1 iii (n (D :. :-* e; 3' ' t- "V r ’. v. '. ’ " v \ I § Ms 't 'fi-u ‘ '- . ’~. -. "7 _,. .~' . #’ rd. w. EMS”. im‘fimmm‘mu. I T' :1: This is to certify that the thesis entitled Complement Channel Formation In Lipid-impregnated Millipore Filters presented by Kevin O'Boyle has been accepted towards fulfillment of the requirements for 1:483:3353 degree in BiODhYSiCS \_ /' \\ Major professor $087 0-7639 Illilllllllllllfllllflflljll 293 OVERDUE FINES: 25¢ per day per item RETURNlm LIBRARY MATERIALS: - Place in book return to remove 93"" charge from circulation records COMPLEMENT CHANNEL FORMATION IN LIPIDhIMPREGNATED MILLIPORE FILTERS By Kevin P. O'Boyle A Thesis Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of BiOphysics 1981 Abstract Complement Channel Formation in Lipid - Impregrated Millipore Filters 13! Kevin P. O'Boyle It is desirable to study the electrical changes induc- ed in the lipid bilayer by the action of complement. A stable planar, membrane system, has been found which allows the ob- ' servation of electrical channels due to complement which ap- pear to support the "hydrophobic doughnut" or transmembrane channel hypothesis. Millipore filters were coated with a lip- id solution containing Forssman Ag and cardiolipin and then in- cubated at 37°C in the presence of the appropriate antiserum , and complement. Step-wise reductions in resistance were ob- served which appear to be due to the insertion of transmemb- rane protein pores. These complement channels were also ob- served when complement was activated via the alternative path- way and directly by the cleavage of C5 by trypsin. Tempera- ture and Concentration studies were done, and controls were run to verify that the electrical responses were due to com- plement. A calculation oi the pore size revealed a channel of 112 A in diameter. This thesis is dedicated to my mother, Regina O'Boyle, whose support has en- sured the progress and success of all my endeavors. Acknowledgments I would like to thank Professor H. T. Tien for the initial suggestion that led to this path of discovery. Table of Contents List of Tables List of Figures Introduction Chapter 1 - Literature Review I. Review of Antigen-Antibody and Comple- ment Reactions on BLM II. Review of the Activation and Attack of C on Liposomes III Other Immunologic Events Observed on ' Planar BLM's IV. Complement V. Classical Pathway of C Activation VI. Alternative Complement Pathway Chapter 2 - Materials and Methods I. Extraction of Sheep Erythrocyte Lipids II. Isolation of Forssman. Ag III. Release of Glucose from Liposomes IV. Preparation of Lipid-Inpregnated Millipore Filters V. Electrical Measurements vi. Other Materials and Methods Chapter 3 - Results and Discussion 1. Introduction II. Glucose Release from Liposomes Formed from Sheep Erythrocyte Lipid Extract III. Sheep Erythrocyte Lipid-impregnated Milli- pore Filters iii P359 vii 10 13 16 17 . 20 27 27 27 ‘29 30 32 32 33 33 34 35 iv Page IV. Millipore Membranes with Purified Forsslan A8 49 V. Concentration Studies 49 VI. Temperature Studies 52 VII. Activation of C by the Cardiolipin Ag-Syphili- tic Ab System - 54 VIII. Reactive Lysis 57 II. Calculation of the Size of the Transnembrane Channel ' 63 1. Evidence in Favor of the Translesbrane Channel Hypothesis ' ' 63 11. How does the Transmembrane Channel Lyse the Cell? ' 65 Chaptor 4 - Summary and Conclusions 68 Bibliography 73 Table Table Table Table Table Table Table Tablo Table Table Table Table Table Table 2. 3. 4. 5. 7. 8. 9. 10. 11. 12. 13. 14. List of Tables Proteins of the Complement System. Sheep Membranes with BOOXRAS + sooxaabbit C + SOOXHuman C + 7OOAVeronal-saline on Each Side. Sheep Membranes with Heat-inactivated Serum (300 Mus, 500 Miabbit c, soomuman c) on Each Side. 1:1 Phosphatidyl Choline: Cholesterol (no Ag) Membranes with BOOXRAS, sooxnabbit c, 500)Human C on Each Side Sheep Membranes with RAS, Rabbit C, and En- man C Added to Each Side. Sheep Membranes with RAS, Rabbit C, and Hu- man C on One Side. Sheep membranes with 200).RAS and soolcpc on Each Side. Sheep Membranes with heat-inactivated Serum (zooxans and 600) are) on Each Side. page 18 37 37 4O 4O 43 43 46 1:1 Phosphatidyl Choline:Cholesterel Membran- es Without Ag But 200 )tRAS + 600 XGPC on Each Side. ' Sheep_Membranes with RAS + GPO on both Sides. Sheep Membranes with RAS 4 GPS on One Side. 46 46 48 1:1 Phosphatidyl Choline:Cholesterel Membran- es with 1:100 Forssman Ag (GL5) and 250nm + SOQ)GPC on Each Side. Temperature Study of Three-Sheep Membranes with 300mm + 500}Rabbit c + SOONiuman c on Each Side. ~ Cardielipin Membranes with 200}Syphilic Ab + sooxcpc on Each Side. 48 55 55 Table 15. Table 16. Table 17. vi PCS. 1:1 Phosphatidyl Choline: Cholesterel Mem- branes Without Ag But with previously Act- ivated Ab + C on Each Side. 55 Reactive Lysis Demonstrated on 1:1 Phos- phatidyl Choline:Cholesterel Membranes with Either Zymosan A or Trypsin on One 61 Side. Millipore Filters with Larger Pore Sizes. 61 Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 2. 3. 4. S. 9. 10. 11. 12. Table of Figures page Schematic Representation of Complement S’Stem o 19 Schematic Diagram of Electrical Appar- atus. 31 Glucose Release from Liposomes. 36 A) Sheep Membrane (#8 from Table 2) show- ing C Channels. B) Heat-inactivated Con- trol from Table 3 (membrane #2). 41 A) Membrane 4 from Table 7 showing C Act- ivity with GPC. B) Without Ag control mem- brane from Table 9. C) Heatinactivated con- trol membrane from Table 8. 47 Membrane 3 (A) and control (B) from Table 12. 50 Reaction Rate Versus Total Serum Concen- tration. Sl Membrane 7 from Table 6 showing the Effect of Increasing the Temperature to 50°C. 53 Arrhenius Plot of Temperature Data. . 56 A) Membrane 2 from Table 14 showing C chan- nels on a cardiolipin membrane. B) Heatinact- ivated control. 58 I/V Plot of Membrane 5 from Table 14 After the C Channels Vere Formed in the Membrane. 59 Membrane 2 from Table 15 showing Reactive Lysis. 62 vii Introduction It has been of interest for some time to study im- munologic reactions on artificial membranes. Several in- vestigators have reported conductance changes on ELM due to attack by complement(C). Most of these studies were not carefully done, however, and the results obtained pro- bably reflect nonspecific resistance changes because the BLM is not stable to the addition of many large proteins. The release of glucose from liposome model membranes has also been studied as a system for observing the attack of complement on artificial lipid membranes. This represents a much more stable system yielding consistent results which have been reproduced often. This system is limited, however, in that it does not allow for the observation of individual complement channels. This thesis presents data on the activation and attack of complement on a more sta- ble planar membrane system, the lipid-impregnated milli- pore filter. This stable planar artificial lipid membrane system permitted the addition of large amounts of serum protein without the production of extraneous resistance changes. When complement was activated, however, dis- crete step-wise reductions in resistance were obtained, indicating the insertion of individual complement channels 2 into the membrane. A calculation of the pore size was made on the basis of the average conductance change per channel observed. This revealed a channel diameter of 112A in excellent agreement with 100A holes seen in negative stain electron micrographs of sheep erythrocyte ghost membranes attacked by complement(1). A unique feature of the studies reported in this thesis which represents a distinct improvement over the earlier ELM studies is that lipid antigens, viz., Forssman antigen(Ag) and cardiolipin antigen (Ag), were used to activate complement by the classical antigen-antibody (Ag-Ab) pathway at the membrane surface. The lipid Ag was incorporated directly into the membrane—forming solu- tion used to coat the millipore filters. Antiserum (anti- boby(Ab) plus complement(C)) and an additional source of complement(e.g., normal guinea pig serum) were then added in concentrations known to give release of glubose from liposomes. The formation of complement channels under these conditions was a reasonable event. The observation of resistance changes in millipore membranes gives rise to a new immunoassay technique for the detection of small lipid haptens(univalent antigens) which are impossible to detect by other means, e.g., precipitation. The chief hypothesis of complement action on cell membranes leading to cell lysis is Mayer's transmembrane channel or "hydrOphobic doughnut" hypothesis. What this hypothesis says is that there are exposed hydrOphobic sites 3 on the activated complement proteins which allow insertion into the lipid bilayer of the cell membrane. These comple- ment proteins form a transmembrane hole in the membrane of about 100A which permits the passage of small molecules and water into the cell but blocks the escape of large macromolecules from the cell. An osmotic gradient then builds up which causes swelling and rupture of the ce11(2). This thesis presents further evidence in support of this hypothesis. Concentration and temperature studies were also car- ried out. Total serum concentration and temperature when increased were both found to have a positive effect on the reaction rate. An Arrhenius plot revealed that the total energy of activation for the activation and attack of com- plement on lipid-impregnated millipore filters was 0.306 keel/mole. A current/voltage plot was also made after complement channels were formed in a millipore filter. This plot showed that the complement channels are vdltage- independent. Complement was also activated via the alternative pathway and directly by the cleavage of C5 by trypsin in order to provide further evidence that the electrical channels observed in the millipore membranes reported here were indeed due to the action of complement.. Also, a proposal is made concerning the cytolytic mechanism of complement in this thesis. It is hypothesized that an electrocapillary event precedes the final osmotic 4 lysis caused by complement. Evidence for this idea comes from two sources: 1)A nonosmotic swelling is seen in sheep erythrocytes before the final osmolytic event but after the transmembrane channels of complement have been intro- duced into the membranes(3). 