Procedures for Identification of Arthropod-borne Viruses*

Bull. Org. mond. SanteX 1961, 24, 723-734Bull. Wid Hlth Org.

Procedures for Identificationof Arthropod-borne Viruses*

JORDI CASALS, M.D.'

Owing to the existence of antigenic groups and to the consequent characteristics ofthe immune response ofa host to grouped viruses, certain logical steps are advisable in theprocess of identification ofan arthropod-borne (arbor) virus. The first of these steps is thedetermination of the arthropod-borne nature of the virus. Since no antigenic property isknown to be common to all arbor viruses, the decision is based on other biological propertiesand on the circumstances of isolation. The second step is determination of the antigenicgroup. For this are used hyperimmune sera with considerable cross-reactivity within thegroup, as well as the test that gives the most overlap. The third step is determination oftypewithin the group. Simple immune sera and the least cross-reactive test are used at this step.

Viruses that belong in minor groups or are ungrouped often constitute a big problemowing to the fact that, once the major groups have been eliminated from further considera-tion, comparative studies must be conducted with practically all the remaining virusesbefore a definitive answer is reached.

The identification of an arthropod-borne (arbor)virus isolate requires that a sufficient number of itsproperties be determined to establish whether theisolate is a new virus or one already described. Of allproperties of the arbor viruses susceptible of in-vestigation, it appears that, in our present state ofknowledge, the serological ones are best suited foridentification owing to the fact that these are stableas well as sensitive enough to result in separation ofdifferent virus " species " and that the methods em-ployed for their study are relatively simple.

While properties other than the serological ones-for example, pathogenicity for various hosts andtype of lesions or disease that these viruses cause-can be most helpful in giving investigators direction,the final generally accepted criterion for identifica-tion is the serological one. In this paper, the termidentification is used to mean serological identifica-tion, i.e., identification on the basis of an antigen-antibody reaction detected by any one or several ofthe following tests: complement-fixation (CF),haemagglutination-inhibition (HI), neutralization

* Paper submitted to the WHO Study Group on Arth-ropod-borne Viruses, September 1960. This paper was pre-sented in part at a meeting on the arthropod-borne virusesheld at the Gould House, Ardsley-on-Hudson, N.Y., USA,in October 1959 under the sponsorship of the RockefellerFoundation.

1 Staff member, The Rockefeller Foundation VirusLaboratories, New York, N.Y., USA.

(NT) either in tissue cultures or in animals (the latterby the central or the peripheral route of inoculation),cross challenge, agar gel precipitin reaction, haemad-sorption and fluorescent antibody technique. Someof these tests are in general use, others are less ex-tensively employed, while still others are only beingdeveloped for use with the arbor viruses.Although the final identification of a virus is based

on serological studies, the initial step in the charac-terization of the isolate-namely, the attempt todecide whether or not it is an arbor virus-isbased not on serological properties, for at presentnone is known common to all, but on a biologicalconcept and a number of properties other thanserological.A process ofidentificationnecessarilyimplies classi-

fication; therefore, the serological identification ofan arbor virus has to be considered and carried outin the light of the antigenic group concept and of theconsequent basic properties of the serological re-sponse ofa host to the grouped viruses (Casals, 1957).It has been shown that arbor viruses occur in groups(see below) defined by the fact that common anti-genic constituents are shared by the members of agroup. When a host is infected or sensitized by onevirus in a given group, antibodies develop not onlyagainst the infecting virus but also, to a greater orlesser degree, against the other viruses in that group.

1007 -723- 4

724 J. CASALS

Consequently, a positive serological reaction be-tween an isolate and a type immune serum need notnecessarily be diagnostic but may be due to im-munological overlap with a related virus.

GENERAL PROCEDURE FOR IDENTIFICATIONOF AN ISOLATE

Given a virus isolate, the process of its identifica-tion within the arbor virus " family " follows alogical, step-wise progression. In practice, however,one does not always adhere to it.

Identification as arbor virus

The first step is to determine whether the isolateis an arbor virus. As mice have been used for theisolation of most arbor viruses, it is not surprisingthat on occasion the isolated virus has proved tooriginate in the mouse rather than in the inoculum;examples are Theiler's, lymphocytic choriomeningitisand hepato-encephalitis viruses. Obviously, theseviruses do not belong in the arbor group nor are theytrue isolations. In addition, genuine isolations ofnon-arbor viruses may be made, especially frombirds; such isolations may include psittacosis andNewcastle disease viruses. Even viruses isolatedfrom wild-caught arthropods are on occasion foundto be non-arbor and presumably represent recentlyingested virus.Proof that an isolate is an arbor virus is, probably,

the most difficult step in its identification, owing tothe fact that no simple experimental procedure un-mistakably answers the question. By working defini-tion, an arbor virus is one which in nature has thecapacity to infect vertebrates and to multiply in thebody of arthropods. As is well known, the uniquerelationship among virus, host and vector, charac-terized by an extrinsic incubation and multiplicationof the virus in the tissues of the arthropod with noapparent damage to the latter, is essential to theconcept of arbor virus; 1 therefore, the crucial orfinal evidence for deciding that a virus belongs in thisfamily is the experimental reproduction of the naturalcycle-a highly impractical procedure. In the ab-sence of this proof-reproduction of the cycle-thenature of an arbor virus is established throughseveral experimental procedures which, in general,are aimed more at the detection of excluding

1 The notion may be put forward that an arbor virus (onserological grounds) may not necessarily be an arthropod-borne virus; this point will not be elaborated here.

properties rather than at identifying exclusiveones.Among the procedures and considerations ap-

plicable to the problem are the following.

