Thrombocytopenia in ExperimentalTrypanosomiasis

Charles E. Davis, … , Richard D. Weller, Abraham I. Braude

J Clin Invest. 1974;53(5):1359-1367. https://doi.org/10.1172/JCI107684.

The effect of experimental trypanosomiasis on coagulation was studied because a patient inthis hospital with Rhodesian trypanosomiasis developed thrombocytopenia withdisseminated intravascular coagulation. Rats injected intraperitoneally with this strain ofTrypanosoma rhodesiense consistently developed trypanosomiasis and severethrombocytopenia without changes in hematocrit or concentration of fibrinogen or fibrin splitproducts. At the time of 50% mortality (4-5 days) mean platelet counts per cubic millimeter ofinfected rats were 18,000±9,000 (±2 SEM) compared to 1,091,000±128,000 in uninfectedcontrols.

In vitro, concentrated trypanosomes and trypanosomefree supernates of disruptedorganisms added to normal rat, rabbit, or human blood produced platelet aggregation within30 min. This platelet aggregation was not blocked by inhibitors of ADP, kinins, or early orlate components of complement. In vivo thrombocytopenia also occurred in infected rabbitscongenitally deficient in C6 and in infected, splenectomized rats.

Although the aggregating substance obtained from disrupted trypanosomes is heat-labile, itis active in the presence of complement inhibitors, suggesting that this trypanosomalproduct may be a protein enzyme or toxin. Since the phenomenon is independent ofimmune complexes, complement, ADP, and kinins, it appears to represent a newmechanism of microbial injury of platelets and the induction of thrombocytopenia.

Research Article

Find the latest version:

http://jci.me/107684-pdf

Thrombocytopenia in Experimental Trypanosomiasis

CHARLEsE. DAVIS, ROBERTS. ROBBINS, RIcIIAIW D. WETmER,andABRAHAMI. BRAUDE

From the Departments of Pathology and Medicine, University of California,San Diego 92103

A B S T R A C T The effect of experimental trypanosomi-asis, on coagulation was studied because a patient in thishospital with Rhodesian trypanosomiasis developedthrombocytopenia with disseminated intravascular coagu-lation. Rats injected intraperitoneally with this strainof Trypanosoma rhodesiense consistently developedtrypanosomiasis and severe thrombocytopenia withoutchanges in hematocrit or concentration of fibrinogen or

fibrin split products. At the time of 50% mortality (4-5days) mean platelet counts per cubic millimeter of in-fected rats were 18,000+9,000 (+2 SEM) compared to

1,091,000--+128,000 in uninfected controls.In vitro, concentrated trypanosomes and trypanosome-

free supernates of disrupted organisms added to normalrat, rabbit, or human blood produced platelet aggrega-tion within 30 min. This platelet aggregation was notblocked by inhibitors of ADP, kinins, or early or latecomponents of complement. In vivo thrombocytopeniaalso occurred in infected rabbits congenitally deficient inC6 and in infected, splenectomized rats.

Although the aggregating substance obtained fromdisrupted trypanosomes is heat-labile, it is active in thepresence of complement inhibitors, suggesting that thistrypanosomal product may be a protein enzyme or toxin.Since the phenomenon is independent of immune com-plexes, complement, ADP, and kinins, it appears to

represent a new mechanism of microbial injury ofplatelets and the induction of thrombocytopenia.

INTRODUCTIONWe studied the effect of experimental trypanosomiasison coagulation factors because a patient in this hospitalwith Rhodesian trypanosomiasis almost died of dissemi-nated intravascular coagulation (DIC)' (1). Various

An abstract of a portion of this work appeared in Clin.Res. 1973. 21: 269.

Received for publication 16 July 1973 and in revisedform 2 November i973.

1 Abbreviations used in this paper: DIC, disseminated

severe infections may cause DIC, but it is usually as-sociated with meningococcemia or septicemia due togram-negative bacilli. Among the protozoan infectionsonly severe falciparum malaria is well documented as acause of this syndrome (2, 3).

Wewere surprised to find that 100% of rats infectedwith this Rhodesian strain developed severe thrombo-cytopenia without evidence of hemolysis or DIC. Thisdissociation of thrombocytopenia from DIC stronglyindicates a direct toxic effect of the trypanosomes onplatelets. Although it has been speculated that Plasmo-dium vivax may produce circulating toxins directlyharmful to platelets (4), most studies suggest that thethrombocytopenia associated with malaria is due to DIC(2, 3), immunologic injury (5), or splenic pooling (6).In bacterial septicemias the thrombocytopenia is clearlyassociated with DIC (7).

In this study we reproduced platelet injury in vitroby whole trypanosomes and with soluble extracts of dis-rupted trypanosomes. Since the process is independentof immune complexes and complement and is not blockedby inhibitors of ADP or kinins, it appears to representa new mechanism of microbial injury of platelets andthrombocytopenia.

