Molecular bases for human complementC7 polymorphisms, C7*3 and C7*4

Takahiko Horiuchi,a,* Hiroaki Nishimukai,b Tatsuyuki Okiura,b Koji Nishimura,b

Hiroaki Nishizaka,a Takeshi Kojima,a Hiroshi Tsukamoto,a

Kenshi Hayashi,c and Mine Haradaa

a Medicine and Biosystemic Science, Kyushu University Graduate School of Medical Sciences, Fukuoka 812-8582, Japanb Department of Legal Medicine, Ehime University School of Medicine, Shigenobu 791-0295, Japan

c Medical Institute of Bioregulation, Kyushu University, Fukuoka 812-8582, Japan

Received 5 September 2002

Abstract

Complement C7 is one of the components of membrane attack complex (MAC) generated by the terminal complement cascade.

C7 protein is polymorphic and most of its polymorphisms have been identified using isoelectric focusing (IEF), which detects protein

charge differences. To date, the molecular bases of the polymorphisms detected by IEF have not been determined. In this paper, we

describe the structural bases of two C7 IEF-detected polymorphisms, C7*3 and C7*4, both of which are common in Asian pop-

ulations. C7*3 resulted from substitution of cysteine (Cys) at amino acid residue 106 by charged arginine (Arg; C106R), while

charged lysine (Lys) at amino acid residue 398 was replaced by neutral glutamine (Gln; K398Q) in C7*4. As C7*3 is hypomorphic, it

is important to study its possible associations with diseases such as immunological disorders and infections. We present genetic bases

for this C7 polymorphism, which we determined using polymerase chain reaction (PCR)-based genotyping, a simple and accurate

method suitable for large-scale studies.

� 2002 Elsevier Science (USA). All rights reserved.

Keywords: Complement C7; Polymorphism; Polymerase chain reaction/single-strand conformation polymorphism (PCR/SSCP)

Complement C7 is one of the five complement pro-

teins that sequentially form a large protein–protein

complex called membrane attack complex (MAC). The

membranolytic activity of MAC is critical for host de-

fense against microorganisms and some pathologicalinflammatory conditions [1]. C7 protein is polymorphic

and its charge differences can be detected on the basis of

two phenotypic features [2]. The first feature can be

identified using isoelectric focusing (IEF), which detects

protein charge differences. In Caucasians, only one

common allele (C7*1) with a gene frequency of 0.99 has

been described and two rare alleles named C7*2 and

C7*3 with genetic frequencies of less than 0.01 have alsobeen reported [3]. In Asians such as Japanese, however,

C7*2, C7*3, and another polymorphism, C7*4, are

common with gene frequencies of 0.10, 0.05, and 0.04,

respectively, in addition to the most common allele C7*1

with a frequency of 0.81 [4]. Other very rare alleles

designated C7*6, C7*7, C7*8, and C7*10 have also beenreported in different populations [2]. C7*5 and C7*9

were registered once but were later demonstrated to be

identical to C7*3 and C7*N, respectively. The second

phenotypic feature of C7 protein is C7*N or C7*M,

which can be detected using the allospecific monoclonal

antibody (mAb) WU4-15 [5]. The C7*M and C7*N gene

frequencies are about 0.8 and 0.2, respectively, and are

similar in Caucasians, Japanese, and South AfricanCape Coloreds.

The genetic basis for the C7 M/N polymorphism has

been determined by epitope mapping using the mAb

WU4-15 and C7 cDNA fragments. A single nucleotide

substitution of the codon for amino acid residue 565 was

Biochemical and Biophysical Research Communications 298 (2002) 450–455

www.academicpress.com

BBRC

* Corresponding author. Fax: +81-92-642-5247.

E-mail address: horiuchi@intmed1.med.kyushu-u.ac.jp (T. Horiu-

chi).

0006-291X/02/$ - see front matter � 2002 Elsevier Science (USA). All rights reserved.

PII: S0006 -291X(02 )02481 -6

found to be the cause of the threonine (Thr) and proline(Pro) substitutions in C7*M and C7*N, respectively [6].

However, the molecular bases of the C7 polymorphisms

detected by IEF have yet to be elucidated. In this study,

we demonstrated the genetic bases of two of these C7

IEF-detected polymorphisms, C7*3 and C7*4, both of

which are common in Asian populations. An important

feature of this study is that we used a simple and rapid

polymerase chain reaction (PCR)-based analyticalmethod to genotype these polymorphisms instead of the

conventional IEF method. A large-scale association

study on patients with diseases such as immunological

disorders and infections is needed, because C7*3 is hy-

pomorphic and C7*4 is believed to be hemolytically

hypofunctional.

