Abstract

HUMAN AND MEDICAL GENETICS

REVIEW ARTICLE

Animal models for human contiguous gene syndromes and other genomic disorders

Katherina Walz^I; Patricia Fonseca^I; James R. Lupski^{I, II, III}

^IBaylor College of Medicine, Department of Molecular & Human Genetics, Houston, Texas, USA

^IIBaylor College of Medicine, Department of Pediatrics, Houston, Texas, USA

^IIITexas Children's Hospital, Houston, Texas, USA

^{Correspondence} Correspondence to James R. Lupski Baylor College of Medicine, Dept. of Molecular & Human Genetics, One Baylor Plaza Room 604B, Houston, TX 77030-34398, U.S.A. E-mail: jlupski@bcm.tmc.edu

ABSTRACT

Genomic disorders refer to a group of syndromes caused by DNA rearrangements, such as deletions and duplications, which result in an alteration of normal gene dosage. The chromosomal rearrangements are usually relatively small and often difficult to detect cytogenetically. In a subset of such conditions the rearrangements comprise multiple unrelated contiguous genes that are physically linked and thus have been referred to as contiguous gene syndromes (CGS). In general, each syndrome presents a complex clinical phenotype that has been attributed generally to dosage sensitive gene(s) present in the responsible chromosomal interval. A common mechanism for CGS resulting from interstitial deletion/duplication has recently been elucidated. The DNA rearrangements result from nonallelic homologous recombination (NAHR) utilizing flanking low-copy repeats (LCRs) as recombination substrates. The resulting rearrangements often involve the same genomic region, a common deletion or duplication, making it difficult to assign a specific phenotype or endophenotype to a single responsible gene. The human and mouse genome sequencing projects, in conjunction with the ability to engineer mouse chromosome rearrangements, have enabled the production of mouse models for CGS and genomic disorders. In this review we present an overview of different techniques utilized to generate mouse models for selected genomic disorders. These models foment novel insights into the specific genes that convey the phenotype by dosage and/or position effects and provide opportunities to explore therapeutic options.

Key words: genomic disorders, contiguous gene syndromes, microdeletion, microduplication.

Contiguous Gene Syndromes and other Genomic Disorders

The term contiguous gene syndrome (CGS) was first introduced by Schmickel (Schmickel, 1986) to describe disorders associated with specific chromosomal microdeletions whose phenotypic manifestations were due to the involvement of multiple, unrelated genes that are physically linked to each other (Shaffer et al., 2001). CGS are a specific subtype of genomic disorders. The concept of "genomic disorders" refers to conditions that result from DNA rearrangements due to regional DNA architecture (Lupski, 1998). Genomic disorders are a result of recurrent DNA rearrangements involving unstable genomic regions. The genomic rearrangements are usually flanked by low copy repeats (LCR) ranging in size from 10 to 400 kb, that have > 97-98% sequence identity between them (Stankiewicz and Lupski, 2002). Through the process of non allelic-homologous recombination (NAHR), DNA rearrangements such as deletions, duplications or inversions, are facilitated (Lupski, 1998; Shaffer and Lupski, 2000; Emanuel and Shaikh, 2001; Shaffer et al., 2001; Inoue and Lupski, 2002; Stankiewicz and Lupski, 2002; Lupski, 2003; Shaw and Lupski, 2004).

