Enzyme linked immunosorbent assay (ELISA) for human IgG (Gm) allotype determination

An inhibition enzyme-linked immunosorbent assay (inhibition-ELISA) was developed for the quantitative determination of human IgG (Gm) allotypes using rabbit anti-Gm antisera, alkaline-phosphatase-conjugated goat anti-rabbit IgG and, as the calibrant, purified human myeloma proteins possessing the relevant Gm allotype. The assay is reproducible and can detect as little as 10 ng/ml of G1m(a), G2m(n) or G3m(st), and 100 ng/ml of G1m(f) or G3m(g). Using this assay, the "gene dosage effect" and "allelic balance" in healthy Japanese were studied.     

A new assay uses monoclonal anti-Rh(D) antibodies to determine rheumatoid factor specificity: reactivity to a monoclonal antibody of the Gm allotype G3m(21) is more frequent in rheumatoid patients negative for G3m(21)

A new method has been developed to determine the specificities of polyclonal rheumatoid factors (naturally occurring antibodies which react with human Fc gamma) (RF) found in sera from patients with rheumatoid arthritis. In this method, monoclonal anti-Rh(D) antibodies of known IgG isotype and allotype are bound to erythrocytes and then act as the target IgG antigen for RF in a direct haemagglutination test. Using two monoclonal anti-D antibodies of the IgG3 isotype and G3m(21) allotype, which were cloned from different donors, we found that a large number of rheumatoid sera reacted with both these G3m(21) proteins. In contrast reactivity of rheumatoid sera with polyclonal anti-D of the G3m(21) allotype in the direct haemagglutination test was rare. A strong correlation was found between reactivities to both G3m(21) monoclonal anti-D antibodies but not with a monoclonal anti-D antibody carrying the alternative allele, namely G3m(5). Haemagglutination inhibition experiments using human paraproteins of known IgG isotype and allotype provided some additional evidence that this method can detect RF with specificity for the G3m(21) allotypic determinant or a related allotypic determinant in polyclonal rheumatoid sera. When each patient's autoantibody response was related to their Gm phenotype, we found that the frequency of reactivity for G3m(21) monoclonal anti-D antibodies was significantly greater in patients negative for G3m(21) than in patients positive for the G3m(21) allotype. IgM preparations from patients' sera were dissociated at acid pH but no 'hidden' antibodies were found. We suggest trans-placental sensitization as one of several possible interpretations of this finding.     

Serum IgG and IgG Subclass Contents in Different Gm Phenotypes

Different serum IgG and IgG subclass levels were found among Gm phenotypes of a normal population. One hundred and fifty-seven Caucasian blood donors were investigated for the reciprocal Gm allotypes on IgG subclass loci namely: for IgG1, Glm(f) and Glm(a); for IgG2, G2m(n) and G2m("); and for IgG3, G3m(b) and G3m(g), and subgrouped in the seven most common Gm phenotypes. The frequencies of Gm phenotypes and haplotypes were given, including numbers of the previously little known G2m(n,") heterozygous individuals. Mean serum quantities ±SD and range of IgG, IgG1, IgG2, IgG3 and IgG4 were given for different Gm phenotypes. The IgG content was significantly lower in the Gm(f,",b/f,",b) phenotype in which the IgG2 levels were also significantly lower, compared with values of the other phenotypes. IgG3 levels were significantly lower in the Gm(a,",g/a,",g) phenotype compared with other phenotypes. These data imply the importance of Gm(f,",b/f,",b) and Gm(a,",g/a,",g) phenotypes causing lower amounts of IgG antibodies. In evaluating IgG subclass deficiency, the range for the low responding Gm(f,",b/f,",b) and Gm(a,",g/a,",g) phenotypes should be considered.     

Immunoglobulin allotypes and IgG subclass antibody response to Pseudomonas aeruginosa antigens in chronically infected cystic fibrosis patients.