Materials and methods

DNA preparation. Peripheral blood mononuclear cells (PBMCs)

were prepared by density gradient centrifugation using Lymphocyte

Separation Medium (LSMR; Organ Teknica) from heparinized blood

obtained from healthy Japanese individuals. DNA was prepared as

described previously [7].

Polymerase chain reaction/single-strand conformation polymorphism

(PCR/SSCP) analysis. The nucleotide sequences of exon-specific

primers for exons 1–17 of the C7 gene were originally designed for the

molecular analysis of C7 deficiency [8]. Due to the lack of information

about the nucleotide sequence flanking exon 0 that encodes the

N-terminal two amino acid residues of the C7 leader peptide, exon 0

was not analyzed. As the polymorphism(s) should reside in the mature

(without leader peptide) C7 protein, analysis of exons 1–17 was suffi-

cient for our study. The PCR/SSCP analytical procedure has been

successfully applied by our group to the detection of molecular defects

in complement C7 deficiency and deficiencies of other MAC compo-

nents, including complement C6, C8, and C9 [8–11]. In brief, radio-

labeled PCR products of each exon were electrophoresed at 25 �Cusing 5% nondenaturing acrylamide gel containing 5% glycerol or at

4 �C using this gel without glycerol. DNA fragments were visualized by

autoradiography using Fuji RX film (Fuji Photo Film).

Nucleotide sequencing of PCR fragments. DNA fragments of in-

terest were excised from the PCR/SSCP acrylamide gels, purified on

SUPREC-01 columns (Takara Shuzo), reamplified using a PCR re-

agent kit (Perkin–Elmer), purified with a Microcon 100 (Amicon), and

directly sequenced using radiolabeled primers and the Amplicycle se-

quencing kit (Perkin–Elmer), as described previously [8,9].

Isoelectric focusing (IEF). IEF was performed as described previ-

ously [4]. Briefly, neuraminidase-treated plasma samples were applied

to agarose gels, electrophoresed, and immunoblotted with antihuman

C7 goat antiserum (Cappel) to detect C7 polymorphisms.

Allele-specific PCR. To detect the polymorphism at codon 106,

allele-specific PCR was carried out with genomic DNA. The nucleotide

sequences for R106-specific and C106-specific primers (reverse) were

50-AGGTTTATCGATATCACA-30 and 50-AGGTTTATCGATATC

ACG-30, respectively. The sequences of the primers were identical

except at their 30-termini which correspond to the single nucleotide

substitution for the polymorphism. The common primer (sense), 50-GT

GTTCAGGTCAGTGCATCA-30, and R106-specific or C106-specific

primers were used for the amplification around the codon 106. DNA

samples were amplified with each primer set, both of which yield

127 bp products. The PCR condition was as follows: initial denatur-

ation at 95 �C for 5min, followed by 30 cycles of 95 �C for 1min, 57 �Cfor 1min, and 72 �C for 1min, and then 72 �C for 5min.

Results

During the process of searching for molecular defects

throughout the coding region of the C7 gene in C7-de-

ficient individuals, we identified altered band patterns in

exons 4 and 9 of several healthy (control) individuals by

PCR/SSCP analysis. Therefore, further study at both

DNA and protein levels was performed.

Molecular basis of C7*3

As shown by the PCR/SSCP results (Fig. 1A), aber-

rant band patterns were identified in exon 4 of the C7

gene in two healthy individuals (lanes 7 and 8). Both the

aberrant (fragment 1) and normal (fragment 2) bands

were excised from the gel and directly sequenced in their

entirety. The nucleotide sequence was identical to that

reported previously [12], except that the third nucleotideof the codon TGT for cysteine (Cys) at amino acid

Fig. 1. (A) PCR/SSCP analysis of exon 4 in healthy individuals. Two

individuals (lanes 7 and 8) displayed an aberrant band (fragment 1).

Electrophoresis was performed using 5% polyacrylamide gel contain-

ing 5% glycerol at 25 �C. (B) Partial DNA sequence and deduced

amino acid sequence (one-letter code, italics) around the polymor-

phism at amino acid residue 106. The arrow indicates the position of

the substitution (T by C at nucleotide 382). Throughout this paper, the

nucleotide and amino acid residues of C7 are numbered according to

DiScipio et al. [12].

