INFECTION AND IMMUNITY, June 1975, p. 1312-1319Copyright i 1975 American Society for Microbiology

Vol. 11, No. 6Printed in U.S.A.

Interactions Between Aerolysin, Erythrocytes, and ErythrocyteMembranes

ALAN W. BERNHEIMER,* LOIS S. AVIGAD, AND GAD AVIGAD

Department of Microbiology, New York University School of Medicine, New York, New York 10016,*and Department of Biochemistry, Rutgers Medical School, Piscataway, New Jersey 08854

Received for publication 30 January 1975

Aerolysin, a hemolytic and lethal exotoxin of Aeromonas hydrophila, wasanalyzed for amino acids. Assuming 8 histidine residues/mol, the purified toxicprotein has, by summation, a molecular weight of 49,000, a value in agreementwith earlier estimates by other methods. Erythrocytes from differentanimal species differ greatly in sensitivity to aerolysin's lytic action. There issome correlation between sensitivity and phosphatidyl choline content. Erythro-cyte membranes of different species bind the toxin, and the efficiency of bindingis a function of sensitivity to lysis. Binding is temperature independent, is notdependent upon membrane sialic acid, and is decreased by prior treatment withphospholipase C and proteases. Preparations of aerolysin convert substantialamounts of membrane phosphorus to water-soluble form; the conversion isconcentration and temperature dependent. Most of the conversion is attributableto contaminating phospholipase(s) that is separable from the toxin. Aerolysinpurified by electrophoresis in polyacrylamide gel retains some phospholipaseactivity, and this activity may or may not be a contaminant.

Aerolysin, a lethal and hemolytic extracellu-lar product of Aeromonas hydrophila, was earlyobserved by Caselitz and Gunther (10) andstudied further by Wretlind et al. (26). Factorsgoverning the production of this protein weresubsequently investigated by Wretlind et al.(25) and by Bernheimer and Avigad (3). Thelatter also described a procedure for purificationof aerolysin, and they partially characterized it.The present report provides information con-cerning its interaction with erythrocytes andtheir membranes.

MATERIALS AND METHODS

Abbreviations. The abbreviations used in thispaper are: Tris, 0.01 M tris(hydroxymethyl)amino-methane, pH 7.2; TS, Tris containing 0.145 M NaCl;TSG, TS containing 0.2% gelatin; and HU, hemo-lytic units as defined by Bernheimer and Avigad (3).

Aerolysin. Aerolysin was prepared as describedearlier (3). A portion of one lot was hydrolyzed invacuo in 6 N HCl for 24 h at 114 C, after which anamino acid analysis was carried out according to themethod of Moore et al. (20). The composition ofaerolysin is shown in Table 1. Assuming 8 histidineresidues/mol, the molecular weight of aerolysin, bysummation, is 49,000, a value that compares favora-bly with earlier estimates of 50,000 and 53,000 ob-tained by gel electrophoresis and gel filtration, respec-tively.Hemolytic activity. Hemolytic activity was as-

sayed by adding decreasing quantities of test solutionto a constant quantity of washed rabbit erythrocytescontained in a final volume of 2 ml of TSG. Afterincubating at 37 C for 30 min and briefly centrifuging,the percentage of hemolysis was estimated from thecolor of the hemoglobin in the supernatant fluids.Further details are given in reference 3. Contrary to astatement by Scholz et al. (23), this lysin in ourexperience does not exhibit the behavior of a hot-coldhemolysin when rabbit, human, or sheep erythrocytesare used.

Preparation of erythrocyte membranes. (MethodA) Eighty milliliters of a 0.7% (vol/vol) suspension ofwashed erythrocytes in 0.077 M NaCl and 0.067 Msodium phosphate (pH 7.0) was centrifuged at about3,000 x g for 10 min. The cells were suspended in 4 mlof TS and lysed by adding 180 ml of distilled water.The lysate was centrifuged at 15,000 x g for 20 min,and the sedimented membranes were suspended in 3ml of Tris.

(Method B) Five milliliters of rabbit blood wasdelivered into 200 ml of 0.077 M NaCl and 0.067 Msodium phosphate (pH 7.0). The cells were centri-fuged at 2,000 x g for 10 min and washed twice, eachtime in 200 to 300 ml of 0.077 M NaCl and 0.067 Msodium phosphate (pH 7.0). The cells were suspendedin 20 ml of TS and lysed by adding them to 800 ml ofdistilled water. The lysate was centrifuged at 15,000 xg for 30 min, and the membranes were collected andwashed in 70 ml of Tris and suspended in 12 ml ofTris.

(Method C) The procedure of Hoogeveen et al. (14)was used as modified and described by Low et al. (18).

Phosphorus. Unless indicated otherwise, total

1312

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.

