Article Text

Original research
Congenital mirror movements are associated with defective polymerisation of RAD51
  1. Oriane Trouillard1,2,
  2. Pauline Dupaigne3,
  3. Margaux Dunoyer4,
  4. Mohamed Doulazmi5,
  5. Morten Krogh Herlin6,
  6. Solène Frismand7,
  7. Audrey Riou8,
  8. Véronique Legros9,
  9. Guillaume Chevreux9,
  10. Xavier Veaute10,
  11. Didier Busso10,
  12. Coralie Fouquet1,
  13. Cécile Saint-Martin11,
  14. Aurélie Méneret2,12,
  15. Alain Trembleau1,
  16. Isabelle Dusart1,
  17. Caroline Dubacq1,
  18. Emmanuel Roze2,12
  1. 1 INSERM, CNRS, Institut de Biologie Paris Seine, IBPS, Neuroscience Paris Seine, NPS, Sorbonne Université, F-75005 Paris, France
  2. 2 Institut du Cerveau—Paris Brain Institute—ICM, Inserm, CNRS, AP-HP, Hôpital Pitié-Salpêtrière, Sorbonne Université, Paris, France
  3. 3 Genome Maintenance and Molecular Microscopy UMR9019 CNRS, Université Paris-Saclay, Gustave Roussy, F-94805 Villejuif Cedex, France
  4. 4 Hôpital Pitié-Salpêtrière, Département de Neurologie, AP-HP, Paris, France
  5. 5 INSERM, CNRS, Institut de Biologie Paris Seine, IBPS, Biological Adaptation and Ageing, B2A, Sorbonne Université, F-75005 Paris, France
  6. 6 Department of Clinical Genetics, Aarhus University Hospital, Aarhus, Denmark
  7. 7 Service de Neurologie, CHRU de Nancy, Nancy, France
  8. 8 Service de génétique clinique & Service de neurologie, CHU Rennes, Rennes, France
  9. 9 CNRS, Institut Jacques Monod, Université Paris Cité, F-75013 Paris, France
  10. 10 Université Paris-Saclay, Inserm, CEA, Stabilité Génétique Cellules Souches et Radiations, CIGEx/iRCM/IBFJ, Université Paris Cité, F-92260 Fontenay-aux-Roses, France
  11. 11 AP-HP, Hôpital Pitié-Salpêtrière, Département de Génétique Médicale, Sorbonne Université, Paris, France
  12. 12 Hôpital Pitié-Salpêtrière, DMU Neuroscience 6, AP-HP, Paris, France
  1. Correspondence to Dr Caroline Dubacq, INSERM, CNRS, Institut de Biologie Paris Seine, IBPS, Neuroscience Paris Seine, NPS, Sorbonne Université, F-75005 Paris, France; caroline.dubacq{at}


Background Mirror movements are involuntary movements of one hand that mirror intentional movements of the other hand. Congenital mirror movements (CMM) is a rare genetic disorder with autosomal dominant inheritance, in which mirror movements are the main neurological manifestation. CMM is associated with an abnormal decussation of the corticospinal tract, a major motor tract for voluntary movements. RAD51 is known to play a key role in homologous recombination with a critical function in DNA repair. While RAD51 haploinsufficiency was first proposed to explain CMM, other mechanisms could be involved.

Methods We performed Sanger sequencing of RAD51 in five newly identified CMM families to identify new pathogenic variants. We further investigated the expression of wild-type and mutant RAD51 in the patients’ lymphoblasts at mRNA and protein levels. We then characterised the functions of RAD51 altered by non-truncating variants using biochemical approaches.

Results The level of wild-type RAD51 protein was lower in the cells of all patients with CMM compared with their non-carrier relatives. The reduction was less pronounced in asymptomatic carriers. In vitro, mutant RAD51 proteins showed loss-of-function for polymerisation, DNA binding and strand exchange activity.

Conclusion Our study demonstrates that RAD51 haploinsufficiency, including loss-of-function of non-truncating variants, results in CMM. The incomplete penetrance likely results from post-transcriptional compensation. Changes in RAD51 levels and/or polymerisation properties could influence guidance of the corticospinal axons during development. Our findings open up new perspectives to understand the role of RAD51 in neurodevelopment.

  • gene expression
  • germ-line mutation
  • human genetics
  • movement disorders

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • To date, only four RAD51 pathogenic variants have been associated with congenital mirror movements, a rare dominant neurodevelopmental condition with incomplete penetrance.

  • The first reported variants were truncating, and the initial hypothesis was RAD51 haploinsufficiency.

  • The discovery of non-truncating pathogenic variants challenges this hypothesis.


  • With the identification of four additional pathogenic variants, we provide new evidence for haploinsufficiency of RAD51 in congenital mirror movements and suggest that compensation at the protein level could be a mechanism for incomplete penetrance.

  • We found that variants alter RAD51 polymerisation and DNA binding properties.


