Jump to
Congenital heart defects are thought to result from a variety of genetic and environmental influences. Although most CHD occurs as a sporadic event, many defects are part of a well-defined genetic syndrome. Particular types of CHD are often overrepresented in the context of a syndrome compared with the general population. Identification of the genetic etiology of the syndrome may lend insight into the etiology of the associated cardiac malformation. Thus, because conotruncal cardiac defects are a cardinal feature of DGS, the specific genetic etiology of this syndrome has become of great interest to the cardiologist and cardiac developmental biologist.
Although the phenotype is highly variable, DGS is typically characterized by aplasia or hypoplasia of the thymus, aplasia or hypoplasia of the parathyroid glands, conotruncal cardiac defects, and mildly dysmorphic facial features.1 2 The most common cardiac defects include truncus arteriosus, interrupted aortic arch, and tetralogy of Fallot.3 Defects in multiple organ systems, which arise concurrently and from common precursors during embryogenesis, have led to the proposal that DGS is a developmental field defect.4 A developmental field refers to a population of embryonic cells that behave as a single coordinated developmental unit. Disruption of this “morphogenetically reactive unit” results in a particular phenotype, which may involve many different end organs and may be phenotypically variable depending upon the timing and nature of the perturbation. Different factors, such as genetic alterations or environmental insults, could disrupt the normal morphogenesis of the reactive unit and result in a similar phenotype.
Preliminary evidence suggests that in the case of DGS, the morphogenetically reactive unit is the cephalic and cranial neural crest. Experiments in the chicken and rodent have demonstrated that neural crest cells from the preotic and postotic caudal hindbrain form the mesenchyme of the third and fourth arches/pouches. The third and fourth branchial arches/pouches give rise to the thymus, parathyroid glands, aorta, and head and neck vessels. Moreover, neural crest cells from the occipital hindbrain (adjacent to somites 1 and 2) migrate into the outflow tract of the heart and are essential for normal aortopulmonary septation.5 6 7 8 Ablation of neural crest cells in the chick embryo at the level of the occipital somites 1 to 3 causes malformations of the aortopulmonary septum, resulting in common arterial outflow channels or transposition of the great arteries.5 These animal experiments first implicated a subpopulation of the cranial neural crest, or the cardiac neural crest, as the “reactive unit” whose perturbation results in a phenotype reminiscent of DGS in humans.
Early reports suggested that DGS was causally heterogeneous (see Lammer and Opitz,4 1986, for review). For example, the DiGeorge phenotype has been seen in association with chromosomal defects and in a proportion of infants exposed in utero to maternal diabetes, alcohol, or retinoids. The first genetic etiology for DGS was identified when cytogenetic studies revealed abnormalities of chromosome 22 in association with DGS.9 10 Subsequently, it was found that ≈10% to 20% of patients with DGS had visible chromosomal abnormalities, the vast majority of which resulted in monosomy of chromosomal region 22q11.11 This finding suggested that chromosome 22 might contain a locus whose loss or deletion was responsible for DGS. More recent molecular studies indicate that nearly 90% of patients with DGS have a submicroscopic deletion of one copy of the region 22q11.21-q11.23, demonstrating that DGS is a microdeletion syndrome.12 13 14 The molecular mechanism by which segmental monosomy results in the DGS phenotype remains to be defined. Presumably, the phenotype results from the diminished dosage (or haploinsufficiency) of a specific gene product(s) at critical points during early development. However, the commonly deleted segment is large and is known to contain numerous genes.15 16 It is unclear whether the phenotype results from haploinsufficiency of one or several critical genes that map to the commonly deleted region. It also remains to be elucidated whether the marked variability in phenotype results from subtle differences in the deletion itself or from modifying factors such as genetic background and/or environmental influences. Efforts to identify the specific genes, the mechanism by which they produce the DGS phenotype, and, in particular, their relationship to cardiac development are under way and will be discussed in the present review.
