Although the overall patterns of anomalies and minor variations in development that are characteristic of chromosomal disorders are quite distinct, these conditions are notable for the variability of their phenotypes. For example, in Down syndrome (DS), which results from trisomy 21, none of the typical physical features is present in all affected individuals, and it is very unlikely that two persons would be phenotypically identical. In the case of the major congenital malformations known to be associated with DS, congenital heart disease occurs only half of the time, and duodenal atresia or stenosis in only 2.5% (Epstein, 2001).
Several possible explanations for this phenotypic variability in aneuploidy have been suggested (Epstein, 1988). One is that allelic differences in the genes that are triplicated might have different effects when present in three copies. For example, heterotrisomy for a gene (or genes) within a 9.6 cM minimal region (D21S167) on human chromosome 21 has been postulated to be a contributing factor to the pathogenesis of ventricular septal defects (VSDs) in DS, either through the presence of three different specific alleles or through the presence of specific combinations of alleles (Baptista et al., 2000). Second, the overall genetic background of the individual in whom the aneuploidy occurs could affect the penetrance and expression of different phenotypic features. Precedents for such background effects, often attributable to the existence of modifier loci, exist for many genetic traits in mice (Nadeau, 2001). Third, environmental factors, whether internal or external, that may influence developmental processes could affect the development of different components of the phenotype. Fourth, developmental processes are subject to random or stochastic processes that could have a similar influence. Finally, the altered expression of genes on the aneuploid chromosome could have a more global effect on the regulation or homeostasis of developmental systems, which could lead to phenotypic variability (Shapiro, 1997).
At the present time, it is not possible to distinguish among these possibilities in human aneuploidies, although it is likely that insight into genetic background effects in the future could be derived from genome-wide analyses. However, mouse models that have been developed for the study of DS provide a useful tool for looking at the effect of background genotype on specific components of an aneuploid phenotype. Mouse trisomy 16 (Ts16) was one of the first such models to be developed (Gropp et al., 1975; Cox et al., 1984).
Although Ts16 is no longer used as a model for DS, its robust phenotype makes it an attractive subject for the investigation of the effects of genetic background on phenotype. Here, we explore the effects of four different genetic backgrounds on the phenotype of Ts16 embryos and fetuses. Particular attention has been paid to development of the cardiovascular system, which is known to be seriously affected in Ts16. Our data demonstrate the complete penetrance of pharyngeal arch artery (PAA) anomalies and subsequent defects in the patterning of the great vessels when the maternal contribution is from the C3H/HeJ strain, but not when other genetic backgrounds are present. The strain-specificity of this feature makes the genetic contribution responsible amenable to mapping. Our data also demonstrate that defects in the patterning of the great vessels can arise at a significant frequency even in the absence of preexisting PAA anomalies, suggesting that there are at least two mechanistic responses to the presence of three copies of MMU16 in the modeling of the cardiovascular system.
Frequency of Trisomy
Although previous studies have shown the prevalence of Ts16 fetuses to be greater on a C3H/HeJ female background (21%) compared with the C57BL/6J female strain (11%; Epstein and Vekemans, 1990), we did not find the incidence of Ts16 to vary significantly with the genetic background of the female parent at E14.5 (P > 0.24; Fig. 1). Discrepancies in the incidence of Ts16 fetuses may be explained by the use of different male strains heterozygous for the Robertsonian translocation chromosome, Rb(16.17). Of interest, we also report a frequency of 20–25% Ts16 at embryonic day (E) 14.5, whereas Epstein and Vekemans found Ts16 frequencies ranging from 10 to 21%, depending on the parental cross. At E17.5, the frequency of Ts16 fetuses was lower in litters with a C3H/HeJ maternal contribution compared with the C57BL/6J, BALB/cJ, and DBA/2J females, but the difference was not significant. The frequency of postimplantation death was determined by dividing the number of moles (deciduomata) and late deaths by the number of implantation sites. Again, no significant differences in the frequencies of moles and fetal deaths were detected at E14.5 (P > 0.05). However, there did appear to be a higher frequency of fetal loss at E17.5 in the C3H/HeJ females (62%) compared with the C57BL/6J (47%), BALB/cJ (41%), DBA/2J (31%) females; but again, these differences did not attain significance (P > 0.05; data not shown).
