Soluble human complement receptor type 1 inhibits complement-mediated host defense.

CLINICALANDDIAGNOSTIC LABORATORYIMMUNOLOGY, Sept. 1994,P.585-589 Vol. 1, No. 5

1071-412X/94/$04.00+0

Copyright C 1994, AmericanSociety for Microbiology

Soluble

Human

Complement

Receptor

Type

1

Inhibits

Complement-Mediated

Host

Defense

ANDREAJ.

SWIFT,'

TIMOTHY S.

COLLINS,'

PETERBUGELSKI,2 ANDJERRY A. WINKELSTEINl* DivisionofImmunology, Departmentof Pediatrics, The JohnsHopkins UniversitySchoolof

Medicine, Baltimore, Maryland

21287-3923,'

andDepartmentof

Toxicology,

SmithKlineBeecham,Kingof Prussia, Pennsylvania19406-09392

Received22April 1994/Returned for modification17May 1994/Accepted 13June 1994

Solublecomplement receptor type1 (sCR1) isapowerful inhibitor ofcomplementactivation. Because of this

ability,

sCR1 may provetobeanimportant therapeutic agentthat canbe usedtoblocktheimmunopathologic

effects ofuncontrolled complement activation ina

variety

ofclinically significantdisorders.Althoughseveral previous studies have examined the ability of sCRi to inhibit complement-mediated immunopathologic

damage, there is no information onits

ability

tointerferewith the host's defense against infection. In the current experiments sCRi exertedaconcentration-dependent

inhibitory

effectonthe

phagocytosis

of

Strepto-coccuspneumoniae by humanpolymorphonuclearleukocytes invitro. NotonlydidsCRI inhibit complement-dependent opsonization of the pneumococcus but athigherconcentrations it also inhibited theingestionof

bacteria whichhad been previously opsonized.Furthermore,when ratswere

injected

withsCR1, it inhibited

both theirserum hemolytic

activity

and serumopsonicactivityinadose-dependentfashion.

Finally,

forrats

treated withsCR1,the509%lethal dose wasalso showntobesignificantlylower thanthat for control animals

after intravenous challenge with S. pneumoniae and Pseudomonasaeruginosa. These data demonstrate that sCRi significantly inhibitscomplement-mediated hostdefenseagainstbacterialinfection.

Thecomplement system is composed ofa series ofserum

proteins and cellular receptors which serve as important

mediators of host defense and inflammation. When the acti-vation ofcomplementis controlled and directedagainst invad-ingmicroorganisms,it playsanimportantrole inresistanceto

infection.However, when itsactivation proceedsinan

uncon-trolledfashionorisdirectedagainstthehost,thecomplement

system can cause immunopathologic damage and can be detrimental to the host.

Complement receptor type 1 (CR1, CD35) is found on

primateerythrocytes andleukocytesandhasspecificityforthe activatedcomplementcomponentsC3band C4b(3).Ithas the

abilitytoinhibit activation of both the classical and alternative pathways of the complement system(1, 2, 7, 12). Recently, a

soluble form of the receptor has been produced by

recombi-nantDNAtechnology(12).Solublecomplement receptor type 1(sCR1)lacks the transmembrane andcytoplasmicdomains of its membrane-bound parent molecule but retains itsabilityto

inhibit thecomplementsystem(12).Invivostudieshave shown thatsCR1 inhibits theimmunopathologic damagemediatedby

complementinanimal models ofmyocardialinfarction(6, 12),

intestinal ischemia (5), cardiopulmonary bypass (4), the re-versedpassive Arthus reaction (15), and the adultrespiratory distress syndrome(9).

Because of its potent ability to inhibit the complement system,sCR1 may provetobeanimportant therapeutic agent that can be used to block the immunopathologic effects of uncontrolled complement activation in a variety of clinically significant disorders. Although several previous studies have examined the ability of sCR1toinhibitcomplement-mediated

immunopathologic damage(4, 5, 6, 9, 12, 15), to date no study

*Correspondingauthor.Mailingaddress: Division of Immunology,

Department of Pediatrics, CMSC 1103, Johns Hopkins Hospital,

Baltimore, MD 21287-3923. Phone: (410) 5883. Fax: (410) 955-0229.

has looked at theability of sCR1tointerfere with host defense against infection.

