N-Deglycosylation of human complement component C9 reduces its hemolytic activity

Molerulur Imn7utdo~~. Vol 26. No. 12. pp. 1125-1132, Printed rn Great Bntain.

0161-5X90:89 $3.00 + 0.00 I$ 1989 Pergamon Presr plc

1989

N-DEGLYCOSYLATION OF HUMAN COMPLEMENT COMPONENT C9 REDUCES ITS HEMOLYTIC ACTIVITY ROLAND KONTERMANN* and ERNST W. RAuTERaERGt Institute

of Immunology,

University

of Heidelberg,

INF 305, D-6900

Heidelberg,

F.R.G.

(First receiced 2 February 1989; accepted in revised form 15 May 1989) Abstract---The effect of enzymatic deglycosylation of human complement component C9 on its hemolytic activity was investigated. Treatment of native C9 (M, 71,000) with glyocpeptidase F (PNGase F) results in a stepwise decrease of the mol. wt. The formation of an M, 67,000 peptide which is further converted to M, 63,000 suggests that there are two N-linked carbohydrate chains per C9 polypeptide. Removal of approximately 88% of the N-linked oligosaccharides results in 80% reduction of the hemolytic activity (CH50). The completely N-deglycosylated M, 63.000 peptide contains a remaining amount of 25% of the total carbohydrates of native C9. These glycans are assumed to be Q-linked and predominantly attached to the C9a part of C9. The electrophoretic mobility of C9 is not affected by endoglycosidase F or A treatments revealing that the two N-linked glycans are of the tri- or tetra-antennary complex type. Cleavage of terminal sialic acids from native C9 by neuraminidase results in an M, 67.000 product with nearly unaltered hemolytic activity. In contrast to other glycoproteins in which deglycosylation remained without major effects on their functional activity, our findings suggest that the N-linked carbohydrates are required for full expression of hemolytic activity of C9.

INTRODUCTION Assembly of the complement components C5b, C6, C7, C8 and C9 occurs either on membranes leading to the formation of the membrane attack complex (MAC) or in the fluid phase resulting in the non-lytic analog SC5b9. The human complement component C9 has a key role for the function of the terminal complement sequence. While the “acceptor” for C9 on target membranes. the C5b-8 complex itself, exhibits even under optimal conditions only a low lytic efficacy, the extent of lysis increases dramatically after its binding of C9. This final step of the terminal sequence is required for the full expression of the bactericidal and lytic potential of the complement system (for review see Podack, 1986; Hlnsch, 1988). One to eighteen C9 molecules may interact with one CSb-8 complex to form the MAC. Thus. C5b-8 of C9. serves as a “primer” for self-polymerization

*Present address: Institute of Molecular Genetics, University of Heidelberg, INF 230, D-6900 Heidelberg, F.R.G. tAuthor to whom correspondence should be addressed at: German Diagnostic Clinic. Aukammallee 33. D-6200 Wiesbaden, F.R.G. Abbreviations: Complement components and their cleavage products are expressed following the WHO recommendations; C9, native human C9; C9,. N-deglycosylated C9 preparation; CHSO. quantity of complement component required to lyze 50% of target cells; EDTA. ethylene diamine tetra-acetic acid: M,, apparent molecular mass; PAGE, polyacrylamide gel -electrophoresis: PAS. periodic acid-Schiff reaction: PNGase F, glycopeptidase F; SDS. sodium dodecylsulfate.

