pq010000268p


A human model for multigenic inheritance: Phenotypic
expression in Hirschsprung disease requires both the
RET gene and a new 9q31 locus
Stacey Bolk*, Anna Pelet†, Robert M. W. Hofstra‡, Misha Angrist*, Remi Salomon†, David Croaker§,
Charles H. C. M. Buys‡, Stanislas Lyonnet†, and Aravinda Chakravarti*¶

*Department of Genetics and Center for Human Genetics, Case Western Reserve University School of Medicine and University Hospitals of Cleveland,
Cleveland OH 44106; †Department of Genetics and Unite Institut National de la Santé et de la Recherche Médicale, U-393, Hopital des Enfants Malades,
75743 Paris, France; ‡Department of Medical Genetics, University of Groningen, Groningen, The Netherlands; and §Department of Surgical Research,
New Children’s Hospital, Westmead, Australia

Communicated by Victor A. McKusick, Johns Hopkins University School of Medicine, Baltimore, MD, October 14, 1999 (received for review May 10, 1999)

Reduced penetrance in genetic disorders may be either dependent or
independent of the genetic background of gene carriers. Hirsch-
sprung disease (HSCR) demonstrates a complex pattern of inheritance
with '50% of familial cases being heterozygous for mutations in the
receptor tyrosine kinase RET. Even when identified, the penetrance of
RET mutations is only 50 –70%, gender-dependent, and varies with
the extent of aganglionosis. We searched for additional susceptibility
genes which, in conjunction with RET, lead to phenotypic expression
by studying 12 multiplex HSCR families. Haplotype analysis and
extensive mutation screening demonstrated three types of families:
six families harboring severe RET mutations (group I); and the six
remaining families, five of which are RET-linked families with no
sequence alterations and one RET-unlinked family (group II). Al-
though the presence of RET mutations in group I families is sufficient
to explain HSCR inheritance, a genome scan reveals a new suscepti-
bility locus on 9q31 exclusively in group II families. As such, the gene
at 9q31 is a modifier of HSCR penetrance. These observations imply
that identification of new susceptibility factors in a complex disease
may depend on classification of families by mutational type at known
susceptibility genes.

A major hallmark of phenotypes displaying a complex pattern ofinheritance is the segregation of multiple factors. In principle,
linkage analysis on a collection of families can identify multiple loci,
although an individual family may segregate only a few of the
relevant genes. Multilocus segregation is compatible with two
distinct hypotheses. Thus, a phenotype could be multigenic or
multifactorial, resulting from the cumulative effects of mutations in
multiple genes; disease expression and penetrance may then be
sensitive to an individual’s genetic background. This hypothesis is in
contrast to the simple association of a disorder with multiple genes,
genetic (locus) heterogeneity, in which mutations in any one of
several genes are sufficient for phenotypic expression and are
independent of genetic background. In either scenario, linkage
heterogeneity exists, but, for the multigenic hypothesis to prevail,
we need to demonstrate multiple gene segregation in the same
families (1). Although unequivocal evidence for multigenic inher-
itance exists in the many rodent models of human disease, and is
strongly suspected in complex human disorders, it remains largely
unproven. However, in the human, in rare instances, retinitis
pigmentosa can arise from digenic inheritance (2); segregation of
insulin-dependent diabetes mellitus depends on interactions be-
tween both HLA and an X-linked gene (3).

Hirschsprung disease (HSCR), or congenital aganglionosis, is a
model complex genetic disorder in which we can test these hypoth-
eses. HSCR is associated with the lack of intrinsic ganglion cells in
the myenteric and submucosal plexuses of the distal gastrointestinal
tract. The phenotype has '100% heritability (4) but shows variable
presentation and reduced penetrance, even in multiplex kindreds,
with affected individuals differing in the length of aganglionosis,
association with pigmentary anomalies and sensorineural deafness,

and severity of intestinal blockage. Segregation analyses (4) of
HSCR have established two genetic hallmarks: (i) recurrence risks
are ' 3–25%, with a non-Mendelian pattern of inheritance; and (ii)
recurrence risk to relatives varies with length of aganglionosis and
gender of the proband. Molecular genetic analyses have identified
five functionally dependent HSCR susceptibility genes in pathways
crucial for the normal development of the enteric nervous system
(5–18).

The major HSCR gene is the receptor tyrosine kinase RET, in
which coding sequence mutations can be identified in 50% of
familial and 35% of sporadic cases (19–21). The functional ligand
for RET is the type b transforming growth factor-like glial cell
line-derived neurotrophic factor (GDNF), which activates RET
through a three-component complex by using the glycosylphos-
phatidylinositol-linked anchor protein GFRA1 (glial cell line-
derived neurotrophic factor-family receptor) (8, 9). Mutations in
GDNF are rare in HSCR (10–13); no mutations in GFRA1 have
been revealed on extensive testing (14). A second biochemical
pathway compromised in HSCR is encoded by the G-protein
coupled endothelin-B receptor EDNRB (15) and its ligand endo-
thelin-3 (EDN3); '5% of familial and sporadic HSCR cases are
explained by mutations in these genes (16). Finally, mutations in the
gene encoding the transcription factor SOX10, which regulates
EDNRB expression (17), have also been identified in multiple
patients with HSCR and pigmentary anomalies (Shah-Waarden-
burg syndrome) (17, 18).

