Evidence for a Mr88000 glycoprotein with a transmembrane association

to a unique flagellum attachment region in Trypanosoma brucei

ANGELA WOODS, ANTHONY J. BAINES and KEITH GULL*

Biological Laboratory, University of Kent, Canterbury, Kent CT2 7NJ, UK

* Present address: Department of Biochemistry & Molecular Biology, University of Manchester, School of Biological Sciences, StopfordBuilding, Oxford Road, Manchester M13 9PT, UK

Summary

We have examined the relationship of externallyaccessible proteins associated with the internalcytoskeleton of procyclic Trypanosoma brucei.Two approaches were taken. First, externally dis-posed glycoproteins were identified with lectins andexamined for their persistence and location inisolated cytoskeletons. Second, proteins containingtyrosine residues available for chemical modifi-cation on the outer surface were identified in iso-lated cytoskeletons and probed for glycosylation.The procyclic form of T. brucei that was employeddoes not express the variable surface glycoprotein.The lectin concanavalin A (ConA) bound to theouter surface of T. brucei in two discrete locations;one a narrow line close to the flagellum attachmentzone on the cell body, the other at the distal tip ofthe flagellum itself. Of these, only the cell bodylabelling was detected when isolated cytoskeletons

were probed with fluorescein isothiocyanate-labelled ConA. When cytoskeletons were preparedfrom cells labelled with gold-conjugated ConA, anarrow line of label was detected parallel to theflagellum attachment zone but was distinct from it.Only one cytoskeletal protein, of Mr 88 000, could belabelled at the cell surface by the 125l/iodogenprocedure. This protein could be precipitated fromSDS-solubilized cytoskeletons with ConA-agarose.These data indicate the existence of a previouslyundetected cytoskeletal structure, situated in thecell body, close to the point of flagellum attach-ment, which has a transmembrane association withan external Mr88000 glycoprotein.

Key words: Trypanosoma brucei, transmembraneglycoprotein, flagella attachment.

Introduction

Trypanosoma brucei has a highly ordered plasma mem-brane-associated cytoskeleton that displays distinct re-gional specialization. In the cell body an array of micro-tubules underlies the plasma membrane and is orientedparallel to the long axis of the cell. Bridges can bediscerned between microtubules and between the plasmamembrane and the microtubules (Vickerman, 1985;Jensen & Smaill, 1986). The nature of the membrane-microtubule cross-bridges is not clear, but two candidateproteins have been described: a yV/r40000 microtubule-binding protein (Schneider et al. 1988) that has a Ca2+-dependent interaction with phospholipids and avV/r60000 microtubule-binding protein that has a phe-nothiazine-sensitive interaction with lipid vesicles(Steiger & Seebeck, 1986). Elsewhere, in the flagellumand in the specialized attachments of the flagellum to thecell body, although distinct cross-bridges have beendetected in the electron microscope their nature remainsunknown.

Journal of Cell Science 93, 501-508 (1989)Printed in Great Britain © The Company of Biologists Limited 1989

One approach to the elucidation of the nature oflinkages between the cytoskeleton and the plasma mem-brane would be to examine proteins accessible on theoutside of the cell that may have a transmembrane linkageto the cytoskeleton. Many examples are now known oftransmembrane proteins that form membrane-bindingsites for the cytoskeleton: amongst others are the integrinfamily, which link microfilaments to the membrane in theadhesion plaque of moving fibroblasts (Tankun et al.1986) and the ankyrin-binding transmembrane proteinsin erythrocytes (Bennett & Stenbuck, 1979) and brain(Srinivasan et al. 1988; Treharne et al. 1988).

Evidence from several lines of approach suggests thatthere are proteins on the surface of Trypanosoma otherthan the variable surface glycoprotein: such proteins canbe detected with lectins (Stricklere/ al. 1978; Frommel &Balber, 1987), cell surface iodination (Gardiner et al.1983; Giovanni-De Simone et al. 1988), metabolic label-ling (Mancini et al. 1982) and proteolysis (Seed et al.1976). In the present paper we investigate the cell surface

501

proteins of the procyclic form of T. brucei with a view todefining such possible transmembrane linkages to thecytoskeleton.