Among the four strains of Ts16 that were generated, there were no significant differences in the ratios of trisomic/euploid fetal weights and lengths at E14.5 and E17.5 (Fig. 2). Although Ts16 weight was reduced by E17.5 compared with control littermates in all four strains, this reduction was not significant. There was, however, a significant difference in Ts16 placental weight relative to euploid controls among the four strains, with DBA/2J placentas weighing significantly less than C3H/HeJ and C57BL/6J (data not shown).
All Ts16 fetuses at E14.5 had mild to severe edema, an increased volume of amniotic fluid, and a thickened neck. By E17.5, the trisomic fetuses exhibited two distinct phenotypes, one with an edematous morphology and the other runt-like (Fig. 3A), but all had open eyelids and displayed placental defects similar to those found at E14.5. There was gross disorganization of the placenta with an increased area of spongiotrophoblast invading the labyrinth, an arrangement that could effectively reduce exchange between the maternal and fetal blood by as much as two thirds (manuscript in preparation). The majority of trisomic fetuses on all backgrounds had malformations of the spinal column and alterations of the vertebral arches that were often accompanied by rib anomalies. The bones of the limbs showed delayed ossification, but there were no malformations (Fig. 3B–D).
Analysis of PAAs at E10.5
We examined the development of the PAAs in a total of 213 E10.5 embryos, 45 (21%) of which were trisomic. At least eight trisomics of each strain were examined. The PAAs were visualized by intracardiac India ink injection and scored as “present” or “absent” (Table 1). All Ts16 embryos derived from C57BL/6J, BALB/cJ, and DBA/2J females, as well as euploid control littermates, displayed normal development of the PAAs (Fig. 4A). However, PAA anomalies were detected in all embryos on the C3H/HeJ maternal background and typically involved abnormal development of at least one of the fourth or sixth PAAs (Fig. 4B,C). Occasionally, there was severe disruption of the entire PAA structure, with absence of the third, fourth, and sixth PAAs and direct connection of the dorsal aorta to the aortic sac (Fig. 4D).
Table 1. Summary of PAA Anatomy in Ts16 Embryos on Different Genetic Backgrounds
PAA anomalies were observed on the left and/or right side of the embryo.
Hearts from five Ts16 fetuses and five euploid controls on each of the four inbred genetic backgrounds were sectioned and examined for cardiac anomalies at E17.5. VSDs were present in all Ts16 fetuses, regardless of genetic background (Fig. 5B). Other defects observed were double outlet right ventricle (DORV), common atrioventricular (AV) canal, and, more rarely, transposition of the great arteries (TGA) and persistent truncus arteriosus (PTA; Table 2). These defects are similar to those described in previous reports (Miyabara et al., 1982; Miyabara, 1990; Buselmaier et al., 1991; Waller et al., 2000). Although variation in the incidence of distinct cardiovascular malformations in Ts16 mice with different genetic backgrounds has been described (Miyabara et al., 1984), we were unable to determine whether there were statistically significant differences in frequency among strains because of the small sample size. No cardiac anomalies were observed in any of the 20 euploid control fetuses (Fig. 5A).
Table 2. Histological Analysis of Cardiovascular Anomalies in Ts16 at E17.5
Usually more than one anomaly was observed. AV, atrioventricular; DORV, double outlet right ventricle; E, embryonic day; PTA, persistent truncus arteriosus; TGA, transposition of the great arteries; VSD, ventricular septal defect.
Common AV canal
Thymic dysplasia was observed in all fetuses with cardiac anomalies (Fig. 5C,D). In most cases the lobes were drop-shaped and occasionally asymmetric.