MATERIALS AND METHODS

Buffers. Afive-times-concentrated stock solution of Vero-nal-buffered saline (VBS; pH 7.4)was prepared as described previously(8). From this stock solution, VBS (ionic strength, 0.147) and VBS containing 0.15 mM CaCl2, 0.5 mM MgCl2, and0.5%bovineserumalbumin(BSA; Sigma,St.Louis, Mo.) (VBS2+-BSA)wereprepared.

RecombinanthumansCR1. Recombinant humansCR1 isa

240-kDa glycoprotein which consists of the extracellular do-mains of humansCR1 (12). sCR1wasexpressedbystably

trans-fected Chinesehamster ovary cells andwaspurified from condi-tioned mediumtogreaterthan99%purity. sCR1 (BRL55730,

TP10-HD) was supplied lyophilized by the Department of

Pharmaceutical ScienceandTechnology, SmithKlineBeecham

Pharmaceuticals. Itwas reconstituted to aconcentration of 5 mg/ml.

Bacteria. Streptococcus pneumoniae type 25 (Pn25) was

obtained from a blood culture of a patient with sickle-cell disease (14). The methods of storage, mouse passage, and quantitation were as described previously (14). To prepare

Pn25, a drop of the stock culture from defibrinated rabbit blood wasplacedin 5 ml of brain heart infusion(BHI; Difco Laboratories, Detroit, Mich.) containing 10% sheep serum

(SS; Pelfreeze, Rodger, AK) (BHI-SS)andincubated for 9 h at

37°C. The culture was then diluted 1/10 in fresh BHI-SS, reincubatedat37°C for 9 h, and rediluted 1/10 in fresh BHI-SS. Afteranadditional 9-h incubation at 37°C, the bacteria were in

log-phase growth and were harvested for use. The bacterial culturestobe used in thephagocytic assays (see below) were washed in VBS three times, counted in a Petroff-Hauser

chamber, and adjusted to 125 x 106 bacteria per 0.025 ml. Bacterial cultures to be used in the 50% lethal dose

(LD50)

studies (see below) were concentrated 40-fold by

centrifuga-585

on August 17, 2020 by guest

http://cvi.asm.org/

tion and were then serially diluted 3-fold in tryptic soy broth

(TSB; Difco, Detroit, Mich.).

Pseudomonas aeruginosa ATCC 19660, isolated from the blood of a burn patient, was purchased from the American Type Culture Collection, Rockville, Md. The bacteria also underwent passage in mice and were stored in tryptic soy broth (TSB) with 20% glycerol at -70°C. In preparation for chal-lenge, a drop of the bacterial stock was added to TSB and the mixture was incubated for 9 h at 37°C. After two successive 9-h passages, the bacteria were prepared for

LD50

studies as described above.

Polymorphonuclear leukocytes. On the day of each

phago-cytic assay, venous blood was collected from healthy human volunteers into a heparinized syringe. Erythrocyte sedimenta-tion with dextran solusedimenta-tion was performed as described previ-ously(14). The leukocyte-rich plasma was layered over lympho-cyte separation medium (Litton Bionetics-Organon Teknika, Irving, Tex.), and the mixture was centrifuged at 300xgfor 30 min at room temperature (11). The pellet containing the polymorphonuclear leukocytes was resuspended in a small quantity ofVBS2+-BSA,and contaminating erythrocytes were lysed by adding 7.5 ml of distilled water. After 20 s of agitation, 2.5 ml of3.6% saline was added to reverse the hypotonicity. Thissuspension was recentrifuged at 300 xgfor 10min, and the pellet was resuspended in VBS2+-BSA and was

subse-quently maintained at 4°C. The cells were counted with a hemocytometer and diluted to the desired concentration.

Animals.All animal experiments were carried out with male Sprague-Dawley rats (weight, 100 to 150 g) obtained from Harlan Sprague-Dawley, Indianapolis, Ind.

Serum collection. Blood wasobtained from the tail vein of rats and was allowed to clot at room temperature, and the

serumwasseparated and stored at -70°C.

Hemolyticassay of complement. The rat sera were tested for total hemolytic complement activity (50% hemolytic

comple-ment

[CH50])

asdescribed previously (8).

Serumopsonizing

activity.