Serial propagation in blood-sucking arthropods byinoculation. Evidence of multiplication with serialtransmission of an agent by inoculation of blood-sucking arthropod tissue suspensions containing thevirus to new arthropods is considered excellent evi-dence that a virus is an arthropod-borne agent. Themethod is, however, outside the scope of all exceptspecialized laboratories; in addition to specializedknowledge, insectaries and colonies of suitablearthropods are required. No viruses other thanarthropod-borne have been found to propagate seri-ally in blood-sucking arthropods, under the requiredexperimental conditions, at The Rockefeller Founda-tion Virus Laboratories.2 On the other hand, failureto propagate a virus may be due to the use of awrong arthropod species.

Action of ethyl ether and sodium desoxycholate.Studies and reports on the action of these chemicalson viruses show a division in resistant and sensitiveagents (Smith, 1939; Bumet & Lush, 1940; Andrewes& Horstmann, 1949; Theiler, 1957; National Found-ation for Infantile Paralysis, Committee on the En-teroviruses, 1957; Dingle & Ginsberg, 1959). As faras reported investigations show, the actions of etherand sodium desoxycholate on any virus are similar,with the exception of psittacosis virus, which has beenreported resistant to the bile salt but sensitive toether. Resistant to these chemicals are: poxviruses,enteroviruses, adenoviruses, reoviruses, mouse en-cephalomyelitis (GD VII), encephalomyocarditis(Mengo). Among sensitive viruses are found thefollowing: arbor viruses, myxoviruses, Sabin's Bvirus, herpes, lymphocytic choriomeningitis. As sug-gested by Theiler (1957), since all arbor viruses so fartested have been found sensitive to sodium desoxy-cholate-and presumably ether-this property couldbe a valuable aid in the characterization of anunknown isolate as a possible arbor virus. Evidently,sensitivity to the chemicals will not uniquely charac-terize a strain as being an arthropod-borne virus; itwill, however, permit exclusion of a number of otherviruses deriving from either the original inoculum orthe inoculated animals, among which mouse ence-phalomyelitis virus is particularly to be borne inmind.

2 L. Whitman-unpublished studies.

PROCEDURES FOR IDENTIFICATION OF ARTHROPOD-BORNE VIRUSES

Circumstances of isolation. The circumstances un-der which a virus was isolated ordinarily give goodindication as to whether it is or is not an arbor virus,provided the possibility of contamination with vi-ruses in the experimental animal, usually the mouse,can be ruled out. Thus, isolates from wild-caughtmosquitos, from sentinel animals exposed in such away that mosquito vectors are the only likely contact,from the blood of human beings in the early days ofa febrile illness and from lower animals (particularlyif in the course of an outbreak which, owing tocircumstances of place, season and presence ofarthropods, is suspected of being arthropod-borne)will probably all belong to this " family ".

Elimination of other viruses by serological tests.As stated earlier, the sodium desoxycholate or ethertest will eliminate from further consideration allenteroviruses. Some of these, as well as otherviruses which are susceptible to the action of thesechemicals, could also be eliminated by simple sero-logical tests, for example, complement fixation, withknown immune sera. Immune sera against the fol-lowing viruses may be recommended: Theiler's virus(mouse encephalomyelitis), encephalomyocarditis,lymphocytic choriomeningitis, Newcastle diseasevirus, psittacosis and herpes.

Behaviour in laboratory animals and tissue cultures.The behaviour of a virus in laboratory animals, in-cluding developing chick embryos, and in tissuecultures may contribute information. Lymphocyticchoriomeningitis virus has shown higher patho-genicity for young adult mice than for newbornmice; Coxsackie A viruses cause a most extensivedegeneration or destruction of muscular tissue insuckling mice, while Coxsackie B viruses in the sameanimals bring about a typical necrosis of the inter-scapular brown fat; herpes virus gives characteristicpocks on the chorioallantoic membrane of chickembryos. These and additional properties of a newisolate can assist a decision.

Serological relationship to established arbor viruses.In the process of identification of an isolate suspectedof being an arbor virus, not infrequently evidence isobtained of serological relationships with an estab-lished virus before other information is at hand show-ing that the new isolate is arthropod-borne; thisrelationship justifies inclusion in the arbor " family ".

Filterability, size. Determination of the size of apresumptive arbor virus is a technique not usually

applied at this stage of its identification. However,certain simple operations are performed in the courseof studying a new isolate, such as filtration through aSeitz EK pad or a tight sintered-glass filter or centri-fugation in a preparatory angle-head centrifuge atspeeds between 10 000 and 12 000 r.p.m., which maygive useful leads concerning the identity of the virus.Considerable loss of titre following such operationsmay lead to the suspicion that the isolate is not anarbor virus but one of a larger size, such as herpes,lymphocytic choriomeningitis or Newcastle diseasevirus.

Determination of antigenic group

The next step in identification is to determine theantigenic group to which the isolate belongs.Not all known arbor viruses have yet been

grouped, nor have some of the more recently un-covered relationships been defined with sufficient pre-cision to warrant consideration except by the morespecialized laboratories. The trend at present ob-served, however, definitely supports the opinion thatserological relationships appear to be a generalproperty of the arbor viruses.

There are, at this writing, at least 110 to 120different arbor viruses. Most of these are listed, byantigenic groups, in Tables 1-6. As the tables show,there are four large groups, each consisting of six ormore viruses, and nine small or minor groups, eachconsisting of two or three agents. In addition, about25 viruses have shown no relationship either amongthemselves or to any of the grouped viruses.Group A (Table 1) has 14 agents; all are mosquito-

borne and all have yielded haemagglutinating anti-gens for goose erythrocytes. In Group B (Table 2)there are 28 viruses, again all capable of producingan agglutinating antigen. The last'seven in the list-excepting Powassan, for which there is not sufficientinformation-have ticks as natural vectors. Theremaining 21 viruses are known or suspected to bemosquito-borne, with the exception of bat salivarygland and Modoc viruses for which evidence is lack-ing concerning their vector.Group C consists of six viruses that so far have

been isolated only in the region of Belem, Brazil(Table 3). These viruses propagate in mosquitos andyield haemagglutinating antigens.