The 22q11 Microdeletion Syndrome: Phenotype and Diagnosis
Phenotype: DGS, VCFS, and Cardiac Defects
In contrast to the heterogeneous etiology originally reported for DGS, VCFS was described as an autosomal dominantly inherited disorder characterized by palatal abnormalities (particularly cleft palate), congenital heart disease, learning disabilities, and a characteristic facial appearance.17 18 The most common types of CHD in VCFS include perimembranous ventricular septal defects, tetralogy of Fallot, and right aortic arch.19 Patients with VCFS were evaluated for alterations at 22q11 because they were reported to share some of the features of DGS, most notably the spectrum of their cardiac defects. Molecular studies have demonstrated that nearly 85% of patients with a presumed diagnosis of VCFS have a 22q11.2 deletion.20 21 The deletion in patients with VCFS is indistinguishable from that seen in association with DGS. Thus, DGS and VCFS share a common genetic etiology and appear to represent different phenotypic manifestations of the same underlying molecular defect.
To better define the phenotype associated with a 22q11 microdeletion, patients with isolated features of DGS or VCFS have been studied for 22q11 deletions. In particular, an early study evaluated patients with conotruncal defects who had not been diagnosed with either DGS or VCFS for the presence of 22q11 microdeletions.22 Five of 17 patients were found to have submicroscopic deletions identical to those of DGS and VCFS. More recently, 50 seemingly nonsyndromic patients with conotruncal defects were studied for 22q11 deletions.23 Four of 10 with truncus arteriosus, 1 of 8 with interrupted aortic arch, and 4 of 32 with tetralogy of Fallot were found to have a 22q11 deletion. Subsequently, we and others have observed that among patients diagnosed with CHD in the neonatal period who were initially thought to be nonsyndromic, there are some who demonstrate mild features of VCFS as they mature. Thus, some of the subtle findings associated with VCFS, such as velopharyngeal insufficiency, mild craniofacial dysmorphia, and speech and learning disabilities, only become apparent past infancy in the school-age child. Some of these findings require early intervention and treatment for improved outcome, suggesting a rationale for early diagnosis.24 25
Microdeletions of chromosomal region 22q11 have been found in patients with other cardiovascular defects and in patients diagnosed with syndromes other than DGS or VCFS. Although we have not found a 22q11 deletion in patients with transposition of the great arteries (unpublished data, 1996), one study reported 22q11 deletions in 4 of 32 patients with transposition of the great arteries.26 Another study found that a high proportion of patients with tetralogy of Fallot and an absent pulmonary valve have a 22q11 deletion.27 Furthermore, investigators have shown that the majority of patients with conotruncal anomaly face syndrome, characterized by conotruncal defects, a typical facial appearance, nasal speech, and mental retardation, frequently have 22q11 deletions.28 29 An occasional patient with features of Noonan’s syndrome associated with DGS or VCFS has been reported to have a 22q11 deletion, although patients with classical Noonan’s syndrome generally do not have the deletion.30 31 Most recently, several patients with autosomal dominant Opitz G/BBB syndrome, characterized by hypertelorism, hypospadius, and laryngotracheoesophageal defects, have been described with a 22q11 deletion.32 Together, these studies demonstrate that the phenotype associated with a 22q11 deletion is highly variable and can be very subtle.
Diagnosis
Standard karyotypic analysis, even with high-resolution banding techniques, will only detect 10% to 20% of 22q11 deletions.11 Currently, FISH is the method of choice for microdeletion detection (Fig 1⇓). In the FISH assay, biotinylated test and control probes are hybridized to metaphase chromosomes. A control probe that hybridizes to the distal end of 22q can be used to confirm the presence of two copies of the long arm of chromosome 22. The test probe is one that maps to the proximal end of the DGCR, eg, D22S75 (N25). In a normal individual, two signals from each probe should be present. The absence of signal from the test probe on one homologue is consistent with a microdeletion of that locus. A diagnostic test can be completed within 3 days using either metaphase chromosomes prepared from lymphocytes or from cultured amniocytes for the purpose of prenatal diagnosis. Alternatively, polymerase chain reaction–based assays that use highly polymorphic short-tandem-repeat polymorphic markers have been developed. This technique identifies individuals who are homozygous at several contiguous markers and are thus likely to have a deletion but does not definitively confirm the presence of a deletion.