2)The electric potential across cultured cell membranes is wiped out by complement, and ouabain, which blocks the Na-K ATPase, inhibits the recovery these cells normally make if the complement is washed away(4). This shows that the normal cell potential is important in preventing lysis, and that a change in that potential may be reaponsible for leading that cell down the path toward irreversible osmotic destruction. Chapter 1 Literature Review I. Review of Antigen-Antibody and Complement Reactions on BLM: The first study of immunologic reactions on an arti- ficial membrane system was that of del Castillo et a1(5). These investigators found that marked transient impedance changes could be observed when a soluble antigen was added to a BLM(bimolecular lipid membrane) already sensitized with the appropriate antiserum. No impedance changes were observed if the antigen(Ag) and antibody(Ab) were mixed before they were added separately. Enzymes and their sub- states gave similar transient effects. This technique was even used in clinical trials to detect antibodies to insu- lin in patients having diabetes(6). Not all phosphdlipid preparations showed the responses, however, and traces of oxidized cholesterol anda-toc0pherol blocked the response es. As a result the clinical utility of such a system re- mains uncertain. . Barfort et a1 (7) found a permanent reduction in the D.C. resistance when homologous antigen and antiserum were added to opposite sides of the lipid bilayer. This effect was attributed to complement because antiserum heated to 56°C for 30 minutes was ineffective in eliciting the re- 5 6 sponse. These results, however, are compromised by two facts. First, the antigen and antiserum(containing the appropriate antibody and complement) were added to opposite sides of the membrane not allowing for the intimate con- tact between Ag and Ab and between the Ag-Ab complex and complement(C) thought to be essential for the activation of C by the classical pathway(8). Secondly, the aqueous medium in Barfort's experiments contained no added Caf+ or Mg++ which are required for complement activity(8). Mueller and Rudin(9) found that mixed brain lipid bilayers do not require the presence of Ag or Ab to inter- act with complement. Complement alone was found to increase the conductance several thousandfold and many membranes showed moderate voltage-dependent conductance decreases resembling EIM(excitability-inducing material). Since such brain lipid membranes are not known to activate complement through the alternative pathway and no Ag or Ab was pre- sent, these results demonstate the unsuitability of the BLM system for the study of complement activation on an artificial membrane surface. Obviously, nonspecific changes in resistance were obtained simply because the BLM is not stable to the presence of complement proteins in the aque- ous phase. Wobschall and McKeon<10) found transient step-wise and permanent changes in membrane resistance and membrane rupture when bovine whole serum and rabbit antiserum a- gainst bovine serum were added to opposite sides of the BLM. No conductance changes were observed if the anti- 7 serum and complement were heat-inactivated at 56°C for I 30 minutes, or if the complement was added to the side Opposite the antibody. Also, no significant reaction oc— curred if proteins or sera other than bovine were added to either the side containing the antibody and complement or the opposite side of the membrane. However, the mem- brane stability was less if both the Ag and Ab were pre- sent on the same side of the BLM. These investigators also plotted the mean step size of the resistance changes at 20mV as a function of the NaCl concentration and reported that the mean pore size caused by C was 22A in diameter. The results of Wobchall and McKeon(10) appear to support the earlier findings of Barfort et al(7). However, as with the previous work there are several questions that must be raised: 1)How can a soluble protein Ag diffuse a- cross a BLM to interact with its Ab and C when the perm- eability of ELM's to even small ions is so low? 2)Why weren't immune responses obtained when Ag-Ab systems other than bovine serum-antibovine serum were used? 3)If the membranes became unstable when Ag and Ab were present on the same side of the BLM, were the C responses observed Just as nonspecific? Also, the investigators reported noise and step changes in resistance for membranes without added Ag, Ab, and C. These events were thought to be rare. However, in the light of the nonspecific responses found for complement by Mueller and Rudin(9), such events must throw doubt on the conclusion that complement activity was actually observed. 4)In the plot of the mean step 8 changes in resistance versus NaCl concentration, step changes were observed for a NaCl concentration of 0.025M which is dubious because the first component of complement (C1) is a euglobulin which is precipitated from serum at a concentration of NaCL less than 0.04M(11). A more convincing report of complement activity on planar BLM's comes from Michael's et al(12). These in- vestigators report that activated CSb,6 combined with C7 and C8 complement components causes'a moderate increase in the ion permeability of BLM's. The resistance was also found to be voltage dependent. The subsequent addition of the complement component C9, however, greatly amplified this resistance change. Also, the more C9 added the great- er the permeability change, until eventually membrane rup- ture occurred. In addition, the channels formed by the addition of C9 to membranes treated with C5b,6,7, and 8, were found to be voltage-independent. No permeability changes occurred when the complement components were added individually or in mixed pairs to the membrane. This data supports the hypothesis that complement attacks membranes by forming a stable protein channel across the lipid matrix of the membrane(2). Michaels et al(13) throw some doubt on their earlier results when they report that BLM's made from oxidized cholesterol or glycerol monoolein give permeability in- creases to CSb,6, and C9 when added individually to the membranes. This report highlights the need for a more ata- 9 ble membrane system which does not yield nonspecific electrical responses. Michaels et al(14) report that oxidized cholesterol and monoolein BLM's increase the sensitivity of the com- plement attack. They suggest that this is due to the de- creased thickness of these membranes. However, it could also be due to increased membrane fluidity which might aid the insertion of C into these membranes. Michaels and Mayer(15) have found that the permea- bility of BLM's increases in discrete steps following the addition of C components, suggestive of membrane channel formation, and that the size and stability of the channels increased with sequential addition of C components. A180,, prior to the addition of C9, the CSb-8 channel opens and closes continuously. Abramovitz et al(16) estimated the size of this complement channel to be greater than 30A. The reports by Michaels et al are a distinct improve- ment over previous BLM studies and give the cleareSt de- monstration of possible C activity on BLM's. The advantage of their system was the employment of purified C compo- nents isolated after activation through the alternative pathway by zymosan A. The most recent paper reporting both Ag-Ab and C effects on BLM's is that of Mountz and Tien(17). They found that heat-inactivated Ag-Ab complexes and fresh complement individually had no effect on the resistance of BLM's formed from oxidized cholesterol and phosphatidyl choline. When 10 combined, however, the resistance was found to decrease within five minutes. It was also discovered that Ab could be titrated against Ag to restabilize BLM's. It was sug— gested that the Ab might be restabilizing the membranes by deabsorbing the Ag from the BLM. The difficulties with this paper include: 1)Phosphate buffer was used to invest- igate C activity, in spite of the fact that it may chelate the Ca++ from solution and inhibit C activation. 2)Ag alone was found to destabilize the BLM. This was the basis for the titration of the Ab. 3)The Ag-Ab complexes were not likely to be at the surface of the membrane for C activa- tion if the Ab caused desorption of the Ag from the mem- brane surface. 4)The complement(guinea pig serum) was di- 10 which is too lit- luted by a factor of approximately 10 tle complement under the best of circumstances. For these reasons this paper represents a throwback to the earlier unbelievable reports of C activity on BLM's. II. Review of the Activation and Attack of C on Liposomes: Haxby et al(18) activated C on liposomes by means of a lipid Ag present in the membrane. They prepared liposomes from a total lipid extract of sheep erythrocyte membranes and observed 65% release of trapped glucose marker in the presence of rabbit antisheep erythrocyte antiserum(RAS) and guinea pig serum(GPS). Kinsky(19) reviewed much 0f the immunology work done with liposomes. Negligible glucose release was found if the CPS was first heated at 56°C for 30 minutes, or if the RAS was replaced by serum from un- 11 immunized rabbits. The release of glucose from the lipo— somes was monitored by an enzyme-catalyzed reaction coupled to the generation of NADPH, which is measured by its ab- sorption at 340nm: hexokinase glucose + ATP : glucose-6-phosphate + ADP G-6-P + NADP + glucose-6-phosphate 'dehydrogenase -—v 6'-PhOSpriogluconate + NADPH + H+ Sheep erythrocyte membrane lipids were employed be- cause the extract contains Forssman Ag and RAS possesses Abs directed against this glycolipid. Liposomes containing purified Forssman Ag(20) were also successful in releas- ing glucose. In addition, purified complement components and specific anticomponent Abs as inhibitors have been used to demonstrate that immune damage of liposomes in- volves the same classical sequence responsible for the Ab-induced cytolysis of cells(21). I Sheep lipid liposomes have also been found to re- lease a variety of other markers upon incubation with RAS and either whole guinea pig, human, or rabbit serum as a source of complement(22). Other naturally occurring lipid Ags have been found capable of sensitizing liposomes to Ab and C. Cardiolipin(23) is one such antigen which is of special interest because of the role it plays in the C-fixation test for syphilis. The phenomenon of "reactive lysis" initiated without Ag-Ab complexes, as in the alternative pathway, has also been studied on liposomes(24). It was found that 05b,6 + 12 C7 + C8 + C9 gave maximal release of trapped markers (glu- cose or chromate) for liposomes that were either positively or negatively charged showing that the complement molecules probably attack the hydrOphobic layer of the membrane di- rectly without regard for the polar groups of the lipids. Liposomes composed of dimyristoyl phosphatidyl cho- line(DMPC), cholesterol, stearylamine(SA), and galactosyl ceramide(25) were found capable of activating C directly, i.e., these lipids themselves were able to activate C via the alternative pathway. All four types of lipids were re- quired for this unique activation. Kinsky(26) has recently been using liposomes contain- ing synthetic haptens to produce an immune response, i.e., to produce Abs to the haptens by injecting the liposomes into rabbits. This is a very important new technique for obtaining a pure yield of Abs against a single antigenic determinant on a small molecule which by itself is not normally immunogenic. Another very useful and interesting application of liposomes has been described by Lewis and McConnell(27). By incorporating spin-labeled lipid haptens into liposomes they are able to study the early recognition and triggering events of C activation as Opposed to the terminal lytic destruction. They have found that the more fluid the mem- brane, the better is the complement fixatioaninding). Presumably, in order to activate C, the Ab molecules must be able to diffuse laterally on the membrane surface. 