Defects in heart and lung development in compound heterozygotes for two different targeted mutations at the N-myc locus

N-myc is a member of the myc family of proto-oncogenes, which includes N-myc, c-myc, and L-myc. Myc proteins are site-specific DNA-binding proteins (Blackwell et al., 1990; Prendergast and Ziff, 1991; Alex et al., 1992), belonging to the basic-helix-loop-helix class of transcription factors, which includes genes that control cell fate determination in such diverse processes as myogenesis, neurogenesis and sex determination (reviewed in Garrell and Campuzano, 1991).

Deregulated expression of myc genes has been implicated in the genesis or progression of a number of naturally occurring tumours, in the transformation of cells in culture and in the formation of tumours in transgenic mice (reviewed in DePinho et al., 1991). In general, the sites of expression of a given myc gene in vivo reflect the types of tumours associated with its elevated expression. Thus N-myc is expressed predominantly in the embryo where it is restricted to undifferentiated subsets of cells in the central and peripheral nervous system, lung, kidney and eye (Mugrauer et al., 1988; Hirning et al., 1991; Zimmerman et al., 1986) and overexpression of N-myc has been associated with tumours of embryonic origin such as neuroblastoma (Kohl et al., 1983; Schwab et al., 1983), small-cell lung cancer (Nau et al., 1986; Wong et al., 1986), Wilm’s tumour (Nisen et al., 1986) and retinoblastoma (Lee et al., 1984). N-myc is also expressed in the skin (Mugrauer et al., 1988), in the epithelial layer of the intestine (Hirning et al., 1991) and, earlier in development, in the heart, sclerotome and visceral arches (Katoh et al., 1991).

The functioning of a Myc protein in vivo should depend not only on its own level of expression, but also on the levels of Max, a protein which, like the Myc proteins, possesses a basic-helix loop helix-leucine zipper (bHLH-LZ) domain (Blackwood and Eisenman, 1991), and which associates with N-Myc, L-Myc and c-Myc proteins in vivo (Blackwood et al., 1992; Wenzel et al., 1991; Mukherjee et al., 1992). Max is required for specific DNA binding by Myc proteins (Blackwood and Eisenman, 1991; Prendergast and Ziff, 1991; Kato et al., 1992; Barrett et al., 1992), and has been shown to be required for transcriptional activation (Amati et al., 1992) and transformation (Amati et al., 1993) by c-myc. Unlike Myc proteins, Max is able to form homodimers in vitro and thereby to bind the myc-binding site (Prendergast and Ziff, 1991; Kato et al., 1992). However Max does not transactivate downstream genes on its own (Amati et al., 1992; Kretzner et al., 1992) because it lacks a transactivation domain (Kato et al., 1992) which Myc proteins possess (Kato et al., 1990). Both transformation of fibroblasts by c-myc and N-myc (Mukherjee et al., 1992; Makela et al., 1992; Prendergast et al., 1992) and transcriptional transactivation by c-myc (Amati et al., 1992; Kretzner et al., 1992) have been shown to be enhanced by low levels of Max and inhibited by excess Max, suggesting that Myc function is indeed influenced by levels of Max. Recently, bHLH-LZ proteins have been isolated which also bind the Myc recognition site and which suppress transcription as heterodimers with Max (Ayer et al., 1993; Zervos et al., 1993). One of these, Mad (Ayer et al., 1993), has been shown to compete with Myc for Max in vitro and in transfected cells. Thus high levels of expression of Max dimerization partners may indirectly affect Myc function in vivo by competing for available Max protein and for DNA-recognition sites. Finally, Myc function in vivo may be affected by the levels of other Myc proteins (Mukherjee et al., 1992; Resar et al., 1993).

In an effort to understand the function of myc genes in embryogenesis, leaky and null mutations were made in N-myc in embryonic stem cells by homologous recombination (Charron et al., 1990; Stanton et al., 1990; Sawai et al., 1991; Moens et al., 1992). These mutations have allowed a description of the function of N-myc at different stages of development. Mice homozygous for the null mutations die at midgestation (Stanton et al., 1992; Charron et al., 1992) while mice homozygous for the leaky mutation survive until birth, when they die due to a defect in lung branching morphogenesis (Moens et al., 1992). The latter phenotype is consistent with the normal expression of N-myc in the developing lung epithelium, and has led us to postulate that N-myc plays a role in the response of the lung epithelium to the signals from the lung mesenchyme that induce epithelial branching.

There have been two detailed reports of the phenotype of mice carrying null mutations in N-myc (Stanton et al., 1992; Charron et al., 1992). In the homozygous condition, these mutations cause embryonic death at around 11.5 days p.c. and cause hypoplasia in a number of components of the embryo, different aspects of which have been emphasized by the two groups. The developing genitourinary system shows a reduced number of mesonephric tubules in mutant embryos, which is consistent with N-myc’s normal expression in the newly induced epithelium of the meso- and metanephric tubules (Katoh et al., 1991; Mugrauer et al., 1988). The genital ridge is also hypoplastic. Stanton et al. (1992) also describe a defect in the development of the stomach and intestine, consistent with N-myc’s expression in the epithelium of these tissues (Hirning et al., 1991). Homozygotes also have reduced cranial and spinal ganglia and a thin neuroepithelium in the telencephalon; both of these are sites of N-myc expression in the embryo. Finally, homozygotes have a defect in heart development in which the heart appears to cease development at 9.5 days p.c. (Charron et al., 1992) and as a result is poorly compartmentalized and appears not to have undergone the normal epithelial-to-mesenchymal transitions that result in the formation of the septa and valves. This is again consistent with a previous description of N-myc expression in the myocardium of the 9.5 day embryo (Katoh et al., 1991).

