Similar mutability pattern in V genes from other species

Using a mutability model that was inferred from mouse sequences to assess properties of human sequences seems justified, given that the features of somatic mutation mechanism in the two species are very similar (1998). It is, however, not clear that the same mechanism is responsible for somatic hypermutation in other species. Evidence for a mechanism that preferentially targets G and C nucleotides in shark (1993) and Xenopus () immunoglobulins suggests that their somatic hypermutation mechanism might be different from the one described in mice and humans. One way of addressing this issue, in light of the optimization features that I found in human genes, is to ask whether evidence for such optimization can be found in these species as well. This in turn would argue that the somatic mutation mechanisms in all these species might be similar. Two other complete germline sequence sets are available, sheep $V_\lambda$ and rainbow trout V_H. Both are tabulated in IMGT database. I isolated these sequences from GenBank (accession numbers taken from IMGT). I added to these sets germline V_H sequences from Heterodontus franciscii (GenBank accession numbers Z11776-Z11778, and Z11780-Z11792). This last set might not be complete, and I did not perform a complete analysis on it.

The results are intriguing. The predicted CDR mutability is higher than FR mutability in these sequences, similar to human sequences (Fig. ).

 

When I analyze the codon usage of the sheep and trout data sets, I find again evidence for codon bias consistent with low FR and high CDR mutability. The normalized rank of the average FR mutability of the sheep $V_\lambda$ sequence set with respect to its translationally neutral variants is 0.0006. For trout, this value is much higher, 0.1. Still, the germline sequence set has an average FR mutability in the lower 10% of what may be obtained with unbiased codon usage. The average CDR mutability of germline sets is signficantly higher than that of variant sets with unbiased codon usage in both these species (normalized rank 0.9761 for sheep, and 0.9825 for trout). Excluding the serine codons from the mutability calculations, the effect on CDRs is similar to the effect we observed in human sequences, namely that the rank of CDR mutability goes down (normalized rank 0.4452 for trout V_H sequences and 0.1691 for sheep $V_\lambda$). Thus, we again find that serine codons contribute to the high mutability of CDR sequences with respect to their translationally neutral variants. The surprise comes when we determine the FR mutability of trout sequences, for which the exclusion of serine codons from the mutability calculation reduces the normalized rank to 0.0411. Thus, in the framework regions of V_H sequences in trout, serine is encoded by the more mutable codons, AGY, rather than the less mutable ones, TCN. This, in fact, may be a situation where the highly mutable serine codons were ``frozen in'' the framework regions, due to the fact that a change to a TCN codon would have involved a change of the amino acid at that position. For the other amino acids, we find codon usage consistent with low FR mutability.

The set of Heterodontus sequences also has codon usage bias consistent with low FR mutability (normalized rank among translationally neutral variant sets 0.0474), and, like trout sequences, the germline sequence set has even lower FR mutability if we exclude the serine codons from the calculation (normalized rank 0.0123). Contrary to all the data sets we analyzed so far, the CDR mutability of Heterodontus sequences, though considerably higher than the FR mutability, is negatively affected by the serine codons (normalized rank 0.6746 with serine codons, 0.8047 without). This is the consequence of serine being encoded mostly by the low mutable codons TCN in Heterodontus CDRs.

I performed a similar test on Xenopus sequences, only to compare their mutability pattern to those of the Heterodontus sequences. As I mentioned previously, there are claims that a different mechanism is responsible for hypermutation in Xenopus and sharks as opposed to mammals (1992). For Xenopus, however, I used cDNA rather than germline sequences. The cDNA is obtained by reverse transcribing the messenger RNA of the cell into DNA. Thus these sequences may have already undergone somatic mutation. I extracted the sequences from Kabat database, accession numbers KADBID004348, KADBID004350-51, KADBID004353, KADBID004356-57, KADBID004359-61,

KADBID004365-66, KADBID004371, KADBID004376, KADBID004386. I translated the nucleotide sequences, and then I aligned the amino acid sequences using ClustalW (1988) algorithm running on the European Bioinformatics Institute server in Hinxton (Cambridge, UK), with the default parameters. This alignment was used to infer the CDR/FR assignments.

