The gene for neuronal apoptosis inhibitory protein is partially deleted in individuals with spinal muscular atrophy

Cell, Vol. 80, 167-178, January 13, 1995, Copyright © 1995 by Cell Press

The Gene for Neuronal Apoptosis Inhibitory Protein Is Partially Deleted in Individuals with Spinal Muscular Atrophy

Natalie Roy, 1, 2, Mani S. Mahadevan, 1.

Michael McLean, 1, 3 Gary Shutler, ~

Zahra Yaraghi, 1, 2 Reza Farahani, ~, 2

Stephen Baird, 1 Anne Besner-Johnston, ~

Charles Lefebvre, ~, 3 Xiaolin Kang, ~, 3

Maysoon Salih, 3 Huguette Aubry, ~

Katsuyuki Tamai, 3 Xiaoping Guan, 4

Panayiotis Ioannou,4t Thomas O. Crawford, 5

Pieter J. de Jong, 4 Linda Surh, 1

Joh-E Ikeda, 3, s Robert G. Korneluk, 1, 3, 7

and Alex MacKenzie ~, 2

~Molecular Genetics Laboratory

Children's Hospital of Eastern Ontario

Ottawa, Ontario K1H 8L1

Canada

2Departments of Biochemistry and Pediatrics

31keda GenoSphere Project

7Departments of Microbiology and Immunology

University of Ottawa

Ottawa, Ontario K1H 8M5

Canada

4Human Genetics Department

Roswell Park Cancer Institute

Buffalo, New York 14263

SDepartment of Neurology

Johns Hopkins University

Baltimore, Maryland 21205

6Department of Molecular and Medical Genetics

The Institute of Medical Science

Tokai University

Kanagawa 259-11

Japan

Summary

The spinal muscular atrophies (SMAs), characterized

by spinal cord motor neuron depletion, are among the

most common autosomal recessive disorders. One

model of SMA pathogenesis invokes an inappropriate

persistence of normally occurring motor neuron apop-

tosis. Consistent with this hypothesis, the novel gene

for neuronal apoptosis inhibitory protein (NAIP) has

been mapped to the SMA region of chromosome

5q13.1 and is homologous with baculoviral apoptosis

inhibitor proteins. The two first coding exons of this

gene are deleted in approximately 67% of type I SMA

chromosomes compared with 2% of non-SMA chro-

mosomes. Furthermore, R T - P C R analysis reveals in-

ternally deleted and mutated forms of the NAIP tran-

script in type I SMA individuals and not in unaffected

individuals. These findings suggest that mutations in

the NAIP locus may lead to a failure of a normally oc-

curring inhibition of motor neuron apoptosis resulting

in or contributing to the SMA phenotype.

*The two first authors contributed equally to this work.

1"Present address: Cyprus Institute of Neurology and Genetics, Nicosia, Cyprus.

Introduction

The childhood spinal muscular atrophies (SMAs) are a

group of autosomal recessive neurodegenerative disor-

ders subclassified as type l (Werdnig-Hoffmann disease),

type II, and type III (Kugelberg-Welander disease) based

primarily upon the age of onset of weakness and clinical

severity (Dubowitz, 1978; Brooke, 1986). There are reports

of mild pathology throughout the central nervous system

(CNS), including the anterior columns, thalamus, dorsal

root ganglia, and lateral geniculate nuclei ('Fowfighi et al.,

1985; Peress et al., 1986; Murayama et al., 1991). How-

ever, the primary abnormality in SMA is a paucity of the

(~ motor neurons, manifested as weakness and wasting

of the voluntary muscles. Type I SMA is the most severe

form, with onset either in utero or within the first few

months of life. Affected children are unable to sit unsup-

ported and rarely survive the first few years owing to respi-

ratory muscle weakness (Dubowitz, 1978; Brooke, 1986).

Type I SMA is the most common monogenic cause of

death in infancy, and, with estimated carrier frequencies

of 1 in 80 for type I and 1 in 100 for types II and III, the

SMAs are among the most common autosomal recessive

disorders (Emery, 1991).

In 1990 genetic linkage with DNA markers situated in

the 5q13 region of chromosome 5 was demonstrated for

all three types of SMA (Brzustowicz et al., 1990; Gilliam

et al., 1990; Melki et al., 1990). These results suggested

that different mutations in the same gene or contiguous

genes account for the three forms of the illness. Since the

mapping of the SMA locus, a series of recombinations

defining a progressively smaller critical SMA-containing

region have been reported (Sheth et al., 1991; Brzustowicz

et al., 1992; Soares et al., 1993; Wirth et al., 1993, 1994;

Francis et al., 1993; Burghes et al., 1994a; Clermont et

al., 1994; Yaraghi et al., submitted). In addition, several

long-range physical maps based on yeast artificial chro-

mosome (YAC) contiguous arrays (Kleyn et al., 1993; Fran-

cis et al., 1993; Wirth et al., 1993; Roy et al., 1995) as

well as a radiation hybrid analysis (Thompson et al., 1993)

of the approximately 4 Mb region have been published.

However, the final step in the search for the SMA gene

has been confounded by the presence of long-range and

variable chromosome 5-specific genomic sequence re-

peats that contain both complex tandem repeat (CTR) poly-

morphisms as well as amplified transcribed sequences map-

ping both inside and outside the critical SMA region

(Francis et al., 1993; Thompson et al., 1993; Roy et al.,

1995; McLean et al., 1994; Burghes et al., 1994b).

Many of the motor neurons observed at autopsy in SMA

spinal cords are characterized by swelling and chromatoly-

sis of cells in a manner that may be consistent with the

process of programmed cell death or apoptosis. Over the

past two decades, a central role for apoptosis in the normal

embryogenesis of the CNS has been delineated (Oppen-

helm, 1991). As many as 50% of cells in the developing

eukaryotic CNS, including motor neurons, undergo apop-

tosis. The motor neuron depletion observed in SMA (Sar-

Cell 168

A L Correlation of Type I SMA 5q13.1 ° marker linkage disequilibrium with ~ L - ~ =. LL \ \ - ~ 1 ~ ~' I

o ~ _ ~ i i ~ i ! i

B 10o~

YACarray , , " " ] ~ I .~,~i '

LI2/:7-;:7; . . . . . . . 7 ' ;

NAIP gene structure lO k b

\

/

23 4 5 6 7 89 10 11 1213 14 1516

k II I I I I I BB B B B B B

t t t CMS I=402'I CA'TTIC 161

Figure 1. Linkage Disequilibrium and Physical Mapping of NAIP Region of 5q13.1

(A) Correlation of the degree of linkage disequilibrium observed in type I SMA families between the disease phenotype and six 5q13.1 markers with the physical map. The SMA- containing interval defined by the key recombinations described in the text is shown. Note the proximity of the disequilibrium peak with the centromeric end of the recombinant- defined SMA interval. (B) YAC contiguous array covering the SMA region of 5q13.1. For both YAC and PAC con- tigs, STSs are denoted by closed triangles, polymorphic tandem repeat polymorphisms by open triangles, and single copy clones by closed squares. Note that our physical map places the CMS1 sublocus containing allele 9 telomeric to the other CMS1 subloci, while the reverse was observed with genetic recombination data, reflecting, we believe, the variation that exists in this region of 5q13.1. (C) PAC contiguous array covering the SMA region of 5q13.1. NAIP gene localization is as depicted; the asterisk denotes a deleted version of the NAIP locus. CBCD541 and the highly homologous SMN gene found by Lefebvre et al. (1995) to be mutated in SMA are also shown. Our mapping is at odds with that found by Le- febvre et al. (1995), who, as a result of the long- range inverted sequence presented in their paper, place NAIP telomeric to SMN. Although we have not observed such a phenomenon, we cannot rule this possibility out conclusively. (D) Gene structure of NAIP.

nat, 1983) has led to suggestions that a genetic defect in

a neuronal apoptotic pathway may result in a pathologic

persistence or reactivation of normally occurring apopto-

sis (Sarnat 1983, 1992; Oppenheim, 1991). In support of

this model, we present a novel gene mapping to the critical

SMA region of 5q13.1 that we believe when deleted may

cause or contribute to the SMA phenotype. The neuronal

apoptotic inhibitory protein (NAIP) contains domains with

sequence similarity to recently characterized baculovirus

proteins that inhibit virally induced insect cell apoptosis

(Clem and Miller, 1993, 1994a, 1994b).

Results

Genetic and Linkage Disequilibrium Analyses Extensive genotyping with a number of 5q13.1 markers

has been conducted upon 96 SMA families (MacKenzie

et al., 1993; McLean et al., 1994; unpublished data). Many

of the markers tested are CTR (CA), polymorphisms inso-

far as they are comprised of several subloci that exist in

a polymorphic number (0-5) of copies dispersed over the

critical SMA region of chromosome 5q13.1 (McLean et al.,

1994; Roy et al., 1995).

