The Identification, Purification, and Characterization of Two Invariant Surface Glycoproteins Located beneath the Surface Coat Barrier of Bloodstream Forms of Trypanosoma brucei”

Two new polypeptides, termed ISG70 and have been found in Trypanosoma brucei, using enzyme-catalyzed radioiodination techniques. Both are exter- nally disposed integral membrane glycoproteins, containing N-linked carbohydrate chains. No structural homology was detected between ISG70, ISGBI, or the variant surface glycoprotein (VSG) when assessed by 1) comparative peptide mapping, 2) immunoprecipitation analysis, and 3) lectin affinity chromatography. LSG7,, occurred in 5.1 X lo4 copies/cell and has been purified 880-fold from detergent extracts of plasma membranes by a procedure that includes gel filtration, lectin affinity chromatography, and preparative SDS- polyacrylamide gel electrophoresis. 1sG.10 was present only in bloodstream forms and was specifically de- tected in six different cloned variants from the Molten0 Institute trypanosomal antigen type (MITat) serodeme of T. brucei and from the single cloned variant of the International Laboratory for Research on Animal Dis- eases trypanosomal antigen type (ILTat) serodeme that was examined. Rabbits stage-specific combined implies


The Identification, Purification, and Characterization of Two Invariant Surface Glycoproteins Located beneath the Surface Coat Barrier of Bloodstream Forms of Trypanosoma brucei"
David G. Jackson$, Henry J. Windle, and H. Paul Voorheis$ From the Department of Biochemistry, Trinity College, Dublin 2. Ireland Two new polypeptides, termed ISG70 and have been found in Trypanosoma brucei, using enzymecatalyzed radioiodination techniques. Both are externally disposed integral membrane glycoproteins, containing N-linked carbohydrate chains. No structural homology was detected between ISG70, ISGBI, or the variant surface glycoprotein (VSG) when assessed by 1) comparative peptide mapping, 2) immunoprecipitation analysis, and 3) lectin affinity chromatography. LSG7,, occurred in 5.1 X lo4 copies/cell and has been purified 880-fold from detergent extracts of plasma membranes by a procedure that includes gel filtration, lectin affinity chromatography, and preparative SDSpolyacrylamide gel electrophoresis. 1sG.10 was present only in bloodstream forms and was specifically detected in six different cloned variants from the Molten0 Institute trypanosomal antigen type (MITat) serodeme of T. brucei and from the single cloned variant of the International Laboratory for Research on Animal Diseases trypanosomal antigen type (ILTat) serodeme that was examined.
Rabbits with chronic infections of T. brucei displayed circulating antibodies against ISGY0. Both the immunogenicity of ISG,, and its invariant nature suggest that it may be useful in the development of an effective serodiagnostic test. Furthermore, its stagespecific location combined with its invariant nature implies that its function is strictly related to a physiological role required for the parasite's residence in its mammalian host.
The African trypanosomes have enormous medical, veterinary, and economic importance. They also serve as good models of eukaryotic cell function. However, compared with other cells, little is known about their numerous surface macromolecules, which in the case of trypanosomes, lie beneath the VSG' or within the partly exposed regions of the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "oduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: VSG, variant surface glycoprotein; VSG,, membrane-bound form of the variant surface glycoprotein; VSG,, released form of the variant surface glycoprotein; ILTat, International Laboratory for Research on Animal Diseases trypanosomal antigen type; ISG, invariant surface glycoprotein; MITat, Mol-ten0 Institute trypanosomal antigen type; PBS, phosphate-buffered saline; BSA, bovine serum albumin; PMSF, phenylmethanesulfonyl fluoride; PAGE, polyacrylamide gel electrophoresis; TES, 2-1 [2-hydroxy-l,l-bis(hydroxymethyl)ethyl]amino)ethanesulfonic acid. flagellar pocket, where they may mediate many of the interesting interactions between host and parasite whose molecular basis remains to be established. In contrast to this situation considerable knowledge is available about the VSG itself (for reviews see . Several early studies identified a series of antigens of unknown function residing on the external surface of the plasma membrane. Several nonvariant surface antigens were restricted to the bloodstream forms of these cells, and limited data indicated that some of these antigens were reached by antibodies bathing live cells (4). Further reports identified an invariant doublet of M, = 22,000 on the cell surface (5) and glycoproteins of M , = 60,000 and 66,000 on the flagellar pocket membrane (6). Further evidence also indicated that the membrane fraction containing these glycoproteins, when used as an immunogen, provided incomplete protection to mice against subsequent challenge with live parasites, whereas an earlier study reported that whole purified plasma membranes provided no such protection in goats or rabbits (7).
A number of specific enzymes has also been localized to the plasma membrane of bloodstream forms of Trypanosoma brucei. For example both adenylate cyclase and the ouabainsensitive Na+/K+-ATPase were found in the plasma membrane (8), whereas about half of the acid phosphatase and acid phosphodiesterase appear to be located on the outer surface of the plasma membrane lining the flagellar pocket Several genes for other possible plasma membrane proteins, termed ESAGs for expression site-associated genes, have been identified in T. brucei and code for amphipathic proteins with typical membrane signal sequences and consensus sites for N-glycosylation (13-15). These proteins are expressed in bloodstream forms (16)(17)(18)(19)(20)(21) and metacyclic forms (20) but not in procyclic forms (17). ESAG 4 has been experimentally localized to that portion of the plasma membrane covering the flagellum (22), and sequence data identify it as the Ca2+activated adenylate cyclase reported previously (23) and now known to function only in bloodstream forms (24,25). ESAG 8 is a Zn2+ finger protein, containing leucine-rich repeats, that is expressed at very low levels (26, 27) and shows homology with the ras-interacting domain of yeast adenylate cyclase. This domain is absent from the Ca*+-regulated adenylate cyclase in T . brucei (19,27), suggesting that ESAG 8 may be the regulator of the ESAG 4 cyclase. The ESAG family of genes appears to be widely distributed among the African trypanosomes, including brucei, rhodesiense, gambiense, euansi, and equiperdum as well as probably present with some sequence alterations in congolense, viuax, and mega (14, 16, 20, 25) but absent from Trypanosoma cruzi, Leishmania tarentolae,and Crithidia fasciculata (16,25). A number of genes located outside of the VSG expression sites, termed GRES-(9-12).

