British Journal of Haematology, 1975,30, 489.

The Blood Group I and i Antigens of Amniotic Fluid I. ASSOCIATION O F I A N D i ANTIGENS WITH BLOOD GROUP A, B A N D H ANTIGENS

TENFEIZI,LARSL. CEDERQVIST AND ROBERT CHILDS T h e Clinical Research Centre, Harrow, and The New York Hospital-Cornell Medical Center, New York (Received 3o]anuary 1975 ; acceptedfor publication 7 February 1975) SUMMARY. Human amniotic fluid has been shown to contain blood group i as well as I antigens. Crude extracts of amniotic fluids at 16-23 weeks of gestation were in general more active than those obtained at term. A pool of amniotic fluids which had A, B, H as well as I and i activity was fractionated with an insolubilized aiiti-I (Group 3 type) immunoadsorbent column. There was evidence for the occurrence of I and i determinants on macromolecules carrying A, B and H determinants, for tlie fraction specifically retained by the column was enriched by 50-60-fold in I and i activity and by at least Io-fold in A, B and H activity. In the fraction not retained by the column there remahed the bulk ofthe A, B, H activity in addition to I activity of Group I type which is known to be distinct from the Group 3 determinant. With the aid of specific immunochemical fractionation procedures, human amniotic fluid should prove useful in structural studies of the several I and i determinants and of their relationship to other blood group determinants. The blood group I and i antigens are the most common targets of the autoantibodies (cold agglutinins) which occur in the chronic cold haemagglutinin disease (Dacie, 1962 ; Marsh &Jenkins, 1960). The traiisiently occurring red cell autoagglutinins in Mycoplasnzn yneunioniae infection and in infectious mononucleosis also have I and i specificities (Chanock et al, 1961 ; Jenkins et al, 1965; Rosenfield et a/, 1965). The I antigens are well represented on the red cells of the vast majority of adults (Wiener et al, 1956); the i antigens are expressed strongly on cord blood cells and in the course of the first year of life a gradual change occurs: the I antigen increases to adult levels and the i antigen diminishes to low levels (Marsh, 1961). The I and i antigens are also present on leucocytes (Franks, 1966; Shuinak et al, 1971; Thomas, 1973).

In the last several years it has become apparent that tlie I and i antigens are a family of determinants and that they occur on certain precursors of the ABH and Lewis blood group substaiiccs (Feizi et a!, 1971a, b; Feizi& Kabat, 1972). So far there is structural information on one of the I antigenic determinants. It involves one of the two carbohydrate chains (Type 2 chain) (Painter et al, 1963 ; Lloyd & Kabat, 1968) of the precursor of ABH and Lewis blood group antigens (Feizi et al, 1971b). The structures of the other five or so I determinants and of the several i determinants are not yet established. Correspoiidcnce: D r Ten Feizi, Division of Commuiiicablc Discascs, Clinical Research Ccntrc, Watford Road, Harrow, Middlesex HA1 3UJ.