We have character ized a recombination event within the

CMS1 CTR (Kleyn et al., 1993), mapping two of the CMS1

subloci and the SMA locus distal to the CMS1 sublocus

containing allele 9 (Yaraghi et al., submitted). This is the

closest recombination we have detected centromeric to

the SMA gene. A recombination between the CMS1 CTR

and SMA has also been observed by van der Steege et

al. (submitted). The closest telomeric marker recombining

with SMA in our families was D5S637 (unpublished data).

However, a closer telomeric recombination between SMA

and D5S557 has been previously reported by Francis et

al. (1993). We estimate the critical 5qcen--CMSl allele

9 sublocus--SMA--D5S557--5qter interval defined by

these recombinations to be approximately 550 kb (Figure

1B; Roy et al., 1995).

A linkage disequilibrium analysis employing six complex

and simple tandem repeats mapping to the SMA region

was also conducted. Two of the polymorphisms employed

in this analysis were the CATT-40G1 and CATT-192F7

subloci of the CATT CTR polymorphism (McLean et al.,

1994; Burghes et al., 1994b). A clear linkage disequilib-

rium peak with the type I SMA phenotype was observed

for the CATT-40G1 sublocus (McLean et al., 1994; Figure

1A). Similar disequilibria have been observed for the A g l

(DiDonato et al., 1994) and C272 (Melki et al., 1994) CTRs.

Physical Mapping of the SMA Region of 5q13.1 Concurrent with our genetic analysis, we constructed a

YAC contiguous array employing clones from three differ-

NAIP, a Spinal Muscular Atrophy Candidate Gene 169

NAIP

P A C s

Exon content of PAC, cDNA and RT-PCR clones

ATG ATPK~TP Occdudin lAP HOMOLOGY bind~g site homology

!11121 ~ 14 12m1,11511.11

125D9

55112

215P15

25017

30B2

238D12

i1121314 ,2m1,11511,]

11 1213 ), 12m1,1151151

12m14115l

Y A C s

Y97

Y98

27H5

24D6

33H10

[ 21314 12mI,11511.1

1213]~ 12m1,11511,l

@12l

121

12mmi.11.1

cDNAs isolated from human fetal brain l i b r a r y

0°23.2 111 215 I ' 1211

CC21.4B

CE3

CD19

CD23

CC14.1

CC20.3

RT-PCR products from SMA tissue

A2, A3, A9, A l l

A3

A2

A9

E []

[]

[]

Figure 2. Exon Content of PAC, Fetal Brain cDNA Clones from Non- SMA Individuals, and RT-PCR Clones from SMA-Affected Individuals

E158 refers to the deletion of a glutamate residue. RT-PCR was performed only between exons 4 and 13; additional undetected deletions may exist outside of this region.

ent YAC libraries (Roy et al., 1995). A minimal representa-

tion from this array, which was correlated with extensive

pulsed field gel electrophoresis analysis, is shown in Fig-

ure lB.

With the initial suggestion of linkage disequilibrium of

the general CATT marker and SMA (Burghes et al., 1994b),

a 220 kb cosmid contiguous array encompassing the five

CATT subloci contained in a monochromosomally derived

flow-sorted chromosome 5 genomic library was constructed

(Roy et al., 1995). More recently, a P1 artificial chromo-

some (PAC) (Ioannou et al., 1994) contiguous array con-

taining the CATT region, comprised of nine clones and

extending approximately 550 kb, was constructed (Fig-

ure 1C).

Our genetic analyses, combined with the physical map-

ping data, indicated that the CATT-40G1 sublocus that

shows the greatest linkage disequilibrium with SMA is du-

plicated and maps to the extreme centromeric end of the

critical SMA interval (Figure 1A). At this point, the 154 kb

PAC clone 125D9 was discovered to contain within 10

kb of its centromeric end the allele 9-containing CMS1

sublocus that defined the centromeric border of the SMA

interval. The determination that the PAC 125D9 insert ex-

tended telomerically to incorporate sequence contiguous

to the CATT-40G1 sublocus as well as roughly one third

of the SMA interval (Figure 1 C) led to its identification both

as a probable site of the SMA locus and as a key resource.

Cloning of the NAIP Gene

Initial isolation of the NAIP transcript was achieved by

probing a human fetal brain cDNA library with the 28 kb

genomic DNA insert of cosmid 250B6 that contains one

of five CATT subloci present in the cosmid library. This

resulted in the detection of a 2.2 kb transcript that ulti-

mately proved to be exon 13 of the NA/P gene.

Libraries were then constructed from complete and par-

tial (average insert size, 5 kb) Sau3AI, BamHI, and

BamHI-Notl digests of PAC 125D9 DNA. Sequencing of

both termini of 200 clones from the 5 kb insert partial

Sau3AI digestion library as well as those of BamHI and

BamHI-Notl clones was conducted in the manner of Chen

et al. (1993), permitting the construction of contiguous and

overlapping genomic clones covering most of PAC 125D9

(data not shown). This has proven instrumental in elucidat-

ing the gene structure of the NAIP locus.

PAC 125D9 is divided into 24 kb centromeric and 130

kb telomeric fragments by Notl digestion that bisects exon

5 of the NAIP gene (see Figures 3 and 4). These Notl

fragments were isolated by preparative pulsed field gel

electrophoresis and used separately to probe human fetal

brain cDNA libraries. Physical mapping and sequencing

of the Notl site region were also undertaken to assay for

the presence of a CpG island, an approach that rapidly

detected coding sequence. PAC 125D9 was also used as a

template in an exon-trapping system (Church et al., 1994)

resulting in the identification of the NA/P gene exons 3,

10, 14, and 15.

These approaches resulted in the rapid identification of

cDNA clones spanning the NAIP locus (Figure 2). Overlap-

ping clones were identified, and chimerism of cDNA clones

was excluded on a number of occasions by the detection

of colinearity of the cDNA clone termini with sequence

from clones of the PAC 125D9 partial Sau3AI digestion

genomic library. Sequence analysis revealed the similarity

between the protein sequence encoded by the NAIP gene

exons 6-12 and two baculoviral inhibitor of apoptosis pro-

teins (lAPs) (Birnbaum et al., 1994). Shortly thereafter,

probing of Southern blots containing DNA from consan-

guinous SMA families with cDNA probes revealed deleted

bands.

NAIP Gene Structure

A schematic of the NAIP transcript with overlapping cDNA

clones is shown in Figure 2, and a physical map of the

Cell 170

4.Skb J

D4dmd r,,a~

m

L ' II N B

Exon 5 Exon 5

NAIP gene s t r u c t u ~ 5 6

B B B E E

p-----4 *.5 ~ I

!

30B2

23 4

I - -

238D12

I 3,o.. t

lkb

EV ~ r E

11213 14 1516

l i r a I II I I I A l l • I IILL E E B B E B E B

~4.5 kt i

exonm encoding protein withlAPhomology

7 85 15

I I # E B B E

23 kb I

~ " ~ 8 9 10 111213 14 1516

i__~11 I [ 1 • I ,],1

I 9.5 kb I

Figure 3. Structure of lntact and InternallyDe- leted/Truncated Versions of the NAIP Gene as Found in the Indicated PACs

Exons are marked as numbered closed boxes. N refers to the Notl site, B to BamHI sites, and E to EcoRI sites. The EcoRV clone that detects the 3 kb and 9.4 kb EcoRI bands referred to in the text is denoted by EV. The 4.8 kb EcoRI- BamHI band deleted in Figure 7 is also depicted. The 6 kb deleted region containing exons 5 and 6 and the 23 kb BamHl fragment resulting from this deletion are both shown. The location of PCR primers is depicted by arrows.

genomic structure of the NAIP locus is depicted in Figure

3. The NAIP cDNA sequence is shown in Figure 4 and the

intron/exon splice sequences in Figure 5. The NAIP gene

contains at least 16 exons comprising a minimum of 5.5

kb spanning approximately 70 kb of genomic DNA. Intron/

exon boundaries were determined both by identification

of trapped exons and by comparison of cDNA sequence

with that of clones from the PAC 125D9 clone libraries.

The initiating NAIP methionine was predicted to be local-

ized to exon 5, 30 bp downstream of an in-frame stop

codon. The NAIP gene-coding region spans 3696 nt and

encodes a 1232 amino acid and 140 kDa protein. The

nucleic and predicted amino acid sequences of NAIP were

used in searches of the DNA and protein data bases using

the BLAST network service (Altschul et al., 1990). No nu-

cleic acid sequence similarity was detected with any pre-

viously characterized gene; however, significant amino

acid sequence similarity was observed with two lAPs,

CplAP (P = 3e-28; 33% identity over 189 amino acids)

and OplAP (P = 2.5e-27; 33% identity over 180 amino

acids) encoded by two baculoviruses, Cydia pomonella

granulosis virus (Crook et al., 1993) and Orgyia pseudotsu-

gata nuclear polyhydrosis virus (Birnbaum et al., 1994),

respectively (Figure 4).