The characteristic ring-like lesions on complementlyzed cells were thought to represent polymerized C9 (Humphrey and Dourmashkin, 1963; Rauterberg and Gebest, 1977; Rauterberg et al., 1979), and poly(C9) was assumed to be the molecular basis of the transmembrane channel responsible for lysis (Tschopp, 1984). However, recent findings have cast some doubts on the polymerization of C9 as a prerequisite for lysis (Martin et al., 1987). The molecular architecture and properties of the single chain globular serum protein C9 are of special interest since (i) it undergoes conformational changes upon interaction with C5b-8 complexes (Silversmith and Nelsestuen, 1986); (ii) it exposes hydrophobic membrane insertion sites in the MAC (Amiguet et al., 1985); (iii) it penetrates the target membrane (Rauterberg and Hollerbach, 1979; Whitlow et al., 1985) i.e. it converts from a serum glycoprotein before activation to an integral membrane protein after its binding to C5b-8; (iv) it is involved in the species restriction of the terminal complement complex (Schonermark et al., 1986, 1988: Zalman et al., 1986); and (v) it is suggested to belong to a protein family comprising at least C7. C8c(. CSP, C9 and perforin (Stanley and Luzio. 1988; Lichtenheld e’t al., 1988), all of which are proteins involved in lytic mechanisms. A variety of techniques has been applied to map the topology and functional domains of the C9 molecule. These include studies on the effects of monoclonal antibodies against C9 on its hemolytic activity (Stanley et al., 1985; Bausback et al., 1988. 1125

ROLANDK~NTEKMANN and ERNSTW. RAUTEKH~KC;

1126

1990; Yoden et cd., 1988) and functional analyses of proteolytic C9 fragments (Biesecker et al.. 1982; Shiver er ul., 1986). However. no generally accepted view as to the nature of the molecular interaction of C9 with either C5b-8 complexes or the target membrane has yet emerged. Like all other complement components, C9 is a glycoprotein. The carbohydrate composition of C9 was estimated by gas-liquid chromatography and found to be in the order of 8% of the total mass (DiScipio and Hugh, 1985). The attachment sites of ‘V-linked oligosaccharides to the protein have been deduced indirectly from the amino acid sequence (Stanley rt al., 1985). The relative contribution of the carbohydrate moiety to the functional activity of C9 is completely unknown. Enzymatic deglycosylation without any dcnaturation or degradation of the protein backbone has proven to be an excellent tool for functional analyses. In the present report WC applied this technique to study the effects of deglycosylation of native C9 on its hemolytic activity using various glycosidases (including glycopeptidase F and neuraminidase). MATERIALS AND Complement

METHODS

components

Native C9 (C9,) was purified from human serum according to Dankert et al. (1985) with minor modifications (Bausback, 1988). Protein concns were determined by the method of Bradford (1976) using BSA as the reference. Serum of a C8-deficient patient was kindly provided by S. Loeke, University of Mannheim (F.R.G.). A preparation containing partially purified C7 and C8 was a generous gift from M. Kirschfink (Institute of Immunology, Heidelberg).

Hemolytic activity was measured using chicken erythrocytes (ChE) as target cells. ChEACl-8 cells were prepared as follows: ChE were washed sequentially in VBS. pH 7.2, containing 0.1% gelatin, 2.75 mM Mg”, 0.75 mM Ca2’ (VBSG), and 10 mM EDTA (VBSG-EDTA), resuspended at a concn of 10’ cells/ml and incubated for 15 min at 30, C with an equal volume of amboceptor (1 : 300 diluted in VBSG-EDTA; Behringwerke, Marburg, F.R.G.). After washing, the cells were incubated with CSdeficient serum (1: 40 diluted in VBSG) for 30 min at 37 C, washed and finally incubated with C7/C8 for 30min at 37’C in VBSG-EDTA. One hundred microliters of ChEACl-8 cells (10’ cells/ml VBSGEDTA) were incubated with 100~1 C9 of varying concns for 30 min at 37C. After addition of 1.3 ml precooled VBSG-EDTA and centrifugation, lysis was determined by measuring the O.D. at 412nm. Competition assays were performed by incubating 100 ~1 ChEACl-8 at 37-C with a mixture of 50 ~1 C9, and 50~1 N-deglycosylated C9(C9,) at a final molarity of 0.5 nM each.

