Endotoxin-Induced Platelet Aggregation and Secretion

J. Cell Sci. a8, 225-236 (1977) 225 Printed in Great Britain © Company of Biologists Limited 1077

ENDOTOXIN-INDUCED PLATELET AGGREGATION AND SECRETION. II. CHANGES IN PLASMA MEMBRANE PROTEINS

KAREEN J. I. THORNE,* RHONDA C. OLIVER,• D. EUAN MACINTYREf AND JOHN L. GORDONJ *Strangeteays Research Laboratory, Worts' Causeway, Cambridge, CBi 4-RAT; f University Department of Pathology, Tennis Court Road, Cambridge, CBz iQP; % AJR..C. Institute of Animal Physiology, Babraham, Cambridge, CB2 \AT, England

SUMMARY Responses of blood platelets to bacterial endotoxin lipopolysaccharide (LPS) have been correlated with changes in the molecular organization and composition of the platelet plasma membrane proteins. Binding of LPS, which occurred in the absence of Ca1+, was distinguished from platelet aggregation and degranulation, which required Cas+ and plasma proteins. Changes in membrane organization were detected by double-labelling with [lt5I] and [1S1I] iodide, mediated by lactoperoxidase and hydrogen peroxide. Changes in total membrane composition were detected by gel electrophoresis of isolated membranes. Binding of LPS was associated with increased accessibility of a protein of mol. wt. 80000 to iodination. After aggregation and degranulation there was, in addition, increased accessibility of proteins of mol. wt. 68000 and 48000. Isolated membranes from LPS-stimulated platelets contained more of a protein of mol. wt. 200000 and less of a protein of mol. wt. 220000 than control membranes prepared from unstimulated platelets in the presence of cAMP and aminophylline. The relationship of the modified plasma membrane proteins to the contractile proteins of the platelet and their possible redistribution in the cell during aggregation and secretion is discussed.

INTRODUCTION Bacterial endotoxin lipopolysaccharide (LPS) stimulates rabbit platelets to aggregate and to release vasoactive amines, ADP and lysosomal enzymes (Des Prez, Horowitz & Hook, 1961; Spielvogel, 1967; Maclntyre et al. 1977). The response of rabbit platelets to bacterial LPS is at least partly complement-dependent (Spielvogel, 1967; Gotze & Miiller-Eberhard, 1971), and requires Ca2+ (Des Prez & Bryant, 1966). Phillips & Agin (1973, 1974) showed that thrombin-induced platelet secretion is associated with characteristic alterations in the platelet plasma membrane. This thrombin-induced platelet activation does not require plasma proteins (Martin, Feinman & Detwiler, 1975) and therefore is presumably not complement-dependent. Changes in organization of the membrane proteins were detected by lactoperoxidase- catalysed iodination of intact cells (Phillips & Morrison, 1971). We used similar procedures to study platelet responses to LPS. The effect of LPS on the total protein composition of isolated platelet plasma membranes was also investigated. 226 K. J. I. Tliorne, R. C. Oliver, D. E. Maclntyre andj. L. Gordon

MATERIALS AND METHODS Rabbit platelets Rabbit blood containing one-tenth volume 3'8% (w/v) Na,cirrate.2H|O in 0-85% (w/v) NaCl as an anticoagulant, was centrifuged for 30 s at 4000 g at room temperature yielding a supernatant consisting of platelet-rich plasma. A pellet of platelets was obtained from this platelet-rich plasma by centrifugation for 8 min at 4000 (J. The pellet was washed, either with Cal+-free Tyrode's salt solution (80 g NaCl, 02 g KC1, 01 g MgCl,.6H,O, 0058 g NaHjPO4.2H,O, 1 -o g NaHCO3, rog glucose in 1 1. of water, pH 76), or with Jamieson's salt solution (1 mM Na,EDTA, 001 M Tris-HCl pH 75, 0-15 M NaCl) (Barber & Jamieson, 1970).

Lipopolysaccharide Lipolysaccharide was obtained by phenol extraction of isolated cell walls of Acinetobacter sp. 199A (Thorne, Thornley & Glauert, 1973). It was assayed by the carbocyanine method (Janda & Work, 1971) using LPS from Salmonella typhimurium (Difco Laboratories, Detroit, Mich., U.S.A.) as standard.