Both lAPs contain in their N-termini an 80 amino acid

baculovirus lAP repeat motif that, after an intervening se-

quence of approximately 30 residues, is duplicated with

33% identity (Clem and Miller, 1993). The same phenome-

non is observed in NAIP; amino acids 185-250 encoded

by exons 5, 6, and 7 are 35% homologous to amino acids

300-370 encoded in exon 9, 10, and 11. The greatest

homology is observed over a 53 amino acid region with

29 identical amino acids.

In addition to the N-terminal lAP domain, there exists

cysteine- and histidine-rich zinc finger-like motifs in the

C-terminus of both CplAP and OplAP. These motifs, which

are proposed to interact with DNA (Birnbaum et al., 1994),

are not seen in NAIP (Figure 4).

Searches of protein domain programs generated the

following results: residues 1-91, an N-terminal domain

with no recognizable motifs; residues 92-110, a hydropho-

bic domain predicted by the MEMSAT program (Jones et

al., 1994) to be a membrane-spanning domain; residues

163-477, a domain that shows homology with baculoviral

lAPs followed by (and immediately upstream of the next

hydrophobic domain of) a GTP/ATP-binding site; residues

479-496, hydrophobic domain predicted by MEMSAT to

be a membrane-spanning domain; residues 497-1232, a

possible receptor domain containing four N-linked glyco-

sylation sites and a prokaryotic lipid attachment site. We

know of at least three exons that comprise 400 bp of the

5' untranslated region (5'UTR); it is possible that more ex-

ist. A striking feature of this region is the presence of a

near-perfect duplication of a 90 bp region (Figure 4). In

addition, the 3'UTR comprising exon 17 has been found

to contain a 550 bp interval that has potential coding region

detected by the GRAIL program with high homology (P =

1.1e-37) to the chicken integral membrane protein oc-

cludin (Furuse et al., 1993). It is noteworthy that this exon

is seen only in deleted forms of the NAIP gene (Figure 2);

thus, the occludin tract does not appear to be a constituent

of the intact NAIP gene. We also believe it is unlikely to


.... ==================================================================================================== 4 5

! O~L~P

ONAP . . . . . . . . . . . .

16 17 ~ 42oo

ocoludln ~ ~o

Figure 4. Complete cDNA Sequence and Pre- dicted Amino Acid Sequence of NAIP

Guanine is represented by the lowercase g to aid in its distinction from uppercase Cs. Exon boundaries are as marked. Arrows underline the perfect 90 nt tandem repeat in the 5'UTR region. The deleted exons 5 and 6 are stippled. The regions of intraprotein repeated amino acid homology in the lAP domain are under- lined. Sequence comparison with baculoviral lAPs is shown; identical residues are darkly stippled, and similar residues are lightly stippled. CplAP and OplAP refer to the lAPs encoded by the baculoviruses Cydia pomonella granulosis virus and Orgyia pseudotsugata nuclear polyhydrosis virus, respectively. The sequence comparison has been extended to the cysteine/histidine putative DNA-interacting region of the baculoviral lAPs; no NAIP homology can be seen. The region showing significant similarity to chicken occludin, which we have found solely in deleted versions of the NAIP

gene, is shown in the 3'UTR.

E x o n # [

0

1

2

3

4

5

61 7

8

9

l0

11

]2

13

14

15

3'SpHc¢ ~nction I 5'Sp~ccluncdon

atgaag/ACAAAA TCATCT/gtatcc

tgcgag/ACGACG GCAAAG/gtgagt ttttag/TTCCTT CCTCAG/gtaagt tttca~/AAAATG TTACAG/stacct atatag/GTAAAC

ttacag/ATGTGA AT]~-ACG/gtaatg cctcag/GGAGAA CATCAG/gtaaaa ttccag/CTTATT ACACAG/gtgagt ttttag/GTATAA TCCCAA/gt~agt acttag/TTGTCC TTACTG/gtaatg

tcaaag/GAAACC TGCCAG/gtaaga ccacag/AAATGG CACAAG/gtatac

ttccag/ACCAAA AACTAG/gtaag~ tcatag/TTGCCA AATTTG/~tatgt ctgca~/CCTACA ~_AATTG/gtgagc

Figure 5. Intron/Exon Splice Sequences of the NAIP Gene

Splice junctions of exons of NAIP mapping to PAC 125D9.

be a chimeric transcript as occludin homologous se-

quence has been detected in PAC 30B2, which contains

a deleted version of the NAIP gene. The possibility of the

occludin sequence representing a coding exon of the NAIP

gene with the putative 3 'UTR actually being heteronuclear

RNA is also unlikely given the consistency with which the

3 'UTR is observed and the presence of in-frame transla-

tional stop codons mapping upstream of the region of oc-

cludin homology. Preliminary reverse transcriptase-poly-

merase chain reaction ( R T - P C R ) analysis indicates that

the occludin tract is transcribed.

T issue Expression

Hybridization of a Northern blot containing adult tissue

m R N A with an exon 13 probe detected bands only in adult

liver (approximately 6 and 7 kb bands) and placenta (7

kb) (Figure 6). Although the level of expression in adult

C N S is not sufficient to result in visible bands on Northern

blot analysis, successful R T - P C R amplification of the

NA/P transcript using spinal cord as well as fibroblast and

lymphoblast RNA suggests transcriptional activity in these

tissues.

Detect ion of Truncated and Internally Deleted

Vers ions of the NAIP Gene

PAC analysis by hybridization of Southern blots containing

PAC DNA with NAIP exon probes and PCR sequence-

tagged site (STS) content assessment revealed that PAC

238D12 contained an NAIP locus with exons 1 -6 deleted

and PAC 30B2 contained an NAIP gene with exons 5, 6,

and 11 -14 deleted. The presence of identically sized bands

in both genomic and PAC DNAs on Southern blot analysis

Cell 172

. ;

k b ~

9.5,-=. ~ "

7.5--',-

4.4--~

~ 6 . 6

~ 4 8

-.r,--4.2

~ 3 . 8

-,=--23

~ 1 4

*--9.6

2.4--~-

1.35"-~

Figure 6. Northern Blot of Adult Tissues Probed with Exon 13 of the NAIP Locus

Tissues are as marked. The filter was washed at 50°C in 0.1 x SSC and exposed for 24 hr. Bands can be seen in liver and placenta in the 6-7 kb range.

as well as PCR results outlined below indicated that the

deletions were real and not cloning artifacts. Probing of

blots containing BamH I-restricted genomic DNA with NAIP

exons 2-10 not only reveals a single band comprised of

equally sized contiguous 14.5 kb BamHI fragments from

the intact NAIP locus (Figure 3) but also two additional

bands at 9.6 and 23 kb (Figure 7), fragments that are seen

in PACs 238D12 and 30B2/25017, respectively. The 9.6

fragment has been subcloned from a cosmid and found

to contain exons 7-10, with truncation occurring just up-

stream of the seventh exon (Figure 3). PCR primers 1882

and 2050 amplify a 2.6 kb fragment containing the trunca-

tion site from both PAC 238D12 and genomic DNA. The 23

kb BamHI band is generated by a 6 kb deletion removing a

BamHI site leading to the replacement of the two contigu-

ous 14.5 kb BamHI fragments with a 23 BamHI fragment

containing exons 1-4 and 7-10 and lacking exons 5 and

6, as depicted in Figure 3. PCR employing primers 1927

and 1933, constructed to amplify a 4.2 kb junction frag-

ment spanning the 6 kb deletion (Figure 3), generated the

appropriate product as shown by size and sequencing in

both genomic DNA and PACs 30B2 and 25017. The vari-

able dosage of both the 9.6 and 23 kb bands seen in geno-

mic DNA from different individuals indicates that the two

deleted versions of the NAIP gene are present in multiple

and polymorphic number in the general population.

A further level of complexity was detected with the identi-

fication of clones from a non-SMA human fetal brain cDNA

library deleted for exons 11 and 12, some of which also

had exons 15 and 16 absent (Figure 2). The fact that these

deletions result in frameshifts and premature protein trun-

cation indicates that they are, rather than normal splicing

variants, more likely the result of transcription of the de-

leted and truncated version of NAIP that are present in

the general population (Figure 3), consistent with amplifi-

cation of a 1.8 kb fragment missing exons 11 and 12 from

genomic DNA using primers 1285 and 2048. In all, a profile

Figure 7. Pedigree and Southern Blot Analysis of Consanguinous French-Canadian Type III SMA Families

(Top) Probing of a filter containing BamHI-EcoRI-digested genomic DNA with a cDNA probe encompassing exons 2-9 of NAIP reveals the loss of the 4.8 kb fragment that contains exons 5 and 6 in all affected individuals, resulting in an in-frame deletion. All others, save for the homozygously normal sister and brother, show half dosage for this band. (Bottom) A BamHI digest of the same family. In affected individuals, two superimposed 14.5 kb contiguous fragments have sustained the 6 kb deletion of sequence containing a BamHI site, resulting in the generation of a 23 kb band (see Figure 3). Note the existence of the 23 kb BamHI band in all individuals in the pedigree in keeping with its general dispersion in the population. Similarly, the 9.6 kb BamHI band representing the deletion of exons 1-6 that is contained in PAC 238D12 and depicted in Figure 3 can be seen in all individuals, including non- SMA carriers.

of a region containing a variable number of copies of inter-

nally deleted and truncated versions of the NAIP locus,

some of which are transcribed, has emerged from our

analysis.

