16. T. Mosquin, Ph.D. thesis (Univ. of Call- 20. R. Ornduff, Univ. Calif. (Berkeley, Los An-foria, Los Angeles, 1961). geles) Pubi. Botany, in press.

17. F. C. Vasek, Evolution 18, 26 (1964); oral 21. H. Lewis, unpublished data.communication. 22. and M. E. Lewis, Univ. Calif.

18. D. W. Kyhos, Evolution 19, 26 (1965). (Berkeley, Los Angeles) PubI. Botany 20, 24119. H. Lewis, American Naturalist 95, 155 (1955).

(1961). 23. H. Lewis, Evolution 16, 257 (1962).

Blood-Group Substances

In the ABO system the genes control the arrangement ofsugar residues that determines blood-group specificity.

Winifred M. Watkins

The first divisions of blood intogroups were based on differencesamong the antigenic substances on thesurfaces of the red cells of a givenspecies of animal. These antigens aredistinguished by means of antibodiesin serum that combine with the redcells. Blood-group characters are in-herited according to simple Mendelianlaws, and the antigens, believed to beproducts of allelic, or closely linked,genes, are classified together in blood-group systems (1). Inherited variationsin serum proteins (2) and enzyme ac-tivities in both serum and red cells (3)are other factors now known to differ-entiate bloods within a species, but theblood-group substances discussed inthis article are those related to thegroups defined by the red cell anti-gens.The first human blood-group sys-

tem, the ABO system, was discoveredby Landsteiner (4) as a result of hisattempts to determine whether specificserological differences existed betweenindividuals of the same species, andthe importance of a knowledge of theABO groups for the safe practice ofblood transfusion was promptly rec-ognized. The second blood-group sys-tem of outstanding clinical significance,the rhesus (Rh)- system, was notdiscovered until 40 years later (5).In the intervening years two other sys-tems, the MN and the P systems, werefound from examination of serums

172

from rabbits injected with differentsamples of human red cells (6). Theantigens in the MN and P systems areof little clinical importance becausethe corresponding antibodies occur in-frequently in man and seldom produceuntoward transfusion reactions. At thetime of their discovery, however, theinheritance of very few normal humancharacters was established, and the Mand N blood-group factors were usedas research tools in genetics and an-thropology. With the outbreak ofWorld War II there was a rapid ex-pansion of blood transfusion servicesin many countries, and the consequentdevelopment of improved blood-grouping techniques, together with thevast increase in the number of bloodsamples examined, led to the discoveryof many new blood-group systems andsubdivisions of the existing groups.Since the end of the war studies on theserology and genetics of blood groupshave continued to flourish; now some14 human systems, which include over60 different blood-group factors, areknown (1).

Although from the point of view ofthe clinician it is not necessary to beacquainted with more than the serologi-cal relationships of the blood-group an-tigens and antibodies to carry out suc-cessful blood transfusions, it is of greatimportance to study their chemistryand genetics in order to understand thebasis of blood-group specificity. The

24. K. R. Lewis and B. John, Chromosoma 10,589 (1959); A. Mtlntzing and S. Akdik,Hereditas 34, 485 (1948).

25. S. Smith-White, Cold Spring Harbor Symp.Quant. Biol. 24, 273 (1959).

26. Research on which this article is based hasbeen supported by NSF.

blood groups are defined by their sero-logical properties; the antigens areidentified by means of antibodies whichmay be derived from the same speciesas that in which the antigen is dem-onstrated, or from a different species.In the simplest instances specific com-bination of antigen and antibodycauses red cells to clump or aggluti-nate; of equal importance in blood-group serology are the so-called "in-complete" or "blocking" antibodiesthat can be demonstrated only by morerefined serological techniques (1).Chemical studies of blood-group anti-gens have not kept pace with the sero-logical and genetical advances, but forthe ABO and closely related Lewissystems a pattern of relationships be-tween gene function, chemical struc-ture, and serological specificity is be-ginning to emerge, and considerationof these systems is the principal topicof this article.

Water-Soluble Blood-Group Substances

Somewhat paradoxically the largestamount of information on the chemicalbasis of blood-group specificity hascome from chemical and immunologi-cal examination of substances that arenot derived from red cells. In the firstthree decades after the discovery ofthe ABO system, attempts to obtainblood-group active materials from redcells met with limited success. A andB substances could not be extractedfrom the cells with water or salt solu-tions, but active preparations were ob-tained by extraction with ethanol; thesubstances were therefore designated as"alcohol-s-oluble." The amounts iso-lated were very small, and reports ontheir chemistry were conflicting (7).Further impetus to chemical studiescame from the discovery that, in hu-mans, substances with A- and B-activ-ity are present not only on the redcell, but they also occur in a water-soluble form in the tissue fluids and

The author is Reader in Biochemistry (Immuno-chemistry), University of London, Lister Insti-tute of Preventive Medicine, London, S.W.1.

SCIENCE, VOL. 152

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

secretions of most individuals (7).In addition, shortly after the discoveryof the human water-soluble blood-group substances it was found thatmaterials with specificities related tothe human A and B substances couldbe isolated in a water-soluble formfrom the proteolytic digests of stomachmucosal linings of certain animals (7).These water-soluble substances are de-tected and assayed by their capacity,when mixed with the appropriate anti-serum, to combine specifically with theblood-group antibody so that aggluti-nation fails to take place upon subse-quent addition of the correspondingred cells. Substances that inhibit blood-group agglutination also occur in cer-tain plants and bacteria, and these haveprovided another source of material forchemical studies on blood-group speci-ficity (8, 9).

For many years after the discoveryof the water-soluble blood-group sub-stances no serious attempts were madeto establish the chemical nature of thered cell antigens, but the term "al-cohol-soluble" was retained to de-scribe them. The application of newerspecial procedures for the separationof complex macromolecules has nowresulted in the isolation, in water-sol-uble form, of the A and B antigensthemselves (10, 11) from the red cells.However, the term "water-soluble" isnevertheless still largely used to describethose blood-group substances derivedfrom sources other than red cells.

Genetics of ABO and Lewis Systems

Five blood-group specificities, A, B,H, Lea, and Leb, are detectable inhuman secretions by the use of thecorresponding antibodies of humanor animal origin. The structures re-sponsible for these specificities are be-lieved to arise from the action of fourindependent, but closely interrelated,gene systems, namely, ABO, Hh, Lele,and Sese (12-14). The relationshipsbetween the gene systems may best beresolved by knowledge of the chemicalstructures that confer specificity, andby understanding the ways in whichthese structures are built up under theinfluence of the genes.The ABO classification of bloods is

based on two red cell antigens, A andB, and the two antibodies to these,respectively, anti-A (a) and anti-B(/3); these antibodies always occur inthe serum or plasma when the cor-

8 APRIL 1966

Table 1. Relation between genotype, antigenson red cells, and antibodies in serum in ABOblood-group systerm.