AEROLYSIN

TABLE 1. Amnino acid composition of aerolysin

A NearestAmino acid Concn (residues integer(uM) [histidine to8 x A

= 1 1)

Aspartic acid ....... 2.232 8.36 67Threonine ......... 1.200 4.49 36Serine ............. 1.320 4.94 40Glutamic acid ...... 1.776 6.65 53Proline ............ 1.128 4.22 34Glycine ............ 1.896 7.10 57Alanine ............ 1.560 5.84 47Half-cystine ....... ? ? ?Valine . ............ 1.140 4.27 34Methionine ........ 0.144 - 1 18Isoleucine .......... 0.717 2.69 22Leucine ............ 1.224 4.58 37Tyrosine ........... 0.645 2.42 19Phenylalanine ...... 0.594 2.22 18Lysine ............. 0.867 3.25 26Histidine .......... 0.267 1.00 8Arginine ........... 0.573 2.14 17Tryptophan ........ ? ? ?

phosphorus was determined by the Bartlett procedure(1) as modified by Bbttcher et al. (7).Enzymes. Neuraminidase was purchased from

Sigma Chemical Co. (St. Louis, Mo.), papain andtrypsin from Worthington Biochemical Corp. (Free-hold, N.J.), and Pronase from Calbiochem (La Jolla,Calif.). Phospholipase C from Clostridium perfringenswas purified from solutions of dried culture filtrate byisoelectric focusing followed by gel filtration in Sepha-dex G-100 (Pharmacia Fine Chemicals, Piscataway,N.J.). The purified enzyme was preserved in 50%glycerol at 4 C.

Blood specimens. Rabbit and guinea pig bloodswere obtained from the local animal facility, sheep,horse, and some rabbit bloods from the Public HealthResearch Institute (New York, N.Y.), and most of theremaining bloods from either the Animal Blood Cen-tre (Syracuse, N.Y.) or Colorado Serum Co. Labs(Denver, Colo.).

RESULTSSensitivity of erythrocytes to lysis by

aerolysin. Suspensions of washed erythrocytesfrom various kinds of animals were prepared tocontain the same amount of hemoglobin as thatin the standard rabbit erythrocyte suspensionused for routine titrations of hemolytic activity.Purified aerolysin titrated with each gave re-sults (Table 2) in which the lytic activity(415,000 HU/mg), obtained with the most sensi-tive erythrocytes, those of the rat, was set at100%. The titers are expressed in proportion tothat obtained with rat cells. Attempts to corre-late these results with quantitative differencesin particular membrane constituents were notnotably successful. A plot of the relative amountof phosphatidyl choline against log sensitivity to

lysis (Fig. 1) suggests that phospholipid mightbe involved as a substrate or receptor for thelysin. If this is true, the results obtained withhorse, ox, and cat cells require explanationbecause they deviate widely from expectedvalues.

Effect of temperature on capacity of eryth-rocyte membranes to inhibit aerolysin-in-duced hemolysis. It was earlier found (3) thatosmotically prepared rabbit erythrocyte mem-branes inhibited aerolysin-induced hemolysis.The effect of temperature on inhibition wasexamined by adding, in replicate, 0.6 ml of TScontaining 120 HU of aerolysin to 0.4 ml ofrabbit erythrocyte membranes (method A) andplacing the mixtures at several different tem-peratures. After 1 h they were centrifuged at

TABLE 2. Relative sensitivity to aerolysin of erythro-cytes of various species

Sensitivity of erythrocytesSpecies compared to that of

rats (%)

Rat...... 100Mouse .. .... 78Dog...... 18Guinea pig .. .... 18Cat...... 16Rabbit...... 3.1Ox...... 2.4Swine...... 1.1Human .. .... 0.9Horse...... 0.5Goat...... 0.4Sheep...... 0.3

0o

2 50-0.0£E 40-

, 30

0 20.O

-cU 10

Q.0-

-c

0~

*Dog Rat

s *Guinea pighpabbit

Aan ~~~CatSwine

Sheep aot *Ox

10 100

Sensitivity to Lysis by Aerolysin

FIG. 1. Proportion of phosphatidyl choline of totalerythrocyte membrane phospholipid, plotted againstlog sensitivity of erythrocytes to aerolysin. Figures forphosphatidyl choline are from van Deenen (In C.Bishop and D. M. Surgenor [ed.], The Red Blood Cell,2nd ed., in press).

I I

VOL. 11, 1975 1313

10.1

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.

BERNHEIMER, AVIGAD, AND AVIGAD

15,000 x g for 20 min. The supernatant fluidswere diluted with 4 volumes of TSG and as-sayed for hemolytic activity by using rabbiterythrocytes. Compared to a no-membrane con-trol, the amounts of activity missing from thesupernatant fluid at 0, 20, and 37 C were 85, 90,and 81%, respectively. The results indicate thatinhibition of hemolysis is not temperature de-pendent and that the inhibition evidently is dueto binding of the toxin to the membranes.