Deglycos~dution

oj’ C9

were removed by incuN-linked ohgosaccharides bating C9,, (150 pg,‘rnl TBS. pH 7.2) with 20 mI_! ml F from Flur:ohr~r.trrill,~~ (final concn) glycopcptidase meningosepticum (Boehringcr. Mannheim. F.R.G.) at 37’C. The PNGase F preparation was free of detectable levels of other glycosidases or of any proteolytic activity. Terminal sialic acids were cleaved overnight at 20 C in TBS. pH 7.2, by treatment with neuraminidase from ilrrhrohuctcr ureq/k%w.r (Boehringer, Mannheim, F.R.G.) at a final concn of 100 mu/ml (containing less than I .6 U protcase i-1 neuraminidasc). Incubation with endoglycosidase I’ or endoglycosidase H (both from Boehringcr. Mannheim, F.R.G.) was performed according to the manufacturer’s protocol at final concns (71‘ 150 mu/ml each. Cleuwge

of’ C9

with

r-thromhin

C9, or C9, (150 pg/ml each) were incubated with purified cc-thrombin from human plasma (Boehringer. Mannheim, F.R.G.) at a final concn al 600mU/ml for 4 hr at 37 C in 100 mM TrissHCI. pH 8.5. lOOmj\/I NaCl. Reactions were stopped by boiling in 1% SDS for 5 min. Gel

elc~tropi7orcsi.s

and

immunohlolting

Protein samples were analyzed on 10% or 1Ok17% linear gradient polyacrylamide gels in the presence of SDS (Laemmli. 1970). Proteins were visualized by staining with Coomassie blue R-250 or silver (Morrissey. 1981). Proteins were electrophoretically transferred onto nitrocellulosc (Towbin et al.. 1979). The filters were pretreated with PBS containing 0.3% Tween-20 and were incubated with purified monoclonal antibodies against C9a (M 10) or C9b (M38) (Bausback, 1988) at a concn of 50Ong~ml. Bound antibodies were detected by horseradish peroxidase conjugated goat anti-mouse immunoglobulins (Dianova. Hamburg. F.R.G.) using diaminobenzidine,!H,O, as substrate. PAS

stuining

Carbohydrates of C9, or C9, were stained following SDS-PAGE by periodic acid-Schiff reaction (PAS). After fixation in 10% trichloroacetic acid overnight, the gels were incubated with I% periodic acid for 2 hr, washed with 15% acetic acid for 2 ht and developed with Schiff’s reagent (Kapitany and Zebrowski, 1973). Stained gels were analyzed on a scanning dcnsitometer (Shimadzu. Japan) at 550 nm for PAS, 595 nm for Coomassie blue or using white light for sibcr stained bands. RESULTS Enzymatic

deglycosylution

of’ humun

CO

To investigate the effects of defined deglycosylation on the hemolytic activity of C9, we first analyzed

Deglycosylation

whether glycopeptidase F (PNGase F) would be able to deglycosylate C9 under native conditions. PNGase F is known to cleave N-linked oligosaccharides from glycoproteins but not carbohydrates O-linked to serine or threonine. The kinetic effects of N-deglycosylation are depicted in Fig. 1: SDS-PAGE analysis shows that incubation of purified human C9 with PNGase F first leads to the conversion of the apparent mol. mass of C9, from 71.000 to an intermediate species with M, 67,000. Further incubation induces the formation of a second product with M, 63,000. Enzymatic Ndeglycosylation thus results sequentially and time dependently in the appearance of two defined peptides with increased electrophoretic mobility. Such behavior suggests that the 67,000 mol. wt band represents a C9 after cleavage of one N-linked oligosaccharide while the 63,000 mol. wt band corresponds to C9 which has lost two N-linked carbohydrate chains. After 2 days of incubation approximately 80% of C9 is devoid of both N-linked glycans. The extent of N-deglycosylation could not be increased by higher concns of the enzyme. Incubation of C9 with PNGase F in the presence of SDS and non-ionic detergent and/or /I-mercaptoethanol at 37°C did not lead to additional cleavage

