Isolation of prostatic kallikrein hK2, also known as hGK-1, in human seminal plasma

ELSEVIER

et Biophysica A~ta Biochimica et Biophysica Acta 1245 (1995) 311-316

Isolation of prostatic kallikrein hK2, also known as hGK-1, in human seminal plasma David Deperthes ~, Pierre Chapdelaine a, Roland R. Tremblay a, Chantal Brunet Jo~lle Berton b, Jacques H6bert b, Claude Lazure c, Jean Y. Dub6 a,*

b,

a Laboratoo' of Hormonal Bioregulation, CHUL Research Center and Laeal Universit3,, Sainte-Foy, Quebec, Canada h Laboratory of Allergy, Rheumatology and Immunology, CHUL Research Center and Laval University. Sainte-Foy, Quebec, Canada c Laboratory Structure and Metabolism of Neuropeptides, Clinical Research Institute of Montreal, 110 Pine Avenue West, Montreal, Quebec, Canada

Received 12May 1995; accepted 25 July 1995

Abstract To demonstrate the presence of kallikrein hK2 in the human prostate and seminal plasma, we used mouse monoclonal antibodies (MAb) against a recombinant hK2-fusion protein. Using one of these MAb 9D5, we detected the presence of several major immunoreactive spots of 22 kDa and minor ones of 31 and 55 kDa in prostate cytosol and seminal plasma. After ion exchange and immunoaffinity chromatography of seminal plasma proteins, the 22-kDa immunoreactive proteins were isolated along with 55- and 75-kDa proteins. The NH2-terminal amino acid sequencing permitted identification of fragments of hK2 and protein C inhibitor, respectively, in the 22- and 55-kDa bands. Furthermore, immunoblotting experiments in one and two-D gels with two different anti-hK2 MAbs and one polyclonal anti-PCI antibody suggested that the major 55- and 75-kDa bands were covalent hK2-PCI complexes containing either the full-length hK2 chain or only its carboxyterminal fragment in the presence of mercaptoethanol. These results demonstrate for the first time the existence of kallikrein hK2 and suggest that PCI may regulate its activity in seminal plasma. Keywords: Kallikrein hK2; Protein C inhibitor; Prostate specific antigen

1. Introduction Some members of the glandular kallikrein gene family are produced abundantly, sometimes exclusively, by the prostate of several mammalian species [1]. In humans, one of the best known kallikreins is the prostate-specific antigen (PSA), also designated as kallikrein hK3 in the new nomenclature of the Kinin 91 meeting in 1991 in Munich [2]. In recent years, this protein has become one of the most useful markers of prostate cancer [3,4]. The human prostate is also known to secrete low amounts of a kallikrein detected by antibodies directed against human urinary kallikrein or hK1 [5]. Furthermore, the human prostate contains another relatively abundant m R N A encoding kallikrein hK2, previously known as hGK-I [6-9], which shares 78% homology with PSA at the amino acid se-

* Corresponding author. Laboratory of Hormonal Bioregulation, CHUL Research Center, 2705 Laurier Boulevard, Sainte-Foy, Quebec, Canada GIV 4G2, Fax: + 1 418 6542714. 0304-4165/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0 3 0 4 - 4 1 6 5 ( 9 5 ) 0 0 1 1 8-2

quence. Although the kallikrein hKLK2 gene was sequenced in 1987 [6] and the presence of the corresponding m R N A in the prostate was demonstrated by our group in 1988 [7], all efforts to show the presence of the protein product of that gene have remained up to now unsuccessful [10,11]. In the present study, we used specific monoclonal antibodies generated against recombinant hK2 protein to detect hK2 kallikrein in prostatic tissue and seminal plasma.

2. Materials and methods

Production of recombinant hk2. Recombinant kallikrein hK2 protein was produced as a fusion protein with glutathione S-transferase (GST) in pGEX-3X vector. The steps leading to the obtainment of the fusion protein included poly (A ÷) RNA preparation by standard procedures, first-strand cDNA synthesis using an oligo (dT)~8 primer, and 200 U of M - M L V reverse transcriptase from Gibco BRL, polymerase chain reaction (PCR), and two