Theopsonic assay was performed

asdescribedpreviously (14).Briefly, aliquots of the leukocyte

suspension containing 12.5 x 10 cells were centrifuged in

polypropylene tubes at 300 x gfor 10 min. The supernatant

was discarded and the cells were resuspended in 20% test

serum in 0.4 ml of

VBS2+-BSA.

Atotal of 125 x 106 Pn25

containedin0.025 mlwasadded,and the mixturewasrotated

at 12 rpmend-over-end in a multipurposerotator (Scientific

Industries, Bohemia, N.Y.) for 60 min at 37°C. Thin smears

weremadeonglassslides, heat fixed,andstainedwith

meth-ylene blue. Serum opsonic activity was measured as percent

phagocytosis,whichwasdetermined bycounting 200

polymor-phonuclear cells and scoring the percentage of cells with

ingested bacteria. Controls included bacteria and phagocytic cells alone and bacteria,

phagocytic cells,

and 20% serum containing0.01%EDTA. Ineachcasethe percent

phagocyto-siswas 1%orless.

In some experiments, aliquots of 0.075 ml of

VBS2+-BSA

containing 125 x 106 bacteria were

preincubated

at 37°C in

20%rat serumcontainingvariousconcentrations ofsCR1orin

20% rat serumalonediluted in 0.35 ml of

VBS2+-BSA.

After 30 min ofrotation,1ml ofVBSwasadded and the mixturewas

centrifugedfor 10 min.The supernatantwasdiscarded and the bacteriawereresuspendedin0.2 ml of

VBS2+-BSA

containing

various concentrations of sCR1 or in 0.2 ml ofVBS2+-BSA

alone. Anadditional 0.2 ml of

VBS2+-BSA

containing

12.5 x

106

polymorphonuclear leukocyteswasadded,and themixture

was rotated at 12 rpm for 60 minat 37°Candwas treated as described above.

Bacterial challenge. Groups of seven animals each were

100'

Cl)

rn

° 75

u

0

-C

a- *-50-0

C

0 -= 25

'6

10

1do

1060

Concentration

sCRI

(,ug/ml)

FIG. 1. Effect of

sCRi

on in vitrophagocytosis ofS.pneumoniae.

injected via the tail vein with the desired dose of either sCR1

or an equivalent volume of saline. Twenty minutes later, the rats werechallenged with 0.3 ml of various concentrations of the desired bacteria. The animals were observed at 12-h intervals for 1 week, and the deaths were recorded. The LD50 wascalculated by the method of Reed and Muench (10).

RESULTS

Effectof

sCRi

on in vitro phagocytosis. Since the comple-mentsystemis known to play a critical role in the phagocytosis of certain bacteria, we examined the effect of sCR1 on the phagocytosis of Pn25. When sCR1 was added to the phagocytic mixtures containing 20% test serum, it exerted an inhibitory

effect on the phagocytosis of Pn25 by human PMNs in a

concentration-dependent fashion (Fig. 1). At all concentra-tions greater than 10

,ug/ml,

nearly complete inhibition of

phagocytosiswasseen.

EffectofsCR1onopsonization and ingestion. Complement-mediated phagocytosis can be separated into two stages:

opsonizationof the bacteriabyactivated C3bandingestionof theopsonizedbacteriabythephagocyticcell. Toinvestigatein which stage sCR1 inhibits phagocytosis, we added sCR1 to

each of the two stages of our phagocytic assay. First, we

examined the effect of sCR1onthe activation ofcomplement

and itsopsonizationof bacteria. Pn25waspreincubatedin20% serumcontainingvarious concentrations of sCR1 for 30 minor in 20% serumwhich didnot contain sCR1, and the bacteria

were washed free of the serum and sCR1 and were then

incubatedbyrotation inthe presence of bufferand

phagocytic

cells. We also

examined

the effect ofsCR1onthe

ingestion

of

preopsonized bacteria by the phagocytic cell. In this case, bacteriawere

preincubated

in 20% test serumwhich did not contain sCR1

(as

described

above),

washed free of serum, and then incubated in the presence of either buffer or various

concentrations ofsCR1 andthe

phagocytic

cells.

sCR1was showntoinhibitboth

opsonization

and

ingestion

in aconcentration-dependentfashion

(Fig. 2).

