Group Title: ninth componet of human complement
Title: The ninth componet of human complement
Full Citation
Permanent Link:
 Material Information
Title: The ninth componet of human complement
Physical Description: Book
Language: English
Creator: Shiver, John W., 1957-
Copyright Date: 1985
 Record Information
Bibliographic ID: UF00102791
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: ltuf - AEH3715
oclc - 14706828

Full Text









LIST OF TABLES.................................................... iv

LIST OF FIGURES................................................... v

LIST ABBREVIATIONS..............................................viii

ABSTRACT....................................................... ix


Introduction.................................................... 1
Materials and Methods................... .............. ........ 11
Purification of human C9..................................... 11
NaDodSO4-PAGE .............. ........ ........................ 13
Determination of C9 hemolytic activity........................ 13
Iodination of C9.................... ......... ... .. .. ... .. 15
Heat polymerization of C9..................................... 15
HPLC elution of heat-treated C9.............................. 15
Vesicle formation............................................ 16
Carboxyfluorescein experiments............................... 16
C9 interactions with SUV: Effects on hemolytic activity...... 17
C9 interactions with SUV: 1H NMR............................ 17
Preparation of a-thrombin fragments of C9..................... 17
Planar lipid bilayer apparatus............................... 18
Protein additions to the bilayer.............................. 22

CHAPTER 2. POLYMERIZATION OF C9.................................. 24

Introduction.......................................... .................. 24
Results.............. ................ ............... . ... 26
Purification of C9........................ ............. 26
Heat polymerization of C9.................................. .. 27
C9 interactions with SUV..................................... 36
Discussion......................... .. ............... 39


Introduction.................. ........................... ......... 45
Results ................................................. ...... 53
Release of carboxyfluorescein front LUV by reactive lysis...... 53
C5b-8-mediated conductances.................................. 55
Cbb-9-mediated conductances................................. 55


MAC-mediated conductances................................... 61
C9, C9n additions to asolectin bilayers....................... 64
Discussion ............................... .................... 64


Introduction................... ....................... ...... 67
Results....................................................... .. 71
Characterization of proteolytic fragments of C9 (C9a, C9b).... 71
C9/planar lipid bilayer interactions ......................... 74
Lipid specificity of C9b/bilayer conductances................. 78
Ion selectivity of C9b/bilayer conductances................... 81
Effect of varying the solubilizing solutions on C9b
channel activity....... .............. ........ .... ... 84
C9a and C9/bilayer interactions.............................. 85
Peptide-mediated marker release from CF/vesicles.............. 87
Peptide-mediated hemolysis by C9a and C9b..................... 95
Discussion.......................... .... .................. ... ... 97


Summary............................ .............. ....... 101
Introduction.......................................... .. ....... 101
Materials and Methods....................................... 102
Source of Materials.......................................... 102
Preparation of 13C-enriched NeuNAc............................ 102
Preparation of CMP-NeuNAc.................................... 102
Instrumentation............................................... 103
Results and Discussion................... ......... ..... ...... 104

LIST OF REFERENCES................................................. 112

BIOGRAPHICAL SKETCH................................................ 119


Table Page

1. Physical Properties of Complement Proteins................. 5

2. Pore Diameter Estimations from Complement-Mediated
Conductances........................ .................... 63

3. Estimation of C9b Pore Diameter from Single Channel
Conductances ........................ .................... 77

4. C9b Lipid Specificity...................................... 80

5. Erythrocyte Hemolysis with C9a and C9b.................... 96

6. 13C NMR Chemical Shifts of 0.5M CMP-NeuNAc................. 1U7


Figure Page

1. Schematic representation of the Classical and
Alternative pathways of complement....................... 2

2. Hypothetical representation of the ultrastructural
appearance of the MAC.................................... 8

3. Diagram of the NaDodSO4-PAGE preparative gel apparatus
used to purify C9a and C9b............................... 19

4. Schematic representation of the planar lipid bilayer
apparatus ................................................ 20

b. Sodium dodecyl sulfate gel electrophoresis of C9
purified from human plasma............................... 28

6. Comparison of the hemolytic activities of C9 purified
from human plasma and Cohn fraction III with unpurified
C9 ..... .................................................. 29

7. HPLC elution profiles of C9 polymerization................ 30

8. Monomer and polymer C9 elution from HPLC relative to
molecular weight markers................................. 31

9. Time course of the loss of monomer C9 and the increase
in polymer C9 at 460C calculated from HPLC elution
profiles.............................................. . 32

10. The loss of hemolytic activity of C9 incubated 72 h at
370C as compared with unheated C9........................ 34

11. The loss of hemolytic activity of C9 incubated 3 h at
460C as compared to unheated C9......................... 35

12. The hemolytic activity of C9 after treatment with egg
lecithin SUV at 2bC and 37C ............................ 37

13. Proton NMR of egg lecithin SUV with and without
addition of C9..................... ....................... 38

14. Release of carboxyfluorescein from LUV mediated by
complement reactive lysis................................. 54

Figure Page

15. Voltage-dependent conductance changes of an asolectin
planar lipid bilayer mediated by C5b-8................. 56

16. Single channel conductances of an asolectin planar lipid
bilayer mediated by C5b-9 at -40 mV...................... 58

17. Conductances of an asolectin lipid bilayer mediated by
C5b-9 a function of voltage.............................. 59

18. Conductances of an asolectin lipid bilayer mediated by
purified MAC as a function of voltage.................... 62

19. Schematic representation of C9 showing the proposed hydro-
philic and hydrophobic domains.......................... 68

20. C9a and C9b elution profile from a preparative NaDodSO4-
PAGE tube gel ........................................... 72

21. Autoradiogram of 1251 radiolabeled C9n and samples
NaDodSO4-PAGE corresponding to fractions eluted from
the preparative gel.................................... 73

22. C9-mediated single channels with asolectin bilayers
at -40 mV .......................................... .. 75

23. Histogram of C9b single channel conductances
at -40 mV ............................................. 76

24. C9b conductances with an asolectin planar lipid bilayer
as a function of voltage............................... 79

25. Ion selectivity of the C9b channel...................... 82

26. Protein additions to an asolectin planar bilayer of
C9b and C9..................................... ...... .... 86

27. Carboxyfluorescein release from egg lecithin LUV after
additions of C9b, C9a, or C9............................. 9

28. Concentration dependence of CF release from egg lecithin
LUV mediated by C9a and C9b.............................. 90

29. Inhibition of C9b-induced marker release from egg
lecithin LUV by monoclonal C9 antibody................... 91

30. Carboxyfluorescein release from egg lecithin LUV
mediated by C9, C9a, C9b solubilized in NaDodS04-PAGE
and SB 3-14............................. ..... ......... 94

31. 13C NMR spectrum obtained at 25.0 MHz of a 020 solution
of CMP-NeuNAc at pD 9.5................................. 105



32. 13C NMR spectrum obtained at 25.0 MHz of a D20
solution of CMP-NeuNAc (101.28 ppm) 90% 13C enriched
at C-2 in the pyronose ring............................. 108

33. 31P NMR spectrum obtained at 40.5 MHz of a 020
solution of CMP-NeuNAc at pD 9.5 containing EDTA......... 110


BLM Black Lipid Membrane

BSA Bovine Serum Albumin

CF 6-Carboxyfluorescein

CMP Cytidine b'-Monophosphate

DPPC Dipalmitoyl L-a-Phosphatidyl Choline

EDTA Ethylenediamine-Tetraacetic Acid

HPLC High Pressure Liquid Chromatography

KDodSO4 Potassium Dodecyl Sulfate

LUV Large Unilamellar Vesicles

MAC Membrane Attack Complex

ManNAc N-Acetylmannosamine

MOPS 2,4-Morpholinepropanesulfonic Acid

NaDodSO4 Sodium Dodecyl Sulfate

NeuNac N-Acetylneuraminic Acid

NMR Nuclear Magnetic Resonance

PMSF Phenylmethylsulfonyl Fluoride

SB 3-12 Dodecyl Dimethyl Ammoniopropanesulfonate

SB 3-14 Myristoyl Dimethyl Ammoniopropanesulfonate

SUV Small Unilamellar Vesicles

TRIS Tris(Hydrnxymethyl) Aminomethane


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



John W. Shiver

August, 1985

Chairman: William Weltner
Major Department: Chemistry

The role of C9 in complement-mediated membranolysis was studied

using lipid bilayers in the form of artificial lipid vesicles, planar

lipid bilayers, and erythrocytes. C9 has been implicated as having a

major role in causing membrane damage as part of the membrane attack

complex, C5b-9. Contrary to reports by other investigators, who found

that C9 could be heat-treated under certain non-physiological conditions

to form a tubular polymer, it was found that heat treatment of C9 caused

little aggregation or loss in hemolytic activity. Protease treatment of

C9 with thrombin produces two peptides, C9a and C9b, which were isolated

by gel elecrophoresis under denaturing conditions. C9b was found to

produce voltage-dependent single channels of 12 1 pS with planar lipid

bilayers which demonstrated ion and lipid specificity. No conductance

changes were observed with either intact C9 or isolated C9a. Similar

single channels (' 12 pS) were observed after C9 additions to a C5b-8

bilayer which in addition included a range of channel sizes from

2.5 pS to 1000 pS. This range of conductances corresponds to calculated

pore sizes of 0.9 to 40 8 for a water-filled channel. C5b-8 alone

produced non-specific, detergent-like conductance fluctuations. It is

concluded that the carboxyl-terminus of C9, C9h, contains a channel-

forming ability which could be responsible for the membranolytic

activity of the C5b-9 complex. The actual mechanism by which the C5b-9

complex exerts its membranolytic activity, whether through the formation

of discrete channels or a detergent-like membrane reorientation, may

depend upon the target membrane and remains to be resolved.



Complement proteins belong to a group of plasma proteins which have

a variety of functions of immunological significance. Among these are

opsonization, chemotaxis, and activation of 3 lymphocytes and macro-

phages (19,50,63,79,116). Complement also kills or inactivates invading

organisms such as bacteria, viruses, fungi, and malignant cells (2U,32,

63,79,86). This cytolytic ability is initiated by one of two mecha-

nisms: the classical pathway (32,63), which was the first to be recog-

nized (31), and the alternative pathway (36,9U,92). These pathways

differ in the manner in which complement activity is initiated but the

cytolytic mechanism is the same for the two pathways. The manner in

which complement is able to kill or inactivate organisms or cells has

oeen studied by a number of researchers and is further investigated in

the experiments described in this text.

The classical pathway is comprised of the sequence of proteins Cl,

C2, C3,..., C9. In this pathway, depicted in Fig. I, complement is

activated by an antigen-antibody reaction at the surface of a target

membrane. This immune complex initiates a series of reactions in which

C1 through C4 are each activated by proteolysis which transforms them

into protein complexes capable of proteolysis and activation of the next

component. The peptide containing the N-terminus of the intact protein

is denoted by the addition of an "a" following the protein name. Other

C2b, C4a
Ig + Cirl
I C2, C4

C3 Convertase C5 Convertase


C3 (H
C3 (HzO)

B, C3

Ba, C3a

Site II

Fig. 1. Schematic representation of the Classical and Alternative path-
ways of complement. The shaded areas represent the membrane surfaces at
which most of these reactions occur while the bars over some of the
protein complexes indicate that these are acting as enzymes (from Esser,


Site III

peptides generated from the parent are denoted by the letters "b", "c",

"d", etc. as the carboxyl terminus of the protein is approached. Each

of the cleaved protein fragments becomes either a protease by itself or

in association with other fragments, or becomes an anaphylatoxin or

chemotaxin (49). These reactions culminate as a protein complex,

C4b,C2a,C3b, capable of cleaving C5 into C5a and C5b. This complex is

called a CS convertase. C5a is an anaphylatoxin and chenotaxin while

Cbb is capable of initiating formation of a protein complex with C6, C7,

C8, and C9 (C5b-9) on the target membrane which effects cytolysis.

C5b-9 is referred to as the membrane attack complex or MAC.

The alternative pathway is independent of antigen-antibody inter-

actions. As depicted in Fig. 1, it is composed of C3 and five other

proteins (Factors B, i], H, I, and properdin) which are activated by

polysaccharides in the form of zymosan, inulin, and lipopolysaccharides

from the membrane surfaces of certain viruses, fungi, bacteria, or para-

sites. C3 continuously hydrolyzes to a small extent in whole olood to

form snall quantities of C3(H20). This complex reacts with Factor B to

form a C3b,B complex which in turn is cleaved by Factor D to form a

C3b,Bb complex. This complex is capable of cleaving another C3 to C3b

thus amplifying the process and forming a third complex, C3b,Bb,C3b,

which is also a C5 convertase. CSb generation leads to C5b-9 (MAC)

formation on the membrane substrate as in the classical pathway.

Factors H and I regulate the alternative pathway by inactivating C3b

complexes while properdin is thought to stabilize the C3b,B complex.

Investigators (66,68) have isolated preformed C5h-6 and subsequently

reacted this species with C7, C8, and C9 in the presence of a membrane

substrate and nave successfully achieved lysis. This lysis, which is

effected by using only the purified terminal complement components is

called "reactive lysis." Table 1 summarizes some of the physical

properties of complement proteins from both pathways.

Reactive lysis on erythrocytes and model membrane substrates such

as lipid vesicles and resealed erythrocyte ghosts is a useful method to

study the manner in which the plasma-soluble terminal complement pro-

teins are converted into the amphiphilic proteins which comprise the

MAC. Cleavage of C5 by the C5 convertase of either pathway generates

C5b. C5b is a very unstable fragment which quickly loses cytolytic

activity unless it associates with C6. C5b6 and C7 form a stable

complex: C5b-7. C7 appears to enhance membrane association of the

proteins with first penetration of the hydrophobic lipid domains

occurring at this step. Addition of C8 and C9 allow formation of the

complexes C5b-8 and C5b-9. Each of these complexes is capable of

membrane damage with vesicles and erythrocytes. C5b-8 mediated lysis

appears to be limited to these substrates although it also kills

Neisseria (43). The addition of C9 greatly accelerates cytolysis with

much smaller quantities of all five proteins required. C5b-9 also lyses

a much greater diversity of target membranes than C5b-8 alone.

