INFECTION AND IMMUNITY, Sept. 1977, p. 607-612Copyright C) 1977 American Society for Microbiology

Vol. 17, No. 3Printed in U.S.A.

Rickettsial Hemolysis: Adsorption, Desorption, Readsorption,and Hemagglutination

HERBERT H. WINKLER

Department ofMicrobiology, University of Virginia Medical School, Charlottesville, Virginia 22901

Received for publication 5 April 1977

The energy-dependent adsorption of radioiodinated rickettsiae to sheep eryth-rocytes was demonstrated. The iodination procedure, however, decreased thehemolytic activity of the rickettsiae. No desorption of rickettsiae from isolatedrickettsia-erythrocyte complexes (prevented from lysing by NaF) could be mea-

sured. On the other hand, rickettsiae desorbed from this complex during or afterlysis and readsorbed and lysed other erythrocytes. Thus, the usual hemolyticassay measures multiple rounds of adsorption and lysis. Although lysis of therickettsia-erythrocyte complex was insensitive to anti-rickettsial rabbit serum,

adsorption and readsorption were completely inhibited by such antiserum. He-magglutination of erythrocytes by rickettsiae was observed (in the presence ofNaF to prevent lysis) and was sensitive to the same inhibitors as adsorption.

Bacterial cell envelope interaction with eu-caryotic cell membrane is a cardinal feature ofany mechanism of bacterial invasiveness andhost defense. Rickettsia prowazeki, an obligateintracellular parasite with a morphologicallytypical gram-negative cell envelope (3, 4, 7), isan excellent (although specialized) model forsuch studies because of the extraordinary evolu-tionary pressures on this organism to perfect itsinteraction with its host.

Lysis of an erythrocyte by a rickettsia is acomplex interaction involving adsorptive andlytic steps that can be experimentally distin-guished (8, 9). Such a model system has manyof the properties involved in initiation of para-sitism of a host cell, and its simplicity is a defi-nite experimental advantage. We now knowthat adsorption is not passive but is dependentupon the energy of the rickettsiae, most likely aprotonmotive force (9, 12). Cholesterol is a keypart of the receptor on the erythrocyte (10).Rickettsiae are able to adsorb to erythrocyteghosts and to both inside-out and right-side-outvesicles derived from ghosts (14). Rickettsia-erythrocyte complexes can be prepared basedon the fact that lysis but not adsorption is in-hibited by fluoride ions (9).

In the present study I investigated the follow-ing questions. (i) Can adsorption of radioiodi-nated rickettsiae to erythrocytes be measureddirectly? (ii) Is there a measurable desorption ofrickettsiae from the erythrocyte in the absenceof lysis? (iii) In the presence of NaF to preventlysis, can a rickettsia interact with more thanone erythrocyte so as to effect cross-bridging oferythrocytes and resultant hemagglutination?

(iv) Is an adsorbed rickettsia able to dissociatefrom the erythrocyte after (or during) lysis andthen lyse another erythrocyte?

MATERIALS AND METHODSRickettsiae preparation and growth. R. prowa-

zeki, Madrid E strain, was propagated in embryo-nated, antibiotic-free hen eggs by inoculation with 0.2ml of a 10-5 dilution of a seed pool (yolk sac passageno. 273 and 274). Rickettsial suspensions were pre-pared from heavily infected yolk sacs by a modifica-tion of the methods of Bovarnick et al. and Wissemanet al. (2, 15) as previously described (13). Only fresh,unfrozen rickettsiae were used.The diluent for the rickettsial inoculum and rick-

ettsial suspension in the purification procedure was asucrose-phosphate-glutamate (SPG) solution origi-nally devised by Bovarnick et al. (1). The diluent forthe sheep erythrocytes and hemolysis activity assayswas SPG-Mg containing 0.01 M MgC12.Adsorption and hemolysis. The hemolysis tests

used were modifications of the method of Snyder etal. (11) as previously described (12). Adsorption wasmeasured after the free unadsorbed rickettsiae wereseparated from rickettsia-erythrocyte complexes bycentrifugation for 450 x g for 7 min at 4°C as previ-ously described (8). The supernatant fluid (unad-sorbed rickettsiae) was then assayed for rickettsial he-molytic activity by using additional erythrocytes.Rickettsia-erythrocyte complexes were prepared bywashing the sedimented fraction.