Group G AntigenGrnou Geno- on Antibodiestype) type red in serumtype)cell

A AA A anti-B (13)

B BB B anti-A (a)

AB AB AB

O 00 anti-A (ca) andanti-B (13)

responding antigen is absent from thered cell (Table 1). The generally ac-cepted mechanism of inheritance, firstadvanced by Bernstein (15), requiresthat there be three allelic genes, A, B,and 0, any one of which can occupya given position on a chromosome.The A and B substances are not foundin secretions unless the correspondingantigen is present on the surface ofthe red cells (16), but about 20 per-cent of those with A or B antigenson their red cells fail to secrete thecorresponding substances. Schiff andSasaki (17), recognizing that secretionis a dimorphic character, suggested thatthe secretion of A or B substance iscontrolled by a pair of allelic genes,now generally called Se and se, whichare inherited independently of the ABOgenes and do not influence the expres-sion of these genes on the red cell.Gene Se in single or double dose re-sults in secretion; se in double doseresults in nonsecretion.The discovery in 1927 (18) of ag-

glutinins in certain cattle serums whichreacted preferentially with group 0cells raised the question whether theO gene is recessive, as was believedearlier, or whether it gives rise to a spe-cific product analogous to the productsof the A and B genes. The secretionsfrom group 0 individuals of the geno-types SeSe or Sese neutralize the groupO cell agglutinins, but saliva from groupA, B, and AB secretors also does so.Since individuals of the genotype ABcannot have an 0 gene these observa-tions strongly indicated that the sub-stance on group 0 cells, and in groupO secretions, detected by the so-called"anti-O" reagents, was not a directproductb of the 0 gene; this substancewas therefore renamed H substance(19).

According to the views of Wat-kins and Morgan (13), Ceppellini(14), and others, the formation of H

substance is controlled by a pair of al-lelic genes H and h (20). The Hgene in single or double dose givesrise to the H character, and the veryrare allele h, when present in doubledose, results in the absence of thischaracter. The H-active material isprobably the substance that, under theinfluence of the A and B genes, ismainly converted into A- and B-activesubstances. The presence of large quan-tities of H substance in group 0 in-dividuals can therefore be attributed tothe absence of the A and B genes.Although the Hh and ABO systemsare inheritable independently, the factthat the formation of H substance isa necessary prerequisite for the forma-tion of A and B inevitably means thatthe two systems are closely related intheir phenotypic effects. Further, thesecretor genes probably do not directlycontrol the expression of the A andB genes but act at the stage of theformation of H substance; that is, insecretory cells, although not on thered cells, the presence of two se genesis in some way incompatible with thefunctioning of the H gene. The sup-pression of this one gene results inthe absence of H, and hence of A orB substances, or both.The Lewis system is defined by two

allelic genes Le and le, which are in-herited independently of the ABO,Hh, and Sese genes (1). The closeinterrelateeship of the ABO andLewis systems, first observed by Grubb(21), is more readily understoodnow because it is known that, at leastinsofar as the secreted substancesare concerned, the changes producedby the Le gene occur on the samemacromolecules as those produced bythe Hh and ABO genes. In single ordouble dose, Le gives rise to Len-spe-cific structures; le in double dose re-sults in their absence. The expressionof the Le gene is not controlled bythe Sese genes. The Leb specificity,once thought to arise from the activityof an allele of the Le gene, is now be-lieved to be an interaction product ofH and Le genes (12-14).

Six main groups can be distin-guished on the basis of ABO andLewis phenotypes on the red cells andthe presence of A-, B-, H-, Lea-, andLeb-specific substances in secretions(Table 2). Individuals of group 4 whofail to secrete A, B, H, Lea, and Lebsubstances have in their secretions asubstance that is chemically very close-ly related to the A, B, H, and Lea sub-

173

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

Table 2. Six groups distinguishable on the basis of the ABO and Lewis (Lea, or Leb)red-cell phenotype and the A, B, H, Lea, and Leb activities present in secretions. The resultsare expressed as follows: +++, Strong specific activity; +, weak specific activity; -, noactivity.

Antigens detectable SpecificitiesProbable detectableGroup gene on red cells in secretions

combinationABH Lea Leb ABH Lea Leb

1 ABO,* H, Se, Le +++ - ++ +++ + ++2 ABO, H, sese, Le +++ +++ - - +++3 ABO, H, Se, lele + + + - - ++4 ABO,.H, sese, lele + ++5 ABO, hh, Le, (Se or sese) - ++ - - +6 ABO, hh, lele (Se or sese) -

*Each individual has two of the ABO genes, one received from either parent. In groups 1 to 4for the red-cell antigens, and groups 1 and 3 for the secreted substances, H activity is strong inindividuals of the genotype 00 and usually much weaker for the other ABO genotypes.

stances, and this substance is probablythe precursor that in other individualsis converted into the blood-group ac-tive materials (13). This precursor hasno detectable blood-group activity andfor convenience is labelled "inactivesubstance." Groups 5 and 6 correspondto the very rare individuals of the"Bombay" phenotype (1) who lackA, B, and H reactivity both in theirsecretions and on the red cell, but donot show abnormalities in the expres-sion of the Le gene.

Chemistry of A, B, H, and

Lewis Substances

The most potent sources of blood-group substances among the normalsecretions of the body are saliva andgastric juice. Meconium is another richsource (7). The substances of humanorigin that have been studied most arethose isolated from ovarian cyst fluids(22). These fluids accumulate withinthe cyst over long periods, and a singlecyst sometimes yields several grams ofactive material (23).

Purified blood-group substances fromsecretions are glycoproteins containinga high percentage of carbohydrate.The average molecular weights of thesubstances from different individualsrange from 3 X 105 to 1 X 106;and probably a preparation havingblood-group activity, even from asingle individual and a single type ofsecretion, contains a family of mole-cules very closely related in generalstructure and composition. Such sub-stances usually contain about 85 per-cent of carbohydrate and 15 percent ofamino acids. The detailed structure ofthese molecules is not yet known, buttheir general properties are consistent

174

with the interpretation that a large num-ber of relatively short oligosaccharidechains are covalently attached at in-tervals to a peptide backbone.

Preparations of A, B, H, and Leasubstances isolated from ovarian cystsand free from contaminants (24), asdemonstrated by physical, chemical,and immunological methods, are iden-tical in qualitative composition. Eachcontains five sugars; a hexose, D-galac-tose; a methyl pentose, L-fucose; twoamino sugars, N-acetyl-D-glucosamineand N-acetyl-D-galactosamine; and thenine-carbon sugar, N-acetylneuraminicacid (sialic acid) (Fig. 1, a-e); thepeptide component in each is composedof the same 15 amino acids. Thus thedistinctive serological properties ofthese substances cannot be traced todifferences in qualitative composition.Some differences in quantitative com-position are observed, but values forany one component vary slightly amongpreparations with the same specificity;differences in composition of specimensof different blood-group specificity aretherefore significant only if they fallwell outside this range. The results ofanalysis of preparations of A, B, H, andLea substances, and a similarly prepared"inactive" glycoprotein from the secre-tion of a person belonging to group 4(Table 2), are given in Table 3. Com-pared with the other substances, Leasubstances usually have a lower fucosecontent, and A substances a higherN-acetylgalactosamine content. The "in-active" glycoprotein has not been sub-jected to the same rigorous purifica-tion procedures and detailed analysisas the active substances, but the re-sults obtained so far indicate that themost outstanding difference in thequantitative composition of this ma-terial is its very low fucose content.

