3!}...
I...

 

.x. .-
{‘5vr...

. It 17..

;
t.l...{.| v...- .
(.1.

‘v‘:
:le

 

 

 

 

.1:
.:| s $9

1!!
.2.
:l...::.

.31. Iv.)-V
. .Ir....:‘.A..t.

a 3“
:3 55:95..
.. J?

 

5“ T
.151..-9In.lv
. rt]. :Yb:

‘..‘ . .n.
. .72.}...

 

RABIES

. \iiiiliiixmw \Y“\\\\\\\\\§\3\\\\\i

This is to certify that the
dissertation entitled

THE IDENTIFICATION OF MUTATIONS
RESPONSIBLE FOR WAARDENBURG SYNDROME

presented by

ROBERT JAMES MORELL

has been accepted towards fulfillment
of the requirements for

 

 

 

 

PhD degree in Zoology
v Major professor
Date ; - f- ?'5

 

MS U i: an Affirmative Action/Equal Opportunity Institution 0~12771

 

 

4- ‘_ _-—V.—‘

4~_, - fie

 

LIBRARY
Michigan State
Unhenfly

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

, n
I

mi) 41:30

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Imnution
emails-9.1

 

j_*

THE IDENTIFICATION OF
MUTATIONS RESPONSIBLE FOR WAARDENBURG SYNDROME

By

Robert James morell

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Zoology

1993

ABSTRACT

THE IDENTIFICATION OF
MUTATIONS RESPONSIBLE FOR WAARDENBURG SYNDROME

BY

Robert James Morell

Waardenburg syndrome (WS) is an autosomal dominant
mutation originally characterized by congenital deafness,
dystopia canthorum, heterochromia irides, poliosis, broad
nasal root, and synophrys. Since the original description of
the syndrome, other clinical characteristics have been
reported in the literature, including aganglionic megacolon,
upper-limb deformities, spina bifida, leukoderma, and
meningomyelocele. WS exhibits reduced penetrance and
variable expressivity for all of the clinical features, even
within families. On the basis of the segregation of discrete
subsets of the clinical features within certain families,
investigators have maintained that the syndrome is
clinically heterogeneous, and have described at least four
types of Waardenburg syndromes: W81, W52, W53 and W54.
Whether the apparent clinical heterogeneity reflects an
underlying genetic heterogeneity, or is an artifact of co-
segregation of modifier genes within families, is a subject

of dispute.

Some WSl mutations have been mapped to chromosome 2q37,
in a region of conserved syntenies between the murine and
human genetic maps. The murine mutation Splotch (Sp) was
hypothesized to be the homolog of the 2q37 WSl mutations by
virtue of its similar phenotype and location on the genetic
map. Several alleles of Sp have been shown to be mutations
of Fax-3. On this basis, mutations in human PAX3 were sought
and identified in individuals with WSl, and are arguably the
primary cause of the WSl phenotype. The molecular events
giving rise to WS are still far from understood since normal
PAX3 function and regulation are not known.

We have characterized two mutations in PAX3 causing
WSl. These mutations offer a better understanding of the.
molecular pathology of WS, and suggest that the WS phenotype
arises as the result of haploinsufficiency of PAX3. We have
also tested for linkage of WS mutations to PAX3 in several
families and found evidence that WS is genetically, as well

as clinically, heterogeneous.

iii

ACKNOWLEDGEMENTS

I would like to thank Tom Friedman and Jim Asher for
their patience, enthusiasm and many interesting discussions;
Tom Barber, Jean Burnett, Peter Crawley, Sue Lootens, Lenny
Robbins, Robin Steinman, Ellen Swanson, Mark Thompson, Lori
Wallrath and Brian Wissink, for many pow—wows, both
scientific and non-scientific; and Pat Venta and Jim
Higgins, members of my committee, for helpful suggestions.
All of the Waardenburg syndrome families described in this
thesis were identified and characterized in whole or in part
by Jim Asher.

I would especially like to thank John and Sue McRorie.
Without their friendship and moral support I would have had
neither the strength nor the desire to endure graduate

school.

W

TABLE OF CONTENTS

List of Tables ........................................
List of Figures .......................................
Glossary and Symbols ..................................
INTRODUCTION ..........................................

I CHAPTER ONE: Genetic linkage mapping of Waardenburg
syndrome to chromosome 2q. .......................

A. Introduction ..................................

B. Results .......................................

1. Linkage analysis of MSU4 .................

2. Linkage analysis of MSUS .................

3. Linkage analysis of MSU6 .................

4. Linkage analysis of UGMl .................

C. Discussion ....................................

II. CHAPTER TWO: Mutations in PAX3 associated with
- Waardenburg syndrome. ............................
A. Introduction ..................................
B. Results .......... .............................
C. Discussion ....................................

III. APPENDIX A: Materials and Methods ...............
Evaluation of Subjects ........................
Preparation of Genomic DNA ....................
Southern Blots ................................
PCR Amplifications ............................
Genotyping for Dinucleotide Repeat Markers

SSCP Analysis .................................
Sequencing ....................................
Linkage Analysis ..............................

OO \J<T\Ln fistelbo F4

IV. APPENDIX B: Additional Figures and Tables ........

V. APPENDIX C: Detection of Changes in Copy Number at
Regions of Candidate Loci By Dot Blot Analysis ...

VI. APPENDIX D: Conventions of Nomenclature for Human
Alleles Used in this Thesis ......................

VII.APPENDIX E: Results of Linkage Analysis for MiOR

VIII. LIST OF REFERENCES ..............................

vi
vii

viii

33
33
36
49

59
59
60
60
61
62
63
63
65

67

74

84

87

9O

Table

Table

Table

Table
Table
Table

Table

Table

Table
Table
Table
Table
Table
Table
Table
Table

Table

Table

Table

10

Al

A2

Bl

82

B3

B4

C1

C2

E1

LIST OF TABLES

Phenotypes of Affected Members of Family
MSU4

MSU4 Linkage Results

Phenotypes of Affected Members of Family
MSUS

MSUS Linkage Results
Phenotypes of Members of Family MSU6
MSU6 Linkage Results

Phenotypes of Affected Members of Family
UGMl

Comparisons of a/b Ratios Between
Indonesians and Caucasians

UGMl Linkage Results

Phenotypes of WS Subjects

Waardenburg Consortium Criteria ..........

PCR and Sequencing Primers
MSU4 Linkage Data File
MSUS Linkage Data File
MSU6 Linkage Data File
UGMl Linkage Data File

Ratios of PAX3 exon Copy Numbers to Other
Loci

Ratios of Non-PAX3 Loci Comparisons

Linkage results for MiOR/+ x +/+ .........

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

000000000000000

OOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOO

12

l4

17

18

22

23

25

26

38

59

66

67

68

69

7O

77

78

88

Figure
Figure
Figure
Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure
Figure

Figure

10

Bl

BZ

B3

LIST OF FIGURES

MSU4 Pedigree ............................... 10
MSUS Pedigree ............................... 15
MSU6 Pedigree ............................... 19
UGMl Pedigree ............................... 24
SSCP Analysis of PAX3 Exon 2 PCR Fragments in
Affected Individuals ........................ 37
Sequence of Mutant PAX3 Allele Responsible

for WSl in Family MSU3 ...................... 39
Confirmation of the Segregation of a Mutant
Allele in Family MSU3 ....................... 41
Sequence of the Mutant PAX3 Allele

Segregating in UGM2 ......................... 45
UGM2 Pedigree and Southern Blot Detection of

a Mutant PAX3 Allele ........................ 47
Summary of all WS mutations in PAX3 published

to date ..................................... 51
MSU4 PAX3 Gel ............................... 71
MSU4 CD3D Gel ............................... 72
MSU4 D11835 Gel ............................. 73

vfi

Aganglionic
megacolon

Dystopia canthorum

Heterochromia
iridis

Leukoderma

PAX3

Fax-3

Poliosis

Ptosis

Sensorineural

deafness

Splotch

Synophrys

WS

SYMBOLS AND GLOSSARY

Distension of the colon due to insufficient
autonomic innervation and consequent difficulty
in voiding. Also called Hirschsprung's disease.

Lateral displacement of the medial aspect of
the canthi, or eye orbits, relative to the rest
of the face.

Different colored sectors of same iris, or
different color of two irides. May be
hereditary as in Waardenburg syndrome or
acquired by damage to sympathetic innervevation
of eye.

Depigmentation of skin. May be hereditary as in
Waardenburg syndrome, or acquired by injury or
infection as in vitiligo. Note that the latter
term is erroneously used as a synonym in the
medical literature.

Human gene, mutant alleles of which are thought
to be the primary cause of some cases of
Waardenburg syndrome.

Murine gene. Splotch mutations are mutant
alleles of Fax-3.

Strictly defined as whitening of the hair. In
medical literature used as a synonym of
congenital white forelock.

Strictly defined as 'drooping' of any organ or
body part. In medical literature means the
drooping of upper eyelid, usually resulting
from defects in sympathetic innervation to eye.

Deafness in reference to the sensory nerve, as
opposed to deafness due to defects in

mechanical transduction.

Murine homolog of Waardenburg syndrome type 1.
Alleles include: Sp, SpZH, Spr, Spa.

Confluent eyebrows.
Waardenburg syndrome. At least four clinical

subtypes described in the literature,
abbreviated WSl, WSZ, W83 and W84.

vfii

I NTRODUC TI ON

The association of deafness with abnormal pigmentation
in domesticated animals has long been known. Darwin.[Yn, for
instance, used the example of deafness and blue eyes in cats
for general discussions on the co-inheritance of traits. And
the prevalence of deafness in Dalmatian dogs and white minks
was generally accepted long before it was scientifically
confirmed [13,23]. Yet, except for isolated reports (for
review see Fraser [23]) , the phenomenon went largely
unrecognized in humans until Waardenburg's 1951 paper [58].

Waardenburg described a form of deafness that co-
segregated with pigmentary and facial abnormalities in 16
families. After first encountering a deaf patient who showed
pronounced dystopia canthorum, or the lateral displacement
of the lachrimael punctae relative to the rest of the face,
he conducted an extensive survey of five Dutch schools for
the deaf to find others with a similar phenotype. He
realized that pigmentary anomalies were also prevalent among
deaf children with dystopia canthorum, and he therefore
defined the syndrome as comprising six clinical
characteristics and estimated the penetrance for each by the
incidence among affected members of the families, excluding
the probands. The clinical manifestations were: l)dystopia
canthorum (99%); 2)broad nasal root (78%); 3)synophrys or
confluent eyebrows (45%); 4)poliosis or white forelock

(17%); S)heterochromia iridum or differently colored irides

(irides either totally mismatched or a single iris
containing differently pigmented sectors)(25%); and
6)Congenital deafness (20%). Many other, less frequent,
attributes were described, and they will be discussed later.
The phenotype segregated as an autosomal dominant, and was
variable both within and between families. Extrapolating
from his sample, he estimated that the syndrome accounted
for 1.43% of the institutionalized deaf population, and the
gene had a population frequency of 1/42,000 and a mutation
rate of l/270,000.

It is important to note that Waardenburg was initially
interested in the co-segregation of deafness with dystopia
canthorum, thus his survey probably represents a biased
sample. In particular, the high penetrance of dystopia
canthorum led him to postulate that a single gene was
responsible, although the reasoning may be tautological in
that the syndrome was primarily defined in terms of the
presence of this trait. However, among his 16 families,
there were two in which dystopia canthorum did not manifest
at all. One arguably represents a sporadic mutation due to
the absence of any other affected members in the pedigree
(family 4). The other (family 8) shows co—segregation of
heterochromia irides with deafness in several members, and
no other manifestations. Other researchers subsequently
reported similar pedigrees and argued that they represented

a distinct syndrome which they Called Waardenburg syndrome

type 2 (WSZ)[1,9,29], and which was defined as Waardenburg
syndrome with the absence of dystopia canthorum.

A third type of Waardenburg syndrome (W83) was proposed
by researchers [26,51] who described patients manifesting
severe musculo-skeletal abnormalities in addition to WSl-
like features, and who otherwise resemble the patient first
reported by Klein and described by Waardenburg [58]. All of
these patients are severely affected and show no family
history for any of the attributes. Therefore, they are
probably the results of sporadic mutations, and most
probably represent examples of contiguous gene syndromes.
However, a family in which WS3 segregates as an autosomal
dominant has recently been described [32,52].

Reports of Waardenburg syndrome type 1 (W81) associated
with aganglionic megacolon [6,23,25,4l,44] seemed to merit a
further subdivision as Waardenburg syndrome type 4 (W84).
AriasU] suggested yet another distinct clinical subtype to
be known as "pseudo—Waardenburg“ syndrome, the
distinguishing feature of which is congenital, hereditary,
unilateral ptosis.

The further subdividing of the phenotypes into clinical
subtypes only emphasizes the challenge presented in
Waardenburg's initial description of the syndrome in that
the phenotype is highly variable in its expressivity and
penetrance. It is important to note that each aspect of the
phenotype has been reported in the literature to segregate

as a single entity [58]. The obvious question raised, and

discussed by Asher and Friedman PH, is whether the primary
cause is a single gene with pleiotropic effects, a single
gene modified by various other loci, or several genes each
of which can give rise to distinct clinical entities when
mutated. The merits of each of these hypotheses will be
discussed in the course of this thesis, but suffice to say
at the outset that each hypothesis suggests a different
strategy for identifying the molecular lesion responsible
for Waardenburg syndrome. Further, any proposed mutation in
a candidate gene would have to be compatible with a highly
variable phenotype, and would have to address the question
posed by Waardenburg as to what could account for the co-
inheritance of a seemingly disparate constellation of

attributes.

I. CHAPTER ONE: Genetic linkage mapping of Waardenburg

syndrome to Chromosome 2q.

A. Introduction

The hypothesis of a single gene exerting pleiotropic
effects gained credence with the elucidation of neural crest
cell development and the realization that all of the
affected tissues in the syndrome are neural crest cell
derivatives [6.35.41]. Accordingly, McKusick [44] proposed that
the pleiotropy in WS can be explained by a single defect
that affects neural crest cell migration and/or
differentiation. This promised to make the task of finding
the genetic cause easier since detecting genetic linkage to
single locus disorders is significantly easier than
detecting linkage in multi-locus disorders.

This optimism was somewhat mitigated by the fact that
there were several genetically distinct mouse mutations that
had been demonstrated to disrupt neural cell migration
and/or differentiation and give rise to pigmentary
abnormalities. These included: dominant spotting (Dom) [36],
lethal-spotted (lsH34h various alleles at the
microphthalmia locus (Mi)(summarized in Lyon and Searlepflfl),
piebald-lethal (Sl)[Ol],SplOtCh (Sp)[5], and Steel (Sl)[43].
Any one of these mouse mutations arguably could be the
murine homolog of Waardenburg syndrome. Despite the number
of candidate loci posed by mouse models for Waardenburg

syndrome, the mouse mutations still suggested that a single

locus segregating in a family could account for the full
extent of the phenotype in that family.

Asher and Friedman BU favored.Mior, Ph (Patch), 5, and
Sp as murine homologs because of the co-segregation of
pigmentary anomalies and microphthalmia in Mior, aganglionic
megacolon in s, and neural tube defects in Ph and Sp. All of
these malformations had been reported in the literature in
association with Waardenburg syndrome [45]. In particular,
Waardenburg [58] described a case of microphthalmia and three
instances of mild cleft face (furrowed forehead) in his
original description of the syndrome.

Taking advantage of the conservation of syntenic groups
between the human and murine genetic maps, Asher and
Friedman predicted human chromosomal locations of 2q, 3p or
3q based on the locations of the predicted murine homologs
on the murine map [48]. In recognition that the Clinical
heterogeneity of Waardenburg syndrome may reflect genetic
heterogeneity, they suggested that linkage to markers in
candidate regions be conducted in single pedigrees to avoid
the false exclusion of regions due to the mixing of data
from a genetically heterogeneous data set.

