Glial Fibrillary Acidic Protein Mutations in Infantile, Juvenile, and Adult Forms of Alexander Disease

Rong Li, MD,1 Anne B. Johnson, MD,2 Gaja Salomons, PhD,3 James E. Goldman, MD, PHD,4 Sakkubai Naidu, MD,5 Roy Quinlan, PhD,6 Bruce Cree, MD, PhD, MCR,7 Stephanie Z. Ruyle, MD,8 Brenda Banwell, MD,9 Marc D’Hooghe, MD,10 Joseph R. Siebert, PhD,11,12 Cristin M. Rolf, MD,13,14 Helen Cox, MB, ChB,15 Alyssa Reddy, MD,16 Luis González Gutiérrez-Solana, MD,17 Amanda Collins, FRCP,18 Roy O. Weller, MD, PhD,19 Albee Messing, VMD, PhD,20 Marjo S. van der Knaap, MD,3 and Michael Brenner, PhD1

Alexander disease is a progressive, usually fatal neurological disorder defined by the widespread and abundant presence in astrocytes of protein aggregates called Rosenthal fibers. The disease most often occurs in infants younger than 2 years and has been labeled a leukodystrophy because of an accompanying severe myelin deficit in the frontal lobes. Later onset forms have also been recognized based on the presence of abundant Rosenthal fibers. In these cases, clinical signs and pathology can be quite different from the infantile form, raising the question whether they share the same underlying cause. Recently, we and others have found pathogenic, de novo missense mutations in the glial fibrillary acidic protein gene in most infantile patients examined and in a few later onset patients. To obtain further information about the role of glial fibrillary acidic protein mutations in Alexander disease, we analyzed 41 new patients and another 3 previously described clinically, including 18 later onset patients. Our results show that dominant missense glial fibrillary acidic protein mutations account for nearly all forms of this disorder. They also significantly expand the catalog of responsible mutations, verify the value of magnetic resonance imaging diagnosis, indicate an unexpected male predominance for the juvenile form, and provide insights into phenotype-genotype relations.

Ann Neurol 2005;57:310-326

Alexander disease is a progressive, usually fatal neuro- logical disorder that displays different clinical and pathological signs depending on the age at onset.1- The disease most frequently presents before 2 years of age with motor and mental retardation, bulbar dys- function, seizures, and megalencephaly, leading to death by 10 years of age. The pathology of this infan- tile form is characterized by marked absence of myelin in the frontal lobes. The juvenile form typically pre-

sents between 2 and 12 years of age and is character- ized by difficulties with coordination, speech, and swal- lowing. Both myelin and mental ability may be relatively intact. The course is slower than the infantile form, with occasional patients living into their 40s. The adult form, with onset from the teens to middle age, may be similar to the juvenile form, or can mimic multiple sclerosis or a brain tumor. Palatal myoclonus is often present. In both the juvenile and adult forms,

From the 1Department of Neurobiology and Civitan International Research Center, University of Alabama Birmingham, Birmingham, AL; 2Departments of Pathology and Neuroscience, Albert Einstein College of Medicine, Bronx, NY; 3Departments of Child Neurology and Clinical Chemistry, VU University Medical Center, Amster- dam, The Netherlands; 4Department of Pathology, Columbia Uni- versity, New York, NY; 5Neurogenetics Unit, Johns Hopkins Uni- versity School of Medicine, Kennedy Krieger Institute, Baltimore, MD; “Department of Biological Science, The University, Durham, United Kingdom; 7Multiple Sclerosis Center, University of Califor- nia at San Francisco, San Francisco, CA; 8Department of Pathology, Children’s Hospital, Denver, CO; 9Division of Neurology, The Hospital for Sick Children, Toronto, Ontario, Canada; 10Depart- ment of Neurology, Hospital Sint-Jan, Bruges, Belgium; 11Depart- ment of Laboratories, Children’s Hospital and Regional Medical Center; 12Department of Pathology, University of Washington, Se- attle, WA; 13Cuyahoga County Coroner’s Office, Cleveland, OH; 14Department of Laboratory Medicine, University of Kentucky, Lexington, KY; 15West Midlands Regional Genetics Unit, Birming-

ham Womens Hospital, Birmingham, United Kingdom; 16Pediatric Neurology, University of Alabama Birmingham, Birmingham, AL; 17Neuropediatrics Unit, Hospital Niño Jesús, Madrid, Spain; 18Wessex Clinical Genetics Service, Southampton, United King- dom; 19Clinical Neurosciences, University of Southampton, Southampton, United Kingdom; and 20Waisman Center and De- partment of Pathobiological Sciences, University of Wisconsin- Madison, Madison, WI.

Received Oct 14, 2004, and in revised form Dec 28. Accepted for publication Dec 28, 2004.

Published online Feb 24, 2005, in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.20406

Address correspondence to Dr Michael Brenner, Department of Neurobiology and Civitan International Research Center, University of Alabama Birmingham, Birmingham, AL. E-mail: aaron@nrc.uab.edu

pathology primarily involves the brainstem and cerebel- lum, sparing the frontal lobes.

Although the infantile, juvenile, and adult forms of Alexander disease can differ markedly, they are united by the abundant presence of Rosenthal fibers: protein aggregates within astrocytes containing glial fibrillary acidic protein (GFAP), &B-crystallin, and heat shock protein 27.5-10 Rosenthal fibers are particularly plenti- ful in perivascular, subpial, and subependymal astro- cytes. Rosenthal fibers are also associated with other neurological disorders, including pilocytic astrocyto- mas,11 multiple sclerosis plaques,12 and syrinx cavities.5 However, in these instances, Rosenthal fibers are more restricted in their distribution and not typically present in subpial or subependymal positions. Nevertheless, be- cause Rosenthal fibers can occur in multiple condi- tions, it is unclear whether the three different forms of Alexander disease arise from a shared cause, or if each is a distinct disease.

Missense mutations were identified in the GFAP gene in 11 of 12 infantile Alexander disease patients.13 The mutations were heterozygous in the patients and absent from the seven parental pairs tested, indicating that the disorder arises from de novo, dominant mis- sense mutations. To further understand the role of GFAP mutations in the juvenile, adult, and atypical in- fantile cases of Alexander disease, we analyzed an addi- tional 44 patients.

Patients and Methods

Selection of Patients

This investigation was performed under institutional review board approvals from the University of Alabama Birming- ham (UAB), Albert Einstein College of Medicine, the Uni- versity of Wisconsin-Madison, and the VU University Med- ical Center (VUMC). Patients were included in this study if a biopsy or autopsy provided pathological evidence of Alex- ander disease, or if magnetic resonance imaging (MRI) was suggestive of the disorder. Patients are listed in Table 1, and their clinical characteristics are summarized in Table 2. Those analyzed at VUMC were Patients 2, 9,11, 16-20, 22, 26-29, 31, 35, 36, 39, 40, 42); all other patients were an- alyzed at UAB.

DNA Preparation

Tissue sources were blood samples (Patients 1-3, 5, 6, 8, 9, 11, 12, 16-22, 25-32, 35, 36, 38-40, 42), immortalized lymphocytes (Patients 10, 14, 33, 34, 37, 41, 44), frozen brain obtained at biopsy (Patient 7) or autopsy (Patients 4, 13, 23, 43), or paraffin-embedded biopsy (Patient 15) or au- topsy (Patient 24) material. DNA analyzed at UAB was pre- pared using the PureGene kit (Gentra Systems, Minneapolis, MN); DNA analyzed at VUMC was prepared using the QIAamp blood kit (QIAGEN, Hilden, Germany). For Pa- tients 15 and 24, the source of DNA was paraffin-embedded tissue obtained at biopsy (brain) or autopsy (liver), respec-

tively. This DNA was prepared as Coombs and colleagues18 described, with the addition of a reverse cross-linking step.19,20

Sequence Analysis

Polymerase chain reaction products obtained from the genomic DNA samples were sequenced using the BigDye Terminator v3.0 Cycle Sequencing Ready Reaction kit and an Applied Biosystems 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA; polymerase chain reaction primers and reaction details available on request). The pres- ence of mutations was confirmed either by sequencing the other strand (VUMC patients and L235P) or by restriction enzyme digestion. These confirmatory methods were also used to analyze the DNA of parents, sibs, and control sub- jects for the presence of a specific mutation where indicated in Table 1. In several instances, exons 1, 4 and/or 8 were analyzed first, because all but 1 of the 41 mutations previ- ously described were present in these segments.21 Analysis was halted if a mutation was discovered. Table 3 provides a summary of the sequenced exons.

Functional Analysis of Mutant Glial Fibrillary Acidic Proteins

The expression plasmids used were pcDNA3.1-hGF(WT), a pcDNA3.1-based vector (Invitrogen, Carlsbad, CA) car- rying the coding region of human wild-type GFAP, or this same plasmid modified to encode the specified amino acid changes. pcDNA3.1-hGF(WT) was constructed by cloning an XbaI/EcoRI fragment from a pET expression vector23 into the corresponding sites of pcDNA3.1; it contains the GFAP complementary DNA sequence from base pair 15 (the initiating ATG) to 1313 (the TGA stop codon).22 Plasmids encoding the altered proteins were generated from pcDNA3.1-hGF(WT) by replacing a unique restriction fragment with one generated by polymerase chain reaction that carried the mutation. To evaluate functionality, we transfected the expression vectors into SW13vim” cells.24 Cells were grown in Dulbecco’s modified Eagle medium (Mediatech/cellgro, Herndon, VA) supplemented with 10% fetal bovine serum, 100U/ml penicillin, and 100mg/ml streptomycin. Two days before transfection, 1 × 104 cells were plated onto glass coverslips in 24-well plates. Cells were transfected overnight using 0.4µg plasmid with Lipo- fectamine Plus (Invitrogen). Two days after transfection, cells on coverslips were washed with phosphate-buffered sa- line and fixed with 4% paraformaldehyde in phosphate- buffered saline for 15 minutes. Primary antibodies (1:2,000) were rabbit polyclonal anti-cow GFAP (DAKO, Carpinteria, CA) and mouse monoclonal anti-vimentin (Sigma, St. Louis, MO); secondary antibodies (1:100) were Alexa Fluor488 conjugated goat anti-rabbit IgG and Alexa Fluor594 conjugated anti-mouse IgG (Molecular Probes, Eugene, OR). Cells were incubated overnight at 4℃ with the primary antibodies, then for 1 hour at room tempera- ture with the secondary antibodies. Vimentin staining was performed to monitor for reversion of the vim” phenotype; none was observed.

