Jaf€¦ · o1.01 ofl 5.0-e 4.0-°3.0-cn 2.0-1.0 0 5.0-e4.0-°3.0-n9 2.0-co 5.01-24.0-°3.0-]. 2.01 o 5.0-e 4.0-°3.0-cy 2.0-a an as jaf 0.01m 002m 0.10 m a. an ago mm 0.01 m a. a.

ISOAGGLUTININS ASSOCIATEDWITH ABOERYTHROBLASTOSIS*By S. KOCHWA,R. E. ROSENFIELD, L. TALLAL AND L. R. WASSERMAN

(From the Department of Hematology, The Mount Sinai Hospital, New York, N. Y.)

(Submitted for publication November 14, 1960; accepted February 2, 1961)

The occurrence and severity of erythroblasto-sis fetalis, due to maternal Rh isoimmunization,can be predicted through simple prenatal sero-logical tests (1, 2). If Rh isoagglutinins are pres-ent in maternal serum, and if the incompatible Rhantigen is transmitted from the father, the new-born will have hemolytic disease, the severity ofwhich is often proportional to the titer of themother's serum Rh isoagglutinins (1, 2). In theevent that the isoimmunized Rh negative motherdelivers an Rh negative infant, the titer of Rhantibodies in the cord serum, except for salineagglutinins, is generally equal to the titer obtainedin the maternal serum (3). For practical pur-poses, almost all maternal Rh isoagglutinins seemto cross the placental barrier freely, and theirtitration value affords guidance toward the ef-fective management of hemolytic disease of thenewborn. This has not been the case for erythro-blastosis fetalis due to ABOincompatibility (1, 2).

ABO erythroblastosis of sufficient severity towarrant exchange transfusion therapy for the con-trol of hyperbilirubinemia, has a frequency equalto that of erythroblastosis due to Rh incompati-bility (2, 4), and yet prenatal tests have been oflittle value thus far in identifying those motherswho may deliver affected infants (4). Since ABOerythroblastosis occurs often in the first incom-patible pregnancy (5, 6), a reliable prenatal testwould be of extreme importance in anticipatingthe occurrence of this disease for intelligentmanagement.

A varying proportion of maternal anti-A (a)and anti-B (,/) isoagglutinins can be demonstratedin the serum of the newborn (7, 8). Most ABOisoagglutinins cannot cross the placenta and there-fore play no role in the pathogenesis of erythro-blastosis. Fudenberg, Kunkel and Franklin (9)

* Presented in part at the Annual Meeting of theAmerican Society of Hematology, November 30, 1960.This investigation was supported in part by Grant H4456from the National Institutes of Health, and by the AlbertA. List, Frederick Machlin and Anna Ruth LowenbergResearch Funds.

suggested that 7S isoagglutinins traverse the pla-cental barrier, thus predisposing to hemolytic dis-ease. If so, diffusion of antibodies across theplacenta may be dependent upon the molecularsize of the antibody molecule, with low molecularweight (about 160,000) Y2 antibodies of the motherreaching the fetus (9).

Of equal importance in the pathogenesis of ABOhemolytic disease of the newborn are the protec-tive effects of nonerythrocytic ABO blood groupsubstances in the fetus. Whereas Rh antigens ap-pear to be restricted to the red blood cells, ABOantigens occur in other tissues, as well as in thesecreted mucopolysaccharides of most persons(10), and ABO isoagglutinins possess varyingdegrees of sensitivity to inhibition by specific solu-ble blood group substances (SSBGS) (11, 12).An excess of ABO secretors has been observedin infants with ABO erythroblastosis (5, 6, 13,14) and in their fathers (14). This excess mayreflect either a sensitivity of ABO secretors, ora resistance of ABO nonsecretors, toward ABOdisease, or even a loss of ABO nonsecretors inutero.

Since blood group substances are readily de-monstrable in the tissues of both secretors andnonsecretors (15), further quantitative and/orqualitative distinctions must exist between thesetwo classes to explain these observations. Re-sistance to neutralization of some a and ,B isoag-glutinins might be related to the clinical problemof ABO erythroblastosis, if such antibodies were7S y2-globulins crossing the placenta (9). Largeantibody molecules (macroglobulins), because theyfail to traverse the placental barrier, are not perti-nent to this problem (9).

Column chromatographic separation of serumproteins on anion exchanger, diethylaminoethyl(DEAE) cellulose, first described by Sober, Gut-ter, Wyckoff and Peterson (16, 17), offers a rela-tively simple means of separating 7S /2-globulinsfrom other serum proteins (18). Such separa-tion of human isoagglutinins has been reported by

874

ISOAGGLUTININS ASSOCIATEDWITH ABO ERYTHROBLASTOSIS

Abelson and Rawson (19) and by Fahey andMorrison (20).

The present report deals with the resistance toneutralization by SSBGSof ABO isoagglutininswhich can traverse the placental barrier. Theseisoagglutinins were obtained by a simplifiedDEAEcellulose column chromatographic method,designed to separate 7S 7y-globulins from the re-mainder of the serum proteins. The results ob-served in the presence and the absence of ABOerythroblastosis suggest that the disease might beanticipated by suitable prenatal tests.

MATERIALS ANDMETHODS

Chromatographic separation was performed accordingto a modified method of Sober and Peterson (17) and ofLevy and Sober (18). In general, the following pro-cedure was used for separation of serum: DEAE cel-lulose 1 was purified according to Sober and co-workers(16) and the slurry, suspended in the initial buffer (0.02M phosphate, pH 6.3), was poured into 5-ml serologicalpipets (25 X 0.5 cm). The ion exchanger was packedunder water pressure of 40 to 50 cm and washed withthe initial buffer overnight. The dead volume of suchcolumns was about 2.5 ml.

