Article Text


ATR-X mutations cause impaired nuclear location and altered DNA binding properties of the XNP/ATR-X protein
  1. Carlos Cardosoa,
  2. Yves Lutzb,
  3. Cecile Mignona,
  4. Emmanuel Compec,
  5. Danielle Depetrisa,
  6. Marie-Geneviève Matteia,
  7. Michel Fontesa,
  8. Laurence Colleauxa
  1. aINSERM U491, Faculté de Médecine de la Timone, 27 Bd Jean Moulin, 13005 Marseille, France, bIGBMC, INSERM/CNRS/ Université Louis Pasteur, Illkirch, CU Strasbourg, France, cINSERM U476, Faculté de Médecine de la Timone, Marseille, France
  1. Dr Fontes, Fontes{at}


Mutations in the XNP/ATR-X gene, located in Xq13.3, are associated with several X linked mental retardation syndromes, the best known being α thalassaemia with mental retardation (ATR-X). The XNP/ATR-X protein belongs to the family of SWI/SNF DNA helicases and contains three C2-C2 type zinc fingers of unknown function. Previous studies have shown that 65% of mutations ofXNP have been found within the zinc finger domain (encoded by exons 7, 8, and the beginning of exon 9) while 35% of the mutations have been found in the helicase domain extending over 3 kb at the C-terminus of the protein. Although different types of mutations have been identified, no specific genotype-phenotype correlation has been found, suggesting that gene alteration leads to a loss of function irrespective of mutation type. Our aims were to understand the function of the XNP/ATR-X protein better, with specific attention to the functional consequences of mutations to the zinc finger domain. We used monoclonal antibodies directed against the XNP/ATR-X protein and performed immunocytochemical and western blot analyses, which showed altered or absentXNP/ATR-X expression in cells of affected patients. In addition, we used in vitro experiments to show that the zinc finger domain can mediate double stranded DNA binding and found that the DNA binding capacity of mutant forms in ATR-X patients is severely reduced. These data provide insights into the understanding of the functional significance of XNP/ATR-Xmutations.

  • mutation
  • zinc finger
  • DNA binding activity

Statistics from

Mutations in the human XNP/ATR-Xgene cause several mental retardation syndromes including α thalassaemia with mental retardation (ATR-X),1Juberg-Marsidi syndrome,2 Carpenter-Waziri syndrome,3 Holmes-Gang syndrome (C Schwartz, personal communication), mental retardation with spastic paraplegia,4 and mental retardation phenotypes without α thalassaemia.5 Despite the variability of phenotype associated with XNP/ATR-X mutations, all male patients have a severe mental handicap associated with microcephaly and craniofacial dysmorphism. This suggests that this gene may be involved in brain development and facial morphogenesis. TheXNP/ATR-X transcript encodes a predicted 2492 amino acid protein which harbours a nuclear localisation signal, three C2-C2 zinc finger motifs, a coiled coil domain, and seven conserved “helicase” motifs found in DNA stimulated ATPase and DNA helicases of the SNF2/SWI2 protein family.6-8 A high proportion (65%) of the mutations described so far are clustered in the zinc finger domain,9 10 suggesting that this region ofXNP/ATR-X is a functionally important region.

Although the precise function of XNP/ATR-Xremains to be elucidated, several lines of evidence suggest that it is involved in gene expression regulation via chromatin remodelling. First, we showed that the XNP/ATR-X protein participates in the formation of multiprotein complexes and is able to associate with the human EZH2 protein,11 a human homologue of the Enhancer of zeste Drosophila gene involved in the regulation of homeotic gene expression through chromatin remodelling.12 Second, a specific interaction has been identified between the mouse heterochromatin protein (HP1) and the mouse protein HP1-BP38,13 a clone which has been shown to correspond to the mouse Xnpgene.14 15 Recently, the interaction between XNP/ATR-X and HP1 has been confirmed by immunoprecipitation in human cells.16 Lastly, it has been reported that XNP/ATR-X is a nuclear protein closely associated with the nuclear matrix at interphase and during mitosis the protein is found close to pericentromeric chromatin.16 17

The aim of this study was to elucidate the function ofXNP/ATR-X further, by analysing the consequence of mutations of the zinc finger domain in affected patients. First, we studied the expression and subcellular localisation of the XNP/ATR-X protein using monoclonal antibodies directed against the XNP protein. We then analysed the biochemical function of the zinc finger domain using a standard in vitro assay to evaluate DNA binding activity in controls and patients with mutations resulting in different mutant proteins.

