Structural characterisation of two forms of procyclic acidic repetitive protein expressed by procyclic forms of Trypanosoma brucei

J. Mol. Biol. (1997) 269, 529±547

Structural Characterisation of Two Forms of ProcyclicAcidic Repetitive Protein Expressed by ProcyclicForms of Trypanosoma brucei

Achim Treumann1, Nicole Zitzmann1, Andreas HuÈ lsmeier1

Alan R. Prescott1, Andrew Almond2, John Sheehan2 andMichael A. J. Ferguson1*

1Department of BiochemistryUniversity of Dundee, DundeeDD1 4HN, UK2Wellcome Trust Centre forCell-Matrix Research, School ofBiological Sciences, Universityof Manchester, Manchester, UK

Present address: A. Treumann, InAbbreviations used: AHM, 2,5-an

albumin; ESI-MS, electrospray ionisJBAM, jack bean a-mannosidase; PAvariant surface glycoprotein; PI, phpropanesulfonate; TX-100, Triton Xionisation time-of-¯ight mass spectrspectrometry; HRP, horseradish pe

0022±2836/97/240529±19 $25.00/0/mb

A procyclic acidic repetitive protein (PARP) fraction was puri®ed fromlong-term cultures of Trypanosoma brucei procyclic forms by a solvent-extraction and reverse phase chromatography procedure. The PARP frac-tion yielded small quantities of a single N-linked oligosaccharide withthe structure Mana1-6(Mana1-3)Mana1-6(Mana1-3)Manb1-4GlcNAcb1-4GlcNAc (Man5GlcNAc2). Fractionation of PARP on Con A-Sepharoserevealed that the majority (80 to 90%) of the PARP fraction did not bindto Con A and was composed of the parpAa gene product that containsrepeats of -Glu-Pro-Pro-Thr- (GPEET-PARP) and that lacks an N-glycosy-lation site. This form of PARP has not been previously identi®ed at theprotein-level. The minor Con-A-binding fraction was shown to be rich inthe previously described form of PARP, encoded by the parpAb and/orparpBa genes, that contains a -Glu-Pro- repeat domain (EP-PARP) and anN-glycosylation site. Analysis of longer and shorter-term culturessuggested that procyclic cells initially express predominantly EP-PARPthat is gradually replaced by GPEET-PARP. Both forms of PARP wereshown to contain indistinguishable glycosylphosphatidylinositol (GPI)membrane anchors, where the conserved GPI core structure is substitutedby heterogeneous sialylated branched polylactosamine-like structures thatare predicted to form a dense surface glycocalyx above which the polya-nionic -Glu-Pro-Pro-Thr- and -Glu-Pro- repeat domains are displayed. Thephosphatidylinositol (PI) component of the GPI anchor was shown to be amixture of 2-O-acyl-myo-inositol-1-HPO4-(sn-1-stearoyl-2-lyso-glycerol) and2-O-acyl-myo-inositol-1-HPO4-(sn-1-octadecyl-2-lyso-glycerol), where theacyl chain substituting the inositol ring showed considerable heterogeneity.Mass spectrometric and light scattering experiments both suggested anaverage mass of approximately 15 kDa for GPEET-PARP, with individualglycoforms ranging from about 12 kDa to 20 kDa, that is consistent with itsamino acid and carbohydrate composition. A measured translational diffu-sion coef®cient of 3.9 � 107 cm2 sÿ1 indicates that this molecule has a highlyelongated shape. The possible functions of these unusual glycoproteins arediscussed.

# 1997 Academic Press Limited

Keywords: Trypanosome; procyclic; glycoprotein;glycosylphosphatidylinositol; PARP*Corresponding author

stitute of Comprehensive Medical Science, Fujita Health University, Japan.hydromannitol ; APAM, Aspergillus phoenicis a-mannosidase; BSA, bovine serumation-mass spectrometry; HPTLC, high-performance thin layer chromatography;RP, procyclic acidic repetitive protein; PNGase F, peptide N-glycanase F; VSG,

osphatidylinositol; Chaps, 3-[(3-cholamidopropyl)-dimethylammonio]-1--100; Gu, gluose units; MALDI-TOF-MS, matrix-assisted laser desorptionometry; GPI, glycosylphosphatidylinositol; GC-MS, gas chromatography±mass

roxidase; FITC, ¯uorescein isothiocyanate.

971066 # 1997 Academic Press Limited

530 The Procyclic Acidic Repetitive Proteins of T. brucei

Introduction

The tsetse ¯y-transmitted African trypanosomeTrypanosoma brucei exists in the mammalian host asbloodstream forms that are covered with a densemonolayer of variant surface glycoprotein (VSG)molecules (Cross, 1990). These parasites undergomorphological and biochemical changes upontransformation into procyclic forms following in-gestion by the tsetse ¯y vector. These changes in-clude the loss of the VSG coat.

In 1987, two independent reports describedcDNA sequences corresponding to abundantmRNAs that were expressed predominantly in theprocyclic (insect-dwelling) forms of T. brucei(Roditi et al., 1987; Mowatt & Clayton, 1987). Thepredicted protein products were named procyclinand procyclic acidic repetitive protein (PARP), re-spectively. Shortly after that Richardson et al.(1988) puri®ed an immunodominant protein fromthe surface of procyclic trypanosomes that con-tained the -Glu-Pro- repeat domain predicted fromthe aforementioned cDNAs. This form of PARPwill be referred to as EP-PARP.

Trypanosomes typically contain groups of twoPARP genes each on four chromosomes (two co-pies of the A locus and one copy of each of the B1and B2 loci; for a review, see Clayton & Hotz,(1996). The B1 and B2 locus genes (parpBa andparpBb) encode two EP-PARP species that differfrom each other in terms of the number of EP-repeats and the presence (parpBa) or absence(parpBb) of an Asn-Xaa-Ser/Thr N-glycosylationsite. One of the A locus genes (parpAb) encodesanother EP-PARP species (with an N-glycosylationsite) and the other (parpAa) encodes a PARPspecies containing a -Gly-Pro-Glu-Glu-Thr- repeatdomain and no N-glycosylation site. This latterform PARP will be referred to as GPEET-PARP.Northern blot analysis (Mowatt et al., 1989) andin vitro translation of total procyclic mRNA(Mowatt & Clayton, 1988) have shown that at leastfour of the parp genes, including the GPEET-PARPparpAa gene, are simultaneously transcribed. How-ever, the expression of the GPEET-PARP at theprotein-level has not been described and the occu-pancy of the predicted N-glycosylation sites of theEP-PARP species has not been determined.

The EP-PARP proprotein is known to be exten-sively post-translationally processed. An N-term-inal signal sequence, that directs the nascentprotein to the endoplasmic reticulum, and a C-terminal signal peptide, that directs the addition ofa glycosylphosphatidylinositol (GPI) membraneanchor, are both removed from the primary trans-lation product (Richardson et al., 1988; Clayton &Mowatt, 1989, Ferguson et al., 1993). The GPI-anchor is unusually complex and contains largeamounts of Gal and GlcNAc, at least partly in theform of N-acetyllactosamine repeats (Fergusonet al., 1993). The anchor is also heavily sialylated,the sialic acid being added by a stage-speci®ctrans-sialidase (Engstler et al., 1993; Pontes de

Carvalho et al., 1993; Schenkman et al., 1994). Thelipid portion of the GPI anchor of EP-PARP hasbeen investigated following metabolic labellingwith 3H-myristic acid. These studies indicated thatthe phosphatidylinositol (PI) moiety is a 1-O-stear-oyl-2-lyso-phosphatidyl-(acyl)-inositol with mainlypalmitate (and some stearate) linked to inositol(Field et al., 1991a). The acyl group was shown tobe attached to the 2 and/or the 3-hydroxyl of themyo-inositol ring (Ferguson, 1992a). The acylationof the inositol accounts for the resistance of the an-chor to the action of bacterial phosphatidylinositol-speci®c phospholipase C (PI-PLC) and trypano-some GPI-speci®c phospholipase C (GPI-PLC;Field et al., 1991a; Roberts et al., 1988). Based on1H-NMR studies of an EP-repeat containing pep-tide (Evans et al., 1986) and molecular modelling(Roditi et al., 1989), the Pro residues of the EP-re-peat domain are predicted to be in an all trans con-formation, restricting the repeat domain to a rigidand extended structure with a diameter of 0.9 nmand a rise of 0.29 nm per residue.

Here, we characterise PARP puri®ed by solventextraction and reverse phase liquid chromatog-raphy from procyclic trypanosomes and show thatGPEET-PARP is expressed at the protein-level bycells from long-term cultures. We present the struc-ture of the N-linked oligosaccharide of EP-PARPand the use of Concanavalin A lectin af®nity chro-matography for the separation of GPEET-PARPfrom EP-PARP. New data on the structure of theGPI anchor of GPEET-PARP are shown, togetherwith mass spectrometric and light scattering stu-dies on GPEET-PARP.

Results

Extraction of PARP from T. bruceiprocyclic cells

T. brucei (strain-427) procyclic cells that hadbeen in continuous culture for about 16 monthswere extracted as shown in Figure 1. Samples ofthe 2% (w/v) 3-[(3-cholamidopropyl)-dimethylam-monio]-1-propanesulfonate (Chaps) total cell ly-sate, the chloroform/methanol extract, the butan-1-ol wash of the 9% (v/v) butan-1-ol extract anda 2% Chaps extract of the ®nal insoluble pelletwere analysed by Triton-X100 PAGE followed bystains-all staining. The characteristic turquoise-staining PARP bands (data not shown) were ex-cised and extracted with formamide and thePARP was quantitated by spectrophotometry ofthe stains-all dye (Ziegelbauer et al., 1990). Theabsorbance values were adjusted to molar valuesof PARP via a calibration curve based on knownquantities of pure PARP (measured by myo-inosi-tol content) run on the same gel. The result indi-cated that: (1) no PARP is lost in the chloroform/methanol extract or in the butan-1-ol wash of the9% butan-1-ol extract, (2) less than 6% of thetotal PARP remains in the insoluble pellet and (3)

Figure 1. PARP puri®cation scheme. Summary of thepuri®cation steps used to produce octyl-Sepharose-puri-®ed PARP (OS-puri®ed PARP). Aliquots of the boxed-fractions were analysed for their PARP content by Tri-ton-PAGE and stains-all staining (see the text).

The Procyclic Acidic Repetitive Proteins of T. brucei 531

that there are approximately 2.2 � 106 PARP mol-ecules per cell.

The 9% butan-1-ol extract (containing > 94% ofthe total PARP) was fractionated on an octyl-Sepharose column. The PARP-containing fractions,which eluted at around 27% (v/v) propan-1-ol,were detected by orcinol-staining of aliquots of thecolumn fractions and by the presence of myo-inosi-tol, as measured by gas chromatography±massspectrometry (GC-MS), (data not shown). The ®nalyield of puri®ed PARP, based on myo-inositol con-tent, was 410(�65) nmol per 1011 cells (mean ofthree separate puri®cations). This ®gure suggests atotal copy number of (2.6 � 0.4) � 106 PARP mol-ecules per cell, in reasonable agreement with the®gure deduced from stains-all staining.