489

490

T. Feizi, L. L. Cederqvist and R. Childs

The occurrence of A, B and H blood group antigens in human amniotic fluid is well known (Freda, 1958). More recently it has been reported to contain I antigen also (Cooper, 1970). The aim of the present studies has been to investigate whether the amniotic fluid I antigens occur on the same or on separate macromolecules as the ABH antigens, and whether they can be isolated in sufficient amounts for structural studies. In the course of this work the i antigen has also been found to occur in amniotic fluid. MATERIALS A N D METHODS Amniotic fluid samples were obtained from patients admitted to the New York HospitalCoriiell Medical Center, for induced abortion at 16-23 weeks of gestation or from patients admitted to Northwick Park Hospital for delivery at term. In the former group amniotic fluid samples were aspirated prior to hypertoiiic saline injection and in the latter they were collected at the time of surgical induction using the Drew-Smythe catheter. The samples were centrifuged at 1400 g for 3 0 mill to remove amniotic cells and they were tested for ABH activity by haemagglutination inhibition immediately or after storage at - 20°C for up to 3 months. Anfisera. Human immune anti-A and anti-B antisera were obtained from Behriiigwerkc AG, Marburg Lahn. An anti-H reagent (extract from Ulcx europaeus) was obtained from Dadc Division, American Hospital Supply Corporation. The anti-I sera Ma, Step and Low and the anti-i serum McDon have been previously described (Feizi ct a/, 1971a, b, 1974; Feizi & Kabat, 1972). These were obtained from patients with chronic cold agglutinin disease. Part of the anti-I Low antibody was cryoprecipitable. The supernatant serum after removal of the cryoglobulin contained substantial residual anti-I activity aiid could be used in quantitative precipitiii assays. O n the basis of quantitative precipitiii reactions with a panel of I-active glycoproteins, Ma and Step have previously been assigned to Group I and Group 3 anti-I specificities respectively (Feizi & Kabat, 1972). By the same criteria, anti-I Low can be assigned to Group 3. The sera from Step and McDon were recent samples kindly provided by Mrs Marie Crookston. Haeniagghtinatioti inhibition assaysfor the dctcctioii of A , B atid H arztigcris were carried out at 20°C by the microtitre method as previously described (Feizi et al, 1971a). In these assays cight haemagglutinating units of anti-A, B or H reagents were used. The inhibitors were either tested neat or after serial two-fold dilutions. Quantitative precipitiiz assays with the anti-I aiid anti-i sera were carried out at 4°C as previously described (Feizi et ul, 1g71b). Preparation o j - m d e extracts of arnriiotic Jziid. The amniotic fluid samples were precipitated with 2.5 volumes of 95% ethanol, left at room temperature overnight and centrifuged at 1400 g for I 11. The precipitates were dried iri vuctto over 'CaCI, and subjected to pepsin digestion, as described by Bendich et a1 (1964) and the ethanol precipitation repeated after the addition of sodium acetate (3.5 g/dl). The precipitates were taken up in distilled water and dialysed against changes of distilled water. The dialysed solutions were clarified by centrifugation at 1400 g overnight and lyophilized. Isolation of I untigens. An anti-I iinmunoadsorbeiit column was made with the cryoglobuliii preparation of anti-I Low which was isolated by repeated precipitation at 4°C followed by

I a i d i Antigens ofAiimiotic Fli4id

49 1

dissolving at 37°C. The IgM kappa cryoglobuliii thus isolated (857 mg) was coupled to Sepharose zB by the CNBr inctliod (Ax& ct nl, 1967; Cuatrecasas ct al, 1968). Isolation of I antigen froin the crude amniotic fluid extracts was carried out by absorption onto the anti-I Low imniunoadsorbeiit column at 4"C, washing at 4'C and elution at 37°C as previously dcscribed (Feizi & Kabat, 1974). RESULTS

Blood group A, B and H antigens in arizniotic.fltiid. The blood group A, B and H antigens detected in 12 amniotic fluid samplcs from which extracts were made, are shown in Tablc I. Threc were obtained at 16-23 weeks of gestation, eight at term and one was a pool of amniotic fluid samplcs obtained at 16-23 weeks. In 17 additional iiistanccs (16-23 weeks of gestation) the typing of rnatcrnal, fetal and amniotic fluid blood groups was carricd out (Table 11). In all instances the amniotic fluid A and B blood groups correlated with those of the fetal red cells. Blood group H activity was detected in 20 out of 27 samples tested (Tables I and 11). The H 'negative' samples included four of blood group 0, two of blood group B aiid onc of blood group A. There was no obvious correlation (Table I) betwccii the presence or absence of A, B and H activities and the I and i activitics of the ainniotic fluid samples. TABLE I. Blood group A, B, H, I and i activitics of crude extracts of amniotic fluid Crude extract*

Blood p o ~ p detected

-

(ld)

A

At 16-23 weeks Moor 0 3 Pa A C h y7 Pooled 0 At term Mor Sni c ) Hol 0 Dic A Man T Am Hop

+

o'c A

* Non-dialysablc

+ i-1-

i-

0 0 0

0

+ +

R

Yield

Anti-i

H

Step

0 i) 0

+

0 0 0 0 0

0

+

+ + + 0 0

+-I+ + + +

+ + - -0 0

Atnoimt.f needed to precipifnfe 3 p g N with

Votrrn1e obtained

McDori

37s 275 275

I00

80 70 92s

1SO

2-24

17s 37s 450 41 3 413

I22

488

210

3 00 300

358 3 40 400 3 20

32s I

solids aftcr pcpsiii digestion. t This is the amount (pg) of crude extract needed to prccipitate 3 p g nitrogen when total antibody is convertcd to 6 pg nitrogen (cf. Fig I). The amount iiccdcd is inversely rclatcd to the activity of the cxtract. 3 Symbols used in Fig I.