Labeling of rickettsiae. To prepare 'nI-labeledrickettsiae, the organisms were suspended in 1 ml ofsodium phosphate buffer (0.12 M, pH 7.4) to whichwas added: glucose, 100 umol; Na1sI, 20 jtCi; lactoper-oxidase, 2 U; and glucose oxidase, 1 U. The mixturewas incubated at room temperature for 20 min, di-luted with 10 ml of sodium phosphate buffer contain-ing NaASSI3 (10 pM), and centrifuged at 12,000 x g for

607

608 WINKLER

10 min. The rickettsiae were washed in sodium phos-phate buffer and then in SPG, resuspended in SPG,layered on top of 10 ml of Renografin-76 (20%[vol/vol] in SPG), and centrifuged at 27,000 x g for 20min. The rickettsial pellet was then washed and sus-pended in SPG-Mg.Chromium labeling. Erythrocytes were labeled

with 5'Cr by incubating them at 50% (vol/vol) in NaCl(0.85%) with 5'Cr (40 uCi/ml) for 30 min at 230C. Thecells were then washed three times in NaCl and twicein SPG-Mg.Antiserum. Rabbits were injected intradermally

at multiple sites with rickettsiae (1 mg of protein)sonically oscillated with complete Freund adjuvant.They were subcutaneously boosted weekly for 3weeks with rickettsiae (1 mg of protein) sonically os-cillated with incomplete Freund adjuvant. The serumfrom three rabbits was pooled, heat inactivated at56°C for 1 h, and stored at -20°C. The serum wastested for antihemolytic activity by incubating serialdilutions of serum (0.1 ml) with 0.1 ml of rickettsiae(about 0.2 mg of protein per ml) for 10 min at 40C,adding 0.4 ml of sheep erythrocytes (25%), and deter-mining hemolysis after incubation at 340C for 1 h.Inhibition was complete at a 1:20 dilution. Proteinswere determined by the method of Lowry et al. (6).

RESULTSAdsorption of radioiodinated rickettsiae.

Table 1 shows the results of experiments inwhich the adsorption of 125I-labeled rickettsiaewas monitored either indirectly by measuringhemolytic activity remaining in the supernatantfluid after sedimenting the '25I-labeled rickett-sia-erythrocyte complexes or directly by count-ing the unadsorbed radioactivity in the superna-tant fluid after sedimentation of complexes. Al-though adsorption could be readily demon-strated by either method, both methods yieldedvalues lower than routinely obtained with un-treated rickettsiae (in which adsorption wasgreater than 90% [8]). This indicated that theiodination process had impaired the ability ofthe rickettsiae to adsorb. In addition, the totalhemolytic activity of the rickettsiae was alwaysdecreased 50 to 70% by iodination. The additionof NaF to the incubation did not inhibit adsorp-tion, whereas KCN almost completely abolishedit. The effect ofKCN could not be examined bythe hemolytic assay because the KCN-treatedrickettsiae were unable to lyse erythrocytes inthe second incubation. However, as previouslydescribed, adenosine 5'-triphosphate was able torestore the ability of KCN-poisoned rickettsiaeto adsorb to and lyse erythrocytes (12).The time of incubation was varied to further

characterize the impaired ability of the iodi-nated rickettsiae to adsorb to erythrocytes (Fig.1). Panels A and B show adsorption measuredby radioactivity and residual hemolytic activity,respectively. Both assays confirmed that at 00C

TABLE 1. Adsorption of iodinated rickettsiae tosheep erythrocytes

Addition to incubation Adsorptionb (mediUMa A,. 125I

None (control) 58 43NaF (l0mM) 67 59KCN (1 mM) ND 10KCN (1 mM) plus ATP 48 38

(1 mM)a 125I-labeled rickettsiae (0.2 mg ofprotein per ml, 30

to 50 kcpm/ml) were incubated with 2 volumes oferythrocytes (25%, [vol/vol]) for 10 min at 340C withthe indicated additions. ATP, Adenosine 5'-triphos-phate.

bThe erythrocyte-rickettsial complexes and freeerythrocytes were separated from the unadsorbedrickettsiae by low-speed centrifugation. Percent ad-sorption was measured by determining either the he-molytic activity of the unadsorbed rickettsiae or theradioactivity of the unadsorbed rickettsiae relative tothat at 00C. The average of two experiments is shown.ND, Not determinable.