METHODSExperimental trypanosomiasis. Male 200-g Holtzman

rats or 2.5-3.0 kg white New Zealand rabbits were injectedi.p. with a 0.15 M NaCl dilution of infected, fresh rat bloodcontaining 10-20,000 trypanosomes, or more than 1,000 in-fective doses for rats. Syringe-passed trypanosomes main-tained in rats were used in all experiments. During thefirst few passages this strain of T. rhodesiense becamemore virulent for rats, but the clinical syndrome and timeof LDwoo has been stable for 1 yr. Control animals receivedeither an equal volume of normal rat blood or nothing.Normal rat blood caused no visible illness and no blood

intravascular coagulation; EM, electron microscopy; FSP,fibrin and fibrinogen split products; HCT, hematocrit;IBS, isotonic buffered saline; PRP, platelet-rich plasma;PSG, phosphate-buffered saline glucose solution.

The Journal of Clinical Investigation Volume 53 May 1974-1359-1367 1359

changes. Studies on splenectomized rats were performedafter wounds were well healed and platelet counts hadreturned toward normal. Rabbits congenitally deficient inC6 were infected in the usual manner. Rats were exsan-guinated by cardiac puncture at the time of study. Rabbitswere bled from the central artery of the ear. Autopsieswere performed on 18 rats at the time of 50% mortality.Serial sections of formaldehyde-fixed kidneys, livers,spleens, lung, hearts, and brains were stained with hema-toxylin and eosin and by the periodic-acid Schiff method.

Hematological studies. Platelets and trypanosomes werecounted by phase microscopy. Occasionally, nonmotile try-panosomes assume a spherical shape but can still be differ-entiated from platelets by careful observation. Microhemato-crits (HCT) were determined by centrifugation of EDTA-treated blood in capillary tubes. Fibrinogen was measuredby the method of Ratnoff and Menzie (8) but with citratedblood, and split products of fibrin and fibrinogen (FSP)were estimated by the staphylococcal clumping technique(9). The details of these procedures have been describedin another publication (10).

In vitro platelet aggregation. Normal rat, rabbit orhuman blood was harvested in 0.008 mg/ml EDTA(EDTA: blood = 1: 9; vol: vol). Blood from infected ratswas harvested either in EDTA or 0.38 mg/ml sodiumcitrate (1: 9; vol: vol). Unless otherwise noted, EDTAand sodium citrate were always used at these standardanticoagulating concentrations. Trypanosomes were har-vested from a visible layer on top of the buffy coat ofinfected rat blood and stored at 40C until use. These con-centrated trypanosomes were platelet-poor and containedno platelet aggregates.

To 1 ml of normal blood harvested in EDTA, 0.1 ml ofconcentrated trypanosomes (W10-107/mm3) or 0.1 ml ofnormal, species-specific plasma was added. The tubes wererocked continuously at 36 oscillations/min on an Amesaliquot mixer (Ames Co., Div. of Miles Lab, Inc., Elk-hart, Ind.) at room temperature, and platelet counts andother determinations were performed at 30 min. All plate-lets that could be counted accurately were incorporated inthe tally, including platelets in small aggregates of fouror five.

Investigation of the mechanism of platelet injury. Theresponse of the platelet counts of infected animals to in-creasing trypanosomemia was measured by infecting 36rats and exsanguinating at least 6 daily. Trypanosomes andplatelets were counted by phase microscopy, and the resultswere plotted as platelet counts versus trypanosome counts.Six littermates were used as the normal controls. The "dose-response" relationship of the aggregation phenomenon wasmeasured by noting the effect of 0.1 ml of serial 10-folddilutions of trypanosomes (8.8 X 10W/mm3, 8.8 X 105/mm3,and 8.8 X 104/mm') on 1 ml of a pool of normal rat blood.The negative control consisted of 0.1 ml of normal plasmaharvested from the same pool of rat blood.

Platelet aggregation was tested under several experi-mental situations designed to elicit the nature of the medi-ator of trypanosome-induced aggregation. It was likely thatfairly large amounts of exogenous ADP from the trypano-somes and rat red cells were added to normal blood duringthe in vitro aggregation experiments. Accordingly, verylarge concentrations of ADP (up to 60 mg/ml) were addedto the EDTA-treated normal blood instead of trypano-somes. The ability of trypanosomes to induce platelet ag-gregation in whole blood or platelet-rich plasma (PRP)pretreated with 50 mMadenosine was determined. Exoge-

nous ADP was also removed by dialyzing the harvestedtrypanosomes overnight against phosphate-buffered salineand by separating the trypanosomes from rat blood in aQAE-50 Sephadex column with a phosphate-buffered saline-glucose (PSG) solution (11). Dialyzed trypanosomes andcolumn trypanosomes, which were completely free of raterythrocytes and platelets, were washed with 10 vol ofPSG, diluted to 8 X 106/mm', and tested for the aggregatingeffect as described above.

In order to test whether complement components wererequired for trypanosome-induced platelet aggregation, theblood of rabbits congenitally deficient in C6 was drawn into0.2 M EDTA, a concentration sufficient to inactivate C4and C2, and the usual concentration of trypanosomes wasadded.

Participation of kininlike substances in the platelet in-jury was checked by adding trypsin (50-500 jug/ml), alpha-chymotrypsin (50-500 ug/ml), or an equal volume of salineto harvested trypanosomes for 30 min at 370C. Lima beantrypsin inhibitor was then added to each sample of try-panosomes before addition of 0.1 ml of the treated trypano-somes to 1.0 ml of rat blood.