Lactoperoxidase-catalysed iodination Platelets from 20 ml of plasma were incubated with LPS (60 /ig/ml) at 37 °C for 5 min under a variety of conditions. The platelets from the incubated sample and from a replicate control sample were collected by centrifugation for 8 min at 4000 g. The cell pellets were washed once with Ca'+-free Tyrode's salt solution, resuspended in 1 ml of the same solution and iodinated for 15 min at o CC after the addition of 10 fig lactoperoxidase (Sigma Chemical Co., Kingston upon Thames, Surrey, KT2 7BH), ico/iCi carrier-free [115I] or [1S1I] iodide (The Radio- chemical Centre, Amersham, Bucks, HP7 9LL) and 4 portions of 20 fi\ of 10 mM H,Oj, added at 30-s intervals. After iodination the platelets were washed 3 times with Ca*+-free Tyrode's salt solution. They were then heated at 100 °C for 10 min with 0-05 ml concentrated SDS sample buffer (see polyacrylamide gel electrophoresis), 025 ml 9 M urea, 0-04 ml mercaptoethanol and 0-15 ml water, in preparation for disk electrophoresis on polyacrylamide gels.

Plasma membrane isolation and characterization Plasma membranes from LPS-stimulated and unstimulated platelets were prepared by the method of Barber & Jamieson (1970). Platelets were suspended in 1 ml Jamieson's salt solution and loaded with glycerol by centrifugation through a glycerol gradient. The glycerol-loaded platelets were lysed by suspension in 1 ml 0-25 M sucrose and the plasma membranes were obtained by centrifuging the suspension for 4 h at 63 500 g at 4 °C on a sucrose gradient com- posed of 4 ml 1-9 M sucrose, 4 ml 1-37 M sucrose, 4 ml I-I M sucrose and 4 ml O'8 M sucrose. In some experiments 005 mM dibutyryl cAMP and 005 mM aminophylline were added to the lysing solution and to the sucrose solutions of the gradient. Bands were collected from the interfaces between 08 and I-I M sucrose (band 1) and between I-I and 137 M sucrose (band 2). They were diluted 4-fold with water and centrifuged at 105000 g for 1 h at 4 °C. The membrane pellets were collected and prepared for electron microscopy as previously described (Maclntyre et al. 1977), and for slab gel electrophoresis by heating at 100 °C for 10 min with 0-05 ml concentrated SDS sample buffer (see polyacrylamide gel electrophoresis), 0-25 ml 9 M urea, 0-04 ml mercaptoethanol and 0-15 ml water. Phosphodiesterase was assayed from the hydrolysis of 1 mM ftu-(4-methylumbelliferyl) phosphate (a gift from Dr N. Crawford) in 0-2 M sodium acetate, pH 5-5. The incubation was terminated by the addition of 2 vol. of 0-5 M glycine/NaOH, pH 10-4, and the fluorescence of liberated 4-methylumbelliferone was measured (excitation 360 nm, emission 448 nm) (Taylor, Williams & Crawford, 1977).

Polyacrylamide gel electrophoresis Disk electrophoresis was performed on 8 and 10% polyacrylamide gels in o-i % sodium dodecyl sulphate (SDS) as described by Weber & Osborn (1969) and on 7-5 % polyacrylamide Endotoxin effects on platelet membranes 227 gels in o-i % SDS and o-i M Tris/Bicine pH 83. Concentrated sample buffer, of 10 times the final concentration, was used in the digestion of platelet samples. After electrophoresis, the gels were sliced into i-mm long segments and left overnight in 0-2 ml H,O, (100 vol.). A scintillation cocktail of s g 2,5-diphenyloxazole and 03 g i,4-6w-(4-methyl-s-phenyloxazole- 2-yl)-benzene in 500 ml Triton X-100 and 1000 ml toluene was added and the samples were counted in a Packard Tricarb scintillation counter set for optimal efficiency for [1I6I] and for tml]. [11SI] counts were corrected for [lllI] overlap. Slab gel ekctrophoresis was performed in 75 % polyacrylamide containing SDS on 8 cm x8cm gels, by the method of Laemmli (1970), but using a stacking gel of 10% (w/v) agarose. Ten- fold concentrated sample buffer was used in the solubilization of plasma membrane proteins. After electrophoresis proteins were detected by staining with Coomassie brilliant blue R (Gurr, High Wycombe, Bucks) and glyco-compounds were detected by the periodate-Schiff method (Fairbanks, Steck & WaJlach, 1971). Molecular weight markers of carbonic anhydrase, ovalbumin, bovine serum albumin, phosphorylaae A and myosin were run alongside the experimental samples.