The manner in which these complexes can destabilize membranes to

effect cytolysis has been investigated by a number of researchers in

recent years. Most agree that the conversion of these proteins from a

hydrophilic state to an amphiphilic one is mediated by the "restricted

unfolding and exposure of hydropnobic domains" (95) which then insert

into a membrane rather than activation by an enzymatic process. How-

ever, it is still a subject of controversy (77,86) whether protease

activity by components within the MAC generate the active lytic forms of

Table 1. Physical Properties of Complement Proteins

Polypeptide Chains






M.W. (Da)




C8 or C9. It has been shown (65) that C6 and C7 can be inactivated for

hemolysis by treatment with the serine esterase inhibitors phenylmethyl-

sulfonyl fluoride (PMSF) and diisopropylphosphofluoridate (DFP) although

the relevance of this observation to complement cytolysis is not

certain. However, these results could not be confirmed by DiScipio and

Gagnon (27).

Several hypotheses (62,68,78) have been offered by different

workers to explain complement lysis. These investigators chiefly dis-

pute the molecular composition of the MAC and the mechanism by which MAC

lyses a membrane. Central to these conjectures over the last 20 years

was the observation made by electron microscopists (10,51) of the

appearance of a circular ring 20 nm in outside diameter on the surface

of lysed membranes. This structure appeared only on membranes which

were lysed by all the terminal complement proteins, i.e., not with

C5b-8. The appearance of the tubular, channel-like structure together

with the observation that it seemed to be filled with negative stain,

implying that it had a hollow interior, made this structure appear

likely to be the cause of membrane destruction by complement. Subse-

quent membrane extractions of the rings by detergents and NaDodSO4-PAGE

analysis revealed that the ring was composed of the terminal complement

proteins. The composition of this ring has been disputed by different

investigators. Kolb et al. (64) reported finding up to six C9 molecules

per C8 in the MAC while Podack et al. (96) and Tschopp et al. (107)

proposed that the MAC is composed of (C5b-8)C912-16. Biesecker et al.

(7) concluded from hydrodynamic analyses of detergent-solubilized

complexes that the MAC is a dimer of C5b-9 with up to six C9 molecules.

Functional lesion sizes have been shown by Boyle et al. (12) to be as


small as 0.72 nm and by Ramm and Mayer (97) to be as large as 5.5 nm.

Boyle also reported that the actual channel size depended upon C8/C9

ratio. Sims and Lauf (104) were unable to confirm this data but found

that the channel size is dependent on the complement doses with larger

doses giving larger channels. Many researchers (77,95,113) now consider

the MAC to be heterogenous in composition with one C5b-8 and a variable

number of C9 molecules per MAC. Fig. 2 shows a schematic of the ultra-

structural appearance of these lesions which have come to be known

synonomously as the MAC.

Mayer (78) and his coworkers proposed in 1972 a model which has

become known as the "doughnut" theory that suggests that C5b-9 forms a

channel with a hydrophilic core that spans the lipid bilayer of a

membrane and allows efflux of cytosol thus disrupting the integrity of

the cell. Data supporting this idea were derived from experiments which

include the observation of "single hit" kinetics in hemolysis (76),

efflux of sized markers from complement-lysed resealed erythrocyte

ghosts (97,98), and C5b-9 mediated planar lipid bilayer conductances

(80-82). "Single hit" theory for immune lysis states that one

complement-induced lesion is sufficient to lyse a cell while "multi-hit"

means more than one is necessary (72). Mayer accounts for the hetero-

geneity of channel size by the addition of more than one C9 molecule to

the C5b-8 complex and not by dimer formation of C5b-9. The evidence for

this hypothesis lies in Mayer's work with C9 dose response experiments

on resealed erythrocyte ghosts containing C5b-8 sites: single hit

kinetics were observed for sucrose (0.9 nm diameter) efflux but multi-

hit kinetics were observed for inulin (3 nm diameter) efflux. This

indicates a second molecule of C9 is needed to enlarge the C5b-9 channel

Fig. 2. Hypothetical representation of the ultrastructural appearance
of the MAC.

for inulin passage. Similar dose response experiments (98) performed

with each of the other terminal components showed one-hit kinetics for

each protein indicting that only a single molecule of C5b6, C7, or C8 is

needed for an effective channel, ruling out dimer formation of C5b-9.

A second model for complement cytolysis was developed by Kinski

(62) and researchers in Lachman's laboratory (47,68) which is known as

the "leaky patch" model. These workers suggested that complement

destabilizes membranes by a detergent-like protein action. This model

was corroborated by Esser and co-workers (34) who found that C5b-9

greatly disordered lipid organization in model membranes. These results

were obtained from experiments utilizing electron spin resonance spec-

troscopy (ESR) with spin-labeled fatty acids incorporated into planar

lipid bilayers to which C5b-9 had been added. These workers envisage

the MAC as a protein "plug" which does not necessarily span the lipid

bilayer but which creates a boundary of disordered lipid at its peri-

phery. It is through these lipid/protein interfaces that exchange of

cellular and extra-cellular material occurs. Heterogeneity of channel

size can be accounted for by the association between more than one MAC

to effectively increase the lesion size between the protein complexes or

by the addition of more C9 to a C5b-9 complex.

There is some general agreement by workers from each group that

detergent-like membrane disruption may be important for bacterial and

viral cytolysis which do display multi-hit kinetics unlike erythrocytes

(32,33). The problem of how complement disrupts membrane integrity

still remains, however, including debate over whether the "doughnut"

lesion is actually the lethal protein complex. This controversy has

been further exacerbated in the last three years by the discovery made

by researchers in Podack and Tschopp's group (96,107,108) that C9 alone

in solution under certain conditions can be induced to polymerize into a

"doughnut" which is similar, ultrastructurally, to the C5b-9 generated

MAC. They also reported that polymerized C9 was capable of lysing

vesicles without other complement components present. This finding con-

tradicts the notion that C5b-8 comprises a part of the "doughnut" and

reduces the role of these proteins to only inducing polymerization of C9

on a membrane surface rather than playing a direct role in membrane

damage (except for substrates lysed by C5b-8). It is also in disagree-

ment with Mayer's analysis summarized above that only one C9 molecule is

needed, with C5b-8, to produce an effective lesion for sucrose leakage.

Another possible explanation may be that the membrane lesions which have

been associated with cytolysis for the last 20 years may not be respon-

sible for cell lysis but represent a side product of reactive lysis.

The experiments presented in this text examine the role which C9

plays in menbranolysis by three approaches. First, the conditions under

which C9 polymerizes were investigated as a follow-up to Podack and

Tschopp's initial reports. Specifically, C9 was heat-treated at various

temperatures (as per workers in Podack's research group (94,107)) or

incubated with lipid vesicles and then physical measurements were made

to determine the degree of C9 polymerization and subsequent changes in

C9 hemolytic activity as a function of these reactive conditions. The

purpose for this study was to characterize C9 polymerization by defining

the conditions and extent that polymerization occurs when C9 is heat-

treated. Two assays were used for this study: chromatography of

unheated and heated C9 to separate monomer from aggregated C9, and

measurement of C9 hemolytic activity before and after heating to


determine whether polymerization of C9 can be correlated to a loss or

gain in its ability to lyse red blood cells. Second, planar lipid

bilayers were used to study the interaction of each individual terminal

complement protein with lipid bilayers and reactive lysis by C5b-8 and

C5b-9. This study would quantitate the differences between C5b-8 and

C5b-9 mediated ion flow across a bilayer membrane and determine whether

either of these two complexes act as protein channels or simply have a

disruptive effect on a membrane so that it is more permeable. This

method would also allow measurement of the heterogeneity of C5b-9

channel sizes. Finally, proteolytic fragments of C9 which were

generated by digestion with thrombin were reacted with lipid vesicles,

erythrocytes, and planar lipid bilayers to determine whether C9 has a

membranolytic or channel-making peptide domain. The existence of a

hydrophobic, lytic peptide is consistent with the proposed amphiphilic

structure of C9 and other proteins, and exposure of such a domain in C9

may be the major role of the C5b-8 complex in effecting complement-

mediated cytolysis.

Materials and Methods

Purification of Human C9. C9 was purified from human plasma by the

method of Biesecker and Mueller-Eberhard (6) with some modifications

(23). Typically, 5 units (1 L) of freshly frozen human plasma were

thawed at 370C and made 5 mM EDTA from a 200 mM EDTA stock. CaC12 was

then added to the plasma to 40 mM, and the solution was incubated at 40C

for 30 min and then centrifuged at 16,000 x g for 10 min. The super-

natant was then precipitated by adding 21% PEG-3350 in 90 mM NaCl, 10 mM

phosphate, 0.5 mM EDTA, 0.5 mM PMSF, 0.02% NaN3, pH 7.4 to form a 7% PEG

solution. After 15 min stirring at 40C, the precipitate was removed by

centrifugation at 16,000 x g for 15 min. The supernatant was then

brought to 20% PEG by addition of solid PEG-3350, stirred for 1 hr at

40C, and centrifuged 15 min at 2,500 x g. The pellet, which contained

C9 activity, was solubilized in 400 ml of 90 mM NaCl, 10 mM phosphate,

0.5 mM PMSF, pH 7.4 buffer (conductance < 11 mmho). This clear, blue

solution was applied to a lysine Sepharose CL-6B column (5 x 16 cm)

which was in tandem with a DEAE-Sephacel (2.6 x 40 cm) column. The

columns were washed with the solubilizing buffer without PMSF at 100

ml/h flow rate until no detectable protein eluted. The DEAE-Sephacel

column was eluted at 60 ml/h flow rate with a linear NaCl gradient

formed from 1 L of elution buffer and 1 L of buffer adjusted to 30 mmho

conductance with NaCl. Fractions were tested for C9 hemolytic activity,

analyzed by NaDodSO4-PAGE, and pooled to minimize contaminating

proteins. This solution was loaded onto a hydroxyapatite column (2.6 x

12 cm) and washed with 50 mM Na/K phosphate, pH 7.8 at 40 ml/h flow rate

until no detectable protein eluted. C9 was eluted with a linear phos-

phate gradient formed from 1 L of 50 mM Na/K phosphate and 1 L of 300 mM

Na/K phosphate at pH 7.8. Fractions were again pooled by hemolytic

activity and NaDodSO4-PAGE analysis, concentrated (Micro-ProDicon

dialysis, Bio-Molecular Dynamics) into buffer containing 200 mM NaCl, 10

mM Tris, 0.02% NaN3 at pH 7.2, and sieved on a Sephacryl S-200 column

(2.6 x 75 cm) that had been equilibrated in the same buffer at a flow

rate of 15 ml/h. The C9 fractions were stored separately at -700C.

Sometimes Cohn fraction III (17), an ethanol precipitate of plasma,

was used instead of plasma as starting material. In this case, 100 g of

this solid was solubilized in 90 mM NaCl, 10 mM phosphate, 1 mM PMSF,

0.02% NaN3, pH 7.4 buffer and centrifuged to remove insoluble material.

The remainder of the C9 purification procedure with this material is the

same as that used with plasma beginning with the BaC12 precipitation.

C9 concentrations were determined using an extinction coefficient of

E280 = 0.96 ml/mg x cm (94).

NaDodSO4-PAGE. The method of sodium dodecyl sulfate polyacrylamide

gel electrophoresis employed was essentially that of Laemmli (69). A

2.5 cm "upper" or stacking gel was used that was 125 mM Tris, 4% acryla-

mide, 0.11% bisacrylamide, 0.1% NaDodS04, and pH 6.8 in composition.

The "lower" or separating gel was 13 cm in length and 375 mM Tris, 0.1%

NaDodS04, 10% or 12% acrylamide, 0.2% bisacrylamide, and pH 8.8 in

composition. The upper and lower electrode solutions were 25 mM Tris

and 200 mM glycine at pH 8.3 with the upper solution also containing

0.1% NaDodS04. Electrophoresis was performed at 120 V (50 mA) for 5

hours. Afterwards, the gel was fixed and stained in a solution of 25%

methanol, 10% acetic acid, and 0.1% Coomassie blue R-250 dye for at

least 30 min. The unbound stain was eluted from the gel with several

changes of a destain solution made from 50% methanol and 10% acetic

acid. Finally, the gel was washed with a second destain solution of 10%

methanol, 7% acetic acid, and 5% glycerol for 30 min to re-swell the gel

before it was dried and stored.

Determination of C9 hemolytic activity. This assay employs the

classical activation of complement by antibodies which have formed

immune complexes to antigens on erythrocytes (red blood cells (RBC)).

RBC with such complexes are referred to as being "sensitized" and are

called "EA". Sheep erythrocytes (SRBC) are the most common choice of

RBC substrate because of their availability and high reactivity with

human complement. The antibody used to sensitize SRBC, hemolysin, is an

IgG generated in rabbits which have been immunized with SRBC. After

formation of EA, the addition of human serum causes lysis of the cells

and release of hemoglobin. The degree of lysis is determined by centri-

fuging the remaining cells in solution to a pellet and measuring the

optical density of the supernatant at either 412 nm or 541 nm. This

absorbance is compared to the optical density of an EA standard solution

completely lysed with water to obtain percent hemolysis.

For the detection of C9 hemolytic activity human sera which was

deficient in C8 (C8D serum) was used to make EA with complement proteins

Cl to C7 on their surfaces (EAC1-7). C8D serum was made by passage of

whole human serum over an affinity column made with antibodies against

C8. EAC1-7 are stable at 40C for a period of days even after the addi-

tion of C8 (forming EAC1-8) in low quantity. EAC1-8 can then be lysed

after addition of C9 and incubation at 370C for 30 min. These cells

were used to test for C9 hemolytic activity.