Characterisation of the N-linkedoligosaccharides of PARP

Monosaccharide analysis of the puri®ed N-linked oligosaccharides released from 10 nmol ofPARP revealed the presence of 2 nmol Man and atrace of GlcNAc. The composition suggested thepresence of oligomannose structures and the low-yield suggested that only a small proportion of thePARP was N-glycosylated.

The NaB3H4-reduced N-linked oligosaccharitolfraction was subjected to Bio-Gel P4 chromatog-raphy. Two radioactive peaks were observed, oneat the void volume and one at 9.0 glucose units(9.0 Gu; data not shown). Samples of the void vo-lume material and the 9.0 Gu material were takenfor strong acid hydrolysis, N-acetylation andHPTLC analysis and the results showed that onlythe 9.0 Gu peak contained authentic 3H-labelled N-linked oligosaccharitol material (see Materials andMethods). The 9.0 Gu hydrodynamic volume ofthe PARP N-linked oligosaccharitol was identicalto that of an authentic standard of Mana1-6(Mana1-3)Mana1-6(Mana1-3)Manb1-4GlcNAcb1-4[1-3H]GlcNAc-ol (Man5GlcNAc2-ol) and bothstructures also co-chromatographed on Dionexcarbohydrate HPLC (eluting at 2.7 Du, data notshown).

The co-chromatography of the PARP N-linkedoligosaccharitol with the Man5GlcNAc2-ol standardsuggested that they were identical. This was con-®rmed by HPTLC analysis of the PARP N-linkedoligosaccharitol, alongside the Man5GlcNAc2-olstandard, before and after acetolysis (that selec-tively cleaves Mana1-6Man bonds) and digestionwith Aspergillus phoenicis (Mana1-2Man-speci®c)a-mannosidase (APAM) and jack bean a-mannosi-dase (JBAM; Figure 2). The two samples co-mi-grated on the HPTLC plate (Figure 2, lanes 5 and6) and the PARP glycan was shown to be resistantto APAM (Figure 2, lane 7), consistent with ab-sence of any terminal a1-2-linked Man residues.Both glycans gave identical digestion productswith JBAM (Figure 2, lanes 8 and 9). Analysis ofthese digests by Bio-Gel P4 (data not shown) re-vealed two peaks at 5.5 and 4.5 Gu, consistent withthe formation of the expected product (i.e. Manb1-4GlcNAcb1-4[1-3H]GlcNAc-ol, 5.5 Gu) and the pro-duct GlcNAcb1-4[1-3H]GlcNAc-ol (4.5 Gu) thatwas presumably generated by a trace of b-manno-sidase activity in the JBAM. Acetolysis of the Man5-

GlcNAc2-ol standard produced a major productmigrating slightly slower than Manb1-4GlcNAcb1-4[1-3H]GlcNAc-ol that is presumably Mana1-3Manb1-4GlcNAcb1-4[1-3H]GlcNAc-ol (Figure 2,lane 10) and the same major product was observedupon acetolysis of the PARP glycan (Figure 2, lane11).

Taken together, the data described aboveestablish that PARP expresses only one N-linkedoligosaccharide structure, namely Mana1-6(Mana1-3)Mana1-6(Mana1-3))Manb1-4GlcNAcb1-4GlcNAc,and that only a small proportion of the PARP mol-ecules in this preparation were N-glycosylated.Given the Man5GlcNAc2 composition of the N-linked oligosaccharide, the yield of 2 nmol Manfrom 10 nmol of PARP would suggest that only 4%of the PARP in this preparation was N-glycosy-lated (although this ®gure does not take into ac-count any losses during oligosaccharide releaseand puri®cation).

Figure 2. Analysis of the N-linkedoligosaccharides released fromPARP. HPTLC analysis and ¯uoro-graphy of the PARP 9.0 Gu oligo-saccharitol (lane 6) alongside anauthentic standard of Man5-

GlcNAc2-ol (lane 5). The productsof the A. phoenicis a-mannosidase(APAM), jack bean a-mannosidase(JBAM) and acetolysis (Ac2O)digests of the PARP 9.0 Gu oligo-saccharitol are shown in lanes 7, 9and 11, respectively. The productsof the jack bean a-mannosidaseand acetolysis digests of the Man5-

GlcNAc2-ol standard are shown inlanes 9 and 10, respectively. A stan-dard of Mana1-2Mana1-2Mana1-6Mana1-4[1-3H]AHM (Man4AHM)(lane 1) was used as a positive con-trol for the APAM, JBAM andAc2O procedures (lanes 2 to 4), asindicated.


Concanavalin A affinity chromatographyof PARP

In order to separate N-glycosylated PARP fromnon-N-glycosylated PARP, material was fractio-nated on a Con A-Sepharose column. About 80%of the material did not bind (the ConAÿ fraction)and the remaining 20% (the ConA� fraction) waseluted at the start of a gradient of a-methyl-man-noside (data not shown). After repuri®cation onoctyl-Sepharose, aliquots of the two fractions weresubmitted to myo-inositol, amino acid and carbo-hydrate analysis and to N-terminal peptide se-quencing (Table 1). The amino acid composition ofthe ConAÿ fraction gave a Glx:Pro ratio of 1.63:1that was not consistent with the predicted aminoacid composition of EP-PARP encoded by theparpAb and/or parpBa and/or parpBb genes(Roditi et al., 1987; Mowatt & Clayton, 1988;Clayton & Mowatt, 1989; KoÈnig et al., 1989;Rudenko et al., 1989, 1990; Dorn et al., 1991;Jackson et al., 1993; see Table 1). However, theamino acid composition was similar to that pre-dicted for the product of the parpAa gene (GPEET-PARP) that contains -Gly-Pro-Glu-Glu-Thr-(GPEET) repeats (Mowatt et al., 1989). The major(ConAÿ) fraction produced only one N-terminal se-quence, corresponding to the predicted N-terminalsequence of GPEET-PARP, whereas the (ConA�)fraction produced two major amino acid deriva-tives per cycle. The sequence of residues was con-sistent with the presence of approximately equalamounts of EP-PARP and GPEET-PARP. A minorseries of amino acid derivatives was also detected(Table 1), consistent with an internal sequence ofGPEET-PARP.

The identities of the PARPs in the two ConAfractions were con®rmed by dot-blot immuno-staining which showed that the ConAÿ PARP frac-

tions stained only with anti-GPEET-PARP anti-bodies, whereas the ConA� PARP fractions stainedwith anti-GPEET-PARP and anti-EP-PARP anti-bodies (data not shown). Samples of the ConAÿ

PARP and the ConA� PARP fractions were ana-lysed by SDS-PAGE and stains-all staining along-side a standard of EP-PARP, provided by Dr TerryPearson, University of Victoria. The ConAÿ

GPEET-PARP displayed signi®cantly faster mi-gration in the gel (apparent molecular mass 20 to35 kDa) compared with EP-PARP (apparent mol-ecular mass 37 to 47 kDa) and the presence of bothforms of PARP in the ConA� fraction was seen(data not shown). These data suggest that the ma-jority (approximately 90%) of the PARP expressedby these particular procyclic cells is GPEET-PARP.Both forms of PARP migrate as broad smears onthe gel, indicating substantial heterogeneity in theircarbohydrate components.

Immunofluorescence microscopy ofT. brucei procyclics

Procyclic cells used in this study, and Nour6cprocyclic cells that lack all parp genes except forthe Aa GPEET-PARP gene (a generous gift fromDr Isabel Roditi), were stained with an anti-EP-PARP monoclonal antibody that recognises the EP-repeat domain (Richardson et al. 1988) and with apolyclonal antibody raised to a synthetic GPEET-containing peptide (also a gift from Dr Isabel Rodi-ti). Immuno¯uorescence microscopy indicated thatall procyclic cells were positive for EP-PARP andGPEET-PARP and that there were no discretepopulations of cells expressing exclusively EP-PARP or GPEET-PARP (Figure 3a and b). TheNour6c cells (Figure 3c and d) serve as a controlfor the speci®city of the anti-EP-PARP monoclonalfor EP-PARP. Interestingly, the immuno¯uorescent

Table 1. Experimental and theoretical sequences and compositions of PARP preparations and parp gene products

Experimental Theoreticalparp parp parp parp parp

ConAÿ ConA� Aa Ab B1a B1b iso-B1a

Sequence VIVKG AEGPE VIVKG AEGPE AEGPE AEGPE AEGPEGKGKE DKGLT GKGKE DKGLT DKGLT DKGLT DKGLTR REDGP KGGKE KGGKE KGGKE KGGKE

----- EEPEE KGGKE KGEKG KGEKE KGEKGTGPEE TKVSD TKVGA TKVQD TKVGA

VIVKG TGPEE DDTNG DDTNG EVEPE DDTNGGKGKE TGPEE TDPDP TDPDP PEPEP TDPDPRE TGPEE EPEPE EPEPE EPEPE EPEPE

TGPEE PEPEP PEPEP PEPEP PEPEP----- TGPEE PEPEP PEPEP PEPEP PEPEP

TEPEP PEPEP PEPEP PEPEP PEPEPG G EPEPE EPEPE EPEPE EPEPE

GKGKE PEPEP PEPEP PEPEP PEPEPREDGP EPEPE EPEPE EPEPE G

PEPEP PEPEP PEPEPEPEPG EPEPE EPG

PEPEPEPEPEPEPG

Ala 1.7 � 0.1 1.7 � 0.2 0 1 2 1 2Cys n.d. n.d. 0 0 0 0 0Asx 3.5 � 0.3 4.0 � 0.5Asp 1 7 6 2 6Asn 0 1 1 0 1Glx 17.6 � 1.4 16.0 � 1.9Glu 20 24 32 31 18Gln 0 0 0 1 0Phe 0.9 � 0.2 0.7 � 0.2 0 0 0 0 0Gly 10.8 � 0.9 7.7 � 1.1 11 10 10 7 10His 0.4 � 0.1 0.3 � 0.0 0 0 0 0 0Ile 1.0 � 0.2 0.7 � 0.1 1 0 0 0 0Lys 3.2 � 0.5 3.2 � 0.5 3 6 6 6 6Leu 1.4 � 0.2 1.2 � 0.2 0 1 1 1 1Met 0.6 � 0.3 0.4 � 0.1 0 0 0 0 0Pro 10.8 � 1.3 13.9 � 1.1 10 25 32 26 18Arg 1.6 � 0.1 0.8 � 0.6 1 0 0 0 0Ser 0.0 0.0 0 1 0 0 0Thr 3.9 � 0.2 3.3 � 0.3 7 4 4 2 4Val 1.6 � 0.2 1.2 � 0.1 2 1 1 2 1Tyr 0.6 � 0.1 0.6 � 0.2 0 0 0 0 0GlcN 26.9 � 3.5 22.0 � 5.7Etn 2.0 � 0.2 1.6 � 0.2MWt (av.) 5874 8464 10045 8527 6878E/G/P 1.63/1/1 2.1/1/1/1.8 1.8/1/1.1 3.2/1/3.2 2.4/1/2.5 4.3/1/3.7 1.8/1/1.8Man 1.1 � 0.6 3.0 � 0.6Gal 16.4 � 9.0 11.9 � 4.0GlcNAc 13.0 � 6.4 10.5 � 4.6Sia 6.3 � 0.5 6.6 � 1.6myo-lno 1.00 1.00

The amounts of amino acids and carbohydrates are expressed relative to myo-inositol. The experimentally determined compositionalvalues for the ConA� and ConAÿ fraction are the averages of ®ve or six independent determinations (�snÿ1). The sequences of theparp genes are taken from Mowatt et al. (1989, parpAa), Roditi et al. (1987, parpAb), Mowatt et al. (1989, parpB1a), Rudenko et al.(1989, parpB1b) and KoÈnig et al. (1989, parp isoB1a).


staining with anti-EP antibodies is uniform(Figure 3b) whereas the staining with anti-GPEET-PARP is punctate (Figure 3a and c). Furthermore,despite the paucity of EP-PARP (and the relativeef®ciency of anti-GPEET PARP antibodies in im-muno-blots) the anti-EP-PARP immuno¯uorescentsignal was stronger than that due to the anti-GPEET PARP antibodies, regardless of the ¯uoro-phore attached to the second antibody.