The effect of washing the trypanosomes before additionto normal blood was tested by washing the harvested try-panosomes from an aliquot of plasma three times in 10vol of 0.15 M NaCl. In the in vitro system, the washedtrypanosomes, the trypanosome-free infected plasma con-centrated five times, and the washings reduced in volumeat 4°C in dialysis tubing to only twice the original plasmavolume were tested for their ability to cause platelet ag-gregation.

1 ml of trypanosomes (l10) harvested from rat bloodwas sonicated in ice by means of a biosonic probe with oneor two 10-s bursts at 30% intensity. This short period dis-rupted approximately 90-95% of the trypanosomes. Thesonicated material was then made trypanosome-free by cen-trifugation at 3,500g for 20 min. 1 ml each of the untreatedsupernate, supernate preheated at 56°C for 30 min, andsupernate treated with trypsin or alpha-chymotrypsin wasadded to 1 ml of normal rat blood.

Electron microscopy (EM). Column-purified trypano-somes or trypanosome-free sonicate were added to tubes ofcitrated rat PRP and gently rocked at room temperature.Samples were withdrawn at 30 min and 90 min for fixationand preparation for EM, along with control samples ofPRP.

For transmission EM the samples were added to equalvolumes of 0.2% glutaraldehyde in isotonic buffered saline(IBS) centrifuged at 1,500g for 15 min to form a loosepellet, fixed in 2% glutaraldehyde in IBS for 2 h, post-fixed in 2% osmium tetroxide in IBS for 1 h, dehydrated,and embedded in Araldite (Ciba Products Co., Summit,N. J.). Sections were cut with a Porter-Blum MT-2Bultramicrotome and stained with lead citrate and uranylacetate. They were examined with a Zeiss 9S electron mi-croscope (Carl Zeiss, Inc., New York).

For scanning EM samples were pipetted onto glasscover slips and fixed in place with 0.2%o glutaraldehyde inIBS for 15 min, followed by 2%o glutaraldehyde in IBS for1 h. After dehydration in an ethanol-Freon series andprocessing in a Bomar SPC-900 critical point drying appa-ratus (The Bomar Co., Tacoma, Wash.), they weremounted on aluminum studs, coated with a 400-600-Alayer of gold-palladium, and examined with an Etec U-1scanning electron microscope (Etec Corp., Hayward, Calif.).

1360 C. E. Davis, R. S. Robbins, R. D. Weller, and A. I. Braude

a

Q0

a-0.

Co

CL._

._0

.0

FIBRINOGEN

Days Post- Infection

FIGURE 1 Effect of experimental trypanosomiasis on the platelet count, fibrinogen, and FSPconcentration. Rats were infected with 10-20,000 trypanosomes intraperitoneally. Each daythree normal and three infected rats were exsanguinated and studied until all rats were dead.Each point represents the mean platelet count and fibrinogen and FSP concentration of threerats.

RESULTS

Experimental trypanosomiasisMore than 500 rats and 50 rabbits were studied. All

infected rats died between 96 and 144 h with no obvioussigns of illness until shortly before death. Rabbits fol-lowed a more erratic course and died after a cacheticillness of 3-6 wk. The effect of experimental trypanoso-miasis on the platelet count, Hcr, and fibrinogen wasdetermined 96 h after infection (24 h before the firstspontaneous death occurred) and at 120 h, the deathpoint in rats not subjected to blood studies. As shownin a typical experiment in Table I, the Hcr and fibrino-gen concentrations of infected rats did not differ fromuninfected controls. The platelet counts, on the otherhand, were markedly reduced in infected rats at 96 hand again fell sharply during the 24 h before death, whenthe number of trypanosomes in the blood was increasingrapidly. In fact an inverse relationship between thenumber of trypanosomes and platelets suggested a directeffect of trypanosomes or their products on the platelets.No platelet aggregates were ever noted in the plateletcounts performed during the in vivo experiments.

Next a group of rats were studied daily to be certainthat DIC with early consumption of fibrinogen andproduction of FSP was not responsible for the thrombo-cytopenia that occurred late in the disease. The rangeof FSP in more than 25 normal rats was 2-16 tg/mlwith a mean of 8 /g/ml +0.68 (2 SEM). When three

different rats were studied daily (Fig. 1) until all in-fected controls died, the only abnormality was a dailyfall in platelets.

Gross and microscopic examinations of the kidneys,livers, spleens, lungs hearts, and brains of infected ratsfailed to demonstrate vascular lesions or fibrin deposition.Splenomegaly was the only finding. If the thrombocyto-penia were due to removal of normal or injured plate-lets by the enlarged spleen, splenectomized animals

TABLE IEffect of Experimental Trypanosomiasis on Platelet

Count, Fibrinogen, and HCT

Platelet count Fibrinogen HCT

X1O3/mm3 mg/100 ml %96 h after infection

Infected 3354114* 210±18 41i3Normal 1,064±106 185±9 41+0.7

120 h after infectionInfected 18±9 257440 37±3.6Normal 1,091 ± 128 207±i 14 41±4-1.6

Male Holtzman rats were infected intraperitoneally with10-20,000 trypanosomes. Normal littermates were not in-fected. Six animals in each group were exsanguinated at thetime 50% mortality was reached in a control group (120 h)and approximately 24 h earlier. All infected rats died within144 h.* ±2 SEM.

