The EMBO Journal vol.14 no. 18 pp.4422-4433, 1995
Glycosylinositol-phosphoceramide in the free-living protozoan Paramecium primaurelia: modification of core glycans by mannosyl phosphate Nahid Azzouzl, Boris Striepen2, Peter Gerold, Yvonne Capdeville3 and Ralph T.Schwarz Zentrum ftir Hygiene und Medizinische Mikrobiologie, Philipps-Universitat Marburg, Robert-Koch-Strasse 17, 35037 Marburg, Gennany and 3Centre de Genetique Moleculaire, UPR 2420, Centre National de la Recherche Scientifique (CNRS), 91198 Gif-sur-Yvette, France 2Present address: Department of Biology, University of Pennsylvania, Philadelphia, PA, USA 'Corresponding author
Glycolipids synthesized in a cell-free system prepared from the free-living protozoan Paramecium primaurelia and labelled with [3H]mannose and [3H]glucosamine using GDP-[3H]mannose and UDP-[3HJN-acetyl glucosamine, respectively, were identified and structurally characterized as glycosylinositol-phosphoceramides (GIP-ceramides). The ceramide-based lipid was also found in the GIP membrane anchor of the G surface antigen of P.primaurelia, strain 156. Using a combination of in vitro labelling with GDP-[3H]mannose and in vivo labelling with 33P, we found that the core glycans of the P.primaurelia GIP-ceramides were substituted with an acid-labile modification identified as mannosyl phosphate. The modification of the glycosylinositolphospholipid core glycan by mannosyl phosphate has not been described to date in other organisms. The biosynthesis of GIP-ceramide intermediates in P.primaurelia was studied by a pulse-chase analysis. Their structural characterization is reported. We propose the following structure for the putative GIP-ceramide membrane anchor precursor of P.primaurelia surface proteins: ethanolamine phosphate-6Man-al-2Manal-Man-(mannosyl phosphate)-al4glucosamineinositol-phosphoceramide. Keywords: biosynthesis/ceramide/GPI-anchor/Paramecium
Introduction Protozoa mainly use the glycosyl-phosphatidylinositol (GPI) structure, linked directly to protein, to anchor their membrane proteins (reviewed by McConville and Ferguson, 1993). The biosynthesis of the GPI-anchored proteins proceeds, via a transamidase reaction, through the attachment of a preformed GPI precursor onto a new C-terminal amino acid, a result of the proteolytic removal of the original C-terminal hydrophobic signal sequence. From intensive studies of parasitic GPIs, the structure and biosynthesis of the anchor GPI precursors have been elucidated. The structure of the GPI precursor core glycan deduced from Trypanosoma brucei (Masterson et at., 1989; Menon et al., 1990a) and Toxoplasma gondii
(Tomavo et al., 1992) cell-free systems is conserved and consists of ethanolamine phosphate-6Man-al-2Man-al6Man-al-4GlcNH2, suggesting a similar biosynthetic pathway. The three mannoses and the ethanolamine phosphate molecules were shown to be transferred from a dolichol phosphorylmannose (dol-P-Man; Menon et al., 1990b) and phosphatidylethanolamine (Menon et al., 1993), respectively. The core glycan can be modified by other molecules during the biosynthesis of the GPI-anchor precursor, such as an additional GalNAc molecule in Tgondii (Tomavo et al., 1992) and a fourth al-2-mannose molecule in Plasmodiumfalciparum (Gerold et al., 1994). Modification of the core glycan can also occur after the transfer of the GPI-anchor precursor to the polypeptide chain, such as a-Gal residues in the case of the Tbrucei variable surface glycoprotein GPI-anchor (Bangs et al., 1988). However, some addition of a-Gal residues to the Tbrucei free GPI precursor has been described (Mayor et al., 1992). Other components such as GalNAc and additional ethanolamine phosphate have been reported to modify rat brain Thy-I GPI-anchor (Homans et al., 1988). In Tbrucei and Trypanosoma cruzi GPI proteins, it has been shown that the anchor is substituted in a different manner (Ferguson et al., 1988; Guther et al., 1992). This peculiarity was also found in the case of Tbrucei bloodstream and Tbrucei procyclic forms, indicating that the modifications of the core glycan can be species- and stage-specific (reviewed by McConville and Ferguson,
1993). In addition to its role in anchoring membrane proteins to the plasma membrane, recent studies have established a potential role for the parasitic and mammalian GPI structures in signal transduction and have shown their importance in cell targeting (reviewed by Field and Menon, 1993). Purified P.falciparum GPI glycolipids and GPI peptides derived from the merozoite surface proteins 1 and 2 have been shown to be involved in the pathology of severe malaria by inducing tumour necrosis factor a and interleukin 1 release by macrophages (Schofield and Hackett, 1993; Schofield et al., 1994). The possibility that they may have other roles in the regulation of the immune response is not excluded as GPI degradation mediated by a specific phospholipase C generates a putative second messenger molecule, the lipid moiety of the GPI (reviewed by Liscovitch and Cantley, 1994). In GPI-anchors, the released lipid moiety is alkyl/acylglycerol or a diacylglycerol. Ceramide-based lipids were found in yeast (Conzelmann et al., 1992; Fankhauser et al., 1993) and Dictyostelium discoideum (Stadler et al., 1989; Haynes et al., 1993). In free-living ciliates, such as Paramecium and Tetrahymena, GPI proteins have been identified (Capdeville et al., 1987; Ryals et al., 1991) and partially characterized. However, none of their anchors has been characterized completely.
4 2 Oxford University Press 4422
Paramecium primaurelia GPI-anchor
Here, we present the structure and the biosynthetic pathway of glycosylinositol-phospholipids (GIPs) synthesized by the free-living ciliate Paramecium primaurelia. This study was stimulated by the regulation and expression of the Paramecium surface antigens (SAgs), which are submitted to the phenomenon of antigenic variation. The surface of Paramecium is coated by a set of surface proteins, designated SAgs. Via mechanisms of mutual intergenic and interallelic exclusion, paramecia can express a repertoire of immunologically distinguishable SAgs which are encoded by a multigene family (Beale, 1952; Capdeville, 1971). The choice of the SAg to be expressed is controlled by environmental conditions (Beale, 1954; Sonneborn, 1974). In fact, the induction and expression of a given SAg can be controlled experimentally. The mechanisms of regulation of the expression of this family of surface proteins are still unknown. Some data suggest that regulation occurs at the level of transcription because only the mRNA corresponding to the expressed antigen was found (Forney et al., 1983; Meyer et al., 1984). The SAg seems to be involved in the control of its own expression by the production of a specific, unknown cytoplasmic factor (Capdeville, 1979). Recent work suggests the prominent role of GPI-anchored surface glycoproteins in cellular signalling and induction events (Robinson, 1991; Stefanova et al., 1991; Brown, 1993). In our previous work on Pprimaurelia GPI-anchor proteins, we were able to show by labelling experiments that one of the purified SAgs contains the components characteristic of a GPI protein (Azzouz and Capdeville, 1992). The anchor was shown to be sensitive to phosphatidylinositol-specific phospholipase C, indicating that the inositol ring is not substituted. Here, we describe the structure of the glycosylinositol-phosphoceramide (GIPceramide) produced in a cell-free system prepared from P.primaurelia. The glycan moiety of the polar glycolipid is substituted with an acid-labile unit, which was identified as a mannosyl phosphate. We propose that the structure of the most polar GPI glycolipid, which consists of ethanolamine phosphate-6Man-al-2Man-al-6Man(mannosyl phosphate)-al-4glucosamine-inositol-phosphoceramide, is destined to anchor the P.primaurelia GPI proteins.