RESULTS Iodination of platelets treated with LPS in Ca2+-free plasma Lacteroperoxidase catalyses the hydrogen peroxide-mediated iodination of proteins accessible from outside the cell. When rabbit platelets were iodinated in plasma containing 3 raM ethylene glycol-6w(/?-amino-ethyl ether) TV.iV'-tetra-acetic acid

E 12 15 E Q. Q. •o

o 10 >-

i 4 5 -2

10 20 30 40 Distance along gel, mm Fig. 1. Iodination of platelets after treatment with LPS in plasma containing 3 mM EGTA. Control platelets were iodinated with [mI] iodide ( ), LPS-treated platelets were iodinated with [1I5I] iodide ( ). The 2 samples were mixed, electrophoresed on an SDS disk gel (7"5% polyacrylamide), sliced into i-mm segments and counted for the 2 isotopes. The major peaks are glycoproteins II and III and proteins i, 2, 3 and 4. Mol. wt. are given in Table 1.

(EGTA) to reduce the Ca2+ concentration and analysed by gel electrophoresis, 2 of the 3 main platelet glycoproteins (designated II and III) and 4 other proteins (1,2,3 and 4) were labelled (Fig. 1). The apparent molecular weights of these proteins are given in Table 1. Occasionally a protein with similar characteristics to glycoprotein I (mol. wt. 150000) was also labelled but, as also observed by others (Okumura & Jamieson, 1976), the appearance of this protein among the labelled species was variable. 228 K. J. I. Thorne, R. C. Oliver, D. E. Maclntyre andj. L. Gordon Changes in protein accessibility induced by LPS were detected using 2 different radioactive isotopes of iodine: [12BI] iodide was used for LPS-treated platelets and f131!] iodide was used for control cells. Electrophoretic separation of the membrane proteins of the combined samples of treated and untreated cells, followed by measure- ment of the levels of the 2 isotopes in each labelled protein, provides a sensitive method for the detection of LPS-induced changes in accessibility to lactoperoxidase-catalysed

Table 1. Lactoperoxidase-catalysed iodination of platelet proteins

Effect of LPS Apparent Protein mol. wt. Plasma Plasma+ EGTA Glycoprotein II 105000 Glycoprotein III 92000 Protein i 80000 Protein 2 68000 Protein 3 55000 Protein 4 48000 + , increased labelling; —, unchanged.

E 12 a a. •o •D

2 'o

o ro I 4 1 °

10 20 30 40 50 Distance along gel, mm Fig. 2. Iodination of platelets after treatment with LPS in Tyrode's salt solution containing 7.5 mM EGTA. Control platelets were iodinated with [115I] iodide ( ), LPS-treated platelets were iodinated with [U1I] iodide ( ). The 2 samples were mixed, electrophoresed on an SDS disk gel (10% polyacrylamide), sliced into i-mm segments and counted for the 2 isotopes. The major peaks are glycoproteins I, II and III and proteins i, 2, 3 and 4. Mol. wt. are given in Table 1. iodination. The only protein with increased labelling in platelets treated with LPS was protein 1 (Fig. 1). Under these conditions LPS binds to platelets but induces no aggregation or secretion response (Maclntyre et al. 1977). When platelets were treated with LPS in Ca2+-free salt solution, in the absence of plasma proteins, no changes were detected (Fig. 2) except in the glycoprotein I region. When this 150000 mol. wt. protein was labelled, LPS treatment enhanced its labelling, even in the absence of Ca2+ and plasma proteins. When the protein was not found LPS did not induce it. Endotoxin effects on platelet membranes 229

Iodination of platelets treated with LPS in plasma The labelling of control platelets from plasma (Fig. 3) differed little from the labelling in Ca2+-free medium. LPS induced increased labelling of proteins i, 2 and 4 when Ca2+ was present (Fig. 3 A, B). In experiments where the protein of mol. wt. 150000 was labelled this protein was also increased by LPS treatment (Fig. 3B). LPS treatment of platelets under these conditions induces aggregation and release of granule contents.