Thrombocytopenia in Experimental Trypanosomiasis 1361

1, 00p00o

2 900E 800S 700

't± 600"E 500

6<,, 400I.-L 300

- 200

X 100

+ ±2SEM

. . . . . . . . -. -_

0 200 400 600 800 1)00 l300 1500 1,700V 00TRYPANOSOMES/mm3(+2S E M )x103

FIGURE 2 Dose response relationship of thrombocytopeniato trypanosomemia. 36 rats were injected on day 0 andat least 6 exsanguinated daily until all were dead. Sixuninfected littermates were used as uninfected controls.Tryparnosomes and platelets were counted by phase micros-copy and the results plotted as platelet counts vs. trypano-some counts. Note the inverse relationship between thenumbers of trypanosomes and platelets.

might not show thrombocytopenia or might have adifferent clinical course with obstruction of small ves-sels and earlier death. As shown in Table II, splenec-tomized animals also displayed thrombocytopenia afterinfection. No platelet aggregates were seen. There wasno difference in the clinical course or the time of 50%or 100% mortality of infected splenectomized or non-splenectomized animals.

To rule out the possibility that the thrombocytopeniaoccurred in vitro by reaction of intact trypanosomesand platelets between the time of exsanguination andthe platelet count, six infected rats were treated withsuramin (3 mg/kg) 24 h before the estimated time ofdeath. When they were exsanguinated 24 h later, thetrypanosomemia had cleared, but striking thrombocyto-penia persisted with a mean platelet count of 128+62X 10' platelet/mm' (±2 SEM) and was nearly as severe

TABLE I IEffect of Trypanosomiasis on Platelet Counts, Fibrinogens,

and HCTof Splenectomized Rats

Rats Platelet count Fibrinogen HCT

X103/mm3 mg/1OO ml %Uninfected 1,200468 227±24 43+2Infected 204±56 251±16 43±5

Holtzman male rats (200-250 g) were splenectomized underether anesthesia. 4 wk later when the platelet counts wereapproaching normal, 15 rats were infected intraperitoneallywith 10,000 trypanosomes. When 50% of the infected ratshad died (5 days), five infected rats and five uninfected,splenectomized rats were exsanguinated and studied. All in-fected, splenectomized rats died within 144 h. Values areAtSEM.

TABLE IIIIn Vitro Effect of Nonmotile Trypanosoma rhodesiense on

Rat Platelets, White Cells, and Hct

Effect of 0.1 ml of Platelet count White cells HCT

XlOS/mm3 %

Rat plasma 998±19* 15,400 37.5Trypanosomes 113±9 14,050 37.0

* ±2 SEM.

TABLE IIIaAggregation of Rabbit and HumanPlatelets by

Trypanosoma rhodesiense

Platelet countsh2 SEM

Effect of 0.1 ml of Rabbit Human

X103/mm'

Homologous plasma 715±136 269±33Trypanosomes 65±4 13 147 ± 15

Trypanosomes were harvested from the top of the buffy coatof infected rats. To 1.0 ml of whole rat blood anticoagulatedwith EDTA, 0.1 ml of normal rat plasma or concentratedtrypanosomes (108) was added. The trypanosome suspensionsare poor in platelets, white cells, and red cells. The tubes wererocked on an Ames aliquot mixer at room temperature andcell counts and HCTSmeasured at 30 min. All platelets thatcould be counted were included in the tally including those insmall aggregates of four or five. Most of the platelets in theblood treated with trypanosomes were in aggregates of greaterthan 10, and aggregates of greater than 50 were notuncommon.

The figures in Table III represent 14 paired determinationsfor'platelets and five paired determinations for white cellsand HCT. The figures in Table Illa represent the effect on theplatelets of five individuals studied simultaneously.

as that of six infected littermates bled 24 h earlier (62±36 X 10) at the time the treated rats were givensuramin. Suramin (3 mg/kg) did not alter the plateletcount of six uninfected rats (1,053±60 X 10') comparedto six untreated littermates (1,042±40 X 10'). Thisexperiment did not rule out an effect of circulatingtrypanosomal products on platelets.

In vitro platelet aggregationAddition of 107 to 10" trypanosomes to 1 ml of normal

rat blood invariably caused platelet aggregation in vitro.Nonmotile trypanosomes stored in the refrigerator forup to 1-2 wk were as effective as fresh, motile trypano-somes. A typical experiment is shown in Table III.After 30 min most of the platelets were found inaggregates of greater than 10, and aggregates of 50were not uncommon. As shown in Table III, therewas no effect on the HCT or white cell count. Multipledeterminations have shown no effect on fibrinogen con-

1362 C. E. Davis, R. S. Robbins, R. D. Weller, and A. 1. Braude

centration. FSP were not studied in vitro. The aggre-gation phenomenon occurred equally in whole blood orPRP at temperatures of 22°-37°C and was effectivein blood anticoagulated with EDTA, sodium citrate, orheparin. Heparin was not used routinely, however,because spontaneous aggregation occurred in heparin-ized rat blood.