Changes in the expression of EP-PARPand GPEET-PARP

The predominant expression of GPEET-PARP,described above, does not appear to be a generalproperty of all T. brucei strain-427 procyclic cell cul-tures. A Western blot of whole cell lysates of thestrain-427 procyclic culture used in this study(made by in vitro transformation of strain-427 var-

Figure 3. Immuno¯uorescence microscopy of procyclicT. brucei cells. Procyclic T. brucei strain-427 (a, b, e, f) orthe deletion-mutant Nour6c expressing only GPEET-PARP (c, d) were ®xed in 4% paraformaldehyde andstained with rabbit anti-GPEET-PARP antibody (a, c)and mouse anti-EP-PARP antibody (b, d). Bound anti-bodies were detected by double staining with a ¯uor-escein isothiocyanate-coupled anti-rabbit IgG secondantibody (detecting the anti-GPEET antibodies; a, c, e)and with a Texas Red-coupled anti-mouse IgG secondantibody (detecting the anti-EP antibodies (b, d, f). eand f show procyclic strain-427 cells which have beenstained with the second antibodies alone.

Figure 4. Analysis of PARP species expressed by differ-ent cultures of strain-427 procyclics. (a) Western blots ofprocyclic cell lysates. SDS sample buffer lysates ofstrain-427 procyclics cultured for 34 months in Scotland(lanes 1 and 3) or cultured in Switzerland (lanes 2 and4) were separated by SDS-PAGE, transferred to PVDFmembrane and stained with rabbit anti-GPEET-PARPantibody (lanes 1 and 2) and mouse anti-EP-PARP anti-body (lanes 3 and 4). Bound antibodies were detectedusing horse radish peroxidase labelled second antibodiesand ECL reagents. The positions of molecular weightmarkers are indicated on the left. (b) Stains-all stainingof PARP species in Chaps lysates of 107 strain-427 pro-cyclics cultured in for one month (lane 3) and Nour6cprocyclics (lane 2) run alongside approximately 20 pmolpuri®ed EP-PARP (lane 4) and 50 pmol puri®ed GPEET-PARP (lane 1).


iant MITat1.4 bloodstream form trypanosomesfrom rat blood) and a strain-427 procyclic culturegrown in another laboratory (supplied by Dr IsabelRoditi, University of Berne), is shown in Figure 4(a)to emphasise this point. In this experiment, bothEP-PARP and GPEET-PARP are clearly evident inthe Swiss culture (Figure 4(a), lanes 2 and 4)whereas the Scottish culture contains GPEET-PARP but insuf®cient EP-PARP for immuno-detec-tion (Figure 4(a), lanes 1 and 3).

The procyclic lysate used in Figure 4(a) (lanes 1and 3), that lacks detectable EP-PARP, was pre-pared from the same continuous culture of procyc-lic cells that was used to prepare the ConAÿ and

ConA� PARP fractions, described above, that docontain detectable EP-PARP. However, the lattercells were in continuous culture for 16 monthswhereas the former were in continuous culture for34 months. This tentatively suggests that EP-PARPmay be gradually replaced by GPEET-PARPduring long-term continuous culture. This sugges-tion received some support when a fresh stabilateof the same procyclic cells (frozen within a fewdays of transformation from bloodstream forms)was placed into culture for 1 month and thePARP-content analysed by a modi®ed SDS-PAGEprocedure and stains-all staining (Figure 4(b), lane3). In this case, the strain-427 procyclics expressedalmost exclusively EP-PARP.

The modi®ed SDS-PAGE procedure described inFigure 4(b), that uses Chaps sample buffer insteadof SDS sample buffer for making the total cell ly-sate, should prove useful for routine monitoring ofthe PARP species in whole cells. Stains-all stainingof SDS sample buffer whole cell lysates does notreveal clear PARP bands, presumably because ofother interfering components in these lysates (datanot shown). This problem was previously solved


by (Ziegelbauer et al., 1990) by performing TritonX-100-PAGE, instead of SDS-PAGE, where onlythe highly-acidic PARP molecules enter the run-ning-gel. However, Triton X-100-PAGE does notresolve EP-PARP from GPEET-PARP. Thus theChaps-lysate/SDS-PAGE system has the advan-tage of simultaneously preventing interferencewith the stains-all staining of PARP while resol-ving EP- and GPEET-PARP species.

Analysis of the GPI anchors of GPEET-PARPand EP-PARP

The PI moieties

It was not possible to obtain a pure preparationof intact EP-PARP from our procyclic cell cultures,even after rechromatography of the ConA� fractionon ConA-Sepharose. The PI analyses describedbelow were therefore carried out on either pureConAÿ GPEET-PARP (Figure 5a and b) or onoctyl-Sepharose puri®ed PARP (90% GPEET-PARP, 10% EP-PARP) (Figure 5c).

The PARP samples were deaminated with ni-trous acid and the released PI components wereextracted into butan-1-ol, dried, redissolved inchloroform/methanol and analysed by electro-spray ionisation mass spectrometry (ESI-MS). Thenegative ion spectrum of the GPEET-PARP derivedsample (Figure 5a) shows the presence of eightpseudomolecular ions corresponding to differentPI species. One of these species (the m/z 837 ion)was investigated by ESI-MS-MS. The daughter ionspectrum (Figure 5b) is consistent with a 2-O-pal-mitoyl-myo-inositol-1-HPO4-(sn-1-stearoyl-2-lyso-glycerol), where the carboxylate fragment ion fromthe glycerolipid (stearate at m/z 283) is character-istically much stronger than the barely visible car-boxylate fragment ion from the 2-position of theinositol ring (palmitate at m/z 255; Treumann et al.,1995). The low abundance of the m/z 241 [inositol-1,2-cyclic phosphate]ÿ ion (an intense daughter ionin conventional PI fragmentation spectra) is con-sistent with the palmitate being located at the 2-position of the inositol ring (Treumann et al., 1995).

A complete assignment of all eight molecular ionspecies was possible by cone voltage induced frag-mentation. Using this technique, all of the pseudo-molecular ions undergo fragmentation in theelectrospray source prior to entering the mass spec-trometer. Thus all the parent and daughter ions areobserved simultaneously (Figure 5c). The daughterions can be related to speci®c parent ions by takinginto account the mass differences between them.For example, the daughter ions at m/z 507, 505,503 and 479 (Figure 5c, inset) can be assigned asthe daughters of the parent ions at m/z 865, 863,861 and 837, respectively. These daughter ions canonly originate from the structure shown inFigure 5d through breakage of the ester bond be-tween the phosphate and the glycerol, with thenegative charge being retained by the phosphate-containing fragment. This shows that the observed

acyl chain heterogeneity of the PARP PI-moiety re-sides in the acyl chain attached to the 2-position ofthe inositol ring. Also, the presence of these ionsexcludes the possibility that the heterogeneity ofthis PI fraction (Figure 5a) is caused by contami-nation with cellular PI species as acylation of theinositol is a GPI-anchor-speci®c modi®cation(McConville & Ferguson, 1993). The pseudomole-cular ions at m/z 851, 849, 847 and 823 are mostlikely due to an octadecyl alkyl chain attached tothe glycerol instead of the stearoyl acyl chain ob-served for the major species. This notion is sup-ported by the stronger intensity of the ion at m/z405 when compared to that at m/z 419. Assumingboth ions correspond to a lysophosphatidate frag-ment ion, the alkyl linkage of sn-1-octadecyl-2-lyso-phosphatidate (m/z 405) can be expected to bemore stable than the acyl linkage of sn-1-stearoyl-2-lyso-phosphatidate (m/z 419).

Based on the aforementioned assignments, therelative proportions of the various PI species foundin GPEET-PARP are listed in Table 2.

The GPI glycans

The compositional analyses of both PARP frac-tions (Table 1) showed the presence of sialic acid,Gal, GlcNAc and Man, with more Man in the EP-PARP enriched fraction, consistent with the pre-sence of the oligomannose N-linked glycan only onEP-PARP. The presence of approximately equalamounts of Gal and GlcNAc suggested the possi-bility of polylactosamine structures attached to theGPI anchors, as previously reported for T. bruceirhodesiense EP-PARP (Ferguson et al., 1993). Thepresence of sialic acid is consistent with PARPbeing the major sialic acid acceptor for the cell sur-face trans-sialidase of T. brucei (Engstler et al., 1993;Pontes de Carvalho et al., 1993; Schenkman et al.,1994). The nature of the GPI glycans from GPEET-PARP and EP-PARP was studied further by pre-paring GPI neutral glycan fractions from each. Un-fractionated PARP was subjected to nitrous aciddeamination and NaB3H4 reduction to introduce alabel into the GPI glycans by converting the GlcNresidue to [1-3H]2,5-anhydromannitol (AHM). Thismaterial was then fractionated by ion exchangechromatography (Figure 6a) and the two pools (Aand B) were further fractionated on ConA-Sepha-rose (Figure 6b and c). Analysis of this radio-labelled material by SDS-PAGE and ¯uorographyrevealed that this procedure produced essentiallypure fractions of labelled GPEET-PARP and EP-PARP (Figure 6d). These fractions were desialyl-ated with neuraminidase and the GPI neutral gly-cans were released from the PARP proteincomponents by aqueous HF dephosphorylation.

