Monoallelic CRMP1 gene variants cause neurodevelopmental disorder
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This manuscript describes de novo dominant toxic mutations in CRMP1 in 3 probands with a shared neurodevelopmental phenotype. The authors show that the mutations lead to reduced protein production from recombinant expression and that the mutations correlate with shorter neurites in cultured cells. This is the first report of mutations in CRMP1 in humans, encoding a cytoskeletal regulator protein. The results could have implications for physicians, geneticists, neurodevelopmental scientists, and cell biologists.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Neuroscience (eLife)
Abstract
Collapsin response mediator proteins (CRMPs) are key for brain development and function. Here, we link CRMP1 to a neurodevelopmental disorder. We report heterozygous de novo variants in the CRMP1 gene in three unrelated individuals with muscular hypotonia, intellectual disability, and/or autism spectrum disorder. Based on in silico analysis these variants are predicted to affect the CRMP1 structure. We further analyzed the effect of the variants on the protein structure/levels and cellular processes. We showed that the human CRMP1 variants impact the oligomerization of CRMP1 proteins. Moreover, overexpression of the CRMP1 variants affect neurite outgrowth of murine cortical neurons. While altered CRMP1 levels have been reported in psychiatric diseases, genetic variants in CRMP1 gene have never been linked to human disease. We report for the first-time variants in the CRMP1 gene and emphasize its key role in brain development and function by linking directly to a human neurodevelopmental disease.
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
- In family 2, the variant was detected by routine trio-based WES diagnostics. Sanger confirmation was not performed. IGV images can be added as supplementary material. Furthermore, median coverage was 75× which might not be sufficient for the identification of all heterozygous variants.
We thank reviewer for pointing it out for clarification. Obviously, at the time (2016) of the reporting of this variant this was our laboratory’s thoroughly validated protocol, which shows that median (!) coverage of 75x with the technology at the time is more than sufficient for robust variant calling. This particular variant was actually below 75X in coverage (at 65x), but Sanger confirmation was not necessary (based on thorough validation of the robustness of calling and GATK scores and other quality …
Author Response
Reviewer #1 (Public Review):
- In family 2, the variant was detected by routine trio-based WES diagnostics. Sanger confirmation was not performed. IGV images can be added as supplementary material. Furthermore, median coverage was 75× which might not be sufficient for the identification of all heterozygous variants.
We thank reviewer for pointing it out for clarification. Obviously, at the time (2016) of the reporting of this variant this was our laboratory’s thoroughly validated protocol, which shows that median (!) coverage of 75x with the technology at the time is more than sufficient for robust variant calling. This particular variant was actually below 75X in coverage (at 65x), but Sanger confirmation was not necessary (based on thorough validation of the robustness of calling and GATK scores and other quality parameters for de novo calling). In addition, when coverage goes below 30-35X Sanger confirmation is warranted.
- Proband 2 (P2) was born as the second child of non-consanguineous parents of Caucasian descent after an uneventful pregnancy and delivery. The boy was macrosomic at birth. Since there was macrosomia, how would the pregnancy be uneventful? At the last assessment at 10 years of age, obesity associated with hyperphagia was of concern; the weight of the patient should be clarified. P2 was diagnosed with autism spectrum disorder but a normal cognitive profile. The identified NM_001014809.2(CRMP1_v001):c.1280C>T variant is very rare and reported in GnomAD exomes with allele frequency 0.0000041.
Routine echographia during pregnancy did not result in any concerns. The pregnancy was indeed uneventful. BMI at last evaluation was 26.1. We included the details in the revised manuscript.
- Proband 3 (P3) is the first of three children of a non-consanguineous family of European descent. There is a familial history of obesity on both parental sides, and the father is macrocephalic (head circumference: 60.5 cm). Macrocephaly can be isolated and benign, such as in benign familial macrocephaly. However, P3 presented with moderate intellectual disability and an autism spectrum disorder. Since P3 has a macrocephaly also, the PTEN gene should be further interrogated by detailed WGS data analysis as well as an additional orthogonal method(s) since it has pseudogenes.
We have not noted any pathogenic variant of the PTEN gene in the genetic analysis.
Reviewer #2 (Public Review):
Weaknesses of the article include:
- Spelling errors and difficult-to-understand language. The use of "variant" is now preferred over mutation. According to current nomenclature, predicted but not experimentally confirmed protein alterations should be written as p.(Phe351Ser) rather than p.Phe351Ser.