In the Bunyamwera group belong eight viruses(Table 4); with the exception of Kairi and Wyeomyiaall give rise to a haemagglutinating antigen and allbut Ilesha have been isolated or shown to multiplyin mosquitos. Recent studies by Whitman (in pre-

725

J. CASALS

TABLE IGROUP A ARTHROPOD-BORNE VIRUSES

Name or laboratory Where isolateddesignation ________________

Aura (Be Ar 10315) Belem (Brazil)

Chikungunya Tanganyika, Uganda, SouthAfrica, Thailand

Eastern equine encepha- USA, Panama, Brazil, Trinidad,litis (EEE) British Guiana

O'nyong-nyong Uganda

Mayaro Trinidad, Belem, Colombia

Middelburg South Africa

Semliki Uganda, West Africa (Kumba),Mozambique

Sindbis Egypt, India, South Africa,Malaya (strain AMM 2215)

Una (Be Ar 13136) Belem

Uruma Bolivia

Venezuelan equine ence- Venezuela, Colombia, Belem,phalitis (VEE) Trinidad, Ecuador

Western equine encepha- USA, Argentina, Mexico,litis (WEE) Canada, British Guiana

AMM 2021 Malayaidentical ?

Saglyama Japan

AMM 2354 Malaya

paration) show that California encephalitis virus(Hammon-Reeves) and several other agents relatedto it (see Table 5) may have some slight relationshipto some of the viruses listed in Table 4 and thusform part of the Bunyamwera group; furthermore,according to Whitman, Wyeomyia virus appears tobe a complex of close but distinguishable agents.Consequently, the Bunyamwera group may include,at the moment, as many as 18-20 viruses, in severalsubgroups or complexes.

Thus, about 65-70 viruses, or between one-halfand two-thirds of the total number of known agents,belong in four large groups, each well studied andcharacterized, with the possible exception of some ofthe viruses in the California and Wyeomyia com-plexes. Placement in one of these groups of a newisolate of an old virus or of a new virus should notpresent great difficulty. Exceptions might be someof the viruses among the newer members of theBunyamwera group and, of course, new agents ofany of these groups that happen to react in a marked-ly different manner in their cross-reacting range.

TABLE 2

GROUP B ARTHROPOD-BORNE VIRUSES

Name or laboratory Where isolateddesignation a Where isolated

Bat salivary gland (Rio~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Bat salivary gland (RioBravo)

Bussuquara

Dengue, type 1

Dengue, type 2

Dengue, type 3

Dengue, type 4

liheus

Japanese B

Modoc

Murray Valley encepha-litis

Ntaya

Spondweni

St. Louis

Turkey meningo-ence-phalitis

Uganda S

Wesselsbron

West Nile

Yellow fever

Zika

AMM 1775

SA H 336

Diphasic meningo-ence-phalitis

orCentral European tick- ormilk-borne

Kyasanur forest disease

Langat (TP 21)

Louping ill

Omsk haemorrhagic fever

Powassan

Russian spring-summerencephalitis

USA

Belem (Brazil), Colombia

Hawaii, New Guinea, Japan,India, Malaya

New Guinea, India, Trinidad,Thailand

Philippine Islands

Philippine Islands

Brazil (liheus, Belem), TrinidadHonduras

Japan, eastern Asian mainlandfrom Soviet Union to Malaya,India, Guam

USA (California)

Australia, New Guinea

Uganda

South Africa

USA, Trinidad, Panama

Israel

Uganda

South Africa

Uganda, South Africa, Egypt,Israel, India

Africa, Central and SouthAmerica, Trinidad

Uganda, Nigeria

Malaya

South Africa (similar toUganda S)

Soviet Union, Central Europe,(Sweden to Balkans)

India

Malaya

Great Britain

Soviet Union

Canada, USA (?)

Soviet Union, Central Europe

a Nairobi sheep disease virus, a tick-transmitted agent ofAfrica, may belong in this group; since confirmation, however,is needed, it is listed among the ungrouped viruses.

726


TABLE 3

GROUP C ARTHROPOD-BORNE VIRUSES

Name or laboratory Where isolateddesignation

Apeu (An 848) Belem (Brazil)

Caraparu (An 3994) it

Marituba (An 15)

Murutucu (An 974)

Oriboca (An 17)

ltaqui (An 12797)

Of the remaining 50 or so viruses, about 25 belongin nine small groups, each consisting of two or threeagents (Table 5); the vector or suspected vector forall is a mosquito, except for the Quaranfil group inwhich a tick is the vector. Finally, there are the 25ungrouped viruses, of which only those with a namedesignation are listed in Table 6. At the moment,these viruses are unrelated to one another or to anyagent with which they have been compared. Strictlyspeaking, African horse sickness and blue tongueviruses should each be considered a group rather thana single virus, since each has immunologically dis-tinct strains. Table 6 also indicates the vector orsuspected vector.Only nine viruses in the minor groups and the

ungrouped category have yielded workable haemag-glutinating antigens with the current methods. Theseare: California encephalitis; sandfly fever, Neapo-

TABLE 4BUNYAMWERA GROUP OF ARTHROPOD-BORNE VIRUSES


Bunyamwera Uganda, South Africa

Cache Valley USA, Trinidad, Belem (Brazil)