Chronic Pseudomonas aeruginosa lung infection is the leading cause of death in patients with cystic fibrosis (CF). Poor prognosis correlates with a high number of anti-pseudomonas precipitins and with high levels of IgG2 and IgG3 anti-pseudomonas antibodies. Reports of several highly significant associations between certain Gm (genetic markers of IgG on human chromosome 14) and Km (k-type light chain determinants on chromosome 2) phenotypes and immune responsiveness to various antigens suggest that allotype-linked immune response genes do exist in man. Furthermore correlation between Gm types and IgG subclass levels has been reported. A group of 143 CF patients were investigated (31 non-infected and 112 chronic infected). The IgG subclass antibodies to three different P. aeruginosa antigens (P. aeruginosa standard antigen (St-Ag), alginate and LPS) were determined. Immunoglobulin allotypes were determined by haemagglutination inhibition. Samples were typed for G1m(1,2,3, and 17), G2m(23), G3m(5,21), and Km(1,3). Statistical analysis of our data demonstrate that IgG3 anti-pseudomonas antibody levels and Gm markers are related. IgG3 antibody levels to all investigated P. aeruginosa antigens are significantly higher in sera homozygous for Gm(3;5), somewhat lower in heterozygous sera, and significantly lower in sera homozygous for Gm(1,2,17;21). We suggest that genetic differences between the patients may explain the present differences in subclass patterns.     

Serum Gm Allotype Development During Childhood

Gm allotypes are genetic variants of the immunoglobulin heavy G chains (IGHG) of IgG molecules, coded from chromosome 14q32, characterized by differences in amino acid epitopes of the constant heavy G chains and inherited in the Mendelian manner. Gm allotypes have influence on IgG subclass levels, and serum Gm allotype levels have been given for different Gm genotypes in adults. Four hundred and thirty healthy children, aged 1-15 years, were examined for serum Gm allotypes and IgG subclasses from the six most common Gm genotypes and different age groups were measured using competitive enzyme-linked immunosorbant assay and radial immunodiffusion methods. Quantities (in g/l) of G1m(a) and G1m(f) of IgG1, G2m(n) and G2m(-n) of IgG2 and G3m(g), and G3m(b) of IgG3 are given. Different maturation rates of the alternative Gm allotypes within IgG1, IgG2 and IgG3 were shown. G2m(n) development was strikingly retarded compared with G2m(-n) from the gamma2 locus. This was found comparing IgG2 levels from homozygous G2m(-n-n) and G2m(nn) individuals, but was also seen in heterozygous G2m(n-n) genotypes. From the gamma1 locus G1m(f) levels dominated significantly, but inconstantly, over G1m(a) levels in heterozygous G1m(af) individuals. In homozygous G1m genotypes, G1m(aa) compared with G1m(ff) of the same age, one or the other dominated, sometimes significantly. Serum levels of G3m(b) from the gamma3 locus of homozygous G3m(bb) individuals were increased significantly compared with G3m(g) levels of homozygous G3m(gg) individuals, in ages over 3 years. However, in heterozygous G3m(gb) individuals G3m(b) dominance was not evident. There is a relatively rapid development of G1m(f) molecules and a retarded development of G2m(n) in the Gm(f;n;b) haplotype. In comparison, G1m(a) is retarded and G2m(-n) is enhanced in the Gm(a;-n;g) haplotype. The retarded serum G2m(n) development is comparable with serum IgA development during childhood. Different maturation rates of Gm allotypes within the same IgG subclass provide further explanation for the variation of the antibody response during childhood. Quantitative Gm allotype determinations give information of the activity from IGHG genes. The genetic variation constitutes an additional basis for evaluation of IgG antibodies in different diseases in childhood.     

Quantitation of Gm Allotypes

A method for quantitation of Gm allotypes is described. Alternative Gm allotypes of the three IgG subclasses, IgG1, IgG2 and IgG3, were investigated for the six most common Caucasian Gm phenotypes, Quantitation of Glm(a), Glm(f) of IgG1, G2m(n) of IgG2 and G3m(b) of IgG3 was performed with specific monoclonal antisera and purified myeloma proteins of different Gm allotypes. Mean±SD are given as percentage of a normal serum pool and in g/1 for the Gm allotypes Glm(a), Glm(f), G2m(n) and G3m(b), For homozygous individuals the G2m(",") values are equal to the IgG2 levels and the G3m(g,g) values equal to the IgG3 levels. For heterozygous individuals the value for G2m(") is calculated as IgG2 minus G2m(n) and for G3m(g) as IgG3 minus G3m(b), Homozygous individuals have about double the amounts of the Gm allotype compared with heterozygous individuals. The gene activity of heterozygous individuals is given by quotients, mean±SD for G1 m(a)/G1 m(f) of IgG 1, G2m(n)/G2m(") of IgG2 and G3m(b)/G3m(g) of IgG3 in different Gm phenotypes. Heterozygous individuals on all three IgG subclass loci have at least six different qualities of IgG molecules compared with three for homozygous individuals.     

