- Email: [email protected]
Comparison of channels formed by poly C9, C5b-8 and the membrane attack complex of complement
Molecular Immunology,
Vol. 21, No.
6, pp. 533-537, 1990
0161-5890/90 $3.00 + 0.00 Pergamon Press plc
Printed in Great Britain.
COMPARISON OF CHANNELS FORMED C5b-8 AND THE MEMBRANE ATTACK COMPLEMENT* LEORA SHALTIEL Z.&MAN?
BY POLY COMPLEX
C9, OF
and HANS J. MUELLER-EBERHARD~
tResearch Institute of Scripps Clinic, Department of Immunology, 10666 North Torrey Pines Road, La Jolla, CA 92037, U.S.A. and $Bernhard-Nocht-Institut fiir Tropenmedizin, Bernhard-Nocht-Strasse 74, D-2000 Hamburg 4, F.R.G.
(First received 10 July 1989; accepted in revised form 28 November 1989) Abstract-The channels formed by poly C9, C5b-8 and C5b-9 were examined using the liposome swelling assay. By plotting the relative rate of swelling of CSb-g-containing liposomes vs the molecular weight of the sugar solute and by applying the Renkin equation, the size of the C5b-8 channel was estimated to be 1.5 mm radius. As increasing amounts of C9 were added during the formation of C5b-9, in C8:C9 ratios of 1: 1, 1: 2, 1 : 6 and 1: 12, the size of the function channel increased. Poly C9 had a pore that was somewhat larger than C5b-9 at a C8: C9 ratio of 1: 12. Using molecular sieving experiments with four different iodinated protein size markers, the channel diameter of poly C9 was estimated at between 90 and 100 A. Monoclonal antibodies to different complement proteins were added to the liposomes to see which might inhibit the channels. C5b-8 containing liposomes could be inhibited by antibodies to C8. Liposomes containing CSb-9 could be inhibited slightly by antibodies to C9 and most strongly by antibodies to the neoantigen of poly C9.
INTRODUCTION membrane attack complex (MAC) of complement consists of five different proteins: CSb, C6, C7, C8 and C9. Assembly of these proteins into the MAC follows cleavage of C5 by the CS convertase of the classical or alternative pathway and generation of C5b. C5b combines with C6 and C7 to form the C5b-7 complex. This complex has a metastable membrane binding site and can attach to the surface of the target membrane. Subsequently, C8 binds to the C5b-7 complex to form CSb-8. The last protein to become incorporated is C9 and the number of C9 molecules that bind varies between 1 and 16 (Miiller-Eberhard, 1988). At high input of C9, the C5b-9 complex forms characteristic rings that can be seen in the electron microscope (Podack et al., 1982). The C5b-8 complex forms a small functional channel, whose formation is time-dependent. It has been shown that C5b-8 can produce channels in a variety of target membranes, including sheep erythrocytes, (E,) (Slolfi, 1968; Tamura et al., 1972), resealed E, ghosts (Ramm et al., 1982), black lipid membranes (Michaels et al., 1976), M21 human melanoma cells (Martin et al., 1987) and gardia lamblia (Deguchi et al., 1985). The size of the channel was estimated by Ramm et al. (1982) to be between 9 and 30 A in diameter. The
*Publication No. 5951-IMM. This research project was supported by National Health Services grants CA47869 and AIl7354, and an Established Investigator Award from the American Heart Association to Dr Zalman.
The size of the MAC pore can vary from 10 to over 100 A depending on the complement dose. With very high concentrations of human complement, a maximum diameter of IlOA was attained in resealed erythrocyte membrane ghosts (Dalmasso and Benson, 1981). Mayer’s group reported that with three C9 molecules per C8, the minimal channel diameter was 3OA and with four molecules of C9 per C8, the minimal channel diameter was 4OA (Ramm et al., 1985). The C9 component is the protein largely responsible for the formation of structural membrane lesion associated with the MAC (Miiller-Eberhard, 1988). Purified C9 can polymerize to form tubular complexes that resemble the MAC when viewed on the electron microscopy (EM). The internal diameter of the tubule as measured from EM micrographs is about 100 A (Young et al., 1986). The purpose of the present work was to determine the size variation of the MAC channel which the C9: C8 molar ratio was increased from 0 to 12. Channel size was assessed using the liposome swelling assay developed by Luckey and Nikaido (1980). In addition, we present data showing that poly C9 produced from isolated C9 in absence of C5b8 is a channel forming structure with functional channel diameters between 90 and 100 A. MATERIALS
AND METHODS
Isolation of proteins C5bA (Podack et al., 1980), C7 (Podack et al., 1979), C8 (Kolb and Miiller-Eberhard, 1976) and C9 533
L.