In addition to finding mutations in single genes, we and others
have presented preliminary evidence of multigenic inheritance in
HSCR. In a large inbred Mennonite kindred segregating HSCR, we
identified a hypomorphic change (W276C) in EDNRB on chromo-
some 13q22 and demonstrated significant nonrandom association
between HSCR and polymorphisms within the RET gene (15, 16).
In addition, the penetrance of HSCR in this kindred depends on
genetic interaction between RET and EDNRB because parent-to-
offspring transmission of W276C is significantly associated with
transmission of multiple polymorphisms within RET (ref. 16; M.
Carrasquillo and A.C., unpublished data). Extensive sequence
analysis of the RET coding sequence has failed to identify a
mutation in the Mennonites. Nevertheless, the strong association
between HSCR and RET, and association between the joint trans-
mission of RET and EDNRB to patients, emphasizes that EDNRB
in conjunction with RET determines HSCR susceptibility in this

Abbreviations: HSCR, Hirschsprung disease; NPL, nonparametric linkage; RT, reverse tran-
scription; cM, centimorgans.

¶To whom reprint requests should be addressed at: Department of Genetics BRB 721, Case
Western Reserve University, 10900 Euclid Avenue, Cleveland, OH 44106-4955. E-mail:
axc39@po.cwru.edu.

The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
§1734 solely to indicate this fact.

268 –273 u PNAS u January 4, 2000 u vol. 97 u no. 1

D
o
w

n
lo

a
d
e
d
 a

t 
C

a
rn

e
g
ie

 M
e
llo

n
 U

n
iv

e
rs

ity
 o

n
 A

p
ri
l 5

, 
2
0
2
1
 



kindred and that the RET mutation is located in the non-coding
regions. Additionally, in some outbred families, GDNF mutations
do not appear to act alone but co-occur with mutations in RET or
trisomy 21 (10, 11). These data underscore the complex inheritance
and provide the first clues that multiple genes may be required to
modulate the expression of HSCR.

To test whether HSCR results from multigene segregation in
outbred families, we have focused attention on 12 families showing
vertical transmission. In such families, few segregating loci are
expected a priori because they largely, not exclusively, segregate
long-segment HSCR and have higher penetrance than short-
segment HSCR (4, 21). We chose families with three or more
affecteds across two or more generations with phenotypic charac-
teristics typical of other reported large families. The variable
expressivity and reduced penetrance in these families may be
explained by environmental effects, stochastic developmental
events, or multigenic inheritance. To distinguish between these
hypotheses, we conducted allele-sharing tests and linkage analysis
to identify multiple segregating genomic regions, followed by
mutation analysis of known genes. We present data that demon-
strate the segregation of RET in the majority (11y12) of families and
additional segregation of a new locus at 9q31. Mutation analysis
identified severe RET mutations in only 50% of families. Impor-
tantly, 9q31 segregation was limited to those six families lacking
coding sequence missense or nonsense RET mutations. Thus, the
nature of the RET mutation in HSCR patients is a strong predictor
of the presence (segregation) of additional genetic changes, sup-
porting multigenic inheritance. A corollary of these observations is
that HSCR patients heterozygous for multiple mutations have weak
(hypomorphic) changes at RET and, thus, explains the reduced RET
penetrance.

Materials and Methods
Family Ascertainment. Multiplex families segregating HSCR were
ascertained under informed consent and were classified into those
with affected siblings only and those with three or more affected
individuals in two or more generations. The latter families show
vertical transmission of HSCR and have higher penetrance. We
identified 12 families for this study, of which families 1– 6 and 8–10
have been reported previously (21, 22); the Mennonite families (15)
are not a part of this study. Genomic DNA was extracted and
lymphoblastoid cell lines were established from participating mem-
bers as described (22).

Linkage Analysis. Polymorphic microsatellite markers were geno-
typed by using standard methods (22), and allele sizes were
determined by using Centre d’Etude du Polymorphisme Humain
(CEPH) individual 1347-02 as a genotyping standard. Parametric
linkage analysis is powerful, but the major hindrance is the lack of
an accurate inheritance model (23) and concomitant reduction of
power of detecting linkage when parameters are specified incor-
rectly. One cannot a priori predict the number of different loci, the
mutant allele frequency at each locus, and the penetrance associ-
ated with each HSCR susceptibility allele. We, therefore, relied on
model-free tests of allele-sharing at marker loci among all affecteds
within each family and searched for regions in which allele-sharing
exceeds Mendelian expectations. All multipoint parametric and
nonparametric linkage analyses were performed by using the
computer program GENEHUNTER 1.1 (24). The genetic markers
used for linkage analysis had intermarker map distances as pub-
lished (25); allele frequencies from CEPH families were used for
these markers and were not different from those estimated from
founders in the 12 families. For lod score analysis, we specified an
autosomal-dominant mode of inheritance, a mutant allele fre-
quency of 1%, and penetrance of 58% in males and 29% in females;
these estimates are based on our previous segregation analysis (22).
Linkage analyses were also performed allowing for heterogeneity,
by using the ‘‘Total Heterogeneity’’ command in GENEHUNTER; we

calculated both a heterogeneity lod score and estimated the pro-
portion of linked families (a). To compute the posterior probability
of linkage on chromosome 10, we placed the disease locus at the
RET intron 5 marker and used Eq. 9.11 in ref. 23; linkage was
declared whenever the posterior probability exceeded the estimated
a. Nonparametric analysis used the nonparametric linkage (NPL)
score and the ‘‘sharing all’’ option.