Chromatography of the GPI neutral glycans on aBio-Gel P4 gel-®ltration column, before and afterdigestion with Bacillus fragilis endo-b-galactosidase(Figure 7a and b), showed that the two fractionswere indistinguishable and that the endo-b-galacto-sidase caused a signi®cant reduction in the average

Figure 5 (legend opposite)


JMB MS1066 [4/6/97]

Table 2. Relative proportions of the PARP PI speciesobserved for the ConAÿ fraction

[M-H]ÿ Proportionm/z R1 R3 ConAÿ (%)

823 C18:0-alkyl C16:0 7.2847 C18:0-alkyl C18:2 4.2849 C18:0-alkyl C18:1 4.1851 C18:0-alkyl C18:0 6.3837 C18:0-acyl C16:0 20.6861 C18:0-acyl C18:2 15.3863 C18:0-acyl C18:1 19.8865 C18:0-acyl C18:0 22.6

The R1 and R3 groups refer to Figure 5d. The relative propor-tion of each species was estimated by integration of the ionintensities shown in Figure 5a.

Figure 6. Fractionation of deaminated and reducedGPEET-PARP and EP-PARP by ion-exchange chroma-tography and ConA-Sepharose chromatography. a, Dea-minated, NaB3H4-reduced PARP was fractionated on aDEAE Sephacel anion exchange column. The am-monium acetate (NH4)Ac) gradient is indicated with abroken line. b and c, Pool A (b) and pool B (c) from awere applied to a Con A-Sepharose column. Thea-methyl mannoside (a-MeMan) gradient for eluting thecolumn is indicated with a dashed line. d, Pools A to Fwere analysed by SDS PAGE and ¯uorography. Lessmaterial was loaded in lane C than in the other lanes.The positions of molecular mass markers are indicatedon the left.


hydrodynamic volume of the neutral glycans. Theinability of the endo-b-galactosidase to produce asmall number of relatively small neutral glycans isthought to be due to the presence of branchingpoints within the polylactosamine domain that areresistant to this enzyme (Scudder et al., 1987). Sup-port for this view was obtained by digestion of atotal PARP GPI neutral glycan fraction with amixture of bovine testes b-galactosidase and jackbean b-hexosaminidase (Figure 7c). In this case, theneutral glycan fraction was reduced to a limitedfamily of core glycan species ranging from 4.2 to7.2 Gu.

The 4.2-7.2 Gu GPI anchor core glycans werefurther analysed by HPTLC following partial acidhydrolysis and a-mannosidase digestions(Schneider & Ferguson, 1995; Figure 8). In all casesa ladder of four bands (comigrating with standardsof AHM, Mana1-4AHM, Mana1-6Mana1-4AHMand Mana1-2Mana1-6Mana1-4AHM) was visible,suggesting that all of the GPI neutral glycans ofPARP contain the underlying conserved sequenceMana1-2Mana1-6Mana1-4AHM. This was con-®rmed by the sensitivity of the Mana1-2Mana1-6Mana1-4AHM band to the Mana1-2Man speci®cA. phoenicis a-mannosidase and the sensitivity ofthe Mana1-4AHM, Mana1-6Mana1-4AHM andMana1-2Mana1-6Mana1-4AHM bands to jack beana-mannosidase.

Taken together, these data indicate that the GPIanchors of PARP are based upon a conventionalcore structure of protein-ethanolamine-HPO4-6Mana1-2Mana1-6Mana1-4GlcN-PI (McConville &Ferguson, 1993).

The molecular mass distribution and shapeof GPEET-PARP

The pure GPEET-PARP material was subjectedto mild alkaline hydrolysis to remove the fattyacids of the GPI anchor, and thus prevent the for-

Figure 5. ESI-MS analysis of the PI fraction isolated from Gof the PI fraction of ConAÿ PARP. b, Negative daughter-ionfragmentation of the PI fraction isolated from octyl-Sepharoions of PARP.

mation of PARP micelles for the light scatteringstudies. This material was analysed by negativeion matrix-assisted laser desorption ionisationtime-of-¯ight mass spectrometry (MALDI-TOF-MS;Figure 9) which showed groups of ions, corre-sponding to the [M ÿ 2H]2ÿ, [M ÿ H]ÿ and

PEET-PARP. a, Negative ion electrospray mass spectrumspectrum of the m/z 837 ion seen in a. c, Cone voltagese puri®ed PARP. d, Fragmentation scheme for the PI

Figure 7. Bio-Gel P4 analysis ofthe GPI anchor neutral glycans ofGPEET-PARP and EP-PARP. a, Pu-ri®ed EP-PARP (Pool D, Figure 6)was dephosphorylated with aqu-eous HF, desialylated with A. urea-faciens sialidase and analysed byBio-Gel P4 chromatography before(thin line) and after (thick line)exhaustive digestion with endo-b-galactosidase. b, Puri®ed GPEET-PARP (Pool E, Figure 6) was trea-ted and analysed as described fora. c, Octyl-Sepharose puri®edPARP GPI-anchor neutral glycansafter digestions with a mixture ofbovine testes b-galactosidase andjack bean b-hexosaminidase. Theline on the top of each panel indi-cates the positions of glucose oligo-mer standards. The peaks arelabelled with their respective glu-cose unit values.


[2M ÿ H]ÿ pseudomolecular ions of GPEET-PARP, around m/z 7750, m/z 14,500 and m/z29,000, respectively. These data are consistent withan average mass of GPEET-PARP of about

cated by AHM, Man1-AHM, Man2-AHM and Man3-AHM, rwith the Man3AHM-AHM standards represent partial acid c

14.5 kDa and a mass range of about 11 kDa to20 kDa. The measured average mass is consistentwith the GPEET-PARP composition data (Table 1)which predicts an average mass of 14,358 Da. The

Figure 8. HPTLC analysis of thePARP GPI anchor core glycans.The GPI neutral glycans of T. bruceisVSG117 (lanes 1 to 3) and thePARP GPI core glycans shown inFigure 7(c) (4.2 Gu glycan, lanes 4to 6; 5.0 Gu glycan, lanes 7 to 9;6.0 Gu glycan, lanes 10 to 12 andthe 7.2 Gu glycan, lanes 13 to 15)were subjected to mild acidhydrolysis followed by digestionwith A. phoenicis a-mannosidase(APAM) or jack bean a-mannosi-dase (JBAM) as indicated. The pro-ducts were separated by HPTLCand the plate was subjected to¯uorography. The migration pos-itions of AHM, Mana1-4AHM,Mana1-6Mana1-4AHM and Mana1-2Mana1-6Mana1-4AHM are indi-

espectively. (Note: the faint bands that do not comigrateleavage products that contain side-chain residues.)

Figure 9. Negative ion MALDI-TOF mass spectrum ofdeacylated GPEET-PARP.


measured mass range would be consistent with apopulation of GPEET-PARP species that containbetween ®ve and 30 lactosamine repeats.

Alkaline-hydrolysed GPEET-PARP was furtherpuri®ed on a Superose 12 gel ®ltration column.During this puri®cation step we observed ashoulder on the main GPEET-PARP peak thateluted at slightly higher molecular mass and thatwas immunoreactive with the anti-GPEET-PARPantibody (data not shown). The nature of thisminor population of PARP molecules, that is dis-continuous with the major population of GPEET-PARP molecules, is under further investigation.The main GPEET-PARP peak from the Superose 12column was rechromatographed on Superose 12and subjected to molecular mass and size determi-nation by multi-angle light scattering. These datasuggested an average mass of 15.4 kDa and a mass

Figure 10. Analysis of GPEET-PARP by gel-®ltrationand light-scattering. The continuous line shows therefractive-index trace for GPEET-PARP eluted from aSuperose 12 gel-®ltration column. The elution volume ofthe top of the peak (12.3 ml) is identical to that of BSA,a 67 kDa globular protein.. The dotted line representsthe Debye-plot of molecular mass, determined by light-scattering, across the GPEET-PARP peak.

range of approximately 12 kDa to 22 kDa, seeFigure 10, in good agreement with the MALDI-TOF-MS data.

The GPEET-PARP sample eluted as a broadpeak on Superose 12 gel-®ltration and the top ofthe peak was found to coelute with bovine serumalbumin (BSA, a 67 kDa globular protein). Thelarge discrepancy between the average molecularweight of GPEET-PARP (about 15 kDa) and its hy-drodynamic volume on gel-®ltration suggest thatGPEET-PARP is a non-globular molecule. This wascon®rmed by measuring the diffusion coef®cient ofGPEET-PARP in aqueous solution by dynamiclight scattering (Berne & Pecora, 1990). The valueobtained (3.9 � 107 cm2 sÿ1) was even lower thanthat of the 67 kDa globular protein BSA(6.0 � 107 cm2 sÿ1; Raj & Flygare, 1974). Taken to-gether, these data indicate that GPEET-PARP has ahighly elongated structure.

Discussion

A detailed analysis of PARP the fraction of T.brucei procyclic cells is warranted because (1)PARP is the most abundant macromolecule on thesurface of these cells and (2) these heavily glycosy-lated proteins are the most likely ligands for thetsetse ¯y lectins that are thought to control themidgut infection and parasite maturation processes(Welburn et al., 1994; Maudlin & Welburn, 1994).

The data reported in this paper would suggestthat, at least for our cell-line, relatively fresh T. bru-cei procyclic cultures express almost exclusivelyEP-PARP while long-term cultures appear to pro-gressively replace EP-PARP with GPEET-PARP.According to the immuno¯uorescence microscopyresults, this slow shift from EP-PARP to GPEET-PARP expression does not appear to be due to achange in the cell population, i.e. from EP-PARP-to GPEET-PARP-expressing cells, but rather to thegradual replacement of EP-PARP with GPEET-PARP on all of the cells. Whether or not this, as yetpoorly characterised, shift in PARP expression inlong-term cultures resembles any process that oc-curs during the maturation of trypanosome infec-tions within the tsetse ¯y remains to bedetermined.

Our ability to detect GPEET-PARP at the pro-tein-level in this study was due to the use of long-term cultures and of a non-selective solvent-extrac-tion, reverse phase chromatography puri®cationsystem. Previous puri®cation strategies have usedCon A-Sepharose af®nity chromatography as thelast puri®cation step (Richardson et al., 1988;Clayton & Mowatt, 1989) and any GPEET-PARP,having no consensus N-glycosylation site, wouldhave been lost at this point. Our Con A-Sepharosefractionation of total PARP suggests that all of theEP-PARP is N-glycosylated (i.e. no anti-EP-PARPimmunoreactivity or EP-PARP peptide sequenceswere observed in the Con A-Sepharose ¯ow-through fractions), suggesting that the non-N-gly-


cosylated EP-PARP species (encoded by the parpBbgene) was not expressed by the cells analysed inthis study. Whether or not this EP-PARP species isexpressed at a speci®c stage of the trypanosomelife-cycle within the tsetse ¯y remains to be deter-mined.

The N-glycosylation of EP-PARP is unusual inthat only one oligosaccharide species (Man5-

GlcNAc2) is found attached to the single glycosyla-tion site. Such a lack of microheterogeneity inN-glycosylation is not unique but it is quite rare.The Man5GlcNAc2 structure is a good ligand forthe lectin Con A and explains the ability of this lec-tin to distinguish between EP-PARP and GPEET-PARP. Interestingly, Con A has recently beenshown to induce apoptosis in procyclic trypano-somes (Welburn et al., 1996) and it seems likelythat the N-linked oligosaccharide of EP-PARP willbe the primary ligand on the cell surface for thislectin.