13 Hsia and Tan(28) have used elctron spin resonance as a membrane immunoassay. Liposomes with lipid-linked anti-- gens have spin markers trapped in their internal aqueous compartments. In the presence of Ab and C the Spin markers leak out so that a sharp, intense signal is detected. The magnitude of this peak intensity in the esr spectrum is directly proportional to the number of spin-label mole- cules released from the liposomes. Thus the degree of lysis can be determined accurately by measuring the change in this signal intensity. Richards et al(29) report that the complement lesion in liposomes may be transient. They have found that the percent of trapped glucose released from multilamellar liposomes(MLL) was inversely related to the aqueous volume of the liposomes. Also, large unilamellar liposomes(LUL) released only 35% of their marker. These data suggest that the complement lesion is Open only for a tranSient period since 100% release would be expected from LUL if the com- plement induced membrane lesion remained open permanently, and a transient complement lesion would result in a greater percentage of release from small liposomes than large ones. III. Other Immunologic Events Observed on Planar BLMs: Henkart and Blumenthal(30) used the BLM technique to study Ab-dependent lymphocyte-mediated cytotoxicity. They found that in the presence of Ab against an and human lymphocytes, dinitrOphenylated lipid bilayers showed rapidly 14 induced increases in membrane conductance of several orders of magnitude without membrane breakage. Such ionic permeability changes were only observed when the membranes' voltage was positive on the lymphocyte side, as wouldbe the case with a target cell membrane. This system is thought to mimic lymphocyte killing of the Ab-coated target cells and the primary event in lymphocyte killing of Ab-coated target cells is the creation of ion-conducting channels in the target membrane. This cell-mediated cytotoxicity is thought to be sim- ilar to cytolysis by complement(31,32). Both processes exhibit one-hit and colloid-osmotic characteristics, and in both cases the lipid bilayer is implicated as the target of attack. Mayer et al(31) prOpose that by analogy with the cytolytic mechanism of complement, the interaction be- tween a lymphocyte and a target cell may activate membrane- associated proteins of the lymphocyte, leading to exposure of hydrophobic proteins, which then_become inserted into the lipid bilayer of the target cell and form transmembrane channels. In Order to test this working hypothesis Mayer et al(33) developed a model system in which multilamellar liposomes containing dimyristoylphosphatidyl choline, ‘ cholesterol, dicetylphosphate and Forssman hapten in the 86Rb+ and 5qu(Na Chromate) in the aqueous lipid phase, and phase, are treated with anti-Forssman IgG and nonadherent cells from mouse spleen. It was found that lymphotoxin, a 45,000 dalton protein, released from activated lymphocytes 15 and responsible for killing target cells nonspecifically, when neutralized with anti-lymphotoxin, is not responsible for the specific killing of target cells which requires in- timate contact between the plasma membranes of the target and killer cells. This "membrane contact" is thought to in~ volve the formation of channels in the target cell membranes, and this hypothesis is supported by the evidence that anti- Forssman Ab and lymphocytes cause direct synergistic marker release from liposomes containing Forssman Ag. Wolf et al(34) Observed that Ab and Ag cause "patch- ing" on BLM's, and that Ab reduces the diffusion coefficient of Ags in the membrane. Similar phenomena were observed us- ing the lectin concanavalin A as the cross-linking probe. The diffusion and distribution of the amphipathic Ags was measured by fluorescence photobleaching recovery. Rosenstreich and Elumenthal<35> have looked at the ionophorous activity of a number of B cell mitogens(pro- teins which cause B lymphocytes to divide), and found that with the exceptions of Keyhole Limpet Hemocyanin(KLH) and EIM, none of the B cell mitogens were ionophorous(i.e., able to increase the permeability of membranes) when tested on BLM's. It may be that EIM and KLH act on lipid receptors and the other mitogens require protein receptors in the membrane to function as mitogens. It is thought that a perm- eability change is required for the stimulation of the cells to divide(36). Deleers et'al(37) have reported that specific glyco- lipids incorporated into the BLM can be detected by conduct- 16 ance increses caused by the lectin concanavalin A. Tosteson (38) showed that the addition of a glchprotein from blood group 0 human red cell membranes to the solution bathing‘the outside of the lipid bilayer membrane led to a steady in- crease of membrane conductance up to 25-fold above that of the control membrane, and when Con A(concanavalin A) was added in the presence of 1mM Ca++ a further 7-8 fold in- crement in conductance was observed. The ELM has, therefore, been used study several types of immunologic reactions. however, the liposome model membrane system has been found to be a much more reliable system for the study of immuno- logic reactions on artificial membranes, IV. Complement: Complement is an important group of proteins in blood serum which when activated can result in cytolytic activity as well as a variety of other biological activities, in- cluding opsonization(enhancement of phagocytosis),'chemo- taxis, leukocyte mobilization, virus neutralization, in- volvement in blood clotting and fibrinolysis, generation of kinins(small polypeptides which increase vascular perm- eability), and activation of B lymphocytes and macrOphages. There are several excellent reviews on complement(8,31.39-. #2). This thesis will concern itself only with the cytolytic function of complement. There are fourteen proteins in the complement system, not counting inhibitors and regulatory enzymes. Eleven of these proteins belong to the "classical" complement system, 17 and these are designated by the letter C and number: C1 (which comprises three distinct proteins, C1q, C1r, and C18 held together by Ca++), 02, C3, and so on up to C9. The other three proteins belong to the "alternative" path- way of the complement system. These are designated by the letters E,D, and P. A complete listing of the proteins is shown in Table 1..A schematic representatiOn.of the reaction sequences of the complement system can be seen in Figure 1. What this diagram shows clearly is that the alternative pathway proteins amplify both Ab-dependent and.Ab-independent C3 cleavage, and the alternative path- way is by no means necessarily of secondary importance to the classical Ag-Ab pathway. v. Classical Pathway of C Activation: In the classical pathway, Ag-Ab complexes activate C1q to C1q', which in turn activates C1r to C1; and C13 to C1§(the overbars indicate activated enzymes). C4 and C2 are cleaved by C13 and the C3-cleaving enzyme CEB:2§ is formed from two of the fragments. This enzyme is converted to the C5-cleaving CEB:2§:33 enzyme by uptake of C3b. 03a and 05a are polypeptide fragments(called anaphylatoxins) that mediate release of histamine from mast cells and produce smooth muscle contraction. 05a is also responsible for the oriented migration of leukocytes toward a site of injury(chemotaxis). Cell-bound C3b conveys the property of immune adherence and renders cells more susceptible to phagocytosis(opsonization). Also, C3b activates E lympho- 18 Table 1. Proteins of the Complement System. Proteins Classical Pathway C1q C1r C13 04 C2 Alternative Pathway P(Pr0perdin) 'D s C3 Attack Sequence 95 C6 027 08 C9 Control Proteins C1INH C3bINA ‘31H 390,000 83,000 83,000 209,000 117,000 220,000 23,500 100,000 190,000 206,000 95.000 .120,000 163,000 79.000 105,000 93.000 150,000 M.W;(Daltons) Cleavage Fragments 04a,C4b,04c,C4d C2a,C2b Eb,Ea C3a,C3b,C3c,C3d,C3e 05a,05b 19 C qu’Ab Clq’ Inflammatory ChN _ and Cytolytic Classical C C1r Products 5 - Pathway C1S' c 33 C4 ’ C4-—.._./ C2 car a Alternative Pathway ‘— Bb, C3 C3 Elm—e C91;L B j Figure 1. Schematic Representation of Complement System. 20 cytes and macrOphages. The formation of 05b-9, which is the cytolytic attack complex of complement is formed by a series of complexing steps not involving enzymatic cleavage. The mechanism of membrane attack by this complex will be discussed further below. VI. Alternative Complement Pathway: Another C3-cleaving enzyme(P,C3b:Eb) is generated on incubation of normal blood serum with certain microbial cells or substances, such as zymosan A, a carbohydrate of the yeast cell membrane, or endotoxic lipopolysaccharide from Gram- bacteria. The assembly of this enzyme starts with the fixation of nascent 03b on a suitable surface, such as that of certain polysaccharides. In the presence of Mg++, Factor E binds to the fixed 03b. Factor‘D, a serine esterase, then activates the bound E by cleaving off a ca. 20,000-dalton fragment designated Ea. The re- sultant complex C33:EE is an efficient C3-cleaving enzyme, but its stability is poor since it is subject to dissocia- tion. The uptake of P(184,00C daltons) via a binding site on C3b, reduces the dissociation tendency and thus pro- duces the relatively stable P,C3E:E3, which is a highly efficient C3-cleaving enzyme. This enzyme is converted to a CB—cleaving enzyme by the uptake of an additional 03b fragment. The alternative pathway is normally blocked by two control proteins, C3bINA(C3binactivator) and 1H. How- ever, on the surface of many microorganisms, the 03- and CS- cleaving enzymes of the prOperdin pathway are assembled 21 because the control proteins are inoperative under these conditions. The 05b-9 Complex and the Mechanism of Complement Attack on Membranes: Three main hypotheses have been proposed for the mechanism of cytolysis by complement: 1)Complement destroys membrane lipids enzymically, e.g., by hydrolysis of phospholipids, i.e., complement may act as a phOSpholipase. 2)Complement generates a detergentlike substance that breaks up the lipid bilayer of the cell membrane. This pr0posal can take two forms. The first form postulates that putative complement-associated phospholipase A gen- erates the detergent lysolecithin by hydrolysis of leci- thin(#3). The second form of this hypothesis postulates that a detergentlike substance may be generated by enzym- ic interaction among the complement proteins themselves. 