Analysis of Defects of the Great Vessels at E17.5
Corrosion casts of Ts16 and euploid control hearts at E17.5 were prepared. The size (relative to body size) and shape of the heart were normal in fetuses recognized to be trisomic, regardless of genetic background (data not shown). Because the abnormalities detected in the E10.5 trisomics were exclusively on the C3H/HeJ background, we first examined the E17.5 C3H/HeJ corrosion cast preparations and found that 100% (10/10) of Ts16 fetuses had abnormalities of the patterning of the great vessels that could be attributable to inappropriate persistence or regression of the fourth and sixth aortic arch arteries. Examination of the corrosion casts revealed that two fetuses had pulmonary atresia (Fig. 6B) with the pulmonary vasculature filling retrograde by means of the patent ductus arteriosus. Two other fetuses had interruption of the aortic arch (IAA), one each of type B and type C. IAA, type B, occurs between the left carotid artery and the left subclavian artery and is caused by interruption of the fourth aortic arch. With this type of IAA, the transverse arch of the aorta is not fully developed and is separated into two, unconnected parts where a single arch would normally be formed. With IAA type C, the interruption occurs between the innominate artery and the left common carotid artery (Fig. 6C).
Two fetuses had the right subclavian artery (RSA) originating from the descending aorta. Normally, the RSA arises from the innominate artery, which bifurcates into the right common carotid artery and the RSA. Another fetus displayed a right aortic arch and anomalous origin of the left subclavian artery (LSA) from a left-sided ductus arteriosus (Fig. 6D). Because the right-side aortic arch travels posterior to the esophagus and the trachea, a vascular ring was present with the potential for compressing these structures. Other defects included severe coarctation of the aorta (Fig. 6E), as well as a truncus with right aortic arch and absent ductus arteriosus (regression of the distal sixth arch; Fig. 6F).
Corrosion casts were prepared for each of the other maternal strains (n = 2). However, because the easy passage of methyl methacrylate across the ventricular septum through a VSD or common AV canal (see above) made the production of informative corrosion casts problematic, we also examined the patterning of the great vessels in additional fixed fetuses that were prepared for subsequent histological analysis. All euploid littermates displayed normal development of the heart, aortic arch, and associated vessels (Fig. 7A). A VSD and/or a common AV canal was present in all Ts16 fetuses. Moreover, in addition to being present in Ts16 fetuses with a C3H/HeJ background, aortic arch defects were also observed in trisomic fetuses with a maternal contribution from C57BL/6J (n = 5), BALB/cJ (n = 4), or DBA/2J (n = 5), which, unlike the C3H/HeJ trisomics, did not have preexisting PAA defects at E10.5.
Because the analysis of both the corrosion casts and fixed fetuses provided comparable data, the data on the vascular malformations associated with Ts16 at E17.5 observed by both methods have been combined in Table 3. All trisomic fetuses with a C3H/HeJ maternal contribution were abnormal (n = 13), as were 85% of those on the other three backgrounds. The most common anomaly was interruption of the aortic arch (type B; Fig. 7B). A right aortic arch with the expected inverted order of arch branches was also observed (Fig. 7C).
Table 3. Frequency of Cardiovascular Anomalies in Ts16a
Type of anomaly
RSA, right subclavian artery; IAA, interrupted aortic arch; DA, ductus arteriosis; PTA, persistent truncus arteriosus.
Number of trisomics in the progeny selected for examination following fixation and production of corrosion casts.
The relationship between genetic background and the incidence and/or variability of a specific phenotype is well documented (Houlston and Tomlinson, 1998; Nadeau, 2001, 2003; Zlotogora, 2003). Previous investigations by Miyabara and collaborators have examined the effect of genetic background on the frequency and nature of cardiovascular malformations in Ts16 mouse fetuses (Miyabara, 1990; Miyabara et al., 1982, 1984). These studies suggest a relationship between genetic background and differences in the frequency of cardiovascular anomalies; however, the interpretation of the results can be complicated by the use of different types of doubly heterozygous males in comparative studies, introducing sources of variability that preclude the assignment of a genetic component. In fact, when different doubly heterozygous males were crossed to an inbred female strain, C57BL/6, significant differences in the frequency of a variety of cardiovascular anomalies were observed, presumably attributable to a paternal genetic component (Miyabara, 1990).