N-Acetylneuraminic acid is the com-ponent present in the most variableamounts in preparations with blood-group activity and the complete re-moval of this sugar does not resultin loss of activity (25). Many prepara-tions show a reciprocal relation betweenthe contents of fucose and N-acetyl-neuraminic acid comparable with thatobserved in urinary and other glyco-proteins (26).The composition of the peptide com-

ponent is distinctive in that four aminoacids, threonine, serine, proline, andalanine, together make up about two-thirds of the amino acids present (27).Sulfur-containing amino acids are vir-tually absent, and aromatic amino acidsare present in only very small amounts.Integrity of the complete macromole-cules is essential for maximum serologi-cal reactivity, and the role of the pep-tide component appears to be that ofmaintaining the correct spacing andorientation of the carbohydrate chains.

Structure and Specificity

Investigations on the structure of theA-, B-, H-, and Lea-active glycopro-teins have been directed more towardfinding the underlying basis of blood-group specificity than toward the eluci-dation of the overall structure of themacromolecules. Immunochemical, en-zymatic, and chemical methods haveeach contributed valuable informationabout structure. The serological inhibi-tion method, based on the observa-tion of Landsteiner (28), that a simplesubstance with a structure closely re-lated to, or identical with, the im-munologically determinant group of anantigen, combines with the antibody andthereby competitively inhibits the re-actions between the antigen and itscorresponding antibody, gave the firstclear indications that the serologicalspecificity of the blood-group sub-stances is associated with the carbohy-drate part of the molecules (29). Anenzymatic inhibition method (30),which made use of the well-knownfact that many enzymes are inhibitedby the products of their own activity,supported and extended the results ob-tained by the serological inhibitiontests.From these indirect methods it

was inferred that at least some of thechains in the A substance have a termi-nal N-acetyl-D-galactosamine residue(Fig. la), joined in a-glycosidic link-age to the next sugar, and that this

SCIENCE, VOL. 152

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

terminal nonreducing sugar unit is animportant part of the structure of theA determinant. The B specificity wassimilarly associated with a terminalnonreducing D-galactose residue (Fig.1 b) joined in a-linkage to the nextsugar. Both the N-acetyl-D-galactos-aminyl and D-galactosyl residues are inthe pyranose form. These first resultstherefore pointed to an important con-clusion, supported by all subsequentwork, that the major difference in Aand B specificity can be traced notonly to the nature of the terminal non-reducing sugar but, more narrowly, tothat of the substituent at carbon atomNo. 2 in this sugar molecule. D-Galac-tose and N-acetyl-D-galactosamine arestructurally identical except that thehydroxyl group in the former at car-bon No. 2 is replaced by an N-acetylamino group in the latter. Ana-L-fucosyl residue (Fig. lc) is an im-portant part of the H-specific determi-nant, as judged by inhibition tests.Thus despite the fact that L-fucose, D-galactose, and N-acetyl-D-galactosa-mine are all components of A, B, andH substances, it is possible in each in-stance to relate specificity to one ofthese sugars and hence to deduce thatthe determinant carbohydrate chainsvary in the nature of the sugar, or its

Table 3. Analytical figures (percentages) forblood-group specific substances, and an "in-active" glycoprotein, isolated from humanovarian cysts.

Specific substance "Inactive"

A B H Lea glycoproteinNitrogen

4.9 4.8 5.0 4.9 5.4

Fucose20 18 19 13 1.6

Hexosamine*32 33 28 29 28

Reducing sugars t51 50 50 56 49*Expressed as glucosamine after acid hydrolysis.tExpressed as glucose after acid hydrolysis.

linkage, or both, at the nonreducingends of the chains.

Sometimes A, B, H, and Lea sub-stances cross-react with horse antibodyto the Type XIV pneumococcal poly-saccharide (7) when in the undegrad-ed state, and they always do so aftermild degradation by acid. On the basisof inhibition tests this cross-reactivitywas attributed to the presence in thesubstances of chains ending in 4-0-/3 -D - galactosyl -N - acetyl-D-glucosamineunits (31).More extensive information about

the Lea determinant was derived from

inhibition tests with fucose-containingoligosaccharides isolated from humanmilk (32). Two oligosaccharides con-taining a branched trisaccharide unitat the nonreducing end, with a-L-fu-cose and /3-D-galactose residues joinedby 1--4 and 1- 3 linkages, respective-ly, to an N-acetyl-ft-D-glucosaminylunit, strongly inhibited the agglutina-tion of Lea-positive red cells by ahuman antiserum to Lea (33). Otherclosely related compounds, that dif-fered only in the point of attachmentof the fucose residue, did not causeinhibition. It was therefore conclud-ed that Lea activity resided in thespatial arrangement of this branchedtrisaccharide (Fig. 2a). The involve-ment of L-fucose in two distinctserological specificities, namely H andLea, indicated that the nature of theterminal nonreducing sugar cannot, byitself, be responsible for specificity,and that the way in which it is linkedto the next sugar, or the nature ofthis second sugar, or both, must alsobe important. Another interesting in-ference from the results with Lea wasthat specificity does not necessarily re-side in the nature and sequence ofsugar units in a straight carbohydratechain, but that branching sugar resi-dues can contribute to the determi-

H2OH *CH20H

OH OH

KHO H.014

NHCOCH3 OHl(a) (b)

*CH20H

,O HOH COH

OH NHCOCH3(c) Id)

(e)Fig. 1 (above). Monosaccharide constituents ofA, B, H, and Lea substances. (a) N-acetyl-D-galactosamine; (b) D-galactose; (c) L-fucose;(d) N-acetyl-D-glucosamine; (e) N-acetylneu-raminic acid.Fig. 2 (right). Structures present in oligosac-charides giving inhibition in (a) Lea hemag-glutination tests and (b) Leb hemagglutinationtests.

8 APRIL 1966

NHCOCH3

a

O-p-l-GALAcTOSYL-(l13)-[O--L-FucOsYL- (1- 4)--- O-N-ACETYL-P-GLUCOSAMINYL-

0-c- L-FUCOSYL-(1c 2)-O-p-D-GALACTOSYL-Ct-.3)-[O-.C-L-FUCOSYL-C1I-4)]-O-N -ACETYL-p-D-GLUCOSAMINYL-

175

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

A-SUBSTANC ESTEP I

"A ENZYME IN-ACETYLGALACTOSAMINE + H-SUBST.