8 16 24 32MINUTES MINUTES

FIG. 1. Kinetics of adsorption of iodinated rickett-siae to sheep erythrocytes. (A) Radioactive rickettsiaeremaining in the supernatant fluid after removal oferythrocytes and erythrocyte-rickettsial complexes bylow-speed centrifugation at the incubation time andtemperature indicated. (B) Residual hemolytic activ-ity in these supernatant fluids. The arrow in (A)indicates the initial radioactivity. For the assay ofhemolytic activity in (B), 1 volume of supernatantfluid was added to 10 volumes of erythrocytes (5%,vol/vol) in SPG plus MgC42 (20 mM) to dilute theinhibitory effects of NaF. Symbols: 0, 0°C; *, 0°Cplus 10 mM NaF; EO, 34°C; U, 34°C plus NaF; A0°C plus 1 mM KCN; A, 34°C plus KCN. BKG,Background radioactivity.

the rickettsiae did not adsorb (8, 9). At 00C (Fig.1B), adsorption in the presence of fluoride ap-peared greater than in its absence because afterdilution some inhibitor remained in the secondincubation. In the presence of NaF at 340C(where adsorption could occur without lysis en-suing [9]), almost complete adsorption occurredin the hemolytic assay but not in the isotopic

INFECT. IMMUN.

RICKETTSIAL HEMOLYSIS 609

assay. This is because the isotopic assay mea-sured both metabolically intact and dead rick-ettsiae, whereas the hemolytic assay measuredonly those able to hemolyze erythrocytes. Theradioactive assay (Fig. 1A) showed that KCNinhibited adsorption, whereas in the hemolyticassay (Fig. 1B) the curve KCN only represents abackground hemolytic level since adsorptioncould not be assayed in this manner in the pres-ence of KCN. The initial rates of adsorptionmeasured with and without NaF were similar.At later times much fewer rickettsiae were asso-ciated with the pelleted complex in the absenceof NaF; thus, the adsorptive capacity of therickettsiae was underestimated. This presum-ably resulted from the loss of rickettsia-erythro-cyte complexes through lysis when NaF was ab-sent.To further characterize dissociation of rick-

ettsiae, the 15I-labeled rickettsia-erythrocytecomplexes were prepared and incubated in theabsence of free rickettsiae (Table 2). In the ab-sence of NaF, hemolysis occurred and rickett-siae rapidly appeared in the supernatant fluid.However, when the erythrocytes were pre-vented from lysing by the continued presence ofNaF, rickettsiae did not desorb from the com-plexes.Lytic dissociation of rickettsiae from

complexes. The dissociation of rickettsiaefrom lysed rickettsia-erythrocyte complexesshown in Table 2 suggested that multiplerounds of adsorption and lysis might occur. Thenumber of rounds would be limited by the insta-bility of rickettsiae and by the accumulation ofa sink of erythrocyte fragments or ghosts thatadsorbed rickettsiae but could not release them(14). To investigate this possibility, rickettsia-erythrocyte complexes were formed in the pres-ence of fluoride to prevent lysis. Native (not io-dinated) rickettsiae were used because of thediminished adsorptive and lytic capacities of thelabeled rickettsiae. The complexes were washedfree from NaF and nonadsorbed rickettsiae at40C. The complexes were then incubated at340C with or without the addition of a twofoldexcess (relative to the unlabeled erythrocytes)of 5lCr-labeled erythrocytes. Hemolysis of thetotal mixture was measured by absorbance at545 nm (A545). The secondary rounds of hemoly-sis were measured by the release of radioactivehemoglobin from the 51Cr-labeled erythrocytes.Hemolysis occurring in the original unlabelederythrocytes and rickettsia-erythrocyte com-plexes was determined by the difference in totalA545 and the calculated A545 contributed by the51Cr-labeled hemoglobin released (Fig. 2). Lysisof the original unlabeled erythrocytes in thepresence of 5'Cr-labeled erythrocytes (curve c,

TABLE 2. Desorption of iodinated rickettsiae fromerythrocyte-rickettsia complexesa

Temp Time Addition to resus- Desorptionb(OC) (min) pension medium (%)

34 60 NaF (lOmM) 434 10 None 3334 20 None 4534 30 None 5134 60 None 72a 125I-labeled rickettsia (0.3 mg of protein per ml)

were incubated with 2 volumes of sheep erythrocytes(25%, vol/vol) in SPG-Mg plus NaF (10 mM) for 15min at 34°C. The complexes were isolated by low-speed centrifugation (1,500 rpm, 7 min) and twowashes in SPG-Mg at 00C.

b After incubation at 34°C of the isolated complexesas described, the desorbed rickettsiae were separatedfrom the complexes by removing the complexes bylow-speed centrifugation and pelieting the rickettsiaein the supernatant fluid at high speed (13,000 x g, 10min).