Detection of a 22q11 deletion by FISH. Metaphase chromosomes are hybridized with labeled test and control probes as described in the text. The control probe is used to identify the distal end of chromosome 22; the test probe is specific for the DGCR of chromosome 22. a, Metaphase chromosomes from a normal individual. Signals from both the test and control probe are seen on both copies of chromosome 22. b, Metaphase chromosomes from an individual with a 22q11 deletion. Signal from the control probe is seen on both homologues of chromosome 22. Signal from the test probe is seen only on chromosome 22. The chromosome with the deletion is indicated by an arrow.
Molecular Analysis of the Commonly Deleted Region
The first association of DGS with abnormalities of chromosome 22 came with the identification of patients with DGS who were noted to have an unbalanced translocation that resulted in the loss of the proximal long arm of chromosome 22 (pter-q11).9 10 Subsequently, cytogenetically visible interstitial deletions of chromosomal region 22q11 were recognized, further narrowing the critical region for the disease locus.11 In order to determine whether patients with DGS and a normal karyotype had a microdeletion, multiple DNA markers mapping to this region were developed. Patients with DGS who demonstrated either a normal karyotype or cytogenetically visible interstitial deletion were tested for the presence of one or two copies of each DNA marker by DNA dosage analysis.14 The presence of only one copy of a DNA marker is consistent with a deletion at that locus. In the initial study, all patients diagnosed with DGS with either a normal karyotype or visible deletion were found to be deleted for a common group of markers within 22q11. The deleted region in different patients was found to overlap, and the approximate size of the deleted region was established. Recent studies have been performed to determine whether more subtle variations in the size of the deletion correlate with the phenotype. However, even those patients with very mild features, such as typical facies and mild palatal abnormalities, have large deletions spanning ≈2 megabases of DNA.33 34 35 The proximal and distal boundaries of the deletion appear to cluster in the majority of patients (Fig 2⇓).36 37
Diagram delineating the map of DGCR. The region deleted in the majority of patients, represented by an open box, spans the deletion breakpoints (BPs) and includes the markers N25 and R32. The common deletion BPs appear to cluster in the shaded areas as pictured. The minimal critical region, pictured as the hatched box above the commonly deleted region, is defined at the telomeric end (Tel) by the unique unbalanced translocation (15;22); it includes the BP of the balanced ADU translocation (2;22) and shares the common deletion end point (shaded) at the centromeric end (Cen). The approximate map location of other unbalanced translocation BPs discussed in the text and additional anonymous and gene specific markers are noted.
In order to determine whether the presence of a single copy of one or multiple genes results in the DGS/VCFS phenotype, several strategies are being pursued. The first uses unique patients with either a subset or all of the features characteristic of DGS/VCFS who do not have the typical deletion but rather a smaller deletion or some other chromosomal rearrangement involving 22q11. Such patients should help to define a minimal critical region (MDGCR) within the commonly deleted region. The second approach is to identify all of the genes in the minimal critical region and/or the larger commonly deleted region and to investigate their potential role in the disease process on a gene-by-gene basis. Third, since it is possible that haploinsufficiency for several genes rather than a single gene may be responsible for the disease phenotype, experiments using mammalian models are being undertaken. Thus, “gene knockout” experiments in the murine model system should be of great utility to investigate the biological consequences of having only one copy of one or several genes.
Defining a Minimal Critical Region
To define a minimal critical region, detailed maps of 22q11 and of the commonly deleted region have been constructed by several investigators. In particular, Halford et al38 used the unique probe KI506 to isolate the cosmid sc11.1 and found that sc11.1 contains low-frequency-repeat elements that flank the proximal and distal portion of the DGCR. Most patients with a deletion are deleted for both repeated regions contained in this cosmid and the unique loci in between. The result is a deletion spanning at least 2 megabases. Driscoll et al14 used unique markers spanning 22q11 to define the DGCR and defined the commonly deleted region of overlap as inclusive of the marker N25 (D22S75) proximally and R32 (D22S259) distally (Fig 2⇑).