The serum sample, dialyzed2 against 3 changes of theinitial buffer in a 4° C cold room, was applied to the col-umn, together with the slight precipitate formed duringdialysis. For each separation, 1 ml serum (correctedfor the increase of volume during dialysis) was appliedto the column at room temperature, and washed with 10ml of the initial buffer, followed by 10 ml of 1 M NaCl.Two-ml samples were collected at a flow rate of 6 to 8ml per hour, at room temperature. The protein contentof the separated fractions was measured by absorptionat 280 mg in a Beckman DU spectrophotometer.

Under these conditions, the 0.02 M phosphate buffer(pH 6.3) fraction (hereafter referred to as 0.02 M frac-tion) contained the first four chromatographic peaks de-scribed by Humphrey and Porter (21), and if isoag-glutinins were present in this fraction, the highest titerswere always found in those tubes which contained mostprotein. No difference was found in the elution patternof the isoagglutinins when 0.01 M phosphate buffer wasfollowed by 0.02 M buffer, or when more than 10 ml ofbuffer was used as first eluent (Figure 1). The 0.02M fraction was not found to contain proteins otherthan y2-globulins when tested by paper electrophoresisand acrylamide electrophoresis (22). Ultracentrifuga-tion of this fraction at a protein concentration of 1 percent (courtesy of Drs. 0. Singher and R. Fenichel,Ortho Research Foundation, Raritan, N. J.) revealedonly a single component having a sedimentation con-

1 Eastman Organic Chemicals, Rochester, N. Y.2 Cellulose casings, Visking Co., Chicago, Ill.

50-E40-0 3.0-N 2.01o1.01

Ofl

5.0-E 4.0-° 3.0-CN 2.0-

1.00

5.0-E4.0-° 3.0-N9 2.0 -

co

5.01-24.0 -° 3.0-]

. 2.01o

5.0-E 4.0-° 3.0-CY 2.0-a

anas

Jaf0.01 M 0 02 M 0.10 M

a. anago mm

0.01 M

a.a.

anNo

0.02 M I M

a

0.02 M I M

U

U

875

I M

0.02 M I M

I LOCATION OFot AND,8 ISOAGGLUTININS

m ONLY PARTIALLY

INHIBITABLE BY SSBGSI READILY INHIBITABLE

BY SSBGS

0o.02 M I MI

0 10 20 30 40ml. OF EFFLUENT

50

FIG. 1. SEPARATION OF ISOAGGLUTININS ON DEAECELLULOSECOLUMNSUNDERVARYING CONDITIONS. Varia-tions of the concentration (0.01 to 0.02 M) of the phos-phate buffer, pH 6.3, and of the volume of effluent col-lected (10 to 30 ml), failed to alter the pattern of 7S Y2-globulins or a and j8 isoagglutinins emerging from thecolumn. The a and is isoagglutinins in the fractionseluted with 1 M NaCl could be further separated byusing 0.1 Mphosphate buffer, pH 6.3, prior to 1 MNaCl.

stant of 7S. Heavier components were not observeddespite special attention for their detection in the earliestparts of the ultracentrifugal run.

The second fraction eluted with 1 M NaCl (hereafterreferred to as 1 M fraction) contained 19S macro-globulins and all other serum proteins, but paper andacrylamide electrophoresis failed to reveal y2-globulins.

I~~aI N

S. KOCHWA,R. E. ROSENFIELD, L. TALLAL AND L. R. WASSERMAN

The isoagglutinin activity of this fraction could be splitinto two fractions using, after 0.02 M phosphate buffer,pH 6.3, first 0.1 M phosphate buffer, pH 6.3, followedby 1 M NaCl (Figure 1). The significance of this ob-servation is under study.

The isotonicity of 0.02 M phosphate buffer fractionswas restored by addition of 0.05 ml of 18 per cent NaClper 1 ml of eluate, and of 1 M NaCl fractions by dilu-tion with 5 vol of distilled water.

Native serum and pooled fractions, the latter concen-trated to the original serum volume by dialysis at 4° Cagainst 20 per cent polyvinylpyrrolidone (PVP), pro-vided materials for titration of antibodies before andafter separation. Parallel titrations showed a good re-covery of the isoagglutinins (Table I).

The agglutination tests for a and fi isoagglutinins weredone by conventional methods, usually in duplicate. Two-fold serial dilutions of serum or fractions were mixedwith equal volumes of 2 per cent washed red cells fromtype A1 and B donors. After incubation at 370 C for 1hour and centrifugation, agglutination was recorded asthe saline titer. An additional volume of 4 per cent acacia(5) was then added and the mixture reincubated for 15minutes at 37° C, centrifuged, and 1 vol of saline addedto abolish rouleaux. The agglutination then observed,usually by two observers, neither familiar with the clini-cal details, was recorded as the acacia titer. This con-centration of acacia was chosen because it afforded re-sults which agreed within one dilution of those observedwith the y-antiglobulin test.

The neutralization tests for a and 8 isoagglutinins wereperformed by mixing serial dilutions of 0.02 M phosphatebuffer fractions with suitable serial dilutions of SSBGS

TABLE I

Recovery of ABOisoagglutinins after fractionationon DEAEcellulose columns

Isoagglutinin titers (reciprocal)Serum Anti-

no. body Native 0.02 M* 1 Mt Pool:

a 80 0 80 803

,8 160 0 80 80

a 1,280 160 640 1,28021

,3 640 40 320 640

a 640 80 320 64042

,B 160 4 80 80

a 160 4 80 80140

160 0 160 160

* Titer of pooled fractions that were eluted with 0.02 Mphosphate buffer, pH 6.3, and concentrated to originalserum volume.

t Titer of pooled fractions that were eluted with 1 MNaCl and concentrated to original serum volume.

t Mixture of * and t corrected for volume change due tomixing.