Results Labelling of glycolipids in P.primaurelia To label the glycolipids, we used membrane preparations from P.primaurelia incubated with GDP-[3H]mannose,
UDP-[3H]N-acetyl glucosamine or CDP-[3H]ethanolamine. Additionally, we performed in vivo labelling with [3H]glucosamine, [3H]palmitic acid and 33P. The labelling was stopped by extraction with chloroform/methanol (C/M; 2:1 by volume), followed by chloroform/methanol/ water (C/M/W; 10:10:3 by volume). Thin layer chromatography (TLC) analysis on silica 60 plates (solvent system A) of C/M/W extractable glycolipids labelled with GDP[3H]mannose (Figure IA), and with UDP-[3H]N-acetyl glucosamine (Figure IB) in a cell-free system, showed two major glycolipids (I and II) labelled by both GDP[3H]mannose and UDP-[3H]N-acetyl glucosamine. These glycolipids were also present after in vivo labelling with glucosamine (Figure IC). The TLC analyses on silica
0
F
5
0
10
distanmce [cm]
Fig. 1. TLC analysis of the Pprimaurelia [3H]mannose- and [3H]glucosamine-labelled glycolipids. Glycolipids were labelled either in vitro using P.primaurelia membranes with GDP-[3H]mannose (A and D) or UDP-[3H]N-acetyl glucosamine (B and E), or in vivo using [3H]glucosamine (C). The labelled glycolipids were extracted with C/M (2:1 by volume) followed by C/M/W (10:10:3 by volume). The C/M/W-extracted glycolipids were dried and partitioned between water and water-saturated n-butanol. The glycolipids recovered in C/M were first subjected to 'Folch' washing and then to butanol-water phase partitioning. The glycolipids recovered in the butanol phase were analysed on silica TLC plates using solvent systems C/M/0.25% KCl (10:10:3 by volume; A-C) or C/M/acetic acid/W (25:15:4:2 by volume; D and E). Radioactivity was scanned using a TLC scanner (Berthold LB 2842). 0, origin; F, front. Glycolipids that migrate close to the origin and the glycolipid indicated by '*' were not characterized further. The latter is probably a plate overloading artefact.
60 (solvent system B) of C/M extracts from GDP[3H]mannose-labelled (Figure ID) and UDP-[3H]N-acetyl glucosamine-labelled (Figure lE) membranes show three glycolipids (III, IV and V) labelled with both GDP[3H]mannose and UDP-[3H]N-acetyl glucosamine. Two glycolipids marked a (Rf value 0.57) and a (Rf value 0.62) were labelled exclusively with UDP-[3H]N-acetyl glucosamine.
Characterization of the CIM and CIMIW extractable glycolipids as GIPs Labelled glycolipids recovered in the C/M and C/M/W extracts after production in the cell-free system were tested for their sensitivity to GPI-specific enzymes and chemical treatments. As summarized in Table I, all glycolipids, except glycolipid VI, were identified as structures containing non-acetylated glucosamine (sensitivity to nitrous acid deamination), inositolphospholipid [sensitivity to phosphatidylinositol-phospholipase C (PI-PLC) and to GPI-phospholipase D (GPI-PLD)]. Further characterization of the mannose-labelled glycolipid VI, which was insensitive to all these treatments, showed it to be dol-P-Man, as discussed below. In addition to labelling with GDP-[3H]mannose and UDP-[3H]N-acetyl glucosamine, glycolipid I was found to be labelled also by CDP[3H]ethanolamine in the cell-free system. None of the other glycolipids incorporated ethanolamine from CDP-
[3H]ethanolamine.
The glucosamine-labelled
glycolipids
4423
N.Azzouz et al. Table I. Characterization of glycolipids
Glycolipid species
Rf values
Labelled via:
Sensitivity to:
GDP-
UDP-
[3H]Man
[3H]GlcNAc
0.05 0.42
+ +
0.10 0.20 0.35 0.90 0.57
-
0.62
-
CDP-[3H]EtN PI-PLC
PLD
PLA2
HNO2
KOH
+ -
+ +
+ +
-
-
+ +
-
+
+
+
-
+
+
-
+
-
+ + +
+ +
-
+ +
+ +
-
+ +
-
-
-
-
-
-
+
-
-
+ +
+
C/M/W extracts, system A I
II C/M extracts, system B III IV V VI a e
+
+ +
-
+
Glycolipids were labelled with UDP-[3H]N-acetyl glucosamine, GDP-[3H]mannose or CDP-[3H]ethanolamine in a cell-free system prepared from P.primaurelia and extracted with C/M (2:1 by volume) followed by C/M/W (10:10:3 by volume). The C/M/W-extracted glycolipids were dried and subjected to butanol-water phase partition. The C/M-extracted glycolipids were subjected to 'Folch' washing and to butanol-water phase partitioning. The glycolipids recovered in the butanol phase were analysed on silica TLC plates. For enzymatic and chemical treatments, the glycolipids were dried, treated and then subjected again to butanol-water phase partitioning. The yield of cleavage was assessed by TLC analysis of the butanol phase and/or liquid scintillation counting of the organic and aqueous phases.
a and P were characterized as GlcNH2-inositol-P-lipid and GlcNAc-inositol-P-lipid respectively, based on their sensitivity to PI-PLC. In contrast to glycolipid a, glycolipid P was found to be resistant to nitrous acid deamination, indicating that the glucosamine is acetylated. These glycolipids were also found to be sensitive to GPIPLD, as shown for the malaria early GPI intermediate GlcNH2-PI (Gerold et al., 1994). The lipid moieties of the mannose-labelled glycolipids I-V and glucosaminelabelled glycolipids a and I were found to be resistant to phospholipase A2 and to mild alkali treatments, indicating that these glycolipids might be GIP-ceramide. We report the structural characterization of glycolipids I-V below.
Structural characterization of the CIM extractable glycolipids To characterize the C/M-extracted glycolipids, the [3H]mannose-labelled glycolipids III-V were purified by TLC. An aliquot from each glycolipid was subjected to nitrous acid deamination, which cleaves GPIs specifically at the level of the glucosamine-inositol bond (Ferguson et al., 1985), and sized on a Bio-Gel P4 size-exclusion column before and after Jack bean a-mannosidase digestion. The remainder of each glycolipid was dephosphorylated prior to nitrous acid deamination and sodium borohydride reduction to yield the core glycan (neutral glycan). The Bio-Gel P4 analysis showed that the watersoluble fragments from glycolipids III-V generated by nitrous acid deamination eluted at 4.2, 3.0 and 2.2 glucose units respectively (Figure 2A-C). Jack bean a-mannosidase treatment resulted in the complete digestion of the three fragments (Figure 2D-F). The neutral glycans generated from glycolipids Ill-V co-eluted on Dionex high pH anion exchange chromatography (HPAEC) with Tbrucei glycan standards Man3-2,5-anhydro-mannitol (Man3AHM), Man2-AHM and Man1-AHM respectively (Figure 2G-I). Together, these results indicate that glycolipids IIIV are the early biosynthetic intermediates for the potential precursor, which should be found in C/M/W extracts. Glycolipid VI was also purified and tentatively identified 4424
as dol-P-Man, because its formation was blocked by amphomycin and by the release of mannose under mild acid conditions (results not shown).
The polar glycolipids I and II contain the conserved ManrAHM core glycan [3H]Mannose-labelled glycolipids I and II, synthesized in the cell-free system and recovered in C/M/W extracts, were purified by preparative TLC, dephosphorylated, deaminated and reduced to generate neutral glycans. Neutral glycans derived from glycolipids I and II co-eluted with the Man3-AHM standard (Figure 3A and D) on Dionex HPAEC. Treatment with Aspergillus saitoi a-mannosidase specific for al-2 linkages generated two fragments. One fragment co-eluted on Dionex HPAEC with the Man2-AHM standard and the second co-eluted with a mannose standard (identical elution position as glucose; Figure 3B and E), indicating the removal of one terminal al-2-linked mannose. Jack bean a-mannosidase treatment of the neutral glycans generated from glycolipids I and II produced a single fragment (Figure 3C and F) which co-eluted on Dionex HPAEC with a mannose standard, indicating the liberation of all the mannose residues. Furthermore, neutral glycans generated from glycolipids I and II have the same size on a Bio-Gel P4 column as the Tbrucei Man3-AHM standard (4.2 glucose units; results not shown), suggesting an identical structure: Man-ax1-2Man-al-6Man-ax1-4AHM.
Identification of an acid-labile modification in the precursor core glycan Additional information concerning the structure of the GDP-[3H]mannose-labelled glycolipids I and II was obtained by an analysis of the water-soluble head groups generated by nitrous acid treatment. The water-soluble fragments generated by this treatment were analysed on a Bio-Gel P4 column, before and after exo-a-mannosidase treatment. Although the structures of the two neutral glycans generated from glycolipids I and II and the Tbrucei Man3-AHM standard are identical, the HNO2-
Paramecium primaurelia GPI-anchor i II
2
cC
c 0 2
C
2 m:m IW
A
15io
2x
4 3 21
D
,543
20)o .- III
2
III
-!
