Compartment-Selective Sensitivity of Cardiovascular Morphogenesis to Combinations of Retinoic Acid Receptor Gene Mutations
Abstract Several aspects of normal cardiovascular development require signaling by the vitamin A metabolite retinoic acid. We have previously established germ-line mutations in mice in the genes that encode the RARα1, RARβ, and RXRα retinoic acid receptors as a means of studying the function of these receptors in vivo. Although mutation of RXRα results in fetal ventricular defects, the RARα1 and RARβ mutations are apparently nonphenotypic in the heart and elsewhere. In this study, we have established and analyzed combinations of these receptor gene mutations. Malformations of the ventricular chamber (chamber hypoplasia and muscular ventricular septal defects), conotruncus (double-outlet right ventricle, transposition, and membranous ventricular septal defects), aortic sac (persistent truncus arteriosus and aorticopulmonary window), and aortic arch–derived arteries were recovered in various combinations of the RARα1, RARβ, and RXRα gene mutations. Depending on the combination of receptor mutations, selective defects were obtained in specific cardiovascular compartments, suggestive of differential expression or function of each receptor within domains of the developing heart.
The vitamin A derivative RA has a critical role in multiple aspects of cardiovascular morphogenesis and in many other aspects of embryonic development. Historically, the biological role of RA in animals has been inferred from the numerous phenotypes that emerge after experimental perturbation of RA status. A broad spectrum of cardiovascular malformations (described below) are prominent consequences of vitamin A deficiency1 2 and vitamin A or RA excess exposure3 4 5 6 in mammals, with a comparable set of malformations observed in avian embryos as well.7 8
The biological effects of RA are mediated by receptors of the nuclear receptor family, which function in the nucleus as ligand-dependent transcription factors.9 There are two subfamilies of RA receptors, the RARs and the RXRs, each with three members: α, β, and γ.10 A fairly sophisticated level of understanding of the molecular biochemistry of RA transcriptional activation has been uncovered over the last several years and has recently been reviewed.11 12 13
As a means to better understand the role of RA and the RA receptors, we and others have initiated a genetic analysis of RA receptor gene function by establishing mutations in these genes in the mouse germ line. Our mutation in the RARα gene14 eliminates the major isoform of this gene, the RARα1 isoform, but leaves intact a second product (RARα2). Our mutations in the RARβ15 and RXRα16 genes completely eliminate all products of these genes. In addition, we have recently generated a complete loss-of-function mutation in the RXRγ gene (M. Chiang, H.M. Sucov, and R.M. Evans, unpublished observations, 1997), and both isoform-specific and complete gene mutations have been established by Kastner et al.17
When these mutations are individually bred to homozygosity, either no apparent consequence is observed (RARα1, RARβ, RARγ2, and RXRγ), or the phenotypes that result are confined to a relatively small number of tissues (RARα, RARγ, RXRα, and RXRβ). In the developing heart, absence of RXRα compromises ventricular morphogenesis (described below) but does not account for many of the known cardiovascular deficiency or excess phenotypes, and no other single RA receptor gene mutation has any cardiovascular developmental consequence. By combining RA receptor gene mutations,17 18 19 20 21 extensive defects are seen throughout the embryo. This suggests that the receptor genes are overlapping in expression and that the receptor proteins are at least partially interchangeable in function, as has been proposed for other multigene families.22 However, an important result to emerge from these studies is the variable frequency at which different malformations are observed with different combinations of RA receptor gene mutations. The clear implication is that each developmental process, in the heart and elsewhere, is differentially sensitive to genetic manipulation.
The long-term goal of these genetic approaches is to explore the cellular and molecular basis of the observed pathologies and of the corresponding aspects of normal development. From an experimental standpoint, different combinations of receptor mutations might prove to be more or less advantageous, in terms of penetrance of phenotypes and the presence of additional copresenting malformations. In the present study, we have established all possible double combinations of the RARα1, RARβ, and RXRα mutations and have analyzed in detail the resultant defects in cardiovascular development.