T. Feixi, L. L. Cederqvist and R. Child5

492

TABLE 11. Maternal and fetal blood groups and A, B and H antigens in 17 amniotic fluid samples at 16-23 weeks of gestation Erythrocyte blood group

No o f cases

Amniotic jhid antigens

Mother

Fetus

0 A 0 A

0

5

0

I

-

-

A A

I

I

-

4 3

4

0

-

-

B B

2 I

-

2

-

I

I

B B A

4

-*

B

l

-

H 3t I I

I out of 2%

3

* - = not detected by haemagglutination inhibition. t Numbers indicate instances of blood group activity

detected in haemagglutination inhibition assays with undiluted amniotic fluid samples. $; Only two of the four amniotic fluid samples in this group were tested for H substance.

I and i activities of the crtide amnioticjluid extracts. There was considerable variation in the yield of crude extract per IOO ml of amniotic fluid and there was overlap between the yields at 1 6 2 3 weeks and at term, although the two highest yields were obtained among the former group (Table 11). Quantitative precipitin assays with two anti-I sera, Ma and Step, and with an anti-i serum, McDon, showed not only I activity but also i activity in all the extracts (Fig I). The activities of the extracts were compared by calculating the amounts required to precipitate 50% of the anti-I and anti-i antibodies in the sera. For coiiveiiience of comparison, the total antibody nitrogen in each serum was converted to 6 pg nitrogen (Table I, cf. Fig I). The 16-23 week amniotic fluid samples had two to three times more I activity per unit weight than the samples at term. The difference in the i activities of the two groups as recognized by anti-i McDon was much less impressive. Studies ofantigens isolated with anti-Ilow irnmunoadsorbent column. 549 mg of the crude extract from the 16-23 week amniotic fluid pool were dissolved in 60 ml isotonic saline and were passed at 4OC over the anti-I Low immunoadsorbent column and after extensive washing with isotonic saline at 4"C, the specifically retained fraction was eluted at 37°C. The effluent fraction and the eluted fraction were dialysed against distilled water and lyophilized. The yields were 401 mg and 7.7 mg respectively. Carbohydrate analyses of the eluate were performed by gas-liquid chromatography (Sawardeker et al, 1965) and mass spectrometry (Golovkina et al, 1966) by Dr Sigfrid Svensson. These revealed galactose 40%, N-acetylglucosamine 27% and small amounts of fucose and N-acetylgalactosamine. The nitrogen content was 6.4%. The quantitative precipitin curves in Fig 2 show the I and i activities of the crude extract, the specifically eluted and the effluent fractions. The eluted fraction was 50-60 times more active than the crude extract. The effluent fraction had negligible activity with the Group 3

I and i Antigens ofArnniotic Fluid Anti-l

Ma 1:2 30p1

Total volume 5 0 0 ~ 1

Anti-l Step 1:4 30pl

493 Total volume 500 pl

Antigen (pg)

FIG I. Quantitative precipitin reactions of crude amniotic fluid extracts when tested with anti-I Ma, anti-I Step and anti-i McDon. The dotted and solid lines demarcate the areas of curves obtained with samplcs a t 16-23 weeks and at term respectively. Symbols for individual sainplesaredefinedin Table I.

anti-I scra Low and Step and with anti-i McDon; however, considerable activity with the Group I anti-I serum remained in this fraction. Thus the molecules retained by the anti-I Low column contained I antigenic determinants of Groups I and 3 and i determinants of McDon type. However, only some of the Group I determinants were in the eluate; a substantial proportion of them was on other molecules which were not retained by this column. Haemagglutination inhibition assays showed that most of the A, B and H active substances were not retained by the anti-I column, for the activity of the effluent fraction was not appreciably different from that of the original material (Table 111). However, there was clear evidence for the occurrence of A, B and H determinants on the molecules retained by the anti-I column, for the eluate was enriched in A, B and H activity by at least ro-fold.