We apologise for the spelling errors and the difficult-to-understand language in the manuscript. We considered the reviewers comments seriously and corrected the errors and rephrased the sentences wherever necessary.
- Inconsistent use of in silico pathogenicity predictors and conservation metrics. These should be standardized for each case and should include at least phylop, CADD, and REVEL.
We have applied consistency in the description of in silico pathogenicity predictors and conservation metrics for each patient.
- CRMP1 is under significant constraint against loss-of-function variation in gnomAD - pLI = 0.99, LOEUF 0.28. Genes in the top decile are highly enriched for haploinsufficiency as a disease mechanism. This should be considered in the interpretation of this data and incorporated into the manuscript.
We thank the reviewer for the comment. As per reviewer’s suggestion, we have included a statement in the revised manuscript under ‘Subjects and Methods’ section.
- I am not convinced the data supports a dominant-negative interpretation. The variants do not oligomerize as well as wild-type CRMP1, and when co-expressed with wild-type CRMP1 there is an increase in monomeric wild-type CRMP1. While this could support a dominant-negative interpretation, an alternative explanation is these are loss-of-function alleles that cannot oligomerize, and at the stoichiometry of this artificial overexpression system, this leads to increased monomeric wild-type CRMP1. The axonal outgrowth studies are more compelling, but without a loss-of-function control allele, it is difficult to interpret.
The experiments in Figure 2 should be replicated, quantitated, and their statistical significance confirmed.
We thank reviewer for raising concern about the experiment and interpretation of the data. We performed size exclusion chromatography experiments and included the data in the revised Figure 2. Unfortunately, we could not reproduce the experiments for Figure 2B. From our current experimental results, we prove that the CRMP1 variants affect the homo-oligomerization process.
Reviewer #3 (Public Review):
- The major weakness is Figure 2, as it is not performed up to high standards like the rest of the paper. Panel A does not show any loading control and does not confirm. Panel B at 720 kDa band is not convincing. Results should be repeated with size exclusion chromatography and/or another method to determine molecular weight and should be quantified from triplicate experiments. Panel C is also not convincing and should be repeated to more carefully show results, and quantified.
We thank reviewer for this important concern raised on our Figure 2 experimental data. We addressed the comments in the revised manuscript. We performed size exclusion chromatography and presented the results in the revised manuscript and discussed accordingly in page 23-24.
Fig. 2A: This panel shows the recombinant CRMP1 wildtype and the variants from E-coli expressing system. We repeated the expression several times and obtained similar partially cleaved proteins. Fig. 2A is Coomassie Brilliant Blue staining. Protein size marker and loading control (BSA) were applied on the same gel as shown in Fig.2A original.
Fig.2B: Due to limited protein expression of T313M and P475L mutants, we could not repeat the gel-filtration experiments.
Fig. 2C, 2D: It is difficult to adjust the expression level of each construct (CRMP1 wildtype, T313M, or P475L) in HEK293T cells (input). Therefore, we measured the signal intensity of myc-IP band and input ratio of V5 blot in each condition. Fig. 2D shows the ratio from four independent experiments.
-
Evaluation Summary:
This manuscript describes de novo dominant toxic mutations in CRMP1 in 3 probands with a shared neurodevelopmental phenotype. The authors show that the mutations lead to reduced protein production from recombinant expression and that the mutations correlate with shorter neurites in cultured cells. This is the first report of mutations in CRMP1 in humans, encoding a cytoskeletal regulator protein. The results could have implications for physicians, geneticists, neurodevelopmental scientists, and cell biologists.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
In this study, Ravindran et al. describe heterozygous de novo variants in the CRMP1 gene in three unrelated kindreds with muscular hypotonia, autism spectrum disorder, and/or intellectual disability. Based on in silico analysis these variants are predicted to affect the CRMP1 structure. The effect of these variants on the protein structure/levels and cellular processes was analyzed. The authors show that the identified CRMP1 variants are dominant-negative and impact the oligomerization of CRMP1 proteins. Moreover, overexpression of mutant-CRMP1 variants affects neurite outgrowth of murine cortical neurons. It has been known that maternal autoantibody reactivity to CRMP1 significantly increases the odds of a child having a higher Autism Diagnostic Observation Schedule (ADOS) severity score (PMID: 33483694), …
Reviewer #1 (Public Review):
In this study, Ravindran et al. describe heterozygous de novo variants in the CRMP1 gene in three unrelated kindreds with muscular hypotonia, autism spectrum disorder, and/or intellectual disability. Based on in silico analysis these variants are predicted to affect the CRMP1 structure. The effect of these variants on the protein structure/levels and cellular processes was analyzed. The authors show that the identified CRMP1 variants are dominant-negative and impact the oligomerization of CRMP1 proteins. Moreover, overexpression of mutant-CRMP1 variants affects neurite outgrowth of murine cortical neurons. It has been known that maternal autoantibody reactivity to CRMP1 significantly increases the odds of a child having a higher Autism Diagnostic Observation Schedule (ADOS) severity score (PMID: 33483694), and increased CRMP1 mRNA levels were identified in individuals with schizophrenia and autism spectrum disorder (PMID: 22798627). Mice lacking Crmp1 expression manifest hyperactivity, impaired learning, memory, and prepulse inhibition (PMID: 24409129). Previous findings strongly support the involvement of CRMPs in neurodevelopmental disorders. It is known that the mammalian CRMP family consists of five cytosolic family members (CRMP1-5) and are highly expressed in the developing and adult nervous system. Monoallelic CRMP5 variants can cause Ritscher-Schinzel syndrome 4 (MIM#619435).