The conditions employed here for hemolysis are essentially those of

Boyle et al. (13). EAC1-7 were standardized to 108 per 1.5 ml of solu-

tion to insure single hit kinetic conditions. (The optical density at

541 nm of 5 x 108 SRBC per ml is 0.188 when 100 ul of these cells are

completely lysed in 2.4 mL of water.) C8 (32.6 ng) was added to each

1.5 ml sample of EAC1-7 forming EAC1-8. C9 which was concentrated at 1

mg/ml for the experiments described in this text was diluted serially

from 1/1000 to 1/10,000 and 10 ul added to each EAC1-8 sample (10 ng to

1 ng of C9) to effect hemolysis as a function of C9 concentration. All

RBC solutions were stored in GVB which is a solution of 142 mM NaCl, 4.9

mM sodium Veronal, 0.1% gelatin, 0.15 mM CaCl2, and 1 mM MgC12 at pH


lodination of C9. C9 was iodinated using lodo-Beads (chloramine-T

derivatized polystyrene. Pierce). Usually, 0.1 to 0.2 mg of protein

was reacted with 0.5 mCi of Na125I (New England Nuclear) and two beads

for either 15 min at 250C or 30 min at 40C. Free iodide was separated

from iodinated protein by sieving the material on a Sephadex G-100

column (1 x 5 cm). The specific activity of the radioactive protein

ranged from 0.1 to 0.5 uCi/mg.

Heat Polymerization of C9. C9 purified from plasma or Cohn

Fraction III was incubated at 370C up to 64 h, at 460C for 2-3 h, and at

560C for 2 h. The protein concentration ranged from 0.2 to 1.0 mg/ml

and the buffers used were NaCl from 0.1-0.2 M, 10 mM Tris or phosphate,

and 0.02% NaN3 at pH 7.2. Control C9 samples in the same buffer and at

the same concentration were kept at 40C. After incubation the hemolytic

activity of the samples was measured as described earlier. Samples of

the incubation solutions were assayed for C9 aggregation formation and

loss of monomer by HPLC either as a time course experiment or with

comparison of C9 elution profiles before and after heating.

HPLC elution of heat-treated C9. The aggregation behavior of C9

was measured by injecting samples of C9 solution into an HPLC before

heating and as a time course during heating for experiments at 370C and

at 460C. The HPLC instrument consisted of a Waters pump (model M-45), a

Rheodyne 20 ul constant volume injection loop (model 7023), a Serva

Si200-Polyol HPLC column (4.6 x 250 mm; 3 micron beads), and an Isco

variable wavelength monitor (model 1840) with a 20 ul flow cell. C9

solution (100 ul) was injected at a time while the column was eluted

with 100 mM phosphate buffer at pH 7.0 at either 0.3 ml/min flow rates.

The absorbance was monitored at either 220 or 280 nm. The molecular

weights of the peaks were determined by comparison to protein standards.

Vesicle Formation. The methods used for small unilamellar vesicles

(SUV) and large unilamellar vesicles (LUV) are essentially those of

Dankert (22). For the formation of SUV, 4 mg of lipid dissolved in

hexane/methanol (9:1) was dried under a stream of nitrogen and then

vacuum. The lipid was then vortexed in 0.5 ml of 100 mM KC1, 10 mM

Mops, pH 7.2 and sonicated in a water bath sonicator containing 0.2%

Triton X-100 at a temperature above the phase transition for the lipid.

Sonication usually lasted up to 30 min until the solution was trans-

lucent. For freeze-thawed vesicles (LUV), this SUV solution was frozen

in an ethanol-dry ice mixture, thawed slowly to room temperature, soni-

cated briefly (10 sec), and a second cycle of freeze-thawing performed.

For experiments utilizing a marker entrapped in the vesicles, 100 mM

6-carboxyfluorescein (CF) was included in the solubilizing buffer. The

CF/vesicles were sieved over a 10 ml column of Sephadex G-100 equili-

brated in Mops/KC1 buffer.

Carboxyfluorescein experiments. Carboxyfluorescein released from

LUV was used to monitor lysis by complement. The method employed was

that of Dankert (22) with excitation of the fluorophore occurring at 480

nm and emission monitored at 520 nm using an Aminco-Bowman Spectrophoto-

fluorometer. LUV containing 100 mM CF have low fluorescence at these

wavelengths because CF fluorescence is quenched at this concentration.

When LUV become leaky the dye is released into solution to a much lower

concentration and fluoroescence increases. Total marker release was

measured by the addition of 2 ul of 10% Triton X-100 solution. The

percentage of lysed vesicles was calculated by dividing the fluorescence

measured a given time after complement addition to a vesicle sample by

the total marker release fluorescence. Generally, 2 to 5 ul of CF/

vesicles (,10 ug of lipid) were added to 250 ul of a 100 mM KC1, 10 mM

Mops, pH 7.0 solution in a cuvette. The volume of a protein addition

varied but was generally less than 10 ul.

C9 interactions with SUV: Effects on hemolytic activity. Egg

lecithin SUV were prepared by water bath sonication as described

earlier. SUV (100 ug) were incubated with 10 ug C9 (10/1, w/w; 1000/1,

mole ratio) for 30 min at 370C and 230C. Aliquots of these two incuba-

tion mixtures were withdrawn, diluted 1/1000 fold with buffer, and

tested for loss of hemolytic activity by comparison with C9 controls

which had been similarly heat-treated without SUV.

C9 interactions with SUV: 1H NMR. Egg lecithin SUV were prepared

at 12.5 mg/ml in D20. A 0.5 ml aliquot of this solution was used as a

control vesicle solution (i.e., no protein present) and a second 0.5 ml

portion was added to 235 ug of lyophilized C9 (26/1, w/w; 2700/1, mole

ratio). 1H NMR spectra were measured at 240C of these solutions in 5 mm

diameter tubes at 300 MHz using a Nicolet spectrometer equipped with the

Nicolet 1280 data processor. The spectra were Fourier transformed and

recorded after 16 accumulations using sweep widths of 3000 Hz. A pulse

width of 8.25 usec corresponding to a tip angle of 450 with a delay time

of 2.0 sec was used for each spectrum. All spectra were non-decoupled.

DSS (sodium 2,2-dimethyl-2-silapentane-5-sulfonate), deuterated in all

positions except for the reference methyl protons, was used as an

internal standard.

Preparation of a-thrombin fragments of C9. Typically, 2 my of C9

were incubated with a-thrombin (20:1 w/w) for 5-16 h at 370C in 200 mM

NaCl, 10 mM Tris, 0.02% NaN3, 1 mM CaCl2, pH 7.4 to produce C9n: a form

of C9 in which a single peptide bond has been cleaved yielding

noncovalently associated peptides of near equal masses which are

referred to as C9a and C9b. Proteolysis was ended by either storing the

sample at 40C or as a 1% NaDodSO4 solution.

Preparative NaDodSO4-PAGE was used to separate the two peptides by

employing a tube gel apparatus (BRL) (see Fig. 3). The best separation

of the peptides was achieved using a 1 cm diameter tube with a 4 cm

separating gel (12% acrylamide) operating at 100 V (12 A). Separating

gels with less acrylamide did not resolve C9a from C9b. The peptides

eluted from the bottom of the gel and were collected in 2 ml fractions.

Typically, 5 x 106 cpm of 1251 C9n was added to the C9n loaded onto the

gel in order to characterize the protein elution profile by radio-


The C9a or C9b fractions were characterized by NaDodSO4-PAGE

utilizing either Coomassie blue or silver stain (11), or autoradiography

for protein band visualization. These fractions were pooled and dia-

lysed versus 100 mM NaCl, 10 mM phosphate, 0.02% NaN3, pH 7.2, both with

and without detergent. The detergents used most often were the zwitter-

ionic detergents SB 3-12 or SB 3-14 (41) at their minimum micellar

concentrations of 0.12% and 0.012%, respectively. These detergents are

considered "mild" and were used to insure the solubilization of peptides

which may be hydrophobic. Generally, up to a week of dialysis (>12000

Da retained within the bag) at 40C with several changes of buffer (4 x

500 ml) was necessary to eliminate the NaDodSOs4 incurred during the


Planar lipid bilayer apparatus. These experiments were performed

by the method of Montal and Mueller (85) as adapted by Donovan (28)

using solvent-free membranes (see Fig. 4). The experimental apparatus

Fig. 3. Diagram of the NaDodSO4-PAGE preparative gel apparatus used to
purify C9a and C9b. A represents the elution buffer vessel; B and C are
the upper and lower electrophoretic buffer chambers, respectively; 0 is
the tube gel through which a protein sample is electrophoresed; E is a
peristaltic pump which draws eluting proteins from the bottom of the
gel; and F is a fraction collector.




cis trans

Fig. 4. Schematic representation of the planar lipid bilayer appara-
tus. Proteins were added to the cis compartment which corresponds to
the outside of a cell membrane, and electric potentials were applied to
the trans compartment which corresponds to the inside of a cell mem-
brane. The expanded region is a schematic diagram of a planar lipid
bilayer formed across a hole in the teflon partition which divides the
two compartments. Darkened circles represent the polar head groups of
the lipids, and the tails extending from the circles represent their
hydrophobic fatty acid chains which comprise the interior of the

consisted of a chamber divided into two adjacent compartments (referred

to as cis and trans) separated by a 12 um teflon partition in which a

small hole (2 x 10-4 to 5 x 10-3 cm) had been punched using a sharpened

30 gauge hypodermic needle. The needle was sharpened by electrolysis as

follows. The needle was mounted in a vertical position and connected to

the positive terminal of a 1.5 V battery. The negative terminal was

connected to a carbon electrode immersed in 6 M HC1 solution. The

needle was briefly immersed (2-5 sec) several times into the acid

solution until its tip became beveled and the needle wall was thinned

without the tip becoming ragged. The needle was inspected under a

microscope with 70x magnification to determine its suitability. Two

sizes of teflon boxes were used for these experiments: one held 4.2 ml

per chamber while the other held 2.5 ml. The most common buffer used

was 100 mM NaCl, 10 mM phosphate, 0.2% NaN3, at pH 7.0 unless otherwise

specified. Sometimes Tris was substituted for phosphate. Typically, 40

ug of lipid was added to the adjacent compartments from a 4 mg/ml stock

solution in pentane followed by 2 ul of a 1% squalene solution in pen-

tane to only the cis side. The solvents evaporated quickly forming a

monolayer of lipid at the air/solution interface. The bilayer was

formed by raising the level of the monolayer solutions to cover the

hole. Ag/AgCl electrodes were used to apply the desired voltages. The

electrode located in the trans compartment (see Fig. 4) provided either

negative or positive potentials. For all the experiments presented

here, the proteins were added to the cis compartment. In each experi-

ment voltages were applied at constant negative or positive potentials

of 40 mV or varied linearly up to 100 mV at both polarities. For ion

selectivity experiments the solutions in both compartments were 10 mM

NaC1 initially and then raised to their final concentrations of 200 mM

by addition of either 2 M KC1 or 2 M NaCl. For these measurements salt

bridges made from 2 M KC1 in a 4% acrylamide gel were used to conduct

between the solution and the electrodes. These were necessary to keep

the chloride concentration at the elecrode surface constant.

Protein additions to the bilayer. Complement proteins C5b-6, C7,

C8, and C9 were added to an asolectin bilayer by the sequential addition

of about 5 ug of each purified protein into the cis compartment at 15

min intervals. Asolectin purified from soybean (58) is a mixture of

lipids consisting of about 80% phosphatidyl choline and phosphatidyl

ethanolamine, 15% phosphatidyl inositol, 2% phosphatidyl serine, and 2%

cardiolipin (83). The last three lipids have negatively charged head-

groups which give asolectin membranes a negative surface charge.

In a second series of experiments, C5b-9 was added to bilayers as

preformed MAC associated with vesicles. MACs that had been purified

from complement-lysed erythrocyte membranes by detergent extraction (4)

were incorporated into liposomes by J.R. Dankert using a modification of

the method reported by Mimms et al. (84) for LUV formation by detergent

dilution. Briefly, 40 mg of egg lecithin was solubilized in buffer

containing 300 mg of deoxycholate (1:7.5 w/w) with 2 mg of preformed

MAC. This solution was chromatographed by gel filtration on Sephadex

G-100. The vesicle-containing fractions showed a large number of MACs

associated with the liposome surfaces as visualized by electron micro-

scopy (Dankert, J.R., unpublished observations). Twenty microliters of

this protein/vesicle solution was added to the cis compartment of the

bilayer box, which already had an asolectin monolayer, by slowly

dripping the vesicles down the side of a glass rod onto the solution

surface. This causes up to 10% of the vesicles to form a monolayer

containing lipid-solubilized protein, as described by Verger and Pattus

(112). The asolectin was present because it is difficult to form a

bilayer from pure egg lecithin. Thus, on the trans side of the bilayer

only asolectin was present, while on the cis side asolectin was mixed

with egg lecithin and MAC.

C9a, C9b, and C9 were added to the cis compartment directly from

solutions of purified protein after bilayer formation for the experi-

ments studying this protein and its peptides alone with bilayers.