Materials and methods


The three ATR-X male patients used in this study have been described previously.7 10 These patients were selected for study on the basis of two criteria, (1) a classical ATR-X phenotype including severe mental retardation, microcephaly, characteristic facial dysmorphism (epicanthic folds, flat nasal bridge, small and triangular nose, anteverted nostrils, and a triangular mouth), hypotonia, and α thalassaemia, and (2) mutations in the zinc finger domain of the XNP/ATR-X protein. The mutations in patients 1 and 3 are located in exons 8 and 7 of theXNP/ATR-X gene, respectively, causing replacement of an arginine by a cysteine residue (R246C) in patient 1, and a glycine by a glutamic acid residue (G175E) in patient 3. In patient 2, a mutation was found in exon 7 causing the splicing of 63 nucleotides coding for 21 amino acids (V178 to K198) and the maintenance of the open reading frame thereafter.


Construction in pTL10SFlag vector and purification of polyhistidine fusion protein His-XNP321-884 has been previously described.11 Monoclonal antibodies were obtained by immunising mice against purified recombinant His-XNP321-884 protein. Splenocytes were fused to the myeloma cell line SP2/0 and hybridomas were screened by reactivity to the His tagged XNP/ATR-X protein in an enzyme linked immunoabsorbent assay (ELISA), as well as by immunofluorescence and western blot analysis of COS1 cells transfected with a PSG5 derived expression vector containing the XNP321-884 protein. A monoclonal antibody was raised against the XNP protein region corresponding to amino acid residues 321-884 and antibody specificity was shown by immunocytofluorescence of transfected COS-1 cells and western blot analysis COS1 cells transfected with a PSG5 derived expression vector containing the XNP321-884 protein. One IgG1 k antibody (XNP-2H12) was selected and used in this study after production as ascites fluid. The antibody specificity was shown by the following results. First, the monoclonal antibody XNP-2H12 reacts efficiently with overexpressed XNP/ATR-X protein in yeast but not with another similarly produced control protein (fig 1B). Second, immunoblotting experiments of total lymphoblastoid cell material show a band whose size is compatible with the size of theXNP/ATR-X open reading frame (about 280 kDa) and which is absent in extracts from ATR-X patients (see Results). Third, a cross reacting band of similar mobility was found in different mouse tissues (data not shown).

Figure 1

Characterisation and specificity of the XNP/ATR-X antibody. (A) Schematic representation of the XNP/ATR-X gene showing the region corresponding to amino acid residues 321-884 of the protein used to generate the XNP-2H12 monoclonal antibody (indicated by the double arrow). (B) XNP-2H12 reacts efficiently with overexpressed XNP protein in yeast but not with another similarly produced control protein. (C) Total protein extracts obtained from lymphoblastoid cell lines of a normal subject and ATR-X patients resolved by SDS-PAGE, transferred to nitrocellulose membrane and probed with anti-XNP antibody. Two ATR-X patients (1 and 2) (patient 5, Villard et al10 and Villard et al,7 respectively) showed absence of the XNP/ATR-X protein. The third ATR-X patient (3) (patient 2, Villard et al10) showed reduced levels of XNP/ATR-X and an abnormal form of the protein.


Lymphoblastoid cells, grown in suspension, were washed in PBS, cytocentrifuged onto poly-L-lysin coated glass coverslips using a cytocentrifuge (200 rpm for 10 minutes, Cytospin2, Shandon), and fixed in 4% formaldehyde in PBS (pH 7.2) at room temperature for 15 minutes.