Mutation Detection. Mutation analysis was performed on all 21 exons
of RET comprising 3,474 bp and 1,700 bp of adjacent intronic
sequence by using published human cDNA and partial genomic
sequences (26–29) as wild-type standards. Single-strand conforma-
tional polymorphism, denaturing gradient gel electrophoresis, and
automated nucleotide sequencing using dye terminator chemistry
were performed by standard methods (6, 14, 21).

RNA Isolation and Reverse Transcription (RT)–PCR. Total RNA was
extracted from lymphoblastoid cell lines by using RNAzolB
(CinnayBiotecx Laboratories, Friendswood, TX); extractions were
performed twice to help reduce DNA contamination. RNA (5 mg)
was reverse transcribed by using SuperscriptII (GIBCOyBRL) with
random hexamer primers, and 3 ml of the RT reaction was used for
subsequent 50-ml PCR amplifications. Primers for specific ampli-
fication of RET exon 11 were designed such that forward and
reverse oligonucleotides were in exon 10 and 12, respectively, so that
the product would span introns and control for genomic DNA
contamination. The primers used were RETX10F: 59-TTCCCT-
GAGGAGGAGAAGTGC-39 and RETX12R: 59-CCACTTTTC-
CAAATTCGCCTT-39; these were used in amplification reactions
as follows: 30 s at 94°C; 30 s at 59°C; and 30 s at 72°C for 35 cycles.

Results
RET Is the Major Gene for HSCR. Human chromosome 10 is the site of
two HSCR genes, RET at 10q11 (19, 20) and GFRA1 at 10q25 (14).
We searched for linkage in 12 multiplex families segregating HSCR
(Fig. 1) by using 22 polymorphic microsatellite markers spanning
the entire genetic length of chromosome 10 and at a resolution of
8 centimorgans (cM) (Fig. 2a). We performed both parametric and
nonparametric multipoint linkage analysis using the computer
program GENEHUNTER (24). We detected significant linkage with
the RET locus with a nonparametric linkage (NPL) score of 11.3
(P 5 7.7 3 1028) and a lod score of 9.4 (Table 2; Fig. 2a). No linkage
was evident at the GFRA1 locus consistent with our previous
observations (14). Haplotype analysis using markers flanking and
within RET demonstrated that all families except 12 were linked to
RET (Table 1). Consequently, RET is the major segregating factor
in multigenerational HSCR families, but additional genes must also
exist. We demonstrated significant linkage heterogeneity (x2 5 5.1,
P 5 0.023) and obtained a multipoint heterogeneity lod score of
10.5 at RET with a 5 91% of families linked (23). Calculations of
the posterior probability of linkage for each family showed that 11
of the 12 families have a .97% probability of being linked to RET
(Table 1). There was no evidence for a common RET haplotype in
the linked families.

Diverse RET Mutations in HSCR. To determine the nature of the
mutations in RET, we conducted extensive mutation analysis of all
probands in the linked families (Table 1). We identified eight
putative mutations distributed across the RET gene in the 11 linked
families relative to reference sequence data (26–29) and studies of
control individuals: four probands had missense changes (N394K,
R360W, R77C, and S32L in families 1–4, respectively), two pro-
bands had nonsense mutations (W942X in family 5 and E921X in
family 6), and two families harbored sequence alterations that could
affect normal splicing (S649S in family 7 and IVS17 1 5G 3 A in
family 8). The synonymous Serine change at residue 649 (TCG 3
TCA) in exon 11 of the RET gene has previously been observed (11)
in one patient. With the observation of this second mutation, we

Bolk et al. PNAS u January 4, 2000 u vol. 97 u no. 1 u 269

G
EN

ET
IC

S

D
o
w

n
lo

a
d
e
d
 a

t 
C

a
rn

e
g
ie

 M
e
llo

n
 U

n
iv

e
rs

ity
 o

n
 A

p
ri
l 5

, 
2
0
2
1
 



sought to determine whether this was a polymorphism. The G-to-A
nucleotide change created a DdeI restriction site and allowed a
rapid screen of 60 CEPH control individuals and 100 HSCR
probands; we did not encounter this again in 320 chromosomes. The
second splicing mutation occurs in intron 17 in the consensus splice
donor site, altering the sequence of the splice donor site from
(exonyintron) AGyGTGAGT to AGyGTGAAT. The remaining
three RET-linked families (families 9–11) have been extensively
screened, but no causative sequence changes could be identified.

To assess which changes might be functional mutations, and
which benign rare variants, we also compared the nature of the
residue altered (conservative, nonconservative, or neutral) and the
evolutionary conservation of that amino acid by protein sequence
alignment of human, mouse, chicken, zebrafish, and Drosophila
RET. As shown in Table 1, the changes in families 1–4 are either
nonconservative residue changes or alterations at evolutionarily
conserved residues and likely lead to diminished protein function.
The mutations in families 5 and 6 are nonsense changes at very
highly conserved sites. Thus, we conclude that the sequence alter-
ations in families 1–6 are RET mutations that lead to HSCR. These
mutations are likely loss-of-function variants.