Regarding studies:
In family 1, whole-exome sequencing (WES) was performed on a HiSeq XTen Deep Sequencer (Illumina, CA, USA), with an average coverage of ~36X, which is lower than expected. CRMP1 variant segregation was confirmed by Sanger sequencing.
In family 2, the variant was detected by routine trio-based WES diagnostics. Sanger confirmation was not performed. IGV images can be added as supplementary material. Furthermore, median coverage was 75× which might not be sufficient for the identification of all heterozygous variants.
In family 3, trio-based whole-genome sequencing was performed. Variants >4kb were called using CNVnator (v0.4.1) and annotated with AnnotSV (v2.5.1). An average depth-of-coverage of >50x was obtained. Sanger sequencing was performed to confirm the identified mutation in the CRMP1 gene.Regarding Results:
Proband 1 (P1) was born as the second child of non-consanguineous healthy parents of Caucasian descent after an uneventful pregnancy. At delivery, a singular umbilical artery was noted. P1 has a moderate intellectual disability and behavioral abnormalities. Chromosome analysis and array-CGH were normal in the index patient (P1). The identified NM_001014809.2(CRMP1_v001):c.1766C>T variant has not been reported in publicly available databases.
Proband 2 (P2) was born as the second child of non-consanguineous parents of Caucasian descent after an uneventful pregnancy and delivery. The boy was macrosomic at birth. Since there was macrosomia, how would the pregnancy be uneventful? At the last assessment at 10 years of age, obesity associated with hyperphagia was of concern; the weight of the patient should be clarified. P2 was diagnosed with autism spectrum disorder but a normal cognitive profile. The identified NM_001014809.2(CRMP1_v001):c.1280C>T variant is very rare and reported in GnomAD exomes with allele frequency 0.0000041.
Proband 3 (P3) is the first of three children of a non-consanguineous family of European descent. There is a familial history of obesity on both parental sides, and the father is macrocephalic (head circumference: 60.5 cm). Macrocephaly can be isolated and benign, such as in benign familial macrocephaly. However, P3 presented with moderate intellectual disability and an autism spectrum disorder. Since P3 has a macrocephaly also, the PTEN gene should be further interrogated by detailed WGS data analysis as well as an additional orthogonal method(s) since it has pseudogenes.
Array analysis revealed two maternally inherited deletions: a 668 kb deletion at 3q26.31 and a 371kb at 5q23.1, confirmed by genome sequencing and considered a variant of unknown significance. The identified NM_001014809.2(CRMP1_v001):c.1052T>C variant has never been reported in the publicly available databases.
Regarding the protein purification, the transient expression, the Western analysis (denaturing and native), and neurite length (4 independent experiments) all seem clean experimental data. The Western blots are clean, and band strength supports the authors' claims. Fluorescence images of the neurons (Fig 3) were not provided but the plots of the four experiments support again the authors' conclusions.
To analyze the effect of CRMP1 variants on its protein levels and cellular function, two isoform variants (CRMP1B-P475L (P1) or -T313M (P2)) were chosen for further functional analysis. CRMP1B T313M and/or -P475L expression might perturb the oligomerization of CRMP1B-wildtype in a dominant-negative manner. Based on the data P475L may exhibit a stronger dominant-negative effect than T313M. -
Reviewer #2 (Public Review):
In "Monoallelic CRMP1 gene variants cause neurodevelopmental disorder," Ravindran et al. attempt to demonstrate that de novo variation in CRMP1, the gene encoding collapsing response mediator protein 1, causes human neurodevelopmental disorders by disrupting CRMP1 oligomerization and neurite outgrowth.