The other carbohydrate components of thePARPs are their GPI membrane anchors. Theanchors of EP-PARP and GPEET-PARP have in-distinguishable glycan structures, as judged bythe Bio-Gel P4 pro®les of their GPI anchorneutral glycan fractions before and after endo-b-hexosaminidase digestion. The digestion of thelarge GPEET-PARP GPI neutral glycans with amixture of b-galactosidase and b-hexosaminidaserevealed a set of core glycan species ranging from4.2 to 7.2 glucose units in size, similar to those gen-erated from T. b. rhodesiense EP-PARP material(Ferguson et al., 1993). The 4.2 Gu core glycan hasthe structure Mana1-2Mana1-6Mana1-4AHM andthe other glycans also contain this basic core-motif,as judged by analysis of the their partial acidhydrolysates. Thus, all of the PARP GPI anchorsappear to contain the conserved ethanolamine-HPO4-6Mana1-2Mana1-6Mana1-4GlcNa1-6PI struc-ture found in all eukaryote GPI-protein anchors(McConville & Ferguson, 1993). The major GPI pre-cursor found in procyclic cells (glycolipid PP1) alsohas this structure (Field et al., 1991b), suggestingthat the extensive side-chain modi®cations foundon the PARP GPI-anchors are added after transferof glycolipid PP1 to the nascent PARP polypep-tides, possibly in the Golgi-apparatus. Polylactosa-mine structures have been previously described asterminal structures of complex N-linked oligosac-charides attached to certain T. brucei bloodstreamform VSG variants (Zamze et al., 1991). Thus, it ispossible that some or all of the polylactosamineglycosyltransferase systems are shared between theVSG and PARP biosynthetic pathways.

Detailed analysis of the molecular species of thePI fraction, released from GPEET-PARP by nitrousacid deamination, by ESI-MS con®rmed the pre-sence of palmitoyl-myo-inositol-1-HPO4-(sn-1-stear-oyl-2-lyso-glycerol) (Field et al., 1991a) but revealedthat this represents only 21% of the total PI frac-tion, the remainder being made up of anotherseven molecular species. The glycerolipid com-ponent is relatively homogeneous and is made of

exclusively sn-1-stearoyl-2-lyso-glycerol (75%) andsn-1-octadecyl-2-lyso-glycerol (25%). Alkylglycerol-containing protein GPI-anchor species have notbeen previously been described in T. brucei,although they are predominant forms in other ki-netoplastid organisms (Schneider et al., 1990; Coutoet al., 1993; Heise et al., 1995; Serrano et al., 1995).The acyl chain attached to the inositol ring appearsto be located at the 2-position, as judged by theESI-MS daughter-ion spectra, and appears to bequite heterogeneous (a mixture of C16:0, C18:0,C18:1 and C18:2 fatty acids). Qualitatively similar,though quantitatively different, heterogeneity hasalso been observed for the acyl chain attached tothe inositol ring of glycolipid C, a major GPI inter-mediate in T. brucei bloodstream forms (GuÈ theret al., 1996).

The large and heterogeneous sialylated,branched polylactosamine GPI side-chains arelikely to form a dense glycocalyx immediately ad-jacent to the plasma membrane of the procyclic try-panosomes. Such a glycocalyx is likely to performa protective function by preventing the approachof hydrolytic enzymes in the tsetse ¯y gut to thecell surface. Since EP-PARP and GPEET-PARP ap-pear to contain identical GPI side-chains, this pro-tective layer would not be affected by changes inthe relative proportions of EP-PARP and GPEET-PARP expressed on the cell surface.

The PARP protein components, that are directlyattached to the GPI-anchor glycocalyx, are of un-known function. However, the light-scattering ex-periments on GPEET-PARP suggest that, like EP-PARP, this molecule will have an extended confor-mation projecting a polyanionic rod-like structureinto the external milieu. The suggested extendedpeptide conformations, and the heavy glycosyla-tion, of the PARPs would account for their aber-rant behaviour on SDS-PAGE, where moleculesthat average about 15 kDa have apparent molecu-lar weights of around 30 kDa (GPEET-PARP) and42 kDa (EP-PARP). However, the faster migrationof GPEET-PARP could be interpreted (Figure 11)as suggesting that it has a less extended confor-mation than EP-PARP, as might be expected forthe -GPEET- versus -EP- repeat domains.

The overall similarity of the molecular architec-ture of the T. brucei procyclic surface to that of theLeishmania (sand¯y-dwelling) promastigote surfacehas been previously noted (Ferguson et al., 1993;McConville and Ferguson, 1993). Thus, Leishmaniapromastigotes also present a surface with a denseglycocalyx through which is projected 3 to 5million polyanionic rod-like molecules. In this case,however, the glycocalyx is composed of discreteand abundant glycoinositol-phospholipids and thepolyanionic rods (the lipophosphoglycans) aremade of phosphodisaccharide repeat units ratherthan glutamic acid-rich peptide. This similarity inthe physicochemical properties of the two surfaces,achieved using different types of molecules, isstriking given the evolutionary distance betweenthese two kinetoplastid parasites (Fernandes et al.,

Fig

ure

11

.S

chem

atic

rep

rese

nta

tio

no

fE

P-P

AR

Pan

dG

RE

ET

-PA

RP

.B

oth

form

so

fP

AR

Pco

nta

ina

po

siti

vel

ych

arg

edm

oti

fo

fK

GG

KG

Kn

ear

thei

ram

ino

term

ini

and

are

pre

dic

ted

toen

din

the

C-t

erm

inal

seq

uen

ceP

EP

EG

(see

bo

ldle

tter

s).

Th

ein

terv

enin

gp

oly

anio

nic

do

mai

ns

dif

fer

inse

qu

ence

bu

th

ave

sim

ilar

char

ge

den

siti

es.

Th

eex

ten

ded

stru

ctu

resu

gg

este

dfo

rE

P-P

AR

Pis

bas

edo

nN

MR

and

mo

lecu

lar

mo

del

lin

gd

ata

(Ro

dit

iet

al.,

1989

).G

PE

ET

-PA

RP

isd

raw

nto

ind

icat

ea

gen

eral

lyex

ten

ded

con

-fo

rmat

ion

,co

nsi

sten

tw

ith

the

lig

ht-

scat

teri

ng

dat

are

po

rted

inth

isp

aper

.B

oth

form

so

fP

AR

Pco

nta

inin

dis

tin

gu

ish

able

GP

Ian

cho

rsb

ut

on

lyE

P-P

AR

Pco

nta

ins

asi

ng

leN

-gly

cosy

lati

on

site

that

iso

ccu

pie

dex

clu

siv

ely

wit

ha

sin

gle

Man

5G

lcN

Ac 2

stru

ctu

re.


1993) and may suggest that this type carbohydrateand negative charge-dominated surface architec-ture may provide a general selective advantage formultiplication inside an insect vector. It would bea mistake to draw too many parallels betweenthese two organisms since the biologies of the in-sect infections are quite different (Molyneux &Ashford, 1983). However, it is interesting to notethat the LPGs of L. major and L. donovani undergoprofound structural changes upon differentiationof procyclic promastigotes to infectious metacyclicpromastigotes (McConville et al., 1992; Sacks et al.,1995) and that the existence of multiple forms ofPARP might allow for analogous changes in sur-face architecture during the establishment, main-tenance and maturation of T. brucei in the infectedtsetse ¯y.

In summary, the isolation and analysis ofGPEET-PARP, and of the N-linked oligosaccharideof EP-PARP, has extended our view of the molecu-lar architecture of the T. brucei procyclic cell sur-face, and their biosynthetic capacity, and shouldassist in unravelling the molecular events involvedin the establishment, maintenance and maturationof procyclic trypanosome infections in the tsetse¯y.

Materials and Methods

Materials

Aluminium-backed high performance thin layer chro-matography (HPTLC) sheets were from Merck; jack beana-mannosidase, Arthrobacter ureafaciens sialidase, bovinetestes b-galactosidase and Bacteroides fragilis endo-b-ga-lactosidase were from Boehringer Mannheim. Aspergillusphoenicis a-mannosidase and jack bean b-hexosaminidasewere from Oxford GlycoSystems. Peptide N-glycanase F(PNGase F) was from New England BioLabs. Bovineliver phosphatidylinositol and Stains-all (3-ethyl-2-[3-(1-ethylnaphthol[1,2-d]-thiazolin-2-ylidene)-2-methyl-prope-nyl[naphthol[1,2-d]-thiazolium bromide) were fromSigma. NaB3H4 (10-15 Ci/mmol) was from Du Pont-New England Nuclear; Concanavalin A-Sepharose (ConA-Sepharose) was from Pharmacia. The anti-EP-PARPmonoclonal antibody (mAb 247) was a gift from DrTerry Pearson, University of Victoria, Vancouver. Theanti-GPEET-PARP rabbit polyclonal antibody (pAb Ku)was a gift from Dr Isabel Roditi, Bern. Fetal calf serumfrom Gibco was heat-inactivated by heating for twohours at 60�C.

Carbohydrate standards

The N-linked oligosaccharide Mana1-6(Mana1-3)Mana1-6(Mana1-3)Manb1-4GlcNAcb1-4GlcNAc(Man5GlcNAc2) was prepared from ribonuclease B andwas a generous gift from Prof. Steve Homans, StAndrews University. This compound was NaB3H4-reduced, to yield Mana1-6(Mana1-3)Mana1-6(Mana1-3)-Manb1-4GlcNAcb1-4[1-3H]GlcNAc-ol (Man5GlcNAc2-ol),as described below. [3H]GlcNAc-ol was generated byNaBH4-reduction and N-acetylation of [3H]GlcN(Dupont NEN). The labelled GPI neutral glycanMana1-2Mana1-2Mana1-6Mana1-4[1-H]2,5-anhydro-

mannitol (Man4AHM) was obtained from yeast GPI an-chors (Fankhauser et al., 1993).

Cell culture

Procyclic trypanosomes were grown in 25 ml sterileuniversal containers (Sterilin) in continuous culture at25�C in 5 ml SDM79 medium (Brun & SchoÈnenberger,1979) supplemented with 5% heat-inactivated fetal calfserum. The cultures were maintained by diluting 100 mlevery four to ®ve days with 4.9 ml of fresh medium.Large batches of cells were grown in 500 ml of mediumin 850 cm2 roller bottles to a density of 1 � 107 to 2 � 107

cells/ml (late logarithmic phase of growth).