Capacity of erythrocyte membranes of dif-ferent species to inhibit aerolysin-inducedhemolysis. Membranes were prepared fromerythrocytes of various species according tomethod A. Volumes of membrane suspensionsdecreasing in steps of about 40% were added to0.5 ml of aerolysin diluted in TSG to contain 30HU. The mixtures were made to 1 ml with Tris,allowed to stand at 20 C for 30 min withoccasional mixing, and centrifuged at 15,000 xg for 20 min. The supernatant fluids werediluted with 4 volumes of TSG and titrated forhemolytic activity by using rabbit erythrocytes.The inhibition end point was the quantity ofmembranes needed to inhibit half (15 HU) thetest amount of aerolysin, and was expressed asthe volume (in milliliters) of 0.7% (vol/vol)erythrocyte suspension from which that quan-tity of membranes was derived. The valuesobtained for various species are plotted in Fig. 2against log relative sensitivity to lysis byaerolysin. The results show that the capacity ofmembranes to inhibit lysis is a function of thesensitivity to aerolysin-induced hemolysis of thespecies of erythrocytes. They are interpreted tomean that the sensitivity of cells of a givenspecies to lysis is related to the capacity of thosecells to bind the toxin.

40

32

24.Y

16

*Horse

*Sheep*Goat

*Man Ox

Sn*Rabbit Dog MousRe0.1 1.00io io0

Sensitivity to Lysis XFIG. 2. Capacity of membranes derived from

erythrocytes of different species to inhibit hemolysis,as a function of log sensitivity of erythrocytes to lysis.Y, milliliters of erythrocyte suspension needed toprovide membranes that will inhibit 15 HU ofaerolysin.

Effect of pretreatment of rabbit erythro-cyte membranes (method A) on their capac-ity to bind aerolysin. (i) Divalent cations. To0.3-ml portions of rabbit erythrocyte membranesuspension were added 0.1-ml quantities of 10mM solutions of CaCl2, MgCl2, CuSO,, and Znacetate, separately, in Tris. After 10 min at 25C, the mixtures were centrifuged at 15,000 x gfor 20 min, and the supernatant fluids werediscarded. The membranes were suspended in0.5-ml portions of TSG, and to each was added0.5 ml of TSG containing 30 HU of aerolysin.After 30 min at 20 C, the mixtures were cen-trifuged at 15,000 x g for 20 min. Four volumesof TSG was added to each supernatant fluid,and the resulting solutions were titrated forhemolytic activity. The results showed thatCaCl2 and CuSO, had no effect, whereas MgCl2and Zn acetate gave slightly increased inhibi-tion compared to that obtained with untreatedmembranes. Additional experiments in whichthe effects of 5 mM ethylenediaminetetraace-tate and 5 mM ethylene glycol-bis(f3-aminoethyl ether)-N, N'-tetraacetate were studiedsuggested that Ca2+ might be necessary forbinding of aerolysin to membranes. However,interpretation of the results was complicated bythe fact that ethylene glycol-bis(O-amino ethylether)-N,N'-tetraacetate, in the absence ofmembranes, decreased the activity of aerolysinin one experiment, and also by the fact that thischemical is known not to be very effective inremoving Ca2+ already bound to membranes(17).

(ii) Sodium periodate. To destroy terminalsialyl residues, rabbit erythrocyte membraneswere treated with 1 mM sodium periodate,followed by reduction with sodium borohydrate(5, 16). Capacity to bind aerolysin was notaffected, indicating that intact membrane sialicacid is not required for binding of aerolysin. Inline with this is also the finding that similartreatment of whole erythrocytes did not affecttheir sensitivity to lysis by aerolysin.

(iii) Neuraminidase, phospholipase C, andproteases. To 0.4 ml of rabbit erythrocytemembrane suspension was added 0.4 ml of Triscontaining 0.5 unit of neuraminidase. After 1 hat 37 C, the membranes were centrifuged andtreated as described for divalent cations. Appro-priate controls were included. The inhibitorycapacity of the enzyme-treated membranes wasthe same as that of untreated membranes.

Phospholipase C and several proteases weretested for capacity to abolish inhibition of lysisby erythrocyte membranes by using a proceduresimilar to that described for divalent cations.The phospholipase was allowed to react withmembranes at 25 C for 60 min while the

1314 INFECT. IMMUN.

8

Ot

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.