The RET S649S Synonymous Change Is a Variant Splicing Mutation. The
synonymous substitution S649S is a priori an unlikely candidate
mutation. We hypothesized that the underlying nucleotide change
(TCG3TCA) causes aberrant splicing of RET exon 11 because the
substitution creates a sequence [CTTCATCGTCTCAGT] that
closely resembles the consensus splice acceptor site [(ct)10ncagG]
(Fig. 3a). If this novel site within exon 11 were used in splicing, then
this would result in the deletion of the first 23 amino acids of exon
11. Given that exon 11 encodes the transmembrane domain of RET,

it is likely that this deletion would not allow membrane anchoring
and would be a loss-of-function allele.

We used RT-PCR analysis (Fig. 3b) to show functionally that the
mutation base at residue 649 creates a novel splice acceptor site
within RET exon 11 that is used in splicing in lymphoblastoid cells.
However, these data do not clarify whether the new splice acceptor
site is used exclusively or not. To determine whether the RT-PCR
band of wild-type molecular weight was a single molecular species
or not, we used an expressed BanI polymorphism, G691S, in exon
11 for which carriers of the S649S variant were also heterozygous.
As shown in Fig. 3c, the wild-type band consisted of amplification
products from two sources: the transcript carrying the wild-type
sequence and the transcript carrying the S649S mutation. There-
fore, the novel splice site is only partially used. On this basis, we
conclude that the S649S mutation is a hypomorphic allele and that
carriers will express an intermediate level of RET. We were unable
to assay the effect on splicing caused by IVS17 1 5G 3 A because
of unavailability of cell lines from family 8.

A New HSCR Locus at Chromosome 9q31 Modifies RET Penetrance. To
search for additional susceptibility factors, we have begun a ge-
nome-scan in the 12 pedigrees, initially concentrating on chromo-
some 9 based on a two-point lod score of 1.51, in one family, with
a microsatellite marker within the Hexabrachion locus at 9q32–34.
We genotyped and performed linkage analysis using 16 polymor-
phic microsatellite markers spanning the entire genetic length of
chromosome 9 at a map resolution of 12 cM (Fig. 2b). We obtained
weak evidence of linkage with a peak NPL score of 3.8 (P 5 0.002)
between markers D9S1677 and D9S1828 (Table 2). Parametric
analysis provided a peak lod score '20-cM proximal to the peak
NPL value, probably because of incorrect specification of param-

Fig. 1. HSCR families used in linkage analysis. Individuals with confirmed HSCR and clinical diagnosis of chronic severe constipation are indicated by filled and
shaded symbols, respectively. Asterisks indicate individuals used in genotyping.

270 u www.pnas.org Bolk et al.

D
o
w

n
lo

a
d
e
d
 a

t 
C

a
rn

e
g
ie

 M
e
llo

n
 U

n
iv

e
rs

ity
 o

n
 A

p
ri
l 5

, 
2
0
2
1
 



eters in the analysis. Further, we obtained a peak heterogeneity lod
score of 2.3 with a 5 46% of families linked. These results are
consistent with haplotype analysis (Table 1; Fig. 2b), which shows
segregation of the 9q31 locus in families 7–12 only. There was no
evidence for a common 9q31 haplotype in the linked families.

The results of mutation analysis at RET and linkage analysis on
chromosome 9 suggested classification of the 12 families into two
groups: RET-linked families 1–6 (group I), which carried severe

nonsense and missense coding sequence RET mutations but no
segregation at 9q31; and those without RET missense or nonsense
mutations (families 7–12, group II) but segregation at 9q31. This
association between RET mutation type and 9q31 segregation is
highly significant (x2 5 12, 2 df, P 5 0.0025) and demonstrates that
segregation of the additional 9q31 locus is restricted to families that
do not have missense or nonsense RET mutations. We reanalyzed
the linkage data by the type of RET mutation (Fig. 4) and show
significant linkage in the group II families with a NPL score of 5.3
(P 5 6.8 3 1025) between markers D9S1677 and D9S1828; as
expected, no evidence of linkage was obtained in the group I
families. Analogous parametric analysis showed a peak lod score of
4.8 in group II families at the same map location defined by the peak
NPL score; no significant linkage to 9q31 was evident in group I
families.

To distinguish between multigenic inheritance and locus heter-
ogeneity, we next reanalyzed RET linkage data by family type. As
shown in Table 2, RET linkage is significant in both group I and II
families with NPL scores of 9.1 (P 5 6.2 3 1025) and 5.3 (P 5 4.
5 3 1025), respectively. Under parametric analysis, significant lod
scores were also observed in both groups (lod 5 7.7, a 5 100% in
group I; lod 5 3.9, a 5 86% in group II). In group I families, RET
is the sole contributory factor because there is no evidence of
segregation of the 9q31 locus. In the remaining group II families,
there is overall significant evidence of linkage to both RET and 9q31
using NPL and lod score analysis. One family (family 12) in this
latter group, analogous to those in group I, might harbor a severe
mutation in the 9q31 gene only. These data clearly demonstrate the
existence of at least two types of HSCR families: families with
missense or nonsense RET mutations sufficient to lead to HSCR
and those with weak alleles in which additional genetic changes are
required for disease expression. We suspect that RET-linked fam-
ilies without recognized mutations (families 9–11) either harbor
common polymorphic variants (excluded from being considered as
mutations) or noncoding sequence changes that make partial
contributions to HSCR risk. The demonstration of 9q31 segrega-
tion in these latter families suggest both multigenic inheritance and
that HSCR results from the cumulative genetic effects at RET and
the yet unknown 9q31 locus in some families.