Probings of blots containing DNA from the somatic cell

hybrid HHW1064 (Gilliam et al., 1989) with NAIP exonic

probes indicate that all forms of the NAIP gene are con-

fined to the 30 Mb deleted region of 5ql 1-13.3 contained

in the derivative chromosome 5 of this cell line. This finding

has been confirmed by fluorescence in situ hybridization

probings with a genomic 2.1 kb BamHI clone containing

NAIP exons 2 and 3 as well as with PAC 125D9 (unpub-

lished data).

NAIP Gene Mutational Analysis

Probing of genomic Southern blots with PCR-amplified

NAIP exons 2-9 revealed the absence of a 4.8 kb EcoRI-

BamHI fragment containing exons 5 and 6 in the four af-

fected individuals of consanguinous type III SMA family

24561 (Figures 3 and 7). The same probing of BamHI-

digested DNA from this family revealed the absence of a

14.5 kb band, also in keeping with a loss of exons 5 and

6 as outlined above (Figures 3 and 7). Similar results were

observed in two other consanguinous French-Canadian

SMA families.

To confirm the proposed deletion of exons 5 and 6, we

made primers homologous to these exons (Figure 3). PCR


Family 24561

0 I

nmO

spinal cord lymphoblast

A n n al a2.a3a4 n asa6a-7

435 bp

241 bp

spinal cord lymph. 3[

I~ n n na7a la2a3 n a t

Family 21470

43S bp

241 bp

sp. cord lym.fib.

r] £t ;9,2. a3 a9 a~o alI

Figure 8. Results of PCR Amplification in Type Ill Families 21470 and 24561 Using Primers 1864 and 1863, Which Amplify Exon 5

The reactions were multiplexed with exon 13 primers 1258 and 1343 to rule out PCR failure obscuring the results. Failure of amplification in keeping with the homozygous absence of exon 5 can be seen to cosegregate with the disease phenotype.

amplif ication of exon 5 using DNA from family 24561 and

a second type III SMA consanguinous family revealed co-

segregation of a failure of amplif ication with the SMA phe-

notype (Figure 8). PCR analysis was then extended to 110

SMA families, employing exon 5 and 6 primers. For these

exons, 17 of 38 (45°/o) type I SMA individuals and 13 of 72

(18%) type II and Ill SMA individuals were homozygously

deleted. Assuming random assortment of chromosomes

and therefore taking the square root of the observed fre-

quency of homozygous exon 5 and 6-deleted individuals

yields est imated frequencies for exon 5 and 6-deleted

chromosomes of 67% in type I SMA and 42% in type II

and III SMAs.

To determine whether the exon 5 and 6 NAIP gene dele-

tion was an SMA mutation, we conducted Southern blot

analysis. An 800 bp EcoRV single copy probe that mapped

immediately 3' to the 6 kb exon 5 and 6 deletion was em-

ployed (Figure 3). Hybridization of this marker to EcoRI

Southern blots detected both a 9.4 kb EcoRI fragment

containing exons 5 and 6 from the intact NAIP locus as

well as a 3 kb EcoRI band from the exon 5 and 6-deleted

copy of the NAIP gene. Analysis was conducted on EcoRI

Southern blots containing DNA from over 900 unrelated

members of myotonic dystrophy, autosomal dominant

polycystic k idney disease, and cystic fibrosis families ob-

tained from our DNA diagnostic laboratory. The 9.4 kb

band was seen in all individuals in keeping with the pres-

ence of at least one copy of intact NAIP in each of the

approximately 900 individuals tested. In addition, the 3 kb

band was observed in every individual, reflecting a virtu-

ally complete dispersion of some form of the exon 5 and

D n n a9 a 2 a l a 3 a 4 a 8 a 7 nn G neg

Figure 9. R T - P C R Ampl i f icat ion of RNA on T issues from SMA and

Non-SMA Indiv iduals

The letter n refers to RNA from a non-SMA individual and the letter

a to RNA from an SMA individual. The tissue source is shown above each panel. Lym refers to lymphoblast and fib to fibroblast. All samples were from type I SMA patients with the exception of a5, which is from an affected member of the consanguinous type III SMA family 24561 shown in Figure 8. RNA was reverse transcribed from exon 13. Primary PCR of products shown in (A) and (B) was with exon 1 primer 1884 and exon 13 primer 1285 or 1974 and those in (C) with exon 6 primer 1919 and exon 13 primer 1285. Secondary PCRs for (A) used exon 4 primer 1886 and exon 13 primer 1974; for (B), exon 5 primer 1864

and exon 11 primer 1979; and for (C), exon 9 primer 1844 and exon 13 primer 1974.

(A) Failure or amplification of reduced products from spinal cord and lymphoblast tissue from individuals a2, a3, a4, a5, a6, and a7. (B) Amplification of reduced size bands in a2 and a3; in a7, a larger product is shown in keeping with an insertion.

(C) Reduced band size in keeping with deletions of exons 11 and 12 in a2, a3, a9, and a11.

(D) RT-PCR amplification of .~-actin RNA from SMA spinal cord mRNA

utilizing random hexanucleotides and oligo(dT) primers as described in Experimental Procedures.

6-deleted NAIP gene in the general population. Moreover,

the variable band dosage observed for the 3 kb band sug-

gested that the number of copies of the exon 5 and 6-deleted

NAIP gene is polymorphic, possibly ranging as high as

four or five copies per genome.

Next, PCR analysis conducted on 168 parents of SMA

Cell 174

children revealed failure of amplification, suggesting ho-

mozygous deletion of exons 5 and 6 in three individuals.

This finding was confirmed by Southern blot analysis in

the two cases with sufficient DNA for this assay. The two

individuals, aged 28 and 35 and both parents of type I

SMA children, when interviewed by telephone described

themselves to be physically well, reporting no symptoms

suggestive of SMA. It was thus concluded that the deletion

of NAIP exons 5-6 in isolation, while possibly reflecting

more extensive deletions in individuals with SMA as out-

lined below, can be clinically innocuous, associated either

with an exceedingly mild SMA or even normal phenotype.

Clinical assessment of these individuals is currently being

undertaken.

RT-PCR Amplification of RNA from SMA

and Non-SMA Tissue

The results of RT-PCR amplification using RNA from both

non-SMA and SMA individuals as template are shown in

Figure 9. We have established that at least some of the

internally deleted and truncated NAIP versions are tran-

scribed. To isolate transcripts primarily from the intact

NAIP gene that would produce a functional protein and

not those from deleted NA/P, we made an effort to RT-

PCR amplify transcripts that were as large as possible.

Given that exon 13 was 2.2 kb, the largest transcript possi-

ble was found to be one that encompassed exon 2 and

the 5' end of exon 13.

A representative subset of RT-PCR experiments is

shown in Figure 9. PCR amplification of reverse-transcribed

product using RNA from non-SMA tissues as template and

reverse transcribing from exons 10 or 13 consistently am-

plified product of the expected size. In contrast, similar

RT-PCR experiments on RNA from SMA tissue revealed

no amplification in five cases in keeping with the marked-

down regulation or complete absence of the intact tran-

script in such individuals (Figure 9A). The RNA obtained

from the SMA tissues was no more than 16 hr postmortem.

As we had no difficulty in amplifying intact NAIP transcript

from normal tissue that is 24 hr postmortem, we do not

believe the difficulty in amplification arises from RNA deg-

radation. The amplification of I~-actin RNA from the same

SMA tissues is consistent with this interpretation (Figure

9D). In the cases in which amplification was observed,

sequencing of RT-PCR products revealed the following

findings: in-frame deletion of codons 153-190 from the 3'

end of exon 5 in sample a9; deletion of exon 6 resulting

in a frameshift with a stop codon occurring 73 nt into exon

7 in a product amplified by exon 5 primer 1864 and exon

8 primer 1843 in sample a2; an approximate 50 nt insertion

in a product amplified by exon 4 primer 1886 and exon

13 primer 1974 in sample a7, which is currently being

sequenced; deletion of glutamic acid codon number 158

in exon 5 in association with deletion of exons 11 and 12

in a product amplified by exon 5 primer 1864 and exon 8

primer 1843 in sample a3; deletion of exons 11 and 12

introducing a frameshift and a stop codon 14 nt into exon

13 in a product amplified by exon 9 primer 1844 and exon

13 primer 1974 in samples a2, a3, a9, and a11. In all,

employing PCR on material reverse transcribed from exon

13, we have observed successful amplification of the ap-

propriate product from all 18 non-SMA tissues attempted

and in only 3 of 12 SMA tissues. In the latter cases, amplifi-

cation was from exons 4 to 13 only; whether the transcript

also incorporates exons 2-3 or 14-17 is unknown. We

believe that these data are consistent with defects in NAIP

either resulting in or contributing to the SMA phenotype.