312

D. Deperthes et al. / Biochimica et Biophysica Acta 1245 (1995) 311-316

cloning steps. The PCR reaction was performed with the following set of primers: 5'-GGATCCGGATTGTGGGAGGCTGGG-3' and 5'-GAATTCGCACTCAGGGGTTGGCTG-3'. These primers corresponded, respectively, to nucleotide positions 1318 to 1337 and 5133 to 5114 of the DNA sequence of hKLK2 gene [5]. In addition, they contained, respectively, a Bam HI or an Eco RI sequence at their 5'-portion. The PCR reaction was conducted with 35 cycles of denaturation (1 min at 95°C), annealing (30 s at 55°C) and polymerization (1.5 rain at 72°C). The 720-bp amplified cDNA was purified by low-melting agarose gel electrophoresis, ligated with T4 DNA ligase to pT7 Blue vector from Novagen, and transformed into DH5c~-competent cells. Positive recombinant clones ascertained by sequencing were doubly digested with Bam HI and Eco RI, and the cDNA fragment was purified by low-melting agarose gel electrophoresis. This fragment was inserted into pGEX-3X vector (provided by Dr Paul Rennie of the British Columbia Cancer Control Agency) and transformed into DH5~-competent cells. For the production of recombinant GST-bK2 fusion proteins, Escherichia coli transformed cells were grown as described by Smith and Johnson [12]. Harvested cell pellets from 200 ml of cell culture were resuspended in 30 ml of cold STE buffer containing 3 mg of lysozyme and incubated for 15 min at 0°C. The suspension was adjusted to 5 mM DTT and 1.5% sarkosyl and centrifuged at 100,000 g for 30 rain. The supernatant was adjusted to 2% Triton X-100 with vortexing and mixed at room temperature with 4 ml of glutathione-Sepharose from Sigma Chemical Co. (St. Louis, MO). After 30 rain, the suspension was centrifuged at 600 X g, and the gel was washed 6 times with PBS buffer containing 0.4% Triton X-100 and 4 times with STE buffer. The GST-hK2 fusion protein was eluted from glutathione-Sepharose with 50 mM of Tris-HCl buffer, pH 8.0, containing 6 M urea and 5 mM DTT. The eluted material was dialysed extensively against 50 mM Tris-HCl buffer, pH 8.0, containing 100 mM NaCI and 0.2 mM DTT. Under these conditions, the GST-fusion protein remained soluble. 2. I. Production of monoclonal antibodies against hK2 The production of mouse monoclonal antibodies against hK2 was done as described previously [13] using the GST-hK2 fusion protein as the immunogen. The selection of clones reacting specifically with hK2, and not with PSA or GST, was done by direct binding ELISA on fixed GST-hK2 fusion protein, purified PSA [14], and an unrelated fusion protein containing GST and the DNA binding domain of the androgen receptor [15] provided by Dr Paul Rennie of British Columbia Cancer Control Agency in Vancouver. In the present study, monoclonal antibodies (MAbs) 9D5 and 8B6 were used for the detection of the protein by Western blotting and MAbs 1D3, 7B5, and 8B6 for the immunoaffinity procedure (see below). In early

experiments, we also used a rabbit PSA antiserum [14] to detect hK2 because of the cross-reactivity between these two proteins. 2.2. Puri~cation of hK2 by immunoaffini~, chromatography To enrich the hK2 protein, we used a pool of 85 ml of seminal plasma. The plasma was first dialyzed against 50 mM Tris-HCl buffer, pH 7.0. The dialyzed material was then applied on CM-Sepharose CL-6B and eluted with a linear gradient of 0 to 0.3 M NaC1. Fractions eluted between 0.15 and 0.2 M NaCI contained the 22-kDa protein(s) immunoreactive to MAb 9D5 by Western blotting. Next an immunoaffinity column in which hK2-specific MAbs 1D3, 7B5, and 8B6 were coupled to CNBr-activated Sepharose 4B was prepared. The CM-Sepharose fractions of the previous step were made 0.5 M NaCI and adsorbed on 5 ml of the affinity gel suspension. The column was washed with 50 mM Tris HCI buffer, pH 7.0, containing 0.5 M NaC1 and eluted with 0.1 M glycine HC1 buffer at pH 2.6. The eluted proteins were immediately neutralized with 1 M Tris-HC1, pH 8.0, and analyzed by 1-D gel electrophoresis and Western blotting. 2.3. Polyaco'lamide gel electrophoresis The proteins were characterized by SDS polyacrylamide gel electrophoresis (PAGE) in one-dimension (l-D) as described by Laemmli [16] and by two-dimensional (2-D) gel electrophoresis according to O'Farrell [17] using 2% Bio-lyte 3-10 from Bio-Rad Laboratories, Mississauga, Ontario, in the first dimension. Immunodetection by Western blotting was done with the ECL Western blotting system of Amersham using either monoclonal antibodies directed against hK2 or rabbit polyclonal antibodies directed against human PCI (purchased from Cedarlane Laboratories, Hornby, Ontario). The prostatic tissues used in these experiments were obtained by transurethral prostatectomy in patient with benign prostatic hypertrophy while the other tissues were autopsy specimens. The cytosols were prepared by homogenizing the tissues in 10 vol of 50 mM Tris-HCl buffer at pH 7.4 with a Polytron PT-10 homogenizer. The homogenates were centrifuged at 50 000 × g' for 1 h and the supernatants were used as cytosol. The seminal plasma pool came from the andrology laboratory of our hospital. 2.4. Amino acid sequencing Determination of the amino acid sequence was done as follows. The partially purified proteins obtained after the immunoaffinity chromatography step were separated by 1-D SDS-PAGE. They were then electroblotted onto Immobilon membranes. After the transfer, the proteins were stained with Ponceau red, and the bands of interest were