Lack of the G2m(n) Allotype in IgG Subclass Deficiency, in IgG2 Deficiency Together with Lack of G1m(a) and G3m(g), and in IgG3 Deficiency Together with Lack of G1m(f) and G3m(b)

Lack of G2m(n) was demonstrated in both IgG2-deficient and IgG3-deficient Caucasian patients. Lack of G2m(n) or G2m(",") was found together with homozygosity for both G1m and G3m allotypes as the dominant finding, i.e. for IgG2-deficient patients together with G1m (f,f) and G3m(b,b), constituting the Gm(f,",b) phenotype, and for IgG3-deficient patients together with G1m(a,a) and G3m(g,g), constituting the Gm(a,",g) phenotype. The group with IgG2 deficiency and the selected patients with the Gm(f,",b) phenotype expressed characteristically very low or undetectable IgG4, significantly increased IgG3, and normal IgG1. The group with IgG3 deficiency and the selected patients with the phenotype Gm(a,",g) expressed instead normal IgG4 and nearly normal IgG2 and IgG1 levels. The lack of G2m(n) together with lack of one or the other of the alternative G1m genes and corresponding G3m genes give different IgG2 levels and different IgG subclass patterns. The frequency of G1m allotypes and corresponding G3m allotypes also deviated significantly when the IgG2 deficiency and IgG3 deficiency groups were compared with each other. Most IgG subclass-deficient patients are homozygous in the Gm system and lack genetic variants in the three IgG subclasses, IgG1, IgG2, and IgG3.     

Evaluation of monoclonal antibodies having specificity for human IgG sub-classes: results of an IUIS/WHO collaborative study

Seventy-four monoclonal antibodies (McAb) of putative specificity for human IgG (11), the IgG sub-classes (59) or Gm allotypes (4) have been evaluated for reactivity and specificity in eight laboratories employing different assay techniques or protocols. For the IgG, IgG3, IgG4, G1m(f) and G3m(u) specificities McAb have been produced that can be satisfactorily applied in most methodologies employed and have potential as reference reagents. The IgG1 and particularly IgG2 specificities proved problematical with all McAb evaluated demonstrating apparent assay restriction and whilst performing well in some assays proved to be poor or inactive reagents in others. However, the study identifies McAb individually suited to application within most commonly employed methodologies. Epitope display is the probable variability rather than capricious behaviour by the McAb. IgG1 and IgG2 were the least immunogenic of the sub-class proteins and there is evidence that epitope display is influenced by the physical and chemical procedures used to immobilize or fix antigen - a common requirement in the assay systems studied.     

Monoclonal antibodies against IgG allotypes G1m(z), G1m(a), G1m(f), G3m(b1/u) and G3m(g1): their usefulness in HAI and capture ELISA

Monoclonal antibodies (McAbs) were produced against the IgG allotypes G1m(z), G1m(a), G1m(f), G3m(b1/u) and G3m(g1). Four out of the six McAbs described in this paper showed in the haemagglutination assay cross-reactivity with some or all IgG-coated cell samples. In the haemagglutination inhibition assay, all six McAbs are useful as typing reagent for the above allotypes. In this assay, two of the McAbs show two different specificities, which depend on the Ig-coated cell sample used. Five McAbs are useful for allotyping in a capture ELISA. The results with four of these are promising for the development of a quantitative determination of Gm allotypes.     

Direct antiglobulin reactions in Gambian children with P. falciparum malaria. III. Expression of IgG subclass determinants and genetic markers and association with anaemia.

The IgG subclass and Gm allotype distribution of red cell-bound IgG molecules from Gambian children with past or present falciparum malaria has been determined. The results show that the antibody is polyclonal with some predominance of the IgG2 and IgG4 subclasses. A restriction towards IgG2 antibodies may indicate specificity for a schizont-derived antigen which is carbohydrate in nature. Not all the allotypes normally carried by the G3m(b) allele could be demonstrated on IgG3-sensitized cells, a finding which remains unexplained. Expression of G3m(10), (11) and (14) allotypes of the IgG3 molecule was noted. Sensitization of red cells with IgG1 molecules correlated with the presence of anaemia but red cells sensitized with IgG2, IgG3 or IgG4 usually came from children with haematological findings within the normal range for that population. The implications of the results are discussed with reference to the few reports on the subclass and Gm allotype of malaria-specific IgG.     