Exclusion of genetic linkage of WSl to regions
predicted by the M10r [3] and SI [22] models preceded the
report by Ishikiriyama et al. [33] of a child exhibiting WSl
and having a de novo inversion (2)(q35q37.3)--a region
predicted by Asher and Friedman on the basis of conserved

syntenies surrounding the Sp locus. Foy et al.[22] and Asher

et al.Efl subsequently demonstrated genetic linkage of WSl to
markers on 2q, giving further support to the proposal that
some WSl mutations are human homologs of Sp.

In order to identify and clone the gene on 2q37 giving
rise to WSl in these families, a higher resolution genetic
map was needed. In addition, the issue of possible genetic
heterogeneity was still outstanding, especially given the
fact that a negative LOD score in one family was reported by
Foy et al.[22]. Both problems could be more readily addressed
by pooling the data sets of the various laboratories that
had done linkage analyses in Waardenburg syndrome families.

A consortium effort involving several laboratories,
including ours at Michigan State University, demonstrated
linkage heterogeneity among 44 kindreds segregating for
Waardenburg syndrome, but failed to produce a genetic map
with high enough resolution to make cloning efforts
practical BAH. The closest linked marker, ALPP (Alkaline
'Phosphatase - Placental), mapped to approximately 7.6 cM
away.

When the Sp phenotype was demonstrated to be the result
of mutations in the gene Pax-3 [18], efforts centered on
investigating the homologous sequences in humans, HuP2 —- a
lambda genomic clone which shows sequence similarities to
exons 2, 3 and 4 of Fax-3 U4]. Prior to the isolation of a
cDNA of the human Pax-3 homolog [32], a locus assignment of
PAX3 was given for these three exons. Thus, the literature

variously refers to the human gene as HuP2 or PAX3.

A dinucleotide repeat polymorphism was identified in a
genomic clone containing the presumed exon 1 of the human
PAX3 gene, and was shown to map to 2q [62]. While the
discovery that Sp was an allele of Fax-3, and the subsequent
identification of mutations in the PAX3 gene of Waardenburg
syndrome patients [7,32,47,54,55] (discussed in Chapter two)
obviated the need for a high resolution genetic map of 2q37,
the PAX3-(CA)n dinucleotide repeat polymorphism became the
preferred marker to assess the degree of genetic

heterogeneity.

B. Results
1. Linkage analysis of MSU4.

The proband of MSU4 (Figure 1) was identified by virtue
of congenital deafness and the presence of heterochromia
irides and facial hair hypopigmentation in the father.
Neither individual exhibits dystopia canthorum. Subsequent
evaluation of the immediate family revealed several
instances of pigmentation anomalies with no dystopia
canthorum (Table 1). There is a history of heterochromia
irides among first-degree relatives of MSU4-l. This family

was diagnosed, therefore, as segregating for W82.

Table 1. Phenotypes of Affected
Members of Family MSU4

‘WP/EG Other

*

+

 

O
ll

Deafness late onset; mother had white forelock; sibs have
HetI.

Possible right sided ptosis.

Unilateral pelvic kidney; other kidney rotated; undescended
testes.

**

9*.

8H = Sensorineural deafness.

not! = Heterochromia irides.

Hype = Hypopigmentation of the skin.

'WPIDG == White forelock and/or early greying.
syn = Synophrys.

Two caveats need mentioning before a discussion of the

linkage analysis can be made. First, these phenotype

lO

designations are tentative. No audiometric analyses has been
made of any individuals in MSU4. The diagnosis of
heterochromia irides in individual MSU4-9 is by hearsay.
During interview, MSU4-9 reported that an ophthalmologist

had told her she had

 

 

 

I810 11 12?
3. 46 48 44 2? 48 7

13 14 15 16 19 2o 21 22 23
47 CT 7. 734 O. 44 4? 47 4? 4? 7?

Figure 1. MSU4 pedigree. Affected individuals are indicated by darkened
symbols. A grey shaded symbol indicates uncertain phenotype. The proband
is indicated by arrow. Beneath each symbol is the number corresponding
to the Linkage data file. Genotypes for the PAX3 dinucleotide repeat are
given in small bold numbers.

heterochromia irides some years before. Dr. David Kaufman,
Dept. of Ophthalmology, School of Medicine, MSU, could not
confirm this diagnosis after inspection of a color slide of
this individual taken recently. Dr. Kaufman did confirm the
heterochromia irides of individual MSU4-5 from a color

slide. The condition in MSU4—5 is reported to have been more

11

pronounced at an earlier age. It is plausible, then, that
the diagnosis of heterochromia in individual MSU4-9 was
accurate some years ago, but the appearance has diminished
to the point where it is not recognizable in a photograph.
In the absence of ophthalmological and audiometric exams,
the two most penetrant signs in W82 (deafness and
heterochromia irides) have not been reliably diagnosed in
this family [29]. The second caveat is that individual MSU4—5
has the appearance of right—sided ptosis in a still 3
photograph. A diagnosis of "pseudo—Waardenburg syndrome“U]
cannot be ruled out.

Having stated these reservations, the results of the
linkage analyses are given below (Table 2). The W82 gene was
assigned a penetrance of 57%, based on the estimate of
penetrance of deafness for W82 given by Hageman and.Delleman
[29]. If one assumes that the diagnosis of W82 could be made
by the presence of characters other than deafness, this
figure is an underestimate. If one allows that there is a
bias of ascertainment in the literature in favor of
identification of individuals that may have W82 mutations
and exhibit deafness, then this figure is an overestimate.

It is quite possible, then, that the failure to
reliably estimate the probability of non-penetrance in
individuals whose genotypes represent obligate
recombinations with PAX3 disallows interpretation of the
data in this family. However, an obligate recombination

occurs between PAX3 and W82 when one considers only those

12
individuals with a positiye diagnosis of W82. An inspection
of the genotypes of all offspring of MSU4-1 and MSU4-2
reveals that MSU4-2 had an obligatory genotype of {4 7}
(Fig. 1).

Of all of these offspring, the most reliably

diagnosed are MSU4—5 (for reasons cited above) and MSU4-3
(because of the presence of a prominent white forelock). If
both individuals are heterozygous for the W82 gene, then an
obligate recombination occurred between W82 and PAX3. MSU4-3
must have inherited the 6 allele from the ”affected“ parent,

while MSU4-5 must have inherited the 4 allele.

Table 2. MSU4 Linkage Results

 

 

 

 

 

 

 

 

 

 

 

LOD score at Recombination Fraction
Marker 0.00 0.05 0.10 0.20 0.40
PAX3 * -2.30 -1.20 -O.68 -0.20 0.05
PAX3 ** -6.78 -1.41 -O.87 -O.36 -0.02
PAX3 # -7.50 -3.11 -2.08 —1.02 -0.15
CD3D * -2.74 -2.39 -1.72 -0.88 -0.15
CD3D # -8.48 -2.90 -1.80 -0.80 -0.08
D11835 * -2.21 -2.01 -1.60 -0.98 —0.24
D11835 # -4.92 -3.65 -2.60 -1.44 -0.31

 

 

 

 

 

 

Calculations made assigning individuals 7, 17 and 18 a phenotype of
'unknown'; W82 parameters.

Individuals 7, 17 and 18 “unknown" phenotype, and W81 parameters.
Individuals 7, 17 and 18 phenotypes as Table 1; W82 parameters.

Linkage of W82 to PAX3 is excluded even if the
penetrance for the +/+ genotype (i.e. the probability of a
phenocopy) is given as 1/1000, the penetrance for W82/+ is
given as 57%, and the gene frequency is assumed to be 1/2100

U]. These relaxed parameters have the effect of allowing

13

more latitude in accounting for apparent recombinations,
while necessarily sacrificing power to detect linkage. For
comparison, a LOD score estimate is also given for the PAX3
to W81 evaluation, where the penetrance for W81/+ genotype
is set to 95%, the penetrance for +/+ is set to O, and the
gene frequency is assumed to be l/42,000|58L

This family was subsequently assayed for two chromosome
11 markers, CD3D and D11835, both of which map near to the
N-CAM (Neural - Cell Adhesion Molecule) locus [39,60]. The
reasons for this are two—fold. First, control of N—CAM
expression is a prominent event in the differentiation and
migration of neural crest cells, and thus is a likely
downstream target of the PAX3 gene product. This hypothesis
is supported by the observation that N-CAM expression is
altered in the neural crest cells of Sp/Sp mice pun. Second,
Tom Strachan and co-workers reported a LOD score of 2.0 for
linkage between W82 and markers near N-CAM in a British
family at the Genetics of Hearing Impairment Meeting in San
Diego, CA in May, 1992. Linkage analysis with both markers,
CD3D and D11835, yield negative LOD scores in MSU4 (Table
2). Further linkage analysis of this family was suspended

due to the lack of reliable diagnoses

2. Linkage analysis of MSUS.
A summary of the phenotypes of MSUS individuals is
given in Table 3, and the pedigree structure is illustrated

in Figure 2. A more thorough discussion of the clinical

l4

characteristics of affected individuals in M8U5 is given in
Asher et al.flfl. The W81 locus was previously mapped to 2q37
in this family prior to the identification of PAX3 as a
candidate locus. A multi—point analysis with several RFLP
markers, including one for the Alkaline Phosphatase -
Placental locus (ALPP), was used to increase the
informativeness. ALPP showed the greatest evidence for

linkage with a LOD(Z) = 2.09 at 8 = 0.0.

Table 3. Phenotypes of Affected
Members of Family M805

.5,

Bed

+ +-++-+ +-+

...
+
...
+
+
+
+
+
+
+
+
...
+

 

fwd: = Dwmwheuflmmm(&flmmasdb>Qflmm
SN = Sensorineural deafness.
Hal = lkwflxmomhhflms
HypS = Hypopigmentation of the skin.
“mum; = WhmfimumXamMXQMygmmm.
Syn = Synophrys.

Individuals have a/b measurements that lie in the region of overlap
hmmwnmemmmnmmamwmdmmMNMnmwmmmmsHmmxmd
both would be the offspring of non-penetrant carriers.

15

PAX3 and ALPP show linkage in MSUS, with a LOD(Z):
1.226 at 9 = 0.11 (Table 4). A LOD(Z)= 1.32 or 1.97 at 9 =
0.0 is generated for comparisons between W81 and ALPP,
depending on whether the phenotype is assigned or a
quantitative measurement is used for the disease locus; in
this case the a/b measurement. This ratio of the inter-
inner-canthal distance (a) over the inter-pupillary distance
(b) has a mean of ~0.535 in normal Caucasians, and ~0.695 in
those affected with W81 [3,16]. The convention suggested by
Cotterman [NH has been to consider a/b measurements greater

than 0.600 to be evidence of dystopia canthorum.
II (DIE II

2 I 1
17 18 5

"$591331?

1
25 in 26 27 28 30 31 32 34 35 36 37

 

Figure 2. MSUS pedigree. Conventions are as for Figure 1.

Phenotypes are assigned by Waardenburg consortium criteria
(see Appendix A). The a/b measurements are given in Table 82.

16

The enhanced ability to detect linkage using a
quantitative measure has been demonstrated before by Lange
et aifl[3n, who used measures of creatine kinase levels to
more accurately map the Cystic Fibrosis locus. Part of the
power of this method is in situations where individuals have
measurements that lie at the extreme end of a normal
distribution, such as individuals M8U5—37 and M8U5—39. Both
individuals have a/b measurements that would classify them
as dystopic according to Cotterman's criterion [Hfl. However,
both are also the offspring of normal parents, necessitating
two non—penetrant obligate carriers in the pedigree (M8U5-13
and M8U5—15). Assignment of M8U5-37 and MSU5-39 as affected
forces M8U5-13 and M8U5-15 to be classified as an obligate '
carrier, and MSUS-IS (but not necessarily M8U5-13) thus
represents an obligate recombination of W81 with respect to
ALPP. The overall result is a LOD(Z): 1.03 at O = 0.08. An
analysis using the a/b measurement as the marker for the
disease locus yields a higher LOD score partly because the
Linkage program is able to include in the calculation the
probability that M8U5-37 and M8U5-39 are actually normal
with a/b measurements at the extreme end of the normal
range. The only obligate recombination between W81 and ALPP
in the pedigree is then mitigated.

The added robustness of a quantitative measure due to
the ability to resolve possible mistypings is illustrated
again in the W81 to PAX3 comparisons. The LOD(Z) score with

regard to PAX3 also is substantially higher (3.6 at 9 = 0.0

17

vs. 1.58 at 0 = 0.013) if the W81 phenotype is measured as a
function of a/b rather than assigned. In this case, both
MSUS-13 and M8U5-15 would have been designated as obligate
recombinants with respect to PAX3, due to the greater
informativeness of the PAX3 marker, if M8U5-37 and M8U5—39

were assigned as affected.

Table 4. MSUS Linkage Results

 

 

 

 

 

 

 

 

LOD Scone at Recombination Fraction
Comparbon 0.0 0.05 0.10 0.20 0.40
PAX3 - Aff. -2.88 0.18 0.65 0.91 0.00
PAX3 - a/b 3.60 3.33 3.05 2.76 0.95
PAX3 - ALPP .oo 1.00 1.22 1.05 0.26
ALPP - Aff. 092 0.87 0.90 0.66 0.08
ALPP - a/b 1.97 1.75 1.52 1.04 0.15

 

 

 

 

 

 

 

3. Linkage analysis of MSU6.

The assignment of phenotypes in MSU6 is particularly
difficult. This is reflected in the three different versions
of the pedigree displayed in Figure 3. As is reported in
Table 5, two individuals have a/b ratios over 0.600
indicating dystopia canthorum. By virtue of these two (MSU6-
2 and MSU6-20) this would be classified as a W81 pedigree.
Applying the Waardenburg consortium rules regarding
classification (see Appendix A, Table A1) [20], would result
in a pattern of inheritance represented in Fig. 3A.
Individual MSU6-9 was not directly inspected, and thus his

phenotype is classified as “unknown“.

18

Table 5. Phenotypes of Members of M806

a/b W Other
1.75 a

.646 2.30 b

500 1.63 -

.500 1.63 -

.564 1.95

.491 1.56

.561 1.87

.484 1.59

? ?

.583 2.05

.459 1.46

.518 1.69

.508 1.66

.518 1.69

.500 1.58

.545 1.80

.509 1.66

.509 1.67

.521 1.66

.653 2.22

.583 1.97

\DmxlOKUthJNt-tg

 

db = a/b ratio with measures that qualify for dystopia canthorum in bold. ‘
W = Measures in dystopic and non-apparent dystopic range in bold.
SN = Sensorineural deafness.
WFIEG = White forelock and/or early greying.
Skin = Skin hypo- or hyper- pigmentation.
Other:

2 sibs with poliosis; mother had 3 miscarriages.
Dizziness; 3 sibs died in infancy.

Hearing loss progressive; frequent constipation.
Hypopigmentation on right leg possibly due to trauma.
Multiple sclerosis.

Distinctly pale eyes, not seen by ophthalmologist
Severe constipation in infancy.

Curved spine.

Double hernia at birth.

Hearing loss progressive; frequent constipation.
Frequent constipation; Treacher-Collins like facies.
Ears rotated.

However, the assignment of phenotype using this method

is problematic in this family due to the ambiguities

19

 

 

 

 

A.
1 2
4 3 5 6 8 7 9 10 11 13 12 r 14
15 17 16 18 19 20 21
B .
1 2
4 3 5 6 8 7 9 10 11 13 12 7 14
15 ‘ 17 16 18 19 20 21
C .
1 2
I 8 7 9 10 11 13 12 7 14
15 18 19 20 21

 

Figure 3. MSU6 pedigree structures. A. Phenotype assignment according to
rules established by Waardenburg consortium (see Appendix A, Table A1).
MSUG-9 is classified as "unknown“, but reportedly has dystopia
canthorum. B. Phenotypes assigned according to W—statistic and including
the NAD category as proscribed by Arias [2]. c. Phenotypes are assigned
by combination of A and B. MSU6-10 is “unknown” because she meets the W-
statistic criterion only. M8U6-15 is 'unknown' because of some
indication of mild aganglionic megacolon but otherwise normal phenotype.