8085
AGs for genes related to ESAGs, are expressed in both procyclic and bloodstream forms (24, 26, 28). Their protein products represent additional adenylate cyclases with extracellular domains different from that encoded by ESAG 4 and not regulated by CaZ+ (22, 25). Recently, a second family of genes, the BS1 family, has been identified that may also code for plasma membrane proteins (29).
Three plasma membrane receptors for ligands that circulate in their mammalian host have also been identified in bloodstream forms of T. brucei: the receptor for low density lipoprotein (30), a transferrin-binding protein which is the product of t h e ESAG 6 gene (31), and possibly a functional EGF receptor (32). Most recently a number of invariant surface glycoproteins or ISGs (33) have been identified on the surface of the bloodstream forms of T. brucei (34, 35). Two ISGs of unknown function have been cloned and sequenced, and their structure has revealed no obvious homology with other membrane glycoproteins (34, 35).
In this report we describe two novel invariant surface glycoproteins that possess extensive reactivity during cell surface radioiodination without similar reactivity during surface radiolabeling with [isethi~nyl-~HIacetimidate. One of these antigens, ISG70, has been partially purified, characterized, shown to be stage-specific, to occur in low copy number, and to be uniformly distributed over the surface of t h e cell.
Purification of ISG70 from T. brucei-An 880-fold purification of ISGTO from whole bloodstream forms of T. brucei was accomplished using the procedure detailed below. All steps were carried out between 0 and 5 "C unless specified otherwise. A suspension of freshly harvested trypanosomes ( 2 3 X 101Ocells) was mixed with surface-radioiodinated cells (4 X lo8) to provide a tracer or marker for ISGT0 during its purification. The mixed suspension was centrifuged (10,000 X g, 2 min), and the pellet was resuspended in 10 ml of TES buffer (20 mM, pH 7.5), containing NaCl (140 mM), KC1 (5 mM), glucose (10 mM), EDTA (1 mM), PMSF (1 mM), and leupeptin (50 pg/ml) and lysed by the addition of 10 volumes of water followed by incubation (37 "C, 5 min) to release the VSG. After centrifugation (10,000 X g, 2 rnin), the pellet of crude membranes was treated with DNase I exactly as described previously (8) before recentrifugation and resuspension in 5 ml of Tris.HC1 buffer (50 mM, pH 9.5) containing NaCl (150 mM) EDTA (1 mM), PMSF (0.5 mM), and leupeptin (50 pg/ml).
ISGs were extracted by the addition of an equal volume of Tris. HCl buffer (50 mM, pH 9.5), containing sodium deoxycholate (20%, w/v), and followed by storage on ice for 1 h prior to centrifugation (50,000 X g, 2 h). The clarified extract was applied to a column (3 X 50 cm) of Ultrogel AcA 34 that had been equilibrated with Tris'HCl buffer, pH 9.5, containing sodium deoxycholate (10 mM). After elution with the same buffer, fractions that contained 'z51-labeled ISGW (as assessed by SDS-PAGE/autoradiography) were pooled and their hydrogen ion concentration adjusted to pH 8.0 with Tris. HCI (1 M, pH 6.8). Pooled fractions were chromatographed on a column (2 ml) of concanavalin A-Sepharose that had been equilibrated in Tris'HCl (50 mM, pH 8.5) containing NaCl(l50 mM) and sodium deoxycholate (10 mM). The column was washed with the same buffer and bound glycoproteins were eluted with a gradient (0-0.5 M) of a-methylmannoside. 1*51-labeled ISGTO eluted over almost the entire gradient of amethylmannoside (as assessed by SDS-PAGE/autoradiography). The fractions were pooled, electrophoresed in parallel lanes on a 10% (w/ v) polyacrylamide SDS-PAGE gel, and the faint band in each lane corresponding to ISG7, (as detected by brief staining with Coomassie Blue) was excised, re-equilibrated with SDS-PAGE sample buffer, and re-electrophoresed on a 7.5% (w/v) polyacrylamide SDS-PAGE gel. The resulting ISG,, bands were located by Coomassie Blue staining and after excision the gel slices were washed in water and stored at -70 "C until required for immunization.
Source and Purification of Trypanosomes-Cloned variants of T. brucei (MITat 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, and 1.8) were gifts from Dr. G. A. M. Cross and ILTat 1.21 was the gift of Dr. M. J. Turner. Bloodstream forms of these organisms were grown in laboratory rats and purified as described previously (37). Procyclic forms of T. brucei were produced by transformation of MITat 1.1 in medium SDM-79 containing heat-inactivated (60 "C, 15 min) fetal calf serum (lo%, v/ v) and sodium citrate as described by Brun and Schonenberger (38) and then maintained in the same medium.
Metabolic Labeling of Trypanosomes with P5S]Methionine-Cells (2 X 107/ml, final volume 5 ml) were incubated with gentle stirring in methionine-free medium (minimum Eagle's medium) supplemented with L-glutamine (0.3 mg/ml) in the presence of [35S]methionine (500 pCi, >lo00 Ci/mmol) at 37 "C for 2.5 h. The incubation was terminated by washing the cells five times by centrifugation (9,000 X g for 30 s) and resuspension in ice-cold phosphate-buffered saline, pH 7.4.
Measurement of Iz5I Covalently Linked to Protein-Samples (5-50 p1) of radioiodinated cells or of subcellular fractions derived from radioiodinated cells were extracted into 20 volumes of a mixture of chloroform/methanol (2:1, v/v), and the resulting emulsions were filtered under negative pressure through Whatman GF/C glass fiber filter disks. The filters were then washed thoroughly with chloroform/ methanol (10 ml/filter), dried, and their content of radioactivity measured in a Packard y spectrometer. The recovery of radioiodinated protein on the filters was linear within the range 0.025-100 pg of protein and was independent of the type or concentration of detergent present or of the duration of the extraction or the temperature of the solvent.