0. 10
0. 5
0~~~~~~~~~~~'IIIIIIm 0I
300t E
B
, .
v .
IV
v
EIV
I
200 +
a-
91
111 31
liH l 11
II
v v
v
-1
51
Fraction number
20
40
Fraction number
Fig. 2. Analysis of the core glycans generated from glycolipids recovered in C/M. TLC-purified [3H]mannose-labelled glycolipids III, IV and V were subjected to nitrous acid deamination and the generated hydrophilic moieties were sized on Bio-Gel P4 chromatography before (A-C) and after Jack bean a-mannosidase (D-F) treatment. (G)-(I) show the Dionex HPAEC analysis of the dephosphorylated, deaminated and reduced core glycans generated from glycolipids III-V, respectively. The elution positions of the co-injected glucose oligomer standards and Man1-AHM, Man2-AHM and Man3-AHM standards derived from Tbrucei GPI glycolipids analysed separately are indicated at the top of (A) and (G). IP, injection pic.
generated fragments showed differences in their elution positions on a Bio-Gel P4 column, with 10.5 and 9.5 glucose units respectively (Figure 4A and D) compared with 6.7 glucose units for the corresponding fragment generated from Tbrucei glycolipid P2. The two glycans generated from glycolipids I and II are both partially sensitive to A.saitoi a-mannosidase (Figure 4B and E) and to Jack bean a-mannosidase (Figure 4C and F). The presence of a terminal ethanolamine, or of any terminal modification except mannose, would render the GPI core glycan completely insensitive to a-mannosidases. Glycolipid I was shown to contain a terminal ethanolamine because it could be labelled in a cell-free system using CDP-[3H]ethanolamine, in contrast to glycolipid II which could not be labelled with CDP-[3H]ethanolamine. The sensitivity of glycolipid I to a-mannosidase treatment (Figure 4B and C) suggests the presence of a fourth mannose molecule linked indirectly to the core glycan.
The limited sensitivity of glycolipid II (not labelled with CDP-[3H]ethanolamine) to Jack bean a-mannosidase indicates that its core glycan is blocked by an acid-labile molecule (a-mannosidase blocking molecule) different from ethanolamine phosphate. The results suggest that the core glycans of glycolipids I and II are substituted by a fourth mannose molecule, which is linked to the core glycan in an acid-labile manner.
Additional mannose is linked to the core glycan via a phosphodiester bond To investigate the nature of the acid-labile modification, we used a combination of Mono Q anion-exchange chromatography of the nitrous acid-generated hydrophilic fragments of glycolipids I and II and an analysis of the products of in vivo labelling with 33P. Mono Q analysis showed that the fragments generated from glycolipid II, which is not labelled by CDP-[3H]ethanolamine, co-eluted
4425
N.Azzouz et al. glycolipid
I
glycolipid
glycolipid II
glycolipid II
I
C-., 0 0
C, 0
E (A
C-) CL
E C-)
Q,
ar '0.
w
4-
m
a
9 x 61
10.
I
i 01
la
a
Fraction number
Fig. 3. Dionex HPAEC analysis of the core glycans generated from glycolipids I and II. TLC-purified [3H]mannose-labelled glycolipids I and II were dephosphorylated, deaminated and reduced, as described in Materials and methods. The resulting neutral glycans were analysed by Dionex HPAEC before (A and D) and after A.saitoi al-2-specific mannosidase (B and E) or after Jack bean a-mannosidase (C and F) treatments. The elution positions of the co-injected glucose oligomer standards are indicated at the top of each profile. Labelled Man1-AHM, Man2-AHM and Man3-AHM standards are indicated at the top of (A) and (D). IP, injection pic.
with the fragment generated from the Tbrucei glycolipid P2 (Figure 5A), which contains one negative charge caused by the presence of ethanolamine phosphate in its structure. The elution position of the same fragment generated from glycolipid I (Figure 5B) indicated that it contains more than one negative charge (one charge belongs to ethanolamine phosphate). However, the latter eluted with the same percentage of NH4OAc as the fragment generated from the mammalian glycolipid described by Kamitani et al. (1992), which contains two negative charges. To determine the nature of this negative charge, we performed in vivo labelling with 33P and, as a control, in vitro labelling with GDP-[3H]mannose (Figure 6A). The 33P-labelled glycolipids were extracted as described above and analysed by TLC (Figure 6B). Glycolipid II labelled with 33P was purified by TLC and subjected to nitrous acid treatment to liberate the inositol-[33P]lipid moiety. The resulting hydrophilic fragment was then analysed on a Bio-Gel P4 column. The co-elution of the fragments generated from glycolipid II labelled with GDP-[3H]mannose (Figure 6C) and 33P (Figure 6D) indicates that glycolipid II contains an extra phosphate group. Together, Mono Q analysis and 33P-labelling provide evidence that the acid-labile modification present in glycolipids I and II is a mannosyl phosphate unit. In contrast to Aspergillus phoenicis exo-acl-2-mannosidase which is unable to cleave mannose linked to phosphate in yeast oligosaccharides linked via phosphate (Hernandez et al., 1989), we show that the A.saitoi exo-al-2-mannosidase is able to cleave mannose.
4426
Fraction number Fig. 4. The core glycans of glycolipids I and II contain an additional mannose molecule. TLC-purified [3H]mannose-labelled glycolipids were deaminated with nitrous acid. The generated hydrophilic moieties were sized on Bio-Gel P4 chromatography before (A and D) and after A.saitoi al-2-specific mannosidase (B and E) or after Jack bean ax-mannosidase (C and F) treatments. The elution positions of the co-injected glucose oligomer standards are indicated at the top of (A) and (D).
11
21
31
41
Fraction number Fig. 5. The core glycan of glycolipid II contains one negative charge. TLC-purified [3H]mannose-labelled glycolipids I (B) and II (A) were subjected to nitrous acid deamination. The water-soluble fragments were analysed on a Mono Q anion-exchange column. The elution position of the corresponding fragment generated from the GPI-anchor precursor of Tbrucei P2 is indicated at the top of (A) by an asterisk.
The activity was verified further by testing the ability of this enzyme to cleave GDP-[3H]mannose (results not shown).
Paramecium primaurelia GPI-anchor
0
F
c
L,
A
v
-9 !v
9.
II
150
100
0 c
50-
9~ ~ ~ ~ ~
us
0 >.20 41
E
|D1 0~~~~~~~~~
a. U
80
*
-
S
VV
V.q
9
.
6040
A
-
I20 0
dI
0
510 distance [cm]
15
20
31
51
71
91
III
Fraction number
Fig. 6. The water-soluble fragment generated from glycolipid II by nitrous acid deamination treatment contains an extra phosphate group. In vivo 33P-labelling was performed and glycolipids were extracted. The C/M/W-extracted glycolipids were subjected to butanol-water phase partition and analysed on silica TLC plates (B), along with a GDP-[3H]mannose-labelling as a control (A). (C and D) An analysis on Bio-Gel P4 of the glycans generated by nitrous acid deamination from glycolipid II labelled with [3H]mannose (C) or 33P (D), showing the fragment generated from TLC-purified [33P]glycolipid II still labelled with 3 P. (*) Uncharacterized material.