D. Deperthes et al. / Biochimica et Biophysica Acta 1245 (1995) 311-316

cut out. The membrane was washed several times in distilled water. The dry membrane was stored at room temperature until microsequencing was performed using an Applied Biosystem gas-phase sequanator (Model 470A) as previously described [ 18].

66-

313

k,L~

45-

313. Results

Large amounts of recombinant fusion protein were produced in Escherichia coli cells. However, less than 1% was recovered in a soluble form after cell lysis in the usual buffers. The inclusion of 1.5% sarkosyl and of 2% Triton X-100 after the lysosyme step permitted the recovery of 18 mg of soluble recombinant protein from 200 ml of cell culture suspension. This material was further purified by affinity chromatography on glutathione-Sepharose. The recovered proteins contained a major 55-kDa band (presumably GST-hK2 fusion protein) that was positive to the PSA polyclonal antiserum by Western blotting. The affinitypurified preparation was used without further processing for the production of murine monoclonal antibodies. We obtained several hybridomas that produced specific MAbs reacting only with the hK2 portion of the fusion protein but not with either PSA or GST. MAb 9D5 was used routinely for the detection of hK2 by Western blotting of human prostatic cytosol and seminal plasma. Several immunoreactive spots could be observed by 2-D gel electrophoresis (Fig. 1). The major spots were found in the 20to 25-kDa range and two other fainter groups of spots in the 31- and 55-kDa ranges. The 55-kDa spots were not observed in seminal plasma, but occasionally, we could see 75-kDa spots with higher isoelectric points than the 55-kDa spots of prostatic cytosol. All three groups of immunoreactive proteins were quantitatively minor proteins as evidenced by the absence of corresponding Coomassie blue staining. The immunoreactive proteins could not be detected in the cytosol of any other tissues (brain, heart, liver, lymph node, kidney, and testis), cells

2114664531-

2114Fig. 1. Demonstration of the presence of immunoreactive hK2 proteins after two-dimensional gel electrophoresis of crude human prostatic cytosol proteins (180 /xg) and seminal plasma proteins (145 /xg) and Western blotting with MAb 9D5. The nitrocellulose membranes containing the transfered proteins were incubated for 55 min at room temperature with 50 ml of PBS buffer containing 120 /xg of purified 9D5 IgGs and treated with the ECL reagents thereafter. The numbers on the left represent the molecular weights (in thousands) of protein standards.

(leucocytes), or fluid (saliva) examined (Fig. 2). MAb 9D5 clearly did not recognize major 34-kDa forms of PSA, which were readily revealed by polyclonal or monoclonal antibodies directed against PSA (results not shown). In the next experiments, we tested several MAbs alone or in combination for the purification of hK2 by im-

926645312114Fig. 2. Tissue distribution of immunoreactive hK2 in crude human tissue cytosols, cells, or fluid including brain (lane 1), heart (lane 2), liver (lane 3), lymph node (lane 4), polymorphonuclear leukocytes (lane 5), prostate with benign prostatic hypertrophy (lane 6), kidney (lane 7), saliva (lane 8), and testis (lane 9). The proteins (50-60 /.~g) from each source were separated by SDS polyacrylamide gel electrophoresis in a minigel and processed for Western blotting as in Fig. 1 with MAb 9D5.