  • Deciphering the molecular pathogenesis of RAD51 variants in congenital mirror movements is a new step in understanding the various phenotypes related to RAD51 pathogenic variants in humans.

  • It also provides a new perspective for studying the role of RAD51 in neurodevelopment.


Congenital mirror movements (CMM) (OMIM #614508) are characterised by involuntary movements of one hand that mimic and accompany voluntary movements performed by the opposite hand.1 2 Patients with CMM cannot perform pure unimanual movements and they have difficulties making bimanual movements which require independent activity of both hands.3 They often develop muscle and tendon pain in the upper limbs related to manual activities. The pathogenesis of CMM involves defects in the lateralisation of motor control.4 Patients with CMM have corticospinal tract abnormalities at the level of the decussation, altered interhemispheric communication and bilateral activation of primary motor areas during unimanual movements.5 6

CMM is an autosomal dominant condition with incomplete penetrance. The established culprit genes are DCC (deleted in colorectal carcinoma (OMIM *120470)), RAD51 (RAD51 recombinase (OMIM *179617)) and NTN1 (netrin 1 (OMIM *601614)).7–12 DCC encodes a receptor for netrin 1, which promotes attraction and guidance of developing axons across the body’s midline.13 RAD51 is well-known for its nuclear role in homologous recombination with a critical function in DNA repair.14 Its involvement in CMM was thus unexpected, and its possible role in the development of the motor system remains to be clarified.

Although pathogenic variants in RAD51 represent the second genetic cause of CMM, little is known about the molecular link between these pathogenic variants and mirror movements (MMs). The actual consequence on transcription and protein expression has been poorly studied, as have the potential causes for incomplete penetrance. The first reported CMM variants were truncating and the initial hypothesis was RAD51 haploinsufficiency (figure 1A, upper panel).9 Since then, non-truncating pathogenic variants have also been reported in patients with CMM.15 16 Interestingly, dominant negative RAD51 non-truncating variants have been identified in three patients with atypical Fanconi anaemia and in one patient with premature ovarian insufficiency, raising the possibility of distinct mechanisms in different RAD51 pathogenic contexts.17–20

Figure 1

RAD51 pathogenic variants in congenital mirror movements (CMM) families. (A) Location of pathogenic variants on a schematic representation of the human RAD51 protein. Previously reported variants are in the upper panel. Novel pathogenic variants are in the lower panel. AAA+, ATPase domain; circles, truncating variants; diamonds, non-truncating variants; HhH, helixhairpin-helix domain. (B) Pedigrees of the CMM families included in this study. m, mutated allele; +, wild-type allele. Black symbols represent individuals with CMM, symbols with an embedded black circle indicate asymptomatic carriers, white symbols indicate unaffected individuals. Note that all the genotyped relatives affected with CMM harbour the heterozygous RAD51 pathogenic variant (m/+). Symbols with a diagonal line represent deceased individuals. Squares represent males, circles represent females and diamonds stand for additional family members who did not participate in this study. (C) cDNA sequences showing exon skipping in CMM index cases of two families. RNAs were extracted from the patient's lymphoblasts, and RAD51 fragments spanning the presumed skipped exon from the mutant allele were amplified by RT-PCR. For each index case, the upper and lower panels show the Sanger sequencing of the wild-type and exon skipping alleles, respectively. (D) Representative result of in vitro analysis of the splicing pattern of minigenes containing these two exon-skipping variants. Wild-type (WT) and mutant constructs in the pCAS2 vector were transfected into HEK293 cells and the minigenes’ transcripts analysed by RT-PCR and agarose gel electrophoresis (n=3). The RT‐PCR products consist in exons A and B of the pCAS2 vector (grey boxes) +/− exon 8 (family B) or 9 (family C) of RAD51 (white box). Empty: control construct with no RAD51 exon; c.876+1G>A: control construct with a donor splice mutant resulting in RAD51 exon 9 skipping.

Here, we studied seven CMM-causing RAD51 variants, truncating or non-truncating, including four new variants. We investigated RAD51 mRNA and protein levels in lymphocytes of both symptomatic and asymptomatic subjects from CMM families. The consequence of CMM variants on RAD51 polymerisation and DNA binding function was also tested.