STEP 2 H" ENZYME

B-SUBSTANCE

IB ENZYME

H-SUBST. + GALACTOSE

I" ENZYME

Lea & TYPE XIV ACTIVE SUBST. + FUCOSE

STEP 3 Le"'ENZYME

TYPE XIV ACTIVE SUBST.Fig. 3. Specificities revealed by the sequential enzymatic degradation of A and Bsubstances.

nant structure. Further indications ofthe importance of branching fucose resi-dues came from inhibition experimentswith Leb (33). Oligosaccharides con-taining two fucose residues attached toadjacent sugars on a backbone chain(Fig. 2b) were active inhibitors ofLeb agglutination, whereas analogs

lacking either fucose unit were inac-tive.The structure of A, B, H, and Lea

substances may be further elucidatedby the use of certain enzymes thatbring about controlled degradation ofthe molecules. Enzymes from bacteriaand protozoa that destroy the serologi-

cal properties of blood-group substanceshave been known for many years (7),but they have proved surprisingly dif-ficult to purify and characterize interms of the chemical changes that theycatalyze. Several of them are nowknown to split off the terminal non-reducing sugars from the ends of thecarbohydrate chains; that is, they areexoglycosidases. By the action of theseenzymes the carbohydrate chains in theblood-group substances can be de-graded sequentially and serologicalspecificities uncovered which werenot detectable, or only weakly so,in the undegraded substances. Thefact that more than one specificity canbe present on the same macromoleculewas known from precipitation experi-ments with selected monospecific anti-serums (34). When, for example, apurified blood-group preparation froman AB subject is precipitated with anantiserum to A, both A and B activitiesare carried down in the precipitate.Both H and Lea structures are demon-strable by similar methods in macro-molecules with either A or B activity,or both. Enzymatic degradation dem-onstrated that, in addition to the vari-ous groupings available for reactivityin the undegraded substances, struc-tures corresponding to other specificities

(a)

O-0F-L-FUCOSYL-(1-2)-O-P-D-GALACTOSYL-(1-4 )-N-ACETYL-D-GLUCOSAMINE O-O4N-ACETYL-D-GALACTOSAMINYL- (1-3)-CO -o -L- FUCOSYL-Cf-2)-]-

0-f-D-GALACTOSYL-C1@4)-N-ACETYL-D-GLUCOSAMINE

NHCOCH3

lb)

O-otL-FUCOSYL- (1-2 )-O-p - D-GALACTOSYL-(d-o3)-N-ACETYL-D-GLUCOSAMINE O-at-D-GALACTOSYL-(1-3)- CO- L-FUCOSYL- C1 '2)-3 -

O-f-D-GALACTOSYL-( -4)-N-ACETYL-D-GLUCOSAMINE

Fig. 4 (left). H-active trisaccharides isolated from (a) alkaline degradation and (b) acid hydrolysis products of human H sub-stance. Fig. 5 (right). Serologically active tetrasaccharides isolated from alkaline degradation products of (a) human A sub-stance and (b) human B substance. The arrows indicate the only groupings that differ in the two structures.

SCIENCE, VOL. 15216

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

are latent in the macromolecules. WhenA or B substances are treated withpurified preparations of certain A- orB-decomposing enzymes they lose N-acetylgalactosamine or D-galactose, re-spectively, and H activity is revealed(35). Treatment of the residual H-ac-tive substance with a specific H-decom-posing enzyme results in the liberationof fucose and the development or en-hancement of cross-reactivity withhorse antiserum to Type XIV pneumo-coccus. Moreover, when the origi-nal A or B preparation is derived froman individual belonging to group 1(Table 2), that is, from someone withan Le gene, loss of H activity is alsoaccompanied by a development of Leaactivity (36). Finally, when the Leaactivity is destroyed by a specific Lea-decomposing enzyme the residual sub-stance retains its cross-reactivity withType XIV antiserum (Fig. 3). It ap-peared theTefore that structures relatedto H, Lea, and Type XIV pneumococ-cus determinants are present in A andB substances and that, as the releaseof a single' sugar unit suffices to un-mask a new specificity, the structuresrevealed were initially part of the samecarbohydrate chains.The identification of carbohydrate

fragments from the products of partialacid hydrolysis (37) and alkaline deg-radation (38) supplied the details ofstructure necessary to interpret thecomplex interrelationships of specifici-ties revealed by the enzyme degrada-tion. Fucose and N-acetylneuraminicacid are readily cleaved from theblood-group substances by conditions ofmild acid hydrolysis and are notfound, except in very small amounts,attached to fragments isolated from theproducts of partial hydrolysis with acid.Since the two sugars most probablyoccur only as terminal nonreducing endgroups, or branching units, the maincarbohydrate chains may be recon-structed without knowledge of thepoints of attachment of the fucose orN-acetylneuraminic acid residues. Onthe basis of the di- and trisaccharidesobtained from partial acid-hydrolysisproducts the structures given in Table4 were suggested for the two maintypes of carbohydrate chain in A, B,H, and Lea substances (39). Thechains in A substance differ from thosein B substance only in the nature ofthe terminal nonreducing sugar; N-acetyl-D-galactosamine in A substanceand D-galactose in B substance. Eachof these terminal sugars is joined by ana-1-;-3-linkage to a subterminal galac-8 APRIL 1966

Table 4. Partial structures proposed for the carbohydrate chains in blood-group A, B, H, andLea substances.

Chain Structure*

A-substanceType 1 a-GaINAc-(1- 3)-f--Gal-( l->3)-,8-GNAc-( l-3)-,--Gal-(1-e3) -GalNAc-Type 2 a-GaINAc-(l-->3)-,-Gal-(l1- 4)-fi-GNAc-(1->3)-,-Gal-(l1-e3)-GalNAc-

B-substanceType 1 a-Gal-( 1-3)-,-Gal-( l-*3)->-GNAc-( 1-3 )-,--Gal-( 1-3)-GalNAc-Type 2 a-Gal-(l1-e3)-,B-Gal-(1-e4)-,8-GNAc-(1I-3)-,8-Gal-(1->3)-GalNAc-

H- and Lea-substancesType 1 P-Ga1-(1- 3)-,B-GNAc-(1-3)-,8-Gal-(1I-3)-GalNAc-Type 2 8-Gal-(le498 NAc-(le3S,8 al-(1 >3-GalNAc-

* Abbreviations: GaINAc, N-acetyl-D-galactosaminopyranose, Gal, D-galactopyranose. GNAc, N-acetyl-D-glucosaminopyranose.

tose residue. All the other linkages inthe chains are in the p8-form. Thechains in H and Lea substances areidentical with those in A and B sub-stances after removal of the terminalnonreducing sugar. The presence oftwo types of carbohydrate chain in eachsubstance was indicated by the isola-tion of two serologically active trisac-charides from both single specimens ofA and B substances, each pair differingonly in the nature of the linkage, (1-*3)or (1-*4), of the subterminal galactoseto the third sugar in the chain, namelyN-acetylglucosamine (40).The point of attachment of some of

the fucose units was ascertained byidentification of fragments isolatedfrom alkaline degradation products.The identification (Fig. 2a) of an ac-tive trisaccharide from Lea substance(41) confirmed the prediction from theserological inhibition tests that thisunit was responsible for Lea activityand established that the fucose in theLea-active chains is on carbon No. 4of the subterminal N-acetylglucosamineresidue. The identification of a sero-logically active fucose-containing tri-saccharide (Fig. 4a), obtained from thealkaline degradation products of H sub-stance, and a second active trisaccha-ride (Fig. 4b), isolated in very lowyield from an acid-hydrolyzate (42),indicated that fucose is linked to car-bon No. 2 of the terminal galactosein both of the types of chain in Hsubstance. A highly active, reducedpentasaccharide fragment isolated fromproducts of alkaline degradation of Asubstance carried out in the presenceof borohydride (43), and a structur-ally similar tetrasaccharide (Fig. 5a)isolated from the products of deg-radation with aqueous methanolic tri-ethylamine (44), each had a L-fucoseresidue attached to the carbon No.2 of the galactosyl unit penulti-mate to the terminal nonreducinggroup. A similar tetrasaccharide was

isolated from the alkaline degradationproducts of B substance (Fig. Sb)(44). On the basis of the structuralinformation gained from serologicalinhibition tests, enzymatic degrada-tion, and partial hydrolysis, it is thuspossible to assign a defined structureto the four blood-group specificities,A, B, H, and Lea, and to begin tosee how these structures are related oneto the other.The points of attachment of the N-

acetylneuraminic acid residues are notyet determined, but there is no evi-dence that this sugar plays any partin A, B, H, Lea, or Leb specificity.