314

D. Deperthes et al. / Biochimica et Biophysica A cta 1245 (1995) 311-316

CB

WB

92-

1----

66-:

2

z-

45-

31-

......3 21-

14Fig. 3. Characterization by one-dimensional gel electrophoresis of the seminal plasma proteins isolated after sequential CM-Sepharose and immunoaffinity chromatography using MAbs 1D3, 7B5, and 8B6 coupled to Sepharose CL-4B. The proteins eluted with 0.1 M glycine-HC1 buffer at pH 2.8 were analyzed by SDS polyacrylamide gel electrophoresis and Coomassie blue (CB) staining followed by Western blotting (WB) with MAb 9D5. The major Coomassie blue-stained proteins were numbered 1 to 3.

munoaffinity chromatography. Only one combination tested (1D3, 7B5 and 8B6) was found to effectively retain immunoreactive hK2 proteins. Fig. 3 shows that a few protein bands or groups of bands were present in the purified preparation. Three of them, numbered 1 to 3, and having respective molecular masses of 75, 55 and 22 kDa are particularly interesting either because of their immunoreactivity to MAb 9D5 (bands 1 and 3) or because of their abundance by Coomassie blue staining (bands 1 and 2). The 31-kDa immunoreactive proteins detected previously in crude samples (Fig. l) were also detected after longer exposures. After the transfer of the 1-D gel protein to an Immobilon membrane, the protein bands 1, 2 and 3 were cut out and subjected to NH2-terminal amino acid sequencing. The sequence of band 1 proteins could not be determined because an insufficient amount of material could be transferred to the Immobilon membrane. The major sequence of band 2 proteins corresponded to amino acids 10 to 29 of the previously published sequence of protein C inhibitor (PCI) [19]. A minor sequence starting at position 9 of PCI was also found. Finally, the sequence (30 cycles) obtained with band 3 protein corresponded to amino acids 1 to 30 of the previously published sequence of hK2 protein deduced from the genomic DNA sequence [6]. The

only unknown residues in our experiment were those of cycles 7 and 26, which corresponded to cysteine residues in the hK2 sequence [6]. The observed sequence diverged at several positions from the ones of PSA (hK3) and human urinary kallikrein (hKl). Based on the NH2-terminal sequence and the observed 29-kDa molecular mass of hK2 in in vitro translation studies [9], we concluded that the 22-kDa protein spots immunoreactive to MAb 9D5 (Fig. 1) were the heterogeneously glycosylated NH2-portion of truncated hK2. We also predicted that a --7-kDa COOH-terminal fragment linked by disulfide bridges to the NH2-terminal fragment should be present in the affinity-purified preparation. Such a fragment ( = 10-kDa) was indeed revealed by another MAb, 8B6, which did not recognize the 22-kDa band detected with MAb 9D5 (Fig. 4). In this experiment, a small amount of 30-kDa band was detected by MAbs 9D5 and 8B6. Both MAbs also immunoreacted with the 75-kDa doublet proteins. MAb 8B6 produced additional strong immunoreactions at the position of the 55-kDa band of PCI and also in the 40-kDa region of the blot. These results strongly suggested that the major form of kallikrein hK2 in seminal plasma had an internal cleavage and that a

9D5

8B6

--qz 75 :~!ii¸~!i !~i~i!i'~i!!i~!

--gz 55

ii!i!i!

--4z 40

-.,4 3 0 22 z,,-

~ ~ii~!i:iiil~ :ii>ii:!if!¸¸!i/i~iiii

~!i~:ii~~ii!~i!~!~i!!~i:~: . . . . . . . . . . . .

------ 1 0

Fig. 4. Characterization of immunoaffinity-purified proteins by Western blotting with two anti-hK2 MAbs of different specificities, 9D5 and 8B6. The preparation of proteins used in this experiment differed from that used in Fig. 3 in that crude seminal plasma was adsorbed directly on the IgG-Sepharose CL-4B without prepurification by CM-Sepharose. The molecular masses in kDa of the major immunoreactive bands (or groups of bands) are indicated. The 10-kDa mass is approximative as the smallest marker used was the 14-kDa hen lysozyme.