12r A 1 2 4 II III a. A •D 11 H1- 2 rr (\\ I''l"l i i - 11 i i \\ o ; i i l A \ i i > 11 f U\»if *J\ 3 Ct

2 4 - 1 VI \

1 10 20 30 40 Distance along gel, mm

D. !« 12 •D

8 ivi u O o T3 4 CD

10 20 30 40 50 Distance along gel, mm Fig. 3. Iodination of platelets after stimulation with LPS in plasma. A, control platelets were iodinated with L131I] iodide ( ), LPS-treated platelets were iodinated with [11SI] iodide ( ). The samples were analysed by electrophoresis on a 75 % polyacrylamide gel. B, control platelets were iodinated with [mI] iodide ( ), LPS- treated platelets were iodinated with [131I] iodide ( ). The samples were analysed by electrophoresis on a 10 % polyacrylamide gel. The major peaks are glycoproteins I, II and III and proteins 1, 2, 3 and 4. Mol. wt. are given in Table 1.

Isolated plasma membranes from LPS-treated platelets Changes detected by iodination with lactoperoxidase and hydrogen peroxide may be due either to changes in the total protein composition of the platelet plasma membrane, or to alterations in membrane organization. To detect changes in total protein.composition, plasma membranes were isolated from glycerol-loaded platelets *$© K. J. I. Thorne, R. C. Oliver, D. E. Maclntyre andj. L. Gordon

"^\^U/

Fig. 4. Electron micrograph of a thin section of isolated plasma membrane from control platelets. Band 2 (p I-IS-I-I8) from a sucrose gradient. Bar line represents 1 fim.

by the method of Barber & Jamieson (1970). Two membrane-containing bands were obtained from sucrose density gradients, an upper band 1 (p i*io-i#i4) and a lower band 2 (p i-i5~i-i8). Electron micrographs of band 2 showed smooth-membrane vesicles, about one-fifth of the diameter of intact platelets (Fig. 4). Barber & Jamieson (1971) isolated 2 plasma membrane fractions from human platelets, but these were of density 1-09 and 1-12. The fractions MI (p i-io-i-i2) and Mil Endotoxin effects on platelet membranes Mol. wt. , . _, Proteins (thousands) fgk ' (see table 1)

150 - 90 - i i i " ! • 2

50 - mmm ijT- 4

30 - J •'

Fig. 5. Disk gel electrophoresis of control platelet plasma membranes, A, band i, and B, band 2 stained with Coomassie brilliant blue for protein; c, band 1, and D, band 2 stained for glyco-compounds. The bands corresponding to glycoproteins I, II and III and proteins 2, 4 and 5 are indicated (see Table 1).

(p i-i3-i-i5), isolated from pig platelets by Taylor & Crawford (1976), had densities closer to the membrane fractions obtained from rabbit platelets in the present study. Differences between the proteins of the two membrane bands, similar to those reported by Taylor & Crawford (1976) for pig platelets, were detected (Fig. 5). Two-thirds of the platelet phosphodiesterase activity assayed with 6w(4-methyl umbelliferyl phosphate) was found in band 2. This fraction had a specific activity of i34nmol hydrolysed/h/mg protein. Both membrane fractions prepared from pig platelets by Taylor et al. (1977) after mechanical breakage, contained this phosphodiesterase but, in contrast to the present observations, the activity of the enzyme was greater in the lighter fraction than in the heavier. Analysis of plasma membranes isolated from rabbit platelets treated with LPS under a variety of conditions did not reveal any consistent LPS-induced changes in the membrane proteins of either band 1 or band 2. However, the preparation of plasma K. J. I. Thorne, R. C. Oliver, D. E. Maclntyre and J. L. Gordon Mol. wt. Proteins (thousands) •W^^^^B^ I _A «_ «*««• ~ D 200 - - — F — I

100-

80-!