Purification of PARP

PARP was puri®ed according to (Ferguson et al., 1993)with minor modi®cations. Brie¯y, freeze-dried cell pel-lets 0.5 � 1011 to 1.3 � 1011 cell) were homogenised usinga pestle and a mortar. The powder was extracted fourtimes with chloroform/methanol/water (1:2:0.8, v/v)with thorough mixing and sonication. After this delipi-dation procedure, the pellet was dried under N2 until itwas almost dry and subsequently extracted four timeswith 9% butan-1-ol in water. The 9% butan-1-ol extractswere pooled and dried by rotary evaporation. The ex-tract was redissolved in 1 ml of buffer A (5% propan-1-olin 100 mM ammonium acetate) and applied to an octyl-Sepharose column (130 mm � 10 mm) at 3.4 ml/h. Thecolumn was washed with 20 ml of buffer A and elutedwith a linear gradient from buffer A to 60% propan-1-olover 100 ml. Aliquots (0.5 ml) of each 0.8 ml fraction wereapplied to an HPTLC plate and carbohydrate contentwas estimated by staining the plate with orcinol reagent.The myo-inositol contents of aliquots of the fractions inthe orcinol-positive peak region were determined by GC-MS (Smith et al., 1996). Inositol-containing fractions werepooled and dried by rotary evaporation.

Chaps lysates of procyclic T. brucei

Trypanosome cell pellets were washed repeatedly inphosphate-buffered saline (PBS; 50 mM sodium phos-phate, pH 7.4, 150 mM NaCl), lysed with 2% Chaps(Sigma), 0.1 mM tosyl-L-lysine chloromethyl ketone,10 mg/ml leupeptin and heated for 90 seconds at 100�C.After centrifugation, 15 minutes, 13,000 g, 4�C the super-natant was transferred to an equal volume of 0.2 M Tris-HCl (pH 6.8), 10 mM dithiothreitol, 4% Triton X-100(TX-100) and 20% (v/v) glycerol (2 � -concentrated TX-100 sample buffer), ®nal concentration 5 � 108 cells/ml(Ziegelbauer et al., 1990).

PAGE analysis of PARP

SDS-PAGE: SDS-PAGE was carried out on 15% or12% (w/v) polyacrylamide minigels under denaturingconditions (Laemmli, 1970).

Triton-PAGE: PARP was electrophoretically separatedfrom other proteins using a 3% polyacrylamide stackergel and a 7.5% separation gel and the Laemmli buffersystem in which SDS was replaced by TX-100 (Ziegel-bauer et al., 1990). Electrophoresis at 20 mA per gel wasterminated 2.5 hours after the marker dye, bromophenolblue, had left the gel.

PARP-containing SDS-PAGE or TX-100 PAGE gelswere stained with Stains-all (Dahlberg et al., 1969; Green


et al., 1973). For the quantitation of PARP in TX-100PAGE gels, the stained bands were cut out, eluted inthe dark with 700 ml formamide and the absorbanceat 590 nm measured using a spectrophotometer (Ziegel-bauer et al., 1990).

Western blotting of T. brucei whole cell SDS-lysates

Trypanosome cell pellets were washed repeatedly inPBS and lysed with an equal volume of boiling 2� con-centrated SDS-sample buffer, ®nal concentration 5 � 108

cells/ml. Aliquots (10 ml) of the lysate were resolved in12% polyacrylamide minigels and electrophoreticallytransferred (Towbin et al., 1979) to an Immobilon P mem-brane (Millipore) using a semidry blot apparatus (Hoe-fer) at 20 mA per gel for 20 minutes. PARP which hadbound to the membrane was then detected as describedbelow for the detection of dot-blots.

Dot-blot analysis of PARP column fractions

Aliquots (1 ml) of the octyl-Sepharose fractions ofPARP were spotted onto a nitrocellulose membraneand blocked overnight at 4�C with Tris-buffered saline(TBS; 50 mM Tris-HCl, pH 7.4, 150 mM NaCl) contain-ing 0.5% (v/v) BSA. The membrane was washed twotimes (ten minutes each at room temperature) withTBS, 0.05% (v/v) Tweenÿ20 (TBS-Tween) and incu-bated for one hour at 37�C with anti-EP-PARP mAb427, diluted 1:1000 with TBS-Tween containing 0.3%BSA, or with anti-GPEET-PARP pAb Ku, diluted1:2000 with TBS-Tween containing 0.3% BSA. After in-cubation with the primary antibody the membraneswere washed twice as described above and incubatedwith horseradish peroxidase (HRP) conjugated anti-mouse antibody (Jackson Laboratories), diluted 1:2000,or HRP-conjugated anti-rabbit antibody (Bio-Rad), di-luted 1:3000. The immunostained dots were detectedusing ECL detection reagents (Amersham) according tothe manufacturers instructions.

Immunofluorescence microscopy of procyclic cells

Procyclic cells were ®xed by suspending cell pelletsin 4% (w/v) paraformaldehyde in PBS (106 cells/ml, 20minutes at room temperature). The ®xed cells were in-cubated with 3% BSA in PBS (16 hours at room tem-perature) and washed three times with PBS, 0.2% BSAand 0.02% (w/v) sodium azide. The ®xed and washedcells were incubated for one hour in TBS-Tween, 3%BSA with and without a mixture of 1:1000 dilutedanti-EP-PARP mAb427 and 1:1000 diluted anti-GPEET-PARP pAb Ku. The cells were washed three times, asdescribed above, and incubated for one hour with 1:50diluted ¯uorescein isothiocyanate (FITC)-conjugatedgoat anti-rabbit antibody and 1:50-diluted Texas-Redconjugated goat anti-mouse antibody (Jackson Labora-tories) in PBS, 0.2% BSA, 0.02% sodium azide. Thecells were washed three times, mounted on slides withmowiol (Sigma) and observed with a BioRad MRC 600confocal microscope.

Release and labelling of N-linked oligosaccharides

A sample of total ocyl-Sepharose-puri®ed PARP(10 nmol) was dried, denatured in 80 ml 1% (w/v) SDS,100 mM dithiothreitol (ten minutes, 100�C) and mixedwith 10 ml 0.5 M sodium citrate (pH 7.5), 10% (v/v) Non-

idet-P40 and 3 ml (3000 units) of PNGase F. After threehours at 37�C, a further 2000 units of PNGase F wereadded and incubation was continued for 16 hours. Thedigest was made 5% (v/v) with respect to propan-1-oland applied to an octyl-Sepharose column, as describedabove. The released N-linked oligosaccharides, thateluted in the void-volume, were desalted by passagethrough a column (1 cm � 20 cm) of Bio-Gel P4 elutedwith water. Samples of the void-volume of the Bio-GelP4 eluate were taken for monosaccharide analysis andfor NaB3H4-reduction.

NaB3H4 reduction

Samples of the released PARP N-liked oligosacchar-ides, and a standard of authentic Mana1-6(Mana1-3)Mana1-6(Mana1-3)Manb1-4GlcNAcb1-4GlcNAc (Man5-

GlcNAc2), were dissolved in 30 ml 100 mM sodium bo-rate buffer (pH 10.5), mixed with 5 ml 36 mM NaB3H4 in100 mM NaOH and incubated at room temperature for90 minutes. Subsequently, 65 ml of 1 M NaB2H4 wereadded and, after three hours, 1 M acetic acid was addedto destroy excess reductant. The products were desaltedby passage through 0.2 ml of AG50X12(H�) followed byrepeated drying from methanol followed by toluene. Thelabelled oligosaccharitols were puri®ed from radiochemi-cal contaminants by downward paper chromatography(developed for 60 hours with butan-1-ol/ethanol/water,4:1:0.8, v/v/v), elution from the origin with water andchromatography on a 1 ml column of microgranular cel-lulose. The sample was applied to the microgranular cel-lulose column in butan-1-ol/ethanol/water (4:1:1, v/v/v)and the column was washed with 5 ml of the same sol-vent followed by 1 ml methanol. The labelled oligosac-charitols were eluted with 5 ml 200 mM ammoniumacetate in water and freeze-dried to remove the am-monium acetate.

Bio-Gel P4 gel-filtration chromatography

Bio-Gel P4 gel-®ltration of labelled glycans was per-formed using a GlycoMap (Oxford GlycoSystems)equipped with a 1 cm � 50 cm column. Chromatog-raphy was performed in the presence of 10 ml of a glu-cose oligomer internal standard (50 mg/ml dextranpartial acid hydrolysate) and the hydrodynamic vo-lumes of 3H-labelled glycans are expressed in glucoseunits (Gu).

Identification and thin layer chromatography ofradiolabelled N-linked oligosaccharitols

For the identi®cation of radiolabelled N-linked oligo-saccharitols, samples were hydrolysed in 100 ml 2 M tri-¯uoroacetic acid (two hours, 100�C), dried, dissolved in100 ml saturated sodium bicarbonate at 0�C and re-N-acetylated by three additions of 2.5 ml acetic anhy-dride (ten minutes apart). The products were desaltedby passage through 0.25 ml Dowex AG50X12(H�) fol-lowed by drying and removal of acetic acid by coeva-poration with toluene. The products were subjected tochromatography alongside a standard of [3H]GlcNAc-olon Silica Gel 60 HPTLC sheets, using two developmentswith butan-1-ol/ethanol/water (4:3:3, v/v/v). Radio-labelled material was detected by ¯uorography atÿ70�C after the sheets were sprayed with En3Hancespray (DuPont-NEN). Radiolabelled N-linked oligosac-


charitols were identi®ed by the presence of [3H]GlcNAc-ol in the N-acetylated acid hydrolysates.

Intact, puri®ed radiolabelled N-linked oligosacchari-tols were subjected to chromatography on the sameHPTLC sheets using butan-1-ol/ethanol/water (4:3:3,v/v/v) for two developments. Radiolabelled glycanswere detected as described above.

Dionex carbohydrate HPLC

NaB3H4-reduced oligosaccharitols were mixed with3 ml of the dextran partial acid hydrolysate and appliedto a Dionex Carbopac PA1 column. The column waseluted at 0.6 ml/min with a gradient of 20 to 200 mM so-dium acetate in 0.1 M NaOH over 70 minutes. Aliquotsof the 0.6 ml column-fractions were taken for scintillationcounting. The unlabelled dextran oligomers were de-tected with a pulsed-amperometric detector. The elutionpositions of the labelled oligosaccharitols are expressedin ``Dionex Units'' (Du) by linear interpolation of theirretention times between those of adjacent dextran oligo-mer standards (Ferguson, 1992b).

Generation of deaminated, reduced PARP

Octyl-Sepharose puri®ed PARP (10 nmol) was deami-nated with 60 ml of 50 mM sodium acetate buffer (pH4.0), containing 250 mM sodium nitrite (two hours, roomtemperature). After deamination, an amount (24 ml) of0.4 M boric acid was added and the phosphatidylinositolwas extracted three times with 85 ml of water-saturatedbutan-1-ol. The aqueous phase was reduced by the ad-dition of 10 ml of 36 mM NaB3H4 in 100 mM NaOH (1.5hours, room temperature). The deaminated, reducedPARP was desalted by passage through 0.2 ml of AG-50X12(H�), rotary evaporation to dryness and dryingtwice from 250 ml of 5% (v/v) acetic acid in methanol,twice from 250 ml of methanol (to remove boric acid) andtwice from 100 ml of toluene (to remove acetic acid). Thedeaminated, reduced PARP was re-dissolved in 200 ml ofwater and applied to a Sephadex G10 column(10 mm � 175 mm, V0 � 7.5 ml) and eluted at 10 ml/h.Fractions of 1 ml were collected and 1 ml aliquots fromeach fraction counted using a liquid scintillation counter.The three radioactive fractions corresponding to the voidvolume were pooled and freeze-dried.

