Proc. Nat. Acad. Sci. USAVol. 71, No. 9, pp. 3588-3592, September 1974

Genetic Control of Basal Serum Immunoglobulin E Level and Its Effecton Specific Reaginic Sensitivity*

(regulator gene/Ir genes/HL-A antigens/allergens/atopy)

DAVID G. MARSHtI, WILMA B. BIAS§, AND KIMISHIGE ISHIZAKAT

Divisions of $ Clinical Immunology, § Medical Genetics and '1 Basic Immunology, Department of Medicine, The Johns HopkinsUniversity School of Medicine, Baltimore, Maryland 21205; and t O'Neill Laboratories, The Good Samaritan Hospital,Baltimore, Maryland 21239

Communicated by Victor A. McKusick, May 29, 1974

ABSTRACT Studies of the distribution of total serumimmunoglobulin. E levels in nonallergic and allergicpopulations defined a cut-off point between low and highimmunoglobulin E at 95 U/ml, based on Mendelian reces-sive inheritance of high immunoglobulin E level. Subse-quent investigations of the distribution of total serumimmunoglobulin E levels in 28 allergic families confirmedthe recessive hypothesis. The results of quantitative skintests in eight families, performed with between five andeight highly purified grass and ragweed pollen allergensper family, demonstrate that the immunoglobulin E-regulating gene exerts a profound effect on specific im-munoglobulin E-mediated sensitivity, often masking theeffect of HL-A associated immune response genes.

Atopic allergy in man is a complex disease associated withthe presence of immunoglobulin E (IgE) reaginic antibodies(1). Since 1872, many authors have reported a familial pre-disposition toward atopic manifestations in general, as well asto each of separate major symptom types-allergic rhinitis,asthma, and eczema (reviewed in refs. 2-4). A further variableis the relative degree to which different allergic subjectsdevelop IgE-mediated sensitivities to different allergencomplexes (e.g., pollens) or, more specifically, to highlypurified allergens that may be isolated from these complexes(4-8). From these and other considerations, one would an-ticipate that the genetics of IgE-mediated allergy would beextremely complicated (2-4, 8).As a first approach to this problem, we have considered

two broad classes of immunogenetic control over IgE-mediated allergy: (i) regulation of the biosynthesis and/ormetabolism of all molecules of the IgE class, and (ii) theeffects of specific immune response (Ir) genes. Family andtwin studies of total serum IgE levels in man, by Hamburgerand associates (9), strongly suggested that the basal levelof IgE is under a fairly simple genetic control, the mechanismof which was not elucidated. Further genetic experiments inanimals showed that certain Ir genes, controlling primarily(if not exclusively) T-cell function, are closely linked to theanimal's major histocompatibility type (10, 11). The effectsof these genes can only be observed under highly limitingconditions of immunization, a situation which pertains inhuman allergy to inhaled pollen antigens (4, 5, 8).

Abbreviations: IgE, immunoglobulin E; Ir, immune response(gene); Creg, crossreacting group (of HI-A antigens).* Presented in part at the 29th Meeting of The AmericanAcademy of Allergy, Washington, D.C., February, 1973.Direct reprint requests to Dr. David G. Marsh at The GoodSamaritan Hospital.

3588

We have recently demonstrated a highly significant associa-tion between IgE and IgG antibody production to Ra5, aragweed pollen allergen [molecular weight = 5000 (12)],and the possession of a major histocompatibility antigen ofthe HL-A7 crossreacting group (Creg) (7, 5, 13). In studieswith more complex allergens, further genetic factors seemed tobe masking associations analogous to those observed forRa5 as well as genetic linkage in families between a specificallergic sensitivity and a specific HL-A haplotype (8). Thepresent communication will- show that one important factorthat modulates the development of specific IgE antibodyresponse is an autosomal gene that regulates basal serum IgEconcentration.

MATERIALS AND METHODS

Unrelated Subjects. The distribution frequency of totalserum IgE levels was ascertained on samples taken from205 unrelated adults (121 men, 84 women) who were highlyallergic to ragweed and/or grass pollens and who had notreceived immunotherapy for at least 2 years. Most samples(70%) were taken when the boosting effect of seasonal al-lergenic stimulation was minimizedl1. None of the remainderwas taken during the probable period of maximal IgE levell.The distribution frequency of IgE levels in nonallergic sub-jects was derived from data on 106 unrelated adults, whowere completely devoid of symptoms. Intradermal skintests on 46 unselected nonallergic subjects, using very highconcentrations (100lg/ml) of both dialyzed ragweed andgrass pollen extracts, gave negative, trace, or one-plus (0.5-cm wheal) reactions.

