15 September 1972, Volume 177, Number 4053

Lectins: Cell-Agglutinatiland Sugar-Specific Protei

Lectins provide new tools for studying polysacchariglycoproteins, and cell surfaces, and for cancer resea:

Nathan Sharon and Halina

Proteins that possess the remarkableability to agglutinate erythrocytes andother types of cell are widely distrib-uted in nature. Such agglutinins arefound predominantly in the seeds ofplants, in -particular those of the le-gumes, but they are also present inother parts of plants such as roots,leaves, and bark (1-5). In addition, theyoccur in invertebrates, for example, insnails and the horseshoe crab, and inlower vertebrates such as fish (6). Plantagglutinins are commonly referred toas phytohemagglutinins or phytoagglu-tinins. However, as cell-agglutinatingproteins also occur in organisms otherthan plants, the term "lectins," pro-posed by Boyd (2), appears more suit-able and will be used in this article.Prominent examples of lectins are con-canavalin A from jack bean (7), soy-bean agglutinin (8, 9), and wheat germagglutinin (10-12).As well as the ability to agglutinate

red blood cells, which makes easy theirdetection, lectins exhibit a host of otherinteresting and unusual biological andchemical properties (Table 1). Some ofthe lectins are specific in their reactionswith human blood groups (ABO andMN) and subgroups (A1) and havetherefore been used in blood typing andin investigations of the chemical basisof blood group specificity (1, 2, 13, 14).The authors are members of the department

of biophysics, Weizmann Institute of Science,Rehovoth, Israel, where Dr. Sharon is the Sadieand Joseph Danciger Professor of MolecularBiology.15 SEPTEMBER 1972

Certain lectins are mthat they can stimulatetion of lymphocytes fiing" cells into largewhich may ultimatelydivision. A most usefulstudy of lectins has beof cytogenetics and tlderstanding of relati4chromosome abnormadiseases. The stimulalcytes by lectins also prtant tool for the exabiochemical events inviversion of a resting celgrowing one.The interactions of

can, in many instancspecifically by simpleThis finding has led tthat lectins bind specharides on the surfahas provided a newvestigation of the arcsurfaces. Lectins alsooligosaccharides andcipitate polysaccharid4teins; the precipitatiorsugars, as in the casetion reaction. This pIcan be used for tipurification of carbohpolymers, and for thchemical structure. Itthe development of n(matography" techniqufication of the lectinsThe interaction of

SCIESNCE:

and glycoproteins with lectins may belikened to the antigen-antibody reac-tion and can serve as a most usefulmodel for the study of this reaction.Because of the high specificities of

ng some lectins, and the homogeneity oftheir saccharide-binding sites, they pro-

ins vide a unique opportunity for the in-vestigation of homogenous combiningsites on antibody-like molecules. An

des, added advantage of lectins is that theircombining sites are small, which makes

rch. the investigation of these sites simpleand attractive. Furthermore, the reac-

Lis tion of lectins with saccharides mayserve as a general model for protein-carbohydrate interactions.The recent observations on lectins

specific for tumor cells (10-12, 17, 18)itogenic (15) in have led to a great surge of interest inthe transforma- these compounds, especially among sci-rom small "rest- entists engaged in cancer research.blast-like cells Some lectins preferentially agglutinateundergo mitotic mammalian tissue culture cells that

I outcome of the have been transformed by oncogenicen the expansion viruses (Fig. 1) or by chemical carcino-he increased un- gens, as well as spontaneously trans-onships between formed cells. These and related find-lity and human ings indicate that the surface of ation of lympho- transformed cell differs from that of*ovides an impor- its untransformed counterpart, andmination of the iraise the hope that studies with lectinsolved in the con- may lead to a better understanding of.1 into an actively cancer. Moreover, they have prompted

/ investigations to use lectins as inhibi-lectins with7ells tors of growth of malignant cells in.es, be inhibited vitro and in vivo (19), and lectin-bind-sugars (4, 16). ing synthetic polymers for the immu-

to the conclusion nization of mice against tumors (20).scifically to sac- Lectins are frequently toxic to ani-Lce of cells, and mals, a property that may in part betool for the in- tfsponsible for the poor nutritive value;hitecture of cell of some plant proteins (21). Because ofbind mono- and the great economic value of lectin-richspecifically pre- plant seeds, such as the soybean and

es and glycopro- castor bean, they have for a long timen is inhibited by attracted the attention of food scien-of the agglutina- tists, food technologists, and nutrition-roperty of lectins ists.- & ---.

ie isolation andkydrate-containingke study of theirhas also led to

ew "affinity chro-ies for the puri-themselves.

I polysaccharides

Although hemagglutinating activityhas been detected in extracts obtainedfrom over 800 plant species (1-5, 22),and from numerous species of snailsand other animals (6), only a smallnumber of lectins have been isolated inpurified form and investigated to anyextent. However, as lectins have be-

949

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

on J

une

15, 2

014

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

Fig. 1. (a) Normal rat cells and (b) polyma virus-transformed rat cells after incubation for 30 minutes with 100 ug of soybeanagglutinin per milliliter (X 80) (18).

come a subject of intense interest inmany laboratories, this article seemstimely. In addition to reviewing thepresent status of research on lectins,we demonstrate their great potential inelucidating the structures of carbohy-drate-containing polymers, and in par-ticular their use in investigations of thearchitecture of cell surfaces. We alsodiscuss briefly the possible role of lec-tins in nature. Information about themitogenic action of lectins is availableelsewhere (15), and there are reviewsand books (1-6, 21) that give moredetailed information on various otheraspects of lectins.

History

The study of lectins was initiated byStillmark who, in 1888, was the first todescribe the phenomenon of hemag-glutination by plant extracts. Stillmarkstudied the toxicity of castor beans(Ricinus communis) and of the presscake formed during the production ofcastor oil (1, 2). He found that castorbeans contained a highly toxic proteinwhich he named ricin, and which wascapable of agglutinating the red cells ofhuman and animal blood. Soon afterthe discovery of ricin, another lectinwas found, abrin from the jequiritybean (Abrus precatorius). Ricin andabrin immediately drew the attentionof Ehrlich, who chose to work withthem on problems of immunology rath-er than with bacterial toxins, such asthe diphtheria toxin, which were sopopular among bacteriologists at thattime. Using ricin and abrin, Ehrlich

950

discovered during the 1890's some ofthe most fundamental principles of im-munology. For instance, in 1891 hereported that mice were rendered im-mune to a lethal dose of ricin by re-peated small subcutaneous injections ofthe toxin. He also showed that theserum of an immune mouse could neu-tralize the toxicity of ricin and, fur-thermore, that this action was specific,since the "antiabrin," which developedin the serum of an animal immunizedwith abrin, would not neutralize thetoxic effects of ricin nor would "anti-ricin" neutralize abrin. Other classicstudies in immunology, such as one ofthe first experiments on the reversibil-ity of the antigen-antibody reaction,were also conducted with these twolectins (23).

In 1908, Landsteiner and Raubit-schek (24) established that the relativehemagglutinating activities of variousseed extracts were quite different whentested with red blood cells from differ-ent animals, and compared this speci-ficity with that of antibodies of animalblood serum (25). Nevertheless, forseveral decades it was presumed thatplant hemagglutinins were nonspecific.It was not until the end of the 1940'sthat the independent studies of Ren-konen (26) and of Boyd and Reguera(27) led to the discovery that certainseeds contain agglutinins specific forsome human blood group antigens.Thus, it was found that crude extractsof the lima bean (Phaseolus limensis)are virtually specific for the A antigen,although when concentrated they reactweakly with type B cells as well. Sim-ilarly, seeds of the tufted vetch (Vicia

cracca) contain powerful agglutininsthat act much more strongly on A thanon B or 0 cells. Other lectins specificfor type A cells, as well as lectins spe-cific for blood types H(O), M, and Nare now known.

Most of the studies described abovewere conducted with crude extracts ofplant origin. The first lectin that wasobtained in crystalline form was con-canavalin A from the jack bean (Ca-navalia ensiformnis). Although crystal-lized by Sumner as early as 1919, itwas only in 1936 that Sumner andHowell (7) identified the hemagglutininof jack bean with concanavalin A, andshowed that it agglutinated horse eryth-rocytes at concentrations as low as0.1 microgram per milliliter. They ob-served that the protein reacted withglycogen and suggested that hemagglu-tination may be a consequence of thereaction of concanavalin A with carbo-hydrate in stroma proteins. Ricin waspurified by Osborne and his co-workersin 1905. Further purification and crys-tallization of ricin was achieved in thelate 1940's (28). It was observed, how-ever, that the material, even after sev-eral crystallizations, consisted of morethan one compound.