AEROLYSIN

protease-membrane mixtures were kept at 20 Cfor 30 min. The results (Table 3) indicate thatexposure of membranes to phospholipase C,papain. Pronase, or trypsin diminished thecapacity of membranes to bind aerolysin. In a

separate experiment, and in agreement withearlier findings (3), proteases had no effect on

the activity of aerolysin in the absence of addedmembranes.Failure of certain sugars and sulfhydryl

inhibitors to affect hemolytic activity. A solu-tion of aerolysin was titrated for hemolyticactivity in the presence, separately, of 5 mMgalactose, lactose, and melibiose. The titers didnot differ significantly from that of a controlsolution. This result is consistent with theassumption that terminal galactosyl moieties ofmembrane glycoproteins and glycolipids are notinvolved in the binding of aerolysin; it is con-

sistent also with the lack of involvement ofsialyl residues.A solution of aerolysin was titrated for hemo-

lytic activity in the presence, separately, of 0.1mM N-ethylmaleimide, so4ium iodoacetate,and sodium p-chloromercuribenzoate. No dif-ference was found among the titers obtained inthe presence and absence of these agents, in-dicating that sulfbydryl groups apparently are

not involved in lysis.Phospholipase C-like effect of aerolysin

preparation on erythrocyte membranes. Ad-dition of phospholipase C from C. perfringens(12) or Bacillus cereus (6) to human erythrocyteghosts causes shrinkage of the ghosts withformation of phase-dense droplets ("blackdots") consisting of diglyceride and ceramidearising from hydrolysis of phospholipids. De-creasing volumes of aerolysin solution in TSG

TABLE 3. Effect of several enzymes on the capacity oferythrocyte membranes to bind aerolysin

SupernatantRabbit fluid from

erythrocyte Additiona aerolysinmembrine membrane

prepn mixture(HU/ml)

None None 350.3 None 70.3 PLCb 10X 240.1 None 220.1 PLC 10 X 330.3 Papain (100 ug) + 24

mercaptoethanol(0.5 gg)

0.3 Pronase (100 g) 280.3 Trypsin (100jug) 21

aAerolysin (35 HU) also added to each.PLC, Phospholipase C.

were brought to 0.25 ml with 10 mM sodiumborate buffer, pH 8.2. After addition of 0.1 ml ofosmotically prepared membranes, the mixtureswere allowed to stand at 25 C for 1 h or longerand then examined by phase contrast micros-copy. The results of the experiments carried outin the absence and presence of 7 mM calciumacetate are in Table 4. They show that: (i)phase-dense droplets develop in rabbit andhuman ghosts but not in sheep cell ghosts; (ii)rabbit erythrocyte ghosts are more sensitivethan human; and (iii) calcium ions intensify theeffect, allowing it to manifest itself in apprecia-bly lower concentrations of aerolysin than whencalcium ions are not added. Additional data,not shown in the table, suggested that the orderof sensitivity to the droplet-producing activityof the aerolysin preparation is guinea pig >rabbit 5 horse > human > sheep.Absence of turbidity-producing activity in

diluted egg yolk. Some phospholipases C pro-duce turbidity in dilute egg yolk. Amounts ofaerolysin as great as 100 jig failed to produceturbidity in egg yolk solution with conditionsdescribed elsewhere (4) in the presence of eitherCa2+ alone or Ca2+ and Mg2+.Liberation of water-soluble phosphorus

from rabbit erythrocyte membranes. (i) As afunction of aerolysin concentration. A stocksolution containing 1 mg of aerolysin per ml in0.01 M sodium borate buffer, pH 8.2, wasprepared. Decreasing volumes were added totubes (18 by 102 mm) having a wall thickness of2 mm, and the volumes were brought to 50 Alwith borate buffer. To each tube were added1 25 ml of 3 mM CaCl2 and 1 ml of rabbit eryth-rocyte suspension (method B). After standing at

'rLE 4. Formation ofphase-dense droplets in osmoti-cally prepared erythrocyte ghosts in absence and pres-

ence of added Ca2+

Phase-dense dropletsFinal concn of Ca2+ in ghosts

aerolysm (MM)(HU/ml) (mM) Rabbit Human Sheep

300 oa ++ b + + 0180 0 ++ ++ 090 0 + + 030 0 A 0 0o 0 0 0 030 7 +++c ++ 09 7 +++ 0 03 7 + 0 00 7 0 0 0

a O signifies no ghosts showed one or more phase-dense droplets.

b + +, +, and signify intermediate degrees.c +++ signifies every ghost showed one or more

phase-dense droplets.

VOL. 11, 1975 1315

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.

BERNHEIMER, AVIGAD, AND AVIGAD

20 C for 30 min, the tubes were centrifuged at15,000 x g for 20 min. The supematant fluidswere removed as completely as possible, andtheir total phosphorus as well as that of thepellets was determined. The results (Fig. 3)show that about 40% of the membrane phos-phorus can be converted to water-soluble formunder the conditions used.

(ii) As a function of temperature. Theconditions in (i) were followed except that aconstant amount (20 jig) of aerolysin was used,and the temperature of the reaction mixturewas varied. The results (Fig. 4) show that theliberation of phosphorus from membranes istemperature dependent.