Families. The genetics of total IgE levels was studied in26 Maryland families and 2 other families (1 PennsylvaniaAmish and 1 Wisconsin) having a total of 108 children (56

11 The minimal (basal) IgE levels occurred between mid-Februaryand mid-May for individuals highly sensitive to grass and rag-weed pollens. The total serum IgE levels in 17 untreated allergicsubjects followed over 2 years (unpublished) showed a year-to-year variation of 16% (414%). Seasonal allergenic exposureproduced a mean rise of 85% (451%) above the basal level forthe year, maximal levels occurring between mid-September andmid-October. On the other hand, ongoing ragweed immuno-therapy (plus contributing seasonal exposure) caused a mean riseof 120% (493%) in 20 allergic subjects during the first year, buta depression after several years of therapy. Serum levels of non-allergic subjects varied by no more than 10% throughout theyear.

Dow

nloa

ded

by g

uest

on

May

9, 2

021

IgE Regulation and Specific Reaginic Sensitivity 3589

boys, 52 girls). All but two children were at least 7 years ofage; the two younger children were included because theyhad unequivocably high IgE levels (>900 U/ml). The meanadult IgE level is reached before 7 years of age (14). At leastone child in each Maryland family was allergic to grass and/orragweed pollens. Maryland serum samples were drawn be-tween March and early May,1 except for Families B7, B9,and E3 (Fig. 2), which were bled during late November andDecember. The Amish (C4) and Wisconsin (C5) familieswere bled during August. Allergy histories were taken fromalmost all subjects and confirmed by skin testing with crudeand purified grass and ragweed allergens in most cases. His-tories of immunotherapy, which can modify IgE levels,1 aregiven in the footnote to Fig. 2.

Measurement of Total Serum IgE Levels. Sera were separatedfrom clotted blood samples as soon as possible and werestored at -20° until used. Total IgE concentrations weremeasured by double antibody radioimmunoassay (15), withduplicate samples in each of two separate experiments. Assayswere repeated further where necessary, particularly in thecritical region of 50-200 U/ml, until replication was withinin--%.

Quantitative Intradermal Skin Testing. Eight immuno-chemically distinct allergens, isolated from grass and rag-weed pollens, were used (Fig. 3). All samples except group IVwere at least 95% pure by published physicochemical andimmunological criteria (4-6, 12, 16, 17). Serial 10-fold dilu-tions (10-7-1 ,ug/ml) of each allergen were prepared in sterileTris-buffered physiological saline stabilized with 0.03%(w/v) human-serum albumin. The concentration of eachallergen eliciting a 2-plus reaction (0.8- to 1.0-cm wheal;2.0- to 4.0-cm erythema) 15-20 min after intradermal in-jection (0.05 ml) was taken to be the person's sensitivity tothe allergen (7).

HL-A Typing. The leukocytes of all family members weretyped, in duplicate,, by standard leukocytotoxic microassay(18), with 90 or more defined sera capable of recognizing 31different HL-A serotypes.

RESULTS AND DISCUSSION

Definition of Low and High IgE and Calculation of GeneFrequencies. The general tendency for allergic people to havehigh and nonallergic people to have low serum IgE levels iswell established (19, 20), but some nonallergic people haveatypically high and some allergic people have atypically lowIgE levels. We will define the cut-off between high and lowIgE as the level at which the combined percentages of suchatypical subjects is minimized. First, the cumulative percentfrequencies of IgE levels for nonallergic and allergic subjectswas plotted (Fig. 1). The difference curve for these two setsof data was then calculated (dashed line). The maximum ofthis curve, at 95 ±t 5 U/ml, gives our best estimate of theIgE cut-off level. Only 21% of the nonallergic subjects haveIgEs above and only 22% of the allergic subjects have IgEsbelow this level. Our estimate of 95 U/ml is in excellentagreement with a value of 91 ± 5 U/ml, calculated by re-plotting data of Gleich et al. (20), taking 1 U/ml 2.42 ng ofIgE/ml (21).In preliminary family studies (22), the mode of inheritance

of hih IgE levels, with or without demonstable allergy, mostclosely approximated a Mendelian recessive pattern. Dr.T. A. Waldman, N.I.H., kindly supplied data on the serum

COa.