Specificity

Lectins differ widely in their speci-ficity in the cell-agglutination reactions.and in the susceptibility of this reactionto the inhibition by saccharides. Mostlectins agglutinate the erythrocytes ofall human blood groups, acting at ap-proximately the same dilution with

SCIENCE. VOL. 177

different blood types, and are usuallyreferred to as nonspecific lectins, orpanagglutinins, or lectins with broadspecificity. The abundance of nonspecif-ic lectins is evident from a recent study(22) in which extracts of 2663 varietiesof plants were tested for human bloodgroup activity. Of these, 711 were non-specific and 90 were blood group spe-cific. An additional 227 extracts hemo-lyzed red cells of all groups, whereasthe remaining 1635 extracts wereinactive.The specific lectins preferentially

agglutinate human erythrocytes of agiven blood type and form precipitateswith corresponding soluble blood groupsubstances. Some of the crude plantextracts which exhibit blood groupspecificity are shown in Table 2. Cer-tain lectins act equally well on erythro-cytes of blood types A and B (A + Blectins), but no lectin is available thatis specific for type B or for Rh. Theunavailability of a potent type B lectinimposes a serious limitation on theroutine use of lectins for blood typing.Some lectins are nevertheless used inhematology. In particular, a crudepreparation from seeds of the gorse(Ulex europeus) is used in the identifi-cation of "secretors," human beingswho secrete blood group substances A,B, or H in saliva or other body fluids(1, 2).

Although it would be desirable touse highly purified lectins for bloodtyping, crude preparations are equallysuited for this purpose. Only rarelydoes an extract from a single sourcecontain two or more lectins with dif-ferent blood group specificities. Ad-sorption with one type of humanerythrocyte of a lectin preparation thatagglutinates two different types oferythrocyte (for example, A and 0)nearly always removes both types ofactivity. Different species of the samegenus may, however, contain lectins ofdifferent specificities: Vicia cracca isspecific for blood group A, Viciagraininea is specific for blood group N,while Vicia faba is nonspecific (5).Type specific lectins are inhibited

best by saccharides which serve as partof the immunodeterminant of the cor-responding blood group substance:lectins specific for group A antigen byN-acetyl-D-galactosamine (or its a-gly-cosides) and lectins specific for groupH(O) antigen by L-fucose. Indeed, thefirst information concerning the roleof sugars as determinants of bloodgroup specificity was obtained in 1952by Watkins and Morgan (13; see also1 5 SEPTEMBER 1972

Table 1. Properties and uses of lectins.

Property Application

Specificity for human blood groups Blood typing; structural studies of blood groupsubstances; identification of new blood types;diagnosis of secretors

Toxicity in animals Studies of nutritional value of animal food-stuffs

Induction of mitosis in lymphocytes Studies of chromosomal constitution of cellsand detection of chromosome abnormalities

Agglutination of malignant cells Investigation of architecture of cell surfaces,and its change upon transformation

Precipitation of polysaccharides and Isolation, purification, and structural studies ofglycoproteins carbohydrate-containing polymers; model for

antigen-antibody reactionsBinding of sugars Studies of specific combining sites on proteins

14) who studied type A-specific agglu-tinins of plant origin (Vicia cracca andPhaseolus limensis) and type H(O)-specific agglutinins from the serumof the eel (Anguilla anguilla) and theseeds of Lotus tetragonolobus. SinceN-acetyl-D-galactosamine specifically in-hibited type A-specific lectins, Watkinsand Morgan (13) concluded that thissugar serves as a determinant ofhuman blood group A specificity.Similarly, the agglutination of groupO cells by the type O(H)-specificlectins was best inhibited by methyla-L-fucopyranoside, indicating that thea-L-fucOsyl residue is a determinant ofH(O) specificity. Both conclusions havebeen fully substantiated in subsequentstudies.Among the large group of lectins

not specific for human blood groups,a few are known to be highly specificwith respect to the binding of saccha-rides. The sugar specificity of thesenonspecific lectins is not necessarilyrelated to that of blood group deter-minants, as shown by the followingexamples. Concanavalin A, soybeanagglutinin, and ricin are powerful,nonspecific lectins, but they exhibitdistinct sugar specificities: concanavalinA toward a-D-mannopyranosides, a-D-glucopyranosides, and a-N-acetyl-D-glu-cosaminides (29); soybean agglutinintoward N-acetyl-D-galactosaminides (aand /B) (30); and ricin toward D-galac-tose t(6, 31). We may therefore con-clude that these lectins interact witherythrocytes by binding to saccharidereceptors different from the blood typedeterminants, and that such receptorsare present on all red blood cell sur-faces. This would explain the caseof concanavalin A, since as far as isknown, none of the soluble bloodgroup substances contain D-mannose orD-glucose. Moreover, although bloodgroup substances are now known tocontain terminal nonreducing N-acetyl-

a-D-glucosaminyl residues as an im-munodeterminant, this determinant isnot blood type specific (32). In thecase of soybean agglutinin and ricin,the lack of blood type specificity maybe the result of their inability to dis-tinguish between a- and /3-linked N-acetyl-D-galactosamine or D-galactose.In blood group substances, specificitydeterminants are a-linked; for example,N-acetyl-D-galactosamine in type A andD-galactose in type B, whereas on theerythrocyte surface we may assumethat these saccharides occur both ina and /B linkages, with the latter typeof linkage perhaps predominating.A more complex pattern of specific-

ity emerges when the agglutination oferythrocytes, as well as other cells fromdifferent animals, by a variety oflectins is examined and compared withthe sugar specificity of these lectins(33) (Table 3). Lectins that do notact on erythrocytes but will agglutinatesarcoma-i 80 cells (34) or leukocytes(35) are present in the red kidney beanand appear to be distinct from thehemagglutinin found in the same bean.The susceptibility of erythrocytes

and of other types of cell to agglutina-tion by lectins is markedly enhancedby mild treatment with proteolytic en-zymes, such as trypsin or Pronase (4,6). For example, the concentration ofsoybean agglutinin (0.1 to 0.2 ttg/ml)sufficient for the agglutination oftrypsinized rabbit red blood cells isabout 200 times lower than that whichwill cause, agglutination of the un-treated cells (30, 36). Therefore, asearch for lectins should include in-vestigations of cells from a variety ofsources, both before and after theyhave been treated with proteolytic en-zymes. Moreover, the agglutinationtests should be carried out in the pres-ence of different additives. This is im-portant, since some lectins behave asif they were "incomplete" in that they

951

do not agglutinate erythrocytes insaline, but do so in the presence ofbovine serum albumin or syntheticpolymers such as polyvinylpyrrolidone(4). Metal ions (for example, calciumand magnesium ions) should also beincluded in the assay mixtures, becausecertain lectins have been shown to re-quire such ions for their activity (seebelow). An interesting incompletelectin is the agglutinin from the seedsof Lotus tetragonolobus. This lectin,when purified, agglutinates humantype 0 cells at a concentration of 38jug/ml. Upon the addition of a crudeextract of the same seeds, diluted to apoint at which the extract by itselfwould not cause agglutination, 2jug of the purified lectin per milliliteris sufficient for the agglutination oftype 0 cells (37).

Purified Lectins

Lectins can be purified from plantextracts or other sources by conven-tional techniques of protein chemistrysuch as salt fractionation and chroma-tography on ion exchangers or othertypes of adsorbent (for example, hy-droxylapatite). More recently, usehas been made of the specific bindingproperties of lectins to develop newand convenient "affinity chromatog-raphy" methods for their purification.Thus, concanavalin A can be purifiedby adsorption on cross-linked dextran(Sephadex) gels (38, 39) and elutionwith either D-glucose or a buffer oflow pH. Specific adsorption on Sepha-dex gels has also been used for thepurification of lectins from the lentil(Lens esculenta) (40, 41) and the seedsof the garden pea (Pisum sativum)(42). In the case of the type A-spe-cific lectins of the seeds of the horsegram (Dolichos biflorus) (43) and thevineyard snail (Helix pomatia) (44)purification was achieved, essentiallyin a single step, by adsorption of thecrude extracts on a column of in-soluble polyleucyl blood group A sub-stance and subsequent elution withN-acetyl-D-galactosamine. A similarstep was used in the purification of thelectins from lima bean (45), whereasthe type A-specific lectin from Viciacracca was isolated by specific adsorp-tion to blood group A substancecoupled to Sepharose B (46). The L-fucose-binding protein of the seeds ofLotus tetragonolobus was specificallyprecipitated by a trifunctional fucosyldye and the dye removed from the

952

Table 2. Human blood group specificity ofsome crude plant lectins (1, 2).Blood group . .agglutinated Origi of lecti

A Phaseolus limensisVicia craccaDolichos biflorus*Crotalaria aegyptiaca

A + B Sophora japonicaCalpurina aurea

H Cytisus sessilifoliusLaburnum alpinumLotus tetragonolobusUlex europeust

M Iris amaraN Vicia graminea*

Bauhinia purpurea* In use in blood banks. t Used for theidentification of secretors.

lectin by treatment with an ion ex-change resin (37). By means of thesetechniques, it is often easy to obtainlarge quantities of purified lectins,several hundred milligrams from a kilo-gram of starting material.The limited number of lectins that

have been obtained in highly purifiLedform, and whose chemical and physicalcharacteristics have been investigatedin greater or lesser detail, form a some-what variable group of substances(Table 4) and it is difficult to singleout properties that may be character-istic of the group as a whole. Most ofthe purified lectins contain covalentlybound sugars, and can thus be classi-fied as glycoproteins. A notable ex-ception is concanavalin A [and possiblyalso the lectin of the garden pea (42)],which is devoid of sugar residues (39,47) and is, therefore, not a glycopro-tein. The carbohydrates found in lectinsinclude mannose, glucosamine, and ga-lactose, which are typical constituentsof animal glycoproteins; other sugarssuch as xylose and arabinose are some-times found, but in smaller amounts.Most lectins are relatively rich in

aspartic acid, serine, and threonine,which may comprise as much as 30percent of their amino acid contentand are very low in, or completelydevoid of, sulfur-containing aminoacids. Such a pattern of amino acidsis characteristic of many plant pro-teins. In the snail lectin (44), asparticacid and serine are the most abundantamino acids, but a significant concen-tration of cysteine and methionine (23and 12 residues, respectively, per mole)is also present. Cysteine is also foundin the lectins of the meadow mush-room (Agaricus campestris) (48), ofthe lima bean (45, 49) and of thehorseshoe crab (Limulus polyphemus)

(50). Whenever tested, lectis havebeen found to contain the metal ionsMn2+ and Ca2+, and it has beenshown that there is a requirement ofmetal for the carbohydrate-binding andagglutinating activities of these lectins(41, 45, 51, 52).