534
%ZALMANand H.J. MILLER-EBERHAR~
(Biesecker and Miiller-Eberhard. 1980) were isolated by established procedures. Poly C9 was made by incubating purified C9 with 50 ,&J &Cl, at 37’C for 2 hr. Monocional
antibodies
Murine monoclonal antibodies to human C5, C6, C7, C8, C9 and poly C9 were produced according to established procedures (Tamerius et al., 1982). Liposome
swelling assay
These experiments were carried out largely as described by Nikaido and Rosenberg (1981). Acetone-extracted egg phosphatidylcholine (Sigma) (3 pmol) and dicetyiphosphate (0.075 pmol) were dried onto the bottom of a large test tube (2cm diameter) under a stream of Nz and evacuated for 1 hr. The mixture was brought up to 200~1 with water and sonicated for about 30 min. The sonicated solution was dried onto the bottom of the tube and dessicated under reduced pressure. Fourteen per cent dextran (Sigma, average mol. wt = 42,000) was then added in 0.6 ml of 10 mM Tris, pH 7.4. Liposomes were formed by manually shaking the tube and incubation at 37°C for about 2 hr. To assemble the MAC on the liposomes, the components were added in the sequence C5b-6, C8, C9 and C7. The final mixture was incubated at 37°C for 1hr. To assemble the CSb-8 complex, the same procedure was applied, however, the C9 was omitted. The assay itself involved quickly mixing 30~1 of the liposome solution with 0.7 ml of an isotonic sugar solution in 1OmM Tris-HCI, pH 7.4, in a cuvette. The isotonic concentration was determined by using raffinose, a large sugar and control liposome that had only the lipid materials (no protein). At the isotonic concentration (typically about 50 puM), the optical density at 500 nm remains constant. If the pore in the protein-containing liposome was permeable to a specific sugar solute, the rate of influx of the sugar and water was recorded as a decrease in optical density at 500nm (Nikaido and Rosenberg, 1981). The rates of sugar influx obtained in the liposome swelling assay with different sugars were used to determine the size of the protein channel by using the Renkin equation (1954). To determine the effect of antibodies on the rates of liposome swelling, 10 pg of monoclonal antibody solution in a volume of 12 ~1 was added to a portion of the liposome solution (100~1). The mixture was incubated at 37°C for 15 min, then at 4°C for 1 hr, then allowed to come to room temperature.
formed by hand-shaking and incubation at 37’C, the mixture was loaded onto a Sepharose 4B column (Pharmacia) (15 x 1 cm) and eluted in 20 mM Tri-HC1, pH 7.4. The liposomes (and any trapped protein) were collected in the void volume, while the free “‘I-protein was in the included volume. If the iodinated marker protein was small enough to penetrate through the poly C9 channel, very little radioactivity remained with the liposomes. If, however, the protein could not penetrate through the pore, the radioactivity remained trapped within the liposomes eluted in the void volume.
RESULTS
We examined the effect of different amounts of C5b-8 on the rate of swelling of one sugar (arabinose) in the liposome swelling assay. The control liposome containing only C5b-7 had a rate that was essentially 0 (Fig. 1). As increasing amounts of CSb-8 were added, the rates of swelling with arabinose also increased. This indicates that CSb-8 does form a channel and that the degree of liposome swelling is dose-de~ndent. It has been reported that complete marker release from resealed erythrocyte ghosts by C5b-8 requires several hours (Ramm et al., 1982). This suggests that secondary rearrangements of the complex occur or that two or more CSb-8 complexes have to aggregate to produce a channel. It was therefore of interest to examine the effect of time on the rates of swelling using different amounts of C5b-8. When 15 or 30 pmol of C5b-8 was used with the standard amount of lipid (3 pmol phosphatidyl-
.07
I
.U6-
.05-
Li~~some leakage The liposomes were made as above, with or without preformed poly C9 incorporated into the phospholipid bilayer. The dried liquid (with or without the protein) was resuspended in 20mM Tris-HCI, pH 7.4, containing iodinated proteins of various molecular weights. After the liposomes were
10
20
30
40
50
60
C5b.8 Input tpmol) Fig. 1. Rate of liposome swelling as a function of the amount of C5b-8 added. The rates of liposome swelling with different amounts of C5l-A (0-6Opmoi) were compared using the sugar arabinose whose mol. wt is 150. The rate of swelling of C5b-7 is included as a control
Channels
formed
by poly
C9,
C5b-8
and
535
MAC
1IO0 70 = i
'i 50
is E B g 'Z = d
\ IOL
/ I 150 200
\\ E. coli porin /O.6nm \
300
500
I
radwd 101
I
400
30
t
600
"
200
' 400
I
/ 600
L
BOO
I
I I ' 1200 1000
Molecular Weight of Solute
Molecular Weight of Solute
Fig. 2. Effect of time of incubation of liposomes containing C5bP8 on the rate of liposome swelling. The rates of liposome swelling were evaluated for three different amounts of C5b-8 (15, 30 and 60 pmol) at four different time points. The sugar solute used was arabinose.
choline), the rates of swelling of the liposomes were low and did not increase with time (up to 4 hr) (Fig. 2). When a larger amount of CSb-8 (60pmol) was assembled on the membrane, however, the initial rates were low, but these rates increased with time, doubling within 2-3 hr. These results suggest that some sort of time-dependent aggregation may be taking place on the membrane. The liposome swelling assay can be used not only to determine whether a pore is formed or not, but also to estimate the size of the pore using the Renkin equation (1954). The Renkin equation measures the effect of solute size on the rate of penetration through pores. a/a0 = [1 - r/R]‘[l
I
Fig. 3. Size of pore formed by C5b-8. The relative rates of swelling (using the rate of liposome swelling with arabinose as 100%) for C5b8 containing liposomes were plotted. The sugars used were arabinose mol. wt = 150), glucose (mol. wt = 180) sucrose (mol. wt = 342), raffinose (mol. wt = 504), maltoletraose (mol. wt = 666), maltopentaose (mol. wt = 829), maltohexaose (mol. wt = 990) and maltoheptaose (mol. wt = 1153). For comparison the relative rates for a E. co/i porin with a pore of 0.6 nm radius is also shown, along with a hypothetical curve showing a pore of 1.7 nm radius,
molar ratios, increasing amounts of C9 were added to five different sets of CSb8 containing liposomes. C9: C8 ratios of 0, 1, 2, 6 and 12 were chosen. As the ratio increased, the size of the pore increased also, as indicated by the relative rate of liposome swelling (Fig. 4). The actual rates of swelling also increased from 0.32 for C5b8 to 0.059 for CSb-9, 0.065 for C5b9,, 0.070 for C5b9, to 0.092 for C5b-90,,.