60-

A B cAMP LPS (control) Fig. 6. Slab gel electrophoresis of plasma membranes isolated from LPS-stimulated platelets. A, band 2 membrane isolated from unstimulated platelets in the presence of 005 mM dibutyryl cAMP and 0-05 mM aminophylline; B, band 2 membrane isolated from LPS-stimulated platelets. 7-5 % polyacrylamide gel containing SDS, stained with Coomassie brilliant blue for protein. The amount of protein A and F is increased relative to protein D in the LPS-stimulated platelets. Glycoprotein I is also slightly increased. Endotoxin effects on platelet membranes 233 membranes involves both physical maltreatment of the cells and changes in the chemical environment. These artificial conditions may be sufficient to trigger the platelet into an activated state. One possible way of protecting the membrane would be with cAMP. cAMP is believed to protect platelets from stimulation (Salzman & Weinberger, 1972). The bands observed after slab gel electrophoresis of control membranes prepared in the presence of cAMP (Fig. 6 A) were compared with those from membranes from platelets stimulated with LPS in plasma (Fig. 6 B). The proteins of band 1 were unaffected, but those of band 2 showed marked differences in the large molecular weight region. Six proteins of mol. wt. 200000 or over (A-F) were detected. The control membranes had protein D (mol. wt. 220000) as the major band, while those from platelets stimulated with LPS contained more protein A (250000) and protein F (200000) relative to protein D. Glycoprotein I also seemed to be slightly increased in concentration in the LPS-stimulated cells. Control membranes prepared without cAMP were highly variable and gave results intermediate between Fig. 6 A and B.