To help define the boundaries of a smaller or minimal critical region (MDGCR), several patients with features of DGS/VCFS and unique chromosomal rearrangements resulting in deletions of 22q11 have been used. In earlier studies, the cell lines GM05878 and GM00980 were used to define the distal or telomeric boundary (Fig 2⇑). GM05878 was established from the unaffected father of a patient with DGS. The father carries a balanced (10;22) translocation, whereas the affected proband inherited an unbalanced complement of chromosomes, resulting in monosomy of the centromeric portion of the DGCR (pter-q11.2). GM00980 was derived from a patient with VCFS and an unbalanced (11;22) translocation that results in a deletion of 22pter-22q11.2. The rearrangement breakpoints of these cell lines were positioned within the DGCR by FISH using members of an ordered array of cosmids from the DGCR to the cell lines (Fig 2⇑).33 39 More recently, Jaquez et al40 studied a patient with classic features of VCFS and an unbalanced (15;22) translocation that results in monosomy 22pter-q11.2. The (15;22) translocation breakpoint has been mapped proximal to that of GM00980, further narrowing the distal boundary of the MDGCR.
The centromeric boundary of the MDGCR is less clearly defined. The repeat containing cosmid sc11.1 has been used to define the most centromeric boundary of the commonly deleted region by some investigators.39 The unique marker N25, which is distal to the sc11.1 repeat, has been present in a single copy in all DGS/VCFS patients with a deletion thus far in our experience.14 Thus, the centromeric boundary of the MDGCR should map between the common deletion breakpoints and the unique marker N25 (Fig 2⇑). In addition, the centromeric boundary of the MDGCR has been presumed to include the breakpoint of the only known balanced translocation in a patient with DGS.41 This is the ADU (2;22) translocation, whose translocation breakpoint was hypothesized to disrupt a gene or control region critical to the development of the DGS phenotype.42 More recently, Levy et al43 described a patient (patient G) with an interstitial deletion whose proximal boundary is telomeric to the ADU (2;22) translocation breakpoint. Therefore, patient G appears to narrow the centromeric boundary of the MDGCR, excluding the ADU breakpoint region. Perhaps the ADU breakpoint affects the expression of downstream genes via a position effect. This hypothesis remains to be examined.
Identification of Genes
Multiple methods are being used to identify genes in both the MDGCR and the commonly deleted region (DGCR). Initially, much interest focused on the unique patient ADU with partial DGS and a balanced (2;22) translocation in hopes that the translocation disrupts a gene or control region critical to the development of the DGS phenotype. To date, it has been difficult to identify a complete gene disrupted by the translocation breakpoint. Complete cloning, sequencing, and analysis of the translocation breakpoint has been accomplished by Budarf et al.42 Two transcripts on opposite strands have been identified in the vicinity of the translocation breakpoint. One of these transcripts is expressed predominantly in the adult heart and skeletal muscle. The open reading frame of the second transcript is disrupted by the translocation. Efforts to clone the full-length cDNAs corresponding to these partial transcripts are under way. Further experiments will be required to determine whether one or both of these transcripts are biologically significant in this disease.
Several groups have recently cloned a gene (LAN, DGCR2, IDD) that is not disrupted by the (2;22) translocation but maps ≈10 kb telomeric of the breakpoint.43 44 45 Analysis of the cDNA sequence suggests a gene that is an integral membrane protein with possible ligand binding domains. Given that faulty neural crest cell migration is implicated in the pathophysiology of DGS, a potential adhesion receptor gene may be important. However, the gene is oriented 5′ to 3′ toward the (2;22) translocation, the coding region of this gene is not disrupted by the translocation, and mutations have not been identified in nondeleted patients with the DGS/VCFS phenotype. Thus, if this is a critical gene, one would have to hypothesize that the (2;22) translocation disrupts a control element 3′ of the coding region rather than within the open reading frame.