Mild Trypsin Treatment of Radioiodinated Cells-A minor modification of the method of Taylor and Cross (41) was used. Freshly radioiodinated cells ( 108/ml) in phosphate buffer were incubated with trypsin (40 pg/ml) at 37 'C for 15 min prior to the addition of an equal volume of ice-cold phosphate buffer, containing soybean trypsin inhibitor (80 pg/ml). The cells were centrifuged (9,000 X g for 2 min) and the resulting pellets and supernatants analyzed by SDS-PAGE.
Osmotic Lysis of Cells-Cells (108/ml) suspended in phosphate buffer that had been surface-radioiodinated were mixed with 10 volumes of distilled-deionized water, containing leupeptin (50 pglml), PMSF (0.5 mM), and EDTA (1 mM) and incubated at 37 "c for 10 min. The cell lysates that resulted were centrifuged (10,000 X g for 2 min), and the pellets (containing crude plasma membranes) and supernatants (containing impure VSG,) were analyzed by SDS-PAGE or processed further as described in individual experiments.
Purification of Plasma Membranes-Plasma membranes were isolated from whole cells exactly as described by Voorheis et al. (8).
Protein Measurements-Protein was measured by the method of Markwell et al. (42) with bovine serum albumin as the standard.
Preparation of Detergent Lysates and Immunoprecipitates-The general approach of Anderson and Blobel (43) was used. Detergent lysates were made by mixing freshly surface radioiodinated cells (lo9/ ml) with an equal volume of a solution of SDS (4%, w/v) in phosphate buffer and boiling for 2 min. The boiled extracts were diluted with a 4-fold excess of phosphate buffer, containing Triton X-100 (2.5%, w/ v), PMSF (0.5 mM), leupeptin (50 pg/ml), and EDTA (1 mM) and then maintained at 0 "C for an hour, followed by centrifuging at 9,000 X g for 5 min. Immunoprecipitations were performed by mixing samples (0.1 ml) of the supernatants of centrifuged detergent-lysed cells (lo') with 10-200 p1 (determined by titration in each case) of the appropriate antiserum or IgG fraction of the antiserum. Antibodytreated lysates were then mixed with a slurry (l:l, w/v) of protein A-Sepharose CL 4B in phosphate buffer in a ratio of 5 volumes of slurry for each volume of serum or serum fraction used. This mixture was agitated for 2 h before centrifuging (9,000 X g) for 30 s to sediment the resin-bound immune complexes. The sedimented resin was washed in sequence by centrifuging and resuspending two times in Tris-Triton buffer (Tris. HCl, 50 mM, adjusted to pH 7.5; 100 mM NaCl, 1 mM EDTA, and 0.1% Triton X-100, w/v), once in the same buffer supplemented with 0.5 M NaCl and, finally, twice in Tris-Triton buffer omitting the Triton X-100. The immune complexes were eluted from the washed resin by boiling in SDS-PAGE sample buffer for 2 min.
Competitive Immunoprecipitation Assay for ZSG,,-Samples (5 pl) of anti-ISGVo serum were incubated with portions (175 pl) of a detergent lysate of surface-radioiodinated bloodstream forms (10'1 ml) of 2' . brucei (MITat 1.1). Sufficient detergent lysate was added to half-saturate the antibody with 1251-labeled ISGio. This lysate-antibody mixture was further mixed with a 15-fold excess (unlabeled cells/labeled cells) of either unlabeled bloodstream forms (MITat 1.1) or unlabeled procyclic forms, originally derived from bloodstream forms of MITat 1.1, and then incubated overnight at 5 "C, prior to recovery of the immune complexes (see previous section). Radioactivity in the complexes was measured using a Packard y spectrometer.
Lectin Affinity Chromatography of ZSGs-Preparations of radioiodinated ISGTo and ISGG4 in deoxycholate extracts of crude plasma membranes were chromatographed on small columns (0.5-1 ml bed volume) of a variety of different lectins, including concanavalin A, lentil lectin, wheat germ agglutinin, and Ricinlls communis agglutininlzo, under the same conditions described for the purification of 1251-labeled ISG,, on concanavalin A-Sepharose. The competing sugars (0.2 M in each case) used for the elution of bound glycoproteins were a-methylmannoside (concanavalin A-Sepharose and lentil lectin-Sepharose), N-acetylglucosamine (wheat germ agglutinin-sepharose) and D-galactose (R. communis agglutininIzo-agarose). Deglycosylation of ZSGs-Two alternative procedures were used. In the first procedure 1251-labeled ISGs present in Triton X-100 extracts of crude plasma membranes were partially purified on concanavalin A-Sepharose prior to treatment with N-glycopeptidase F (44). Crude plasma membranes, derived from osmotically lysed surface radioiodinated trypanosomes (10' cells), were extracted into 0.5 ml of Triton X-100 (176, w/v) in Tris.HC1 buffer (20 mM, pH 7.5) supplemented with NaCl (100 mM). Following centrifugation (9,000 X g, 5 min), the clarified extract was incubated (1 h, 5 "C) with 0.1 ml of concanavalin A-Sepharose in a Minifuge tube subjected to constant agitation. At the end of the incubation the resin beads were washed extensively in Tris-Triton buffer (see above) by centrifugation (9,000 X g for 30 s) and resuspension prior to boiling the affinity resin in SDS (l%, w/v; 0.1 ml) for 2 min to elute bound glycoproteins. The SDS extract was diluted 5-fold into deglycosylation buffer (250 mM Na2HP04, 10 mM EDTA, 1%, v/v, 2-mercaptoethanol, 1%, w/v, Triton X-100, 50 pg/ ml leupeptin, adjusted to pH 8.5) and incubated at 37 "C for 1 h with N-glycopeptidase F (0.1 unit/ml). In the second procedure partially purified 1251-labeled ISGs (0.35 mg protein/ml) from the deoxycholate extract of the crude plasma membrane fraction of cells were treated with N-glycopeptidase F prior to chromatography on concanavalin A-Sepharose. In this case treatment with the enzyme was carried out in a modified deglycosylation buffer (50 mM Tris. HC1,150 mM NaC1, 10 mM EDTA, 1%, v/v, 2-mercaptoethanol, 10 mM sodium deoxycholate, 50 pg/ml leupeptin, adjusted to pH 8.5) for 24 h at 20 "C. At the end of this incubation period, EDTA was removed from the deglycosylated sample by centrifugation/gel-filtration on Sephadex G-25 contained in a disposable plastic syringe (10 ml bed volume), using the method of Penefsky (45). The EDTA-free samples were then chromatographed on concanavalin A-Sepharose as described earlier in this section.