Values reported in the literature forthe molecular weight of different lectinsvary markedly from 26,000 for wheatgerm agglutinin (11, 12) to 269,000for the lima bean lectin (49) and to400,000 for the horseshoe crab lectin(50). Moreover, markedly differentvalues have sometimes been obtainedfor the same lectin, for example,55,000 (53) to 100,000 (54) for con-canavalin A. This may be partly be-cause many (see Table 4), and perhapsall, lectins are made up of subunits andundergo association-dissociation reac-tions. Moreover, the subunits them-selves may be comprised of polypeptidefragments which are complementaryand which assemble to form the intactsubunit, as has been shown in con-canavalin A (55).The presence of several very similar

lectins has been observed in a numberof seed extracts (40-42, 47, 56, 57),as well as in the snail (44). Thesemultiple molecular forms of aggluti-niins, or isolectins, differ in their electro-phoretic mobilities. They may be theproduct of closely related genes, orthey may be formed prior to or duringisolation, as a result of side chainmodifications, such as hydrolysis of theamide group of glutamine or aspara-gine in the protein. In the isolectinsthat are glycoproteins, the differencesmay reside in the carbohydrate sidechains. The two glycoprotein lectinsisolated from the seeds of Ulex euro-peus (58, 59) cannot be consideredas isolectins, however, because they dif-fer markedly in their composition andhave distinct sugar specificities (L-fu-cose and di-N-acetylchitobiose, respec-tively) although, for reasons which arenot clear, they both exhibit the sameblood type specificity [H(O)]. Anotherexceptional example of different lectinsbeing isolated from the same sourcehas been reported by Kalb (60). Heisolated from Lotus tetragonolobusthree distinct lectins which have incommon the capacity for binding L-fu-cose, but which differ in molecularweight, number of binding sites andvalues of binding constants for L-fu-cose, and in chemical composition.

For the agglutination of cells, or forthe formation of precipitates with com-plementary macromolecules, a lectin

SCIENCE, VOL. 177-Jklij,- A

Table 3. Agglutination titers* of erythrocytes and murine tumor cells by crude lectins [adapted from Tomita et al. (33)].Rat Mouse Human erythrocyte

Orgnof lectin Ehrlich Leu- SugarOrigin Yoshida Eryth- ascites kIemia Eryth- A, B O specificitysarcoma rocyte tumor (L 1210) rocyte

Phaseolus limensis 0 0 0 0 0 128 0 0 NoneSophora japonica 1 1 0 0 0 32 16 16 NoneUlex europeus 0 0 0 0 0 16 4 16 L-Fuc,(D-GlcNAc)4ALaburnum alpinum 0 0 0 0 0 0 0 4 NoneLens culinaris 128 16 64 32 16 4 4 8 D-Man, D>GlcVicia faba 128 32 32 16 4 4 1 4 D-Man, D-GlcNPisum sativum 256 32 16 16 32 4 4 2 D-ManWistaria floribunda 64 1 32 32 1 16 S 4 D-GaINSesanum indicum 32 2 8 8 4 32 16 16 D-GlcRicinus communis 32 256 256 256 32 1024 1024 1024 D-GalMomordia charantia L. 8 64 32 32 4 256 512 512 D-Gal, n-GaINTriticum vulgaris 128 64 64 64 4 16 8 8 (D-GIcNAc),Solanum luberosum 0 8 32 16 16 64 128 256 (D-GlcNAc),Phaseolus vulgaris 1024 256 512 256 256 256 256 256 None* Highest dilution of crude lectin preparation (16 mg/ml) causing detectable agglutination after 1 hour at room temperature. t Saccharides testedfor inhibitory action: D-glucose (D-Glc), o-galactose (D-Gal), D-mannose (o-Man), o-glucosamine (D-GlcN), D-galactosamine (D-GaIN), N-acetyl-D-galactosa-mine (D-GaINAc), N-acetyl-D-glucosamine (D-GlcNAc), di-N-acetylchitobiose [(D-GlcNAcJ,], N-acetyl-D-mannosamine (o-ManNAc), L-fucose (L-Fuc), sialicacid (up to a concentration of 2 mg/ml). * Data from Matsumoto and Osawa (58, 59).

Table 4. Chemical and biological properties of highly purified lectins. Abbreviations: Xyl, xylose; Ara, arabinose; other abbreviations as inTable 3. Roman numerals indicate different lectins isolated from the same plant. For further information see text.

Carbohydrates Mito- Specificity No. MetalSoreMolecular Sub- Pr ao genic Hu- of re- Rfrneweights units cent- constit- activ- man Sugar sig quire-

ae unsity blood mentage uents typesieHigher plants: Leguminosae

Canavalia enisformis 55,000 2 0 + None ot5D-Man 2 + (7, 29, 51, 33, 90)(jack bean)

Dolichos biflorus 140,000 3.8 GlcN, Man - A a-D GalNAc (43,105)(horse gram)

Glycine max 110,000 5 D-GlcNAc, D-Man - None D-GalNAc 2 + (8,9, 30, 62)(soybean)

Lens culinaris* 42,000- 2 2 GlcN, Glc + None aD-Man 2 + (40,41,52,106)(common lentil) 69,000

Lotus tetragonolobus 120,000 9.4 GlcN, Hexose H(O) a-L-Fuc 4 (13, 14,37,60)II 58,000 4.8 GlcN, Hexose H(O) a-L-FUc 2 (13, 14,37, 60)III 117,000 9.2 GlcN, Hexose H(O) a-L-FuC 4 (13,14,37, 60)

Phaseolus lunatust I 269,000 8 4 GlcN, Man, Fuc A + (27, 45, 49)(lima bean) II 138,000 4 4 GlcN, Man, Fuc A +

Phaseolus vulgaris 128,000 5.7 Hexosamine (107)(black kidney bean) Man, Xyl

Phaseolus vulgarist I 138,000 8 8.9 GlcN, Man + (108)(red kidney bean) II 98,000- 4 4.1 GlcN, Man + o-GaINAc (35)

138,000Phaseolus vulgaris 10.4 GlcN, Man, Glc, (109)

(yellow wax bean) AraPisum sativum 53,000 (0.3) (Glc) None D-Man + (42,52)

(garden pea)Robinia pseudoacacia 90,000 10.7 GlcN, Man, Fuc, (110)

(black locust) XylUlex europeus I 170,000 5.2 GlcN, Man H(O) L-Fuc (58,59)

(gorse) II 21.7 GlcN, Man, Gal, H(O) (D-GlcNAc), (58,59)Ara

Other high plantsRicinus communis 98,000 None D-Gal (6,101)

(castor bean)Solanum tuberosum (20,000) 5.2 Ara (D-GlcNAc) (6, 111)

(potato)Triticum vulgaris 26,000 4.5 None (D-GlcNAc), (6, 10-12, 93)

(wheat)Lower plants

Agaricus campestris 64,000 4 4 + (48)(meadow mushroom)

InvertebratesHelix pomatia 100,000 7.3 Gal, Man A a-D-GalNAc 6 (44)-(vineyard snail) (hexcosaniine)

Lbnulus polyphemus 400,000 18 Sialic acid + (6,50)(horseshoe cmb)

* Also known as Lens esculenta. t Also known as Phaseolus limensis. t Two (or more) lectins with distinct activities appear to be present inthe red kidney bean, an erythroagglutinin (I) and a leukoagglutinin (II).

1S SEPTEMBER 1972 953

molecule must possess at least twobinding sites as is found in immuno-globulins. Concanavalin A has onebinding site for methyl a-D-manno-pyranoside or methyl a-D-glucopyrano-side per subunit of molecular weightof about 32,000 (51), but this subunitdoes not exist as such; concanavalinA usually occurs as a dimer (molecularweight 55,000) with two binding sites(53, 61). Two binding sites for N-acetyl-D-galactosamine have been foundin soybean agglutinin (62), two formethyl a-D-glucopyranoside in the lentillectin (63), two or four for L-fucosein the lectin from Lotus tetragonolobus(37), and six sites for N-acetyl-D-galactosamine in the lectin from thevineyard snail (44). The associationconstants for the binding of the abovesaccharides to the corresponding lectinsare usually in the range of Ka = 103 to104 (with the exception of the lentillectin, which binds a-D-mannose withKa=2.3 x 102). These values arelower than the values of the associationconstants found for the interaction ofoligosaccharides with antibodies to thecorresponding carbohydrates (64). Inall cases, the combining sites are ho-mogenous and identical with respect totheir distribution in different moleculesof the same lectin. In this propertylectins differ from immune antibodiesin that the combining sites of the latterare neither identical with one anothernor are they homogenous.