Materials and methods

TrypanosomesProcyclic Trypanosoma bnicei brucei, stock 427 were grown intissue-culture flasks in semi-defined medium 79 (Brun &Schonenberger, 1979).

LectinsThe lectins concanavalin A (ConA), wheat-germ agglutinin(WGA) and peanut agglutinin (PNA), their fluorescein isothio-cyanate (FITC) and gold conjugates were obtained fromSigma. [125I]WGA was prepared by the iodogen method(Bainese/ al. 1982).

Lectin fluorescenceCells were harvested by centrifugation and washed in phos-phate-buffered saline (PBS): 0-14M-NaCl, 2-5mM-KCl, 8mM-Na2HPO4, 1 niM-KH2PO4, pH7-4. They were then processedvia one of the following protocols.

Labelling of ivhole cells. Cells were incubated at roomtemperature for 30 min with FITC-conjugated lectin(SO^gml"1), bovine serum albumin (BSA) (68ftgml~ ) inPBS, with slow rotation, washed three times in PBS, fixedbriefly in suspension in 3-7% formaldehyde, washed in PBSthen settled on poly-L-lysine-coated slides and mounted inMowiol 488 (Harco, Harlow, UK) containing lragral"/>-phenylenediamine.

Preparation of cytoskeletons from labelled cells. Cells afterlabelling and washing were extracted in 0-5 % Nonidet P40(NP40), lOOmM-Pipes, pH6-9, 2mM-EGTA, 1 niM-MgSO4,OlmM-EDTA (PEME), on ice for 5 min, washed, fixed in3 7 % formaldehyde, washed in PBS and settled onto poly-L-lysine-coated slides. The cytoskeletons were then incubated for40 min in anti-tubulin monoclonal antibody (KMX-1 ascites:Birkett et al. 1985) diluted 1:500 in PBS; unbound antibodywas removed by washing three times in PBS. The cells werethen incubated in rhodamine-conjugated rabbit anti-mousesecondary antibody (Dakopatts) diluted 1:10 in PBS. Slideswere then washed in PBS and mounted as above.

Labelling of cytoskeletons. Cells after harvesting and washingwere extracted on ice in PEME/0'5 % NP40, washed and fixedbriefly in 3 7 % formaldehyde/PBS before being incubated withlectin, washed and settled as above. The cytoskeletons werethen labelled with KMX-1 and rhodamine second antibody, andmounted as before.

In control experiments 0-2M of the appropriate competingsugar was added during labelling: methylo-D-mannopyranoside for ConA, A'-acetylglucosamine forwheat-germ agglutinin and D-galactose for peanut agglutinin.

Lectin—goldCells were washed and settled onto Formvar-filmed carbon-coated grids and processed according to one of the followingprotocols.

labelling of ivhole cells. Whole cells were blocked in 1 %BSA in PBS (BPBS), incubated in 20nm gold-conjugatedConA (Sigma) (diluted ten times in PBS containing 1 mM-CaCl2, lmM-MnCl2, 0-5% BSA), washed three times inBPBS, twice in 0 1 % BSA/PBS and once in PEME. The gridswere then fixed in 2-5% glutaraldehyde in PEME and nega-tively stained in 2% ammonium molybdate, pH7.

Preparation of cytoskeletons from labelled cells. After label-ling and washing as above, the cells were extracted in 0 '5%NP40/PEME, washed in PEME before being fixed and stainedas above.

Labelling of cytoskeletons. After settling on grids, cells wereextracted in NP40/PEME and fixed in 3-7% formaldehyde,0-7% glutaraldehyde before labelling, washing, refining andstaining as above.

SDS-PAGE and blotting

SDS-PAGE was done as described by Laemmli (1970). Gelswere stained with Coomassie Blue R250 or the proteins weretransferred to nitrocellulose paper and probed with ConAaccording to the method of Treharne et al. (1988).

Cell surface labellingThis was carried out essentially by the method of Markwell &Fox (1978). l,3,4,6-tetrachloro-3,6-diphenyl glycouril (Sigma)was dissolved in chloroform and used to coat the reaction vesselsby evaporation of the solvent. Washed cells in 100 /.tl PBS,pepstatin (S^gniF1), chymostatin (5/igmF1), leupeptin(50^(gml~') and phenylmethylsulphonyl fluoride (PMSF)(SOjUgmP1) were added to the reaction vessel. 50ftCi of Na12sl(Amersham: IMS30) was added, the tubes were mixed well andleft for 10 min before removing the labelled cells from thereaction tubes. The cells were washed free of unbound 1Z5I withPBS.