Patients and methods


We studied six index cases with typical CMM due to pathogenic variants in the RAD51 gene, and their families. Each patient and available family members underwent a standardised neurological assessment and DNA sampling. The severity of the MM was evaluated with the Woods and Teuber rating scale.21 Family history, MM location and reported functional disability were noted (online supplemental table S1). Blood samples were obtained after informed and signed consent according to local ethics regulations. DNA was extracted using a standard protocol.11

Supplemental material

DNA sequence analysis and in silico analysis

Coding and flanking intronic regions of RAD51 (GenBank accession no. NM_002875.5) were amplified from the patients’ DNA samples, and Sanger sequenced as previously described.9 10 All the primers used in this study are listed in online supplemental table S2). No pathogenic variant was found in other CMM genes. Possible pathogenicity of missense variants was assessed in silico with Combined Annotation Dependent Depletion (V.1.6 with genome annotation GRCh38), Polymorphism Phenotyping V.2 and Sorting Intolerant From Tolerant. Potential splicing effects of I292I and splice variants were predicted using Alamut Visual Plus software V.1.4 (Sophia Genetics). All variants were classified according to the American College of Medical Genetics and Genomics guidelines (ACMG).22

Cell cultures and transfection

The patients’ lymphoblastoid cell lines were established by Epstein-Barr virus transformation of the peripheral blood mononuclear cells and cultured using conventional methods.9

Human embryonic kidney cells (HEK293) were transfected with indicated plasmids using Lipofectamine 2000 Reagent (Thermo Fisher Scientific). RAD51 constructs were derived from the wild-type (WT) human RAD51 cDNA (NCBI Reference Sequence: NM_002875 on GRCh38), or from the patients’ gDNAs for RAD51 hybrid minigenes, as detailed in the online supplemental experimental procedures.

RNA splicing analysis and quantification

Total RNAs were extracted using RNeasy Mini kit (Qiagen) and reverse transcribed with Superscript IV VILO kit (Thermo Fisher Scientific).

For splicing analysis, cDNAs were amplified by PCR separated on agarose gel, and then each band was individually purified by gel extraction and PCR clean-up kit (Macherey-Nagel) and Sanger sequenced. Splicing pattern of RAD51 minigenes was analysed as previously described.23 Briefly, the minigenes were transfected into HEK293 cells and the transcripts were analysed by RT-PCR and agarose gel electrophoresis 24 hours later.

Transcript quantification was performed by Allele-Specific Reverse Transcription-quantitative PCR (AS RT-qPCR) with GAPDH as the housekeeping gene and using the primers listed in online supplemental table S2.

Protein preparation and quantification

Recombinant His-SUMO-RAD5124 was expressed in Escherichia coli strain and purified as detailed in the online supplemental experimental procedures. Human RPA (Replication Protein A) protein was purified as previously described.25

Whole cell protein extracts were prepared, separated by dodium dodecyl-sulfate (SDS) polyacrylamide gel electrophoresis and analysed by western blot analysis following standard procedures (see online supplemental experimental procedures). All antibodies used for this study are listed in online supplemental table S3.

Mass spectrometry analysis of WT and mutant peptides from the patients’ protein samples was performed as described in the online supplemental experimental procedures.

RAD51 co-immunoprecipitation

Forty-eight hours after transfection, whole cell extracts of HEK293 cells were prepared and 150 µg were incubated with 40 µL of Pierce Anti-Myc Magnetic Beads (Thermo Fisher Scientific). The beads were washed with immunoprecipitation buffer (500 mM Tris HCl (pH 7.5), 1% Tween-20, 3 M NaCl). Proteins were eluted at 95°C with non-reducing Lane Marker Sample Buffer (Thermo Fisher Scientific); 5 μg of whole cell extracts (Input) and 25% of the IP fraction were analysed by western blot analysis.

RAD51 filament assembly and DNA displacement loop assay

Nucleotides (nt) of DNA (15 µM, NEB) were incubated with 2.5 or 5 μM RAD51 (1 protein per 6 or 3 nt, respectively) in reaction buffer (10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 2 mM MgCl2, 2 mM CaCl2, 1.5 mM ATP, 1 mM DiThioThreitol (DTT)) 15 min at 37°C. For single-stranded DNA (ssDNA), 0.15 μM RPA (1 protein per 100 nt) was also added 3 min after RAD51. The reaction was quickly diluted 120-fold in reaction buffer without ATP and DTT, and 5 µL were deposited on a 600-mesh copper grid previously covered with a thin carbon film and pre-activated by glow discharge in the presence of amylamine (Sigma-Aldrich).26 27 The grids were rinsed and positively stained with aqueous 2% (w/v) uranyl acetate, dried and observed in the annular dark-field mode in zero-loss filtered imaging using a Zeiss 902 transmission electron microscope. Images were captured at a magnification of 85 000× with a Veleta CCD camera and analysed with iTEM software (Olympus Soft Imaging Solution).

Six μM in nt of a 400-nt long Cy5-labelled ssDNA substrate were incubated with 2 μM RAD51 (1 protein per 3 nt) in reaction buffer 15 min at 37°C; 5 µL of the reaction was crosslinked 5 min at room temperature in 0.01% glutaraldehyde, then separated on agarose gel; 10 µL of the reaction was added to 25 nmol in molecules of homologous pUC19 plasmid double-stranded DNA (dsDNA) donor. After 20 min at 37°C, the reaction was stopped and deproteinised using 1% SDS, 12.5 mM EDTA and 0.5 mg/mL proteinase K during 30 min at 37°C, then separated on agarose gel. The DNA displacement loop (D-loop) yield of three independent experiments was quantified using ImageJ software (National Institutes of Health).