5.0- 12000~~~~~~~~W z~~-

6 2.00 / 806

1.0- 40

0.02 M I M

5.02 MOTHER2 200

4.0- FROM C

o0 aN~~~~~~~~

6 2.0- 8Oi

repesete 40reircl ausofog A) nd orsI'B rgi. Th-eselswr

.0 0

0. 2 4 6 8 10 1214 16 1820ml. OF EFFLUENT

FIG. 2. THE RECOVERYOF MATERNALAND FETAL ABOISOAGGLUTININS FROMDEAECELLULOSECOLUMNS. Bothmother (no. 268) and newborn (no. 265) type 0; 1 mlserum samples fractionated. Isoagglutinin titers ofwhole serum: mother, a 1: 512 and t 1: 512; newborn,a 1:t160 and u 1:a16. Isoagglutinin titer in figure isrepresented by reciprocal values.

of hog (A) and horse (B) origin.3 The test cells wereadded after 15 minutes' incubation at room temperature.After further incubation for 1 hour at 370 C and centri-fugation, agglutination was recorded as saline titer. Then4 per cent acacia was added to the tests which werefurther incubated for 15 minutes at 370 C. The agglu-tination observed after centrifugation and addition of sa-line was recorded as the acacia titer. The acacia titerwas always found to be more reliable and is giventhroughout this report. Titers with and without SSBGSwere frequently checked for reproducibility at a laterdate and consistent results were always obtained.

RESULTS

1. The separation of cord serum a and /3 isoag-glutinins on DEAEcellulose columns. ABOiso-agglutinins have been demonstrated in cord bloods,especially those from type 0 offspring of type 0mothers. Cord serum ABO isoagglutinins could

3 Biosales Inc., 300 W. 43rd St., New York, N. Y.

876

ISONAGGLUTININS ASSOCIATED WITH ABO ERYITHROBLASTOSIS

7T'Ie iecIz,, VoX cot'cod bltoidf'roms DEA,

I().}4462

1 1 714314414514614 71485(

15. 215.154155S1 5O1 8(1

IhI( o ttI w

MIothler (Chtild

C) ()(~)(I (0I(1 (1)0) (1() (I

() 0)(I (I() (I() Is

* (ord sera selcte(l to haxt Card blond isoitggluttittitl

of these atnd 17 ot her cord St

+ Incoatcentrateil sitigle 1)wh'lichi conttaitned the tost i-ajustted taor diluttion over tlh

lbe of 75 characterllete reflections of cill the sertim of the

Thirty-three cordiSoagg1,lutilllins, wvere

coltumns, 7I11)1 elluateM1etho(ls. The i5l07recovered ((nll- ill tIfractions, un(ler theIlOt reveal i s oelakRarely. alnl only- NV

ill ex;eNss of I 200conlceniltra tionl of I

irregular trace amlapprlximatcly 1 1P c

ierhaps to contauniSixteen sera wve:

titer of 1: 20 or gre

lel 1,v titration of1111ic 0.0 2 A frac

the isoatgglutilninlto 1)o11l or to conc(anldtilIollies invariabhavillng MlSt of thetionl Clvavs ceontaina titer (qtlite simi'lairtim ( Figure 2).f(o)rilm illI pacrallelsonmetlilles va,.,rried.

TABLE II results and shows the good recovery of cord se-iso(ioglutlfinli in 0.02 1j c/autes t rum Magglutinins in single 0.02 AM fractions fronm

E cellulose cclumns + DEAEcellulose columns.2 ?. Co011pari'sont of thv

S a (and 13 'so(yyiugtldiniIOaggltitinliii titer- ,reciprocall) f m tt-tcrtSrO matornal anid cord scra-. If the a and /3Na.tive corde 0).02 M\sraction isoanrlutinils of cord sera are 7S y2-globulins

\nti-A\ .\ti'ti Atnti-A Aniti-l that can be recovered in 0.02 MI fractions from

1(1( 1( 40 <1( DEA\E cellulose colidumus, and if these isoagglu-8(o 160 40 4080) <10) 20 <1(1 tintins are of maternal origin, haviigdiffuseld40 <11) 40 <1(1680 <1( 40 <1( across the placental barrier, the 0.02 -I fractions160 <10 811 <10810 <1(4 80( <1( ot maternal and cord serum palirs (collected at the160 <1(1 8(1 <1i o-) should l)e exl)ecte(l to have ileI)-40 4(0 4(1 40 time (If delivery)shol eepce ohv d

I(s 111 140 <1(0 tical titer values if e 1ilibritim is attainee(.2( <1 20<1(1 <10 Ight pairs of maternal aid cor(i sera, all typ)e4(0 2(0 41) 2(181) <1( (1 <1( () all( Contalillilnl a an(l /3 isoagglutinins, were8( 4(0 fractionated on I)EAE cellulose columlis. The

ve >I:10 atti-A. 0.02 Al fractiolns aindl the naitive miaternal seratIn~t nbervel in te 0.1 tn 1.0 M eluates -vere titratedl for a an(l 13 isoagglutilnis. and theratested. L)~1

(.02 MI phosphiate hbtffer, pH 6.3. eltiatae 1eut /simrzdi ale11 hwta h-zoagglitits were titrateil atid the \alites resultS isumm arized in Table 111) sho that theleX-litiie ot the 1,tiaVe SetiTli. 0.0 MAI fraction of the (a and /3 isoagglutinin ac-

tivitv in maternal serumi is dlupllicate(l in the cord(9), because they are incom- serum. The quanititative data o)f one of these ma-

an(l /3 isoagglutininls present ternal-cord pairs, showvn in Figure 2. reveals thatmother. these isoagglutinins, as well as other 7S 72-globu-

sera. each containing a and /3 lins without isoagglutiniin activity, exist in equi-sel)arate(l oIl DEAEcellulose librium On both sides of the l)1acenta.

