CHD3 helicase domain mutations cause a neurodevelopmental syndrome with macrocephaly and impaired speech and language

Chromatin remodeling is of crucial importance during brain development. Pathogenic alterations of several chromatin remodeling ATPases have been implicated in neurodevelopmental disorders. We describe an index case with a de novo missense mutation in CHD3, identified during whole genome sequencing of a cohort of children with rare speech disorders. To gain a comprehensive view of features associated with disruption of this gene, we use a genotype-driven approach, collecting and characterizing 35 individuals with de novo CHD3 mutations and overlapping phenotypes. Most mutations cluster within the ATPase/helicase domain of the encoded protein. Modeling their impact on the three-dimensional structure demonstrates disturbance of critical binding and interaction motifs. Experimental assays with six of the identified mutations show that a subset directly affects ATPase activity, and all but one yield alterations in chromatin remodeling. We implicate de novo CHD3 mutations in a syndrome characterized by intellectual disability, macrocephaly, and impaired speech and language.


Supplementary Figure 2: RNA analysis of the c.5802_5803insGAAC mutation (p.(Phe1935Glufs*108))
To study the effects of the frameshift mutation in the penultimate exon of the CHD3 gene [c.5802_5803insGAAC (p.(Phe1935Glufs*108))] in mRNA from individual 35, lymphoblastoid cell lines were generated from peripheral blood cells by Epstein-Barr virus transformation following standard procedures. To check for the occurrence of nonsense mediated decay, mRNA was isolated from cells that were cultured in the presence (a) and absence (b) of cyclohexamide. A negative control was also included (c). CHD3 mRNA was analyzed by the synthesis of cDNA and Sanger sequencing according to standard protocols. Sanger analysis shows no difference between the RNA analysis from untreated cells and the cells that were treated with cyclohexamide. These data indicate that the alternative transcript that is a result of the frameshift mutation is not degraded by nonsense mediated decay. In conclusion, in this individual (individual 35) two transcripts are present: the wild-type transcript and the transcript with the frameshift, that leads to a stop codon after 108 amino acids.

Supplementary Figure 3: Subcellular localization of wildtype and mutant CHD3
Fluorescence microscopy images of HEK293 cells transfected with wild-type and synthetic CHD3 variants fused to mCherry (shown in red). Nuclei were stained with Hoechst 33342 (blue). The localization of CHD3 is nuclear in all pictures (wild-type and mutations), no difference in subcellular localization was seen between wild-type and mutants.  As no experimentally solved 3D-structure of CHD3 exists, we performed homology modeling using the modeling option with standard parameters in the YASARA 3 & WHAT IF 4 twinset. Several models of the ATPase/helicase domain were created. The best scoring model was based on template PDB-file 5JXR (M. thermophilia MtISWI, sequence identity 41% over the aligned residues), which shows a closed conformation of the ATPase/helicase domain. We also studied the model based on PDB-file 3MWY (yeast Chd1, sequence identity 45%) which shows a more open conformation but contains an ATP substitute. Both templates represent an auto-inhibited form, however, it is impossible to say which of these models best represents the real biological form of CHD3, since movement of the domains is probably important for correct functioning of the protein. Therefore, both models were studied.

Introduction of models and mutations
The complete CHD3 protein (2000 amino acids in isoform 1) is even bigger than the modeled domains shown here. Model 1 (closed conformation, based on 5JXR) contains residues 500-1290, while model 2 (open conformation, based on 3MWY) contains residues 445-1413.  In the open conformation, a wide gap exists between the Helicase ATP-binding domain (yellow) and Helicase C-terminal domain (green), while in the closed conformation no gap is seen. The existence of at least two different conformations indicates that it is necessary to study the effects of the mutations in these two different models.
It is unclear what triggers the conformational change. The authors of the original articles that belong to the templates 5,6 speculate that the Chromodomains are important for auto-inhibition and differentiation between naked DNA and DNA with nucleosomes. If that is true, both models here represent an autoinhibited state. However, one of them contains ATP and the other one shows more interaction within the ATP binding domains.
Also, it might very well be possible that the closed conformation contains ATP while the open conformation does not (the opposite of what is shown here). This is one of the limitations of modeling; substrates present or missing in the template will also be present or missing in the model.
The missense mutations were mapped on the models as is shown in figure 2. We performed a detailed analysis of the three-dimensional modeling of the mutations. As the CHD3 mutations are found largely clustered in the conserved motifs characteristic of the SF2 superfamily of helicases/translocases, we combined the 3D modeling analysis with information from the literature on these conserved motifs. A summary of this analysis is provided below.