3)Complement forms a transmembrane channel by inser- tion of hydrophobic peptide chains into the lipid bilayer. The first hypothesis and the first form of the second hypothesis can be ruled out on the basis of experiments in which it was shown that complement releases marker from the aqueous compartments of liposomes made from lecithin analogs with ether linkages that are not susceptible to hydrolysis by phospholipases(44). The second form of the second hypothesis, i.e., de- tergent generation by enzymic interaction among the com- 22 plement proteins themselves is thought to require the 8~101o such lytic molecules per production of about 10 cell in order to lyse by detergent action. This is be- cause a critical value in their concentration must be reached before lysis can occur. One must ask, therefore, whether it is possible to conceive of that number of de- tergent molecules from the quantity of complement protein that becomes fixed on the surface of a cell. The third hypothesis is by far the most viable hypothesis. This transmembrane channel hypothesis was first proposed by Mayer in 1972(45). This concept, termed the "doughnut model" was based on the fluid mosaic model of biomembranes and envisages an annular assembly, made up of complement proteins 05b-9, which penetrates into the lipid bilayer forming a stable transmembrane channel. It is assumed in this hypothesis that the terminal com- plement proteins undergo conformational changes when they 'react with one another and that these changes lead to ex- posure of hydrophobic peptides from their interior. Evid- ence which supports this hypothesis includes the follow- ing: 1) The cytolysis by complement is a two-stage event. When a cell is attacked by complement it swells until the membrane is exposively ruptured and the cell contents spill out. The cause of the swelling is the Donnan effect (46,47). The simplest explanation for this is that comple- ment makes holes in the membrane that are large enough to permit salt and water and other small molecules to pass, but are too small to allow macromolecules to leave the cell. 23 2) The one-hit theory of complement lysis. This theory was developed during the 1950's from kinetic and statist-‘ ical studies(48). Recently(48949) this theory has been con- firmed, indicating that the production of a single lesion in the cell membrane is sufficient for complement lysis of a cell. 3) Electron micrographs using the negative staining technique technique show lesions in cell membranes as a light ring with a dark central portion. Those pro- duced in red blood cells by guinea pig complement range between 8.5 and 9.5 nm internal diameter; those produced by human complement range between 10 and 11 nm(1). In negative stain electron microscopy, the electron-dense stain accumulates in regions accessible to water, and so it has been thought that the dark central portions of these lesions represent hydrOphilic pores surrounded by the complement proteins making up the light ring. However, freeze fracture electron microscopy(51) has failed to reveal lesions penetrating right across the bilayer. The alterations due to complement seen in freeze-fracture all appear to be limited to the extracellular fracture face. 4) The demonstration that complement releases phos- pholipid from membranes(52,53) lends support to the trans- membrane channel hypothesis because it indicates that hydrOphobic peptides are exposed on the interaction of the terminal complement proteins. Some of the exposed peptides become inserted into the membrane bilayer and form chan- nels. Those that do not become inserted may be responsi- ble for the phospholipid removal from liposomes. 0r, 24 phospholipids may be released by displacement from the lips id bilayer when the hydrophobic membrane-attack unit of complement components, CSb-9, inserts itself into the bi- layer. In any case, complement may cause changes in the lipid moiety of membranes, aside from the formation of channels. Further support for Mayer's "doughnut" hypothesis comes from the finding that the C5b-9 complex is an am- phipathic molecule that posesses apolar, detergent-bind- ing surfaces. This evidence was obtained from charge— shift crossed-immunoelectrophoresisC54). "Neoantigenic" determinants were also found on the CSb-9 complex, which are absent on native 05-C9 molecules. This indicates that the assembly of 05-09 into the terminal membrane CBb-9 complex is accompanied by conformational changes in the individual components that lead to the exposure of a- polar molecular regions in the complex. Recently, Shin et al(55) reported that increasing acyl chain length of lecithin liposomes containing chol- esterol was fOund to decrease the release of aqueous marker, suggesting that channel formation is affected by (membrane thickness. Similarly, phospholipid release . diminished with increasing acyl chain length. Increasing cholesterol concentration decreased the amount of aque- ous marker released, but increased the amount of phos- pholipid released. It is difficult to interpret these results because the lipid composition of a membrane not 25 only has an effect on the final attack complex of com- plement, CSb-9, but the membrane fluidity influences the activation of complement on the membrane surface as mentioned earlier(27). Cholesterol might increase phospholipid release by facilitating binding on the mem- brane surface, but decrease aqueous marker release by increasing membrane thickness and inhibiting channel formation. 'Electron spin resonance studies(56,57) of the effect of complement attack on membrane fluidity detected, de- pending on the probe used, either a C-induced decrease, increase, or no change in membrane fluidity. Therefore, the changes in fluidity are not essential for the de- ve10pment of the C lesion. They not represent C-induced leaky patches in the membrane. The C-lesion, which is moat probably due to the formation of a transmembrane channel by one or more C proteins, may or may not cause modifications in membrane fluidity. Ramm and Mayer(58) have found using rescaled ery- throcyte ghosts containing trapped protein markers and treated with anti-Forssman Ab and large doses of guinea pig complement that the channels produced have a long, although finite, life-span and an effective diameter of at least 55A on the basis of ovalbumin and hemo- globin release, and not more than 150A, since serum albu-' min was not released. Using osmotic blockers of differing Stokes' radii, Boyle et al(59) have found that the lesions produced in the red cell membrane by C are not uniform but vary in 26 size depending on the concentration of C9. A most important and interesting study used a glass microelectrode pushed up against the cell membrane of a living cell to record individual complement channels(60). The current recorded by the pipette underwent discrete jumps indicating the Opening and closing of channels. The channel currents were uniform in amplitude and suggested a C channel of approximately 8A in diameter. Control exper- iments with heat-inactivated C or cells which were not prior Ab-coated had no effect. Chapter 2 Materials and Methods I. Extraction of Sheep Erythrocyte Lipids: The following procedure was adapted from Haxby et al(1s): ' 1) Centrifuge sheep blood(purchased from Microbiological Associates) for 20 minutes at 1000x g. Wash erythrocytes three times in 0.85M NaCl. 2) Lyse the cells by suspending and mixing in 10x volume of distilled H20. 3) Spin down erythrocyte ghosts at 20,000x g for 40 min. 4) Wash membranes with distilled H 0 and spin at 25,000x g 2 for 20 minutes. 5) Extract lipids with 2:1 CHCl -Me0H. Filter through 3 0.22um millipore filter, dry and redissolve 1% in n-octane. II. Isolation of Forssman Ag(GL5): This method was adapted from Esselman et al(61): 1) Wash 100gm of canine small intestine with H2 . 2) Cut tissue into small pieces and put in Waring blender. 3) Add 500ml 2:1 CHCl :MeOH, homogenize 5 seconds at low 3 speed, 15 seconds at high speed. 4) Filter material through Euchner funnel with Whatman #4 paper. Dry filtrate in rotary evaporator. Put residue and filter paper back into blender. 27 28 5) Repeat steps 3 and 4. 6) Repeat steps 3 and 4 using 500ml 1:1 CHCl :MeOH. 7) Discard residue and redissolve combined :ried filtrates in 400ml 2:1 CHCIB:Me0H. Add 80ml H20. Shake and emulsify lipids. Store overnight at 4°C in a separatory funnel. 8) Take off lower phase. Discard interface and upper phase. Dry lower phase in rotary evaporator and leave overnight in an evacuated dessicator if necessary. Dry as much as possible. 9) Add 9ml pyridine + 6ml acetic anhydride to dissolve lipids. Let the mixture stand at room temperature for 18 hours.or until the lipids are acetylated. 10) Add 50ml toluene(redistilled) and dry lipid solution. Repeat 2x. 11) Dissolve the dried residue in dichloromethane(DCE) and put onto a Florisil column(made by pouring a slurry of 30gm of Florisil and 100ml DCE into a column with a glass wool plug at the bottom). 12) Elute the column with 300ml DCE to remove cholesterol. 13) Elute the column with 200ml 9:1 DCE:Me0H. Collect this fraction which contains the glycolipids. 14) Dry the solution of acetylated glycolipids on a rotary evaporator. Redissolve in a couple of mls of 2:1 CHCl5: MeOH. 15) Add 0.5% sodium methoxide in methanol until the solue tion is strongly basic(test with pH paper). 16) After 30 minutes, neutralize the reaction mixture With ethyl acetate, and evaporate to dryness. 29 17) Emulsify in about 10ml H20 in a water bath sonicator, and dialyze overnight at 4°C. 18) Freeze dialyzed material in a shell with isopropanol- dry ice bath and lyophilize glycolipids. 19) Redissolve lyOphilized material in 2:1 CHCl :MeOH, and streak on silicon thin layer plates. 5 20) Run plates in 42:21:3 CH013zMeOH:H2O for one hour. 21) Allow plates to dry, then spray with 0.8% orsinol in 4N H2804. Place plate in oven at 100°C for about 10 min- utes or until the lipids turn purple. 22) Scrape off GL5 band and redissolve in 2:1 CHCl :MeOH. 3 23) Dry CHCla:He0H and mix with membrane-forming solution for eXperiments. III. Release.of Glucbse from Liposomes: This technique was similar to that descibed by Haxby et al(18): 1).Liposomes were prepared by drying sheep erythrocyte lipids in a round-bottomed flask. The dried lipid film was then dispersed in a solution of 300mM glucose with the aid of a vortex mixer. 2) The glucose assay was run on a Eeckman spectr0photo- meter. The standard assay reagent consisted of 2mM ATP, 1.0 mM TPN+, 3.5mM MgCle, 145mM NaCl, 0.15mM CaC12, 4.9mM veronal(Na barbital) buffer at pH 7.4, 80ug/ml hexokinase, and 41 ug/ml glucose-6—phosphate dehydro- genase. The 2ml cuvette contained 200%(ul) rabbit serum, 150A human serum, 150) RAS, 500Xveronal-saline, and 3O 1ml standard assay reagent. The control cuvette, run as the blank in the spectrOphotometer, contained the same ingredients except the sera were heat-inactivated at 56°C for one hour before use. 10 or 20)\of liposomes were added to the cuvettes and mixed by inversion immediately before the absorbance at 340nm was recorded was recorded with time. IV. Preparation of Lipid-Impregnated Millipore Filters: Millipore filters were first used to form model lipid membranes by Tobias et al(61) in 1962. More recently, lipid impregnated millipore filters were used as a system for the assay of photopotentials from bacteriorhodopsin- containing liposomes(62). The procedure described below, however, is somewhat different from those used by other investigators: 1) 0.22um millipore filters were attached to specially prepared plastic chambers by inserting one part of the plastic chamber into dichloromethane and pressing it onto the millipore filter. The slightly melted plastic binds to the millipore filter and the excess filter is cut off. A diagram of the chamber used can be seen in Figure 2. 2) 10 of membrane-forming solution, e.g., 1% sheep eryé throcyte lipids in n-octane, is then spread on the milli- pore filter. After 30 seconds or when the surface drys, the millipore filter is inserted into the other half of the chamber apparatus and the bathing solution is added. 31 ‘4 recorder .-—, __.-._w I electrometer .____ __l lv‘rfivl millipore __J 43 F membrane short 105 10‘ 107 108 109 10’‘0 (3 o o o o 0 o ey—uJ; _L __ 150.1L. 5OAflL l .3. Figure 2. Schematic Diagram of Electrical Apparatus. 