SDS-PAGE-All samples were prepared and electrophoresed exactly as described previously (46). Gels were stained either with Coomassie Blue R-250 or with the silver staining procedure of Morrissey (47).
Densitometric Scanning of Gels-Vertical lanes were cut from Coomassie Blue-stained SDS-PAGE gels and were dried onto Whatman 3MM electrophoresis paper and then scanned using a Joyce Loebel scanning densitometer/integrator operated in the reflectance mode.
Autoradiography-Gels were dried under vacuum onto sheets of Whatman 3MM paper and then exposed to Kodak X-Omat RP x-ray film in Kodak X-Omatic regular intensifying screen cassettes at -20 "C for appropriate periods of time. Exposed films were developed in Kodak LX-24 developer and fixed in FX-40 fixer.
Determination of the Content of Radioactiuity in Dried Gels-Gels (SDS-PAGE) that had been dried onto paper were cut vertically to separate all of their constituent electrophoretic lanes, each of which was then cut cross-wise into 1-2-mm wide strips. The radioactivity present in each strip was measured in a y spectrometer.
Peptide Mapping-The N-chlorosuccinimide method of Lischwe and Ochs (48) was used with minor modifications. Briefly, bands of interest were located on dried SDS-PAGE gels by autoradiography and then excised with a scalpel blade, followed by rehydration with several changes of distilled water. The protein present in the rehydrated gel slices was digested with N-chlorosuccinimide (15 mM) in acetic acid/urea/water (l:l:l, v/w/v) for 30 min at 20 "C and the treated gel slices then washed and equilibrated with SDS-PAGE sample buffer exactly as described by Lischwe and Ochs (48). Finally, the gel slices were placed in the sample wells of a 15% (w/v) polyacrylamide SDS-PAGE gel and electrophoresed. The completed peptide maps were visualized by autoradiography.
Western Blotting-Transfer of proteins from SDS-PAGE gels to nitrocellulose membranes and subsequent incubation with antibody were performed as described previously (49) with the exception that the primary antibodies were detected using a 1/1,000 dilution of either anti-mouse or anti-rabbit IgG antisera (horseradish peroxidaseconjugated) rather than lz51-labeled protein A. The antigen-antibody complexes were detected on washed membranes following their incubation in a solution of 4-chloro-l-naphthol(O.O5%, w/v), hydrogen peroxide (0.015%, w/v) and methanol (0.17%, v/v) in phosphatebuffered saline as described by Towbin et al. (50).
Raising Antisera-Rabbit anti-VSG (MITat 1.1) and rabbit antibloodstream form trypanosomal plasma membranes (MITat 1.1) were prepared using the general procedures for the preparation of polyclonal antisera described by Owen (51). Rabbits were primed with either 150 pg of purified MITat 1.1 VSG, or 3 mg of purified MITat 1.1 plasma membranes injected intradermally in Freund's complete adjuvant, followed by two booster injections administered subcutaneously in Freund's incomplete adjuvant on day 14 and day 26. Antisera were harvested 42 days after the initial injection. Rabbit anti-ISG70 antiserum was prepared by initially inoculating microgram quantities of purified ISGTo in phosphate buffer directly into the popliteal lymph nodes as described by Siege1 et al. (52). Otherwise conventional procedures were followed for the production of rabbit anti-ISG70 antiserum.
Hyperimmune antiserum was collected from a rabbit that had a chronic relapsing parasitemia approximately 3 months after an initial inoculation of the rabbit with lo4 live bloodstream forms of 2' . brucei (MITat 1.2).
Mouse anti-ISGi, antiserum was prepared by repeated (~3 ) intraperitoneal immunization of male outbred mice with homogenized SDS-PAGE gel slices containing microgram quantities of purified ISGVO in phosphate-buffered saline.
Immunoglobulin Fractionation-The IgG fraction of whole polyclonal antiserum was prepared using standard methods (53).
Immunofluorescence Microscopy-Bloodstream forms of T. brucei (MITat 1.1) were fixed in 2% paraformaldehyde in PBS at room temperature for 20 min. The fixed cells were washed in PBS containing 5% (w/v) BSA (PBS/BSA). Excess paraformaldehyde was neutralized by washing the cells with PBS containing glycine (1 mg/ml). The cells were resuspended in PBS/BSA and applied to glass slides and air-dried. Nonspecific protein binding sites were blocked by treating the slides with PBS/BSA for 20 min. After washing the slides with PBS, the cells were treated for 14 h at 4 "C with the appropriate antiserum (anti-ISG70, preimmune) diluted 1:50 in PBS/ BSA. The slides were washed with Tris buffer (50 mM Tris, pH 8.0, 150 mM NaCI) prior to treatment with alkaline phosphatase-conjugated anti-rabbit immunoglobulin for 1.5 h a t room temperature diluted 1:500 in Tris buffer containing BSA (5%, w/v). The slides were then prepared for staining with the alkaline phosphatase substrate Vector Red according to the manufacturer's instructions. Prior to mounting in glycerol, the slides were washed with ethanol. Cells were viewed with a Nikon optiphot fluorescence microscope equipped with a x 100 oil immersion Planapo objective.