CD0

CDw-LJ-iz0zaz

,C

-LJn

0z

w0.w

0

I

0 125 25 100 400 1600 6400LOG, TOTAL SERUM IgE (U/ml)

-100

0

cc C

- 80 i'.142L eOGoD00

- 60Z-

- 40Q.OF J)

20 z a-

U z)

00

~-cc5

FIG. 1. Cumulative distribution frequencies for basal serumIgE levels in nonallergic and allergic populations, and a dif-ference curve for the two distributions.

IgE levels of 74 unselected adults living in the nearby Wash-ington, D.C. area. A plot of his data showed that 27.5% ofthe population possessed IgE levels over 95 U/ml, from whichwe calculated gene frequencies of hypothetical alleles, R andr, to be 0.48 and 0.52, respectively, assuming Hardy-Wein-berg equilibrium (23). Using analogous data obtained byDr. G. J. Gleich, Mayo Clinic, Minnesota (ref 3; and data on79 unselected adults, personal communication), we determinedthe frequencies of R and r to be 0.45 and 0.55. The closeagreement between the data, for different populations livingunder different exposure conditions, is encouraging.

Test for Recessive Inheritance of High IgE. We analyzedthree types of mating: type A, low X low giving at least onehigh IgE (Rr X Rr); type B, low X high giving at least onehigh IgE (Rr X rr), and type C, high X high (Fig. 2). TypesA and B were analyzed by the "bias of ascertainment" method(23). In four families (types D and E) having all low children,it is not possible to determine whether the low parents are RRhomozygotes or Rr heterozygotes.

Individuals with IgE levels near the 95 U/ml cut-off pointcould not readily be assigned as "high" or "low" for thefollowing reasons: uncertainty in the precise value of the cut-off; year-to-year variation in basal IgE levels in allergicpeopled; effect of known allergenic stimulation in allergicmembers of families B7, B9, E3, C4, and C5; accuracy ofindividual IgE determinations. For samples taken at themost appropriate time of yearl I, we estimate that there is anuncertain "gray region" extending about 10 U/ml below(arising from possible errors) and 26 U/ml above (arisingfrom errors plus annual variation in basal IgE) the cut-offpoint. Allergic individuals falling in this region were assignedas "low," and nonallergic individuals as "high" (see legendto Fig. 2). Of the five families bled at less appropriate timesof the year, IgE levels could readily be assigned as high orlow in B7, C4, and C5. Family B9 had one allergic and one(as yet) nonallergic child, having IgE levels of 102 U/ml and88 U/ml, respectively (Fig. 3). In view of the boosting effectof allergenic stimulation, and the clearly defined distributionof IgE levels in this family, these two levels were taken to below. All the children of family E3 were allergic and all, in-

Proc. Nat. Acad. Sci. USA 71 (1974)

Dow

nloa

ded

by g

uest

on

May

9, 2

021

Proc. Nat. Acad. Sci. USA 71 (1974)

MATING TYPE A(Low xLow-Al astney High IgE)

00

00 .

so t0 a0- *'l . 0

. no t *

oZ5 25 50 1 200, 4- , *Fo12. 25 so l00 200 400 800 9800'

MATING TYPE B(Low x High -At last o High IgE)

a 0I 0 0

a33 . aIIo o0 a

I I

1: a: _a ' I a , 0 a

0 0E s

I-rI I I F A

0125 25 50 900 200 400 800 960>900

TOTAL SERUM Ig E (kiternot. Units/ml)

a 0,Parents *eOffspngI

MATING TYPE C--(Highx-High-All K01h9E)

Cl

C2-C3C4Cs5

0a2

DlB00

25 50 90 260 40 860 00>9600MATING TYPE D

(Low xLow-AlI Low bE)

X --o -!| fa* 0a oI I 4 o I0

0912.5 25 50 900 200 400 80SM 0>I600

MATING TYPE E(LowxHgh -.-AII Low Ig E)

m

. I-.GM a

25 50 Po

0

200 4008 o M00 >00TOTAL SERUM IgE (Inteot Urits/ml)

