Structure of METTL3-METTL14 with an m6A nucleotide reveals insights into m6A conversion and sensing

Reviewer #1 (Public review):

Summary:

In this manuscript submitted by Qi et al., the authors study the RNA methylation mechanism by the METTL3-METTL14 complex. This complex catalyzes the major epitranscriptome methylation mark of nuclear RNA, including mRNA and lncRNAs. They catalyze the transfer of methyl group from SAM to convert the N6 of adenosine in RNA to m6A. Mutations in this complex have been associated with several diseases, such as type 2 diabetes and several types of cancer. The primary focus of this study was to understand the post-catalytic state of the METTL3-14 bound to a structural mimic of a reaction product known as N6-methyladenosine monophosphate (m6A) using X-ray crystallography. The authors show that the m6A occupies a novel pocket at the interface of the METTL3-14 complex and identified that residues interacting with m6A are mutated in several cancers. Furthermore, the authors demonstrate that the mutations lead to a significant loss in catalytic activity, alter RNA binding, and hinder the proper positioning of the substrate adenine in the active site. Lastly, the authors perform supervised molecular dynamics simulations to understand the effect of the mutations on the interaction network with m6A. The evidence for this study is good, with the combination of X-ray, functional assays, and molecular dynamics justifying their overall conclusions. This structure is significant as it provides new insights into the structural determinants of known cancer-associated mutations of this important class of enzymes. However, some issues need to be addressed.

Strengths:

(1) The X-ray structure is well determined, and the density map has the quality to observe all the interactions of the METTL3-14 complex with m6A.

(2) The structure reveals a novel 'cryptic pocket' in the complex that is 16 Å away from the SAM binding site. It is a functional m6A-sensor, illustrating a mechanism where the complex switches its functionality from an m6A writer to a reader.

(3) The structure illustrates that the residues forming cryptic pockets are found in multiple Cancer-associated mutations and are well conserved across several organisms.

(4) The functional assays (methyl transferase, RNA binding, kinetic, and SPR assays) provide a complete picture of the effect of the mutations on the activity of the METTL3-14 complex.

(5) Molecular dynamics simulations were done to understand the impact of the mutations on the pocket structure and its dynamics and support the X-ray structure findings.

Weaknesses:

(1) Although the X-ray structure is well determined, the statistics are a bit troubling, particularly the Ramachandran, Sidechain and RSRZ outliers. It is well above the average for structures at that resolution. Maybe the use of alternative software such as ISOLDE may be adequate to improve those parameters.

(2) The authors should expand their discussion as to why the affinity for the product is higher than the substrate and the implications on the mechanism.

(3) The SPR profiles of the association kinetics look to have several minor association-dissociation events occurring. Multiple binding sites? Authors should provide an explanation for such behavior. Also, what is the structural explanation of the difference in binding modes between the wt vs. mutant (one vs. two-state binding modes)?

(4) In materials and methods, it shows the data in Figure 2a was fitted to a Michaelis-Menten equation, however, the Y axis shows Normalized methylation and not initial rates. The authors should elaborate on their approach. In addition, more than three initial velocity rate points per protein are needed to fit a Michaelis-Menten curve confidently. Additionally, where can the Michaelis-Menten parameters be found?

Reviewer #2 (Public review):

Summary:

Qi et al. determined the X-ray crystallographic structure of the methyltransferase core of the obligate heterodimeric complex METTL3-METTL14 in complex with methyladenosine monophosphate (m6A), a product mimic for the methylation of adenosine, to a resolution of 2.5 Å. Their structure appears to reveal a cryptic binding pocket for m6A that had not previously been identified. Using full-length protein produced in insect cells, Qi et al. determined the methyltransferase activity of wildtype METTL3-METTL14 and compared it to that of mutant forms of the protein that have been implicated in cancer. In addition to methyltransferase activity, the authors used both fluorescence polarization assays and surface plasmon resonance to investigate the affinities and kinetics of RNA binding to wildtype and mutant forms of the full-length complex. The results indicate that mutations in the methyltransferase core of two separate arginine residues alter the dynamics of RNA binding and enzyme specificity of METTL3-METTL14. The authors go on to use a combination of supervised molecular dynamics simulations and comparisons to recently published structures to propose a "swivelling" mechanism for the transfer of the methylated substrate from the catalytic site of the complex to the novel cryptic pocket.

Strengths:

I appreciated the inclusion of supplementary data showing the purity and monodispersity of the protein used for crystallization as well as the omit map and other electron density maps to support the placement of the product mimic in the cryptic site. The authors use a combination of complementary biophysical techniques to test the effects of mutations that were identified in the literature as being clinically important and to develop a hypothesis for the large-scale translocation required for the enzymatic product to move from the catalytic site to the cryptic pocket. The use of molecular dynamics simulations to attempt to indirectly visualize how this translocation might occur in vivo was well done.

Weaknesses:

Even taking into account the 2.5 Å resolution of the structure, the model is not refined to the point that it could be. Some waters seem to be built into blobs of density that aren't particularly convincing, and other seemingly obvious waters aren't built at all. The structure validation report supports this and shows that overall, and in the context of 2.5 Å resolution, this is not a great model. A good many parts of the structural analysis don't seem consistent with what I see when I look at the model and density in terms of proposed interactions in the cryptic pocket. Much of the language used in the manuscript is too strong when the model is quite speculative.