A Molecular Pathway Revealing a Genetic Basis for Human Cardiac and Craniofacial Defects

See allHide authors and affiliations

Science  19 Feb 1999:
Vol. 283, Issue 5405, pp. 1158-1161
DOI: 10.1126/science.283.5405.1158


Microdeletions of chromosome 22q11 are the most common genetic defects associated with cardiac and craniofacial anomalies in humans. A screen for mouse genes dependent on dHAND, a transcription factor implicated in neural crest development, identified Ufd1, which maps to human 22q11 and encodes a protein involved in degradation of ubiquitinated proteins. Mouse Ufd1 was specifically expressed in most tissues affected in patients with 22q11 deletion syndrome. The human UFD1L gene was deleted in all 182 patients studied with 22q11 deletion, and a smaller deletion of approximately 20 kilobases that removed exons 1 to 3 ofUFD1L was found in one individual with features typical of 22q11 deletion syndrome. These data suggest that UFD1Lhaploinsufficiency contributes to the congenital heart and craniofacial defects seen in 22q11 deletion.

Congenital heart defects (CHDs) are the most common of all human birth defects and are the leading cause of death in the first year of life (1). CHDs involving the outflow tract of the heart and the vessels arising from it are due to abnormal development of neural crest–derived cells that populate the heart (2, 3). The branchial arches, which give rise to craniofacial bones, the thymus, and the parathyroid glands are also populated by neural crest cells (3). Some 90% of individuals with cardiac and craniofacial defects [DiGeorge, velo-cardio-facial (VCFS), and conotruncal anomaly face syndromes (CAFS)] have monoallelic microdeletion of chromosome 22q11.2 (4).

Among heart defects, 22q11 deletions are found in 50% of patients with interruption of the aortic arch, 30% with persistent truncus arteriosus (failure of septation of aorta and pulmonary arteries), and 15% with tetralogy of Fallot (malalignment of aorta and pulmonary artery with ventricles) (5). Such cardiac defects are common after neural crest ablation in chick embryos (6), suggesting that one or more genes regulating neural crest cells may map to 22q11. A region 2.0 megabases in length is most commonly deleted and is called the DiGeorge Critical Region (DGCR). Extensive mapping, positional cloning, and sequencing of the human and syntenic mouse DGCR have been performed (7, 8); however, mutation analyses of candidate genes in humans and deletion studies in mice have failed to identify any genes responsible for the 22q11 deletion syndrome (7).

The basic helix-loop-helix transcription factor dHAND is required for survival of cells in the neural crest–derived branchial and aortic arch arteries and the right ventricle (9–11). Mice lacking endothelin-1 (ET-1) have cardiac and cranial neural crest defects typical of 22q11 deletion syndrome and display down-regulation of dHAND (11, 12), suggesting that a molecular pathway involving dHAND may be disrupted in this syndrome. The genes for dHAND, ET-1, or the ET-1 receptor, however, do not map to 22q11 in humans (11, 12).

To investigate the potential mechanisms through which dHAND might function, we identified dHAND-dependent genes by suppressive-subtractive hybridization (13), a procedure that yielded genes expressed in wild-type but not dHAND mutant embryos at embryonic day 9.5 (E9.5). One of the dHAND-dependent genes (Ufd1) was the mouse homolog of a yeast gene involved in degradation of ubiquitinated proteins (14). Ufd1 is essential for cell survival in yeast (14) and is highly conserved from yeast to humans (15). Quantitative reverse transcriptase–polymerase chain reaction (RT-PCR) (16) confirmed that Ufd1 was down-regulated in dHAND-null hearts at E9.5 (Fig. 1A).

Figure 1

dHAND-dependent expression of Ufd1 and localization of humanUFD1L to the DGCR of 22q11. (A) Down-regulation of Ufd1 transcripts in dHANDmutants was detected by subtraction cloning and confirmed by quantitative RT-PCR after 31 and 40 cycles of amplification. Complementary DNA in wild-type (+/+) and dHAND mutants (–/–) was comparable as demonstrated by glyceraldehyde 3-phosphate dehydrogenase (G3PDH) mRNA amplification. Positive (+) and negative (–) controls are shown. (B) The position of humanUFD1L in the 2-Mb DGCR is shown relative to neighboring genes. Arrows indicate direction of transcription. FISH analysis of a normal individual (D) and patient with 22q11 deletion (C) using a UFD1L probe. The patient (C) has only one allele of UFD1L, seen in blue (white arrows). Chromosome 22 was labeled at 22q13.3 with a red fluorescent marker (yellow arrows). Bars, 2.5 μm.