Totals, 28 08 64

FIG. 2. Distribution of total serum IgE levels in families. The cut-off point between low and high IgE (95 U/ml) is shown by a solidvertical line, and the limits of the uncertain "gray region" around this point (85-121 U/ml) are denoted by dashed lines. The followingallergic people having IgE levels in this region were classified as "low": fathers of families A4 and A9, mother of B5, and one child in each

of A10, B3, B9, and E3. A nonallergic, male child with negative skin test of C1, with an IgE level of 97 U/ml was classified as "high." Thefemale child of C2 with a level of 124 U/ml (marginally above the gray region) was not allergic. A further asthmatic child of family C3died after an allergic reaction to a bee sting. The following people have been receiving immunotherapy during the past several years

(IgE levels probably suppressed): mother and only male child of B4, the male children of B6 and B8 who had IgE > 700 U/ml, three outof four of the children of B9 who had IgE > 700 U/ml, and the mother, father, and one male child (IgE = 645 U/ml) of C1.

cluding the highest IgE level (89 U/ml), were taken to below.For families of types A, B, and C, we found a good agree-

ment between the observed and predicted incidence of highIgE levels in the siblings (Fig. 2), especially considering theuncertainty in assigning people whose IgE levels were in thegray region, 85-121 U/ml. As predicted, all 25 siblings ofthe 5 type C families had high IgE levels. The incidences ofhigh children in types A and B families are both slightlyhigher than one standard deviation above the theoreticallypredicted values. This may be due to a bias in selecting severalfamilies (especially B6 and B9) because they had more thanone highly allergic child, or may result from the relativelysmall number of families studied. The apparent imbalance ofmale children with high IgE levels (42 boys compared with 22girls) results from selecting most families from an allergystudy in a boys' school. We saw no sex difference in the dis-tribution of IgE levels in the children's parents or the 205allergic adults studied, in agreement with other reports on

adults and children (14, 20). Thus, we have every reason tobelieve that the IgE gene is autosomal. The small number offamilies with mating types D and E reflect the selection onlyof families with at least one allergic child.

Thus, the inheritance of a high serum IgE level appears toapproximate closely to a simple Mendelian recessive trait.

We also suggest that the same conclusion is directly applicableto the experiments of Schwartzman et al. (24) in dogs. Whilethey did not have reagents to measure canine IgE levels,they clearly demonstrated that matings of dogs having multipleallergic symptoms and skin reactivities (characteristic ofhigh IgE levels in man) gave rise only to offspring with similarallergic manifestations. Analysis of their data on matingsanalogous to type B also gave results consistent with our

hypothesis.The gene product of the dominant R allele may regulate the

rate of IgE -biosynthesis in or secretion of IgE from plasmacells. Other possibilities for control include: regulation at theT-cell level (see ref. 25), differentiation of B-cells into IgE-producing plasma cells, and removal of circulating IgE bydegradation or cellular adsorption. Whatever the mechanism,only the rr homozygotes would have their serum IgE levelsdetermined by a less efficient regulator, the gene product ofthe r allele **.

Control of Specific IgE Response. The induction of specificIgE antibody synthesis (and resulting allergy) will depend

** It is quite possible that more than two alleles of this gene, and/or that further IgE-regulating genes, may be involved in determin-ing IgE levels in certain people, particularly those having veryhigh or very low levels.

do o

a

II, *c

a

*coo

I glmft a *

A IA2 -

A3 -

A4 -

A5 -

A6 -

A7 -

ABA9AlO -

BB2B3B4B5B6B7B8B9 TYPE Ncof! Total Noofcdewfith HiMIQE

A 10 41 I [ 153+22B 9 131 1 17.5 +24C 55 25 j25j MO0CD 2 __ __

E 3 9 __

3590 Immunology: Marsh et al.

-- -- -- --- ---

I

Dow

nloa

ded

by g

uest

on

May

9, 2

021

IgE Regulation and Specific Reaginic Sensitivity 3591

on whether the individual has the necessary Ir genes (includ-ing those associated with HL-A), which determine recognitionof the stimulating allergen in immunogenically limitingdosage. In order to ascertain the relative importance of (a)HL-A associated Ir genes and (b) the IgE-regulating gene,in determining specific IgE responsiveness in man, we made adetailed study of 24 pairs of highly allergic siblings from sevenfamilies. Three families were tested with eight allergens andthe remainder with either five or six allergens. A sibling pairwas considered to be concordant in sensitivity to a particularallergen when the ratio of skin sensitivities in the two mem-bers was less than or equal to 100, and discordant when thisratio was 1000 or greater.