Statistical analysis

Normality in variable distributions and homogeneity of variances across groups were assessed with the Shapiro-Wilk and Levene’s tests, respectively. Variables were analysed using one-way or two-way analysis of variance followed by Tukey’s or Dunnet’s post hoc test for multiple comparisons. The statistical tests are described in the legend of each figure.


Novel RAD51 pathogenic variants result in CMM

We further analysed the consequences of the RAD51 nonsense variant, (c.760C>T, p.R254*) found in the previously reported family A,9 focusing on three individuals with various status: non-mutated (III.1, nA +/+), asymptomatic carrier (II.3, nA m/+) and affected (III.10, A m/+) (figure 1B, online supplemental figure S1A). We also analysed another CMM family, of Norwegian origin, harbouring the same truncating variant.11 In addition, we identified five heterozygous variants that could be the cause of CMM in nine affected individuals from five newly identified CMM families (families B–F; figure 1B), and two asymptomatic carriers (I.2 in family B and I.1 in family C), in line with the known incomplete penetrance in RAD51-CMM.9 The characteristics of the patients from the six families are summarised in online supplemental table S1). All affected individuals had typical MMs, and five of them reported difficulties in fine bimanual activities. No other neurological or non-neurological phenotype cosegregated with the RAD51 variants.

Supplemental material

In the index case of family B, we found a variant in the canonic donor site of intron 8, (c.774+1G>C, p.?), transmitted by his asymptomatic mother (figure 1B, online supplemental figure S1A). This variant was absent from the genome aggregation database (gnomAD) and was classified as pathogenic according to the ACMG guidelines (table 1). In silico analysis predicted exon 8 skipping, which was confirmed by cDNA analysis (figure 1C). An in vitro splicing assay confirmed that the c.774+1G>C variant did induce exon 8 skipping (figure 1D). Exon 8 skipping resulted in a frameshift and premature stop codon (PTC) (r.645_774del, p.R251Sfs*4). The resulting mutant mRNA is expected to be degraded by nonsense-mediated mRNA decay (NMD), as observed in family A.9

Table 1

Characteristics of patients with RAD51 pathogenic variants

In family C, the (c.876C>T, p.I292=) synonymous variant was identified in a son with CMM, transmitted from his asymptomatic father (figure 1B, online supplemental figure S1A). This variant was classified as a variant of uncertain significance (VUS) according to the ACMG guidelines (table 1). However, it was absent in gnomAD. We thus performed an in silico analysis: only the ESE Finder and EX-SKIP algorithms predicted a disruption of the splicing donor site in exon 9 with possible exon skipping. Study of cDNAs from the patient’s lymphoblasts did indeed reveal the presence of a RAD51 transcript lacking exon 9 (figure 1C). An in vitro splicing assay confirmed that the c.876C>T synonymous variant did induce exon 9 skipping but in a very limited manner, with the full-length transcript remaining the major transcript (figure 1D). Exon 9 skipping resulted in a frameshift and a PTC occurring in the last exon of RAD51, (r.775_896del, p.F259Ifs*23), suggesting that the resultant mRNA may escape NMD.

In two affected subjects (mother and daughter) of family D, we found a truncating variant corresponding to a deletion of 2 nt, resulting in a frameshift and PTC in the last exon of RAD51, (c.954_955del, p.C319Sfs*3) (figure 1B, online supplemental figure S1A). It was absent in gnomAD and was classified as likely pathogenic according to the ACMG guidelines (table 1). The variant mRNA may escape NMD as in family C.

In three affected subjects (two children and their mother) of family E, we identified an in-frame deletion leading to the loss of two amino acids (c.261_266del, p.A89_T90del) (figure 1B, online supplemental figure S1A). This new variant deleted two highly conserved amino acids (online supplemental figure S1B), was absent in gnomAD and was classified as likely pathogenic according to the ACMG guidelines (table 1).

In two affected subjects (mother and daughter) of family F, we found a missense RAD51 variant, (c.749G>A, p.R250Q) (figure 1B, online supplemental figure S1A). The same missense variant has been reported in a CMM family from North America.16

In summary, NMD of the mutant mRNAs was probably present in families A and B, whereas mutant proteins could exist in families C–F.

Pathogenic variants induce CMM through RAD51 haploinsufficiency

We challenged the hypothesis of RAD51 haploinsufficiency inducing CMM. We investigated whether the RAD51 variants impact WT RAD51 levels and then we assessed the levels of mutant RAD51 (when present). To this purpose, we studied RAD51 expression in the lymphoblasts of the patients with CMM by performing AS RT-qPCR, RT-PCR after emetin treatment, ddPCR (see online supplemental experimental procedures) to specifically quantify WT and mutant RAD51 mRNAs (figure 2, online supplemental figures S2, S3, S4), and western blot analyses to quantify proteins (figure 3).