- 2.104 r/R +2.09(r/R)3
- 0.95(r/R)5],
where a, a,, r and R are the effective area of the pore, the total cross-sectioned area of the pore, the radius of the solute, and the radius of the pore, respectively. This method involves plotting the relative rates of liposome swelling with various different sized sugars versus the molecular weight of the sugars. A steep slope indicates a small pore. This approach has been used to measure the channel diameter of the Exherichiu coli porin (Nikaido and Rosenberg, 1981), and several bacterial pore-forming toxins (Zalman and Wisnieski, 1984). Figure 3 shows the relative rates of swelling for C5b8 containing liposomes plotted versus the molecular weight of the sugars. For comparison, a curve obtained for E. coli porin (Nikaido and Rosenberg, 1981) and a theoretical curve for a pore with a radius of 1.7 nm are also shown. Using this approach, we estimated that the size of the C5b-8 channel (R in the Renkin equation) was about 1.5 nm radius. This figure agrees quite closely with the upper limit found by other groups using resealed ghosts (Ramm et al., 1982). In order to compare the size of channels formed with the MAC containing C8 and C9 in different
/
60 pmol
.14-
.12P E -g
.I 1
E
s
$
.06
19/
.06 .04
. .
t* .02
JOpmol
. .
ibpmol 1
t
I
.5
1
3
2
Time at 37°C
4
(hr)
Fig. 4. Size of pore produced by C5b9 as a function of C8:C9 molar ratio. The relative rates of swelling (with the rate of swelling with arabinose as 100%) for liposomes containing C5bP8 with various amounts of C9 are shown.
L. S. ZALMAN
536
and H. J. MUELLER-EBERHARD
Poly C9 can be incorporated into liposomes in two ways. Preformed poly C9 is added to phosphohpids and the mixture is then sonicated extensively to ensure thorough mixing of the protein and lipid. When such a suspension was dried down and then rehydrated to form liposomes, the rates obtained in the liposome swelling assay were very high, the oDj,,/min with arabinose was 0.37 for 12 pg protein (Table 1). If C9 is added to preformed liposomes and the solution brought to 50 PM ZnCI,, the C9 polymerized into poly C9 on the outer bilayer of the liposomes. In this case, the rates obtained were somewhat lower with the o~,,,/min being 0.16. This lower rate probably reflects the fact that when the protein was added during the formation of the liposome, the poly C9 incorporated into all of the layers of the multilamelar liposome, but when the protein was added to preformed liposomes, it could only bind to the outer layer. When preformed poly C9 was added to preformed liposomes, the rates were essentially zero (oD,,/min = 0.018). In contrast to forming poly C9, preformed poly C9 was unable to insert into a membrane bilayer. In order to estimate the size of the poly C9 channel, four different batches of liposomes were made with poly C9 incorporated in them. These liposomes were preloaded each with a different radioactive protein and passed over a gel filtration column to remove the free radioactive protein. If the channel formed by poly C9 was larger than the Stokes radius of the iodinated marker protein, the protein leaked out of the liposomes, and the liposomes, which were eluted in the void volume, retained very little radioactivity. If, however, the iodinated protein could not pass through the poly C9 pore, the liposomes that eluted would contain substantial amounts of radioactivity. Liposomes made in the absence of any protein retained about 10% of each of the radioactive markers tested, so 10% retention of marker is the maximum amount that can be incorporated into these liposomes. Alcohol dehydrogenase, with a molecular weight of 150,000 and a Stokes radius of 45 A passed through the poly C9 pore (Fig. 5). C3, [I-amylase and catalase, with molecular weights of 180.000, 200,000 and 230,000, respectively, did not pass through the channel. These results indicate that the channel diameter of poly C9 is between 90 A and 100 A (the diameter of C3). Finally, we wished to determine whether the channels seen with the liposome swelling assay could Table
llr
-Amvlase
1 150
180
200
230
MolecularWeightx 10.’
Fig. 5. Exclusion limit of the poly C9 channel. Four sets of liposomes, all containing poly C9 were pre-loaded with four different iodinated proteins (alcohol dehydrogenase, fi-amylase, or catalase). Each liposome preparation
be inhibited by monoclonal antibodies to the different precursors of the MAC. Liposomes containing CSb8-created pores could be inhibited by antibodies to C8 and slightly by antibodies to C7 (Table 2). Liposomes with C5b9 could be inhibited slightly by antibodies to C9. The CSb-9 channel could be inhibited most strongly by antibodies to the neoantigen of poly C9. DISCUSSlON
The size of the C5b-8 channel as estimated by the liposome swelling assay was small, on the order of 3 nm in diameter. Experiments with planar lipid bilayers indicate that the C5b-8 complex forms a channel with a diameter of between 1 and 2 nm (Michaels et a/., 1976; Jackson et al., 1981). With a low dose of C5bF-8 (15 or 30 pmol per 3 pmol phosphatidylcholine) the relative rates of swelling were low indicating minimal channel formation, At higher doses (60 pmol C5b8) the initial rates of swelling more than doubled and the rates increased as a function of time, doubling within 2-3 hr. This suggests that there is a minimum dose required to form an
I. The poly C9 channel detected by liposome (LP) swelling assay
Conditions of formation of poly COSLP complex Preformed poly C9 (12 pg) incorporated into LP Poly C9 formed (12 pg) I” presence of LP Preformed poly C9 (IO pg) added to preformed LP Amount of LP = 3 uM.