Materials and Methods
Derivation and Analysis of Double-Mutant Combinations
Mice that are homozygous for the RARα1 and RARβ mutations are normal and have been maintained as homozygous stocks on a mixed genetic background in our colony. The RXRα mutation is embryonic lethal when homozygous and has been maintained as a heterozygous line of adults. Crosses were established that yielded normal and healthy adults of the following genotypes: [RARα1−/−, RARβ−/+], [RARα1−/+, RARβ−/−], [RARα1−/−, RXRα−/+], and [RARβ−/−, RXRα−/+], as well as all combinations of double heterozygotes. Appropriate animals were crossed, and embryos were isolated at E14.5, paraffin-embedded, horizontally sectioned, and stained with hematoxylin/eosin. Serial sections were taken from a level above the thymus to below the bottom of the heart, and in some cases, they were taken through the entire embryo. Yolk sac tissue was taken for determination of genotype.
Embryos homozygous for the RXRα mutation, either alone or in combination with other receptor mutations, are embryonic lethal around E14.5, whereas RARα1/RARβ double-mutant embryos survive to term. The definitive late fetal cardiovascular organization is largely established in normal mouse embryos by E14.5. Combination mutant embryos in this study were therefore isolated at E14.5 and serially sectioned to reveal defects in embryonic cardiovascular anatomy. Cardiovascular defects identified in this study are summarized in Table 1⇓. All embryos that were double heterozygous for any combination of receptor mutations were normal and were not included in the data presented in Table 1⇓.
Atrial, Valvular, and Venous Organization
Atrial chamber defects are not among the known fetal vitamin A deficiency phenotypes, although RA excess induces atrial septal defects with low frequency. Defects in atrial chamber septation were not observed, although these would have been difficult to identify at the early time point of this study (E14.5). Atrioventricular valve defects are also not among the vitamin A deficiency phenotypes, although subtle valvular defects have been documented in RXRα−/− embryos by microdissection and SEM analysis.23 These malformations would be extremely difficult to identify in histological sections and were not categorized in this study. Gross abnormalities, such as persistent atrioventricular canal, which was documented in vitamin A deficiency and observed in other combinations of receptor mutations,18 were not observed in the present study. No defects in the organization of venous inflow into the heart were noted in any receptor-deficient embryos, and such defects are not part of the known vitamin A deficiency or excess phenotype.
Ventricular Chamber Morphogenesis
Embryonic vitamin A deficiency causes hypoplasia of the developing ventricular chamber, resulting in a persistently thin-walled chamber with a poorly formed interventricular septum. This phenotype is embryonic lethal. An excess exposure to RA in mammals has not been reported to affect ventricular morphogenesis, although RA excess in chick embryos results in a thickened myocardium.8
All RXRα−/− embryos exhibit ventricular cardiomyocyte hypoplasia identical to the vitamin A deficiency phenotype, as documented in several previous studies.16 21 23 Almost all RXRα−/+ embryos at E14.5 are anatomically normal in the ventricular chamber, and RXRα−/+ adults are viable, although isolated cases of hypoplasia in E14.5 RXRα−/+ embryos have been observed.
All of the embryos analyzed in this study that were homozygous for the RXRα mutation, in combination with any additional mutations in the RARα1 or RARβ genes, revealed ventricular chamber hypoplasia, both in the chamber wall and in the ventricular septum (Table 1⇑; see Figure 1D⇓). The severity of hypoplasia was comparable to our previously described RXRα−/− embryos. One embryo of six that was heterozygous for the RXRα mutation and homozygous for the RARα1 mutation was hypoplastic in the ventricular chamber, comparable in severity to an RXRα−/− embryo, but as noted above, this has also been seen occasionally with RXRα−/+ embryos. Thus, addition of the RARα1 or RARβ mutations does not intensify or attenuate the ventricular phenotype that is obtained by the RXRα mutation alone.
