The latest available data concerning XLMR have been compiled by the European XLMR Study Group and released on the XLMR Genes Update Web site. In all, 209 families with XLMR (140 with MRXS and 69 with MRX) have been entered into the database so far, and 118 loci associated with XLMR have been mapped on the X chromosome. Of those, 34 have yielded specific genes as causes for mental retardation. Seven genes responsible for MRX have been cloned; several among them—PAK3 (p21 activating kinase 3 [MIM 300142]), OPHN1 (oligophenin-1 [MIM 300127]), and ARHGEF6 (a guanine nucleotide exchange factor for Rho GTPases [MIM 00267])—are involved in the Rho-GTPase cycle, which mediates organization of the cytoskeleton, cell shape, and motility in neuronal cells (Tonilo ²⁰⁰⁰; Chelly and Mandel ²⁰⁰¹). In addition, mutations in GDI1, one of the Rab GDP-dissociation inhibitors that control recycling of Rab GTPases, have been revealed in two families with MRX (MRX41 and MRX48) (^{D’Adamo et al. 1998}).

On the other hand, 27 genes responsible for MRXS have been identified. Among them are FGD1 (faciogenital dysplasia protein 1), an activator of Rho GTPase (Cdc42) (Zheng et al. ¹⁹⁹⁶), and DDP (deafness/dystonia peptide), which shows similarity to yeast mitochondrial proteins and plays an important role in the mitochondrial protein-import system (Jin et al. ¹⁹⁹⁶). Moreover, a missense mutation in the RSK2 (ribosomal S6 kinase 2 [MIM 300075]) gene, which was first identified as the element responsible for Coffin-Lowry syndrome, has also been reported in one family (MRX19) (Trivier et al. ¹⁹⁹⁶). The combined evidence indicates that genes participating in various stages of intracellular signaling, especially members of the GTP-binding family, are likely to play critical roles in the development of cognitive functions, and impairment of their activities may result in mental retardation (Tonilo ²⁰⁰⁰; Chelly and Mandel ²⁰⁰¹).

Recently we detected a constitutional chromosomal aberration in a 16-year-old male patient with Duchenne-type muscular dystrophy (DMD [MIM 300377]), an inv(X) (p21.2q22.2). This patient’s DMD symptoms were accompanied by profound mental retardation, athetosis, and nystagmus. Such profound neurological symptoms are quite rare in DMD; however, loci associated with MRXS in 10 families and with MRX in 8 others have been mapped around Xq22-23. We hypothesized that the breakpoints rearranged in this patient’s inv(X)(p21.2q22.2) might have disrupted not only the DMD gene but also additional, unidentified gene(s) that could account for his neurological symptoms.

The patient was the third child of nonconsanguineous Japanese parents; he was born by spontaneous normal delivery, with a birth weight of 3,000 g. No history of neuromuscular anomalies had been described in his family. As a newborn, he exhibited less-than-normal movement and activity, as well as weakness in suckling. A muscle biopsy at age 4 mo indicated that these features were compatible with muscular dystrophy. His condition was subsequently diagnosed as congenital muscular dystrophy accompanied by mental retardation, congenital nystagmus, and athetosis. He was never able to speak any meaningful words or to sit without support. When he was 15 years old, DMD was diagnosed by use of DNA analysis that indicated deletion of DMD exons 43–60. Subsequent cytogenetic analysis of peripheral lymphocytes revealed karyotypes in the patient and his mother of 46,Y,inv(X)(p21.2q22.2) and 46,X,inv(X)(p21.2q22.2), respectively. By this time (age 16 years), he was severely hypotonic, with extreme weakness of the head and neck muscles, but retained smooth-pursuit eye movements. He presented rigidity of the spine, scoliosis, and increased lordosis as well. His mental retardation was too severe for its actual degree to be assessed. At the age of 18 years, he died of suffocation due to tracheal obstruction by viscid sputum. Before the present study, appropriate informed consent was obtained from the patient’s mother.

Epstein-Barr virus–transformed lymphoblastoid cell lines (LCLs) were established from peripheral blood lymphocytes of the patient, his mother, and normal volunteers and were employed in the present study. We first performed FISH, as described elsewhere (Ariyama et al. ¹⁹⁹⁵; Saito et al. ¹⁹⁹⁵), to investigate the breakpoints involved in the inv(X). A total of 22 CEPH YACs representing the regions of interest at Xp21-p22 (YACs 931B3, 965C6, 808E8, 874A8, 940C10, 779G5, 954F3, 912B8, 939F7, and 821F5) and Xq21-q22 (YACs 768A10, 756A5, 886H6, 896F12, 874B5, 223D3, 883D1, 768E1, 857G9, 962C9, 821C4, and 940G5), and a BAC clone (RP11-208J19) containing the human PLP gene were used as probes. Among the 22 YACs examined, we observed no specific signals for 6 (808E8, 940C10, 939F7, 779G5, 954F3, and 874A8), and the signals for YACs 912B8 and 965C6 were reduced on inv(X) chromosomes (fig. 1A) from both the patient and his mother. However, strong and specific signals of all eight of those YACs were detected on Xp21 in normal X chromosomes from the mother and from normal volunteers, indicating that submicroscopic regions of DNA involved in the inv(X) were deleted between DXS1219 and WI-2651 and that the distal site of the deletion lay within YAC 965C6. When FISH analysis was performed using a BAC that contained human PLP sequence, an extra-copy signal was detected at Xp21 in addition to a normal copy of PLP on inv(X) in both the patient (fig. 1B) and his mother (data not shown).