Determination of the Isoelectric Points of ISG,, and ISGu-Plasma membranes were prepared from T. brucei (MITat 1.1) exactly as described previously (8) and subjected to two-dimensional electrophoresis (isoelectric focussing/SDS-PAGE) as described by OFarrell (54). After electrophoresis in the second dimension the ISG~O and ISGu were detected by Western blotting. respectively, that were not detectable by Coomassie Blue or silver staining were radiolabeled by enzymatic radioiodination while only the VSG was radiolabeled by radioacetimidation (Fig. 1A). It is important to note that the M, = 67,000 marker was bovine serum albumin, which is known to migrate in SDS-PAGE at an anomalously high molecular weight value and did not lie on the calibration curve constructed from the remaining markers in this study. Consequently, the

M, values
of the ISGs and the VSG were calculated without reference to this particular marker. Longer exposure of the autoradiograms revealed the presence of a small number of additional radioiodinated proteins in the range of M, = 45,000-150,000.
However, these minor components have not been characterized further in this present study. It was interesting to note that the intensity of labeling by radioiodination of ISG,, in particular was almost as intense as that observed for the VSG despite the large excess of VSG as assessed by Coomassie Blue staining. It seems likely that this observation reflects some unusual aspect of the structure of ISG70 such as a relatively high tyrosine content of that portion of the polypeptide exposed to the outer surface of the cell. Neither ISG70 nor ISGs, Is Structurally Related to the VSG-Following SDS-PAGE of whole cells (MITat l.l), the bands of radioiodinated ISGYo, ISGs4, and VSG were cut from the gel, and each protein was peptide mapped using N-chlorosuccinimide under conditions that have been shown previously to promote specific cleavage of tryptophanyl peptide bonds (48). Comparison of the resulting peptide maps (Fig.  1B) reveals that each map was essentially unique; no obvious homology between any of the three glycoproteins was observed. However, homology within regions devoid of sites cleavable by N-chlorosuccinimide (tryptophan) could not be entirely eliminated. Nevertheless, this last possibility was considered unlikely, because similar results were also obtained when peptide mapping was carried out by limited proteolysis with either bovine chymotrypsin or Staphylococcus aureus V8 proteinase (data not shown).
In a second independent approach antiserum raised against purified VSG (MITat 1.1) precipitated only the VSG and did not precipitate either ISG7o or ISGs4 from detergent extracts of whole cells. In addition immunoprecipitation of ISGyo, radiolabeled with ["S]methionine in vivo, from a detergent lysate prepared from bloodstream forms of T. brucei demon-  ( M , 120,000) in track 1 is a cross-linked dimer of the VSG produced by a side reaction during formation of the labeled imidoesters (68)(69)(70). B, slices (1 mm) of SDS-polyacrylamide gels that contained '251-labeled ISG70, '2sII-labeled ISGu, or '251-labeled VSG, (MI-Tat 1.1) were prepared as described under "Experimental Procedures" and then treated with N-chlorosuccinimide (15 mM) for 30 min at 20 "C prior to repeat SDS-PAGE followed by autoradiography. The strated unequivocally that ISG70 was a genuine trypanosoma1 protein (Fig. 2) rather than a host protein adsorbed to the surface coat of the cell. Moreover, antisera from a rabbit with chronic active trypanosomiasis that had sustained repeated cycles of antigenic variation initiated with live cells of a different variant (MITat 1.2) also precipitated both ISGs but failed to recognize the VSG. These results demonstrate the structural differences between the ISGs and the VSG from both homologous and heterologous variants of T. brucei and indicate the capacity of ISGs to elicit an immune response independent of the VSG during the course of a natural infection.
Purification of ISG70-Essentially all of the ISGm present in whole cells was recovered in the initial crude membrane fraction and after extraction of the membranes with deoxycholate (Table I). The majority of the VSG was released into the supernatant following rupture of the cells as reported previously (10). However, significant amounts of VSG contaminated the membrane fraction (see Fig. 3A) despite extensive washing with high salt buffers and probably represents material trapped within membrane vesicles that were created during the osmotic lysis procedure. Removal of this contaminating VSG was usually achieved during the final preparative SDS-PAGE step (Fig. 3B), but alternative methods such as chromatography on anti-VSG-Sepharose were also found to be effective.
Gel-filtration of the deoxycholate extract of membranes of T. brucei 8089 (not shown) separated ISGyo from the bulk of the membrane proteins which chromatographed as a large skewed peak close to the excluded volume of the Ultrogel AcA 34 column (exclusion limit > M , = IO6). Interestingly, ISG70, which migrates with an apparent M, = 70,000 on SDS-PAGE gels (Fig. 3B), eluted in the same volume as IgG ( M , = 150,OO) during gelfiltration (not shown) and was clearly separated from the VSG which migrates as a monomer (M, = 60,000) in the presence of deoxycholate (55). One possible explanation of this result is that native ISGioforms dimers in solution. Other possibilities such as the formation of disulfide-linked aggregates containing ISG70 are less likely, since we have detected no difference in the mobility of this glycoprotein under either fully reduced or nonreduced conditions when assessed by SDS-PAGE (not shown). SDS-PAGE analysis of the ConA-purified material (Fig. 3, track e) shows that a faint Coomassie Blue-stained band in the same position as '*'I-labeled ISG7o was clearly visible.
was also detected in this same fraction by autoradiography (Fig. 3, track f ) and appears to copurify with ISGm throughout the purification procedure up to the final step of preparative SDS-PAGE. However, was not easily visible by Coomassie Blue staining due to the presence of relatively large quantities of additional proteins, including the VSG, which migrates in the same position as during SDS-PAGE (Fig. 3, track e).
Following preparative SDS-PAGE, a sample of the material recovered from the gel corresponding to was re-electrophoresed on an analytical SDS-PAGE gel (Fig. 3B). The purified migrated as a diffuse band as visualized by Coomassie Blue staining. Analysis of the radioactivity profile confirmed that most or all of this stainable material was labeled with Iz5I and further revealed that the stained band consisted of two or more very closely migrating species. Similarly, ISG7o metabolically labeled with ["S]methionine also migrated as a doublet (Fig. 2). The possibility that these preparations of ISG7o contain low levels of other polypeptide species must also be considered.
The final yield of purified ISG7o was low (6% , Table I) but sufficient for the production in mice of small quantities of specific antisera. The major losses of ISG7o occurred mainly during the two column chromatography steps and may have been due to nonspecific adsorption of ISG70 onto the relatively large surface area of the matrix. A similar type of loss was reported by Fox et al. (56) during the purification of the VSGspecific glycosylphosphatidylinositol phospholipase C from T.