Discussion
Segregation of Hirschsprung disease is decidedly non-Mendelian;
in most families, the phenotype is variable and the mutant genotype

Fig. 2. Linkage analysis of HSCR. (a) Chromosome 10. (b) Chromosome 9. The
x axis indicates the genetic map of the indicated chromosome with intermarker
distances in centimorgans (cM), and the y axis indicates the linkage score. (Solid
line, NPL score; dashed line, parametric lod score. Although depicted on the same
graph, NPL and lod scores have distinct properties and are not synonymous; thus,
they should not be numerically compared against each other.)

Table 1. Segregation of RET and a chromosome 9q31 locus in HSCR families

Family
Posterior probability

of RET linkage*

RET†
9q31 haplotype

shared‡Mutation Type Evolutionary conservation

1 1.00 N394K n m, c, z no
2 0.99 R360W nc m no
3 0.99 R77C nc m, c no
4 1.00 S32L1 nc m no
5 0.99 W942X1 — m, c, z, d no
6 0.99 E921X1 — m, c, z, d no
7 0.99 S649S — — yes
8 0.98 IVS17 1 5G . A — — yes
9 0.97 ND — — yes

10 1.00 ND — — yes
11 0.99 ND — — yes
12 0.005 ND — — yes

*Posterior probability of RET linkage (see Materials and Methods): haplotype sharing among all affecteds within each family, based on
markers D10S141-RETint5-D10S1099, was consistent with posterior probability estimates (shared in families 1–11; not shared in family 12).

†RET mutations in exons [1, reported previously (7); ND, none detected]; listed are mutational change, residue change classified as
nonconservative (nc) or neutral (n), and evolutionary conservation of human residue (m, c, z, and d denoting conservation in mouse,
chicken, zebrafish, and Drosophila, respectively).

‡Haplotype sharing within each family based on markers D9S176-D9S1832-D9S1677-D9S1828.

Bolk et al. PNAS u January 4, 2000 u vol. 97 u no. 1 u 271

G
EN

ET
IC

S

D
o
w

n
lo

a
d
e
d
 a

t 
C

a
rn

e
g
ie

 M
e
llo

n
 U

n
iv

e
rs

ity
 o

n
 A

p
ri
l 5

, 
2
0
2
1
 



displays reduced penetrance (4). The central genetic question is,
therefore, whether the reduced penetrance is random and depen-
dent on environmental or stochastic developmental factors, non-
random and dependent on an individual’s genetic background, or
both. We have presented significant evidence that the latter is true
because HSCR shows multigenic inheritance with RET-mediated
susceptibility being the major theme. Moreover, additional genes
with large effect, such as the one we have mapped to 9q31, must also
exist.

Significant evidence argues that RET is the major gene for HSCR
and is involved in '80% of families (6, 19–21, 33). Nevertheless,
investigators have failed to find coding sequence mutations in
.50% of familial cases (6–8, 21). This inability to detect RET
coding sequence mutations in linked families is not attributable to
mutation detection technology but is a persistent finding that
requires alternative explanations, such as: (i) RET linkage is
spurious in small families; (ii) a gene closely linked to RET harbors

susceptibility mutations; (iii) mutations other than those in the RET
coding sequence, including introns and the promoter regions,
compromise RET function; and (iv) RET is under epigenetic
regulation that can affect its normal expression and function.

Scenario i is undoubtedly true for some families but cannot be the
major explanation because the sum total of families without iden-
tified RET mutations still demonstrates linkage to chromosome
10q11 (Table 2). Indeed, family 12 reported here is the first
description of a multiplex family in which RET-mediated suscepti-
bility can be definitively excluded. Scenario ii, in which a gene
different from, but tightly linked to, RET contributes to suscepti-
bility, cannot be currently ruled out. However, our preliminary
analysis of '500 kilobases of nucleotide sequence proximal to RET
(GenBank accession no. AL022344) has not identified any specific
transcripts that might be implicated. Rigorous identification and
testing of all genes in this region is required and is currently
underway. We conclude that the related scenarios iii and iv are the
most likely explanations, both of which suggest misregulation of
RET in HSCR.

In this study, we have demonstrated the diverse nature of RET
mutations in HSCR (Table 1). Although functional tests have not
been performed on each mutation, the genetic data strongly argue
for loss-of-function mutations in group I families. The two nonsense
mutations we observed, E921X and W942X, occur in the tyrosine
kinase domain and are expected to abolish kinase activity as
suggested by in vitro functional studies of similar mutants (34).
Additional studies imply that the four missense mutations, S32L,
R77C, R360W, and N394K, occur in the extracellular domain at

Fig. 4. Nonparametric linkage (NPL) analysis of chromosome 9 markers in
HSCR on subdividing families by RET mutation status. Dashed line, those with
RET coding sequence mutations (families 1– 6); solid line, those without RET
coding sequence mutations (families 7–12).