1.5V 32 V. Electrical Measurements: The resistance measurements folowed the procedure outlined for BLM's described by Tien(63). A diagram 0f the set-up can be seen in Figure 2. The resistance was calculated from the following equation: . Rm - Em R1 1- where Ei=50mV in applied potential, Em=recorded or meas- ured potential across the membrane, and Ri was the series resistance whose value was varied between 105 and 1040JL. until Em was approximately half of Bi or 25mV. VI. Other Materials and Methods: Complement-fixation saline(veronal-saline) was pure chased from BEL or prepared according to the following recipe: 0.145M NaCl, 0.15mM CaCl2, 0.5mM MgCl , and 4.9mM 2 sodium barbital at pH 7.4. Rabbit anti-sheep erythrocyte serum(RAS) and guinea pig complement(GPC) were purchased from Colorado Serum Company. Rabbit complement was purchased from Microbio- logical Associates. Human serum was collected from the author. The Kolmer-Wassermann cardiolipin Ag and positive syphilitic serum were purchased from Sigma Company. The temperature was controlled, usually at 372100 for the millipore membranes, by means of a water bath on a hot plate. Chapter 3 Results and Discussion I. Introduction: Chapter 2 illustrated the fact that C activity has not been sufficiently demonstrated on BLM's. The reason for this is probably due to the limitation of the ELM system, i.e., its lack of stability to the addition of proteins in the aqueous phase. However, the most direct approach to the activation of C on artificial membranes was not taken on planar BLM's. The best method for acti- vating C to attack a membrane is to have an Ag in the membrane. This was done with liposomes when Haxby et al (18) used a sheep erythrocyte lipid extract containing Forssman Ag. C activity was also observed on liposomes containing purified Forssman Ag(20) and cardiolipin(23) ad the lipid haptens. It was therefore logical for this author to try to activate C on BLM's made from 1:1 sheep erythrocyte lipid extract:ph03phatidyl choline or purified Forssman Ag and cardiolipin 1:100 in 1:1 phosphatidyl choline:cholesterol. The above ratios are molar ratios and the lipids were dissolved 1% w/v in n-octane. Antiserum and complement were then added to the bathing solution up to a serum 33 34 concentration of 1:2000. No complement specific reductions in resistance were found. Nonspecific reductions in re- sistance and membrane rupture were occasionally observed, but they were as common in the heat-inactivated and with- out-Ag controls as in the experiment. Above a serum con- centration of 1:2000 the BLM's were unstable and ruptured quickly or would not go black. A more stable membrane system was therefore required so that serum concentra- tions equivalent to those used on liposomes could be employed on a planar artificial membrane system, where electrical measurements could be made to look for possible channel formation. II. Glucose Release from Liposomes Formed from Sheep Erythrocyte Lipid Extract: In order to confirm the validity of the liposome work and the quality of my own reagents, this author prepared liposomes from a sheep erythrocyte lipid extract as described in Chapter 2. Rabbit and human serum were used as the sOurce of complement in concentrations sim- ilar to that in some of the millipore membrane experiments described below. Figure 3 Show the increase in absorbance at 340nm, corresponding to an increase in NADH, with time as glucose leaks out of the liposomes attacked by Ab and C. The glucose is phosphorylated by the hexokinase and ATP I 1 present in the external solution, and the resulting glu- cose-6-phosphate is oxidized while NAD+ is reduced by 35 glucose-6phosphate dehydrogenase, also present in the external solution. The reduction of MAD+ is measured as an increase in NADH which has an absorption peak at 340nm. The glucose assay reagent is described in Chapter 2. A standard 2ml cuvette contained 10X liposomes, 200X rab- bit 0, 150).human c, 150ARAS, 500>.veronal-salinc buffered at pH 7.4, and 1ml glucose assay reagent. The blank or reference cuvette run in the Eeckman spectrOphotometer contained the same ingredients except the sera were heat- inactivated at 56°C for 1 hour before use. Curve A.had only half the concentration of serum used in the other curves. That is why the shape of the curve is sigmoidal. It in- dicates a multiprocess event, i.e., the enzymatic cascade of C activation. Curve D had twice the amount of liposomes, 20), , and as a result curve D Shows the fastest rise in NADH production. These results are substantially the same as those reported by others(18,20,40). III. Sheep Erythrocyte Lipid-impregnated Millipore Filters: Millipore membranes prepared according to the recipe in Chapter 2 were found to be stable to the high concentra- tions of serum required for,C activation. Table 2 shows the results for ten membranes formed with 3002\RAS, 500A rabbit o, and 500).human c in each of the 2mls of bathing solution on each side of the membrane. the remainder of . the bathing solution was veronal-saline at pH 7.4. The temperature was maintained at 37: 1°C. Channel formation, 36 9 lb l w 3 4 5 6 Time (minutes) u- q- u- l g 1 mai- Figure 3. Glucose Release from Liposomes. 37 Table 2. Sheep Membranes with 300k RAS + 500}. Rabbit o e 500 )(Human C + 700) Veronal-saline on Each Side. Membranes \OQQO‘U‘I-PWN-S .4 0 average Time Lag 17 mins. 5 n 31 " 14 " 5 n 33 " 2 n 30 " 75 " 1 u 21‘ n LEE. 1 1.9x108 7.2 " 3.5 " 4.5 " 4.5 " 6.2x107 4.0x1o8 4.6x107 9.0 " 1.3x108 Rmfcflo 1.0X10 5.5 II 1.4 " 6 1.0 " 1.0 " 1.1 " 4.0 " 1.0x105 2.1 " 1.7 " 2.9 " 1.5x106 Duration 12 mins. 5 n 50 " 10 " 4.5 " 37 " 9 u 17 " 62 " 29 " 24 " l N 0.5 1.7 0.1 0.8 2.9 0.1 0.7 0.4 0.1 0.3 0.8 Chans 16 23 16 6 17 20 8 9 13 G 0.2 0.2 0.2 0.7 0.3 0.1 0.4 0.6 0.2 0.3 Table 3. Sheep Membranes with Heat-inactivated Serum(300)t RAS, 500>tRabbit c, 500>.Human 0) on Each Side. Membranes Rm;(Jl) 1 2 3 average 5.6x’10‘7 1.5x108 3.0x107 7.9x107 11 5.6x107 8 Rm 1.9x10 3.0x107 9.2x107 Duration 120 mins. 120 " 120 " 120 " 38 i.e., step-wise reductions in resistance, was seen in all ten membranes and is presumed to be responsible for all the resistance decreases observed on millipore mem- branes reported in this thesis. Only channels for which the resistance changes could be measured accurately are listed in the tables of this thesis, however. There was usually some time lag before the channel formation was observed, and this is listed in the tables. Also listed is the duration of the response, or the time over which the channel formation was observed, and the in- itial membrane resistance before channel formation(Rmi) and the resistance of the membrane after it stabilized at some new value(Rmf). The resistance was measured con- tinuously by recording the potential across the membrane on chart paper. The series resistance, Ri, was adjusted periodically to keep the membrane potential at approx- imately half of the applied potential. I The rate of the reaction for each membrane, K, was calculated as the fractional change in membrane resistance per minute, i.e., Ksantilog(Log(Rmi/Rmf)/duration)-1. The average conductance increase per channel, G, was calculate ed in a similar manner: G=antilog(Log(Rmi/Rmc/#channels)-1, where the #channels is the number that could be measured accurately before the resistance changes became too small, and Rmc is the membrane resistance after those channels were formed in the membrane. Sometimes Rmc corresponded to Rmf, but usually Rm was less than Rmc. f The meaning of K and G is made clear by looking at 39 a few examples from Table 2. Membrane 1 has a.reaction rate, K, of 0.5. This means that in one minute of the C reaponse(indicated by channel formation), the conduc- tance(which is the reciprocal 0f the reSistance).increased by 50% on the average. Likewise, the average conductance increase per channel, G, for the sixteen channels measured accurately for membrane 1 was 0.2. This indicates that the average channel in that membrane increased the con- ductance by 20%. Similarly, the average reaction rate for the ten membranes in Table 2 was 0.8, which means that the conductance increased by 80% per minute on the average during a response. The average conductance in- crease per channel for the ten membranes was 0.3 or 30%. The channel formation is shown to be due to C activ» ity by control membranes. Table 3 shows three sheep ery- throcyte lipid-impregnated millipore membranes under the same conditions as the membranes in Table 2 except that the sera were heated at 56°C for one hour before use. No reductions in resiStance were seen under these circum- stances. No step-wise or significant changes in resist- ance were observed either if there was no Ag in the mem- brane to activate the C. These results are indicated in Table 4 which lists five membranes made from 1:1M Phos- phatidyl choline:cholesterol without Ag formed in the presence of 300); RAS, 500), rabbit c, 500), human 0 on each side of the membrane. These controls clearly demonstrate the fact that C has been activated on millipore membranes 40 Table 4. 1:1 Phosphatidyl Choline:Cholesterol(no Ag) Mem- branes with 300)RAS, soolaabbit c, 500AHuman c on Each Side. Membranes Rm.(JL) 1 2 3 4 5 average 2.3x108 1.7 " 3.1 " 2.1 " 2.1x107 1.9x108 RmEQL) 7.8x108 1.6 " 3.1 " 8.5x107 1.8 " 2.7x108 Duration 120 mins. '1 130 mins. 31 5 II 161 " Table 5. Sheep Membranes with RAS, Rabbit c, and Human 0 Added to Each Side. Membranes Time Lag‘ 75ARAS 2504Rab 2 SOXHum 2OOARAS 250hRab 250AHum 350KRAS 400hRab 3OOAHum average 21 mins. 100 mins o 5 MinSo 'Rm.qu 2.1x107 5.2x107 2oOX1O 8 RmECflQ 1.0x105 19 mins..0.3 12 2.0x105 40 mins. 0.1 15 1.0xflO Duration _I_(_ gChans. §_ 6 5 mins. 1.9 15 0.3 3mm 41 Rum) 4- 105 : : p , , , 1o 20 3'0 46 50 ‘60 Time (minutes) Figure-4.A) Sheep Membrane(#8 from Table2) showing C Channels. E) Heat-inactivated Control from Table 3(membrane #2). 42 made from sheep erythrocyte lipid extract containing Forssman Ag. The C is presumably activated on the surface of the membrane by the Ag-Ab complexes which form there. This activation is via the classical pathway, and is the first demonstration of such activation on a planar arti- ficial membrane system. The attack of C on these membranes is seen as step-wise reductions in resistance which pro- bably indicate the formation of stable transmembrane chan- nels in the membrane. A calculation Of an average hole size and consideration of the various hypotheses of C action will be seen later in the thesis. Figure 4 shows membrane 8 from Table 2 illustrating the step-wise reductions in resistance corresponding to channel formation(A). Figure 4 also shows a heat-inacti- vated control membrane from Table 3(membrane 2) which shows no channel formation(E). In fact, the resistance in Figure 4B went up slightly. Other sheep membranes with RAS + rabbit C + human 0 added to both sides are presented in Table 5. The average conductance change per channel for these membranes was 30”. Table 6 shows membranes with RAS + rabbit C + human C on only one side..The average conductanceichange per channel for these membranes was 20%. There is variation not only in the average size of channels from one membrane to an- other, but there was wide variation in the size of indiva idual complement channels. This variation is probably de- pendent upon the number of C9 molecules incorporated into 43 Table 6. Sheep Membranes with RAS, Rabbit C, and Human C on One Side. Membranes Time Lag Hugh: JRmEM-J ‘Duratigij I; fiChans 9 3.118.. 