By combining leaky and null alleles of N-myc in a compound heterozygote, we hoped to investigate N-myc function in the mouse embryo at a stage intermediate to those identified by the null and leaky alleles individually. We find that compound heterozygotes indeed survive longer than null homozygotes, but that they die before birth. The mutant phenotype includes a more severe defect in branching morphogenesis of the lung, as well as a cardiac myocyte hypoplasia in the compact subepicardial layer of the ventricular myocardium. We show that, consistent with this phenotype, N-myc expression in the developing myocardium is restricted to the compact layer. A role for myc genes in the control of myocyte differentiation in the heart is discussed.

Yolk sacs from dissected embryos were washed several times in fresh PBS and placed in 100-200 μl of proteinase K buffer with non-ionic detergents (50 mM KCl, 10 mM Tris pH 8.3, 2.0 mM MgCl2, 0.1 mg/ml gelatin, 0.45% Nonidet P-40, 0.45% Tween-20). 10 μg of proK was added to each tube and samples were incubated at 55°C overnight. 10 μl of the mixture was placed in a PCR tube with 30 μl sterile water and heated to 94°C for 10 minutes before cooling to 62°C. 0.1 μg of each primer, 200 μM dNTPs, 10× reaction buffer and Taq polymerase (Promega) were added and PCR was run with an annealing temperature of 62°C (1 minute), an extension temperature of 72°C (2 minutes), and a denaturing temperature of 94°C (1 minute). Each consecutive cycle was extended by 2 seconds. The oligonucleotide primers used for the PCR detection of 9a, BRP and wild-type alleles of N-myc are shown in Fig. 1. Their sequences are as follows: primer a (N-myc intron 1): 5′-GGTAGTCGCGCTAGT AAGAG-3′; primer b (N-myc exon 2): 5′-GGCGTGGGCA GCAGCTCAAAC-3′; primer c (N-myc intron 2): 5′-CCGAGCATCTGTCGCCAAGTC-3′; primer d (neo): 5′-GACCGCTATCAGGACATAGCG-3′.

An anti-human N-myc monoclonal antibody, NCMII 100 (Ikegaki et al., 1986), was generously provided by Dr R. Kennett. 11.5 days p.c. embryos from 9a/+×BRP/+ crosses were dissected and homogenized in several volumes of 1× sample buffer without bromophenol blue (60 mM Tris, pH 6.8, 2% SDS, 0.1 M DTT, 0.32% pharmalyte 3-10 (Pharmacia)). Samples were boiled for 5 minutes and chromosomal DNA was sheared by repeated passage through a 26-gauge needle before freezing at −20°C for future use. Yolk sacs were washed several times in PBS and were prepared for and typed by PCR as described above. After typing, extracts of embryos of each genotype were pooled and the concentration of protein in each sample was determined by the Bradford assay (protein assay reagent from Biorad). Approximately 20 μg of protein from each genotype in 2× loading buffer (0.1 M Tris pH 6.8, 20% glycerol, 4% SDS, 0.2% bromophenol blue, 0.2 M DTT) was run in a 10% SDS-PAGE. Protein was transferred onto PVDF membrane (Immobilon-P, Millipore). Application of primary antibody and fixation of the primary antibody to the blot with 0.2% glutaraldehyde was performed as described (Ikegaki and Kennett, 1989). Horseradish peroxidase-conjugated goat anti-mouse secondary antibody and reagents for the chemiluminescent detection of HRP activity were obtained in the ECL western blotting kit (Amersham) and were used as suggested by the manufacturer.

Dissected embryos were fixed in 10% buffered formalin overnight at room temperature, then were dehydrated and cleared by soaking sequentially in 70%, 80%, 90% and 95% ethanol, 1:1 ethanol:xylene, xylene, 1:1 xylene:paraffin wax, (TissuePrep 2, Fisher Scientific) each for 1 hour. Embryos were then oriented and embedded in paraffin wax. Mutant embryos and control littermates were oriented either for sagittal or frontal sections. 5 μm sections were placed on slides, which were dried overnight at 37°C. Slides were dewaxed, rehydrated, and stained with hematoxylin and eosin. To analyse mutant phenoypes, sections throughout wild-type and mutant embryos were carefully examined in order to find those sections that passed through corresponding regions of the different embryos. Mutant embryos were always compared to non-mutant littermates rather than to non-mutant embryos from other litters, in order to rule out differences due to age.