The mannosyl phosphate is attached to the first mannose
The position of the mannosyl phosphate modification in the core glycan was investigated using the nitrous acidgenerated hydrophilic fragment from [3H]mannoselabelled glycolipid II, because this glycolipid does not contain the terminal ethanolamine phosphate blocking molecule. The hydrophilic fragment generated by nitrous acid treatment was first subjected to Jack bean a-mannosidase to remove all the mannose residues that are accessible to the enzyme. After this enzymatic treatment, the generated phosphorylated fragment was purified by passing the reaction mixture over a Bio-Gel P2 column, equilibrated and eluted with water (Figure 7A). The phosphorylated fragment was then subjected to alkaline phosphatase treatment to remove the phosphate molecule (the a-mannosidase blocking molecule). The resulting material was reduced, desalted on anion- and cation-exchange columns, filtered and analysed on Dionex HPAEC. Figure 7B shows that the fragment generated by a-mannosidase and alkaline phosphatase treatment co-eluted at a glucose unit corresponding to the position of a Man-AHM standard. Therefore, we conclude that the mannosyl phosphate modification is attached to the core glycan at the mannose linked to the glucosamine. The base-resistant lipid moiety of the P.primaurelia GPI is sensitive to sphingomyelinase treatment C/M- and C/M/W-extracted glycolipids were found to be resistant to both mild alkali treatment and phospholipase
A2 (PLA2), indicating that the anchor may not be a phosphatidylinositol. To characterize the nature of the lipid moiety, we performed in vivo labelling using [3H]palmitic acid. An aliquot of cells was used to isolate the Rprimaurelia 156G SAg and the remainder was used to isolate the GPI glycolipids. [3H]Palmitic acid-labelled glycolipids I and II were purified and tested for their sensitivity to mild alkali and sphingomyelinase treatment. A comparison of the lipids released from glycolipids I and II, by sphingomyelinase (Figure 8A and B) or by PTPLC treatment in the case of glycolipid TI as a control (Figure 8C), showed that both released lipids have the same Rf value (0.76) on TLC, using C/M (9:1 by volume) as a solvent system in which the intact glycolipids I and IT do not migrate (results not shown). The sensitivity to sphingomyelinase suggests that the lipid actually consists of ceramide. Figure 8D shows that the lipid released from the 156G SAg by PI-PLC treatment has the same Rf value as the lipid released from glycolipids I and II following sphingomyelinase or PI-PLC treatment. Thus, these results show that the PI-PLC-sensitive anchor component of 156G SAg, glycolipids I and II, is the same ceramidebased lipid.
Glycolipid I is the mature GPI precursor To provide evidence that glycolipids IT-V are intermediates in the biosynthesis of glycolipid I, we performed in vitro pulse-chase labelling using GDP-[3H]mannose. As shown in Figure 9, at 0 min of chase the radioactivity was incorporated mainly in dol-P-Man (glycolipid VI); the 4427
N.Azzouz et aL F
0 100
E 0
5
1
9
13
17
Fraction number a 0n
BPv
I2
I
I
I
C 0
0
1
2
v
:i:
3
v
4
v
v
15
)o Ea
10
(.
5 I 4 __
I a_,--
-..d.d.
distance [cm]
0
1
21
41
61
Fraction number Fig. 7. The mannosyl phosphate modification is linked to the core glycan at the mannose adjacent to the molecule of glucosamine. TLC-purified [3H]mannose-labelled glycolipid II was deaminated and treated with Jack bean a-mannosidase. The liberated mannose residue was discarded by passing the reaction mixture over a Bio-Gel P2 column (A). The fractions eluting in the void volume were pooled, treated with calf intestine alkaline phosphatase, reduced and analysed by Dionex HPAEC (B). The elution position of the co-injected glucose oligomer and the elution positions of Man1-AHM, Man2-AHM and Man3-AHM standards are indicated.
Fig. 8. Sensitivity of glycolipids I and II to sphingomyelinase treatment. Cells were labelled in vivo with [3H]palmitic acid and glycolipids were extracted as described in the legend to Figure 1. (A and B) Analysis of the lipid moieties released by sphingomyelinase from glycolipids I (A) and 11 (B). (C and D) Analysis of the lipid moieties released by PI-PLC from glycolipid II (C) and P.primaurelia [3H]palmitic acid-labelled 156G SAg (D). The purification of the 156G SAg was performed as described by Azzouz et al. (1990). The released lipids were extracted as described in Materials and methods and analysed by TLC on silica HPTLC plates using the solvent system C/M (9:1 by volume). On the basis of the Rf values, the ceramidebased lipid resulting from PI-PLC or sphingomyelinase treatment are identical.
radioactivity disappeared progressively from dol-P-Man with the appearance of Man1-GlcNH2-inositol-phosphoceramide (glycolipid V), Man2-GlcNH2-inositol-phosphoceramide (glycolipid IV), Man3-GlcNH2-inositolphosphoceramide (glycolipid III), Man3(P-Man)-GlcNH2inositol-phosphoceramide (glycolipid II) and EtN-PMan3(P-Man)-GlcNH2-inositol-phosphoceramide (glycolipid I) successively during the chase. From the pulse-chase analysis, glycolipid I appears to be the mature glycolipid which is transferred to nascent protein in P.primaurelia. The biosynthesis of glycolipid VI is inhibited by the addition of amphomycin (results not shown), which is known to block the formation of dol-P-Man (glycolipid VI) from GDP-mannose and thus block the biosynthesis of glycolipid I. We conclude that dol-P-Man, as in other organisms, is the direct donor of at least
4428
the first mannose residue during the biosynthesis of glycolipid I.
Discussion The ciliated protozoan Paramecium has provided a model for the study of surface membrane proteins (reviewed by Capdeville al., 1993) and the cellular role of calcium and trichocyst exocytosis (reviewed by Cohen and Kerboeuf, 1993). Because the major membrane proteins (termed SAgs or immobilization antigens) of this protozoan are anchored in the surface membrane via a glycolipid structure with the characteristics of a GPI-anchor structure (Capdeville et al., 1987), we have used Paramecium as a model to study the biosynthesis of GPI-anchors in ciliates. Here, we report the structure of a spectrum of GIP-
Paramecium primaurelia GPI-anchor 0
III v IV
11
VI F
45 sec 0.
u
g
0
1mh
mms
lo
20
distance [cm] Fig. 9. Pulse-chase experiment. Membranes were pulse-labelled with GDP-[3H]mannose for 15 s. They were chased with an excess of unlabelled GDP-mannose. The glycolipids were extracted at different times in one step with C/M/W (10:10:3 by volume) by adding C/M (1:1 by volume) to the aqueous reaction mixtures. The resulting extracts were processed, as described in Materials and methods. The glycolipids were analysed by TLC (silica 60, solvent system A). The
reduction in intensity of dol-P-Man (glycolipid VI) during the chase with the appearance of the different intermediates (V, IV, III, II and I) reflects the different biosynthesis steps of the glycolipid I.
ceramides synthesized by Pprimaurelia. Glycolipids were labelled in vitro with GDP-[3H]mannose and UDP-[3H]Nacetyl glucosamine and then extracted by the addition of C/M (2:1 by volume) followed by C/M/W (10:10:3 by volume). A TLC analysis of the glycolipids recovered in C/M and C/M/W showed five glycolipids (glycolipids I-V) labelled with both GDP-[3H]mannose and UDP[3H]N-acetyl glucosamine, two glycolipids labelled with UDP-[3H]N-acetyl glucosamine (glycolipids a and f) only and one glycolipid (glycolipid VI) labelled with GDP-[3H]mannose only. The latter was dol-P-Man. Supplementary in vitro labelling with CDP-[3H]ethanolamine revealed that glycolipid I recovered in the C/M/W extracts was the only glycolipid labelled with [3H]ethanolamine. Glycolipids I and II were also obtained after in vivo labelling with [3H]glucosamine, indicating that both in vitro and in vivo labelling could be used for their structural characterization. We followed two approaches to determine the structure of the glycan part of these glycolipids, as summarized in Figure 10. The first approach involved a structural comparison of the neutral glycans, generated by hydrofluoric acid, deamination and reduction treatments, with the standard neutral glycans generated from the Tbrucei GPI glycolipids (Mayor et al., 1990) on Dionex HPAEC. The second approach involved the structural analysis of the hydrophilic fragments generated by nitrous acid deamination on Bio-Gel P4 columns. The neutral glycans generated from glycolipids I-V were analysed by Dionex HPAEC and shown to co-elute with Man3-AHM, Man3AHM, Man3-AHM, Man2-AHM and Man1-AHM standards, respectively. A.saitoi al-2-mannosidase treatment of Man3-AHM generated from glycolipids I and II resulted in the cleavage of one mannose residue, indicating the presence of an exo-mannose molecule in a 1-2 linkage. Based on the elution position on Dionex HPAEC of the intact fragments or the fragments generated by al-2mannosidase and Jack bean a-mannosidase, compared with standards generated from Tbrucei GPI glycolipids, we conclude that the structure of the Man3-AHM neutral glycan of the Paramecium glycolipids is consistent with the commonly conserved core glycan structure in all GPIs characterized so far: Man-al-2Man-al-Man-al-4AHM. Bio-Gel P4 analysis showed that fragments generated by specific cleavage with nitrous acid from glycolipids I and II were partially sensitive to a-mannosidase treatment, even though glycolipid I has a terminal mannosyl molecule blocked by ethanolamine. Taking into consideration the fact that glycolipid II is a biosynthetic intermediate for the biosynthesis of glycolipid I, we provide evidence for the presence of an acid-labile modification containing mannose. The acid-labile modification was shown to contain an extra phosphate molecule by additional in vivo labelling with 33P. The results obtained for the neutral glycans and the water-soluble fragments generated by nitrous acid treatment when combined suggest that the core glycan of the Paramecium glycolipids I and II is modified by a mannosyl phosphate unit linked to the mannose adjacent to glucosamine via a phosphodiester bond. Glycosyl phosphate (mannosyl phosphate and glucosyl phosphate) substitutes have been described in yeast N-linked oligosaccharides (Mill, 1966; Hernandez et al., 4429
N.Azzouz et al.