D. Deperthes et al. / Biochimica et Biophysica Acta 1245 (1995) 311-316

315

complexes in which hK2 was either full-length (75 kDa) or contained an internal cleavage (55 kDa).

92665-

4. Discussion

1-

These results unequivocally show that kallikrein hK2 is present in human prostate and seminal plasma. Our preliminary tissue distribution study suggests that kallikrein hK2 expression may be similar to that of PSA, that is, restricted to prostatic tissue and secretions. Further studies with a larger variety of normal and pathologic human tissues will be necessary to ascertain that point. The new kallikrein hK2 appears to be present at a very low concentration in the prostate cytosol and seminal plasma. This conclusion was based on both the low recovery rate of immunoreactive hK2 proteins after immunoaffinity chromatography and the absence of corresponding Coomassie-blue-stained protein spots after 2-D gel electrophoresis of crude prostatic cytosol and seminal plasma. These observations may explain why kallikrein hK2 was isolated 8 years after the publication of the nucleotide sequence of hKLK2 gene [6]. The relatively low amounts of hK2 kallikrein in human prostate is somewhat surprising as that tissue contains hKLK2 mRNA levels equivalent to as much as 15 to 50% of PSA levels [7-9]. At the protein level, the difference between hK2 kallikrein and PSA concentrations in seminal plasma are probably 100- to 1000-fold in favour of PSA; however, the reasons for this finding are not immediately apparent because the in vitro translatability of PSA and hK2 synthetic RNAs was quantitatively similar in a rabbit reticulocyte lysate system [9]. However, it is possible that the in vivo translatability of hKLK2 mRNA might be influenced by prostatic factors not present in the reticulocyte lysate system. Alternatively, the degradation of hK2 along the secretory pathway may be higher than that of PSA in prostatic epithelial cells. Another important observation in our study was the association of hK2 with other proteins during the immunoaffinity chromatography procedure. Indeed, besides hK2, the major Coomassie blue stained proteins were the 75- and 55-kDa immunoreactive PCI bands. In a control experiment using Sepharose CL-4B without conjugated IgG, no 75- and 55-kDa bands stained with Coomassie blue were observed. These results strongly suggest that the association of hK2 and PCI is not fortuitous. This conclusion is strengthened by the observation that the COOHterminal fragment of hK2 is bound covalenfly to a protein comigrating with PCI. In the case of PSA, it was recently reported that less than 5% of PSA in the ejaculate could form complexes with PCI [20]. Our results suggest that the proportion of hK2 that forms such complexes with PCI may be much higher. In the prostatic cytosol, our results show the presence of 55-kDa immunoreactive spots having an electrophoretic

21149266-

MAB 9D5 anti h K - 2

4531-

2114Fig. 5. Colocalization of hK2 and PCI using immunoaffinity purified proteins analyzed by two-dimensional gel electrophoresis followed by Western blotting either with anti hK2 MAb 9D5 or anti PCI rabbit polyclonal antibody. The arrow indicates the position of the major immunoreactive species with each antibody.

large proportion of it was covalently bound by its carboxy-terminal portion to a 55-kDa protein co-migrating with PCI, The internal cleavage hypothesis was verified by sequencing the -- 10-kDa band after its transfer to an Immobilon membrane. The NH2-terminal amino acid sequence obtained (12 cycles) was the following: SLQXVSLHLLSN. That sequence corresponded to amino acids 146 to 157 of the published hK2 sequence [6] and diverged from the PSA sequence at positions 146, 151, 154 and 155 and from the urinary kallikrein sequence at positions 146, 151, 154 and 156. In order to determine whether the 75-kDa immunoreactive band could also represent a PCI-hK2 complex, we performed a two-dimensional gel electrophoresis of an affinity-purified material similar to the one used in Fig. 3. Thereafter, Western blotting was conducted with MAb 9D5 and also with commercial polyclonal antibodies directed against PCI. Fig. 5 shows that the major 75-kDa immunoreactive forms of both PCI and hK2 migrated exactly at the same position by two-dimensional gel electrophoresis. Immunoreactive PCI spots of 55 kDa were also present. Based on these results and those of Fig, 4, we concluded that the 75- and 55-kDa bands were PCI-hK2

316

D. Deperthes et al. / Biochimica et Biophysica Acta 1245 (1995) 311-316

mobility quite different from the 55-kDa P C I - h K 2 complexes in seminal plasma. The identity of these spots remain to be determined. A n o t h e r important goal in future studies will be to obtain sufficient a m o u n t s of u n c o m p l e x e d biologically active hK2 protein in order to study its e n z y m a t i c and other activities in seminal plasma. Finally, the specific m o n o c l o n a l antibodies that we have obtained for both hK2 and P S A (results not shown) will be extremely useful in comparing the expression of hK2 and PSA in prostatic tissue as well as in the serum of patients with prostate disorders.