Figure 2

Expression of wild-type (WT) RAD51 mRNA in congenital mirror movements (CMM) families. RNAs were extracted from the patients’ lymphoblasts and RAD51 mRNA expression was quantified by allele-specific RT-qPCR (n≥3). For each family, the WT RAD51 mRNA levels were normalised to a non-mutated relative. Data are represented as means±SEM. Significance was calculated by two-way analysis of variance followed by Tukey’s post hoc test (ns, non-significant; *p<0.05; ***p<0.001). A, affected; m, mutated allele; nA, non-affected; +, WT allele.

Figure 3

Wild-type (WT) and mutant RAD51 proteins expression in congenital mirror movements (CMM) families. (A) For each family, the upper part shows a representative immunoblot performed on whole cell protein extracts from the patients’ lymphoblasts. The lower part shows the quantification of n≥3 experiments: RAD51 protein levels were normalised to ⍺-tubulin, used as a loading control, and to the WT protein level of a non-mutated relative. Histograms in black correspond to full-length WT RAD51 protein, in white to mutant RAD51 protein, in grey to undistinguishable WT and mutant RAD51 proteins (total RAD51 protein). Data are represented as means±SEM. Significance was calculated by two-way analysis of variance followed by Tukey’s post hoc test (ns, non-significant; *p<0.05; **p<0.01; ***p<0.001). (B) For one family with a non-truncating RAD51 pathogenic variant (family E), immunoprecipitation of RAD51 protein was performed on whole cell protein extracts from affected patients’ lymphoblasts, followed by tandem mass spectrometry analysis. Both the WT and the A89_T90del mutant RAD51 proteins were identified by generating extracted ion chromatograms (XIC) of their characteristic peptides LVPMGFT(TA/)TEFHQR, which distinguish the WT from mutant RAD51 protein. The relative expression ratio of the two isoforms was assessed by calculating the ratio of XIC peak areas. A, affected; nA, non-affected; m, mutated allele; +, WT allele; RT, retention time; AA, peak area.

For all the CMM families, WT RAD51 mRNA levels were 44%–78% lower in mutated individuals as compared with their non-mutated relatives (figure 2). The levels of WT RAD51 mRNA were similar in the three asymptomatic carriers and their affected relatives (figure 2, online supplemental figure S2). In families A–D, the WT and mutated (if present) proteins were expected to be easily distinguished in western blot analysis. Quantification of WT RAD51 proteins for these four families demonstrated a significant decrease of 33%–51% in affected individuals as compared with non-mutated relatives (figure 3A). WT RAD51 protein levels were only 14%–26% lower in asymptomatic carriers, two of them being significantly different from their affected relatives (figure 3A). For families E and F, WT and mutant proteins were not distinguishable using western blot analysis. The levels of total RAD51 proteins (WT+mutant) were significantly lower only for one of the affected individuals (II.1 in family E) as compared with non-mutated relatives (figure 3A). For family E, mass spectrometry analysis of the characteristic peptides of WT and mutant RAD51 proteins found that the proportion of WT protein was 61%–73% of total RAD51 proteins in affected subjects (figure 3B). The combination of the WT/total ratio estimated by mass spectrometry and the total RAD51 protein levels measured by western blot analysis estimated a 46%–58% reduction in the WT protein in affected individuals of this family (online supplemental figure S5). In family F, although the characteristic peptides of the recombinant WT and mutant proteins were detected by mass spectrometry, the peptides could not be detected in the patients’ cells.

Mutant mRNAs were detected at nearly undetectable levels in families A and B (online supplemental figures S2, S3). The increase in the mutant mRNA levels after emetin treatment (in the study by Depienne et al 9 for family A and online supplemental figure S4 for family B) confirmed the degradation of the mutant mRNAs by NMD in these families. In family C, the mutant I292I mRNA was almost undetectable (online supplemental figures S2, S3). The presence of the exon 9 skipping mRNA was detected at similar levels in the affected individual and the asymptomatic carrier, and was also detected in the non-mutated relative (online supplemental figure S2). In family D, the mutant mRNA was detected in the affected individual (online supplemental figures S2, S3). The exon 9 skipping or mutant mRNA levels in families C and D were unchanged after emetin treatment (online supplemental figure S4) confirming that these mRNAs escaped NMD. For families E and F (non-truncating variants), mutant mRNAs were detected at low levels (online supplemental figures S2, S3).

In families A–C, no mutant proteins were detected (figure 3A). Very low levels of proteins at the expected size of the exon 9 skipping isoform of RAD51 were present in the three members of family C as well as in the non-mutated relative of other families (online supplemental figure S6). Altogether, this exon 9 skipping isoform of RAD51 did not appear to be a specific product from the I292I allele, in line with the physiological expression of this isoform.28 In family D, the mutant protein was detected in the affected individual (figure 3A). For family E, the mutant protein was detected in the patients’ cells by mass spectrometry, contributing 27%–39% of total RAD51 proteins in affected subjects (figure 3B).