Genetic Control of Biosynthesisof Blood-Group Substances

The combined serological, genetical,and biochemical results support theidea that the blood-group genes do noteach control the complete synthesis ofthe macromolecules that carry specific-ity, but impose the specific pattern onlyat a late stage in synthesis (13). Theconversion in vitro of an inactive gly-coprotein into an active blood-groupsubstance has not yet been achieved.Nevertheless the simplest explanationof the role of the Le, H, A, and Bgenes is that they control the forma-tion, or functioning, of specific gly-cosyl transferase enzymes that addsugar units from a donor substrate tothe carbohydrate chains in a pre-formed glycoprotein molecule. An al-ternative mechanism by which thesugars are added to oligosaccharidesthat are subsequently joined to thepeptide moiety can also be envisaged,but the argument for the role of theblood-group genes is essentially thesame for the two mechanisms, provid-ed growth is from the terminal non-reducing ends of the carbohydratechains. The alleles of Le, H, A, andB, namely le, h, and 0, are thought

177

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

cSTRUCTURES FORMED

IGAL

CHAIN SPECIFICITY

(2) T-- TYPE XIV

ILI GENE

CADOITION OFL-FUOOSE TOPRECURSOR"3

GALIGALC1GA

(1) ---- Lea

(2) ----TYPE XIV

H h GENES-..

A GENE

ECADlO N OFN-ACETYLGALACT-OSMINE TO H-ACTIVE CHAINS3

|j GENE

CADOITION OFALACTOSE TOH-ACTIVE CHARS2

SYMBOLS D-GALACTOSE

-- )Mnd(2) ---A

N-ACETY LGALACTOSAMINE

L L-FUCOSE N -ACETYLGLUCOSAMINE

Fig. 6. Representation of the structures proposed for the carbohydrate chains in theprecursor glycoprotein and the additions to these chains controlled by the H, Le, A,and B genes.

178

GENES

drate chains in the precursor substanceis believed to be essential for the func-tioning of the transferases controlledby the A and B genes. N-acetyl-D-galactosamine is added in a-linkage to

SCIENCE, VOL. 152

UNIDENTIFIEDGENES GIVING'PRCUSOR'

H GENF.CADDITION OF--L -FUCOSE TOPRECURSOR

OD

not to control any changes in the gly-coproteins and, therefore, from thepoint of view of the biosynthesis ofblood-group active structures, can beconsidered inactive genes.

The presence in the secretions ofcertain individuals (group 4, Table 2)of a glycoprotein, chemically similarto A, B, H, and Lea substances butlacking any of these blood-group acti-vities, supports the idea that, in the ab-sence of the changes controlled by theblood-group genes, it is not the glyco-protein that is missing, but only thosegroupings that confer serological ac-

tivity. At least some of the m.ain car-

bohydrate chains in this glycoproteinare probably complete as far as theterminal nonreducing pl-galactosyl units-as indicated in the structures pro-posed for the chains in H and Leasubstances (Table 4). The type-2chain ending in an unsubstituted O-3-D-galactosyl- (1 -- 4)-N-acetylglucosaminylunit would confer the cross-reactivitywith antiserum to Type XIV pneu-mococcus observed with these so-

called "inactive" preparations. The con-version of the chains in the precursorinto Lea, Leb, and H, A, and B activegroupings could occur as follows (Fig.6): The transferase controlled by theH gene adds L-fucose in a-linkage tothe carbon No. 2 position of the termi-nal galactose in either chain to givean H-active structure. The transferasecontrolled by the Le gene, also a

fucosyl transferase, adds L-fucose ina-linkage to the carbon No. 4 positionof the subterminal N-acetylglucosamineunit in chains of type 1. The sub-terminal N-acetylglucosamine units inchains of type 2 are already substitutedat the carbon No. 4 position so thataddition of fucose at this position can-not take place. When both H and Legenes are present, addition of twofucosyl units to adjacent sugars onthe chains of type 1 results in a struc-ture different from that produced byeither gene alone. This structure is con-sidered to be the Leb determinant, al-though it should be emphasized thatan oligosaccharide containing thegrouping shown in Fig. 2b has yet tobe isolated and characterized from thedegradation products of a blood-groupsubstance.The presence of the fucosyl unit con-

ferring H specificity on the carbohy-

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

carbon No. 3 of the terminal galac-tosyl units of the H-active structureunder the influence of the A gene; D-

galactose is added, also in a-linkageand to carbon No. 3 of the terminalgalactosyl unit, under the influence ofthe B gene. The addition of theseterminal nonreducing sugars masks theserological reactivity of the H-activegroupings. The substitutions at the endsof the chains controlled by the A, B,and H genes similarly mask the Lee-active groupings. Under the control ofthe Le gene, fucose is added as abranch to the penultimate N-acetyl-glucosaminyl residue, whereas the sugaradditions controlled by the H, A, andB genes are made to the terminal non-reducing galactose residues in the pre-cursor chains; the Le gene is thereforenot competing for the substrate. Thenumber of Lea groupings available forreactivity does, however, depend onthe activity of the A, B, and H genes,and the strongest Lea-active substancesare found in nonsecretors in whom sup-pression of the H gene by the sesegenes prevents the formation of H-active structures, and hence of Leb, A-or B-active structures.

A and B Substances from Red Cells

Extraction of group A and B redcells, or stroma, with organic solventsand purification of the extract bychromatography on cellulose and silicicacid yields active materials that areglycolipids (10, 11), that is, com-pounds containing a carbohydrate moie-ty joined, through sphingosine, to fattyacids. The most active materials fromgroup A cells so far described hadabout 50 percent of the activity of apurified glycoprotein preparation of Asubstance. The activity of some verymuch less active fractions is increasedto this same degree of activity by com-bination with serologically inactive lipidisolated from the same red cell prepara-tons. Koscielak (11) suggests that theactivity of blood-group substancesfrom red cells depends largely on theirstate of aggregation in aqueous solu-tion. The active glycolipids contain, inaddition to fatty acids, sphingosine, andglucose, the same five sugars that arepresent in the glycoprotein blood-group substances, namely, galactose,glucosamine, galactosamine, fucose, andsialic acid. The analysis of two blood-group A-specific glycolipids is given inTable 5. The preparation insoluble incold methanol contains mainly ligno-8 APRIL 1966