20

regarding the inheritance of certain of the characteristics.
In particular, individuals MSU6-18 and MSU6-19 could be
classified as affected because each exhibits congenital
deafness and has a first-degree relative who is otherwise
diagnosed as W81 (MSU6—7, who has a white forelock and
reports progressive hearing loss). However, the father,
MSU6-8, has hearing impairment and reports a family history
of deafness that may be genetic. Thus, the source of the
deafness in MSU6-18 and MSU6-19 is questionable, and so then
is the assignment of genotype, since that assignment relies
upon the presence of hearing impairment.

Similarly, the presence of a white forelock in MSU6—7
and a first degree relative otherwise diagnosable as having
W81 (MSU6-2) would indicate that MSU6—7 is affected.
However, several sibs of MSU6—1 also have a white forelock,
thus this trait may be segregating in the family
independently of a W81 gene. Alternatively, this suggests
the possibility that MSU6-1 is a non-penetrant carrier for
W81, and that M8U6—2 is a phenocopy. There is some merit to
this hypothesis, since MSU6—2 also has hypertelorism (c
measurement = 90mm) , which Waardenburg [58] suggested was an
entity distinct from dystopia canthorum.

A resolution to these complications would be to rely on
a quantitative characteristic as the measurement of
phenotype. As was the case in the analysis of MSUS, this
creates the opportunity for a computer program to weigh the

relative probabilities that an individual is affected versus

21

assigning an all-or-none genotype for the disease locus.
However, as noted above, there are only two individuals who
exhibit a/b ratios in the affected range. A linkage analysis
using the a/b measurement would not be robust enough.

Arias and Mota EH described a tri—modal population
distribution of the W-statistic, and were able to use the W-
statistic to reliably differentiate between affected (W >
2.07) and non-affected (W < 1.89) individuals in previously
diagnosed families. Furthermore, the non-penetrant obligate
carriers in these families invariably had W—values between
2.07 and 1.89, constituting the third modality of the
distribution. This population was designated as Non-Apparent
Dystopic (NAD). When W-values are calculated as suggested by
Arias and Mota, and phenotypes are assigned according to
their criteria, the result is a pattern of inheritance
represented by Fig. 3B. This raises the possibility of a
more robust analysis, with more ”affected“ individuals, but
creates the additional problem of a second source for the
mutation in individual MSU6—10.

A third method to more reliably assign phenotypes is to
combine the consortium criteria with the W-statistic
evaluation. The pattern of inheritance that results, with
two modifications, is represented in Fig. 3C. Under this
scheme, MSU6-10 is assigned an "unknown“ phenotype since she
meets the criteria for W81 by the W-statistic evaluation but
not by any other criteria. MSU6-15 is assigned an “unknown“

phenotype since she reportedly has had severe episodes of

22

constipation, which may be indicative of some degree of
aganglionic megacolon, but no other indications of WS. Three
other members of the family also report repeated episodes of
constipation as a clinical feature (MSU6-7, MSU6—19, M8U6-
20), although this is a common complaint and need not have
any relation to aganglionic megacolon. These individuals

were not diagnosed using this as a criterion.

Table 6. MSU6 Linkage Results

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LOD Score at Recombination Fraction
Comparison 0.0 0.05 0.10 0.20 0.40
PAX3 - Aff. (D -10.73 -3.21 -1.92 -0.76 -0.01
PAX3 - Aff. (II) -4.94 -2.37 -1.57 -0.69 0.00
PAX3 - Aff. (III) -1.72 -1.66 -1.54 -0.98 0.10
PAX3 - Aff (IV) -1.73 -1.53 -l.13 -0.53 0.00
PAX3 - W (V) -6.06 -2.59 -l.55 -0.60 0.00
I = Phenotype assigned according to consortium std.;WS1 parameters.
11 = Phenotype assigned according to consortium std.; W82 parameters.
111 = Expanded phenotype with Father as source of mutation; W82 parameters.
IV = Expanded phenotype with Mother as source of mutation; W82 parameters.
V = Phenotype assigned according to W-statistic; WSl parameters.

Linkage analyses of the disease locus to PAX3 were
performed under the three models discussed above. In
addition, the analysis was conducted using the more relaxed
W82 parameters; allowing more latitude for non-penetrant
genotypes and phenocopies. Further linkage analyses were
also conducted assuming that MSU6-1, and not MSU6-2, was the
source of the mutation. Under all hypotheses and conditions,
obligate recombinations between W81 (W82?) and PAX3 occur

(Table 6).

23

4. Linkage analysis of UGMl.

UGMl is an Indonesian family identified by Dr. James
Asher, Jr. during a survey of schools for the deaf on the
island of Java (Figure 4). The designation 'UGM" reflects
the collaborative efforts of individuals from the University
of Gadjah Madah, Indonesia in identifying and diagnosing
individuals with Waardenburg syndrome, and in obtaining DNA
samples for linkage analysis. UGMl is an extensive pedigree,
comprising 101 individuals. Only a portion of the pedigree
is represented in Figure 4. This “clan“ of 25 individuals is
the only subgroup of UGMl that segregates for dystopia
canthorum (Table 7), and was thus chosen to be the

experimental group for all linkage analyses.

Table 7. Phenotypes of Affected
Members of Family UGMl

Skin

 

+
+
+
+
+
+
+
+
+

Dmmh= Dfimmhcmanmrmb>Oefll
Bet] = Heterochromia irides.
SHm=:Ekmhnxboflupypmmammmt
SN = Sensorineural deafness.
* = Married into family. Classified unaffected by consortium standards.

45TH.

W

a ..
:12; 5.5.55

85 85 67 58 69 70 7172 73 747

 

 

 

mo—
30—

Figure 4. UGMl pedigree. Phenotype is indicated by shaded quadrants:
upper right = dystopia canthorum; lower right = heterochromia irides;
lower left = hypo- or hyperpigmentation; upper left = sensorineural
deafness.

Given that this is a family with mongoloid ancestry,
the question immediately raised is whether the shortening of
the palpebreal fissures observed is within the normal range
in a population that would tend to exhibit pronounced
epicanthal folds, or is a manifestation of dystopia
canthorum. To address this question, the a/b measurements of
69 normal Indonesians who married into family UGMl and
another Indonesian family, UGM2 (discussed in Chapter 2),
were compared to those of 516 normal Caucasoid EH. The
results of two-tailed, unpaired T-tests reveals that there
is a significant difference between the a/b measurements of
normal Indonesians and normal Caucasians (Table 8). There is
still a significant difference, however, between Indonesians
otherwise classified as affected for W81 and normal

Indonesians. Linkage analyses were performed, therefore,

using a quantitative measurement for dystopia canthorum

25

with means and standard deviations for the a/b measurement

calculated for the Indonesian population.

Table 8. Comparisons of a/b Ratios Between
Indonesians and Caucasians

 

 

 

 

Affected Normal Normal
Indonesian Indonesian Caucasian P - V3111.
(n = 15) (n = 69) (n : 516)
mean (std.dev.) mean (std.dev.) mean (std.dev.)
-- 0.553 (0.037) 0.532 (0.034) < 0.0001
0.684 (0.037) 0.553 (0.037) -- < 0.0001

 

 

 

 

 

Estimates of linkage between W81 and ALPP were largely
uninformative, but yielded negative LOD scores for absolute
linkage whether the disease state was assigned or calculated
from the a/b measurement (Table 9). The smaller LOD score of
-1.53 obtained from the a/b measurement analysis versus —0.8
obtained from the assigned phenotype analysis is presumably
attributable to individual UGMl-28, who has an a/b
measurement of 0.677, but is assigned an unaffected status
due to the lack of any other of the criteria, including
first degree relative diagnosed with W81. But a closer
inspection of the genotype data suggests that the results
are actually counterintuitive. UGM1—28 has the 3181 genotype
for the ALPP RFLP marker, and thus does not contribute
information to the linkage analysis, yet the magnitude of
the LOD score nearly doubles with the change of affection
status. The most likely explanation for this phenomenon is
that the probability of detecting an obligate recombination

with regard to individual UGM1-63 actually increases using

26

the a/b measurement, while the status of UGMl—28 is not
important. This is confirmed by repeating the W81(Affection)
- ALPP linkage analysis assigning UGM1—28 the status of
affected. and by repeating the W81 (a/b) - ALPP analysis
with UGMl-63 assigned the measure of 0.00 (unknown). In the
former case, the LOD score at 0: 0.0 is -0.86, exactly the
same as if the affection status were not changed. In the
latter case, however, the LOD score at 0 = 0.0 is +0.90.
This suggests that the critical feature in the data set is
the a/b measurement of UGM1-63, and that the reliance on a
quantitative measure for the disease state increases the
power to detect an obligate recombination resulting in the

genotype for UGM1-63.

Table 9. UGMl Linkage Results

 

 

 

 

 

 

 

 

 

LOD Score at Recombination Fraction
Comparison 0.00 0.05 0.10 0.20 0.40
Aff. - ALPP -0.80 -0.14 0.01 0.07 0.00
a/b - ALPP -1.53 -0.19 0.01 0.06 0.01
Aff. - PAX3 2.70 2.48 2.25 1.74 0.57
a/b - PAX3 2.68 2.46 2.22 1.72 0.56

 

 

 

 

 

Based on this analysis, UGMl would have been classified
as one of the 55% of W81 families that do not show tight
linkage between the disease locus and ALPP EHH. However, the
results of the analysis using the PAX3 marker indicate
strong evidence for linkage between W81 and PAX3 (Table 9).
The LOD(Z) at 0 = 0.0 is hardly affected whether the disease

state is assigned (LOD(Z) = 2.7) or calculated from the a/b

27

measurement (LOD(Z) = 2.68). This result emphasizes the
importance of not relying too heavily on the estimates for
genetic heterogeneity made by Farrer et al.EHH since the

marker locus used for those analyses is clearly inadequate.

28

C. Discussion

As Asher and Friedman BU stated in consideration of
mouse models for Waardenburg syndrome(s), there are two
possible hypotheses for the observed clinical heterogeneity
and variable expressivity in W8: 1) there are multiple loci
for W8, each of which when mutated can give rise to the
phenotype; 2) there is a single W8 locus, and the phenotype
is modified by multiple loci. Separate linkage analyses in
single families would at once avoid the pitfall, suggested
in the first hypothesis, of pooling heterogeneous data,
resulting in false exclusion of chromosomal regions, while
simultaneously testing the validity of both hypotheses.

The evidence of the several mouse models mentioned in
the Introduction best supports the first hypothesis. Each
mouse mutation segregates independently and causes some
aspect of the clinical constellation defining the syndrome,
but does not in itself account for the full range of the
phenotype (to the extent that any have been assayed for the
full range of clinical features). However, it must be
remembered that laboratory mouse strains are highly inbred,
and it is likely that the alleles at putative modifying loci
would be fixed in the population. The fact that only a
subset of the possible traits segregates in a strain may be
due more to a fixed set of modifying loci in each strain
than it is to the presence of multiple "W8" genes.

Support for the second hypothesis comes from the

observation by Dr. James Asher (unpublished data) that when

29

the Spd allele (which originated in a MUS musculus
background) is crossed into a.MUs spretus background, the
Iresulting hybrids exhibit dystopia canthorum, a trait not
previously observed. Thus, the penetrance of the gene for
one of the traits depends on the genetic background in a
strain. Since the various W8 clinical types are defined by
the inclusion or exclusion of certain traits comprising a
rather broad set, the particular subset of clinical features
exhibited by a given family may be largely dependent on
allelic states at other loci.

The identification of Fax-3 as the Sp locus did not
resolve this issue [18,19]. As a probable transcription factor
acting on numerous downstream genes expressed in neural
crest cells [27], mutations in PAX3 might be responsible for
the full spectrum of WS clinical features. In addition to
the several mutations in PAX3 that are purported to give
rise to W81, there are now PAX3 mutations that presumably
give rise to W82 [55] and W83 [32]. The reported W82 case is
arguably W81, since one of the family members has a W-index
value > 2.07. The W83 diagnosis, however, is clearly
accurate, since several individuals in the family have
upper—limb deformities.

With regard to the relationship between PAX3 and the
various W8 subtypes, there are three possibilities: 1) It is
possible that the variance and clinical heterogeneity of WS
are due purely to the normal chance events of a mutant gene

exerting its effects on a large population of.cells that

30

must traverse through, and interact with, a complex
environment. 2) It is also possible that mutations of PAX3
are the primary causes of all W8, and the variance and
clinical heterogeneity is due to the allelic states at
various downstream target genes. The former and latter
possibilities are, of course, not mutually exclusive. But it
also remains a possibility that 3) a mutation at one or
several of the hypothesized downstream targets could be
necessary and sufficient to account for some instances of
W8.

Observed recombinations between a PAX3 allele and the
"disease locus" when the resulting phenotype is normal
cannot distinguish among these three possibilities. The
recombination may not be real, but only apparent because of
non-penetrance or difficulty in diagnosis. Or the phenotype
may be normal due solely to chance events suggested under
possibility number 1 despite the presence of a PAX3
mutation. But if a recombination occurs between a PAX3
allele and the disease locus in the case that the resulting
phenotype is unambiguously classifiable as W8, then the best
explanation (barring a new mutation or phenocopy) is that
another W8 locus exists.

The linkage analyses of MSU4 and MSU6 both provide
evidence of the latter sort. The evidence in MSU4 is.
stronger because of the difficulties in assigning
phenotypes, or even determining the number of mutant genes

segregating, in MSU6. In MSU4, an obligate recombination

31

occurs when one considers the PAX3 genotypes among reliably
diagnosed affected individuals. When a more complete
clinical examination is made of the individuals in MSU4
additional recombination events may be detected. This
pedigree might then be suitable for detecting positive
evidence for linkage of genetic markers to a W82 locus.

To date, the best evidence for genetic heterogeneity of
WS is Farrer et al.EflH. This analysis of a pooled data set
comprising 44 kindreds, including MSUS, found evidence for
genetic heterogeneity when considering linkage of W81 to the
RFLP marker for ALPP. However, there are at least two
weaknesses to the analysis that argue for withholding
judgement on this issue. Both are illustrated in the linkage
analyses of MSUS and UGMl.

First, the consortium analysis relied on the
recombination information provided solely by ALPP. It now
appears that ALPP is ~10 cM from the W81 locus. As can be
seen from a comparison of the analyses of UGMl and MSUS, it
is quite possible to obtain a negative LOD score with
respect to ALPP in one family and a positive one in another
family, when both families give positive and significant LOD
scores with respect to PAX3. Also, the RFLP marker for ALPP
is often uninformative. Moreover, there is a pseudogene of
ALPP whose location is unknown [42], and an ALPI locus on
2q35 that presents a similar restriction pattern on Southern
blots to ALPP. There exists, then, considerable opportunity

for mistyping of the ALPP locus. The consortium is currently

32

collecting PAX3 genotyping data for the entire data set, as
well as data for other highly informative markers thought to
be near PAX3.

The second weakness of the consortium analysis is that
the criteria for assignment of phenotype may not have been
consistently applied. This is unavoidable in any
collaborative effort, but the diagnoses for W8 are
particularly difficult since few of the characteristics can
be quantified. Ideally, the phenotypes would have been
quantified, both as a means of ensuring uniformity in
clinical classification, and as a means of adding robustness
to the linkage analysis. As demonstrated in the linkage
analysis of MSUS, the ability to assign probabilities for
individuals to be either affected or at the extreme end of a
normal distribution can have a significant impact on the
analysis. Two obligate recombinations with respect to PAX3
in MSUS were mitigated by consideration of the amount of
overlap between the a/b measurement distributions of the
normal and affected populations.

The best assessment, then, of the degree of genetic
heterogeneity in W8 will be the consortium analysis
currently being done. Ideally, this analysis will include a
quantitative measure as a marker for the disease locus
whenever possible. Most important will be the identification
of obligate recombinations involving individuals who clearly

have the W8 phenotype.

33

II. CHAPTER TWO: Mutations in PAX3 associated with

Waardenburg syndrome.

