis, V. A., Cartee, R. T., Fry, M. R., and Kagiwada, S. (1996) inPhos- pholipid Transfer Proteins: Emerging Roles in Vesicle Trafficking, Signal Transduction, and Metabolic Regulation, pp. 51–72, Springer, New York 13. Abe, A., and Sasaki, T. (1985)J. Biol. Chem.260,11231–11239

14. Brown, R. E., Jarvis, K. L., and Hyland, K. J. (1990)Biochim. Biophys. Acta 1044,77– 83

15. Yamada, K., Abe, A., and Sasaki, T. (1985)J. Biol. Chem.260,4615– 4621 16. Yamada, K., Abe, A., and Sasaki, T. (1986) Biochim. Biophys. Acta879,

345–349

17. Brown, R. E., Stephenson, F. A., Markello, T. M., Barenholz, Y., and Thompson, T. E. (1985)Chem. Phys. Lipids38,79 –93

18. Abe, A. (1990)J. Biol. Chem.265,9634 –9637

19. Abe, A., and Sasaki, T. (1989)Biochim. Biophys. Acta985,45–50 20. Abe, A., and Sasaki, T. (1989)Biochim. Biophys. Acta985,38 – 44

21. Rosenfeld, J., Capdevielle, J., Guillemot, J., and Ferrara, P. (1992)Anal.

Biochem.203,173–179

22. Smith, D. E., and Fisher, P. A. (1984)J. Cell Biol.99,20 –28

23. Mattjus, P., Molotkovsky, J. G., Smaby, J. M., and Brown, R. E. (1999)Anal.

Biochem.268,297–304

24. Mattjus, P., Pike, H. M., Molotkovsky, J. G., and Brown, R. E. (2000)Biochem- istry, in press

25. Gammon, C. M., Vaswani, K. D., and Ledeen, R. W. (1987)Biochemistry26, 6239 – 6243

26. Kyte, J., and Doolittle, R. F. (1982)J. Mol. Biol.157,105–132 27. Nakai, K., and Kanehisa, M. A. (1992)Genomics14,897–911

28. Saupe, S., Descamps, C., Turcq, B., and Be´gueret, J. (1994)Proc. Natl. Acad.

Sci. U. S. A.91,5927–5931

29. Chen, J.-Q., Zhang, H.-T., Hu, M-H., and Tang, J.-G. (1995)Appl. Biochem.

Biotech.55,5–15

30. Wirtz, K. W. A. (1997)Biochem. J.324,353–360

31. Moreau, P., Bessoule, J. J., Mongrand, S., Testet, E., Vincent, P., and Cassagne, C. (1998)Prog. Lipid Res.37,371–391

32. Mahuran, D. J. (1998)Biochim. Biophys. Acta1393,1–18 33. Kolter, T., and Sandhoff, K. (1998)Brain Pathol.8,79 –100 34. Bloj, B., and Zilversmit, D. B. (1981)J. Biol. Chem.256,5988 –5991 35. Fu¨ rst, W., and Sandhoff, K. (1992)Biochim. Biophys. Acta1126,1–16 36. Conzelmann, E., Burg, J., Stephan, G., and Sandhoff, K. (1982)Eur. J. Bio-

chem.123,455– 464

37. Hiraiwa, M., Soeda, S., Kishimoto, Y., and O’Brien, J. S. (1992)Proc. Natl.

Acad. Sci. U. S. A.89,11254 –11258

38. Soeda, S., Hiraiwa, M., O’Brien, J. S., and Kishimoto, Y. (1993)J. Biol. Chem.

268,18519 –18523

39. Vogel, A., Schwarzmann, G., and Sandhoff, K. (1991)Eur. J. Biochem.200, 591–597

40. Be´gueret, J., Turcq, B., and Clave`, C. (1994)Trends Genet.10,441– 446 41. Wu, J., Saupe, S. J., and Glass, N. L. (1998)Proc. Natl. Acad. Sci. U. S. A.95,

12398 –12403

42. Brown, R. E., Jarvis, K. L., Vandenburgh, A. W., Hyland, K. J., and Zwizinski, C. W. (1988)J. Cell Biol.107,A2045

43. Ichikwawa, S., and Hirabayshi, Y. (1998)Trends Cell Biol.8,198 –202 44. Rani, C. S. S., Abe, A., Chang, Y., Rosenzweig, N., Saltiel, A. R., Radin, N. S.,

and Shayman, J. A. (1995)J. Biol. Chem.270,2859 –2867 45. Datta, S. C., and Radin, N. S. (1988)Lipids23,508 –510

46. Marsh, N. L., Elias, P. M., and Holleran, W. M. (1995)J. Clin. Invest.95, 2903–2909

47. Huang, X., and Miller, W. (1991)Adv. Appl. Math.12,337–357

48. Lin, X., Mattjus, P., Pike, H. M., Windebank, A. J., and Brown, R. E. (1999) Mol. Biol. Cell10,175 (abstr.)