In 1981 Tschopp and Podack reported that C9 is capable of self-

association, in the absence of other complement proteins, into a tubular

ring which appears ultrastructurally identical to membrane lesions

caused by complement (108). Polymerization of C9 was induced by heating

C9 solutions at either 460C for two hours or 370C for 3 days. In subse-

quent reports (93-96,106,107) they extended their observations with data

showing that polymerized C9 lost its hemolytic activity. Heating C9 in

the presence of egg lecithin SUV containing carboxyfluorescein led to

substantial marker release indicating membranolysis. SUV showed a large

degree of association with C9 even in the presence of 2 M NaC1. This

group of workers concluded from these data that C9 is primarily respon-

sible for complement-mediated lysis, that lysis occurs as the result of

the polymerization of C9 in the presence of a membrane, and that the

role of the other terminal components of complement (C5b-8) is to induce

polymerization of C9 at a membrane surface. A possible mechanism for

C5b-8 polymerization of C9 suggested that this protein complex lowered

the energy of activation for polymerization. In the absence of these

proteins C9 polymerization could still be achieved at increased tempera-


Tschopp and Podack's work is of great interest to investigators

studying the mechanism of complement lysis. First, it has led to a

reexamination and subsequent revision of the existing hypotheses about

the molecular composition of the MAC. Prior to this work, it was

thought by most researchers that the "doughnut" was composed of all the

terminal complement proteins rather than composed mainly of C9 as was

suggested by Tschopp and Podack. Furthermore, Tschopp and Podack's

results imply almost total responsibility for C9 in complement-mediated

lysis rather than C9 simply enlarging and stabilizing a lesion already

begun with C5b-6 attachment to a membrane surface.

Several questions arise about Tschopp and Podack's work upon exami-

nation of their data. The lipid/protein molar ratios they used for SUV

lysis experiments are about 8:1. At these concentrations nonspecific

lipid/protein interactions could be responsible for membrane damage.

There are about 5-6000 lipids per SUV which means there were as many as

500 C9s per SUV in these experiments. Tschopp and Podack also quanti-

tated C9 binding to SUV by incubating the two and the centrifuging the

mixture on a cushion of sucrose. The low-density SUV floated on the

sucrose with bound C9 while unbound C9 became separated and penetrated

further into the sucrose solution of greater density. These experiments

were conducted with radiolabeled C9 and protein/lipid molar ratios of

about 6:1, or about 50 times more protein than was used in the SUV lysis

experiments. This increases the likelihood that these protein/lipid

interactions are nonspecific.

It is apparent that further studies on lipid-induced and heat-

induced polymerization of C9 are necessary. The prominence of the

"doughnut" in the models for complement lysis requires that the condi-

tions which promote C9 polymerization into the "doughnut" be carefully

defined and quantitated. The experiments presented in this text began

as an attempt to reproduce Tschopp and Podack's polymerization of C9 by

heat-treatment of the protein at 46C. In these experiments no loss of

hemolytic activity of C9 was measured and only few poly-C9 "doughnuts"

were observed when heated C9 solutions were examined with an electron

microscope (electron microscopy of all samples was performed by J.R.

Dankert and A.F. Esser). These preliminary results increased the impor-

tance of studying C9 polymerization carefully in order to determine the

reasons for the apparent discrepancies in our results. This study

begins with a careful purification of C9 to produce the cleanest C9

possible, followed by heat-treatment of C9 solutions at 370C, 46C, and

560C. Experiments by other workers in this laboratory (see Dankert et

al. (23,24)) examined the interaction of C9 with SUV and LUV. Two brief

experiments which complement Dankert's work are included in this text:

The effect of SUV on the hemolytic activity of C9, and proton NMR

spectra of SUV both with and without C9 additions.


Purification of C9. C9 was purified as described in the methods

section of this text. In particular care was taken to minimize proteo-

lytic degradation of C9 by adding protease inhibitors to all buffers

until elution from the DEAE-Sephacel column while the BaC12 precipita-

tion removed vitamin K-dependent enzymes. A more shallow phosphate

gradient was used for the hydroxyapatite column than was used by

Biesecker and Mueller-Eberhard (6) and a gel filtration step was

included afterwards which helped remove higher molecular weight contami-

nants from C9 which had been difficult to separate without these modifi-

cations. It was also important to use NaDodSO4-PAGE in addition to C9

hemolytic activity for pooling eluted C9 fractions after each

chromatographic step. This helped to minimize the inclusion of protein

impurities to a C9 pool which did not separate well from C9 with subse-

quent chromatography.

For these preparations either freshly-frozen plasma or Cohn

Fraction III was used as a source of C9. NaDodSO4-PAGE of the purified

C9 obtained from Cohn always showed minor quantities (< 1%) of proteo-

lytic fragments of C9. Purified plasma C9, on the other hand, did not

contain these fragments and consistently produced unproteolysed, pure C9

as shown by Coomassie-stained NaDodSO4-PAGE gels (see Fig. 5). For this

reason plasma C9 was preferred for studying C9 polymerization although

Cohn C9 was occasionally used as well.

Figure 6 shows that the hemolytic activities of C9 obtained from

plasma and Cohn Fraction III were identical when tested on EAC1-8 and

showed no loss in activity compared to unpurified C9 in whole human


Heat Polymerization of C9. Figure 7 shows HPLC elution profiles

following the time course incubation of C9 at 460C over a three hour

period. The aggregated protein eluted first in the excluded volume of

the column indicating a molecular weight > 660,000 Da according to the

molecular weight standardization profile for this column (Fig. 8). The

second peak eluted in a volume expected for monomeric C9 (71,000 Da).

The aggregated protein peak increased over the course of the experiment

while the monomer C9 peak decreased as shown in Fig. 9 with 16% loss of

monomer occurring in three hours incubation. Fractions collected over

the elution profile were tested for hemolytic activity with EAC1-8 and

showed that only the monomeric C9 peak contained activity. The aggre-

gated C9 fractions never showed any activity. An additional elution


1234567 89

; *4. ,-* -'-__ .
'* -s -..- 'i.. .".. ....,-- -. ....... ..... ",
,, '; .. ; L ,:' ' ,

09 f .. 0. C reduced

dye front


dye front C9b

Fig. 5. Sodium dodecyl sulfate gel electrophoresis of C9 purified from
human plasma. The first five lanes are from consecutive fractions of
protein eluting from a Sephacryl S-200 sieving column. Lanes 7-9 are
the same as lanes 2-4 but electrophoresed under reduced conditions with
dithiothreitol. At the bottom is a densitometer scan of unreduced C9
from lane 3.




0.2 0.5 1.0 2.0

5.0 10.0

50 100

PROTEIN (nmoles)

Fig. 6. Comparison of the hemolytic activities of C9 purified from
human plasma and Cohn fraction III with unpurified C9 were calculated
using an extinction coefficient at 280 nm of 0.96 ml/mg/cm, and the
concentration used for unpurified C9 in whole human serum was 60 ug/ml
(see Table I). "Z" values represent the extent of hemolysis for each
sample and is calculated from the following relationship: Z = -In
(fraction of cells lysed).

C9, p




- 00


O 46C

0 mL

Fig. 7. HPLC elution profiles of C9 polymerization. 100 ul of the C9
incubation solution (0.21 mg/ml in 100 mM Tris, 200 mM NaC1, pH 7.2) was
injected into the HPLC using a Rheodyne 20 ul constant volume loop
injector. The proteins were eluted with 100 mM phosphate buffer at pH
7.0 at 0.3 ml/min.

1 x106
poly C9 thyroglobulin


MW aldolase



1x104 .-
1.0 2.0 3.0


Fig. 8. Monomer and polymer C9 elution from HPLC relative to molecular
weight markers. All proteins were chromatographed as described in Fig.







Fig. 9. Time course of the loss of monomer C9 and the increase in
polymer 9 at 460C calculated from HPLC elution profiles.

I I I I I I I I I ~ I


poly C9

I I I..........

profile in Fig. 7 shows the degree of C9 polymerization after 72 h incu-

bation at 370C. The loss of monomeric C9 was less than 10% with the C9

hemolytic activity again occurring only with the monomer peak.

The effect of heating on the hemolytic activity of C9 was also

measured without separating the aggregated from monomeric C9, i.e., on

the incubation "pot". After heating C9 for 3 h at 46C, 3 h at 56C, or

72 h at 370 aliquots were withdrawn from each incubation solution and

tested for hemolytic activity for a series of protein concentrations

ranging from 1 to 10 ng of C9 per sample of EAC1-8. Figures 10 and 11

show that C9 heated at 370C and 460C resulted in < 10% loss of hemolytic

activity at all dilutions as compared to C9 which was stored at 40C. C9

which was heated at 560C lost all it hemolytic activity and subsequent

experiments showed that all activity was lost within 30 min at this

temperature indicating, perhaps, that at some critical temperature

between 460C and 560C C9 undergoes irreversible heat denturation.

Ultrastructural examination (23,24,103) of the different incubation

mixtures using electron microscopy revealed very few "doughnut"

formations resulting from the above heating experiments at any of the

three temperatures which indicates that heat polymerization of C9 into

"doughnuts" occurs only to a small extent under these conditions (data

not shown). These experiments were performed using C9 which was puri-

fied from both plasma and Cohn Fraction III. No differences were

evident in the response of C9 from either source to the heat treatments

described above. Varying either the concentration of C9 from 0.2-1.0

mg/ml or the NaCl concentration from 0.1-0.2 M in 10 mM Tris or phos-

phate buffers did not have any effect upon these experiments.




,, 40


.: : . ...

10.0 5.0 2.5 1.7 1.25

Fig10. The loss of hemolytic activity of C9 incubated 72 h at 370C as
compared with unheated C9 stored at 4C for the duration of the experi-
ment. The open bars represent the percent hemolysis obtained with
unheated C9 concentrated at 1 mg/ml for both samples, and then tested
for hemolytic activity at 5 different dilutions. The unheated and
heated C9 henolytic activities are depicted by the open bars and cross-
hatched bars, respectively.






10.0 5.0 2.5 1.7 1.25


Fig. 11. The loss of hemolytic activity of C9 incubated 3 h at 460C as
compared to unheated C9 stored at 40C for the duration of the experi-
ment. As in Fig. 10, C9 was concentrated at 1 mg/mi for both samples,
and then tested for hemolytic activity at 5 different dilutions. The
unheated and heated C9 hemolytic activities are depicted by the open and
cross-hatched bars, respectively.

C9 interactions with SUV. Figure 12 shows the effect on C9 hemo-

lytic activity by incubating C9 with egg lecithin SUV at 1000/1 mole

ratio of lipid to protein. No loss in hemolytic activities was observed

after 30 min at either 230C or 37C compared with C9 samples incubated

in the absence of lipid at similar temperatures or 40C.

Proton NMR of egg lecithin SUV in the presence and absence of C9 at

2000/1 lipid to protein mole ratio show no differences in the linewidths

of resonances corresponding to the fatty acid methylene protons at 1.25

and 0.85 ppm, respectively, or the headgroup choline methyl protons

(N-CH3) at 3.2 ppm (37) (Fig. 13). Restriction of the mobility of a

molecule, or a portion of the molecule, results in the broadening of the

linewidth of the resonance corresponding to the restricted region, if

that region contains detectable nuclei (8,9). Proton NMR allows the

observation of almost all regions of the lipid molecules which comprise

the SUV because of the ubiquity of hydrogen in lipids. Proteins which

bind strongly to the outer vesicle surface could restrict the motion of

the choline methyls, while proteins which insert into the fatty acid

regions of the bilayer would alter fatty acyl chain order, intermolec-

ular associations of the lipids, and rotations of a fatty acid about the

axis of its carbon backbone (8,9,74). The absence of such effects

observed in the spectra presented in Fig. 13 implies that C9 does not

strongly interact with lipid head groups at the surface of egg lecithin

SUV and that C9 does not penetrate into the fatty acid core of the

bilayer at these concentrations of lipid and C9. These results are

inconsistent with experiments (23,24,103) which show that C9 binds to

SUV strongly enough to survive gel chromatography. No resonances from

C9 protons were visible in spectra of C9 solutions (0.47 mg/ml) without


80 250C

> 60
w 40


C9 C9/SUV C9 C9/SUV C8

Fig. 12. The hemolytic activity of C9 after treatment with egg lecithin
SUV at 250C and 37C. C9 (10 ug) was added to 100 ug of freshly soni-
cated egg lecithin SUV and incubated 30 min at each temperature while
control C9 samples were similarly diluted into buffer without SUV and
incubated at each temperature. The hemolytic activity of each sample
was then assayed. Open bars represent C9 hemolytic activity without SUV
treatment and cross-hatched bars represent C9 hemolytic activity after
SUV treatment. The solid bar shows background lysis of cells without C9
additions (C8 lysis).


. AV, (Hz)

EL/C9 10.6 36.9 30.0

EL 10.3 36.0 29.7


3 2
3 2


1 0 PPM

Fig. 13. Proton NMR of egg lecithin SUV with and without addition of
C9. The spectra were recorded at 300 MfHz at 24C in b mn diameter
sample tubes. Spectrum A represents a sample of 6.25 ig egg lecithin
SUV with 0.235 mg of C9 (26/1, w/w; 2700/1, mole ratio) dissolved in 0.5
ml of 020 containing 100 mMi NaCl, 10 mMl phosphate, 1 mM EDTA at pH 7.0.
Spectrum B is the same as A but without C9. Both spectra were obtained
using a pulse width of 8.25 usec (45'C), a delay time of 2.0 sec, with-
out decoupling using sweep widths of 3UO Hz. Spectrum A was Fourier
transformed and recorded after 16 acquisitions while spectrum 3 was
recorded after 10U acquisitions giving A somewhat more noise. PI, P2,
and P3 refer to N-metnyl choline, fatty acid methylene, and fatty acid
terminal methyl protons, respectively. V, is the linewidth at half
peak height. 1.0 ppm is equivalent to 3UU Hz.


lipid present after up to 100 acquisitions which means that only lipid

protons were observed in the spectra shown in Fig. 13.

These experiments were intended to accomplish two basic objectives:

first, to gain an understanding of the conditions in which C9 self-

associates into a tubular polymer which may be synonomous with the

classical complement lesion; and, second, to observe whether C9 associ-

ates with liposomes, and, if so, to describe the nature of this

protein/lipid interaction in the context of the terminal complement

proteins' destabilizing effect upon membranes. As explained in the

introduction to this text, this work was intended to verify and to quan-

titate the observations initially made by Tschopp and associates

(107,108) concerning C9 polymerization.