Discussion

Relation of the NAIP Gene to SMA

We present characterization of the gene for a novel pro-

tein, NAIP. A variable number of copies of intact and par-

tially deleted forms of the NAIP gene have been found to

map to 5q 13.1. In contrast with most autosomal recessive

diseases, in which causal mutations are usually detected

in the single copy of a given gene, we propose that an

SMA chromosome is characterized by a paucity or, for

severe SMA mutations, an absence of the intact NAIP

gene, possibly resulting from unequal crossing over.

This model must accommodate the apparently unaffected

status of the three parents of individuals with SMA who

do not have a copy of exons 5 and 6 in their genome.

Judging both by the cDNA clones identified both in normal

fetal brain libraries as well as the make up of RT-PCR

NAIP products (Figure 2), many and possibly all truncated/

deleted copies of the NAIP gene appear to be transcribed.

We believe that there is a good possibility that the exon

5 and 6-deleted version of NAIP is also translated. In keep-

ing with this model, removal of exons 5 and 6 results in

an in-frame deletion that extends the longest NAIP open

reading frame upstream to a start methionine in exon 2

at nucleotide 211 (Figure 4). Furthermore, the protein se-

quence encoded by the deleted exon 5-6 lAP motif is

approximately 35% homologous to the lAP motif encoded

in exons 10 and 11; the presence of the latter motif may

account for the absence of discernible phenotype in the

three exon 5 and 6-deleted individuals. One model is that

a single copy of exon 5 and 6-deleted NAIP on each chro-

mosome results in the mild SMA phenotype, while individ-

uals with greater than three or four copies of the exon

5 and 6-deleted NAIP locus are clinically unaffected. A

duplication of the SMA gene underlying the disease has

recently been proposed by DiDonato et al. (1994).

The delineation of an SMA genotype in a given individual

is complicated by the unusual amplification of the NAIP

gene in the 5q13.1 region. Probings of Southern blots con-

taining genomic DNA with NAIP exon probes invariably

reveal bands resulting from copies of internally deleted

and truncated versions of the NAIP gene. The presence

of variable numbers of the different forms of the NAIP loci

in the general population is therefore the norm and not

diagnostic of an SMA mutation per se, complicating the

mutational analysis of the NAIP gene. If the detection of

genomic DNA containing altered NAIP loci affords no proof

of an SMA chromosome, then, by default, the search must

be for the absence of the normal NAIP gene. However,

given the individuals with no copies of exons 5-6 in their

genome who are clinically unaffected, the identification of


an SMA chromosome is contingent on the absence of both

the intact as well as forms of NAIP in which only exons

5-6 are deleted. Assaying for their absence is complicated

by the presence of segments of normal NAIP gene in each

of the other (more extensively deleted) forms of the NAIP

locus. We believe that many and perhaps most of the nu-

merous exon 5 and 6-deleted SMA individuals we have

observed contain neither the intact NAIP loci nor version

in which exons 5-6 are deleted, but rather some other

combination of more extensively truncated/deleted ver-

sions of the locus with resultant absence of intact NAIP

translation. Support for this interpretation comes from the

difficulty encountered in amplifying normal NAIP tran-

scripts employing RT-PCR on RNA from type I SMA tissue

and preliminary long-range Southern blot analysis using

an SMA genomic DNA restricted with rare cutting endonu-

cleases (unpublished data).

The presence of a variable number of copies of trun-

cated and internally deleted versions of the NAIP gene is

similar to the situation seen with the chromosome 6 CYP21

gene that encodes steroid 21-hydroxylase (Wedell and

Luthman, 1993; Collier et al., 1993). CYP21, which when

mutated causes an autosomal recessive 21-hydroxylase

deficiency, has been observed in 0-3 copies in different

individuals. There also exists in the region a variable num-

ber of inactive pseudogene copies of CYP21 known collec-

tively as CYP21P. The majority of the CYP21 mutations

observed in 21-hydroxylase deficiency can also be found

in some form of CYP21P, and it is thought that the pseu-

dogenes act as a source of the mutations observed in

CYP21. The truncated and internally deleted NAIP genes

might be viewed as analogous to CYP21P and, as has been

postulated for CYP21/CYP21P, unequal crossing over may

result in chromosomes deleted for forms of the NAIP gene

that encode functional protein. We believe that the exis-

tence of a polymorphic number of mutated NAIP genes

on 5q13.1 suggests a credible mechanism for generation

of SMA chromosomes in this fashion.

In all, the evidence in support of mutations in or the

absence of the NAIP gene playing a role in the genesis

of SMA includes the following. First, there is the strong

possibility that NAIP, given its homology with baculoviral

lAPs, functions as an inhibitor of apoptosis. This charac-

teristic is wholly compatible with the pathology of SMA. It

is noteworthy that mutations in a regulator of apoptosis

have been previously suggested as a cause of SMA (Op-

penheim, 1991 ; Sarnat, 1992). Second, there are the map-

ping of the NAIP locus within the recombination-defined

critical SMA interval and the fact that polymorphic markers

have been shown to be in strong linkage disequilibrium

with type I SMA; C272 (Melki et al., 1994) and Agl (DiDo-

nato et al., 1994) both map to PAC 125D9 just distal to

the NAIP locus (Figure 1C). Third, there is the nature of

linkage disequilibrium observed between the type I SMA

phenotype and the 5q13.1 markers. We have shown that

the CATT-40G1 CTR sublocus, which is frequently dupli-

cated on non-SMA chromosomes (Roy et al., 1995), is

deleted in 80% of type I SMA chromosomes compared

with 45% of non-SMA chromosomes (McLean et at.,

1994). This finding is in keeping with a depletion of the

number of NAIP genes on SMA chromosomes. In a similar

fashion, Melki et al. (1994) have observed a heterozygote

deficiency consisting of a reduced number of bands for

the C272 CTR in type I SMA, reflecting, they propose,

chromosomal deletions. DiDonato et al. (1994) have also

seen a striking reduction in the number ofAgl CTR subloci

in type I SMA individuals when compared with non-SMA

individuals. We believe that the observation by three

groups of the depletion of these markers on type I SMA

chromosomes fits well with the proposed model of a lack

or absence of both the intact and exon 5 and 6--deleted

form of the NAIP gene underlying the disease. Fourth,

there is the markedly increased frequency of NAIP exon

5-6 deletions observed in SMA chromosomes (approxi-

mately 67% of type I SMA chromosomes and 42% of type

II/111 SMA chromosomes) compared with that detected for

non-SMA chromosomes (2%-3°/0). As outlined above, we

believe that this phenomenon reflects the rarity or absence

of both the intact NAIP gene as well as the NAIP version

with only exons 5-6 deleted in the SMA chromomsomes,

leaving only the more significantly internally deleted and

truncated forms of the NAIP gene present. Fifth, there is

our consistent inability to RT-PCR amplify appropriate

size transcripts from RNA obtained from 9 of 12 SMA indi-

viduals despite success with 18 of 18 RNAs from non-SMA

individuals. Furthermore, sequencing of those RT-PCR

products that could be obtained from type I SMA material

revealed a variety of mutations and deletions.

Baculoviral lAPs

NAIP shows significant homology with the two baculoviral

gene products, CplAP and OplAP, that are capable of

inhibiting insect cell apoptosis. Insect cell apoptosis fol-

lowing baculoviral infection has been well documented

and is postulated to be a defense mechanism. Premature

death of infected insect cells results in an attenuation of

viral replication (Clem and Miller, 1993, 1994a). CplAP

and OplAP are thought to represent baculoviral responses

to this apoptotic mechanism. Both act independently of

other viral proteins to inhibit host insect cell apoptosis,

thereby permitting increased viral proliferation (Clem and

Miller, 1994a, 1994b). They are known to be strongly simi-

lar only to each other; no sequence similarities with cross

phyla proteins have been reported. Their mode of action

is unknown, although some interaction with DNA has been

postulated.

Relation of the NAIP Gene to SMN

Given the comparatively large regions of genomic DNA

spanned by the various forms of NA/P, it is conceivable

that neighboring genes from the region might also be in-

volved with the depletion of NAIP from a chromosome.