Is Collected as (lescriled under 3.iic11 Occilbothices of a (1*1(/ /3ipo(ZQceliftiilla iSlLggltitinins of these sera w-ere 1ic 0.02 1Il fractiols of 110 prcnlata/ spc(cimnelns.lie 0.02 M fractions; the 1 ', Study of the cord sera of compatible pregn-lancies

stan(larl test conditions, did (7) lhas furnished some information concerninigtininl activ ity (Figuire 2) . lplacenita-crossiig ABC) i soagglutinins, but there

hen some cord sera contalined lhas been no method thus far to study such anti-titewr of)t ABO is(aoIT~ltltitlillsS lfb es in unselecte( 1)reon-xncies ill adv-ance of1 NaEl fraction~s :onltained (leliverv. The 0.02 'M fractions of DEAEK cellu-

Liunts (1lof Iagglutllills uip tor cent of the native titer, due TABLE Illnation with 7 ,-globulins.Ct2omparison of the (a (mrid 0 isoagglultinin titers * of mtalte(rnal/

re found to have anti-A in a (and corld sera obtained (It delivery t

.ater, andc these sera w-ere stud-the native material and that

tion \vhich contained most ofctivitv. There was no need

nintrate fractions, because these'v appeared witlh the effluent

/ S y/-globilihns, and one frac-ed most of the isoagglutinins inr to that of the native cord se-

'itrations Were seldom per-and the test red cells wNereTlal)le IT summarizes these

'1 it('ill'll110.

26 '4556 '6127 '61107 1'10

135 '43139 '157169 171265 '268

Isiatglutitiit titer recipr)ioc al)

MaIatertnalnifait 0.02 SI (.()2 MI Nati

Irast t II f ract i1n terna

16(1 S(1 1 6(1 16( 6404(1 4 4(1 21) 16020 <2 2(0 <2 80s() 40 4(0 80( 32(020 20 20 20 16020 40 20 40 1604(1 2 40 4 64064 8 64 8 512

itSt l-iii1l

0640160

80640160160

80512

* Itconcetttrated single 0.)2 M phosthalitre bUffer, plI. 6.3. eliltiteswhich cotitaitted the Itiost isoagglttinins were titrated, and the vtliieSadjit sted tar diltiotn over the valumne of the native seritit.

t t dienrsttiewleiborit all t\ ase (').

I 7

II

S. KOCHWA,R. E. ROSENFIELD, L. TALLAL AND L. R. WASSERMAN

TABLE IV

The distribution of the ABOblood types of the newborn of 110mothers of types 0, A and B

Blood type, motherBlood type,

newborn 0 A B

0 45 5 7A 20 20 0B 8 1 1

AB 0 2 1

Total 73 28 9

lose columns offered such an opportunity, and 110prenatal specimens of type 0, A and B patientswere fractionated. Table IV shows the distribu-tion of the blood types of the newborn subsequentlydelivered, and Table V reveals the distribution ofthe a and ,/ isoagglutinins in maternal 0.02 Mfrac-tions. To be detected by the screening tests em-

ployed, a titer of at least 1: 2 was required.In this series, isoagglutinins were not observed

in the 0.02 M fractions of the sera of 28 type Amothers and 9 type B mothers, but were observedin the sera of 53 out of 73 type 0 mothers. Devi-ation from our earlier results in study of compati-ble gestation (7) may be due to the dilution atwhich these tests were now performed. Whereasthe cord serum of compatible newborn can betested without dilution, the unconcentrated 0.02 Mfractions represented a dilution of at least 1: 2over the volume of the native serum that was

placed on the cellulose columns. A further slightdilution occurred when concentrated NaCl was

added for isotonicity, and some loss of antibodywas expected, because only one of several anti-body-containing, or potentially antibody-contain-ing, fractions was tested. This small study of0.02 M fractions revealed only that ABO isoag-glutinins in a titer of at least 1: 2 were observedin 69 per cent of 73 type 0 mothers and were

TABLE V

The occurrence of a and ,B isoagglutinins of at least 1:2 titerin the 0.02 Mphosphate buffer fractions of 110 mothers

of blood types 0, A and B

Blood type, mother

00 A B

a +18 24 0 0a only 27 0 0# only 2 0 0No isoagglutinins 20 28 9

not observed in 37 mothers of type A or B. Thisis consistent with the strong association of ABOerythroblastosis with mothers of type 0 (16).

Furthermore, it is of interest to note (Table V)that anti-B was observed in only 26 of 73 type 0mothers, whereas anti-A was found in 51 of thesame mothers. In addition, /8 without a was ob-served only twice, but a without ,8 was encountered27 times. This preponderance of a activity mightwell be associated with the preponderance oferythroblastosis due to A incompatibility (7).

4. Relative inhibition by specific soluble bloodgroup substances of the isoagglutinins in the 0.02Mfractions of 53 type 0 mothers. Table V re-veals that the placenta-crossing a and /8 isoagglu-tinins contained in 7S 72-globulin fractions oc-cur commonly, but clinical ABO erythroblastosisis observed in only 1 of 200 newborn (23). Thisdifference could be due to the protective effect ofextraerythrocytic ABOantigens.