Chittoor India, Malaya

Germiston South Africa

Guaroa Colombia, Belem

llesha West Africa

Kairi Trinidad, Belem

Wyeomyia Colombia

Additional agents (seetext)

OTHER GROUPS OFTABLE 5

ARTHROPOD-BORNE VIRUSES


California encephalitis a USA (California)

Trivittatus USA

Melao Trinidad

Be Ar 8033 Belem (Brazil)

Guama Belem

Catu Belem

Bimiti Trinidad

Bwamba UgandaPongola South Africa

Simbu South Africa

Oropouche Trinidad

Sathuperi India

Turlock USA (California)

Umbre India

Anopheles A Colombia

Anopheles B Colombia

Tr 10076 Trinidad

AMM 2549 MalayaAMM 2325 Malaya

Tr 7994 Trinidad

Tr 8762 Trinidad

Tr 9223 Trinidad

Quaranfil Egypt

Chenuda Egypt

Eg Ar 1304 Egypt

a See text; this group may be part of the Bunyamwera group.

litan strain; sandfly fever, Sicilian strain; Witwaters-rand; Manzanilla; Tacaiuma; Rift Valley fever;AMM 2549; and AMM 2325.

Obviously, in such a rapidly developing field asthat of the arbor viruses, changes must be antici-pated; new groups may be formed, heretofore un-grouped viruses may be placed in groups, or some ofthe existing smaller groups may be fused. At any

727

J. CASALS

TABLE 6UNGROUPED ARTHROPOD-BORNE VIRUSES

Name or laboratory designation Where isolated Vector

African horse sickness a Africa and Eastern Medi- Culicoidesterranean area

Blue tongue a Africa, North America, CulicoidesSpain, Portugal, Israel

Colorado tick fever USA Tick

Crimean haemorrhagic fever Soviet Union Tick

Nairobi sheep disease Africa Tick

Sandfly fever, Neapolitan strain Italy Phlebotomus

Sandfly fever, Sicilian strain Italy, Egypt Phlebotomus

Hart Park USA (California) MosquitoManzanilla (Tr 3587) Trinidad Mosquito

Rift Valley fever Africa Mosquito

Tacaiuma (Be An 73) Belm (Brazil) Mosquito

Witwatersrand (SA Ar 1062) Africa MosquitoArgentinian haemorrhagic fever (Juninvirus) Argentina Mite (?)

a See text.

rate, the trend of recent results has been towards in-creasing the proportion of grouped viruses, even inthe face of an increase in the total number of recog-nized arbor viruses.

Placement of a newly isolated virus in one of thegroups is of practical value, because it speeds iden-tification by eliminating the need for further com-parison with viruses outside the group. Placing anisolate in one of the four major groups is, with fewexceptions, the easiest part of the identification. Thevalue of some of the tests enumerated earlier as ameans for grouping or, conversely, for specificitydetermination cannot yet be appraised; such is thecase with agar gel precipitin reaction, fluorescentantibody technique, haemadsorption and NT testsin tissue culture. Of the remaining, heretofore moregenerally used tests-CF, HI, NT in mice-the oneto be chosen at this particular step in the identifica-tion is that which results in the greatest overlap. Norule can be given, however, because the tests varyconsiderably in this respect from one group toanother. In a general way, HI is more inclusive ingroups A and B than is CF, while in groups C (Casals& Whitman, 1961) and Bunyamwera (Casals &Whitman, 1960) the opposite is true. In all four

groups, the NT test by the intracerebral route of in-oculation into mice is generally more specific thaneither the HI or CF test; on the other hand, theperipheral NT test is markedly overlapping, oftenmore than either the HI or the CF test. The cross-challenge test, particularly when carried out by peri-pheral inoculation of the virus, has been found, incertain instances, to result in marked overlap; how-ever, since this test is costly in time, space and labour,its use is of limited value in identification problems.An outline of the procedure for identification of a

new isolate which seems the least costly in time andlabour follows. A haemagglutinating antigen is pre-pared from infected newborn-mouse brain tissue or,with certain viruses, from blood serum collected atthe peak of viraemia. A complement-fixing antigencan be prepared, usually without difficulty, from theinfected brain tissue, and in fact the same brain tissueextract can, in most instances, be used both as com-plement-fixing and as haemagglutinating antigen.The antigens are tested by the HI or the CF test, orboth, against known positive sera, either widelycross-reacting for group determination or more spe-cific for type determination; the results of the in vitrotests are subsequently checked by the NT test to the

728


extent that is considered necessary. An integral partof the identification is to prepare an immune serumagainst the new isolate and to test it against standardantigens.An important practical question is whether it is

possible to prepare for each of the four major groupsa serum, or a limited number of sera, that will reactin some way or other with all known (and future)members of the group. If this were possible, withfour, or not many more, sera one could group about60 different viruses.

Attempts to prepare such reference group sera canbe carried out in two ways. The first is by poolingseveral immune sera, each prepared against one virusof the group. The resulting mixtures show, in addi-tion to direct reactions against the represented vi-ruses, a greater or lesser degree of cross-reaction withagents not represented in the mixture. Assuming thatthe cross-reactions by the test selected cover a widespectrum, the pool can react with a considerablenumber of viruses. The disadvantage of serum poolsis that unless the cross-reactions are sufficientlypronounced, too many type sera are needed in one

or several pools; furthermore, with each additionalserum, all are further diluted and hence less effective.The second method of preparing widely cross-

reactive sera is based on a reported basic property(Casals, 1957) of the immune response with thegrouped arbor viruses, particularly when investigatedby haemagglutination inhibition. It has been shownthat when an animal previously immunized with avirus of a group is subsequently inoculated withanother virus of the same group, the immune re-

sponse that follows has marked anamnestic (i.e., ac-

celerated and intense) as well as synergistic (i.e.,exaggeratedly cross-reactive) qualities. Studies on

the group immune response with arbor viruses are

being conducted in this laboratory; some of theresults are briefly reported in the following para-

graphs for their possible relevancy to the matter ofdiagnostic group sera.