Figure 1 Results of investigations of Xp21. A, Summarized genomic structure at Xp21 around one breakpoint in the patient with inv(X). YACs and STSs between markers WI-7235 and WI-5231 were positioned according to information archived in the Whitehead Institute/MIT (more ...)

PCR amplification was performed with genomic DNA as a template and with primer-pairs designed to amplify the STS markers or the additional seven primer-pairs indicated in figure 1A. (PCR primer sequences and reaction conditions are available on request from the corresponding author.) Amplified products were obtained at loci between DXS1097 and DXS985, as well as in DMD exon 41, whereas no PCR products were generated at loci from DXS726 to WI-9121 or in DMD exon 42 (fig. 1A). These results indicated that the proximal boundary of the Xp21 deletion of inv(X) lay between exons 41 and 42 (intron 41) of the DMD gene and that the telomeric boundary would lie between DXS726 and DXS1097. Additional PCR analyses, using primer pairs Pr1–Pr5 (fig. 1) that were prepared according to the genomic DNA sequence database for this portion of the X chromosome, yielded products only with primers Pr1–Pr3. Therefore, the proximal breakpoint at Xp21 in the inv(X) lies between Pr3 and Pr4 (fig. 1A). We estimated the length of the deletion to be ~1.3 Mb (Nataraja et al. ¹⁹⁹⁸).

For further analysis of the Xp21 breakpoint, we performed Southern blot analysis. Genomic DNAs extracted from LCLs derived from the patient, his mother, and normal individuals were digested with HindIII, HindIII plus EcoRI, or HindIII plus BglII, as described elsewhere (Imoto et al. ²⁰⁰¹). Southern blotting detected rearranged bands only when the PCR product from the Pr3 primer set was used as a probe (fig. 1C), indicating unambiguously that the Xp21 breakpoint lay in the vicinity of the Pr3-amplified sequence. Next, we constructed and performed screening of a cosmid library from genomic DNA derived from the patient, as described elsewhere (Inazawa et al. ¹⁹⁹³). Clones containing the DMD (Xp21) breakpoint were screened by colony hybridization, using an [³²P]-dCTP–labeled PCR product Pr3 (fig. 1A) as a probe, and we obtained five clones by screening the cosmid library of genomic DNAs derived from the patient with the Pr3 PCR product as the screening probe. When FISH analysis was performed, using these five clones as probes, signals for cos11T3 appeared split into two loci (i.e., Xp21 and Xq22) on the normal chromosome but were seen only at Xp22 on the inv(X) (data not shown). Therefore, cos11T3 was most likely to contain sequence harboring the Xq22 breakpoint of the inv(X). Subsequently, subcloned HindIII/BglII fragments of cos11T3 into pBluescript SK plasmid vector (Stratagene) were sequenced using PRISM 377 ABI autosequencers (Applied Biosystems Japan); sequences were assembled with the Auto-Assembler program (Applied Biosystems) and were characterized using the ^BLAST program.

By sequencing, we identified two clones (pc1105 and pc1117) that contained parts of DMD intron 41 and also a sequence from cosmid U237H1 (Xq22; ^GenBank accession number Z95624) that we identified as a novel ras-like GTPase gene and designated “RLGP.” On the basis of the genomic database of the U.S. National Center for Biotechnology Information, physical distance between PLP and RLGP genes is ~836 kb (^GenBank accession number NT_027217). Complete sequence analyses of these two plasmids revealed that the breakpoint at Xq22 lay 143–145 bp upstream from the putative start codon of RLGP (figs. 1D and 2). The region surrounding this breakpoint contained sequence (−250 to −120 bp from the putative start codon) that a computer search (^PROSCAN) predicted as a promoter. The BLAST program predicted that the amino acid sequence of RLGP protein would contain a GTP-binding domain similar to RAS small GTPase, as well as an SOCS (suppressors of cytokine signaling) domain. The putative gene product showed 99%, 88%, and 72% identities, respectively, to RAR2A (Xq22.1), RAR (17q25.3), and RAR-like protein (16p13.3), all of which are human homologues of Saccharomyces cerevisiae SEC4 proteins (data not shown).