of C9

I127

products or a higher yield of the 63,000 mol. wt species (data not shown). Thus, under native conditions all removable N-linked oligosaccharides were cleaved from C9. C9 incubated with PNGase F for 1 day was analyzed further by additional digestion with zthrombin cleaving the C9 molecule into the C9a and C9b fragments. The cleavage products were analyzed by immunoblotting using monoclonal antibodies specific for the C9a or C9b fragment, respectively (C9a-MAb MIO; C9b-MAb M38; Bausback, 1988). Figure 2 demonstrates that deglycosylation of C9 by PNGase F exclusively alters the mobility of the C9b part of C9. While in C9, this fragment exhibits an apparent mol. wt of 38,000 it is reduced to 34,000 or 30,000, respectively, in the C9,. The mobility of the C9a fragment (mol. wt 28,000 under non-reducing conditions), however, was not affected by the enzymatic treatment. This experiment indicates that the N-linked carbohydrates cleaved from C9 by PNGase F are restricted to the C9b part. Since only intermediate products of mol. wt 67,000 (or 34,000 after a-thrombin cleavage) were observed, we propose that both N-linked carbohydrate chains possess nearly identical mol. wts (approximately 4000 as determined on SDS-PAGE).

1234567 kd

94-

67-

-71 -67 ‘ -63

Fig. 1. Incubation of human C9 (150pg/ml) with PNGase F (20mU/ml) at 37°C. Samples were taken at the following times: (1) 0 min, (2) 5 min. (3) 30 min, (4) 1hr, (5) 6 hr, (6) 1 day and (7) 2 days. C9 (0.5 lg) was analyzed on a 10% SDS-polyacrylamide gel under non-reducing conditions and proteins were visualized by silver staining.

ROLAYI) KOVTERMANNand

1128

ERNSTW. RAITERREKG

Comparison of the carbohydrate content of PNGase F treated C9 with C9, reveals a reduction of the PAS-stainable glycans of about 40% for the 67,000 mol. wt band or 75% for the 63.000 mol. wt band, respectively (Fig. 3). Incubation of C9, with neuraminidase resulted in an increase of its electrophoretic mobility to M, 67,000 when analyzed on 10% SDS-PAGE (Fig. 4A). Furthermore, the apparent mol. wt of the two C9, species shifted to 65,000 or 62,000, respectively, after subsequent neuraminidase cleavage (Fig. 4B). These findings indicate that sialic acids are present as terminal residues on N-linked as well as on O-linked carbohydrate chains attached to C9. Digestion of native or denatured C9 by either endoglycosidasc F (endo F) or endoglycosidasc H (endo H) did not affect its apparent mol. wt in SDS-PAGE analyses. Hemoiytic

uctioity

of degl~cosylated

C9

The hemolytic activity of N-deglycosylated C9 preparations was markedly reduced (Fig. 5). This effect depended on the degree of N-deglycosylation

B

1

71-N

2

#A57 '63

-34 m-30

Fig. 2. Reaction of two MAbs against C9a and C9h. respectively, with native C9 (lane 1) or PNGase F cleaved C9 (lane 2) after partial cleavage with a-thrombin. Figure 2 shows a western blot after lo-17% SDS PAGE analysis: incubation with monoclonal antibody Ml0 recogmzing C9a (A) and with Mi8 directed against C‘9h (H).

Fig. 3. PAS stained native (‘9 (A) and PNGasc f‘ cleaved CO (H). Four micrograms of each sample ~vcre applied to 10% SDS PAGE. ataincd hy either the periodic ,~cldSchiff reaction (P) or Coomas~ blue (C) and analyd h> xxnnins

dwritomctry.