The major strengths of the article include:
- Strong genomic data. The identification of de novo, ultra-rare, predicted damaging missense CRMP1 missense variants in three probands with neurodevelopmental disorders is compelling evidence of a novel disease-gene association.
- Thorough phenotypic characterization of all subjects demonstrating a non-syndromic neurodevelopmental disorder characterized by developmental delay, behavioral abnormalities, intellectual disability, and autism.
- Strong functional data …
Reviewer #2 (Public Review):
In "Monoallelic CRMP1 gene variants cause neurodevelopmental disorder," Ravindran et al. attempt to demonstrate that de novo variation in CRMP1, the gene encoding collapsing response mediator protein 1, causes human neurodevelopmental disorders by disrupting CRMP1 oligomerization and neurite outgrowth.
The major strengths of the article include:
- Strong genomic data. The identification of de novo, ultra-rare, predicted damaging missense CRMP1 missense variants in three probands with neurodevelopmental disorders is compelling evidence of a novel disease-gene association.
- Thorough phenotypic characterization of all subjects demonstrating a non-syndromic neurodevelopmental disorder characterized by developmental delay, behavioral abnormalities, intellectual disability, and autism.
- Strong functional data supporting the variant impact of gene function and neuronal properties.
Weaknesses of the article include:
- Spelling errors and difficult-to-understand language. The use of "variant" is now preferred over mutation. According to current nomenclature, predicted but not experimentally confirmed protein alterations should be written as p.(Phe351Ser) rather than p.Phe351Ser.
- Inconsistent use of in silico pathogenicity predictors and conservation metrics. These should be standardized for each case and should include at least phylop, CADD, and REVEL.
- CRMP1 is under significant constraint against loss-of-function variation in gnomAD - pLI = 0.99, LOEUF 0.28. Genes in the top decile are highly enriched for haploinsufficiency as a disease mechanism. This should be considered in the interpretation of this data and incorporated into the manuscript.
- I am not convinced the data supports a dominant-negative interpretation. The variants do not oligomerize as well as wild-type CRMP1, and when co-expressed with wild-type CRMP1 there is an increase in monomeric wild-type CRMP1. While this could support a dominant-negative interpretation, an alternative explanation is these are loss-of-function alleles that cannot oligomerize, and at the stoichiometry of this artificial overexpression system, this leads to increased monomeric wild-type CRMP1. The axonal outgrowth studies are more compelling, but without a loss-of-function control allele, it is difficult to interpret.
- The experiments in Figure 2 should be replicated, quantitated, and their statistical significance confirmed.
-
Reviewer #3 (Public Review):
This manuscript identifies specific dominant-negative mutations in the CRMP1 gene encoding Collapsing response mediator protein 1 involved in cytoskeletal remodeling. The authors identified 3 independent probands, each with a de novo CRMP1 mutation-based upon unbiased exome or genome sequencing. Family 1 showed p.P589L/p.P475L, family 2 showed p.T427M/p.T313M and family 3 showed p.A351S/p.A237S. CRIMP1 is known to homo-oligomerize, and the paper attempts to show defects in this ability with the incorporation of patient mutations. Finally, forced expression of patient mutations into neuronal cells show defects in the length of the longest neurite.
Major weakness:
The major weakness is Figure 2, as it is not performed up to high standards like the rest of the paper. Panel A does not show any loading control …
Reviewer #3 (Public Review):
This manuscript identifies specific dominant-negative mutations in the CRMP1 gene encoding Collapsing response mediator protein 1 involved in cytoskeletal remodeling. The authors identified 3 independent probands, each with a de novo CRMP1 mutation-based upon unbiased exome or genome sequencing. Family 1 showed p.P589L/p.P475L, family 2 showed p.T427M/p.T313M and family 3 showed p.A351S/p.A237S. CRIMP1 is known to homo-oligomerize, and the paper attempts to show defects in this ability with the incorporation of patient mutations. Finally, forced expression of patient mutations into neuronal cells show defects in the length of the longest neurite.
Major weakness:
The major weakness is Figure 2, as it is not performed up to high standards like the rest of the paper. Panel A does not show any loading control and does not confirm. Panel B at 720 kDa band is not convincing. Results should be repeated with size exclusion chromatography and/or another method to determine molecular weight and should be quantified from triplicate experiments. Panel C is also not convincing and should be repeated to more carefully show results, and quantified.
-