This

inhibitory

effectwas more

pronounced

whensCR1wasadded

during

the

opsonization of bacteria. For

example,

whereas a concentra-tion of 10 ,ug/ml added during

opsonization

caused a 93%

on August 17, 2020 by guest

http://cvi.asm.org/

sCRI INHIBITS COMPLEMENT-MEDIATED HOST DEFENSE 587

4)

-c 0

E

0 0

0~

4-0

ol 100'

l0:

. . .11. . 11i I .

2060 4 8 12 1 2 3 4

Minutes Hours Days

Concentration sCRI

(p.g/ml)

FIG. 2. Relative inhibitory effect of sCR1 onopsonization of S. pneumoniae (a) and ingestionofopsonized S. pneumoniae (0).

inhibition ofphagocytosis,a100-foldgreaterconcentration of sCR1 addedafteropsonizationresulted inonly 76%inhibition. In vivo effect of sCRi on serum hemolytic and opsonic activities. Groupsoffiverats eachwereinjectedwith various doses of sCR1 and were bled at given time intervals after

injection, the serumwas pooled, and the CH50swere

deter-mined(Fig. 3). One hour after administration ofthesCR1,the sera ofthose rats thatreceived the higherdose ofsCR1 (25 mg/kg) had a CH50which was only 6% of the pretreatment

level. One dayfollowingtreatment, theCH50was still signifi-cantly depressed (29%of thepretreatmentlevels)and didnot approach normal levels until day 4 (85% of pretreatment levels). The sera of rats receiving a lower dose of sCR1 (10 mg/kgofbody weight)showedasmaller decrease inCH50after

a)

U)

-J

E

0

0~

0-CK

t.

100

l0

0 2060 4 8 1 2 3 4

Minutes Hours Days

Time

FIG. 3. Serumhemolytic activity(CH50) inratstreated with 10mg

(0)or25mg(0)ofsCR1perkg.

Time

FIG. 4. Serum opsonic activity for the pneumococcus in rats treatedwith 10mg (0)or25 mg(0) of sCR1perkg.

1 h(17%ofpretreatmentlevels)andapproachednormal levels

muchmorequickly.

Theabilityofpooledserumobtained from thegroupsofrats

that had been treated with sCR1 to support opsonization of Pn25was also tested. The sera ofratsreceiving sCR1 at the

highest dose (25 mg/kg) had only 3% of their pretreatment opsonizing activity at 1 h after injection, and the opsonizing activity ofserum didnot approach normal levels until 1 day

later (Fig. 4). Theopsonizing activity of thesera ofrats that

receivedalowerdose(10 mg/kg)wasinhibitedmostat20min (23% ofpretreatment levels) andbegan to approach normal

levels by24 h.

Bacterial challenge ofsCR1-treated animals with S. pneu-moniae. SincesCR1inhibitedbothopsonization and ingestion

of thepneumococcusboth in vitro and invivo,we investigated whethersCR1-treatedratswere moresusceptibletoinfection.

Groupsofsevenratseachwereinjectedwiththe desired dose ofsCR1andwerethenchallenged intravenouslywithPn2530 minlater.Table 1 liststhe

LD5Os

found foreach dose ofsCR1. At 10mg/kg therewas afourfold difference between theLD50s for thoserats receiving sCR1 and those rats receivingsaline

(P<0.05). This effectwas evenmorepronounced atahigher

dose ofsCR1.For thoseratsreceiving 25mg ofsCR1perkg, the LD50 was eightfold lower than that for control animals (P< 0.05) (Table 1 andFig. 5).

Bacterial challenge of sCR1-treated animalswith P. aerugi-nosa.Todeterminewhetherratstreated withsCR1were more

TABLE 1. LD50s forratstreated withsCR1 andchallenged with bacteria

Bacterium sCR1dose LD50(±2

SD)-(mg/kg) Control rats sCR1-treated rats

Pneumococcus 10 (8.5±2.8)x 108 (2.1± 1.9)x 108 25 (8.9± 1.0)x 108 (1.1± 1.1) x108 Pseudomonas 25 (11.0±2.6)x 107 (4.0±2.2)x107

a P<0.05 for

comparisons

betweenboth groups.

c 0

._ g

0._0 VOL. 1, 1994

on August 17, 2020 by guest

http://cvi.asm.org/

-O.