Fig. 3. Functional consequences of the S649S variant in RET splicing. (a)
Normal and mutant splicing events of RET exon 11 and its resulting translation.
The site of the G 3 A transition observed in HSCR family 7 does not alter the
Serine649 residue but does create a sequence closely resembling the consen-
sus splice acceptor site [(ct)10ncagG]. (b) RT-PCR on RNA isolated from a
lymphoblastoid cell line. One band of the expected size (383 bp) on normal
splicing and a smaller band (314 bp) resulting from aberrant exon 11 splicing
are amplified in affected carriers of the Serine variant (lanes 2– 6). The unaf-
fected grandmother (lane 1) who does not carry the S649S variant has only one
RT-PCR product of the expected size (383 bp) on normal splicing. (c) BanI
restriction digest of RT-PCR products. An expressed polymorphism in RET,
which alters a BanI restriction site in exon 11, was used to analyze RT-PCR
products. The wild-type band in lane 3 (labeled as 31) and the mutant band
in lane 3 (32), as well as the band in lane 1 (11), were gel-extracted. Two bands
(239 bp and 144 bp) are observed after the digestion of lane 1, indicating
homozygosity for the BanI site. Lane 31 contains three bands: one undigested
band (383 bp) and two bands resulting from the BanI digest (239 bp and 144
bp), indicating that the wild-type band observed in b has been amplified from
both the wild-type and mutant alleles. The novel acceptor splice site is thus
only partially used when present in an affected individual.

Table 2. Linkage analysis of chromosome 10 and 9 genetic
markers and family classification by RET mutational type

Family type

RET 9q31 locus

NPL
(P value)

lod
score
(a)

NPL
(P value)

lod
score
(a)

Missenseynonsense mutations, 9.1 7.7 0.4 0.0
families 1–6 (6.2 3 1025) (1.00) (0.5) (0)

Splice variantsyno detected 5.3 3.9 5.3 4.8
changes, families 7–12 (4.5 3 1025) (0.86) (6.8 3 1025) (1.0)

All families 11.3 10.5 3.8 2.3
(7.7 3 1028) (0.91) (0.002) (0.46)

Families were classified according to presence (families 1– 6) or absence
(families 7–12) of missense or nonsense mutation. Nonparametric analysis
provided the NPL score and P value; parametric analysis provided the heter-
ogeneity lod score and the estimated proportion of linked families (a) (see
Materials and Methods).

272 u www.pnas.org Bolk et al.

D
o
w

n
lo

a
d
e
d
 a

t 
C

a
rn

e
g
ie

 M
e
llo

n
 U

n
iv

e
rs

ity
 o

n
 A

p
ri
l 5

, 
2
0
2
1
 



residues that are either highly conserved andyor lead to radical
substitutions, are expected to alter protein folding, and inhibit
transport of RET protein to the cell surface (35–37). These obser-
vations, together with microdeletions of chromosome 10q11.1
including RET in rare patients, have supported the view that HSCR
results from haplo-insufficiency because 50% RET cannot support
normal enteric development in the human. In group I families, these
severe mutations may be sufficient to lead to HSCR.

The group II families hold the key to further understanding of
HSCR. Although these families show significant linkage to RET,
only two splice variants, IVS17 1 5G 3 A and S649S, of which
S649S is a weak RET allele (Fig. 3), have been identified. Such weak
alleles compromise but do not abrogate RET activity and thus
require additional genetic changes for HSCR expression. The
segregation of 9q31 in group II families adds weight to this
conclusion. In three of the remaining five (group II) families,
despite a .97% probability of RET linkage, we have failed to
identify any coding sequence RET mutations. We conclude that
mutations in group II families are hypomorphs and harbor muta-
tions in the noncoding regulatory regions. In addition, polymor-
phisms within the gene, which have been excluded as disease-
causing, may contribute to disease risk as well.

Genetic disorders with variable expressivity and incomplete
penetrance can arise from either structural or regulatory hypomor-
phic mutations. In familial retinoblastoma, structural mutations
that result in only partial loss of protein function, rather than
complete abrogation of normal gene function, are associated with
incompletely penetrant forms of the disease (38, 39). In addition,
splicing mutations with variable efficiency in usage of novel splice
sites have been proposed as the molecular basis for reduced
penetrance in cystic fibrosis (40). Epigenetic (regulatory) modifi-
cation of a gene, such as through position effects, is another type of
mutation that may be responsible for reduced penetrance. First,
position effect mutations can occur hundreds of kilobases away
from the target gene, explaining both linkage and failure to observe
coding mutations (41). Second, position effect changes have phe-

notypes similar to loss-of-function mutations; several examples are
known in the human and mouse (41). The best-studied is the
receptor tyrosine kinase c-kit, known to be subject to position effect
variegation as a result of specific mutations affecting chromosomal
rearrangement and, as a consequence, control elements of the gene.
In this case, gene function is not completely abolished, but the effect
of the mutation represses transcription in a tissue-specific manner
(42). Similar factors may be operating in HSCR.

