Heart development in fibronectin-null mice is governed by a genetic modifier on chromosome four.

Abstract

Absence of the fibronectin (FN) gene leads to early embryonic lethality in both 129S4 and C57BL/6J strains due to severe cardiovascular defects. However, heart development is arrested at different stages in these embryos depending on the genetic background. In the majority of 129S4 FN-null embryos, heart progenitors remain at their anterior bilateral positions and fail to fuse at the midline to form a heart tube. However, on the C57BL/6J genetic background, cardiac development progresses further and results in a centrally positioned and looped heart. To find factor(s) involved in embryonic heart formation and governing the extent of heart development in FN-null embryos in 129S4 and C57BL/6J strains, we performed genetic mapping and haplotype analyses. These analyses lead to identification of a significant linkage to a 1-Mbp interval on chromosome four. Microarray analysis and sequencing identified 21 genes in this region, including five that are differentially expressed between the strains, as potential modifiers. Since none of these genes was previously known to play a role in heart development, one or more of them is likely to be a novel modifier affecting cardiac development. Identification of the modifier would significantly enhance our understanding of the molecular underpinning of heart development and disease.

Full text

Page 1

Heart development in fibronectin-null mice is governed
by a genetic modifier on chromosome four
Sophie Astrof
a,1, Andrew Kirby
Mark Daly
b, Kerstin Lindblad-Toh
b,c, Richard O. Hynes
c,
a,*
aHoward Hughes Medical Institute, Center for Cancer Research, Department of Biology, Massachusetts Institute of Technology,
77 Massachusetts Avenue, Cambridge, MA 02139, USA
bCenter for Human Genetic Research, Massachusetts General Hospital, Boston, MA, USA
cBroad Institute of Harvard and MIT, Cambridge, MA, USA
Received 21 February 2007; received in revised form 16 May 2007; accepted 28 May 2007
Available online 2 June 2007
Abstract
Absence of the fibronectin (FN) gene leads to early embryonic lethality in both 129S4 and C57BL/6J strains due to severe cardiovas-
cular defects. However, heart development is arrested at different stages in these embryos depending on the genetic background. In the
majority of 129S4 FN-null embryos, heart progenitors remain at their anterior bilateral positions and fail to fuse at the midline to form a
heart tube. However, on the C57BL/6J genetic background, cardiac development progresses further and results in a centrally positioned
and looped heart. To find factor(s) involved in embryonic heart formation and governing the extent of heart development in FN-null
embryos in 129S4 and C57BL/6J strains, we performed genetic mapping and haplotype analyses. These analyses lead to identification
of a significant linkage to a 1-Mbp interval on chromosome four. Microarray analysis and sequencing identified 21 genes in this region,
including five that are differentially expressed between the strains, as potential modifiers. Since none of these genes was previously known
to play a role in heart development, one or more of them is likely to be a novel modifier affecting cardiac development. Identification of
the modifier would significantly enhance our understanding of the molecular underpinning of heart development and disease.
? 2007 Elsevier Ireland Ltd. All rights reserved.
Keywords: Fibronectin; Heart development; Genetic modifier; Mapping
1. Introduction
The heart is the first organ to form in the developing
vertebrate embryo and the survival of the embryo and
the well-being of the adult are critically dependent on the
integrity of the developmental program guiding heart for-
mation (Srivastava, 2006). One of the earliest events during
cardiogenesis is the coalescence of the bilateral cardiac pri-
mordia toward the midline to form a single heart tube
(Buckingham et al., 2005) – the failure of this to happen
leads to cardia bifida and lethality.
Genetic analyses in zebrafish and mice have uncovered a
number of genes and processes required for the formation
of a single heart tube from the paired cardiac primordia.
First, cardiac mesoderm is derived from the early wave of
cells migrating through the primitive streak, and the dele-
tion of genes that affect this process (Mesp1 and Nap1)
gives rise to cardia bifida (Rakeman and Anderson, 2006;
Saga et al., 1999). Second, endoderm provides an essential
signal for the migration of the cardiac lateral plate meso-
derm toward the midline, and genes required for endoderm
specification (cas, bon, oep) (Alexander et al., 1999; Kiku-
chi et al., 2000; Schier et al., 1997) or morphogenesis
(fau, GATA4, Hrs, Furin, Foxp4) (Komada and Soriano,
0925-4773/$ - see front matter ? 2007 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.mod.2007.05.004
*Corresponding author. Tel.: +1 617 253 6422; fax: +1 617 253 8357.
E-mail address: rohynes@mit.edu (R.O. Hynes).
1Present address: Center for Molecular Cardiology, Weill Medical
College of Cornell University, New York, NY, USA.
www.elsevier.com/locate/modo
Mechanisms of Development 124 (2007) 551–558