This finding that Ufd1 was down-regulated indHAND mutants was intriguing because human UFD1Lis located in the DGCR (Fig. 1B) and was shown to be deleted in 13 patients with 22q11 deletion (15). To determine the frequency of UFD1L deletion, we studied 182 patients with 22q11 deletion, as detected by fluorescence in situ hybridization (FISH) (17). Our FISH analysis revealedUFD1L deletion in 182 of 182 patients (Fig. 1, C and D).

To determine if Ufd1, like dHAND, might play a role in cardiac and cranial neural crest development, we examined the embryonic expression pattern of mouse Ufd1 (18). Ufd1 was expressed in the first through fourth branchial arches, abnormalities of which are the basis of much of 22q11 deletion syndrome, with enrichment in the tips of the branchial arches, similar to dHAND expression (Fig. 2, A, B, and G).Ufd1 expression in the palatal precursors and frontonasal region was prominent (Fig. 2, B, D, and G), an important finding because cleft palate and facial anomalies are common features of 22q11 deletion syndrome. The developing limb bud, another site ofdHAND expression, revealed Ufd1 expression in a pattern similar to dHAND at E10.5 and E12.5 (Fig. 2, D and F). Ufd1 expression was also detected in the developing ear (otocyst) as previously reported (Fig. 2, A and F) (15). Within the brain, Ufd1 was expressed with marked specificity in the medial telencephalon that forms the hippocampus (Fig. 2E). A role for Ufd1 in the hippocampus, which is involved in long-term memory, would be consistent with the learning impairment that often accompanies 22q11 deletion.

Figure 2

Expression of Ufd1 in mouse embryos, detected by in situ hybridization. (A) Expression ofUfd1 in the branchial arches (ba), limb bud (lb), and otocyst (ot) was evident at E9.5 in a lateral view. Arrowheads mark first through fourth branchial arches. (B) Clearing of the embryo revealed expression of Ufd1 in the fourth aortic arch artery (aa), the branchial arch, and fronto-nasal (fn) area. (C) Ufd1 was down-regulated in the branchial and aortic arches of dHAND mutants, although expression was detected in the otocyst and limb bud. (D) At E10.5,Ufd1 expression was seen in the branchial arches, palatal (p) precursors, and limb buds. (E) Frontal view of an E12.5 embryo revealed Ufd1 transcripts in the medial telencephalon (t) and in the hair follicles (hf) of the vibrissae (whiskers). (F) Lateral view demonstrated expression in the developing limb and ear (e). (G) Histologic analysis revealedUfd1 in the distal mesenchyme of an E10.5 branchial arch and in the palatal mesenchyme. (H) Sagittal section showed expression of Ufd1 in the mesenchyme (m) of the proximal aorta (ao). (I) E10.5 sagittal section through the heart revealed Ufd1 transcripts in the conotruncus (ct), fourth aortic arch artery, and the branchial arch. Bars, 500 μm. h, head; ht, heart.

Within the heart, neural crest–derived cells are required for septation of the cardiac outflow tract into the aorta and pulmonary artery (19) and for remodeling of the bilaterally symmetric aortic arch arteries that form the mature aortic arch and proximal pulmonary arteries (20). Ufd1 transcripts were detected in the conotruncus (cardiac outflow tract) just as neural crest cells condensed and underwent ecto-mesenchymal transformation (Fig. 2I). At E9.5 to E10.0, Ufd1 expression was most evident in the fourth aortic arch artery (Fig. 2, B and I), which is responsible for formation of the segment of the aortic arch that lies between the left carotid and subclavian arteries (Fig. 3E). This segment does not form in interrupted aortic arch type B, one of the most common cardiac defects associated with 22q11 deletion (Fig. 3F). Vascular mesenchymal cells surrounding the proximal aorta also express Ufd1 (Fig. 2H), similar to dHAND. In dHAND-null embryos,Ufd1 was down-regulated in the branchial arches and conotruncus, but was detected at lower levels in the limb bud and was not affected in the developing ear (Fig. 2C). The expression ofUfd1 in numerous tissues affected in 22q11 deletion syndrome and its involvement in a molecular pathway regulating neural crest development suggest that Ufd1 may play a role in many features of this disease.