Table 1 summarizes concordance of specific sensitivitieswith HLC-A haplotype within pairs of highly allergic siblings.Each comparison represents one specific sensitivity in oneof the pairs. Since not every allergic child is sensitive toevery allergen, we made two types of comparison: column 1,irrespective of whether both children were negative to theallergen in question, and column 2, where at least one memberof the pair gave a 2-plus reaction with the allergen solutionat a concentration of 10-2 jg/ml or lower. We found nosignificant differences between the levels of concordance inoverall specific sensitivities for HILA identical pairs (groupA), pairs sharing one HL-A haplotype (group B), and HL-Adistinct pairs (group C).

In Table 2, the concordance in skin sensitivities to specificallergens is examined for pairs of highly allergic siblingswho have either similar or very different IgE levels. Columns1 and 2 present the same types of evaluation as presented inTable 1. In Table 2, we find a much greater overall con-cordance in specific sensitivities in group F, where bothmembers of the pair have high IgE levels, than in group E,where the IgE levels are very different. For the more criticalcomparisons in column 2, the difference is highly significantby chi-squared analyis with Yates' correction (P = 0.001).Low-low pairs (group D) are sensitive to fewer allergens thanhigh-high pairs, but there is not yet sufficient data to permita valid comparison of group D with the other groups.The importance of genetic regulation of total IgE level,

relative to HL-A associated Ir genes, in determining specificsensitivity is further illustrated by Family B9 (Fig. 3). Offive highly allergic siblings, two have inherited HLA haplo-types A and D; the other three have the alternative haplo-types B and C. For siblings nos. 3, 4, and 5 (genotypes BC,AD, and BC), there is a similar overall pattern of specificskin sensitivity to the eight highly purified allergens. IfHL-A associated Ir genes were predominant determinants of

TABLE 1. Influence of HL-A genotype on the concordanceof specific skin sensitivities

within sibling pairs of eight families

No. of Concordance in specificNo.of skin sensitivityfHaplotypes of siblingsibling pair pairs 1 2

Group A Identical* 9 46/66 (70%) 14/34 (41%)Group B One different* 6 32/45 (71%) 11/24 (46%)Group C Both different 9 48/65 (74%) 19/36 (53%)

* Groups include data from one family where the father was

ASYMPTOMATIC PARENTS

(4t) (41)

A B C D- 2 Morgtool stinorectivily No skin reoctivity

o0 Allergy in fomity No ollergy in familyGRASS RAG GRASS RAG

Ig E = 500U/tl I E = 31 U/nl

LI I

) (18-/2)

A D-6--4 --2 - I

GRASS RAG

IgE =33U Ant

A 1,8

B 9,12

C 9,W-10Do X, 12

I I

(0(16-/4) El (t5-3) ai (14-9) (0 (11-9)/\/\~ /\ /

B C B C A D B C

GRASS RAG GRASS RAG GRASS RAG GRASS RAG

IgE 1O2UM IqEa1440U/inl IgE 704UAMt IgE 1056Umnl

Other sibs. [i (10) Q(8) i (6) [iJ(2)

/\ /\ /'\B C B C B D N.D

Negative or Weakly Skin Reoctive aAsymppomotic Asymptomotic No testedIgE= 24OU Nl 88UAI 960 U.m

FIG. 3. Distribution of specific IgE-mediated skin sensitivities,HL-A type, and total serum IgE levels in one family (B9).Siblings are numbered 1 through 9; their ages and the ages ofonset of allergic rhinitis (italicized) in allergic siblings are givenin parentheses. Vertical bars represent the logo (allergen concen-

tration, ug/ml) that elicits 2-plus skin reactions. The followingimmunochemically distinct, highly purified pollen allergenswere used: * group I (molecular weight 27,000), 1 group II(molecular weight 11,000), egroup III (molecular weight 11,000),and LIMgroup IV (molecular weight about 50,000; not completelyhomogeneous), all from Lolium perenne (rye grass) (4, 5); MAgB(molecular weight 10,500) from Phleum pratense (timothy grass)(16); EIAgE (molecular weight 37,800), a Ra3 (molecularweight 11,000) and B Ra5 (molecular weight 5000), all fromAmbrosia elatior (short ragweed) (17, 6, 12) and supplied by Drs.T. P. King and L. Goodfriend. All grass pollen allergens were

prepared in the author's laboratory (reviewed in ref. 5).