brucei.
ISG7o Cell Copy Number-Each cell was calculated to contain 5.1 X IO4 copies of ISG7o based upon the data in Table I. Consequently, ISG7o accounts for 0.12% of the total cell protein, which is 190-fold less than the VSG.
Isoelectric Points of ISG7" and ISGG4-The isoelectric points of ISG70 and KGfi4 were determined by two-dimensional electrophoresis (isoelectric focussing/SDS-PAGE) followed by Western blotting and were found to be 6.8 and 7.2, respectively.
Both ISG70 and ISG,, Are Located at the Outer Surface of the Plasma Membrane-Initial experiments demonstrated that radiolabeled ISGm and ISGfi4 were associated exclusively with the particulate fraction of the cell, whereas the VSG was released into the soluble fraction, following osmotic rupture of the cells (Fig. 4, track e-f) due to activation of the VSG releasing enzyme under these conditions (39). Subsequent

TABLE I Recouery ojISG70 at each stage ofpurification
The amount of ISG7o present a t each stage during its purification from whole bloodstream forms was determined by subjecting samples to SDS-PAGE followed by y spectrometry of gel slices. The protein content of each sample was measured by the method of Markwell et al. (42). Each value given is the mean f S.E. of the number of separate determinations shown in parentheses. "This value was determined from the ISG, band cut from the subsequent preparative SDS-PAGE gel that had been loaded with the combined 1SG; o fractions from the ConA step rather than from a separate analytical SDS-PAGE gel as used for assaying each of the previous steps.
The assumption was made that the recovery of IS& would be the same on analytical and preparative SDS-PAGE gels.
* In this case the protein concentration was estimated by densitometric scanning of Coomassie Blue-stained SDS-PAGE gels. experiments revealed that both ISG7" and ISGM copurified with plasma membranes during further fractionation of the crude particulate fraction of cells (Fig. 4, truck g; the low intensity of the ISGs4 band was caused by relatively inefficient radioiodination during surface labeling in this particular experiment). This low intensity did not simply reflect how well the membranes were washed, because it has been observed many times that preparations of cells which showed high intensity labeling of ISGs4 in pellets of water-lysed cells also showed high intensity labeling of this protein in purified membranes after various treatments and multiple washes. Conversely, when the initial labeling of ISGG4 was low, the same low content of labeled ISGs4 was likewise found in the purified membranes. Furthermore, subjecting surface-radiolabeled cells to a second step of cell purification on a DEAEcellulose column (see "Experimental Procedures") failed to alter the relative proportions of ISGs4, ISG70, and the VSG. Finally, repeated washing of membranes with 0.5 M KC1 did not remove detectable amounts of ISGs4 or ISGTo, whereas detergents readily solubilized both proteins, making it unlikely that either of these proteins were loosely attached peripheral trypanosoma1 proteins or adsorbed host proteins.

Invariant Surface
Mild trypsin treatment of whole intact radioiodinated cells removed essentially all of the radioactivity associated with ISG7,,, ISGs4, and the VSG (Fig. 4, truck c). Even following trypsin treatment under the conditions used in these experiments, the cells remained fully motile and phase dense as assessed microscopically, leading to the inevitable conclusion that the cells also remained intact. Consequently, this result also supported a second conclusion, based on the results of the surface labeling and cell fractionation experiments, that both ISGs in common with the VSG were located at the external surface of the cell. SDS-PAGE analysis of the material released by trypsin (Fig. 4, truck d ) revealed the presence of only a major doublet of radioactive Coomassie-stained bands of M, = 45,000-50,000 which were identified subsequently as fragments of the VSG by immunoprecipitation with VSG specific antiserum (data not shown). The failure to detect radioiodinated tryptic fragments of either IsGPO and ISG, suggests that these fragments may have been too small to be retained within the polyacrylamide gel during fixation and staining.

Both ISG70 and ISGfi4 are N-Glycosyluted Glycoproteins-
The lectin binding affinities of radioiodinated ISGs present in deoxycholate extracts of purified plasma membranes (VSG absent in these extracts) were studied using concanavalin A-Sepharose (ConA), wheat germ agglutinin-Sepharose (WGA), R. communis aggl~tinin~~~-agarose (RCA,,,), and lentil lectin-Sepharose (LL). Comparison of the elution profiles obtained in each case demonstrated that each of the lectin columns with the exception of lentil lectin bound significant amounts of ISGs in the order of decreasing affinity: ConA > RCA,,, = WGA >> LL. A typical elution profile for one of these columns, the ConA affinity column, is shown in Fig. 5. Interestingly, large quantities of ISGs failed to bind any of the lectins as evidenced by the high levels of radioactive ISG eluting with the flow-through fraction in each case, despite the fact that the ISG concentration was well below the binding capacity of the immobilized lectins. In addition, elution of bound ISGs from columns of ConA and elution of crude Ultrogel-filtered ISG7, from ConA resulted in a broad elution profile which appeared to consist of more than one distinct peak. These results suggested the presence of carbohydrate microheterogeneity in one or both ISGs. Such carbohydrate microheterogeneity would be expected to quantitatively affect lectin binding but not to influence the ability of endoglycopeptidase F to remove all N-linked glycosyl chains (Fig. 6). Quantitative analysis of the lectin chromatography data re- FIG. 5. Analysis of ISGVO and ISGsl by lectin affinity chromatography. Deoxycholate extracts of crude plasma membranes prepared from surface-radioiodinated cells were chromatographed on columns of immobilized lectins (see "Experimental Procedures"). A shows the elution profile of ISG,, obtained from the ConA column as an example. Column fractions were assayed for protein bound '"I. The arrow indicates the point a t which elution with the appropriate competing sugar (0.2 M) was initiated (see "Experimental Procedures''). B shows the recoveries for both 1251-labeled ISG70 and '"1labeled ISGa during the chromatographic separation on the ConA column depicted in A. The amounts of each glycoprotein present in the fractions was determined by subjecting samples of each fraction to SDS-PAGE followed by measuring the radioactivity present in slices of the gel with a y spectrometer. In each case the blackened areas on the bar diagrams represents the specifically eluted ISG, the hatched areas represent the material in the flow-through fractions obtained before initiating the specific elution, and the clear areas represent the material loaded that was not recovered. Treatment of ISG,, and ISG, with N-glycopeptidase F destroyed the ability of both glycoproteins to bind ConA and reduced the apparent molecular mass of ISG70 by 8,000 daltons to M, 62,000 as assessed by SDS-PAGE (Fig. 6). Either there was no change in the apparent molecular weight of ISGM or it comigrated with ISG,, following treatment with N-glycopeptidase F. These latter results indicate that both ISGs contain N-linked sugar chains and suggest that ISG~O is more heavily glycosylated than ISG,.