T. Horiuchi et al. / Biochemical and Biophysical Research Communications 298 (2002) 450–455 451

residue 106 was substituted by C in the aberrant band.

This substitution led to the nonconserved amino acid

substitution of Cys by arginine (Arg; C106R; Fig. 1B).

This indicates that these individuals are heterozygousfor the substitution. As C106R was associated with the

change in protein charge, this aberrant band was con-

sidered to represent one of the C7 polymorphisms de-

tected by IEF. As PCR/SSCP analysis showed that the

other exons were normal (data not shown), no other

polymorphisms appeared to exist in these individuals.

Next, we performed IEF analysis of these individuals

carrying Arg at amino acid residue 106 (R106). Asshown in Fig. 2, the electrophoretic pattern of the in-

dividual carrying R106 (lane 2) was heterozygous for

C7*3 and C7*1. The C7*3 phenotype was hypomorphic,

as described previously [13], which can be explained by

the finding that structurally important Cys was replaced

by Arg and resulted in the conformational change in

C7*3. The amino acid substitution of Cys by positively

charged Arg also explains why C7*3 displayed morecathodal movement when subjected to IEF analysis than

the C7*1 phenotype. PCR/SSCP analysis was then used

to estimate the C7*3 allele frequency in 56 unrelated

Japanese healthy individuals. This allele was observed in

6, all of whom, according to their PCR/SSCP patterns,

Fig. 2. C7 phenotype revealed by IEF of desialized samples and sub-

sequent immunoblotting. The anode is at the top. Lane 1, 1 (homo-

zygous for C7*1); lane 2, 3–1 (heterozygous for C7*3 and C7*1); lane

3, 2–1; lane 4, 4–2; lane 5, 4–1. The samples in lanes 1, 3, and 4 are

reference samples.

Fig. 3. (A) PCR/SSCP analysis of healthy individuals. Each of the five fragments was excised from the gel and sequenced in its entirety, as described

in Materials and methods. Electrophoresis was performed using 5% polyacrylamide gel without glycerol at 4 �C. (B) Nucleotide and deduced amino

acid sequences at the polymorphic site (A to C at nucleotide 1258, C to G at nucleotide 1166; arrowed). (C) Allele 1 (T367/Q398) was composed of

fragments 2 and 5 and was responsible for the C7*4 phenotype. Allele 2 (S367/K398) was composed of fragments 1 and 4, while allele 3 (T367/ K398)

consisted of fragments 3 and 5. The gene frequencies of these alleles are shown at the bottom.

452 T. Horiuchi et al. / Biochemical and Biophysical Research Communications 298 (2002) 450–455

were heterozygous for C7*3 and the allele frequency wascalculated to be 0.054, which agrees well with previously

reported C7*3 allele frequencies [2,4]. These results

suggest strongly that R106 in exon 4 of the C7 gene is

the cause of polymorphism C7*3 at the protein level.

Molecular basis of C7*4

PCR/SSCP analysis also demonstrated aberrant

bands in exon 9 in a healthy individual (Fig. 3A). Thesamples in lanes 2–5 displayed identical patterns con-

sisting of four bands. The migration mobilities of two

each of these bands were the same as those in lane 6

(fragments 1 and 4) and lane 7 (fragments 3 and 5).

These results indicate that there was a common poly-

morphism in exon 9 and the samples in lanes 2–5 were

heterozygous for this polymorphism. Presumably, the

samples in lanes 6 and 7 were homozygous for thepolymorphism. Lane 1 showed a unique band (fragment

2), which indicates another polymorphism was present.

All five bands were excised from the gel and subjected to

direct nucleotide sequencing (Fig. 3B).

A novel polymorphism was identified at amino acid

residue 398, which was coded by AAG for lysine (Lys) in

fragments 1, 3, and 4 and was identical to the previously

reported sequence [12]. In fragment 2, the first nucleo-tide of the codon AAG was replaced by C, resulting in

substitution of the nonconserved amino acid by gluta-

mine (Gln; K398Q). Due to the technical limitations of

the SSCP analysis, fragment 5 was not separated into

two fragments and, therefore, at the sequencing level, it

contained both genotypes at amino acid residue 398.

This amino acid change at residue 398 from positively

charged Lys to neutral Gln was associated with thechange in the charge of C7 protein, suggesting that this

fragment corresponded to one of the C7 polymorphisms

detected by IEF.