The cell preparations were treated as previously described.19 After fixation, the cells were permeabilised with 0.1% triton X-100 in PBS for 15 minutes at room temperature. Non-specific staining was blocked with 1% bovine serum albumin (BSA) in PBS for one hour before immunofluorescence staining. The cell preparations were incubated with monoclonal XNP antibody 2H12 (ascites fluid diluted 1:50 in PBS, 1% BSA) for one hour at room temperature. Following three consecutive five minute washes in PBS, cells were incubated for a further 30 minutes with Cy3 conjugated goat anti-mouse antibody at 1:40 dilution in PBS, 1% BSA. Slides were washed three times with PBS before counterstaining of nuclear DNA with Hoechst stain (10 mg/ml in PBS, three minutes) or DAPI stain (60 ng/ml, five minutes) and mounting in antifade solution (Vectashield). Preparations were observed under a fluorescence microscope (Zeiss Axioplan 2) with 100 × magnification. The images were captured with a CCD camera (photometrics Sensys) treated using IPLab spectrum software (Vysis).


The following oligonucleotide pairs,XNP dh11 (CCAGTGCTGAATGAAGACAAAGATG) andXNP 5′R (GTGCGGAATAAGAGTAGGTTAC), were used to amplify by PCR XNP exons 7 to 9 (accession number U75653) which were cloned into vector pRSET C (Invitrogen). Sequence integrity of the amplified fragment was checked before use. DNA binding was assayed using radiolabelled protein fragments corresponding to aa 131-338 from either wild type XNP/ATR-X protein or from XNP protein found in ATR-X patients and generated using the T7 translation transcription (TnT) rabbit reticulocyte lysate system (Promega) with translation grade [35S]methionine (NEN). In vitro translated products were standardised by analysis on both SDS-PAGE before binding experiments and by quantitation of incorporated [35S]. A total of 50 μg of single stranded DNA or double stranded DNA from calf thymus coupled to agarose beads (SIGMA) were reacted with TnT XNP in 0.5 ml of binding buffer (50 mmol/l NaCl, 50 mmol/l Tris (pH 7.5), 2 mmol/l DTT, 2.5 mmol/l MgCl2, 0.2% Nonidet P-40) for one hour at 4°C. The bound fraction was then rinsed five times at 4°C for 10 minutes in 1 ml of pre-chilled binding buffer supplemented with NaCl. Bound radiolabelled fractions were finally released from the agarose beads by boiling in SDS dye and resolved by 15% SDS-PAGE. To prevent artefactual results, binding experiments were independently repeated twice.



We raised a monoclonal antibody (XNP-2H12) against a fusion protein corresponding to amino acid residues 321-884 of the XNP protein (fig 1A) and showed that this antibody specifically recognised the XNP protein. Using immunoblot analysis, we showed that the monoclonal antibody XNP-2H12 reacts efficiently with overexpressed XNP protein in yeast but not with another similarly produced control protein (fig 1B). By western blot experiments of total lymphoblastoid cell material, we showed a band in the normal control whose size was compatible with the size of the XNP open reading frame (about 280 kDa) (fig 1C). We then tested the protein expression in lymphocytes from three ATR-X patients described above. As shown in fig 1C, patient 1 with a missense mutation described above as a “ hot spot” mutation in the zinc finger domain (R246C),10 and patient 2 carrying a deletion of 63 bp (ΔV178-K198),7 10 did not show a band of the correct size on western blotting, suggesting degradation or instability of the protein. Patient 3 (G175E)10 showed one band of the correct size, but with a reduced intensity and another band corresponding to a unknown 180 kDa product.


Considering that the subcellular localisation of transcription factors is essential for their biological activity, we speculated whether an abnormal subcellular distribution of mutant XNP/ATR-X proteins could account for the disease. To address this issue, the localisation of XNP protein in cells from ATR-X patients was examined by immunofluorescence microscopy. As previously described,16 17 we found that in control cells XNP/ATR-X localises exclusively to the nucleus and is mainly concentrated in interphase into large discrete foci that vary in number (between two and 15 per nucleus) and size (fig 2A). Similar profiles of staining were obtained with fibroblasts cells during interphase (data not shown).

Figure 2

Alteration of the distribution of the XNP/ATR-X protein in lymphoblastoid cells of the patients. Lymphocytes from a normal subject and from ATR-X patients were immunolabelled with XNP antibody (red signals). Cell nuclei are shown enlarged × 2000 (20 mm=10 μm). In a normal control, XNP/ATR-X protein localised exclusively within the nucleus, with discrete foci of intense staining numbering between two to 15 per nucleus (A). The same experiments were performed on lymphocytes of previously reported XNP missense mutations in the zinc finger domain, R246C (B) and G175E (D), and a 21 amino acid deletion ΔV178-K198 (C) showed either absence or a modified pattern of nuclear location.