Aggregation significantly lowered the platelet countby 15 min and was virtually complete by 30 min. Therewas no significant change between 30 and 180 min.Continued oscillation at 36/min or rapid vortex mix-ing for 30 s (Vari Whirl, Van Waters and Rogers,Inc., San Francisco, Calif.) did not change the plateletcount or the number of aggregates. Vigorous agitationon a Burrell Wrist Action Shaker (Burrell Corp.,Pittsburgh, Pa.) at 210 cycles/min for 5 min physicallysheared some aggregates, increasing both the numberof aggregates and the number of platelets. When a 30-min preparation with a platelet count of 28 X 103/mm3and 26 X 103 large aggregates/mm3 was subjected to 5min agitation on the wrist action shaker, the plateletcount increased to 184 x103/mm3 and the number ofaggregates to 76 X 103/mm3.

As shown in Table IIIA, trypanosomes also causedaggregation of human and rabbit platelets.

Mechanism of platelet injuryBecause platelet aggregation was not detected at any

stage of infection, it was possible that the in vivothrombocytopenia was not related to in vitro aggrega-tion. For example, the in vitro phenomenon might beADP-induced aggregation that would not occur in vivobecause the smaller circulating concentrations of ADPavailable in vivo at any given time might be hydrolyzedto inactive metabolic products. It did not seem likely,however, that the aggregation was due to ADP becauseit occurred in the presence of strong chelating agentssuch as EDTA, known to interfere with ADP-inducedaggregation (12). Other more likely explanations thatwould relate the in vivo and in vitro phenomena in-cluded immunologic injury and platelet damage by a

trypanosomal enzyme or toxin. Immunologic injurywould most likely be mediated by antigen-antibodycomplexes with platelet injury by immune adherence,late components of complement, or pharmacologicallyactive substances such as kinins.

Dose-response relationships. An inverse relationshipbetween the numbers of trypanosomes and platelets inboth the in vivo and in vitro models first suggestedthat thrombocytopenia in vivo and aggregation in vitromight be due to a direct effect of the trypanosomes ortheir products on the platelets. Further exploration ofthis relationship showed interesting dose-response curves

(Figs. 2 and 3). The number of trypanosomes neces-

1,1000oo0900

° 800

"E 700

E- 600cn

500u 400-i 300

200100

0 200 400 600 800 P00 k200 1400 1,00 M0O 000

TRYPANOSOMES/mm3x 103

FIGURE 3 Dose response relationship of the aggregationphenomenon to the concentration of trypanosomes. Serial10-fold dilutions of concentrated trypanosomes were eachadded to 1 ml of a pool of normal rat blood and the plateletcounts determined after 30 min. The negative control con-sisted of the same volume (0.1 ml) of normal plasma har-vested f rom the same pool of rat blood. Note that thenumber of trypanosomes needed to cause aggregation (about800 X 103/mm3) is the same as that necessary to causethrombocytopenia in vivo (Fig. 2).

sary to cause minimal but significant in vivo thrombo-cytopenia was the same (about 800 X 103) as the num-ber necessary to initiate the in vitro aggregation phe-nomenon. Further, the shapes of the curves and thepoints of "maximal" effect were similar.

ADP-induced aggregation. The possibility that thein vitro phenomenon was due merely to the additionof exogenous ADP from trypanosomes or hemolyzedred cells was tested and disproved in the followingways: (a) ADP concentrations 6,000 times higher thanthe usual aggregating dose of 0.3-1.0 Ag/ml (13) didnot cause aggregation in EDTA-treated whole blood orrat PRP; (b) trypanosome-induced aggregation wasnot inhibited by 50 mMadenosine, either in EDTA-

TABLE IVIn Vitro Aggregating Effect of Trypanosoma rhodesiense on

Rabbit Platelets in a Decomplemented System

C6 deficientrabbit blood(collected in

Effect of 0.1 ml of 0.2 MEDTA)

X10'/mm'

Trypanosomes harvestedin 0.2 MEDTA 194

Normal plasma harvestedin 0.2 MEDTA 730

Trypanosomes were harvested from rat blood drawn into0.2 M EDTA and added to C6-deficient rabbit blood alsoharvested in 0.2 M EDTA. In this system complementmediated immune adherence and C lysis were both blocked.The platelets were counted at 30 min as outlined in the textand in Table III.

Thrombocytopenia in Experimental Trypanosomiasis

--- --- --- -.- ---- -.- .-.- .-.- ;I-

1363

TABLE VPlatelet Aggregation by Trypanosomes in the Presence of Kinin Inhibition

Trypanosomes pretreated with:

Lima beanSaline control Trypsin Alpha-chymotrypsin trypsin inhibitor

(no trypanosomes) Untreated (50 jtg/ml) (50 /g/ml) (125 /Lg/ml)1,100 168 132 136 162

Platelet count X 103/mm3Platelet counts from pooled blood of five rats. After harvesting, trypanosomes were treated with the indicated con-centration of trypsin or alpha-chymotrypsin for 30 min at 370C. Trypsin was inactivated by lima bean trypsin in-hibitor before trypanosomes were added to the rat blood. Note that the lima bean trypsin inhibitor also had no effecton aggregation. Platelets were counted at 30 min as outlined in the text and in Table III.

plasma or heparinized plasma. At this concentrationadenosine is a powerful inhibitor of ADP-induced plate-let aggregation (13); (c) removal of ADP from try-panosome concentrates by dialysis and Sephadex QAE-50 did not alter the aggregating effect. At the sameconcentration of trypanosomes (8 X 106/mm3), aggre-gation resulted in platelet counts of 28 X 10'/mm' withthe usual buffy coat trypanosomes, 34 X 103 with di-alyzed trypanosomes, and 26 X 103 with column trypano-somes. The plasma control was 1,100 X 103/mm3.