An increasing number of rearrangement-based disorders have been recognized (Figure 1). The DNA rearrangements are usually small on a cytogenetic scale (< 5 Mb), thus molecular technologies, such as PCR, pulsed-field gel electrophoresis (PFGE) and fluorescence in situ hybridization (FISH), are necessary for the identification and diagnosis of several of these disorders. A listing of representative genomic disorders can be found in Table 1; they range from X-linked deletions (that result in structural and functional nullisomy of the genes involved in males), through autosomal deletions (that cause reduction of gene dosage, hence causing structural and functional monosomy, and phenotypes secondary to haploinsufficient genes), to autosomal duplication (giving an increase of gene dosage, resulting in structural and functional segmental trisomy). Each syndrome is characterized by a specific and complex phenotype. Some patients lacking the classical rearrangement, but harboring point mutations in a candidate gene prompted the identification of the dosage sensitive gene responsible for some of these disorders. Examples of such single predominant haploinsufficient genes in CGS include: mutations in the transcriptional co-activator CBP for Rubinstein-Taybi syndrome (Petrij, et al., 1995); p57(KIP2) for Beckwith-Wiedemann Syndrome (Hatada et al., 1996); JAG1 for Alagille syndrome (Oda et al., 1997; Li et al., 1997); UBE3A for Angelman syndrome (Kishino et al., 1997); NSD1 for Sotos (Kurotaki et al., 2002) and RAI1 for Smith-Magenis Syndrome (Slager et al., 2003). Interestingly, several of these genes encode transcription factors, which can potentially explain the pleiotropic phenotypes present in the patients.

Establishing a correlation between each phenotype and the particular gene that is dosage sensitive has often been a major challenge given the facts that the regions involved are usually large (megabases in size) and include several genes. Furthermore, the same size rearrangement usually occurs in most patients because of the molecular mechanism that results in fixed breakpoints reflecting genome architecture. In cases wherein it was not possible to identify patients harboring varying-sized deletions/duplications, or single gene mutations were not found, or a translocation disrupting a particular gene in the critical region was not available, murine models have proven to be extremely valuable for elucidating the predominant gene(s) responsible for the phenotype. Murine models are also essential for evaluating the effects of modifier loci.

Generation of Chromosome Rearrangements

The mouse shares physiologic, anatomic and genomic similarities with humans and can be readily manipulated genetically. It has thus become an important animal model for studying human disease. The completion of the sequencing of the human and mouse genomes enables comparative genomic analyses. Many genomic regions between these species maintain conserved syntenic relationships. Chromosomal deletions, duplications, inversions and translocations can be produced in mice using radiation or chemicals mutagens (Russell, 1951; Rinchik et al., 1993). The disadvantage of these methods is that the rearrangements are created at random, without the ability to predetermine their endpoints. Another disadvantage is the relatively large size required for cytogenetic detection, therefore skewing toward a selection of large deletions in gene poor regions of the genome. You et al. (1997) could generate deletions in the genome combining irradiation and ES cell technology. The advantage of this methodology relies on the fact that ES cells are evaluated which reduces the amount of work involving the whole animal. However, since these deletions are still generated at random, a significant amount of work is required for the molecular characterization of each deletion.

Only recently have strategies been developed to introduce defined chromosomal rearrangements in the mouse genome by engineering them in embryonic stem cells (ES) using the Cre-loxP site-specific recombination system. Allan Bradley and colleagues described the first generation of megabase genomic rearrangements in mice utilizing chromosomal engineering (Ramirez-Solis et al., 1995).

To generate a specific genomic rearrangement between two endpoints, two sequential gene-targeting steps are required in order to prepare each endpoint for selectable Cre-loxP recombination (Ramirez-Solis et al., 1995)(Figures 2 and 3). The double-targeted ES cell is then transiently transfected with Cre recombinase (an enzyme that catalyzes site specific recombination between the specific 34 bp loxP sites) that facilitates the recombination between the two loxP sites, achieving the corresponding genomic rearrangement. The type of chromosome rearrangement obtained will be a direct consequence of the relative initial loxP configurations. If the loxP sites are in the same or direct orientation, the region between them will be deleted or duplicated. If the two loxP sites are inverted or in opposite orientation, then the genomic region between them will be inverted (Mills and Bradley, 2001; Yu and Bradley, 2001). Another versatility of this methodology is the possibility to choose the desired endpoints, with the extra advantage of the availability of two complementary libraries (the 5'Hprt library and the 3'Hprt library) for the initial targeting events. About 100,000 clones from these libraries have recently been sequenced and indexed on the mouse genome as a Mutagenic Insertion and Chromosome Engineering Resource (MICER) (Adams et al., 2004). Each of the libraries contain one half of the Hprt cassette in the vector backbone that when reassembled can be selected for in culture due to the resistance to hypoxanthine aminopterin thymidine (HAT) conferred by the reconstructed Hprt gene generated after the Cre mediated recombination. The vector backbones also include coat color markers and selectable markers (Zheng et al., 1999). If the targeted endpoints are in the same chromosome, or in homologous chromosomes, then a deficiency, duplication or an inversion can be obtained; if the endpoints are in different chromosomes, a balanced translocation can be generated.