Page 2

1999; Kuo et al., 1997; Li et al., 2004; Reiter et al., 1999;
Roebroek et al., 1998) are required for the formation of
a central heart tube. Hypoxia-regulated processes requiring
HIF1a are also involved. Finally, genes important for cell
migration (paxillin, fibronectin, S1P receptor, NAP1)
(George et al., 1993; Hagel et al., 2002; Kupperman
et al., 2000; Rakeman and Anderson, 2006), cell polarity
(e.g. fibronectin, NAP1) (Rakeman and Anderson, 2006;
Trinh and Stainier, 2004), extracellular matrix synthesis
(mtx1) (Sakaguchi et al., 2006), or protein prenylation
(hmgcr1b) (D’Amico et al., 2007) are essential for the for-
mation of a centrally located heart tube.
Interestingly, the genetic makeup of laboratory mouse
strains affects the phenotypes of mice carrying various
knockout alleles. The most dramatic variations, from
embryonic lethality in one strain to survival past birth in
a different strain, are seen in mice carrying deletions in
genes encoding an Rb family member p130 or EGF recep-
tor (LeCouter et al., 1998; Threadgill et al., 1995). Heart
function is also affected by strain-specific genetic variation.
For example, adult wild-type mice of C57BL/6J strain pos-
sess the ‘‘heart of an athlete’’ compared with the hearts of
A/J mice (Hoit et al., 2002). Genetic differences between
strains also affect the degree of cardiac defects in some
knockout mice. A deletion in the HIF1a gene results in car-
dia bifida in 32% of the embryos on a mixed 129X1-Swiss
background while a central looped heart develops in all
embryos from a mixed 129Sv/J · C57BL/6J strain (Compe-
rnolle et al., 2003; Iyer et al., 1998).
FN-null embryos have cardia bifida in the 129S4 strain of
mice but develop a central looped heart in the C57BL/6J
strain (George et al., 1997). Interestingly, the point mutant
natter in zebrafish, which introduces a translational stop
codon at the beginning of the FN1 message and abrogates
the expression of fibronectin, also shows a strain-depen-
dent cardia bifida phenotype (Trinh and Stainier, 2004).
These phenomena suggest that there exist genetic factors
whose identity and importance for heart development
could be uncovered by performing genetic mapping analy-
sis. Identification of new genes involved in early cardiogen-
esis may also facilitate our understanding of congenital
heart defects. So far, mutations in a number of genes essen-
tial for embryonic heart development (e.g. GATA4,
Notch1, Nkx2.5, Tbx1, Tbx5) have been found to underlie
familial cases of congenital heart disease (Garg et al., 2003,
2005; Li et al., 1997; Merscher et al., 2001; Schott et al.,
1998).
To identify new genes involved in heart development, we
decided to map factor(s) governing heart formation in the
FN-null embryos derived from F2 intercrosses between
129S4 and C57BL/6J strains. We used high-throughput
genotyping of single-nucleotide polymorphisms (SNPs)
and a dense SNP haplotype map of 129S4 and C57BL/6J
strains (Wade and Daly, 2005) to identify a 1-Mbp interval
on mouse chromosome four containing a modifier of
cardiogenesis in FN-null embryos. Furthermore, by per-
forming microarray analysis on wild-type, heterozygous,
and FN-null embryos, we identified 21 candidate genes
expressed in this interval, five of which were differentially
expressed between 129S4 and C57BL/6J strains. This is
the first example, to our knowledge, of mapping a quanti-
tative trait locus of an embryonic lethal phenotype.
2. Results and discussion
2.1. Phenotypic analysis of FN-null embryos
All FN-null embryos from the C57BL/6J background
developed a centrally located and looped heart by e9.3
(Fig. 1c and f). However, in 72% of 129S4 FN-null
embryos, heart development stopped following the migra-
tion of cardiac primordia to their anterior–lateral positions
(Fig. 1a and d). In the remaining 129S4 FN-null embryos,
the heart primordia migrated and fused to form a central
heart ball (Fig. 1b and e); however, none of the 129S4
FN-null embryos developed a looped heart tube (Table
S1). Analysis of 62 FN-null embryos from F1 crosses
between 129S4 and C57BL/6J, showed that, similar to
129S4 embryos, the largest group among the FN-null F1
embryos (37 out of 62) had bilateral cardiac primordia.
2.2. Genetic mapping
To perform genetic mapping analyses, we collected two
independent sets of FN-null embryos from F2 intercrosses
(there were 175 and 165 embryos in each set) and classified
them into three groups according to their heart phenotypes
(Fig. 1). Group 1 comprised embryos with unfused bilat-
eral heart primordia, group 2 included embryos with a
small- or medium-sized heart in the shape of a ball, and
group 3 comprised embryos that developed a looped heart.
Both sets of F2 embryos had similar distributions of phe-
notypes (Table S2). Each embryo was then genotyped
using a genome-wide panel of SNP markers spaced about
20 cM apart on each chromosome. Each set of embryos
was genotyped with a different set of genetic markers.
These genome-wide scans indicated a potential linkage to
the distal arm of chromosome four (Fig. 2a). We per-
formed linkage analysis on the first and second sets of F2
embryos independently, and observed linkage to chromo-
some four in similar locations in each case (Fig. S1). An
additional panel of SNP markers placed on chromosome
four allowed us to narrow down the position of the modi-
fier to a 5-Mbp interval (Fig. 2a and b). Other genomic
regions did not show significant linkage (Fig. 2a). Further-
more, we did not detect genetic interactions between the
locus on chromosome four and other chromosomal loci
with LOD scores above 2.5 (Table S3).
We reasoned that the genetic modifier on chromosome
four could affect either of two processes: migration of the
lateral primordia to the midline or subsequent morphogen-
esis of the heart tube. In an attempt to distinguish between
these two possibilities, we performed linkage analysis using
different combinations of phenotypic groups. For example,
552
S. Astrof et al. / Mechanisms of Development 124 (2007) 551–558