A. Introduction

The mapping of W81 mutations to 2q37 [3,22] gave greater
weight to the proposition that the mutation Sp was the
murine equivalent of W81.PH When Epstein et al.U8]
subsequently demonstrated that the allele Sp2H was a
mutation of the Fax-3 gene, the problem of identifying the
W81 gene was transformed from one of positional cloning to
that of_testing a candidate locus. Tassabehji et al.[&fl and
Baldwin et al.[U demonstrated sequence changes in human PAX3
in individuals with classical W81 soon after. Both mutations
created missense messages: an 18 bp deletion retaining the
reading frame and a proline to leucine replacement. Both
mutations occur in the highly conserved paired box, which
encodes a DNA binding domain [57], while presumably leaving
intact the rest of the protein, including the two other
motifs common in DNA-binding proteins (an octapeptide
sequence and a homeo box) [27]. The effects that these
mutations would have on normal PAX3 protein function are
necessarily speculative, since the target genes of this
transcription factor have not been identified. The question
remains open, then, as to whether the mutations exerted
their dominant effects by: 1) producing DNA-binding proteins
with altered specificity; 2) producing mutant proteins that

interact with normal proteins to produce a negative-dominant

effect; or 3) producing functionally negligible proteins
resulting in haploinsufficiency in a concentration sensitive
milieu. The most direct method for addressing this question
is to characterize additional mutations that may suggest the
mode of molecular pathology.

One strategy for detecting mutations is to sequence the
PAX3 gene in affected individuals. The effort involved would
be considerable given that the PAX3 cDNA is expected to be
approximately 1,437 nucleotides long based on the murine
cDNA [27]. However, the only human sequences available at the
time of this study were represented by HuP2, a genomic clone
containing three of the predicted five exons, or 134 of the
predicted 479 codons [hfl. Given the fact that an assay for
possible mutations in all of the coding and non-coding
sequences of PAX3 was not possible, and given the estimate
that mutations in 55% of families are not linked to 2q37
markers BAH, a more efficient method for detecting possible
mutations in the three available exonic sequences of PAX3
was necessary.

Our strategy for identifying W81 mutations in PAX3 was
three-fold. DNA fragments containing exons 2, 3 and 4 of
were amplified by the polymerase chain reaction (PCR) and
either: 1) electrophoresed on native polyacrylamide gels to
detect heteroduplex DNA, which are DNA fragments formed from
the mismatched strands of wild type and mutant alleles, and
exhibit different mobility from homoduplex DNA due to

looping of the mismatched sequences; or 2) denatured and

35

then electrophoresed on native polyacrylamide gels to detect
single strand conformation polymorphisms (SSCPs), which
occur due to the assumption of sequence-specific secondary
structures as the denaturing solution carrying the single
strands becomes increasingly diluted during migration
through the gel. A major pitfall of any PCR-based technique
to detect mutations is the possibility that the mutant
allele will not PCR amplify because of a deletion or
mutation in the primer sequences. Therefore 3), 32P — dATP
labeled PCR fragments amplified from wild-type exons of PAX3
were used as probes to Southern blots of endonuclease
restricted genomic DNA in order to detect large
rearrangements, potentially involving intronic sequences or

deletions encompassing the PCR primer sites.

36

B. Results

Representative DNA samples from thirteen different
families segregating for Waardenburg syndromes were included
in the SSCP analyses (Figure 5). In each case, the proband
of a family was used, with the exception of MSU6. For MSU6
both the parents were used (MSU6-54 and MSU6-55) because of
uncertainty of the proband's phenotype and the uncertainty
of the origin of the mutant allele. For MSU4 a distantly
related individual (MSU4-25) was included in addition to the
proband (MSU4-4) because of the severity of the phenotype
and the possibility that more than one mutant allele was
segregating in the family. Table 10 summarizes the
phenotypes of individuals whose DNA was used in this
analysis.

Initial results comparing heteroduplex gels with SSCP
gels demonstrated the SSCP technique to be more sensitive in
detecting sequence variations, and heteroduplex assays were
discontinued. SSCP gels run at 8 Watts at room temperature
were comparable to gels run at 12.5 Watts at 49C, and only
the former are shown in Figure 5. Single strand conformation
polymorphisms for the exon 2 containing fragment (PCR
primers TF30 & TF32) were observed for MSU3-9, MSU6-55, and
UGM2-24. The polymorphism for MSU6-55 was shown to be due to
an intronic Hae III polymorphisnlfn by restriction analysis.
No polymorphisms were detected for exon 3 or exon 4 PCR

fragments.

37

”a? 5‘3. 5‘ ,(s ,0 1:1” ~31“ 5,0 ,
«9 “1° s2» 19°" .19“ .8 £18 1°18. a s‘” .689 i" 2‘"

' ' i
9". M 1 : ...-... .... p _ ._ p
”a a , < , ‘ ,
, '5‘“; b:— 5 , i _ , > ' 7 .11" H W ‘ 1"
Aha-‘9 . -..»; 1 .. _ L . .34 u . . .' __.>.,_
H t . in“

 

Figure 5. SSCP Analysis of P313 Exon 2 PCR Fragments in Affected
Individuals. The 35S-dATP labelled PCR fragments were generated from
genomic DNA using primers TF30 and TF32 (see Table A2). Individual IDs
are indicated above lanes. All individuals are the probands for the
given W8 family, with the exceptions of: MSU4-25, which is a severely
affected third cousin of MSU4-4; and M806-54 and MSU6—55, which are the
parents of the proband M8U6-70, and are substituted because of
uncertainties regarding phenotype for MSU6—70 and uncertainty of the
origin of the mutation. The normal 535 bp PCR fragments migrate in a
simple two—band pattern when electrophoresed through 1x MDEE gels (AT
Biochem). Fragments generated from other alleles create additional bands
indicated by arrows. The faint additional bands from MSU3—9 are due to
an allele containing a 1 bp insertion. The extra band from UGM2-24 is
due to an allele containing a 14 bp deletion. The extra set of bands
from MSU6—55 are due to an allele containing a nucleotide substitution
in the intron sequences of the PCR fragment.

38

Table 10. Phenotypes of WS Subjects

Person ID a/b netI lr/so Broad Nasal SN
Root

MSUl-3 . X
usuz-7 . X. X
MSU3-9 . X X
MSU4-4
MSU4-25'
-21
6-54
usue-Ss
MSU7-6
usu9-c X
UGM1-66 . X“
UGM2-24 . X”
UGM3-1l . X" X
com-1 . X" X
_1 . x0.
* Other characteristics include: cleft lip and palate, microcephalus,
absence of nasal bone, left hydronephrosis, small penis, mental
retardation. 1
** Indicates isochromiac hypoplastic blue eyes.
# Mentally retarded, abnormal facies, abnormal gait.

>4

>4

 

>41K >41” >438 N154

The exon 2 PCR fragments for MSU3-9 and UGM2-24, exon 3
fragments for UGM2—24, and exon 2, 3 and 4 fragments for
M8U5-21 and UGM1-66 were cloned into the pCR 1000
(Invitrogen) vector and both strands of multiple clones were
sequenced.

Exon 2 of the mutant allele in MSU3-9 contains a
guanine insertion at the first nucleotide of the paired
domain (Figure 6). This insertion would presumably cause a
shift of the reading frame and the creation of a premature
translation codon in exon 3, resulting in a truncated
protein product lacking any functional paired domain,
octapeptide sequence, or homeodomain. This insertion creates

an AvaI restriction site (CTCGGG) in the mutant allele.

39

Figure 6. Sequence of Mutant PAIB Allele Responsible for W81 in Family
MSU3. The left panel shows a partial sequence from the normal allele
PAX3 PCR amplified from individual MSU3-9. The right panel is the
sequence from the mutant allele PCR amplified from M8U3—9. The guanine
insertion is indicated by an asterisk. Both strands of multiple
independant clones representing the mutant allele were sequenced.

40

G0

GCAT

JJ: 2

.3 .5
*
TGCGCCGGGA CCGGG CTCCCC

1., _

O
. ...” L . _
. f .. r
. .... . ... 1.. s , ._ ....
.u . . . ., , I.
.. . H l w. ...
.... . ... W fl. . mm. a
...u. . , . ._ . .. .... ... 1: ,
i a .... . . . ,. t. I... .
_ . . 1 l. I. .
._ .. ..., 1 .
. . ... u

. TGCEGGACCGG CT. CCCC.
3 5

41

Figure 7. Confirmation of the Segregation of a Mutant Allele in Famdly
MSU3. A. The pedigree structure of MSU3 with individual symbols
overlying the corresponding lanes for.B and C. Phenotypes are
represented as filled quadrants in the pedigree diagram: upper right =
dystopia canthorum; lower right = heterochromia iridis; lower left =
hypo pigmentation; upper left = sensorineural deafness. 8. Silver-
stained gel showing the SSCPs of the exon 2 PCR fragment amplified from:
1) MSU3-10; 2) MSU3-9; 3) M8U3-8; 4) MSU3-15; 5) MSU3-16 (blank); 6) an
individual previously demonstrated to be heterozygous for a HaeIII
polymorphism described by Hoth et al.[32]. Thus, MSU3-10 exhibits the
normal pattern; MSU3-9 exhibits an additional band presumeably
representing the fragment from the mutant allele; MSU3-8 also exhibits
the additional band, and a lower band which is a result of the
polymorphism at an intronic HaeIII site; MSUB-IS is heterozygous for the
HaeIII polymorphism. The extra G occurs at the first nucleotide of the
paired box and creates an AvaI digestion site. C. PAX3 exon 2 fragments
PCR amplified from genomic DNA samples from affected and unaffected
individuals in family MSU3. Fragments were digested with AvaI and
electrophoresed on a 3% Nusieve 3:1 gel in TBE buffer. PCR fragments
from the normal allele migrate at 535 bp. Fragments from the mutant
allele migrate at 289 bp and 247 bp. Lane 5 contains a molecular weight
marker with sizes indicated at right.

42

 

43

Therefore, we confirmed that this insertion is not a random
PCR or a cloning artifact by AvaI digestion of total PCR
amplified DNA and electrophoretic separation on an agarose
gel. Genomic DNA amplified from unaffected individuals and
digested with Ava I displayed only a single band of 535 bp,
while that of affected individuals display a 535 bp fragment
representing the normal allele, and two fragments of 289 bp
and 247 bp representing the mutant allele (Figure 7).

Exon 2 of the mutant allele in UGM2-24 contains a 14
base pair deletion beginning at nucleotide 167 of the paired
domain (Figure 8A & B). The deleted segment includes a RsaI
restriction site beginning at nucleotide 168. All available
DNA samples from the UGM2 family were subsequently assayed
on a Southern blot (Figure 9) probed with 32P—dATP labeled
exon 2 PCR fragments. Perfect cosegregation of a unique 2200
bp fragment with the mutant phenotype was observed (Lod(Z) =
3.82, 0 = 0.0). The 2200 bp fragment was not seen in 18
normal unrelated Indonesians nor in 16 normal unrelated
Americans in subsequent experiments. This deletion causes a
frame shift in exon 2 which results in a UGA stop codon in
exon 3 (Figure 8C). The position of the stop codon was
confirmed by sequencing exon 3 from the same affected
individual (Figure 8D). The deduced translated product would
lack 57% of the paired domain and all of the octapeptide

sequence and homeo domain.

No SSCPs, heteroduplex formations, or sequence
alterations were detected in any of the exons PCR amplified

from individuals M8U5-21 and UGMl-66.

45

Figure 8. Sequence of the Intant PAI3 Allele Segregating in UGM2.

A. A comparison of the mutant and wild type sequences of the paired box
of PAX3. The nucleotide sequence begins with the first codon of the PAX3
paired box according to Burri. et a1 [hfl. The wild type amino acid
sequence is given in the top line, and the mutant amino acid sequence is
given in the bottom line. Dashes indicate amino acids identical to the
wild type sequence. The 14 bp deletion is indicated by arrows and
intervening ellipses. The new reading frame created by the deletion is
indicated by brackets beneath the nucleotide sequence. The exon 2/exon 3
border is indicated by <>. The RsaI restriction site deleted in the
mutant allele is shaded. The putative first alpha helix region is
underlined. B. A partial nucleotide sequence from the wild type exon 2.
C. Nucleotide sequence from the mutant exon 2. D. The exon 3 sequence
from an affected individual. The new termination codon in exon 3 created
by the frame shift mutation is indicated by *. The sequence data
represented in B, C and D were each confirmed for both strands of three
independent clones.

A

Normal Gly Gln Gly Arg Val Asn Gln Lcu (Ely (in Vul Phc llc Asn Gly Arg Pro
1 GCC CAC GGC CGC GTC AAC CAC CTC GGC GGC GT1" lll ATC AAC GGC AGO CCG
Mutant - - - - - - - - - - - - - - - - -

Lcu Pro Asn His llc Arg His Lys llc Val Glu Mct Ala His His Gly llc
52 CTG CCC AAC CAC ATC (‘GC CAC AAG ATC CTG GAG ATG GCC CAC CAC GGC ATC

Arg Pro Cys Val llc Scr Arg Gln Lcu Arg Val Scr His Gly Cys Val Scr
103 COG CCC TGC OTC ATC TCG CGC CAG CTG CGC GTG TCC CAC GGC TGC OTC TCC

Lys llc Lcu Cys Arg Tyr Gln Glu Thr Gly Scr llc Arg Pro Gly Ala llc
154 AAG ATC CTG TGC AOG TAC CAG GAG ACT GQC TCp A'EA CQT CCJ‘ GCfl‘ GCC AT,C

 

 

 

- - - - AI rg Lcu His Thr Scr Trp Cys His
Gly (311' So: Lys Pro Lys <> Gln Val Thr Thr Pro Asp Val '°°
205 GQC GQC AGQ AAQ C(‘p AAQ Cap 010 AQA AQG COT 06C CTG ...
Arg Arg Gln Gln Ala Gln A la Gly Asp Asn Ala Slop

U7
0
U

C A 234 5
(T: C G T A C G T A C G T A c
0
1' ..— — m T
0 ‘ fin. ' _ =-_-'= 6
C —l -. ~ A
A - — c

m 0 ° .‘II, a _—

C I” 5‘ G 167 A
c — ~ -.J‘ ‘— A
1‘ — .- 1- o c
C = “‘ G C -— C
A - . ' — ...-l" C c ru- 1'
G v v A A h...- .
a ‘ a.“ G G c.
‘ _, "" — ' G G A
6 ~ ’ C A .' II C
A _ ‘r G _ G
C a = — “ C A '7'" T
II) 1‘ ~ "" C C G
G ‘ "°" A T ll) - G
G —' ‘ —.- T — — 4
c I" A — — °
1- - — C I” 3' _ A
—- - ~--" A
C . e...— - .- fi 6
i _ 0 a“ A
A
7 A

A A 253 3

J I” C

47

Figure 9. UGM2 Pedigree and Southern Blot Detection of a Mutant PAI3
Allele. A Southern blot of RsaI restricted genomic DNA from members of
family UGM2, probed with 32P-dATP labelled exon 2 PCR fragments from a
wild type PAX3. This PAX3 probe hybridizes to a 1.8 kb fragment from the
wild type allele, and to a 2.2 kb fragment from the mutant allele.
Pedigree members are illustrated above the corresponding lanes.
Phenotypes are represented as filled in quadrants: upper right =
dystopia canthorum; lower right = heterochromia iridis; lower left =
hypo- or hyperpigmentation; upper left = sensorineural deafness.

48

 

 

 

“teatime

 

16 1O 17 10' 19

1O 11 12 13 14

-.-

1.8kb- . - ‘”

2.2 kb-

49

C. Discussion

The spatial and temporal pattern of expression of the
murine gene Pax-3 in developing mouse embryos is coincident
with regions in which neural crest cells originate and
migrate [27], which is consistent with the neural crest
origin of tissues that are altered in W81. Fax-3 contains
both a paired box and a homeo box, which are highly
conserved sequences that encode DNA binding domains, and is
thus thought to encode a transcription factor responsible
for pattern formation in the developing embryo. Thus, the
discovery that Sp2H is an allele of Fax-3 [18] promised a
possible resolution to the problem ofvariable phenotype in
Waardenburg syndrome. As a transcription factor, PAX3
proteins would interact with a variety of other genes, many
of which would be expected to be involved in normal neural
crest cell differentiation. The manner and degree in which a
mutant PAX3 would interact with these downstream target
genes could be heavily influenced by different alleles at
the downstream loci or different alleles of PAX3. Thus, a
gene affecting neural crest cell populations prior to their
migration and differentiation offers an explanation for the
pleiotropy of W81, and the candidate locus' status as a
transcription factor interacting with many other genes could
account for the variation in expressivity. This model could
possibly account for all of the clinical heterogeneity.
Conversely, the identification of the downstream targets for

PAX3 could provide the best candidates for other W8 loci if

50

the clinical heterogeneity in fact reflects genetic
heterogeneity.