Interactive effect of Gm allotypes and HLA-B locus antigens on the human antibody response to a bacterial antigen.

Two hundred healthy adults were immunized with 1 microgram of the bacterial antigen monomeric flagellin from Salmonella adelaide, and grouped as responders and non-responders on the basis of a rise in titre of antibody 2 weeks after immunization. Immunoglobulin allotypes G1m(a), G1m(z) and G3m(g) were more frequent among responders who made immunoglobulin (Ig)G antibody (P less than 0.02), and HLA-B12 was more frequent among responders who made IgM antibody (P less than 0.05). The mean log titre of IgG antibody was higher in females (P less than 0.001), in subjects with T1m(a), G1m(z) and G3m(g) allotypes (P less than 0.05), and in Gm heterozygotes (P less than 0.01). The mean log titre of the IgG antibody response in subjects with particular Gm phenotypes was also dependent on the HLA-B locus phenotypes HLA-B7, B8 and B12 (P less than 0.005); for example, among those with the phenotype Gm(a-x-b) subjects with HLA-B7 were low responders and those with HLA-B8 were high responders. These findings are consistent with the hypothesis that there are immune response genes within the major histocompatibility complex (MHC) which interact with Gm-linked genes in determining levels of serum antibodies of different isotypes and specificities.     

The Importance of G1m and 2 Allotypes for the IgG2 Antibody Levels and Avidity Against Pneumococcal Polysaccharide Type 1 Within Mono- and Dizygotic Twin-Pairs

Eighty-two mono-or dizygotic Caucasian twins vaccinated with a 23-valent pneumococcal vaccine, who had previously had their IgG2 antibody levels to pneumococcus type 1 determined before and after vaccination, were included in this study. Their IgG2 antibody levels were related to their G1m and G2m allotypes/phenotypes and their Gm amounts.
Eight different Gm phenotypes were found and characteristically IgG2 antibody levels were related to them. G2m (n) homozygotic twins had significantly higher IgG2 levels than heterozygotic twins who had significantly higher levels than G2m (-n) homozygotic twins (P <0.05). The G1m allotype, on the other hand was without influence on the IgG2 levels and so were the Gm amounts among G2m (n) heterozygotic twins. The IgG2 antibody avidities were not related to Gm allotypes but significantly correlated to IgG2 levels (P = 0.05). Finally, a highly significant intra-pair correlation was found for avidity in the monozygotic twins supporting a genetic regulation of avidity (P <0.002).
These results may explain our earlier findings that IgG2 antibody levels after pneumococcal vaccination are significantly more closely correlated within mono-compared to dizygotic twins.     

A quantitative method for determining gamma G allotype antigens. 3. Gm genes and gene complexes in different human populations

The sera expression of human immunoglobulin heavy chain genes and gene complexes was analyzed in different ethnic groups by quantitative determination of Gm allotype antigens. In all populations tested, IgG3 subclass heavy chains of Gm(b+) allotype had greater sera expression than the allele Gm(g). The percentage of IgG3 allotype molecules was higher in Chinese and Easter Islanders than in Caucasians due to higher Gm(b) values. In general, IgG1 allotypes did not show quantitative differences correlated with Gm genotype. A possible exception was the Gm^{za, g} gene complex which had a lower Gm(a) concentration than comparable Chinese sera of different genotype. Certain similarities in the regulation of Gm genes were consistently observed among the populations tested; these included the restricted range of percentage of IgG1 allotype antigens, a gene dosage effect, and differential gene activity of Gm^b and Gm^g IgG3 Gm genes.     

Computer localization of some Gm markers on the surface of the Fc region of human immunoglobulin G.

The surface localization of some Gm markers on the Fc fragment of IgG has been identified from previously published amino acid sequences associated with known Gm markers using the atomic coordinates described by Deisenhofer, INSIGHT software and a Digital VAX 11/785 computer, which together permit a study of the three-dimensional structure of the Fc fragment. The G1m(x)-associated amino acid residue 431, the G3m(s)- and G3m(u)-associated residue 435 and the nG4m (a)- and (b)-associated residue 309 are all localized in the interface between the CH2 and CH3 domains. Furthermore, it is postulated that the G1m(a)-associated residue 356 (Asp, Glu) influences the interface formation through an ion pair interaction to Lys 439. Finally, G3m(b) and G3m(g) are associated with the interface via residues 435 and 436. The data explain why sera from patients with rheumatoid arthritis are useful tools for the detection of some Gm markers and support the view that rheumatoid factors from these patients are internal images of microbial Fc-binding proteins.     