I find that the high predicted CDR mutability of Xenopus sequences (normalized rank 0.9909 among the variant sets with the same translation) is due to a large extent to the usage of highly mutable serine codons, AGY. When I exclude the serine codons from the mutability calculation, the CDR mutability decreases (normalized rank 0.6303 among the translationally neutral variants). There is no evidence for codon usage bias consistent with low FR mutability in these sequences. The normalized rank among the variant sets with identical translation, and unbiased codon usage is 0.4227, or 0.6954, depending on whether I do or do not include the serine codons in the mutability calculation.

The set of sheep $V_\lambda$ sequences merits particular attention. In sheep, somatic hypermutation seems to be used as a diversification mechanism involved in generating the primary repertoire (1995), presumably without stringent antigen selection. Testing the functionality of the immune receptors that are generated in this manner is probably delayed, allowing a number of mutations to be introduced in the gene. These are likely to render the sequence non-functional. We expect that selection pressure for undergoing a minimal number of FR mutations is operating in these sequences. I thus decided to analyze these sequences individually, looking for evidence for codon usage bias that would render the framework regions of these sequences resistant to replacement mutations. Indeed, I find that the predicted replacement mutability of FR nucleotides is rendered extremely low by the codon bias (Table ).

 

Table 3.3: Normalized ranks of individual sheep $V_\lambda$ sequences.

	Translationally invariant variants
Gene	$\mu_F$	$\mu_C$	$\mu_C$/$\mu_F$
SHPIGJVB	0.003	0.911	0.992
AF040900	0.002	0.987	1
AF040901	0.01	0.999	1
AF040902	0.004	0.976	0.998
AF040904	0.031	0.427	0.685
AF040905	0.005	0.999	1
AF040907	0.002	0.975	0.999
AF040908	0.003	0.87	0.982
AF040909	0.015	0.673	0.919
AF040911	0.007	0.913	0.988
AF040913	0.006	0.804	0.96
AF040914	0.022	0.995	1
AF040915	0.008	0.978	0.998
AF040916	0.019	0.673	0.909
AF040917	0.001	0.91	0.991
AF040918	0.001	0.976	0.999
AF040919	0.01	0.672	0.932
AF040920	0.014	0.674	0.921
AF040921	0.006	0.288	0.745
AF040922	0.013	0.701	0.914
AF040923	0.001	0.975	0.999
AF040924	0.003	0.975	0.998

 

Summarizing these results:

• Higher predicted mutability of CDR than of FR nucleotides is a general feature in all germline sequence sets used in this study. In addition, I tested that this property holds for two germline sequences of nurse shark antigen receptor (1998), as well as by V_H cDNA sequences from Xenopus.
• With the exception of Xenopus, all data sets that I analyzed show evidence for codon bias consistent with low FR mutability. For trout and Heterodontus V_H sequences, this bias is stronger if I exclude the serine codons from the mutability calculation. This indicates that the highly mutable serine codons, AGY, are used in the FR regions in these species at high frequencies relative to the low mutable ones, TCN. This seems an interesting situation, in which the highly mutable serine codons have been ``frozen into'' the sequence during evolution.
• The predicted CDR mutability is invariably high, although different factors contribute to it in different species.
• Contrary to the current view (1998), the mutability pattern of Heterodontus V_H sequences resembles that of mammalian, rather than Xenopus sequences.

The finding that codon usage bias consistent with low FR and high CDR mutability is present in all but one of the species that I studied, supports the hypothesis that the components of the somatic hypermutation mechanisms are similar in these species. The selection pressures that operate in these different species would be difficult to estimate. It is generally believed that selection is weaker in B cells of Xenopus (), and it is possible that the different mutability pattern in this species comes from lower selection pressure on FR mutability. After all, the CDR mutability is already higher than FR mutability in Xenopus, similar to all the other species. We cannot, however, exclude the possibility that different mechanisms are responsible for somatic mutation in Xenopus, as opposed to mammalian species.