ISG,, Is a Common Invariant Antigen of Bloodstream Forms of T. brucei-The possibility that ISG,, was present in a number of different variant populations of T. brucei other than MITat 1.1 was tested by immunoprecipitation from detergent lysates prepared from seven different cloned variants, MITat 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, and ILTat 1.21. The antiserum used in these experiments was raised in mice against highly purified (880-fold) ISG7,. The antiserum proved to be specific for ISG7, and precipitated radiolabeled ISG,, from extracts of five of the six variants of the MITat serodeme tested in this particular experiment and from the single variant of the ILTat serodeme available in our laboratory (Fig. 7). In the one case where little or no was precipitated (MITat 1.7), the glycoprotein was detected readily by Western blotting at levels that were similar to those found in both MITat 1.1 and ILTat 1.21 where immunoprecipitation was not a problem. These results suggested that the efficiency of radioiodination of ISG,, in intact cells rather than the level of its expression may differ between different variants. These data also do not eliminate the somewhat unlikely possibility that ISG,, is expressed but not incorpo- and analysis by SDS-PAGE and autoradiography (see "Experimental Procedures"). A shows the autoradiograms of the SDS-PAGE gels of the detergent lysates of whole surface-radioiodinated cells (approximately 1 X lo7) of each of the seven variants. B shows the autoradiograms of the SDS-PAGE gels of the immunoprecipitates obtained by treating the detergent lysates of each variant with either control nonimmune serum (even-numbered tracks) or with anti-ISG70 antiserum (odd-numbered tracks) as indicated on the figure. The molecular weight markers used were the same as those described in the legend to Fig. 1 (38), were examined for the presence of ISG70, using a sensitive competitive immuno-precipitation assay. The results of a typical experiment are shown in Table 11. ISG, was detected in bloodstream forms but not in procyclic forms of T. brucei.
Immunofluorescence Microscopy-Immunofluorescent antibody staining of paraformaldehyde fixed trypanosomes revealed that ISG, was uniformly distributed over the entire surface of the plasma membrane covering both the cell body and the flagellum (Fig. 8). It was not possible to detect ISG70 when live trypanosomes were first incubated in a solution of anti-ISGYo antibody prior to fixing the cells.

DISCUSSION
Two novel and previously undescribed invariant surface glycoproteins of M , 70,000 (ISG70) and 64,000 (ISG,) that were detected during the surface labeling of bloodstream forms of T. brucei with lzSI have been characterized.
The discovery of ISG70 and ISGs4 by cell surface radioiodination was somewhat surprising, since previous studies by other workers who used a variety of different surface labeling reagents, including [f~rmyl-~~SS]methionine sulfone methylphosphate (57), galactose oxidase/NaB3H, (58), [isethionyl-3H]acetimidate (39), as well as the lactoper~xidase/Na'~~I technique (59), detected the presence of only the VSG. However, it appears that the two ISGs are particularly susceptible to iodination. The basis for the apparently selective and extensive reactivity of ISG70 and ISG, with under the conditions of the present study was not established with complete certainty but is thought to reflect some unusual aspect of the structure of these glycoproteins, such as a high tyrosine content. This conclusion is supported by the finding that the majority (81%) of the recovered radioactivity in "' 1labeled ISG70 was found in iodotyrosine residues following enzymatic digestion of the protein and separation of the released amino acids by thin layer chromatography. No other discrete spots of radioactivity on the chromatogram were found. This result eliminates the possibility that any histidine residues were labeled as has been observed frequently when using more vigorous iodination protocols under oxidizing conditions and also eliminates the further possibility that some non-amino acid component of the mature processed protein, such as a carbohydrate or lipid residue that could be iodinated was labeled. Furthermore, the extensive reactivity of the ISGs was not dependent upon the use of the lactoperoxidase/ glucose oxidase method for surface radiolabeling, since equally high levels of iodination were observed when radiolabeling was carried out using the chemical catalyst IODO-GEN.' The failure to detect ISGs in bloodstream forms of T. brucei in some of these other studies, even when the same conditions for enzymatic radioiodination as one of those described here were used (59), was almost certainly due to the VSG in a number of different variants masking the presence of ISGm and ISGe4. This masking effect occurs because the relatively large amounts of the VSG from some but not all variants migrates as a dense band within the same region of SDS-PAGE gels as that occupied by the ISGs. The less likely possibility that ISGTo is expressed but not incorporated into the plasma membrane of some variants must also be considered.
A considerable body of evidence emerges from this study to support the contention that these two ISGs are integral *D. G. Jackson, H. J. Windle, and H. P. Voorheis, unpublished observation.
plasma membrane proteins of trypanosomal origin and not adsorbed host proteins. First, both proteins remain associated with the cells following passage through DEAE-cellulose (pH 8.0, Krebs-Ringer phosphate buffer) during both the initial separation from host elements as well as subsequent to a second identical chromatography step following a buffer wash. Second, both proteins are associated with the particulate fraction following osmotic lysis of the cell and none can be detected in the supernatant fraction. Third, both proteins copurify with plasma membranes from the initial crude pellet following cell lysis and remain with the purified plasma membranes following several high salt (0.5 M KCl) washes. In fact the mildest procedure that removes both proteins from purified membranes is extraction with neutral detergent (1% Nonidet P-40, not shown). Fourth, both proteins are highly immunogenic in rabbits as evidenced by the presence of circulating antibodies in animals infected with live trypanosomes and the absence of such antibodies in noninfected animals. Fifth, both proteins can be immunoprecipitated from detergent extracts of surface '*'I-labeled parasites using either hyperimmune sera from parasitemic rabbits or sera raised against highly purified trypanosomal plasma membranes but not when using control sera from uninfected and unimmunized animals. Sixth and last, ISG7", which has been most studied and characterized, can be immunoprecipitated from detergent extracts of ["S]methionine biosynthetically labeled parasites. This last experiment has not yet been carried out in the case of which a t present is less well characterized than ISGio. Certainly all of the existing evidence indicates that both of these ISGs are integral plasma membrane proteins of trypanosomal origin.