-0 cs

-00

Pn25

FIG. 5. Cumulative mortality ofratstreated with 25 mgofsCR1

perkg(0)orsaline (0) andsubsequently challenged intravenously withS. pneumoniae.

susceptibletoothertypesofbacterial infections,P. aeruginosa, agram-negative rod, was chosen forchallenge. At 10 mg of sCR1 perkg, only slight differences in the LD50swere seen (P>0.05) (datanotshown). However,atthe higher dose of 25 mg/kg,therewas afourfold difference in the

LD50s

(P<0.05).

DISCUSSION

sCR1 has beenshowntobe apowerfulinhibitorof comple-mentactivation(1, 2, 7, 12). It inhibits theactivation of C3 and C5 by interferingwith theassemblyandexpressionofthe C3-andC5-cleaving enzymesof both theclassicalandalternative

pathways. Becauseit is such anefficient inhibitor of comple-mentactivation, itmaybeof value as a pharmacologic agent thatcanbe usedtoinhibitcomplement-mediated damageina numberof diseasestates. Infact,ithasalreadybeen shownto

be effective in suppressing complement-mediated

immuno-pathologic damagein animalmodels ofpostischemic

myocar-dial infarction(6, 12),intestinal ischemia(5), cardiopulmonary bypass (4),thereversedpassiveArthus reaction(15),andacute respiratorydistresssyndrome (9). Thepresent studyprovides

evidence that sCR1 may also have detrimental effects on a patient'scomplement-mediatedhostdefenseagainstbacterial infections.

Complement plays a critical role in the opsonization ofa variety ofbacteria, including the pneumococcus (13). It has

previouslybeen established thatsCR1 inhibits the hemolytic activityof both the classical and alternativepathways (1, 2, 7, 12). Our study extends these observations by showing that sCR1 inhibitsopsonizationof thepneumococcusbothin vitro

and in vivo. When ratswere treated withsCR1 invivo, their serum opsonizing activitywas inhibited in a dose-dependent fashion. Whereasadoseof 10mg ofsCR1 perkgresultedin a 67% inhibition of serum opsonizing activity 1 h after

administration, the higher dose of 25 mg/kg inhibited the serumopsonizing activity nearly completelyatthe sametime point.The factthatsCR1inhibits thecomplementsystemina dose-dependent manner will undoubtedly prove to be an important consideration in the dose of the drug thatmay be

administered, particularly to patients already at an increased riskfor infection, i.e., bum patients and postoperative patients. The results of the present study also indicate that sCR1 interfereswith the ingestion of bacteria that carry opsonically active C3b on their surfaces. When sCR1 was added to phagocytic mixtures containing bacteria that had already been preopsonized with C3b, the ingestion of the bacteria by granulocytes was inhibited in a dose-dependent fashion. It should be noted, however, that much higher doses of sCR1 werenecessary toinhibit ingestion of opsonized bacteria than were needed to inhibit opsonization. It is possible that sCR1 binds to the C3b molecule on the surface of the bacteria and competitively inhibits the recognition of opsonized bacteria by

thephagocytic cell.

Finally, rats treated with sCR1 had increased susceptibilities tochallenge with two different bacterial species, S. pneumoniae and P. aeruginosa. In each instance, for animals treated with sCR1, theLD50swerelower when the animals were challenged with bacteriaintravenously and the effect was more prominent with the higher dose of sCR1 (25 mg/kg). Although the studies described here demonstrated that sCR1 interferes with the host's defenseagainst bacterial infection, it is possible that the effect would be less pronounced if the animals were challenged

by differentroutes, e.g.,byinhalation.

It is not clear if the effect of sCR1 on host defense will

interferewith itspotential clinical utility in preventing comple-ment-mediated immunopathologic damage. It is difficult to

predicttowhat extent sCR1would interfere with host defense inhumans who might receive this pharmacologic agent. In this regard, it is interesting that human sCR1 inhibits the lysis of

antibody-sensitized sheep erythrocytes by human serum to a greaterdegreethanit does rat serum(12).Thus, it is possible thatitseffects on host defense will be greater in humans than in rats. Nevertheless, the results of the current study do not necessarily preclude the use of sCR1 in humans. Its effect on

thecomplementsystemisrelativelyshort-lived,is dose

depen-dent, andshouldnotinterfere with other mechanisms of host defense. Thus, itsbenefit incertain clinical situations charac-terized by complement-mediated damage mayoutweigh any

potentialrisks.