To summarise, except for family F where information was not available, WT RAD51 protein levels were lower in the lymphoblasts of all affected individuals in RAD51-CMM families compared with the non-mutated relatives. These pathogenic RAD51 levels tended to be lower than those observed in asymptomatic carriers (online supplemental figure S5). As predicted, NMD of the mutant mRNAs was confirmed in families A and B, and mutant proteins were not detectable in these two families, or in family C. In families D and E, the mutant proteins were detected and contributed to more elevated levels of total RAD51 proteins (online supplemental figure S5). In family F, the WT and mutant proteins could not be distinguished and the high total RAD51 levels might be a result of pathogenic levels of WT RAD51 complemented by the mutant protein. Altogether, these results imply that the unknown neurodevelopmental function of RAD51 revealed by CMM is impaired in the mutated proteins of families D–F (p.C319Sfs*3; p. A89_T90del and p.R250Q, respectively).

Pathogenic variants alter polymerisation of RAD51

The three-dimensional structure of the filament formed by RAD51 polymerisation indicated that the affected residues of CMM-mutated RAD51 proteins (A89, T134,15 R250 and C319) were located at the interface of adjacent RAD51 protomers (figure 4A). The A89 residue was in the main RAD51-RAD51 interaction motif.29 The T134 and the C319 residues participated in ATP binding in the Walker A domain and the facing ATP cap,30 respectively, the two domains forming together a secondary RAD51-RAD51 interaction motif. The R250 residue was also involved in a secondary interaction motif, capping the end of an alpha helix in the adjacent protomer. Thus, these four RAD51 pathogenic variants might theoretically influence RAD51-RAD51 interaction.

Figure 4

Effect of RAD51 pathogenic variants on RAD51-RAD51 interaction. (A) The congenital mirror movements (CMM)-mutated residues are located at the interface between RAD51 protomers. Cartoon representation of the structure of human RAD51-ATP filament (PDB=5NWL; each colour corresponds to one protomer) showing in the insets the A89, T134, R250 residues as sticks, the C-terminal end of the protein starting from C319 in yellow. A89 is located at the interface between two RAD51 protomers. T134 is forming a hydrogen bond with the second phosphate of the ATP. C319 is in the ATP cap (D316-P321). ATP is sandwiched between the ATP-binding site and the ATP cap of the adjacent protomer. R250 faces the C-terminal end (containing D231) of an alpha-helix of the adjacent protomer. In the form where RAD51 binds single-stranded DNA (ssDNA) (PDB=5H1B), R250 forms a salt-bridge with D231. (B) Representative immunoblots of RAD51 co-immunoprecipitation (coIP) experiments. The upper panel (input) shows whole cell extracts from HEK293 cells transfected with the indicated wild-type (WT) or mutant HA-tagged constructs and WT Cmyc-tagged constructs (ratio 1:2). The lower panel shows RAD51 coIP with anti-Cmyc beads. A89E and R250A were used as experimental controls. EGFP-Cmyc is used as a negative control. (C) Quantification of the relative coIP efficiency of the RAD51-WT-Cmyc and mutant RAD51-HA (white bars), normalised to the relative coIP efficiency of the RAD51-WT-Cmyc and RAD51-WT-HA condition (black bar; n≥3). For each condition, the IP efficiency was calculated as the ratio of IP normalised to input RAD51 densities both for the prey (HA) and the gait (Cmyc) RAD51 proteins. The relative co-IP efficiency of each condition, that is, the prey (HA) normalised with the gait (Cmyc) IP efficiencies, further normalised with the WT prey (HA) condition, was then calculated for each experiment. Data are represented as means±SEM. Significance was calculated by one-way analysis of variance followed by Dunnett’s post hoc test (*p<0.05, **p<0.01, ***p<0.001).

We therefore tested the ability of mutant proteins to interact with WT protein by co-immunoprecipitation. The C319Sfs*3 mutant protein is destabilised when expressed in heterologous systems and, therefore, could not be included in this part of the study (online supplemental figure S7). An experimental variant altering one of the CMM-mutated residues and studied in the literature, A89E, was included as a positive control of altered RAD51-RAD51 interaction.29 All mutant RAD51 proteins tested showed a reduced relative co-immunoprecipitation efficiency compared with the WT RAD51 protein (figure 4B–C).