DISCUSSION Lactoperoxidase-catalysed iodination has revealed that the responses of rabbit platelets to LPS are associated with specific changes in the proteins of the plasma membrane. Under conditions of low Ca2+ concentration, where LPS binds to the platelet surface without inducing aggregation or secretion, only protein 1 becomes more accessible to iodination after LPS treatment. When adequate Ca2+ is present, platelets degranulate and form aggregates in response to LPS, and this is associated with an increase in the labelling of proteins 1, 2 and 4. The requirement for Ca2+ may be for activation of the classical complement pathway of LPS. The appearance of protein 1 at the cell surface would be independent of complement-activation while the appearance of proteins 2 and 4 would be consequent upon complement activation. The occurrence of a protein co-migrating with glycoprotein I amongst the iodinat- able species is somewhat variable. There is disagreement as to whether or not glycoprotein I can be iodinated in intact platelets. Okumura & Jamieson (1976) found that glycoprotein I was not accessible to iodination, while Phillips (1972) obtained some labelling of glycoprotein I. Evidently glycoprotein I is highly sensitive to experimental conditions. Alternatively the iodinated protein may not be platelet glycoprotein I, but may originate from contaminating granulocytes. Thorne, Oliver & Lackie (1977) have shown that the only iodinated protein on the granulocyte surface has an apparent mol. wt. of 150000. Contamination of the platelet-rich plasma with io4 granulocytes in 10 ml plasma would be sufficient to produce the observed labelling of this protein. In addition the granulocyte protein responds to LPS even in the absence of plasma proteins (Thorne et al. 1977). This protein in the platelet preparation behaved similarly (Fig. 2). Identification of the membrane proteins which are modified when LPS interacts with platelets in plasma and induces degranulation and aggregation can at this time only be tentative. The presence of plasma proteins allows LPS-induced complement 234 K- 3- I- Thome, R. C. Oliver, D. E. Maclntyre and J. L. Gordon activation, and some of the changes in cell surface proteins may therefore be attribut- able to complement components. Another explanation of the observed modifications in the platelet membrane may be changes in organization of the cell's contractile proteins. The mol. wt. of protein 4 suggests that it is probably the protein identified by Taylor & Crawford (1976) as actin. Microfilaments containing actin may have a role to play in platelet degranulation and aggregation. Electron-microscopic evidence for a rapid extrusion of platelet ' cytogel' containing contractile proteins during platelet aggregation has been presented by Booyse, Hoveke, Kisieleski & Rafelson (1972). A process of this type might explain the increased accessibility of an actin-like protein to iodination after stimulation with LPS. Alternatively actin microfilaments may play a role in degranulation. Taylor & Crawford (1976) and Hagen, Olsen & Solum (1976) have found actin in platelet granules. Fusion of granule membrane with plasma membrane could increase the amount of actin in the plasma membrane. Induction of platelet release by collagen is associated with phosphorylation of platelet actin (Haslam & Lynham, 1976). If membrane actin is phosphorylated this could affect its accessibility to iodination. Changes in contractile proteins could also explain the observed modifications in the large molecular weight proteins of the isolated plasma membrane. Control plasma membranes were prepared in the presence of cAMP, since otherwise they exhibited considerable variation in their protein composition. It was argued that cell lysis and membrane purification introduces such considerable modification to the membrane environment that this may itself 'activate' the membrane. The proteins which are changed by LPS are of molecular weight 200000 and greater. Protein F has the same mol. wt. as the myosin heavy chain. Proteins A and D, mol. wt. 220000 and 250000, are similar in size to actin-binding protein (Fairbanks et al. 1971). The observed increase in F and decrease in D might therefore be explained by an increased binding of myosin to the membrane and a decreased binding of one of the components of actin-binding protein. Actin-binding protein is believed to inhibit the polymerization of actin, so that filaments form only when actin-binding protein is removed (Tilney & Detmers, 1975). The changes induced in the platelet plasma membrane by LPS may be compared with changes induced by thrombin. Thrombin stimulation of platelets differs from LPS stimulation in not requiring plasma proteins and therefore probably being independent of complement. Phillips & Agin (1973, 1974) used lactoperoxidase- catalysed iodination to demonstrate that after thrombin treatment glycoprotein I and proteins of mol. wt. 70000 and 55000 became more accessible to the cell surface. We have already discussed the ambiguities of glycoprotein I labelling. The other 2 proteins differ somewhat in apparent mol. wt. from the proteins discussed here. A protein which becomes less accessible to the surface after thrombin treatment is glycoprotein II (Steiner, 1973). This has been interpreted as hydrolysis of glycoprotein II by thrombin. In summary, the changes in the platelet induced by bacterial LPS can be correlated with changes in the molecular organization and composition of the plasma membrane. Aggregation and secretion can only be induced by bacterial LPS in the presence of Endotoxin effects on platelet membranes 235 plasma proteins and Ca2+. This is associated with the appearance of 3 new proteins at the cell surface, one of which resembles actin. If Ca24" is absent, LPS binds to the platelet surface, but induced no shape change, aggregation or secretion. Under these conditions only one of the 3 proteins, protein 1 of mol. wt. 80000, appears at the cell surface. In the absence of plasma proteins there was no binding of LPS and no change in the cell surface proteins. Results obtained from plasma membranes isolated in the presence of cAMP emphasize the difficulties of ensuring that 'baseline' values for membrane composition are not artificially altered by the preparative procedures. Under conditions which minimize artificial modifications proteins of mol. wt. 200000 and greater seem to be modified by LPS stimulation. Experiments are in progress to identify these proteins and to investigate their relationship to the contractile proteins actin-binding protein and myosin.

We acknowledge the support of the Medical Research Council.

