On 2016 Mar 22, Joshua L Cherry commented:

This excellent study provides much useful information, but some important points are not apparent from the article. An important claim of the article is that the unmutated common ancestor (UCA) antibody binds the founder ENV protein with high affinity. However, several amino acids in the important CDR3 of the heavy chain “UCA” cannot have been encoded by germline sequences. These include most of the RGQLVN sequence that comprises six of the twelve paratope residues in the heavy chain. Either the nucleotide positions in question represent mutated germline residues, or they represent random additions by TdT. In the first case, the “UCA” that bound antigen with high affinity actually included mutations, presumably the product of affinity maturation. In the second case, we cannot know the identities of the original, unmutated nucleotides, and it is reasonable to suspect that the true UCA had a different sequence and a lower affinity for ENV. In any case, the fact that so many important residues are not encoded by the germline might mean that antibodies of this type are rare in the naive repertoire and therefore difficult to elicit.

The reconstructed UCA sequence contains 27 bases between the V and J regions. The longest perfect match to any germline D segment is just five bases. Five bases would be an unusually small size for a D region; less than 1% of the naive B cell repertoire contains a D region of five or fewer bases. Furthermore, a total of 22 N nucleotides, though not extraordinary, would be larger than typical. This suggests that some of the “UCA” bases in this region, including some of those encoding the critical RGQLVN sequence, are actually mutated D nucleotides, presumably selected for higher affinity to ENV.

Suppose, though, that the 22 bases that do not match the germline indeed represent N nucleotides, randomly added by TdT (with perhaps a small contribution from P nucleotides). In that case, we cannot know with any certainty what these bases were in the unmutated ancestor. If these bases changed early in the affinity maturation, perhaps greatly increasing the more modest affinity of the actual unmutated ancestor, there would be no way to know this.

The authors did construct three other candidate UCAs that differed at one or two amino acids. These, however, encompass only a small component of the uncertainty in the UCA sequence. Furthermore, results with these variants illustrate the concern raised here. Two of the variants bore an E at the third position of the critical hexapeptide, which is consistent with a longer D region. These more plausible UCA antibodies exhibited a 3.8- to 7-fold reduction in ENV affinity compared to the candidate containing RGQLVN.

If the UCA sequence is correct, the fact that several of the most important residues are encoded by randomly added nucleotides may have implications for the ease of eliciting similar antibodies through vaccination. Sequences encoded by commonly used germline segments will be present at reasonable frequencies in the naive B cell repertoire. Sequences encoded by N nucleotides will be rarer. If they must be associated with unusually short D regions, the combination will be rarer still. To the extent that a vaccination strategy is aimed at eliciting antibodies of this type, this may be a problem.

On 2016 Mar 22, Joshua L Cherry commented:

This excellent study provides much useful information, but some important points are not apparent from the article. An important claim of the article is that the unmutated common ancestor (UCA) antibody binds the founder ENV protein with high affinity. However, several amino acids in the important CDR3 of the heavy chain “UCA” cannot have been encoded by germline sequences. These include most of the RGQLVN sequence that comprises six of the twelve paratope residues in the heavy chain. Either the nucleotide positions in question represent mutated germline residues, or they represent random additions by TdT. In the first case, the “UCA” that bound antigen with high affinity actually included mutations, presumably the product of affinity maturation. In the second case, we cannot know the identities of the original, unmutated nucleotides, and it is reasonable to suspect that the true UCA had a different sequence and a lower affinity for ENV. In any case, the fact that so many important residues are not encoded by the germline might mean that antibodies of this type are rare in the naive repertoire and therefore difficult to elicit.

The reconstructed UCA sequence contains 27 bases between the V and J regions. The longest perfect match to any germline D segment is just five bases. Five bases would be an unusually small size for a D region; less than 1% of the naive B cell repertoire contains a D region of five or fewer bases. Furthermore, a total of 22 N nucleotides, though not extraordinary, would be larger than typical. This suggests that some of the “UCA” bases in this region, including some of those encoding the critical RGQLVN sequence, are actually mutated D nucleotides, presumably selected for higher affinity to ENV.

Suppose, though, that the 22 bases that do not match the germline indeed represent N nucleotides, randomly added by TdT (with perhaps a small contribution from P nucleotides). In that case, we cannot know with any certainty what these bases were in the unmutated ancestor. If these bases changed early in the affinity maturation, perhaps greatly increasing the more modest affinity of the actual unmutated ancestor, there would be no way to know this.

The authors did construct three other candidate UCAs that differed at one or two amino acids. These, however, encompass only a small component of the uncertainty in the UCA sequence. Furthermore, results with these variants illustrate the concern raised here. Two of the variants bore an E at the third position of the critical hexapeptide, which is consistent with a longer D region. These more plausible UCA antibodies exhibited a 3.8- to 7-fold reduction in ENV affinity compared to the candidate containing RGQLVN.

If the UCA sequence is correct, the fact that several of the most important residues are encoded by randomly added nucleotides may have implications for the ease of eliciting similar antibodies through vaccination. Sequences encoded by commonly used germline segments will be present at reasonable frequencies in the naive B cell repertoire. Sequences encoded by N nucleotides will be rarer. If they must be associated with unusually short D regions, the combination will be rarer still. To the extent that a vaccination strategy is aimed at eliciting antibodies of this type, this may be a problem.