Precipitation of ConA binding proteins from 1251-labelledcellsCytoskeletons were prepared from 125I-surface-labelled cells byextraction on ice for 5 min with 0 5 % NP40/PEME and washedwith PBS. The samples were made 1 % in SDS, boiled for2 min and diluted with lOvols of PBS. ConA-Sepharose beads(a 50 % slurry in PBS) were added, in the presence or absence of0-2M-<r-methylniannoside, and mixed for 1 h. Unbound ma-terial was removed by centrifugation of the beads through 10%sucrose, PBS, 1 % SDS. The pellets were boiled in gel samplebuffer and analysed by SDS-PAGE and autoradiography.

Fractionation of isolated cytoskeletonsThe subpellicular microtubules may be separated from theflagellum and flagellum attachment zone (FAZ) by salt extrac-tion. Isolated cytoskeletons were prepared from 12sI-surface-labelled cells and incubated on ice for 40 min in PEME, 1 M-KC1, in order to solubilize the subpellicular microtubules. Theflagellum/FAZ fraction was collected by centrifugation at11600#av for lOmin.

Results

T. brucei cytoskeleton isolation

In this study cytoskeletons were required that satisfiedthe following criteria. First, the morphology of the intactcell should be preserved; and second, the cytoskeletonsshould be lipid-free to prevent the possible artifact ofproteins being retained in the isolated cytoskeleton viaassociation with residual lipid. Of the detergents tested,only Nonidet P40, Triton X-100 and CHAPS (in higherconcentrations) satisfied these two criteria: at concen-trations of 0-1-1% lipid-free, morphologically intactcytoskeletons were obtained. Other detergents either didnot produce lipid-free cytoskeletons (Tween 20, Tween80 and Brij 58) or resulted in the destruction of the

502 A. Woods et al.

morphology (Surfynol). In this paper, all cytoskeletonswere prepared using 0 5 % Nonidet P40, as described inMaterials and methods.

Binding of labelled lectins to whole cells andcytoskeletonsThe lectins concanavalin A (ConA: specificity, a-D-glucosyl or mannosyl residues), wheat-germ agglutinin(WGA: specificity, TV-acetyl-D-glucosamine or sialic acidresidues) and peanut agglutinin (PNA: specificity,iV-acetylgalactosamine or D-galactose) were all tested fortheir ability to bind to intact cells and isolated cytoskel-etons. No binding of FITC-PNA or FITC-WGA toisolated cytoskeletons could be detected by fluorescencemicroscopy. Using radioligand binding, [ I]WGA gaveno specific binding to cytoskeletons. However, both[12SI]WGA and FITC-WGA bound to intact cells, butthe binding was weak and only partially inhibited by0-2M-iV-acetylglucosamine (in agreement with Mutharia& Pearson, 1987). Thus, no distinct glycoprotein associ-ation with the cytoskeleton could be detected with theselectins.

By contrast, FITC-ConA bound strongly to bothcytoskeletons and intact cells: representative labellingpatterns are shown in Fig. 1. On intact cells (Fig. 1A,B)FITC-ConA bound preferentially to an area of the cellbody close to the flagellum, the tip of the flagellum itselfand also gave a more diffuse labelling pattern over theremainder of the cell. When isolated cytoskeletons werelabelled (Fig. 1F,G,H), only the line close to the flagel-lum was preserved, the flagellum tip and cell bodylabelling seeming to be removed by NP40 extraction.However, if whole cells were first labelled with F I T C -ConA and then extracted with NP40 (Fig. 1C,D,E) theflagellum tip labelling was preserved in about 50 % of thecytoskeletons. No labelling of any structure was detectedin the presence of 0-2M-a-methylmannoside.