(iii) Effect of divalent cations. The condi-tions in (i) were followed except that a constantamount (10 Ag) of aerolysin was used, the

4040Q0-

L30c

-aC

320

U-_

o 10

u

i0

X- I10 20 30 40 50

Lig AerolysinFIG. 3. Liberation of water-soluble phosphorus (P)

from rabbit erythrocyte membranes by aerolysin.

5014)

0C

4L)0.

3-oC

0

0

0

4)U.

40-

30-

20-

10 .- no aerolysin

10 20 30 40

Temperature in OCFIG. 4. Effect of temperature on liberation by

aerolysin of water-soluble phosphorus (P) from eryth-rocyte membranes.

divalent cation was varied, and the temperatureof the reaction mixture was 40 C. The phos-phorus values were corrected for liberation ofphosphorus in the absence of aerolysin. Theresults (Table 5) show that Ca2+ and Mg2+caused a small but significant increase in theamount of phosphorus released from mem-branes, Zn2+ inhibited nearly completely, andethylenediaminetetraacetate substantially re-duced the amount of phosphorus released.

Fractionation of aerolysin preparation inpolyacrylamide gels. An aerolysin preparationwas subjected to electrophoresis in polyacryl-amide gel, and then the gel was sliced. Theslices were eluted, and the eluates were exam-ined for hemolytic activity, black dot-formingactivity, and capacity to release phosphorusfrom rabbit erythrocyte membranes.

Seven percent polyacrylamide gelg were pre-.pared in pH 9.5 buffer according to directionsprovided by the manufacturer of the Canalcomodel 12 electrophoresis apparatus (Canalco,Rockville, Md.) but omitting stacking gel. Aer-olysin (200 ,ug) plus 3 tl of 0.05% bromophenolblue were electrophoresed at 4 mA per gel untilthe tracking dye traveled to within 3 mm of thebottom. The gel was chilled and sliced with adevice (19) yielding 54 slices, each slightly lessthan 1 mm thick. Pairs of adjacent slices wereeluted by placing each pair in 0.33 ml of 0.02 MTris, pH 6.7, and allowing them to standovernight in the refrigerator. The supernatantfluids were assayed for the several activitiesmentioned according to methods already de-scribed. An additional gel was stained for pro-tein with Coomassie brilliant blue (Colab Labo-ratories, Glenwood, Ill.) according to Weber andOsborn (24) and scanned with a model 2410-Slinear transport scanner (Gilford InstrumentLaboratories, Inc., Oberlin, Ohio). The resultsare shown in Fig. 5, in which allowance has beenmade for 8% swelling of the stained gel.Degradation of erythrocyte membrane

phospholipids by aerolysin preparation and

TABLE 5. Effect of divalent cations and ethylenedi-aminetetraacetate on liberation of water-soluble phos-

phorus by aerolysin

Total Pa of reactionAdditions (10 mM) mixture (%) released

into supernatant fluid

None... 26MgCl2... 31CaCl2... 32ZnCl2... 2EDTA5... 18

a P, Phosphorus.bEDTA, Ethylenediaminetetraacetate.

1316 INFECT. IMMUN.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.

AEROLYSIN

1oor

_-

80

60

40~

20)

0

3r-2

_ _

O; 0

'1 Ev- i n TF JK

o <S E F11 -l1 :1 i'11~H II IIA

10 20 30 40 50 60

mm from origin

FIG. 5. Distribution of (a) stained protein, (b)capacity to release phosphorus from erythrocyteghosts, (c) black dot-forming activity, and (d) hemo-lytic activity of aerolysin preparation fractionated byelectrophoresis in polyacrylamide gel.

by two major electrophoretic fractionsthereof. Aerolysin preparations (50,g) were

subjected to electrophoresis in polyacrylamidein each of 11 gels according to the conditions ofthe preceding experiment. One gel was stained,and ten were frozen and sliced. From 10 gels,slices between 17 and 23 mm from the originwere pooled and eluted three times each with 1ml of 0.03 M sodium borate buffer, pH 8.2. Thethree eluates were pooled and designated eluateA. Slices between 31 and 41 mm, from the same10 gels, were similarly treated and designatedeluate B. Thirty percent of the input hemolyticactivity was recovered in eluate A and none ineluate B. Recombination of portions of eluatesof all slices along the entire length of the gel didnot result in greater recovery of hemolyticactivity.Mixtures having the composition shown in

Table 6 were prepared in tubes (18 by 102 mm)with a wall thickness of 2 mm. After 1 h at 40 C,the tubes were chilled and centrifuged at 25,000x g for 20 min. After removing the supernatantfluids, the lipid of each pellet was extracted bythe method of Reed et al. (22). Each prepara-tion of dried extracted lipid was dissolved in 0.4ml of chloroform, and about 0.2 ml of each ofthe resulting solutions was analyzed by two-dimensional thin-layer chromatography accord-ing to Broekhuyse (8, 9). Spots were stainedwith iodine vapor and identified by comparison

with those of authentic phospholipids and bycomparison with published R, values. Thespots were scraped from the glass, 0.5 ml ofconcentrated HCl was added, and the mixtureswere evaporated to dryness in a 110 C oven.