322.9 2122.2 *4 15o)150)~ 150). 40 mins. 9.0x1071 6.1x104 10 mins. 1.1 150M150) 200). 104 " 4.0x’107 1.0x105 16 " 0.4 14 0.2 150%500). 500). 0.2 " 1.2::108 4.3x’lO6 25 " 0.1 11 0.5 200A500>~ 0 " 7.22:108 5.5x10“ 15 " 0.9 200).500>~ 500k 15 " 5.2x107 1.0x105 15 " 0.6 15 0.2 500)». 500). 500). 1 " 1.7x107 4.0x105 17 " 0.2 6 0.7 500). 500). 200). 10 " 2.1x107 5.6x106 51 " 0.05 15 0.1 500>u+00)~ 500). 6 " 5.2x107 1.6x105 58 " 0.2 4 0.7 500)» 500). 500)\ 0 " 1.0::108 1.0x105 5 " 5.0 550A 200A 250M 1 '1 5.6x10 1.9x104 59 n 0.2 10 0.1 avwerage 1 0.2 4 Table 7. Sheep Membranes with 200).RAS and 600)\GP0 on Each Side. 1 average _Membranes Time Lag 1 1 mins. 2 11 " 5 2.5 " 4 1 4 S 5.7 " 6 19 " 7 5 II 8 ' 5 " 9 2 II 10 0 " S I! 1 FML (l 4.6x107 1.6x108 5.2x107 4.0 n 7.2 u 4.0 N 9.0 N 1.5x10 9.0x’107 5.5x1o6 7.5x107 8 Ihn Jt 1.0x105 6.4 " 1.0 " 1.0 " 8.6 " 4.2x10 1.9x10 6.1x10 1.5x10 2.1x10 6.6x10 JDuration 9 N 6 5 4 5 5 ll .p n3.4 n).# U1-P n>xn 3 mins. _I_{_ fiChans. fl 9 5079 24 0.3 0.85 19 003 2.78 34 0.2 19.0 9 0.7 2.2 13 0.2 0.57 5 0.2 0.48 45.2 599 4.02 69.4 0.5 44 the "hydrophobic doughnut" of the transmembrane pore(59). The data in Tables 5 and 6 will be considered again in the section on concentration studies. The variation exhibited in the reaction rates of membranes studied under the same conditions, as in Tables 2 and 7, precluded meaningful quantitative concentration studies, however. The combination of rabbit C and human C was used only because at first the best results were obtained with that system. However, when fresh frozen guinea pig com- plement(GPC) as Opposed to lyophilized GPO was used very good results were obtained. This is also the system of choice because RAS and CFO are the standard reagents in the C-fixation assay. Table 7 shows the results for ten sheep millipore membranes incubated at 57°C in the presence of 200).RAS, 600). CFO, and 1200) veronal—saline on each side of the membranes. There is wide variation in the results al- though the conditions for each membrane were identical. The time lag for the channel formation to begin varied from 0 to 19 minutes. The duration of the response varied from 1 to 5 minutes. The average conductance increase per channel was 30%. Enormous variation was seen in the reac- tion rates. The reaction rates varied from 0.48 min-1 to 599 min"1 tative concentration or temperature studies. The impor- . This kind of data obviously precludes quanti- tance of these experiments is that C channels are observed and a calculation of the average pore size can be made. which can help in the determination of the correct hypo- 45 thesis concerning the cytolytic mechanism of 0. As with the rabbit plus human C experiments, were run to verify that the channels observed with GPC were due to C activity. Table 8 shows three membranes formed under identical conditions as in Table 7 except that the RAS and CFO were heat-inactivated at 56°C for 1 hour before use. Table 9 shows three membranes made from 1:1 phosphatidyl choline:cholesterol without Ag in the presence of 2OOXRAS and 600)\GPC on each side of the membranes. Ione of the controls showed any channel formation or significant reductions in resistance. Figure 5. shows the dramatic resistance change and channel production in membrane 4 from Table 7 compared to two control membranes, membrane 3 of Table 8 and mem- brane 3 of Table 9. Table 10 shows sheep membranes with other concentra- tions of RAS and CFO on each side. These two membranes have an average conductance increase per channel of 50%. Table 11 shows eight membranes formed from sheep erythrocyte lipid extract with RAS and CFO on only one side of the membranes. The average conductance increase per channel for these membranes was also 30%. The data from Tables 10 and 11 will be combined with that from Tables 5 and 6 to make up the concentration studies. 46 Table 8. Sheep Membranes with Heat-inactivated Serum (200).RAS and 600 AGPC) on Each Side. Membranes Rm.(lL2 1 1.6x108 2 7.2x107 3 7.8x107 Rmsgltz 1.5x108 5.6x107 4.5x107 Duration 120 mins. Table 9. 1:1 Phosphatidyl Choline:Cholesterol Membranes without Ag But 200). RAS + 600,\ GPC on Each Side. Membranes Rm.glt2 1 2.1x107 2 1.4x107 5 5.6x107 Rm (IL 2.8x107 1.5x107 1.9x107 Duration 1 20 mins 0 fl Table 10. Sheep Membranes with RAS + GPO on Both Sides. Membranes FTime Lag BA§ 2132 150X100). 20 mins. 100).200)s 18 " average Rm FEE 1.1x106 4.0x10 6 Duration IN 59 mins 0 .0.1 0.1 140 .. 1 1 kflChan 15 '19 m) 0.4 0.2 0.3 R 60-) 47 DD (3 b 107 #1_1 illil' l 1 J 4 - I41 “05 : a : 2 4 6 Time (minutes) Figure 5.A) Membrane 4 from Table 7 showing 0 Activity 8 10 with GPC. B) Without Ag control membrane from Table 9- C) Heat-inactivated control membrane from Table 8. 48 Table 11. Sheep Membranes with RAS + GPS on One Side. Membranes {Time Lag 12:11.91) ngéfi Duration] .I_{_ Chans Q RAS 020 ' 40). 2003 120 mins. 4.0x107 1.0x105 56 mins. 0.1 4 0.8 50). 250)» 14 " 4.0 " 5.0::106 52 " 0.1 4 0.7 350) 1ml 2 n 2.5 " 1.6x105 6 " 0.6 4 0.2 150A1m1 15 n 6.2 " 1.0 u 40 " 0.2 12 0.4 200A400). 8 " 7.2 " 5.2::106 52 n 0.1 16 0.2 200). 400A 20 " 5.2 " 2.5 " 50 " 0.1 500). 600)\ 22 " 7.5 " 1.6 " 4 1' 1.6 6 0.9 400)» 500) 26 " 5.2 " 1.0 " 20 " 0.5 52 0.2 average 0.3 Table 12. 1:1 Phosphatidyl Choline:Cholesterol Membranes with 1:100 Forssman Ag(GL5) and 250ARAS + 500x020 on Each Side. Membranes Time Lag Rm.alz [Rmell Duration ,5 fiChans. Q 1 25 mins. 2.6x107 4.7x106 7 mins. 0.5 9 0.2 2 5 " 2.6 " 7.2 " 18 " 0.1 . 12 0.1 3 5 " 1.5 " 1.9x105 55 " 0.1 21 0.2 4 0 N 2.6 I! 8.5 fl 2 II 4.5 S 8 " 2.1 " 5.5 " 22 " 0.2 6 I 4 n 1.9 " 2.6 " 66 " 0.1 average ' ‘ , 0.2 heat-inactiv. sontr. 2.1x107 «9:107 510 " -' 49 IV. Millipore Membranes with Purified ForsSman Ag(GLS): Membranes were formed with purified GL5 from canine small intestine 1:100 in 1:1 Phosphatidyl choline:chol- esterol in to verify that a single glycolipid antigen incorporated into a millipore membrane could activate C and cause channel formation. The purified GL5 is the same Ag present in the sheep erythrocyte lipid extract. The results of six millipore membranes with 250XRAS and 500). GPC on each side and one heat-inactivated control membrane are listed in Table 12. The resistance changes of the six membranes with fresh serum varied from 10 to 450% per minute for the responses. The average conductance increase per channel was 30%. The important finding of this data is that complement channel formation can be demonstrated on the millipore membrane system for a single purified glyco- lipid antigen. Figure 6.shows membrane 3 from Table 12 along with the the heat-inactivated control membrane which contained the same amount of Ab and C as the other membranes except that the sera were heated at 56°C for one hour before use. V. Concentration Studies: The variability of the complement response prevented careful studies of the effect of various concentrations of Ab and C, but the general effect of increasing serum concentration can be seen in Figure 7. The data was com- piled from Tables 5,610, and 11. A slope of +0.00125 was 50 108 B 1hr 107 __ 1; .L q. 1- Rmm) .- ~1- 106 __ q- d. 4r .L 105 t J. 4 5 4 10 20 50 40 50 Time(minutes) Figure 6. Membrane 5(A) and contr01(B) from Table 12. 50 108 __ B 107 __ 1: db A qu- Rmm) up -1- 6 10 .— .1." a. d- l .41. 105 t ' : : a 4 1O 20 30 40 5O Time(minutes) Figure 6. Membrane 3(A) and contr01(B) from Table 12. 51 2.0 ‘- qb K(min'1) db 1.0 .- q» 41- db 0-0 J 1 ? : .L : . . q 1ml 2m1 (Total serum) Figure 7. Reaction Rate Versus Total Serum Concentration. 52 obtained for a least squares line, and the correlation coeffi- Icient 0.72 for the data. The X-intercept was 460X.of serum for a reaction rate of zero, suggesting that a minimal threshold of serum concentration might be required for the reaction to go. However, a linear regression line can not be assumed. A more reasonable shape to be expected would be a sigmoidal curve since the enZymatic cascde of C activation is a multiprocess event. The curve in Figure 7 says nothing about the one-hit theory of C lysis. Lysis is not even assayed by the millipore technique. Only the trans membrane channels which presumably lead to C lysis are seen in the millipore membranes. The rate of reaction that is plotted in Figure 7 is a logarithmic function of the number of C channels introduced into the membrane. Therefore, Figure 7 is a plot of the logarithm of the number of channels produced in the membrane per minute versus the total serum concentration in the bathing solution. This curve shows a moderate to good cor— relation between the number of channels formed in the membrane per minute and the serum concentration. This helps to confirm the fact that the channels are really due to C activity. VI. Temperature Studies: The effect of increasing the temperature on the rate of a complement reaction already in progress can be seen in Figure 8. The temperature was increased from 57°C to 50°C within 2 min- utes and a clear change in the reaction rate can be seen. There was a 6-fold change in the reaction rate for a 13°C temperature 53 q- 107 __ J. J. .J. Rmm) {b '1' JL 3 wb 105 A ~ :e—ee : : : - at 20 4O 6O 80 100 Time(minutes) Figure 8. Membrane 7 from Table 6 Showing the Effect of Increasing the Temperature to 50°C. 54 rise. This corresponds to a doubling of the reaction rate for a 5°C increase in temperature. Extensive quantitative temperature studies were out of the question given the variability of the C reponse on the millipore membrane system. The only other temperature study conducted consisted of three membranes run under the conditions as those in Table 2 except for the change in ' temperature. The data is listed in Table 15. One membrane was run at 50°C, one at 57°C, and one at 44°C. Over the 14°C rise in temperature the reaction rate increased 7-fold from 0.1 to 0.7. Again,a doubling of the reaction rate was seen for a 5°C rise in temperature. A semilog plot of the three reaction rates(expressed in sec-1) versus 1000/T is seen in Figure 9. From the slope of this curve the energy of activation(Ea) was calculated to be 0.506 Kcal/mole. This number is the total energy of activation and attack of C on the membrane. It is imposs- ible to say what the rate-limiting step is. It could even be the thinning of the millipore membranes. VII. Activation of C by the Cardiolipin Ag-Syphilitic Ab System: ‘ Cardiolipin Ag, a lipid hapten from beef heart, is of interest to test on the millipore membrane system because of its clinical usefulness in the Wassermann C-fixation test for syphilis. Millipore membranes were formed from a 1:1 mixture of Massermann Ag(containing 1:100 cardiolipin 55 Table 15. Temperature Study of Three Sheep Membranes with 500) RAS + 500} Rabbit C + 500), Human C on Each Side. Membrane 50°C 57°C 4490 Time Lag 40 mins. 1 n 5 n {Rmi@fl) 1.0x107 1.5x108 7.2x107 Ihn 11 1.2x106 1.7x105 Duration 25 mins 0 29 “ 4.0x1051 10 " .E 0.1 0.5 0.7 22 Charis o E 11 0.2 Table 14. Cardiolipin Membranes with 200)Syphi1io Ab + 500) GPC on Each Side. Membrane Time Lag RmiUl) [ngéfli [Duration' 1 21 mins. 2.1x107 1.0x10S 19 mins. 2 22 " 9.4 , 1.0 " 55 u 3 40 " 5.5 " 1.4 " 9O " 4 25 " 6.2 " 1.9x10“ 70 " 5 20 " 1.0x108 1.8x105 25 " average 26 " 6.6x107 1.1x105 47 " heat—inact. contr. 6.2x107 6.2x107 180 " Table 15. 