Valdivia et al. in their article titled “Myelin Basic Protein Phosoholipid Complexation Likely Competes with Deimination in Experimental Autoimmune Encephalomyelitis Mouse Model” explore the specific lipid components that undergo complexation with myelin basic protein (MBP), a major myelin protein of the central nervous system with relevance in multiple sclerosis (MS). The authors identify four specific lipid species that bind MBP, one of which is 18:1 lysophosphatidylcholine (LPC). Increased levels of citrullination, a known modification of MBP catalyzed by peptidyl-arginyl deiminases (PADs), has been linked to MS. The authors perform a multidisciplinary study where they find that LPC can block MBP citrullination and thus maintain MBP in a functional form. The authors propose a model where MBP-LPC complexation drives healthy myelin formation and prevents MS pathology.

The current study by Valdivia and colleagues is of general interest and provides an intriguing, novel biological mechanism that might have relevance in scavenging MBP and regulating its activity in achieving compact myelin maturation. After studying the article, a few matters were discovered that I would kindly ask the authors to clarify or revise (major points, see below), along with smaller changes that should be considered to improve the text as such (minor points).

Major points

1. The effect of MBP citrullination- and lipid complexation-induced folding is based on circular dichroism (CD) measurements in section 2.5. (data shown in figure 4C). However, these results appear to have been at least partially misinterpreted, and as such will affect the conclusions presented in the article. In the text, the authors claim that LPC does not induce any change in secondary structure, although it is very clear that in the presence of LPC the spectral minimum at ~200 nm shifts to ~205 nm, and a distinct maximum at 190 nm emerges. This is a sign of major secondary structure formation. The same effect occurs in the presence of PAD and LPC. In addition, the authors claim that in the presence of PAD, MBP folds (so presumably when it becomes citrullinated). However, this does not seem to be the case based on the CD spectrum, where the observed minimum remains at 200 nm. The intensity of the signal decreases, which suggests that there is either less protein in the sample or some of the protein aggregates/precipitates and thus does not contribute to the CD signal. These would be easy to verify from the HT voltage and absorbance curves. The authors also present deconvolution results in figure 4D, but the deconvolutions appear to be very similar for each sample, despite large differences in the spectra themselves – how were the deconvolutions performed? Finally, the Experimental Section does not describe whether vesicles or other aqueous lipid preparations were used for sample preparation – only that proteins were mixed with lipid.

2. The MD simulations that the authors undertook reveal a flexible protein, yet compacted, in the absence of any lipids. While no regular secondary structure elements are present, earlier data, based on small-angle X-ray and neutron scattering studies, suggests that MBP is highly extended and chain-like, with radii of gyrations (Rg) ranging between 3-4 nm and maximum dimensions (Dmax) between 8-18 nm (Haas et al. (2004) Biophys J. 86(1): 455–460; Wang et al. (2011) PLoS One. 6(5):e19915; Stadler et al. (2014) J Am Chem Soc. 136(19): 6987–6994; Raasakka et al. (2017) Sci Rep. 7: 4974). This is a substantial difference to the authors’ simulations and should be explained.

3. A close examination of the Experimental Section revealed that either phosphate buffer or phosphate-buffered saline were used in the experiments involving MBP and phospholipid complexation. Moreover, when PAD2 was used to deiminate MBP, CaCl2 was included. Potential issues arise in using both phosphate and Ca2+ in the described experiments. As MBP is known to bind membranes mostly through phospholipid-mediated electrostatics, inclusion of phosphate buffers can substantially interfere with MBP-lipid binding. The presence of Ca2+ has also been shown to alter MBP-lipid interactions (Raasakka et al. (2019) Biochem Biophys Res Commun. 511(1): 7–12.), and in vitro, Ca2+ concentrations in the millimolar range can result in Ca2+-phospholipid complexation and agglomeration, which decreases the amount of either ionic species in solution that MBP can interact with. Therefore, the authors are encouraged to perform control measurements for each relevant experiment in the absence of phosphate and Ca2+.

4. Finally, a more general point arises on which the authors could potentially expand their discussion. As far as current understanding goes, the deimination of MBP has to occur before it binds membranes and forms compact myelin, as PADs are unable to fit in between the tightly compacted cytoplasmic membranes of the major dense line. This means, that MBP deimination must occur very early during myelination (during post-natal development and early childhood) and also very rapidly after MBP translation (MBP is locally translated where it is needed), as most of compact myelin remains relatively stable after MBP has bridged membranes together. This is supported by data that shows that MBP binds irreversibly to membranes in in vitro systems (see e.g. Raasakka et al. (2017) Sci Rep. 7: 4974). The question that arises is: how does the authors’ deimination hypothesis translate to multiple sclerosis that develops after most of myelin has matured, for example during adulthood? Most MS cases arise between the ages of 20-40, and as such this question specifically involves most new annual MS cases.

Minor points

1. The authors mention in the introduction that studies mapping the individual lipid components of myelin that interact with MBP have not been performed prior to the current study. However, several rather recent reports exist where efforts were made to specifically determine which lipids mediate the interaction between myelin membranes and MBP, and even reports that discuss how changes in lipid composition can induce MBP-related MS-pathogenicity. I encourage the authors to study these reports and refer to them and their conclusions briefly in the introduction and discussion. These studies include the following: Shaharabani et al. (2016) J Am Chem Soc. 138 (37): 12159–12165.; Raasakka et al. (2017) Sci Rep. 7: 4974.; Widder et al. (2018) Langmuir 34(21): 6095–6108.; Widder et al. (2020) BBA Biomembr 1862(2): 183077.; Träger et al. (2020) Cells 9(3): 529.

2. In Figure 2C, it appears the amount of PE is clearly increased in EAE animals vs. controls, although in section 2.3. text the authors claim that PE-levels remain unchanged. This might be due to their statistical test, but a change in PE-content has also been reported earlier in EAE animals (Min et al. (2009) Proc Natl Acad Sci USA 106(9): 3154–3159). How was the statistical test performed and does the authors’ MS approach only regard PE in general or does it distinguish between PE with esteric fatty acid tails and PE plasmalogens? Plasmalogens are the major form of PE in normal myelin membranes (Farooqui & Horrocks (2001) Neuroscientist 7: 232–245). The (potential) change in PE content should be carefully considered, especially since MBP-membrane interactions are influenced by PE (Widder et al. (2018) Langmuir 34(21): 6095–6108). Perhaps a shift in ester-ether linkage ratio might be present between native and EAE myelin PE?

3. Figure 7 is an excellent summary of the current study and how it ties to myelin compaction and maintenance in general, but even despite the authors’ comments on uncertainty, the figure in its current form is misleading: MBP is drawn as a C-shaped molecule in all panels, despite lipid membranes being absent. It is well established that MBP adopts a C-shaped conformation only when it binds to lipids, whereas in solution, even in as a highly deiminated form (Wang et al. (2011) PLoS One. 6(5):e19915), it remains intrinsically disordered and highly extended. I kindly ask the authors to clarify the figure by drawing any membrane elements where necessary in all panels. Otherwise, MBP should be drawn as an extended chain when not bound to lipids. The same changes should also be applied to the graphical abstract.