In these studies, guinea-pigs were inoculated intra-cerebrally with 0.1 ml of a 10-1 or 10-2 dilution ofnewborn-mouse brain infected with a given virus ofa group, as indicated in Tables 7-9. About 40-50 dayslater, a sample of blood serum was taken and theanimals were reinoculated by the same route with a

similar amount of a different virus of the same group.

The guinea-pigs were bled at intervals thereafter, butfor simplicity's sake only the results of tests donewith sera taken a few days before and two weeks afterinoculation of the second virus are given in the tables.

TABLE 7

GROUP IMMUNE RESPONSE IN GUINEA-PIGSINOCULATED IN SUCCESSION WITH TWO DIFFERENT

VIRUSES OF GROUP A, AS SHOWNIN HAEMAGGLUTINATION-INHIBITION TEST

Guinea-pig No. and virus inoculatedAntigen 222 a 228 a 22(8 units) 2a282Mayaro|+ Sindbis Mayarol+ Sindbis Sindbis

Mayaro 2 560 b 5 120 20 480 10 240 0

Chikungunya 80 320 40 1 280 0

Semliki 160 1280 80 1280 0

AMM 2021 320 1 280 80 2 560 0

AMM 2354 20 640 40 640 0

Sindbis 0 640 20 1 280 640

WEE 20 1 280 80 2 560 320

Aura(Ar 10315) 0 320 0 1 280

Middelburg 0 160 20 640 0

EEE 0 320 0 1280 0

VEE 0 160 0 1280 0

a Guinea-pigs No. 222 and 228 were inoculated first withMayaro virus, later with Sindbis; the titres under the columns" Mayaro " and "+ Sindbis " were those of the serum samplestaken after injection of one or two viruses respectively.

b The titre of the serum was 1: 2560; 0 indicates no reactionat dilution 1: 20, the lowest used.

Table 7 is an example of the response withGroup A viruses. Following inoculation with Mayarovirus, two guinea-pigs, No. 222 and 228, reacted to a,high titre with this antigen, to a moderate degreewith certain viruses, most of which, with Mayaro,form part of a complex or subgroup within Group A,and very slightly, or not at all, with EEE, VEE andMiddelburg. Following reinoculation with Sindbisvirus, the immune response of these guinea-pigs, asdetected by HI, was characterized by a marked over-lap far greater than that expected in Group A whenrepeated injections of the same virus are given(Casals, 1957). The type of response shown byguinea-pigs inoculated with Sindbis virus only isexemplified by the result with serum from animalNo. 22.

In Group B, Japanese B encephalitis and SA H 336viruses were injected in succession; the results shownin Table 8 are typical. As before, the first bleeding is40 days after inoculation of Japanese B encephalitisvirus and the second, 14 days after reinoculation

729

J. CASALS

TABLE 8GROUP IMMUNE RESPONSE IN GUINEA-PIGS

INOCULATED IN SUCCESSION WITH TWO DIFFERENTVIRUSES OF GROUP B, AS SHOWN

IN HAEMAGGLUTINATION-INHIBITION TEST

Guinea-pig No. and virus inoculated a

Antigen (8 units) 232 239Japanese + SA H Japanese + SA H

B 336 B 336

Japanese B ence-phalitis 640 b 5 120 640 5 120

SA H 336 80 5120 80 5120

Russian SSE 40 1280 40 1280

Powassan 20 640 0 640

Yellow fever 80 2 560 160 2 560

Zika 320 10 240 320 10 240

Dengue 1 40 1 280 40 1 280

Dengue 2 80 2 560 40 1 280

Dengue 3 80 2 560 40 1 280

Spondweni 20 640 40 640

Bussuquara 320 10240 320 10240

a Guinea-pigs were inoculated first with Japanese B ence-phalitis virus, later with SA H 336 virus.

b The titre of the serum was 1: 640; 0 Indicates no reactionat dilution 1: 20, the lowest used.

with SA H 336. Although the sera were testedagainst only a few of the nearly 30 different antigensnow available in Group B, the antigens selected wererepresentative of the several subgroups or complexesin the group. In particular, Powassan antigen wasincluded, since previous experience (Casals, 1960)had shown that this antigen cross-reacted poorly.In Table 8 it can be seen that guinea-pigs No. 232and 239 responded to the second stimulus with atypical group response which, in fact, was at ahigher titre to the Zika and Bussuquara antigens,viruses not inoculated to these animals, than to thetwo viruses inoculated.The results with Group C viruses were obtained

with a different schedule of immunization andbleeding from that followed in the previous examples.The interval between inoculations was approximately200 days. In Table 9, the samples of sera fromguinea-pigs inoculated twice-either with differentviruses or with the same virus-were taken one daybefore and seven days after the second inocula-tion. The serum from animal No. 43, inoculated

TABLE 9GROUP IMMUNE RESPONSE IN GUINEA-PIGS

INOCULATED IN SUCCESSION WITH TWO DIFFERENTVIRUSES OF GROUP C, AS SHOWN

IN HAEMAGGLUTINATION-INHIBITION TEST

Guinea-pig No. and virus inoculatedAntigen 5a4(4 units) 54 a 58 b 43

Caraparu Oriboca CCarPparu araparu Oriboca

Oriboca 20c 2 560 20 20 640

Caraparu 640 5 120 1 280 10 240+ 0

Apeu 320 2 560 640 5120 0

Marituba 40 640 80 320 0

Murutucu 40 640 80 160 0

a Guinea-pig No. 54 was inoculated with Caraparu virus first,later with Oriboca virus.