GAL-(1-3)-GALNAc-(C1-m4)-GAL-(1 -*4)-G LUC-1-*R

NANA-(2-. 8)-NA NAFig. 7. Structure proposed by Dodd, Bigley, Johnson, and McCluer (56) for the im-munological determinant of the Rh. (D) antigen. Gal, D-galactose; Gluc, D-glucose,GalNAc, N-acetylgalactosamine; NANA, N-acetylneuraminic acid.

ceric acid residues, whereas that solublein methanol contains a higher propor-tion of short-chain fatty acid residues.The hemagglutination inhibiting ac-tivity of the methanol-soluble materialis less than 1 percent of that of themethanol-insoluble glycolipid, but itsactivity is increased to that of themethanol-insoluble glycolipid by com-bining it with inactive lipid (11).Glycolipids specific for blood-group B,essentially similar in composition andproperties to the A-specific glycolipids,have been isolated from B cells (45).The red cell A and B substances

thus appear to belong to a differentmolecular species from the A- and B-active glycoproteins from secretions.Nevertheless the possibility remainsthat the groupings responsible for thespecific blood-group properties areidentical in the two types of mole-cule. The structures of the glycolipidshave not yet been studied directly, butserological and enzymatic inhibitiontests (46) indicate that N-acetylgalacto-samine and D-galactose are the majorserological determinant units in the redcell A and B antigens, respectively.Moreover, in precipitin tests by doublediffusion in agar gel the secreted andred cell A substances give a reaction ofidentity with antisera produced in rab-bits against either group A red cells

Table 5. Analysis ofglycolipids (I l).

blood-group A active

Solubility of glycolipidConstituent in methanol

Insoluble Soluble

Hexosamine (%) 15.8 12.1Reducing sugar (%) 43.0 41.4Sialic acid (%) 10.4 10.9Fucose (%) 1.2 2.3Nitrogen (%) 2.46 2.48Sphingosine N (%) 0.81 0.91Glucosamine:galactosamine 3.0:1 3.1:1

Galactose:glucose 3.1:1 2.8:1Hemagglutination

inhibition* 50 <1* Expressed as percentage of activity of glyco-protein A substance.

or the purified A substance from anovarian cyst (46). On the assumptionthat the steps controlled by the blood-group genes take place only at theends of the carbohydrate chains inboth the glycoprotein and the glyco-lipid substances, and that both havethe same basic carbohydrate struc-tures, the association of A and Bspecificities with different classes ofmacromolecules on the red cells andin secretions is understandable. The rea-son for this difference is not clear,especially because glycoproteins not dis-similar in general composition to thesecreted blood-group A, B, H, and Leasubstances have been isolated fromred cells and shown to be associatedwith M and N specificities (47, 4&)Another rather baffling finding is thatmaterial isolated from group 0 cellsby techniques similar to those usedfor the isolation of A and B substancesyields only inactive glycolipids, and H-active substances have yet to be ob-tained from red cells (10, 11).

M, N, P1, and Rh. (D) Specificities

Evidence that carbohydrate struc-,tures are responsible for blood-groupspecificities other than A, B, H, andLewis is rapidly increasing. There isnow a fairly extensive body of workon the M and N substances from redcells (9, 47, 48); and, at least insofaras the specificities detected by humanand rabbit antiserums to M and N areconcerned, terminal nonreducing N-acetylneuraminic acid residues (Fig. le)are an important part of the serologicaldeterminant structures (49, 48). Thepresence of N activity in M prepara-tions derived from cells of the geno-type MM, the finding that both theM and N activity are associated withthe same macromolecule (50), and theexposure of common specificities in Mand N preparations by removal of N-acetylneuraminic acid (9, .51) sug-gest that sequential, gene-controlledchanges in a common precursor sub-

179

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

strate may also form the pattern ofsynthesis for antigens associated withthe MN system.The P1 antigen of the P blood-group

system (1) has not yet been isolatedfrom red cells. However, in a substancewith a similar specificity obtainedfrom the fluid of the sheep hydatid cyst(52), such activity is associated witha carbohydrate structure, and D-galac-tose in a-linkage is probably the mostimportant part of the serological de-terminant grouping (53). Attempts todetermine the nature of the serological-ly active structures in antigens in theRh system by the method of hemag-glutination inhibition with low-molecu-lar-weight inhibitors have involved somany compounds of diverse chemicalnature that interpretation of the re-sults is extremely difficult. Specific in-hibition of Rh. (D) (54) antibody bycompounds containing neuraminic acidhas been claimed, however, by morethan one group of workers (55); andrecently Dodd, Bigley, Johnson, andMcCuer (56), on the basis of inhibi-tic,a by gangliosides isolated from hu-man brain, proposed that the structuregiven in Fig. 7 represents the serologi-cal determinant in the Rh. (D) antigen.The similarity of this structure to thoseproposed for A, B, H, and Leaspecificities makes it an exciting pos-sibility, but Springer reports (57) thathe has been unable to confirm the in-hibition by gangliosides, and there-fore further developments must beawaited.

Summary and Conclusions

The picture that emerges from thechemical, serological, and geneticalanalysis of the ABH and Lewis blood-group substances from human secre-tions is one in which patterns of syn-thesis resulting from the sequential ac-tion of the products of different blood-group genes on a common precursorglycoprotein lead to the three-dimen-sional patterns of sugar residues re-sponsible for the serological specificity.The addition of a sugar residue to theend of, or as a branch on, a carbo-hydrate chain produces a new serologi-,cal specificity primarily determined bylie added sugar, but also dependent onthe nature, sequence, and linkage ofthose sugars already present in thechain. This new specificity may bemasked by the addition of anothersugar residue. Thus H-specific struc-

180

tures become part of the A- and B-specific groupings when N-acetyl-D-galactosamine or D-galactose, respec-tively, are added to the ends of thechains. It is not yet clear whether theLea groupings contribute to the com-plete A and B determinants in thechains to which fucose is added underthe control of the Le gene, but theLea groupings in the A and B chainsare not available for reactivity withLea antibodies.

In group AB, when both A and Bgenes are present, there is, presumably,competition for the carbohydratechains; but any one chain can onlybe completed either as an A- or asa B-active structure. Therefore A andB groupings can be present on thesame molecule without much apparentinteraction. On the other hand it isproposed that, in certain of the carbo-hydrate chains, the additions of L-fu-cose controlled by the H and Le genestake place on adjacent sugars in thesame chain, and that the presence oftwo fucosyl units gives rise to an in-teraction product, Leb, which has aserological specificity different fromthat of the structure produced byeither the H or Le gene acting in theabsence of the other gene.How far the picture for the secreted

blood-group substances is applicable tothe A and B substances on the redcell is not yet clear. Nevertheless, itis reasonably well established that N-acetyl-D-galactosamine and D-galac-tose, respectively, are important partsof the serological determinants of theglycolipid A and B substances. Thedifference in A and B specificity, ofsuch great importance in blood group-ing and transfusion, thus appears toreside ultimately in a relatively smallstructural variation, namely, in the na-ture of the substituent at carbon No.2 in a sugar with a D-galactose con-figuration (Figs. 1, a and b). If theoriginal gene at the ABO locus wereone that controlled the formation, orfunctioning, of an enzyme that trans-fers a D-galactosyl type of structurein a-(1--3) linkage to a ft-linked galac-tose residue substituted in the 2-positionwith fucose, then the mutationalchange which led from A to B, orvice versa, can be envisaged as achange which permits recognition bythe gene products of a hydroxyl groupat carbon No. 2 in the galactose con-figuration in place of an N-acetyl-amino group. Expressed in these terms,the extent of the dissimilarity between

the products of the A and B genesmay appear less great, and hencemore compatible with the kind of dis-similarity expected in the products ofallelic genes, than if the statement thatthe A and B genes control the plac-ing of two different sugar residues istaken to mean that configurationallydistinct structures are involved.