In the studies reported here, a single strain of doubly heterozygous males was crossed with four different inbred strains of females. This strategy permitted the striking all-vs.-none difference that we observed in the presence of PAA defects at E10.5 to be attributed specifically to an allele or alleles present in the C3H/HeJ strain and not in C57BL/6J, BALB/cJ, or DBA/2J. The nature of the responsible gene(s) is unknown at this time, but it could be amenable to mapping and identified by techniques currently being used for the isolation of modifier loci.
The one other mouse aneuploidy in which the effect of genetic background on phenotype have been investigated is the Df1 deletion in proximal mouse chromosome 16, which is a model for the 22q11 deletion that causes the DiGeorge and related syndromes in man. The Df1 deletion results in conotruncal and aortic arch anomalies similar to those found in Ts16, and their penetrance ranges from 16 to 50%, depending on background (Taddei et al., 2001). In contrast to our findings with Ts16, the genetic background did not affect the frequency of PAA defects, in this instance involving the 4th arches, that occurred in all E10.5 embryos regardless of genotype. Rather, the background appears to have affected the degree of “recovery” of the PAAs from these impairments in development (Lindsay and Baldini, 2001; Taddei et al., 2001). Therefore, quite different mechanisms appear to be at work in Ts16 and Df1.
The PAA system is transient and initially involves a symmetrical array of vessels that regress or are extensively remodeled into adult vascular structures, including the subclavian arteries and arch of the aorta. During gestation, the PAAs link the heart with paired dorsal aortae and develop symmetrically until approximately E11.5. The PAAs then undergo a dramatic remodeling process that requires the presence of neural crest cells to establish the embryonic asymmetric circulatory system (reviews Creazzo et al., 1998; Sucov, 1998; Epstein and Buck, 2000). The aberrant persistence and/or regression of different components of the PAA system is the presumed mechanism leading to specific congenital vascular malformations (Momma et al., 1999). For example, IAA, type B, is caused by the absence or regression of the 4th aortic arch.
In our studies at E10.5, we found that fully penetrant PAA abnormalities ranging from a reduction in size to the complete absence of the 4th and 6th PAAs, either unilaterally or bilaterally, occurred on a C3H/HeJ background. These findings can explain the defects in the patterning of the aortic arch and great vessels that subsequently occur at E17.5 (Momma et al., 1999). However, our data also demonstrate that cardiovascular anomalies can occur without preexisting defects of the PAAs, at least before E11, implicating additional mechanisms in the maldevelopment of the aortic arch and associated vessels. It has been suggested, for example, that VSDs may play a role in aortic arch defects by altering hemodynamic parameters and causing the aortic arch and pulmonary segments to compete for priority, with one segment being reabsorbed as a redundant feature. This phenomenon has been referred to as the “arch-rival” hypothesis (Buselmaier et al., 1991). However, although all fetuses with a female genetic component from C57BL/6J, BALB/cJ, or DBA/2J had a physiological anomaly, VSD and/or common AV canal, only 85% manifested aortic arch defects. Therefore, if this hypothesis is correct, other factors operative in development, such as stochastic events, may affect the ultimate phenotypic outcome.
Cardiac neural crest anomalies have been described in Ts16 (Webb et al., 1996; Waller et al., 2000) and other mouse mutants (Conway et al., 1997; Kochilas et al., 2002; Ohnemus et al., 2002) in association with outflow tract and aortic arch abnormalities. Immunohistochemical studies on serially sectioned hearts from normal and Ts16 mouse embryos, ranging from E8.5 to E14.5, strongly suggest that abnormal neural crest cell behavior is involved in the pathogenesis of the conotruncal malformations in the Ts16 mouse (Waller et al., 2000). Based on the PAA defects at E10.5 and the frequent involvement of the thymus gland, our results support the neural crest hypothesis. Although the significance of the similarities in cardiovascular defects between Df1 and Ts16 remains unclear, evidence suggests that the migration and distribution of the neural crest cells in early embryonic development are critical pathways that are sensitive to gene dosage effects.