Table 1. Description of Cases
Case No.	Mutation			Sex/ Form	Age		Clinical Symptoms						Pathology	MRI	Parents or Controls
					Age		Macro- cephaly	Seizures	Spasticity	Bulbar and/or pseudo- bulbar signs	Ataxia	Cogni- tive defect
					Onset	Death (+) or last report
	Protein	cDNA	Protein Structure		Onset	Death (+) or last report
1	K63Q	187A>C	Precoil	F/adlt	21y	43y	No	No	Yes	Yes	Yes	No	Biopsy	Atyp	Mo, 2 sisters and 50 controls
2	M73T	218T>G	1A	M/inf	3m	2y	Yes	Yes	Yes	Yes	Yes	Yes	No	Typ	WT
3	L76F	226C>T	1A	F/inf	1w	9yt	No	Yes	Yes	Yes	Yes	Yes	Autopsy	Typ	WT
4	L76V	226C>G	1A	F/inf	5m	10y 7m+	No	Yes	Yes	Yes	NA	Yes	Autopsy	Typ	WT
5	N77S	230A>G	1A	F/inf	7m	2y 3m™	Yes	Yes	No	Yes	No	Yes	Autopsy	Typ	WT
6	R79C	235C>T	1A	M/juv	7y	14y	Yes	No	Yes	Yes	No	Yes	No	Typ	PT
7	R79C	235C>T	1A	F/inf	1y 6m	6y 11m+	Yes	Yes	No	No	Yes	Yes	Biopsyª	Nob	PT
													& au-
8	R79C	235C>T	1A	F/inf	7m	19y	Yes	Yes	Yes	No	Yes	Yes	topsy No	Typ	PT
9	R88C	262C>T	1A	M/juv	6y	10y	No	Yes	No	No	No	Yes	No	Typ	PT
10	L97P	290T>C	1A	M/inf	Birth	6y 5m+	Yes	Yes	Yes	No	No	Yes	No	Typ	PT
														rbo
11	E207K	619G>A	1B	M/juv	10y	17y	No	No	No	Yes	No	No	No	Atyp	WT
12	E207Q	619G>C	1B	M/juv	11y 6mº	16y	No	No	Yes	No	No	No	No	Atyp	WT
13	E210K	628G>A	1B	F/adlt	24y	33 yr1	No	No	No	Yes	Yes	No	Autopsy	Atyp	Mo and 50 controls
														rbod
14€	L235P	704T>C	2A	M/juv	3y	10 yr+	No	No	Yes	Yes	Yes	No	No	Atyp	WT
15€	L235P	704T>C	2A	M/juv	3y	8y	No	No	Yes	Yes	Yes	No	Biopsy &	Atyp	WT
													autopsy
16	L235P	704T>C	2A	M/juv	2y 6m	12y	Yes	Yes	Yes	Yes	Yes	Yes	No	Typ	WT
17	R239C	715C>T	2A	M/juv	2y	3y	No	Yes	Yes	Yes	Yes	Yes	No	Typ	WT
18	R239C	715C>T	2A	F/inf	1y 2m	2y	Yes	Yes	Yes	No	No	No	No	Typ	WT
19	R239C	715C>T	2A	M/inf	8m	8y	Yes	Yes	No	Yes	Yes	Yes	No	Typ	WT
20	R239C	715C>T	2A	M/inf	1y 6m	8y+	No	Yes	No	Yes	Yes	Yes	No	Typ	WT
21	R239C	715C>T	2A	M/inf	1m	2y 5m+	Yes	Yes	No	Yes	NA	Yes	Autopsy	Typ	PT
22	R239C	715C>T	2A	F/juv	4y	13y	No	No	No	No	Yes	Yes	Biopsy	Atyp	PT
23	R239H	716G>A	2A	M/inf	3m	11.5m+	Yes	Yes	Yes	Yes	NA	Yes	Autopsy	Typ	PT
24	R239Hf	716G>A	2A	F/inf	2m	11.5m+	Yes	Yes	Yes	Yes	NA	No?	Autopsy	Typ	PT
25	R239H	716G>A	2A	M/inf	4m	5y 3m1	Yes	Yes	Yes	Yes	NA	Yes	No	Typ	PT
26	R239H	716G>A	2A	M/inf	6m	3y	No	No	Yes	Yes	NA	Yes	No	Typ	WT
27	R239H	716G>A	2A	M/inf	3m	6m	No	Yes	Yes	No	NA	NA	No	Typ	WT
28	R239P	716G>C	2A	M/juv	2y	18y	Yes	No	No	Yes	Yes	Yes	Biopsy	Atyp	WT
29	A244V	731C>T	2A	F/juv	9y	17y	No	No	No	Yes	No	No	No	Atyp	Mo and controls PT
30	A253G	758C>G	L2ª	M/inf	<ly	8y	No	No	No	Yes	No	No	No	Atyp	Mo and 50 controls
31	K279E	835A>G	2B	M/juv	5y	13y	No	Yes	No	Yes	No	Yes	No	Typ	WT
32	349HLins		2B	F/inf	1m	3.5m+	No	Yes	No	Yes	NA	NK	Autopsy	Typ	WT
33	L352Pf	1055T>C	2B	M/inf	Birth	38d+	Yesh	Yes	No	No	NA	NA	Autopsy	Typ	WT
34	L359V	1075C>G	2B	M/juv	3y	20y	No	No	Yes	Yes	No	Yes	No	Typ	WT
					6m
35	A364P	1090G>C	2B	M/inf	Birth	4m+	Yes	Yes	No	No	NA	NA	Autopsy	No	WT
36	Y366H	1096T>C	2B	M/inf	1m	1.5y	No	Yes	No	No	NA	Yes	No	Typ	WT
37	E373K	1117G>A	tail	F/inf	6w	6m1	Yes	Yes	No	Yes	NA	Yes	Autopsy	Typ	WT
38	E373K	1117G>A	tail	M/inf	2m	2y	Yes	NK	NK	Yes	NK	Yes	Biopsy	Typ	PT
														rbo‘
39	E373Q	1117G>G	tail	M/inf	3m	6m	Yes	Yes	No	No	No	No?	No	Typ	WT
40	E374G	1121A>G	tail	F/inf	1y 6m	4y	No	Yes	Yes	No	Yes	Yes	No	Typ	WT
41	R416Wf	1246C>T	tail	M/juv	10y	25y	No	No	Yes	Yes	Yes	No	Biopsy	Typ	PT
42	V115	343G>T	1B	M/inf	<1w	8m	No	Yes	Yes	Yes	NA	Yes	No	Atyp	Mo and 50
															controls = WT Fa
															= V1151
43	no	na	na	M/adlt	45y	59y+	No	No	Yes	Yes	Yes	No	Autopsy	Atyp	NA
														rbok
44	no	na	na	F/juv	9y	13y 9m	No	No	No	No	Yes	Yes	No	Typ	NA

Results

Patients were included in this study if pathology or MRI analysis was judged consistent with Alexander dis- ease. An overview of the clinical, pathological, and MRI features of the patients analyzed is provided in Table 1. Of the 44 patients studied, 41 are newly re- ported and 3 were previously described clinically (see Table 1, footnote f).

Neuropathology

Typical pathology for infantile patients was abundant Rosenthal fibers in perivascular, subpial, and sub-

ependymal regions, as well as throughout the white matter. There was also markedly diminished myelin, especially in the frontal lobes. In later onset patients, the Rosenthal fibers tended to be primarily in the brainstem, cerebellum, or both. In the cerebellum, they were mainly subpial and in white matter; in the brainstem, they were subpial, but they also were present below the floor of the fourth ventricle, in teg- mental regions, and scattered through otherwise normal-appearing brain tissue. They were also found in patchy, focal regions in hemispheric white matter and in the spinal cord. All the pathologically proven

Table 1. Continued

Explanations of column headings are as follows: Mutation: previously described mutations are indicated by standard font, those involving novel amino acids changes are shown in boldface, and those occurring at novel sites are underlined; nucleotide numbering is for the coding region of the cDNA (the A of the initiating ATG = position 1); Protein Structure shows the location in the protein structure based on themdel of Strelkov and colleagues14; Classification: inf = infantile (< 2 yr at onset), juv = juvenile (2-12 yr at onset), adlt = adult = ≥ 13 yr at onset; Age: ages are in days (d), weeks (w), months (m), or years (y); Pathol: indicates pathology-based diagnosis after autopsy or biopsy; Macro- cephaly: yes = >95th percentile; MRI: MRIs are listed as typ = typical, atyp = atypical (atypical cases are discussed in the text), or no = no MRI available. All MRIs were reviewed by the authors (MvdK or SN) unless noted as rbo = reviewed by others; Parents or controls: WT = parents tested and found to be wild type, PT = parents for another patient with this mutation were previously tested and found to be WT, controls = number of control individuals tested and found to be WT, Mo = mother, Fa = father. Other symbols used are NA = not applicable (eg, spasticity for patients who never walked), NK = not known, ? = likely but uncertain based on information available. Restriction enzymes used to confirm mutations were as follows: K63Q, ScrFl; L76F, Sacl; L76V, MnlI; N77S, Ddel; R79C/H, Acil; L97P, BsaHI; E207Q, BslI; E210K, Hpy188I; R239C/H, Acil; A253G, Bs/I; Y349_Q350insHL, ScrFI; L352P, AlwNI; E373K, Mnll; L359V, ScrFI; R416W, Acil.

aPathology at the EM level for this patient was previously published.7

bNo MRI, but computed axial tomography scan reported as typical.