Concanavalin A: PhysiochemicalPropertiesThe relative abundance of con-

canavalin A in jack bean meal (2.5 to3 percent by weight) (7), the ease ofits preparation, and the availability ofa great variety of low and high molec-ular weight saccharides with which itcan interact, have led to a large num-ber of studies on this lectin. The paceof research on concanavalin A hasbeen markedly accelerated followingthe demonstration that cells trans-formed by DNA tumor viruses or car-cinogens are agglutinated by this lectinmore readily than are normal cells(17). As a result, we have considerablymore knowledge of the physical, chemi-cal, and biological properties of con-canavalin A than of any other lectin.

Although concanavalin A is biologi-cally homogenous, as judged by thefact that 95 to 98 percent of its weightcan be precipitated by a dextran (47),it is not a homogenous molecular spe-

954

cies in the pH range where it bindssaccharides. Physical heterogeneity ofhighly purified concanavalin A has, forexample, been observed upon examina-tion by free boundary electrophoresis(47) or isoelectric focusing (57). Veryrecently, it was found that crystallineconcanavalin A can be separated instrongly dissociating solvents into sev-eral molecular species (55). Thesehave been isolated and identified as anintact subunit or "protomer" of con-canavalin A (molecular weight 27,000)and two main fragments which accountfor most of the intact protomer. It wasconcluded that the intact chain is theprimary gene product and that the twosmaller chains (NH2-terminal, molecu-lar weight 12,500; and COOH-terminal,molecular weight 14,000) were derivedfrom it by chemical or enzymatic hy-drolysis. The chains associate stronglyand the results suggest that the con-canavalin A protomer may consist of asingle peptide chain of 27,000 molecu-lar weight or a mixture (ratio 1: 1)of complementary fragments. In solu-tion below pH 6, concanavalin A formsa molecular species made up of twoprotomers, whereas at pH 6.7 andabove, it may consist of four protomers.These association-dissociation reactionswould account for the different valuesfor the molecular weight of concana-valin A (55,000 and 100,000) whichhave been reported. Other intermediatevalues probably reflect mixtures of thetwo-protomer and four-protomer units(65).These conclusions are in agreement

with recent x-ray crystallographic stud-ies of concanavalin A (66) as well asearlier studies of the binding of sac-charides and metal ions to the lectin(51). The crystallographic studies, car-ried out to 4-A resolution, clearly showthe molecular shape and the packingof the concanavalin A molecules. Theasymmetric unit (or protomer) has amolecular weight of 27,000 and formsan elliptical dome with a base of ap-proximately 46 by 26 A and a heightof 42 A. The protomers are paired toform dimers which are in turn pairedto form tetramers of roughly tetra-hedral shape. Each protomer has a de-pression located on the surface whichcould be the site of saccharide binding.The molecular structure of the pro-tomers is not clear, since the aminoacid sequence of concanavalin A is notyet known. A symmetrical structureand even number of binding sites maybe a common feature of many lectins.

Concanavalin A has specific binding

sites for transition metal ions and cal-cium ions, in addition to the saccharidebinding site. Each set of sites is asso-ciated with a protein unit that has amolecular weight of 32,000 and thereare interactions between the three sites:calcium ions can be bound only afterthe transition metal site is occupiedand occupancy of both sites is requiredbefore the binding of a saccharide suchas methyl a-D-glucopyranoside can oc-cur (51). The circular dichroic spec-trum of concanavalin A shows changesin the near ultraviolet region uponspecific binding of methyl a-D-manno-pyranoside, which suggests that sugarbinding at physiological pH is accom-panied by a conformational change inthe protein (67).The results of studies conducted by

Agrawal et al. (68) and Hassing et al.(69) have led to the conclusion thatfree carboxyl groups, but not freeamino or phenolic hydroxyl groups areessential for the precipitation of dex-trans by concanavalin A. It was sug-gested that carboxyl groups may bepresent in or near the saccharide-bind-ing sites of concanavalin A (68, 69);however, chemical modifications mayinfluence the saccharide-binding activ-ity by affecting the metal-binding sites.There is evidence of a hydrophobicregion on the protein in close proximityto the carbohydrate-binding region (70).Slight changes in the ultraviolet absorp-tion spectra of the tryptophyl andtyrosyl residues could be observed uponthe addition of saccharides to con-canavalin A, suggesting that these resi-dues may be perturbed although notnecessarily situated within the sac-charide-binding site (71).

Concanavalin A: Specificity and Uses

The specificity requirements of theinteractions between concanavalin Aand saccharides have been thoroughlyinvestigated by Goldstein and his co-workers (16, 29). Concanavalin A canform a precipitate with numerous poly-saccharides that contain multiple termi-nal nonreducing g-D-ganosyl (orits 2-acetamido de r, a-D-man-nopyranosyl, /-D-fructofuranosyl, or a-D-arabinofuranosyl residues. These in-clude glucans such as glycogens,amylopectins, and dextrans; yeast man-nans and phosphomannans as well assynthetic D-mannans; D-fructans suchas bacterial and plant levans; and amycobacterial arabinogalactan. It willalso form a percipitate with many gly-

SCIENCE, VOL. 177

coproteins, including soybean agglu-tinin which has been shown to possessterminal a-D-mannopyranosyl residues(9, 72), and with certain teichoic acids(73).

Further insight into the stereo-chemical specificity of concanavalin Ahas been obtained by examining theextent to which a large number ofmonosaccharides, oligosaccharides, andmodified sugars inhibit the precipita-tion reaction (the Landsteiner hapteninhibition technique). The carbohy-drate-combining site of the concanava-lin A molecule appeared to be directedprimarily toward unmodified hydroxylgroups at the C-3, C-4, and C-6 posi-tions of the six-membered a-D-manno-pyranoside or a-D-glucopyranoside ring(positions 3, 4, 5, 6 of the D-arabinoconfiguration) (Fig. 2). Any modifica-tion at these positions completelyabolished the inhibitory action of thesaccharide. On the other hand, con-siderable modification of the group atC-2, as long as it was equatorial, af-fected the inhibitory activity to a smallextent only. Thus 2-deoxy-D-glucose,2-0-methyl-D-glucose and N-acetyl-D-glucosamine exhibited almost the sameinhibitory activity as D-glucose. It wasalso concluded that the oxygen atom ofthe C-2 position of D-mannose servesas a polar binding locus for the inter-action of the saccharide with the pro-tein (29). It appears that D-mannose,D-glucose, and their derivatives bind toconcanavalin A in the C-I chair con-formation and that the protein pos-sesses a specific locus capable of inter-action with the anomeric oxygen atomof the a-linked glycosides of thesesugars (Fig. 2). A /3-anomeric oxygenatom interferes with binding, since/3-D-mannopyranosides or /3-D-glucopy-ranosides are very poor inhibitors. In-deed, of the large number of glucansand mannans tested, only those con-taining a-glycosidic linkages interactedwith concanavalin A, whereas the /B-linked polysaccharides failed to inter-act. The inhibition studies indicate thatpolar interactions, such as H bonds,and charge-dipole (but not charge-charge) interactions are the predomi-nant stabilizing forces in the concana-valin A-polysaccharide complexes andthat the formation of these complexesinvolves chiefly the nonreducing chainends of the reacting polysaccharides.

In addition to glycopyranosides, thebinding site of concanavalin A can alsoaccommodate certain five-memberedrings, such as /3-D-fructofuranosidesand D-arabinofuranosides (74); this15 SEPTEMBER 1972

"I 4 CHpO0 )H o

,a )a

.) Me

(b)HO CH,OH Q

HO ,

OMe

HO

Fig. 2. Methyl a-D-mannopyranoside (a)and methyl a-D-gluCopyranoside (b), thebest monosaccharide inhibitors of the pre-cipitation of polysaccharides by concana-valin A; and methyl [3-D-mannopyranoside(c), which is noninhibitory (29). Thesugars are drawn in their most stable C-1chair conformation, and in (a) the groupsthat interact with the protein are circled.

interaction is considerably weaker thanthe concanavalin A-glycopyranosidereaction. The binding of furanoidsugars to concanavalin A has been ex-plained on the basis of common con-figurational features with sugars pos-sessing the pyranoid ring. Binding offuranose ring structures to pyranose-specific sites has also been observedwith the lectin of Lotus tetragonolobus(75), and has been demonstrated byx-ray crystallography for another car-bohydrate-binding protein, the enzymelysozyme of hen egg white (76).