A first step in dissecting the molecular pathology is
to assess the functional consequences of mutations on normal
protein function, and correlate mutant genotypes with
phenotypes.

Exons 2, 3 and 4 of PAX3 encode a paired domain of the
prd - gsb type. Exon 4 also encodes an octapeptide sequence
conserved in DNA-binding proteins, and exon 5 encodes a
homeodomain [27,32]. In other species, it has been
demonstrated that the paired and homeo domains have DNA—
binding properties, and are frequent motifs of transcription
factors involved in developmental regulation [27,28,50,57]. The
mechanisms by which these domains bind DNA and affect
transcription control have not been completely determined.

The paired box encodes three putative alpha helices U2]
of variable amino acid sequence, while the regions
surrounding the helices are highly conserved. Treisman et
afln[5fl demonstrated in vitro that disruption of the first
helix in the Drosophila prd gene abolished the DNA binding
capacity of the paired domain to a specific recognition
sequence, while various mutations in either the second or
third helices did not impair DNA binding of the paired
domain. Substitution of a glycine residue for serine at
position 48 in the highly conserved region 5' to the first
helix also abolished DNA binding. This is the same

substitution that occurs in the mouse gene Pax-l and causes

51

the undulated (un) phenotype [8,15]. Treisman et al. reasoned
that amino acid variation in the highly conserved regions
would result in loss of DNA binding activity due to changes

in the three dimensional folding of the protein.

Norma
{1th3-4;F8113TER}
{14bp689-93;F8108TER}
«upbeat-swarm;
{2hpA186;FS202TER}
{AsM7l-is}

«iflflflfl

flhdfllsu}

 

{Arm}
«Meme:

D_Imm
Dmimm

Figure 10. summery'of all‘WS mutations in PAI3 published to date.
Diagram represents the translated sequences of PAX3 according to
Goulding et al.[27]. Grey shading represents paired domain. The
octapeptide coding sequence in exon 4 is represented by hatched shading.
Black shading in exon 5 represents the homeo domain. Asterisks above
diagrams indicate points of deletions or insertions. Reading frame
shifts are represented by stippled shading, ending in new translation
termination codons. Vertical lines through the diagram indicate
positions of missense mutations. Allele designations for each diagram
are given at left. The nomenclature conventions are explained in
Appendix D. References for each mutation can be found with their
descriptions in the text.

In PAX3, the putative first alpha helix region of the
paired box extends from position 56 to 64 (numbering from

amino acid sequence determined by Hoth et al.[32]. The

52

deletion described by Tassabehji et aiJLSH extends from
codon 62 to 68 (PAX3{18pr62-68D. The missense mutation
reported by Baldwin et al.[H occurs at codon 50
(PAX3{ProSOLeu}) where, by analogy to the prd and Pax-l/un
mutations, it may effectively abolish the DNA binding
capacity by altering the tertiary structure surrounding the
first alpha helix. Because both of these PAX3 mutations
maintain the downstream amino acid Sequences, the functional

impairments of the PAX3 protein are not immediately

 

apparent.

The W81 mutation segregating in family UGM2 deletes 14
nucleotides beginning at codon 89 causing a frame shift and
thus a translation termination signal 19 codons later
(PAX3(14pr89-9LF8108TER}). The first alpha helix is left
intact in the mutant protein, while the remainder of the
paired domain, including the two helices comprising amino
acids 113 to 138, and the entire predicted homeo domain are
removed. The deduced mutant PAX3 peptide contains only 163
amino acids of the predicted total of 479.

It is possible that the truncated PAX3 protein in W81
individuals of UGM2 is unstable and is degraded rapidly, or
lacks the nuclear targeting sequences. In either case, the
result would be haploinsufficiency of the PAX3 protein.
However, if the truncated protein is transported to the
nucleus, it might retain its DNA binding ability by virtue
of an intact first alpha helix in the paired domain, and

could conceivably interfere with normal PAX3 functions. The

53

mutant PAX3 allele in UGM2 would then represent a trans
dominant negative mutation. A series of binding assays of
the normal and truncated proteins to known paired domain DNA
recognition sites would be required to assess the likelihood
of the latter proposition.

Since a full length cDNA for PAX3 is not currently
available, electrophoretic mobility shift binding assays for
normal and truncated PAX3 proteins are not yet possible.
However, Goulding et aJHEZH performed such assays for the
Pax—3 protein and demonstrated that it binds the e5 sequence
of the Drosophila even-skipped gene which contains
recognition sites for both the paired domain and homeo
domain. While both recognition sequences were necessary for
high affinity binding by Fax-3 and Pax-l proteins, truncated
Pax—l proteins are capable of binding DNA in vitro at low
efficiency [15,27]. The various truncated proteins they
assayed did not, however, include one in which only the
first helix in the paired domain was left intact. Thus, it
is conceivable that the truncated PAX3 protein lacking the
entire homeo domain could occupy normal binding sites, but
the question remains whether a single helix in the paired
domain is sufficient for DNA sequence recognition and
binding. It is possible that the truncated protein product
may compete with normal PAX3 proteins for binding with
associated factors that are necessary either for normal DNA

binding or for exerting normal transcription control. The

effects of reduced PAX3 production would thus be compounded
by reduced availability of associated factors.

If high affinity binding of the PAX3 protein to its DNA
target sequences requires the interaction with additional
factors, it is most likely that the truncated
(PAX3{l4pr89-91F8108TER}) protein lacks the necessary
binding domains. Whether this would result simply in the
loss of normal transcription control, or in the gain of
novel transcription control is impossible to assess at this
point.

The mutation segregating in MSU3 adds insight to this
question. It also causes a shift in the reading frame
(PAX3{1pr34,F8113TER}). However, the one base-pair
insertion occurs at the first codon of the paired box,
eliminating the reading frame for the entire paired domain.
The resulting peptide would lack paired domain as well as
lacking an octapeptide sequence and homeo domain. It is
highly improbable that this peptide retains any DNA—binding
capacity, although it is still possible that it can interact
with associated factors via the peptide coded for by the
first exon.

It is instructive to note that the Splotch phenotype in
mice can be caused by the complete deletion of a Fax-3
allele (Spr/+ H10J8L Another allele of Fax-3 that causes
the Splotch phenotype (SpZH) codes for a truncated protein
[Hfl. In this case, a deletion causing a frame shift resulted

in a premature termination codon in the homeo box downstream

55

from the paired box. The predicted protein product has a
normal paired domain and a partial homeo domain. The
phenotype of SpZH/+ mice appears to be the same as that of
Spr/+.

This relation of genotype to phenotype is consistent
with other known mutations in PAX genes. For example, two of
the three mutations that cause the dominant small eye
phenotype in mouse are nonsense mutations in Fax-6 creating
truncated proteins, and the third is a deletion heterozygote
[3”. Again, the phenotypes are indistinguishable. Aniridia,
a condition in humans homologous to small eye in mouse, is
the result of a homozygous deletion in one case, and a
heterozygous nonsense mutation in another [56]. These
observations suggest that the observed traits in the mutant
phenotypes are caused by insufficient dosage of functional
proteins.

The two mutations described in this thesis,
'PAX3{1pr34,FSll3TER} and PAX3{14pr89-93,F8108TER}, produce
virtually the same phenotype as those produced by
PAX3{ProSOLeu} and by PAX3{l8pr62-68}, even to the extent
that both the UGM2 family and the Brazilian family
investigated by Baldwin et al. have an unusually high
penetrance for hearing impairment (75% and >78%
respectively). All of the affected individuals show
classical W81 characteristics. The addition of the two
mutations described in this thesis, and others recently

published [19,55], argue strongly for haploinsufficiency to be

56

the primary cause of the W81 and Sp phenotypes. The
mutations PAX3{18pr62-68} and PAX3{ProSOLeu}, therefore,
probably result in functionally inactive proteins. The
variable penetrance and expressivity could then be
attributed to the chance events of organogenesis inherent in
a system where a population of migratory cells interacts
with a complex extracellular environment, which is in turn
influenced by multiple genes.

The reduced dosage effect could be magnified if PAX3
normally autoregulates its own expression, as has been
demonstrated for other homeo domain proteins such as be in
Drosophila HRH. The mutant allele might dilute the up-
regulation effects by failing to bind factors that promote
the further PAX3 production. In addition, a threshold level
of PAX3 activity may be needed before a cell can commit
itself to a particular developmental pathway. Reduced dosage
of PAX3, magnified by obstruction of normal up-regulation
mechanisms, might delay the attainment of threshold levels
beyond the boundaries of narrow developmental windows. This
would result in a defect of heterochrony, or the timing of
developmental events. In fact, neural crest migration from
neural tube explants is delayed in Sp/Sp mice [46]. It is
possible that the neural crest cells of mutant embryos
retain the capacity for normal migration and
differentiation, and that the surrounding tissues retain the
capacity to support neural crest immigration, but that the

tissues develop out of phase with each other. A similar

57

situation is seen in the white axolotl mutant, in which the
subepidermal matrix develops out of phase with migrating
melanoblasts [49].

As mentioned in Chapter 1, the clinical description in
the literature of Waardenburg syndrome allows for the
interpretation of a single gene defect with epistatic
influences. An observation in support of this idea is the
appearance of dystopia canthorum, a trait not previously
known to be associated with splotch, in mice when Spd is
crossed into a different genetic background than which it
originated in (J. Asher, unpublished data).

Given that the differentiation of numerous tissues are
affected by PAX genes, it nay be that differences between
organisms heterozygous for PAX point mutations and organisms
heterozygous for PAX deletions would be revealed upon closer
inspection of other traits in the phenotype. It is still
possible that the high variation in expressivity among W81
families is not due to complex epistatic interactions, nor
stochastic events, but rather to allelic mutations at PAX3,
some of which result in haploinsufficiency of PAX3 while
others result in varying degrees of a trans dominant
negative effect. One might expect that missense mutations at
PAX3 would produce more severe phenotypes, since the mutant
protein might have altered specificity thereby affecting
numerous target genes. The mutation PAX3{Asn47His} described
by Hoth et al.[32] lends weight to this hypothesis, since it

occurs in an individual with type 3 Waardenburg phenotype.

58

The is contrary to expectations, since most of the
previously described W83 cases were most likely examples of
a contiguous gene syndrome (see Introduction). But such
speculation should be tempered by the consideration of the
other four missense mutations and the classical type I
phenotypes that they produce.

In summary, there does not yet appear to be a clear
correlation between genotype and phenotype for the mutations
so far described. This both reflects the paucity of
information on normal PAX3 function and regulation, and the
likelihood of epistatic effects from other, unidentified,

loci.

59

III. Appendix A.

Materials and Methods

1. Evaluation of Subjects.

Unless otherwise specified, subjects were classified
according to the criteria established by the Waardenburg
consortiunltflfl. To be considered affected, an individual
must exhibit either: a) at least two of the major criteria
or b) one of the major criteria and two of the minor

criteria.

Table A1. Waardenburg Consortium Criteria

 

thor:

 

1. Sensorineural deficit

2. Iris pigmentary abnormality:
a. heterochromia irides
b. segmented iris/iris bicolor
c. "sapphire blue“ iris confirmed with slit lamp

3. Hair hypopigmentation
a. white forelock, at any age, need not be permanent
b. white hairs elsewhere on body

4. Dystopia canthorum, defined as W index > 2.07
(Note that this does not include the non apparent
dystopic {NAD} class defined by Arias and Mota to be
2.(Y7 < ll > l..89)

 

 

 

5. First degree relative otherwise classified as W81
Minor:

14 Congenital skin hypopigmentation

2. Synophrys

3. Broad high nasal root

4” Hypoplasia of lateral cartilages of the nose

ES. Premature greying, predominance of grey scalp hair

before age 30.

 

Other recognize traits that do not contribute to the assignment of
phenotype are: spina bifida; cleft lip or palate; Hirschsprung disease;
limb defects; congenital heart abnormalities; vestibular defects; low
anterior hair line; broad square jaw.

 

2. Preparation of Genomic DNA.

DNA was obtained from 20 ml of heparinized blood
samples. Blood cells were lysed by the addition of an equal
volume of 0.1 M Tris—HCl pH7.9, 1 mM EDTA, 20 mM NaCl, 4%
SDS. After a 30-min. incubation at room temperature, the
lysate was extracted with an equal volume of salt-saturated—
phenol (SS—phenol) with shaking for 10 min. Following a 20-
min. centrifugation at 1,800 g (2523), the aqueous phase was
removed and the DNA was precipitated with the addition of
1/10 vol. of 3'M NH4OAc and 2.5 vol. of 100% ethanol.
Samples were again centrifuged 10 min. at 1,800 g (25W3).
The DNA was then dissolved in 1.6 ml of 50 mM Tris-HCl pH
7.9, 0.5 M EDTA, 10 mM NaCl. The samples were digested with
9 units RNAase A (Sigma) for 30 min. at 37°C; brought to
0.5% SDS, and digested with 17 units of Pronase (Sigma) for
2 hours at 37°C. Samples were re—extracted with SS-phenol;
ethanol precipitated, washed in 70% ethanol, and solubilized
in 0.5 ml 10 mM Tris-HCl pH 8.0, 1 mM EDTA, yielding 200-600

ug of high molecular weight DNA.

3. Southern Blots.

Genomic DNA (3-5 ug) was digested with the appropriate
restriction endonucleases and separated by electrophoresis
through 0.8% or 2.0% agarose gels. The gels were denatured
and neutralized, then blotted by capillary transfer to
either Hybond—N nylon (Amersham) or Immobilon-N (Millipore)

membranes. DNA was bound to the membranes either by UV

61

irradiation (if membrane was nylon) or baking at 80°C for 1
hour under vaccuum (if membrane was Immobilon-N). Membranes
were prehybridized at 42°C > 4 hours in 50% formamide, 50 ug
sheared salmon sperm DNA/ml, 0.1 M PIPES pH 7.04, 0.1 M
NaCl, 0.1% Sarkosyl, 0.1% Ficoll, 0.1% PVP-40 (mol. wt.
360,000), 0.1% BSA. Blots were hybridized > 16 hours 42°C in
a solution identical to the prehybridization solution with
the exceptions that the formamide concentration was reduced
to 40%, the concentration of salmon sperm DNA was reduced by
half (25ug/ml), and 0.32P-dATP (10mCi/ml) labelled probe was
added to a concentration of 1 x 106 CPM/ml. Probes were
random primer labelled to a specific activity of ~ 1 x 109
according to the method of Feinberg and Vogelstein [21].
Blots were washed in 2x 88C, 0.05% N-laurylsarkosine, 0.02%
NaPPi at room temperature followed by two 1-hour washes with
0.1x 88C, 0.05% N-laurylsarkosine, 0.02% NaPPi at 50%:.
Blots were exposed to X—ray film for 1-5 days with

intensifying screens.

4. PCR Amplifications.

All PCR fragments were amplified from 100 ng of genomic
DNA in 50 ul reactions containing 0.25 uM of each primer,
200 uM dNTPs, 2 units Taq polymerase (Boehringer Mannheim),
the manufacturer's reaction buffer, and a 50 ul overlay of
mineral oil (USB). All reactions were performed in a PTC-lOO
thermocycler (MJ Research). Fragment identities, primer

pairs and conditions for amplification are presented in

62

Table A2. All samples were denatured at 950C for 5 min prior
to the addition of Taq polymerase, and the final extension
time was increased to 5 minutes. Mineral oil was removed
from the PCR product by separation on Parafilm M. In the
case of PCR amplification of (CA)n regions for genotyping
(CD3D, D11835, and PAX3 (CA)n), and amplification for
subsequent SSCP analysis, PCR parameters were modified as
follows: genomic DNA was amplified for 25 cycles in a 25 ul
volume with 100 uM dNTPs. This reaction mixture was then
supplemented to make a final volume of 50 ul; 200 uM dCTP,
dTTP, dGTP; 50 uM dATP; and 0.4 ul 353—dATp (lOmCi/ml). The

reaction continued for 10 more cycles.