His886Arg
Located on interaction surface of Helicase ATP-binding domain. Might be responsible for correct interactions to facilitate ATP-binding. Mutation into an Arginine, which is bigger and positively charged, might affect the function of the protein domain.
Residue also located on surface, but does not interact with other half of Helicase domain. However, still in ATPbinding domain, and responsible for stabilizing interactions, which might be lost due to the mutation.
His886 is part of the Walker B motif in motif II. This motif coordinates the catalytic Mg++ involved in ATP hydrolysis.

Leu915Phe
Leu915 residue is semiburied, and in close contact with His886 and Glu921. It seems to make hydrophobic interactions that stabilize ATP-binding domain. A bigger residue like Phe will not fit and destabilizes the protein.
Leu915 is close to Glu921 and Gly961 (in this model His886 is a bit further away). Leu915 makes still hydrophobic interactions, probably important for interaction surface with other helicase domain. A Phe residue will not fit here without causing reorganisation of surrounding residues.
Leu915 is located in conserved motif III.

Glu921Lys
This residue is surrounded by His886, Gly961 and Leu915. It can be found at the surface but makes interactions that will be lost when mutated into Lys with opposite charge.
This residue is semiburied and makes hydrogenbonds and saltbridges, thereby stabilizing the domain. Close to Leu915 and His886. Mutation into Lys will destabilize the area since Lys carries an opposite charge and has a different shape.
Glu921 is located just next to motif III.

Gly961Glu
This residue clusters with His886, Leu915 and Glu921; although these other three are closer together. Gly is small and flexible, and located close to a Proline at the end of a helix. Mutation into Glu will introduce a much bigger and less flexible residue, this will cause a structural change that might affect the interaction surface and protein function.
Gly961 is still close to Leu915. It is also more clearly located on the surface that interacts with the other half of the helicase domain. Mutation might affect the local structure.
Gly961 is part of conserved motif IV.

Conclusion
These four residues are the only four mutated residues in the Helicase ATP-binding domain. In model 1 it is clearly visible that these four residues cluster together and mutation of one of them could affect the position of the others. Clustering is not so clear in the closed model (model 2), but the residues still appear in the same area. The two Helicase lobes are closer together in model 1, and therefore a close interaction with other mutated residues becomes visible as well. For example, in this model Asp1120 and Arg1121 are close to Gly961 and Leu915. Figure 3 and 4 below illustrate this clustering.

Arg985Trp / Arg985Gln
Arg985 is located on the surface of the Helicase C-terminal domain. It is not closely located to the ATP-binding or interaction surface. It makes a saltbridge and could be responsible for overall stability of this part of the protein. It clusters with Arg1187. Mutation of Arg into Trp might cause folding problems. Mutation into Glu would be easier, but stabilizing saltbridges would still be lost.
In this model the last C-terminal tail has a completely different conformation, which might be caused by the crystallization process. Originally, this model is a dimer with the last C-terminal tail swapped, and as a result Leu1236 is not close to Arg985 and Arg1187. Also, this model does not contain the C-terminal bridge (see Leu1326Pro), and Arg only seems to add some stable interactions to the area, contributing to general stable protein folding. Mutation into either Gln or Trp will affect this folding.
This mutation is located outside the canonical helicase motifs.

Arg1187Pro
This residue clusters with Arg985 on the surface of the protein's C-terminal helicase domain. It might mediate interaction with a regulatory unit. The residue makes a saltbridge, which will be lost due to the mutation. Mutation into Pro will change the backbone conformation. Regardless of interactions with a regulatory unit, this mutation will change local protein structure and might affect stability and function.
This residue clusters with Arg985 on the surface of the protein's C-terminal helicase domain. Mutation into Pro will change the backbone conformation, and will change local protein structure and might affect stability and function.
Arg1187 is part of motif VI, a motif that contains multiple mutations, and contributes to ATP binding and hydrolysis.

Leu1236Pro
In the ATP bound model, the residue is located closely to Arg985 and Arg1187 in a C-terminal bridge that is suggested to be important for regulation 5 .
Leu1236 is found in a helix that is used for domain swapping in the crystal. The biological function of this helix is unclear. However, it is known that any mutation into Pro will affect the local backbone structure, because Pro is the only amino acid that will force the backbone into a rigid turn. Regardless of the exact position of the residue, this mutation can affect the protein's structure.
This mutation is located outside the canonical helicase motifs.

Conclusion
Residues Arg985, Arg1187 and maybe Leu1236 are found close together. It is unclear what the function of the last C-terminal tail is, although it has been suggested to have a regulatory effect 5 . In that case, interactions with the tail are important. If this is not the case, both mutations Arg1187Pro and Arg985Trp/Gln are expected to affect the structure and thereby maybe affect the function as well.