As the RAD51-RAD51 interaction is essential for RAD51 polymerisation, we then tested this function in vitro. Whereas the WT RAD51 assembled in a long, continuous and homogenous nucleofilament on dsDNA or ssDNA (figure 5Aa–b, 5Ba–b), the three CMM-mutant RAD51 proteins failed to form a proper nucleofilament on dsDNA or ssDNA: rare fixation to DNA and no filament formation for A89_T90del (figure 5Ac,Bc), formation of discontinuous and irregular filaments for T134N and R250Q (figure 5Ad–e, 5Bd–e), with large aggregates for T134N. Electrophoretic mobility shift assays confirmed that the A89_T90del mutant protein failed to form stable complexes with ssDNA, whereas the T134N mutant protein could interact with DNA but was highly prone to aggregation, and the R250Q mutant protein formed less stable complexes with ssDNA than the WT protein (figure 5D). Furthermore, CMM-mutant RAD51 also lacked the capacity to form a D-loop (figure 5E–G), a homologous recombination intermediate DNA structure which is dependent on RAD51 activity (see WT on figure 5F–G). An equimolar mix of the WT and each of the mutant proteins did not noticeably change the aspect of the filament formed on ssDNA (figure 5C), the ssDNA binding (figure 5D) or the D-loop activity (figure 5F–G) of the same subsaturating quantities of the WT protein only. Overall, the CMM-mutated RAD51 proteins included in this biochemical study showed loss-of-function in terms of RAD51-RAD51 interactions, nucleofilament formation, DNA binding and/or D-loop activity, but did not significantly alter the properties of the WT protein in an equimolar mix.

Figure 5

Functional impact of RAD51 pathogenic variants on RAD51 polymerisation. (A–C) Nucleofilament formation. Dark-field electron microscopy (EM) images of filaments formed by wild-type (WT) and/or mutant recombinant RAD51 proteins on (A) double-stranded DNA (dsDNA, linearised pBR322—4361 bp) or (B, C) single-stranded DNA (ssDNA, PhiX174 virion—5386 nt). Insets in Ab, e show bright-field images of filaments formed in the same condition. In A–B, the nt/protein ratio was 3:1 for WT or mutant RAD51; in C, the nt/protein ratio for each indicated RAD51 protein was 6:1, resulting in equimolar ratio of WT and mutant RAD51 in Cb–d; in B–C, RPA (Replication protein A) protein (nt/protein ratio 100:1) was added to promote unfolding of the ssDNA substrate and allow elongation of the nucleofilament. Inset in Ba shows ssDNA without RPA. Scale bar represents 100 nm. (D) DNA binding assay. Representative electrophoretic migration shift assay (EMSA) of a Cy5-labelled ssDNA substrate after incubation with WT and/or mutant RAD51 recombinant proteins (same nt/protein ratio as in B–C). (E–G) Displacement loop (D-loop) activity. (E) Scheme of the D-loop assay. Filaments are formed on a 400-nt long ssDNA in presence of WT and/or mutant RAD51 recombinant proteins (as in D), followed by addition of a homologous supercoiled dsDNA and deproteinisation of the sample. (F) Representative electrophoretic gel of D-loop assay. (G) Quantification of the percentage of D-loop yield (n=3). Data are represented as means±SEM. Significance was calculated by one-way analysis of variance followed by Dunnett’s post hoc test (*p<0.05, **p<0.01, ***p<0.001).


We describe here four new pathogenic variants of RAD51 in CMM families, in addition to the four previously reported variants. We provide evidence supporting that RAD51 haploinsufficiency is the key molecular mechanism causing CMM. We propose that incomplete penetrance might be the result of a compensation at the protein level. When present, CMM-causing mutant RAD51 proteins result in abnormal polymerisation. We thus speculate that a change in RAD51 levels and/or polymerisation properties could influence axon guidance of the corticospinal tract during development of the motor system.

In a previous study, we identified two truncating variants causing CMM: (p.R254*) and (p.P286Tfs*37) (figure 1A, upper panel).9 For the (p.R254*) nonsense variant, we demonstrated a downregulation of total RAD51 mRNA by NMD resulting in the absence of mutant protein. This raised the possibility of RAD51 haploinsufficiency in CMM.9 11 Here, we identified four new variants, namely two truncating variants (p.C319Sfs*3 and p.R215Sfs*4), a synonymous variant (p.I292=), and an in-frame deletion (p.A89_T90del) (figure 1A, lower panel). A total of eight variants, including three non-truncating (p.A89_T90del; p.T134N15 and p.R250Q16), have thus been associated with CMM so far (figure 1A). No functional study has been performed on the non-truncating variants, except for the overexpression of the (p.R250Q) variant in primary cultures of mouse cortical neurons suggesting a loss-of-function of the mutant protein.31 The possible presence of the mutant RAD51 proteins and the functional consequences in patients’ cells remain a matter of debate.

We studied the expression of WT-RAD51 for the four new CMM-related variants and two previously reported pathogenic variants. We found lower levels of WT-RAD51 in all affected patients with CMM compared with non-mutated subjects at both mRNA and protein levels (when available). Moreover, the asymptomatic carriers had intermediate WT-RAD51 protein levels, lower than those of their non-mutated relatives but higher than affected individuals of the same family (online supplemental figure S5). We further demonstrated either absence of the mutant proteins in three families or loss-of-function of the mutant proteins in three other families. This formally demonstrates that RAD51 haploinsufficiency causes CMM.