5. Genotyping for Dinucleotide Repeat Markers.

Six ul of 355-dATp labelled PCR mixture was diluted
with 4 ul of sequencing loading buffer, heated at 95%: for >
5 min., then chilled on wet ice. 1-3 ul of this mixture was
then loaded on 6% polyacrylamide sequencing gels containing
8 M Urea and 12% formamide. A dideoxy labelled M13 sequence
was usually run on the same gel in order to estimate allele
sizes (assigned according to the darkest band in the set of
bands generated for each dinucleotide containing fragment).
Samples were electrophoresed for 3-6 hours at 50 Watts (50
C). Gels were fixed in 15% methanol, 10% acetic acid for 45
min., then dried on a slab dryer. Dried gels were exposed to

X-ray film for 1 to 10 days at room temperature.

A

63

6. SSCP Analysis.

35S-dATP labelled PCR fragments were diluted and
denatured as for genotyping as described above. Samples were
electrophoresed on 20 x 45 cm native 0.5x MDE Hydrolink gels
(AT Biochem) in 0.6x TBE buffer, at 8 Watts for 14 hours at
room temperature, or at 12 Watts for 8 hours at 49C. No
substantial difference was seen between the two methods of
electrophoresis. Gels were dried on a slab drier without

fixing and exposed to X-ray film for 1—10 days.

7. Sequencing.

PAX3 PCR fragments from affected individuals that
showed an SSCP, and all PAX3 PCR fragments from the probands
of UGMl and MSUS, were cloned directly into pCR 1000
(Invitrogen). The exon 2 fragments from UGM2-18 were first
enriched for the mutant fragment by digestion with RsaI and
electrophoresis on 4% NuSeive 3:1 agarose gels (FMC).
Following restriction with RsaI, the wild type fragments
migrated at 416 bp and 119 bp, while fragments from the
mutant allele migrated at 535 bp.

All potential mutant fragments were excised from the
gel and eluted using the Geneclean system (Bio 101), and
then cloned into pCR 1000 (Invitrogen). Plasmid DNA from
clones was isolated using Magic Miniprep (Promega), and then
alkali denatured prior to sequencing according to the method
of Wang et al.[59]. For each sequencing reaction, 2 ug double

stranded plasmid DNA and 6 ng of primer were used in the

Sequenase system v.2 (USB). In all sequencing experiments 3
independent clones were sequenced with primers for both
strands. Sequencing primers were the same as those for PCR,
with the addition of JA4 and JAS, which prime between the
TF30 and TF32 sites in exon 2.

For all detected sequence changes representing
potential mutations, a restriction site change was created.
That the mutation was not a PCR artifact was then confirmed
by digestion of PCR fragments with the appropriate
restriction endonuclease. In the case of the 14 bp deletion
segregating in UGM2, the existence of the mutation was
confirmed by the loss of an Rsa I site in fragments
generated from affected individuals. In the case of the 1 bp
insertion segregating in MSU3, the existence of the mutation
was confirmed by the gain of an Ava I restriction site in
fragments generated from affected individuals. A 1 bp
substitution that would have created a Tha I site in an
allele from MSU5-21 could not be confirmed with Tha I
digestion. Subsequent attempts to isolate another clone
containing this “missense mutation“ were not successful, and
it was concluded that the 1 bp substitution was a PCR
artifact. The SSCP detected in the unaffected individuals
MSU2-6 and M8U5-18 was confirmed to be due to a Hae III
polymorphism (described by Baldwin et al.[fl) in the intron

0
5' to exon 2 by Bee III digestion.

65

8. Linkage Analysis.

Linkage analyses were performed using the Linkage (v.
5.10) program package kindly supplied by Dr. Jurg Ott,
Columbia University Medical School. Full data files for all
families are presented below. When affection status was
assigned the following parameters were used: a) for W81 the
penetrance for WSl/+ and W81/W81 genotypes was 95%; the
penetrance for the +/+ genotype (i.e. probability of a
phenocopy) was 0.001; gene frequency was 1/42,000. b) for
W82 the penetrance for W82/+ was 57%, W82/W82 was 95% and
the penetrance for +/+ was 0.001. The gene frequency was
1/2100. The latter figure is often cited in the literature
and attributed to Arias et al.Efl. In fact, no data has ever
been presented to support this figure. Arias extrapolated
the gene frequency for W82 from a consideration of the
frequency of hereditary white forelock in the population.
The overall effect of overestimating the gene frequency in
the linkage analysis is to make obligate recombinations
involving individuals classified as affected more difficult

to detect, and to weaken the overall efficiency.

66

Table A2. PCR and Sequencing Primers

 

 

 

 

 

 

 

 

 

 

 

 

LOCOS ID Sequence (5'-3') Cycling parameters‘

CD3D TF41 TAGCTGCTGCATAAGCTCAC 94°C (L45
TF42 G'I'I‘AGTGGAAGAGCAGAGC 55° C 1:00
72° c 2:00
011835 TF39 ACAATTGGATTACTACTAGC 95°C (L45
TF40 TGTATTTGTATCGATTAACC 52°C (L45
72° C 0:45
PAX3 (CA)n TF26 CAGGGAGATGGCAGTT 94°C 0:45
TF38 CAGAGGCACAGAAAGA 50° c 1:00
72° C 1:00
PAX3 Exonl TF58 CCG'I'I‘TCGCC'I'I‘CACC'I'GGA 94° C 0:45
TF59 GCGCTGAGGCCCTCCCTTAC 60°C (L45
72° C 1:00
PAX3 Exon2 TF30 ATTTTGCCCCATTTGCTGTC 9T’C 0:45
TF32 CCGGTCTTCCCCAACACAGG 61°C (L45
72° C 1:00
PAX3 Exon3 TF33 CCTGCCCGCCTGTTCTCT 95°C 0:45
TF34 CGACTGACTGTCGCGCCT 60°C (L45
72° C 1:00
PAX3 Exon4 TF35 AGCCCTGCTTGTCTCAACCATGTG 94°C (L45
TF36 TGCCCTCCAAGTCACCCAGCAAGT 66°C (L45
72° c 0:45
RHO Exonl TF60 AGCTCAGGCC'I'I‘CGCAGCAT 94° C 0:45
TF61 GAGGGC'I'I'IGGATAACA'I'TG 60° G 1:00
72° C 0:45

PAX3 Exon2 JA4 CCAACCACA'I‘CCGCCACAAG Sequencing Primer

PAX3 Exonz JAS ACTGCGAAA‘I'I‘ACGTGCTGC Sequencing primer

 

 

 

 

* = All PCR reactions for 30 cycles plus a final 5 min. extension.

 

 

 

IV. Appendix B

67

Additional Figures and Tables

Table Bl. MSU4 Linkage Data File

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ID# Father Mother Aff. Sex PAX3 D1 1835 CD30
1 1 0 0 2 2 64 44 25
1 2 O 0 1 1 00 00 00
1 3 2 1 2 2 64 34 23
1 4 0 0 1 1 78 22 26
1 5 2 1 1 2 74 24 22
1 6 0 0 2 1 83 23 56
1 7 2 1 1 0 64 24 25
1 8 0 0 2 1 49 25 56
1 9 2 1 2 2 44 34 23
110 0 0 l 1 27 12 23
111 2 1 l l 64 34 23
112 0 0 2 1 74 24 36
113 4 3 2 1 47 23 36
114 4 3 1 1 76 00 36
115 5 6 2 1 78 24 26
116 5 6 2 1 34 34 25
117 7 8 2 0 69 22 26
118 7 8 1 0 00 25 25
119 10 9 l 1 74 23 23
120 10 9 1 l 74 24 23
1 21 11 12 1 1 7 4 4 4 3 6
1 22 11 12 1 l 7 4 4 4 2 6
1 23- 11 12 2 1 00 00 00
Aff.: 1 = normal; 2 = affected.
Sex: 1 = male; 2 = female.

PAX3: Dinucleotide marker for locus.
Dlls35: Dinucleotide marker for locus.
CD3D: Dinucleotide marker for locus.

 

Table 32. 14805 Linkage Data File

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

S A
E F a/b
ID F M X F ALPP D253 (x10) FNl D2855 TNPl PAX3 028211
1 0 0 1 1 00 0000 0.00 00 00 00 00 00
2 0 0 2 2 11 0011 7.26 01 11 01 89 22
3 1 2 2 2 11 0010 7.00 01 11 01 59 26
4 1 2 1 2 11 0010 6.94 01 11 01 59 26
5 0 0 1 1 00 0000 0.00 00 00 00 00 00
6 5 3 1 2 10 1010 6.67 01 11 01 49 56
7 5 3 2 2 10 1010 7.26 01 01 01 49 56
8 9 3 1 1 01 1010 5.36 11 11 00 00 66
9 0 0 1 1 01 1100 4.76 11 11 01 44 36
10 0 0 2 1 11 1010 5.45 10 10 00 79 46
11 4 10 1 2 11 1010 7.27 11 11 11 79 26
12 4 10 1 1 01 1010 4.82 11 10 00 57 66
13 4 10 2 1 11 0010 5.00 11 11 11 57 66
14 4 10 1 2 11 1010 7.17 11 10 11 99 24
15 4 10 2 1 01 1010 5.49 11 11 00 S7 66
16 0 0 1 1 10 0000 5.38 01 00 01 99 45
17 0 0 2 1 11 0000 5.00 10 00 01 16 47
18 16 17 2 1 11 1010 5.82 11 11 01 69 45
19 0 0 1 1 01 0010 5.22 10 01 11 37 48
20 0 0 2 1 01 0000 4.84 00 10 00 00 56
21 0 0 2 1 11 0101 5.32 10 11 11 45 36
22 0 0 2 1 01 0011 5.00 10 11 01 46 18
23 0 0 1 1 11 0101 4.33 11 10 01 34 36
24 0 0 1 1 11 0110 5.56 10 01 11 99 48
25 6 18 1 2 11 0010 6.78 11 11 01 69 45
26 6 18 1 2 10 1010 6.73 01 10 01 99 56
27 6 18 2 1 10 0010 5.66 01 11 01 00 56
28 19 7 2 1 11 1010 5.00 11 01 11 47 68
29 8 20 1 1 01 0000 5.56 00 00 00 00 66
30 11 21 1 1 01 0000 5.38 00 00 00 00 00
31 11 21 2 2 10 0110 6.85 11 11 01 59 26
32 11 21 2 2 11 0011 7.00 11 11 11 49 23
33 12 22 2 1 01 1001 5.09 11 10 01 47 16
34 12 22 2 1 01 0010 5.56 11 11 01 56 68
35 23 13 1 1 01 0011 5.36 01 11 01 35 66
36 23 13 1 1 01 0011 5.56 01 10 01 00 66
37 23 13 1 2 00 0000 6.00 00 00 00 00 00
38 24 15 2 1 00 0110 5.36 11 01 00 79 68
39 24 15 2 2 11 1010 6.22 10 11 10 79 46

 

 

69

 

Table B3. MSU6 Linkage Data File

 

(aff.)

 

EU¥K3

1

l
l

1

1
l
5

3

 

7'

.75

.31

.63

.63

1.95
1.57
.87

.59

.00

.05

.46

.69

.66

.69

1.58
1.80

1
l

.66

.67

.66
.22

.97

1

 

a/b

(:10)

.38
.46

.00

.00
.64
.91

.61

.84
.32

.83

.59
.18
.08
.35
.00

.45

.09

.09
.21

5

.53

.83

 

SAG

 

PHI

1

 

D2s55

l

l
l

 

l

l

 

 

Sex Aft ALPP

 

10
12

 

13

 

 

ID

 

 

 

 

 

 

 

 

 

10
11
12
13
14
15
16
17
18
19
20

 

 

 

 

 

 

 

 

 

 

 

21

 

 

 

70

 

Table 34. UGKl Linkage Data File

 

PAXB

9
9
9
9

5

9

 

TNPI

0 0

 

ALPP

1

 

D2355

1

 

a/b

.00
.08

.52

.50

.00

5.23

.65
.57

.52
.10

.14
.51

.67

.67

15
.90

.93

.00

 

Aft.

 

8611

 

29

29

29

30
30
3O
32
32
32
35
35
35
35
35

 

28

28

28
31

31

31
33
33

33
34
34
34
34
34

 

 

ID

 

 

28

 

 

29

30
31

 

 

32

 

33

 

34
35
63

 

 

 

64

 

65
66

 

 

67

 

68
69

 

 

70

 

71

 

 

72
73

 

74

 

75

 

76

 

 

 

71

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

r fi.',,m_ ..

 

Figure Bl. M804 PAX3 Gel.

Dinucleotide repeat of PAX3 PCR amplified from members of
family MSU4. Lanes 1—21 correspond to individuals MSU4—1
through MSU4-21. Assigned genotypes are given in Table B1.
PCR amplification parameters are given in Table A2.

72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

 

Figure B2. MSU4 CD3D Gel.

Dinucleotide repeat of CD3D PCR amplified from members of
family MSU4. Lanes 1—21 correspond to individuals MSU4-1
through MSU4-21. Assigned genotypes are given in Table B1.
PCR amplification parameters are given in Table A2.

73

1 2 3 4 5'6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

 

Figure B3. MSU4 D11835 Gel.

Dinucleotide repeat of D11S35 PCR amplified from members of
family MSU4. Lanes 1-21 correspond to individuals MSU4-1
through MSU4-21. Assigned genotypes are given in Table B1.
PCR amplification parameters are given in Table A2.

74

V. Appendix C
Detection of Changes in Copy number at Regions of Candidate

Loci By Dot Blot Analysis

l.Introduction

The methods relied on in this thesis for the detection
of mutations in PAX3 have a serious weakness. Because
fragments of alleles must first be PCR amplified before they
can be assayed for mutations or sequenced, the possibility
exists that mutations that would result in the inability to
amplify that particular fragment would be missed. A mutation
in the primer target sequence, or various size deletions,
could result in inability to amplify the mutant allele, with
.the consequence of false identification of a +/(del) set of
fragments as being +/+. This type of error has been
demonstrated before by Fuj imura et al.[24] and more recently
by Beutler et al.[LH. While any PCR—based method of
detection of alleles has this limitation, it is particularly
important to recognize this potential source of error in
detecting WS mutations because deletions and null alleles
have been demonstrated to give rise to the WS and Sp
phenotypes.

Accordingly, all of the probands of the WS families in
this study were assayed for changes in copy number at PAX3
and for markers from the 3p and 3q candidate regions PH. Two
methods were chosen. First, Southern blots of EcoRI digested

genomic DNA were probed with PAX3 (exon 2) probes that had

75

been ae32P-dATP random primer labelled. No fragments of
unusual length were detected with this method. Since this
method would only detect deletions that removed part, but
not all, of sequences within a restriction fragment or
changes at the restriction site itself, it was decided to be
too inefficient to continue. Changes in copy number were
then assayed for by hybridization of probes to genomic dot
blots. The results of these experiments are summarized

below.

2. Results
Initially, dot blots were made using nitrocellulose
filters (Schleicher and Schuell). The methods were as

_follows:

a) Genomic DNA from each proband of the WS families was
used for analysis, with the exceptions of MSU6 in
which case DNA of both of the parents were used and
MSU4 where DNA from the severely affected individual
MSU4-25 was included. DNA from a randomly selected
individual (HK27) was included as a normal control.
The characteristics, race, and sex of this
individual are not known.

b) 2 ug of DNA was diluted with TLE to make a final

volume of 10 ul. The sample was boiled for 5 min.,
then chilled on wet ice.

C) 40 ul of chilled 20x SSC was added to each sample.

d) A nitrocellulose filter was pre—wetted with 10x SSC,
and placed in a dot blot apparatus (Schleicher and
Schuell) over 2 Whatman filter papers which were
soaked in 10x SSC.

