FEBS Lett. 2017 Sep;591(18):2936-2950. doi: 10.1002/1873-3468.12774.

Dynamics behind affinity maturation of an anti-HCMV antibody family influencing antigen binding.

Di Palma F1, Tramontano A1,2.

1 Department of Physics, Sapienza – Università di Roma, Italy.
2 Istituto Pasteur Italia – Fondazione Cenci Bolognetti, Roma, Italy.


The investigation of antibody affinity maturation and its effects on antigen binding is important with respect to understanding the regulation of the immune response. To shed light on this crucial process, we analyzed two Igs neutralizing the human cytomegalovirus: the primary germline antibody M2J1 and its related mature antibody 8F9. Both antibodies target the AD-2S1 epitope of the gB envelope protein and are considered to establish similar interactions with the cognate antigen. We used molecular dynamics simulations to understand the effect of mutations on the antibody-antigen interactions. The results provide a qualitative explanation for the increased 8F9 peptide affinity compared with that of M2J1. The emerging atomistic-detailed description of these complexes reveals the molecular effects of the somatic hypermutations occurring during affinity maturation.

PMID: 28771696



In order to introduce this work, we would like to point out few bottom lines on which the project has been thought and designed. First of all, affinity maturation increases antibody (Ab) effectiveness through many cycles of somatic hypermutation of the genes responsible of the immunoglobulin (Ig) encoding in B cells, along with a strict selection in the germinal center based on the ability to recognize and bind the antigen. Thanks to this process, Abs increase their affinity, resulting in their competence in the neutralization of infections and acting as an anti-pathogen driving force [1]. The just described heavy pressure on the germline Igs produces by means of an uncommon mutation-rate very different functional mature Abs (when compared with their germline-encoded counterparts). An important outcome of this process is an increase of many orders of magnitude in the affinity of the mature Abs for antigen compared to the corresponding Ig receptor exposed on naïve B cell surface [2].

In the second place, Human Cytomegalovirus (HCMV) remains one of the major opportunistic pathogens that causes a large spectrum of clinical manifestations not just limited to immunosuppressed patients [3]. Therefore, a great effort has been made on the investigation of possible strategies to take down such a pathological cause, which poses high risk to mankind [4]. Several anti-HCMV vaccine attempts have been made using different viral peptides as target, but none, more than the experimental trials, has passed the first steps of the clinical tests yet [5]. Thus, alternative approaches must be considered in order to change the point of view aiming at an in-depth understanding of the cell-mediated immune response against HCMV. Ultimately, in this regard, a motion picture of the interactions occurring between the viral epitopes and the different antibodies elicited by the immunological system during the affinity maturation process against following HCMV infections is still lacking.

The atomistic details of the antibody-antigen (Ab-Ag) complexes sampling multiple conformations can be obtained using a full-atom molecular dynamics (MD) [6] approach considering a germline immunoglobulin and its directly related mature antibody. This is exactly the case of our work in which we considered for our simulations two anti-HCMV antibodies: the germline Ig M2J1 and the somatically mutated 8F9 (Fig. 1) [7]. Their antigen is the most abundant HCMV envelope protein (the glycoprotein B, gB). In particular, the Abs target site 1 of the antigenic domain 2 (AD-2S1) on the gB extracellular surface fragment, a linear decapeptide. After the first encounter with AD-2S1, the B-lymphocytes exposing M2J1 undergo the affinity maturation process, thus producing 8F9. All the mature amino acids contacting the antigen are conserved from the germline sequence, thus the somatic mutations seem not to involve the residues that directly contact the antigen. As a consequence, the arising hypothesis (by McLean at al. [8] and by Thompson et al., [7]) that we explored in our in silico study, is that the 8F9 mutations were selected since they indirectly stabilize the antigen binding. Our main objective in this study was to sample the differences in the inter- and intra-molecular interactions by simulating the two Ab-Ag complexes in order to be able to explain the germline Ab weaker affinity compared to the mature one. Furthermore, the performed analysis provided a residue by residue clarification of the rationale behind the key mutations occurring during the affinity maturation process in the anti-AD-2S1 Ab mini-family.

Analyzing the root-mean-square-fluctuation (RMSF), we can generally observe that in the Mature simulation the standard deviation of the residues from their average position is lower than the corresponding ones in the germline Ab (Fig. 2). This behaviour is immediately apparent for the amino acids of the complementarity-determining regions (CDRs, responsible for the Ig interaction with the antigen) in both Light (L) and Heavy (H) chains. In particular especially the L chain (top panel Fig. 2), in which the curve trend of L1, L2 and L3 in 8F9 is steadily under the same loops of M2J1.