After ashing with 70% perchloric acid, phospho-rus was determined by the ascorbic acid modifi-cation (11) of the Fiske-Subbarow method.The results (Table 7) can be interpreted as

follows: (i) the unfractionated aerolysin prepa-

b ration (C) degraded all the phosphatidylserine,all the phosphatidylethanolamine, and most ofthe phosphatidyl choline. Since no lysophos-phatides were found the degradation is presum-

c ably due to a phospholipase C. (ii) Eluate A,containing all of the recovered hemolytic activ-ity, produced slight phospholipid degradation

d (13.6%) compared to that (30.7%) of eluate Band that (67.1%) of the unfractionated prepara-

tion. (iii) Phosphatidylserine-degrading, phos-

TABLE 6. Composition of reaction mixtures prelimi-nary to thin-layer chromatographic analysLs

Packed 10mMrabbit 100mM sodium

Addition erythrocyte MgC1, boratemembranes

(ml) buffer,(method (ml) pH 8.2

ml) (ml)

(A) Eluate A (1 ml) .... 0.8 0.1 3.1(B) Eluate B (1 ml) .... 0.8 0.1 3.1(C) Aerolysin prepn

(50kg) .......... 0.8 0.1 4.1(D) None .............. 0.8 0.1 4.1

TABLE 7. Distribution of phospholipids in membraneresidues

Total phospholipidPhospholipid (mol/100 mol)0

At, Bb C D'

Phosphatidylserine 5.2 2.2 0 6.7Sphingomyelin ......... 17.2 17.2 17.2 17.2Phosphatidyl choline ... 29.1 18.6 15.7 41.0Phosphatidyl-

ethanolamine ........ 17.9 7.5 0 28.3Lysophosphatidyl choline 4.4 11.9 0 3.0Lysophosphatidyl-

ethanolamine ........ 8.2 11.9 0 3.7Lysophosphatidylserine 4.4 0 0 0

Total phospholipid foundrelative to D (%) .... 86.4 69.3 32.9 100

Total phospholipid de-graded relative to D, bydifference (%)..... . 13.6 30.7 67.1 0

a Assuming no sphingomyelin was degraded.bReaction mixture plus addition as described in

Table 6.

VOL. 11, 1975 1317

Il

1000-u-r- E

9 'E'7: :D 0

a

iI

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.

BERNHEIMER, AVIGAD, AND AVIGAD

phatidylethanolamine-degrading, and phospha-tidyl choline-degrading activities were found inboth eluates, but there was substantially morein eluate B than in eluate A. (iv) Lysophos-phatidyl choline was produced by eluate B butnot by eluate A, suggesting that the formercontains a phospholipase A. (v) Lysophosphati-dyl serine was produced by eluate A, suggestingthe presence of a second phospholipase A.

In general, the results can be explained byassuming (i) that eluate A contains a phospholi-pase A active on phosphatidylserine and phos-phatidylethanolamine, and (ii) that eluate Bcontains a phospholipase A active on phos-phatidylethanolamine and phosphatidyl cho-line, and also a phospholipase C, or combina-tion of other enzymes, active on phosphatidyl-serine.

DISCUSSIONOur results show that erythrocytes from dif-

ferent mammals differ greatly in sensitivity tolysis by aerolysin, those from the rat beingabout 300 times as sensitive as those from thesheep, whereas erythrocytes from other speciesexhibit intermediate degrees of sensitivity orresistance. These findings contrast sharply withthe results of Wretlind et al. (26), and theydiffer also from those of Scholz et al. (23) and ofCaselitz and Gunther (10), which more or lessresemble Wretlind's. Wretlind et al. found onlya 16-fold difference in sensitivity between cellsof the most sensitive and resistant species. Theyfound, moreover, that rat and sheep erythro-cytes were the most resistant among the kindstested, whereas we find that sheep cells are themost resistant but that rat cells are the mostsensitive. Erythrocytes from two additionalstrains of rats exhibited sensitivity comparableto that given in Table 2. The reason for thediscrepancies is not obvious but one explana-tion among others is that the apparent sensitiv-ity may depend upon the state of purity of theaerolysin. This and other possible explanationsare amenable to experimental test.The possibility that sialic acid is involved in