Investigators have also used multiple cloning techniques to develop a complete transcription map of the MDGCR. Lindsay et al15 constructed an ordered array of overlapping DNA fragments spanning the MDGCR. To identify the maximum number of genes in the region, they used a variety of approaches. They screened a human placenta–arrayed cDNA library, performed exon trapping, and used cDNA selection methodology with cloned DNA fragments mapping to the MDGCR. They identified six transcripts from a genomic region of ≈270 kb, two of which had been previously reported. Gong et al16 also used a cosmid contig covering the MDGCR to isolate region-specific cDNAs by cDNA selection. Minimally overlapping cosmids were used to screen cDNAs from several sources to create a cDNA sublibrary. The individual clones were grouped according to their location in the MDGCR and were organized into individual cDNA contigs or transcription units. Expression analysis of each transcription unit was completed. Where possible, complete open reading frames were cloned using 5′ and/or 3′ rapid amplification of cDNA ends (5′ and 3′ RACE). The genomic structure and direction of transcription were derived by comparing the cDNA sequences with the genomic sequence of the MDGCR as it became available (see below). Using this approach, they identified 11 transcripts.
Large-scale sequencing and sequence analysis are also being used to identify genes within the MDGCR. A cosmid contig covering the MDGCR is presently being sequenced in its entirety (B. Roe, GenBank L77570, L77569, and unpublished data). The genomic sequence is being subjected to searches against multiple databases for either known genes or for homologous sequences in another species. The sequence is also being subjected to analysis by GRAIL, which will identify potential coding regions.46 These two methods (database searches and GRAIL analysis) can be used in conjunction to predict genes. In order to confirm whether the sequence predicted to be coding is an expressed gene, laboratory-based molecular assays are designed to detect expression by Northern blot analysis and/or reverse-transcription polymerase chain reaction. If there is evidence of expression, a cDNA library can be screened to isolate additional clones that contain or permit the assembly of the entire open reading frame. The genomic structure can be deduced by comparing the cDNA and genomic sequence. Using this methodology, we have identified several genes that are under further investigation for their potential role in this disorder. Among these genes is a putative mitochondrial transport protein.47
Many other expressed sequences have been mapped to the larger DGCR using a variety of approaches. Several genes, including catechol-O-methyltransferase, T10, LZTR-1, and glycoprotein Ibβ, map within the commonly deleted region but lie outside of the minimal critical region.48 49 50 51 A zinc finger gene, ZNF74, mapping to 22q11.2 has been found to be deleted in 23 of 24 patients with DGS.52 A putative transcriptional regulator (TUPLE1/HIRA) has also been identified within the commonly deleted region.39 53 Most recently, a second human clathrin heavy chain gene and a gene with homology to the Drosophila melanogaster gonadal protein and the laminin γ-1 chain have been identified.54 55 56 The potential role of these genes in the pathogenesis of the 22q11 haploinsufficiency syndrome remains to be fully evaluated.
Picking the Right Gene(s)
Since multiple expressed sequences have been identified, the challenge is to decipher which gene or genes are biologically responsible for the DGS phenotype. Presumably, the presence of only a single copy of the critical gene(s) results in a diminished dosage of the corresponding gene product at a critical moment in development. However, it is still not known whether the absence of a single copy of just one gene will be responsible for the entire phenotype or whether the diminished dosage of several gene products results in abnormal development. Given the phenotypic variability associated with a 22q11 deletion, modifying factors are also likely to play a role. Similarly, it is unknown whether these modifying factors will include genes mapping to the DGCR, status of the nondeleted allele, genetic background, or environmental influences, including the in utero environment.
In order to determine which genes in the DGCR are of significance to this phenotype, one might hypothesize that certain types of genes might be of particular interest, such as those likely to be developmentally regulated or those proposed to participate in neural crest migration. Ultimately, several lines of investigation will have to be pursued to determine which are the significant genes. First, the pattern of expression of each candidate gene during embryonic development should be determined in a mammalian model system. A candidate gene should be expressed at the right time and place. Second, targeted gene disruption experiments, or so-called gene “knockout” experiments, of both individual and multiple genes might mimic the apparent haploinsufficiency seen in the human. The mouse mammalian model system has been chosen by most laboratories, since early organogenesis resembles that of the human, and gene knockout experiments can be performed. The mouse homologues to the human genes mapping to the DGCR are being isolated. Evidence thus far indicates that the proximal portion of mouse chromosome 16 is syntenic to 22q11 and the DGCR and that this region is distinct from the distal portion of mouse 16, which is syntenic to other human chromosomes.57 58 59 Thus, knockout experiments targeting the proximal end of mouse chromosome 16 are likely to yield important information.