2.4-a) Cr

1.8-I-1.2-/ b)Orig

~31.2 £0c)OrigCr

0.6-o d)Cr OdC7- ~~~~~~~~~~~~~e)Or'

0 5 10 15 20 30MINUTES

FIG. 2. Readsorption and second-round hemoly-sis. Rickettsiae (3 ml at 0.8 mg/ml) were incubatedwith erythrocytes (6 ml of 12.5%, vol/vol) in SPG-Mgplus 10mMNaF for 30 min at 34 or 0°C. The mixturewas centrifuged (450 x g, 7min), washed once at 0°Cin SPG-Mg plus 10 mM NaF and twice at 0°C inSPG-Mg, and resuspended at 0°C to 4.5 ml in SPG-Mg. The cells originally incubated at 34°C (a, b, c) or0°C (d, e) were then added to equal volumes of 5'Cr-labeled erythrocytes (5(t6, vol/vol) for curves a, c, d,and e or SPG-Mg for curve b. AU mixtures wereshifted to 34°C, and samples were taken at the indi-cated times to determine the Aw (total hemoglobin)and counts per minute (Cr-hemoglobin) of the super-natant to generate the curves, as described in thetext. A conversion factor of 33,700 cpm/A5 unit wasused in this experiment. Lysis of the original unla-beled erythrocytes is depicted in curves b, c, and e,and that of Cr-labeled cells is depicted in curves aand d.

Orig + Cr) was lower than the lysis shown in theabsence of 5'Cr-labeled erythrocytes (curve b,Orig). This suggested that in curve c the secondround of lysis was now occurring principally in

VOiL 17, 1977

610 WINKLER

the 5'Cr-labeled erythrocytes, which were in ex-cess (whereas in curve B the second round musthave occurred solely in the unlabeled erythro-cytes). Indeed, the 51Cr-labeled erythrocytes(curve a) were beginning to lyse at just the timethat curves b and c diverged. The control exper-iments, curves d and e, showed that when theoriginal incubation occurred at 0°C, there waslittle lysis of either the complexes or the added51Cr-labeled erythrocytes since rickettsiae couldnot adsorb at this temperature and no com-plexes were formed.This phenomenon was further elucidated by

using rabbit anti-rickettsiae hyperimmune se-rum. This antiserum when added initially to ahemolytic mixture of rickettsiae and erythro-cytes inhibited hemolysis by more than 90%.Phase-contrast microscopy demonstrated agglu-tination of rickettsiae. However, when the anti-serum was added to such a hemolytic mixture,not initially, but after 30 min (or to rickettsia-erythrocyte complexes), there was much less in-hibition. There was little inhibition of lysis ofthe original unlabeled complexes due to antise-rum (Fig. 3, curves b and c). Curves d and eshow the lysis of the unlabeled complexes in thepresence of 5'Cr-labeled erythrocytes with and

- 2.0-ItIn.4

>. 1.5-

1.0-cti-0

0.5-

0

wi 12 18 24 30

MINUTESFIG. 3. Effect of antiserum on primnary and sec-

ondary rounds of hemolysis. Protocol was similar tothat for Fig. 2, except aU the ceU incubations were at34°C and antiserum (1/10 volume) was added to thewashed complexes, where indicated, before the finalincubation. Curves a (0) and e (A) were calculatedfrom an incubation in which 5ICr-labeled erythrocytesbut no antiserum was added; curves d (-) and f (5)were calculated from one in which both 51Cr-labelederythrocytes and antiserum were added; curve c (V)was calculated from one in which antiserum but no51Cr-labeled erythrocytes were added; and curve b(0) was calculated from one in which neither wasadded. Thus, curves a and f represent lysis of the51Cr-labeled erythrocytes, and curves b, c, d, and erepresent lysis of the original unlabeled erythrocytein the presence of the indicated additions. The con-version factor in this experiment was 31,500 cpm/A545unit for generation of the curves as described in thetext.

without added antiserum. These two curves

were almost identical to each other and to curve

c, in which antiserum but no 5'Cr-labeled eryth-rocytes was added. As shown by a comparisonof the lysis of 5'Cr-labeled erythrocytes withand without antiserum (curves a and f), the sec-

ondary rounds of lysis (those that occurred inthe 5'Cr-labeled erythrocytes) were completelyinhibited by the antiserum. These data indicatethat the preformed complexes could lyse andthat antiserum had no effect on this process.However, the secondary rounds were com-pletely inhibited, presumably by agglutinationof the dissociated rickettsiae before they read-sorbed.Hemagglutination of sheep erythrocytes.