Page 3

to determine whether the modifier on chromosome four
affected the progression from the bilateral cardiac primor-
dia stage to a central heart, we analyzed embryos from
group 1 against embryos from combined groups 2 and 3.
While our analysis showed that the peak on chromosome
four was mainly due to the genetic differences between
groups 1 and 3 (Table S4), we obtained the highest LOD
score, 5.5, when the three phenotypic classes were com-
bined into two groups, one group contained embryos with
unfused lateral heart primordia (group 1), while the other
group consisted of embryos in which fusion of the primor-
dia occurred (groups 2 and 3). This suggested that we were
mapping a modifier affecting the event of coalescence of the
two heart primordia into a single heart tube, a migration
process that ordinarily happens during normal heart devel-
opment (Buckingham et al., 2005).
The peak on chromosome four spanned five megabases
(Mbp) (Fig. 2b) between 134 and 140 Mbp (2004 genome
assembly). The LOD score of 5.5 corresponds to a point-
wise (two-sided) p value of 2.5 · 10?6and a genome-wide
significance level of 0.003 (0.003 is the number of times that
a linkage of LOD 5.5 could have been found fortuitously).
These values comprise significant evidence (Lander and
Kruglyak, 1995) that this region of chromosome four is
linked to a gene(s) affecting heart development in FN-null
embryos, and we term this locus modifier influencing single
heart assembly (Misha-1).
The highest predicted LOD score (5.5) within the peak is
2 cM away from the closest genotyped marker, w4_13753,
at 138 Mbp (Fig. 2b). Those embryos having two C57BL/
6J alleles at the w4_13753 locus are approximately twofold
more likely to develop a centrally located heart than to
arrest at the bilateral primordia stage, and embryos homo-
zygous or heterozygous for the 129S4 allele at w4_13735
have a 50/50 chance either to develop a central heart or
to arrest at the bilateral primordia stage (Table S5). This
suggests that, while a locus within the 5-Mbp region on
chromosome four contains the major modifier of FN-null
phenotype, there exist other quantitative trait loci, which
contribute to strain-specific differences in heart develop-
ment of FN-null mice.
While our experiments suggested that the locus on
chromosome four contains one or more modifiers affect-
ing the coalescence of the bilateral primordia into a sin-
gle heart tube, the heart phenotype is not the only
difference between these two groups, another notable
a
Group1
A, V
P
Anterior View
c
Group 3
A
P
Lateral View
V
b
Group 2
A, V
P
Dorsal View
def
Fig. 1. Phenotypes of FN-null embryos. (a and d) A whole-mount view and a transverse section through an embryo representative of the majority of FN-
null embryos from the 129S4 strain and of F2 embryos belonging to group 1. Arrows point to the two cardiac primordia, arrowheads point to the head
folds. Anterior (A) and ventral (V) are at the bottom and posterior (P) is at the top. (b and e) A whole-mount view and a transverse section through a
representative F2 embryo from group 2. Arrow points to the single heart ball outlined in B and encompassed by the bracket (e). Arrowheads point to
headfolds. Anterior is to the right in (b) and at the bottom in (e). (c and f) A whole-mount view and a sagittal section through an embryo representative of
FN-null embryos from C57BL/6J strain and F2 embryos belonging to group 3. Anterior is to the right in (c) and at the top right corner in (f). Arrow points
to the central heart which is also delineated by a bracket in (f). Arrowhead points to the head. Scale bars in (d, e and f) are 65 lm.
S. Astrof et al. / Mechanisms of Development 124 (2007) 551–558
553

