If a humanized V region consisting of rodent CDRs transplanted into highly conserved human framework regions has been produced (which didn't require the inclusion of donor framework residues for the maintenance of binding affinity), can anything further be done to ensure minimum antigenicity? Possibly, by thinking of the reshaped V region in terms of the potential number of foreign peptides that can be derived from it, rather than in terms of the number of individual donor amino acid residues which it still contains. This makes sense, as it is peptides rather than single amino acids, presented to the immune system in association with MHC molecules, which drive immune responses (see next section). In addition, species specific motifs are carried by strings of amino acids with characteristic sequences, not by individual residues taken out of the context of those around them. Reducing the overall number of "potentially foreign" peptides can be achieved by extending the human sequences in the reshaped V region from the framework regions into the CDRs, and this is quite often possible in the case of framework 4 and CDR3.
As has been discussed earlier, choosing germline VH genes as providers of acceptor framework regions requires that a separate comparison be made between the donor V region and human J segments, in order to choose a suitable sequence for framework 4 of the reshaped product. Extending the region of comparison into CDR3 of the donor V region, so as to cover the entire J segment coding region, may avoid the unnecessary production of a stretch of amino acids which are still typically non-human (i.e. rodent), or which are neither human nor rodent. This situation is illustrated by looking at the reshaping of the VH region of the rat CD3 monoclonal antibody, YTH12.5 (Figure 6). Framework 4 of the rat monoclonal antibody VH region is already identical to equivalent part of human JH6. Therefore, if one is only considering the framework regions, there is no need to alter framework 4 as it is already "humanized". However, when account is taken of the adjacent CDR3 residues which are contributed by the JH segment, it becomes apparent that using human JH4 is a better option, even though it requires the alteration of one of the rat framework 4 residues from Thr to Leu. This extends the stretch of "human" sequence from framework 4 into CDR3.
In the above example, although the length of human sequence was effectively extended into CDR3, it didn't require that any of the donor CDR3 residues be altered. One can imagine that in some instances it may be possible to extend the human sequences by altering CDR residues, depending on the influence they have on the binding properties of the monoclonal antibody. This is most likely to be successful for CDRs 1 and 2, in cases where the donor and acceptor V regions have high degrees of homology both in the frameworks and CDRs, and so with the hypervariable regions belonging to the same canonical structure classes [62]. The replacement of any non-essential donor residues with their acceptor V region counterparts may be worthwhile because, despite their hypervariability, CDRs 1 and 2 probably carry species specific motifs like the framework regions. CDR3 is less likely to contain recognisable motifs due to the great diversity of amino acid sequence generated in this part of the V region.
Figure 7 shows a comparison between the CDRs 1 and 2 of the rat YTH12.5 VH region and those of the chosen human acceptor VH gene, VH26. In CDR1, 3 out of 5 residues are different in the two genes. The YTH12.5 VH region contains Phe, Pro and Ala at positions 32, 33 and 35 respectively. As far as can be determined on currently available data, all three of these residues are atypical of human germline VH genes [31]. The question is therefore, are they the result of somatic mutation within the YTH 12.5 VH gene, or are they present in its germline counterpart. The sequence of the germline counterpart is required to make the distinction, but, if present, they could be said to represent a rat specific motif, and their substitution with the corresponding human residues would be desirable. If they are the product of somatic mutation they may be especially important in maintaining high affinity antigen binding. The same points apply to the Thr and Arg residues at positions 50 and 60 respectively in the YTH 12.5 VH CDR2, and ultimately the only way to see if their substitution is feasible to perform the experiment. Given that the donor and acceptor CDRs have hypervariable loops of the same canonical structure classes (H1 regions are both class 1, H2 regions are both class 3) and that several of the proposed changed are conservative or semi-conservative [63], there are grounds for expecting at least partial success if this were attempted.
The new approaches to producing human antibodies which have been emerging over recent years will bypass some of the problems associated with monoclonal antibody humanization (for recent reviews see [64]). Antibodies of described specificity which are derived directly from rearranged human V gene segments, either by conventional hybridoma technology or by phage antibody technology, will not suffer from "non human" and motifs in their CDRs, or from the presence of irrelevant framework region somatic mutations. V region gene allotypic variations will still be a relevant concern. To some extent, the reshaping approach has an advantage over "human V gene rescue" methods, as it allows the V region haplotype contributing the framework regions of the final humanized antibody to be pre-defined. Although not yet tried, producing human monoclonal antibodies by the combinatorial V gene library approach in association with mutagenesis and affinity selection, beginning with a limited selection of human V genes which are well represented in the population and which exhibit a cross-section of the available structural repertoire, should be possible [65],[66],[67],[68].