In contrast to the wild type, mutated XNP/ATR-X protein showed a drastically altered cellular distribution pattern (fig 2B, C, D). The number of intense nuclear dots was severely decreased in cells from the patient with a deletion of 63 bp (ΔV178-K198) in the zinc finger domain (fig 2C). In cells from patients with missense mutations R246C or G175E, a speckled pattern of organisation could not be distinguished and a diffuse XNP/ATR-X staining was observed (fig 2B, D). A few cells showed a more surprising pattern with staining showing cytoplasmic perinuclear signals.


Zinc finger domains have been associated with either DNA binding activity or protein/protein interactions (PHD fingers). To discriminate between these two possibilities, we investigated if XNP could directly bind DNA using an in vitro assay. Difficulties in overexpression of the proteic region corresponding to the zinc finger domains inE coli or in yeast compelled us to synthesise 35S-radiolabelled peptides corresponding to the three C2-C2 zinc finger motifs (aa 131-338) using a coupled transcription translation system in rabbit reticulocytes (TnT). These synthetic peptides were then assayed for their ability to bind DNA homopolymers coupled with agarose.20 21 As shown in fig3A, we found that XNP/ATR-X peptide associates very efficiently with DNA. To study the effect of ionic strength on the stability of the XNP/DNA interaction, protein bound to the single stranded (ss) or double stranded (ds) DNA was washed in a buffer with increasing NaCl concentrations (100, 250, and 500 mmol/l). Association with ss-DNA was detected only when low stringency washing conditions were used, probably ruling out the potential in vivo binding capacity of this domain to ss-DNA. However, we found that binding to ds-DNA was still detectable at 500 mmol/l NaCl, suggesting that hydrophobic interactions rather than ionic contacts may be involved in the formation of the XNP/DNA complex. XNP/ATR-X is therefore, at least in vitro, a DNA binding protein and this activity is mediated by the zinc finger region.

Figure 3

In vitro DNA binding assays of wild type and mutant XNP/ATR-X protein. Five percent (30 000 cpm) of the TnT added to each binding reaction is shown on the left of the panel (input). Bound fractions remaining after washes in 100, 250, or 500 mmol/l NaCl were resolved by 15% SDS-PAGE. The two XNP/ATR-X species observed are probably the result of either an internal initiation event or premature termination. Binding assay was performed as described above using in vitro generated wild type XNP/ATR-X: control (A) and mutated XNP/ATR-X protein with a missense mutation G175E (B) or with a 21 amino acid deletion ΔV178-K198 (C). (D) Quantification of the double strand DNA binding activity of the normal and mutant proteins The percentage of35S XNP peptide remaining bound to the DNA after washes (Y axis values) was estimated using Kodak digital science software (1D Image analysis).

Most of the mutations detected in patients affect the zinc finger domain. To investigate whether these mutations affected the DNA binding capacities of the corresponding XNP/ATR-X protein, we cloned two mutated domains originating from two patients and synthesised the corresponding 35S radiolabelled peptides, as described above. One missense mutation (G175E) and one deletion mutation (ΔV178-K198) were tested. Equivalent amounts of in vitro generated wild type or mutated protein fragments were incubated with single or double stranded DNA cellulose, and bound fractions were analysed as described above (fig 3B, C). The data shown in fig 3D indicate that all mutations result in decreased binding efficiency. The deletion of 21 amino acids in the first zinc finger (ΔV178-K198) as well as introduction of the G175E missense mutation almost completely abolish the in vitro DNA binding capacity.