in Wtro and in wvo labelling
extraction and TLC purification of glycolipids dephosphorylation (HF) deamination (HNO, ) reduction (NaBH45)
deamination (HNO )/ 2
HNO~-generated fragments Jack bean a-mannosidase/
Bio-Gel P2 analysis
neutral core glycans
+/- A. saitoi or J. bean a mannosidases - Omm
Bio-Gel P4 analysis Mono Q AEC analysis
Dionex HPAEC analysis
rnannose
phosphor) lated fragment alkaline phosphatase
Dionex HPAEC analysis Fig. 10. Schematic representation of the strategy used for the structural determination of the P.primaurelia GIPs.
1989) and also in lipophosphoglycans (LPG) from Leishmania major (McConville and Homans, 1992). The structure of the core glycan of the Paramecium GPI described here differs from the phosphorylated free GPI glycolipid described for another free-living ciliate Tetrahymena mimbres. The glycan core of this glycolipid was shown to contain a phosphorylated Man-al-2Manal-3Man--al-4GlcNH2 (Weinhart et al., 1991). The second mannose and the third mannose adjacent to the glucosamine were shown to be linked in a 1-3 instead of a 1-6 linkage. Interestingly, the lipid moieties of the Paramecium glycolipids were shown not to be susceptible to mild alkali treatment, indicating the absence of ester-linked fatty acids, but were found to be sensitive to sphingomyelinase treatment, suggesting a ceramide-based lipid. The ceramide-based lipid was also shown to anchor the 156G GPI SAg of P.primaurelia. This finding suggests that the lipid moieties are not remodelled, in contrast to the Trypanosoma lipid remodelling during the biosynthesis of the GPI precursor (Masterson et al., 1990). In yeast, the ceramide lipid was identified only in the GPI-anchor of proteins (Conzelmann et al., 1992; Fankhauser et al., 1993) and not in the GPI-anchor precursor (Sipos et al., 1994), also suggesting a kind of lipid remodelling. The lipid moieties of the early Paramecium GIP intermediates, containing GlcNAc-inositol (glycolipid P) and GlcNH2-inositol (glycolipid a), were also found to be ceramide-based lipids. The structural characterization of the Paramecium GIP glycolipids, along with the pulsechase experiment results, led us to propose the biosynthetic pathway of the P.primaurelia GPI precursor shown in Figure 11. The pathway for the biosynthesis of glycolipid I involves the sequential addition of glucosamine, mannose, mannosyl phosphate and ethanolamine to a probable inositol-phosphoceramide derived from phosphatidylinositol (Becker and Lester, 1980). As in other organisms, the three mannose molecules
4430
Inositol-P-Ceramide * GIcNAc-lnositol-P-Ceramide GIcN-lnositol-P-Ceramide Man-GlcN-lnositol-P-Ceramide
Man-Man-GIcN-Inositol-P-Ceramide Man-Man-Man-GIcN-lnositol-P-Ceramide Man-Man-Man (-P-Man)-GIcN-lnositol-P-Ceramide EtN-P-Man-Man-Man (-P-Man)-GIcN-lnositol-P-Ceramide Fig. 11. Biosynthetic pathway for P.primaurelia GIP-anchor. This pathway is based on the structural characterization of the different intermediates and the pulse-chase experiment. The pathway differs from those described for other biological systems: (i) the presence of ceramide-based lipid suggests that the first step involves the addition of GlcNAc to inositol-phosphoceramide; and (ii) the presence of mannosyl phosphate is a side-chain modification. The hypothetical inositol-phosphoceramide (indicated by an asterisk) has not yet been identified.
that constitute the GPI precursor core glycan in Paramecium are derived from dol-P-Man synthesized from GDPmannose (Menon et al., 1990b). In Tbrucei, all GPI biosynthetic lipid intermediates were located at the cytoplasmic face of the endoplasmic reticulum (Vidugiriene and Menon, 1993, 1994). It has been shown that in Tbrucei the biosynthesis of GPI proteins is initiated by the biosynthesis of the GPI precursor on the cytoplasmic side of the endoplasmic reticulum, which is then trans-
Paramecium primaurelia GPI-anchor
located across the membrane into the lumenal side of the endoplasmic reticulum to be transferred to a protein. Because our data on Paramecium GIP glycolipids suggest that the mannosyl phosphate modification is added onto the manosylated core glycans before transfer of the GPI precursor onto the polypeptide chain, it is necessary to search for the donor of this modification. GDP-mannose is a potential donor, as in yeast mannan biosynthesis in which the mannosyl phosphate units are derived directly from GDP-mannose (Karson and Ballou, 1978). This would have direct implications for the membrane topology of the assembly process during GPI biosynthesis, because unlike dol-P-Man, GDP-mannose is located only on the cytoplasmic side of the endoplasmic reticulum, supporting the data obtained from Tbrucei. The SAgs of P.primaurelia are encoded by a repertoire of genes that are expressed via mutually exclusive mechanisms, known as antigenic variation. In general, one type of antigen is expressed under a given set of conditions, and antigen switching can occur in response to a change in environmental conditions, most notably shifts in temperature. The mechanisms which control this phenomenon in ciliated protozoa are still unknown. We have already described a protocol for the purification of the Rprimaurelia 156G SAg, expressed at a lower temperature (Azzouz et al., 1990). The same protocol is used to purify the 168D SAg expressed at a higher temperature. The availability of relatively large amounts of the purified SAgs expressed under different conditions and in different strains will now permit extensive structural studies. The GPI glycolipids characterized here should facilitate these studies. The elucidation of the structures of the GPI-anchor of different Paramecium SAgs expressed under different conditions, and the role of their anchor or its products (glycan and lipid parts) generated by an endogenous phospholipase C-like hydrolase (Capdeville et al., 1987), may also provide new insights into the regulation of expression of SAgs in ciliates. The Paramecium system is also suitable for the purification and characterization of enzymes implicated in this post-translational modification of membrane proteins.
Materials and methods Materials 9,10-(n)-[3H]Palmitic acid (53.4 Ci/mmol), [33P]phosphoric acid (0.2 Ci/ mmol) and [3H]glucosamine hydrochloride (26.0 Ci/mmol) were obtained from Amersham Corporation. Guanosine diphosphate-3-4[3H]mannose (15.1 Ci/mmol) and uridine diphospho-N-acetyl-D-6[3H]glucosamine (18.9 Ci/mmol) were purchased from DuPont-New England Nuclear.