1:1 Phosphatidyl Choline:Cholesterol Membranes .5 0.5 0.2 0.1 0.1 0.5 0.2 fiChans 23 km 0.2 0.5 0.5 Without Ag But with Previously Activated Ab + C on Each Side. Membrane Time Lag 1 15 mins. 2 5 " 5 . 1O " average 10 " FmL 1L 7.2x107 5.2 " 5.5x108 1.6 " ’Rmsm) 1.1x105 5.2 " 6.2 " 4.2 " Duration 9 mins. 25 II 16 " >17 II we 1.0 0.2 0.5 0.6 .11 mChans 18 13 24 km 0.3 0.2 0.2 0.2 56 ‘2? 1r q. .3 J. i 0.01 -— 1. 7. 4. K(sec'4),. 1 l q- t .L 24 ~~ e : 5.1 5.2 5.5 1000/T(°K) Figure 9. Arrhenius Plot of Temperature Data. 57 with phosphatidyl choline and cholesterol) and phosphatidyl choline. Table 14 shows the results when 200) syphilitic Ab and 500 >\GPC were added to the bathing solution. Complement channels were observed in all five membranes with fresh serum, but no channels were observed in the heat-inactivated sera control membrane. These results clearly indicate.that cardio- lipin can be used in millipore membranes to activate C, and that the millipore technique can be used to detect syphilitic serum in a more direct way than the C-fixation test. Figure 10 shows membrane 2 and the heat-inactivated control membrane from Table 14. A current-voltage plot was made of membrane 5 from Table 14 after the C channels were inserted. This was done to de- termine the voltage-dependency of the 0 channels once they were in the membrane. Figure 11 shows the I/V plot which is ohmic indicating that the transmembrane channels of C are voltage-independent. This is in agreement with the re- . sults of Michaels et al(12) on BLM. The correlation coeffi- cient for the data was 0.999 indicating an almost perfect fit. VIII. Reactive Lysis: It was important for these studies to explore the poss- bility of observing channel formation in the millipore sys- tem without activating C on the membrane surface. Membrane damage due to the activation of C in the aqueous phase is known as "reactive lysis". This was attempted on the millipore system by three separate methods: 1) C was activated on one 58 '-1O .— l rVVTfivl Rmcna 4 14 (4414' rI-Wl L. A. A l —' j I 10 20 . 50 40 50 1r 4% Time(minutes) Figure 10. A) Membrane 2 from Table 14 showing C channels on a cardiolipin membrane. B) Heat-inactivated control. 59 Figure 11. I/V Plot of Membrane 5 From Table 14 After the C Channels Were Formed in the Membrane. 6O millipore membrane and then transferred to another. 2) C was activated via the alternative pathway. 5) C was activated directly with trypsin which cleaves C5(64). Table 15 shows three 1:1 phosphatidyl choline mem- branes with 500XRAS + 500%Rabbit C + 500). Human. c on each side which was transferred within 10 minutes from a sheep membrane with demonstrated C channels. Mem- brane one had its serum taken from membrane 2 of Table“2; membrane 2 had its serum taken from membrane 5 of Table 2; and membrane 5 had its serum taken from membrane 7 of Table 2. All three membranes demonstrate dramatic resist- ance decreases due to channel formation. Table 16 shows that when 500).GPc is added to 1:1 phosphatidyl choline:cholesterol membranes in the presence of 2.5mg of zymosan A(from yeast cell wall extract) or 100)\of 1% trypsin on one side of the membrane the resist- ance is significantly reduced in a step-wise fashion. All three methods of activating C away from the mem- brane surface employed here clearly demonstrate that a soluble lytic factor is generated which is capable of diffusing to the membrane surface to cause disruption. This factor is presumably C5b,6(39). The heat-inactivated controls in Table 16 confirm that the resistance changes were due to C. Figure 12 shows the 0 channels seen in membrane 2 of Table 15. 61 Table 16. Reactive Lysis Demonstrated on 1:1 Phosphatidyl Choline:Cholesterol Membranes with Either Zymosan A or Trypsin.on One Membrane 2.5mg zymosan A .heat-inact. cntr 100 1% trypsin heat-inact. cntr Table 17. Millipore Filters with Larger Membrane 2 from Table 6 2 I! ll N-‘NCDUI average 10 11 11 12 14 14 500)GPC Activated with Pore Size 0.45 um 0.8 " 0.8 " 0.45 " 0.45 " 0.45 " 0.45 '" Side. Time lag Rmiaj) RmfGO) Duration 20 mins. 4.4x106 5.4x105 70 mins. 50 " 2.6 " ' 2.4 " 80 " . 5.2 " 5.6x106 240 " 20 mins. 5.4x107 1.0x105 140 " . 1.1x108 9.0x1071240 " Pore Sizes. fiChannels ,Q 14 0.2 19 '0.2 16 0.2 52 0.2 12 0.1 7 0.2 25 0.3 0.2 "N 0.04 0.03 0.04 62 10 107 j lJlJl ‘1 'ff' Ran9- 10 L 4 l L' ‘5? 10 15 20 25 Time(minutes) Figure 12. Membrane 2 from Table 15 Showing Reactive Lysis. 63 IX. Calculation of the Size of the Transmembrane Channel: The average conductance increase per channel for all the channels in this thesis(612) was 28%. This means Rmi/Rmfa1.28, where Rmi= the resistance before the insertion of the average channel and Rmf= the resistance after the insertion of that channel. For the first channel intro- . duced into the membrane, Rmi=ylxlm/A where 71: resistivity of sheep erythrocyte lipid extract and lm= thickness of membrane=100A and A: area of membrane=5.14(0.55cm)2=0.585cm2 and 1/Rmf=(A-a)4rllm + aéfslm where a= area of channel andysz resistivity of salt or serum solution. Therefore, Rmi/Rmf-1-a/A+afl/A,S or O.28=a/A(,1/’S-1) or a=O.28A/Spl/’S-1) But fl/ys=Rmixle/Rsxlm where le= distance between electrodes and Rs: resistance measured between electrodes when millipore filter was not coated with lipid. Rmi(ave. of 60 membranes) =1.7x108 4 M and Rs=1.9x10 and le=6cm, thereforeyl{PS=1.1x10 and a=9.8x10-15cm2 and r(radius of the channel)=56A and the diameter of the channel-112A. This is in excellent agreement with the 100A holes seen in negative stain elec- tron micrographs of sheep erythrocyte ghost membranes at- tacked by complement(1). X. Evidence in Favor of the Transmembrane Channel Hypothesis: For certain membranes in this thesis a larger millipore filter size than 0.22um was used. The membranes listed in Table 17 employed millipore filters of 0.45 or 0.8um. The purpose of varying the pore size was to see if larger channels 64 were formed. If larger channels were observed it might indicate that miniature membranes formed in parallel in the millipore membrane system were being ruptured by the action of C. This would be evidence in favor of a deter- gent theory of C action as opposed to the "hydrOphobic doughnut" or transmembrane channel hypothesis of Mayer(2). The seven membranes in Table 17 have an average _ conductance increase per channel of 20%. This is less than that found for the average channel on the 0.22um millipore filters. This is strong evidence in favor of the transmembrane hypothesis which says that stable pro- tein pores are inserted in the membrane. Two other experiments were carried out to try to distinguish between the two main hypotheses of 0 action. In one experiment, the light scattering of liposomes(the same as those used for the glucose release assay) exposed to heat-inactivated RAS + GPC was measured as absorption in a Beckman spectrophotometer. A complete absorption spectrum was made from 250 to 700nm. No change in this spectrum was seen when fresh Ab and C was added under conditions that cause release of glucose. This probably indicates that the liposomes were not being destroyed or dispersed as they might be by detergent. Finally, octadecylamine was added to the bathing solution as a detergent to see if it could induce chan- nel formation. It was found not to. Either high resistance membranes were formed or no membranes were formed. There was no intermediate state of individual step-wise resist- ance decreases indicating channel formation. 65 These three pieces of evidence may be taken as strong evidence in favor of the transmembrane channel hypothesis. The pore size of 112A is also evidence in favor of this hypothesis since it is about 20x less than the pore size of the 0.22um millipore filters. This is further evidence against the detergent theory which should predict rupture of the miniature BLM's presumably formed in parallel in - the millipore membrane system. XI. How does the Transmembrane Channel Lyse the Cell? Green, Barrow, and Goldberg(47) showed that cells damaged by Ab and C exhibited rapid loss of potassium and influx of sodium, followed shortly by loss of amino acids and ribonucleotides. However, macromolecular loss (RNA and Protein), cell swelling, and lysis were pre- vented if the cells were osmotically protected by media containing high concentrations of macromolecules, suCh as albumin. The interpretation of this work was that the primary or direct complement-produced membrane lesion was of limited size, and that lysis followed only if secondary osmotic water influx and swelling were permitted to occur. Valet and Opferkuch(5) have distinguished three steps in the lysis of sheep erythrocytes:1)Fixation of C9 to EAC1-8 cells(sheep erythrocytes with Ab and C1-8 bound to their surfaces), a process which can proceed at 0°C; 2) A temperature-dependent non-osmotic swelling of the EAC1-9 cells; and 5) osmotic lysis of the cells. Lauf(65) has 66 found that complement lysis can proceed without a classic 4colloid osmotic swelling phase. In view of these findings, an additional force for the lysis of cells will be proposed in this thesis as follows. The formation of the transmembrane channel, an energy- requiring, temperature-dependent process(42), results in a rapid loss of membrane potential(4). This change in the - electric potential across the cell membrane may reSult in a reduction of the interfacial tensionC?) of the cell mem- T,P,u I brane according to the Lippman equation:(33) -Q "if where I = interfacial tension of the cell membrane,g is the electric potential across the membrane and Q/A is the surface charge density of the membrane. A reduction in the interfacial tension due to a change in the membrane potential would result in a non-osmotic swelling of the cell membrane. The proposal can be visualized in the following way. C inserts itself into the membrane. A transmembrane chan- nel is formed. Normally the inside of the cell is negative, and the transmembrane channel eliminates this electric po— tential, which is the equivalent of an influx of Na+ ions. These counterions cause the inside of the cell membrane to become overly positively charged, resulting in a repul- sion of the lipid molecules from one another, i.e., the interfacial tension(‘) is reduced. This movement of the lipid molecules away from one another is manifest as a non-osmotic swelling of the Cell. This may then increase 67 the cell‘s susceptibility to osmotic lysis by making the cell membrane more leaky to small ions and molecules(5). Chapter 4 Summary and Conclusions This thesis has conclusively demonstrated for the first time that it is possible to observe the formation of discrete channels with time in an artificial planar membrane system due to C activated at the membrane surface by Ag-Ab reactions or in solution by zymosan A or trypsin. The uniqueness of the system results from the use of lipid Ags, Forssman Ag and cardiolipin Ag, to activate C via the classical pathway on the membrane surface. Complement was also activated via the alternative pathway by zymosan A and directly by the cleavage of 05 by trypsin. Lipid-impregnated millipore filters were made by coat- ing millipore filters with a lipid solution in n-octane containing Forssman Ag or cardiolipin. Forssman Ag was iso- lated either as a crude lipid extract from sheep erythro- cytes