So far considerable attention has been paid to the engineering of the immunoglobulin isotype but very little to problems of immunoglobulin allotype. A number of allelic forms of human immunoglobulins have been identified [46] (reviewed by van Loghem [47]). The epitopes have been mapped to amino acid substitutions for some but not all of these differences. Of interest is the classification of markers into 'allotypes' and 'isoallotypes' [46],[47]. These are distinguished on different serological bases dependent upon the strong sequence homologies between isotypes. Allotypes are sequence differences between alleles of a subclass whereby the antisera recognises only the allelic differences. An isoallotype is an allele in one isotype which produces an epitope which is shared with a non-polymorphic homologous region of one or more other isotypes and because of this the antisera will react with both the relevant allotypes and the relevant homologous isotypes.
There are several reasons to be concerned about immunoglobulin allotypes with regard to the engineering of antibodies for therapy. Firstly, when a patient has a different immunoglobulin allotype from the therapeutic monoclonal then there is an increased potential for an antiglobulin response, directed to both the constant region and the variable region (discussed in more detail in chapter 2). The antiglobulin response to allotypes is the major concern here - an anti-isoallotype response is less likely to occur because of the cross reactivity with the patients own immunoglobulins of different isotype. One way to avoid this problem would be to create a panel of matched therapeutic antibodies with the different allotypes and then to match the therapeutic antibody to the patient. Commercially this approach is unfeasable because of the complications in manufacture and testing of several similar products each with a restricted application. An alternative approach is to select the most common allotype for general use. Unfortunately, the frequency of different alleles varies between different racial groups [46],[47]. A possible solution to this is to exploit the idea of isoallotypes by artificially creating a new allele, consisting entirely of isoallotypes [48]. An example of such an allele is described in Figure 4. This antibody has been constructed with the CAMPATH-1 specificity and does not react with the tested anti-G1m allotypic antisera whilst still retaining its effector function activity in-vitro. A clinical trial of this antibody will be necessary to establish whether most patients will fail to a make an antiglobulin response to the constant region.
Another question concerning allotypes relates to polymorphisms affecting the function of the antibody. At present there is very little evidence of gross differences in effector functions of human IgG allotypes when compared to the big differences seen between subclasses, for example IgG1 versus IgG4. However a detailed comparison between matched sets of allotypes has not yet been carried out. In the NP specific series of antibodies described above, two different alleles of IgG3 were compared and there were differences in titres in complement activation and lysis [15],[28], see Figure 1 . In a recent set of experiments a matched set of IgG1 allotype antibodies with specificity for CAMPATH-1H have been compared in complement mediated lysis and no significant differences were seen for the "wild type" forms [48] (see Figure 4 and Figure 5 ). However when the allotype differences were combined with a Asn/Ala 297 mutation which results in an aglycosyl form of the antibody then a difference was observed (see Figure 4 & Figure 5 ). The carbohydrate on antibodies has been shown to be important for most effector functions [34],[35],[49],[50] and aglycosyl antibodies may have a therapeutic use in situations where activation of effector functions need to be minimised [51]. It is interesting to speculate on the functional differences seen between these two aglycosyl mutants in Figure 5 since the only structural differences are in the allotypic residues in the C H 1 and C H 3 region (see Figure 4 ) and do not involve any of the residues from the C H 2 and hinge region discussed above. Studies on an aglycosyl mutant of IgG3 specific for NP have also been described and in this case there was also some residual activity in complement activation observed [50]. The IgG3 isotype has an Arg at position 214 which is the same as the allotype G1m(3) and different to the Lys found at this position in the allotype G1m(17).
Allotype differences between individuals have also been implicated in disease associations [52] although the linkage is often very weak, suggesting either that the allotype is only a minor contributing factor or that it is in linkage disequilibrium with a much more important locus, possibly even the immunoglobulin V-gene segments. Polymorphisms of the immunoglobulin genes are further complicated by polymorphisms in the Fc receptors (reviewed by Ravetch & Anderson [53]) and complement components (reviewed by Campbell et al [54]) with which they interact and so it is not yet possible to make clear statements about the role of any given antibody.