Figure 2 Results of investigations of RLGP. A, Genomic structure of the RLGP gene in relation to the breakpoint located in its 5′ region. B, The nucleotide and deduced amino acid sequences of the RLGP-encoding cDNA. The nucleotide sequence is numbered (more ...)

To gain insight into the distribution of RLGP mRNA, we performed northern blots containing RNA from different human tissues (human 12-lane MTN blots) (Clontech). As shown in figure 3, a single RLGP transcript was detected in brain, heart, skeletal muscle, kidney, and liver but was undetectable in seven other tissues, including peripheral blood leukocytes. Adjustment of the signal to that of GAPDH mRNA indicated that the brain had the highest level of expression among the 12 tissues examined. In addition, we confirmed RLGP-specific RT-PCR product, containing a putative coding region and 5′ as well as 3′ UTR of this gene, could be detected in human brain tissue (data not shown). However, because RNA from the patient or his mother was available only from LCLs established from peripheral lymphocytes and because RLGP was not expressed in these materials (data not shown), we were unable to evaluate expression of this gene in the patient.

Figure 3 Expression of RLGP mRNA in normal human tissues. The same blot was rehybridized with a GAPDH probe, as a control, for RNA loading and transfer.

To further determine the subcellular localization of RLGP, we performed indirect fluorescent immunocytochemistry using a pEGFP-RLGP (green) construct (fig. 4). Expression constructs of RLGP were assembled by cloning the coding sequence of this gene in-frame with green fluorescent protein (GFP) in the pEGFP-C1 or N1 vector (Clontech), and transfected into COS7 cells, using Lipofectamine Plus (Invitrogen). Cells fixed with 2% paraformaldehyde were incubated with mouse monoclonal anti–human mitochondria (1:100 dilution; Chemicon International) or anti-Grp78 (1:200 dilution; Stressgen Biotechnologies) antibodies. Binding was detected by rhodamine-conjugated anti–mouse IgG (1:1,000 dilution; Molecular Probes). Fluorescence was observed by means of a confocal laser-scanning microscope (Personal Confocal System, Olympus). The stained RLGP overlapped with antimitochondrial staining but showed no specific accumulation in the endoplasmic-reticulum compartment labeled by anti-Grp78. The staining patterns of both N- and C-terminal–tagged RLGP constructs were similar. Nuclear accumulation of green fluorescence was occasionally observed in cells transfected with pEGFP-RLGP (fig. 4). However, since this color was detected predominantly in the nucleus of cells transfected with pEGFP mock-vector only (data not shown), the apparent nuclear accumulation may have been nonspecific background due to the proteolytic cleavage of the GFP-RLGP fusion protein.

Figure 4 Subcellular localization of RLGP. COS7 cells were transiently transfected with pEGFP-RLGP (A and D), fixed, and immunocytochemically stained using mouse monoclonal anti-human mitochondria (B) or anti-Grp78 (E) antibodies. Note that the localization of (more ...)

Mild mental retardation is a clinical feature in about one-third of patients with DMD; studies of large cohorts of patients with DMD have revealed a mean IQ ~1 SD below the normal mean (Worden and Vignos ¹⁹⁶²; Dubowitz ¹⁹⁶⁵). The relationship between mental retardation and specific deletions of the DMD gene has been studied, with a significant correlation now recognized between intellectual impairment and macrodeletions of DNA in the distal part of DMD (Bushby ¹⁹⁹⁵; Bardoni et al. ²⁰⁰⁰; Felisari et al. ²⁰⁰⁰). However, those studies were conducted only on patients with DMD who had mild mental retardation. Although very severe mental retardation is quite rare as a clinical complication of DMD, a few patients do have severe mental retardation; this group seems to represent a different entity (i.e., mental retardation–complicated DMD) (Bandoh et al. ¹⁹⁸⁷; Bushby et al. ¹⁹⁹⁵). So far as we know, very severe mental retardation has been reported in one patient with a complete deletion of DMD, including its 5′ and 3′ noncoding regions (Alimsardjono et al. ¹⁹⁹⁵), and in two unrelated families in which DMD exon 66 was skipped in transcription (Wibawa et al. ²⁰⁰⁰). However, the patient presented here merely showed a common type of DMD deletion, involving exons 41–79.