Asp1120His
Asp1120 is part of a helix in the Cterminal helicase domain, and clusters with Arg1121. The residues are not especially close to the ATP, although Asp1120 seems to be located on the possible interaction surface. Asp is negatively charged, which is needed to make correct interactions, mutation into His will affect these interactions. Mutation might affect interaction between domains and/or ATP binding.
In this model Asp1120 gets close to the other half of the helicase domain (Helicase ATP-binding domain), for example close to mutated residue Gly961. Asp is negatively charged, which is needed to make correct interactions, mutation into His will affect these interactions. Mutation might affect interaction between domains and/or ATP binding.
Asp1120 is part of helix integral to conserved motif V. Mutations in Motif V in the context of yeast SNF2 abrogate ATP hydrolysis and remodeling activity.

Arg1121Pro
This residue is located in the same helix in the C-terminal helicase domain as Asp1120, although its sidechain points in a different direction. The Arg sidechain is positively charged and makes hydrogen bonds and saltbridges. The Arg 1121 residue is probably not involved in ATP binding or interaction with the other domains, but the interactions it makes might be important for a stable structure. Mutation into Pro will surely affect the structure because interactions will be lost, and Pro will change the backbone conformation.
This residue is located in the same helix in the C-terminal helicase domain as Asp1120, although its sidechain points in a different direction. The Arg sidechain is positively charged and makes hydrogen bonds and saltbridges. The Arg 1121 residue is probably not involved in ATP binding or interaction with the other domains, but the interactions it makes might be important for a stable structure. Mutation into Pro will surely affect the structure because interactions will be lost, and Pro will change the backbone conformation.
Arg1121 is part of helix integral to conserved motif V. Mutations in Motif V in the context of yeast SNF2 abrogate ATP hydrolysis and remodeling activity.

Thr1136Ile
This residue is located in the core of the protein, making a hydrogenbond with its -OH group and hydrophobic interactions with the methyl in its sidechain. The tight packing does not allow a bigger residue here. Also, the mutation will cause loss of the hydrogenbond and thereby destabilize the local structure. The surrounding residues seem important for correct shape of the interaction site.
This residue is located in the core of the protein, making a hydrogenbond with its -OH group and hydrophobic interactions with the methyl in its sidechain. The tight packing does not allow a bigger residue here. Also, the mutation will cause loss of the hydrogenbond and thereby destabilize the local structure. The surrounding residues seem important for correct shape of the interaction site.
This amino acid is located within Motif V.
Mutations in Motif V in the context of yeast SNF2 abrogate ATP hydrolysis and remodeling activity.

Trp1158Arg
This residue is clearly important for the hydrophobic core of the protein. In both models it is (semi) buried and makes hydrophobic interactions. Mutation into anything else will affect the stability and the structure of this domain.
This residue is clearly important for the hydrophobic core of the protein. In both models it is (semi) buried and makes hydrophobic interactions. Mutation into anything else will affect the stability and the structure of this domain.
This Trp residue has recently been shown to be of critical importance in remodeling, by binding nucleosomal DNA in the minor groove. Mutation of this residue impacts remodeling, not ATP hydrolysis 9,10 .

Asn1159Lys
This residue is located in the same helix as some of the following mutations (see below) and seems to form the interaction surface with ATP and the other domain. It makes a few hydrogenbonds. Mutation into something larger and positively charged might affect interactions.
This residue is located in the same helix as some of the following mutations(see below). This residue is buried and makes many interactions, some of the interactions are made with residues in a N-terminal helix. This helix might be important for autoinhibition 6 . Mutation into something larger and positively charged might affect interactions with surrounding residues or with the inhibition helix.
This amino acid is located adjacent to the Trp residue at position 1158, and might alter the environment of this critical Trp residue.

His1161Arg
This residue follows a similar story to the mutations above. It is located in the same helix as Asn1159 and seems to be important for the surface interactions. Mutation into Arg will change the amino acid properties drastically.
This residue follows a similar story to the mutations above. It is located in the same helix as Asn1159 and seems to be important for the surface interactions. In this model the residue makes interactions with the putative inhibition helix. Mutation into Arg will change the amino acid properties drastically.
This amino acid is located close to the Trp residue at position 1158, and might alter the environment of this critical Trp residue.

Conclusion
Based on recent literature, it is known that the Trp residue at position 1158 is of critical importance in remodeling. The Trp binds nucleosomal DNA in the minor groove, and mutation of this residue will impact chromatin remodeling. The other two residues (1159 and 1161) are probably altering the environment of this critical Trp residue. The W1158 is buried and makes many hydrophobic interactions. N1159 and H1161 become buried in the closed model and seem to interact with residues in a putative inhibition helix (green) 6 . Recent articles show that W1158 is of critical importance in binding nucleosomal DNA 9,10 .