Probes for in situs were as follows. N-myc: a 541 bp Pst1-Sca1 fragment of the N-myc genomic clone N7.7 (DePinho et al., 1986), including largely 3′-untranslated sequence; c-myc: a 351 bp HaeIII fragment including the untranslated first exon (Bossone et al., 1992); flk-1: an 800 bp fragment spanning the transmembrane domain and part of the extracellular ligand-binding domain (Yamaguchi et al., 1993) and α-cardiac actin, a 130 bp BamHI fragment including the first untranslated exon of the mouse αcardiac actin gene (Sassoon et al., 1988). In situ hybridization was carried out essentially as described (Frohman et al., 1990) with the following modifications: in vitro-transcribed, α-³⁵S-labeled RNA probes were used at 5×10⁴ disintegrations/minute per 1 μl of hybridization solution for the α-cardiac actin probe, and at 1×10⁵ disintegrations/minute per 1 μl of hybridization solution for c-myc, N-myc and flk-1 probes. Slides were hybridized at 53°C overnight, and were washed in the presence of 0.1% β-mercaptoethanol as a reducing agent. High stringency washes were in 50% formamide, 0.1% β-mercaptoethanol, and 1× SSC at 65°C. Final rinses in 2× and 0.1× SSC were at 65°C for 15 minutes. Slides were exposed for either 6 days (for α-cardiac actin probe) or for 14 days (for N-myc, flk-1 and c-myc probes). After developing and photographing with dark-field illumination, slides were stained with hematoxylin and eosin and were rephotographed with bright-field illumination.

The leaky mutation that was generated in N-myc (Moens et al., 1992) was termed N-myc^9a and the null allele of N-myc used in this study was named N-myc^BRP (Stanton et al., 1992). Both the N-myc^9a mutation and the N-myc^BRP mutation are lethal in the homozygous condition. Therefore, N-myc^9a/BRP embryos were generated by crossing N-myc^9a/+ females to N-myc^BRP/+ males, with the expected ratio of embryos being 1 N-myc^+/+: 1 N-myc^9a/+: 1 N-myc^BRP/+: 1 N-myc^9a/BRP. The reciprocal cross was also performed: no differences in mutant phenotypes were observed.

In order to expedite the genotypic analysis of offspring from this cross, a PCR strategy was designed that used four oligonucleotide primers to distinguish 9a, BRP, and wild-type alleles of N-myc (Fig. 1). 161 live offspring from N-myc^9a/+×N-myc^BRP/+ crosses were typed by PCR between 1 day and 3 weeks after birth. Among these pups, 59 were N-myc^+/+, 53 were N-myc^9a/+ and 49 were N-myc^BRP/+. No live animals were N-myc^9a/BRP. No pups were observed to die postnatally as occurred in the case of N-myc^9a/9a newborns, suggesting that the N-myc^9a/BRP phenotype was indeed more severe than the N-myc^9a/9a phenotype.

We dissected embryos from N-myc^9a/+×N-myc^BRP/+crosses at various stages of gestation in order to determine the time during gestation when N-myc^9a/BRP embryos died. Between 8.5 and 11.5 days p.c., N-myc^9a/BRP embryos were phenotypically normal and constituted approximately 25% of the total number of embryos, by PCR analysis. Of 268 embryos dissected between 8.5 and 11.5 days p.c., 61 (23%) were N-myc^9a/BRP embryos. This is clearly different from the situation in N-myc^BRP/BRP embryos, which were all found to be dead by 11.5 days p.c. (Stanton et al., 1992). The observed differences in phenotype among the various mutants were unlikely to be due to different genetic back-ground effects, since all mutant phenotypes were analysed in outbred mice.

Occasionally, N-myc^9a/BRPembryos at 11.5 days were slightly smaller than their littermates but were otherwise indistinguishable from their wild-type or single heterozy-gote littermates. However, by 12.5 days, most N-myc^9a/BRP embryos were distinguishable from their littermates either because they were dead and necrotic, or because they were alive but had a characteristic edema of the body wall in the neck area (Fig. 2). Subsequent histological analysis identified this swelling as likely being the result of extravasion of fluid from the vascular compartment into the connective tissue, since the jugular veins are dilated in these embryos (Fig. 7g,h). Of 230 embryos dissected at 12.5 days p.c., 42 (18%) were N-myc^9a/BRP, but among these, 12 were dead and necrotic, 23 had the characteristic edema described above, two were smaller than normal and five were pheno-typically normal. After 12.5 days p.c., live N-myc^9a/BRP embryos were progressively less frequent and all were phenotypically abnormal. N-myc^9a/BRP embryos that survived until 14.5 days p.c. were always smaller than normal and had a large edema in the neck area. Live N-myc^9a/BRP embryos were not observed after 14.5 days p.c. Coincident with the loss of N-myc^9a/BRP embryos there was an increase in the number of resorptions visible in the uterus. These resorptions were not typed, but presumably were derived from N-myc^9a/BRP embryos, since the sum of the number of N-myc^9a/BRP embryos and the number of resorptions added up to approximately 25% of the total number of embryos and resorptions at each stage of gestation. These data are shown graphically in Fig. 3.

