PREreview of Dynamics and Activation of Membrane-Bound B Cell Receptor Assembly

Peer Review: Dynamics and Activation of Membrane-Bound B cell Receptor Assembly

Kevin Alexander Estrada Alamo, University of California San Francisco

Summary

In this study, the authors performed MD simulations of the IgM B cell Receptor (BCR) in the presence or absence of an antigen. Additionally, the simulations were performed in an all-atom (AA) and course-grain (CG) representation. The authors’ initial model is based on the cryo-EM structure of the human IgM BCR (PDB:7XQ8), where the fragment antigen binding domains (Fabs) of the receptor were replaced with Fabs that target HIV-1 Env protein (VRC-CH31), and the model was generated using SWISS-MODELLER. The BCRs were simulated in a membrane mimicking the lipid composition of a B cell membrane. The MD simulations suggest: 1) regions of increased/decreased flexibility within the BCR upon antigen binding and 2) rearrangements of the receptor’s transmembrane domains (TMs) and lipid composition upon antigen binding.

A significant success of the paper is identifying some mode of allostery/conformational change in the BCR upon antigen recognition and simulating the change in lipid composition (suggested in previous studies) from disordered to ordered lipids upon antigen binding. A significant weakness of the paper is that the simulations were not done with multiple BCRs (at least two). Thus, the authors cannot make claims that support or oppose any BCR activation models. Currently, it is disputed which BCR activation model is the best representative, whether it is oligomerization, dissociation, conformational changes, or a mixture of these models. MD simulations of multiple BCR and antigen molecules have the potential to provide insightful hypotheses for later experimental validation. 

Major points

The representative antigen-bound BCR model needs a stronger caveat attached to it.

• Looking at Figure 1, both Fabs appear to bind to a single monomer of the antigen. Additionally, in Figure 2, the ΔRMSF shows that the Fabs' variable regions are equally rigid. This suggests that both Fabs are bound to the same antigen. In the main text, the authors say, “HIV-1 Env primarily interacted with the CDR3 regions of one of the Fab domains while it had minor interactions with the other Fab domain as well”. I find this explanation insufficient, as the model generated goes against what we would expect. We expect one Fab to bind to one antigen and one antibody to bind to two antigens. However, the authors depict two Fabs binding to one antigen, which suggests that the model generated is not physiologically relevant. 

• Visually comparing the crystal structure of VRC-CH31 Fab bound to HIV-1 Env (PDB:6NNJ) to the model proposed, it is clear that the authors’ model disagrees with the crystal structure. The authors need to provide alignments of the authors’ model of antigen-bound BCR to 6NNJ along with RMSD values and explain why their model differs from the solved structure.

• Since the mode of binding displayed in the authors’ model differs from the experimentally characterized one, the authors should validate their model using other docking methods or AlphaFold2/3  multimer and report confidence metrics for the model.

Minor points

Figure 1

• In Figure 1, the authors show the models of the BCR not bound and bound to HIV-1 Env. In the current figure, it is hard to tell the difference between the two models, and showing an alignment of the BCR and the antigen-bound BCR will clearly illustrate any conformational changes that occurred. 

Figure 2

• In Figure 2, the authors show a close-up of the antigen-bound BCR model, where the BCR is colored based on the ΔRMSF. In sub-figure b, the authors label the Fabs as “Fab1” and “Fab2*”, and I think the figure caption needs to be updated to address what the authors are trying to convey with the asterisk. In the main text, the authors explain that “HIV-1 Env primarily interacted with the CDR3 regions of one of the Fab domains while it had minor interactions with the other Fab domain as well”, but it does not state which one is Fab1 and Fab2, so something like “interacting mostly with the second IgM subunit (containing Fab2)” as done in the caption of Figure S1, would be helpful. 

Figure 3

• In Figure 3, the authors show a diagram of how BCMM calculates the tilt angle of the TMs and a table of those tilt angles of the two BCR models. Without images, this figure is abstract. The authors can improve the legibility of the figure by adding alignments of the two models focusing on the TMs and showing those tilt angle changes directly.

Methods

• Under the “Simulation system setups” section, the authors mention the reason for choosing the SWISS-MODELLER model of the BCR over the AlphaFold models. The authors state that “SWISS-MODELLER did not model the intracellular loops of the Igα/β heterodimer” and “observed that AlphaFold predicted the Igα/β heterodimer intracellular loop to ‘U-turn’ into the membrane”, and because of AlphaFold trying to pack the unstructured intracellular tails up against the rest of the structure and that was their reason for choosing the SWISS-MODELLER model. Instead, the authors could truncate the intracellular tails from the AlphaFold models before running the MD simulations. To justify using SWISS-MODELLER, the authors must provide a stronger rationale for the choice through RMSDs to 7XQ8 or some other metric. Even providing some evidence that the SWISS model is comparable to those generated using AlphaFold would be sufficient and provide a stronger rationale. 

Main text

• In the introduction, the authors wrote, “They are a complex of IgM containing dimer of antigen-binding the Fab domains, Fc domains, transmembrane helices (TM), and the Igα/β heterodimer signaling component”.

  • If a word count limits the authors, this can be made more concise. The “IgM containing dimer of antigen-binding the Fab domains, Fc domains, transmembrane helices (TM)” part can be shortened to membrane-bound immunoglobulin (mIg) or membrane-bound IgM antibody.

  • The authors should be careful with the wording here because this sentence makes it seem that there is only one isotype of BCR (the IgM isotype) instead of five.

• As a result of the simulations containing one BCR molecule, the language used in the results and discussion sections of the main text is too strong to fit what the results suggest. For the following sentences, the authors can tone down the language, stating that the authors’ data “support” any of the BCR activation models. Instead, I would focus on how the data supports the apparent conformational change in the IgM BCR upon antigen binding, which is an important finding. 

  • “Seemingly, antigen binding to the BCR at the extracellular Fab domains … supporting the conformation-induced oligomerization model”

  • “Therefore, antigen binding at the extracellular Fab domains … and dissociation activation models of BCR activation” 

  • “This result supported the dissociation activation model … their intracellular ITAM phosphorylation motifs” 

  • “Findings from our work support either … opposing the classical cross-linking model of BCR activation”

  • Broadly, the whole discussion section

Competing interests

The author declares that they have no competing interests.