Several naturally occurring mouse mutants and several mutants resulting from gene knockout experiments have resulted in phenotypes that are reminiscent of the DGS/VCFS phenotype in humans. Most result in a phenotype consistent with disruption of neural crest cell development. For example, homozygous Splotch mutants with mutations of Pax3, as well as homozygous Patch mutants with deletion of the platelet-derived growth factor α-subunit receptor gene demonstrate multiple findings consistent with disruption of neural crest development, including truncus arteriosus.60 61 More recently, mouse pups homozygous for targeted disruption of endothelin-1 were found to have aortic arch anomalies and ventricular septal defects reminiscent of those seen in DGS and VCFS.62 These mouse mutants are not exact models of DGS or VCFS, given that the disrupted genes do not map to 22q11. However, they begin to reveal the complexity of the developmental pathways that are presumably disrupted in DGS/VCFS. They also identify gene products that potentially modify the DGS/VCFS phenotype or interact with disease-causing genes.
In addition to mouse models, genes critical to DGS/VCFS could be identified by screening nondeleted patients with the characteristic phenotype for mutations in candidate genes. As noted, nearly 90% of patients with DGS and 85% of patients with VCFS have a 22q11 microdeletion.63 Thus, some patients have the DGS/VCFS phenotype but do not have the 22q11 deletion by FISH analysis. Some of these patients might have a smaller deletion, a different genetic rearrangement, or even a point mutation in one of the genes mapping to the DGCR that is not detected by the standard FISH assay. Thus, such patients are being screened for mutations in the candidate genes. If a nucleotide alteration that is not found in normal individuals is detected in an affected patient, then the alteration most likely represents an inactivating point mutation. The identification of a mutation in a candidate gene from within the DGCR in a nondeleted affected individual would be good evidence that the gene is important. Of note, however, DGS is known to be etiologically heterogeneous. Therefore, some of the nondeleted patients are likely to have DGS on a different etiologic basis, such as deletion of chromosomal region 10p.64
Summary
The phenotype associated with a 22q11 deletion is highly variable and still under investigation. Of particular interest to cardiologists and cardiac developmental biologists is the finding that many patients with a 22q11 deletion have conotruncal cardiac defects and aortic arch anomalies. Despite the phenotypic variability, the vast majority of patients have a similar large deletion spanning ≈2 megabases. The low-frequency repeated sequences at either end of the commonly deleted region may be responsible for the size of the deletion and account for the instability of this chromosomal region. Molecular studies of patients with the DGS/VCFS phenotype and unique chromosomal rearrangements have allowed a minimal critical region for the disease to be defined. Multiple genes have been identified in the minimal critical and larger deleted region. These genes are being investigated for their potential role in the disease pathophysiology by screening for mutations in nondeleted patients with the phenotype and by analysis of the pattern of expression in the developing mouse embryo. Further experimentation in the mouse mammalian model system will be of great utility to help determine whether haploinsufficiency of one critical gene or several genes within the DGCR results in the disease phenotype. Modifying factors, both genetic and environmental, must also be considered. Further investigation into the disease mechanism leading to the DGS/VCFS phenotype will hopefully further our understanding of cardiac development and disease.
Selected Abbreviations and Acronyms
| CHD | = | congenital heart disease |
| DGCR | = | DiGeorge chromosomal region |
| DGS | = | DiGeorge syndrome |
| FISH | = | fluorescence in situ hybridization |
| GRAIL | = | Gene Recognition and Analysis Internet Link |
| MDGCR | = | minimal DiGeorge critical region |
| VCFS | = | velocardiofacial syndrome |
Acknowledgments
This study was supported by National Institutes of Health grants DC-02027 (Dr Emanuel), HL-51533 (Dr Emanuel), HD-26979 (Dr Emanuel), and HL-03191 (Dr Goldmuntz).
- Received August 21, 1996.
- Accepted December 18, 1996.
- © 1997 American Heart Association, Inc.
References
This Issue
Jump to
Article Tools
Share this Article
- Genetic Disorders of Cardiac Morphogenesis