T. Feizi, L. L. Cederqvist and R. Childs

494

Anti-I Low I: 6 2 0 4

r

:F1-

Total volume 500p1

Anti4 Ma 1:2 30p1

Totol volume 5 0 0 ~ 1

3

2 0I .0

ea Anti-I Step 1:4 30p1

E

Anti4 McDon 1:16 lop1

Total volume 500p1 r

4 !

2 I I I I 0 0 10 20 30 40 50 60 0

~

500

1000

1500

Total volume 250pl r

L

0 10 20 30 40 50 60

0

500

1000

1500

Antigen (pg)

FIG 2. Quantitative precipitin reactions of the pooled 16-23 week amniotic fluid material when tested with anti-I sera Low, Step and Ma and with anti-i McDon. 0,Original material applied to anti-Low immuno-adsorbent column; +, effluent fraction; 0 , specific eluate. TABLE 111. Blood group A, B and H activity of amniotic fluid fractions obtained by affinity chromatography with the anti-I Low immunoadsorbent column ~

~~

Amniotic Jtrid,fiactiotts Material applied Effluent fraction Specific eluate

Blood group activity*

Weight (mg)

s 49 401 7.7

310

640 64

94 5 640 16

1260 1300

127

* Activity is expressed as the lowest concentration (pg/ml) giving complcte inhibition of eight haemagglutinating units of antibody or lectin. DISCUSSION The source of the blood group substances of amniotic fluid deserves detailed investigation. There is general agreement that they are largely of fetal origin but there is conflicting information on the possible maternal contribution if the mother is a secretor. This matter has been reviewed by Arcilla & Sturgeon (1972) and the importance of avoiding contamination by maternal secretions stressed. Our data with the 16-23 week amniocentesis samples would suggest that these blood group substances are fetal in origin. Among the 17 mother-fetus

I and i Antigens of Amniotic Fluid

495

pairs studied there were five instances of maternal-fetal heterospecificity (mother A, fetus 0;mother B, fetus 0; or mother A, fetus B); in no instance was the mother’s blood group represented in the amniotic fluid. It is not yet known whether the amniotic fluid blood group substances are products of secretory epithelial cells or of degraded erythrocytes and leucocytes, or a mixture from both sources. W e have previously presented evidence for the occurrence of I antigenic determinants in the interior of the blood group A, B, H and Lewis antigens: certain I determinants were revealed after partial degradation of A, B and H substances (Feizi et al, 1971b) and the I (and i) antigens were present on an ovarian cyst glycoprotein which behaved like a blood group precursor substance (Feizi et al, 1971a, b; Feizi & Kabat, 1972). The present studies show that a proportion of the A, B and H antigens occur on the same molecules as the I and i antigens. However, a substantial proportion of the A, B and H substances was not retained by the anti-I Low column. It is possible that some of these are more complete molecules on which the A, B and H determinants conceal the Group 3 I and i (and other associated) determinants. A similar situation was encountered in our studies of the blood group I and PI-active glycoproteins of sheep hydatid cyst fluid (Feizi & Kabat, 1974). Affinity chromatography with an anti4 immunoadsorbent column yielded an I t PI-active fraction which was retained by the column and a PI-active fraction without I activity which was not retained. Human erythrocyte extracts with A, B, H and I activities have been obtained in several laboratories (Hamaguchi & Cleve, 1972; Aiistee & Tanner, 1974; Gardas & Kokielak, 1974). However, attempts were not made to isolate the 1 active components immunochemically and it is not yet known if the I and the ABH determinants occur on the same molecules. Recently we have studied 1-active extracts from human erythrocytes in quantitative precipitin assays (Feizi et al, 1975). These materials were lacking A and B activity in haemagglutination inhibition assays (Roelcke, personal communication). However, the work of Gardas & Kokielak (1973) suggests that these membrane antigens may present special problems in haemagglutination inhibition systems: they have evidence that these may be unusual glycoproteins associated with small amounts of lipid and that they fail to show activity in haemagglutination inhibition assays unless they are complexed with ‘carrier lipids’. Assays with the effluent fraction of the anti-I Low column (Fig 2 ) show that a proportion of the I determinants of Ma type (Group I) occur on separate molecules which are devoid of the Group 3 and other associated I and i antigenic determinants. Such a glycoprotein with Group I activity and lacking in the majority of other I and i determinants was isolated from human milk (Feizi & Marsh, 1970; Feizi et al, 1971a; Feizi & Kabat, 1972) and the Group I determinant was shown to involve a non-reducing PDGal (I +4)~ ~ D G ~ N A c 6)-structurc. (I-+ This structure makes up the type z chain (Painter et a / , 1963 ; Lloyd & Kabat, 1968) of the precursor of ABH and Lewis blood group antigens and cross reacts with the pneumococcus type XIV antigen. We are currently using an anti-I Ma immuno-adsorbent column to isolate the Group I antigens from the above effluent fraction from the Low column. Among the various I-active glycoproteins so far studied, the ovarian cyst precursor-like glycoprotein (OG) is unique in that it contains all the I and i determinants. Only certain I determinants are detectable on the other glycoproteins (Feizi et al, 1971b; Feizi & Kabat, 1972). Our screening studies currently underway indicate that the majority of the I and i antigenic determinants are present in amniotic fluid. Although these antigens coiistitute minor