DEAE-Sephacel chromatography of deaminated,reduced PARP

The deaminated and reduced PARP was re-dissolvedin 500 ml water and applied to a DEAE-Sephacel column(10 mm � 100 mm). The column was eluted with a lineargradient from 0 to 1 M ammonium acetate over 100 mlat 4 ml/h. Fractions of 1 ml were collected and 1 ml ali-quots from each fraction were counted using a liquidscintillation counter. The two radioactive peaks werepooled and lyophilised repeatedly to remove the am-monium acetate. Small aliquots from each pool were setaside for SDS-PAGE analysis.

Concanavalin A lectin affinity chromatographyof PARP

Native, non-radioactive, PARP or the DEAE-Sephacelseparated fractions of 3H-labelled deaminated, reducedPARP were applied in Con A-buffer (50 mM Tris-HCl,pH 7.2, 150 mM NaCl, 1 mM MgCl2, 1 mM CaCl2, 1 mM

MnCl2) to a Con A-Sepharose column (bed volume 3 ml)at 4 ml/h. The column was washed with 25 ml Con Abuffer and eluted with a linear gradient of 0-200 mMa-methyl-D-mannopyranoside (a-MeMan). The elution ofPARP was followed either by liquid scintillation count-ing of radioactive fractions (in the case of deaminated,reduced PARP) or by measuring the absorbance of col-lected fractions at 220 nm and by dot-blotting aliquotsof the collected fractions with anti-PARP antibodies (inthe case of native PARP).

Generation of the PARP GPI-anchor neutral glycan

Deaminated and NaB3H4-reduced and Con A-Sepha-rose separated EP-PARP and GPEET-PARP sampleswere desialylated using Arthrobacter ureafaciens sialidase(see below), dephosphorylated (using 50% aqueous HFat 0�C for 48 hours), N-acetylated and desalted as de-scribed in (Ferguson, 1992b).

Thin layer chromatography of labelled GPIneutral glycans

Radiolabelled GPI anchor neutral glycans were sub-jected to chromatography on Silica Gel 60 HPTLC sheetsusing propan-1-ol, acetone, water (9:6:4 v/v) for one de-velopment (Schneider et al., 1993). Radiolabelled materialwas detected by ¯uorography at ÿ70�C after the sheetswere sprayed with En3Hance spray (DuPont-NEN).

Composition analyses and N-terminal sequencingof PARP

Carbohydrate and inositol contents were measuredusing GC-MS, as described previously (Ferguson, 1992b).Samples for amino acid analysis were hydrolysed in 6 MHCl vapour at 110�C and analysed using a Waters Pico-Tag system. N-terminal sequencing of PARP was carriedout on an Applied BioSystem model 477A pulsed-liquidsequencer.

Enzymatic and chemical cleavages

Exoglycosidase digestions and acetolysis

Aspergillus phoenicis a-mannosidase (APAM) diges-tions, jack bean a-mannosidase (JBAM) digestions andacetolysis reactions were performed as described in(Ferguson, 1992b). The conditions of the Arthrobacter ur-eafaciens sialidase and the Bacteroides fragilis endo-b-galac-tosidase digestions were as described in (Treumann et al.,1995).

Mixed exoglycosidase digestion

PARP GPI-anchor neutral glycans were digested witha combination of jack bean b-hexosaminidase (JBBH) andbovine testes b-galactosidase (BTBG). Samples were dis-solved in 68 ml of 100 mM sodium citrate/phosphate buf-fer (pH 4.5) and added to a mixture of 1 unit of JBBH in22 ml 20 mM sodium citrate/phosphate buffer pH 6.0and 0.05 unit of BTBG in 31 ml 20 mM Tris-HCl (pH 7.5),100 mM KCl, 1 mM EDTA, 0.05% NaN3 (total digest vo-lume approximately 120 ml). After 16 hours at 37�C thesamples were boiled for ®ve minutes and desalted bypassage through a column of 0.2 ml AG50X12(H�) over0.2 ml AG3X4(OHÿ) over 0.1 ml QAE-Sephardex-A25(OHÿ).


Partial acid hydrolysis

GPI neutral glycans were hydrolysed in 100 mM tri-¯uoroacetic acid at 100�C for four hours (Schneider &Ferguson, 1995). After hydrolysis the samples were driedin a Speedvac and dried twice from 50 ml water to re-move residual tri¯uoroacetic acid.

Mild base hydrolysis of GPEET-PARP

GPEET-PARP (400 nmol) was hydrolysed in 2 ml of12.5% (w/v) NH3 in 20% propan-1-ol for 16 hours at37�C. The solvents were removed by rotary evaporationand the released fatty acids were removed by partition-ing the products between water and butan-1-ol. The aqu-eous phase was centrifuged in a microfuge and thesupernatant, containing deacylated PARP, was dried.

Mass spectrometry

Electrospray ionisation mass spectrometry

ESI mass spectra were recorded on a Micromass Quat-tro triple-quadrupole mass spectrometer (Micromass,Altrincham, U.K.). Samples (10 to 100 pmol/ml) were in-troduced into the electrospray source at 10 ml/min usinga Michrom HPLC pump. For the negative-ion modemass spectrometry of phosphatidylinositols 1 mM NH3

in methanol/chloroform (3:2, v/v) was used as solvent.For tandem MS experiments the pseudomolecular parentions were accelerated into a collision cell containingargon (2.3 � 10ÿ3 mbar) through a potential difference ofbetween 70 and 110 V. For cone-voltage induced frag-mentation the cone voltage was 95 V. All the mass spec-tra were background-subtracted and smoothed usingMassLynx software. The source and collision-cell con-ditions were optimised with a standard of bovine liverphosphatidylinositol.

Matrix-assisted laser desorption ionisation time-of-flightmass spectrometry

The MALDI-TOF mass spectra of GPEET-PARP wereacquired on a LaserTOF (Micromass, Altrincham) instru-ment; 10 pmol of GPEET-PARP were co-crystallised withsinapinic acid in 1 mM ammonium acetate as a matrix.Spectra were recorded in negative mode.

Light scattering analyses of GPEET-PARP

Base-hydrolysed GPEET-PARP (150 nmol in 150 mlwater) were analysed at a ¯ow rate of 0.4 ml/min on aSuperose 12 gel ®ltration column (Pharmacia) which hadbeen equilibrated with phosphate buffered saline con-taining 1 mM EDTA and 0.05% NaN3. The eluant waspassed into a DAWN DSP-F multi-angle light scatteringdetector (Wyatt Technology Corp., Santa Barbara CA,USA) and subsequently through a Wyatt Optilab refrac-tive index detector. The laser wavelength was 488 nm.The data were analysed with the Wyatt ASTRA version4.00 software using the method of Zimm and assuminga speci®c refractive increment (dn/dc) of 1.60 for PARP.This was done on the assumption that two thirds of themolecular weight of PARP is carbohydrate and one thirdis protein and that the authentic (dn/dc) can be approxi-mated linearly between 1.80, the speci®c refractive incre-ment of BSA, a protein which does not containcarbohydrate and 1.50 which would be the speci®c re-fractive increment of a polysaccharide.

Dynamic light scattering of PARP eluted from theSuperose 12 gel-®ltration column was performed on a S4700 Version PCS 1.26 (Malvern) spectrometer. The datawere ®tted to the correlation function and the diffusioncoef®cient calculated as described in (Berne & Pecora,1990).

Acknowledgements

This work was supported by a Wellcome Trust pro-gramme grant to M. A. J. F. A. T. was supported by anEU-fellowship (contract no. ERBCHBICT940944). M. A. J.F. is a Howard Hughes Medical Institute internationalresearch scholar. We are grateful to Isabel Roditi andPeter BuÈ tikofer, for exchanging unpublished data and forsupplying the Nour6c cells and anti-GPEET-PARP anti-bodies, to Terry Pearson for supplying pure EP-PARPand anti-EP-PARP antibodies, to Steve Homans for sup-plying the Man5GlcNAc2 standard and to Barry Caud-well and Nick Morrice for amino acid analyses andpeptide sequencing. We thank Ian Maudlin, Sue Wel-burn, Paul Englund, Kuo-yan Hwa and Angela Mehlertfor helpful discussions.

References

Berne, B. & Pecora, R. (1990). Dynamic Light Scatteringwith Applications to Chemistry, Biology and Physics,Robert E. Krieger Publishing Company, Malabar,FL.

Brun, R. & SchoÈnenberger, M. (1979). Cultivation andin vitro cloning of procyclic culture forms of Trypa-nosoma brucei in semi de®ned medium. Acta Tropica,36, 289±292.

Clayton, C. E. & Mowatt, M. R. (1989). The procyclicacidic repetitive proteins of Trypanosoma brucei.Puri®cation and post-translational modi®cation.J. Biol. Chem. 264, 15088±15093.

Clayton, C. E. & Hotz, H.-R. (1996). Post-transcriptionalcontrol of PARP gene expression. Mol. Biochem.Parasitol. 77, 1±6.

Couto, A. S., Lederkremer, R. M., Colli, W. & Alves,M. J. M. (1993). The glycosylphosphatidylinositolanchor of the trypomastigote-speci®c Tc-85 glyco-protein from Trypanosoma cruzi. Metabolic labelingand structural studies. Eur. J. Biochem. 217, 597±602.

Cross, G. A. M. (1990). Cellular and genetic aspects ofantigenic variation in Trypanosomes. Annu. Rev.Cell Biol. 8, 83±100.

Dahlberg, A. E., Dingman, C. W. & Peacock, A. C.(1969). Electrophoretic characterisation of bacterialpolyribosomes in agarose-acrylamide compositegels. J. Mol. Biol. 41, 139±147.

Dorn, P. L., Aman, R. A. & Boothroyd, J. C. (1991). Inhi-bition of protein synthesis results in superinductionof procyclin (PARP) RNA levels. Mol. Biochem. Para-sitol. 44, 133±140.

Engstler, M., Reuter, G. & Schauer, R. (1993). Thedevelopmentally regulated trans-sialidase from Try-panosoma bruceisialylates the procyclic acidic repeti-tive protein. Mol. Biochem. Parasitol. 61, 1±14.

Evans, J. S., Levine, B. A., Trayer, I. P., Dorman, C. J. &Higgins, C. F. (1986). Sequence-imposed structuralconstraints in the TonB protein of E. coli. FEBSLetters, 208, 211±216.


Fankhauser, C., Homans, S. W., Thomas-Oates, J. E.,McConville, M. J., Desponds, C., Conzelmann, A. &Ferguson, M. A. J. (1993). Structures of the glycosyl-phosphatidylinositol membrane anchors from Sac-charomyces cerevisiae. J. Biol. Chem. 268, 26365±26374.