Although we have previously shown reduced N-myc mRNA levels in N-myc^9a/9a embryos (Moens et al., 1992), N-Myc protein levels have not been determined in mice bearing either the leaky or the null N-myc alleles. In the present study, we have used western blot analysis to assess N-Myc protein levels in embryos with the various N-myc genotypes. Fig. 4 shows a western blot performed on extracts of whole 11.5 days p.c. embryos from a N-myc^9a/+×N-myc^BRP/+ cross, using a monoclonal antibody that was raised against human N-Myc protein (Ikegaki et al., 1986), but which also recognizes mouse N-Myc protein. N-Myc protein runs at approximately 65×10³Mr. This N-myc antibody also showed binding to a protein of approximately 100×10³Mr. The relative intensities of this crossreacting band between lanes matched the relative intensities of a ubiquitously expressed protein phosphatase, syp (not shown), so this band was used as a loading control in scanning densitometry. The amount of N-Myc protein in N-myc^9a/BRP embryos was considerably lower than in their wild-type or singly heterozygous littermates. While N-myc^9a/9a embryos had approximately 25% of wild-type levels of N-Myc protein (not shown), N-myc^9a/BRP embryos had approximately 15% of wild-type levels, as determined by scanning densitometry of this and several other, similar Western blots. This is consistent with the more severe phenotype of N-myc^9a/BRP embryos. N-myc^BRP+ embryos had approximately 50% of wild-type levels of N-Myc protein, consistent with the N-myc^BRP mutation being a null mutation.

In order to determine more precisely the phenotype of N-myc^9a/BRP embryos, histological analysis of hematoxylin and eosin-stained, sectioned embryos was performed at various stages of development. We examined a total of 42 N-myc^9a/BRP embryos and 50 N-myc^+/+ littermates in this manner (5 N-myc^9a/BRP embryos at 14.5 days p.c., 28 at 12.5 days p.c., 3 at 11.5 days p.c., and 6 at 10.5 days p.c.). Younger embryos were analyzed without sectioning, after whole-mount RNA in situ hybridization with various probes (data not shown). Occasionally, N-myc^9a/+ embryos were used as controls since we have previously shown that the N-myc^9a mutation has no phenotypic effect in the heterozygous condition. N-myc^9a/BRP embryos not used for histology were used for N-Myc protein analysis and RNA in situ analysis.

Before 12.5 days p.c., N-myc^9a/BRP embryos were apparently normal and could not generally be distinguished from their littermates either by gross morphology or by histological examination of sectioned embryos. The first signs of lethality occurred at 12.5 days, when 12 out of 42 N-myc^9a/BRP were clearly in the process of resorption. Aside from the edema in the neck area, described above, and their slightly smaller size, live N-myc^9a/BRP embryos at 12.5 days were still not grossly abnormal. However careful examination of specific organ systems did reveal particular defects.

N-myc^9a/9a mice die at birth due to a defect in lung branching morphogenesis, consistent with the observation that N-myc is expressed in the lung epithelium at early stages of lung development (Moens et al., 1992). Specifically, at 12.5 days p.c., when wild-type lungs have tertiary and quaternary branches, N-myc^9a/9a lungs had only the beginnings of tertiarybranches. The further reduction in the amount of N-Myc protein in N-myc^9a/BRP embryos led to an enhancement of this phenotype, as predicted. Branching morphogenesis in N-myc^9a/BRP lungs was all but blocked (Fig. 5A,B), N-myc^9a/BRP embryos having only the beginnings of secondary branches. Interestingly, the pattern of branching was normal, as can be observed by the presence of a rudimentary right postcaval lobe extending to the left side of the N-myc^9a/BRP embryo (Fig. 5B). The lungs begin their development by budding from the trachea into the surrounding mesenchyme and are visible as two pouches at 10.5 days of development. This early phase of lung development occurred normally in N-myc^9a/BRP embryos (not shown).

In order to determine whether aspects of lung development other than the branching of the lung epithelium were affected by the reduction in N-Myc protein levels, we performed RNA in situ hybridizations on wild-type and N-myc^9a/BRP lungs at 12.5 days p.c. (Fig. 6). Flk-1, which encodes a tyrosine kinase receptor for vascular endothelial growth factor, marks endothelial cells in the developing capillaries and blood vessels in the mouse (Yamaguchi et al., 1993; Millauer et al., 1993), and as such is expressed in a punctate manner throughout the lung mesenchyme where the lung vasculature is developing (Fig. 6C). The pattern and intensity of flk-1 expression was not affected in N-myc^9a/BRP lungs (Fig. 6G), indicating that the development of the lung vasculature is not affected by the reduction in N-Myc protein observed in N-myc^9a/BRP embryos. Fig. 6B,F shows the pattern of expression of N-myc in wild-type and N-myc^9a/BRP lungs, respectively, at 12.5 days p.c., reconfirming that expression is restricted to the lung epithelium and showing that the reduced N-Myc protein levels in N-myc^9a/BRP embryos did not lead to a complete loss of N-myc-expressing cells in the lung epithelium. N-myc^9a/BRP lungs appeared to express more than 15% of normal levels of N-myc RNA (Fig. 6F). This is because the N-myc^BRP allele encodes a transcript that includes the entire N-myc open reading frame, but which contains stop codons that prevent translation of any part of the N-myc protein (Stanton et al., 1990). Fig. 6D,H demonstrate that levels of c-myc mRNA at the cellular level were not affected by the reduction in N-Myc protein in the lungs of N-myc^9a/BRP embryos. C-myc is normally expressed in the lung mesenchyme (Fig. 6D, Hirning et al., 1991) and the pattern and level of c-myc expression was not altered in the N-myc^9a/BRP lung, despite its reduced size (Fig. 6H).