Mouse Models for Complex Syndromes Due to DNA Rearrangements

Chromosomal engineering mediated by the Cre-loxP recombination system is particularly useful for modeling large deletions and their reciprocal duplications, as well as inversions. Other techniques have been explored to refine the critical region and to identify dosage sensitive gene(s) responsible for different phenotypic features in several CGS deletions (Figure 4). These include the generation of nested chromosomal deletions, targeted mutagenesis and transgenesis. Although extremely informative relative to a potential role for two or more different genes in the etiology of the disease, or position effects as they relate to phenotypic expression, the generation of nested deletions is time-consuming and may not always reproduce the full clinical spectrum of the disease.

Targeted mutagenesis is the most logical approach when there is evidence that a single gene is responsible for the observed clinical features. It has been valuable in the case of genomic disorders wherein it was possible to identify non-deleted patients with the full/partial phenotype associated with point mutations in a specific gene. However, it is possible that a single gene is responsible for the predominant clinical features, while other genes in the critical region convey modifying effects. In this case, a double knock-out may be necessary to completely reproduce the clinical spectrum. Targeted mutagenesis can be combined with nested deletions, if available, or simply be supplemented by transgenesis to elucidate phenotypic effects of different allelic combinations.

Transgenesis has been used extensively to: i) narrow down the critical region in certain CGS, ii) to produce overexpressing animal models that mimic duplication syndromes, and iii) to complement the phenotype (either fully or partially) when a murine model constructed by either targeted mutagenesis or chromosome engineering of the disease is available. This technique is rapid relative to the previous ones described and it is particularly powerful to assess gene dosage effects. Initially, relatively small transgenes integrated in the mouse genome showed low levels of expression, but with the recent availability of large genomic clones (YAC/BAC/PAC), transgene expression is usually spatially and temporally correct, since the regulatory regions crucial for full expression are included. Levels of expression mimic the endogenous gene and correlate with the number of copies integrated in the genome. The copies are usually integrated in tandem at a unique site within the genome, although rearrangements accompanying transgenesis have been described.

When analyzing a complex phenotype in the mouse and attempting to recapitulate a CGS, several factors must be considered that may complicate the assessment of correlation between a specific gene and a specific phenotypic feature. Rearrangements can occur in transgenic mice; the probability of such increases with the size of the fragment integrated. Dosage sensitive genes can be modified by strain background effects, thus different strains must be evaluated. These caveats can add substantial time and expense to the analysis. Nevertheless, these techniques have enabled the development of mouse models for several genomic disorders. Here we discuss selected examples in which insights into the disease process were gleaned from mouse models.

Charcot-Marie-Tooth Disease - Human Hereditary Neuropathy with Liability to Pressure Palsies

Peripheral myelin protein-22, PMP22, is a major component of myelin expressed in the compact portion of essentially all myelinated fibers in the peripheral nervous system and is produced predominantly by Schwann cells. A duplication on chromosome 17p12 was identified as the basis of Charcot-Marie-Tooth disease type 1 (CMT1A) (Lupski et al., 1991), whereas an interstitial deletion including the PMP22 gene was observed in patients with Hereditary Neuropathy with Liability to Pressure Palsies (HNPP) (Chance et al., 1993). These distinct conditions represent prototypical disorders for the elucidation of the mechanism responsible for genomic disorders, as low copy repeats (LCRs) were identified in the region involved and nonallelic homologous recombination (NAHR) using these repeats as recombination substrates results in either the duplication or deletion (Pentao et al., 1992; Chance et al., 1994, Reiter et al., 1996; Reiter et al., 1998).