These results suggest that a structure close to the zoneof flagellum attachment to the cell body has a stabletransmembrane association with the cytoskeleton. Theflagellum tip labelling is a more complex case, with theinteraction only preserved if the determinants are boundto (and presumably cross-linked with) the tetravalentConA before NP40 extraction. This type of modulatedinteraction is well known in other organisms; forexample, transmembrane glycoproteins become cytoskel-eton-associated in lymphocytes (Turner et al. 1987) anderythrocytes (Chassis et al. 1988) when cross-linked bylectins.

A higher-resolution view of the ConA binding sites wasobtained using ConA conjugated to 20 nm gold particles.Whole cells, settled onto electron microscope grids andincubated with ConA-gold, showed a clear line of labelseemingly at the point where the cell body meets theflagellum (Fig. 2A,B), confirming the results in Fig. 1A.However, only a few of the cells treated this waydisplayed the flagellum tip labelling. We cannot saywhether this is a result of limited access of the largeConA-gold particle to the flagellum tip, or if, forexample, cell surface determinants are restricted in theirdiffusion on the surface of the settled cell and so are

unable to form a cap structure. Fig. 2C and D shows thatbinding of ConA to the cell surface is abolished by thespecific competing sugar, cv-methylmannoside.

Cytoskeletons prepared from settled cells incubatedwith ConA-gold retain the line of label along the cell.Labelling appears to be closely associated with flagellumattachment (Fig. 3), and overlays the junction of the cellbody and flagellum. However, if the flagellum is detachedfrom the cell body, the line of label consistently remainswith the cell body. Previous work (cf. the accompanyingpaper, Woods et al. 1989) has identified a punctatestructure in the cell body that appears to be the flagellumattachment zone (FAZ). The line of ConA-gold label isclearly distinct from this: it covers the length of the cellparallel to, but disconnected from, the FAZ. We areunable to detect a specific structure in this area bynegative staining of cytoskeletons.

Glycoproteins associated with the cytoskeletonGlycoproteins containing a'-D-glucosyl or a'-D-mannosylresidues were analysed by SDS—PAGE and Westernblotting with ConA and peroxidase (Treharne et al.1988). More than 50 ConA positive bands were found inextracts from whole cells (Fig. 4, lane B2), of which onlyabout seven were retained in isolated cytoskeletons:Mr88000, 86000, 82 000, 80000, 64000, 62 000 and10000 (Fig. 4, lane B3). None of these bands wasrecognized by Coomassie Blue staining of replicate gels orAmido Black staining of replicate blots, a commonproperty of glycoproteins (Dzandu et al. 1984). Thebinding of ConA and peroxidase appeared specific since:(1) it could be completely abolished by addition of 0-2 M-cr-methylmannoside during the ConA binding step (notshown); (2) tubulin and other non-glycosylated proteinsshowed no binding; and (3) ovalbumin, the Mr standardat 42 000 (Fig. 4, lane Bl) bound ConA strongly.

Any or all of the cytoskeleton-retained glycoproteinscould potentially be present in the structure observedusing FITC-ConA and ConA-gold. To clarify this, itwas necessary to determine whether any of these proteinswere surface-exposed. The iodogen/ I method (Mark-well & Fox, 1978) can be used to label selectively cellsurface proteins with accessible tyrosine residues. Pro-teins that are candidates for components of the structurewe have observed should: (1) be accessible from theoutside of the cell to iodogen/ I or proteases; or (2) beprecipitable with ConA. Fig. 5 shows the results of cellsurface labelling. Two predominant species have tyrosineavailable to iodogen/ I at the cell surface, Mr 88 000 andMT10000 (Fig. 5, lane 2). The Mr88 000 band remainsassociated with the isolated cytoskeleton (Fig. 5, lane 3),and is accessible to proteases in whole cells (Fig. 5, lane10). As an additional control to test that no internalproteins were being labelled in intact cells, cells werelysed with 0-5 % NP40/PEME before labelling: lysis, asexpected, permitted labelling of many internal proteinsincluding tubulin (Fig. 5, lane 4). ThevV/r88000 speciesappears as a broad band on the autoradiographs (Fig. 5),but by using shorter exposures we have been unable toresolve this into more than one band.