._ OD,,, jmin 0.374 0.160 0.018
C3.
was passed over a Sepharose 4B column. The amount of radioactivity remaining with the liposome preparation as it eluted at the void volume of the column was noted.
Table 2. Inhibition of rate of liposome swelling by mono clonal antibodies LPC5b9 Anti-C5 (I) Anti-C6 (I) Anti-C7 (I) Anti-C8 (5) Anti-C9 (5) Anti-poly C9 (I)
LPCSt+B
+ ++(3 *(I of 5) ++
of 5) *
Channels formed by poly C9, C5b-8 and MAC efficient channel with CCSb8. These results also imply that there may be a time-dependent rearrangement leading to larger and/or more efficient channels. When C9 is added to the (33-8 complex on the liposome, the size of the channel increased as measured by relative rates of swelling. Step-wise addition of increasing amounts of C9, from a C5&8:C9 ratio of 1:0 to l:l, 1:2, 1:6-1:12 led to a corresponding increase in the size of the channel formed. At a ratio of 1: 12, the size of the (3~9 channel approached that of the poly C9 channel. Also, this C5b-9 channel was most effectively blocked by antibodies to the neoantigen of poly C9. Although these results are not conclusive, they imply that at least at a C9 : C8 ratio of 12: 1, the channelforming portion of the membrane attack complex may resemble poly C9. REFERENCES
Biesecker G. and Miiller-Eberhard H. J. (1980) The ninth component of human complement: Purification and physicochemical characterization. J. Immun. 124, 1291-1296. Dalmasso A. P. and Benson B. A. (1981) Lesions of different functional size produced by human and guinea pig complement in sheep red blood cell membranes. J. Immun. 127, 221&2218. Deguchi M., Gillin F. and Gigli I. (1985) Killing of Giardiu lambia trophozytes by complement. Complement 2, 21. Jackson M. B., Stephens C. L. and Lecar H. (1981) Single channel currents induced by complement in antibodycoated cell membranes. Proc. Natn. Acad. Sci. U.S.A. 78, 64214425. Kolb W. P. and Miiller-Eberhard H. J. (1976) The membrane attack mechanism of complement: The three polypeptide chain structure of the eighth component (C8). J. exp. Med. 143, 1131-1139. Luckey M. and Nikaido H. (1980) Specificity of diffusion channels produced by I phage receptor protein of Escherichia coli. Proc. Natn. Acad. Sci. U.S.A. 77, 161-171. Martin D. E.. Chiu F. S., 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. clin. Invest. 80, 226-233. Michaels D. W., Abramovitz A. S., Hammer C. H. and Mayer M. M. (1976) Increased ion permeability of planar
531
lipid bilayer membranes after treatment with the C5b-9 cytolytic attack mechanism of complement. Proc. natn. Acad. Sci. U.S.A. 73, 2852-2856. Miiller-Eberhard H. J. (1988) Molecular organization and function of the complement system. A. Rev. Biochem. 57, 321-347. Nikaido H. and Rosenberg E. Y. (1981) Effect of solute size on diffusion rates through the transmembrane pores of the outer membrane of Escherichia coli. J. gen. Physiol. 77, 121-135. Podack E. R., Esser A. F., Biesecker G. and MiillerEberhard H. J. (1980) Membrane attack complex of complement. J. exp. Med. 151, 301-313. Podack E. R., Kolb W. P., Esser A. F. and Miiller-Eberhard H. J. (1979) Structural similarities between C6 and C7 of human complement. J. Immun. 123, 1071-1077. Podack E. R., Tschopp J. and Miiller-Eberhard H. J. (1982) Molecular organization of C9 within the membrane attack complex of complement induction of circular C9 polymerization by the C5b-8 assembly. J. exp. Med. 156, 268-282. Ramm L. E., Whitlow M. B. and Mayer M. M. (1982) Size of the transmembrane channels produced by complement proteins C5b-8. J. Immun. 129; 1143~1146. _ Ramm L. E.. Whitlow M. B. and Maver M. M. (1985) The relationship between channel size and the number of C9 molecules in the C5b-9 complex. J. Immun. 134, 25962599. Renkin E. M. (1954) Filtration, diffusion and molecular sieving through porous cellulose membranes. J. gen. Physiol. 38, 225-243. Slolfi R. L. (1968) Immune lytic transformation: A state of irreversible damage generated as a result of the reaction of the eighth component in the guinea pig complement system. J. Immun. 100, 46-54. Tamerius J. D., Pangburn M. K. and Miiller-Eberhard H. J. (1982) Selective inhibition of functional sites ofcell-bound C3b by hybridoma-derived antibodies. J. Immun. 128, 512-514. Tamura N., Shimada A. and Chang S. (1972) Further evidence for immune cytolysis by antibody and the first eight components of Complement in the absence of C9. Immunology 22, 131-140. Young D-E., Cohn Z. A. and Podack E. R. (1986) The ninth component of complement and the pore-forming protein (perforin 1) from cytotoxic T cells: Structural, immunological and functional studies. Science, N. Y. 233, 184-190. Zalman L. S. and Wisnieski B. J. (1984) Mechanism of insertion of diphtheria toxin: Peptide entry and pore size determinations. Proc. natn. Acad. Sci. U.S.A. 81, 3341-3345.