Complement inhibition. Because immune complexesin the presence of complement may cause platelet ag-gregation, we tested this possibility in vitro by thefollowing experiment. The blood of rabbits congenitallydeficient in C6 was drawn into 0.2 M EDTA, a con-centration sufficient to inactivate C4 and C2. In thissystem immune lysis requiring all late components of Cand immune adherence requiring the first four com-ponents would be completely blocked. As shown inTable IV, platelet aggregation is not mediated by com-plement.

Proof that the in vivo thrombocytopenia was also in-dependent of the late components of complement at leastwas obtained by checking for thrombocytopenia in in-fected C6-deficient rabbits. Two rabbits congenitally

TABLE VIEffect of Particle-Free Supernate from Disrupted

Trypanosomes on Rat Platelets

Platelet counts X 103/mm3 30 min after 1.0 mlof rat blood was treated with 0.2 ml of:

Heatedsupernate

Normal 56°C forplasma Supernate 30 min

1,038 452 896

Normal blood anticoagulated with EDTAwas obtained froma pool of five rats. Platelets were counted at 30 min as outlinedin the text and in Table III.

deficient in C6 were infected in the usual manner. 3 wklater, when the animals began to show signs of clinicalillness, platelet counts were compared to those taken2 days before infection. The platelet counts before in-fection were 650 X 103 and 852 X 103/mm'. The 21-daypost-infection counts were 122 X 103 and 210 X 103/mm3,respectively.

Kinin inhibition. Kininlike substances, however,might be produced and released directly from trypano-somes, from platelets, or from rat mast cells. Therabbit plasma inactivator of kinin, kininase, is destroyedby EDTA. Such inactivation of trypanosomal kininmight explain the persistence of the in vitro plateletaggregation by trypanosomes in the presence of thischelating agent. Cochran and Wuepper (14) have shownthat kinin activity is destroyed by carboxypeptidase-Bor alpha-chymotrypsin but not by trypsin. The experi-ment shown in Table V exploited this observation. Asshown in this and several other experiments, neitheralpha-chymotrypsin or trypsin inhibited tryanosome-in-duced platelet aggregation at the usual concentrationsshown in Table V (50 Ig/ml) nor in concentrations of500 ltg/ml.

Studies implicating a possible trypanosonmal enzymeor toxin. Because the thrombocytopenia was not dueto immune complexes, ADP, or kinin activity, it mightbe mediated by a trypanosomal enzyme or toxin. Weitz(15) found an exoantigen of Trypanosoma brucei inthe serum of infected animals and in washings of har-vested trypanosomes. This antigen was necessary forinfectivity, because infectivity for mice was lost bywashing trypanosomes and restored by suspending theorganisms in their washings or trypanosome-free in-fected serum. Platelet aggregation, on the other hand,was not affected by washing our strain of Trypanosomarhodesiense. Harvested trypanosomes washed three timesin 10 vol of 0.15 M saline caused platelet aggregationequal to fresh organisms. Concentrated, infected plasmaand the concentrated washings failed to cause aggrega-tion. Since the aggregating factor appeared to be bound

1364 C. E. Davis, R. S. Robbins, R. D. Weller, and A. I. Braude

to the trypanosome, we next tried to isolate the activefactor by disruption of the organisms. This manipula-tion was successful only after very short periods ofsonication. When 0.2 ml of the trypanosome-free super-nate of the disrupted trypanosomes was added to 1.0ml of normal rat blood, aggregation always occurred. Atypical experiment is shown in Table VI. Like the wholetrypanosome, this aggregating factor is effective inEDTA-treated blood and is resistant to kinin inhibitors.It is labile, however, to heating at 56°C for 30 min(Table VI). This heat lability suggests that the aggre-gating substance may be a protein toxin or enzyme. Itseffectiveness in the presence of EDTAand the previousstudies with the whole trypanosome seem to rule outthe participation of complement.

The aggregating substance was not removed fromthe particle-free supernate by dialysis. Column-purifiedtrypanosomes sonicated and made particle-free by cen-trifugation were divided into two portions. After over-night dialysis of one portion against PSGbuffer, equalvolumes of the two portions caused equal platelet ag-gregation in citrated rat PRP. After 30 min the plateletcounts were; normal rat plasma control 1,450 X 103,untreated supernate 346 X 103, and dialyzed supernate378 X103/mm3.

EM. Scanning and transmission electron micro-graphs were made of aggregates induced by wholetrypanosomes and trypanosome-free supernate. Fig. 4shows the typical appearance on scanning EM of anaggregate induced by the trypanosome-free supernate.Fig. 5 compares normal platelets and aggregated plate-lets by transmission EM. These aggregates show typi-cal changes of aggregation, including degranulation,centralization of granules, presence of micelles, and re-duction in cytoplasmic density.