specific sensitivity, one would expect a much greater overallconcordance in sensitivity between HL-A-identical siblings,nos. 3 and 5, than between the nonidentical pairs, which is notthe case. The common feature is that all three -have veryhigh IgE levels (>700 U/ml). Conversely, siblings nos. 1 and2, who have genotypes AD and BC and much lower IgElevels, differ from one another in some specific sensitivities;but each differs more markedly from other allergic siblingshaving identical HL-A types but high IgE levels. Thus, themost important factor determining overall concordance ofspecific IgE-mediated skin sensitivity does not appear to be Irgenes associated with inherited HL-A haplotypes, but the genecontrolling serum IgE level.Four further points should be made about this family,

which are typical of our findings in several other "allergicfamilies" (4, 5, 26). First, some individuals (the father and

TABLE 2. Influence of genetic controlof basal IgE level on the concordance

of specific skin sensitivities within sibling pairs

Concordance in specificNo. of

IgE levels of sibling skin sensitivitysibling pair pairs 1 2

Group D Low-low 3 18/21 (86%) 3/6 (50%)Group E Low-high 11 52/83 (63%) 12/43 (28%)Group F High-high 10 56/72 (78%) 29/45 (64%)

probably homozygous (see Fig. 4 of ref. 8). If this family isexcluded, the analyses are not significantly different.

t See text.

Proc. Nat. Acad. Sci. USA 71 (1974)

Dow

nloa

ded

by g

uest

on

May

9, 2

021

3592 Immunology: Marsh et al.

children nos. 6 and 8) with high IgE levels may not haveclinical allergy or significant skin sensitivity to grass andragweed pollen allergens. The young children may not yet

have received sufficient allergenic exposure (see ages of onsetof symptoms in siblings nos. 1, 2, 4, and 5); on the other hand,some members (particularly the father) may lack essentialIr genes or other "allergy genes." Second, some individualshave very low total IgE levels and yet are allergic (see siblingno. 1), but usually such individuals are sensitive to fewerallergens than people with high IgE levels. Third, there isno evidence of linkage between HL-A haplotype and IgE level.Finally, there is a clear failure to demonstrate parent-to-childtransmission of IgE-mediated sensitivity to any allergen,ruling out the possibility of a single gene model to explainallergy.

Thus, in most allergic families, a gene regulating basalserum IgE level appears to mask the role played by hypoth-esized Ir genes linked to HL-A haplotype, in controlling theexpression of specific IgE-mediated sensitivities to a numberof different allergens. This conclusion contradicts the recentfindings of Levine et al. (27), claiming to demonstrate geneticlinkage between HL-A haplotype and skin sensitivity toragweed antigen E in eight allergic families. We have severalreservations about their results, discussed in detail else-where (3, 4). Despite an attempt to rule out the influenceof allergic "expressivity" genes, we feel that most of theirfindings (particularly on several five-member families withtwo allergic members) could well be explained primarily bygenetically determined differences in IgE level rather thanHL-A haplotype. Also, the use of the highly complex multi-determinant allergen, antigen E, which crossreacts with otherragweed allergens (4), complicates analysis of specific Irgenes.

We are left with the following dilemma: our family studiesemphasize the importance of the IgE-regulating gene relativeto the hypothesized HL-A-linked Ir genes in determiningspecific IgE-mediated response, but previous populationstudies (7, 5) showing association of HL-A7 Creg with re-

sponse to the low molecular weight allergen, Ra5, point to thereverse. We believe that particularly for the more complexand more abundant allergens, the products of different allelesof Ir genes (perhaps controlling recognition of different carrierdeterminants at the T-cell level) may well be involved, espe-cially where an individual has a high capacity to synthesizeIgE. In such cases, the effects of Ir genes associated withsingle (or a limited number) of HL-A types would be maskedby the influence of the IgE-regulating gene. This predictionis borne out in a recent population study showing a clearinterrelation between Ir genes associated with HL-A8 Cregand the IgE-regulating gene in determining sensitivity to therye grass group I allergen (28, 4). Future family studiesinvestigating chromosomal linkage between HL-A type andspecific IgE response to complex allergens should most ap-propriately be directed toward the relatively rare allergicfamilies where all members have low IgE levels.