Vol. 21, No.
6, pp. 533-537, 1990
0161-5890/90 $3.00 + 0.00 Pergamon Press plc
Printed in Great Britain.
COMPARISON OF CHANNELS FORMED C5b-8 AND THE MEMBRANE ATTACK COMPLEMENT* LEORA SHALTIEL Z.&MAN?
BY POLY COMPLEX
C9, OF
and HANS J. MUELLER-EBERHARD~
tResearch Institute of Scripps Clinic, Department of Immunology, 10666 North Torrey Pines Road, La Jolla, CA 92037, U.S.A. and $Bernhard-Nocht-Institut fiir Tropenmedizin, Bernhard-Nocht-Strasse 74, D-2000 Hamburg 4, F.R.G.
(First received 10 July 1989; accepted in revised form 28 November 1989) Abstract-The channels formed by poly C9, C5b-8 and C5b-9 were examined using the liposome swelling assay. By plotting the relative rate of swelling of CSb-g-containing liposomes vs the molecular weight of the sugar solute and by applying the Renkin equation, the size of the C5b-8 channel was estimated to be 1.5 mm radius. As increasing amounts of C9 were added during the formation of C5b-9, in C8:C9 ratios of 1: 1, 1: 2, 1 : 6 and 1: 12, the size of the function channel increased. Poly C9 had a pore that was somewhat larger than C5b-9 at a C8: C9 ratio of 1: 12. Using molecular sieving experiments with four different iodinated protein size markers, the channel diameter of poly C9 was estimated at between 90 and 100 A. Monoclonal antibodies to different complement proteins were added to the liposomes to see which might inhibit the channels. C5b-8 containing liposomes could be inhibited by antibodies to C8. Liposomes containing CSb-9 could be inhibited slightly by antibodies to C9 and most strongly by antibodies to the neoantigen of poly C9.
INTRODUCTION membrane attack complex (MAC) of complement consists of five different proteins: CSb, C6, C7, C8 and C9. Assembly of these proteins into the MAC follows cleavage of C5 by the CS convertase of the classical or alternative pathway and generation of C5b. C5b combines with C6 and C7 to form the C5b-7 complex. This complex has a metastable membrane binding site and can attach to the surface of the target membrane. Subsequently, C8 binds to the C5b-7 complex to form CSb-8. The last protein to become incorporated is C9 and the number of C9 molecules that bind varies between 1 and 16 (Miiller-Eberhard, 1988). At high input of C9, the C5b-9 complex forms characteristic rings that can be seen in the electron microscope (Podack et al., 1982). The C5b-8 complex forms a small functional channel, whose formation is time-dependent. It has been shown that C5b-8 can produce channels in a variety of target membranes, including sheep erythrocytes, (E,) (Slolfi, 1968; Tamura et al., 1972), resealed E, ghosts (Ramm et al., 1982), black lipid membranes (Michaels et al., 1976), M21 human melanoma cells (Martin et al., 1987) and gardia lamblia (Deguchi et al., 1985). The size of the channel was estimated by Ramm et al. (1982) to be between 9 and 30 A in diameter. The
*Publication No. 5951-IMM. This research project was supported by National Health Services grants CA47869 and AIl7354, and an Established Investigator Award from the American Heart Association to Dr Zalman.
The size of the MAC pore can vary from 10 to over 100 A depending on the complement dose. With very high concentrations of human complement, a maximum diameter of IlOA was attained in resealed erythrocyte membrane ghosts (Dalmasso and Benson, 1981). Mayer’s group reported that with three C9 molecules per C8, the minimal channel diameter was 3OA and with four molecules of C9 per C8, the minimal channel diameter was 4OA (Ramm et al., 1985). The C9 component is the protein largely responsible for the formation of structural membrane lesion associated with the MAC (Miiller-Eberhard, 1988). Purified C9 can polymerize to form tubular complexes that resemble the MAC when viewed on the electron microscopy (EM). The internal diameter of the tubule as measured from EM micrographs is about 100 A (Young et al., 1986). The purpose of the present work was to determine the size variation of the MAC channel which the C9: C8 molar ratio was increased from 0 to 12. Channel size was assessed using the liposome swelling assay developed by Luckey and Nikaido (1980). In addition, we present data showing that poly C9 produced from isolated C9 in absence of C5b8 is a channel forming structure with functional channel diameters between 90 and 100 A. MATERIALS
AND METHODS
Isolation of proteins C5bA (Podack et al., 1980), C7 (Podack et al., 1979), C8 (Kolb and Miiller-Eberhard, 1976) and C9 533
L.