8) Wells were rinsed with 200 ul of chilled 20x ssc.
f) Samples were added under vacuum.
g) Wells were rinsed again with 200 ul of chilled 20x

SSC.

i)

j)

k)

76

The filter was UV crosslinked in a Stratalinker
(Stratagene) while damp and allowed to air dry for
at least 3 hours.

Filters were probed and washed exactly as for
Southern blots. Probes were made for PAX3 using PCR
fragments including exon 2 (primers TF30-TF32) exon
3 (TF33-TF34) and exon 4 (TF35-36); a murine genomic
clone Erba-Z which should hybridize to the
homologous sequence THRB on human 3p21; a murine
cDNA clone of Rho (MOPS-M2) which should hybridize
to the RHO locus on human 3q21—24; a human cDNA
clone for carbonic anhydrase II gene located on 8q.
Filters were exposed to a phosphor plate for 12 to
18 hours, and the plate was then scanned on a
PhosphorImager (Molecular Dynamics). The signals
were quantified and comparisons made using the
ImageQuant software system (Molecular Dynamics).
Filters were stripped with three 15 min. washes of
boiling 0.1x SSC/0.1% SDS. Filters were then exposed
to autorad film for > 5 hours to confirm they had
been adequately stripped of probe before being re-
probed.

Three dot blots totalling 5 repetitions of each sample were

analyzed. The data from the first blot was discarded due to

the ommission of the normal control (HK27) from that blot.

All pair-wise comparisons between total counts for probes

were first normalized to the HK27 ratio in each series, and

then averaged over the four repetitions for each sample.

The results of comparisons from the 4 series are summarized

in the following tables.

77

Table C1. Ratios of PAX3 exon Copy Numbers to Other Loci

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total ratios to Erba2 Total ratios to Rho Total ratios to CarblI

id # Exonz Exon3 Exon4 Exonz Exon3 Exon4 Exonz Exon3 Exon4

msu1-3 0.806 1.964 1.669 1.383 2.2 2.384 0.993 1.934
msu2-7 0.929 1.804 2.01 0.903 1.687 2.441 0.906 1.938
msu3-9 1.01 0.921 1.127 1.167 0.961 1.37 1.153 1.289
msu4-4 0.844 0.942 0.914 1.253 1.412 1.363 0.838 0.855
msu4-25 0.861 0.885 0.808 0.88 0.939 0.86 0.857 0.82
msu5-21 0.94 1.684 0.85 1.032 1.339 1.048 0.953 0.92
msu6-54 0.879 0.665 0.671 0.83 0.829 0.634 0.822 0.608
msu6—55 0.889 1.05 0.802 0.898 0.923 0.815 0.878 0.782
msu7-6 0.975 1.143 1.135 1.316 1.622 1.626 1.443 1.69
msu9-c 0.96 2.371 1.343 1.328 2.264 1.821 1.057 1.35
ugm1-66 0.622 0.894 0.659 1.12 1.564 1.232 0.593 0.628
ugm2-24 0.947 1.136 0.895 0.97 1.289 0.928 0.966 0.874
ugm3-11 1.076 1.541 1.02 1.123 1.514 1.156 1.131 1.121
ugm4-10 0.879 0.983 0.794 1.044 1205 0.955 0.916 0.865
ung-l 0.986 0.874 0.909 1.204 1.325 1.083 0.966 0.903
HK27 1.002 1 1 1 1 1 0.999 1

 

 

 

 

 

 

 

 

There are significant deviations from 1:1 ratios represented
in the table. The most prominent are highlighted. The
greatest deviations are invariably seen with the exon 3
probe. However, the standard deviations within series are
also highest with this probe -- typically ~O.5 or higher
while standard deviations within series with the exon 2
probe are ~0.2. This is probably due to the high proportion
of paired domain sequence in the exon 3 probe while the
exon2 probe has nearly 50% intronic sequence. Thus, the exon
3 probe may be hybridizing to several PAX genes. For this
reason, the exon 3:CA2 comparisons were not even made.

If one discounts the data for the exon 3 probe then it

(appears that individuals MSU1-3 and MSU2-7 have duplications

78

for exon 4 (the exon 4 probe also has a high proportion of
unique sequence and standard deviations in the series
comparable to the exon 2 probe); individual MSU6-54 appears
to have a deletion of PAX3; and individual UGM1-66 appears
to have a deletion of PAX3 when compared to the Erba-2 and
CA2 probes, but not when compared to the Rho probe. A
resolution to the latter contradiction is suggested by

pairwise comparisons of the non—PAX3 probes.

Table C2. Ratios of Non-PAX3 Loci Comparisons

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CAflEflmZ CAflRMl Rdeflmz
id # avg stdev avg stdev avg stdev
msul-3 0.9 0.255 1.42 0.35 0.67 0.24
msu2-7 1.08 0.187 1.16 0.26 0.95 0.23
1118139 0.89 0.128 1.06 0.3 0.89 0.25
nme4 105 OJ&1 162 024 (ME 008
msu4-25 0.98 0.181 1.02 0.13 0.98 0.26
msu5-2l 0.99 0.185 1.09 0.24 0.97 0.34
msu6-54 1.08 0.204 1.04 0.26 1.13 0.48
msu6-55 1.02 0.079 1.06 0.12 0.98 0.15
msu7-6 0.75 0.266 0.95 0.08 0.78 0.24
msu9-c 0.95 0.22 1.29 0.23 0.77 0.24
ugml-66 1.03 0.069 1.97 0.33 0.54 0.09
ugm2-24 1.02 0.078 1.1 0.18 0.95 0.15
ugm3-1 1 0.96 0.162 1.03 0.15 0.96 0.22
ugm4-10 0.94 0.119 1.16 0.12 0.82 0.1
ung-l 1 0.127 1.27 0.23 0.83 0.27
hk27 1 7E-05 1.01 0.01 1 0

 

 

 

 

 

 

 

 

All of the ratios are consistently near 1:1 with the
notable exceptions of those that are highlighted. In
particular, individuals MSU1-3, MSU4-4 and UGMl-66 appear to

have reduced copies of RHO, which may be indicative of a

79

deletion on 3q. Pairwise T-test comparisons of the measures
of these three samples with the other samples gave evidence
of significant differences (p = 0.0016). A deletion
encompassing RHO would explain why the PAX3:RHO ratios were
1:1 for UGM1—66 (assuming that UGMl-66 has a deletion of
PAX3 as well) and why the PAX3:RHO ratio was exceptionally
high in MSUl-3 (assuming that MSUl-3 has a duplication of
PAX3). Potentially the most interesting observation is that
MSU4-4 may have a deletion of RHO and no obvious change in
copy number at PAX3. MSU4 is segregating for W82, and
linkage analysis detected an obligate recombination between
WS2 and the PAX3 dinucleotide marker (see Chapter 1). These
observations suggested that there may be another locus for
the WS phenotype on 3q which serves to modify the WSl
phenotype and may be sufficient to cause the WS2 phenotype.
At first, the coincidence of perhaps three individuals
(MSU1-3, MSU2-7, UGM1-66) in a sample of 15 having changes
of copy number at both the 2q and 3q regions may seem too
unlikely. However, it should be remembered that the families
analyzed were chosen precisely because they presented
obvious phenotypes. If a major modifying locus of WS1 is
located on 3q, it is in the most obviously affected
individuals that mutations would be expected at both loci.
In other words, the sample set may be biased towards finding
coincident mutations. Also, it is possible that the

cieletions and duplications of 2q and 3q are causally

80

related, perhaps by non-homologous recombination during
meiosis.

In order to confirm these conclusions, the experiments
were repeated on a dot blot containing DNA from all of the
members of UGMl to test for Mendelian inheritance of the
deletion, and correspondance to phenotype. Ideally, this
analysis would have been repeated for families MSU4 and
MSU6, but the uncertainties regarding the phenotypes in
these families (see Chapter 1) render dubious any
conclusions from such analyses.

When ratios of PAX3 to Erba-2 were compared in UGMl, it
was observed that not only did the apparent deletions not
correspond 1:1 with the WSl phenotype, but the deletions did
not segregate in Mendelian fashion. Because of this
observation, and the high standard deviations in some of the
other samples on the original dot blots, it was decided that
all experiments should be repeated with modifications
designed to make each experiment more reproducible.

Numerous changes in the procedure were tried. Instead
of presenting the results with each modification, I will
summarize all of the modifications that I believe were

useful.

an Dot blots were made using Hybond-N nylon
membranes (Amersham). This allowed probes to be
removed using a 30 min. wash with 0.4 N NaOH.
Probe removal was uniform and complete at each
step, which was not true of the boiling SDS
procedure used with the nitrocellulose filters.

b)

d)

e)

81

A PCR fragment of exon 1 of the human RHO gene
was substituted for the murine Rho probe. This
allowed for more comparable hybridization
kinetics to the PAX3 probe both because it
contains human sequence and because the probe
lengths are similar (535 bp vs. 585 bp).

A different hybridization solution was used which
contains no salmon sperm DNA. Hybridization
solution is 37% formamide; 1% SDS; 10% dextran
sulfate; and 5x SSPE. 15 mls. of this was used as
a pre-hyb. The pre-hyb was then substituted with
a 15 ml. solution containing probe. The absence
of salmon sperm DNA allows for careful control of
the ratio of probe DNA to target DNA, which will
affect the kinetics of hybridization. Unless
linear kinetics are achieved, it will not be
possible to reliably detect changes in copy
number.

Six normal control samples were substituted for
the one HK27 control sample. They were MSU5-19,
MSU5-18, MSUS-Zl, MSU5-22, UGM2-22 and UGM2-41.
These samples have the advantage that the
individuals have all been inspected and
determined to be clinically normal. They all have
married into the respective families and have no
family history of WS themselves. By using the
average measurements for these six as the
normalizing factor an error of compounding
variation is removed from the analysis. There may
be variation between repetitions of samples that
are due purely to the limitations of the
technique. By normalizing a test sample to a
control sample that itself is subject to those
inherent variations the overall variation is
compounded. By normalizing to an average of
control samples, the only source of variation
should be in the test samples. In fact, the
standard deviations within series did improve
with this technique.

A Southern blot should be included in the
hybridization chamber with each dot blot to
confirm that any signal detected is due to
specific hybridization to a discrete sequence,
and as a rough estimation of the number of
sequences in the genome that the probe is
hybridizing to.

82

f) The probes for CA2, Erba-2, and exon 3 of PAX3
should be discontinued to minimize the number of
hybridization and stripping steps for each blot.
Matched sets of blots should be probed with RHO,
and PAX3 exon 2, and an X—linked marker in
scrambled order.

gm A unique sequence X-linked marker should be used
to first determine whether males and females can
be distinguished with the technique. The DXsl4
marker was tried but is not suitable since it
recognizes many bands ( >15) on a Southern blot
and gives exceptionally high background signals
on dot blots.

1U The total amount of probe in the hybridization
solution should be calculated to give at least a
ten-fold excess of probe per ml to target
sequence. If you assume that the target sequence
is approximately 105bp in length, then it
represents 105/109 of the total sequence. Thus,
the total amount of genomic DNA on a blot should
be divided by 109, and this figure multiplied by
10 to give the total amount of probe in solution.

For example: if a blot contains 160 ug of
genomic DNA, then the probe should contain ten
times 0.16 ng of DNA per ml, or 16 ng in a 10 ml
hybridization solution. For several experiments I
added labelled probe up to 2 x 109 CPM/ml and
then supplemented the total with unlabelled probe
to achieve 20 ng/ml. I now regard this as a
mistake since this meant having 4 to 5 times as
much unlabelled probe as labelled probe in
solution. Since kinetics of solution
hybridization are favored over hybridization to a
2 dimensional matrix, linear kinetics with regard
to the target sequences on the membrane was
probably not achieved. The hybridization solution
should contain only labelled probe fragments,
with no other DNA in solution.

Analysis with the above modifications had to be
terminated due to time constraints and the unexpected
characteristics of the DXs14 probe. Once a suitable X—linked
(probe is found, I feel confident that male to female

differences can be detected with the dot blot method. The

initial observations can then be reliably tested.

83

If the original changes in copy number are confirmed,
but the deletion of PAX3 shows non-Mendelian segregation in
UGMl, I would strongly urge that the entire UGMl sample set
be typed for a group of highly polymorphic markers not
linked to 2q. A possible resolution of the non-Mendelian
inheritance pattern originally seen would be if several DNA
samples were mislabelled. It should be pointed out that the
Mendelian pattern of inheritance observed with the PAX3
marker may be coincidental, or may be a reflection of the
tendency for this observer to assume that the pedigree
structure and sample labels are all correct and to resolve
ambiguous genotype assignments in favor of Mendelian
inheritance.

If the original observations cannot be repeated with
the modification of the dot blot method described here, my
suggestion would be to attempt to show changes in copy
number with in situ hybridization techniques. It is
possible, however, that the original observations were due
to inconsistencies inherent in the methodology. Given the
reproducibility of the UGMl-66 data, there may be a
component of the DNA sample that renders it prone to removal

from a nitrocellulose filter by the boiling SDS method.

VI. Appendix D
Conventions of Nomenclature for Human Alleles Used in this

Thesis

The allele designations used in this thesis are a
modification of McKusick H3. The goal is to develop a
short-hand method for referring to particular mutations
while still conveying as much information as possible about
the mutation.

Allele designations follow the locus name in brackets
{}. The field inside the bracket should be thought of as
having a center, where the codon involved in the mutation is
located, and fields to the left and right. The field to the
left is reserved for either a description of the nucleotide
change or the identity of the codon in the normal allele.
The field to the right is reserved for either a description
of the new translation codon or the identity of the codon
after the mutation. Certain symbols for deletions,
insertions, etc. should be agreed upon for the sake of

brevity. I propose:

A = deletion
V = insertion
FS = frame shift

TER = new translation termination codon

85

The alleles are named according to the type of mutation
and the functional consequence on the predicted translation
product. All mutations are numbered in reference to the
codon number in the predicted reading frame. Thus, a
missense mutation in PAX3 that results in a Leucine for
Arginine substitution at position 50 is designated:
PAX3{Arg50Leu}. A 1 base pair insertion that causes a shift
in the reading frame, and early translation termination
resulting in a peptide 113 amino acids in length would be
designated: PAX3{1pr34;F8113TER}. For ease of reference,

I suggest that the translation termination codon be numbered
in reference to the position in the mutant allele. The new
translation termination codon in the 14 bp deletion mutant
allele straddles 112 and 113 codons in the wild-type allele,
but is unambiguously defined as the 108 codon in the mutant
allele (PAX3{14pr89-93;FSIO8TER}).

Modifications of the system are needed to account for
more complex mutations such as tri—nucleotide repeat
expansions or splice site mutations, especially when such
mutations involve nucleotides in non-coding regions. When
deciding on an allele designation it should be remembered
that the goal is to achieve a unique, short, and readily
idetifiable name and not a full description of the molecular
pathology. The intronic mutation in Fax—3 that results in at
least four aberrant splicing products (Sp) is a case in
point UQL Instead of attempting to describe each translated

product it may be sufficient to simply indicate that a

86

nucleotide change in intron 3 results in aberrant splicing.
Perhaps PAX3{IBsplice} would suffice (if this were a human

mutation)?

87

VII. Appendix 8

Results of Linkage Analysis for M103

1. Introduction

Asher and Friedman UH identified several mouse
mutations that could be the murine homologs of Waardenburg
syndrome. As a positional cloning strategy, they argued that
the identification and cloning of the murine gene first
would greatly facilitate the cloning of the human gene given
the advantages of mapping and cloning a gene from an inbred
experimental animal. The murine geneM’iOR was the strongest
of the candidate genes arguing strictly from phenotypic
characters since hearing impairment had been demonstrated in
.MiOR/+ animals (J. Asher, unpublished observations), whereas
no such impairment had been demonstrated for the other mouse
models, including Sp[53]. Unfortunately, the MiOR model
predicted two different human locations based on conserved
syntenies between the murine and human genetic maps,
depending on whetherMiOR mapped closer to the genes Raf-1
or Rho. Given the relative paucity of informative genetic
markers available for the human genome at the time, an
efficient strategy was to better resolve the murine map and
then concentrate on conducting a multi—locus analysisflflfl in
the human candidate region which appeared more likely based
on the higher resolution murine map. The time and effort
expenditure in conducting a multi-locus analysis was

considerable, and necessary given the poor informativeness

88

of the human map, which was composed almost entirely of RFLP
and VNTR markers in 1990. A multilocus strategy was
necessary in initially demonstrating linkage of WSl to 2q37
[3].