ConA is known to retain its activity in moderate

Flagellum attachment region in T. brucei 503

1A

Fig. 1. Lectin-fluorescence images, using ConA conjugated to F1TC, of T. bntcei. Phase-contrast (A) and fluorescence (B)images of whole, fixed cells labelled with FITC-ConA. Phase-contrast (C) and fluorescent (E) images of detergent-extracted,labelled cells. Phase-contrast (F) and ConA-FITC fluorescence staining (H) patterns of cells detergent-extracted beforelabelling. D and G represent the image of the cytoskeleton after labelling with rhodamine-labelled anti-tubulin antibody, KMX,thus showing the cytoskeleton to be intact.

concentrations of SDS, and thus can precipitate SDS-denatured glycoproteins. To determine whether theMr88000 species is glycosylated, cytoskeletons wereprepared from 125I-surface-labelled cells and denaturedin SDS. Glycoproteins were precipitated with ConAbound to Sepharose and analysed by SDS—PAGE andautoradiography. The jV/r88000 band was precipitated

with ConA-Sepharose (Fig. 5, lane 5) the precipitationbeing inhibited by 0'2M-a'-methylmannoside (lane 6):underivatized Sepharose likewise produced no precipi-tation (lane 7). We conclude that the MT88 000 species isan externally disposed glycoprotein that is associated withthe cytoskeleton.

In isolated cytoskeleton preparations, the microtubules

504 A. Woods et al.

ftB

Fig. 2. A,B- Electron micrograph showing lectin-gold-labelling pattern of whole, unfixed cells labelled with gold-conjugatedConA. B. An enlarged area of A. C,D. ConA binding inhibited by labelling in the presence of 0-2M-a-inethylmannoside.D. Enlarged area of cell shown in C.

3A

B

Fig. 3. A. An electron micrograph of a ConA-gold-labelledcell that has been detergent extracted. B. An enlarged versionof box in A.

of the cell body can be separated from the residues of theflagellum and flagellum attachment zone by extractionwith the 1 M-KC1 solutions (cf. the accompanying paper,Woods et al. 1989). The A/r88 000 band remained withthe flagellum/attachment zone fraction after KC1 extrac-tion, consistent with the possibility of an attachment zoneassociation (Fig. 5, lane 8).

In summary, ConA uniquely recognizes a structure inthe cytoskeleton of T. bntcei related to flagellum attach-ment. This structure seems to contain an externallyaccessible Mr 88 000 glycoprotein.

Discussion

One of the critical factors dictating the organization anddynamics of the plasma membrane is its association withan underlying cytoskeleton. The nature of this interac-tion varies from case to case, and it is now becoming clearthat cells with highly differentiated surfaces use differentmechanisms to meet specific requirements at variouspoints on the plasma membrane. In this paper we haveaddressed the question of whether the cytoskeleton of thetrypanosome cell has a transmembrane linkage with thesurface.

Flagellum attachment region in T. brucei 505

205-

116-97-

66-

45-

2 9 -

The results presented here suggest that the plasmamembrane of T. bnicei shows more regional specializ-ation than was previously supposed. The lectin concan-avalin A reveals two areas of the cell surface that possessdeterminants with a transmembrane linkage to the cyto-

1 8

Fig. 4. ConA-labelled glycoproteins in T. brucei.A. Coomassie Blue-stained gel; and B, thecorresponding blot probed with ConA andhorseradish peroxidase. Lanes 1, molecular weightmarkers; lanes 2, whole-cell lysate of T. brucei; andlanes 3, T. brucei cytoskeletons. Lanes 2 and 3 ineach case were loaded with extracts prepared fromequal numbers of cells.

skeleton, one being on the cell body close to the zone offlagellum attachment (hereinafter called the glycoproteintrack (GT)), the other at the tip of the flagellum itself.The flagellum tip labelling is only preserved in thecytoskeleton if the cells are reacted with ConA before

10

Fig. 5. Autoradiogram of iodinated cellextracts. Lanes 1, unlabelled cells;2, iodinated whole cells; 3, detergent-insoluble extract of iodinated cells;4, iodinated cytoskeletons; 5, solubilizedlabelled cells precipitated withConA-Sepharose; 6, iodinated cellsprecipitated with ConA-Sepharose in thepresence of O2M-methyl-a/-D-mannopyranoside; 7, iodinated cellsprecipitated with uncoupled Sepharose;8, flagellum containing pellet from labelledwhole cells after detergent and saltextraction; 9, supernatant from saltextraction of iodinated cells; 10, cellstrypsinized before iodination.