534
%ZALMANand H.J. MILLER-EBERHAR~
(Biesecker and Miiller-Eberhard. 1980) were isolated by established procedures. Poly C9 was made by incubating purified C9 with 50 ,&J &Cl, at 37’C for 2 hr. Monocional
antibodies
Murine monoclonal antibodies to human C5, C6, C7, C8, C9 and poly C9 were produced according to established procedures (Tamerius et al., 1982). Liposome
swelling assay
These experiments were carried out largely as described by Nikaido and Rosenberg (1981). Acetone-extracted egg phosphatidylcholine (Sigma) (3 pmol) and dicetyiphosphate (0.075 pmol) were dried onto the bottom of a large test tube (2cm diameter) under a stream of Nz and evacuated for 1 hr. The mixture was brought up to 200~1 with water and sonicated for about 30 min. The sonicated solution was dried onto the bottom of the tube and dessicated under reduced pressure. Fourteen per cent dextran (Sigma, average mol. wt = 42,000) was then added in 0.6 ml of 10 mM Tris, pH 7.4. Liposomes were formed by manually shaking the tube and incubation at 37°C for about 2 hr. To assemble the MAC on the liposomes, the components were added in the sequence C5b-6, C8, C9 and C7. The final mixture was incubated at 37°C for 1hr. To assemble the CSb-8 complex, the same procedure was applied, however, the C9 was omitted. The assay itself involved quickly mixing 30~1 of the liposome solution with 0.7 ml of an isotonic sugar solution in 1OmM Tris-HCI, pH 7.4, in a cuvette. The isotonic concentration was determined by using raffinose, a large sugar and control liposome that had only the lipid materials (no protein). At the isotonic concentration (typically about 50 puM), the optical density at 500 nm remains constant. If the pore in the protein-containing liposome was permeable to a specific sugar solute, the rate of influx of the sugar and water was recorded as a decrease in optical density at 500nm (Nikaido and Rosenberg, 1981). The rates of sugar influx obtained in the liposome swelling assay with different sugars were used to determine the size of the protein channel by using the Renkin equation (1954). To determine the effect of antibodies on the rates of liposome swelling, 10 pg of monoclonal antibody solution in a volume of 12 ~1 was added to a portion of the liposome solution (100~1). The mixture was incubated at 37°C for 15 min, then at 4°C for 1 hr, then allowed to come to room temperature.
formed by hand-shaking and incubation at 37’C, the mixture was loaded onto a Sepharose 4B column (Pharmacia) (15 x 1 cm) and eluted in 20 mM Tri-HC1, pH 7.4. The liposomes (and any trapped protein) were collected in the void volume, while the free “‘I-protein was in the included volume. If the iodinated marker protein was small enough to penetrate through the poly C9 channel, very little radioactivity remained with the liposomes. If, however, the protein could not penetrate through the pore, the radioactivity remained trapped within the liposomes eluted in the void volume.
RESULTS
We examined the effect of different amounts of C5b-8 on the rate of swelling of one sugar (arabinose) in the liposome swelling assay. The control liposome containing only C5b-7 had a rate that was essentially 0 (Fig. 1). As increasing amounts of CSb-8 were added, the rates of swelling with arabinose also increased. This indicates that CSb-8 does form a channel and that the degree of liposome swelling is dose-de~ndent. It has been reported that complete marker release from resealed erythrocyte ghosts by C5b-8 requires several hours (Ramm et al., 1982). This suggests that secondary rearrangements of the complex occur or that two or more CSb-8 complexes have to aggregate to produce a channel. It was therefore of interest to examine the effect of time on the rates of swelling using different amounts of C5b-8. When 15 or 30 pmol of C5b-8 was used with the standard amount of lipid (3 pmol phosphatidyl-
.07
I
.U6-
.05-
Li~~some leakage The liposomes were made as above, with or without preformed poly C9 incorporated into the phospholipid bilayer. The dried liquid (with or without the protein) was resuspended in 20mM Tris-HCI, pH 7.4, containing iodinated proteins of various molecular weights. After the liposomes were
10
20
30
40
50
60
C5b.8 Input tpmol) Fig. 1. Rate of liposome swelling as a function of the amount of C5b-8 added. The rates of liposome swelling with different amounts of C5l-A (0-6Opmoi) were compared using the sugar arabinose whose mol. wt is 150. The rate of swelling of C5b-7 is included as a control
Channels
formed
by poly
C9,
C5b-8
and
535
MAC
1IO0 70 = i
'i 50
is E B g 'Z = d
\ IOL
/ I 150 200
\\ E. coli porin /O.6nm \
300
500
I
radwd 101
I
400
30
t
600
"
200
' 400
I
/ 600
L
BOO
I
I I ' 1200 1000
Molecular Weight of Solute
Molecular Weight of Solute
Fig. 2. Effect of time of incubation of liposomes containing C5bP8 on the rate of liposome swelling. The rates of liposome swelling were evaluated for three different amounts of C5b-8 (15, 30 and 60 pmol) at four different time points. The sugar solute used was arabinose.
choline), the rates of swelling of the liposomes were low and did not increase with time (up to 4 hr) (Fig. 2). When a larger amount of CSb-8 (60pmol) was assembled on the membrane, however, the initial rates were low, but these rates increased with time, doubling within 2-3 hr. These results suggest that some sort of time-dependent aggregation may be taking place on the membrane. The liposome swelling assay can be used not only to determine whether a pore is formed or not, but also to estimate the size of the pore using the Renkin equation (1954). The Renkin equation measures the effect of solute size on the rate of penetration through pores. a/a0 = [1 - r/R]‘[l
I
Fig. 3. Size of pore formed by C5b-8. The relative rates of swelling (using the rate of liposome swelling with arabinose as 100%) for C5b8 containing liposomes were plotted. The sugars used were arabinose mol. wt = 150), glucose (mol. wt = 180) sucrose (mol. wt = 342), raffinose (mol. wt = 504), maltoletraose (mol. wt = 666), maltopentaose (mol. wt = 829), maltohexaose (mol. wt = 990) and maltoheptaose (mol. wt = 1153). For comparison the relative rates for a E. co/i porin with a pore of 0.6 nm radius is also shown, along with a hypothetical curve showing a pore of 1.7 nm radius,
molar ratios, increasing amounts of C9 were added to five different sets of CSb8 containing liposomes. C9: C8 ratios of 0, 1, 2, 6 and 12 were chosen. As the ratio increased, the size of the pore increased also, as indicated by the relative rate of liposome swelling (Fig. 4). The actual rates of swelling also increased from 0.32 for C5b8 to 0.059 for CSb-9, 0.065 for C5b9,, 0.070 for C5b9, to 0.092 for C5b-90,,.