RARα1/RARβ double-mutant embryos almost invariably appeared normal in the ventricular chamber, with well-formed muscular interventricular septum and chamber walls, although one atypical double-mutant embryo was affected by ventricular hypoplasia. One [RARα1−/−, RARβ−/+] embryo exhibited a very mild manifestation of ventricular hypoplasia (not shown), probably of marginal physiological consequence.
Thus, in general, combinations of the RARα1 and RARβ gene mutations do not result in ventricular chamber defects, although such defects are reproducibly observed with any combination in which the RXRα gene product is absent.
Morphogenesis of the Aortic Sac
PTA represents a series of defects in the formation of the A/P septum that leave a single outflow vessel arising from the heart. PTA occurs rarely and only partially after vitamin A deficiency and is infrequently, if ever, seen after vitamin A or RA excess treatment.
RXRα−/− embryos have a low incidence (≈10%) of PTA (see legend to Table 1⇑). In this study (Table 1⇑ and Fig 2⇓), one of four RARβ/RXRα double-mutant embryos exhibited PTA, and one of five [RARβ−/+, RXRα−/−] embryos displayed a localized partial defect in the A/P septum (A/P window, Fig 2L⇓). Because the RXRα mutation alone yields such defects in ≈10% of cases, the frequencies in these classes of combination mutants do not appear to differ significantly from those observed with RXRα alone. However, PTA was recovered in two of six [RARα1−/+, RXRα−/−] embryos and in five of six RARα1/RXRα double-mutant embryos. Thus, combining the RARα1 mutation with RXRα deficiency significantly increased the incidence of A/P septal malformation.
PTA was also seen in all RARα1/RARβ double-mutant embryos. This was true as well for embryos examined at E18.5,20 indicating that the underlying cause of PTA is not a delay but rather a failure in the formation of the A/P septum. Embryos homozygous for the RXRα mutation could not be taken past E14.5 to demonstrate this same principle because of the lethality caused by ventricular hypoplasia.
Pediatric clinical diagnoses of PTA are subdivided into four categories24 (see also below). All four subtypes of PTA were recovered in the various double-mutant embryos studied, and with the possible exception of PTA type A1, which appeared preferentially in [RARα1−/+, RXRα−/−] embryos, no obvious asymmetric distribution of PTA subtypes with respect to RA receptor mutant genotype was apparent in this study (Table 2⇓). This is not necessarily surprising, as the classification of PTA subtypes is based on features relevant to surgical correction as well as underlying developmental mechanisms.
Morphogenesis of the Conotruncus
Conotruncal defects seen after fetal vitamin A deficiency include membranous VSDs and DORV. Membranous VSDs, DORV, and transposition (transposition-type DORV and transposition of the great arteries) are known consequences of RA excess exposure, although transposition is not one of the described vitamin A deficiency conditions.
Approximately 20% of RXRα−/− embryos exhibit VSDs representative of an incompletely formed conus septum (see legend to Table 1⇑). The frequency of the more severe DORV phenotype is generally low in RXRα−/− embryos, and RXRα heterozygotes are only rarely affected by conotruncal defects. A higher frequency of more subtle conotruncal defects can be visualized in RXRα−/− embryos by microdissection and scanning electron microscopy,23 but these would generally not have been observed by histological analysis as in the present study.