496

T. Feixi,L. L. Cederqvist and R. Child3

components of the total non-dialysable solids of amniotic fluid, they hold promise for structural studies since they can be specifically isolated by imniunoadsorbent columns. Furthermore, it will be important to study isolates from individuals of known blood groups. These are likely to be less heterogeneous than those from pooled samples and they may provide useful information on blood group genetics and biosynthesis. ACKNOWLEDGMENTS

We thank Mr I. McFadyen and his staff at Northwick Park Hospital for the amniotic fluid samples obtained at term. For the carbohydrate analyses we are indebted to Dr S. Svensson of Klin Kem Lab, Lasarettet, Lund. The assistance of Mr P. Gough and Mr R. Sidebotham is gratefully acknowledged. A part of this work was carried out in the laboratory of Dr H. G. Kunkel at the Rockefeller University, New York. REFERENCES ANSTEE, D.J. & TANNER, M.J.A. (1974) Blood-group serology of fractions obtained from the human erythrocyte membrane. European Journal of Biochemistry, 45, 31. ARCILLA, M.B. & STURGEON, P. (1972) Lewis and ABH substances in amniotic fluid obtained by amniocentesis. Pediatric Research, 6, 8 5 3 . A x ~ NR., , PORATH, J. & ERNBACK, S. (1967) Chemical coupling of peptides and proteins to polysaccharides by means of cyanogen halides. Nature, 214, 1302. BENDICH,A., KABAT,E.A. & BEZER,A.E. (1946) Immunochemical studies on blood groups. 111. Properties of purified blood group A substances from individual hog stomach linings. Journal of Experimental Medicine, 83, 485. CHANOCK, R.M., RIFKIND,D., KRAVETZ,H.M., KNIGHT,V. &JOHNSON, K.M. (1961) Respiratory disease in volunteers infected with Eaton agent; a preliminary report. Proceedings of the National Academy of Sciences of the United States ofAmerica, 47, 887. COOPER, A.G. (1970) Soluble blood group I substance in human amniotic fluid. Nature, 227, 508. CUATRECASAS, P.,WILCHEK,M. & ANFINSEN, C.B. (1968) Selective enzyme purification by affinity chromatography. Proceedings of the National Academy ofsciences of the United States ofAmerica, 61, 636. DACIE,J.V. (1962) The Haemolytic Anaemias, and edn, Part 11. Grune & Stratton, New York. FEIZI,T., EBERT, W. & ROELCKE, D. (1975) Studies on the blood group I and i activities of human erythrocyte glycoproteins. (In preparation). FEIZI, T. & KABAT,E.A. (1972) Immunochemical studies on blood groups. LIV. Classification of anti-I and anti-i sera into groups based on reactivity patterns with various antigens related to the blood