b Guinea-pig No. 58 was inoculated twice with Caraparu virus.c The titre of the serum was 1: 20; 0 indicates no reaction a

dilution 1: 10.

with one virus only, was taken seven days frominoculation. It is evident from the results in thetable that inoculation of different viruses, Caraparuplus Oriboca, led to a wider group reactivity thandid inoculation of the same virus, Caraparu, twice.Furthermore, although a single inoculation ofOriboca virus to guinea-pig No. 43 resulted in aserum having no cross-reactivity with the rest of theantigens in Group C, inoculation of Oriboca toguinea-pig No. 54, previously sensitized withCaraparu, resulted not only in a greater response toOriboca antigen than was shown by guinea-pigNo. 43, but also in a marked boosting action on theantibody titres against all other antigens as well.While the experimental evidence so far has indi-

cated that by HI tests sera from guinea-pigs inocu-lated with only two selected viruses of a group arewidely cross-reactive, the behaviour of such sera inCF and NT tests is only now being investigated.Preliminary results with Groups A and B indicatethat by neither of these tests is cross-reactivity asmarked as by the HI test.

Conceivably, polyvalent or group sera should havetheir maximum applicability at the level of a regionalor local laboratory for the purpose of rapid andsimple grouping of local isolates. Under thesecircumstances, preparation of group sera reactive byCF and NT as well as by HI may be accomplishedwith greater ease by inoculation into animals of a

730

PROCEDURES FOR IDENTIFICATION OF ARTHROPOD-BORNE VIRUSES 731

few selected viruses known to exist in the area, foreach group. This procedure has been followed andfound helpful by Causey et al. (1961) in their studiesof isolates in the Amazon Valley.To summarize this phase of the identification,

group determination can be easily carried out bypreparing an in vitro antigen-haemagglutinating orcomplement-fixing-with the unknown virus, whichis tested against polyvalent sera-or assorted sera-for each of the four major groups and others asavailable. At the same time, an immune serumagainst the isolate is prepared for subsequent use inthe identification.

Determination of type (or " species")Regardless of the result of attempts to group an

isolate, the next step in its identification is determina-tion of specificity or type. As previously stated, ifthe virus is placed in a group, only the other membersof the group need be considered in subsequent com-parative studies; if no group affiliation has beenproved, further comparison is made with ungroupedviruses only, including those that belong in the, sofar, " minor " groups, as well as the strictly un-grouped.The same methods are used for specificity deter-

mination as were used in the preceding step, withthe understanding that an immune serum against thenew isolate is now indispensable. For the sake ofconvenience, specificity determination will be treatedin two sections, as applied to (1) grouped viruses inthe four major groups, and (2) viruses in the minorgroups and those strictly ungrouped.

Grouped viruses. The choice of method is dictatedby the consideration that specificity, rather thancross-reactivity, is desired. The new isolate-and itsserum-is to be compared with all the remainingmembers of the group, or as many as are deemednecessary. From these results, a conclusion isreached as to whether the isolate is or is not a newvirus. Of the methods at present in general use, thepreferable ones (see earlier under group determina-tion) are CF and intracerebral NT 1 tests for GroupsA and B, and HI and intracerebral NT tests forGroups C and Bunyamwera.

In specificity determination, it should be borne inmind that within an antigenic group there are smallersubdivisions or complexes, consisting of viruses more

1 Although this cannot be entirely substantiated, it wouldappear that the intracerebral NT tests gain in specificity whencarried out in newborn mice.

closely related among themselves than to the re-maining members of the group. Thus, in Group A,for example, Mayaro, Semliki, Chikungunya and afew other viruses are more closely related amongthemselves than any of them is to either WEE orSindbis, which, in turn, are similar. In Group B,among various other examples, Japanese B and WestNile viruses are close, while only distantly related toseveral tick-borne viruses or to the dengue viruses.Although clear-cut establishment of such subgroupsmay have its limitations, knowledge of their existencenevertheless has a definite advantage. At this pointin the identification, it is advisable to include in thetests type sera from representative viruses of eachsubgroup, assuming that it is not feasible or practicalto have all viruses in the group represented.

In type determination, particularly when the workinvolves closely related viruses in Groups A and B-less work has been done with the other groups-thenature of the typing sera is important.Whether or not sera from different animal species

are more or less cross-reactive cannot be conclusivelystated; this point is definitely worth further explora-tion. However, it has been established that repeatedinjections of a virus into an animal result generally inan increase of cross-reactivity of the serum relative tothe homologous reaction and that in the course ofimmunization or infection there is an early periodwhen the cross-reactivity of a serum is at a minimum.Consequently, sera for determination of specificityshould be prepared as far as possible by earlybleeding of the experimental animal following asingle injection.A method of great value for this step in the iden-

tification of an isolate, in addition to other applica-tions, is that developed by Clarke (1960) and calledby her the antigenic analysis technique. It consistsin using in the HI test immune sera from which thecross-reacting group antibodies have been removedby adsorption with another virus of the group; byselecting the adsorbing virus and the proper ex-perimental conditions, sera are prepared with onlyspecific antibodies left. This technique has beenapplied by Clarke mainly to Group B viruses, sinceit is in this group that diagnostic difficulties havemore often been found in the past. It seems logicalthat the method should be generally applicable.