Witebsky (11) suggested that "non-inhibitable"ABO isoaglutinins might be related to ABOerythroblastosis, but Zuelzer and Kaplan (6)noted a poor correlation. However, the study oftotal serum isoagglutinins (12) would not be in-formative, since the bulk of a and ,/ antibodies are19S (20) (Table V), which not only fail to crossthe placental barrier, but could confuse the results.

The erythrocytic, tissue, and mucopolysac-charide ABO antigens, even of animal origin,share common structural groupings. On theother hand, agglutinins which are resistant toneutralization by large quantities of SSBGSper-mit a distinction between erythrocytic and extra-erythrocytic ABOantigens. Therefore, diffusiblenoninhibitable maternal isoantibodies, even of lowtiter, could be responsible for clinical ABOeryth-roblastosis. The 0.02 M cellulose column frac-tions of the isoagglutinins of 53 type 0 mothers,presented an opportunity to attack this problemdirectly.

Titration of the ABO isoagglutinins in the0.02 M fractions of these 53 mothers yielded val-ues ranging from 1: 4 to 1: 640, and little if anycorrelation with the native serum titers (Table II).Of more importance were duplicate titrations towhich SSBGS were added to each tube of thetitration series. The concentration of the SSBGSwas varied from 1: 10 to 1: 500 for A substance,and from 4: 1 to 1: 500 for B substance, to deter-

878

ISOAGGLUTININS ASSOCIATEDWITH ABO ERYTHROBLASTOSIS

TABLE VI

Relative inhibition by specific soluble blood group substances of the isoagglutinins in the0.02 Mphosphate buffer fractions of 53 type 0 mothers

Number of mothers with noninhibitable isoagglutininsBlood No. of No. of isoag- Highest dilution of SSBGSwhich fails to inhibittype, mothers glutinins tested 1 1 1 1 1 1 1

newborn tested Soo 100 TO20 0 T 4

a 11 7 3 1A 13

(3 5 2 3

a 4 3 1*B 4

,B 4 3 1*

a 36 21 10 2 3t0 36

(8 17 2 7 5 3t

All 53 a 51 31 13 1 2 4 0 0(3 26 7 10 0 0 5 0 4

* Same case; child had severe ABOerythroblastosis requiring transfusion.t Only four sera represented; two contained both noninhibitable a and (3, one contained only a and another only (3.

mine that concentration which just failed to in-hibit visible agglutination. These results, cor-related with the ABO type of the newborn, aresummarized in Table VI. The horse B substanceused for these tests inhibited anti-B much less ef-fectively than the hog A substance inhibited anti-A.

Five prenatal sera so studied for placenta-cross-ing antibodies were found to contain a and/or pagglutinins, neither of which could be neutralizedby SSBGSat a concentration of 1: 10 for A and4: 1 for B. Four of these patients delivered typeO newborn, but one delivered a type B child whichrequired exchange transfusion therapy for se-vere ABO erythroblastosis.

In 200 deliveries, approximately 90 type 0mothers can be anticipated, and in the small seriesshown in Tables III, IV and V, 73 type 0 motherswere studied. Therefore, 0.8 case of ABO dis-ease was expected (23), and one case was en-countered. Type 0 mothers have incompatibleoffspring of types A, and B in approximately 30per cent of instances (the sum of the gene fre-quencies of Al and B), so that for every case oferythroblastosis, an additional two to three un-selected type 0 women should have similar ABOisoagglutinins which are of no importance to new-born of types 0 and A2 (23). In the present se-ries, four such women were encountered. One ofthese patients happened to be primagravida, con-sistent with the fact that ABOdisease may be ob-served in the first incompatible pregnancy.

Table VI also shows that incompatible infantswithout evidence of hemolytic disease were de-livered from 17 type 0 mothers who had demon-strable isoagglutinins in their 0.02 M fractions.In 15 instances these antibodies had the specificityof the A or B antigen found on the cord erythro-cytes, and the titration values ranged from 1 :4to 1: 160. The mothers of 10 incompatible new-born were found to have such incompatible isoag-glutinins with a titer of at least 1: 20. However,only one infant of this entire group had clinicaldisease, and the isoagglutinins in the 0.02 M frac-tion of this particular mother were highly resistantto inhibition by SSBGS, whereas the other ninewere easily neutralized. Thus the inhibitable iso-agglutinins in 0.02 Mfractions, regardless of titer,did not appear to cause erythroblastosis, whereasthe noninhibitable isoagglutinins so separated,even of low titer, furnished the expected correla-tion.

5. The nature of the isoagglutinins in the seraof mothers who have had infants with severe ABOerythroblastosis. Isoagglutinins eluted in 0.02 Mfractions and resistant to inhibition by SSBGSarenot common (Table VI), having been observedin only 5 of 110 prenatal specimens of blood types0, A and B. However, if such isoagglutinins areresponsible for ABO erythroblastosis, they mustoccur in the sera of all women who have a his-tory of children with severe clinical disease. Sevenpatients with typical histories, including the case

879

S. KOCHWA,R. E. ROSENFIELD, L. TALLAL AND L. R. WASSERMAN

TABLE VII

Inhibition by hog (A) and horse (B) specific soluble blood group substances of the 0.02 Mphosphate buffer fractionsof maternal a and j3 isoagglutinins when there is a history of severe ABOerythroblastosis *

Titer (reciprocal) of 0.02 M fraction in the presence of SSBGS

a agglutinins agglutinins

ABO Amount of A substance added Amount of B substance added Titer oftype of native serumaffected Case 1 I1 I I I I 1 4

child no. °t 500 100 40o 2-0 1-0 Ot 00 -100- 4-0 20 10 I a

A 21 160 ndt 40 40 20 4 40 nid 40 40 20 20 0 1,280 320A 22 80 20 20 nd nd 4 2 0 0 nd nd 0 0 160 20A 42 160 160 80 nd nd 20 4 2 2 nd nid 0 0 640 160A 64 80 80 20 nd nd 6 20 20 6 nd nd 2 0 2,560 160A 94 160 nd 80 20 4 4 2 0 0 0 0 0 0 2,560 640A 184 160 nd 20 10 4 2 160 nd 80 80 40 20 0 1,280 2,560B 141 40 20 10 nd nd 4 40 40 20 nd nd 4 2 320 320

* All mothers, type 0.t No SSBGSadded; titer of isoagglutinins.t Not done.

from Table VI, were recalled for study, and theresults are shown in Table VII.