Mutations in the XNP/ATR-X gene have been associated with a large and varied group of X linked mental retardation syndromes with associated anomalies in multiple systems, suggesting its importance as a fundamental regulatory component in multiple tissues during development. XNP is a large gene (7 kb of ORF) but most mutations detected in familial ATR-X cases and related syndromes are clustered in a 300 bp region, encoding three putative zinc fingers.7 This region is also frequently mutated in sporadic ATR-X cases.10 In this study, we have shown by western blot and immunocytochemical analysis an altered or absent expression of the XNP protein in ATR-X patients with mutations of the zinc finger domain. These domains are classically involved in DNA binding activity, yet it has been proposed that the XNP/ATR-X zinc fingers are homologous to those present in plant homeodomain (PHD) finger domains,9 a structure known to be involved in protein/protein interactions.22

Results obtained with normal controls show localisation of the XNP/ATR-X protein to the nucleus where it is concentrated in a distinct speckled pattern. This is consistent with recent findings showing XNP/ATR-X to be a nuclear protein associated with pericentromeric heterochromatin during interphase and mitosis.16 17 In addition, our findings show that in cells from ATR-X patients with missense mutations or deletions of the zinc finger domain, the XNP protein is mislocalised. This is a similar finding to that of mutants with autoimmune polyendocrinopathy candidiasis ectodermal dystrophy (APECED), who lack PHD domains and consequently show an altered subcellular distribution of AIRE protein fragments.23 24In both disorders, however, the exact mechanism of protein mislocalisation is uncertain, and may be a consequence of failure of protein-protein interaction or a failure of DNA binding activity.

In regard to this question, we and others have previously been unable to find mutations or deletions in ATR-X patients in the domain of protein-protein interaction (aa 321-1201), which is close to the zinc finger domain. Our previous studies using a two hybrid screen failed to show an interaction between proteins associated with heterochromatin and the zinc finger domain.11 In addition, it has recently been shown that mutations in ATR-X patients can cause a change in the pattern of methylation of several repeated sequences including ribosomic DNA, a Y specific sequence, and subtelomeric repeats.25 This may suggest a link between the methylation machinery and the multiprotein complex containing the XNP/ATR-X protein and thus a role of XNP/ATR-X in the recognition and expression of specific genes (such as the α globin gene).

In the present study, we addressed this question further by performing in vitro binding experiments to study the potential nucleic acid binding properties of the XNP/ATR-X protein. We showed that the zinc finger region is sufficient to bind double stranded DNA in vitro with strong affinity, suggesting that XNP could be a DNA binding protein. DNA binding properties of SWI/SNF complexes have already been reported25 for members of the CHD subfamily, which exhibit both helicase/ATPase domains and a DNA binding domain.26However, XNP is the only member of the SWI/SNF family known to contain a zinc finger domain for which DNA binding activity has been documented.

These results are derived from in vitro studies. Support for our hypothesis of impaired DNA binding in patients withXNP/ATRX mutations will require further experiments using purified XNP/ATR-X protein to show that XNP binds DNA in vivo. Owing to the size of the XNP/ATR-X protein, such experiments are technically difficult using available methods. The data presented here describe only non-sequence specific DNA binding and characterisation of DNA recognition motifs is currently being investigated to identify genes targeted to XNP/ATR-X regulation. However, despite these limitations, the reduced DNA binding activity shown in our patients with XNP missense and deletion mutations supports the hypothesis that nucleic acid binding by XNP/ATR-X is functionally significant.

Although compelling evidence now exists that different types of mutations in the XNP/ATR-X gene account for common clinical features, the precise biological basis of the phenotype observed in ATR-X patients is yet to be elucidated. The present study was thus undertaken to determine how mutations in the XNP/ATR-Xgene may affect protein function. For this, we first investigated the subcellular localisation of mutant XNP/ATR-X proteins and found that the cellular distribution of the mutant protein present in ATR-X patients is severely affected. Secondly, we produced a series of recombinant proteins carrying mutations identified in ATR-X patients and tested these proteins for DNA binding. We found that most mutations in the zinc finger domain alter DNA binding. We thus conclude that an alteration of DNA binding is probably the major abnormal biochemical mechanism leading to pathogenicity. Whether the same situation exists for mutations located outside the zinc finger domain, particularly for mutations in the helicase domain, is currently under investigation.


We thank the patients, families, and clinicians for their collaboration. We are grateful to R J Leventer, J L Mandel, D Devys, and L Villard for critical reading of the manuscript and helpful discussions. We also thank the AFM/AP-HM Cell, DNA library, and C E Schwartz who provided ATR-X patient specimens. These studies were funded by grants from the Institut National de la Sante et de la Recherche Medicale and from the Association pour la Recherche contre le Cancer.


View Abstract

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.