CDP-[3H]ethanolamine (28.8 Ci/mmol) was synthesized according Menon et al. (1993).
to
Culture and in vivo biosynthetic labelling Cells of RPprimaurelia strain 156 were grown monoaxenically at a temperature at which the G SAg (156G SAg) was expressed exclusively (24°C) in hay grass medium inoculated with Enterobacter aerogenes on the previous day and supplemented with ,-sitosterol (0.8 tg/ml). The identity of the expressed antigen and the homogeneity of the population were verified by immobilization tests in the presence of specific 156G antibodies (Capdeville, 1971). Cultures were used when 100% of the cells expressed the G SAg. Before labelling, exponentially grown paramecia were pelleted at 250 g in a GGT centrifuge (Jouan) and washed in mineral water. Washed paramecia (2.5X I 5) were incubated in 4 ml of mineral water with [3H]palmitic acid-bovine serum albumin complex (1.25 mCi/ml) or [3H]glucosamine hydrochloride (1.25 mCi/ml) for 3 h. 33P-labelling was performed by feeding cells with bacteria
labelled previously with 33P. Bacteria were grown in Neidhardt medium (Neidhardt et al., 1974), which was phosphate limited (0.9 mM) and supplemented with [33P]phosphoric acid (1.25 mCi/ml). Washed paramecia were mixed with pelleted 33P-labelled bacteria in mineral water supplemented with f-sitosterol.
Extraction of labelled GPI glycolipids in vivo Washed cells were pelleted at 250 g in a GGT centrifuge (Jouan). Less polar glycolipids were extracted twice using 1 ml C/M (2:1 by volume), followed by C/M/W (10:10:3 by volume) to obtain the more polar
glycolipids.
Preparation of Pprimaurelia lysate were prepared as described above, except that the medium was supplemented with 10 ,ug/ml tunicamycin to block the N-glycosylation pathway of proteins. After 1 h at 37°C, cells were hypotonically lysed using the method described by Masterson et al. (1989). Cells were resuspended in ice-cold water containing 0.1 mM TLCK and I tg/ml leupeptin prior to disruption with 50 strokes in a Dounce homogenizer. An equal volume of 100 mM Na-HEPES (pH 7.4), 50 mM KCI, 10 mM MgCI2, 0.1 mM TLCK, 1 tg/ml leupeptin and 20% glycerol was added to the cell lysate prior to freezing at -80°C. The cell lysate was used for in vitro radiolabelling experiments, as described below.
Cells
Biosynthesis and extraction of GPI glycolipids in vitro
The lysate (106 cell equivalent) was washed three times with 50 mM Na-HEPES (pH 7.4), 50 mM KCI, 10 mM MgCI2, 0.1 mM TLCK and I tg/ml leupeptin. After centrifugation (Beckman J21; 10 000 r.p.m., 20 min at 4°C), the pellet was resuspended in the washing buffer supplemented with 5 mM MnCI2, 0.2 ,ug/ml tunicamycin (Calbiochem), 1 mM ATP, 1 mM CoA and 2 ,uCi 3H-labelled nucleotide sugars or 18 gCi CDP-[3H]ethanolamine. Assays were supplemented with 1 mM GDP-mannose for experiments involving UDP-[3H]N-acetyl glucosamine or with 1 mM UDP-N-acetyl glucosamine for experiments involving GDP-[3H]mannose. In the case of CDP-[3H]ethanolamine, assays were supplemented with GDP-mannose and UDP-N-acetyl glucosamine. After incubation at 37°C for 1 h, labelled glycolipids were extracted twice with 1 ml CIM (2:1 by volume), followed by C/M/W (10:10:3 by volume). For pulse-chase experiments with GDP-[3H]mannose, the membranes were labelled for 15 s at 37°C and then supplemented with 1 mM GDP-mannose (final concentration) at 0 min of chase. After different incubation periods, lipids were extracted in one step with C/M/W (10:10:3 by volume) and analysed by TLC.
TLC Labelled glycolipids recovered in C/M extracts were subjected to repeated 'Folch' washing (Scharma et al., 1974). The resulting lower phases (organic) were pooled, dried in a speed-Vac (Savant Inc.) and partitioned between water and water-saturated n-butanol. Labelled glycolipids recovered in the C/M/W extracts were dried and partitioned between water and water-saturated n-butanol. The glycolipids recovered in the butanol phases were analysed on silica 60 plates (Merck) using solvent systems [system A, C/M/0.25% KCI (10:10:3 by volume) or system B, C/M/acetic acid/W (25:15:4:2 by volume)] or on silica 50 000 HPTLC plates (Merck) with C/M (9:1 by volume). After chromatography, the plates were dried and scanned for radioactivity with a Berthold LB 2842 automatic scanner. For structural analysis, glycolipids were purified by scraping the corresponding areas and extracted for 15 min under sonication with C/M/W (10:10:3 by volume) or C/M (2:1 by volume).
Enzymatic and chemical analysis of glycolipids
The butanol-extractable glycolipids were treated with PI-PLC, GPIPLD, PLA2, nitrous acid (HNO2) and mild base (0.1 N KOH). Glycolipids were dried and redissolved in 100 ,ul 0.1 M Tris-HCl (pH 7.4), 0.1% sodium deoxycholate containing 1 U Bacillus cereus P1-PLC (Boehringer-Mannheim) or 10 RI normal rabbit serum as a source of GPI-PLD (Davitz et al., 1988). After incubation for 12 h at 370C, the reaction was terminated by adding 2 gl acetic acid. Bee venom PLA2 (Boehringer-Mannheim) treatment (50 U) was performed in the same buffer as described for PI-PLC supplemented with I mM CaCl2 overnight at 37°C. For HNO2 deamination, glycolipids were dried and dissolved in 0.1 M sodium acetate (pH 3.5) containing 0.1% SDS/0.5 M NaNO2 (v/v), and incubated for 4 h at room temperature. The reaction was stopped by the addition of 5 g. acetic acid. For mild base treatment, 50% aqueous glycolipids were dissolved in 200 gl 0.1 N KOH intreatments were methanol and incubated for 4 h at 50°C. The enzymatic stopped by heating the reaction mixtures to 100°C for 1 min. After each
4431
N.Azzouz at aL. chemical or enzymatic treatment, the reaction mixtures were partitioned between water and water-saturated n-butanol. The cleavage rate was estimated by TLC analysis and/or liquid scintillation counting of the organic and aqueous phases. For HPLC (Mono Q anion-exchange chromatography) analysis, the deamination step (HNO2 treatment) was carried out in the presence of 0.05% octyl glucoside (Sigma) instead of 0.1% SDS.
Generation and analysis of the neutral glycans
The TLC-purified glycolipids were dephosphorylated, deaminated and reduced according to Mayor and Menon (1990). Glycolipids were dephosphorylated by treatment with ice-cold 48% aqueous HF for 60 h at 0°C. After neutralization with frozen saturated LiOH, the reaction mixture was desalted, deaminated by HNO2 treatment for 4 h, and reduced by sodium borohydride for 5 h. The resulting material was desalted on AG3X3 (OH-) and AG50X12 (H+) tandem ion-exchange columns and filtered through a 0.2 tm filter. The neutral glycans were analysed by HPAEC and Bio-Gel P4 size-exclusion chromatography.
HPAEC analysis of labelled neutral glycans
Neutral glycans were analysed by HPAEC on a Dionex Basic Chromatography System. The analysis was accomplished using a gradient elution program (Mayor and Menon, 1990) on a carbopakTM PAl (4X250 mm) column [100% buffer A (0.1 M NaOH), 0% buffer B (0.1 M NaOH, 0.25% M NaOAc) to 6 min after injection, then an increase of buffer B to 30% over 36 min with a flow rate of 1 mlmin]. Co-injected glucose oligomer standards were detected by pulsed amperometric detection.
Bio-Gel P4 analysis
hydrophilic moieties of glycolipids generated by HNO2 treatment were sized on a Bio-Gel P4 column (I x 130 cm, -400 mesh) equilibrated and eluted with 0.2 M ammonium acetate containing 0.02% sodium azide (Masterson et al., 1989). Fractions were collected at a rate of one fraction per 24 min (850 pu). Glucose oligomers from partially hydrolysed dextran were included as internal standards and detected by oxidation after adding an aliquot of 25 gl from each fraction to 100 gl orcinol (2 mg/ml in concentrated sulfuric acid). Radioactivity was monitored by liquid scintillation counting.
The
Preparation of GPI standards from T.brucei
Washed preparations of lysed trypanosomes (Tbrucei MIT variant clone 118) purified from infected rat blood were labelled in the presence of tunicamycin with GDP-[3H]mannose (Masterson et al., 1989). After labelling, the Tbrucei variable surface glycoprotein (VSG) GPI-anchor precursor, termed P2, and its biosynthetic intermediates were extracted and purified according to Menon et al. (1988). The neutral glycans were generated as described above.