Combining the RARα1 mutation with RXRα deficiency synergistically increased the frequency and severity of conus malformations (Table 1⇑). All of the four diagnosable [RARα1−/+, RXRα−/−] embryos displayed deficiencies in the conus, three of which were severe (ie, comparable to that shown in Fig 1C⇑), and the fourth was of intermediate severity (Fig 1D⇑). The additional two embryos analyzed of this genotype class were affected by PTA, which obscures evaluation of conus defects and which were therefore not independently scored in this study. The single RARα1/RXRα double-mutant embryo that did not have PTA exhibited transposition-type DORV, in which both outflow vessels originated from the right ventricle, with the aorta passing ventral to the pulmonary trunk (Fig 1E⇑ and 1F⇑). Synergy between the RARα1 and RXRα genes was further evidenced by the recovery of conus defects in two of six embryos homozygous for the RARα1 mutation and heterozygous for the RXRα mutation (Fig 1B⇑ and 1C⇑), although such defects are never seen in RARα1−/− embryos and only rarely in RXRα−/+ embryos.
To a lesser extent, the RARβ mutation also synergized with the RXRα mutation. Thus, conus deficiencies were evident in two of the five [RARβ−/+, RXRα−/−] embryos and in two of the three diagnosable RARβ/RXRα double-mutant embryos. However, three of the [RARβ−/+, RXRα−/−] embryos and all of the [RARβ−/−, RXRα−/+] embryos were normal in the conus, demonstrating a significantly lower synergy compared with the corresponding combinations of the RARα1 and RXRα mutations.
Conotruncal malformations were not observed in homozygous/heterozygous or double-heterozygous combinations of the RARα1 and RARβ mutations, and independent conus defects were not obvious when both RARα1 and RARβ were lacking. A membranous VSD was seen in all RARα1/RARβ double-mutant embryos, but this is at least in part an obligate secondary consequence of PTA.
Aortic Arch Artery Anomalies
A broad spectrum of aortic arch arterial defects, involving inappropriate persistence or regression of the third, fourth, and sixth arch arteries and the pulmonary arteries, has been described for both fetal vitamin A deficiency and excess syndromes. No individual defect is seen uniformly, or even regularly, although some anomalies occur more frequently than others.
The types of aortic arch artery anomalies observed in the receptor-deficient embryos of this study were varied (Table 2⇑). Persistence of the right dorsal aorta, either in continuity with the innominate artery (ie, a vascular ring) or as a retroesophageal right subclavian artery (Fig 2G⇑ through 2I), was seen in several genotype classes. A right arch of the aorta was seen in four RARα1/RARβ double-mutant embryos (Fig 2J⇑ and 2K⇑); in two of these, the left subclavian artery was supplied by a retroesophageal persistent left dorsal aorta, whereas in the other two, a left innominate artery was present as a derivative of the left fourth arch artery. Abnormalities of the pulmonary arteries, either involving an inappropriate origin of the right pulmonary artery (from the left side of the truncus arteriosus [Fig 2J⇑ and 2K⇑] or from the innominate artery) or a missing left pulmonary artery, were noted in several genotype backgrounds. An absent left pulmonary artery generally copresented with a missing left lung, but in one example the left lung was present and was supplied by collateral vessels from the dorsal aorta, suggesting that the primary defect might be in the formation of the left pulmonary artery rather than of the left lung.
In two embryos with PTA, the left sixth arch artery, rather than the left fourth artery, became the arch of the outflow tract, and the left fourth arch artery regressed. In sections, this was ascertained by the presence of a single vessel ascending from the arch of the aorta from which the right and left carotid arteries both branched (Fig 2G⇑ through 2I). This organization is referred to as interrupted aortic arch type B, and PTA with a left sixth aortic arch is defined as PTA type A4. In PTA types A2 to A4, where only a single outflow vessel connects to peripheral circulation, it has been speculated24 that hemodynamic growth of the fourth and sixth arch arteries are competitive processes. Accordingly, when the left sixth arch becomes the definitive arch of the outflow tract (type A4), the left fourth arch loses blood flow and regresses. Similarly, when the fourth arch artery becomes the aortic arch (types A2 and A3), the sixth arch artery regresses, evident as an absent ductus arteriosus. In the two observed cases of PTA type A1, where a shortened pulmonary trunk was formed distal to the outflow valves, the ductus arteriosus was maintained.