“Age of onset is presentation for scoliosis; chronic bed-wetting began at an earlier age.

dRead by attending physician. Reported at age 29 yr as atypical with symmetrical abnormal signal intensity within the periventricular deep white matter, most prominent in parietal and occipital regions, but also involving frontal and temporal areas and accompanied by symmetrical brainstem atrophy and abnormal contrast enhancement.

ePatients 14 and 15 are identical twins.

Clinical findings have been previously published for Cases 23,15 32,16 and 4017 (second patient in this article).

§Present in the interhelical linker region between coils 2A and 2B.

h75th percentile, but rapidly expanding prior to death.

¡Reported as typical, but with additional increased signal intensity bilaterally within the cerebral peduncles, crossing fibers of the pons, and the dentate nuclei.

‘This patient is heterozygous for a V115I change which is also present in his unaffected father and is provisionally classified as a polymorphism (see Discussion).

KA computed tomography scan at age 46 reportedly showed mild cerebral cortical atrophic changes and an MRI at age 56 was considered unremarkable except for a mild change around the left middle cerebral artery.

“Three MRI criteria were fulfilled, one criterion could not be assessed (no contrast administered), and one criterion was not met (brainstem involvement), making the MRI pattern suggestive of Alexander disease, but not fully typical.

Table 2. Summary of Patient Characteristics
Sex or Clinical Sign	Number Displaying Clinical Signa (yes/total = %)
Sex or Clinical Sign	Infantile	Juvenile	Adult
Male	16	12	1
Female	10	3	2
Macrocephaly	16/26 = 62%	3/15 = 20%	0/3 = 0%
Seizure	23/25 = 92%	4/15 = 27%	0/3 = 0%
Spasticity	13/25 = 52%	8/15 = 53%	1/3 = 33%
Bulbar and/or pseudobulbar	16/26 = 62%	11/15 = 73%	3/3 = 100%
Ataxia	7/12 = 58%	7/15 = 47%	3/3 = 100%
Cognitive function defect	18/22 = 82%	9/15 = 60%	0/3 = 0%

“Total numbers do not sum to 44 for each of the clinical signs, because some data were unavailable.

cases displayed these typical diagnostic features except for Patient 43. This patient did have Rosenthal fibers in the brainstem in the expected positions, but they were unusually sparse.

Neuroimaging

Available MRIs were evaluated (M.S.v.d.K.). Patients meeting previously published criteria25 are labeled as “typical” in Table 1. Those labeled “atypical” did not meet these criteria, yet were suggestive of Alexander disease. These latter patients are either described in footnotes to Table 1 (Patients 13, 43, 44) or discussed in the accompanying article26 (Patients 1, 11, 12, 14, 15, 22, 28-30, 42).

Clinical Features

Based on age at onset, patients were classified as infan- tile (<2 years old, 26 patients), juvenile (2-12 years old, 15 patients), or adult (≥13 years old, 3 patients). Age at onset for the infantile patients ranged from birth to 1.5 years; outcomes ranged from death at 38 days old to surviving at 19 years old. For the juvenile patients, onset ranged from 2 to 11.5 years; outcome ranged from death at 3 years old to surviving at 23 years old. Onsets for the 3 adult patients occurred at 21, 24, and 45 years; outcomes were surviving at 43 years old and death at 33 and 59 years old, respec- tively. The age categories impose arbitrary boundaries on what is actually a graded continuum of clinical pre-

Table 3. Exons Sequenced
Exon	Amino Acidsª	Patient No.
1	1-154	1-20, 22, 26-29, 31-40, 42-44
2	155-174	2, 9, 11, 16-20, 22, 26-29, 31- 33, 35-40, 42-44
3	175-206	2, 9, 11, 16-20, 22, 26-29, 31- 33, 35-40, 42-44
4	207-260	2-6, 9, 11-29, 31-40, 42-44
5	261-302	2, 9, 11, 16, 18-20, 22, 26-29, 31, 32, 35-37, 39, 40, 42-44
6	303-376	2, 9, 11, 14-20, 22, 26, 28, 29, 31-40, 42-44
7	376-391	2, 9, 11, 16-20, 22, 26, 28, 29, 31-33, 35, 36, 39, 40, 42-44
8	391-419	2, 6, 9, 11, 12, 14-20, 22, 26, 28, 29, 31-44
9	420-432	2, 9, 11, 17-20, 22, 26, 28, 29, 31-33, 35, 36, 39, 40, 42-44

ªSee Brenner and colleagues.22 If a codon is split between two ex- ons, the corresponding amino acid is listed for both.

sentations, but they are useful to illustrate the changes in disease features with time of onset. For example, macrocephaly, seizures, and impaired cognitive func- tion were relatively common in infantile patients but progressively less common in juvenile and adult pa- tients (see Table 2).

Most of the infantile patients presented with seizures and failure to meet mental and motor milestones, fol- lowed by a downward progression leading to death. However, several had special characteristics. Patient 33 was the most severe case, with feeding problems noted at birth, and death at 38 days.16 Patient 32 had a dev- astating course, with onset at 1 month of age accom- panied by severe vomiting and gastroesophageal reflux. Intractable epilepsy subsequently developed, and she died at 3.5 months. Patient 3 presented with a seizure at 1 week of age and is notable for being microcephalic rather than macrocephalic. She died at 9 years of age during surgery to correct severe scoliosis. Patient 30 presented at younger than 1 year of age with recurrent vomiting and was below the third percentile in height and weight. However, this patient had features sugges- tive of a later onset form, including normal cognitive function and abnormal signal in the brainstem on MRI. The patient also had autonomic signs including labile hypotension and anhydrosis. Patient 24 displayed the psychomotor retardation and macrocephaly com- mon in the infantile form, as well as bulbar signs usu- ally associated with the juvenile form. MRI and pathol- ogy correspondingly showed abnormalities in both the frontal lobes and brainstem (see Gingold and col- leagues15 for details).

Several juvenile patients also had unusual features. Patients 14 and 15 were identical male twins who had

normal births and a healthy brother. They presented at 3 years of age with malnutrition, were below the 3rd percentile in height and weight, and because of self- induced regurgitation, were diagnosed with anorexia. They experienced development of progressive bulbar dysfunction, including central apneas, and difficulty with swallowing and walking. Patient 14 also had se- vere scoliosis. Both were treated for suspected low- grade gliomas by radiotherapy and chemotherapy. Pa- tient 15 died at 8 years of age and Patient 14 died at 10 years of age from cardiorespiratory failure. Autopsy of Patient 15 showed massive accumulation of Rosenthal fibers. MRI results for these patients were atypical, and they are described in the accompanying article.26 Patient 12 had scoliosis, urinary urgency, and enuresis but is without cognitive impairment. Patient 44 also had urinary urgency, in addition to occasional incontinence, bowel urgency, frontal headaches, abnor- mal gait, difficulty with balance, and a “high stocking” distribution of numbness in her legs. She has dimin- ished cognitive function without macrocephaly or epi- lepsy.

Each of the three adult-onset patients has unique features. Patient 1 presented at 21 years of age with nystagmus and ataxia. She had no seizures or macro- cephaly and received a tentative diagnosis of multiple sclerosis. She experienced development of dysphagia, miosis, parkinsonism, generalized hyperreflexia, cere- bellar signs, and insidiously progressive gait impair- ments. Although the patient became wheelchair bound 19 years after symptomatic onset, her cognitive func- tion remains normal. At 40 years old, MRI indicated brainstem and cerebellar abnormalities, and a brain- stem biopsy to test for glioma instead showed numer- ous Rosenthal fibers. No neurological problems were reported for either parent or her two sisters; an older brother has an undefined behavioral disorder. The mother and two sisters do not carry the patient’s mu- tation, the father is deceased, and the brother refused to provide a DNA sample for analysis.

Patient 13 had a normal childhood and development until 24 years of age, was without macrocephaly, sei- zures, or spasticity, and was of above-average intelli- gence. Her first symptom was loss of balance and leg weakness. She had intermittent diplopia and dysarthria. As the disease progressed, she experienced development of episodic daytime somnia and sleep apnea and re- quired a wheelchair for locomotion and a gastric tube for feeding. She had increasing respiratory difficulties until her death near 34 years of age.

Patient 43 had difficulty with balance, hoarseness, and “noises in my head” at 45 years of age. A com- puted tomography scan performed the following year showed mild cerebral and cerebellar cortical atrophy. At 47 years old, he underwent coronary revasculariza- tion surgery. At 51 years old, he experienced develop-

ment of marked elevation of the right diaphragm and atelectasis of the right lung and began to have respira- tory symptoms and sleep apnea, dysarthria, and dys- phagia. He experienced development of progressive weakness and clumsiness, right foot drop, and spastic- ity, especially on the right side. Over the next few years, his symptoms progressed. The respiratory symp- toms were particularly disabling, and oxygen inhalation became necessary until his death at 59 years of age. An autopsy demonstrated scattered Rosenthal fibers throughout much of the brain. The patient had four siblings, three of whom had clinically similar disorders. A sister was thought to have multiple sclerosis, but at autopsy was diagnosed with Alexander disease; her identical twin and another sister were dysarthric and ataxic and had epilepsy. All three sisters died between the ages 48 and 55 years. An apparently unaffected male sibling remains well. Both parents had diabetes; their neurological statuses were unknown.

Sequencing Results

Coding changes that are candidates for causing Alex- ander disease were detected in 42 of our 44 patients (see Table 1). Consistent with previous findings, each of the mutations was heterozygous, indicating a domi- nant effect. Figure 1 displays examples of the sequenc- ing and restriction enzyme digestion data obtained.

Strong evidence that a mutation is disease causing is its absence from both parents, because the chance that a mutation would arise at random in an average sized gene such as GFAP is only about 1 in 20,000.27 For most of the previously detected mutations, at least one case has been described for which both parents were tested and found to be wild type (indicated by PT for previously tested in Table 1). For the novel mutations, or ones previously described but for which both par- ents were not tested, parental DNA was analyzed if available (eg, see Fig 1C, D). DNA from both parents was available from all but four patients in this group,

Fig 1. Representative sequencing and restriction enzyme results. (A) Patient 12 has a heterozygous G to C transition at base pair 619 (arrow) (3' strand sequence is shown). (B) Patient 32 has a heterozygous six-base insertion at position 1047 (arrow), resulting in double peaks in the sequencing chromatogram for most positions from the point of the insertion onward. (C, D) Restriction en- zyme digestion patterns for polymerase chain reaction (PCR) fragments spanning the mutation sites of Patients 12 and 32, respec- tively. In each case, the heterozygous mutation eliminates a restriction site from half of the PCR products, resulting in the presence of an additional digestion product. The parents (F = father; M = mother) are wild type.