(Table I) and was optimally visible at one C’H50 unit of C9,> or C9,,. Cleavage of about XX% of the N-linked oligosaccharides diminished the CH50 activitv i to 20%. This effect was not due to a direct action of the PNGase F in the C9, preparation clcac-ing carbohydrates of the target crythrocytes. since addition ol PNGasc F to the hemolytic assay (ChEACI 8 plus CS,,) did not influence the degree of hcmolysis. In addition. incubation of C9,, for 2 days at 37 C without enzyme or with heat inactivated enzyme did not alter its hcmolytic activity. Since we could not detect any proteolytic frapments in silver stained gels even when 1 p g of CO,,n as applied either under non-reducing or reducing condotions. the marked reduction in hcmolytic activit! of C9 was most certainly not caused by protcolytlc degradation of C9. In contrast to PNGasc F treatmenl. removal 01 terminal siaiic acids from native C9 by ncuraminidasc had only marginal effects on its hemolytic acti\ilq (Fig. 6). Approximately 0. IS nM dcsialatcd Cc) corresponded to 0.12 mZ;I C9,,. IX. one C‘H50 unit. By competition assays we investigated whether CO,. preparations would inhibit or modulate the binding or cfficicncy of Cc),, Mixtures of eqimolar amount ot C9, and 0, had a hemolytic actixlty comparable to C9, alone (Fig. 7). Thus, X-dcglycosylatcd CY ncithcr increased the C9,, dependent Iysis- thus confirming that the remaining hemolytic activity 01‘ <“I,, LILIS

Deglycosylation

A

1

2

1129

B123

kd

kd

94-

94-

67-

of C9

iw\., *= -71 -67

67-

43-

-65 -62

43-

30-

30-

Fig. 4. (A) Cleavage of 0.5 pg of C9 by neuraminidase (100 mu/ml) at 20-C for 1 day. Native C9 (lane 1) and neuraminidas~ treated CY (lane 2) were analyzed on 10% SDS-PAGE under non-reducing conditions. (B) C9 pretreated with PNGase F (20mU/ml) for I day at 37’C (lane 2), incubated with neuraminidase (I 00 mUimi) for another 24 hr at 20-C (lane 3). Lane 1 shows untreated C9. Samples were analyzed as described above

0.001

0.01

0.1

10

700

Fig. 5. Hemolytic assay of native C9 and N-deglycosylated C9 preparations. ChEACl-8 cells (100 ~1) were incubated with 100 ~1 of C9 of varying concns. C9 degiycosylated by PNGase F for 6 hr (0, lane Z), 1 day (8, lane 3) or 2 days (A, lane 4) shows a stepwise reduction of its hemolytic activity as compared to native C9 (0, lane 1) depending on the degree of deglycosylation.

ROLAND KONWRMANN and ERNST W. RAUTERBEKC;

1130

Table I. Hemolytic activity of natwe C9 and PNGase F treated C9 Ratio of M, cleavage products (%) 7 I ,000 67,000 63,000 Native c9 C9 deglycosylated for. 6 hr I day 2 days

Remaining N-linked carbohydrates (%)

c9 (IlM) reqwred SW 50% lys1s

100

0.12

100 IO

80 40 25

IO 60 7s

50 20 12.5

0.18 0 30 0.60

Ratms of cleavage products were estimated on the basis of the peak area of the protein band\ in silver stained SDS-polyacrylamide gels as determmed by scanning densltometry. Remainmg N-linked carbohydrates were calculated assuming two Wlinked glycans per C9 molecule.

low--nor was lysis inhibited to a greater extent, thus suggesting the C9, exhibits a reduced binding capacity to C5b--8 complexes.