We postulate that the degree of RET expression is critical for
HSCR expression: severe alleles lead to HSCR directly; weak alleles
also require the effects of a 9q31 gene. A corollary is that the 9q31
gene is a modifier of HSCR penetrance. Consequently, the 9q31
gene could be a member of a signaling pathway that affects RET
expression, and a ‘‘hit’’ at multiple locations within the pathway may
be the cause of HSCR in families with weak RET mutations.
Proving this hypothesis will require identification of this gene;
unfortunately, the current linkage location is still too broad (95%
support limits are '10 cM) to allow positional cloning. It is possible
that this 9q31 gene is the same as that which is mutant in Riley-Day
syndrome [Mendelian Inheritance in Man (MIM) 223900]. The
latter, also known as familial dysautonomia, is a hereditary neu-
ropathy that results from a loss of subsets of autonomic neurons and
is associated with lack of tearing, emotional lability, paroxysmal
hypertension, increased sweating, cold hands and feet, corneal
anesthesia, and blotching of the skin. Riley-Day syndrome, common
in Ashkenazi Jews, has been shown to co-occur with HSCR and has
been mapped to an '150-kilobase segment in 9q31 (43). This gene,
when identified, will be an excellent candidate gene for HSCR.

We are indebted to the HSCR families for participation and the American
Pseudo-obstruction & Hirschsprung Disease Society, Inc., for cooperation.
We gratefully acknowledge the work of Jennifer Scott for family studies,
James Gabriel for advice with RNA studies, Kimberly Bentley and Carl
Kashuk for excellent technical assistance, and Drs. Hunt Willard and Evan
Eichler for comments on the manuscript. This work was supported by a
grant from the National Institutes of Health (HD28088) to A.C.

1. Chakravarti, A. & Lander, E. S. (1990) in Genetics and Biology of Alcoholism, ed. C. R.
Cloninger, H. Begleiter (Cold Spring Harbor Lab. Press, Plainview, NY), pp. 307–315.

2. Kajiwara, K., Berson, E. L. & Dryja, T. P. (1994) Science 264, 1604 –1608.
3. Cucca, F., Goy, J. V., Kawaguchi, Y., Esposito, L., Merriman, M. E., Wilson, A. J.,

Cordell, H. J., Bain, S. C. & Todd, J. A. (1998) Nat. Genet. 19, 301–302.
4. Badner, J. A., Sieber, W. K., Garver, K. L. & Chakravarti, A. (1990) Am. J. Hum. Genet.

46, 568 –580.
5. Edery, P., Pelet, A., Mulligan, L. M., Abel, L., Attie, T., Dow, E., Bonneau, D., David,

A., Flintoff, W., Jan, D. et al. (1994) J. Med. Genet. 31, 602– 606.
6. Angrist, M., Bolk, S., Thiel, B., Puffenberger, E. G., Hofstra, R. M., Buys, C. H. C. M.,

Cass, D. T. & Chakravarti, A. (1995) Hum. Mol. Genet. 4, 821– 830.
7. Attie, T., Till, M., Pelet, A., Amiel, J., Edery, P., Boutrand, L., Munnich, A. & Lyonnet,

S. (1995) Hum. Mol. Genet. 4, 2407–2409.
8. Treanor, J. J. S., Goodman, L., De Sauvage, F., Stone, D. M., Poulsen, K. T., Beck, C. D.,

Gray, C., Armanini, M. P., Pollock, R. A., Hefti, F., et al. (1996) Nature (London) 382, 80–83.
9. Trupp, M., Arenas, E., Fainzilber, M., Nilsson, A.-S., Sieber, B.-A., Grigoriou, M.,

Kilkenny, C., Salazar-Grueso, E., Pachnis, V., Arumae, U., et al. (1996) Nature (London)
381, 785–789.

10. Angrist, M., Bolk, S., Halushka, M., Lapchak, P. L. & Chakravarti, A. (1996) Nat. Genet.
14, 341–344.

11. Salomon, R., Attie, T., Pelet, A., Bidaud, C., Eng, C., Amiel, J., Sarnacki, S., Goulet, O.,
Ricour, C., Nihoul-Fekete, C., et al. (1996) Nat. Genet. 14, 345–347.

12. Hofstra, R. M., Osinga, J. & Buys, C. H. (1997) Eur. J. Hum. Genet. 5, 180 –185.
13. Ivanchuk, S. M., Myers, S. M., Eng, C. & Mulligan, L. M. (1996) Hum. Mol. Genet. 5,

2023–2026.
14. Angrist, M., Jing, S., Bolk, S., Bentley, K., Nallasamy, S., Halushka, M., Fox, G. &

Chakravarti, A. (1998) Genomics 48, 354 –362.
15. Puffenberger, E. G., Hosoda, K., Washington, S. S., Nakao, K., deWitt, D., Yanagisawa,

M. & Chakravarti, A. (1994) Cell 79, 1257–1266.
16. Chakravarti, A. (1996) Hum. Mol. Genet. 5, 303–307.
17. Southard-Smith, E. M., Angrist, M., Ellison, J., Agarwala, R., Baxevanis, A., Chakravarti,

A. & Pavan, W. J. (1999) Genome Res. 9, 215–225.
18. Pingault, V., Bondurnad, N., Kuhlbrodt, K., Goerish, D. E., Prehu, M. & Puliti, A. (1998)

Nat. Genet 18, 171–173.
19. Romeo, G., Ronchetto, P., Luo, Y., Barone, V., Seri, M., Ceccherini, I., Pasini, B., Bocciardi,

R., Lerone, M., Kaarianen, H. & Martucciello, G. (1994) Nature (London) 367, 377–378.
20. Edery, P., Lyonnet, S., Mulligan, L. M., Pelet, A., Dow, E., Abel, L., Holder, S., Nihoul