The other clearly visible defect in N-myc^9a/BRP embryos examined at 12.5 days p.c. was in the morphology of the heart. While N-myc^9a/BRP hearts had the normal fourchamber structure, normal endocardial cushions, valves and septa, the heart was small and the myocardium was abnormally thin (Fig. 7A-D). The latter defect was particuarly apparent in the ‘compact’, or subepicardial layer of the ventricular myocardium while the atrial myocardium and the inner trabecular layer of the ventricles were less severely affected. In mammals, trabeculation occurs early during heart development and, at later stages, growth occurs largely in the compact layer (Rumyantsev, 1991). In N-myc^9a/BRP embryos, this growth of the compact layer appeared not to occur, so that the N-myc^9a/BRP myocardium at 12.5 days p.c. was no thicker than it had been at 10.5 days p.c.

Occasionally, N-myc^9a/BRP embryos survived until 14.5 days p.c. These embryos were always grossly abnormal, with large edemas in the neck area, and were smaller than their littermates. Fig. 7E and F compares wild-type and N-myc^9a/BRP hearts at this stage, showing that the compact layer of the myocardium of the compound heterozygote was still no more developed than it had been at 10.5 days.

It seems likely that this failure in the proliferation of the ventricular myocardium leads to a failure of the circulatory system and ultimately in fetal death. The inefficient function of the heart was demonstrated by the hugely expanded jugular veins of N-myc^9a/BRP embryos (Fig. 7G,H), since the lack of a highly muscularized ventricular myocardium pumping blood from the heart could lead to a back-up of blood in the major veins. However, it should be noted that we did not see this marked distension in the other major veins that lead into the heart.

We wished to determine whether the observed defect in the development of the compact layer of the ventricular myocardium correlated with N-myc expression in the heart. Expression of N-myc in the heart at 9.5 days p.c. has been described (Katoh et al., 1991), and we have confirmed this by whole-mount RNA in situ hybridization (data not shown). However the differentiation of trabecular and compact layers has only just begun at this stage, and N-myc expression was not shown to be restricted to one specific layer. Our experiments indicate that, by 10.5 days p.c., N-myc expression in the heart is largely confined to the compact layer and not to the trabecular layer of the developing ventricular myocardium (Fig. 8B). For comparison, α-cardiac actin, a marker of differentiated myocytes in the developing heart (Sassoon et al., 1988), is expressed in both the trabeculae and the compact layer (Fig. 8C). We observed continued N-myc expression in the compact layer at the time when a mutant phenotype was first observed in the heart (12.5 days p.c., Fig. 9B,G). This higher level of of N-myc expression in the compact layer compared to the trabecular layer is consistent with a direct role in the development of the mutant phenotype we have observed in N-myc^9a/BRP embryos.

Considerable N-myc expression was detected in the compact layer of 12.5 days p.c. N-myc^9a/BRP hearts (Fig. 9B,G) because the N-myc^BRP mutation prevents translation but not transcription of N-myc, as noted above. However, as expected, the layer of N-myc-expressing cells was narrower in the ventricular myocardium of compound heterozygotes than in the wild-type myocardium (compare Figs 9B and 8B). The observation that the mutant alleles continue to be transcribed in the N-myc^9a/BRP myocardium suggests that cells that normally express N-myc are present but in reduced numbers.

Flk-1, the endothelial cell marker described above, is expressed in the endocardium (Yamaguchi et al., 1993), which is the mesodermally derived inner lining of the heart that later contributes to the endocardial cushion and heart valves through an epithelial-to-mesenchymal transition (Fig. 9D,I; Markwald et al., 1990). The differentiation of the endocardium was not affected in N-myc^9a/BRP embryos, as demonstrated by the continued expression of flk-1 in the endocardium (compare Figs 10D,I and 8D,I). The N-myc^9a/BRP genotype also did not prevent the differentiation of cardiac myocytes in the myocardium, as indicated by the strong expression of α-cardiac actin in both the compact and trabecular layers of the mutant myocardium (compare Figs 10C,H and 9C,H).

The finding that reduction in N-myc expression in the compact layer of the heart caused a myocyte hypoplasia evident at 12.5 days p.c. was interesting, in that overexpression of the closely related c-myc proto-oncogene in the heart of transgenic mice has been described to cause a hyperplasia of the myocytes which is also evident during development (Jackson et al., 1990). The normal expression pattern of c-myc in the heart has not been described, so the relative roles of these two genes in myocyte proliferation in vivo was not clear. Fig. 9E,J demonstrates that c-myc is expressed at low levels in the heart, and that its pattern of expression is different, and to some extent complementary to that of N-myc. While N-myc is expressed in the compact layer of the ventricular myocardium, c-myc expression is largely restricted to cells adjacent to and on the outside of the compact layer, and to isolated cells or clusters of cells adjacent to and on the inside of the compact layer. The expression on the outer surface of the heart is similar to that of flk-1 (Fig. 8D,I) and probably represents expression in the capillaries that run between the myocardium and epicardium (Viragh and Challice, 1981). The small foci of grains on the inner surface of the myocardium also appears to represent c-myc expression in endocardial cells, as they are found specifically in the vicinity of red-blood-cell-containing capillaries. c-myc mRNA was also observed in endothelial cells lining the endocardial cushion. c-myc is only expressed at low levels in the myocardium itself, in spite of the fact that when it is expressed ectopically in the myocardium it causes myocytic hyperplasia (Jackson et al., 1990). In N-myc^9a/BRP hearts, the expression of c-myc did not appear to be affected (Fig. 10E,J).