DISCUSSIONSevere thrombocytopenia occurs in rats and rabbits in-fected with this human strain of Trypanosoma rho-desiense. Since it occurs without consumption of fib-rinogen or release of FSP (Table I and Fig. 1), itis clear that the thrombocytopenia is not secondary toDIC. Instead a direct injury of platelets by trypano-somes appears responsible. Direct injury was first sug-gested by the inverse relationship of parasitemia tothrombocytopenia (Fig. 2). The degree of platelet ag-gregation caused by the addition of trypanosomes invitro to normal rat, rabbit, or human blood (Table IIIand IIIA) was directly related to the number of trypan-osomes (Fig. 3) and stimulated a search for a trypan-osomal factor capable of causing direct platelet dam-age. After it was shown that the in vitro aggregationwas not caused by ADP derived from trypanosomes orred cells, this system became a simple assay for

FIGURE 4 Platelet aggregate (at 90 min) induced by try-panosome-free supernate X 2,900 (scanning EM).

studying the mechanism of platelet injury. A series ofexperiments with this system, supplemented by the invivo model, has ruled out immunologic injury as thecause of platelet damage and shown that the aggre-gation is caused by a substance found in heat-labilesupernates of disrupted trypanosomes.

It was necessary to study immunologic phenomenacarefully because a series of reports have implicated theparticipation of antigen-antibody complexes in the path-ogenesis of the lesions of trypanosomiasis. First, inexperimental rabbit trypanosomiasis, each wave of para-sitemia is caused by trypanosomes with antigens dif-ferent from those of the predecessors, against whichantibody has already been produced (16). Second,Goodwin and Hook (17) have shown that the majorpathologic changes in rabbits infected with Trypano-sonta brucei are vascular lesions, thrombus formation,and increased vascular permeability. Finally, kinin con-centrations are increased in experimental infections oflaboratory animals (18) and volunteers (19). Thus ithas been postulated that the vascular pathology is re-lated to the release of kinins by antigen-antibody com-plexes. Our studies suggest an alternative mechanismof vascular damage by showing that trypanosome medi-ated platelet injury is independent of immune complexesas mediated by immune adherence (Table IV), comple-ment-mediated lysis (Table IV), and kinins (Table V).Independence from late components of complement wasalso shown by demonstrating that infected C6-deficientrabbits also developed thrombocytopenia. It is notknown whether the trypanosomes in Goodwin and

Thrombocytopenia in Experimental Trypanosomiasis 1365

as ~ to1mlo ra PRP Cotroswr R to whc . lo ltltpo lsafo h6x O A. Conro platlet X5,800A~~~. Sam as A 1250 Not magia bandof mirtuue (MB in logiuinln

granles (lrg arrow) s.

-,-a~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Am s

FIGURE( Comparison of normal rat platelets and platelet aggregates induced by trypano-some-free supernate. Aggregates were induced by adding 0.2 ml of trypanosome-free supernateto I ml of rat PRIP. Controls were PRP to which 0.2 ml of platelet-poor plasma f rom thesame pool was added. Samples were obtained after 90 min as noted in the text.

A. Control platelets X 5,800.B. Same as A X 12,500. Note marginal band of microtubules (MB) in longitudinal and

cross-sections and surface-connected membrane system (SCS).C. Platelet aggregate X 5,800.D. Same as C X 12,500. Note degranulation and reduction of cytoplasmic density in several

platelets, with some micelles ( M) and f ree granules present and centralization ofgranules (large arrow).

Hook's rabbits (17) or Boreham's volunteers (19) described by Goodwin and Hook could also contributecaused thrombocytopenia, because platelets were not to thrombocytopenia. Gross and microscopic examina-studied. Deposition of platelets in the vascular lesions tion of the kidneys, livers, spleens, lungs, hearts, and

1366 C. E. Davis, R. S. Robbins, R. D. Weller, and A. 1. Braude

brains of rats infected with our strain of T rhodesiensehas not demonstrated such vascular lesions or fibrindeposition. Splenomegaly is the only finding.

As shown in Table VI, remarkable platelet aggrega-tion is produced by trypanosome-free extracts of dis-rupted organisms. The aggregating substance is so labilethat it is inactivated by sonication lasting as little as20 s beyond the initial 20-s period. The fact that it isheat-labile but not blocked by inhibitors of complementstrongly suggests that it is a protein toxin or enzyme.The identity of this substance has not been determined.Among the agents known to cause platelet aggregation,ADP, antigen-antibody complexes, and FSP have beenruled out. Serotonin and epinephrine are unlikely be-cause of the marked lability of the phenomenon to heatand sonication. Long-chained saturated fatty acids seemto be eliminated because the aggregating substancewas not removed from supernates by extraction withpetroleum ether.2 The trypanosomes may contain a pro-tein with thrombin, fibrinogen, or collagen activity or atrypanosomal protein unrelated to other aggregatingsubstances. Larger quantities of the thrombocytopenicfactor are being accumulated for studies in vivo withthe aggregating substance and in vitro with proteinand enzyme inhibitors and isolation of the aggregatingsubstance.