We16 23 and others21 have previously demonstrated that one of the fetal vitamin A deficiency cardiovascular phenotypes, ventricular chamber hypoplasia, is replicated in RXRα−/− embryos with complete penetrance. Additional deficiency and excess phenotypes, involving malformations of the conotruncus and aortic arch arteries, were observed in RXRα−/− embryos with low penetrance, and no defects whatsoever have been observed in individual RARα1−/− and RARβ−/− embryos. In the present study, the severity and frequency of conotruncal and aortic arch artery defects were greatly intensified by combining the RXRα mutation with the RARα1 mutation in particular and with the RARβ mutation as well. Combining the RARα1 and RARβ mutations resulted in PTA plus aortic arch artery defects with complete penetrance. Most of the known fetal cardiovascular vitamin A deficiency and excess syndromes have been replicated in combinations of our RXRα, RARα1, and RARβ mutations, mostly at high or complete penetrance.
Vitamin A Deficiency and Receptor Deficiency
In general, a similar spectrum of cardiovascular defects is obtained by RA receptor mutation and nutritional vitamin A deficiency, demonstrating that removal of ligand or receptor compromises the same developmental processes. However, in the vitamin A deficiency studies, aortic arch artery defects, membranous VSDs, and ventricular hypoplasia were the prominent malformations; PTA was seen rarely and only partially; and severe conus defects were observed only rarely. Most likely, each aspect of cardiovascular morphogenesis has a different threshold to nutritional deficiency, such that ventricular and aortic arch artery defects occur first, with more severe deficiency then affecting the conotruncal septa. As such, because the ventricular phenotype is midgestationally lethal and because severe vitamin A deficiency is incompatible with pregnancy, more substantial conotruncal septal defects might not have been possible to recover in the nutritional studies. In fact, persistent atrioventricular canal and pulmonary atresia were only seen in vitamin A deficiency studies after midgestation restoration of vitamin A,2 suggesting that these phenotypes require extreme vitamin A deficiency to occur and were only recovered if lethality was averted. Thus, each compartment of the developing heart uses RA through pathways that are quantitatively unique.
Receptor mutation distinguishes these cardiovascular developmental processes independently. As described above, different compartments of the developing heart are differentially compromised by different combinations of receptor mutations, with a spectrum of genetic sensitivity differing substantially from that of nutritional sensitivity. For example, A/P septal defects were only marginally recovered in nutritional studies but uniformly seen in RARα1/RARβ mutant embryos. A possible explanation for these observations might involve different levels of receptor expression in different compartments of the developing heart or different functional properties of different receptors relevant to specific aspects of morphogenesis in individual cardiovascular compartments.
Vitamin A Deficiency and Excess Phenotypes
Transposition, a consequence of malrotation of the conotruncal septum, is most commonly associated with RA excess and was not originally reported with vitamin A deficiency. However, the recovery of transposition in this study, albeit in a single RARα1/RXRα double-mutant embryo, suggests that alignment of the conotruncal ridges may also require endogenous vitamin A signaling. The severity of vitamin A deficiency in the nutritional studies might not have been sufficient to induce this phenotype (see comments above).
Excess exposure to RA during defined periods of embryonic development leads to conotruncal and aortic arch malformations that overlap (although not completely so) with the vitamin A deficiency syndrome. Transposition may also be one of these phenotypes, and concordant phenotypes in other developing tissues have also been obtained after vitamin A deficiency and excess (ie, cleft palate4 25 ). This could indicate that both types of manipulation perturb the same process or that different consequences are initiated by deficiency and excess, which ultimately result in convergent dismorphogenesis. For morphogenesis of the aortic arch arteries, the temporal sensitivities to excess and deficiency are noncoincident,2 4 suggesting that different consequences are elicited by each treatment, although little evidence is available for other malformations.