The sera of all of these women with a historyof severe ABOerythroblastosis in previous preg-

nancies revealed type-specific "noninhibitable"ABO isoagglutinins in a titer of at least 1: 2 intheir 0.02 Mfractions. All newborn in this seriesrequired exchange transfusion treatment or de-veloped kernicterus because treatment was notavailable. One mother (no. 42) had delivered herlast child eight years previously, thereby revealinghow persistent these antibodies may be.

6. The nature of the isoagglutinins in the serum

of unselected type 0 donors immunized withSSBGS. Increasing severity of ABO erythro-blastosis in later deliveries is seen occasionally.Significant increase in maternal ABO isoagglu-

tirlins has not been observed during the course ofan incompatible gestation nor after the delivery ofa nonsecretor infant (6). On the other hand, titerin reases have been noted post partum followingthe delivery of some ABO incompatible secretors(6), presumably due to the SSBGS content ofamniotic fluid. This raises the question as towhether SSBGScan incite noninhibitable isoag-gluitinins that can be eluted in 0.02 M fractions.

For information on this problem, the sera offive unselected type 0 donors, who had been im-munized with both A and B SSBGSfor prepara-

tion of anti-AB reagents (high titer type 0 sera),were fractionated on DEAE cellulose columns.The SSBGS which had been used for the im-munization of these donors were of the same

source and manufacture as those employed in the

TABLE VIII

Inhibition by hog (A) and horse (B) specific soluble blood group substances of the 0.02 Mphosphate bufferfractions of the a and,3 isoagglutinins of unselected donors previously immunized with the same SSBGS

Titer (reciprocal) of 0.02 M fraction in the presence of SSBGS

a agglutinins P agglutinins

Amount of A substance added Amount of B substance added Titer ofnative serum

Case 0* 1 1 1 1 1 * 1 1 1 1 1 4no. 500 1w00 30 O2 10 ° - 100 40i 2i0 1i0 I a

250 40 ndt 16 12 8 4 80 nd 80 80 80 40 10 2,048 2,048251 40 nd 20 16 16 10 80 nd 30 30 30 20 10 1,024 1,024253 60 nd 20 20 10 10 80 nd 80 40 40 10 8 512 2,048254 > 160 nd 160 40 20 20 > 160 nd > 160 > 160 < 160 > 160 8 2,048 2,048255 > 160 nd 30 20 20 4 > 160 nd > 150 80 80 30 8 >2,408 1,024

* No SSBGSadded; titer of isoagglutinins.t Not done.

880

ISOAGGLUTININS ASSOCIATEDWITH ABO ERYTHROBLASTOSIS

present inhibition studies. The 0.02 M fractionsof all five donor sera were found to contain non-

inhibitable a and ,/ isoagglutinins, resembling thosein the sera of the mothers of erythroblastic in-fants. The results are summarized in Table VIII.

DISCUSSION

Severe ABO erythroblastosis is familial andoften observed in the first ABOincompatible child(5, 6). Although almost all subsequent incom-patible infants are also affected, increasing se-

verity of the disease is not seen regularly (12, 14).Thus ABOdisease resembles Rh erythroblastosisin its familial incidence, but differs by not re-

flecting so much evidence of specific maternal iso-immunization by the products of conception.

Schiff (12) postulated that ABOerythroblasto-sis was to be expected when the maternal serum

contained a special ("Type VI") pattern of anti-bodies highly resistant to neutralization bySSBGS; but, to explain the absence of hemolyticdisease in the incompatible children of some typeO mothers of this description, he proposed pro-

tection against the cross-reacting antibody de-scribed by Rosenfield and Ohno (7). SinceSchiff studied whole maternal serum, his conclu-sions must be reviewed in the light of the prop-

erties of 7S maternal isoagglutinins.If maternal isoagglutinins are to be responsible

for erythroblastosis, they must not only cross theplacental barrier but combine with fetal erythro-cytes. In the case of ABO isoagglutinins, extra-erythrocytic antigens occur in many tissues inde-pendent of the ABOsecretor status (15) and ex-

ert a protective effect of considerable magnitude,as evidenced by the low frequency of erythroblas-tosis despite the common occurrence of diffusibleisoagglutinins (6, 8).

At least four components, with sedimentationconstants of 7S, 19S, 28S, and 44S, have beenfound in the y-globulin fractions of normal humanserum (24), and recently Fahey and Morrison(20) showed that all of the ABO isoagglutininshave a sedimentation value of either 7S or 19S.The "non-y-globulin" isoagglutinins reported byAbelson and Rawson (19) appear to have beendue to conditions of the separation (20). Abelsonand Rawson also reported that noninhibitableABO isoagglutinins were almost all located in 7Sfractions; but in all cases, and at the same con-

centration of SSBGS that they employed, wefound only a small part of the total isoagglutininsto be noninhibitable and to occur in both the 0.02M and the 1 M fractions if isoagglutinins werepresent at all in the former.