DISCUSSION

Obviously, the N-linked oligosaccharides are of functional importance for the full hemolytic activity of C9. The double shift of its apparent mol. wt after treatment with PNGase F suggests that two N-linked carbohydrate chains are present in C9. The finding is in accord with amino acid sequence data deduced from cDNA clones as reported by Stanley et al. (1985). The authors described two potential attachment sites for N-linked glycans at positions 256 and 393. respectively. In contrast, on the basis of their sequence data, DiScipio and co-workers (1984) predicted only one possible attachment site at position 256. It was suggested that this descrepancy is due to a polymorphism (DiScipio and Hugli, 1985). Residual 25% carbohydrates are associated with the double N-deglycosylated C9 species (M, 63,000). We assume that the remaining oligosaccharides are

O-linked to C9. The r-thrombin cleavage cxperiments revealed the location of the N-linked glycans at the C9b part. Biesecker and co-workers (19X2) observed earlier that the C9a part contains 24% of the total carbohydrates of CY. This is about the same amount which appears to be O-linked to C9 as deduced from our data. Carbohydrates of C9a might. therefore, represent nearly all oligosaccharides Olinked to the protein backbone. A possible signihcance of these O-linked oligosaccharides for the function of C9 has yet to be elucidated. The majority of terminal sialic acid residues seem to be attached to N-linked carbohydrates. This conclusion is based on our finding that N-deglycosylatcd C9 (M, 63,000) further treated with neuraminidasc resulted in a small reduction of the apparent mol. mass by only 1000, while neuraminidase treatment 01 the native molecule reduced its apparent mol. wt by 4000. Terminal sialic acid residues on both types of oligosaccharides (i.e., 0- or ‘V-linked) seem to have no effect on the function of C9 since neuraminidasc treated C9 has nearly the same hemolytic activity as C9,.

C9 Fig. 6. Hemolytic

activity

of native C9 (0,

(nM)

lane 1) and of neuraminidase concns.

treated

C9 (A, lane 2) at varying

Deglycosylation

Since neither cndo F nor endo H released oligosaccharides from C9 to a measurable extent these experiments gave indirect evidence that the N-linked oligosaccharides of C9 are of the tri- or tetra-antennary complex type. based on the specificity of both enzymes (Tarentino et ul., 1985; Timble and Maley, 1984). Thus, such a structure of the carbohydrate chains would be in good agreement with a high number of terminal sialic acid residues present on X-linked oligosaccharides. The marked reduction in the hemolytic efficiency of N-deglycosylated C9 points to a functional importance of one or both of these N-linked oligosaccharides. Both carbohydrate chains are bound near the presumed region of membrane insertion within amino acids between positions 275 and 350 (Stanley et al., 1986). The two attachment sites for N-linked oligosaccharides are found at related positions in the mouse C9: one is also present in the C9 of trout (Stanley and Herz, 1987). This conservation indirectly supports our conclusion that at least one of the N-linked carbohydrate chains on C9b is important for C9 function. These oligosaccharides might (i) either influence the orientation and the anchorage of C9 in the membrane; (ii) interact with proteins or carbohydrates on the target membranes; or (iii) modulate the interaction of C9 with the C5b-8 complex. The latter possibility would be in accord with our assumption that C9, has a stronger affinity to C5b8 bearing target cells than C9,. Reduction of hemolytic function of C9 could also be caused by changes of its tertiary structure. A stabilizing effect of carbohydrates was described for other glycoproteins (Ryan et al., 1988). However, in those examples deglycosylation did not necessarily affect the functional activity of the proteins. The

ioo-

i-_

;

5

15

30

time

45

60

(min)

Fig. 7. Competition assay testing the effect of N-deglycosylated C9 (treatment was for 2 days, i.e. identical to lane 4, Fig. 5) on hemolytic efficiency of native C9. N-deglycosylated C9 (A, 1.0 nA4; or a, 0.5 nM) alone revealed the lowest hemolytic activity. A mixture of 0.5 mM native C9 and 0.5 mM N-deglycosylated C9 (m) resulted only in a minor reduction of hemolysis as compared to 0.5 nM native C9 (0) and did not have the efficiency of 1.0 nM C9 (0).