Mono 0 anion-exchange chromatography
hydrophilic moieties of glycolipids generated by HNO2 treatment chromatographed on a Mono Q resource column (Pharmacia) linked to an HPLC system (Waters). Separation was accomplished by gradient elution as described by Kamitani et al. (1992): 100% water, 0% 0.5 M NH4OAc, pH 5.5, to 5 min after sample injection, followed by a linear gradient to 75% water, 25% NH4OAc over 30 min, then to 50% water, 50% NH4OAc to 40 min, and finally to 100% NH4OAc to 45 min. The flow rate was 1 ml/min, and 1 ml fractions were collected for liquid scintillation counting. a-Mannosidase digestion The
were
The neutral glycans or the water-soluble glycans obtained after deamination were dissolved in 100 gl 50 mM sodium acetate (pH 4.5) containing 0.2 mM ZnCl2, 0.025 sodium azide and 2 U Jack bean a-mannosidase (Sigma). They were incubated at 370C for 24 h. For A.saitoi a-mannosidase (Oxford glycosystem), digestion was performed in 30 ,ul 0.1 M sodium acetate (pH 5.0) containing 10 gU of enzyme followed by incubation for 14 h at 37°C. Digestions were stopped by heating the reaction mixtures to 100°C for 3 min.
Alkaline phosphatase digestion
The water-soluble fragment obtained after HNO2 deamination of glycolipids was first treated with Jack bean a-mannosidase. The resulting fragment was purified and desalted on a Bio-Gel P2 (1 X45 cm) column equilibrated and eluted with water. Calf intestine alkaline phosphatase (Boehringer-Mannheim) treatment was performed in 40 p1 25 mM TrisHCI, pH 8.9, containing 10 mM magnesium chloride, 0.1 mm zinc chloride and 10 U enzyme. The reaction mixture was then passed through
4432
a column (1 ml) of AG3X3 (OH-) and AGSOXl2 remove the uncleaved material.
(H+) to desalt and
Sphingomyelinase treatment Glycolipids or the Pprimaurelia 156G SAg protein were incubated with sphingomyelinase from Staphylococcus aureus (Sigma) according to Morrison (1969). 10 gl of Triton X-100 were added to the dissolved sample in C/M (2:1 by volume). The sample was dried and dissolved in 20 pl chloroform, 300 ,ul diethylether, 40 pl 1 M Tris-acetate, pH 7.6, and 20 pl 0.4 M MgCl2 in the presence of 1 U enzyme. Samples were incubated at 37°C overnight under constant stirring. The organic solvent was evaporated under nitrogen. The ceramide was extracted with diethylether and analysed on HPTLC plates with C/M (9:1 by volume) as the solvent system.
Acknowledgements The authors would like to thank Drs L.P.Aggerbeck, V.Eckert and L.Sperling for their critical reading of this manuscript, and M.Eppinger and A.Eichhorn for technical assistance. This work was supported by Deutsche Forschungsgemeinschaft (SFB 286), Fonds der Chemischen Industrie and Stiftung P.E.Kempkes. N.A. acknowledges support by EMBO and PROCOPE from ANRT/DAAD for fellowships. B.S. thanks the Friedrich-Ebert-Stiftung for a fellowship. P.G. thanks the Hessische Graduiertenfordung for a fellowship.
References Azzouz,N. and Capdeville,Y. (1992) Structural comparisons between the soluble and the GPI-anchored forms of the Paramecium temperaturespecific 156G surface antigen. Bio. Cell, 75, 217-223. Azzouz,N., Ranck,J.L. and Capdeville,Y. (1990) Purification of the temperature-specific surface antigen of Paramecium primaurelia with its glycosyl-phosphatidylinositol membrane anchor. Protein Exp. Purif., 1, 13-18. Bangs,J.D., Doering,T.L., Englund,P.T. and Hart,G.W. (1988) Biosynthesis of a variant surface glycoprotein of Trypanosoma brucei. Processing of the glycolipid membrane anchor and N-linked oligosaccharides. J. Biol. Chem., 263, 17697-17705. Beale,G.H. (1952) Antigenic variation in Paramecium aurelia variety 1. Genetics, 37, 62-74. Beale,G.H. (1954) The antigens. In Salt,G. (ed.), The Genetics of Paramecium aurelia. Cambridge University Press, Cambridge, UK, pp. 77-123. Becker,G.W. and Lester,R.L. (1980) Biosynthesis of phosphoinositolcontaining sphingolipids from phosphatidylinositol by a membrane preparation from Saccharomyces cerevisiae. J. Bacteriol., 142, 747754. Brown,D. (1993) The tyrosine kinase connection: how GPI-anchored proteins activate T cells. Curr Opin. Immunol., 5, 349-354. Capdeville,Y. (1971) Allelic modulation in Paramecium aurelia heterozygotes. Mol. Gen. Genet., 112, 306-316. Capdeville,Y. (1979) Regulation of surface antigen expression in Paramecium primaurelia. (II) Role of the surface antigen itself. J. Cell Physiol., 99, 383-394. Capdeville,Y., Cardoso de Almeida,M.L. and Deregnaucourt,C. (1987) The membrane-anchor of Paramecium-specific surface antigens is a glycosylinositol phospholipid. Biochem. Biophys. Res. Commun., 147, 1219-1225. Capdeville,Y., Charret,R., Antony,A., Delorme,J., Nahon,P. and Adoutte,A. (1993) Ciliary and plasma membrane proteins in Paramecium: description, localisation, and intracellular transit. In Plattner,H. (ed.), Advances in Cell and Molecular Biology of Membranes. JAI Press, Hampton Hill, UK, Vol. 2A, pp. 181-226. Cohen,J. and Kerboeuf,D. (1993) Calcium and trichocyst exocytosis in Paramecium: genetic and physiological studies. In Plattner,H. (ed.), Advances in Cell and Molecular Biology of Membranes. JAI Press, Hampton Hill, UK, Vol. 2A, pp. 61-81. Conzelmann,A., Puoti,A., Lester,R.L. and Desponds,C. (1992) Two different types of lipid moieties are present in glycophosphoinositolanchored membrane proteins of Saccharomyces cerevisiae. EMBO J., 11, 457-466. Davitz,M.A., Hereld,D., Shark,S., Krakow,J., Englund,P.T. and
Nussenzweig,V. (1988) A glycan-phosphatidylinositol-specific phospholipase D in human serum. Science, 238, 81-84.