12F

12M

STD

32M

32F

190

145

109

131

114

118

and all were wild-type except for the father of Patient 42, who also harbored the V115I change, despite being neurologically normal. For the other four patients in this group, Patients 1 (K63Q), 13 (E210K), 29 (A244V), and 30 (A253G), the mother tested wild type, but paternal DNA was not available.

To gain further information about the disease relat- edness of these four amino acid substitutions and the V115I change, we analyzed control DNA for their presence, and a functional study of the altered protein was performed. For all five, the corresponding alter- ation was not found in 100 control chromosomes, sug- gesting that the coding changes are either disease- causing or rare polymorphisms. In the functional tests, vectors expressing the candidate proteins were tran- siently transfected into SW13vim- cells, which lack endogenous intermediate filaments,24 and their ability to form normal-appearing filament networks was com- pared with that for the wild-type protein. Both the V115I and A244V proteins formed filaments indistin- guishable from wild-type, whereas the K63Q, E210K, and A253G proteins produced only diffuse background staining, aggregates, or both (Fig 2).

Presumptive polymorphisms discovered in the course of sequencing are presented in Table 4. These are nu- cleotide changes present in an unaffected parent, as well as in the patient. However, particularly for changes not found in control subjects, these could be disease-related mutations with incomplete penetrance. Three such mutations, V115I, D157N, and E223Q, are considered further in the Discussion.

Discussion

Numbers and Types of Cases

A primary goal of this research was to better define the association of GFAP mutations with Alexander disease, particularly for unusual infantile cases and for the later onset forms. A total of 44 patients were investigated, of which 42 were believed to have Alexander disease (see later for a discussion of the two exceptions, Patients 42 and 44). These patients included 26 with the infantile form, 15 (including a pair of identical twins) with the juvenile form, and 3 with adult onset. GFAP coding changes were found for all patients except one juvenile and one adult patient. As has been previously found for Alexander disease patients, all the changes were het- erozygous, indicating a dominant effect. One might ex- pect equal numbers of male and female individuals to be affected because the mutations arise de novo on a somatic chromosome (GFAP is on chromosome 17).29 However, a male predominance has been suggested for pathologically diagnosed infantile cases.1 When our new infantile patients are combined with those already described with GFAP mutations,13,30-35 male patients (n = 38) do outnumber female patients (n = 28), but

this difference is not statistically significant (p = 0.268 by the x2 test of proportions; inherited cases are not included, and identical twins are counted as a single case). However, the difference for all juvenile patients with GFAP mutations13,30,36-40 (15 male vs 5 female patients) does border on statistical significance (p = 0.044). This raises the possibility of a sex-related ge- netic or environmental factor in the juvenile form of Alexander disease.

Review of Later Onset Cases

While this work was in progress, other groups pub- lished cases of juvenile and adult Alexander disease with GFAP mutations (these studies are summarized in Table 5). Most of these studies concerned individual patients or families, which does not provide firm infor- mation about the frequency of GFAP mutations in sus- pected cases because negative findings are unlikely to be reported. However, a group of five juvenile patients was identified by clinical and MRI criteria, and GFAP coding mutations were found in all patients.3º Simi- larly, we found GFAP mutations in 17 of 18 later on- set patients (see Table 1). Thus, although the infantile and later onset forms can differ markedly in their clin- ical signs, GFAP mutations are present in a high pro- portion of each, indicating a common cause.

Disease-Relatedness of Novel Mutations

As suspected, a high proportion of the mutations in the later onset patients are novel (both adult muta- tions and 7 of the 15 juvenile mutations). Surpris- ingly, 10 of 26 of the infantile mutations are also novel, suggesting that many additional GFAP changes leading to Alexander disease may yet be discovered. To obtain further insight into their potential impor- tance, we scored these new mutation sites for conser- vation among related proteins. Table 6 shows they have a high level of conservation among other GFAP species and with the related type III intermediate fil- ament proteins vimentin and desmin. Only the K63 position, which is in the more variable precoil do- main of the protein, displays a nonconservative sub- stitution in desmin. Homology is also evident with more distantly related intermediate filaments for many of these novel sites. Disease-causing mutations in other intermediate filaments were found at 7 of 12 of the highly homologous positions (see Table 6). Later onset patients and the most severe infantile pa- tient account for the sites not associated with other disease mutations. Because all these proteins are pre- sumed to share a similar central rod structure and po- lymerization mechanism, it will be of interest whether mutations at these sites are eventually found in other intermediate filament family members.

The strong conservation of nearly all of the novel mutations suggests they are important for GFAP

Fig 2. Functional testing of glial fibrillary acidic protein (GFAP) mutants. GFAP expression vectors were transfected into SW13vim cells, and, 2 days later, the cells were examined for GFAP filament formation as described in Patients and Methods. (A) Cells transfected with wild-type GFAP displayed a normal-appearing filamentous network extending throughout the cytoplasm. (B) The K63Q mutant yielded protein aggregates and some ringlike structures (arrows). (C) The V115I GFAP yielded normal- appearing filaments. (D) E210K produced only short, thick, needle-like filaments. (E) The A244V amino acid change had no effect on normal filament assembly. (F) The A253G mutation resulted in aggregate formation.

function. In most instances, a role for the new mu- tations in Alexander disease was confirmed by finding that the change arose de novo (M73T, L76V, N77S, E207K, E207Q, L235P, K279E, Y349_Q350insHL,

L352P, L359V, A364P, Y366H, E373Q, and E374G; see Table 1), or the mutation was absent from control samples and the protein was defective in filament for- mation (K63Q, E210K, A253G; see Fig 2).

Table 4. GFAP Polymorphisms Detected
Polymorphismª	Frequencyb	Present (patient no.)	Absent (patient no.)
Exon 1
c140C > T (P47L)	1/76 (1.3%)	20	1-19, 22, 26-29, 31-40, 42-44
c141G>A (P47P)	2/76 (2.6%)	19, 29	1-18, 20, 22, 26-28, 31-40, 42-44
c343G>A (V115I)	1/176 (0.6%)	42	1-20, 22, 26-29, 31-40, 43, 44 and 50 controls
Exon 2
c469G>A	6/150 (4.0%)	35, 5 controls	2, 9, 11, 16-20, 22, 26-29, 31-33, 36-40,
(D157N)			42-44 and 45 controls
Intron 3 IVS3-12C>T	6/76 (7.9%)	9, 12, 18, 26, 39, 40	2-6, 11, 13-17, 19-25, 27-29, 31-38, 42-44
Exon 4 c667G>C (E223Q)	1/76 (1.3%)	36	2-6, 9, 11-29, 31-35, 37-40, 42-44
Exon 5 c858G>A (R286R)	6/46 (13.0%)	17, 19, 22, 26, 31, 32	2, 9, 11, 16, 18, 20, 27-29, 35-37, 39, 40, 42-44
Intron 5 IVS5-100A>C	4/28 (14.3%)	9, 18, 39, 40	11, 16, 19, 20, 28, 29, 31, 35, 36, 42
Intron 6
IVS6-66 C>G	1/8 (12.5%)	33	32, 43, 44
Intron 8
IVS8-86 C>T	7/36 (19.4%)	11, 28, 29, 31, 32, 33, 42	2, 8, 17-20, 26, 35, 36, 39, 40
3'UTR		11, 28, 29, 31, 32, 33, 42
+28C>G	4/18 (22.2%)	19, 28, 36, 42	16, 29, 32, 33, 44
+33C>G	1/8 (12.5%)	43	32, 33, 44

aPositions of polymorphisms are indicated by the nomenclature of den Dunnen and Antonarakis.28 For polymorphisms located in exons, the c# represents the cDNA position. For polymorphisms located in introns, the IVS# is the intron, followed by the number of nucleotides after the preceeding exon (positive numbers) or before the next exon (negative numbers). For example, IVS3-12C>T means there is a C to T change in intron 3, 12 base pairs before exon 4.

bFrequency is for chromosomes (not individuals) calculated from the cases and controls listed under the Present and Absent columns. All cases showing polymorphisms were heterozygous, and the change was found it an unaffected parent. GFAP = glial fibrillary acidic protein.

Glial Fibrillary Acidic Protein Polymorphisms

Putative polymorphisms found in this study are listed in Table 4. In several instances, it is not clear if a coding change is a harmless polymorphism or is disease related. Although the V115I change was absent from tested con- trol subjects, we tentatively conclude it is a polymor- phism because it was present in the patient’s unaffected father, the change of an isoleucine for valine is present in goldfish and zebrafish GFAP (see Table 6), and the V115I protein assembled into normal-appearing fila- ments in SW13vim- cells (see Fig 2C). The significance of the A244V change found for Patient 29 is also un- certain, because the parents were not available for test- ing, and the altered protein also formed normal- appearing filaments in transfected cells (see Fig 2E). The MRI for our patient had features suggestive of Alexander disease but did not fully meet the published criteria. However, an A244V change was found in another clin- ically diagnosed Alexander disease patient with typical MRI results.39 In that patient, both parents also were not available, but the mother tested wild type and the change was not present in 130 control chromosomes. The site is conserved among other GFAP species but not

among other intermediate filaments. Because of the typ- ical MRI in the previous case and the absence of the change in a normal parent, the A244V is tentatively clas- sified as disease causing. Two other changes of uncertain consequence were present in Patients 35 and 36. Patient 35 had both a de novo A364P mutation and a D157N change that was also present in his neurologically normal father, and Patient 36 had both a de novo Y366H mu- tation and an E223Q change that was present in his neurologically normal mother. We tentatively classify both D157N and E223Q as polymorphisms by their presence in normal parents, but incomplete penetrance is equally plausible. E223Q was previously found in a putative Alexander disease patient,43 and the site is highly conserved among intermediate filaments (see Ta- ble 6). However, interpretation of both MRI and clinical signs for that patient are complicated by his alcoholism and hypertension, and the patient’s mother carries the same change.