Several modifications were made to the standard C9 purification

method as was published by Biesecker (6) resulting from suggestions by

A.F. Esser and from personal observations made over the course of

several C9 preparations. The first of these was to include a BaC12

precipitation of plasma by making it a 40 mM solution of this salt from

a 1 M stock prior to the usual PEG precipitations (Esser, private

communication). This precipitation removes vitamin K-associated

enzymes. A second procedural change involved narrowing the phosphate

gradient from 80 to 400 mM as was used by Biesecker to one employing a

40 mM to 300 mM linear gradient. A larger volume gradient was used as

well, increasing the total gradient volume from 150 ml to 2 L. This

effected a slower, more gentle gradient which allowed better resolution

of the C9 from contaminating protein which elutes prior to C9 in the

gradient. Experiments to optimize C9 purification also via

hydroxyapatite chromatography by Esser and Ervin (unpublished obser-

vations) showed that C9 is best purified at pH > 8.0. Based upon this

observation, pH 8.0 was used for my phosphate gradient rather than pH

7.7 as used by Biesecker but not higher because C9 is known to lose

hemolytic activity at pH > 8.0

Finally, a gel filtration chromatographic step was added at the

end of the procedure employing a Sephacryl S-200 column which removes

any high molecular weight species, including any endogenous poly C9,

prior to protein storage. These modifications afforded very pure

protein for our work. It was preferred to use C9 for the polymerization

experiments just after sieving as freezing and thawing C9 solutions were

observed to produce some polymerized C9. C9 could be stored at -20C in

50% glycerol solution without loss in hemolytic activity and could be

dialyzed versus a desired buffer before use. No difference in hemolytic

activity was observed with C9 obtained from either Cohn Fraction III or

plasma although Cohn C9 always contained some proteolyzed C9. This

verification helped to resolve the question of whether discrepancies in

our results from that of other groups resulted from the source of C9.

No differences in the ability of C9 from Cohn or plasma to polymerize by

either heating, vesicle interaction, or reaction with MC5b-8 have been

observed to date.

Heat treatment of C9 under the conditions described in this text

revealed that the aggregation of C9 occurs only to a small extent under

these conditions (< 15%). Purer C9 appeared to aggregate and lose hemo-

lytic activity to lesser extents than C9 which showed some protein

contamination with NaDodSO4-PAGE. This could be the basis, at least in

part, for the discrepancies in these results compared to those of

Tschopp and Podack (107). Electron microscopy of these heated samples

show two types of this polymerization exists: the tubular, doughnut

poly C9 which resembles the MAC; and "string-like" aggregates (Dankert

et al. (23)). Separation of these aggregated forms of C9 from monomeric

C9 by HPLC or gel filtration revealed that only monomeric C9 retained

hemolytic activity. Because of the dramatic loss in hemolytic activity

which occurs when increasing the incubation temperature of C9 from 460C

to 560C, it is probable that an irreversible thermal denaturation occurs

in this temperature range.

Experiments by Dankert et al. (23) have shown that aggregated C9

strongly scatters light rendering optical density measurements at either

220 or 280 nm, as was used in this study to monitor protein chromato-

graphic elution, unsuited for the quantitation of the degree of poly-

merization. The loss of monomer peak measured at these wavelengths is

the better method to determine the extent of the reaction. The HPLC

column used in this work was standardized with protein molecular weight

markers and was shown to almost exclude thyroglobulin with this column

showing that these forms of C9 contain at least 10 molecules. However,

this represents only the minimum size of these aggregates as electron

microscopy has shown (23) that not only are there "strings" and "rings"

of C9, but that these also associate with each other forming super-

molecular complexes. Some of these complexes are not NaDodSO4 disso-

ciable as shown by NaDodSO4-PAGE and, in fact, barely penetrate even

2.5% acrylamide gels (data not shown), confirming the strength of the

molecular associations and their apparent large size.

Incubation of C9 with egg lecithin SUV at either 240C or 370C at

1000:1 lipid/protein molar ratios did not cause loss of hemolytic

activity of C9. Observations reported (23,24,103) showed that C9 asso-

ciates with SUV made from egg lecithin or DPPC and is polymerized, up to

30% extent, to large aggregates which appear to be composed df associ-

ated "doughnuts" and "strings" with the remainder of C9 in an unaggre-

gated association with SUV. Most of the vesicle population eluted from

a Sepharose 2B column as SUV (with associated C9) but some of the

vesicles eluted in the column's void volume with aggregated C9. Inter-

estingly, vesicles from both fractions still formed membrane potentials

with valinomycin which indicated that despite the presence of aggregated

or unaggregated C9, little or no membrane damage had occurred to most of

the vesicle population. One explanation for why C9 did not lose hemo-

lytic activity after vesicle treatment might be that only a subpopu-

lation of C9 is hemolytically active and that these C9s do not polymer-

ize (thus losing activity) under these conditions. Hemolysis experi-

ments in this laboratory have shown that at least 100 C9 per erythrocyte

are needed for lysis, i.e., the most active C9 preparations have, at

best, only 1% hemolytically active molecules. A second explanation may

be the size variation of vesicles produced by sonication. Dankert et

al. (23) demonstrated that LUV do not interact at all with C9 and no C9

polymerization is produced with these vesicles. Thus, there may be

vesicle size dependence for C9 interaction. SUV have much greater

curvature than LUV. The SUV I prepared by sonication to test their

effect on C9 hemolytic activity were not characterized by gel chromato-

graphy to determine whether they were of SUV size. However, the

sonicated lipid solutions were almost completely translucent which is

indicative of small vesicle size. Also, Dankert did not include tests

of C9 hemolytic activity after treatment with SUV either before or after

gel chromatography. Future experiments should include both sizing of

SUV and C9 by gel chromatography and tests of C9 hemolytic activity

after SUV treatment in order to resolve these data.

Proton NMR spectra of similar solutions in deuterated water showed

no differences in spectra of SUV or SUV with C9. Proton NMR of SUV with

C9 show no line shifts or line-broadening for any of the prominent lipid

resonances. C9 associated with SUV probably does not penetrate the

hydrophobic inner core of the bilayers. This result is in agreement

with Dankert's report that SUV associated with C9 are still able to form

membrane potentials with valinomycin and do not leak carboxyfluorescein

entrapped within the SUV. However, C9 bound to an SUV would be expected

to immobilize choline methyl protons at the outer vesicle surface. The

ratio of lipid to protein molecules used in this experiment (2700/1)

would give two C9s per SUV. According to Dankert's experiments only 30%

of the C9 present may be bound to SUV giving less than one C9 per SUV.

Because 40% of the choline methyls from the phosphatidyl choline head-

groups are located in the inner bilayer of the vesicles these methyls

are never exposed to bound C9. Only the outer choline methyls can asso-

ciate with C9 and with only one C9 per vesicle insufficient numbers of

immobilized choline methyl protons may be present to cause a difference

in the proton NMR spectra. Again, this experiment would be improved

upon by characterizing the SUV with gel chromatography and by using only

the sieved SUV with bound C9 for the NMR experiments.

In summary, these experiments show that C9 aggregates and loses

hemolytic activity to only a small extent upon heating at 370C for 3

days or 46C for 3 hours; aggregated C9 has lost hemolytic activity;

aggregated C9 has a molecular weight greater than 660,000 Da indicating


at least 10 C9/protein complex; heating C9 at temperatures greater than

560C completely inactivates C9 within a few minutes without "doughnut"

formation; and, C9 does not lose hemolytic activity after treatment with

egg lecithin SUV and does not cause perturbations to proton NMR spectra

of SUV after addition of the protein.



Planar lipid bilayers provide a unique opportunity to measure the

functional activity of a single molecule. The basic premise of this

experiment is that a single lipid bilayer can be formed across an ori-

fice separating two compartments filled with buffer solution (see Fig.

4). Because a lipid bilayer has a hydrophobic interior composed of

fatty acid chains, almost no current (ion flow across the bilayer) flows

between the two compartments when an electric potential is formed

between the compartments. When molecules are introduced into a compart-

ment after a membrane has been formed, currents result if the molecules

interact with the membrane in a way that allow ion flow across the bi-

layer. For molecules which do mediate current, a current is produced

for each molecule (or molecular complex) interacting with the bilayer,

usually in the form of discrete increases in current or channels. More

than one such channel occurring simultaneously is seen as stepwise

increases in current each of which has the same magnitude. Some of the

information which can be obtained about such channels is whether the

channel shows ionic selectivity, pH dependence, voltage dependence,

"gating", i.e., whether channels only conduct at some minimum potential

or close at some greater potential, lipid specificity, and an estimation

of the pore size.

Several variations on this kind of apparatus for membrane current

measurement are in use in different laboratories. The apparatus used

for the work presented in this text is based on the method of Montal and

Mueller (85) which employs "solventless" lipids to give a membrane of

known composition. Bilayers formed by this method do have the hydro-

carbon squalene present but this molecule is considered not to be

present within a bilayer but rather serves to anchor the bilayer at the

teflon edge. A second bilayer method is referred to as a black lipid

membrane or BLM (39). This membrane is formed by painting a lipid

solution containing non-volatile solvents such as decane or undecane

across an orifice which also separates two compartments. The membrane

subsequently thins to form a non-conducting planar bilayer composed of

both lipid and solvent within the hydrophobic matrix. This technique

forms bilayers which have uncertain composition and is farther removed

from a cell membrane than planar membranes without solvents due to the

effects of the solvents on the fatty acid mobilities, packing, order

parameters, etc. A third method of measuring conductance across

membranes utilizes a patch clamp electrode on the surface of a cell thus

measuring an in vivo current (also referred to as voltage clamping)

(87). This technique is restricted to certain cell types such as muscle

tissue and nerve axons. Disadvantages of this technique include the

presence of endogenous ion channels within the membrane and the

inability to study compositionally well-defined systems. One of the

objectives of our experiments was to use membranes without substances

which do not occur in real cell membranes such as hydrocarbon solvents.

Another goal was for only the proteins we wished to study for channel-

making activity to be present with a membrane. These conditions would

allow a study of only the complement proteins' effects on a bilayer

membrane's permeability to ions. For these reasons, solventless planar

lipid bilayers were selected over BLM and voltage-clamping techniques.

The kinds of molecule/membrane interactions of most interest here

are those which produce small currents (> 10-12 amp) by forming trans-

membrane channels or by acting as ion carriers across the bilayer rather

than those which produce large currents (> 10-3 amp) due to general

membrane damage, i.e., breaking the membrane (a property usually associ-

ated with complement-mediated lysis). Some of the types of molecules

which have been studied with bilayers are antibiotics such as gramicidin

and alamethicin (ion channels), valinomycin (an ion carrier), bacterial

toxins, anesthetics, melittin (a protein found in bee venom), and sodium

channel proteins (44,48,105). It is apparent that many different

species have been studied with bilayer membranes, including complement.

The first experiments of complement interactions with planar lipid

bilayers were performed by del Castillo et al. (26) and Barfort et al.

(1) in 1966 and 1968, respectively. These experiments were conducted

using antigen and antibody additions to one side of a membrane (or even

with antigen on one side with antibody on the other) with subsequent

addition of serum as the source of complement. This produced "small

holes" in the membrane apparently due to the attack of complement.

Wobschall and McKeon (117) used antibody/antigen activation of rabbit

and guinea pig sera on bilayers to observe complement-mediated conduc-

tances. They also placed antigen and antibody on opposite sides of the

bilayer, noting that membrane stability was greater in that case.

Heating serum at 560C for one hour caused loss of activity; a result

corroborated by other investigators who noted heat-inactivation of

complement hemolytic activity (57). Low concentrations of antigen/

antibody/complement were necessary to prevent membrane breakage.

Greater noise accompanied the onset of conductance which occurred as

non-uniform steps. They estimated a 22 A pore produced by complement in

a membrane composed of sphingomyelin/cholesterol/tocopherol. They

suggested this technique as a method to rapidly determine antibody or

antigen concentrations.

In 1981 Jackson et al. (53) used an extracellular patch electrode

on rat skeletal muscle to study complement-mediated conductances across

a cell membrane. The muscle tissue was first derivatized with TNP

(trinitrophenol) and then exposed to pure anti-TNP antibody. The elec-

trode was filled with rabbit serum diluted 1:3 with saline solution as a

source of complement. They observed single channel conductances of

about 90 pS occurring on a time scale of tens to hundreds of msec along

with rarer 40 pS channels with a mean lifetime of 2 msec. A pore dia-

meter of 8 R was estimated by these workers based upon the larger of the

two channels. These conductances showed no dependence on applied vol-

tage (a linear change in conductance as the voltage is changed linearly)

up to voltages of 60 mV of either positive or negative polarity when the

conductances plateaued upon further voltage increase. In addition to

these data they noted that occasionally larger conductances were

observed (although no data describing them were offered); that more

noise occurred when channels were open; that up to 30 min usually passed

after the experiment was begun before conductances began; and that

channels could close permanently (no more conductance observed). As

controls, they used membranes which had not been derivatized with TNP,

or left out the antibody, or both. However, in each of these cases they

still observed the same channels although with somewhat less activity.

They suggested these results meant that either antibody-independent

complement activation had occurred or that heterophil antibodies was

present in the rabbit serum to the rat membrane proteins. The only case

in which channels were not observed was after heat treatment of the

serum at 560C for 30 min.