or in purified form from canine small intestine. Cardiolipin was purchased as Wassermann Ag in a mixture with phosphatidyl choline and cholesterol. When these millipore membranes were incubated in a bathing solution containing the appropriate antiserum and complement dis- crete step-wise reductions in resistance were observed with time indicating the formation of transmembrane chan- nels. These channels were shown to be the result of 0 action by control membranes with heat-inactivated sera 68 69 which showed no step-wise reductions in resistance. The lipid Ags were shown to be required in the classical path- way by control membranes without Ag. These membranes dem- onstrated no channel formation. A soluble lytic substance was shown to be produced through the classical pathway by transferring the antiserum and C from a membrane with Ag to one without Ag. Complement channels were observed in these membranes. The soluble lytic factor is presumably C5b,C6 which when it combines with C7—C9 can generate a transmembrane channel(59). This type of "reactive lysis" was also demonstrated with zymosan A and trypsin and helps to confirm that the conductance channels observed are indeed due to the activation of 0. Concentration and temperature studies of the reaction rate were done indicating a general increase in the reac- tion rate with both increasing total serum concentration and temperature. This supports the notion of C activation. Quantitative studies were impossible due to the variability of the reaction rate under constant conditions. A calcu- lation of the total activation energy for the activation and attack of C on these membranes was made from an Arrhenius plot of the reaction rates with temperature. The total activation energy for the process was 0.506 kcal/mole. The reaction rate was found to double for a 5°C rise in temperature. The size of the 0 channel was found to be 112A in dia- meter. Since this is about 20x smaller than the 2200A pore size of the millipore filters this indicates that the C 70 channels are not due to the rupture of miniature BLM's formed in parallel on the 2200A pores of the millipore filters. The 112A electrical channels observed presumably represent the insertion of transmembrane protein channels across the membrane according to the "doughnut hypothesis". Further evidence for this hypothesis was obtained by varying the pore size of the millipore filters. The size _ of the C channels was found not to increase with the pore size of the millipore filters. This is evidence against the idea that miniature BLM's are being ruptured by the action of C as would be expected with the detergent theory of 0 attack. Also, no G channels were found in millipore membranes formed in the presence of octadecylamine, a water-soluble detergent, and the light scattering(measured as absorption) of liposomes attacked by Ab and 0 did not change. These two pieces of data also help to rule out the detergent mechanism of 0 action, and help to confirm the transmembrane channel hypothesis. The 112A pore size calculated from the observed elec- trical channels is in excellent agreement with the 100A holes seen in negative stain electron micrographs of sheep erythrocyte ghost membranes attacked by 0(1). It is:alSO in good agreement with the pore size calculated from liposome studies. Raman and Mayer(58) reported a pore size of 55A based on ovalbumin and hemoglobin release from liposomes, and not more than 150A, since serum albumin was not re- leased. It is thought that the transmembrane pore of C results 71 in the lysis of a cell by allowing for the diffusion of small molecules and water into the cell, but not the re- lease of macromolecular components from inside the cell, resulting in an osmotic swelling of the cell(59). This has been confirmed by Green, Barrow, and Goldberg(47) who blocked the lysis of damaged cells with high concentra- tions of macromolecules, like serum albumin, in the extra- cellular medium. Valet and Opferkuch(5), however, have noted a non-osmotic swelling in addition to the osmotic lysis of sheep erythrocytes. A possible explanation for this non-osmotic swelling is promulgated in this thesis. It is pr0posed that the C lesion eliminates the transmem— brane potential which might result in a reduction'in.the interfacial tension of the cell membrane according to the Lippman equation. The reduced attraction among the lipid molecules would account for the non-osmotic spreading. The elimination of the transmembrane electric potential by C has been measured by microelectrode techniques in cultured cells(4). Stephens and Henkart(4) also report that ouabain by inhibiting the Na-K ATPase inhibits the recovery theSe cells would normally experience after washing the C away. This indicates that the Na-K ATPase may be responsible for maintaining the electric potential in order to keep the cells viable. It seems less likely that the Na-K ATPase would be able to reStore an osmotio' imbalance. It also seems implausible that one C lesion (capable of lysing a cell according to the one-hit theory) 72 would be capable of allowing the passage of enough mole- cules to establish an osmotic gradient in the time that lysis occurs(especially for erythrocytes where lysis occurs in a few minutes). It, therefore, seems much more likely that there is an electrical phenomena responsible for the initial swelling which preceeds the osmotic lysis. This noneosmotic swelling probably increases the leakiness of the cell membrane and precipitates the osmotic event. An I/V plot of the C channels revealed that they are voltage independent. This is in agreement with previous results(12). The C channels were also observed to be stable and long-lasting. This is also in agreement with other reports(58). The importance of the data in this thesis is that it helps to confirm the "hydrophobic doughnut theory" of C lysis, it estimates the size of the transmembrane channels, and it presents a new research tool for the detection of lipid. A33 0 Bibliography 1. 2. 5. 4. 5. 6. 7. 8. 9. 10. 11. 12. 15. 14. 15. 16. 17. Bibliography Humphre , J.H. and Dourmashkin, R.R., Adv. Immunol. y Mayer, M.M., Proc. Natl. Acad. Sci. USA 69, 2954 (1972)- Valets G. and Opferkuch, J., J. of Immunol. 115, 1028 1975 . Stephens, C.L. and Henkart, Fed. Proc. 58, 1468 (1979). del Castillo, J., Rodriguez, A., Romero, C.A., Sanchez, J,, Science 15 , 185 (1966). Toro-Goyco, E., Rodriguez, A., and del Castillo, J., Biochem. Biophys. Res. Commun. 25, 541 (1966). Barfort, P., Arquilla, E.R., Vogelhut, P.O., Science Muller-Eberhard H.J., Ann. Rev. Biochem. 44, 697 (1975). Mueller, P. and Rudin, D.0. (1969) in Current Topics in Bioenergetics (Sanadi, D.r., ed.) Vol. 5. P.157. Wobschall, D. and McKeon, 0., Biochim. Biophys. Acta Pillemer, L., Ecker, E.E.,.Oncley, J.L., and Cohn, E.J., J. Exp. Med. 4, 297 (1941). Michaels, D.W., Abramovitz, A.S., Hammer, C.H., and Mayer, M.M., Proc. Natl. Acad. Sci. USA 25, 2852 (1976). Michaels, D.W., Abramovitz, A.S., Hammer, C.H., Fed. Proc. 55, 1762 (1976). Michaels, D.W., Abramovitz, A.S., Hammer, C.H., and Mayer, M.M., Biophys. J. 12, 82a (1977). Michaels, D.w. and Mayer, M.M., Biophys. J. 21, 125a (1978). Abramovitz, A.S., Hammer, C.H., and Mayer, M.M., Fed. Proc. 55, 1762 (1976). Mountz, J.D. and Tien, H.T., J. BIoenerget. and Biomem- branes 69, 1 (1979)- 75 18, 19. 20. 21. 22. . 23. 24. 25. 26. 27. 28. 29. 50- 51. 52. 53- 54. 55. 56. 74 Haxby, J.A., Kinsky, C.B., Kinsky, S.C., Proc. Natl. Acad. Sci. 61, 500 (1968). Kinsky, S.C. and Nicolotti, R.A., Ann. Rev. Biochem. £5. 49 (1977) Kinsky, S.C., Haxby, J.A., Zopf, D.A., Alving, C.R., Kinsky, C.B., Biochemistry 8, 4149 (1969). Haxby, J.A., Gotze, 0., Muller-Eberhard, H.J., Kinsky, S.C., Proc. Natl. Acad. Sci. USA 64, 290 (1969). Knudson, K.C., Bing, D.E., Kater, L., J. Immunol. 106, 258 (1971). Humphries, G. and McConnell, M.K., Proc. Natl. Acad. Sci. USA 2;, 2485 (1975). Lachmann P.J., Mann, E.A., Weissman, G., Immunology ‘19, 985 (1970). . Mold, 0. and Gewurz, H., J. Immunol. 124, 1552 (1980). Kinsk , S.C., Annals New York Acad. Sci. 508, 111 (1979i. Lewis, J.T. and McConnell, H.M., Annals New York Acad. SCio 208, 124 (1979). Hsia, J.C. and Tan, C.T., Annals New York Acad. Sci. 22§9,159 (1979). Richards, R.L., Alving, C.R., and Scher, I., Fed. Proc. 2Q. 1468 (1979). Henkart, P. and Blumenthal, R., Proc. Natl. Acad. Sci. USA 2, 2789 (1975). A Mayer, M.M., Hammer, C.H., Michaels, D.W. and Shin, M.L., Immunochemistry 15, 815,(1978). Mayer, M.M., Hammer, C.H., Michaels, D.W., and Shin, M.L., Transplantation Proceedings 19, 707 (1978). Mayer, M.M., Gately, MoKo, OkamOto, Mo, Shin, M.L., and Willoughby, J.B., Annals New York Acad. Sci, 552, 595 (1980). Wolf, D.E., Schlesinger, J., Elson, E.L-, Webb, W.W., Blumenthal, R., and Henkart, P., Biochemistry 16, 5476 1977 . Rosensteich, D.L. and Blumenthal, R., J. of Immunol. 118. 129 (1977). Crumpton, M.J., Allen, D., Auger, J., Green, M.M., and Naino, V.C., Phil. Trans. R. Soc. Lond. B 2 2, 175 (1975). 37- 58. 59. 4o. 41. 42. 45. 44. 45. 46. 47. 48. 49. 50- 51. 52. 53- 54. 75 Deleers, M. Poss, A., and Ruysschaert, J.M., Biochem. BiOphys. Res. Commun. 2, 709 (1976). Tosteson, N.T. Lau F. and Tosteson D.C. Nature 245 112 (1975). 9 $ 9 S 9 , Mayer, M.M., Harvey Lect, Ser. 22, 159 (1978). Mayer, M.M., J. Immunol. 112, 1195 (1977). Benacerrat, B. and Unanue, E.R..(1979) in Textbook of Immunology(Williams and Wilkins) Chapter 12, 218. Osler, A.G. (1976) Complement, Mechanisms and Functions (Prentice-Hall, Inc.). von Dungern, F. and Coca A., Muench. Med. Wodensohr. 2i, 2517 (1907). Kinsky, S.C., Bunsen, P.P.M., Kinsky, C.B., van Deegan, L.L.M., and Rosenthal, A.F., Biochim. BiOphys. Acta. 253. 815 (1971). Mayer, M.M., Proc. Natl. Acad. Sci. USA 69, 2954 (1972). Green, H., Fleischer, R.A., Barrow, P., and Goldberg, B., J.Exp. Med. 110, 699 (1959)- Green, H. Barrow, P., and Goldberg, B., J. Exp. Med. 110. 699 (1959). Mayer, M.M. (1961) in Immunochemical Approaches to Problems in Microbiology (M. Heidelberger and 0.J. Plescia, eds.) p. 268, Rutgers University Press, New Brunswick, New Jersey. Rommel, F.A. and Mayer, M.M., J. Immunol. 110, 657 (1973)- Kitamuraz H., Itakura, H., and Inai, S., Immunochemistry 129 771 1976). Iles, G.H., Seeman, P. Naylor, D., and Cinader, B., J. Cell Biol. 2g, 528 (1975). Inoue, K., Kinoshita T., 0kada, M., and Akiyama, Y., J. Immunol. 119, 65 (1977). Kinoshita, T., Inoue K., Okada, M., and Akiyama, Y., J. Immunol. 119, 75 (1977). . - Bhahdi, S., Bjerrum, 0.J., Bhakdi-Lehnen, B., and Tranum-Jensen, J., J. Immunol. 121, 2526 (1978). Shin, M.L., Pazneka, w.A., and Mayer, M.M., J. Immunol. 57- 58. 59. 60. 61. 62. 65. 64. 65. 76 A Nakanura, M., Ohnishi, 6., Kitamura, H., and Thai, D., Biochemistry 12, 4858 (1976). Giaveioni, E.B., Mason, R.P., and Dalmasso, A., J. Immunol. 120, 2005 (1978). Ramm, L.E. and Mayer, M.M., J. Immunol. 124, 2281 (1980). Boyle, M.D.P., Gee, A.P., and Borsos, T., J. Immunol. Stephens, C.L., Jackson, M.B., and Lecar, H., J. Immunol. 124, 1541 (1980). Tobias, J.M., Agin, D.P., Pawlowski, R., J. Gen. Physiol. 329 989 (1962). Shieh, P.K., Lanyi, J., and Packer, L., Methods in anymology, 55, 604 (1979). Tien, H.T. (1974) BLM: Theory and Practice, Marcel Dekker, N.Y. Minta, J.0. and Man, D.P., J. Immunol. 112, 1597 (1977). Lauf, P.K.Z., Immunituetsfursch 142, 514 (1974).