Figure 3

Deletion of UFD1L and clinical features of patient JF. (A) Southern blot analysis of Hind III–digested genomic DNA from patient JF, hybridized to a UFD1L cDNA probe, revealed a 58% decrease in intensity of a 12.3-kb DNA band compared to her mother (M), father (F), and an unrelated control (C). This band corresponds to exons 1 to 3 of UFD1L and is deleted from one allele of UFD1L. (B) Northern analysis of thymic RNA revealed diminished UFD1L mRNA transcripts (1.1 kb) in JF compared to control (C). (C) Levels of mouse cdc45 transcripts, assessed by RT-PCR, were comparable in wild-type (+/+) and dHAND-null (–/–) hearts. (D) Genomic organization of UFD1L andCDC45 with Hind III restriction sites (black vertical bars) and expected DNA fragment sizes (brackets at bottom, in kilobases) encompassing exons of UFD1L (open boxes) andCDC45 (shaded boxes) are shown. DNA breakpoints of the deletion in patient JF are shown with vertical arrows. (Eand F) Ultrasound image of JF demonstrated interrupted aortic arch (IAA) type B (F) in comparison to image of a normal aortic arch (E). In JF, the ascending aorta (aa) and descending aorta (da) were discontinuous between the left carotid (lc) and subclavian artery (sca), marked by an asterisk. A patent ductus arteriosus (pda) supplied blood to the descending aorta during fetal and perinatal life. Cartoons depict the heart anatomy, including persistent truncus arteriosus (PTA), with vessels shown in black. Bars in (E) and (F), 1 cm.

To determine if UFD1L haploinsufficiency is responsible for part of the 22q11 deletion phenotype, we searched for UFD1L deletions in individuals with cardiac and craniofacial defects who did not have detectable 22q11 deletions (21). Southern analysis of genomic DNA hybridized with a UFD1L cDNA probe revealed one individual (JF) with monoallelic deletion of exons 1 to 3 of UFD1L (Fig. 3, A and D), leaving exons 4 to 12 intact. Deletion of exons 1 to 3 was not detected in the parents, both phenotypically normal, suggesting that the UFD1Ldeletion in JF occurred de novo. Deletion of UFD1L was not detected in 100 normal unrelated individuals. As expected, expression of UFD1L mRNA in the thymus of patient JF (22) was diminished compared to controls without cardiac neural crest defects (Fig. 3B), confirming haploinsufficiency associated with this deletion.

The phenotype of patient JF encompassed nearly all features commonly associated with the 2-Mb 22q11 deletion. Four days after birth, she was diagnosed with interrupted aortic arch (Fig. 3F), persistent truncus arteriosus, cleft palate, small mouth, low-set ears, broad nasal bridge, neonatal hypocalcemia, T lymphocyte deficiency, and syndactyly of her toes. Although other chromosome deletions can also result in a similar phenotype (23), the genotype of JF and theUFD1L expression pattern suggest that UFD1Lhaploinsufficiency can contribute to many features observed in 22q11 deletion syndrome.

There is substantial variability in the phenotype associated with 22q11 deletions. Thus, it is possible that other genes in this region, distant modifier genes, or environmental factors could contribute to distinct features of 22q11 deletion syndrome. HIRA, the human homolog of a yeast histone regulatory gene (24), is expressed in the cardiac neural crest and maps 50 kb centromeric ofUFD1L. However, HIRA was not deleted and was expressed normally in the thymus of patient JF (25).CDC45, the human homolog of a yeast cell cycle protein, is immediately telomeric to UFD1L (26) and was used to define the 5′ breakpoint of the deletion in patient JF (21,25) to the region between exons 5 and 6 of Cdc45 (Fig. 3D). Although the ubiquitous nature of Cdc45 expression (27) and its normal expression in dHAND mutants (Fig. 3C) (16) make it an unlikely candidate gene for the 22q11 deletion syndrome, it is conceivable that the deletion ofCDC45 may act as a modifier of patient JF's phenotype.

Our results support a role for ubiquitin-mediated mechanisms in controlling neural crest development. Ubiquitin-specific proteases are essential for regulating numerous critical cellular pathways, including those involving p53-related cell survival and NF-κB (nuclear factor–κB) activity (28). In vitro overexpression or inhibition of ubiquitin-specific proteases results in programmed cell death, indicating that their activity is dose-dependent (29). Yeast lacking Ufd1 exhibit a cell survival defect that is incompletely rescued by one allele of Ufd1(14), consistent with the notion that haploinsufficiency of UFD1L may contribute to the phenotype seen in 22q11 deletion syndrome. We speculate that UFD1Lhaploinsufficiency leads to accumulation of certain proteins and defective survival of cardiac and cranial neural crest cells, resulting in premature thymic apoptosis and loss of cells that contribute to the transverse aortic arch, palate, and craniofacial structures. Further mutation analysis of UFD1L in humans and elucidation of the cellular pathways regulated by UFD1L may provide new directions in understanding basic mechanisms of neural crest development and congenital cardiac and craniofacial defects.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: dsriva{at}


View Abstract

Navigate This Article