Noteworthy New Mutations

The 42 patients reported here nearly double both the number of Alexander disease patients shown to harbor

Table 5. Later Onset Cases
Type	Mutation		Sex	Age		Pathol	MRI	Parents or Controls	Comments	Ref
Type	Amino acid	Prot Struc	Sex	Onset	Death (+) or last report	Pathol	MRI	Parents or Controls	Comments	Ref
Juv	M73R	IA	M	9y	15y	No	Yes	96 controls	Symptoms include bulbar signs, seizures, spasticity, and cognitive defects	30
Juvª	D78E	IA	M/Fª	a	a	Yes	Yes	a	Familial case; see details in footnote a	41
Adult	R79C	IA	F	12.5y	14y+	Yes	Yes	PT	CSF lactate elevation	37
Adult	V87G	IA	F/F/M	53y	58y	No	Yes	200 controls	Familial case; see details in footnote b	42
Juv	R88C	IA	M	7y	9y	No	Yes	PT	Symptoms include bulbar	30
Juv	R88C	IA	M	c	4y	No	Yes	PT	signs and cognitive defects Asymptomatic	30
Adult	E223Qd	IB/2A linker	M	40y	40y	No	Yes	75 controls	Fluctuating left hemiplegia; mutation also present in mother	43
Juv	A244V	2A	M	2y	10y	No	Yes	Mother and 65 con- trols	Seizures, mental retardation, dysarthria, clumsiness	39
Adult	R276L	2B	M/M	33ye	53y+	Yes	Yes	78 controls	Two brothers; see footnote e	44
Juv	E362D	2B	M	13y	21y	No	Yes	Yes	Progressive physical and mental deficits	40
Juv	R416W	Tail	F	5y	7y	No	Yes	PT	Symptoms were bulbar signs	30
Juv	R416W	Tail	F		14y	No	Yes	PT	Initially asymptomatic; now displaying some cognitive decline	30
Adult	R416W	Tail	M	24y	30y	No	Yes	PT	Bulbar signs, muscle weak- ness, autonomic disorders; megalencephaly	45
Adult	R416W	Tail	F/M	42y	45y+	No	Yes	Yes	Mother and son; see foot- note f	46

Abbreviations are the same as for Table 1.

aPatients were from a large family with affected individuals in three generations, two of which had a pathological diagnosis of Alexander disease on autopsy. There were a variety of symptoms, but all had bulbar signs, dysautonomia (severe constipation), sleep disturbances, and dysmor- phism. Onset was from 5 to 10 years based on occurrence of the severe constipation, but ranged to over 50 years for neurological symptoms other than sleep disturbances. Unlike our adult-onset cases, several patients had seizures. MRIs were typical of later onset cases. A D78E mutation was found in all five affected patients tested, including members of each generation. It was also found in one unaffected family member in the direct blood line of 16 tested. This person is currently 30 years old and has not had a physical examination; however, the absence of both dysautonomia and a sleep disorder suggests incomplete penetrance.

bAges are for the proband. Also carrying the mutation were her daughter (disease onset at 38 yr) and son (neurologically normal at 32 yr except for lower extremity hyperreflexia and Babinski signs; he was considered preclinical). Mother and daughter show bulbar signs, unsteady gait, lower limb weakness, scoliosis, and palatal myoclonus but no cognitive impairment. Parents of the proband were not available for testing but presumed unaffected.

^Asymptomatic; GFAP was sequenced when MRI performed for another condition suggested Alexander disease. dAs discussed in the text, there is some question whether the E223Q mutation is disease-causing and whether this patient has Alexander disease. e Ages shown are for the elder brother; a 3-year-old younger brother had onset at 48 yr and survives. The elder brother had bulbar signs, spastic tetraplegia, and urinary retention; the younger brother developed a progressive spastic gait and Babinski signs. Parents (not tested) were neurologically normal and died in their 70s.

“Ages are for themther; although she presented at the clinic at age 42, she had scoliosis from age 12 and a progressive gait disturbance starting at age 37. She subsequently developed palatal tremor, dysphagia, and dementia. Her affected son has mild spastic paraparesis and some cognitive impairment.

GFAP mutations and the number of unique mutations. Figure 3 presents the location within the GFAP pro- tein of the newly identified (right side) and previously reported (left side) mutations. A total of 42 different coding changes are now associated with Alexander dis- ease, involving 31 different sites along the protein. Sev- eral of the newly discovered mutations are noteworthy.

K63Q is the first example of an alteration in the N-terminal portion of the protein before the start of coil 1A. E207K, E207Q, and E210K are the first mu- tations in the 1B rod domain. A253G is the first mu- tation in the linker joining coils 2A and 2B. The Y349_Q350insHL insertion is the first mutation in Al- exander disease other than a simple substitution. The

Table 6. Conservation of GFAP Mutation Sites and Related Disease-Associated Mutations
Site	Conservationª			Homologous Mutations in Other IF Diseasesb
Site	GFAPs	Vim/Des	Other IFsc	Homologous Mutations in Other IF Diseasesb
K63Q	zR	dL	K, A, 2D, 3G, 2Q, S, 3X	none
M73T	Conserved	vL, dL	7M, 4I, 2L	K9-M156V47; K9-M156T48; K10-M150R49; K10-M150T50; K12-M129T51; K13- M108T52; K14-M119T53; K14-M119V53; K14-M119I54; K17-M88T55
L76V	Conserved	Conserved	Conserved	K5-L175F56; K9-L159V57; K10-L153V58; K14-L122F59; K16-L124R60
N77S	Conserved	Conserved	Conserved	K1-N188S61; K1-N188T62; K2e-N192Y63; K2e-N192D64; K2e-N192K65; K5- N176S6; K6a-N171K67; KhHb6- N114H68; KhHb6-N114D68; K9- N160Y69; K9-N160K7º; K9-N160S71; K10-N154H72; K14-N123S66; K13- N112S73; K16-N125S74; K17-N92D75, K17-N92S76, K17-N92H76
V115Id	gL, zLl	vL, dL	2A, C, 4L, 2P, Q, 3Y	None
D157Nd	Conserved	vE, dE	3D, A, 3E, H, 2N, M, 2T	None
E207K/Q	Conserved	Conserved	Conserved	None
E210K	mD	Conserved	3E, 3A, I, D, 3Q, S, V	None
E223Qª	gD, zD	vD	8E, D, 2K, 2S	None
L235P	Conserved	Conserved	9L, 4/L	None
A253G	mT, rT, gT, zT	Conserved	11A, 1L	None
K279E	Conserved	Conserved	9K, A, 2R, H	None
L352P	Conserved	Conserved	11L, 2V	None
L359V	Conserved	Conserved	Conserved	K10-L442Q72, K14-L408M77 K5-L463P78
A364P	Conserved	Conserved	9A, 2E, H, Q	K14-A413T79
Y366H	Conserved	Conserved	Conserved	K12-Y429D80 K14-Y415H81, K14-Y415C82, K1-Y482C49
E374G	Conserved	Conserved	6E, A, 4D, S, Q	K1-E490Q83, K2eE494K84

Except for E223Q, data are presented only for new mutation sites reported in this article.

aAbbreviations used: m = mouse (accession # VEMSGF), r = rat (NP_058705), z = zebrafish (AAR19286), g = goldfish (AAA49166), v = vimentin (CAA26571), d = desmin (AAC50680); e.g., zR = arginine in zebrafish.

bHomologies are designated by the name of the intermediate filament (K: keratin) followed by the amino acid change and the reference. “Other intermediate filaments surveyed include K1, K5, Hb6, Hb1, K9, K10, K12-14, lamin and neurofilament light chain. Numbers refer to the number of times a given amino acid (letter) is found in the homologous position in the other intermediate filaments surveyed; eg, 2A, C, 4L means alanine is present twice, cysteine once, and leucine four times. 3X for the K63Q site indicates no alignment in this region with 3 of the other IFs. Boldface letters indicate the identical amino acid as in human GFAP, and italics a conserved subsitution. dTentatively classified as a polymorphism (see text).

inserted segment is an exact match of an HL coding sequence commencing just 15 nucleotides 5’ of the in- sertion site, suggesting that the mutation resulted from a copying error. The finding of a precise insertion of amino acids, rather than a frameshift, further suggests that Alexander disease mutations act through a domi- nant gain of function. In contrast, truncated, frame- shifted proteins, such as found in keratin diseases, re- sult in loss of function caused by polymerization failure.86,87

Genotype-Phenotype and Structure- Function Relations

Our results support the previous observation that mu- tations at the R239 locus are typically severe, with those involving an R239H change being especially dev- astating (see Fig 3).31 All five of our R239H patients

had an onset before 6 months of age, and all died by 5 years of age. Among this group is a case (Patient 24) diagnosed at autopsy as Alexander disease, but initially thought to have Leigh’s syndrome based on increased serum lactate, pyruvate and alanine and increased cere- brospinal fluid protein and pyruvate content.15 The in- creased lactate content, however, may actually be a consequence of Alexander disease, because it has also been reported for a case involving an R79C muta- tion,37 and increased white matter lactate was observed in multiple Alexander disease patients by magnetic res- onance spectroscopy (M.S.v.d.K., unpublished observa- tions). Surprisingly, three juvenile cases at the R239 site were found, two with an R239C mutation (cases 17 and 22) and one with an R239P mutation (case 28). Symptom onset for these patients ranged from 2 to 4 years. Patient 22 survives at 13 years of age with

Fig 3. Location of Alexander disease-associated mutations in glial fibrillary acidic protein in relation to the protein domain structure of intermediate filaments. The vertical red boxes indicate the four a-helical segments within the central rod domain, with the initial filled segment indicating the helical precoil domain, which is not conserved among intermediate filaments. The red lines between the boxes represent the nonhelical linkers, and the vertical black lines represent the nonhelical N-terminal head and C-terminal tail domains. Previously published mutations are shown to the left, and mutations described in this article are shown to the right. The high frequency of mutations at R79, R88, R239, and R416 may result from methylation of the CpG-containing codons.85 Each symbol represents an individual patient, or a family in the case of familial patients or identical twins. Multiple independent occurrences of a mutation are indicated by the number of symbols (a single symbol with an inter- nal asterisk is used for the set of cases from a single family). Symbols are color-coded for clinical category based on age at onset (infantile, juvenile, or adult). The two colors in the symbol for D78E indicate that this family had both juvenile- and adult- onset patients. If more than one amino acid change has been reported at the same codon, the change for each case is shown within the colored symbol.