of C9

I131

results of our experiments with neuraminidase treated C9 suggest that the relative contribution of the negatively charged terminal sialic acids in the stabilization of tertiary structure-measured as functional activity-is nearly negligible. As revealed by sequence data and immunological studies all components of the terminal complement complex (except C5) exhibit structural and functional similarities and all of them are glycoproteins. However, C9 is the only component converting during MAC formation from a globular serum protein to an integral membrane protein penetrating the entire target membrane. Therefore, it would be of special interest to further elucidate functional differences of the carbohydrate moieties within these proteins of the terminal complement complex. Acknowledgements-The M. Reichel is gratefully for helpful discussions of the manuscript.

excellent technical assistance of acknowledged. We thank K. Rother and G. Petersen for critical reading

REFERENCES Amiguet P., Brunner J. and Tschopp J. (1985) The membrane attack complex of complement: lipid insertion of tubular and non-tubular polymerized C9. Biochemi.rtry 24, 1328-1334. Bausback J. (1988) Epitopkartierung und Charakterisierung funktionell bedeutsamer Domlnen der humanen Komplementkomponente C9 mit monoklonalen und monospezifischen polyklonalen Antikiirpen. Dissertation, Universitat Heidelberg. Bausback J., Kontermann R. and Rauterberg E. W. (1988) Functionally important domains of C9 as defined by monocional antibodies to C9 inhibiting lysis. fmmunohiology 178, 58 (abstract). Bausback J., Kontermann R. and Rauterberg E. W. (1990) Characterization of ten new monodonal antibodies against human C9. (Manuscript in preparation.) Biesecker G., Gerard C. and Hugh T. E. (1982) An amphiphilic structure of the ninth component of human complement. J. biol. Chem. 257, 25842590. Bradford M. M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Analyt. Biochmm. 72, 248254. Dankert J. R., Shiver J. W. and Esser A. F. (1985) Ninth component of complement: self-aggregation and interaction with lipids. Biochemistry 24, 27542762. DiScipio R. G., Gehring M. R., Podack E. R., Kan C. C., Hugli T. E. and Fey G. H. (1984) Nucleotide sequence of cDNA and derived amino acid sequence of human complement component C9. Proc. natn. Acad. Sci. U.S.A. 81, 7298-7302. DiScipio R. G. and Hugh T. E. (1985) The architecture of complement component C9 and poly(C9). J. hiol. Chem. 260, 14,802-14,809. Hlnsch G. M. (1988) The complement attack phase. In 7%~ Complemenf System (Edited by Rother K. and Till G. 0.). pp. 202-230. Springer Verlagt Berlin. Humphrev J. H. and Dourmashkin R. R. (1963) Discussion on’ mechanism of blood cell damage produced by immunological reactions. In Proceedings of The Third International Symposium on Immunopathologv, p. 369. Schwabe and do.; Basel. Kaoitanv R. A. and Zebrowski E. J. (1973) A high resolution PAS stain for polyacrylamide gel ‘electrophoresis. Analyf. Biochem. 56, 361-369.