Although lysis of erythrocytes by typhus rick-ettsiae has been known for many years (5, 11)and adsorption of rickettsiae to erythrocyteshas been extensively studied in this laboratory(8, 10, 12, 14), agglutination of erythrocytes bytyphus rickettsiae has never been reported.Since irreversible adsorption of rickettsiae toerythrocytes is not a passive process but, rather,one that requires metabolic activity of the rick-ettsiae, by analogy, hemagglutination wouldalso require metabolic energy. To observe he-magglutination, erythrocyte lysis must be pre-vented. This can be accomplished by addingNaF, which prevents lysis without affecting ad-sorption (respiratory inhibitors or low tempera-ture would inhibit both lysis and adsorption).Figure 4 demonstrates agglutination of eryth-

rocytes by R. prowazeki. The process was in-hibited by cyanide (1 mM), 2,4-dinitrophenol (1mM), and N-ethylmaleimide (1 mM). The detec-tion of hemagglutination required large quanti-ties of rickettsiae, quantities that, in the absenceof fluoride, could have lysed many times over,the amount of erythrocytes agglutinated. Arti-facts (false positive) arose from the settling ofthe rickettsiae, which obscured the erythrocytepellet when even larger quantities of rickettsiaewere used. Decreasing the amount of erythro-cytes to almost the limit of detection onlyslightly increased the sensitivity with respect torickettsiae concentrations, so that under no con-ditions has this phenomenon been a practicalassay for rickettsiae.

DISCUSSIONThis study defines several new aspects of

rickettsial hemolysis. For example, rickettsiaeare able to adsorb to more than one erythrocyteand display the resultant hemagglutination.Thus, the rickettsial envelope does not have asingle unique site for adsorption, and the occu-pancy of an adsorptive site has not precludedthe function of other such sites. However, once

a) Cr

b) Orig

, c) Orig+Ab, )d)Orig+Cr+Ab

, A ,A A AA

e) Orig+Cr

1n alO f) Cr+Ab

INFECT. IMMUN.

RICKETTSIAL HEMOLYSIS 611

., .i.

...... ... ....... i , : ... .* .j,, . .. -i .; . ... . .. -.... ; < . ...... ..jY *.,. t * i * xs .. .-KS.W ,, i . R . .. : . .:* - W .. ffi.g. s .-..... ... . ..Y 't .4 r. 2#>. M. ,ts... .. ;

.. .. s,

8 SE iy A

FIG. 4. Hemagglutination of sheep erythrocytes. Sheep erythrocytes were washed and suspended in SPG-Mg as for a hemolysis assay. Sodium fluoride was present throughout at 10 mM to inhibit lysis, and fetalcalf serum (1%) was added throughout to aid in settling of the erythrocytes. Rickettsiae were in SPG-Mgbuffer; 50 ,1 at 4 mg/ml was added to the first well, and serial twofold dilutions were made. No rickettsiaewere in the last two wells. Erythrocytes were added as 50-,Il portions of a 1% suspension. N-ethylmaleimide,cyanide, and dinitrophenol were all present at 1 mM where indicated. The components were mixed andincubated for 1 h at 23°C and then overnight at 4°C.

adsorbed, the rickettsiae have a very low proba-bility of desorbing and readsorbing onto anothererythrocyte without lysis taking place. Sincethese studies had to be done in the presence ofNaF, it was possible that NaF could have pre-vented not only lysis, but also desorption. How-ever, previous studies with ghosts, which couldnot be lysed even in the absence of NaF, showedthat once rickettsiae were attached they wouldnot desorb and lyse added erythrocytes, despitethe fact that the erythrocytes adsorbed rickett-siae better than ghosts (14). On the other hand,rickettsiae do dissociate from erythrocyteseither during or after lysis. It is not knownwhether the rickettsia separates completelyfrom the lysed erythrocyte. It may retain asmall patch (the receptor) of the erythrocytemembrane still attached to its adsorptive site.These dissociated rickettsiae are capable of ad-sorbing to and lysing other erythrocytes. Thus,the hemolytic assay as usually done does notmeasure a single-step process but the result ofmultiple rounds of lysis. However, completelysis of erythrocyte populations is not usuallyseen, since a large excess of erythrocytes is usedand the number of cycles is limited by the labil-ity of the rickettsiae and the creation of nonlys-able sinks for rickettsiae, namely, the ghostsformed during lysis.The use of anti-rickettsial serum allows the