Thrombocytopenia has not been reported previouslyin experimental trypanosomiasis and indeed in onlytwo cases of naturally acquired human trypanosomiasis(1, 20). Both patients were Americans who visited en-demic areas of Africa. It is important to know whetherthrombocytopenia is a uniform accompaniment of Rho-desian trypanosomiasis, because the usual cause ofdeath in this fulminating disease is unknown. Our datasuggest that thrombocytopenia may be an importantmanifestation of this illness.

The platelet injury described here is independent ofhemolysis, DIC, or immunologic injury related to im-mune adherence, complement-mediated lysis, or releaseof kinins. Aggregation of platelets in vitro by heat-labile supernates of disrupted trypanosomes is notblocked by inhibitors of complement or ADP. If theplatelet effect is due to a protein toxin or enzyme, thiswill be the first infectious agent shown to have a directtoxic effect on the platelet without DIC. Release ofprocoagulants from injured platelets might then resultin intravascular clotting in some animal species andaccount for the DIC observed in the patient treated inthis hospital.

Note added in proof. After this manuscript was com-pleted, our observation of thrombocytopenia in humantrypanosomiasis was confirmed in rhesus monkeys by Sadun,Johnson, Nagle, and Duxbury. 1973. Am. J. Trop. Med.Hyg. 22: 323.

2 Unpublished observation.

REFERENCES1. Barrett-Connor, E., R. J. Ugoretz, and A. I. Braude.

1973. Disseminated intravascular coagulation in trypan-osomiasis. Arch. Intern. Med. 131: 574.

2. Dennis, L. H., J. W. Eichelberger, M. M. Inman, andM. E. Conrad. 1967. Depletion of coagulation factorsin drug-resistant Plasmodium falciparumi malaria. BloodJ. Hematol. 29: 713.

3. Borochovitz, D., A. L. Crosley, and J. Metz. 1970.Disseminated intravascular coagulation with fatal hae-morrhage in cerebral malaria. Br. Med. J. 2: 710.

4. Hill, G. J., II, V. Knight, and G. M. Jeffrey. 1964.Thrombocytopenia in vivax malaria. Lancet. 1: 240.

5. Neva, F. A., J. N. Sheagren, N. R. Shulman, and C. J.Canfield. 1970. Malaria: host-defense mechanisms andcomplications. Ann. Intern. Med. 73: 295.

6. Skudowitz, R. B., J. Katz, A. Lurie, J. Levin, and J.Metz. 1973. Mechanisms of thrombocytopenia in ma-lignant tertian malaria. Br. Med. J. 2: 515.

7. Hoffman, T. A., and E. A. Edwards. 1972. Group-specific polysaccharide antigen and humoral antibodyresponse in disease due to Neisseria meningitidis. J. In-fect. Dis. 126: 636.

8. Ratnoff, 0. D., and C. Menzie. 1951. New method fordetermination of fibrinogen in small samples of plasma.J. Lab. Clin. Med. 37: 316.

9. Leavelle, D. E., B. F. Mertens, E. J. W. Bowie, andC. A. Owen, Jr. 1971. Staphylococcal clumping onmicroliter plates. A rapid simple method for measuringfibrinogen split products. Am. J. Clin. Pathol. 55: 452.

10. Braude, A. I., H. Douglas, and C. E. Davis. 1973.Treatment and prevention of intravascular coagulationwith antiserum to endotoxin. J. Infect. Dis. 128: S157.

11. Lanham, S. M. 1968. Separation of trypanosomes fromthe blood of infected rats and mice by anion-exchangers.Nature (Lond.). 218: 1273.

12. Born, G. V. R. 1967. Mechanism of platelet aggregationand its inhibition by adenosine derivatives. Fed. Proc.26: 115.

13. Born, G. V. R., A. J. Honour, and J. R. A. Mitchell.1964. Inhibition by adenosine and by 2-chloroadenosineof the formation and embolization of platelet thrombi.Nature (Lond.). 202: 761.

14. Cochrane, C. J., and K. A. Wuepper. 1971. The kinin-forming system in plasma. Excerpta Med. Int. Cong.Ser. 229: 137.

15. Weitz, B. 1960. The properties of some antigens ofTrypanosoma brucei. J. Gen. Microbiol. 23: 589.

16. Gray, A. R. 1962. The influence of antibody on sero-

logical variation in Trypanosoma brucei. Ann. Trop.Med. Parasitol. 56: 4.

17. Goodwin, L. G., and S. V. M. Hook. 1968. Vascularlesions in rabbits infected with Trypanosoma (Trypano-zoon) brucei. Br. J. Pharmacol. Chemother. 32: 505.

18. Richards, W. H. G. 1965. Pharmacologically activesubstances in the blood, tissues, and urine of mice in-fected with Trypanosoma brucei. Br. J. Pharmacol.Chemother. 24: 124.

19. Boreham, P. F. 1970. Kinin release and the immunereaction in human trypanosomiasis caused by Trypano-soma rhodesiense. Trans. R. Soc. Trop. Med. Hyg. 64:394.

20. Ottoman, P., J. Zumwalt, J. Chin, and R. Roberto. 1970.African trypanosomiasis-California. Morbid. Mortal. 19:233.

Thrombocytopenia in ExperiMental Trypanosomiasis 1367