Characterization of the ISGs demonstrated a number or important similarities and differences between ISGm and ISGR4. First, both of the ISGs were relatively minor cellular components; for example each individual cell was estimated to contain between lo4 and 10" copies of ISGm, whereas there are over 10' copies of the VSG (49). Second, both ISGs displayed properties that are typical of integral membrane proteins. Third, peptide mapping and differential immunoprecipitation experiments revealed that both ISGs had unique peptide maps and contained unique immunological determinants; neither ISG showed any similarity of primary structure with each other or with the VSG. Both ISGio and 1 %~~ were found to contain N-linked carbohydrate chains that were sensitive to cleavage by Nglycopeptidase F. Furthermore, these N-linked carbohydrate chains were responsible for the strong affinity for ConA that was characteristic of both glycoproteins. 1%" appears to be much more heavily glycosylated than based on the finding that its apparent molecular weight shifted by approximately 8,000 following deglycosylation, whereas little or no shift was apparent for Consequently, it may be estimated that ISG7" contains approximately 11% N-linked carbohydrate by weight, which is comparable with that found for the VSG (60). However, it is interesting to note that the Nlinked carbohydrate chains of ISG7o may not be freely accessible from the exterior of intact cells where they are probably covered by the VSG. Certainly, the carbohydrate chains of the ISGs are not cleaved when whole cells are treated with N-glycopeptidase F (not shown), but they are cleaved when the purified glycoproteins are treated. In addition the carbohydrate composition of the two ISGs was different. This conclusion was shown by the finding that only ISGm bound in any amount to either wheat germ agglutinin or RCAlm. The fact that not all of the ISG7o bound to these two lectins further reveals that ISG7o is differentially glycosylated and demonstrates considerable carbohydrate microheterogeneity.
The marked affinity of ISG7o for ConA provided a convenient means for the purification of this glycoprotein on a small scale. However, it was necessary to minimize the number of steps used in the purification procedure because of the losses of ISGs encountered, particularly during column chromatography. Extensive losses of other membrane proteins from T. brucei during their purification have been reported (56) and probably results from the marked hydrophobic character of these proteins. The purified ISG,, migrated as a broad diffuse band on analytical SDS-PAGE gels due probably to the carbohydrate microheterogeneity detected by the lectin binding experiments. However, the possible presence of other unrelated glycoproteins cannot be eliminated completely a t this stage.
ISG,, appears to be an invariant surface glycoprotein, because it was detected unchanged in eight different variants from the MITat serodeme as well as in the only variant from the ILTat serodeme of T. brucei that was tested. In contrast to the results from all of the bloodstream forms examined, both ISG7o and ISGs4 could not be detected in procyclic forms of the parasite, leading to the conclusion that these ISGs are stage-specific proteins restricted to bloodstream forms. Although we have not yet investigated the presence of ISG,, in a wide variety of variants, it seems likely that it also may be expressed ubiquitously in bloodstream forms, since antibodies recognizing in MITat 1.1 are produced when the infection in rabbits is initiated with MITat 1.2.
The finding that anti-ISGTo and anti-ISGs4 antibodies circulate in animals infected with small numbers of trypanosomes suggests that the ISGs must become exposed to the host immune system at some stage during the infection even if the exposure only occurs as a result of the death and breakdown of trypanosomes that precedes the appearance of each new variant during the undulating course of the parasitemia.
The physiological role(s) of ISGm and ISGs4 are as yet unknown. They may of course mediate some function that is restricted to the bloodstream stage of the life cycle, since they are absent from procyclic forms. However, it is clear that ISG7, is not the catalytic unit of the VSG-specific glycophosphatidylinositol phospholipase C that releases the VSG from the plasma membrane of bloodstream forms, since that enzyme does not copurify with ISGTo.' This conclusion is also supported by the recent reports (56, 61, 62) that the VSGreleasing enzyme is not glycosylated and has an apparent molecular weight of about 39,000. Nevertheless, the possibility cannot be ruled out that either ISG,, or ISG,, is involved in regulating the release of the VSG. The relatively low copy number of ISG7o compared with the VSG suggests that this glycoprotein may either fulfill an enzymatic role or, alternatively, act as a surface receptor. Preliminary experiments3 have shown that a phosphatase activity, that can be detected in purified plasma membranes, separates from the ISGs during the lectin affinity chromatography step. Several particulate phosphatase activities have been described in both the genus, Trypanosoma (6, 9, 63, 64), and in other members of the order, Kinetoplastida (55, 65-67), but none of these phosphatase enzymes have been either fully purified or characterized, and based upon our preliminary experiments they appear to be separate antigens.
There are a number of similarities between the ISGs described in this paper and those reported recently by Ziegelbauer and Overath ( membrane-bound surface glycoproteins). However there are a number of important differences also, particularly between ISG70 described in this paper and ISG75 described previously (34). For example, the apparent molecular weights differ by 5,000. The degree of glycosylation is significantly different also (ISG,, contains approximately 11% N-linked carbohydrate by weight, whereas ISG75 (34) contains only 2.5% carbohydrate. Furthermore the isoelectric point of ISG,, is 6.8, whereas ISG,, has a calculated PI of 5.3 (34). The major difference between ISG64 and is the degree of glycosylation of both proteins. We estimate that ISG,, contains a maximum of 3% carbohydrate, whereas ISG,, contains 12% carbohydrate (34). The issue of the relationship between these proteins will be resolved once sequence information is available. These studies are currently in progress.