Acknowledgements W e are indebted to Dr. Jean Paul Valet for the preparation of the i m m u n o a f f i n i t y c o l u m n , Dr R6jean Delisle, Dept of Urology, l ' H 6 t e l - D i e u de Roberval for providing h u m a n prostatic tissues, and Mrs Lucie Turcotte for typing the manuscript. This work was supported by a grant from the Medical Research Council of Canada.

References [1] Dub6, J.Y. (1994) Comp. Biochem. Physiol. 107C, 13-20. [2] Berg, T., Bradshaw, R.A., Carretero, O.A., Chao, J., Chao, L., Clements, J.A., Fahnestock, M., Fritz, H., Gauthier, F., MacDonald, R.J., Margolius, H.S., Morris, B.J., Richards, R.I. and Scicli, A.G. (1992) In Agents and Actions, Supplements, Vol. 38/1, Recent

Progress on Kinin. (Fritz, H., Luppertz, K., and Turk, V. eds.), pp. 19-25, Birkhauser, Basel. [3] Papsidero, L.D., Kuriyama, M., Wang, M.C., Horoszewicz, J., Leong, S.S., Valenzuela, L., Murphy, G.P. and Chu, T.M. (1981) J. Natl. Cancer Inst. 66, 37-42. [4] Oesterling, J.E. (1991) J. Urol. 145,907-923. [5] Fink, E., Schill, W.B., Fiedler, F., Kassnigg, F., Geiger, R. and Shimamoto, K. (1985). Biol. Chem. Hoppe-Seyler 366, 917-924, [6] Schedlich, L.J.. Bennetts, B.H. and Morris, B.J. (1987) DNA 6, 429-437. [7] Chapdelaine, P., Paradis, G., Tremblay, R.R. and Dub& J,Y. (1988) FEBS Lett. 236, 205-208. [8] Henttu, P., Lukkarinen, O. and Vihko, P. (1990) Int. J. Cancer 45, 654-660. [9] Young, C.Y.F., Andrews, P.E., Montgomery, B.T. and Tindall, D.J. (1992) Biochemistry 31, 818-824. [10] Paradis, G., Tremblay, R.R. and DubE, J.Y. (1989) Prostate 15, 343-353. [11] Christensson, A., Laurell, C.B. and Lilja, H. (1990) Eur. J. Biochem. 194, 755-763. [12] Smith, D,B. and Johnson, K.S. (1988) Gene 67, 31-40. [13] Mourad, W., Pelletier, G., Boulet, A., Islam, N., Valet, J.P. and H~bert, J. (1986)J. Immunol. Methods 89, 53-59. [14] DubS, J.Y., Gaudreault, D. and Tremblay, R.R. (1989) Andrologia 21, 136-139. [15] Rennie, P.S., Bruchovsky, N., Leco, K.J., Sheppard, P.C., McQueen, S.A., Cheng, H., Snoek, R., Hamel, A., Bock, M.E., MacDonald, B.S., Nickel, B.E., Chang, C., Liao, S., Cattini, P.A. and Matusik, R.J. (1993) Molec. Endocrinol. 7, 23-26. [16] Laemmli, U.K. (1970) Nature 227, 680-688. [17] O'Farrell, P.H. (1975) J. Biol. Chem. 250, 4007-4021. [18] Lazure, C., Saayaman, H.S., Naude, R.J., Oelofsen, W. and Chr~tien, M. (1989) Int. J. Peptide Prot. Res. 33, 46-58. [19] Sozuki, K., Deyashiki, Y., Nishioka, J., Kurachi, K., Akira, M., Yamamoto, S. and Hashimoto, S. (1987) J. Biol. Chem. 262, 611-616, [20] Christensson, A. and Lilja, H. (1994) Eur. J. Biochem. 220, 45-53.