506 A. Woods et a I.

cytoskeleton preparation: thus it is possible to preserveuniquely the GT as the only structure containing ConAbinding proteins, by Nonidet P40 extraction of otherwiseuntreated cells.

Only one cytoskeleton-associated protein can belabelled at the cell surface by the iodogen/12sI procedure:since this protein is precipitable with ConA, it seems verylikely that this is a component of the GT. What is thenature of the structure we have detected? Despite exhaus-tive searches in isolated cytoskeleton preparations, wehave not been able to detect a definitive structure in theGT area by negative staining procedures. It is, however,interesting to note that in this area there is a structuralspecialization of four microtubules (Vickerman & Pres-ton, 1976). These microtubules, which often remain withthe flagellum after salt extraction, may be involved incoordinating the GT zone that we have described.

There is no evidence as yet for whether the 88000A/r

glycoprotein itself is a transmembrane protein. We havebeen unable to solubilize it without using denaturingdetergents or chaotropes: these would be expected toperturb the results of, for example, tests for hydrophobi-city by the Triton X-114 procedure. However, Giovanni-De Simone et al. (1988), using the iodogen/125I pro-cedure with T. cruzi, labelled a surface-exposed glyco-protein that appears to have an identical Mr to the one wehave seen: in their hands it appeared hydrophobic by theTriton X-114 partition method, and would possibly betransmembrane. It will be of interest to determinewhether this is the case with our 88 000 MT protein andwhether it has tubulin-binding activity. Gardiner et al.(1986) also noted the presence of an 88000 Mr proteinexposed on the surface of T. rhodesiense; Frommel &Balber (1987) found a ConA binding protein of similarmolecular weight in a number of different isolates of T.brucei brucei, T. brucei gambiense and T. rhodesiense.These disparate observations suggest the possibility thatthe 88 000 MT protein may be highly conserved amongvarious trypanosomes. It will be necessary to developspecific antibodies to test this. Balber & Frommel (1988)described FITC-succinyl-ConA labelling of both cellsurface and flagellum in procyclic T.b. gambiense andT.b. rhodesiense. We observe much lower levels of cellbody labelling with T.b. brucei, scarcely above ar-methyl-mannoside controls. It is possible that the labelling thatthey attribute to the flagellum is in fact the GT.

It was most striking that the flagellum tip label wasonly preserved in cytoskeletons if they were preparedfrom cells with ConA bound to them. Such a phenom-enon has been observed elsewhere. Binding of an extra-cellular ligand to erythrocyte glycophorin A, for example,promotes its interaction with the cytoskeleton (Chassis etal. 1988; Bourguingnon & Bourguingnon, 1984). Theflagella of Chlamydamonas contain a glycoprotein ofMr 240 000, that is associated with the cytoskeleton (Rein-hardt & Bloodgood, 1988): it is distributed over the entireflagellum surface. On being cross-linked by monoclonalantibody or ConA, it forms a cap on the tip of theflagellum and is eventually shed into the medium (Blood-good et al. 1986). In contrast to this, we have neverobserved ConA binding over the entire flagellar surface of

T. brucei, arguing against the possibility that what we seeis a result of an analogous process. Equally, the pro-portion of cells displaying the flagellar tip is not greatlyaltered by preincubation of the cells in the cold or withsodium azide (not shown). Analysis of silver-stainedSDS-containing gels of cytoskeletons prepared fromConA-prelabelled cells shows that several bands areenhanced relative to unlabelled controls (not shown): it ispossible that these are components of the flagellar tip.These data suggest that the components of the T. bruceiflagellum tip have a modulated interaction with thecytoskeleton, which depends on external ligand bindingrather than cross-linking per se.

It is very striking that, whereas ConA-HRP blotsdetect more than 50 bands in whole cells, [125I]iodogenonly detects two. These results are consistent with thoseof other investigators (contrast the ConA blots obtainedby Frommel & Balber, (1987) with the results of cellsurface 125I labelling obtained by Giovanne-De Simoneet al. (1988)). Clearly, the two reagents detect differentchemical groups: it also seems likely that many of theglycoproteins must be intracellular (in lysosomes, Golgi,etc) and/or few of the cell surface proteins have tyrosineresidues accessible to [12sI]iodogen.