Ferguson, M. A. J. (1992a). Site of palmitoylation of aphospholipase C-resistant glycosyl-phosphatidylino-sitol membrane anchor. Biochem. J. 284, 297±300.

Ferguson, M. A. J. (1992b). Chemical and enzymicanalysis of glycosyl-phosphatidylinositol anchors. InLipid Modi®cation of Proteins: A Practical Approach(Turner, A. J. & Hooper, N., eds), pp. 191±230, IRLPress, Oxford.

Ferguson, M. A. J., Murray, P., Rutherford, H. &McConville, M. (1993). A simple puri®cation of pro-cyclic acidic repetitive protein and demonstration ofa sialylated glycosyl-phosphatidylinositol mem-brane anchor. Biochem. J. 291, 51±55.

Fernandes, A. P., Nelson, K. & Beverley, S. M. (1993).Evolution of nuclear ribosomal RNAs in kinetoplas-tid protozoa: perspectives on the age and origins ofparasitism. Proc. Natl Acad. Sci. USA, 90, 11608±11612.

Field, M. C., Menon, A. K. & Cross, G. C. (1991a).A glycosylphosphatidylinositol protein anchor fromprocyclic stage Trypanosoma brucei: lipid structureand biosynthesis. EMBO J. 10, 2731±2739.

Field, M. C., Menon, A. K. & Cross, G. A. M. (1991b).Developmental variation of glycosylphosphatidyli-nositol membrane anchors in Trypanosoma brucei.J. Biol. Chem. 266, 8392±8400.

Green, M. R., Pastewka, J. V. & Peacock, A. C. (1973).Differential staining of phosphoproteins on poly-acrylamide gels with a cationic carbocyanin dye.Anal. Biochem. 56, 43±51.

GuÈ ther, M. L. S., Treumann, A. & Ferguson, M. A. J.(1996). Molecular species analysis and quanti®cationof the glycosylphosphatidylinositol intermediateglycolipid C from Trypanosoma brucei. Mol. Biochem.Parasitol. 77, 137±145.

Heise, N., Almeida, M. L. C. D. & Ferguson, M. A. J.(1995). Characterisation of the lipid moiety of theglycosylphosphatidylinositol anchor of Trypanosomacruzi 1G7-antigen. Mol. Biochem. Parasitol. 70, 71±84.

Jackson, D. G., Smith, D. K., Luo, C. & Elliot, J. F.(1993). Cloning of a novel surface antigen from theinsect stages of Trypanosoma brucei by expression inCOS cells. J. Biol. Chem. 268, 1894±1900.

KoÈnig, E., Delius, H., Carrington, M., Williams, R. O. &Roditi, I. (1989). Duplication and transcription ofprocyclin genes in Trypanosoma brucei. Nucl. AcidsRes. 17, 8727±8739.

Laemmli, U. K. (1970). Cleavage of structural proteinsduring the assembly of the head of bacteriophageT4. Nature, 277, 680±685.

Maudlin, I. & Welburn, S. C. (1994). Maturation of try-panosome infections in tsetse. Expt. Parasitol. 79,202±205.

McConville, M. J. & Ferguson, M. A. J. (1993). The struc-ture, biosynthesis and function of glycosylatedphosphatidylinositols in the parasitic protozoa andhigher eucaryotes. Biochem. J. 294, 305±324.

McConville, M. J., Turco, S. J., Ferguson, M. A. J. &Sacks, D. L. (1992). Developmental modi®cation oflipophosphoglycan during the differentiation ofLeishmania major promastigotes to an infectiousstage. EMBO J. 11, 3593±3600.

Molyneux, D. H. & Ashford, R. W. (1983). The Biology ofTrypanosoma and Leishmania Parasites of Man andDomestic Animals, Taylor Francis Ltd, London.

Mowatt, M. R. & Clayton, C. E. (1987). Developmentalregulation of a novel repetitive protein of Trypano-soma brucei. Mol. Cell. Biol. 7, 2838±2844.

Mowatt, M. R. & Clayton, C. E. (1988). Polymorphismin the procyclic acidic repetitive protein gene familyof Trypanosoma brucei. Mol. Cell. Biol. 8, 4055±4062.

Mowatt, M. R., Wisdom, G. S. & Clayton, C. E. (1989).Variation of tandem repeats in the developmentallyregulated procyclic acidic repetitive proteins of Try-panosoma brucei. Mol. Cell. Biol. 9, 1332±1335.

Pontes de Carvalho, L. C., Tomlinson, S.,Vandekerckhove, F., Bienen, E, J., Clarkson, A. B.,Jiang, M.-S., Hart, G. W. & Nussenzweig, V. (1993).Characterization of a novel trans-sialidase of Trypa-nosoma brucei procyclic trypomastigotes and identi®-cation of procyclin as the main sialic acid acceptor.J. Expt. Med. 177, 465±474.

Raj, T. & Flygare, W. H. (1974). Diffusion studies ofbovine serum albumin by quasielastic lightscattering. Biochemistry, 13, 3336±3340.

Richardson, J. P., Beecroft, R. P., Tolson, D. L., Liu,M. K. & Pearson, T. W. (1988). Procyclin: an unu-sual immunodominant glycoprotein surface antigenfrom the procyclic stage of Africa trypanosomes.Mol. Biochem. Parasitol. 31, 203±316.

Roberts, W. L., Myher, J. J., Kuksis, A., Low, M. G. &Rosenberry, T. L. (1988). Lipid analysis of the gly-coinositol phospholipid membrane anchor ofhuman erythrocyte acetylcholinesterase. Palmitoyla-tion of inositol results in resistance to phosphatidy-linositol-speci®c phospholipase C. J. Biol. Chem. 263,18766±18775.

Roditi, I., Carrington, M. & Turner, M. (1987). Ex-pression of a polypeptide containing a dipeptiderepeat is con®ned to the insect stage of Trypanosomabrucei. Nature, 325, 272±274.

Roditi, I., Schwarz, H., Pearson, T. W., Beecroft, R. P.,Liu, M. K., Richardson, J. P., BuÈ hring, H.-J., Pleiss,J., BuÈ low, R., Williams, R. O. & Overath, P. (1989).Procyclic gene expression and loss of the variantsurface glycoprotein during differentiation of Trypa-nosoma brucei. J. Cell Biol. 108, 737±746.

Rudenko, G., Bishop, D., Gottesdiener, K. & Ploeg,L. H. T. (1989). Alpha-amanitin resistant transcrip-tion of protein coding genes in insect and blood-stream form Trypanosoma brucei. EMBO J. 8, 4529±4263.

Rudenko, G., Blanco, S. I., Smith, J., Lee, M. G. S.,Rattray, A. & Ploeg, L. H. T. (1990). Procyclic acidicrepetitive protein (PARP) genes located in an unu-sually small a-amanitin-resistant transcription unit.Mol. Cell. Biol. 10, 3492±3504.

Sacks, D. L., Pimenta, P. F., McConville, M. J.,Schneider, P. & Turco, S. J. (1995). Stage-speci®cbinding of Leishmania donovani to the sand ¯y vectormidgut is regulated by conformational changes inthe abundant surface lipophosphoglycan. J. Expt.Med. 181, 685±697.

Schenkman, S., Eichinger, D., Pereira, M. A. &Nussenzweig, V. (1994). Structural and functionalproperties of Trypanosoma trans sialidase. Annu.Rev. Microbiol. 48, 499±523.

Schneider, P. & Ferguson, M. A. J. (1995). Microscaleanalysis of glycosylphosphatidylinositol structures.Methods Enzymol. 250, 614±630.


Schneider, P., Ferguson, M. A. J., McConville, M. J.,Mehlert, A., Homans, S. W. & Bordier, C. (1990).Structure of the glycosyl-phosphatidylinositol mem-brane anchor of the Leishmania major promastigotesurface protease. J. Biol. Chem. 265, 16955±16964.

Schneider, P., Ralton, J. E., McConville, M. J. &Ferguson, M. A. J. (1993). Analysis of the neutralglycan fractions of glycosyl-phosphatidylinositolsby thin-layer chromatography. Anal. Biochem. 210,106±112.

Scudder, P., Lawson, A. M., Hounsell, E. F., Carruthers,R. A., Childs, R. A. & Feizi, T. (1987). Characteris-ation of oligosaccharides released from human-blood-group O erythrocyte glycopeptides by theendo-b-galactosidase of Bacteroides fragilis. Eur. J.Biochem. 168, 585±593.

Serrano, A. A., Schenkman, S., Yoshida, N., Mehlert,A., Richardson, J. M. & Ferguson, M. A. J. (1995).The lipid structure of the glycosylphosphatidylino-sitol-anchored mucin-like sialic acid acceptors ofTrypanosoma cruzi changes during parasite differen-tiation from epimastigotes to infective metacyclictrypomastigote forms. J. Biol. Chem. 270, 27244±27253.

Smith, T. K., Cottaz, S., Brimacombe, J. S. & Ferguson,M. A. J. (1996). Substrate speci®city of the dolicholphospahte mannose:glucosaminyl phosphatidylino-sitol a1-4 mannosyltransferase of the glycosylpho-

sphatidylinositol biosynthetic pathway of Africantrypanosomes. J. Biol. Chem. 271, 6476±6482.

Towbin, H., Staehelin, T. & Gordon, J. (1979). Electro-phoretic transfer of proteins from polyacrylamidegels to nitrocellulose sheets: procedure and someapplications. Proc. Natl Acad. Sci. USA, 76, 4350±4354.

Treumann, A., Lifely, M. R., Schneider, P. & Ferguson,M. A. J. (1995). Primary structure of CD52. J. Biol.Chem. 270, 6088±6099.

Welburn, S. C., Maudlin, I. & Molyneux, D. H. (1994).Midgut lectin activity and sugar speci®city in ten-eral and fed tsetse. Med. Vet. Entomol. 8, 81±87.

Welburn, S. C., Dale, C., Ellis, D., Beecroft, R. &Pearson, T. W. (1996). Apoptosis in procyclic Trypa-nosoma brucei rhodesiense in vitro. Cell Death Different.3, 229±236.

Zamze, S. E., Ashford, D. A., Wooten, E. W.,Rademacher, T. W. & Dwek, R. A. (1991). Structuralcharacterization of the asparagine-linked oligosac-charides from Trypanosoma brucei type II and typeIII variant surface glycoproteins. J. Biol. Chem. 266,20244±20261.

Ziegelbauer, K., Quinten, M., Schwarz, H., Pearson,T. W. & Overath, P. (1990). Synchronous differen-tiation of Trypanosoma brucei from bloodstream toprocyclic forms in vitro. Eur. J. Biochem. 192, 373±378.

Edited by I. B. Holland

(Received 16 December 1996; received in revised form 14 March 1997; accepted 18 March 1997)

Note added in proof: An independent study (BuÈ tikofer, P., Ruepp, S., Boschung, M. Roditi, I. (1997).Biochem. J. In the press) also describes the expression of GPEET-PARP by procyclic trypanosomes.

Documents

Structural characterisation of two forms of procyclic acidic repetitive protein expressed by procyclic forms of Trypanosoma brucei