Ungrouped viruses. Until the additional smallergroups are more precisely defined, it may be moreexpedient for the identification procedure to includein this section all the viruses not in the four larger

J. CASALS

groups; there are approximately 50 of them. Sinceonly about nine yield a haemagglutinating antigenwith the current methods, and since the newermethods of study are even less developed with theseviruses than with the grouped agents, the diagnosticserological methods are, at present, and, for allpractical purposes, only the CF and NT tests in mice.The ordinary procedure for identification of an

ungrouped isolate is to prepare a CF antigen and testit against type immune sera for each of the knownviruses; the positive findings are then checked by theNT test. With the large number of types now in thiscategory, the task becomes increasingly cumbersome.On the other hand, a positive reaction with a strictlyungrouped type serum is diagnostic.

Pools of type sera (as described in another section)have been used occasionally in CF tests (Causey etal., 1961), and perhaps their use should be genera-lized. In order not to dilute each serum to excess, thepools used consisted ordinarily of four sera; if areaction was positive, the specific answer was to befound in a second test in which the sera were testedseparately.The problem with the ungrouped viruses is evident-

ly one of availability of a complete type serum collec-tion which, in turn, requires that all viruses be athand; similarly, testing by CF or by other tests animmune serum against the isolate requires all theantigens and therefore all the viruses in this category.While maintaining such a collection of viruses andsera and performing CF and NT tests on the scalerequired is, admittedly, not a simple proposition, atpresent it appears to be the only practical procedure.

EVALUATION OF RESULTS:

CONCLUSIONS CONCERNING IDENTIFICATION

The interpretation of immunological test resultsmay lead to one of several conclusions concerningthe identity of the new isolate:

(1) The isolate is unlike and unrelated to any of theestablished viruses; it is, therefore, a new virus. Thisis the simplest case, if not in terms of effort at leastas regards the absence of reservations about the con-clusion reached.

(2) The isolate is indistinguishable, within experi-mental limits, from an established virus. The con-clusion, again simply reached, is that this is a newisolation of an old virus.

(3) The isolate is related to, but distinguishablefrom, an established virus or viruses, This is the case

in which difficulties of interpretation begin to appear,depending on the degree of distinctness of the newisolate.

It is obvious that the greater the differences be-tween the unknown virus and previously describedviruses, the easier it will be to say that the unknownis truly a new agent. The real problem facing us isthe proper evaluation of minor immunological dif-ferences. Should these be considered of sufficientimportance to justify specific separation, or wouldit be better to accept them as reflecting strainvariants ?From the purely diagnostic point of view, it would

certainly be advisable to disregard small differencesand thus keep the number of arbor viruses withinsmaller limits. However, if the facts show that thereare in the arbor virus class closely related but distinctviruses, this knowledge must be accepted.The situation becomes particularly complicated

when the new isolate seems to fall between two pre-viously described, close viruses (the possibility ofvirus mixtures can also be entertained). This appearsto be the case with some of the strains isolated inIndia which are being investigated by T. H. Work(personal communication); his studies show thatsome of the isolates are closely related to bothJapanese B and West Nile viruses. If close proximityjustifies considering a new isolate as a strain of anold virus, it is easy to see how a situation similar tothat of Work's strains could lead, in the extreme,to considering as a single agent viruses now admittedto be different.

This aspect of the problem of identification-whether, as has been said repeatedly, to split or tolump-is well worth consideration, if for no otherreason than because instances of new isolates closeto but not identical with established viruses occurwith increasing frequency. Among instances of suchclose antigenic similarity between viruses and recentisolates, the following, taken from the literature orobserved at the Rockefeller Foundation laboratoriesboth in the field and in New York, can be mentioned:AMM 2021 and Sagiyama; Mayaro Tr 4675, MayaroTr 15335 and Uruma; Sindbis (Ar 339), AMM 2215,IA 1036 and SA Ar 86; Chikungunya (Ross), Ba H306 and TH 35; Uganda S, Makonde and SA H 336;Bwamba and Pongola; dengue type 2, strains NewGuinea B and Tr 1751; West Nile (Smithburn), IG2266, Egypt 101 and some recent Indian strains;Wyeomyia prototype and several Wyeomyia-com-plex strains isolated in Trinidad and Brazil. Some ofthe isolates have been considered by the workers who

732


isolated them as new viruses, others as strains of oldviruses; with some, a decision has not yet beenreached. In the resolving of these problems, con-siderable aid can be anticipated from the use of ad-sorbed sera (Clarke, 1960) in order to concludewhether strains are different-although closely allied-4or identical. Whatever methods of immunologicalanalysis are employed, whether those at present inroutine use-CF, NT and HI-or others that arenow being developed, it would appear that the defini-tion of serological identity between strains shouldundoubtedly be based on the concept that a virustype-or " species "-is a cluster of different in-dividualities grouped around and resembling a pro-totype or model, rather than a number of strains allidentical with a prototype. Failure to adopt thisconcept might well lead to a situation in which nearlyevery strain would be unique.

There is a final consideration which applies to allsteps of serological identification-namely, the possi-bility that the isolate is a mixture of two viruses. Ifthe two viruses are unrelated, there should be nodifficulty in separating them. However, if the twoviruses are closely related, for example, Japanese Band West Nile, the problem of separation may begreat and require a combination of methods, such asinoculation of final dilutions, passage in the presence

of immune sera and, when possible, plating on tissueculture monolayers.