A mouse model for CMT1A was constructed by pronuclear injection of a YAC containing the human PMP22 gene and a large proportion of the region duplicated in CMT1A (Huxley et al., 1996). This mouse developed a peripheral neuropathy closely resembling that observed in human CMT1A, with progressive weakness of the hind legs, severe demyelination in the peripheral nervous system, and the presence of onion bulb formations on neuropathology. Transgenic lines harboring a much greater Pmp22 copy number (Magyar et al., 1996) revealed complete absence of myelin. A transgenic rat model with increased expression of PMP22 was described by Sereda et al. (1996). PMP22-transgenic rats develop gait abnormalities caused by a peripheral hypomyelination, Schwann cell hypertrophy (onion bulb formation), muscle weakness and reduced nerve conduction velocities. Homozygous transgenic animals completely fail to elaborate myelin.

To model HNPP, the Pmp22 gene was targeted (Adlkofer et al., 1995). Homozygous null mice develop a neuropathy similar to HNPP with very slow conduction velocities, while heterozygous mice show a less severe phenotype. The generation of a hypomorphic allele of the gene by injection of antisense RNA (Maycox et al., 1997) showed that even with a relatively low decrease in levels of endogenous expression, a phenotype resembling HNPP is observed, though at a somewhat slower rate.

Since it was proposed that impaired Schwann cell differentiation was the mechanism responsible for hereditary peripheral neuropathies, Perea et al. (2001) generated a transgenic mouse model for CMT in which mouse Pmp22 overexpression occurs specifically in Schwann cells of the peripheral nerve and is switched off when the mice are fed tetracycline. When the gene is overexpressed throughout life, demyelination occurs. In contrast, myelination is nearly normal when Pmp22 overexpression is inactivated. Interestingly, if the gene is switched off in adult mice, abrogation of the demyelination begins within 1 week and myelination is well advanced by 3 months. The observation of correction of the neuropathology suggests that even after the initiation of demyelination, therapeutic intervention may be of substantial benefit, which is encouraging towards a therapy using drugs or gene transfer. In previously normal adult mice, overexpression of Pmp22 is followed by active demyelination. The authors hypothesized that even adult mice are sensitive to the level of expression of Pmp22 with respect to homeostasis of the myelin sheath (Perea et al., 2001).

The steroid hormone progesterone has been shown to stimulate Pmp22 gene expression both in cultured Schwann cells and in adult mice (Melcangi et al., 1999). Recently, utilizing the rat model of CMT1A with extra copies of the Pmp22 gene (Sereda et al., 1996), male mice that were treated with the progesterone antagonist onapristone showed reduced overexpression of Pmp22 and clinical improvement in the CMT phenotype (as indicated by maintenance of large axons and improved motor performance) (Sereda et al., 2003). Pmp22 mRNA was decreased by 15% in onapristone-treated animals, suggesting that even a small reduction in Pmp22 transcription may have a beneficial effect on the disease course (Sereda et al., 2003). The availability of different models expressing different levels of the PMP22 protein (overexpressing transgenic mice, hypomorph, heterozygous and homozygous null) allowed the establishment of a correlation between the severity of the phenotype, assessed by peripheral myelin ultrastructure and nerve conduction, and protein expression. These mouse models confirmed that gene dosage is a critical etiological factor for both CMT1A and HNPP; and provide fantastic tools for the exploration of therapeutics for these diseases. It is interesting to speculate that progesterone agonist may be useful in the treatment of HNPP.

DiGeorge Syndrome

DiGeorge syndrome (DGS) is a genomic disorder caused by a heterozygous deletion of chromosome 22q11. DGS patients present with several phenotypic abnormalities, including cleft palate, cardiovascular defects, thymic hypoplasia, hypothyroidism, and learning disabilities. The triad that defines the syndrome is congenital heart disease characterized by right-sided lesions (most notably interrupted aortic arch), hypocalcemia resulting from parathyroid hypoplasia, and frequent infections due to immunodeficiency secondary to thymic aplasia. Each of these structures (heart, thymus and parathyroid) is derived from the 3^rd and 4^th pharyngeal pouches.