Epitope characterization of pre-existing and developing antibodies to an aglycosylated monoclonal antibody therapeutic of G1m17,1 allotype

Allotypes of IgG1 molecules can influence the immunogenicity of therapeutic monoclonal antibodies and may account for the presence of some pre-existing antibodies. An electrochemiluminescent (ECL) bridging immunoassay was used to characterize the binding epitopes of anti-therapeutic antibodies (ATAs) in a Phase 1 single ascending dose clinical trial of a therapeutic aglycosylated IgG1monoclonal antibody (mAb). There was no evidence for ATAs specific for a possible neo-epitope created due to the lack of glycosylation. ATAs that developed post-treatment were specific for the F(ab')2, whereas, pre-existing ATAs were specific to the Fc region. Further characterization of the pre-existing ATAs identified the specific epitope to be the G1m1 allotype determinant in the Fc of the therapeutic. A novel competitive bridging assay was developed to verify that serum IgG1 from subjects with pre-existing anti-G1m1 antibodies was homozygous for the antithetical allotype (G1m3). The endogenous G1m allotype of all subjects was assessed and correlation to ATA incidence and adverse events was evaluated. Interestingly, the pre-existing anti-allotype antibody in subjects persisted but was not augmented after dosing, indicating the lack of a secondary immune response to this epitope. These studies indicate the relationship of the therapeutic allotype and the corresponding allotype of subjects is an important component to further understand the impact of immunogenicity on the safety and efficacy of therapeutic antibodies.     

Influence of the G2m(n) allotype and age on IgG subclass distribution in antibodies to dietary proteins in children with coeliac disease

In 47 children with coeliac disease, IgG1 was the dominant IgG subclass (mean 61%) in antibodies to gliadin, a protein component of wheat; IgG2 (22%) and IgG3 (15%) were also found. Very little IgG4 (3%) was detected. Both G2m(n)allotype and the age of the patient had an independent effect on the subclass distribution. The effects of these two factors seem similar to those described earlier on responses to polysaccharides; both G2m(n) and age increase the expression of IgG2 antibodies and concomitantly the share of IgG1 is decreased. The average increase in the share of IgG2 brought about by n/n-genotype was 25.3% (95% confidence limits 12.2-38.4%; P less than 0.001) and that brought about by each additional year of age of the patient 1.1% (0.3-1.9%; P = 0.006). IgG antibodies to beta-lactoglobulin and ovalbumin consisted mainly of IgG1 and IgG4. The titres of the IgG subclasses were low and in many cases below detection level. No association with G2m(n) genotype and share of IgG1 or IgG4 could be demonstrated. Age was inversely correlated with the shares of IgG1 in antibodies to beta-lactoglobulin and ovalbumin.     

IgG (Gm) allotypes and multiple sclerosis in a French population: phenotype distribution and quantitative abnormalities in CSF with respect to sex, disease severity, and presence of intrathecal antibodies

The association of a given Gm allotype or phenotype with MS susceptibility, as previously described in some Caucasian populations, was not observed in a large French MS group, whether or not considering the possible influence of sex or disease severity. This result could be related to variations in geographical distribution of Gm alleles and MS susceptibility gene(s) or suggests the simultaneous involvement of Gm and other genetic system(s). In contrast, the corresponding CSFs exhibited already known MS-associated abnormalities of IgG1 (G1m) allotype contents, which therefore did not merely result from a Gm-associated MS susceptibility. These quantitative abnormalities were not sex dependent, but may fluctuate with MS severity. The G1m allotype levels in each CSF were not correlated with titers of various intrathecal antibodies but with the number of antibody specificities detected, a picture arguing for a polyclonal, non-antigen-specific activation of G1m allotype-producing B cells present in MS brain.     