Careful examination of N-myc^9a/BRP embryos between 10.5 and 14.5 days p.c. gave no evidence of defects in the kidney or brain, both of which are major sites of N-myc expression in the embryo, and both of which are affected in N-myc^BRP/BRP embryos (Stanton et al., 1992; Charron et al., 1992). In four N-myc^9a/BRP embryos that survived until 14.5 days p.c., the kidneys were small but were structurally normal (not shown). N-myc^BRP/BRP mice have also been described as having reduced genital ridges (Stanton et al., 1992) and cranial and spinal ganglia. However, these structures were indistinguishable in N-myc^9a/BRP embryos from those of their wild-type littermates (not shown).

We have generated mice that carry two different mutant alleles of the N-myc proto-oncogene in order to identify functions for N-myc that were not revealed in mice homozygous for each mutation individually. The N-myc^9a allele is a leaky mutation which in the homozygous condition causes perinatal lethality due to a defect early in lung branching morphogenesis. The levels of residual N-Myc protein (approximately 25% of normal levels) are presumably sufficient to support the normal development of other tissues in which N-myc normally functions. The N-myc^BRP allele is likely to be a mutation which, in the homozygous condition, results in embryonic lethality at approximately 11.5 days p.c., at which point the epithelial component of the mesonephros, brain, lung, stomach and intestine are all hypoplastic (Stanton et al., 1992). The phenotype of N-myc^9a/BRP compound heterozygotes is intermediate to the two homozygous mutant phenotypes, consistent with the observation that the N-myc^9a/BRP heterozygotes contain lower levels of N-Myc protein than are present in N-myc^9a/9a embryos. N-myc^9a/BRP embryos die between 12 and 15 days p.c., and a number of tissues that are affected in N-myc^BRP/BRP homozygotes, such as the kidney, brain and cranial and spinal sensory ganglia, appear to be normal. The lungs, which are the main organ affected in N-myc^9a/9a homozygotes, are even more severely affected in N-myc^9a/BRP embryos. Compound heterozygotes also showed a defect in the development of the compact subepicardial layer of the heart, and appeared to die from a failure of heart function caused by a hypoplasia of ventricular myocytes in the compact layer. These results indicate a critical role for N-myc in the development of the heart.

The reduced levels of N-Myc protein found in N-myc^9a/BRP embryos result in a considerable thinning of the subepicardial compact layer of the myocardium by 12.5 days of development. Consistent with this phenotype, N-myc expression is expressed much more strongly in the compact layer of the heart at 10.5 and 12.5 days p.c. than in the trabecular layer of the myocardium or in the endothelium. Trabeculation, or formation of the myocardial projections that form a lattice of contractile cells throughout much of the ventricles of the embryonic heart, occurs early during heart development in the mouse, beginning around day 9.5 of gestation (Challice and Viragh, 1973). By a number of ultrastructural and cytochemical criteria, the myocytes within the trabeculae are more highly differentiated than the myocytes in the compact layer (Rumyantsev, 1991). Thus compact layer myocytes are more basophilic and richer in RNA, while trabecular myocytes contain more mitochondria, ribosomes and granular endoplasmic reticulum. Myofibrils in trabecular myocytes are thicker and more highly organized than in compact layer myocytes, where myofibrils are present but are scattered randomly relative to one another in the cytoplasm. Consistent with this picture is the observation that the rate of cell proliferation in the compact layer is 2-to 3-fold higher than in the trabeculae (Rumyantsev, 1977; Tokuyasu, 1990). The highly differentiated myocytes in the trabecular layer have been postulated to be responsible for the early beating of the heart while, at later stages, the thickened compact layer becomes the major contractile force. The cardiac hypoplasia that we observe in N-myc^9a/BRP embryos is more apparent in the compact layer. This may explain our observation that embryonic lethality does not occur until later during heart development, when heart function may depend more on the compact layer than on the trabecular layer.

Our observation of N-myc expression in the compact layer and the absence of development in the compact layer in N-myc^9a/BRP mice suggests that N-myc is required either for the proliferation of myocytes in the compact layer, and/or for preventing the differentiation of these cells, although there are other possible explanations, such as that they are dying prematurely. The simplest explanation is the former. However, the second hypothesis, in which N-myc expression prevents the terminal differentiation of compact layer myocytes into trabecular-type myocytes, is consistent with previous descriptions of N-myc expression in the embryo, in which N-myc expression is correlated with cells in an undifferentiated state in the kidney, brain and skin, regardless of their proliferative state, and the further differentiation of these cells is correlated with down-regulation of N-myc (Mugrauer et al., 1988).