Renier et al. (¹⁹⁸¹) recognized three types of Pelizaeus-Merzbacher disease (PMD [MIM 312080]), a mental retardation syndrome. The classic type, with onset in infancy and death in late adolescence or young adulthood, is the most common form and is characterized by initial signs of nystagmus and extrapyramidal effects, including rolling head movements or head tremor. The second, “connatal” type shows rapid progression and is fatal in infancy or childhood; the third, a transitional form, is intermediate. Duplication of the PLP gene is detected in 60%–70% of patients with PMD (Sistermans et al. ¹⁹⁹⁸; Inoue et al. ¹⁹⁹⁹), and such patients invariably show the classic form with a milder phenotype (weaker demyelination) than that of patients who carry PLP point mutations (Sistermans et al. ¹⁹⁹⁸; Inoue et al. ¹⁹⁹⁹). Patients with PLP duplications commonly maintain head control and can speak in sentences, and some are able to walk with assistance. FISH analysis in the present patient unambiguously revealed duplication of PLP in the inv(X); some of his clinical features, such as nystagmus, extrapyramidal athetosis, and relatively long course, met the criteria for the classic type of PMD. Moreover, pathological findings in brain sections, stained with Kluver-Barrera and Holzer stains after his death, showed weak demyelination in the cerebral white matter, especially in both temporal lobes (data not shown), an observation that was compatible with the classic type of PMD involving PLP duplication (Sistermans et al. ¹⁹⁹⁸; Inoue et al. ¹⁹⁹⁹). Among our patient’s clinicopathological features, only the degree of mental retardation was discordant with the classic type of PMD. We inferred that additional molecular mechanisms might be implicated in the pathogenesis of his very severe mental retardation.

We successfully isolated a novel gene, RLGP, from the inv(X) breakpoint at Xq22.2 in this patient and showed that the breakpoint disrupted its putative promoter region. So far, only seven genes have been identified with mutations in families with MRX, and three of the genes were isolated from breakpoints of X;autosome translocations (Billuart et al. ¹⁹⁹⁸; Kutsche et al. ²⁰⁰⁰; Zemni et al. ²⁰⁰⁰). Of these seven genes, six (GDI1, OPHN1, PAK3, RPS6KA3, IL1RAPL, and TM4SF2) have been shown to participate in various stages of intracellular signaling (Tonilo ²⁰⁰⁰; Chelly and Mandel ²⁰⁰¹). For example, GDI1 and OPHN1 regulate Rab and Rho GTPases involved in vesicle cycling, neurotransmitter release, cell migration, and neurite outgrowth (Chelly and Mandel ²⁰⁰¹). PAK3 is a downstream effector of Rho GTPases (Rac and Cdc42), carrying signals to the actin cytoskeleton (Sells et al. ¹⁹⁹⁷) and mediating MAP kinase cascades, including JNK1 and p38 (Bagrodia et al. ¹⁹⁹⁵). These circumstances strongly support our view that interruption of RLGP expression might have affected cognitive development in the subject of the present study.

Immunostaining of RLGP protein in cells, by use of expression constructs tagged with GFP, clearly demonstrated colocalization of RLGP with mitochondria. A high-affinity GDP/GTP-binding site is present on mitochondrial ribosomes, which may mediate protein synthesis by regulating the initiation complex (Denslow et al. ¹⁹⁹¹). Moreover, several GTP-binding proteins, mainly small G proteins, have been identified at points of contact between the inner and outer mitochondrial membranes (Thomson et al. ¹⁹⁹⁸; Sleer and Hall ²⁰⁰⁰). RLGP protein may work in a similar manner. Moreover, many diseases resulting from defects in mitochondrial DNA involve neurological symptoms. Nuclear genetic disorders such as Friedreich ataxia (FRDA [MIM 229300]), Wilson disease (WND [MIM 277900]), and autosomal hereditary spastic paraplegia (SPG7 [ MIM 602783]) also involve mitochondrial dysfunction as a direct or indirect cause of their neurological degenerative phenotypes (Chinnery et al. ¹⁹⁹⁹). Taking these observations together, we consider it possible that RLGP may play a critical role in the development or function of the CNS, by transducing signals or transporting molecules across mitochondrial membranes.

In keeping with the marked progression of the Human Genome Project, numerous genes associated with MRX and/or MRXS have been identified to date. Here, we have shown that a Ras-like GTPase gene, RLGP, was disrupted by the inv(X)(Xp21.2;q22.2) in a patient with very severe mental retardation, although the actual function of this gene remains unclear. To understand the role that RLGP might have played in intellectual development of the patient, as well as in his other clinical features, we acknowledge a need to conduct further clinicopathological characterizations of RLGP, including mutational screening of this gene in a large group of patients with XLMR. It should be borne in mind that genes associated with at least six MRXSs and eight MRXs have been mapped on Xq22-23 but remain unidentified.

We thank Dr. T. Toda, for valuable advice about the clinical features of the patient, and Dr. A. Iwaki, for technical advice about FISH experiments using BACs as probes. This work was supported by grants from the Ministry of Education, Science, Sports and Culture of Japan (J.I. and I.I.) and by the Special Program of Integrated Genomics and Advanced Medical Frontier Research (J.I.), and the Research Grant 13B-1 for Nervous and Mental Disorders from the Ministry of Health, Labor and Welfare of Japan (F.S.-O.).