Several groups attempted the generation of an animal model that recapitulates the syndrome. Lindsay et al. (1999) created an animal model for DGS using Cre-loxP chromosome engineering in ES cells (Ramirez-Solis et al., 1995). They engineered a deletion of a portion of mouse chromosome 16B that is syntenic with human chromosome 22q11 and contains many of the orthologous genes. They reported that 26% of the 18.5 days postcoitum embryos carrying the heterozygous deletion had cardiovascular abnormalities, and 18% of adult mice harboring the deletion or deficiency, had cardiovascular abnormalities. However, such mice had normal levels of calcium, phosphorus, and parathyroid hormone, normal percentages of B and T cells, and the thymus was normal in size. In addition, none of the mice harboring the deletion had cleft palate or gross palatal abnormalities. Deficiency mice showed abnormalities in sensorimotor gating, learning, and memory (Paylor et al., 2001).

Using a combination of chromosome engineering and P1 artificial chromosome transgenesis the Tbx1 gene, that encodes a member of the T-box transcription factor family that maps to 22q11 in humans, was shown to be required for normal development of the pharyngeal arch arteries in a gene dosage-dependent manner (Lindsay et al., 2001). These observations indicated that haploinsufficiency of Tbx1 is sufficient to generate at least one important component of the DGS phenotype in mice. Null mutations of the Tbx1 gene were also produced (Jerome and Papaioannou, 2001). Heterozygous Tbx1^+/- mice had a high incidence of cardiac outflow tract anomalies, thus modeling one of the major abnormalities of the human syndrome. Mice that were hemizygous for a 1.5-Mb deletion syntenic to that on 22q11 in DGS patients were generated by chromosome engineering using a Cre-loxP strategy (Merscher et al., 2001). These mice exhibited heart conotruncal and parathyroid defects. The conotruncal defects could be partially rescued by a human BAC containing the TBX1 gene again implicating TBX1 as having a major role in the molecular etiology of DGS. However, definitive proof that haploinsufficiency of TBX1 is causative for DGS in humans was only recently obtained, when mutations in the TBX1 gene were found in nondeleted patients with features of DGS (Yagi et al., 2003).

The CRKL gene that encodes an SH2-SH3-SH3 adaptor protein maps within the common deletion region for DGS. Mice homozygous for a targeted null mutation at the Crkl locus exhibit defects in multiple cranial and cardiac neural crest derivatives including the cranial ganglia, aortic arch arteries, cardiac outflow tract, thymus, parathyroid glands, and craniofacial structures. Thus, Crkl may be involved in the molecular mechanism underlying DGS in the case of deletion. Another gene that seems to be involved in the cardiac phenotype is RALDH2 encoding retinoic acid-synthesizing enzyme. Hypomorphic alleles in mice result in prenatal death of the mutant mice with heart outflow tract septation defects and anomalies of the aortic arch-derived head and neck arteries, laryngeal-tracheal cartilage defects, and thyroid/parathyroid aplasia or hypoplasia (Vermont et al., 2003). These observations are consistent with a decreased level of embryonic retinoic acid (through genetic and/or nutritional causes) potentially representing a major modifier of the expressivity of human 22q11del-associated DGS. Several gene mutations have been found in mice affecting the development of the pharyngeal apparatus (for a recent review see Vitelli and Baldini, 2003). As most of the DGS patients have the same chromosomal deletion, but a great degree of clinical variability, the presence of modifiers is strongly suggested. Further investigations are required to determine if any of these genes play a role in the DGS phenotype.