None of the lectins tested, or direct cell surfacelabelling with fluorescein isothiocyanate (not shown),revealed a general plasma membrane pattern of labellingthat was retained with the cytoskeleton. It now seemsvery unlikely that transmembrane proteins have a generalrole in linking with the cytoskeleton over the entireplasma membrane. However, this possibility cannot beexcluded because the reagents we used may not be able todetect such proteins, and/or their interaction(s) with thecytoskeleton may be disrupted by Nonidet P40. Never-theless, we would favour the view that proteins such asthe Afr40000 and 60000 proteins detected by Seebeck'sgroup (Schneider et al. 1988; Seebecketal. 1988), whichare tubulin- and lipid-binding proteins, may be theprincipal components linking the cytoskeleton to themembrane.

This work was supported by a project grant to A. J. Bainesand K. Gull from the Science and Engineering ResearchCouncil. This investigation received financial support from theUNDP/WORLD BANK/WHO Special Programme for Re-search and Training in Tropical Diseases.

References

BAINES, A. J., BANGA, J. P. S., LYNCH, D., HUEHNS, E. R. &GRATZER, W. B. (1982). Red cell membrane protein anomalies incongenital dyserythrpoietic anaemia, type II (HEMPAS). Br.J.Haem. 50, 563-574.

BALBER, A. E. & FROMMEL, T. O. (1988). Trypanosome bniceigambiense and T.b. rhodesiense: Concanavalin A binding to themembrane and flagellar pocket of blood stream and procyclicforms. J . Pwtozool. 35, 214-219.

BENNETT, V. & STENBUCK, P. (1979). The membrane attachmentprotein for spectrin associates with band 3 in human erythrocytes.Nature, Loud. 280, 463-473.

BIRKETT, C. R., FOSTER, K. E., JOHNSON, L. & GULL, K. (1985).

Use of monoclonal antibodies to analyse the expression of multi-tubulin family. FEBS Lett. 187, 211-218.

Flagellum attachment region in T. brucei 507

BLOODGOOD, R. A., WOODWARD, M. P. & SALOMENSKY, N. A.

(1986). Redistribution and shedding of flagellar membraneglycoproteins visualized using an anti-carbohydrate monoclonalantibody and concanavalin A. J . Cell Biol. 102, 1797-1812.

BOURGUINGNON, L. Y. N. & BOURGU1NGNON, G. J. (1984). Cappingand the cytoskeleton. Int. Rev. Cytol. 87, 195-224.

BRUN, R. & SCHONENBERGER, M. (1979). Cultivation and in vitrocloning of procyclic culture forms of Trypanosoma bnicei in asemi-defined medium. Acta tivpica 36, 289-292.

CHASSIS, J. A., REID, M. B., JENSEN, R. H. & MOHANDAS, M.(1988). Signal transduction by glycophorin A: role of extra-cellularand cytoplasmic domains in a modulate process. J'. Cell Biol. 107,1351-1368.

DZANDU, J., DEH, M. E., BARRATT, D. L. & WISE, G. E. (1984).Detection of erythrocyte membrane proteins, sialoglycoproteinsand lipids in the same polyacrylamide gel using a double stainingtechnique. Pmc. natn. Acad. Sci. U.S.A. 81, 1733-1737.

FROMMEL, T. 0 . & BALBER, A. E. (1987). Trypanosoma bniceibnicei, T. bnicei gainbiense and T. bnicei rhodesiense: commonglycoproteins and glycoprotein oligosaccharide heterogeneityidentified by lectin affinity blotting and endoglycosidase Htreatment. Expl Parasitol. 63, 32-41.

GARDINER, P. R., FINERTY, J. F. & DWYER, D. M. (1983).Iodination and identification of surface membrane antigens inprocyclic Trypanosoma rhodensiense.J. Immiin. 131, 454-457.

GIOVANNI-DE SIMONE, S., BONALDO, M. C , PINHO, R. T., PONTES-DA-CARVAUHO, L. C , GALVAO-CASTRO, B. & GOLDBERG, S.(1988). A study on the amphiphilic proteins of three Trypanosomacmzi populations. Brazilian J. med. Res. 21, 435-443.