REFERENCES BARBER, A. J. & JAMIBSON, G. A. (1970). Isolation and characterisation of plasma membranes from human blood platelets. J. biol. Chem. 245, 6357-6365. BARBER, A. J. & JAMIESON, G. A. (1971). Isolation of glycopeptides from low and high density platelet plasma membranes. Biochemistry, N.Y. io, 4711-4717. BOOYSE, F. M., HOVBKE, T. P., KISIELHSKI, D. & RAFBLSON, M. E. (1972). Mechanism and control of platelet-platelet interaction. I. Effects of inducere and inhibitors of aggregation. Microvascular Res. 4, 179-198. DES PREZ, R. M. & BRYANT, R. E. (1966). Effects of bacterial endotoxin on rabbit platelets. IV. The divalent ion requirements of endotoxin-induced and immunologically induced platelet injury. J. exp. Med. 124, 971-982. DES PREZ, R. M., HOROWITZ, H. 1. & HOOK, E. W. (1961). Effects of bacterial endotoxin on rabbit platelets. I. Platelet aggregation and release of platelet factors in vitro. J. exp. Med. 114, 857-873- FAIRBANKS, G., STECK, T. L. & WALLACH, D. F. H. (1971). Electrophoretic analysis of the major polypeptides of the human erythrocyte membrane. Biochemistry, N.Y. 10, 2606—2616. GOTZE, O. & MOLLER-EBERHARD, H. J. (1971). The C3-activator system: an alternate pathway of complement activation. J. exp. Med. 134, 90S—108S. HAGEN, I., OLSEN, T. & SOLUM, N. O. (1976). Studies on subcellular fractions of human platelets by the lactoperoxidase iodination technique. Biochim. biophys. Acta 455, 214-225. HASLAM, R. J. & LYNHAM, J. A. (1976). Increased phosphorylation of specific blood platelet proteins in association with the release reaction. Biochem. Soc. Trans. 4, 694-697. JANDA, J. & WORK, E. (1971). A colorimetric estimation of lipopolysaccharide. FEBS Letters, Amsterdam 16, 343~345- LAEMMLI, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, Lond. 227, 680-685. MACINTYRE, D. E., ALLEN, A. P., THORNE, K. J. I., GLAUERT, A. M. & GORDON, J. L. (1977). Endotoxin-induced platelet aggregation and secretion. I. Morphological changes and phar- macological effects. J. Cell Set. (in Press). MARTIN, B. M., FEINMAN, R. D. & DETWILER, T. C. (1975). Platelet stimulation by thrombin and other proteases. Biochemistry, N.Y. 14, 1308-1314. OKUMURA, T. & JAMIESON, G. A. (1976). Platelet glycocalicin. I. Orientation of glycoproteins of the human platelet surface. J. biol. Chem. 251, 5944-5949. PHILLIPS, D. R. (1972). Effect of trypsin on the exposed polypeptides and glycoproteins in the human platelet. Biochemistry, N.Y. 11, 4582-4588. PHILLIPS, D. R. & AGIN, P. P. (1973). Thrombin-induced alterations in the surface structure of the human platelet plasma membrane. Ser. Haemat. 6, 292-310. 236 K. J. I. Thorne, R. C. Oliver, D. E. Maclntyre andj. L. Gordon PHILLIPS, D. R. & AGIN, P. P. (1974). Thrombin substrates and the proteolytic site of thrombin action on human-platelet plasma membranes. Biochim. biophys. Acta 352, 218-227. PHILLIPS, D. R. & MORRISON, M. (1971). Exposed protein on the intact human erythrocyte. Biochemistry, N.Y. 10, 1766-1771. SALZMAN, E. W. & WEINBERGER, H. (1972). Role of cyclic AMP in platelet function. Advances in Cyclic Nucleotide Research, vol. 1 (ed. P. Greengard & G. A. Robinson), pp. 231-247. New York: Raven Press. SPIKLVOGKL, A. R. (1967). An ultrastructural study of the mechanisms of platelet-endotoxin interaction. J. exp. Med. 126, 233-250. STEINER, M. (1973). Effect of thrombin on the platelet membrane. Biochim. biophys. Acta 323,

TAYLOR, D. G. & CRAWFORD, N. (1976). Enzymatic and chemical analysis of pig platelet membrane subfractions isolated by zonal centrifugation. Biochim. biophys. Acta 436, 77-94. TAYLOR, D. G., WILLIAMS, V. M. & CRAWFORD, N. (1977). jBw-(4-methylumbelliferyl) phosphate as a substrate for the surface membrane-associated phosphodiesterase activity of pig platelets. Biochim. biophys. Acta 465, 667-671. THORNE, K. J. I., OLIVER, R. C. & LACKIE, J. M. (1977). Changes in the surface properties of rabbit polymorphonuclear leucocytes, induced by bacteria and bacterial endotoxin. J. Cell Sri. 27, 213-225. THORNE, K. J. I., THORNLEY, M. J. & GLAUERT, A. M. (1973). Chemical analysis of the outer membrane and other layers of the cell envelope of Arinetobacter sp. J. Bact. 116, 410—417. TILNEY, L. G. & DETMERS, P. (1975). Actin in erythrocyte ghosts and its association with spectrin. J. Cell Biol. 66, 508-520. WEBER, K. & OSBORN, M. J. (1969). The reliability of molecular weight determinations by dodecyl sulfate polyacrylamide gel electrophoresis. J. biol. Chem. 244, 4406—4412.

(Received 27 April 1977)