CONCLUSION

An attempt has been made to present in a generalway the problems involved and the methods em-ployed in the identification of arthropod-borne vi-ruses. It is hardly necessary to add any conclusionsexcept to underline the essential need in a diagnosticlaboratory of as complete a collection of type im-mune sera as possible. At the local or field laboratorylevel, immune sera against the local isolates can beproduced, by means of which it is possible to deter-mine the number of distinct virus types existing asnew isolates are brought in. Furthermore, at thislevel, or at the regional laboratory level, considerableprogress can be made towards the ultimate serolo-gical identification of a virus by means of a limitednumber (four to eight) of group immune sera havinga wide degree of overlap. These could be preparedregionally according to the recognized arthropod-borne viruses in the geographic region in question.Through use of such sera, it should be possible toplace, or tentatively exclude, viruses in the majorantigenic groups, a considerable step in the processof final identification.

ACKNOWLEDGEMENT

The author wishes to express his deep gratitude to Dr Max Theiler for continued advice and helpful sug-gestions in the preparation of this manuscript.

RItSUMIt

Pour identifier un virus transmis par les arthropodes,c'est-A-dire capable d'infecter les vertebres et de se mul-tiplier chez les arthropodes, il faut preciser un nombresuffisant de ses proprietes, afin de savoir s'il est nouveauou s'il a deja ete decrit. Les proprietes serologiques sont,en l'etat actuel des connaissances, celles qui conviennentle mieux pour l'identification, en raison de leur stabilite,de leur sensibilit6 qui permet de distinguer differentstypes, et de la relative facilite de leur recherche. Lesepreuves serologiques les plus couramment utilisees sontles tests de neutralisation, de deviation du complement etd'inhibition de l'Femagglutination.On identifiera un virus transmis par les arthropodes a

la lumiere des connaissances actuelles des groupes anti-geniques, manifestes par les caracteres fondamentaux dela r6ponse immunologique d'un h6te, a l'un ou l'autregroupe de virus. L'identification obeit a une progression

logique. Le premier point est de savoir si la souche etudi6eappartient a la famille des virus transmis par les arthro-podes. Ce ne sont pas les proprietes immunologiques quipermettent de l'affirmer, car on n'en connait pas quisoient communes a tous les virus transmis par les arthro-podes. On se fondera donc sur d'autres caracteres biolo-giques et les circonstances de l'isolement. Parmi cescaracteres, on peut citer la propagation en serie chez lesarthropodes hematophages, la sensibilite a l'ether ethyli-que et au desoxycholate de sodium, l'action sur les ani-maux de laboratoire et les cultures tissulaires.Le second point consiste A preciser a quel groupe anti-

genique appartient la souche a l'etude. On connaitactuellement 110-120 virus transmis par les arthropodes.Ils sont repartis en quatre groupes principaux: A (14 sou-ches); B (29); C (6); Bunyamwera (8 au moins, peut-etre18-20), puis en neuf groupes secondaires comprenant

733

734 J. CASALS

25 souches, A raison de 2-3 par groupe. II y a en outre25 souches qui n'ont pu etre rattachees, jusqu'ici, a aucundes groupes pr&cedents.

I1 est relativement facile de classer un virus dans l'undes quatre groupes principaux, si l'on dispose de serumsimmuns representatifs de chaque groupe. II est habituelle-ment simple de determiner le groupe auquel un virusappartient, en utilisant des serums a large spectre serolo-gique a l'interieur du groupe, et en utilisant le test immu-nologique qui donne le plus facilement des reactionscrois6es. Cela fait, il reste a comparer le virus a l'etudeavec les autres virus constituant le groupe; les recherchessont ainsi simplifiees.

Lorsqu'on a determine le groupe, on utilise les serumsles plus specifiques a 1'int6rieur du groupe et les reactionsdonnant le moins de reactions croisees, car il s'agit alorsde distinguer et non plus de rapprocher.

L'identification d'un virus qui n'appartient pas 'a l'undes principaux groupes demande plus de temps. II fautalors le comparer avec de nombreux virus, sinon avectous ceux qui sont etrangers a ces quatre groupes.

II est essentiel pour un laboratoire de diagnostic dedisposer d'une collection aussi complete que possible dedifferents serums immuns types, surtout de ceux qui cor-respondent aux virus dont la presence dans la regionconsideree est connue.

REFERENCES

Andrewes, C. H. & Horstmann, D. H. (1949) J. Micro-biol., 3, 290

Burnet, F. M. & Lush, D. (1940) Aust. J. exp. Biol. med.Sci., 18, 141

Casals, J. (1957) Trans. N. Y. Acad. Sci., Series 2, 19, 219Casals, J. (1958) Antigenic classification of arthropod-

borne viruses. In: Sixth International Congresses onTropical Medicine and Malaria, 1958. Abstracts of thepapers, Lisbon, Instituto de Medicina Tropical, p. 160

Casals, J. (1960) Canad. med. Ass. J., 82, 355Casals, J. & Whitman, L. (1960) Amer. J. trop. Med.

Hyg., 9, 73Casals, J. & Whitman, L. (1961) Amer. J. trop. Med.

Hyg., 10 (in press)

Causey, 0. R., Causey, C. E., Maroja, 0. M. & MacedoD. G. (1961) Amer. J. trop. Med. Hyg., 10 (in press)

Clarke, D. H. (1960) J. exp. Med., 111, 21Dingle, J. H. & Ginsberg, H. S. (1959) The adenovirus

group. In: Rivers, T. M. & Horsfall, F. L., ed., Viraland rickettsial infections of man, 3rd ed., Philadelphia,Lippincott, p. 613

National Foundation for Infantile Paralysis, Committeeon the Enteroviruses (1957) Amer. J. publ. Hlth, 47,1556

Smith, W. (1939) J. Path. Bact., 48, 557Theiler, M. (1957) Proc. Soc. exp. Biol. (N. Y.), 96,

380Theiler, M. & Casals, J. (1959) Klin. Wschr., 37, 59

Documents

Procedures for Identification of Arthropod-borne Viruses*