We identified another polymorphism in exon 9, which

has already been reported [12,14]. The codon for amino

acid residue 367 was either ACT for Thr or AGT for

serine (Ser), which resulted from single nucleotide sub-

stitutions. Fragments 2, 3, and 5 were carrying Thr atcodon 367 (T367) and fragments 1 and 4 were carrying

Ser (S367). Thus, according to the PCR/SSCP patterns

and sequencing results, three genotypes existed in exon 9

(Fig. 3). Allele 1 carries T367/Q398, which should cor-

respond to one of the C7 polymorphisms detected by

IEF, whereas alleles 2 and 3 represent S367/K398 and

T367/K398, respectively. PCR/SCP analysis was used to

assess the gene frequencies of these alleles in 56 healthyJapanese individuals, most of whom were either het-

erozygous or homozygous for alleles 2 or 3. Only five

were heterozygous for allele 1. The T367/Q398 gene

frequency was 0.045, whereas those of S367/K398 and

T367/K398 were 0.57 and 0.38, respectively. The T367S

polymorphism is common in Japanese, but it is not as-

sociated with protein charge alteration and, therefore, isnot responsible for the polymorphisms detected by IEF.

The K398Q polymorphism was associated with a

protein charge difference and its gene frequency of 0.045

was almost the same as the reported C7*4 gene fre-

quency [2,4], strongly suggesting that Q398 is the cause

of the C7*4 polymorphism. Therefore, we subjected

plasma from the individual carrying Q398 to IEF (Fig.

3A, lane 1) to confirm this at the protein level. As shownin Fig. 2, IEF showed that this individual was in fact

carrying C7*4 at the protein level (lane 5). As PCR/

SSCP analysis showed that this individual�s other exonswere normal, we concluded that Q398 is responsible for

the C7*4 phenotype.

To further confirm that R106 in exon 4 and Q398 in

exon 9 account for C7*3 and C7*4, respectively, we

determined the nucleotide sequence of exons 4 and 9 ofthree individuals carrying C7*3 heterozygously, three

with C7*4 heterozygously, and four with other IEF

polymorphisms in addition to the cases described above.

As shown in Table 1, the three heterozygotes for C7*3

and C7*1 (C7*3–1) were all carrying both amino acids

R and C at codon 106. On the other hand, the three

individuals heterozygous for C7*4 (two were C7*4–1

and the other was C7*4–2) were all heterozygous foramino acids K and Q at codon 398. The individuals who

are not carrying C7*3 or C7*4 (two were C7*1–1 and

the other two were C7*2–1) were all homozygous for C

at codon 106 and K at codon 398. We next performed

allele-specific PCR analysis to detect specifically either

the alleles carrying R106 or C106. DNA samples from

the 10 individuals carrying C7*3 allotype as determined

by IEF were all amplified by the R106-specific primerset, while those from 40 individuals without C7*3 were

all amplified only by the C106-specific primer set (data

not shown).

Table 1

Genotyping of C7 allotypes at codon 106 and 398

C7 allotype Codons

106 398

3–1a R/Cb K/K

3–1 R/C K/K

3–1 R/C K/K

4–1 C/C Q/K

4–1 C/C Q/K

4–2 C/C Q/K

1–1 C/C K/K

1–1 C/C K/K

2–1 C/C K/K

2–1 C/C K/K

a 3–1: a heterozygote for C7*3 and C7*1; 4–1: a heterozygote for

C7*4 and C7*1; 4–2: a heterozygote for C7*4 and C7*2; 1–1: a ho-

mozygote for C7*1; 2–1: a heterozygote for C7*2 and C7*1.bOne letter amino acid code.

T. Horiuchi et al. / Biochemical and Biophysical Research Communications 298 (2002) 450–455 453

Discussion

Extensive C7 protein typing by IEF has been per-

formed in Caucasian and several Asian populations, in-

cluding Japanese, Chinese, and Koreans [2]. In Asian

populations, four common and several very rare alleles

have been detected, which contrasts with Caucasians who

have only one predominant allele, C7*1, with a gene fre-

quency of 0.99 [2]. As typing using IEF is based on de-tection of protein charges, these C7 polymorphisms are

considered to reflect amino acid substitutions that alter

charges. However, the molecular bases of the C7 poly-

morphisms detected by IEF have yet to be elucidated.

Thepresent study is,webelieve, the first to demonstrate

the genetic bases of two of the commonpolymorphisms of

C7, C7*3 and C7*4. In comparison with the most com-

mon allele, C7*1, the C7*3 allotype shows more cathodalmovementwhen subjected to IEF,while theC7*4 allotype

displays more anodal movement (Fig. 2). Another im-

portant characteristic of C7*3 is its hypomorphic feature.