distinction of the primary and secondary roundsof lysis. The antibody has no effect on lysis ofthe prefonned rickettsia-erythrocyte complexbut agglutinates free rickettsiae and preventstheir adsorption or readsorption.

Iodination of rickettsiae to label them forphysiological experiments seems ill advised.With the very gentle lactoperoxidase-glucoseoxidase technique (or harsher procedures),about 50% of the hemolytic activity was lost.Moreover, the fractions that remained hemolyt-ically active were not native in that their rateof adsorption was markedly inhibited.

ACKNOWLEDGMENTSThis work was supported by Public Health Service re-

search grant AI-10164 from the National Institute of Allergyand Infectious Diseases.

I thank Beth Miller and Kathy Pettiss for expert technicalassistance.

LITERATURE CITED

1. Bovarnick, M. R., J. C. Miller, and J. C. Snyder. 1950.The influence of certain salts, amino acids, sugars, andproteins on the stability of rickettsiae. J. Bacteriol.59:509-522.

2. Bovarnick, M. R., and J. C. Snyder. 1949. Respirationof typhus rickettsiae. J. Exp. Med. 89:561-565.

3. Brinton, L P., and W. Burgdorfer. 1971. Fine struc-ture of Rickettsia canada in tissues of Dermacentorandersoni Stiles. J. Bacteriol. 105:1149-1159.

4. Burton, P. R., J. Stueckemann, and D. Paretsky.1975. Electron microscopy studies of the limiting layersof the rickettsia CoxieUa burneti. J. Bacteriol.122:316-324.

5. Clarke, D. H., and J. P. Fox. 1948. The phenomena ofin vitro hemolysis produced by typhus fever with anote on the mechanism of rickettsial toxicity in mice. J.Exp. Med. 88:2541.

6. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J.Randall. 1951. Protein measurements with the Folinphenol reagent. J. Biol. Chem. 193:265-275.

7. Murray, R. G. E., P. Steed, and H. E. Elson. 1965. Thelocation of the mucopeptide in sections of the cell wallof Escherichia coli and other Gram-negative bacteria.

VOL. 17, 1977

INFECT. IMMUN.

Can. J. Microbiol. 11:547-560.8. Ramm, L E., and H. H. Winkler. 1973. Rickettsial he-

molysis: adsorption of rickettsiae to erythrocytes. In-fect. Immun. 7:93-99.

9. Ramm, L E., and H. H. Winkler. 1973. Rickettaisi he-molysis: effect of metabolic inhibitors upon hemolysisand adsorption. Infect. Immun. 7:550-555.

10. Ramm,L E., and H. H. Winkler. 1976. Identification ofcholesterol in the receptor site for rickettsiae on sheeperythrocyte membranes. Infect. Immun. 13:120-126.

11. Snyder, J. C., M. R. Bovarnick, J. C. Miller, and R. 8.

Chang. 1954. Observations on the hemolytic propertiesof typhus rickettsiae. J. Bacteriol. 67:724-730.

12. Winkler, H. H. 1974. Inhibitory and restorative effects ofadenine nucleotides on rickettsial adsorption and he-molysis. Infect. Immun. 9:119-126.

13. Wilner, H. H. 1976. Rickettsial permeability: an ADP-ATP transport system. J. Biol. Chem. 251:390-396.

14. Winkler, H. H., and L E. Ramm. 1975. Adsorption oftyphus rickettsiae to ghosts of sheep erythrocytes. In-fect. Immun. 11:1244-1251.

15. Wisseman, C. L, Jr., E. B. Jackson, F. E. Hahn, A. C.Ley, and J. E. Smadel. 1951. Metabolic studies ofrickettsiae. L. The effects of antimicrobial substancesand enzyme inhibitors on the oxidation of glutamate bypurified rickettsiae. J. Immunol. 67:123-136.

612 WINKLER