Fankhauser,C., Homans,S.W., Thomas-Oates,J.E., McConville,M.J.,
Paramecium primaurelia GPI-anchor Desponds,C., Conzelmann,A. and Ferguson,M.A.J. (1993) Structures of glycosylphosphatidylinositol membrane anchors from Saccharomyces cerevisiae. J. Biol. Chem., 268, 26365-26374. Ferguson,M.A.J., Low,M.G. and Cross,G.A.M. (1985) Glycosyl-sn-1,2dimyristylphosphatidylinositol is covalently linked to Trypanosoma brucei variant surface glycoprotein. J. Bio. Chem., 260, 14547-14555. Ferguson,M.A.J., Homans,S.W., Dwek,R.A. and Rademacher,T.W. (1988) Glycosyl-phosphatidylinositol moiety that anchors Trypanosoma brucei variant surface glycoprotein to the membrane. Science, 239, 753-759. Field,M.C. and Menon,A.K. (1993) Glycolipid anchoring of cell surface proteins. In Schlessinger,M.J. (ed.), Lipid Modification of Proteins. CRC Press, Boca Raton, USA, pp. 83-134. Forney,J.D., Epstein,L.M., Rudman,B.M., Wildmayer,D.J., Klein,W.H. and Pree,J.R. (1983) Structure and expression of genes for surface proteins in Paramecium. Mol. Cell. Biol., 3, 466-474. Gerold,P., Dieckmann-Schuppert,A. and Schwarz,R.T. (1994) Glycosylphosphatidylinositols synthesized by asexual erythrocytic stages of the malarial parasite, Plasmodium falciparum. J. Biol. Chem., 269, 2597-2606. Guther,M.L.S., Cardoso de Almeida,M.L., Yoshida,N. and Ferguson, M.A.J. (1992) Structural studies on the glycosylphosphatidylinositol membrane anchor of Trypanosoma cruzi 1G7-antigen. The structure of the glycan core. J. Biol. Chem., 267, 6820-6828. Haynes,P.A., Gooley,A.A., Ferguson,M.A.J., Redmond,J.W. and Williams,K.L. (1993) Post-translational modification of the Dictyostelium discoideum glycoprotein PsA. Glycosylphosphatidylinositol membrane anchor and composition of 0-linked oligosaccharides. Eur. J. Biochem., 216, 729-737. Hemandez,L.M., Ballou,L., Alvarado,E., Tsai,P.-K. and Ballou,C.E. (1989) Structure of the phosphorylated N-linked oligosaccharides from the mnn9 and mnnlO mutants of Saccharomyces cerevisiae. J. Biol. Chem., 264, 13648-13659. Homans,S.W., Ferguson,M.A.J., Dwek,R.A., Rademacher,T.W., Anand,R. and Williams,A.F. (1988) Complete structure of the glycosyl phosphatidylinositol membrane anchor of rat brain Thy- 1 glycoprotein. Nature, 333, 269-272. Kamitani,T., Menon,A.K., Hallaq,Y., Warren,C.D. and Yeh,E.T.H. (1992) Complexity of ethanolamine phosphate addition in the biosynthesis of glycosylphosphatidylinositol anchors in mammalian cells. J. Biol. Chem., 267, 24611-24619. Karson,E.M. and Ballou,C.E. (1978) Biosynthesis of yeast mannan. J. Biol. Chem., 253, 6484-6492. Liscovitch,M. and Cantley,L.C. (1994) Lipid second messengers. Cell, 77, 329-334. Masterson,W.J., Doering,T.L., Hart,G.W. and Englund,P.T. (1989) A novel pathway for glycan assembly: biosynthesis of the glycosylphosphatidylinositol anchor of trypanosome variant surface glycoprotein. Cell, 56, 793-800. Masterson,W.J., Raper,J., Doering,T.L., Hart,G.W. and Englund,P.T. (1990) Fatty acid remodeling: a novel reaction sequence in the biosynthesis of trypanosome glycosyl-phosphatidylinositol membrane anchors. Cell, 62, 73-80. Mayor,S. and Menon,A.K. (1990) Structural analysis of glycosylinositol phospholipid anchors of membrane proteins. Methods: Compan. Methods Enzymol., 1, 297-305. Mayor,S., Menon,A.K. and Cross,G.A.M. (1990) Glycolipid precursors for the membrane anchor of Trypanosoma brucei variant surface glycoproteins. II. Lipid structures of phosphatidylinositol-specific phospholipase C sensitive and resistant glycolipids. J. Biol. Chem., 265, 6174-6181. Mayor,S., Menon,A.K. and Cross,G.A.M. (1992) Galactose-containing glycosylphosphatidylinositols in Trypanosoma brucei. J. Biol. Chem., 267, 754-761. McConville,M.J. and Homans,S.W. (1992) Identification of the defect in lipophosphoglycan biosynthesis in a non-pathogenic strain of Leishmania major. J. Biol. Chem., 267, 5855-5861. McConville,M.J. and Ferguson,M.A.J. (1993) The structure, biosynthesis and function of glycosylated phosphatidylinositols in the parasitic protozoa and higher eukaryotes. Biochem. J., 294, 305-324. Menon,A.K., Mayor,S., Ferguson,M.A.J., Dusezenko,M. and Cross,G.A.M. (1988) Candidate glycophospholipid precursor for the glycosylphosphatidylinositol membrane anchor of Trypanosoma brucei varaint surface glycoproteins. J. Biol. Chem., 263, 1970-1977. Menon,A.K., Schwarz,R.T., Mayor,S. and Cross,G.A.M. (1990a) Cellfree synthesis of glycosyl-phosphatidylinositol precursors for the
glycolipid membrane anchor of Trypanosoma brucei variant surface glycoproteins. J. Biol. Chem., 265, 9033-9042. Menon,A.K., Mayor,S. and Schwarz,R.T. (1990b) Biosynthesis of glycosyl-phosphatidylinositol lipids in Trypanosoma brucei: involvement of mannosyl-phosphoryldolichol as the mannose donor. EMBO J., 9, 4249-4258. Menon,A.K., Eppinger,M., Mayor,S. and Schwarz,R.T. (1993) Phosphatidylethanolamine is the donor of the terminal phospho-
ethanolamine group in trypanosome glycosyl-phosphatidylinositols. EMBO J., 12, 1907-1914. Meyer,E., Caron,F. and Guiard,B. (1984) Bloking of in vitro translation of Paramecium messenger RNAs is due to messenger RNA primary structure. Biochimie, 66, 403-412. Mill,J.P. (1966) Phosphomannans and other components of flocculent and non-flocculent walls of Saccharomyces cerevisiae. J. Gen. Microbiol., 44, 329-341. Morrison,W.R. (1969) Polar lipids in bovine milk. I. Long-chain bases in sphingomyelin. Biochim. Biophys. Acta, 176, 537-546. Neidhardt,F.C., Bloch,P.L. and Smith,D.F. (1974) Culture medium for enterobacteria. J. Bacteriol., 119, 736-747. Robinson,P.J. (1991) Signal transduction by GPI-anchored membrane protein. Cell. Biol. Int. Rep., 15, 761-767. Ryals,P.E., Pak,Y and Thompson,G.A.,Jr (1991) Phosphatidylinositollinked glycans and phosphatidylinositol-anchored proteins of Tetrahymena mimbres. J. Biol. Chem., 266, 15048-15053. Scharma,C.B., Babczinski,P., Lehle,L. and Tanner,W. (1974) The role of dolicholmonophosphate in glycoprotein biosynthesis in Saccharomyces cerevisiae. Eur. J. Biochem., 46, 35-41. Schofield,L. and Hackett,F. (1993) Signal transduction in host cells by a glycosylphosphatidylinositol toxin of malaria parasites. J. Exp. Med., 177, 145-153. Schofield,L., Gerold,P., Schwarz,R.T. and Tachado,S. (1994) Signal transduction in host cells mediated by glycosylphosphatidylinositols of the parasitic protozoa, or why do the parasitic protozoa have so many GPI molecules? Braz. J. Med. Biol. Res., 27, 249-254. Sipos,G., Puoti,A. and Conzelmann,A. (1994) Glycosylphosphatidylinositol membrane anchors in Saccharomyces cerevisiae: absence of ceramides from complete precursor glycolipids. EMBO J., 12, 2789-2796. Sonnebom,T.M. (1974) Paramecium aurelia. In King,R.C. (ed.), Handbook of Genetics. Plenum Press, New York, Vol. 2, pp. 469-594. Stadler,J., Keenan,T.W., Bauer,G. and Gerisch,G. (1989) The contact site A glycoprotein of Dictyostelium discoideum carries a phospholipid anchor of a novel type. EMBO J., 8, 371-377. Stefanova,I., Horejsi,V., Ansotegui,I.J., Knapp,W. and Stockinger,H. (1991) GPI-anchored cell-surface molecules complexed to protein tyrosine kinases. Science, 254, 1016-1021. Tomavo,S., Dubremetz,J.F. and Schwarz,R.T. (1992) Biosynthesis of glycolipid precursors for glycosylphosphatidylinositol membrane anchors in a Toxoplasma gondii cell-free system. J. Biol. Chem., 267, 21446-21458. Vidugiriene,J. and Menon,A.K. (1993) Early lipid intermediates in glycosyl-phosphatidylinositol anchor assembly are synthesized in the ER and located in the cytoplasmic leaflet of the ER membrane bilayer. J. Cell Biol., 121, 987-996. Vidugiriene,J. and Menon,A.K. (1994) The GPI anchor of cell-surface proteins is synthesized on the cytoplasmic face of the endoplasmic reticulum. J. Cell Biol., 127, 333-341. Weinhart,U., Thomas,J.R., Pak,Y, Thompson,G.A.,Jr and Ferguson,M.A.J. (1991) Structural characterization of a novel glycosylphosphatidylinositol from the protozoan Tetrahymena mimbres. Biochem. J., 279, 605-608. Received on
April 18, 1995; revised on July 5,
1995
4433