Inherited variations in protein struc-tures, or other antigenic noncarbohy-drate components of the red cell sur-face, may possibly account for someof the observed blood-group divisions.The pattern of synthesis and geneticalcontrol outlined for the ABO andLewis systems may have wider appti-cation since there is evidence thatcarbohydrate structures are concernedin M, N, P1, and possibly Rh. (D)specificity and that many of the sero-logical observations in the MN, P,and Rh systems can be explained ifthe specificities arise from the sequen-tial action of genes on a precursorsubstance (1, 58). A large variety ofstructural patterns can be formed witha relatively limited number of sugarunits, and if the role of many of tf~blood-group genes is the control ofthe arrangement of these carbohydratebuilding blocks it is possible to per-ceive how great diversity of the cell sur-face can be produced with consider-able economy of means.

References and Notes

1. For details of the serology and genetics ofall the human blood-group systems see R. R.Race and R. Sanger, Blood Groups in Man(Blackwell, Oxford, 1962).

2. 0. Smithies and G. E. Connell, in Symposi-um on the Biochemistry of Human Genetics,G. E. W. Wolstenholme and C. M. O'Connor,Eds. (Churchill, London, 1959), p. 178; R.Grubb, ibid., p. 264; J. Hirschfeld, ActaPathol. Microbiol. Scand. 47, 160 (1959).

3. H. Harris, in International Conference oiCongenital Malformations, 2nd, New York(International Medical Congress, 1964), pp.135-144.

4. K. Landsteiner, Zentr. Bakteriol. Parasitenk.27, 357 (1900); , Wien. Kiln.Wochschr. 14, 1132 (1901).

5. P. Levine and R. E. Stetson, J. Amer. Med.Assoc. 113, 126 (1939); K. Landsteiner andA. S. Wiener, Proc. Soc. Exp. Biol. Med.43, 223 (1940); A. S. Wiener and H. R.Peters, Ann. Internal. Med. 13, 2306 (1940).

6. K. Landsteiner and P. Levine, Proc. Soc.Exp. Biol. Med. 24, 941 (1927); J. Exp.Med. 48, 731 (1928).

7. E. A. Kabat, Blood-Group Substances (Aca-demic Press, New York, 1956).

8. G. F. Springer, in Symposium on the Chem-istry and Biology of Mucopolysaccharides,G. E. W. Wolstenholme and M. O'Connor,Eds. (Churchill, London, 1958), p. 216;S. Iseki, E. Onuki, K. Kashiwagi, GunmaJ. Med. Sci. 7, 7 (1958).

9. G. F. Springer, in Colloquium der Gesell-schaft fUr Physiologlsche Chemie (Mosbach,Berlin, 1964), p. 90.

10. T. Yamakawa and S. Suzuki, J. Biochem.(Tokyo) 39, 393 (1952); J. Koscielak and K.Zakrezewski, Nature 182, 516 (1960); S. I.Hakamori and R. W. Jeanloz, J. Biol.Chem. 236, 2827 (1961).

11. J. Koscielak, Biochim. Biophys. Acta 78,313 (1963).

SCIENCE, VOL. 152

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

12. R. Ceppellini, Proc. Intern. Congr. BloodTransfusion, 5th, Paris (1955), p. 207; W.M. Watkins, Proc. Intern. Congr. BloodTransfusion 7th, Rome (1958), p. 206; W.M. Watkins, in The Red Blood Cell, C.Bishop and D. M. Surgenor, Eds. (Aca-demic Press, New York, 1964), p. 359.

13. W. M. Watkins and W. T. J. Morgan,Vox Sang. 4, 97 (1959).

14. R. Ceppellini, in Symposium on the Bio-chenmistry of Human Genetics, G. E. W.Wolstenholme and C. M. O'Connor, Eds.(Churchill, London, 1959), p. 242.

15. F. Bernstein, Klin. Wochschr. 3, 1495 (1924).16. There are a few very rare exceptions to this

rule, described by Race and Sanger (1).7. F. Schiff and H. Sasaki, Klin. Wochschr. 34,

1426 (1932)..8. F. Schiff, ibid. 6, 303 (1927).19. W. T. J. Morgan and W. M. Watkins,

Brit. J. Exp. Pathol. 29, 159 (1948).'I). The Hh genes correspond to the Xx genes

described by P. Levine, E. Robinson, M.Celano, 0. Briggs, L. Falkinburg, Blood 10,1100 (1955).R. Grubb, Nature 162, 933 (1948).

22. W. T. J. Morgan and R. van Heyningen,Brit. J. Exp. Pathol. 25, 5 (1944).D. Aminoff, W. T. J. Morgan, W. M. Wat-kins, Biochem. J. 46, 426 (1950); E. F. Anni-son and W. T. J. Morgan, ibid. 50, 460(1952); and 52, 247 (1952); R. A. Gibbonsand W. T. J. Morgan, ibid. 57, 283 (1954).

`4. Details of isolation and purification ofblood-group specific substances from ovariancyst fluids are given by W. T. J. Morgan,Methods Carbohydrate Chem. 5, 95 (1965).

25. A. Pusztai and W. T. J. Morgan, Biochem.J. 78, 135 (1961); H. M. Tyler and W. M.Watkins, unpublished observations.

26. Z. Dische, Ann. N.Y. Acad. Sci. 106, 259(1963).

27. A. Pusztai and W. T. J. Morgan, Biochem.J. 88, 546 (1963).J1.Landsteiner, Biochem. Z. 104, 280 (1920);the Specificity of Serological ReactionsHarvard Univ. Press, Cambridge, 1947),pp. 182-196.

29. W. M. Watkins and W. T. J. Morgan,Nature 169, 852 (1952); W. T. J. Morgan

and W. M. Watkins, Brit. J. Exp. Pathol.34, 94 (1953); E. A. Kabat and S. Lesko-witz, J. Amer. Chem. Soc. 77, 5159 (1955).

30. W. M. Watkins and W. T. J. Morgan, Na-ture 175, 676 (1955).

31. , ibid. 178, 1289 (1956); P. Z. Allenand E. A. Kabat, J. Immninol. 82, 340 (1959).

32. R. Kuhn, Angew. Chemn. 60, 23 (1957).33. W. M. Watkins and W. T. J. Morgan,

Nature 180, 1038 (1957); Vox Sang. 7, 129(1962).

34. W. T. J. Morgan and W. M. Watkins,Nature 177, 521 (1956); W. M. Watkins andW. T. J. Morgan, Adca Genet. Statist. Med.6, 521 (1957); P. C. Brown, L. E. Glynn,E. J. Holborow, Vox Sang. 4, 1 (1959).