The idea is that to a wider residue fluctuation could correspond a less stable or tight interaction network when one or more amino acid among the pool of mutations occurred during the affinity maturation are involved. To deeply investigate this interesting hypothesis we reconstructed all the different networks of interaction involving directly the Ag and the residues of the CDRs, in addition to the residues of the frameworks that could cooperate to indirectly stabilize the direct contacts creating the required inter-molecular webs necessary to support them.

The outcomes of the average distances matrix analysis (Fig. 3) were coherent with the results provided by the interaction networks analysis described in details in our paper. The residues of the mature CDR loops H1, H3, L1 and L3 were more tightly interacting, thus more stably, with the Ag amino acids than the germline ones (highlighted ellipsis in Fig. 3). Furthermore, if in the Germline simulation the average distances between the residues of one loop and another were bigger than their corresponding one in the Mature simulation, it can be said that probably these loops were more mobile and flexible in the former than in the latter. In this framework, the interplay between the H- and L-chain loops resulted in a less stable interaction with the antigen, and thus resulting in a lower binding affinity for the antigen of the germline Ab than the mature one.

We were able to reconstruct the behaviour of all the important network of interactions influencing and/or influenced by the Ag-binding. The hypothesis arising from our analysis would be able to explain the possible reasons behind the well-known increased affinity of 8F9 for AD-2S1 when compared with M2J1. This led to the recognition of the possible underlying cause-and-effect relationship, thus shedding light on both some structural and dynamical features of these antibodies.

Even if the findings described in our paper are specific for the particular anti-HMCV Abs that we analyzed, we used a methodological approach that could be applied more generally to any germline/mature Abs in order to answer relevant questions about the maturation mechanisms.

In addition, it needs to be remarked that in the literature there is no other example of such long equilibrium simulations (500 ns each) describing the differences in dynamics of a germline Ab and its somatically mutated mature Ab both in complex with their viral Ag as detailed in this work.



Figure 1. Three-dimensional structure of the anti-HCMV antibody-antigen complexes. Mature antibody variable fragment (8F9, in shades of black cartoon) in complex with the antigen, AD-2S1 (in yellow licorice representation) in sodium chloride aqueous solution (in pale-yellow Na+, green Cl, and navy beads water); the CDRs of the H-chain are colored respectively in cyan, light-blue and blue, the CDRs of the L-chain are colored in orange, magenta and red.



Figure 2. RMSF for the simulations of the two Ab-Ag complexes: Germline (magenta) and Mature (blue). In the top-panel we show the data for the L-chain and in the bottom-panel those for the H-chain. On the x-axis the amino acid numbering following the Chothia scheme [9] with the CDR loops labeled and highlighted in yellow ellipsis in the plots.



Figure 3. Differential distance matrix obtained subtracting the Mature from the Germline matrix. Along the axis the antibody Heavy and Light chain, as well as the complementarity-determining region (CDR) loops and the antigen peptide, AD-2S1, are labeled. The CDR residues are identified by asterisks. The dashed-lines crossing the matrix define the important regions of the matrix where the residues belonging to the different CDR loops interact each other. Moreover, the dark-green ellipsis point out the intermolecular contacts between the CDR loops and the antigen having a significant difference in the two antibodies. A positive value (yellow to red tone scale) means that the average distance during the simulation between two residues in the Mature is smaller than in the Germline, and vice-versa in the presence of negative values (cyan to blue tone scale).



  1. MacLennan IC. Germinal centers. Annu Rev Immunol. 1994;12:117–139.
  2. Chan TD, Brin R. Affinity-based selection and the germinal center response. Immunol Rev. 2012;247:11–23.
  3. Boeckh M, Geballe AP. Cytomegalovirus: pathogen, paradigm, and puzzle. J Clin Invest. 2011;121:1673–80.
  4. La Rosa C, Diamond DJ. The immune response to human CMV. Future virol. 2012;7:279–293.
  5. Anderholm KM, Bierle CJ, Schleiss MR. Cytomegalovirus Vaccines: Current Status and Future Prospects. Drugs. 2016;76:1625–1645.
  6. Tuckerman ME. Statistical mechanics: Theory and molecular simulation. Oxford University Press 2010, Oxford (UK).
  7. Thomson CA, Bryson S, McLean GR, Creagh AL, Pai EF, Schrader JW. Germline V-genes sculpt the binding site of a family of antibodies neutralizing human cytomegalovirus. EMBO J. 2008;27:2592–602.
  8. McLean GR, Olsen OA, Watt IN, Rathanaswami P, Leslie KB, Babcook JS, Schrader JW. Recognition of human cytomegalovirus by human primary immunoglobulins identifies an innate foundation to an adaptive immune response. J Immunol. 2005;174:4768–4778.
  9. Al-Lazikani B, Lesk AM and Chothia C. Standard conformations for the canonical structures of immunoglobulins. J Mol Biol. 1997;273, 927–948.