K63Q

☒

M73R

M73T

☒ L76F

L76

V ☒

☒ N77Y

N77S

☒

O ☒ D78E

C C

G H

R79

R79C

☒ ☒

V87G

COC C

R88

R88C

☒

L90P

☒

L97P

☒

E207

K ☒ Q

E210K

☒

☒ E223Q

L235P ☒ C C

R239

☒

Y242D

A244V

☒

R258P

A253G

☒

R276L

K279E ☒

HL349-50ins

☒

☒ infantile

☒

L331P

L352P

☒

L359V

☒ juvenile

☒

E362D

A364P

☒

☒ adult

Y366H

☒

☒ E373K

O inherited

E373

☒ K

E374G

☒

K ☒

☐ asymptomatic

☒ R416W

R416W

☒

minimal cognitive impairment and an ataxic gait. Pa- tient 28 remains ambulatory at 18 years of age.

Another clustering of severe cases is present near the end of the 2B helical region. Patient 33, carrying an L352P mutation, was the most fulminant case of any described, with onset at birth and death at 38 days of age.16 Others within this group (Patients 32, 35-39) had onsets between birth and 3 months of age, and died between 4 months and 2 years of age. The lone milder case in this region (Patient 34) may be a reflection of the highly conservative change in

the L359V mutation. In addition to these severe cases, a clustering of juvenile and adult forms is present, extending from the 1B region to the start of 2A (Patients 11-16).

Accumulating data about the mechanism of interme- diate filament assembly allow some structural correlates of the above phenotype-genotype relations. The first stages in polymerization, formation of coiled-coil dimers and their association into tetramers, are believed to be stabilized by interactions of specific “trigger” se- quences within the 1B (amino acids 191-203) and 2B

rod domains (amino acids 357-369).88 Three muta- tion sites were found in the 2B trigger sequence, but none were found within the 1B trigger sequence. In- termediate filament polymers are believed to elongate through head-to-tail interactions involving an overlap of the 1A and 2B end regions.89 This interaction likely explains the especially strong sequence conservation of these segments and the clustering of disease-causing mutations in these regions in both Alexander disease and other genetic disorders of intermediate filaments.89 In GFAP, the conserved N-terminal region spans amino acids 72 to 86, and contains four mutation sites, including the R79 hot spot. A change as conservative as L76V in this region can be devastating (Patient 4). The conserved C-terminal region extends from amino acid 363 to 376, overlapping the 2B trigger sequence. This dual function may explain the special severity of the GFAP mutations in or near this region. The cluster of mutations near the beginning of coil 2A, including the R239 hot spot, may reflect a role for this region in the alignment of coiled-coils as demonstrated by cross- linking data for vimentin.90

Like other a-helices, the polypeptide backbone in the rod domain completes two corkscrew turns about every seven residues. If the amino acid positions in these heptad repeats are designated a,b,c,d,e,f,g, resi- dues a and d are usually hydrophobic, often leucines, resulting in a hydrophobic stripe that slowly winds around the surface of the helix. Interaction of two monomers along their stripes produces the initial coiled-coil dimer. About half of the mutations de- scribed in this article affect residues in these a or d positions: M73, L76, L97, E207, L235, L352, L359, and Y366. It is notable that the E207Q mutation pro- duces disease, albeit a relatively mild form, because this amino acid substitution would be expected to strengthen the hydrophobic interactions in the dimer. This result suggests that the occasional charged residue in the a or d position serves an important structural role. In contrast, the L352P mutation may be particu- larly pernicious because prolines disrupt «-helical structures. Similarly, the Y349_Q350insHL mutation may be severe because insertion of the two amino acids misaligns the heptad repeats. The dimer is also stabi- lized by intrachain and interchain ionic bonds formed between polar residues. For example, amino acids in the b and e positions will also align along one side of the helix, allowing their side chains to interact. Amino acids mutated in these positions among our cases in- clude K63, N77, R88, R239, K279, and A364. Two mutation sites occur at the g position, R79 and E210. Changes at R79 can be severe, probably because it is part of the 1A initiator region, but the E210K muta- tion has relatively mild effects, even though it involves a reversal of charge. Only one of the mutations occurs in the c position, A244V, and it also produces a rela-

tively mild phenotype; none of our mutations occurs in position f. Interestingly, the only previously reported Alexander disease mutation in position c was also an A244V in a juvenile patient,39 and the only mutation in position f is D78E, which occurs in an extended family with adult-onset Alexander disease.41

In contrast to these genotype-phenotype-structure relations, the effects of some mutations are poorly un- derstood. For example, the patient with an R239P change, which should disrupt the a-helix, had a milder disease than those with either R239C or R239H mu- tations. More pointedly, in several instances, the same mutation produces distinctly different outcomes, dem- onstrating that other genetic or environmental factors can influence the phenotype. For example, the R416W mutation has been found in all three forms of Alex- ander disease. 13,30,46

Diagnosis and Prognosis

The diagnoses for 19 of our patients were confirmed by pathology (14 at autopsy, and 5 by biopsy). Patho- logical findings of abundant Rosenthal fibers, espe- cially in subpial and subependymal regions, were con- sidered the definitive diagnosis of infantile Alexander disease, whereas in later onset cases, the Rosenthal fi- bers may be more sparse and largely confined to the cerebellum and brainstem. At their extremes, the findings for the infantile and later onset forms are quite distinct; however, several of our cases simulta- neously showed both types of pathology and clinical features, indicating a continuum of one form into the other. For example, Patient 15 is a juvenile case that displayed the intense presence of Rosenthal fibers in the frontal lobes characteristic of infantile patients, whereas Patient 24 is an infantile case that showed lesions in the brainstem, as well as in the frontal lobes. Another infantile case (Patient 30) displayed only the brainstem lesions found in later onset cases. Eighteen of 19 cases with pathology were found to have disease-causing mutations. The pathology for the single exception (adult-onset Patient 43) was reexam- ined by two of us (A.B.J., J.E.G.) and confirmed to display the typical brainstem distribution of Rosenthal fibers, but was unusual for their sparseness. However, the diagnosis is supported by a reported pathological determination of Alexander disease in the patient’s sister.

Another 25 of our patients were diagnosed by MRI. Eighteen of the MRI-diagnosed patients in our study met established MRI criteria,25 and all proved to have a disease-causing GFAP mutation. One patient (Patient 44) met three of the requested four MRI criteria, but she did not receive contrast; therefore, the criterion of enhancement of certain structures could not be as- sessed. She lacked the fifth criterion of brainstem in- volvement. Although she did not fully meet the MRI

criteria, she was believed to have MRI results highly suggestive of Alexander disease. No GFAP mutation was found in this patient, and it is therefore unlikely that she has Alexander disease. Six other patients (Pa- tients 11, 12, 14, 29, 30, 42) did not meet the criteria but were included because the images had some fea- tures suggestive of Alexander disease (see the accompa- nying article26). GFAP missense changes were found in each of these cases, but, as noted, it is unlikely that the V115I change in Patient 42 is disease related. Although MRIs can have high diagnostic efficacy,30,31 it is our experience that the images need to be interpreted by radiologists or neurologists experienced with leukodys- trophies. During the early phase of this investigation, we sequenced DNA from six patients considered to have Alexander disease based on their clinical signs and evaluation of MRIs by their attending radiologists. Of this group, only one patient was found to have a GFAP mutation. However, when the MRIs were evaluated by one of us (M.S.v.d.K., blinded to DNA results for all but one case), only the case yielding a mutation met the criteria for Alexander disease, and therefore was re- tained in this study.

The patients showed great variation in presentation, particularly among the later onset patients (see Tables 1 and 2). Of the 16 juvenile and adult patients with GFAP mutations, 12 displayed 3 or fewer of the 6 listed symptoms, and Patient 12 initially showed none, presenting instead with scoliosis and bed-wetting. Ataxia and bulbar or pseudobulbar signs were the most consistent symptoms of the later onset patients, and in the infantile form, nearly all patients displayed seizures. However, individually these are relatively nonspecific characteristics, and the data do not show a systematic combination of symptoms that could sharpen the di- agnosis. Accordingly, for cases lacking pathology, eval- uation of an MRI by a physician experienced in inter- preting leukodystrophies is recommended before commencing DNA sequencing.

The age at onset proved a fairly good indicator of the prognosis for our patients, with infantile patients averaging a survival of about 3.6 years after onset, ju- venile patients averaging 8.1 years, and adult patients averaging 15.0 years (these are minimal estimates, be- cause they include ages of surviving patients). A re- markable exception to this generality is Patient 8, an infantile case with onset at 7 months of age who is surviving at 19 years of age. However, despite her long survival, her language skills are at an 18-month-old level. She requires a wheelchair, and she is not toilet trained. Excluding this case, the average survival after onset for infantile cases is 3.0 years.

Concluding Remarks

The results of this study greatly expand the number of Alexander disease patients found to have GFAP

coding mutations, particularly for the later onset pa- tients, confirming the usefulness of DNA testing for diagnosis of all Alexander disease forms. It is now clear that most cases of all three forms are manifesta- tions of the same disease, and that dominant GFAP missense mutations are the primary underlying cause. The dominant effect of the mutations appears to be caused by a gain of function rather than a loss of function, because human null mutations have not been found, and GFAP null mice do not show signs of the disorder (see review by Li and colleagues21). How the many different GFAP mutations, distributed throughout the protein, produce the same gain of function remains a puzzle that is the subject of con- tinuing research.