Prader-Willi / Angelman Syndrome

Prader Willi syndrome (PWS) and Angelman syndrome (AS) are clinically distinct disorders, and represent important examples of genomic imprinting in humans. Each results from deletion of 15q11-13, paternal deletion or maternal uniparental disomy (UPD) gives PWS whereas maternal deletion or paternal UPD leads to AS. The cardinal features associated with PWS are decreased fetal activity, neonatal hypotonia, obesity and psychomotor and mental retardation. The clinical manifestations of AS include severe cognitive impairment, absence of speech, ataxia, seizures, and inappropriate laughter. Cattanach et al. (1992) described a putative mouse model of PWS, wherein the phenotype was associated with maternal duplication (partial maternal disomy) for the region of mouse chromosome 7 syntenic to human 15q11-q13. The Snrpn gene was not expressed with the maternal duplication. A paternal duplication of the same region (Cattanach et al., 1997) exhibited characteristics of AS, with reduced growth rate during the first 4 to 5 weeks of life, and obesity by 6 months of age, a smaller overall skeletal size, and neurobehavioral differences. Mice with the heterozygous paternal duplication also displayed a mild gait ataxia with slight eversion of the hind limbs and showed marked behavioral hyperactivity relative to their normal sibs in open field testing. Neuropathologic examinations revealed that total brain weight was diminished by about 10%. Electrocorticographic (ECoG) recordings on paternally duplicated mice showed a striking diffuse cortical excitability disturbance that was identical in all animals. However, mice harboring an intragenic deletion in Snrpn are phenotypically normal (Yang et al., 1998) suggesting that mutations of Snrpn are not sufficient to produce the PWS phenotype. Mice with larger deletion, including both Snrpn and the putative PWS-IC (PWS- imprinting center), lack expression of the imprinted genes Zfp127, Ndn and Ipw, and manifest several phenotypes of PWS patients (Yang et al., 1998).

Mice with the maternal or paternal Ube3A genes knocked out were generated and compared with their wildtype littermates (Jiang et al., 1998). Mice with paternal deficiency (p-/m+) were essentially similar to wild type mice. Maternal deficiency (m-/p+) mice resemble human AS with motor dysfunction, inducible seizures, and a context-dependent learning deficit. The absence of detectable expression of Ube3A in hippocampal neurons and Purkinje cells in m-/p+ mice, indicated that imprinting, with silencing of the paternal allele, correlated well with the neurologic and cognitive impairments. Long-term potentiation in the hippocampus was severely impaired. The cytoplasmic abundance of p53 was found to be greatly increased in Purkinje cells and in a subset of hippocampal neurons in m-/p+ mice, as well as in a deceased AS patient. Jiang et al., 1998 suggested that failure of Ube3A to ubiquitinate target proteins and promote their degradation could be a key aspect of the pathogenesis of AS.

Smith-Magenis Syndrome - dup(17)(p11.2p11.2)

Smith-Magenis syndrome (SMS) is a genomic disorder associated with a deletion within sub-band p11.2 of chromosome 17. Recently, mutations in RAI1 (retinoic acid-induced gene 1) were found in rare patients with many features consistent with SMS and without a recognizable deletion (Slager et al., 2003). The SMS clinical phenotype has been well described and includes craniofacial abnormalities, brachydactyly, self injurious behavior, sleep abnormalities, and mental retardation. Less commonly reported is cleft palate, congenital heart defects, seizures, hearing impairment, and urinary tract anomalies. Patients harboring the predicted reciprocal duplication product, dup(17)(p11.2p11.2), have also been described (Potocki et al., 2000). The phenotype for this independent syndrome associated with the duplication rearrangement seems to be less severe than SMS with mild to borderline mental retardation and behavioral difficulties in the few reported patients. Mice carrying a deletion or duplication of the syntenic region on mouse chromosome 11 that s the genomic interval commonly deleted in SMS patients were constructed, using Cre-loxP chromosome engineering. Heterozygous deleted mice exhibit craniofacial abnormalities, seizures, marked obesity, and male-specific reduced fertility (Walz et al., 2003). Compound heterozygous (i.e., Del/Dup) mice did not display any of these major phenotypes consistent with the trait expression being related to gene dosage effects and not consistent with position effects playing a major role in their manifestation. Behavioral testing of these mice indicates that heterozygous male mice carrying the engineered deletion are hypoactive and show circadian rhythm alterations (Walz et al., 2004).