The most extensive planar lipid bilayer studies with complement

were conducted by researchers in Mayer's laboratory (79-82). This work

investigated reactive lysis by C5b-8 and C5b-9 with solvent-containing

(decane/hexadecane) black lipid membranes (BLM) using either lecithin,

oxidized cholesterol, or glycerol monoolein as lipid. The complement

proteins were either purified from guinea pig serum or were functionally

(but not biochemically) pure human components. The results they

reported showed dependence upon the type of membrane used. With

lecithin no conductance occurred during sequential additions of C5b-6,

C7, C8 and C9 until C8 was added. No currents were observed when each

protein was added to the bilayer alone. The C5b-8 conductances opened

and closed continuously and showed voltage dependence (the current was

about three times greater at positive compared with negative poten-

tials). When C9 was added to a C5b-8 containing bilayer much greater

conductances of varying size resulted which showed voltage indepen-

dence. Prior to the addition of C9, C5b-8 channels opened and closed

continuously but after addition of C9 the channels remained open. They

also noted that if the terminal complement components were mixed before

addition to the bilayer no conductances occurred suggesting that an

inactive complex formed in the absence of lipid. Reaction between C5b-6

with C7 and C5b-8 with C9 occurred only when added on the same side of

the bilayer while C5b-7 could react with C8 from the opposite side of

the membrane. Thinner membranes (35-45 A) made from oxidized choles-

terol produced conductances with each addition of complement starting

with C5b-6.

From the above data Mayer and his coworkers estimated the size of

the channels with sequential addition of the proteins to C5b-6 (6 A),

C5b-7 (12 R), C5b-8 (16 R), and C5b-9 (25 R). They suggest that the

C5b-9 channel may be as large as 40 R. Although they report that C5b-9

conductance occurred as discrete jumps which are characteristic of

proteins that make transmembrane channels, no figures show single

channels in any of Mayer's papers. The only figures they have shown are

plots of time-dependent conductances with points reported at two minute

intervals. Without this type of data, their reports of various channel

sizes for the stepwise formation of C5b-9 cannot be verified. In addi-

tion to these omissions, no information is given regarding the sensi-

tivity of their measurements to smaller, fast channel formations (with

second or millisecond lifetimes). Also, the addition of proteins is

described as "units" of activity rather than by mass without information

given to reconcile the two conventions. Finally, guinea pig and human

complement were used for these investigations without information given

as to which was used for a given experiment. While cross-reactivity

exists between complement of the two species differences in reactivity

do exist between them and the molecular details of these differences

have not been examined.

The most recent work was performed by O'Boyle et al. (89). They

used a novel method of bilayer formation on Millipore and Nucleopore

filters. The pore size of these filter is about 0.24 um and the

bilayers are supported across these uniform holes. Bilayer membranes

were formed with lipids extracted from sheep erythrocytes. Antibodies

to these lipids were added to the formed membranes and guinea pig serum

added to the solution for complement activation by the classical path-

way. They observed conductance increases which did not occur in control

experiments in the absence of antigen or antibody or when heat-inacti-

vated serum was used. The conductances reported varied widely from one

experiment to another even under conditions which were identical. The

current-voltage curves of the currents showed voltage independence and

they observed channel formations. These channels opened in a stepwise

fashion without closing with apparent variability in channel size. From

this data they report a channel diameter of 112 R. Criticisms of this

work include the use of serum as the source of complement and these

workers' choice of using filters to form bilayers. This raises the

question of whether the formation of a channel of the size they report

(112 R) is distinguishable from the spontaneous breaking of bilayers

across some of the filter holes: the membrane pores are only about 20

times larger. This point is particularly important because their data

depict macroscopic conductances rather than single channels.

There exists both a relative wealth and poverty of information

contained in the work described by the above workers in addition to

large amounts of contradiction and ambiguity. Each group which has

published on complement/lipid bilayer interactions has observed either

conductances or single channels under conditions similar to those known

to activate the terminal complement proteins for cytolysis, i.e., by

either the classical means employing membrane antigen/antibody inter-

actions with serum or by reactive lysis using pure or functionally pure

terminal components.

Except for Mayer, each of the other groups used serum as their

source for complement and activation with antibodies in either patch

clamp electrode experiments or BLM. The use of serum does not allow

differentiation between C5b-8 and C5b-9 conductances and clearly does

not allow one any certainty that complement-mediated conductances are

actually being observed rather than currents produced by other species

in serum. This possibility may be realized by the inability of each of

these groups to have clear-cut control experiments. Either channel-

making activity continued when no antibody or antigen was present

(conditions in which complement should not be activated) or they

observed conductances under unlikely conditions such as when antibody

and antigen or serum were on different sides of the bilayer. The only

control which gave no currents for all of these experiments was when the

serum was heat-treated at 560C which may inactivate other species

present besides complement proteins.

Presented here are the results of experiments on complement's

interaction with solventless planar lipid bilayers. The source of

complement was purified human terminal components C5b-6, C7, C8, and C9

added sequentially to an asolectin bilayer. While C8 and C9 were very

pure as analyzed by NaDodSO4-PAGE, the C5b-6 and C7 used for these

experiments were less well characterized. These proteins appeared free

of C8 and C9 functional activities, however, over the course of the

planar bilayer experiments. An additional experiment utilized isolated

MAC which had been incorporated into lipid vesicles and then spread into

a monolayer for bilayer formation. This result is included because of

the similarity of the results obtained with C5b-9 added sequentially to

the bilayer, and because of the novelty of the method. Voltage

dependence for each conductance experiment are demonstrated as well as

single voltage conductance values at both negative and positive polar-

ity. From these data hypotheses are offered which attempt to describe

how complement-mediated membrane channels could destabilize a cell



Release of carboxyfluorescein from LUV by reactive lysis. LUV

composed of asolectin, egg lecithin, or egg lecithin/cholesterol were

tested for release of entrapped CF after treatment with C5b-6, C7, C8,

and C9 added sequentially (see Fig. 14). Similar experiments have been

performed previously by other workers (52, 102) with the possible excep-

tion of asolectin vesicles. They are included here for comparison with

the planar lipid bilayer experiments described below. Figure 14 shows

that minor marker release occurred after addition of C5b-6 and C7 to

asolectin LUV while little or no release occurred with the other lipid

compositions. Most of the CF was released from the LUV after C8 addi-

tion to the C5b-7/LUV. Subsequent addition of C9 caused further CF

release for egg lecithin/cholesterol vesicles but not for asolectin or

egg lecithin vesicles. Not all of any of the three types of vesicles

were lysed at this point as is indicated by the addition of Triton X-100

to achieve full marker release. (Shown at position E of Fig. 14.) In

other experiments (data not shown) with asolectin and egg lecithin LUV,

decreasing the amount of C8 from 1 ug to 0.2 ug before addition of C9

reduced C5b-8 induced lysis but did not yield additional CF release upon

addition of C9. The reasons for the dependence on cholesterol for C5b-9

mediated marker release from LUV are unclear but probably means that the

more fluid vesicles (without cholesterol) lyse with only C5b-8 present

5 min.
* |

n E

a /

UJ asolectin
< B
uJ D E

o A B

EL/chol C E
C5b-6 C7 C8


Fig. 14. Release of carboxyfluorescein from LUV mediated by complement
reactive lysis. LUV composed of asolectin, egg lecithin, or egg
lecithin/cholesterol were treated with the sequential addition of 1 ug
each of C5b-6, C7, C8and C9enoted by A,B,C, and D, respectively. The
increase in relative fluorescence is due to release of Cf from the LUV
and is indicative of membrane damage. Total marker release was achieved
by addition of 2 ul of 10% Triton X-100 as denoted by E.

while less fluid vesicles (containing cholesterol) require C5b-9 for

lysis. Figure 14 also shows that asolectin behaves similarly to egg

lecithin with C5b-8 and C5b-9. Asolectin was chosen for the planar

lipid bilayer experiments with C5-8 and C5b-9 because it forms much more

stable membranes than egg lecithin. It was also interesting to test

whether differences in C5b-8 and C5b-9 mediated conductances would occur

with this lipid in view of the results from these LUV experiments.

C5b-8-mediated conductances. When C5b-6 and C7 were added to an

asolectin bilayer, no change in conductance was observed. After C8 was

added a rapidly fluctuating conductance began without observable single

channels. Fig. 15 shows this conductance as the voltage was varied from

-100 mV to +100 mV. This figure shows that the current mediated by

C5b-8 rectified with respect to applied voltage polarity, i.e., positive

conductances were about three times greater than negative conductances

at potentials greater than 40 mV. The current continued to increase

above +40 mV but became increasingly hyperbolic (super-ohmic) while the

current increased but plateaued significantly as the voltage increased

at negative polarity. Thus, C5b-8 conductances through planar lipid

bilayers are voltage-dependent. In general, larger net currents

resulted when larger voltages were applied, and the current continued to

increase at all potentials throughout the course of the experiment.

CSb-9-mediated conductances. For the experiments described in this

section C5b-8 was added in quantities (1-3 ug of each protein in a 4.2 ml

chamber) which did not produce significant conductances at a constant

voltage of 40 mV. This was done in order to help prevent the bilayer

from breaking due to excessive C5b-9 conductances. Initially, no

changes occurred when C9 was introduced to a bilayer containing C5b-8

-100 mV

100 mV

10 pA

20 mV

Fig. 15. Voltage-dependent conductance changes of an asolectin planar
lipid bilayer mediated by C5b-8. Both the cis and trans compartments
contained 100 mM NaCl, 10 mM Tris, pH 7.2 solutions.


under these conditions. But within a few minutes conductances began

which had different conductance characteristics from those observed with

C5b-8 (see Figs. 16 and 17). Figure 16a shows discrete channels of

widely different single conductances. First, a few sporadic channels

began opening and closing ranging in size from 5 to 30 pS with common

values of 5.0, 10, 12 and 28 pS (with an uncertainty of about 1 pS).

These channels usually remained open for less than a minute although a

few were observed to remain open for as long as 5 min. During this time

the baseline conductance remained the same. Figure 16b shows the onset

of much larger conductances starting as a series of very rapid (< 1 sec)

increases in the hundreds of pS (125 to 550 pS) of which a majority

remained opened over a period of minutes. The net conductances were

sometimes as large as 2000 pS. At this point the membranes often broke

although sometimes they did not. If the bilayer was broken, it usually

could be reformed, sometimes still with large currents and sometimes

with only the small channels shown in Fig. 16a. At these large membrane

currents the baseline fluctuated rapidly. Figure 16c shows that with

greater time resolution (60x) these fluctuations appeared to be composed

of rapidly opening and closing channels of about 125 pS which had life-

times on the order of 1 sec.

Figure 17 shows the changes in the complement-mediated currents

when a triangular voltage wave was applied. The C5b-9 conductances

demonstrated a linear, or ohmic, relationship thus showing another

departure from C5b-8 conductances. Although Fig. 17 is not the best

example of the C5b-9 conductance voltage independence due to the changes

in current depicted in this figure, other experiments consistently

demonstrated this characteristic (for example Fig. 18). Voltage

u %L-U. IpS

500 pS

5 min


5 sec

Fig. 16. Single channel conductances of an asolectin planar lipid
bilayer mediated by C5b-9 at -40 mV. Both the cis and trans
compartments contained 10 mM NaC1, 10 mM phosphate, pH 7.0 solutions.
See text for further-figure explanation.


-100 pA

100 mV

I i

-100 mV

10 pA

-10 pA

Fig. 17. Conductances of an asolectin lipid bilayer mediated by C5b-9
as a function of voltage. Both the cis and trans compartments contained
10 mM NaCl, 10 mM phosphate, pH 7.0 solutions. See text for further
figure explanation.


independence of C5b-9 currents was also observed at constant voltage

(usually 40 mV) whenever the polarity of the applied potential was

reversed. Figure 17a shows a voltage sweep which was begun when a large

current already existed at -40 mV. A net conductance can be seen of

2400 pS at 100 mV at the end of the first sweep through positive voltage

which then began to decrease in several large steps starting at -80 mV.

The current had decreased to 150 pS by the end of the second sweep

(Fig. 17b). This decrease occurred as a series of large current drops

of about 500 pS (within a margin of perhaps 50 pS). At this point the

sensitivity of the y axis was increased five fold (50 pA/cm to 10 pA/cm)

and the voltage sweeps continued. These tracings (Fig. 17c) show the

smaller current, still ohmic in nature, to be composed of the opening

and closing of channels which ranged from 100 pS to less than 10 pS:

similar to channels observed before the large (> 100 pS) current jumps

had begun. Certain portions of these curves show the fluctuation of

current which had earlier been demonstrated by the C5b-8 mediated

conductances. The smaller channels also continued to close without

reopening until a line with zero slope resulted without channels.

Equation (1) from Finkelstein (38) was used to calculate the

diameter of some of these channels based upon the conductances of single


G xr r2 x C
(1) g =-

where g is the conductance of a channel (S); r is the radius of the

channel (cm); 1 is the distance spanning the bilayer (30 A or 3 x 10-7

cm for an asolectin bilayer); c is the salt concentration of the

solution (mol/L); and G is the equivalent conductivity of NaCl (126.4

Scm2 eq-1 (14)). This relationship is based upon the simplest assump-

tions treating a protein channel like a cylinder spanning a bilayer with

the same conductivity as water but was shown to provide an accurate

model for gramicidin (38). This equation assumes a neutral membrane

although this is not the case for asolectin (which has a net negative

charge on its headgroups) but provides a basis for relating channel

conductances to pore diameter. Table 2 summarizes some of the observed

channel conductances for C5b-9 and estimations of the diameters of the

corresponding protein pores.

MAC-mediated conductances. After forming a bilayer from monolayers

with a spread vesicle/protein mixture on one side small channels ranging

from 2.5 pS up to 12 pS were observed at 40 mV over a period of an hour

or more without larger conductances. These channels lasted from several

seconds to several minutes with more than one occurring at the same time

occasionally, i.e., having the same appearance as the channels shown in

Fig. 16a. After a period of time (ranging from one to several hours)

large conductances (hundreds of pS) began to occur. Figure 18 shows the

voltage dependence upon the onset of these large conductances. Again, a

range of channels are observed including single openings and closings of

about 1000 pS down to channels less than 20 pS. In some regions of

these curves the fluctuating currents are observed that were noted

earlier and then associated with C5b-8 conductances. Such a statement

cannot be confidently offered in this case due to the magnitude of these

fluctuations (up to 500 pS) and the unknown composition of the protein

complexes incorporated into the bilayers. However, channel-like conduc-

tances are observed as before over the same size range and with similar

lifetimes. Table 2 also summarizes the MAC/bilayer channel data along

with estimates of pore size based upon Equation (1).