Although it was originally postulatedthat the specificity of concanavalin Ais directed exclusively toward a singlesaccharide residue, there is now evi-dence that this lectin may possess amore extended binding site. Thus, ithas been shown (77) that the disac-charide D-Man-a(l -12)-D-Man andthe trisaccharide D-Man-a(1- 2)-D-Man-a( - 2)-D-Man are much strong-er inhibitors of the concanavalin A-dextran interaction than is methyla-D-mannopyranoside, the disaccharideby a factor of 5 and the trisaccharideby a factor of 20. These findings mightalso be explained by the possibility ofthere being two or more concanavalinA molecules associated with extended

sequences of a( I 22)-D-mannopy-ranosyl residues.The results of many other studies of

concanavalin A are in perfect agree-ment with its specificity requirements.The order of reactivity of the varioussugars used in inhibition studies isthe same, regardless of the polysac-charide used for combining with con-canavalin A, and the method of assayof the concanavalin A-polysaccharideinteraction (turbidimetry, precipitation,or agar gel diffusion) (29). Onlysugars that are good inhibitors displaceconcanavalin A from Sephadex gels towhich it has been adsorbed and therelative efficiencies of the sugars paral-lel their inhibitory activity (38). More-over, direct measurements by equilibri-uni dialysis of the binding to concana-valin A of methyl a-D-mannopyranoside-and methyl a-D-glucopyranoside giveconstants [Ka = 1.4 x 104 liter/moleand Ka = 0.3 x 104 liter/ mole, respec-tively (61)] which are inversely pro-portional to the capacity of these sugarsto inhibit precipitation [0.34 [Lmole/mland 1 .3 ,umole/ ml, respectively, wererequired for 50 percent inhibition ofprecipitation in the concanavalin A-levan B-512 system (29)].

Knowledge of the specificity require-ments of concanavalin A has resultedin its being used as a reagent in ana-lytical and preparative biochemistry.We will mention only a few examplesof the increasing uses of this lectin. Adextran from Streptococcus bovis, orig-inally believed to be a linear polymerof a(l -- 6) linked D-glucose residues,gave a precipitate with concanavalin A.As a result of this finding, furtherchemical studies were undertaken andthe dextran was shown to be abranched polymer with a(1 -- 3) anda(I -* 6) linkages (78). An a-D-Con-figuration was assigned to mannosyl-glycerol of Micrococcuis Iysodeikticus,because this glycoside inhibited theconcanavalin A-dextran reaction to thesame extent as methyl-a-D-mannopy-ranoside; subsequent investigationshave supported this assignment (79).With the aid of concanavalin A, a newantigenic determinant, unrelated toABO blood group specificity and hav-ing a terminal nonreducing N-acetyl-a-D-glucosaminyl residue, was detectedand identified in hog and some humangastric mucin blood group A + Hsubstances (32). Precipitation by con-canavalin A of the antihemophilic fac-tor (factor VIII) of blood proved tobe a useful step in purification of thisfactor (80). The structure of the car-

955

bohydrate moiety of rabbit gammaglobulin has been investigated withthe use of concanavalin A (81). It hasalso been suggested that concanavalinA and appropriate carbohydrates beutilized in the classroom for demon-strating the principles of the antigen-antibody reaction, eliminating the needof antigens and antiserums (82).The recently prepared, insoluble,

biologically active derivatives of con-canavalin A (83) greatly extend therange of application of this lectin. In-soluble concanavalin A can be used forthe isolation of immunoglobulins andblood group substances (83), and forthe purification of glycoprotein en-zymes such as glucose oxidase (84). Ina similar manner, partial purificationof concanavalin A receptors from thelymphocyte plasma membrane of thepig has been achieved (85).

In addition to the great increase inour knowledge of the chemical proper-ties of concanavalin A, a wealth of in-formation is now available concerningits biological properties. The toxicity toanimals of concanavalin A has beenrecognized for many years (7). Re-cently, concanavalin A has been shownto agglutinate embryonic cells (86),spermatozoa (87), and certain viruses(88), to stimulate the migration oftumor cells (89), to exhibit mitogenicactivity (90), and to confer cytolyticactivity on lymphal node cells inter-acting with fibroblasts (91). Whenevertested, these activities were inhibited bythe same sugars that inhibit the precipi-tation of dextrans by concanavalin A.

In spite of its high specificity, con-canavalin A may precipitate polyelec-trolytes and polysaccharides which aredevoid of terminal mannopyranoside(or related) residues (92). It is, how-ever, easy to distinguish between thespecific and nonspecific reactions ofconcanavalin A, because the latter arenot inhibited by saccharides.

Interaction with Malignant Cells

In 1963, Aub and his co-workers(10) reported that wheat germ lipasepreparations agglutinated mouse tumorcells more readily than cells from nor-mal tissues. Subsequently, they showedthat the lipase preparations containedas an impurity a lectin that was re-sponsible for this agglutinating activity.Wheat germ agglutinin was purified byBurger and Goldberg (11) and Burger(12). These workers concluded thatthe agglutinin was a glycoprotein with

956

a molecular weight of 26,000. It ag-glutinated several lines of virally trans-formed cells at concentrations that didnot cause agglutination of the untrans-formed parent cells. Burger and Gold-berg therefore suggested that the sur-face of the transformed cells containeda component that was not present onthe surface of the normal cells. It wasproposed that this component is N-acetyl-D-glucosamine or a closely re-lated derivative, because studies ofhapten inhibition showed that N-acetyl-D-glucosamine and its disaccharide,di-NV-acetyl-chitobiose, specifically in-hibit the agglutination reaction (11,12). The addition of inhibitory sugarsto the clumped cells resulted in dis-sociation of the aggregates, showingthat the agglutination was readily re-versible. Ovomucoid, a glycoproteincontaining a high proportion of N-acetyl-D-glucosamine, inhibited the ag-glutination at very low concentrations.Binding to insoluble ovomucoid andelution with an excess of N-acetyl-o-glucosamine was used in a modifiedprocedure for the purification of wheatgerm agglutinin (11, 12). The specific-ity of wheat germ agglutinin for N-acetyl-D-glucosamine was confirmed instudies of the agglutination of erythro-cytes in which it was also shown thatthis lectin is nonspecific for humanblood types (93).

Although normal somatic cells arenot agglutinated by low concentrationsof wheat germ agglutinin, they can beagglutinated after brief treatment withproteolytic enzymes. The agglutinationof normal cells treated with proteasesis similar in many respects to that ofthe untreated virally transformed cells,and the chemical nature of the recep-tors appears to be identical in bothtypes of cell (11, 12).Wheat germ agglutinin is not the

only lectin that preferentially aggluti-nates malignant cells, or normal cellsafter mild proteolysis. Inbar and Sachshave shown (17) that concanavalin Aat a concentration of 250 ,ug/ml agglu-tinates leukemic cells and tissue-culturecells that have been transformed bythe polyoma virus, simian virus 40,chemical carcinogens, and irradiationwith x-rays. Concanavalin A does notagglutinate normal cells under the sameconditions. The agglutination is re-versed by methyl a-D-glucopyranoside,but not by saccharides that do not bindto concanavalin A. Removal of bivalentmetal ions from concanavalin Aabolishes its cell-agglutinating capacity,in accord with the metal requirements

of concanavalin A for the binding ofcarbohydrates. After treatment of nor-mal cells with trypsin they are aggluti-nated by concanavalin A. Soybean ag-glutinin, which is specific for N-acetyl-D-galactosarnine (both a- and s-linked)(30) also agglutinates mouse, rat, andhuman cell lines transformed by viralor chemical carcinogens (18). In con-trast, transformed hamster cells are notagglutinated, whereas normal hamstercells can be agglutinated after mildproteolytic treatment, as is the casewith the other normal cell lines. Theagglutination is, in all instances, specif-ically inhibited by N-acetyl-D-galactos-amine, suggesting that this saccharide,or a closely related one, is the terminalnonreducing moiety of the receptor forsoybean agglutinin on the surfacemembrane of the agglutinable cells.The receptor site of wheat germ ag-

glutinin was isolated from membranes*of transformed cells (94). Concentra-tions of the receptor glycoprotein,which fully inhibited the agglutinationof cells by wheat germ agglutinin, hadno effect on the agglutination withother lectins, such as soybean agglu-tinin or concanavalin A. These findingssuggest that the receptor sites presenton the surface of the transformed cellsfor the different lectins are both chemi-cally and topographically distinct. Onthe other hand, the glycoprotein recep-tor for the kidney bean lectin, isolatedfrom human erythrocytes, is a potentinhibitor of the agglutination of humanerythrocytes by the lentil lectin, al-though the two lectins are known topossess different carbohydrate specilici-ties (95). It is therefore possible thatthe lentil lectin and the kidney beanlectin bind to different portions of thesame oligosaccharide on the humanerythrocyte surface.

Although it was originally suggestedthat the difference between cells whichare agglutinated and those which arenot is in the number of saccharide re-ceptor sites, it now appears that bothtypes of cells may contain approxi-mately the same number of receptors.This conclusion is based on studies withradioactively labeled lectins f3H--wd1251-labeled concanavalin A; 125I-labeledwheat germ agglutinin (96); and 1251-labeled soybean agglutinin (97)], whichshow that there is no direct correlationbetween the amount of lectin'lboundand the agglutinability of the cbls.Often normal and transformed cells, aswell as cells which have undergone pro-teolysis, bind more or less the sameamount of lectin. This surprising find-

SCIENCE, VOL. 177

ing has been confirmed in studies ofthe interaction of soybean agglutininwith untreated and trypsinized erythro-cytes (36).