This study was supported by the NIH (National Institute of Neu- rological Disorders and Stroke, P01NS42803, M.B., J.E.G., R.Q., A.M .; RO1NS39055, M.B .; Mental Retardation Research Center, (P30HD38985, M.B.), the Lei Foundation (M.B.), and the United Leukodystrophy Foundation (A.B.J.).

Human tissues and case information were provided by Drs S. Schochet and G. Raymond; they were also obtained from the Na- tional Institute of Child Health and Human Development Brain and Tissue Bank for Developmental Disorders under National In- stitute of Child Health and Human Development contracts N01- HD-43368 and NO1-HD-8-3284; and from the National Neuro- logical Research Specimen Bank, VAMC, sponsored by National Institute of Neurological Disorders and Stroke/National Institute of Mental Health, the National Multiple Sclerosis Society, the He- reditary Disease Foundation, and the Veterans Health Services and Research Administration, Department of Veterans Affairs.

We thank R. Anderson for technical assistance in all phases of this research, including preparation of the manuscript, Dr M. Su for assistance with cell culture and immunohistochemistry, Dr M. Lind- strom for statistical analysis, and J. Willett for preparation of Figure 2. We also thank the Albert Einstein College of Medicine Human Genetics Program for preparation of lymphoblasts and DNA sam- ples, the UAB MMRC Molecular Biology Core for recombinant DNA support, and the UAB Genomics Core Facility of the Howell and Elizabeth Heflin Center for Human Genetics and the VUMC Metabolic Unit of the Department of Clinical Chemistry for DNA sequencing. We are indebted to the patients and their families for participating in this study.

References

1. Russo LS Jr, Aron A, Anderson PJ. Alexander’s disease: a report and reappraisal. Neurology 1976;26:607-614.

2. Messing A, Goldman JE, Johnson AB, Brenner M. Alexander disease: new insights from genetics. J Neuropathol Exp Neurol 2001;60:263-273.

3. Johnson AB, Brenner M. Alexander disease: clinical, patholog- ical and genetic features. J Child Neurol 2003;18:625-632.

4. Messing A, Goldman JE. Alexander disease. In: Lazzarini RA, ed. Myelin and its diseases. San Diego: Academic Press, 2004: 851-866.

5. Rosenthal W. Über eine eigenthümliche, mit Syringomyelie komplizierte Geschwulst des Rückenmarks. (Bietr) Betrieb Pathol Anat 1898;23:111-143.

6. Alexander WS. Progressive fibrinoid degeneration of fibrillary astrocytes associated with mental retardation in a hydrocephalic infant. Brain 1949;72:373-381.

7. Johnson AB, Bettica A. On-grid immunogold labeling of glial intermediate filaments in epoxy-embedded tissue. Am J Anat 1989;185:335-341.

8. Tomokane N, Iwaki T, Tateisha J, et al. Rosenthal fibers share epitopes with &B-crystallin, glial fibrillary acidic protein, and ubiquitin, but not with vimentin. Immunoelectron microscopy with colloidal gold. Am J Pathol 1991;138:875-885.

9. Iwaki T, Kume-Iwaki A, Liem RK, Goldman JE. Alpha B-crystallin is expressed in non-lenticular tissues and accumu- lates in Alexander’s disease brain. Cell 1989;57:71-78.

10. Iwaki T, Iwaki A, Tateishi J, et al. Alpha B-crystallin and 27-kd heat shock protein are regulated by stress conditions in the cen- tral nervous system and accumulate in Rosenthal fibers. Am J Pathol 1993;143:487-495.

11. Grcevic N, Yates PO. Rosenthal fibers in tumors of the central nervous system. J Pathol Bacteriol 1957;73:467-472.

12. Schlote W. [Piloid astrocytes (spongiocytes) and Rosenthal fi- bers in multiple sclerosis (author’s transl)]. Acta Neuropathol (Berl) 1975;33:35-44.

13. Brenner M, Johnson AB, Boespflug-Tanguy O, et al. Mutations in GFAP, encoding glial fibrillary acidic protein, are associated with Alexander disease. Nat Genet 2001;27:117-120.

14. Strelkov SV, Herrmann H, Aebi U. Molecular architecture of intermediate filaments. Bioessays 2003;25:243-251.

15. Gingold MK, Bodensteiner JB, Schochet SS, Jaynes M. Alex- ander’s disease: unique presentation. J Child Neurol 1999;14: 325-329.

16. Bassuk AG, Joshi A, Burton BK, et al. Alexander disease with serial MRS and a new mutation in the glial fibrillary acidic protein gene. Neurology 2003;61:1014-1015.

17. Deprez M, D’Hooghe M, Misson JP, et al. Infantile and juve- nile presentations of Alexander’s disease: a report of two cases. Acta Neurol Scand 1999;99:158-165. (Erratum in: Acta Neu- rol Scand 1999;100:354.)

18. Coombs NJ, Gough AC, Primrose JN. Optimisation of DNA and RNA extraction from archival formalin-fixed tissue. Nu- cleic Acids Res 1999;27:e12.

19. Ip YT, Jackson V, Meier J, Chalkley R. The separation of tran- scriptionally engaged genes. J Biol Chem 1988;263: 14044-14052.

20. Solomon MJ, Varshavsky A. Formaldehyde-mediated DNA- protein crosslinking: a probe for in vivo chromatin structures. Proc Natl Acad Sci U S A 1985;82:6470-6474.

21. Li R, Messing A, Goldman JE, Brenner M. GFAP mutations in Alexander disease. Int J Dev Neurosci 2002;20:259-268.

22. Brenner M, Lampel K, Nakatani Y, et al. Characterization of human cDNA and genomic clones for glial fibrillary acidic pro- tein. Brain Res Mol Brain Res 1990;7:277-286.

23. Sekimata M, Tsujimura K, Tanaka J, et al. Detection of protein kinase activity specifically activated at metaphase-anaphase tran- sition. J Cell Biol 1996;132:635-641.

24. Hedberg KK, Chen LB. Absence of intermediate filaments in a human adrenal cortex carcinoma-derived cell line. Exp Cell Res 1986;163:509-517.

25. Van der Knaap MS, Naidu S, Breiter SN, et al. Alexander disease: diagnosis with MR imaging. Am J Neuroradiol 2001; 22:541-552.

26. Van der Knaap MS, Salomons GS, Li R, et al. Unusual variants of Alexander disease. Ann Neurol 2005;00:00-00.

27. Nachman MW, Crowell SL. Estimate of the mutation rate per nucleotide in humans. Genetics 2000;156:297-304.

28. den Dunnen JT, Antonarakis SE. Nomenclature for the de- scription of human sequence variations. Hum Genet 2001;109: 121-124.

29. Bongcam-Rudloff E, Nister M, Betsholtz C, et al. Human glial fibrillary acidic protein: complementary DNA cloning, chromo- some localization, and messenger RNA expression in human glioma cell lines of various phenotypes. Cancer Res 1991;51: 1553-1560.

30. Gorospe JR, Naidu S, Johnson AB, et al. Molecular findings in symptomatic and pre-symptomatic Alexander disease patients. Neurology 2002;58:1494-1500.

31. Rodriguez D, Gauthier F, Bertini E, et al. Infantile Alexander disease: spectrum of GFAP mutations and genotype-phenotype correlation. Am J Hum Genet 2001;69:1134-1140.

32. Shiroma N, Kanazawa N, Kato Z, et al. Molecular genetic study in Japanese patients with Alexander disease: a novel mu- tation, R79L. Brain Dev 2003;25:116-121.

33. Meins M, Brockmann K, Yadav S, et al. Infantile Alexander disease: a GFAP mutation in monozygotic twins and novel mu- tations in two other patients. Neuropediatrics 2002;33: 194-198.

34. Shiroma N, Kanazawa N, Izumi M, et al. Diagnosis of Alex- ander disease in a Japanese patient by molecular genetic analy- sis. J Hum Genet 2001;46:579-582.

35. Trollmann R, Kraus C, Orlova N, et al. In vivo diagnosis of infantile Alexander disease by molecular genetic analysis of GFAP gene [German]. Monatsschrift fur Kinderheilkunde 2003;151:311-314.

36. Imamura A, Orii KE, Mizuno S, et al. MR imaging and 1H-MR spectroscopy in a case of juvenile Alexander disease. Brain Dev 2002;24:723-726.

37. Probst EN, Hagel C, Weisz V, et al. Atypical focal MRI lesions in a case of juvenile Alexander’s disease. Ann Neurol 2003;53: 118-120.

38. Shiihara T, Kato M, Honma T, et al. Fluctuation of computed tomographic findings in white matter in Alexander’s disease. J Child Neurol 2002;17:227-230.

39. Aoki Y, Haginoya K, Munakata M, et al. A novel mutation in glial fibrillary acidic protein gene in a patient with Alexander disease. Neurosci Lett 2001;312:71-74.

40. Sawaishi Y, Yano T, Takaku I, Takada G. Juvenile Alexander disease with a novel mutation in glial fibrillary acidic protein gene. Neurology 2002;58:1541-1543.

41. Stumpf E, Masson H, Duquette A, et al. Adult Alexander dis- ease with autosomal dominant transmission: a distinct entity caused by mutation in the glial fibrillary acid protein gene. Arch Neurol 2003;60:1307-1312.

42. Okamoto Y, Mitsuyama H, Jonosono M, et al. Autosomal dominant palatal myoclonus and spinal cord atrophy. J Neurol Sci 2002;195:71-76.

43. Brockmann K, Meins M, Taubert A, et al. A novel GFAP mu- tation and disseminated white matter lesions: adult Alexander disease? Eur Neurol 2003;50:100-105.

44. Namekawa M, Takiyama Y, Aoki Y, et al. Identification of GFAP gene mutation in hereditary adult-onset Alexander’s dis- ease. Ann Neurol 2002;52:779-785.

45. Kinoshita T, Imaizumi T, Miura Y, et al. A case of adult-onset Alexander disease with Arg416Trp human glial fibrillary acidic protein gene mutation. Neurosci Lett 2003;350:169-172.