35. S. Iseki and S. Masaki, Proc. Japan Acad.29, 460 (1953); W. M. Watkins, Biochem. J.64, 21P (1956); S. Iseki, K. Furukawa, S.Yamamoto, Proc. Japani Acad. 35, 513(1959); W. M. Watkins, M. L. Zarnitz,E. A. Kabat, Nature 195, 1204 (1962); G. J.Harrap and W. M. Watkins, Biochem. J.93, 9P (1964).

36. W. M. Watkins, Bull. Soc. Chim. Biol. 42,1599 (1960); , Immunology 5, 245(1962); K. Furukawa, K. Fujisawa, S. Iseki,Proc. Japan Acad. 39, 540 (1963).

37. R. C6t6 and W. T. J. Morgan, Nature178, 1171 (1956); G. Schiffman, E. A.Kabat, S. Leskowitz, J. Amer. Chem. Soc.84, 73 (1962); T. J. Painter, W. M. Wat-kins, W. T. J. Morgan, Nature 193, 1042(1962).

38. G. Schiffman, E. A. Kabat, W. Thompson,Biochemistry 3, 113, 587 (1964); V. P.Rege, T. J. Painter, W. M. Watkins, W. T. J.Morgan, Nature 203, 360 (1964); , ibid.204, 740 (1964).

39. V. P. Rege, T. J. Painter, W. M. Watkins,w. r. J. Morgan, Nature 200, 4906 (1963).

40. I. A. F. L. Cheese and W. T. J. Morgan,ibid. 191, 149 (1961); T. J. Painter, W. M.Watkins, W. T. J. Morgan, ibid. 199, 282(1963).

41. V. P. Rege, T. J. Painter, W. M. Watkins,W. T. J. Morgan, ibid. 204, 740 (1964).

42. , ibid. 203, 360 (1964).43. K. Lloyd and E. A. Kabat, Biochem.

Biophys. Res. Commun. 16, 5 (1964).

44. T. J. Painter, W. M. Watkins, W. T. J.Morgan, Nature 206, 594 (1965).

45. J. Koscielak and K. Zakrewski, in Inter-national Symposium on Biologically ActivleMucoids (Polish Acad. of Sciences, Warsaw,1959), p. 21; J. Koscielak, Nature 194, 751(1962).

46. W. M. Watkins, J. Koscielak, W. T. J.Morgan, Proc. Intern. Congr. Blood Trans-fusion, 1962, 9th, Mexico City (1964), p. 213.

47. T. Baranowski, E. Lisowska, A. Morawiecki,E. Romanowska, K. Strozecka, Arch. Im-munol. Terapii Doswiadczalnej 7, 15 (1959);K. Stalder and G. F. Springer, Fed. Proc.19, 70 (1960); R. H. Kathan, R. J. Winzler,C. A. Johnson, J. Exp. Med. 113, 37 (1961);G. M. W. Cook and E. H. Eylar, Biochim.Biophys. Acta 101, 57 (1965).

48. E. Klenk and G. Uhlenbruck, Z. Physiol.Chemw. 319, 151 (1960).

49. G. F. Springer and N. J. Ansell, Proc. Nat.Acad. Sci. U.S. 44, 182 (1958); 0. Makelaand K. Cantell, Ann. Med. Exp. Biol. Fen-niae Helsiniki 36, 366 (1958); E. Romanow-ska, Arch. Imtmunol. Terapii Doswiadczalnel7, 749 (1959).

50. G. Uhlenbruck, Z. Immunitatsforsch. 121,420 (1961).

51. - and M. Krupe, Vox Sang. 10, 326(1965); E. Romanovska, ibid. 9, 578 (1964).

52. G. L. Cameron and J. M. Staveley, Nature179, 147 (1957).

53. W. T. J. Morgan and W. M. Watkins,Pr-oc. Intern. Congr. Blood Transflusion, 9thMexico, 1962 (1964) p. 225; W. M. Watkinsand W. T. J. Morgan, ibid., p. 230.

54. Two different systems of notation are incommon usage to describe the antigens inthe Rh system. In one system Rh, is equiva-lent to D in the other system. For a fulldescription of Rh nomenclature see Race andSanger (1, p. 135).

55. M. C. Dodd, M. J. Bigley, V. B. Geyer,Science 132, 1398 (1960); W. C. Boyd andE. Reeves, Nature 190, 1123 (1961).

56. M. C. Dodd, N. J. Bigley, G. A. Johnson,R. H. McCluer, Nature 204, 549 (1964).

57. Personal communication.58. R. R. Race, Ann. N.Y. Acad. Sci. 127, 884

(1965).

In the debate over the proposed newcopyright law and its possible impacton the future of computer-based infor-mation systems, only two points havethus far emerged clearly and incontro-vertibly: (i) new legislation is very badlyneeded; (ii) the new law, though itshould adequately protect the ownersof copyrights, must not be so stringent

The author is chairman of the board of theMcGraw-Hill Book Company, 330 West 42Street, New York, New York.

8 APRIL 1966

as to restrict the development of com-puterized information systems, particu-larly in science and applied science.Nevertheless, the Congress may soondeal decisively with this matter in act-ing upon the new copyright bill whichis now before both houses.To date there has not been enough

debate, either public or private, on theproblems and issues involved. It maybe useful, therefore, to have an analy-sis of them by a book publisher, though

an admittedly biased interest in the mat-ter may be displayed here. But evena publisher can strive for objectivity inconsidering certain long-range involve-ments of the future welfare of scienceinformation and hence of science itself.A basic requirement of the new law

is to provide for copyright security ina work first produced by means of, orwith the aid of, an automated mech-anism such as a computer. This re-quirement has to be dealt with de novobecause there is nothing in the presentcopyright law-enacted in 1909 andnot overhauled since-that recognizesthis kind of production. The pressingneed to satisfy this requirement is sug-gested in a paragraph in the AnnualReport (draft copy) of the Registerof Copyrights for the Fiscal Year 1965:

As computer technology develops andbecomes more sophisticated, difficult ques-tions of authorship are emerging. In pastyears the Copyright Office has received anapplication for registration of a musicalcomposition created by computer. Thisyear copyright was claimed for an abstractdrawing, and for compilations -of variouskinds, which were at least partly the"work" of computers. It is certain that

181

Computers and Copyrights

Restrictions on computer use of copyrighted materialwould protect authors, publishers, and even users.

Curtis G. Benjamin

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from

Blood-Group SubstancesWinifred M. Watkins

DOI: 10.1126/science.152.3719.172 (3719), 172-181.152Science 

ARTICLE TOOLS http://science.sciencemag.org/content/152/3719/172

REFERENCES

http://science.sciencemag.org/content/152/3719/172#BIBLThis article cites 69 articles, 8 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

trademark of AAAS. is a registeredScienceAdvancement of Science, 1200 New York Avenue NW, Washington, DC 20005. The title

(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for theScience

1966 by the American Association for the Advancement of Science

on Decem

ber 2, 2020

http://science.sciencemag.org/

Dow

nloaded from