In their description of embryos homozygous for a null mutation in N-myc, Charron et al. (1992) noted a defect in the development in the heart, visible as early as 9.5 days p.c., in which development was apparently slowed as evidenced by the absence of endocardial cushion tissue and of interatrial and interventricular septa. It was postulated that N-myc is involved in the generation of the inductive signal sent by the myocardium to the endocardium to induce the epithelial-to-mesenchymal transition of the endocardium, which forms these anlagen of the cardiac valves and septa. Our results confirm a function for N-myc in the development in the heart, but tend to support a role for N-myc within the myocardium itself, although this could presumably have a secondary impact on endocardial differentiation in more severe mutants. We have not observed defects in the formation of septal or endocardial cushion tissue in N-myc^9a/BRP embryos.

When c-myc is overexpressed in the heart of RSV/c-myc transgenic mice, these mice develop a fetal cardiac myocyte hyperplasia and at birth have more than twice the normal number of cardiac myocytes (Jackson et al., 1990). This has suggested that endogenous c-myc may play a role in cardiac myocyte proliferation in vivo. However, we have demon-strated that c-myc is only expressed at low levels in the heart at 12.5 days p.c. and that this expression is largely in endothelial cells and not in the myocardium. Background levels of c-myc expression in the heart were also observed in mouse embryos at 13.5 days p.c. (Stanton et al., 1992) and in first trimester human embryos (Pfeifer-Ohlsson et al., 1985). N-myc, in contrast, is expressed at high levels in the compact layer of the ventricular myocardium at 10.5 and 12.5 days. Combined with the hypoplasia of the myocardium that we have observed in N-myc^9a/BRP embryos, these data suggest that N-myc rather than c-myc may play a primary role in the regulation of proliferation and/or differentiation of ventricular myocytes during heart development, and that the effect of c-myc in these transgenic mice may reflect the possibility that in this instance c-myc can mimic the normal effects of N-myc. It has previously been observed that different myc family genes can cause the same tumour types in transgenic mice when they are overexpressed using identical promoters, even though this expression may be ectopic (Rosenbaum et al., 1989; Dildrop et al., 1989; Adams et al., 1985). Furthermore, high levels of expression of one myc gene in transgenic mice can repress transcription of itself and of other myc genes (Rosenbaum et al., 1989; Dildrop et al., 1989; Adams et al., 1985). These results have suggested that at high levels, the various myc genes may be able to mimic each other’s effects on downstream targets. This hypothesis has been strengthened by the observations that N-Myc and c-Myc proteins bind the same core DNA sequence in vitro, and that N-, L- and c-myc all form heterodimers with Max in vivo (Blackwood et al., 1992; Wenzel et al., 1991; Mukherjee et al., 1992), and all transform cells in culture through their interaction with Max (Mukherjee et al., 1992). We are presently crossing RSV/c-myc mice with the N-myc mutant mice to generate N-myc^9a/BRP embryos that also express this c-myc transgene in the heart. If c-myc in these transgenics truly mimicks the normal role of N-myc, the cardiac myocyte hypoplasia of compound heterozygous embryos is expected to be rescued by the transgene.

Recently, Davis et al. (1993) have described the phenotype of mice that bear a null mutation in c-myc. These embryos die before 10.5 days p.c. and exhibit, among other abnormalities, an enlargement of the heart and a dilated, fluid-filled pericardium. This is unexpected in light of the observation, described above, that mice with ectopic expression of c-myc in the heart have enlarged hearts. However, it is still unclear whether the heart abnormality in the c-myc null mutants is a direct result of the mutation or is secondary to other defects that are causing the embryo to die.

In spite of the possibility that c-myc expression may be able to replace N-myc function, we observe no up-regulation of c-myc in either the compact layer of the myocardium or in the lung epithelium of N-myc^9a/BRP mice. This may be because cross-regulation of the myc family genes does not normally occur in these tissues as it does when they are over-expressed in transformed cells or in transgenic mice. Stanton et al. (1992) showed that c-myc is expressed in the telencephalon of N-myc^BRP/BRP embryos, and this observation was interpreted to indicate cross regulation of N-myc and c-myc in the neuroepithelium.

Mice homozygous for the N-myc^9a mutation die at birth due to a defect in lung morphogenesis which is visible as early as 12.5 days p.c. (Moens et al., 1992). We have postulated that N-myc is required for the lung epithelium to respond to local inductive signals emanating from the lung mesenchyme, which cause branching to occur. N-myc^9a/BRP embryos have more severely affected lungs, with only a rudimentary branching pattern at 12.5 days. However, the earliest events of lung development, in which two buds are induced to grow from the trachea by surrounding mesenchyme (Spooner and Wessells, 1970), occur normally in N-myc^9a/BRP and indeed in N-myc^BRP/BRP embryos (Stanton et al., 1992). We and others (Hirning et al., 1991; Moens et al., 1992) have shown that N-myc is expressed in the lung epithelium and, further, that expression is largely restricted to bronchioles and is present at very low levels in the trachea and bronchi. These results suggest that N-myc is involved in branching morphogenesis in the lung but not in the initial induction of budding of the tracheal epithelium. Experimental manipulations in vitro have suggested that there are different mechanisms for budding versus branching of the lung epithelium. A number of different stimuli, including salivary gland mesenchyme and bronchial mesenchyme, can induce supernumerary buds in tracheal epithelium, but only bronchial mesenchyme is able to induce those buds to branch (Wessells, 1970; Spooner and Wessells, 1970). The phenotypes of mice bearing mutations in N-myc provide genetic evidence for such a mechanism for lung development and provide a candidate gene that is involved in the control of one process (branching) and not the other (budding).