Consistent with this hypothesis, a second gene, called

SMN (for survival motor neurons), has also been reported

as being mutated in SMA (Lefebvre et al., 1995 [this issue

of Ceil]). There are two copies of the gene BCD541, a

centromeric version called CBCD541 and the telomeric

TBCD541 or SMN. Mutations in exon 7 and 8 of the telo-

meric SMN locus, detected by single strand conformation

polymorphism, are observed in the significant majority of

Cell 176

SMA individuals regardless of the severity of the pheno-

type. It is possible that both the NAIP and SMN proteins are

involved in the pathogenesis of SMA with SMN mutations

underlying the majority of SMA mutations and a superim-

posed NAIP deficiency accounting for the more extensive

SMA mutations, thereby resolving the absence of appar-

ent genotype-phenotype correlation seen with SMN mu-

tations. If both genes are found to be involved in the

pathogenesis of SMA, the defined and restricted SMA

phenotype with its underlying cellular specificity suggests

a mutual interaction of the gene products. One possibilty

is that NAIP-SMN forms a heterodimer. We have mapped

the intact NAIP locus into the interval defined by the two

copies of SMN approximately 60 kb proximal to the telo-

meric SMN locus (Figure 1). A number of examples of

contiguous loci encoding distinct subunits for heterodim-

ers, including murine nerve growth factor (Evans and Rich-

ards, 1985) and human y-aminobutyric acid receptors (Glatt

et al., 1994), have been reported. Studies of the gene

products for such an interaction will be helpful in this

regard.

In conclusion, we have identified a region of 5q13.1 that

contains, in addit ion to the complete NAIP gene, a variable

number of copies of internally deleted and truncated forms

of the gene. We believe that a lack or absence of both

the intact NAIP gene and the NAIP locus with exons 5 and

6 deleted from the genome of a given individual is likely

to cause or (in concert with SMN mutations) contribute to

the SMA phenotype. Analysis of NAIP at the protein level

in SMA tissues will be helpful in clarifying its status in the

pathogenesis of SMA and, possibly, aid in the formulation

of conventional and genetic therapies for these debil i tating

conditions. Furthermore, the identification of genes show-

ing homology with the NAIP locus and proteins that interact

with NAIP may help in the continuing elucidation of apop-

totic mechanisms in mammalian cells.

Experimental Procedures

Family Material Clinical diagnoses were conducted as described in MacKenzie et al. (1993) with all patients fulfilling the diagnostic criteria given therein. DNA was isolated from peripheral leukocytes as described previously

(MacKenzie et al., 1993).

Genetic and Linkage Disequilibrium Analyses Genotyping with microsatellite markers was as outlined in MacKenzie et al. (1993) and McLean et al. (1994). The following 5q13.1 loci were used as described: D5S112 (Brzustowicz et al., 1990); D5S351 (Hud- son et al., 1992); D5S435 (Soares et al., 1993); D5S557 (Francis et al., 1993); D5S629 and D5S637 (Clermont et al., 1994); D5S684 (Brahe et al., 1994); Y98T, Y97-r, Y116T, Y122T, and CMS1 (Kleyn et al., 1993); CATT (Burghes et al., 1994b; McLean et al., 1994); and MAP1B (Lien et al., 1991).

Linkage disequilibrium analyses were conducted using parameters

that can accommodate the multiple alleles seen with microsatellite repeats. Given the complexities inherent in disequilibrium analyses, a total of four different parameters for which multiple alleles may be used were employed. These were Dij, Dij', and D', as defined in Hedrick (1987), and the ~2 test. Two of these, Dij and Dij', have given the best a posteriori positional information in a previous study on myotonic dystrophy (Podolsky et al., 1994). The patient and control populations are as outlined in McLean et al. (1994).

Cosmid, YAC, and PAC Arraying Cosmid and YAC contig assembly was as outlined in Roy et al. (1995). PACs were constructed as outlined in Ioannou et al. (1994). Three PAC libraries have been constructed using these procedures, with a

combined total of 175,000 clones, and propagated as individual clones in microtiter dishes (P. I. et al., unpublished data). Pools derived from the three libraries (designated LLN L PAC1, RPCI1, and RPCI2) were screened with 5q 13.1 STSs. Positive PACs were arranged into contiguous and overlapping arrays by further analysis with additional STSs combined with probings of Southern blots containing PAC DNA by single copy genomic DNA and cDNA probes.

DNA Manipulation and Analysis

Four genomic libraries containing PAC 125D9 insert were constructed by BamHI, BamHI-Notl, and total and partial Sau3AI (selected for 5 kb insert size) digestions of the PAC genomic DNA insert and subcloned into Bluescript vector. Sequencing of approximately 400 bp of both termini of 200 clones 5 kb in length from the partial Sau3AI digestion library as well as those of the BamHI and BamHI-EcoRI libraries in the manner of Chen et al. (1993) was undertaken.

Coding sequences from the PACs were isolated by the exon amplification procedure as described by Church et al. (1994). PACs were digested with BamHI or with BamHI and Bglll and subcloned into pSPL3. Pooled clones of each PAC were transfected into COS1 cells. After a 24 hr transfection, total RNA was extracted. Exons were cloned into pAMPt 0 (GIBCO BRL) and sequenced utilizing primer SD2 (GTG AAC TGC ACT GTG ACA AGC TGC).

DNA sequencing was conducted on an ABI 373A automated DNA sequencer. Two commercial human fetal brain cDNA libraries in Xgt (Clontech) and XZAP (Stratagene) were used for candidate transcript isolation. The Northern blot was commercially acquired (Ciontech), and probing was performed using standard methodology.

In general, primers used in the paper for PCR were selected for Tins of 60°C and can be used with the following conditions: 30 cycles of 94°C for 60 s, 60°C for 60 s, and 72°C for 90 s. PCR primer map- pings are as referred to in the figure legends and text. Primer sequences are as follows: 1258, ATg CTT ggA TCT CTA gAA Tgg; 1285, AgC AAA gAC ATg Tgg Cgg AA; 1343, CCA gCT CCT AgA gAA AgA Agg A; 1844, gAA CTA Cgg CTg gAC TCT TTT; 1857, CAT TTg gCA TgT TCC TTC CAA g; 1863, CTC TCA gCC TgC TCT TCA gAT; 1864, AAA gCC TCT gAC gAg Agg ATC; 1882, CTg AgT CAg ACA CTT ACA ggT AA; 1884, CgA CTg CCT gTT CAT CTA CgA; 1886, TTT gTT CTC CAg CCA CAT ACT; 1893, gTA gAT gAA TAC TgA TgT TTC ATA ATT; 1910, TgC CAC TgC CAg gCA ATC TAA; 1919, TAA ACA ggA CAC ggT ACA gTg; 1923, CAT gTT TTA AgT CTC ggT gCT CTg; 1926, TTA gCC AgA TgT gTT ggC ACA Tg; 1927, gAT TCT ATg TgA TAg gCA gCC A; 1933, gCC ACT gCT CCC gAT ggA TTA; 1974, gCT CTC AgC TgC TCA TTC AgA T; 1979, ACA AAg TTC ACC ACg gCT CTg; 2048, ATg gAg gAg TgT CAT ATA CgC ACT; 2050, TgC TCC TCA CTC TTC TAC CIF TTC.

RT-PCR cDNA was synthesized in a 20 p_l reaction utilizing 7 p.g of total RNA. The RNA was denatured for 5 min at 95°C and cooled to 37°C. Reverse transcription was performed at 42°C for 1 hr after addition of 5 I~1 of 5x reverse transcription buffer (GIBCO BRL), 2 p.I of 0.1 M DTT, 4 p.I of 2.5 mM dNTPs, 8 U of RNasin, 25 ng of cDNA primer (1285), and 400 U of MMLV (GIBCO BRL). cDNA (1 p.I) was utilized as template in subsequent 50 p.I PCRs. Of this primary PCR, 1 p] was utilized as template for secondary PCR amplification using nested primers.

In addition, 11 spinal cord RNA samples (5 p.g of total RNA) were reverse transcribed using MMLV reverse transcriptase (GIBCO BRL) and random hexamer plus oligo(dT) primers in 20 p.I reactions. PCR amplification of 661 bp ~-actin cDNA from 1 i~1 aliquots of reverse transcription reactions using Stratagene primers at 65°C annealing temperature from these samples was conducted.

Sequence Analysis Primary DNA sequence data were edited with the TED program (Glee- son and Hillier, 1991). As many of the partially sequenced 200 clones 5 kb in length from the partial Sau3AI digestion library as possible


were arranged into overlapping arrays using the XBAP Staden pack- age (Dear and Staden, 1991). Sequence data were also assembled and analyzed using the GCG sequence analysis (University of Wiscon-

sin Genetics Computer Group). Protein domain homologies were found by searching the PROSITE protein data base (Bairoch and Bucher, 1994). The MEMSAT program was also used to search for

transmembrane domain regions (Jones et al., 1994).