Paradigms of Congenital Cardiovascular Disease
The recovery of the known vitamin A deficiency and excess cardiovascular phenotypes in the double-receptor mutants described here illustrates the somewhat obvious conclusion that vitamin A signaling is mediated by the vitamin A receptors. The real significance of this effort, however, is that genetically defined experimental models are available to uncover the mechanistic basis of these phenotypes and of the normal processes that underlie them. Thus, ventricular chamber defects are specifically recovered with complete penetrance in RXRα−/− embryos, and A/P septal defects are recovered with complete penetrance and specificity in RARα1/RARβ embryos. Conotruncal defects are recovered in high penetrance in [RARα1−/+, RXRα−/−] embryos (although with the added experimental complications of ventricular chamber and A/P septal defects as well). We have recently determined (M. Chiang, H.M. Sucov, and R.M. Evans, unpublished observations, 1997) that combining a newly established mutation in the RXRγ gene with the RXRα mutation yields conotruncal and ventricular defects with complete penetrance and without an increase in the frequency of PTA. Other investigators have also described additional combinations of RA receptor gene mutations that yield cardiovascular defects at various frequencies.18 21 Through further analysis of these lines of mice, molecular and cellular paradigms might be obtained to explain the etiology of the observed murine defects and possibly of the corresponding defects in humans as well. These genetic models should be of great usefulness in pursuing the fundamental bases of normal and pathological cardiovascular development.
Selected Abbreviations and Acronyms
|DORV||=||double-outlet right ventricle|
|E (with number)||=||embryonic day|
|PTA||=||persistent truncus arteriosus|
|RAR, RXR||=||RA receptor subfamilies|
|VSD||=||ventricular septal defect|
This study was supported in part by grants to Dr Sucov from the American Heart Association (No. 96012100) and the American Heart Association, Greater Los Angeles Affiliate, Inc (No. 1118-GI1); to Dr Giguere from the Medical Research Council of Canada (No. MA-10561) and the National Cancer Institute of Canada (No. 1662); and to Dr Evans from the National Institutes of Health (GM-26444), the Mathers Foundation, and the Howard Hughes Medical Institute.
- Received October 15, 1996.
- Accepted March 6, 1997.
- © 1997 American Heart Association, Inc.
Kalter H, Warkany J. Experimental production of congenital malformations in strains of inbred mice by maternal treatment with hypervitaminosis A. Am J Pathol. 1961;38:1-21.
Pexieder T, Blanc O, Pelouch V, Ostadalova I, Milerova M, Ostadal B, Clark EB, Markwald RR, Takao A, eds. Developmental Mechanisms of Heart Disease. Armonk, NY: Futura Publishing Co; 1995:297-307.
Evans RM. The steroid and thyroid hormone receptor superfamily. Science. 1988;240:889-895.
Li E, Sucov HM, Lee KF, Evans RM, Jaenisch R. Normal development and growth of mice carrying a targeted disruption of the alpha 1 retinoic acid receptor gene. Proc Natl Acad Sci U S A. 1993;90:1590-1594.
Sucov HM, Dyson E, Gumeringer CL, Price J, Chien KR, Evans RM. RXR alpha mutant mice establish a genetic basis for vitamin A signaling in heart morphogenesis. Genes Dev. 1994;8:1007-1018.
Mendelsohn C, Lohnes D, Decimo D, Lufkin T, LeMur M, Chambon P, Mark M. Function of the retinoic acid receptors (RARs) during development, II: multiple abnormalities at various stages of organogenesis in RAR double mutants. Development. 1994;120:2749-2771.
Lohnes D, Mark M, Mendelsohn C, Dolle P, Dierich A, Gorry P, Gansmuller A, Chambon P. Function of the retinoic acid receptors (RARs) during development, I: craniofacial and skeletal abnormalities in RAR double mutants. Development. 1994;120:2723-2748.
Hale F. The relation of vitamin A to anophthalmos in pigs. Am J Opthalmol. 1935;18:1087-1093.