Only a subset of the tissues that normally express N-myc are visibly affected in N-myc^9a/BRP embryos. N-myc^9a/9a embryos have approximately 25% of wild-type levels of N-Myc protein and, in these embryos, the lungs and spleen are the only tissues affected (Moens et al., 1992). N-myc^9a/BRP embryos have approximately 15% of wild-type levels of N-Myc protein and, in these embryos, the heart is also affected. However, 15% of normal levels of N-Myc protein appear to be sufficient for normal genitourinary and nervous system development. The molecular basis for this remains to be determined. It is possible that different tissues within the embryo make different amounts of normal protein relative to the amounts in wild-type embryos. We have previously attempted to correlate the relative levels of normal N-myc mRNA in different tissues of N-myc^9a/9a embryos with the presence or absence of a mutant phenotype (Moens et al., 1992) and, although there were differences among the tissues examined, no strong correlation could be established. Another explanation for the tissue-specific effects of N-myc mutant alleles is that different tissues are affected differently by approximately the same reduction in N-Myc protein because of differences in the ratio of N-Myc protein to Max (Blackwood and Eisenman, 1991), to Max-associated proteins such as Mad and Mxi1 (Ayer et al., 1993; Zervos et al., 1993), or to other Myc proteins.

Myc proteins require dimerization with Max for DNA binding (Blackwood and Eisenman, 1991; Prendergast and Ziff, 1991; Kato et al., 1992). A number of lines of evidence have suggested that Max overexpression can inhibit transformation (Mukherjee et al., 1992; Makela et al., 1992; Prendergast et al., 1992) and transactivation (Kretzner et al., 1992; Amati et al., 1992) by myc genes. Mad, cloned by virtue of its ability to dimerize with Max, has been shown to compete with Myc for binding to Max, and to thereby inhibit transactivation by Myc (Ayer et al., 1993). Mxi1 (Zervos et al., 1993) and other, as yet unidentified, Max partners presumably act in a similar manner and are also likely thereby to inhibit N-myc function. Also, L-Myc, a poorly transforming member of the Myc family, has been shown to prevent transformation by other myc genes, presumably by competing for and forming less active DNA-bound complexes with Max (Mukherjee et al., 1992). In cell types where a number of Max-associated and Myc proteins compete with N-Myc protein for dimerization with Max and sites on DNA, an 85% reduction in N-Myc protein is expected to reduce the response of downstream targets of N-Myc more strongly than in cells where there are no competitors for Max binding. Interestingly, L-myc is co-expressed with N-myc in the lung, perhaps in the same cell type (Zimmerman et al., 1986), but the two genes are not co-expressed in the kidney (Mugrauer and Ekblom, 1991), where neither the N-myc^9a/9a nor the N-myc^9a/BRP embryos have an abnormal phenotype. It will be interesting to compare the detailed expression patterns of the various interacting factors with the tissues affected by the N-myc mutations.

We have generated a third N-myc mutant phenotype by combining leaky and null alleles in a compound heterozygote. These mice have allowed us to study the function of N-myc at a stage in development that is not reached in embryos homozygous for the null allele (Stanton et al., 1992) and that is not affected in embryos homozygous for the leaky allele (Moens et al., 1992). Classical genetic studies of development in a number of systems have demonstrated the importance of studying the phenotypic effects of different mutant alleles and combinations of mutant alleles in a given gene in order to determine its multiple roles in the course of development. Our results have shown that the technique of gene targeting by homologous recombination in the mouse can be used to the same ends. The clear delineation of a function for N-myc in both lung branching morphogenesis and myocardial development also provides target tissues in which to search for the elusive downstream genes in the N-myc signaling pathway.

Recently, a third description of embryos homozygous for a putative null allele of N-myc has been published (Sawai et al., 1993). The overall phenotype of these mutant embryos is very similar to those described by Stanton et al. (1992) and Charron et al. (1992), but the defect in heart development is shown to be primarily in the myocardium.

We wish to thank Dr R. Kennett for the anti-N-myc antibody, Dr M. Buckingham for the α-cardiac actin probe, T. Yamaguchi for the flk-1 probe, Dr C. Asselin for the c-myc probe and Dr R. DePinho for the N-myc genomic clone from which probes for RNA in situ hybridization were subcloned. Our gratitude also to Dr Arch Perkins for his help in the phenotypic analysis of N-myc^9a/BRP mutants, to Alexandra Joyner for her critical reading of the manuscript, and to Valerie Prideaux, Chi-Chong Hui, Benny Motro and Ester Ivanyi for their generous assistance in various aspects of this work. This work was supported by a Terry Fox program project grant from the National Cancer Institute of Canada. C. B. M. was supported by a Natural Sciences and Engineering Research Council of Canada ‘Centennial’ Scholarship and a Medical Research Council of Canada Studentship. J. R. is an International Scholar of the Howard Hughes Medical Institute and a Terry Fox Cancer Research Scientist of the National Cancer Institute of Canada. B. R. S. and L. F. P. are sponsored by the NCI-DHHS under contract NO1-CO-74101 with ABL.