46. Thyagarajan D, Chataway T, Li R, et al. Dominantly-inherited adult-onset leukodystrophy with palatal tremor caused by a mu- tation in the glial fibrillary acidic protein gene. Mov Disord 2004;19:1244-1248.

47. Hennies HC, Zehender D, Kunze J, et al. Keratin 9 gene mu- tational heterogeneity in patients with epidermolytic palmo- plantar keratoderma. Hum Genet 1994;93:649-654.

48. Covello SP, Irvine AD, McKenna KE, et al. Mutations in ker- atin K9 in kindreds with epidermolytic palmoplantar kerato- derma and epidemiology in Northern Ireland. J Invest Derma- tol 1998;111:1207-1209.

49. Syder AJ, Yu QC, Paller AS, et al. Genetic mutations in the K1 and K10 genes of patients with epidermolytic hyperkeratosis. Correlation between location and disease severity. J Clin Invest 1994;93:1533-1542.

50. Paller AS, Syder AJ, Chan YM, et al. Genetic and clinical mo- saicism in a type of epidermal nevus. N Engl J Med 1994;331: 1408-1415.

51. Corden LD, Swensson O, Swensson B, et al. Molecular genetics of Meesmann’s corneal dystrophy: ancestral and novel muta- tions in keratin 12 (K12) and complete sequence of the human KRT12 gene. Exp Eye Res 2000;70:41-49.

52. Rugg E, Magee G, Wilson N, et al. Identification of two novel mutations in keratin 13 as the cause of white sponge naevus. Oral Dis 1999;5:321-324.

53. Cummins RE, Klingberg S, Wesley J, et al. Keratin 14 point mutations at codon 119 of helix 1A resulting in different epi- dermolysis bullosa simplex phenotypes. J Invest Dermatol 2001;117:1103-1107.

54. Chen H, Bonifas JM, Matsumura K, et al. Keratin 14 gene mutations in patients with epidermolysis bullosa simplex. J In- vest Dermatol 1995;105:629-632.

55. Celebi JT, Tanzi EL, Yao YJ, et al. Mutation report: identifi- cation of a germline mutation in keratin 17 in a family with pachyonychia congenita type 2. J Invest Dermatol 1999;113: 848-850.

56. Nomura K, Shimizu H, Meng X, et al. A novel keratin K5 gene mutation in Dowling-Meara epidermolysis bullosa simplex. J Invest Dermatol 1996;107:253-254.

57. Endo H, Hatamochi A, Shinkai H. A novel mutation of a leucine residue in coil 1A of keratin 9 in epidermolytic pal- moplantar keratoderma. J Invest Dermatol 1997;109: 113-115.

58. Ishiko A, Akiyama M, Takizawa Y, et al. A novel leucine to valine mutation in residue 7 of the helix initiation motif of keratin10 leads to bullous congenital ichthyosiform erythro- derma. J Invest Dermatol 2001;116:991-992.

59. Yamanishi K, Matsuki M, Konishi K, Yasuno H. A novel mu- tation of Leu122 to Phe at a highly conserved hydrophobic res- idue in the helix initiation motif of keratin 14 in epidermolysis bullosa simplex. Hum Mol Genet 1994;3:1171-1172.

60. Smith FJ, Fisher MP, Healy E, et al. Novel keratin 16 muta- tions and protein expression studies in pachyonychia congenita type 1 and focal palmoplantar keratoderma. Exp Dermatol 2000;9:170-177.

61. Yang JM, Chipev CC, DiGiovanna JJ, et al. Mutations in the H1 and 1A domains in the keratin 1 gene in epidermolytic hyperkeratosis. J Invest Dermatol 1994;102:17-23.

62. Arin MJ, Longley MA, Kuster W, et al. An asparagine to thre- onine substitution in the 1A domain of keratin 1: a novel mu- tation that causes epidermolytic hyperkeratosis. Exp Dermatol 1999;8:124-127.

63. Smith FJ, Maingi C, Covello SP, et al. Genomic organization and fine mapping of the keratin 2e gene (KRT2E): K2e V1 domain polymorphism and novel mutations in ichthyosis bul- losa of Siemens. J Invest Dermatol 1998;111:817-821.

64. Takizawa Y, Akiyama M, Nagashima M, Shimizu H. A novel asparagine-aspartic acid mutation in the rod 1A domain in ker- atin 2e in a Japanese family with ichthyosis bullosa of Siemens. J Invest Dermatol 2000;114:193-195.

65. Whittock NV, Ashton GH, Griffiths WA, et al. New muta- tions in keratin 1 that cause bullous congenital ichthyosiform erythroderma and keratin 2e that cause ichthyosis bullosa of Siemens. Br J Dermatol 2001;145:330-335.

66. Sorensen CB, Ladekjaer-Mikkelsen AS, Andresen BS, et al. Identification of novel and known mutations in the genes for keratin 5 and 14 in Danish patients with epidermolysis bullosa simplex: correlation between genotype and phenotype. J Invest Dermatol 1999;112:184-190.

67. Smith FJ, McKenna KE, Irvine AD, et al. A mutation detection strategy for the human keratin 6A gene and novel missense mu- tations in two cases of pachyonychia congenita type 1. Exp Dermatol 1999;8:109-114.

68. Korge BP, Hamm H, Jury CS, et al. Identification of novel mutations in basic hair keratins hHb1 and hHb6 in monilethrix: implications for protein structure and clinical phe- notype. J Invest Dermatol 1999;113:607-612.

69. Torchard D, Blanchet-Bardon C, Serova O, et al. Epidermo- lytic palmoplantar keratoderma cosegregates with a keratin 9 mutation in a pedigree with breast and ovarian cancer. Nat Genet 1994;6:106-110.

70. Reis A, Hennies HC, Langbein L, et al. Keratin 9 gene muta- tions in epidermolytic palmoplantar keratoderma (EPPK). Nat Genet 1994;6:174-179.

71. Bonifas JM, Matsumura K, Chen MA, et al. Mutations of ker- atin 9 in two families with palmoplantar epidermolytic hyper- keratosis. J Invest Dermatol 1994;103:474-477.

72. Chipev CC, Yang JM, DiGiovanna JJ, et al. Preferential sites in keratin 10 that are mutated in epidermolytic hyperkeratosis. Am J Hum Genet 1994;54:179-190.

73. Terrinoni A, Rugg EL, Lane EB, et al. A novel mutation in the keratin 13 gene causing oral white sponge nevus. J Dent Res 2001;80:919-923.

74. Shamsher MK, Navsaria HA, Stevens HP, et al. Novel muta- tions in keratin 16 gene underly focal non-epidermolytic pal- moplantar keratoderma (NEPPK) in two families. Hum Mol Genet 1995;4:1875-1881.

75. McLean WH, Rugg EL, Lunny DP, et al. Keratin 16 and ker- atin 17 mutations cause pachyonychia congenita. Nat Genet 1995;9:273-278.

76. Smith FJ, Corden LD, Rugg EL, et al. Missense mutations in keratin 17 cause either pachyonychia congenita type 2 or a phe- notype resembling steatocystoma multiplex. J Invest Dermatol 1997;108:220-223.

77. Schuilenga-Hut PH, Vlies P, Jonkman MF, et al. Mutation analysis of the entire keratin 5 and 14 genes in patients with epidermolysis bullosa simplex and identification of novel muta- tions. Hum Mutat 2003;21:447.

78. Dong W, Ryynanen M, Uitto J. Identification of a leucine-to- proline mutation in the keratin 5 gene in a family with the generalized Kobner type of epidermolysis bullosa simplex. Hum Mutat 1993;2:94-102.

79. Chao SC, Yang MH, Lee SF. Novel KRT14 mutation in a Taiwanese patient with epidermolysis bullosa simplex (Kobner type). J Formos Med Assoc 2002;101:287-290.

80. Nishida K, Honma Y, Dota A, et al. Isolation and chromo- somal localization of a cornea-specific human keratin 12 gene and detection of four mutations in Meesmann corneal epithelial dystrophy. Am J Hum Genet 1997;61:1268-1275.

81. Hut PH, v d Vlies P, Jonkman MF, et al. Exempting homol- ogous pseudogene sequences from polymerase chain reaction amplification allows genomic keratin 14 hotspot mutation anal- ysis. J Invest Dermatol 2000;114:616-619.

82. Ciubotaru D, Bergman R, Baty D, et al. Epidermolysis bullosa simplex in Israel: clinical and genetic features. Arch Dermatol 2003;139:498-505. (Erratum in: Arch Dermatol. 2003;139: 1084.)

83. Rothnagel JA, Dominey AM, Dempsey LD, et al. Mutations in the rod domains of keratins 1 and 10 in epidermolytic hyper- keratosis. Science 1992;257:1128-1130.

84. Jones DO, Watts C, Mills C, et al. A new keratin 2e mutation in ichthyosis bullosa of Siemens. J Invest Dermatol 1997;108: 354-356.

85. Cooper DN, Youssoufian H. The CpG dinucleotide and hu- man genetic disease. Hum Genet 1988;78:151-155.

86. Rugg EL, McLean WH, Lane EB, et al. A functional “knock- out” of human keratin 14. Genes Dev 1994;8:2563-2573.

87. Jonkman MF, Heeres K, Pas HH, et al. Effects of keratin 14 ablation on the clinical and cellular phenotype in a kindred with recessive epidermolysis bullosa simplex. J Invest Dermatol 1996;107:764-769.

88. Wu KC, Bryan JT, Morasso MI, et al. Coiled-coil trigger mo- tifs in the 1B and 2B rod domain segments are required for the stability of keratin intermediate filaments. Mol Biol Cell 2000; 11:3539-3558.

89. Fuchs E. The cytoskeleton and disease: genetic disorders of intermediate filaments. Annu Rev Genet 1996;30: 197-231.

90. Steinert PM, Marekov LN, Parry DA. Molecular parameters of type IV alpha-internexin and type IV-type III alpha-internexin- vimentin copolymer intermediate filaments. J Biol Chem 1999; 274:1657-1666.

Quartz 4

Explorer

15732097