- 2.104 r/R +2.09(r/R)3
- 0.95(r/R)5],
where a, a,, r and R are the effective area of the pore, the total cross-sectioned area of the pore, the radius of the solute, and the radius of the pore, respectively. This method involves plotting the relative rates of liposome swelling with various different sized sugars versus the molecular weight of the sugars. A steep slope indicates a small pore. This approach has been used to measure the channel diameter of the Exherichiu coli porin (Nikaido and Rosenberg, 1981), and several bacterial pore-forming toxins (Zalman and Wisnieski, 1984). Figure 3 shows the relative rates of swelling for C5b8 containing liposomes plotted versus the molecular weight of the sugars. For comparison, a curve obtained for E. coli porin (Nikaido and Rosenberg, 1981) and a theoretical curve for a pore with a radius of 1.7 nm are also shown. Using this approach, we estimated that the size of the C5b-8 channel (R in the Renkin equation) was about 1.5 nm radius. This figure agrees quite closely with the upper limit found by other groups using resealed ghosts (Ramm et al., 1982). In order to compare the size of channels formed with the MAC containing C8 and C9 in different
/
60 pmol
.14-
.12P E -g
.I 1
E
s
$
.06
19/
.06 .04
. .
t* .02
JOpmol
. .
ibpmol 1
t
I
.5
1
3
2
Time at 37°C
4
(hr)
Fig. 4. Size of pore produced by C5b9 as a function of C8:C9 molar ratio. The relative rates of swelling (with the rate of swelling with arabinose as 100%) for liposomes containing C5bP8 with various amounts of C9 are shown.
L. S. ZALMAN
536
and H. J. MUELLER-EBERHARD
Poly C9 can be incorporated into liposomes in two ways. Preformed poly C9 is added to phosphohpids and the mixture is then sonicated extensively to ensure thorough mixing of the protein and lipid. When such a suspension was dried down and then rehydrated to form liposomes, the rates obtained in the liposome swelling assay were very high, the oDj,,/min with arabinose was 0.37 for 12 pg protein (Table 1). If C9 is added to preformed liposomes and the solution brought to 50 PM ZnCI,, the C9 polymerized into poly C9 on the outer bilayer of the liposomes. In this case, the rates obtained were somewhat lower with the o~,,,/min being 0.16. This lower rate probably reflects the fact that when the protein was added during the formation of the liposome, the poly C9 incorporated into all of the layers of the multilamelar liposome, but when the protein was added to preformed liposomes, it could only bind to the outer layer. When preformed poly C9 was added to preformed liposomes, the rates were essentially zero (oD,,/min = 0.018). In contrast to forming poly C9, preformed poly C9 was unable to insert into a membrane bilayer. In order to estimate the size of the poly C9 channel, four different batches of liposomes were made with poly C9 incorporated in them. These liposomes were preloaded each with a different radioactive protein and passed over a gel filtration column to remove the free radioactive protein. If the channel formed by poly C9 was larger than the Stokes radius of the iodinated marker protein, the protein leaked out of the liposomes, and the liposomes, which were eluted in the void volume, retained very little radioactivity. If, however, the iodinated protein could not pass through the poly C9 pore, the liposomes that eluted would contain substantial amounts of radioactivity. Liposomes made in the absence of any protein retained about 10% of each of the radioactive markers tested, so 10% retention of marker is the maximum amount that can be incorporated into these liposomes. Alcohol dehydrogenase, with a molecular weight of 150,000 and a Stokes radius of 45 A passed through the poly C9 pore (Fig. 5). C3, [I-amylase and catalase, with molecular weights of 180.000, 200,000 and 230,000, respectively, did not pass through the channel. These results indicate that the channel diameter of poly C9 is between 90 A and 100 A (the diameter of C3). Finally, we wished to determine whether the channels seen with the liposome swelling assay could Table
llr
-Amvlase
1 150
180
200
230
MolecularWeightx 10.’
Fig. 5. Exclusion limit of the poly C9 channel. Four sets of liposomes, all containing poly C9 were pre-loaded with four different iodinated proteins (alcohol dehydrogenase, fi-amylase, or catalase). Each liposome preparation
be inhibited by monoclonal antibodies to the different precursors of the MAC. Liposomes containing CSb8-created pores could be inhibited by antibodies to C8 and slightly by antibodies to C7 (Table 2). Liposomes with C5b9 could be inhibited slightly by antibodies to C9. The CSb-9 channel could be inhibited most strongly by antibodies to the neoantigen of poly C9. DISCUSSlON
The size of the C5b-8 channel as estimated by the liposome swelling assay was small, on the order of 3 nm in diameter. Experiments with planar lipid bilayers indicate that the C5b-8 complex forms a channel with a diameter of between 1 and 2 nm (Michaels et a/., 1976; Jackson et al., 1981). With a low dose of C5bF-8 (15 or 30 pmol per 3 pmol phosphatidylcholine) the relative rates of swelling were low indicating minimal channel formation, At higher doses (60 pmol C5b8) the initial rates of swelling more than doubled and the rates increased as a function of time, doubling within 2-3 hr. This suggests that there is a minimum dose required to form an
I. The poly C9 channel detected by liposome (LP) swelling assay
Conditions of formation of poly COSLP complex Preformed poly C9 (12 pg) incorporated into LP Poly C9 formed (12 pg) I” presence of LP Preformed poly C9 (IO pg) added to preformed LP Amount of LP = 3 uM.