1132

ROLAND

KOXTERMANN

Laemmli U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. ,Vuture 227, 680685. Lichtenheld M. G., Olsen K. J., Lu P.. Lowrey D. M.. Hameed A., Hengartner H. and Podack E. R. (1988) Structure and function of human perforin. Vurrrrc~ 335, 448450. Martin D. E.. Chiu F. .I., Gigli I. and Miiller-Eberhard H. J. (1987) Killing of human melanoma cells by the membrane attack complex of human complement as a function of its molecular composition. J. dirz. Itwe.sr. 80, 226- 233. Morrissey J. (1981) Silver stain for proteins in polyacrylamide gels: a modified procedure with enhanced uniform sensitivity. Anaiyr. Biochm. 117, 307 310. Podack E. R. (1986) Molecular mechanisms of cytolysis by complement and by cytolytic lymphocytes. J. Cell. Biochrm. 30, 133-170. Rauterberg E. W. and Gebest H. J. (1977) Demonstration of C9 at complement induced membrane “holes” by immunoferritin electron microscopy. .I. Inzmrm. 116, 1747 (abstract). Rauterberg E. W. and Hollerbach H. (1979) Studtes on the mechanism of complement-induced membrane damage: further evidence for the penetration of C9 through the entire lipid bilayer. Z. fntmunFor.~ch. esp. 71~er. 155, 52 (abstract). Rauterberg E. W.. Ungemach B. and Gebest H. J. (1979) Quantitative measurement of C9 sites and their association to ring-like “lesions” on complement-lysed membranes: a morphometric immunoferritin study. J. Immun. 122, 3555365. Ryan R. J.. Charlesworth M. C.. McCormick D. J.. Mihus R. P. and Keutmann H. T. (1988) The glycoprotein hormones: recent studies of structure ~function relationships. FASEB J. 2, 2661-2669. Schonermark S., Filsinger S.. Berger B. and Hansch G. M. (1988) The C8-binding protein of human erythrocytes interaction with the components of the complementattack phase. Immwzolog~ 63, 585 590. Schonermark S., Rauterberg E. W.. Shin M. K., Liike S.. Roelcke D. and Hansch G. M. (1986) Homologous species restriction in lysis of human crythrocytes: a membrane-derived protein with CR-binding capactty functions as an inhibitor. J. fnrnrun. 136, 1772 1776. Shiver J. W., Dankert J. R., Donovan J. J. and Esser A. F. (1986) The ninth component of human complement (C9):

and ERNST W. RAIJTERBERG functional activity of the b fragment. J. hid. C/WV. 261: 962999636. Silversmith R. E. and Nelsestuen G. L. (1986) The assembly of the membrane attack complex of complement on SITId unilamellar phospholipid vesicles. Biochemisrrr 25, 852-860. Stanley K. K. and Hcrz J. (1987) Topological mapping of complement component C9 by recombinant DNA techniques suggests a novel mechanism for its insertion into target membranes. EMBU J. 6, 1951 -1957. Stanley K. K., Kocher H.-P., Luzto J. P.. Jackson P. and Tschopp J. (1985) The sequence and topology of human complement component C9. EMBU J. 4, 375 ~382. Stanley K. and Luzio P. (1988) A family of killer proteins. !V0lure 334, 475-476. Stanley K. K., Page M.. Campbell A. K. and Luzio J. P. (1986) A mechanism for the insertion of complement component C9 into target membranes. M&c,. Immun. 23, 451 458. Tarentino A. L., G6mez C. M. and Plummer ‘T. 11 , Jr (19X5) Deglycosylation of asparagine-linked glycans by peptidc: ,V-glycostdase F. Biochemi.trrr 24, 4665 4671. Timble R. B. and Maley F. (1984) Optimizing hydrolysis of .V-linked high-mannose oligosaccharides by endo/IV-acetylglycosaminidase H. .4nu/yr. Bio&ern. 141, 515 522. Towbin H., Staehlin T. and Gordon J. (1979) Electrophoretic transfer of proteins from polyacrylamtde gels to nitrocellulose sheets: procedure and some applications. Proc. t~utn. Acd SC;. C’.S.A. 76, 4350.4354. Tschopp J. (1984) Ultrastructure of the membrane attack complex of complement: heterogeneity of the complex caused by different degrees of C9 polymerization. .I. hid Chcm. 259, 7857 7863. Whitlow M. B., Ramm L. E. and Mayer M. M. (1985) Penetration of the C5b- 9 complex across the erythrocyte membrane into the cytoplasmic space. J. hiol. Chem. 260, 998 -1005. Yoden A., Moriyama T., lnoue K. and Inat S. (19X8) The role of the C9b domain in the binding of C9 molecules to EACI--8 defined by monoclonal antibodies to C9. J. !rrlmun. 140, 2317 2321. Zalman L. S., Wood L. M. and Miiller-Eberhard H. J. (1986) Isolation of a human erythrocyte membrane protein capable of inhibiting expression of homologous complement transmembrane channels. Proc,. n~t)l. ilc,trt/. %i. C’..S.A. 83, 6975 ~6979.