Washio et al. [13] reported that both the protein concen-

tration (34%) and hemolytic activity (47%) of the C7*3

phenotype were significantly lower than those of the most

common phenotype C7*1. They also suggested that the

quantity of the product of C7*4 is normal, but its hemo-lytic activity is lower than that of C7*1. The amino acid

substitution of Cys by Arg at amino acid residue 106,

which we identified in the individual carrying C7*3, ex-

plains both these features of the C7*3 protein well. Sub-

stitution of uncharged Cys by positively charged Arg in

C7*3 results in an increase in the positive charge of the

protein, leading to amore cathodal IEFposition than that

of the common C7*1 allotype. As C7 contains 56 cyste-ines, which presumably form 28 disulfide bonds, the cys-

teine residues ofC7would be expected to be important for

the folding and function of this protein [12]. It is thus

conceivable that the substitution of Cys at amino acid

residue 106 might influence the secretion and/or stability

of C7 protein. In fact, even a single amino acid substitu-

tion in the C7 gene has been shown to result in deficiency

of the protein [2,8,15]. The gene frequency of R106 was 6heterozygotes out of 56 healthy subjects (0.054), which

was consistent with that of C7*3 [2,4]. Nucleotide se-

quencing of exon 4 of the C7 gene revealed that three

additional individuals carrying C7*3 allotype by the IEF

analysis were all shown to have amino acid residue R at

codon 106 (Table 1). Seven other individuals without

C7*3 were all carrying C106 homozygously. Moreover,

allele-specific PCRanalysis of the genomicDNA revealedthat 10 additional C7*3 heterozygotes assessed by IEF

analysis were all carrying both R106 and C106, while 40

individuals without C7*3 were not carrying R106 (data

not shown). These results suggest strongly that R106 is

responsible for the C7*3 polymorphism detected by IEF.

The structural basis of the C7*4 allotype was also

elucidated in the present study. The substitution of

positively charged Lys by neutral Gln at amino acidresidue 398 reduced the positive charge of the protein,

thereby leading to anodal migration when subjected to

IEF. We found that the individuals carrying Q398 dis-

played C7*4 at the protein level and the gene frequency

of Q398 was almost identical to that of C7*4. As shown

in Table 1, all the three individuals carrying C7*4 allo-

type had amino acid residue Q at codon 398, while those

without C7*4 were not carrying Q398. These observa-tions suggest strongly that Q398 is responsible for the

C7*4 allotype. Both polymorphisms, C106R for C7*3

and K398Q for C7*4, reside in the cysteine-poor per-

forin-like domain of C7 (Fig. 4). Although the other

domains are cysteine-rich and presumably are critical

for protein–protein interactions in the MAC complex

[12], the function of the perforin-like domain is not well

understood. The molecular basis of C7*2 has yet to bedetermined.

The T367S polymorphism has been reported to be

common in Caucasians and US Blacks [14]. It was also

detected in an Asian population of Malays, although the

number (n ¼ 6) was too small to estimate the allele

frequency [14]. In the present study, we found that the

T367S polymorphism is also common in Japanese. The

gene frequency of S367 was 0.37 in Caucasians and 0.57in Japanese. These results indicate that this polymor-

phism, which cannot be detected either by IEF or an

ELISA using the mAb WU4-15, is common in different

ethnic groups.

Individuals deficient in complement C7 frequently

suffer from infections caused by Neisseria meningitidis or

Neisseria gonorrhea [16]. In addition, some cases of C7

deficiency are associated with immunological disorderssuch as systemic lupus erythematosus [16,17]. A signifi-

cant association of C7 phenotypes with various diseases

has also been reported. Nishimukai et al. showed strong

associations of C7*3 (also called C7*5) with IgA ne-

phropathy (p < 0:001, relative risk; RR¼ 12.71) and

Fig. 4. Diagram of the molecular structure of C7 (adapted from [19]).

The positions of the amino acid alterations responsible for the C7*3

and C7*4 polymorphisms are marked at the bottom. Modules are

designated according to the published recommendations [20] as fol-

lows: T1, thrombospondin, type 1; LA, receptor, type A; EG, epi-

dermal growth factor-like; CP, complement control protein; FM,

complement factor I, MAC proteins.