Heterozygous duplicated animals are underweight and do not have seizures, craniofacial abnormalities, or reduced fertility (Walz et al., 2003). Heterozygous male mice carrying the engineered duplication are hyperactive and they have impaired contextual fear conditioning (Walz et al., 2004). Of interest to note is that the behavioral phenotypes seem to be complementary when comparing the Del versus the Dup mice; hypoactive with deficiency while hyperactive with duplication. Is this caused by a gene dosage effect or are position effects involved? This question will be answered by the analysis of Del/Dup animals in the same battery of behavioral tests. The penetrance of the different phenotypes varies with the genetic background of the mice and deletion size (Yan et al., submitted) Disruption of several genes included in the genomic region containing the common deletion found in SMS patients have been reported, Srebp-1 (Shimano et al., 1997), Pemt (Walkey et al., 1997), Myo15 (Probst et al., 1998), Taci (Yan et al., 2001), Top IIIá (Li et al., 1998) and Fliih (Campbell et al., 2002) and no phenotypes have been described in the heterozygous animals indicating that these genes are not responsible for the phenotypes observed in SMS patients. Csn3 , the subunit 3 of the signalosome, deficiency seems not to be responsible for the haploinsufficient phenotypes observed in mice, and homozygous Csn3 -/- results in lethality (Yan et al., 2003).

To clarify the extent to which each gene present in the region contributes to the phenotype observed in SMS, nested deletion encompassing different segments of the initial deletion mouse are being analyzed in our laboratory. Also, individual gene knockouts within the critical region are being developed. Alternatively, BAC transgenesis of candidates genes are being analyzed in order to determine which gene(s) is responsible for the dup(17)(p11.2p11.2) phenotype. Is the same gene(s) involved in the phenotypes expressed with deletion and duplication? Are there multiple genes causing each syndrome? Or different single genes involved? RAI1 seems to be the major player in the SMS phenotype (Slager et al., 2003), but is it responsible for the phenotype present in dup(17)(p11.2p11.2)? Mouse models that recapitulate two independent human syndromes that result from either decreased (deletion) or increased (duplication) gene dosage involving the identical genomic interval enable a powerful means to investigate the phenotypic consequences of gene dosage imbalance.

Adrenal Hypoplasia Congenita - Sex Reversion

Adrenal Hypoplasia Congenita (AHC) and the phenotype resulting from the reciprocal duplication, Dosage-Sensitive Sex Reversal (DSS) are examples of conditions caused by an X-linked dosage sensitive gene, DAX1. DAX1 encodes a nuclear hormone receptor that is a negative regulator of retinoic-acid receptor. Deletion and point mutations of DAX1 result in both AHC and hypogonadotrophic hypogonadism (Muscatelli et al., 1994). When duplicated, DAX1 causes male-to-female sex reversal. Identification of males deleted for the same region suggested that the locus is not required for testis differentiation, but is involved in ovarian development or functions as a link between ovarian and testicular formation (Bardoni et al., 1994). A logical approach to dissect the molecular role of DAX1 gene in mice was to generate a null mutant by targeted mutagenesis (Yu et al., 1998) and an overexpressing transgenic model (Swain et al., 1998), which revealed some surprises.

XY mice carrying extra copies of the transgene show delayed testis development but no sex reversal. The sex reversal phenotype occurs only when crosses are made with mice harboring weak alleles of the sex-determining Y-chromosome gene (Sry). Therefore, the authors hypothesized that Dax1 acts as an anti-testis gene by antagonizing Sry action and that the transgenic mouse model indicates that the dosage-sensitive sex reversal observed in humans is caused by the duplication of the gene. On the other hand, the disruption of the gene showed decreased testicular size and progressive degeneration of the germinal epithelium, however not as a consequence of the hormonal dysfunction. Females had neither development nor fertility affected. Attempts to rescue the phenotype with a Dax1 transgene were only partially successful. Jeffs et al., (2001) and Meeks et al., (2003) generated tissue-specific Dax1 transgenic mice that were crossed with the deficient model. The partial recovery from infertility and testicular pathology suggests that the receptor has a functional role in different gonadal cell types.