._ OD,,, jmin 0.374 0.160 0.018
C3.
was passed over a Sepharose 4B column. The amount of radioactivity remaining with the liposome preparation as it eluted at the void volume of the column was noted.
Table 2. Inhibition of rate of liposome swelling by mono clonal antibodies LPC5b9 Anti-C5 (I) Anti-C6 (I) Anti-C7 (I) Anti-C8 (5) Anti-C9 (5) Anti-poly C9 (I)
LPCSt+B
+ ++(3 *(I of 5) ++
of 5) *
Channels formed by poly C9, C5b-8 and MAC efficient channel with CCSb8. These results also imply that there may be a time-dependent rearrangement leading to larger and/or more efficient channels. When C9 is added to the (33-8 complex on the liposome, the size of the channel increased as measured by relative rates of swelling. Step-wise addition of increasing amounts of C9, from a C5&8:C9 ratio of 1:0 to l:l, 1:2, 1:6-1:12 led to a corresponding increase in the size of the channel formed. At a ratio of 1: 12, the size of the (3~9 channel approached that of the poly C9 channel. Also, this C5b-9 channel was most effectively blocked by antibodies to the neoantigen of poly C9. Although these results are not conclusive, they imply that at least at a C9 : C8 ratio of 12: 1, the channelforming portion of the membrane attack complex may resemble poly C9. REFERENCES
Biesecker G. and Miiller-Eberhard H. J. (1980) The ninth component of human complement: Purification and physicochemical characterization. J. Immun. 124, 1291-1296. Dalmasso A. P. and Benson B. A. (1981) Lesions of different functional size produced by human and guinea pig complement in sheep red blood cell membranes. J. Immun. 127, 221&2218. Deguchi M., Gillin F. and Gigli I. (1985) Killing of Giardiu lambia trophozytes by complement. Complement 2, 21. Jackson M. B., Stephens C. L. and Lecar H. (1981) Single channel currents induced by complement in antibodycoated cell membranes. Proc. Natn. Acad. Sci. U.S.A. 78, 64214425. Kolb W. P. and Miiller-Eberhard H. J. (1976) The membrane attack mechanism of complement: The three polypeptide chain structure of the eighth component (C8). J. exp. Med. 143, 1131-1139. Luckey M. and Nikaido H. (1980) Specificity of diffusion channels produced by I phage receptor protein of Escherichia coli. Proc. Natn. Acad. Sci. U.S.A. 77, 161-171. Martin D. E.. Chiu F. S., 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. clin. Invest. 80, 226-233. Michaels D. W., Abramovitz A. S., Hammer C. H. and Mayer M. M. (1976) Increased ion permeability of planar
531
lipid bilayer membranes after treatment with the C5b-9 cytolytic attack mechanism of complement. Proc. natn. Acad. Sci. U.S.A. 73, 2852-2856. Miiller-Eberhard H. J. (1988) Molecular organization and function of the complement system. A. Rev. Biochem. 57, 321-347. Nikaido H. and Rosenberg E. Y. (1981) Effect of solute size on diffusion rates through the transmembrane pores of the outer membrane of Escherichia coli. J. gen. Physiol. 77, 121-135. Podack E. R., Esser A. F., Biesecker G. and MiillerEberhard H. J. (1980) Membrane attack complex of complement. J. exp. Med. 151, 301-313. Podack E. R., Kolb W. P., Esser A. F. and Miiller-Eberhard H. J. (1979) Structural similarities between C6 and C7 of human complement. J. Immun. 123, 1071-1077. Podack E. R., Tschopp J. and Miiller-Eberhard H. J. (1982) Molecular organization of C9 within the membrane attack complex of complement induction of circular C9 polymerization by the C5b-8 assembly. J. exp. Med. 156, 268-282. Ramm L. E., Whitlow M. B. and Mayer M. M. (1982) Size of the transmembrane channels produced by complement proteins C5b-8. J. Immun. 129; 1143~1146. _ Ramm L. E.. Whitlow M. B. and Maver M. M. (1985) The relationship between channel size and the number of C9 molecules in the C5b-9 complex. J. Immun. 134, 25962599. Renkin E. M. (1954) Filtration, diffusion and molecular sieving through porous cellulose membranes. J. gen. Physiol. 38, 225-243. Slolfi R. L. (1968) Immune lytic transformation: A state of irreversible damage generated as a result of the reaction of the eighth component in the guinea pig complement system. J. Immun. 100, 46-54. Tamerius J. D., Pangburn M. K. and Miiller-Eberhard H. J. (1982) Selective inhibition of functional sites ofcell-bound C3b by hybridoma-derived antibodies. J. Immun. 128, 512-514. Tamura N., Shimada A. and Chang S. (1972) Further evidence for immune cytolysis by antibody and the first eight components of Complement in the absence of C9. Immunology 22, 131-140. Young D-E., Cohn Z. A. and Podack E. R. (1986) The ninth component of complement and the pore-forming protein (perforin 1) from cytotoxic T cells: Structural, immunological and functional studies. Science, N. Y. 233, 184-190. Zalman L. S. and Wisnieski B. J. (1984) Mechanism of insertion of diphtheria toxin: Peptide entry and pore size determinations. Proc. natn. Acad. Sci. U.S.A. 81, 3341-3345.