Fekete, C., Ponder, B. A. & Munnich, A. (1994) Nature (London) 367, 378–380.
21. Attie, T., Pelet, A., Edery, P., Eng, C., Mulligan, L. M., Amiel, J., Boutrand, L., Beldjord,

C., Nihoul Fekete, C., Munnich, A., et al. (1995) Hum. Mol. Genet. 4, 1381–1386.
22. Angrist, M., Kauffman, E., Slaugenhaupt, S. A., Matise, T. C., Puffenberger, E. G., Bolk, S.,

Washington, S. S., Lipson, A., Cass, D. T., Reyna, T., et al. (1993) Nat. Genet. 4, 351–356.
23. Ott, J. (1991) Analysis of Human Genetic Linkage (Johns Hopkins Univ. Press, Baltimore).
24. Kruglyak, L., Daly, M. J., Reeve-Daly, M. P. & Lander, E. S. (1996) Am. J. Hum. Genet.

58, 1347–1363.
25. Dib, C., Faure, S., Fizames, C., Samson, D., Drouot, N., Vignal, A., Millasseau, P., Marc,

S., Hazan, J., Seboun, E., et al. (1996) Nature (London) 380, 152–154.
26. Ceccherini, I., Bocciardi, R., Luo, Y., Pasini, B., Hofstra, R., Takahashi, M. & Romeo,

G. (1993) Biochem. Biophys. Res. Commun. 196, 1288 –1295.
27. Takahashi, M., Buma, Y., Iwamoto, T., Inaguma, Y., Ikeda, H. & Hiai, H. (1988)

Oncogene 3, 571–578.
28. Takahashi, M., Buma, Y. & Hiai, H. (1989) Oncogene 4, 805– 806.
29. Myers, S. M., Eng, C., Ponder, B. A. & Mulligan, L. M. (1995) Oncogene 11, 2039 –2045.
30. Kusafuka, T., Wang, Y. & Puri, P. (1997) J. Pediatr. Surg. 32, 501–504.
31. Hofstra, R. M. W., Osinga, J., Tan-Sindhunata, G., Wu, Y., Kamsteeg, E.-J., Stulp, R. P.,

van Ravenswaaij-Arts, C., Majoor-Krakauer, D., Angrist, M., et al. (1996) Nat. Genet.
12, 445– 447.

32. Edery, P., Attie, T., Amiel, J., Pelet, A., Eng, C., Hofstra, R. M., Martelli, H., Bidaud,
C., Munnich, A. & Lyonnet, S. (1996) Nat. Genet. 12, 442– 444.

33. Seri, M., Yin, L., Barone, V., Bolino, A., Celli, I., Bocciardi, R., Pasini, B., Ceccherini,
I., Lerone, M., Kristoffersson, U., et al. (1997) Hum. Mutat. 9, 243–249.

34. Pasini, B., Borrello, M. G., Greco, A., Bongarzone, I., Luo, Y., Mondellini, P., Alberti,
L., Miranda, C., Arighi, E., Bocciardi, R., et al. (1995) Nat. Genet. 10, 35– 40.

35. Iwashita, T., Murakami, H., Asai, N. & Takahashi, M. (1996) Hum. Mol. Genet. 5,
1577–1580.

36. Carlomagno, F., De Vita, G., Berlingieri, M. T., de Franciscis, V., Melillo, R. M.,
Colantuoni, V., Kraus, M. H., Di Fiore, P. P., Fusco, A. & Santoro, M. (1996) EMBO
J. 15, 2717–2725.

37. Pelet, A., Geneste, O., Edery, P., Pasini, A., Chappuls, S., Attie, T., Munnich, A., Lenoir,
G., Lyonnet, S. & Billaud, M. (1998) J. Clin. Invest. 101, 1415–1423.

38. Sakai, T., Ohtani, N., McGee, T. L., Robbins, P. D. & Dryja, T. P. (1991) Nature
(London) 353, 83– 86.

39. Otterson, G. A., Chen, W., Coxon, A. B., Khleif, S. N. & Kaye, F. J. (1997) Proc. Natl.
Acad. Sci. USA 94, 12036 –12040.

40. Rarve-Harel, N., Kerem, E., Nissim-Rafinia, M., Madjar, I., Goshen, R., Augarten, A.,
Rahat, A., Hurwitz, A., Darvasi, A. & Kerem, B. (1997) Am. J. Hum. Genet. 60, 87–94.

41. Bedell, M. A., Jenkins, N. A. & Copeland, N. G. (1996) Nat. Genet. 12, 229 –232.
42. Duttlinger, R., Manova, K., Chu, T. Y., Gyssler, C., Zelenetz, A. D., Bachvarova, R. F.

& Besmer, P. (1993) Development (Cambridge, U.K.) 118, 705–717.
43. Slaugenhaupt, S., Blumenfeld, A., Liebert, C. B., Mull, J., MacCormack, K., LeBel, A.,

O’Leary, K., Breakefield, X. O., Maayan, C., McCormick, M. K., et al. (1996) Am. J. Hum.
Genet. 59, 55 (abstr.).

Bolk et al. PNAS u January 4, 2000 u vol. 97 u no. 1 u 273

G
EN

ET
IC

S

D
o
w

n
lo

a
d
e
d
 a

t 
C

a
rn

e
g
ie

 M
e
llo

n
 U

n
iv

e
rs

ity
 o

n
 A

p
ri
l 5

, 
2
0
2
1