We thank Drs. S. H. Hsu, L. M. Lichtenstein, and P. S. Nor-man for providing some of the sera and clinical data, Drs. T. P.King and L. Goodfriend for giving ragweed allergens, and E.Hill, E. Jarrett, E. Weathers, P. Black, E. Siekierski, M. Askey,and the staff of the Immunogenetics Laboratory for assistance.Supported by NIH Grants Al 09565 and GM 10189 and a grantfrom the John A. Hartford Foundation, and by RCDA AI50304 (to D.G.M.). This is Publication no. 108 of the O'NeillResearch Laboratories.1. Ishizaka, K. & Ishizaka, T. (1968) J. Allergy 42, 330-363.2. Bias, W. B. (1973) in Asthma, eds. Austen, F. & Lichtenstein,

L. (Academic Press, New York), pp. 39-53.3. Black, P. L. & Marsh, D. G. (1974) Bronchial Asthma, eds.

Segal, M. & Weiss, E. (Little, Brown, Boston), in press.4. Marsh, D. G. (1974) The Antigens, ed. Sela, M. (Academic

Press, New York), Vol. III, in press.5. Marsh, D. G. (1974) Proceedings of the VIIIth International

Congress of Allergology (Excerpta Medica, Amsterdam), inpress.

6. Underdown, B. J. & Goodfriend, L. (1969) Biochemistry 8,980-989.

7. Marsh, D. G., Bias, W. B., Hsu, S. H. & Goodfriend, L.(1973) Science 179, 691-693.

8. Marsh, D. G., Bias, W. B., Hsu, S. H. & Goodfriend, L.(1973) in Mechanisms in Allergy. Reagin-mediated Hyper-sensitivity, ed. Goodfriend, L. (Marcel Dekker, New York),pp. 113-129.

9. Hamburger, R. N., Orgel, H. A. & Bazaral, M. (1973) inMechanism in Allergy. Reagin-mediated Hypersensitivity,ed. Goodfriend, L. (Marcel Dekker, New York), pp. 131-139.

10. Benacerraf, B. & McDevitt, H. 0. (1972) Science 175, 273-279.

11. Vaz, N. M. & Levine, B. B. (1970) Science 168, 852-854.12. Lapkoff, C. B. & Goodfriend, L. (1974) Int. Arch. Allergy 46,

215-229.13. Marsh, D. G., Bias, W. B., Goodfriend, L. & Ishizaka, K.

(1973) Fed. Proc. 32, 1000.14. Johansson, S. G. 0. (1968) Int. Arch. Allergy 34, 1-8.15. Ishizaka, K., Tomioka, H. & Ishizaka, T. (1970) J. Im-

munol. 105, 1459-1467.16. Malley, A., Reed, C. E. & Lietze, A. (1962) J. Allergy 33,

84-93.17. King, T. P., Norman, P. S. & Connell, J. T. (1964) Bio-

chemistry 3, 458-468.18. Amos, D. B., Bashir, H., Boyle, W., MacQueen, M. &

Tiilikainen, A. (1969) Transplantation 7, 220.19. Berg, T. & Johansson, S. G. 0. (1969) Int. Arch. Allergy 36,

219-232.20. Gleich, G. J., Averbeck, A. K. & Swedlund, H. A. (1971)

J. Lab. Clin. Med. 77, 690-698.21. Bazaral, M. & Hamburger, R. N. (1972) J. Allergy Clin.

Immunol. 49, 189-191.22. Marsh, D. G., Hsu, S. H. & Bias, W. B. (1973) Presented at

the Meeting of The American Academy of Allergy, Washing-ton, D.C.; Reported in Hosp. Prac., June, 1973.

23. McKusick, V. A. (1969) Human Genetics (Prentice Hall,Englewood Cliffs, N.J.), 2nd ed., pp. 125-141.

24. Schwartzman, R. M., Rockey, J. H. & Halliwell, R. E.(1971) Clin. Exp. Immunol. 9, 549-569.

25. Tada, T., Okumura, K. & Taniguchi, M. (1973) in Mecha-nisms in Allergy. Reagin-mediated Hypersensitivity, ed.Goodfriend, L. (Marcel Dekker, New York), pp. 43-61.

26. Bias, W. B., Marsh, D. G. & Ishizaka, K. (1973) Amer.J. Human Genet. 25, 16 abstr.

27. Levine, B. B., Stember, R. H. & Fotino, M. (1972) Science178, 1201-1203.

28. Marsh, D. G. & Bias, W. B. (1974) Fed. Proc. 33, 774.

Proc. Nat. Acad. Sci. USA 71 (1974)

Dow

nloa

ded

by g

uest

on

May

9, 2

021