group A, B, H, Lea and Leb and precursor substances. Journal of Experimental Medicine, 135, 1247. FEIZI, T. & KABAT,E.A. (1974) Immunochemical studies on blood groups. LVI. Purification of glycoproteins with different I determinants from hydatid cyst fluid and from human milk on insoluble anti-I immunoadsorbents. Journal of Immunology, 112, 145FEIZI,T., KABAT,E.A., VICARI,G., ANDERSON, B. & MARSH,W.L. (1971a) Immunochemical studies on blood groups. XLVII. The I antigen complexprecursors in the A, B, H, Lea, and Lebblood group system-hemagglutination-inhibition studies. Jownal of Esperimental Medicine, 133, 39. B. & FEIZI,T., KABAT,E.A., VICARI,G., ANDERSON, MARSH, W.L. (1971b) Immunochemical studies on blood groups. XLIX. The I antigen complex: specificity differences among anti-I sera revealed by quantitative precipitin studies; partial structure of the I determinant specific for one anti-I serum. Journal of Immunology, 106, 1578. H.G. & ROELCKE, D. (1974) Cross FEIZI,T., KUNKEL, idiotypic specificity among cold agglutinins in relation to combining activity for blood grouprelated antigens. Clinical arid Esperintental Inimunology, 18, 283. FEIZI, T. & MARSH,W.L. (1970) Demonstration of I-anti-I interaction in a precipitin system. VOS Sanguinis, 18, 379. FRANKS,D. (1966) Antigenic markers on cultured human cells I. Ii, Tj", Donath-Landsteiner and 'nonspecific' autoantigens. Vox Sanguinis, 11, 674. FREDA, V.J. (1958) A-B-O(H) blood group substances in the human maternal-fetal barrier and amniotic fluid. American Journal of Obstetrics and Gynecology, 76, 407.

I and i Antigens ofAmniotic Fluid GARDAS, A. & KO~CIELAK, J. (1973) New form of A-, B-, and H-blood-group-active substances extracted from erythrocyte mcmbrancs. Enropean Journal qf Biochemistry, 32, 178. A.& K O ~ I E L AJ.K(1974) , Megaloglycolipids GARDAS, -unusually complex glycosphingolipids of human erythrocyte membrane with A, B, H and I blood group specificity. FEBS Letters, 42, 101. L.S., CHIZHOV, O.S. & WULFSON, N.S. GOLOVKINA, (1966) Mass spectrometric investigations of carbohydrates. Izvestiya Akadentii Nank S.S.S.R., Seriya Khimecheskaya, 1915. H. & CLEVE, H. (1972) Solubilization of HAMAGUCHI, human erythrocyte membrane glycoproteins and separation of the M N glycoprotein from a glycoprotein with I, S, and A activity. Biachimica et Biophysica Acfa, 278, 271. JENKINS,W.J., KOSTER,H.G., MARSH,W.L. & CARTER, R.L. (1965) Infectious mononucleosis: an unsuspected source of anti-i. British Journal of Haematology, 11, 480. LLOYD,K.O. & KABAT,E.A. (1968) Immunocheniical studies on blood groups. XLI. Proposed structures for the carbohydrate portions of blood groups A, B, H, Lewis" and Lewisb substances. Proceedings of the National Academy ofSciences ofthe United States of America, 61, 1470. W.L. (1961) Anti-i: a cold antibody defining MARSH,

497

the li relationship in human rcd cells. Britisk jonrnal of Hacmatology, 7, zoo. W.J. (1960) Anti-i: a new MARSH,W.L. &JENKINS, cold antibody. Natrire, 188, 753. W.M. & MORGAN, W.T.J. PAINTER, T.J., WATKINS, (1963) Isolation of two scrologically active trisaccharides from human blood-group B substancc. Nature, 199, 282. ROSENFIELD, RE., SCHMIDT,P.J., CALVO,R.C. 8; MCGINNISS,M.H. (1965) Anti-i, a frequent cold agglutinin in infectious mononucleosis. Vox Sangthis, 10, 631. A. (1965) SAWARDEKER, J.S., SLONEKER, J.H. & JEANES, Quantitativc determination of monosaccharidcs as their alditol acetates by gas liquid chromatography. Analytical Chemistry, 37, 1602. R.A., CROOKSTON, SHUMAK,K.H., RACHKEWICH, J.H. (1971) Antigens of the M.C. & CROOKSTON, Ii systcm on lymphocytes. Nature: N e w Biology, 231, 148. D.B. (1973) Antibodies specific for human THOMAS, T lymphocytes in cold agglutinin and lymphocytotoxic sera. Europeanjournal oflmmunology, 3, 824. L.J., COHEN,L. & FELDMAN, J. WIENER, A.S., UNGER, (1956) Type-specific cold auto-antibodies as a cause of acquired hemolytic anemia and hemolytic transfusion reactions: biologic test with bovine red cells. Annals of Infernal Medicine, 44, 221.