-100 mV

Fig. 18. Conductances of an asolectin
purified MAC as a function of voltage.
compartments contained 100 mM NaCl, 10

100 m
100 mV

100 pA

20 mV

lipid bilayer mediated by
Both the cis and trans
mM phosphate, pH 7.2 solutions.

Table 2. Pore Diameter Estimations from Complement-Mediated


C5b-9 (pS)a


MAC (pS)b


pore diameter (A)

estimated pore

diameter (R)


a[NaCl] = 0.01 M
b[NaC1] = 0.10 M

C9, C9n additions to asolectin bilayers. As a supplement to the

above experiments and to those in the next chapter of this text, the

interaction of C9 alone was tested with asolectin bilayers under the

same conditions as with C5b-9 additions. In experiments lasting up to

six hours, C9 was added in quantities reaching 130 ug/ml or 307 ug total

protein added to one compartment of the bilayer apparatus. This quan-

tity corresponds to about 60 times the amount of C9 used in the C5b-9

experiments. This experiment was also performed separately with C9n in

the same quantities. No channels or changes in the baseline conductance

of the bilayer were observed in either case at 40 mV or under conditions

of linear voltage variation up to 100 mV at negative and positive polar-

ities (see Fig. 26 in Chapter 4).


The data presented here indicate that with asolectin planar lipid

bilayers: i) the fewest terminal complement components which mediate

conductances are C5b-8; ii) C5b-8 conductances do not show single

channel characteristics and are voltage dependent; iii) C5b-9 conduc-

tances include single channels of various sizes (5 to 1000 pS) which

open an close; and iv) C5b-9 conductances are voltage independent.

In contrast to experiments published by other researchers which

were reviewed in the Introduction, these experiments were performed

using purified proteins. Although the C5b-6 and C7 used were less pure

than C8 and C9, these were at least functionally pure because no conduc-

tances were observed after addition of these two proteins until C8 was

added. The C9 and C8 used were pure as determined by NaDodSO4-PAGE and

did not contain functional activities of each other based upon the

markedly different conductances these proteins demonstrated as either

C5b-8 or C5b-9. Except for Mayer, other workers have usually used serum

or functionally purified complement components. With a technique as

sensitive as the bilayer apparatus one of its potentially greatest weak-

nesses also lies in its strength: its ability to detect reactions of a

single molecular entity. Thus, compounds present in the reaction mix-

ture other than those of interest, even in very minute quantities, might

possibly be causing the effects attributed to substances under study.

These possibilities are minimized when the use of purified proteins is

employed. Indeed, whenever serum was used as the source of complement

by other researchers, channel activities were usually observed even in

control experiments.

Using purified components also allowed the observation of C5b-8

currents independent of those produced by C5b-9. Mayer and his co-

workers (79,81) were the only other group thus far to distinguish these

conductances. However, their work did not present conductances for

either of these species for a full range of voltages. Neither did they

show any single channels in their data. Another point of difference is

their assertion that addition of C9 to C5b-8/membranes "stabilizes" the

lesion, i.e., the channels are irreversibly open in contrast to results

presented here that large and small C5b-9 channels do close.

The results with CF release from egg lecithin and egg lecithin/

cholesterol vesicles by reactive lysis represent work by other

researchers (52,102). One of the main observations from these experi-

ments is that additional marker release, after C5b-8 mediated CF

release, does not occur when C9 is added with egg lecithin vesicles.

When cholesterol is included in the lipid composition, there is addi-

tional CF release when C9 is added after marker release is produced by

C5b-8. The reasons why this occurs are not certain at this point.

Probably, C5b-8 alone on the surface of an asolectin or egg lecithin

vesicle without cholesterol is sufficient to leak carboxyfluorescein

while C5b-9 is needed for CF release from vesicles containing choles-

terol. It is certain that MAC formation does occur on the surfaces of

both types of vesicles from EM inspection of these complement-treated

vesicles (Dankert, personal observations). In Fig. 14 the release of CF

from asolectin vesicles mediated by complement reactive lysis is shown

to be similar to the results with egg lecithin. A bilayer study of the

effects of membrane composition on complement reactive lysis may provide

some of the answers to the questions discussed here.

An interesting experiment presented here was the addition to a

bilayer of egg lecithin vesicles reconstituted with purified MAC. The

-results indicate that similar conductances were produced to those

observed with reactive lysis: both large channels (< 1000 pS) as well

as small channels (< 12 pS). These channels also exhibited voltage

independence as C5b-9 currents did. Both of the results from each

method of introducing complement to a bilayer indicate that a hetero-

geneity of channel sizes are produced. Whether these different conduc-

tances are mediated by a single form of the MAC which appears ultra-

structurally as a "doughnut" or by a heterogeneous assortment of MAC

which are intermediate to the "doughnut" both in appearance and in pore

size is not known.

From these results it is apparent that channel activity resides in

C9. The probable role of C5b-8, in reactive lysis and lysis requiring

C9, is the reorientation of hydrophobic regions within C9 in the

presence of a membrane so that the channel-forming functional domains

may become membrane-inserted, conducting channels.



The addition of C9 to EAC1-8 greatly accelerates the rate and

increases the extent of lysis. Bacteriolysis by complement has been

shown to require C9 in most, if not all, cases studied to date. These

results have led many researchers to conclude that the major function of

membrane-bound C5b-8 may be to interact with C9 in a manner which trans-

forms C9 from its plasma-soluble, non-membrane associated state to one

which demonstrates great membrane affinity that is capable of causing

membrane damage. Any hypothesis(es) offered to explain the role of C9

in a complement membrane lesion would require that some correlation be

made between the membranolytic functional activity expressed by C9 and

its molecular structure. Biesecker and Mueller-Eberhard (6) first

published the amino acid composition of C9 and in a subsequent paper (5)

reported the effects of different proteolytic enzymes on C9. Their

findings showed that human thrombin treatment of C9 selectively cleaves

one peptide bond generating two fragments of C9 which are non-covalently

associated (see Fig. 19). This treatment did not result in any loss of

C9 hemolytic activity. They referred to this "nicked" form of C9 as C9n

and named the 34,000 Da and 37,000 Da peptides C9a and C9b, respectively.

Amino acid composition analysis of purified C9a and C9b showed that C9 is

a highly amphiphilic protein with the carboxy-terminal C9b fragment less

hydrophilic than the NH2-terminal C9a fragment (5). To quantitate

(300-310 ) ac-Thrombin Site
'OOH NHiGly-Lys-Gly-(SAC)-Phe

13.9% CHO)


34.000 Mr,Hydrophilic domain


37,000 Mr, Hydrophobic domain


Fig. 19. Schematic representation of C9 showing the proposed hydro-
philic and hydrophobic domains, respectively. The site of a-thrombin
cleavage is shown near the center of the protein as well as the
carbohydrate contents of each fragment. (from Biesecker et al. (5)).



this difference, a normalized average hydrophobicity index was calcu-

lated for each peptide based on its amino acid composition using a func-

tion derived by Barrantes (2). These calculations showed that C9a is a

highly charged, acidic peptide while C9b has a hydrophobicity index

approaching integral membrane proteins.

These observations were followed by a communication by Ishida et

al. (52) in which C9 and C9n were used with C5b-8 to lyse vesicles.

These vesicles contained a membrane-restricted, radiolabeled fatty acid

which could be coupled to a membrane-inserted protein via a photoacti-

vatable probe (12(-4-azido-2-nitrophenxy)-stearoyl glucosamine). Their

results showed that C9 was labeled by the probe to a much greater extent

than the other terminal complement proteins and that all of this label

was in C9b, implying that lytic activity by complement probably resides

in a functional domain of this peptide.

The work of these two research groups indicated that there might be

functional domains within C9 which are exposed by C5b-8 on a membrane

surface. This model of C9 is made more attractive because of the prece-

dent of some bacterial toxins as molecules with several functional

domains. Examples of such molecules include diphtheria toxin, the

colicin toxins E2 and E3, and cloacin DF 13 (40). Studies relating

functional activities of these toxins to their structure revealed that

each of these molecules have a membrane receptor binding region, an

enzymatic region which causes the toxic effect upon a target cell (such

as inhibition of protein synthesis), and a region which is thought to

help transport the active domain to its target. Proteolysis and isola-

tion of fragments of these toxins have yielded peptides which show each

of the above functions in vitro. Another group of toxins such as

colicins El, K, la, and A appear to effect their bacteriocidal actions

via energy-depolarization, presumably through the insertion of the

protein or a peptide across the energy-transducing bacterial inner

membrane (25,29,67,115). Isolation of proteolytic fragments of these

toxins have yielded peptides which were capable of binding to bacteria

but could not kill them, and peptides which could not bind to bacteria

but showed in vitro energy-dependent depolarization or ion flux with

inner membrane vesicles, artificial lipid vesicles, and planar


It has not been demonstrated with these toxins that proteolysis is

required for their in vivo activity except for diphtheria toxin. The

proteases typcially employed in these domain studies have not been those

of in vivo significance, e.g., the use of trypsin with diphtheria toxin

and colicin El. Still, the use of proteases to generate functional

peptides from toxins has yielded useful information about intoxication

mechanisms and membrane/protein interactions. A similar study with C9

proteolytic fragments could yield important information about whether a

domain structure of C9 is important to complement-mediated cytolysis.

The next section will present the generation and purification of two

c-thrombin peptides of C9 which will then be characterized for lytic

activities with the same assays employed in the last chapter of this

text: lysis of erythrocytes both with and without other complement

proteins present, release of 6-carboxyfluorescein from lipid vesicles,

and conductance changes of planar lipid bilayers (35).


Characterization of proteolytic fragments of C9 (C9a, C9b).

Figure 20 shows the elution profile of C9a and C9b from a preparative

NaDodSO4-PAGE tube gel apparatus. The first peak from the left repre-

sents free 1251 present in the C9n pool loaded onto the gel. The second

peak is a smaller peptide of C9 than C9a or C9b which eluted with the

dye front (bromphenol blue). The leading shoulder of the C9a peak is

also probably a smaller peptide of C9. The peptides were identified as

C9a and C9b by NaDodSO4-PAGE utilizing Coomassie blue and silver

staining techniques, or by autoradiography when 1251-C9n was used as the

starting material. Figure 21 is an autoradiogram of an NaDodSO4-PAGE

gel characterizing the elution profile and the separation of C9a and C9b

obtained with this system. Because during the course of these experi-

ments it was observed that 1251-C9 is a poor substrate for thrombin

cleavage, i.e., it resisted proteolysis into C9a and C9b remaining

uncleaved C9 instead, C9 was first thrombinized into C9n and then radio-

iodinated for these experiments.

The pooled peptide fractions contained NaDodSO4 incurred during the

purification process and were dialyzed from 2 to 7 days with several

changes of buffer in order to minimize the amount of residual NaDodSO4.

Removal of NaDodSO4 was desired because NaDodSO4 inactivates C9, and in

order to minimize the amount of NaDodSO4 introduced into the bilayer

apparatus. Small volumes of 2 M KC1 were occasionally added to the

dialyzing solutions to monitor the amount of NaDodSO4 remaining in

solution by precipitating KDodSO4 (the potassium salt is much less

soluble than the sodium salt). Greater than 90% of the peptides

remained in solution (judged by loss of radioactivity in aliquots

050 00 150 200


Fi. 20. C9a and C9b elution profile from a preparative NaDodSO4-PAGE
tube gel. The first peak from the left corresponds to free 1251; the
second peak contains C9 peptides smaller than C9a and C9b and elutes
with the bromphenol blue dye front; and the last two peaks are C9a and
C9b. Another minor peptide of C9, smaller than C9a, elutes just prior
to C9a and is seen as a leading shoulder of C9a-containing fractions.
to C9a and is seen as a leading shoulder of C9a-containing fractions.




dye front

Fig. 21. Autoradiogram of 1251 radiolabeled C9n and samples
NaDodSO4-PAGE corresponding to fractions eluted from the preparative
gel. Approximately 30,000 cpm were loaded in each lane and the film
exposed to the gel for 20 h before development.


periodically withdrawn from the dialyzing solutions) after dialysis into

buffers with or without detergent. Well-dialyzed solutions demonstrated

less detergent-like (i.e., "noisy") conductances with planar lipid bi-

layers (see below) and this observation was utilized as well in order to

judge the adequacy of dialysis.

C9b/planar lipid bilayer interactions. When C9b, solubilized in SB

3-14, was added to an asolectin planar lipid bilayer conductances were

observed in the form of discrete channels which remained open as long as

a minute before closing (see Fig. 22). At -40 mV the most prominent

single channels were about 11 1 pS but smaller channel sizes were also

observed. The smallest of these appeared to be about 2.5 pS. Fig. 23

shows a histogram of the range of channel sizes observed at -40 mV.

This figure was made by measuring single conductances of 100 consecutive

channels following the addition of <1 ug of C9b to a bilayer. An esti-

mation of the channel diameter for each single channel conductance

observed was calculated by equation (1) and the results summarized in

Table 3. The pore diameters estimated for C9b conductances appear to be

unrealistically small to mediate ion flow through a membrane. Sodium

ions have a diameter of 1.92 R and a hydrated diameter of 5.55 R. These

discrepancies may be due to inadequacies of equation (1) which repre-

sents a very simple model for a protein channel, and because charged

lipids were used to form the membrane. As mentioned above, estimations

of channel sizes using equation (1) would probably be more realistic

with a neutral membrane. However, the pore size estimations are

probably correct within an order of magnitude and are useful to illus-

trate relative channel sizes for different conductances.

VOLTAGE = -40 mV

protein addition


5 min

Fig. 22. C9-mediated single channels with asolectin bilayers at -40 mV.
Approximately 0.1 ug of C9b was added where indicated as determined by the
method of Lowry (73). Both compartments contained 100 mM NaCl, 10 mM
phosphate, pH 7.2 solutions.