Table III reveals that most ABO isoagglutininsare not y2-globulins and do not cross the placenta.Poorly inhibitable isoagglutinins eluted in 1 Mfractions, although not shown in this report, wereobserved frequently and would contribute mis-leading information.

Placental permeability and resistance to in-hibition by SSBGS are certainly not the soleparameters of the problem of ABOerythroblasto-sis, but their contribution must be of primaryconsideration. Other contributing factors in-clude the expression of erythrocytic ABOantigensbecause A2 children tend to be spared (6) ; thesecretor status because erythroblastotic infantstend to be secretors (5, 13, 6, 14) ; and the abilityof the affected newborn to cope with a hemolyticsyndrome (4). These and, probably, other vari-ables must be studied. If the noninhibitable iso-agglutinins of 0.02 M fractions are responsiblefor ABOerythroblastosis, there is now an oppor-tunity to attack these secondary problems.

The present report raises an immediate prob-lem concerning the nature of noninhibitable iso-agglutinins shown to be associated with erythro-blastosis. If these antibodies can be incited bySSBGS (Table VIII), they should not representantigenic structural grouping differences betweenerythrocytes and SSBGS. Instead, these dif-ferences could reflect the heterogeneity of anti-body molecules with respect to the size of theirspecific combining sites (25, 26). Kabat (25, 26)has demonstrated that antidextran molecules varyconsiderably in the size of their combining sites,with the upper limit being sites complementary toterminal nonreducing chains of six- or seven glu-cose units in a-1,6 linkage. Thus, althougherythrocytes of types A, and B might agglutinatewith antibody molecules having combining sitesof any size, SSBGSmight form reversible com-plexes with some antibody molecules which couldbe released in sufficient amount to support visibleagglutination under selected circumstances. Al-ternatively, nonspecific anamnestic response is re-quired to explain the observations shown in TableVIII.

881

S. KOCHWA,R. E. ROSENFIELD, L. TALLAL AND L. R. WASSERMAN

The term "immune" for agglutinins which crossthe placenta, as opposed to "natural" saline ag-glutinins (8, 27), is not supported by the presentdata. All ABOagglutinins appear to be immuneresponses to antigenic structural groupings re-sembling those defined by Kabat (10) for humanblood group substances.

The nature of the distinction between A, and A2red cells is also subject to review. Anti-A1 re-agents of human, animal or plant origin are highlysusceptible to inhibition by SSBGS, and precipi-tate with SSBGS in common with anti-A (28).The distinction made by these reagents thus ap-pears to be on a quantitative basis: there is moreA on Al cells than on A2 cells. On the other hand,anti-A inhibited with SSBGStoward A2 cells, maypersist in agglutinating A1 cells. The problem ofthe specificity of noninhibitable anti-A, as relatedto the restricted occurrence of erythroblastosis intype A, offspring, is now under investigation.

The column chromatographic isolation of iso-agglutinins which can cross the placenta appearsto be a valuable method for the study of all formsof maternal isoimmunization in relation to erythro-blastosis. This method is now being used for itsapplicability to Rh problems, and the results ap-pear promising, but not identical with the reportof Abelson and Rawson (29).

The column chromatographic method requiresat least two days to perform and is not feasible forrapid diagnosis in the newborn period. For pre-natal tests, the method appears to be practical.

SUMMARYAND CONCLUSION

1. ABO isoagglutinins associated with the bulkof 7S 72-globulins were separated from other se-rum proteins by anion exchange column chroma-tography.

2. The ABO isoagglutinin activity of 33 cordblood sera was found to be restricted to 7S 72-glob-ulin-containing fractions.

3. Maternal serum ABOisoagglutinins were en-countered chiefly in macroglobulin-containingfractions, but paired specimens of maternal andcord sera obtained at the time of delivery revealedthat the 7S 72-globulins as well as the associatedisoagglutinins existed in equilibrium on both sidesof the placenta.

4. Out of 110 unselected prenatal sera, isoag-glutinins, in a titer of at least 1: 2 in acacia, were

eluted in 0.02 Mphosphate buffer, pH 6.3, in 53of 73 type 0 mothers, but not in 28 type A and 9type B women. Anti-A (a) was noted in 51 ofthese fractions, but anti-B (,8) in only 27. Thesefindings were considered to be consistent with therarity of ABO erythroblastosis in offspring ofmothers not of type 0, and with the excess oferythroblastosis due to anti-A.

5. When these 53 examples of chromatographi-cally separated isoagglutinins were tested for re-sistance to inhibition by specific soluble blood groupsubstances (SSBGS) (hog A and horse B), only5 contained markedly resistant antibodies. Fourof these mothers delivered type 0 children, but onemother delivered a type B child who had erythro-blastosis and required exchange transfusion ther-apy. An additional nine mothers in this seriesalso delivered ABO-incompatible newborn, andtheir 0.02 M fractions contained type-specific iso-agglutinins in a titer ranging from 1: 20 to 1: 160,but in each instance these antibodies were readilyinhibitable with SSBGS, and clinical erythroblasto-sis was not observed.

6. Six other womenwith a history of having hadone or more children with severe ABO erythro-blastosis, and five unselected donors who had beenimmunized previously with SSBGSwere all foundto have such type-specific noninhibitable isoag-glutinins.

7. The use of these methods in prenatal tests isimplied.

ACKNOWLEDGMENT

The authors wish to express their appreciation toProfessor E. A. Kabat for his valuable suggestions inthe preparation of this manuscript, and to Mrs. NatalieSchneiderman, Miss Shirin Hakim and Miss CharlotteSpitz for their technical assistance.