Page 4

difference is the presence of vessels in embryos with at
least a central heart ball, in contrast to those embryos
without a heart (George et al., 1997). Therefore it is pos-
sible that the modifier locus affects an earlier stage in
embryogenesis, correlating with the extent of the heart
development. In order to understand the role of this
locus in embryogenesis and heart formation, we need
to identify the gene within it that acts as modifier.
2.3. Further analyses of the region
To narrow down the region containing the candidate
modifier(s), we determined those areas within the 5-Mbp
interval that contain polymorphisms between 129S4 and
C57BL/6J strains, using a mouse SNP chip (Wade and
Daly, 2005). This experiment showed that none of the
298 genotyped markers located between 134.8 and
139.1 Mbp was polymorphic between 129S4 and C57BL/6J
strains (http://www.broad.mit.edu/?mjdaly/mousehapmap).
Fig. 3. Gene expression analysis of wild-type, heterozygous and FN-null
embryos from 129S4 and C57BL/6J strains. Hierarchical unsupervised
clustering of gene expression profiles distinguishes embryos from 129S4
and C57BL/6J strains and they cluster according to the number of wild-
type FN alleles and according to genetic background.
0
1
2
3
4
5
6
118126134.7138.2 140.1
Mbp
152.7
LOD
a
b
Fig. 2. A genetic modifier of heart development in FN-null mice is linked to the distal arm of chromosome four. (a) Genomescan of 341 F2 samples
identified a linkage peak on chromosome four. For each chromosome, positions of the SNP markers in cM are plotted on the x-axis and LOD scores are
plotted on the y-axis. Penetrance scan is shown, representing analysis of embryos from group 1 against embryos from combined groups two and three.
Peaks on chromosomes eight and seventeen are not reproducible between the two sets of embryos and probably result from combining two different sets of
markers used for the genome-wide scan. Chromosome four was genotyped with a single set of markers for all of the 341 embryo samples. (b) Linkage peak
and haplotype map of the region on chromosome four. Embryos were pooled into two phenotypic groups (group 1 vs. groups 2 and 3) and analyzed using
penetrance scan. Red marks and red points on the graph indicate positions of genotyped SNPs. Blue points on the curve show the points calculated by the
MapmakerQTL. Blue bar – haplotype block shared between 129X1, 129S1, 129S4 and C57BL/6J strains based on the genotyping of 298 SNPs in this
region (see Section 3). Marker w4_13753 at 138.2 Mb is the only polymorphism detected in this region. Red bars – regions containing SNPs between the
129S4 and C57BL/6J strains.
554
S. Astrof et al. / Mechanisms of Development 124 (2007) 551–558