Association of CSF IgG concentration and immunoglobulin allotype in multiple sclerosis and optic neuritis

The cerebrospinal fluid (CSF) and serum from 64 patients with multiple sclerosis (MS) and 47 patients with monosymptomatic optic neuritis (ON) were analyzed for the distribution of allotypic determinants on IgG and compared to similar samples from 51 patients with other neurological diseases (OND) as well as to serum samples from 97 healthy controls. The results indicate a significantly increased frequency of the haplotypes Gm a;g and Gm a,x;g among MS patients (P = 0.024) with an associated increase in relative risk for MS among individuals with the Gm a,(x);g haplotypes compared to those individuals without them (P = 0.014). Among MS patients, those with the Gm a,(x);g haplotypes had significantly higher CSF levels of IgG than those without (P = 0.016); levels of serum IgG did not covary with Gm haplotype. Two-way analysis of variance indicates that familial cases have significantly higher levels of CSF IgG than nonfamilial cases (P less than 0.001) and that familial cases with the Gm a,(x);g haplotypes have the highest CSF IgG levels (P less than 0.005). There was no correlation between Gm haplotype and CSF or serum IgG levels in patients with ON or OND. The allotype effects were independent of age at onset and duration of disease. In all patients, regardless of disease classification, the phenotypes found in serum samples were identical to those found in CSF samples. The data presented support the hypothesis that the etiology of MS has as one of its parameters an immunoregulatory/immunogenetic factor. The successful analysis of these various parameters will provide useful information not only about MS but also about general principles of human immune responsiveness.     

Linkage of IgA deficiency to Gm allotypes; the influence of Gm allotypes on IgA-IgG subclass deficiency

IgA deficiency (IgAD) is the most common immunodeficiency, characterized by an arrest in B cell differentiation. It has a sporadic occurrence or variable inheritance pattern, and is also linked to the HLA genes. IgA deficiency is sometimes associated with IgG subclass deficiency. In this study the Gm allotypes, as genetic characteristics of the IgG1, IgG2 and IgG3, were analysed in 83 Caucasian IgAD individuals. Half of the patients presented with IgG4 < 0·01 g/l compared with 5% (P < 0·001) in a healthy population. Three of the 83 had significantly low IgG2 and four had significantly low IgG3 levels. Gm allotype frequencies in IgAD deviated compared with a normal population. Of the 83 patients, 44 (53%) showed homozygous G2m(“,”) expression on the IgG2 locus (33% in controls, P < 0·01). In IgAD the Gm(a, g) haplotype was more frequent (43%) compared with controls (31%, P < 0·01). The Gm homozygous phenotype Gm(a', g/a', g) was most common, found in 20 of 83 patients (24%, P < 0·05) compared with controls (14%). On the other hand the Gm(f,n,b) haplotype of IgAD was rare (28%) compared with controls (45%, P < 0·001). The low IgG4, < 0·01 g/l, found in 50% of the patients, was even more frequent (56–69%) among the G2m(“,”) phenotypes. IgG subclass levels were given for different Gm phenotypes of the IgAD group and compared with controls. Significantly low IgG4 was revealed in the Gm(a,“,g/a,”,g) phenotype (P < 0·01) and significantly low IgG2 in the Gm(a,“,g/f,”,b) phenotype (P < 0·01). The Gm(a,“,g/f,”,b) phenotype contained the three patients found with IgG2 levels < - 2 s.d., and the four patients with IgG3 levels < -2 s.d. were present among those with the homozygous Gm(a,“,g/a,”,g) phenotype; both phenotypes with G2m(“,”) on the IgG2 locus. The ‘compensatory’ increase of IgG was significant for both IgG1 and IgG3 in all Gm phenotypes, but in the Gm(a,“,g/f,”,b). Thus, the susceptibility of IgAD with the additional IgG antibody deficiencies, down-regulated IgG4 and IgG2/IgG3, is associated with Gm allotypes, especially the homozygous G2m(“,”) expression on the IgG2 locus.     

Quantitative studies of Gm allotypes. II. G1m(1), G3m(5) and G3m(21) in sera from healthy caucasoid blood donors, and in a family with the inheritance of a gene responsible for a weak Gm1,17;..;21 haplotype.

The quantitative expression of three allotypes--G1m(1), G3m(5) and G3m(21)--has been studied in normal caucasoid sera. A gene dosage effect, previously described for all of these allotypes, is not noticed for G3m(5) in the present study. The different content of G3m(5) and G3m(21) IgG3 molecules in heterozygous Gm5/Gm21 sera has been verified: these contents are respectively 75% and 25% of the IgG3 content. The same allotypes have been studied in a family which shows the inheritance of a weak Gm1,17;21 haplotype. The gene responsible for this abnormality is probably a new regulatory gene, and is transmitted without the 'aimed' haplotype.