Very recently it has been suggestedthat agglutination by lectins dependson the relative distribution of sugarreceptors on the surface of the cell; innormal cells the receptors are dis-persed, whereas after proteolysis ormalignant transformation a redistribu-tion of sites occurs which results intheir clustering, an arrangement favor-able to cell agglutination. This conclu-sion is based on results obtained bymeans of a new electron-microscopicmethod developed for investigating thespecific localization of antigens andsaccharides on cell membrane surfaces;ferritin-conjugated concanavalin A wasused in these studies (98). Rearrange-ment of exposed sites upon malignanttransformation has also been suggestedby Ben-Bassat et al. on the basis ofbinding studies (99). However, thepossibility cannot be excluded that onlya small percentage of the lectin mole-cules bound to the cells is involved inagglutination, whereas the bulk isbound to sites that contain the appro-priate saccharide receptors, but bindingto these receptors does not lead to ag-glutination. If this were true, this"nonproductive" binding could maskany small but real difference in thenumber of "productive" binding sites.That the medium in which the cells aregrown can affect the binding of con-canavalin A has also been observed(99).The significance of the changes in

the surface membranes of cells to theunderstanding of malignant transforma-tion are not clear; there is even somedoubt whether such changes are a spe-cific characteristic of all malignant cells(100). It is clear, however, that lectins,both native and modified, provide anew and useful tool for the study of thechemical architecture of cell surfacesand the changes which they undergoupon malignancy.

Conclusions

Although much progress has beenmade in the study of lectins, a largenumber of questions regarding theirproperties, uses, and functions remainunanswered. One of the main problemsencountered in studies of proteins, es-pecially those proteins that exhibitmultiple biological activities, is)that ofdeciding whether they are huarogenous.15 SEPTEMBER 1972 1

Most, and perhaps all, of the purifiedlectins (Table 4), have not been sub-jected to extensive testing for homo-geneity, and it is therefore difficult toconclude whether their different bio-logical activities reside in the samemolecule. This is true even for the mostthoroughly investigated lectin, concana-valin A, which is hemagglutinating,mitogenic, toxic, and sugar-binding.With some lectins, separation of differ-ent activities has been reported. Thus,the erythroagglutinating activity of thelectin of the red kidney bean has beenseparated from its leukoagglutinatingactivity (35). Purified, crystalline ricinhas been separated into different frac-tions, one of which is toxic and devoidof hemagglutinating activity, whileanother that contains the hemaggluti-nating protein exhibits low toxicity(101). In view of the observationsmade with "incomplete" lectins (4) itwould be of interest to know whetherthe addition of different substancesmight endow the nonhemagglutinatingprotein with agglutinating activity.

In addition to the unresolved prob-lems of chemical and biological homo-geiieity of lectins, there is also the ques-tion of the mechanism or mechanismsby which their biological activities areexpressed. Are all the activities the re-sult, direct or indirect, of binding tospecific saccharide sites on cell sur-faces? Are the receptors for the differ-ent activities identical, chemically andtopologically? Even the relatively sim-ple agglutination reaction is not yet wellunderstood.The saccharide specificity of many

lectins is well established, but for somelectins (Tables 3 and 4) no saccharideshave been found that inhibit their ag-glutinating activity. That the appropri-ate sugars have not been tested is aremote possibility, since it implies thatanimal cell surfaces contain unknownsugars, which is unlikely. Another pos-sibility is that these lectins possess ex-tended sites which require complexsaccharide structures for efficient inter-action. Indeed, in spite of the generalagreement that the saccharide bindingsitesof lectins are smaldl onding

sizRthereatrons o lec-he sits may be more ex-77d( btr-ings us toe

question, still far from being answered,of whether there are structural featurescommon to the binding regions of alllectins. In this connection it is alsortinentte covalently

linkesl present in

most lectins, contribute in any way totheir biological activities. Since thereare lectins that are not glycoproteins(Table 4) it would appear that thesugar moiety does not play an essentialrole in these activities.

Although we have pointed out thatlectins have a role as analytical tools inthe study of carbohydrates, it shouldbe kept in mind that the use of thesereagents is not free of pitfalls, becausethey may interact with apparently un-related saccharides. Thus, the lectins ofeel serum and of the seeds of Lotustetragonolobus are specific in their ag-glutination of type 0 erythrocytes butare inhibited not only by L-fucose or its0-methyl ethers, but also by some 0-methyl ethers of D-fucose (102).The role of lectins in nature-

whether in plants or in other organisms-remains a mystery. It has been sug-gested that they are antibodies intendedto counteract soil bacteria (1, 2, 5);that they serve to protect plants againstfungal attack by inhibiting fungal poly-saccharases (103); that because oftheir affinity for saccharides they areinvolved in sugar transport and storage;or that they serve for the attachment ofglycoprotein enzymes in organizedmultienzyme systems. In view of themitogenic properties of lectins it ispossible that their function is to con-trol cell division and germination inplants. There is, however, no evidencefor or against any of these hypotheses.It is therefore not surprising that it hasbeen proposed that the biological prop-erties of lectins, as observed in thelaboratory, have no relation to theirfunction in nature.The ready availability, great diver-

sity, and the ease of isolation of theseagglutinins and the many interesting-and unusual properties they exhibit,make them a most attractive subject forresearch. They may eventually providea series of most useful probes, com-parable to the carbohydrate-specificantibodies to the pneumococcal bac-teria (104), for the recognition of indi-vidual sugars and specific linkages inpolysaccharides, glycoproteins, andglycolipids, and for the mapping of thearchitecture of cell surfaces. Studies onthe primary, secondary, and tertiarystructures and of the physiochemicalproperties of lectins will undoubtedlylead to the understanding of the molec-ular basis of their sugar specificity andtheir mechanisms of action on cells,whether normal or malignant. Eventu-ally their role in nature may also beunraveled.

957

References and Notes

1. G. W. G. Bird, Brit. Med. J. 15, 165 (1959).2. W. C. Boyd, Vox Sang. 8, 1 (1963); Ann.

N.Y. Acad. Sci. 169, 168 (1970).3. M. Kriipe, Blutgruppenspezifische pflanzliche

Eiwesskorper (Phytagglutinine) (Enke, Stutt-gart, 1956); J. Tobiska, Die Phytohamag-glutinine (Akademie Verlag, Berlin, 1964).

4. 0. Makela, Ann. Med. Exp. Biol. Fenn. 35,suppl. 11 (1957).

5. G. C. Toms and A. Western, in Chemotax-onoiny of the Leguminosae, J. Harborne, D.Boulter, B. L. Turner, Eds. (Academic Press,New York, 1971), chap. 10, pp. 367-463.

6. G. 1. Pardoe and G. Uhlenbruck, J. Med.Lab. Technol. 27, 249 (1970).

7. J. B. Sumner and S. F. Howell, J. Bacteriol.32, 227 (1936).

8. 1. E. Liener and M. J. Pallansch, J. Biol.Chem. 197, 29 (1952); S. Wada, M. J. Pal-lansch, 1. E. Liener, ibid. 233, 395 (1958).

9. H. Lis, N. Sharon, E. Katchalski, ibid. 241,684 (1966); Biochim. Biophys. Acta 192,364 (1969).

10. J. C. Aub, C. Tieslau, A. Lankester, Proc.Nat. Acad. Sci. U.S.A. 50, 613 (1963); J. C.Aub, B. H. Sanford, M. N. Cote, ibid. 54,396 (1965).

11. M. M. Burger and A. R. Goldberg, ibid. 57,359 (1967).

12. M. M. Burger, ibid. 62, 994 (1969).13. W. M. Watkins and W. T. J. Morgan, Na-

ture 169, 825 (1952).14. W. T. J. Morgan and W. M. Watkins, Brit.

J. Exp. Pathol. 34, 94 (1953); W. M. Wat-kins, in Glycoproteins, A. Gottschalk, Ed.(BBA Library, Elsevier, Amsterdam, 1966),vol. 5, pp. 475-483.

15. P. Nowell, Cancer Res. 20, 462 (1960); J. H.Robbins, Science 146, 1648 (1964); Ch. K.Naspitz aind M. Richter, Progr. AUergy12, 1 (1968); N. R. Ling, Lymphocyte Stimu-lation (North-Holland, Amsterdam, 1968).

16. I. J. Goldstein, C. E. Hollerman, E. E. Smith,Biochemistry 4, 876 (1965); E. E. Smithand I. J. Goldstein, Arch. Biochem. Biophys.121, 88 (1967).

17. M. Inbar and L. Sachs, Nature 223, 710(1969); Proc. Nat. Acad. Sci. U.S.A. 63,1418 (1969).

18. B. A. Sela, H. Lis, N. Sharon, L. Sachs,J. Menmbrane Biol. 3, 267 (1970).

19. J. Shoham, M. Inbar, L. Sachs, Nature 227,1244 (1970); M. M. Burger and K. D. Noo-nan, ibid. 228, 512 (1970).

20. W. T. Shier, Proc. Nat. Acad. Sci. U.S.A.68, 2078 (1971).

21. 1. E. Liener, Amer. J. Clin. Nutr. 11, 281(1962); Econ. Bot. 18, 27 (1964); W. G.Jaffe, in Toxic Constituents of Plant Food-stuffs, I. E. Liener, Ed. (Academic Press,New York, 1969), pp. 69-101.