Production of Ts16 Embryos and Fetuses
Rb(6.16)24Lub and Rb(16.17)7Bnr mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and crossed to generate [Rb(6.16)24Lub x Rb(16.17)7Bnr]F1 stocks heterozygous for the two Robertsonian translocation chromosomes, which share monobrachial homology for MMU16. F1 males were naturally mated with females from four inbred strains, BALB/cJ, C3H/HeJ, C57BL/6J, and DBA/2J, obtained from the Jackson Laboratory. We used the same four F1 males throughout the accumulation of embryos for PAA analysis. Females were checked daily for the presence of a vaginal plug, and the day on which it was found was designated 0.5 day of gestation. Females from each inbred strain were housed with the four F1 males for a 1-week period (plugged females were removed daily and replaced with a virgin female) for a period of 4 weeks (1 week/strain). This procedure was repeated for another 4-week cycle. Preliminary data indicating strain-specific PAA anomalies on the C3H/HeJ background prompted us to breed more C3H/HeJ females to accumulate embryos in excess of the other three inbred strains. Husbandry records demonstrated that each F1 male produced both Ts16 and 2N control embryos from each inbred female strain. Pregnant females were killed, and embryos/fetuses were collected at E10.5, E14.5, and E17.5. Karyotype analysis of cultured tail fibroblasts performed by conventional methods was used to determine the genotype of each animal. Whereas euploid embryos/fetuses have 38 acrocentric chromosomes and one metacentric (Robertsonian translocation) chromosome (total of 40 arms), Ts16 embryos/fetuses have 37 acrocentric and two metacentric chromosomes (total of 41 arms).
Fetuses were formalin-fixed and embedded in paraffin using standard protocols. Serial sections were cut at 5 μm and stained with hematoxylin and eosin for light microscopy. Comparative morphometric analyses were performed on equivalent sections through the heart and thymus. Digital images were captured and analyzed by using Spot Software version 2.1 (Diagnostic Instruments, Inc.).
Ink Injection and Corrosion Casts
For India ink injections, embryos collected at E10.5 were injected intracardially with India ink (diluted 1:20 in H2O), using a finely drawn glass pipette, and were then immediately fixed in 95% ethanol:1% acetic acid:1% chloroform. The embryos were dehydrated overnight and cleared in 1:1 benzyl benzoate:methyl salicylate. The aortic arches were scored as present or absent and, if present, for relative thickness.
To identify vascular patterning defects, pregnant females were anesthetized and the fetuses removed at E17.5. The fetal hearts were exposed by a thoracic incision and rib removal. Methyl methacrylate (LADD Research Industries, Inc.) was injected into the right and left ventricles until the entire fetal vasculature was filled. After allowing the resin to harden overnight, the fetuses were placed in maceration solution (Polysciences, Inc.) for 2–3 days to dissolve the soft tissues, leaving a mold of the vasculature. Photographs were digitally captured and processed with Adobe Photoshop software for Mac.
Because of the difficulties involved in producing corrosion casts in fetuses with VSDs, we also examined the heart and associated great vessels in fixed fetuses at E17.5. The thymus was also inspected. Subsequently, the fetuses were serial sectioned and stained with hematoxylin and eosin to expose VSDs.
Fetuses were collected at E17.5 and fixed in 80% ethanol for a minimum of 24 hr before removal of the skin and viscera. The carcasses were dehydrated and stained with Alcian blue for cartilage and with Alizarin red for calcified tissue according to standard procedures (McLeod, 1980). Euploid and Ts16 skeletal preparations in 100% glycerin were viewed in uncovered 60-mm Petri dishes using a Zeiss dissecting microscope. Digital images were captured by using Spot Software v2.1 and processed with Photoshop software.
The frequency of live trisomy fetuses was recorded at E14.5 and E17.5. The weight and head-to-rump length of Ts16 embryos were compared with their euploid littermates using analysis of variance for parametric data and the Kruskal–Wallis test for nonparametric data, with the litter as the unit of comparison. Ninety-five percent confidence intervals for nonparametric data were calculated by using Vassarstat's Web site for statistical computation at http://faculty.vassar.edu/lowry/VassarStats.html. Strains were compared pairwise by using the Tukey t-test for multiple parametric comparisons and the Mann–Whitney test for multiple nonparametric comparisons.
We thank R. Gacayan and M. Doan for excellent animal care. Intracardiac ink injection and corrosion cast protocols were kindly provided by Anne Moon.