The decrease of RAD51 expression associated with the synonymous c.876C>T (p.I292I) variant of family C was puzzling. We excluded a role of alternative splicing and we hypothesise that this synonymous codon could alter mRNA regulatory sequences.32

The three asymptomatic carriers had WT-RAD51 mRNA levels, similar to that of affected individuals, but intermediate protein expression levels. Therefore, post-transcriptional compensation is likely to account for incomplete penetrance by maintaining the WT-RAD51 protein level above a critical threshold. This compensation mechanism in RAD51-CMM families differs from the transcriptional compensation mechanism observed in a DCC-CMM family, in which the WT-DCC mRNA levels of asymptomatic carriers is higher than that of non-mutated individuals.33

In the nucleus, RAD51 polymerises to form a nucleofilament with ssDNA through the assembly of several RAD51 monomers.34 The four residues affected by the non-truncating CMM variants (A89, T134, R250 and C319) are located at the interface between adjacent RAD51 protomers and are probably involved in RAD51-RAD51 interactions. We found an impaired interaction between the A89_T90del, T134N and R250Q RAD51 mutant proteins and WT-RAD51, as previously shown for the experimental A89E variant.29 In addition, all three variants prevented the formation of regular nucleofilaments associated with an inability to interact with DNA and to form D-loops. Of note, in experimental conditions with subsaturating levels of WT protein, addition of equimolar quantities of mutant RAD51 proteins did not interfere with the WT protein DNA binding, filament assembly or D-loop activity. This argues against a dominant-negative effect of the mutant protein, in contrast to what has been described in variants related to atypical Fanconi anaemia or premature ovarian insufficiency.17–20 In cells of patients with CMM, the WT-RAD51 protein is reduced (42%–67% of WT-RAD51 levels in the intrafamilial controls) without any phenotypic evidence for alterations in its homologous recombination activity: in our CMM cohort, none of the patients with RAD51 pathogenic variants presented fertility problems, Fanconi anaemia or cancer. In line with this observation, the cellular levels of RAD51 are thought to exceed the cell’s needs for homologous recombination.35 Our findings strongly support that both truncating and non-truncating RAD51-CMM variants result in a loss-of-function selectively associated with the CMM phenotype.

RAD51 is expressed in the cytoplasm of cortical neurons of neonate mice.9 This critically occurs within a spatiotemporal window corresponding to the midline crossing of the corticospinal tract.9 Impaired midline crossing of the corticospinal tract is a key and consistent feature in patients with CMM.5 6 Together with the present findings, this suggests that a balanced RAD51 polymerisation is important for its cytoplasmic role in the development of the corticospinal tract. This opens up a new framework to investigate the role of RAD51 in the cytoplasm.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

Ethics statements

Patient consent for publication

Ethics approval

This study was approved by ethics committee of the Hôpital de la Pitié-Salpêtrière and the Comité de Protection des Personnes (CPP), Ile-de-France 6, ParisPromotion INSERM RBM 03-482021-A00989-32- Promoteur AP-HP. Participants gave informed consent to participate in the study before taking part.


The authors thank the patients and their families for their cooperation and support of this research. We gratefully acknowledge Ludmila Jornea and Sylvie Forlani for cellular immortalisation. We thank the iGenSeq core facility of ICM for their advice on the ddPCR. Raphaël Guérois and Jessica Andreani contributed for helpful discussions and assistance.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • CD and ER are joint senior authors.

  • Contributors ID, CD and ER designed the work. OT, PD, MDu, MKH, SF, AR, VL, GC, XV, DB, CF, CS-M and AM acquired data. OT, PD, MDu, MDo analysed data. OT, PD, MDu, MDo and AT interpreted data. OT, CD and ER drafted the work. MDu, MDo, MKH, SF, AR, VL, GC, XV, DB, CF, CS-M, AM, AT and ID revised it for intellectual content. CD and ER contributed equally to this paper. All coauthors have approved the final version of the manuscript. Part of this work was carried out on the DNA & cells bank of the Paris Brain Institute and on the real-time PCR facility of IBPS. CD is responsible for the overall content as guarantor.

  • Funding This work was supported by the Fondation Desmarest, Merz-Pharma, Elivie, Orkyn, Djillali Mehri, CNRS, INSERM and Sorbonne Université. This work was also funded by grants from the Agence Nationale de la Recherche (ANR) (ANR-14-CE13-0004-01, ANR-18-CE16-0005-02), from the National Institute of Health NIDCD (R01-DC-017989, USA) and it was performed within the framework of LABEX LIFESENSES (ANR10-LABX-65) supported by French state funds managed by the ANR within the Investissements d’Avenir programme (ANR-11-IDEX-0004-02).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.