Wolf-Hirschhorn Syndrome

Wolf-Hirschhorn syndrome (WHS) is a terminal deletion syndrome caused by segmental haploidy of chromosome 4p16.3. The phenotypic presentation includes a "Greek warrior helmet" facial appearance (prominent glabella, hypertelorism, widely set eyes), various midline closure defects, growth retardation, juvenile seizures, cataracts, iris colobomas, and mental retardation (Lurie et al., 1980; Wilson et al., 1981). Utilizing nested deletions generated by ES cell irradiation (You et al., 1997), five mouse models with different sized deletions in the mouse chromosome 5 (in a region of conserved synteny with human 4p16.3) were produced (Naf et al., 2001). Several phenotypes present in the patients were recapitulated in these animals. These included: growth retardation, susceptibility to seizures, and midline (palate closure, tail kinks), craniofacial, and ocular anomalies (coloboma, corneal opacities). Other phenotypes included cerebellar hypoplasia and a shortened cerebral cortex. Expression of WHS-like traits was variable and influenced by strain background and deletion size.

Conclusion

Genomic disorders refer to a group of syndromes caused by DNA rearrangements, such as deletions and duplications, which result in an alteration of normal gene dosage within a defined genomic interval. Altered dosage of a single gene (e.g. CMT1A and HNPP), or potentially several genes, that are included in the rearranged region is responsible for the phenotype. The chromosomal rearrangements involved in genomic disorders are the result of NAHR utilizing flanking LCRs as recombination substrates. Generally speaking, phenotypes that result from dosage effects may display significant variability. Some variation may reflect modifying effects from i) genetic background, ii) deletion/duplication of other genes in rearranged segment, and/or iii) alterations in allelic gene expression from the nonrearranged allele.

The human and mouse genome sequencing projects are complete. This tremendous genomic information in conjunction with the ability to engineer mouse chromosome rearrangements have enabled the construction of mouse models for CGS and other genomic disorders paving the way to determine the roles of specific genes, allelic interactions, position effects and potential modifier genes in the ultimate phenotypic expression. Single gene manipulations in mice have been ongoing for greater than a decade and many mouse models have been constructed. However, until recently the idea of mouse models for rearrangement based genomic disorders were technically extremely challenging. Fortunately, with the appearance of techniques allowing chromosomal engineering systematic studies to investigate rearrangements and gene dosage effects are being implemented. Moreover, as presented here, in several mouse models for genomic disorders, specific phenotypes present in human patients have been recapitulated in the engineered mice, including learning disabilities and behavioral phenotypes.

An important example of the value of mouse genetics for understanding human disease was shown for DGS. After several attempts to find the gene responsible for this syndrome (Lindsay et al., 1999; Gong et al., 2001), and having a clear clue that Tbx1 was responsible for part of the phenotypes in mice, human patients having the clinical phenotype of DGS, but not a distinguishable deletion were found to harbor mutations in TBX1 (Yagi et al., 2003).

Another advantage of having animal models for these syndromes is clearly the possibility of treating some of the symptoms. We reviewed the improvement seen in a rat model for CMT1A when treated with a progesterone antagonist. Correct dosage of selected genes may be crucial for normal development. The possibility of having different size rearrangements, point mutations, duplications, and transgenic animals will enable systematic investigation to identify dosage sensitive genes and how altered levels of expression can affect phenotypes.

Acknowledgements

We appreciate the critical reviews of Drs W. Bi, J. Yan, P. Stankiewicz and L. Pentao. This work has been supported in part by the National Cancer Institute, NIH (PO1CA75719).

Editor Associado: Angela M. Vianna-Morgante

Received: May 20, 2004;

Accepted: June 14, 2004.

Publication Dates

History