22. N. K. Allen and L. Brillantine, J. Itnmunol.102, 1295 (1965).

23. Further details of the early history of lectinsand their role in the development of immu-nology can be found, for example, in (4)and (5).

24. K. Landsteiner and H. Raubitschek, Zbl.Bakt. 45, 660 (1908).

25. K. Landsteiner, in The Specificity of Sero-logical Reactions, K. Landsteiner, Ed. (Har-vard Univ. Press, Cambridge, Mass., 1945),pp. 4-5.

26. K. 0. Renkonen, Ann. Med. Exp. Biol. Fenn.26, 66 (1948); ibid. 28, 45 (1950); ibid. 38,26 (1960).

27. W. C. Boyd and R. M. Reguera, J. Immu-nol. 62, 333 (1949).

28. E. A. Kabat, M. Heidelberger, A. E. Bezer,J. Biol. Chem. 168, 629 (1947); M. Kunitzand M. R. McDonald, J. Gen. Physiol. 32,25 (1948).

29. I. J. Goldstein and L. L. So, Arch. Biochem.Biophys. 111, 407 (1965); L. L. So and I. J.Goldstein, J. Biol. Chem. 242, 1617 (1967);R. D. Poretz and I. J. Goldstein, Biochem-istry 9, 2890 (1970).

30. H. Lis, B. A. Sela, L. Sachs, N. Sharon,Biochim. Biophys. Acta 211, 582 (1970).

31. R. GC. Drysdale, P. R. Herrick, D. Franks,Vox Sang. 15, 194 (1968).

32. K. 0. Lloyd, E. A. Kabat, S. Beychok, J.Immunol. 102, 1354 (1969); C. Moreno andE. A. Kabat, ibid., p. 1363; M. E. Etzler,B. Anderson, S. Beychok, F. Gruezo, K. 0.Lloyd, N. G. Richardson, E. A. Kabat,Arch. Biochenm. Biophys. 141, 588 (1970).

33. M. Tomita, T. Osawa, Y. Sakurai, T. Ukita,Int. J. Cancer 6, 283 (1970).958

34. M. Tunis, J. Inmmunol. 92, 864 (1964).35. T. H. Weber, Scand. J. Clin. Lab. Invest.,

suppl. 111 (1969); L. W. Allen, R. H. Sven-son, S. Yachnin, Proc. Nat. Acad. Sci.U.S.A. 63, 334 (1969); D. Allan and M. J.Crumpton, Biochemn. Biophys. Res. Commzun.44, 1143 (1971).

36. J. A. Gordon, N. Sharon, H. Lis, Biochim.Biophys. Acta 264, 387 (1972).

37. J. Yariv, A. J. Kalb, E. Katchalski, Nature215, 890 (1967).

38. B. B. L. Agrawal and I. J. Goldstein, Bio-chim. Biophys Acta 133, 376 (1967); ibid.147, 262 (1967).

39. M. 0. J. Olson and I. E. Liener, Biochemis-try 6, 105 (1967).

40. M. Ticha, G. Entlicher, J. V. Kostir, J. Ko-courek, Biochim. Biophys. Acta 221, 282(1970).

41. I. K. Howard, H. J. Sage, M. D. Stein,N. M. Young, M. A. Leon, D. F. Dyckes,J. Biol. Chem. 246, 1590 (1971); N. M.Young, M. A. Leon, T. Takahashi, I. K.Howard, H. J. Sage, ibid. p. 1596.

42. G. Entlicher, J. V. Kottii, J. Kocourek,Biochim. Biophys. Acta 221, 272 (1970).

43. M. E. Etzler and E. A. Kabat, Biochemnistry9, 869 (1970).

44. S. Hammarstrom and E. A. Kabat, ibid. 8,2696 (1969); ibid. 10, 1684 (1971).

45. W. Galbraith and I. J. Goldstein, Fed. Eur.Biochem. Soc. Lett. 9, 197 (1970).

46. L. Sundberg, J. Vorath, K. Aspberg, Bio-chim. Biophys. Acta 221, 394 (1970).

47. B. B. Agrawal and I. J. Goldstein, Arch.Biocheni. Biophys. 124, 218 (1968).

48. H. J. Sage and S. L. Connett, J. Biol. Chemn.244, 4713 (1969).

49. N. R. Gould and S. L. Scheinberg, Arch.Biochem. Biophys. 137, 1 (1970); ibid. 141,607 (1970).

50. I. J. Marchalonis and J. M. Edelman, J. Mol.Biol. 32, 453 (1968).

51. J. Yariv, A. J. Kalb, A. Levitzki, Biochim.Biophys. Acta 165, 303 (1968); B. B. L.Agrawal and I. J. Goldstein, Can. J. Bio-chem. 46, 1147 (1968).

52. M. Paulova, G. Entlicher, M. Ticha, J. V.Kostif, J. Kocourek, Biochim. Biophys. Acda237, 513 (1971); M. Paulova, M. Ticha,G. Entlicher, J. V. Kotif, J. Kocourek,ibid. 252, 388 (1971).

53. A. J. Kalb and A. Lustig, ibid. 168, 366(1968).

54. J. B. Summer, N. Gralen, I.-B. Eriksson-Quensel, J. Biol. Chem. 125, 45 (1938).

55. J. L. Wang, B. A. Cunningham, G. M.Edelman, Proc. Nat. Acad. Sci. U.S.A. 68,1130 (1971); A. B. Edmundson, K. R. Ely,D. A. Sly, F. A. Westholm, D. A. Powers,I. E. Liener, Biochemistry 10, 3554 (1971).

56. H. Lis, C. Fridman, N. Sharon, E. Katchal-ski, Arch. Biochem. Biophys. 117, 301(1966); N. Catsimpoolas and E. W. Meyer,ibid. 132, 279 (1969).

57. G. Entllcher, J. V. Ko8tif, J. Kocourek,Biochim. Biophys. Acta 236, 795 (1971).

58. I. Matsumoto and T. Osawa, ibid. 194, 180(1969).

59.- , Arch. Biochem. Biophys. 140, 484(1970).

60. A. J. Kalb, Biochim. Biophys. Acta 168,532 (1968).

61. L. L. So and I. J. Goldstein ibid. 165,398 (1968).

62. H. Lis, J. A. Gordon, H. W. Siegelman,N. Sharon, unpublished results.

63. M. D. Stein, I. K. Howard, H. J. Sage,Arch. Biochem. Biophys. 146, 353 (1971).

64. F. Karush, Trants. N.Y. Acad. Sci. 20, 581(1958); C. Moreno and E. A. Kabat, J.Exp. Med. 129, 871 (1969); M. Katz andA. M. Pappenheimer, Jr., J. Immunol. 103,491 (1969).

65. W. D. McCubbin and C. M. Kay, Biochem.Biophys. Res. Commun. 44, 101 (1971).

66. J. Greer, H. W. Kaufman, A. J. Kalb, J.Mol. Biol. 48, 365 (1970); A. Jack, J.Weinzierl, A. J. Kalb, ibid. 58, 389 (1971);K. D. Hardman, M. K. Wood, M. Schiffer,A. B. Edmundson, C. F. Ainsworth, Proc.Nat. Acad. Sci. U.S.A. 68, 1393 (1971);F. A. Quiocho, G. N. Reeke, Jr., J. W.Becker, W. N. Lipscomb, G. M. Edelman,ibid., p. 1853.

67. R. Zand, B. B. L. Agrawal, I. J. Gold-stein, ibid., p. 2173; M. N. Pflumm, J. L.Wang, G. M. Edelman, J. Biol. Chem. 246,4369 (1971).

68. B. B. L. Agrawal, I. J. Goldstein, G. S.Hassing, L. L. So, Biochemistry 7, 4211(1968).

69. G. S. Hassing, I. J. Goldstein, M. Marini,Biochimn. Biophys. Acta 243, 90 (1971).

70. R. D. Poretz and I. J. Goldstein, Arch.Biochem, Biophys. 125, 1034 (1968).

'1. G. S. Hassing and I. J. Goldstein, Eur. J.Biochem. 16, 549 (1969).

72. I. J. Goldstein, L. L. So, Y. Yang, Q. C.Callies, J. Immunol. 103, 695 (1969).

73. W. J. Reeder and R. D. Ekstedt, ibid. 106,334 (1971); A. R. Archibald and H. E.Coapes, Biochem. J. 123, 665 (1971).

74. E. A. Lewis, M. J. St. Cyr, F. Smith,Carbohyd. Res. 5, 194 (1967); L. L. Soand I. J. Goldstein, ibid. 10, 231 (1969);I. J. Goldstein and A. Misaki, J. Bacteriol.103, 422 (1970).

75. G. F. Springer, P. R. Desai, B. Kolecki,Biochemistry 3, 1076 (1964).

76. C. R. Beddell, J. Moult, D. C. Phillips, inMolectular Properties of Druig Receptors, R.Porter and M. O'Connor, Eds. (Churchill.London, 1970), p. 85.