10 -8


0 4


2 3 4 5 6 7 8 9 10 11 12


Fig. 23. Histogram of C9b single channel conductances at -40 mV.


Table 3. Estimation of C9b Pore Size from Single Channel Conductances

Channel (pS) Calculated pore diameter (R)

2.4 0.85
5.4 1.28
10.2 1.75
11.4 1.86

When the polarity of the applied voltage was reversed, i.e.,

changing the voltage from -40 mV to +40 mV, the magnitude of the current

and the single channel sizes did not change. To test for voltage depen-

dence of C9b channels across a larger range of potentials, a triangular

voltage wave was applied to the bilayer which varied the potential from

+120 mV to -120 mV in a linear manner. Figure 24 shows the effect of

this voltage change on the C9b-mediated conductances. As the voltage

varied, the net current produced by the channels changed nonlinearly

indicating that the channels are voltage dependent throughout the range

of these voltages. This effect is more pronounced at potentials greater

than 40 mV. There was also current measured at all potentials indi-

cating that the C9b channel does not require some minimum voltage to be

applied before it can conduct ion flow across a membrane. Figure 24

does not show the most typical appearance of the voltage dependence of

C9b channels. It has the appearance that positive currents are smaller

than negative currents at potentials greater 60 mV. Trace A of Fig. 25

is a voltage dependence of C9b channels obtained under similar condi-

tions as Fig. 24 which is more representative of those usually observed:

greater positive current than negative throughout the range of

voltages. The experiment used in Fig. 24 was selected because it

clearly shows single channel formation at all applied potentials.

Lipid specificity of C9b/bilayer conductances. C9b channels also

demonstrated lipid specificity (see Table 4). Four different lipids

were used to form bilayers. Bilayers composed of asolectin showed the

most channel activity and was the lipid chosen for all the other experi-

ments presented in this test which characterize C9b channels. Soybean

phosphatidyl choline (PC) and plant phosphatidyl ethanolamine (PE) were

100 mV

10 pA

20 mV

Fig. 24. C9b conductances with an asolectin planar lipid bilayer as a
function of voltage. Approximately 0.4 ug of C9b was added to the
membrane. Both compartments contained 100 mM NaCl, 10 mM phosphate, pH
7.2 solutions.

-100 mV

Table 4. C9b Lipid Specificity

Lipida Net Current (pA)b C9b (ug)

asolectin 10 0.2
soybean PC 7.5 0.6
plant PE 3.5 0.6
diphytanoyl PC 0.1 0.8

afor each experiment both compartments contained 10 mM NaCl, 10 mM
phosphate at pH 7.2

bat -40 mV

also used to form bilayers in other experiments. C9b-mediated conduc-

tance was observed for each of these lipids but the soybean PC gave

larger conductances with more observable single channels. Diphytanoyl

PC did not demonstrate any channel activity even after the addition of

ten times more C9b (5 ug) than was used to observe channels with the

other lipids.

Ion selectivity of C9b/bilayer conductances. Modifications were

made to the planar bilayer experimental conditions as described in

Methods to determine whether the C9b channel shows an ion selectivity

between Na+ and K+. An asolectin bilayer was formed with symmetric 10

mM NaCI solutions on either side. Trace A in Fig. 25 shows the C9b-

mediated conductance after a 1 ug C9b addition to the cis side of the

bilayer. The conductance of the bilayer was then allowed to reach a

steady state while a triangular voltage wave was being applied from -140

mV to +140 mV. This trace demonstrates the aforementioned voltage-

dependence of c9b channels. Trace B shows the conductance which

occurred after making the solutions on either side of the bilayer asym-

metric by raising the NaC1 concentration in the cis compartment to 200

mM. The entire trace is shifted away from the origin to the left (nega-

tive voltage) with the new "x" intercept at -115 mV and the new "y"

intercept at 18 pA. This shift can be explained if one assumes an

accumulation of Na+ in the trans compartment so that when no potential

is applied by the electrode, current still occurs as if a positive

potential was being applied. If Cl- had accumulated in the trans

compartment then the current would have shifted to the right as though a

negative potential was being applied. This result shows that Na+ is

transported preferentially over Cl- under these conditions. However,

200/ I 0 NaCI

-I 00 mV

0 pA

100 mV

NaCI/ 90 KCI


Fig. 25. Ion selectivity of the C9b channel. A is the C9b conductance
dependence after a 0.3 ug addition of C9b with symmetric 10 mM NaCl
solutions in the cis and trans compartments. B is the conductance with
asymmetric NaCl solutions after a 2 M NaCl addition to the cis
compartment to form a 200 mM NaCl cis solution with a 10 mM NaC1 trans
solution. C is the conductance after a 2 M KC1 addition to the trans
compartment to give a 200 mM NaCl solution in the cis compartment and a
trans compartment with 190 mM KC1 and 10 mM NaCl concentrations. The y
scale (current) sensitivity was increased 4x on trace C to keep the
conductance profile on scale with the other traces.

this result may be expected when asolectin is used to form a bilayer

because the phospholipid headgroups have a net negative charge which

would repel Cl- from the surface of the bilayer and make it less likely

to be transported across the bilayer.

Trace C shows the effect on C9b-mediated conductance when each side

of the bilayer has the same ionic strength but the cis compartment is

200 mM NaC1 while the trans compartment is now 190 mM KC1 and 10 mM

NaC1. In this experiment the Cl- concentrations are symmetric on each

side of the bilayer but the cation concentrations are asymmetric. If

the C9b channel did not selectively transport one cation over the other

the result would be a current trace with "x" and "y" intercepts at the

origin since no potential would arise due to the asymmetric accumulation

of cations on one side of the bilayer. However, Fig. 25 shows that

trace C has shifted to the right of the origin with an "x" intercept of

about +28 mV and a "y" intercept of about -60 pA. This result shows

that K+ is being transported preferentially over Na+ to give an accumu-

lation of positive charge in the cis chamber. In order for no net

potential to exist across the bilayers, the electrode has to apply a +25

mV voltage. The Goldman relationship (equation 2) (49) was used to

calculate a K+ selectivity over Na+ by C9b channels of 3.2 0.2 to 1

from this information:

RT PK (K+t + PNa (Na+)t
(2) E = --In
F PK (K+)c + PNa (Na+)c

where E is the potential difference (V) across the membrane in the

absence of any net ionic current; PK and PNa are the permeability

constants for individual ions; ( )t and ( )c are the ionic activities

for the trans and cis compartments, respectively; R is the ideal gas

constant (8.314 V coulomb T-1 mol-1); T is the absolute temperature (295

K for these experiments); and F is Faraday's constant (96,490 coulombs/

equiv). Several assumptions were made to obtain this calculation: that

the sodium and potassium concentrations near the asolectin membrane

(which is negatively charged) are much greater than that of the chloride

concentration which creates a condition in which cation transport is

much greater than anion transport (as shown in trace B of Fig. 25); and

that the concentrations of sodium or potassium ions near the membrane

are the same, i.e., that each ion experiences the same coulombic attrac-

tion to the negative potential existing at the membrane surface.

Effect of varying the solubilizing solutions on C9b channel

activity. All of the experiments above describing the properties of

C9b-mediated single channels with bilayers were conducted with C9b

solubilized in either SB 3-12 or 3-14 zwitterionic detergents. These

detergents do not cause loss of C9 hemolytic activity with EAC1-8. When

C9b was dialyzed against buffer without detergent (100 mM NaC1, 10 mM

Tris or phosphate, pH 7) little or no channel activity was observed. In

several experiments C9b solutions were divided and dialyzed with either

detergent-containing buffer or 6 M urea. At least two of these experi-

ments yielded C9b solutions in SB 3-14 with very high channel activity

(literally thousands of channel events observed in a three hour experi-

ment) compared to a dozen channels observed for the C9b/urea solution.

In each case the C9b concentration remained constant throughout dialysis

as demonstrated by comparison of the specific radioactivities of ali-

quots from each solution. From this information it would appear that

the activity of C9b is much higher when detergent is present. The

channel-making activity does not appear to be very high in NaDodSO4

solutions either, although direct additions to a bilayer of NaDodSO4-

containing solutions were rare in order to avoid contamination of the

apparatus by this detergent.

Channel-making fragments of C9 were also produced by methods other

than NaDodSO4-PAGE. Another investigator (J.R. Dankert, unpublished

results) in this laboratory conducted experiments using C9n chromato-

focused in 6 M urea to obtain an eluting fraction of protein at pH 6.1

which formed single channels of 12 pS with bilayers. This fraction was

one of two peaks eluting from the column; the second of which eluted at

pH 5.5 (these pHs probably do not correspond to the pIs for these pep-

tides). Characterization of these two peaks by NaDodSO4-PAGE was not

successful due probably to the interference of the focusing "polybuffer"

with the pH of the electrophoretic gel producing anomalous gels (an

effect high resisted precipitation of the peptides with TCA before

electrophoresis). However, it is logical to assume that active C9b was

produced by this method due to the similarity of the channels to others

made with C9b in detergents. This result is important because chromato-

focused C9b was never exposed to NaDodSO4 or electrophoresis and helps

rule out the possibility that these two experimental conditions produced

an artifactual activity with C9b. Another point worth noting with this

experiment is that the activity of the peptide in the "polybuffer"/urea

solution appeared to be greater than when NaDodSO4-PAGE C9b was dialyzed

into urea.

C9a and C9/bilayer interactions. Figure 26 shows the conductance of

an asolectin bilayer after 5 ug of C9a solubilized in buffer containing

SB 3-14 was added. No channels or conductances were observed other than

CO (SB 3-14)

C9a (SB 3-14)

5 min

Fig. 26. Protein additions to an asolectin planar bilayer of C9b and
C9. In each experiment 2 ug of C9a and C9b solubilized in SB 3-14
detergent were added to the membranes. In the topmost profile 300 ug of
C9 was added to the membrane.

48 pSi

L 7:.' I7 ==-~ C ~- --L

those which were observed after additions of a similar volume of deter-

gent solution which did not contain protein. This amount of protein was

ten times the quantity of protein which gave C9b channel activity with


C9 and C9n did not cause single channel formation or conductance

increases when added to asolectin bilayers in substantial quantities

(>300 ug) as noted in Chapter 3. However, because all of the experi-

ments presented in this chapter with C9b channels were conducted with

C9b purified in and then solubilized in detergent solutions, similar

conditions were necessary for C9 experiments. This kind of control

experiment is important to make certain that the channel-making activity

demonstrated by C9b was not due to an anomolous detergent/protein or

detergent interaction with the bilayer, or some other kind of artifac-

tual effect which was a consequence of the manner in which the C9 pep-

tides were isolated.

First, C9 was diluted into SB 3-14 solution to give solutions which

were still > 100 ug/ml of protein, but now containing the same concen-

tration of detergent used with C9b, and added to an asolectin bilayer.

No channels were observed after this addition. A second control experi-

ment involved passing C9 over an NaDodSO4-PAGE prep gel with subsequent

solubilization of the protein into buffers with and without SB 3-14 as

was done with C9a and C9b. Addition of these protein solutions to

bilayers did not result in channel formation either as shown in Fig. 26.

Peptide-mediated marker release from CF/vesicles. C9a and C9b

purified by NaDodSO4-PAGE and dialyzed into SB 3-12 or 3-14 containing

entrapped carboxyfluorescein as another model membrane in addition to

planar lipid bilayers for the demonstration of a lytic fragment of C9.

The concentrations of the peptide solutions were determined by the

method of Lowry et al. (73) using BSA, BSA solubilized in SB 3-14

buffer, C9, and C9 in SB 3-14 was also tested as a control in an analo-

gous manner to that described above.

Figure 27 shows that each of the two peptides elicit marker release

when similar quantities were added to the egg lecithin vesicles which

were above phase transition at room temperature. The amount of marker

release observed was significantly greater than when an equivalent

volume (or up to 5 fold more) of detergent solution alone was used. C9a

and C9b which were still in NaDodSO4 solution or which had been dialyzed

into non-detergent solutions did not effect CF release (see Fig. 30).

C9 showed the same behavior in each of these experiments as C9a and

C9b. Figure 28 shows a dose response curve for peptide-mediated CF

release by both C9a and C9b. These curves show similar concentration

dependence for each peptide.

Because of the similarity of these results it was important to

demonstrate that C9a and C9b were not contaminated with each other.

Care was taken to show by NaDodSO4-PAGE with subsequent Coomassie and

silver staining, and autoradiography that each peptide solution was free

from the other peptide, at least to the sensitivity of these protein

visualization techniques. A purified monoclonal antibody to C9b was

used to inhibit the activity of C9b for marker release. Figure 29 shows

that this antibody did indeed cause loss of activity for C9b with

respect to the percent CF released in a given amount of time. The same

antibody did not inhibit the lytic action of C9a demonstrating the

independence of the two peptide actions.


5 min.


CSb (SB 3-14)

CSa (SB 3-14)

S(SB 3-14)
C9 (SB 3-14)

SB 3-14


Fig. 27. Carboxyfluorescein release from egg lecithin LUV after
additions of C9b, C9a, or C9. Each protein was solubilized in SB 3-14
after elution from an NaDodSO4-PAGE preparative gel. The last profile
is the CF release caused by an addition of detergent alone (10 ul)
concentrated at its cmc of 0.012%. Approximately 0.2 ug (2 ul) of each
protein was added to 8 ug of LUV in 250 ul of 100 mM KC1, 10 nM Mops, pH
7.2 solution.

100 200 .300 400 500 600


Fig. 28. Concentration dependence of CF release from egg lecithin LUV
mediated by C9a and C9b. C9a and C9b were solubilized in SB 3-14
containing solutions. Background lysis by SB 3-14 was subtracted for
each point. Different concentrations of protein were obtained by
dilution of protein solution into SB 3-14 buffer. A constant volume of
10 ul for the protein and 8 ug of LUV was used for each addition.

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.9 - mvs