ACKNOWLEDGMENT

Thisworkwassupportedin partbyNIH grant HL-47191. REFERENCES

1. Fearon, D. T. 1979.Regulationoftheamplification C3convertase

of human complement by an inhibitory protein isolated from

human erythrocyte membrane. Proc. Natl. Acad. Sci. USA 76:

5867-5871.

2. Fearon, D. T. 1980. Identification of the membraneglycoprotein thatis the C3breceptorof the human erythrocyte,

polymorpho-nuclear leukocyte, B lymphocyte, and monocyte. J. Exp. Med.

152:20-30.

3. Fearon, D.T., R. H. Carter,andJ.M.Ahearn. 1990. CR1 and CR2:receptorsmediatingcellularrecognitioninthecomplement system, p. 55-82. In Cellular and molecular mechanisms of inflammation. Academic Press, Inc.,NewYork.

4. Gillinov,A.M.,P. A.DeValeria, J.A.Winkelstein,I. Wilson,W. E.

Curtis,D.Shaw,C. G.Yeh,A.RRudolph,W. A.Baumgartner,A. Herskowitz, and D. E. Cameron. 1993. Complement inhibition with soluble complement receptor type 1 in cardiopulmonary

bypass.Ann.ThoracicSurg. 55:619-624.

5. Hill, J. T., F.Linsay,F.Ortiz,C.G.Yeh,H. B.Hechtman,and F. D. Moore, Jr. 1992. Soluble complement receptor type 1

ameliorates the local and remote organ injury after intestinal ischemia-reperfusion in therat.J.Immunol.149:1723-1728. 6. Homeister, J. W., P. S.Satoh,K.S.Kilgore,and B.R. Lucchesi.

1993.Soluble complementreceptor type 1prevents human

on August 17, 2020 by guest

http://cvi.asm.org/

sCR1 INHIBITS COMPLEMENT-MEDIATED HOST DEFENSE 589 plement-mediated damage of the rabbit isolated heart. J.

Immu-nol. 150:1055-1064.

7. lida, K., and V. Nussenzweig. 1981. Complement receptor isan

inhibitor of thecomplement cascade. J. Exp.Med.153:1138-1150. 8. Mayer, M. M. 1959. Kabat and Mayer's experimental immuno-chemistry, 2nd ed.,p. 133-240. Charles C Thomas, Springfield, Ill. 9. Rabinovici, R C., G. Yeh, L. M. Hillegass, D. E. Griswold, M. J. Dimartino, J. Vernick, K L. Fong, andG. Feuerstein. 1992. Role ofcomplement in endotoxin/platelet activating factor induced lung injury. J. Immunol. 149:1744-1750.

10. Reed, L. J., and H. Muench. 1938. A simple method of estimating fiftypercentendpoints.Am.J.Hyg.27:493-497.

11. Swift, A. J., E. R. Moxon, A. Zwahlen, and J. A. Winkelstein. 1991. Complement-mediatedserumactivitiesagainst geneticallydefined capsular transformants of Haemophilus influenzae. Microb. Pathog. 10:261-269.

12. Weisman, H. F., T. Bartow, M. K.Leppo,H.C. Marsh, Jr., G. R.

Carson, M. F. Concido, M. P. Boyle, K. H. Roux, M. L. Weisfeldt, and D. T. Fearon. 1990. Soluble human complementreceptor type 1: in vivo inhibitor of complement suppressing postischemic myocardial inflammation and necrosis. Science 249:146-151. 13. Winkelstein, J. A. 1981. The role of complement in the host's

defense against Streptococcus pneumoniae. Rev. Infect. Dis. 3:289-298.

14. Winkelstein, J. A., and R H. Drachman. 1968. Deficiency of pneumococcalserumopsonizing activity in sickle-cell disease. N.

Engl. J. Med. 279:459-466.

15. Yeh, C. G., H. C. Marsh, Jr., G. R Carson, L. Berman, M. F. Concino, S. M. Scesney, R E. Kuestner, R Skibbens, K. A. Donahue, and S. H. Ip. 1991. Recombinant soluble human

com-plement receptor type 1 inhibits inflammation in the reversed passive Arthus reactioninrats.J.Immunol. 146:250-256. VOL. 1, 1994

on August 17, 2020 by guest

http://cvi.asm.org/