Acknowledgments

Correspondence should be addressed to A. M. This work was funded by the following grants: to A. M. by the Muscular Dystrophy Association of Canada (MDAC), the Muscular Dystrophy Association of the United

States (MDA), the Children's Hospital of Eastern Ontario Research Foundation, and the Medical Research Council of Canada (MRC); to

R. G. K. and J.-E. I. by the Research Development Corporation of Japan and the Networks of Centres of Excellence (Canadian Genetic

Diseases Network); to P. J. d. J. and X. G. by the Cancer Center

Support Grant at Roswell Park Center Institute (CA-16056), by National Institutes Health grant 1R01RG01165-01, and by Department of En- ergy grant DE-FG02-94ER61883. N. R. is supported by an MDAC

fellowship. M. S. M. is an MRC postdoctoral fellow, and R. G. K. is an MRC scientist. A. M. is an Ontario Ministry of Health career scientist. We wish to thank L. Deaven for the generous gift of the chromosome

5 library. We are grateful to K. Stoddard and A. Sukhedo for their productive work on the project. We wish to thank the following people for technical and other assistance: J. Barcel6, J. Earle-MacDonald,

M. Foitzik, M. Higgins, D. Lahey, S. Leblond, N. Lemieux, L. McAuley, G. Mettler, L. Podolsky, E. Separovic, S. Soder, M. J. Somerville, and

J. Wang. We thank Drs. J. Edwards, H. Sarnat, H. Heick, and A. Hunter

for informed commentary. We are grateful to Drs. N. Stirpe (MDA), M. Conneally(MDA), T. Munsat (MDA), A. Emery(European Neuromuscu-

lar Center), and Mr. Y. Poortman (European Neuromuscular Center)

for their efforts in guiding the international SMA consortium as well as to the following members of the consortium for their exchange of

data prior to publication: C. Brahe, A. Burghes, C. Buys, K. Davies, C. Gilliam, L. Kunkel, J. McPherson, J. Melki, B. Russman, F. Samaha,

L. Simard, B. Wirth, and G. van der Steege. Finally, we wish to thank the various Canadian clinicians and particularly the SMA families who have been so supportive and given so freely to our work.

Received November 17, 1994; revised December 20, 1994.

References

Altschui, S. F., Warren, G., Miller, W., Myers, E. W., and Lipman,

D. J. (1990). Basic alignment search tool. J. Mol. Biol. 215, 403-410.

Bairoch, A., and Bucher, P. (1994). PROSITE: recent developments.

Nucl. Acids Res. 22, 3583-3589.

Birnbaum, M. J., Clem, R. J., and Miller, L. K. (1994). An apoptosis-

inhibiting gene from a nuclear polyhedrosis virus encoding a polypep-

tide with Cys-His sequence motifs. J. Virol. 68, 2521-2528.

Brahe, C., Velon~, U., van der Steege, G., Zappata, S., van de Veen, A. Y., Osinga, J., Tops, C. M. J., Fodde, R., Khan, M., Ruys, C. H. C. M., and Neri, G. (1994). Mapping of two new markers within the smallest interval harboring the spinal muscular atrophy locus by

family and radiation hybrid analysis. Hum. Genet. 93, 494-501.

Brooke, M. (1986). A Clinician's View of Neuromuscular Diseases (Bal- timore: Williams and Williams).

Brzustowicz, L. M., Lehner, T., Castilla, L. H., Penchaszadeh, G. K., Wilhelmsen, K. C., Daniels, R., Davies, K. E., Leppert, M., Ziter, F.,

Wood, D., Dubowitz, V., Zerres, K., Hausmanowa-Petrusewicz, I., Ott, J., Munsat, T. L., and Gilliam, T. C. (1990). Genetic mapping of chronic childhood-onset spinal muscular atrophy to chromosome 5ql 1.2-13.3. Nature 344, 540-541.

Brzustowiez, L. M., Kleyn, P. W., Boyce, F. M., Lien, L. L., Monaco,

A. P., Penchaszadeh, G. K., Das, K., Wang, C. H., Munsat, T. L., Ott,

J., Kunkel, L. M., and Gilliam, T. C. (1992). Fine-mapping of the spinal muscular atrophy locus to a region flanked by MAP1B and D5S6. Genomics 13, 991-998.

Burghes, A. H. M., Ingraham, S. E., Kote-Jarai, Z., Rosenfeld, S., Herta, N., Nadkarni, N., DiDonato, C. J., Carpten, J., Hurko, O., Flor-

ence, J., et ai. (1994a). Linkage mapping of the spinal muscular atrophy gene. Hum. Genet. 93, 305-312.

Burghes, A. H. M., Ingraham, S. E., McLean, M., Thompson, T. G.,

McPherson, J. D., Kote-Jarai, Z., Carpten, J. D., DiDonato, C. J., Ikeda, J.-E., Surh, L., Wirth, B, Sargent, C. A., Ferguson-Smith, M. A., Fuerst,

P., Moysis, R. K., Grasdy, D. L., Zerres, K. I., Korneluk, R., MacKenzie, A., and Wasmuth, J. J. (1994b). A multicopy dinucleotide marker that maps close to the spinal muscular atrophy gene. Genomics 21,394-

402.

Chen, E. Y., Schlessinger, D., and Kere, J. (1993). Ordered shotgun sequencing: a strategy for integrated mapping and sequencing of YAC

clones. Genomics 17, 651-656.

Church, D. M., Stotler, C. J., Rutter, J. L., Murrell, J. R., Trofatter,

J. A., and Buckler, A. J. (1994). Isolation of genes from complex sources of mammalian genomic DNA using exon amplification. Nature

Genet. 6, 98-105.

Clem, R. J., and Miller, L. K. (1993). Apoptosis reduces both the in vitro replication and the in vivo infectivity of a baculovirus. J. Virol. 67,

3730-3738.

Clem, R. J., and Miller, L. K. (1994a). Induction and inhibition of

apoptosis by insect viruses. In Apoptosis I1: The Molecular Basis of

Apoptosis in Disease (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 89-110.

Clem, R. J., and Miller, L. K. (1994b). Control of programmed cell death by the baculovirus genes p35 and lAP. Mol. Cell. Biol. 14, 5212- 5222.

Clermont, O., Burlet, P., BLirglen, L., Lefebvre, S., Pascal, F., McPher-

son, J., Wasmuth, J., Cohen, D., Le Paslier, D., Weissenbach, J., Lathrop, M., Munnich, A., and Melki, J. (1994). Use of genetic and

physical mapping to locate the spinal muscular atrophy locus between two new highly polymorphic DNA markers. Am. J. Hum. Genet. 54,

687-694.

Collier, S., Mayada, T., and Strachan, T. (1993). A de novo pathological

point mutation at the 21-hydroxylase locus: implications for gene con- version in the human genome. Nature Genet. 3, 260-264.

Crook, N. E., Clem, R. J., and Miller, L. K. (1993). An apoptosis-

inhibiting baculovirus gene with a zinc finger-like motif. J. Virol. 67, 2168-2174.

Dear, S., and Staden, R. A. (1991). A sequence assembly and editing program for efficient management of large projects. Nucl. Acids Res. 19, 3907-3911.

DiDonato, C. J., Morgan, K., Carpten, J. D., Fuerst, P., Ingraham,

S. E., Prescott, G., McPherson, J. D., Wirth, B., Zerres, K., Hurko,

O., Wasmuth, J. J., Mendell, J. R., Burghes, A. H. M., and Simard, L.

(1994). Association between Agl-CA alleles and severity of autosomal

recessive proximal spinal muscular atrophy. Am. J. Hum. Genet. 55, 1218-1229.

Dubowitz, V. (1978). Muscle Disorders in Childhood (Philadelphia: W. B. Saunders).

Emery, A. E. H. (1991). Population frequencies of inherited neuromuscular diseases: a world survey. Neuromusc. Disord. 1, 19-29.

Evans, B. A., and R. I. Richards (1985). Genes for the c~ and ~' subunits of mouse nerve growth factor are contiguous. EMBO J. 4, 133-138.

Francis, M. J., Morrison, K. E., Campbell, L., Grewal, P. K., Christo-

doulo, Z., Daniels, R. J., Monaco, A. P., Frichauf, A. M., McPherson,

J., Wasmuth, J., and Davies, K. E. (1993). A contig of non-chimeric YACs containing the spinal muscular atrophy gene in 5q13. Hum. Mol. Genet. 2, 1161-1167.

Furuse, M., Hirase, T., Itch, M., and Nagafuchi, A. (1993). Occludin:

a novel integral membrane protein localizing at tight junctions. J. Cell Biol. 123, 1777-1788.