the binding of aerolysin to the erythrocyte isrendered unlikely by the lack of effect of neu-raminidase and by the failure of periodate treat-ment, followed by reduction, to affect thecapacity of membranes to bind the toxin. Otherresults suggest that sulfhydryl groups probablyare not involved in binding. The effect of phos-pholipase C and of proteases, however, indi-cates that phospholipid and protein are directlyor indirectly involved in attachment of aeroly-sin to the membrane. In contrast to experiencewith certain other lysins such as streptolysin S

and staphylococcal 6-toxin, for which the re-ceptors appear to be phospholipids (2), theaction of aerolysin is not inhibited by phos-pholipids even in high concentration (3).Although the major constituent of the

aerolysin preparations employed is aerolysinitself, the preparations, as found earlier byelectrophoretic analysis in polyacrylamide gel(3), contain numerous impurities. The phos-pholipase C-like effect of aerolysin on osmotic-ally prepared erythrocyte membranes, namely,the production of phase-dense lipid droplets("black dots") in membranes, is an effect that isdue largely to protein(s) other than aerolysinitself in the preparations. The effect could bedue to a phospholipase C or to the sequentialaction of a phospholipase D and a phosphatase.In this context both choline and diglyceridehave been found to be reaction products inmixtures of phosphatidyl choline and enzymeextracts of Aeromonas spp. (21).Our findings show that addition of aerolysin

preparations to erythrocyte membranes releasessubstantial amounts of water-soluble phos-phorus. The release is concentration dependent,but unlike binding of aerolysin to membranes itis also temperature dependent, suggesting theinvolvement of one or more enzymes. Fractiona-tion by polyacrylamide gel electrophoresisshowed that most of the phosphate-liberatingactivity was not associated with the hemolytictoxin, but a minor fraction of the phosphate-liberating activity was. The nonaerolysin (non-toxin) activity evidently involves phospholip-ases hydrolyzing phosphatidylethanolamine,phosphatidylserine, and phosphatidyl choline.The failure of any of our preparations to

produce turbidity in diluted egg yolk is surpris-ing at first sight, in view of the aforementionedblack dot-producing activity and in view of thefact that cultures ofAeromonas spp. and relatedbacteria have been shown by Esselmann andLiu (13), Owens (21), and others to produceturbidity in egg yolk-containing media. Theproduction of turbidity in dilute egg yolk byphospholipase C from Clostridium welchii andfrom Pseudomonas aeruginosa has been shownby Kurioka and Liu (15) to be inhibited by ahemolysin from the latter organism, inhibitiondepending upon the hemolysin's detergent-likeeffect in solubilizing various lipids. It is conceiv-able that a similar mechanism may operate inour system. Another possibility is that theprincipal phospholipid of egg-yolk, phosphati-dyl choline, may be a poor substrate for thephospholipase C-like enzyme(s) present in ourAeromonas preparation. It is of interest thatphospholipase A activity also was detected in

1318 INFECT. IMMUN.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.

AEROLYSIN

the preparation. In the few instances in whichphospholipase A has been found in bacteria, itusually is associated with the cell membrane.Our results do not provide definitive informa-

tion regarding the mechanism by whichaerolysin brings about erythrocyte lysis. Al-though most of the phospholipase activity couldbe separated from the toxin (hemolysin), aminor fraction of the enzymatic activity frac-tionated along with the aerolysin. The resultssuggest that the substances responsible for thelast two activities are closely similar in chargeand molecular weight, and that their separationmay therefore prove very difficult. It is tempt-ing to suggest that the small peak visible on theright limb of the main (presumably toxin) peakof Fig. 5 is contaminating phospholipase.

ACKNOWLEDGMENTSWe are indebted to L. Fishman for carrying out the amino

acid analysis. We are grateful to L. L. M. van Deenen forplacing at the disposal of A. W. B. the resources of hislaboratories at the University of Utrecht where part of thework was done. We are grateful also to J. A. F. Op den Kamp,B. Roelofsen, R. F. A. Zwaal, and P. Comfurious for helpfuladvice.

This study was supported by Public Health Service grantAI-02874 and Public Health Career Program award 5K06-AI-14-198 (to A. W. B.), both from the National Institute ofAllergy and Infectious Diseases.

LITERATURE CMD1. Bartlett, G. R. 1959. Phosphorus assay in column chro-

matography. J. Biol. Chem. 234:466-468.2. Bernheimer, A. W. 1974. Interactions between mem-

branes and cytolytic bacterial toxins. Biochim. Bio-phys. Acta 344:27-50.

3. Bernheimer, A. W., and L. S. Avigad. 1974. Partialcharacterization of aerolysin, a lytic exotoxin fromAeromonas hydrophila. Infect. Immun. 9:1016-1021.

4. Bernheimer, A. W., P. Grushoff, and L. S. Avigad. 1968.Isoelectric analysis of cytolytic bacterial proteins. J.Bacteriol. 95:2439-2441.