454 T. Horiuchi et al. / Biochemical and Biophysical Research Communications 298 (2002) 450–455

minimal-change nephrotic syndrome (p < 0:001,RR¼ 14.20) in a study of 158 Japanese patients with

chronic glomerulonephritis [18]. A significant associa-

tion between idiopathic membranous nephropathy and

C7*4 (p < 0:05, RR¼ 2.42) was also observed in that

study. It is of note that the serum concentration of the

C7*3 phenotype is only one-third of those of other C7

phenotypes [13]. Moreover, the specific hemolytic ac-

tivity of C7*4 appears to be lower compared with theother C7 phenotypes [13].

We have elucidated the molecular bases of the C7

polymorphisms C7*3 and C7*4. In both cases, noncon-

served substitution of an amino acid residue was respon-

sible for the protein charge alterations detected by IEF.

C7 phenotyping at the gene level using a method such as

PCR/SSCP analysis would be much easier and more ac-

curate than conventional IEF analysis. Our data shouldfacilitate the detailed and large-scale study of possible

associations of these C7 hypofunctional allotypes with

diseases such as immunological disorders and infections.

References

[1] H.J. Muller-Eberhard, Annu. Rev. Immunol. 4 (1986) 503–528.

[2] R. Wurzner, K. Witzel-Schlomp, K. Tokunaga, B.A. Fernie, M.J.

Hobart, A. Orren, Exp. Clin. Immunogenet. 15 (1998) 268–285.

[3] M.J. Hobart, V. Joysey, P.J. Lachmann, J. Immunogenet. 5 (1978)

157–163.

[4] H. Nishimukai, Y. Tamaki, Vox Sang. 51 (1986) 60–62.

[5] R. Wurzner, M.J. Hobart, A. Orren, K. Tokunaga, R. Nitze, O.

Gotze, P.J. Lachmann, Immunogenetics 35 (1992) 398–402.

[6] R. Wurzner, B.A. Fernie, A.M. Jones, P.J. Lachmann, M.J.

Hobart, J. Immunol. 154 (1995) 4813–4819.

[7] T. Horiuchi, K.J. Macon, V.J. Kidd, J.E. Volanakis, J. Immunol.

142 (1989) 2105–2111.

[8] H. Nishizaka, T. Horiuchi, Z.-B. Zhu, Y. Fukumori, J.E.

Volanakis, J. Immunol. 157 (1996) 4239–4243.

[9] T. Horiuchi, H. Nishizaka, T. Kojima, T. Sawabw, Y. Niho, P.M.

Schneider, S. Inaba, K. Sakai, K. Hayashi, C. Hashimura,

Y. Fukumori, J. Immunol. 160 (1998) 1509–1513.

[10] H. Nishizaka, T. Horiuchi, Z.-B. Zhu, Y. Fukumori, K. Nagas-

awa, K. Hayashi, R. Krumdieck, C.G. Cobbs, M. Higuchi, S.

Yasunaga, Y. Niho, J.E. Volamakis, J. Immunol. 156 (1996)

2309–2315.

[11] T. Kojima, T. Horiuchi, H. Nishizaka, Y. Fukumori, T. Amano,

K. Nagasawa, Y. Niho, K. Hayashi, J. Immunol. 161 (1998) 3762–

3766.

[12] R.G. DiScipio, D.N. Chakravarti, H.J. Muller-Eberhard, G.H.

Fey, J. Biol. Chem. 263 (1988) 549–560.

[13] K. Washio, K. Tokunaga, G. Dewald, S. Misawa, K. Omoto, Jpn.

J. Hum. Genet. 33 (1988) 41–47.

[14] G. Dewald, M.M. Nothen, K. Ruther, Hum. Hered. 44 (1994)

301–304.

[15] T. Horiuchi, J.M. Ferrer, P. Serra, N. Matamoros, M. Lopez-

Trascasa, C. Hashimura, Y. Niho, J. Hum. Genet. 44 (1999) 215–

218.

[16] S.C. Ross, P. Densen, Medicine 63 (1984) 243–273.

[17] H.J. Zeitz, G.W. Miller, T.F. Lint, M.A. Ali, H. Gewurz, Arthritis

Rheum. 24 (1981) 87–93.

[18] H. Nishimukai, I. Nakanishi, Y. Takeuchi, R. Sumiyoshi, K.

Mizutani, N. Iida, T. Shinomiya, Hum. Hered. 39 (1989) 150–155.

[19] M.J. Hobart, B. Fernie, R.G. DiScipio, J. Immunol. 154 (1995)

5188–5194.

[20] R.F. Doolittle, Ann. Rev. Biochem. 64 (1995) 287–314.

T. Horiuchi et al. / Biochemical and Biophysical Research Communications 298 (2002) 450–455 455