Page 5

Clearly this is a conserved haplotype block shared between
thetwostrains.Weresequenced
polymorphisms in this region – only one, w4_13753, was
a true polymorphism and is the only known polymorphic
severalputative
marker in this region. We initially hoped that w4_13753
might help identify the modifier. However, there are no
known or predicted genes in the near vicinity of
w4_13753. The closest gene to the left of this marker
Padi4
Necap2
D4Ertd22e
RefSeqGenes
Arhgef19
Clcnka
Fblim1
BC025833
139, 500,000 140, 000,000
Rcc2
Padi6
Padi3
Padi1
Padi2
Sdhb
Atp13a2
Mfap2
Crocc
Spata21
D4Ertd22e
Fbxo42
Epha2
Clcnkb
Hspb7
Gm694
Zbtb17
Spen
B330016D10Rik
x
x
x
x
x
ENMUST00000013713.1
x Not Expressed
x No coding mutations
xxDifferentially expressed between 129S4 and C57BL/6J strains
Missense mutations
Expressed in the heart, e10.5
Fig. 4. Candidate modifier genes. There are 26 genes annotated in this 1-Mbp region of chromosome four. Twenty-five of them are annotated by RefSeq
and are shown in the figure; the expression of one additional gene, ENMUST00000013713.1, is detected by the Affymetrix probe 1459714 at 139.67 Mbp.
Genes expressed in the hearts of e10.5 embryos were found by mining the Entrez GEO database. Missense mutations were found by using the Perlegen
database and by direct sequencing of cDNAs (see Section 3 and Table 1). Genes not expressed in the region were determined using Affymetrix absence and
presence calls.
Table 1
Candidate modifier genes – expression and polymorphisms
Affymetrix
probe No.
Gene name/symbol Expressed at
e8.0
Differentially
expressed
Expressed in the heart
at e10.5
Missense mutations
1
2
3
4
5
6
7
8
9
1426897_at
1437084_at
1422760_at
1419767_at
1419323_at
1418252_at
1418005_at
1428340_at
1417359_at
1427338_at
1418961_at
1455718_at
1434482_at
1453382_at
1459714_at
RCC2
PADi6
PADi4
PADi3
PADi1
PADi2
Sdhb*
ATP13a2
Mfap2
Rootletin
NECAP2
Spata21
D4Ertd22e
Fbxo42
ENMUST
00000013713.1*
ArhGEF19*
EphA2
Gm693
Clcnka
Clcnkb
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
No
No
No
No
No
Up in 129S4
No
No
No
No
No
No
Up in C57BL/6J
No
No
Yes
Yes
No
Yes
Yes
Yes
Yes
No
Yes
No
Yes
Yes
Yes
NS
NS
A573V
NS
NS
NS
None*
NS
NS
T289A, E360G, V811M, S845N
I104L
NS
NS
NS
L40I, C106R, L168I, A107E*
10
11
12
13
14
15
16
17
18
19
20
1437629_at
1421151_at
1457076_at
1455677_at
1450340_at
Yes
Yes
Yes
No
No
No
No
No
No
No
G492S*
NS
NS
NS
A380V, I431V, A554T, V571M,
L637P, S667F
NS
NS
NS
NS
I159R*
NS
Up in C57BL/6J
No
No
No
21
22
23
24
25
26
1421290_at
1416224_at
1420397_at
1455492_at
1418569_at
1451528_at
Hspb7
Zbtb17
Spen
B330016D10Rik
Fblim1*
BC025833
Yes
Yes
Yes
Yes
Yes
No
No
No
No
Up in C57BL/6J
Up in C57BL/6J
No
Yes
Yes
Yes
No
Yes
No
There are 26 genes annotated in the 1-Mbp region of chromosome four. Microarray analysis showed that 22 of these genes are expressed in e8.0 embryos
and five of these genes are differentially expressed between the strains. Genes expressed in the hearts of e10.5 embryos were found by browsing Entrez GEO
database. Missense mutations (C57BL/6J –> 129S4) were found by mining the Perlegen database and by direct sequencing. NS, not sequenced; these genes
did not contain coding SNPs in the Perlegen database. Asterisks denote genes whose cDNAs were sequenced at 8· coverage from 129S4 and C57BL/6J
strains.
S. Astrof et al. / Mechanisms of Development 124 (2007) 551–558
555