JENSEN, C. G. & SMAILL, B. H. (1986). A technique for analysingthe spatial organisation of microtubular arms and bridges. Ann.N.Y.Acad. Sci. 446, 417-419.

LAEMMLI, U. K. (1970). Cleavage of structural proteins duringassembly stage of the head of bacteriophage T4. Nature, Land.227, 680-685.

MANCINI, P. E., STRICKLER, J. E. & PATTON, C. L. (1982).Identification and partial characterization of plasma membranepolypeptides of Trypanosoma bnicei. Biocluin. biophvs. Acta 688,399-410.

MARKWELL, M. A. K. & Fox, F. (1978). Surface specific iodinationof membrane proteins of viruses and eukaryotic cells using 1,3,4,6-tetrachloro-3a',6n'-diphenylglycoluril. Biochemistry 17, 4807-4817.

MUTHARIA, L. M. & PEARSON,V. W. (1987). Surface carbohydratesof procyclic forms of African trypanosomes: studies usingfluorescence activated cell sorter analysis and agglutination withlectins. Molec. biochem. Parasitol. 23, 165-172.

REINHART, F. D. & BLOODGOOD, R. A. (1988).

Membrane-cytoskeleton interactions in the flagellum: a 240 000 M,surface exposed glycoprotein is tightly associated with the axonemein Chlamydomonas moewusii. J. Cell Sci. 89, 521-531.

SCHNEIDER, A., EICHENBERGER, W. & SEEBECK, T. (1988). A

microtubule binding protein of Trypanosoma bnicei which containscovalently bound fatty acid. J. biol. Client. 263, 6472-6475.

SEED, T. M., SEED, J. R. & BRINDLEY, D. (1976). Surface

properties of bloodstream trypanosomes (Ttypaiiosoma bniceibnicei). Tropenmedizin und Parasitologie 27, 202-212.

SRINIVASAN, E. L., DAVIS, J., BENNETT, V. & ANGELIDES, K.

(1988). Ankyrin and spectrin associate with voltage dependentsodium channels in brain. Nature, Loud. 333, 177-180.

STEIGER, J. & SEEBECK, T. (1986). Monoclonal antibody against a60kDa phenothiazine binding protein from Trypanosoma bniceican discriminate between different trypanosome species. Molec.biochem. Parasitol. 21, 37-45.

STRICKLER, J. E., MANCINI, P. E. & PATTON, C. L. (1978).

Trypanosoma bnicei bnicei: isolation of the major surface coatglycoprotein by lectin affinity chromatographv. Expl Parasitol. 46,262-276.

TANKUN, S. W., D E SIMONE, D. W., FONDA, D., PATEL, R. S.,

BUCK, C , HORWITZ, F. & HYNES, R. O. (1986). Structure of

integrin, a glycoprotein involved in the transmembrane linkagebetween fibronectin and actin. Cell 46, 271-282.

TREHARNE, K. J., RAYNER, D. & BAINES, A. J. (1988). Identification

and partial purification of ABGP 205, an integral membraneglycoprotein from brain that binds ankyrin. Biochem. J. 253,345-350.

TURNER, C. E., NEWTON, M. R. & SHOTTEN, D. M. (1988).

Cytoskeleton involvement of the sequential capping of ratthymocyte surface glycoprotein. J. Cell Sci. 89, 308-314.

VICKERMAN, K. (1985). Developmental cycles and biology ofpathogenic trypanosomes. Br. med. Bull. 41, 105-114.

VICKERMAN, K. & PRESTON, T. M. (1976). Comparative cell biologyof the kinetoplastid flagellates. In Biologv of the Kinetoplastida(Lumdsden, W. H. R. & Evans, D. A.","eds), p p . 35-130.Academic Press, New York.

WOODS, A., SHERWIN, T., SASSE, R., MACRAE, T. H., BAINES, A.

J. & GULL, K. (1989). Definition of individual components ofTrypanosoma bnicei by a library of monoclonal antibodies. J. CellSci. 93, 491-500.

(Received 27January 1989-Accepted 6 April 19S9)

508 A. Woods et at.