An interspecific cross was initiated where the MiOR
gene was crossed on to a Robertsonian (4;6) chromosome
originating from MUS poschianus. (The original purpose of
locating the MiOR gene on a Robertsonian chromosome was to
aid in flow sorting the chromosome should that prove
necessary.) This chromosome was then crossed into a MUS
musculus line..MIOR/+ x +/+ crosses were then made, and the
progeny typed for white belly spots as a marker for.MiOR,
and genotyped for a variety of RFLP markers including Raf—1,

Rho, and Try-1.

2. Results and Discussion

Genotyping data from Southern blot analysis was coded
in a format for use by the Gene-Link program, which was the
generous gift of Dr. Xavier Montagutelli, Institut Pasteur,
Paris, France. Gene—link calculates liklihood scores,
identifies obligate recombinant animals, and double
recombinations, and determines confidence levels for the

most likely linkage orders.

Table 81. Linkage results for.MiaR/+ x +/+
son Recombinanta/N ‘ Distance (cu)

.Mi - Raf-1 18/94 19.15
.Mi - Rho 16/81 19.75

 

Raf-l - Rho 1/81 1.23

  

89

Of the 18 recombinants with respect to.MiOR - Raf—1, 12
were classified as +/+ and 6 as MiOR/+. Moreover, when
considering all markers typed (Raf-1, MiOR, Rho, Try—1,
.Met), there are 7 animals in which the only recombination is
at.Mi; 6 are classified as +/+ and l as.MiOR/+. Since non-
penetrance for the MiOR phenotype must be a consideration
the analysis was performed considering only those animals

with a positive assignment as MiOR/+. The resulting map is:

uuom ------- 13.33CM ------------ Raf-1 -—-1.23cM ------ Rho

l --------------------- 15.79cM ---------------------- I

This linkage order is 100x more likely than the next most
likely linkage relationship. Other markers cannot be
unambiguously added to the map due to distortion in the map
nearer to the centromere, presumably due to the nature of

the Robertsonian chromosome.

90

LIST OF REFERENCES

1. Arias, S. “Genetic heterogeneity in the Waardenburg
syndrome." Birth Defects Original Article Series (VII): 87-
101, 1971.

2. Arias, S. and M. Mota. "Apparent non-penetrance for
dystopia in Waardenburg syndrome type I, with some hints on
the diagnosis of dystopia canthorum.” J. Genet. Hum. 26:
103—131, 1978.

3. Asher, J. H., Jr., R. Morell and T. B. Friedman.
"Waardenburg syndrome (WS): The analysis of a single family
with a WSl mutation showing linkage to RFLP markers on human
chromosome 2q." Am. J. Hum. Genet. 48: 43-52, 1991.

4. Asher, J. H., Jr., and T. B. Friedman. "Mouse and
hamster mutants as models for Waardenburg syndromes in
humans." J. Med. Genet. 27: 618-626, 1990.

S. Auerbach, R. “Analysis of the developmental effects of
a lethal mutation in the house mouse." J. Exp. 2001. 127:
305-329, 1954.

6. Badner, J. A. and A. Chakravarti. “Waardenburg syndrome
and Hirschsprung disease: Evidence for pleiotropic effects

of a single dominant gene.“ Am. J. Med. Genet. 35: 100—104,
1990.

7. Baldwin, C. T., C. F. Hoth, J. A. Amos, E. 0. da Silva
and A. Milunsky. ”An exonic mutation in the HUPZ paired

domain gene causes Waardenburg's syndrome.“ Nature. 355:
637—638, 1992.

8. Balling, R., U. Deutsch and P. Gruss. “undulated, a
mutation affecting the development of the mouse skeleton,
has a point mutation in the paired box of Pax-l.” Cell. 55:
531-535, 1988.

9. Bard, L. A. “Heterogeneity in Waardenburg's syndrome.“
Arch. Ophthal. 96(2): 1193—1198, 1978.

10. Beechey, C. V. and A. G. Searle. "Mutations at the Sp
locus." Mouse News Lett. 75: 28, 1986.

11. Beutler, E., T. Gelbart and C. West. "Identification of
six new Gaucher disease mutations.“ Genomics. 15: 203—205,
1993.

91

12. Bopp, D., E. Jamet, S. Baumgartner, M. Burri and M.
Noll. “Isolation of two tissue-specific Drosophila paired
box genes Pox meso and Pox neuro." EMBO J. 8: 3445-3457,
1986.

13. Brown, K. S., D. R. Bergsma and M. V. Barrow. "Animal
models of Pigment and hearing abnormalities in man." Birth
Defects: Original Article Series (VII): 102—109, 1971.

14. Burri, M., Y. Tromvoukis, D. Bopp, G. Frigerio and M.
Noll. "Conservation of the paired domain in metazoans and
its structure in three isolated human genes." EMBO J. 8(4):
1183-1190, 1989.

15. Chalepakis, G., R. Fritsch, H. Fickenscher, U. Deutsch,
M. Goulding and P. Gruss. “The molecular basis of the
undulated/pax-l mutation.“ Cell. 66: 873-884, 1991.

16. Cotterman, C. W. "Some statistical problems posed by
Waardenburg's data on dystopia canthorum and associated
anomalies.“ Am. J. Hum. Genet. 3: 254-266, 1951.

17. Darwin, C. "The Origin of Species." 1859 London.

18. Epstein, D. J., M. Vekemans and P. Gros. 'Splotch
(SpZH), a mutation affecting development of the mouse neural
tube, shows a deletion within the paired homeodomain of Pax-
3." Cell. 67: 767-774, 1991.

19. Epstein, D. J., K. J. Vogan, D. G. Trasler and P. Gros.
“A mutation with intron 3 of the Pax-3 gene produces
aberrantly spliced mRNA transcripts in the splotch (Sp)
mouse mutant.“ Proc. Natl. Acad. Sci. USA. 90: 532-536,
1993.

20. Farrer, L. A., K. M. Grundfast, J. Amos, K. S. Arnos,
J. H. Asher, Jr., P. Beighton, S. R. Diehl, J. Fex, C. Foy,
T. B. Friedman, J. Greenberg, C. Hoth, M. Marazita, A.
Milunsky, R. Morell, W. Nance, V. Newton, R. Ramesar, T. San
Agustin, J. Skare, C. Steven, R. G. J. Wagner, E. R. Wilcox,
I. Winship and A. P. Read. “Waardenburg syndrome (WS) type 1
is caused by defects at multiple loci, one of which is near
ALPP on chromosome 2: First report of the WS consortium.”
Am. J. Hum. Genet: 50: 902-913, 1992.

21. Feinberg, A. P. and B. Vogelstein. "A technique for
radiolabeling DNA restriction endonuclease fragments to high
specific activity.“ Anal. Biochem. 132: 266-267, 1983.

92

22. Foy, C., V. Newton, D. Wellesley, R. Harris and A. P.
Read. “Assignment of the locus for Waardenburg syndrome type
I to human chromosome 2q37 and possible homology to the
splotch mouse." Am. J. Hum. Genet. 46: 1017-1023, 1990.

23. Fraser, G. R. “The Causes of Profound Deafness in
Childhood." 1976 Johns Hopkins University Press. Baltimore.

24. Fujimura, F. K., H. Northrup and A. L. Beaudet.
“Genotyping errors with the polymerase chain reaction." N.
Engl. J. Med. 322: 61, 1990.

25. Goldberg, M. F. "Waardenburg's syndrome with fundus and
other anomalies." Arch. Ophthal. 76: 797-810, 1966.

26. Goodman, R. M., G. Oelsner, M. Berkenstadt and D.
Admon. “Absence of vagina and right-sided adnexa uteri in
the Waardenburg syndrome: a possible clue to the
embryological defect.“ J. Med. Genet. 25: 355-357, 1988.

27. Goulding, M. D., G. Chalepakis, U. Deutsch, J. R.
Erselius and P. Gruss. “Pax-3, a novel murine DNA binding
protein expressed during early neurogenesis." EMBO J.
10(5): 1135-1147, 1991.

28. Gruss, P. and C. Walther. "Pax in development." Cell.
69: 719-722, 1992.

29. Hageman, M. J. and J. W. Delleman. "Heterogeneity in
Waardenburg syndrome." Am. J. Hum. Genet. 29: 468-485,
1977.

30. Hayashi, S. and M. P. Scott. ”What determines the
specificity of action of Drosophila homeodomain proteins?“
Cell. 63: 883-894, 1990.

31. Hill, R. E., J. Favor, B. L. M. Hogan, G. F. Saunders,
I. M. Hanson, J. Prosser, T. Jordan, N. D. Hastie and V. van
Heyningen. “Mouse Small eye results from mutations in a
paired-like homeobox-containing gene." Nature. 354: 522-
525, 1991.

32. Hoth, C. F., A. Milunsky, N. Lipsky, R. Sheffer, S. K.
Clarren and C. T. Baldwin. ”Mutations in the paired domain
of the human PAX3 gene cause Klein-Waardneburg syndrome
(WSIII) as well as Waardenburg syndrome type I (WSI).“ Am.
J. Hum. Genet. 52: 455-462, 1993.

33. Ishikiriyama, S., H. Tonoki, Y. Shibuya, S. Chin, N.
Harada, K. Abe and N. Niikawa. “Waardenburg syndrome type I

93

in a child with de novo inversion (2)(q35q37.3)." Am. J.
Med. Genet. 33: 505-507, 1989.

34. Jacobs-Cohen, R. J., R. F. Payette, M. D. Gershon and
T. P. Rothman. "Inability of neural crest cells to colonize
the presumptive aganglionic bowel of ls/ls mutant mice:
Requirements for a permissive microenvironment.“ J. Comp.
Neurol. 255: 425-438, 1987.

35. Kaplan, P. and J.-P. de Chaderevian. 'Piebaldism-
Waardenburg syndrome: Histopathologic evidence for a neural
crest syndrome.“ Am. J. Med. Genet. 31: 679-688, 1988.

36. Lane, P. W. and H. M. Liu. "Association of megacolon
with a new dominant spotting gene (Dom) in the mouse." J.
Hered. 75: 435-439, 1984.

37. Lange, K., M. A. Spence and M. B. Frank. “Application
of the lod method to the detection of linkage between a
quantitative trait and a qualitative marker: a simulation
experiment." Am. J. Hum. Genet. 28: 167-173, 1976.

38. Lathrop, G. M., J. M. Lalouel, C. Julier and J. Ott.
"Strategies for multilocus linkage analysis in humans.”
Proc. Natl. Acad. Sci. USA. 81: 3443-3446, 1984.

39. Litt, M., V. Sharma and J. A. Luty. "A highly
polymorphic (TG)n microsatellite at the D11335 locus.“
Cytogenet. Cell Genet. 51: 1034, 1990.

40. Lyon, M. F. and A. G. Searle. “Genetic Variants and
Strains of the Laboratory Mouse.“ 1989 Oxford University
Press. Oxford.

41. Mallory, S. B., E. Wiener and J. J. Nordlund.
“Waardenburg's syndrome with Hirschsprung's disease: A
neural crest defect.” Pediat. Dermat. 3(2): 119-124, 1986.

42. Martin, D., N. K. Spurr and J. Trowsdale. "RFLP of the
human placental alkaline phosphatase gene (PLAP).' Nucleic
Acids Res. 15: 9104, 1987.

43. Mayer, T. C. ”Site of gene action in steel mice:
analysis of the pigment defect by mesoderm -- ectoderm
recombinations.“ J. Exp. Zool. 184: 345-352, 1973.

44. McKusick, V. A. “Congenital deafness and Hirshsprung's
disease (letter).“ New Eng. J. Med. 288: 691, 1973.

94

45. McKusick, V. A. "Mendelian Inheritance in Man." 1988
Johns Hopkins University Press. Baltimore.

46. Moase, C. and D. Trasler. "N-CAM alterations in splotch
neural tube defect mouse embryos." Development. 113: 1049-
1058, 1991.

47. Morell, R., T. B. Friedman, S. Moeljopawiro, Hartono,
Soewito and J. H. Asher, Jr. ”A frameshift mutation in the
HuP2 paired domain of the probable human homolog of murine
Fax-3 is responsible for Waardenburg syndrome type 1 in an
Indonesian family.“ Human molec. Genet. 1(4): 243-247,
1992.

48. Nadeau, J. H. "Maps of linkage and synteny homologies
between mouse and man." Trends Genet. 5: 82-86, 1989.

49. Perris, R. and M. Bronner-Fraser. "Recent advances in
defining the role of the extracellular matrix in neural
crest development.“ Comm. Dev. Neurobiol. 1: 61-83, 1989.

50. Pierpoint, J. W. and R. P. Erickson. "Facts on PAX."
Am. J. Hum. Genet. 52: 451-454, 1993.

51. Shah, K. N., S. J. Dalal, M. P. Desai, P. N. Sheth, N.
C. Joshi and L. M. Ambani. “White Forelock, pigmentary
disorder of irides, and long segment Hirshsprung disease:
possible variant of Waardenburg syndrome.“ J. Pediat. 99:
432-435, 1981.

52. Sheffer, R. and J. Zlotogora. "Autosomal dominiant
inheritance of Klein-Waardenburg syndrome." Am. J. Med.
Genet. 42: 320-322, 1992.

53. Steel, K. and R. J. H. Smith. ”Normal hearing in
splotch (Sp/+), the mouse homolog of Waardenburg syndrome
type I." Nature Genet. 2: 75-79, 1992.

54. Tassabehji, M., A. P. Read, V. E. Newton, R. Harris, R.
Balling, P. Gruss and T. Strachan. "Waardenburg's syndrome
patients have mutations in the human homologue of the Fax-3
paired box gene.“ Nature. 355: 635-636, 1992.

55. Tassabehji, M., A. P. Read, V. E. Newton, M. Patton, P.
Gruss, R. Harris and T. Strachan. "Mutations in the PAX3
gene causing Waardenburg syndrome type 1 and type 2.“ Nature
Genetics. 3: 26-30, 1993.

56. Ton, C. C. T., H. Hirvonen, H. Miwa, M. M. Weil, P.
Monaghan, T. Jordan, V. van Heyningen, N. D. Hastie, H.

95

Meijers-Heijboer, M. Drechsler, B. Royer-Pokora, F. Collins,
A. Swaroop, L. C. Strong and G. F. Saunders. "Positional
cloning and characterization of a paired box- and homeobox-
containing gene from the aniridia region." Cell. 67: 1059-
1074, 1991.

57. Treisman, J., E. Harris and C. Desplan. “The paired box
encodes a second DNA-binding domain in the paired homeo
domain protein.“ Genes & Dev. 5: 594-604, 1991.

58. Waardenburg, P. J. “A new syndrome combining
developmental anomalies of the eyelids, eyebrows and nose
root with pigmentary defects of the iris and head hair with
congenital deafness.“ Am. J. Hum. Genet. 3: 195-253, 1951.

59. Wang, L.-M., D. K. Weber, T. Johnson and A. Y.
Sakaguchi. “Supercoil sequencing using unpurified templates
produced by rapid boiling." BioTechniques. 6(9): 839-843,
1988.

60. Weber, J. L., A. E. Kwitek and P. E. May. "Dinucleotide
repeat polymorphisms at the Dlls419 and CD3D loci.“ Nucleic
Acids Res. 18(13): 4036, 1990.

61. Webster, W. "Embryogenesis of the enteric ganglia in
normal mice and in mice that develop Congenital aganglionic
megacolon.“ J. Embryol. Exp. Morphol. 30: 573-585, 1973.

62. Wilcox, E. R., N. M. Rivolta, B. Ploplis, S. B. Potterf
and J. Fex. “The HuP2 gene is mapped to chromosome 2,
together with a highly informative CA dinucleotide repeat."
Hum. Molec. Genet. 1: 215, 1992.

IE5

”’11111111111111”