Re-educating myself about the light chain

I have an unconscious habit of personification, and I always see the antibody light chain as lazy for not contributing more residues to binding interfaces (obviously a generalisation – e.g. insertions in CDRL4 in anti-HIV bNAbs [1]). Perhaps this is why I have a personal preference for the more diverse [2] heavy chain with its specificity-determining [3] CDR3. Having written this down, I realised it’s actually pretty weird to consider an antibody chain as a person and I ought to re-educate myself about the role that light chains play.

I had a gander at the non-redundant (100% amino acid identity across VH/VL) set of antibody/antigen complexes from SAbDab (August 2020 – 1612 structures). First, a quick look at number of paratope residues (using a standard cut-off distance of 4.5 angstrom):

Distribution of number of contacts (contact distance of 4.5Å) across the heavy and light chain of 1612 unique VH/VL pairs in complex with protein, peptide or hapten antigens. Most antibodies in SAbDAb co-crystal structures use both light and heavy chain for binding (1510/1612). Most antibodies use more heavy chain paratope residues than light chain paratope residues (79%), but most of the leftover (14%) have more light chain paratope residues than heavy. Only nine structures had exclusively light chain paratope, while 60 structures had exclusively heavy chain paratopes.

Then I looked at the paratopes using Arpeggio [4], which annotates interatomic contacts with a structural interaction fingerprint. I used the subset of 1209 ab/ag complexes from the above with a protein or peptide antigen. Arpeggio extracted chemically sensible interactions for 1091 of those interfaces. The total interaction type breakdown according to chain type is as below:

Arpeggio contact classifications of 1209 antibody-antigen complex interfaces, stratified by chain type. On a chain-by-chain basis (A) the distributions of contact type are similar and random forest classification of heavy/light chains based on contact type usage isn’t much better than random on the test set. (B) The distributions of contact type considering the aggregate of contacts is, unsurprisingly, similar.

This is just a cursory analysis of SAbDab that I did to get the most up-to-date stats possible. For more a more comprehensive analysis of antibody/antigen interfaces (and nicer figures) I refer the reader to [5]. It would be interesting to look at binding motifs with Catherine’s tool Ab-Ligity [6] to see whether there is something special about heavy and light chain binding. My only conclusion, unsurprisingly, is that the heavy chain does contribute on average more residues to the binding interface; a check with Arpeggio reveals that the contact-type usages are not easily separable.

Observational and other experimental evidence:

  • Heavy chain only antibodies (HCAbs) evolved convergently in camelids (VHH) and twice in cartilaginous fish (VNAR) (all also use heterodimeric antibodies) – so HCAbs have evolved three times during vertebrate evolution (a fourth time if you count the chimeric Ig/TCR locus-derived single-domain antibodies in monotremes) [7].
  • Harris et al created a transgenic rat with a diverse heavy chain repertoire but a fixed kappa light chain, called OmniFlic. The platform was used to recover an abundance of antigen-specific antibodies targeting nine antigens, and clonotype diversity was comparable to that observed with unrestricted light chain use [8]. High-affinity, antigen-specific antibodies were also recovered from a transgenic rat expressing heavy-chain only antibodies , UniRat [9].
  • Sidhu et al generated high-affinity antibodies using fixed light chain phage-display libraries, and concluded it may be possible to dispense with the light chain altogether (in the therapeutic antibody context) [10].

In conclusion, heavy-chain antibodies have evolved separately in camelids and cartilaginous fish, and to make matters worse, fixation of the light chain does not preclude the production of high-affinity antibodies in either transgenic models or phage display libraries. While species that have been using HCAbs since before it was cool do also use heterodimeric antibodies, transgenics with silenced light chain loci seem to get on fine too [9]. In the absence of evidence that the light chain is necessary for complementarity to any particular antigen, I found myself reneging on my new found interest in light chains.

That was until I read an interesting paper which concluded that the light chain is so much less diverse than the heavy chain because the maximand of the light chain repertoire is not diversity but the prevention of autoreactivity [2]. The relative transcriptional orientations of lots of the V and J genes in the murine and human kappa loci mean that after recombination, autoreactive sequences get a second chance – the intermediate genes between the chosen V and J are not lost and secondary rearrangements can occur (receptor editing). The lambda loci are also permissive of receptor editing in humans. To return to my annoying habit of anthropomorphism, light chains have basically sacrificed the attention that heavy chains get in order to help the heavy chains out in producing self-tolerant B-cells. I would really recommend this paper for a clear and detailed description of the light chain loci, and a sense of guilt that you haven’t given the light chain the respect it deserves 🙁

In conclusion, I think that my preferential treatment of the heavy chain with respect to binding is somewhat justified – if I had to pick a single chain to defend myself, I would still go down the camelid route. However, I have a new found respect for the light chain. I now know that it isn’t lazy, but is likely quietly performing an important job that I am very thankful for. Cheers to light chains!

Eve

[1] Kelow, Simon P., Jared Adolf-Bryfogle, and Roland L. Dunbrack. “Hiding in plain sight: structure and sequence analysis reveals the importance of the antibody DE loop for antibody-antigen binding.” BioRxiv (2020).

[2] Collins, Andrew M., and Corey T. Watson. “Immunoglobulin light chain gene rearrangements, receptor editing and the development of a Self-Tolerant antibody repertoire.” Frontiers in immunology 9 (2018): 2249.

[3] Xu, John L., and Mark M. Davis. “Diversity in the CDR3 region of VH is sufficient for most antibody specificities.” Immunity13.1 (2000): 37-45.

[4] Jubb, Harry C., et al. “Arpeggio: a web server for calculating and visualising interatomic interactions in protein structures.” Journal of molecular biology 429.3 (2017): 365-371.

[5] Akbar, Rahmad, et al. “A compact vocabulary of paratope-epitope interactions enables predictability of antibody-antigen binding.” bioRxiv (2019): 759498.

[6] Wong, Wing Ki, et al. “Ab-Ligity: Identifying sequence-dissimilar antibodies that bind to the same epitope.” BioRxiv(2020).

[7] Flajnik, Martin F., Nick Deschacht, and Serge Muyldermans. “A case of convergence: why did a simple alternative to canonical antibodies arise in sharks and camels?.” PLoS Biol 9.8 (2011): e1001120.

[8] Harris, Katherine E., et al. “Sequence-based discovery demonstrates that fixed light chain human transgenic rats produce a diverse repertoire of antigen-specific antibodies.” Frontiers in immunology 9 (2018): 889.

[9] Clarke, Starlynn C., et al. “Multispecific antibody development platform based on human heavy chain antibodies.” Frontiers in immunology 9 (2019): 3037.

[10] Sidhu, Sachdev S., et al. “Phage-displayed antibody libraries of synthetic heavy chain complementarity determining regions.” Journal of molecular biology 338.2 (2004): 299-310.

Author