REFERENCES

1. Levine, P., Vogel, P., and Rosenfield, R. E. Hemo-lytic disease of the newborn. Advanc. Pediat. 1953,6, 97.

2. Allen, F. H., Jr., and Diamond, L. K. Erythroblas-tosis Fetalis, Including Exchange TransfusionTechnic. Boston, Little, Brown, 1958, p. 143.

3. Wiener, A. S. Rh factor in immunological reactions.Ann. Allergy 1948, 6, 293.

4. Rosenfield, R. E. Erythroblastosis and maternal iso-immunization in Medical, Surgical and Gyneco-logical Complications in Pregnancy, A. F. Gutt-macher and J. J. Rovinsky, Eds. Baltimore, Wil-liams & Wilkins, 1960, p. 386.

882

ISOAGGLUTININS ASSOCIATEDWITH ABO ERYTHROBLASTOSIS

5. Wiener, A. S., WVexler, I. B., and Hurst, J. G. Theuse of exchange transfusion for the treatment ofsevere erythroblastosis due to A-B sensitization,with observations on the pathogenesis of the dis-ease. Blood 1949, 4, 1014.

6. Zuelzer, W. W., and Kaplan, E. ABO heterospecificpregnancy and hemolytic disease; a study of nor-

mal and pathological variants. I-IV. A. M. A. J.

Dis. Child. 1954, 88, 158.7. Rosenfield, R. E., and Ohno, G. A-B hemolytic dis-

ease of the newborn. Rev. Hemat. 1955, 10, 231.8. Mollison, P. L., and Cutbush, M. Hemolytic dis-

ease of the newborn due to fetal-maternal ABOincompatibility in Progress in Hematology, L. M.Tocantins, Ed. New York, Grune & Stratton,1959, vol. 2, p. 153.

9. Fudenberg, H. H., Kunkel, H. G., and Franklin, E.C. High molecular weight antibodies. Proc.VIIth Congr. Int. Soc. Blood Transfusions, L.Hollander, Ed. Basel, S. Karger, 1959, vol. 2, p.

522.10. Kabat, E. A. Blood Group Substances; Their Chem-

istry and Immunochemistry. New\- York, AcademicPress, 1956, p. 330.

11. \Vitebsky, E. Interrelationship between the Rh sys-tem and the A B system. Blood 1948, Spec. issueno. 2, 66.

12. Schiff, W. Zur Frage der Spezifitit der Reaktions-typen von O/A/B-Seren. II. Mitteilung, sero-

logische Untersuchungen bei 0- und A- Menschenvor und nach Immunisierung mit B-bzw.A-Speichel. Z. Immun.-Forsch. 1958, 117, 1.

13. Crawford, H., Cutbush, M., and Mollison, P. L.Hemolytic disease of the newborn due to anti-A.Blood 1953, 8, 620.

14. \Wiener, A. S., Freda, V. J., Wlexler, I. B., andBrancato, G. J. Pathogenesis of ABO hemolyticdisease. Amer. J. Obstet. Gynec. 1960, 79, 567.

15. Szulman, A. E. The histological distribution of bloodgroup substances A and B in man. J. exp. Med.1960, 111, 785.

16. Sober, H. A., Gutter, F. J., Wyckoff, M. M., andPeterson, E. A. Chromatography of proteins.II. Fractionation of serum protein on anion-ex-

change cellulose. J. Amer. chem. Soc. 1956, 78,756.

17. Sober, H. A., and Peterson, E. A. Protein chro-matography on ion exchange cellulose. Fed. Proc.1958, 17, 1116.

18. Levy, H. B., and Sober, H. A. A simple chromato-graphic method for preparation of gamma globu-lin. Proc. Soc. exp. Biol. (N. Y.) 1960, 103, 250.

19. Abelson, N. M., and Rawson, A. J. Studies of bloodgroup antibodies. I. Fractionation of anti-A andanti-B isohemagglutinins by anion-cation celluloseexchange chromatography. J. Immunol. 1959, 82,435.

20. Fahey, J. L., and Morrison, E. G. Separation of6.6S and 18S gamma globulins with isohemagglu-tinin activity. J. Lab. clin. Med. 1960, 55, 912.

21. Humphrey, J. H., and Porter, R. R. Reagin con-

tent of chromatographic fractions of human gamma-globulin. Lancet 1957, 1, 196.

22. Orenstein, L., and Davis, B. J. Disc electrophoresis.In preparation.

23. Levene, H., and Rosenfield, R. E. ABO incompati-bility. Progr. hum. Genet. In press.

24. Wallenius, G., Trautman, R., Kunkel, H. G., andFranklin, E. C. Ultracentrifugal studies of ma-

jor non-lipide electrophoretic components of nor-mal human serum. J. biol. Chem. 1957, 225, 253.

25. Kabat, E. A. Heterogeneity in extent of the com-

bining regions of human antidextran. J. Immunol.1956, 77, 377.

26. Kabat, E. A. The upper limit for the size of thehuman antidextran combining site. J. Immunol.1960, 84, 82.

27. Zuelzer, W. W., and Cohen, F. ABO hemolyticdisease and heterospecific pregnancy. Pediat. Clin.N. Amer., May, 1957, p. 405.

28. Bird, G. WV. G. Agar gel studies of blood group

specific substances and precipitins of plant origin.I. The precipitins of Dolichos biflorus. Vox Sang.(Basel) 1959, 4, 307.

29. Abelson, N. M., and Rawson, A. J. Studies of bloodgroup antibodies. II. Fractionation of the Rhantibodies by anion-cation cellulose exchangechromatography. J. Immunol. 1959, 83, 49.

883