77. L. L. So and I. J. Goldstein, J. Biol. Chenm.243, 2003 (1968).

78. I. J. Goldstein, R. D. Poretz, L. L. So,Y. Yang, Arch. Biochem. Biophys. 127, 787(1968).

79. W. J. Lennarz and B. Talamo, J. Biol.Chem. 241, 2707 (1966).

80. L. Kass, 0. D. Ratnoff, M. A. Leon, J.Clin. Invest. 48, 351 (1969).

81. B. R. Andersen, Immunochemistry 6, 739(1969).

82. R. J. Doyle and L. S. Cronholm, J. Chem.Educ. 46, 868 (1969).

83. E. H. Donnelly and I. J. Goldstein,Biochem. J. 118, 679 (1970); K. 0. Lloyd,Arch. Biochem. Biophys. 137, 460 (1970).

84. S. Avrameas and B. Guilbert, Biochimie 53,603 (1971).

85. D. Allan, J. Auger, M. J. Crumpton,Biochem. J. 126, 6P (1972).

86. A. A. Moscona, Science 171, 905 (1971).87. G. M. Edelman and C. F. Millette, Proc.

Nat. Acad. Sci. U.S.A. 68, 2436 (1971).88. J. D. Oram, D. C. Ellwood, G. Appleyard,

J. L. Stanley, Nature New BDol. 233, 50(1971).

89. S. Friberg, Jr., A. J. Cochran, S. H. Golub,ibid. 232, 121 (1971).

90. M. 0. J. Olson and I. E. Liener, Biochem-istry 6, 3801 (1967); A. E. Powell and M. A.Leon, Exp. Cell Res. 62, 315 (1970); A.Novogrodsky and E. Katchalski, Biochim.Biophys. Acta 228, 579 (1971).

91. L. Stavy, A. J. Treves, M. Feldman,Nature 232, 56 (1971).

92. R. J. Doyle, E. E. Woodside, C. W. Fishel,Biochem. J. 106, 35 (1968).

93. F. Biddle, A. P. Cronin, F. K._San,

Cytobios 5, 9 (1970).94. M. M. Burger, Nature 219, 499 (1968);

V. K. Jansons and M. M. Burger, Fed.Proc. 30, 692 (1971).

95. R. Kornfeld and S. Kornfeld, J. Biol. Chem.245, 2536 (1970); S. Komfeld, J. Rogers,W. Gregory, ibid. 246, 6581 (1971).

96. H. J. Cline and D. C. Livingston, NatureNew Biol. 232, 155 (1971); B. Ozanne andJ. Sambrook, ibid., p. 156; D. J. Arndt-Jovinand P. Berg, J. Virol. 8, 716 (1971).

97. B. A. Sela, H. Lis, N. Sharon, L. Sachs,Biochim. Biophys. Acta 249, 564 (1971).

98. G. L. Nicolson and S. J. Singer. Proc. Nat.Acad. Sci. U.S.A. 68, 942 (1971); G. L.Nicolson, Nature New Biol. 233, 244 (1971);S. J. Singer and G. L. Nicolson, Science175, 720 (1972).

99. H. Ben-Bassat, M. Inbar, L. Sachs, J.Membrane Biol. 6, 183 (1971).

100. J. C. Aub, B. H. Sanford, L. H. Wang,Proc. Nat. Acad. Sci. U.S.A. 54, 400 (1965);R. R. Gantt, J. R. Martin, V. J. Evans, J.Nat. Cancer Inst. 42, 369 (1969); E. G.Moore and H. M. Temin, Nature 231, 117(1971).

101. T. Takahashi, G. Funatsu, M. Funatsu, J.Biochem. Tokyo 51, 288 (1962); ibid. 52,50 (1962); M. Ishiguro, T. Takahashi, G.Funatsu, K. Hayashi, M. Funatsu, ibid. 55,587 (1964); ibid. 56, 325 (1964); V. E. Wald-schmidt-Leitz and L. Keller, Z. Physiol.Chem. 350. 503 (1969).

SCIENCE, VOL. 177

102. G. F. Springer and P. Williamson, Biochem.J. 85, 282 (1962).

103. P. Albersheim and A. J. Anderson, Proc.Nat. Acad. Sci. U.S.A. 68, 1815 (1971).

104. M. Heidelberger, Fortschr. Chem. Org.Naturst. 18, 503 (1960).

105. J. Font, A. M. Leseney, R. Bourillon,Blochim. Biophys. Acta 243, 434 (1971).

106. I. K. Howard and H. J. Sage, Biochemistry8, 2436 (1969).

107. W. G. Jafft and K. Gaede, Nature 183, 1329(1959); W. G. Jaff6 and K. Hannig, Arch.Biochem. Biophys. 109, 80 (1965).

108. D. A. Rigas, E. A. Johnson, R. T. Jones,J. D. McDermed, V. V. Tisdale, inChromatographie et Mdthodes de SeparationImmediate (Association of Greek Chemists,Athens, 1966), vol. 2, p. 151; D. A. Rigasand E. A. Johnson, Biophys. J. 7, abstr. p.29 (1967); D. A. Rigas and C. Head,Biochem. Biophys. Res. Commun. 34, 633(1969); K Dahlgren, J. Porath, K. Lindahl-Kiessling, Arch. Blochem. Biophys. 137, 306(1970).

109. T. Takahashi, P. Ramachandramurthy, I. E.Liener, Blochim. Biophys. Acta 133, 123

(1967); T. Takahashi and I. E. Liener, ibid.154, 560 (1968).

110. R. Bourrillon and J. Font, Ibid., p. 28; J.Font and R. Bourrillon, ibid. 243, 111 (1971).

111. V. A. Marinkovich, J. Immunol. 93, 732(1964).

112. Work from the authors' laboratory was sup-ported by grants from the U.S. Departmentof Agriculture. This article was written inpart while N.S. was at the Department ofChemical Pathology, St. Mary's Hospital,London, under the Royal Society-lsraelVisiting Research Professorship Scheme.

Coherent Social Groupsin Scientific Change

"Invisible colleges" may beconsistent throughout science.

Belver C. Griffith and Nicholas C. Mullins

Recent studies have revealed a greatdeal about the communication and or-ganizational patterns that underliemajor advances and changes of direc-tion in science. An important featureof these patterns is their consistencythroughout a variety of disciplines,periods of time, and types of research.Biologists on Long Island in the 1940's,ethnomethodologists in southern Cali-fornia in the 1960's, physicists inCopenhagen in the 1920's, and mathe-maticians in Gottingen in the 1900'sacted in very similar ways when, armedwith insights radical for their times anddisciplines, they faced important scien-tific problems. Our data on these andother such groups suggest that a singleset of social mechanisms evolves in re-sponse to the challenge posed by newand major scientific problems. Whenchallenged, some members of a scien-tific specialty become organized to worktoward certain objectives, voluntarilyand self-consciously, as a coherent andactivist group. This article examinesfindings from surveys, individual inter-views, and biographical essays, and dis-cusses the similarities among contem-porary groups that developed intosmall, coherent, activist groups andthat subsequently had major impacts ontheir "home" disciplines.lS SEPTEMBER 1972

Low Levels of Organizationand Communication

Communication and some degree ofvoluntary association are intrinsic inscience, and the important questiontherefore becomes not whether scien-tists organize, but rather how, why, andto what degree? As a background tounderstanding high degrees of commu-nication and organization, the first sec-tion of this article examines the proc-esses entailed in the "loose" networks,the level of communication and organi-zation that appears normal for science.This level has been repeatedly demon-strated by different methodologies forspecialties in various disciplines.

Three groups of psychological re-searchers studied by Griffith and Millerexhibited this effective, loose communi-cation network (1). The workers inthese groups had considerable knowl-edge of the activities of other majorresearchers, and it is clear that individ-uals sought out, and interacted veryeffectively with, one another on thebasis of their current research interests.Mullins' data on biologists and Crane'son rural sociologists also show this kindof loose communication network. Inboth of these studies, respondentsnamed more persons outside their spe-

cialty than inside as having had sig-nificant effects on their work, therebysuggesting that scientists work in andinfluence more than one specialty, anapparently normal condition for highlyactive scientific researchers (2). In ad-dition, the Crane study showed thatproductive scientists were more fre-quently named as the object of contact,indicating, as expected, that communi-cation is more intense around produc-tive researchers. This finding was laterquantified by Griffith et al. (3).The Griffith and Miller study (1)

focused upon persons who were com-paratively productive, each of whomheaded a team of students and juniorcolleagues. These respondents employedseveral special strategies to facilitateinformation exchange within their spe-cialty. For example, one area, speechperception, was small enough that fewcommunication problems developed,even though the area exhibited lowlevels of social organization. Of theother specialties studied, those that ex-hibited loose networks employed, forvarying periods of time, mechanismsthat are used in highly coherent groups(for example, conference series and ex-changes of papers before publication).However, the adoption of a pattern ofcommunication within these specialtieswas in response to a current scientificproblem and to the inadequacy offormal meetings and journals for an-swering communications needs createdby these problems. For example, psy-cholinguists seemed to develop differentpatterns of organization dependingupon whether they were in the processof applying psychological theories andmethodology to studies of language (asthey did in the early 1950's) or modi-fying linguistic theories so they couldbe used in experimental psychology (asthey did after the development of gener-ative grammar). By contrast, research-

Dr. Griffith is professor of library science, DrexelUniversity, Philadelphia, Pennsylvania 19104. Dr.Mullins is associate professor of sociology, In-diana University, Bloomington 47401.

959