5. Blumenfeld, 0. O., P. M. Gallop, and T. H. Liao. 1972.Modification and introduction of a specific radioactivelabel into the erythrocyte membrane sialoglyco-proteins. Biochem. Biophys. Res. Commun. 48:242-253.

6. Bowman, M. H., A. C. Ottolenghi, and C. E. Mengel.1971. Effect of phospholipase C on human erythrocytes.J. Membr. Biol. 4:156-164.

7. Bittcher, C. J. F., C. M. Van Gent, and C. Preis. 1961. Arapid and sensitive sub-micro phosphorus determina-tion. Anal. Chim. Acta 24:202-203.

8. Broekhuyse, R. M. 1968. Phospholipids in tissues of theeye. I. Isolation, characterization and quantitativeanalysis by two-dimensional thin-layer chromatogra-phy of diacyl and vinyl-ether phospholipids. Biochim.

Biophys. Acta 152:307-315.9. Broekhuyse, R. M. 1969. Quantitative two-dimensional

thin-layer chromatography of blood phospholipids.Clin. Chim. Acta 23:457-461.

10. Caselitz, F. H., and R. GUnther. 1960. HMmolysinstudienmit Aeromonasstammen. Zentralbl. Bakteriol. Abt. I.Orig. 180:30-38.

11. Chen, P. S., Jr., T. Y. Toribara, and H. Warner. 1956.Microdetermination of phosphorus. Anal. Chem.28:1756-1758.

12. Coleman, R., J. B. Finean, S. Knutton, and A. R.Limbrick. 1970. A structural study of the modificationof erythrocyte ghosts by phospholipase C. Biochim.Biophys. Acta 219:81-92.

13. Esselmann, M. T., and P. V. Liu. 1971. Lecithinaseproduction by gram-negative bacteria. J. Bacteriol.81:939-945.

14. Hoogeveen, J. T., R. Juliano, J. Coleman, and A.Rothstein. 1970. Water-soluble proteins of the humanred cell membrane J. Membr. Biol. 3:156-172.

15. Kurioka, S., and P. V. Liu. 1967. Effect of the hemolysinof Pseudomonas aeruginosa on phosphatides and phos-pholipase C activity. J. Bacteriol. 93:670-674.

16. Liao, T. H., P. M. Gallop, and 0. 0. Blumenfeld. 1973.Modification of sialogylcoproteins of the human eryth-rocyte membrane. J. Biol. Chem. 248:8247-8253.

17. Lichtman, M. A., and R. I. Weed. 1973. Divalent cationcontent of normal and ATP depleted erythrocytes anderythrocyte membranes, p. 79-93. In M. Bessis, R. I.Weed, and P. F. Leblond (ed.), Red cell shape, physiol-ogy, pathology, ultrastructure. Springer-Verlag, NewYork.

18. Low, D. K. R., J. H. Freer, and J. P. Arbuthnott. 1974.Consequences of sphingomyelin degradation in erythro-cyte ghost membranes by staphylococcal a-toxinsphingomyelinase C). Toxicon 12:279-285.

19. Matsumura, T., and N. Haruhiko. 1973. A simple devicefor transverse slicing of polyacrylamide gel rods. Anal.Biochem. 56:571-575.

20. Moore, S., D. H. Spackman, and W. H. Stein. 1958.Chromatography of amino acids on sulfonated polysty-rene resins. Anal. Chem. 30:1185-1190.

21. Owens, J. J. 1974. The egg yolk reaction produced byseveral species of bacteria. J. Appl. Bacteriol.37:137-148.

22. Reed, C. F., S. N. Swisher, G. V. Marinetti, and E. G.Eden. 1960. Studies of the lipids of the erythrocyte. I.Quantitative analysis of the lipids of normal human redblood cells. J. Lab. Clin. Med. 56:281-289.

23. Scholz, D., W. Scharmann, and H. Blobel. 1974. Leuco-cidic substances from Aeromonas hydrophila. Zen-tralbl. Bakteriol. Abt. I. Orig. Reihe A 228:312-316.

24. Weber, K., and M. Osborn. 1969. The reliability ofmolecular weight determinations by dodecyl-sulfate-polyacrylamide gel electrophoresis. J. Biol. Chem.244:4406-4412.

25. Wretlind, B., L. Heden, and T. Wadstrom. 1973. Forma-tion of extracellular haemolysin by Aeromonashydrophila in relation to protease and staphylolyticenzyme. J. Gen. Microbiol. 78:57-64.

26. Wretlind, B., R. Mollby, and T. Wadstrom. 1971. Separa-tion of two hemolysins from Aeromonas hydrophila byisoelectric focusing. Infect. Immun. 4:503-505.

VOL. 1 l, 1975 1319

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/i

ai o

n 29

Nov

embe

r 20

21 b

y 20

0.12

.31.

108.