Gilliam, T. C., Freimer, N. B., Kaufmann, C. A., Powhik, P. P., Bassett,

A. S., Bengtsson, U., and Wasmuth, J. J. (1989). Deletion mapping of DNA markers to a region of chromosome 5 that cosegregates with schizophrenia. Genomics 5, 940-944.

Cell 178

Gilliam, T. C., Brzustowitcz, L. M., Castillo, L. H., Lehner, T., Penchas- zadeh, G. K., Daniels, R. J., Byth, B. C., Knowles, J., Hislop, J. E., Shapira, Y., Dubowitz, V., Munsat, T. L., Ott, J., and Davies, K. E. (1990). Genetic homogeneity between acute and chronic forms of spinal muscular atrophy. Nature 345, 823-825.

Glatt, K., Sinnett, D., and Lalande, M. (1994). The human 7-aminobutyric acid receptor subunit [33 and ct5 gene cluster in'chromosome 15ql 1- q13 is rich in highly polymorphic (CA)n repeats. Genomics 19, 157- 160.

Gleeson, R., and Hillier, L. (1991). A trace display and editing program for data from fluorescence based sequencing machines. Nucl. Acids Res. 19, 6491-6493.

Hedrick, P. W. (1987). Gametic disequilibrium measures: proceed with caution. Genetics 117, 331-341.

Hudson, T. J., Englestein, M., Lee, M. K., Ho, E. C, Rubenfield,

M. J., Adams, C. P., Housman, D. E., and Dracopoli, N. C. (1992). Isolation and chromosomal assignment of 100 highly informative human simple sequence repeat polymorphisms. Genomics 13, 622-629.

Ioannou, P. A., Amemiya, C. T., Games, J., Kroisel, P. M., Shizuya, H., Chen, C., Batzer, M. A., and de Jong, P. J. (1994). A new bacteriophage Pl-derived vector for the propagation of large human DNA fragments.

Nature Genet. 6, 84-89.

Jones, D. T., Taylor, W. R., and Thornton, J. M. (1994). A model recog-

nition approach to the prediction of all-helical membrane protein structure and topology. Biochemistry 33, 3038-3049.

Kleyn, P. W., Wang, C. H., Lien, L. L., Vitale, E., Pan, J., Ross, B. M., Grunn, A, Palmer, D. A., Warburton, D., Brzustowicz, L. M., Kunkel, L M., and Gilliam, T. C. (1993). Construction of a yeast artificial chromosome contig spanning the spinal muscular atrophy disease gene region. Proc. Natl. Acad. Sci. USA 90, 6801-6805.

Lefebvre, S., BLirglen, L., Reboullet, S., Clermont, O., Burlet, P., Viol- let, L., Benichou, B., Cruaud, C., Millasseau, P., Zeviani, M., Le Paslier, D., Fr~zal, J., Cohen, D., Weissenbach, J., Munnich, A., and Melki, J. (1995). Identification and characterization of a spinal muscular atrophy-determining gene. Cell 80, this issue.

Lien, L. L., Boyce, F. M., Kleyn, P., Brzustowicz, L. M., Menninger, J., Ward, D. C., Gilliam, T. C., and Kunkel, L. M. (1991). Mapping of human microtubule associated protein 1B in proximity to the spinal muscular atrophy locus at 5q13.1. Proc. Natl. Acad. Sci. USA 88,

7873-7876.

MacKenzie, A., Roy, N., Besner, A., Mettler, G., Jacob, P., Korneluk, R., and Surh, L. (1993). Genetic linkage analysis of Canadian spinal muscular atrophy kindreds using flanking microsatellite 5q13 polymor-

phisms. Hum. Genet. 90, 501-504.

McLean, M. D., Roy, N., MacKenzie, A. E., Salih, M., Burghes, A., Simard, L., Korneluk, R. G., Ikeda, J.-E., and Surh, L. (1994). Two 5q13 simple tandem repeat loci are in linkage disequilibrium with type I spinal muscular atrophy. Hum. Mol. Genet. 3, 1951-1956.

Melki, J., Abdelhak, S., Sheth, P., Bachelot, M. F., Burlet, P., Marcadet, A., Aicardi, J., Barois, A., Carriere, J. P., Fardeau, M., Fontan, D., Ponsot, G., Billsette, T., Angelini, C., Barbosa, C., Ferriere, G., Lanzi, G., Ottolini, A., Babron, M. C., Cohen, D., Hanauer, A., Colerget-

Darpox, F., Lathrop, M., Munnich, A., and Fr6zal, J. (1990). Gene for chronic proximal spinal muscular atrophies maps to chromosome 5q. Nature 344, 767-768.

Melki, J., Lefebvre, S., B~rglen, L., Burlet, P., Clermont, O., Millas- seau, P., Reboullet, S., Benichou, B., Zevianai, M., LePaslier, D., Co- hen, D., Weissenbach, J., and Munnich, A. (1994). Denovo and inherited deletions of the 5q13 region in spinal muscular atrophies. Science

264, 1474-1477.

Murayama, S., Bouldin, T. W., and Suzuki, K. (1991). Immunocyto- chemical and ultrastructural studies of Werdnig-Hoffmann disease. Acta Neuropathol. 81,408-417.

Oppenheim, R. W. (1991). Cell death during development of the nervous system. Annu. Rev. Neurosci. 14, 453-501.

Peress, N. S., Stermann, A. B., Miller, R., and Kaplan, C. G. (1986). "Chromatolytic" neurons in lateral geniculate body in Werdnig-Hoff- mann disease. Clin. Neuropathol. 5, 69-72.

Podolsky, L., Tsilfidis, C., Baird, S., Korneluk, R. G., and MacKenzie,

A. E. (1994). An empiric comparison of linkage disequilibrium parameters in disease gene localization: the myotonic dystrophy experience. Am. J. Hum. Genet. 55, A161.

Roy, N., McLean, M., Johnston, A., Lefebvre, C., SaUh, M., Yaraghi, Z., Ikeda, J.-E., Korneluk, R. G., and MacKenzie, A. E. (1995). Refined physical map of the spinal muscular atrophy gene region at 5q 13 based on YAC and cosmid contiguous arrays. Genomics, in press.

Sarnat, H. B. (1983). Histopathology of SMA. In Progressive Spinal Muscular Atrophy, I. Gamstorp and H. B. Sarnat, eds. (New York: Raven), pp. 91-110.

Sarnat, H. B. (1992). Cerebral Dysgeneeis: Embryology and Clinical Expression (Oxford: Oxford University Press).

Sheth, P., Abdelhak, S., Bachelot, M. F., Burlet, P., Masset, M., Hil- laire, D., Clerget-Darpoux, F., Fr~zal, J., Lathrop, G. M., Munnich, A., and Melki, J. (1991). Linkage analysis in spinal muscular atrophy, by six closely flanking markers on chromosome 5. Am. J. Hum. Genet. 48, 764-768.

Soares, V. M., Brzustowicz, L. M., Kleyn, P. W., Knowles, J. A., Palmer, D. A., Asokan, S., Penchaszadeh, G. K., Munsat, T. L., and Gilliam, T. C. (1993). Refinement of the spinal muscular atrophy locus to the interval between D5S435 and MAP1B. Genomics 15, 365-371,

Thompson, T. G., Morrison, K. E., Kleyn, P., Bengtsson, U., Gilliam, T. C., Davies, K. E., Wasmuth, J. J., and McPherson, J. D. (1993). High resolution physical map of the region surrounding the spinal muscular atrophy gene. Hum. Mol. Genet. 2, 1169-1176.

Towfighi, J., Young, R. S., and Ward, R. M. (1985). Is Werdnig-Hoff- mann disease a pure lower motor neuron disorder? Acta Neuropathol.

65, 270-280.

Wedell, A., and Luthman, H. (1993). Steroid 21-hydroxylase (P450c21): a new allele and spread of mutations through the pseudogene. Human Genet. 91,236-240.

Wirth, B., Voosen, B., Rohhrig, D., Knapp, M., Piechaczek, B., Rudnik- Schoneborn, S., and Zerres, K. (1993). Fine mapping and narrowing of the genetic interval of the spinal muscular atrophy region by linkage studies. Genomics 15, 113-118.

Wirth, B., Pick, E., Leuter, A., Dadze, A., Voosen, B., Knapp, M., Piechaczak-Wappenschmidt, B., Rudnik-Schoneborn, S., Schonling, J., Cox, S., Spurr, N. K., and Zerres, K. (1994). Large linkage analysis in 100 families with autosomal recessive spinal muscular atropy (SMA)

and 11 CEPH families using 13 polymorphic loci in the region 5ql 1.2-

q13.3. Genomics 20, 84-93.

GenBank Accession Number

The accession number for the sequence reported in this paper is

U 19251,

Documents

The gene for neuronal apoptosis inhibitory protein is partially deleted in individuals with spinal muscular atrophy