Clinical Science (2000) 98, 9–14 (Printed in Great Britain)
Changes in vitreous concentrations of human
hepatocyte growth factor (hHGF) in
proliferative diabetic retinopathy:
implications for intraocular hHGF production
Masato NISHIMURA*, Tsunehiko IKEDA†, Masaji USHIYAMA*, Shigeru KINOSHITA†
and Manabu YOSHIMURA*
*Department of Clinical and Laboratory Medicine, Kyoto Prefectural University of Medicine, 465 Kajii-cho,
Kawaramachi-Hirokoji, Kamikyo-ku, Kyoto 602-8566, Japan, and †Department of Ophthalmology, Kyoto Prefectural
University of Medicine, 465 Kajii-cho, Kawaramachi-Hirokoji, Kamikyo-ku, Kyoto 602-8566, Japan
A
B
S
T
R
A
C
T
We measured human hepatocyte growth factor (hHGF) concentrations in the original vitreous
and in the artificial vitreous after vitrectomy in 13 patients with proliferative diabetic
retinopathy (PDR) undergoing repeated pars plana vitrectomy, in order to investigate whether
the vitreous hHGF concentration is related to the recurrence of PDR after vitrectomy as well as
to the original occurrence of PDR. We also examined the relationship between vitreous
concentrations of hHGF and transforming growth factor-β2 (TGF-β2), the predominant TGF-β
isoform in the vitreous, in 14 patients with PDR. For the original vitreous, mean hHGF
concentrations were higher (P 0.05) in that from patients with severe PDR (vitreous
haemorrhage, fibrovascular proliferation and tractional retinal detachment) than in that from
patients with vitreous haemorrhage alone. In the artificial vitreous, mean vitreous hHGF
concentrations were higher (P 0.05) in that from patients with severe PDR than in that from
patients with vitreous haemorrhage alone or with vitreous haemorrhage plus fibrovascular
proliferation. No correlation was found between the hHGF concentration in the artificial
vitreous and time between vitrectomies. Vitreous hHGF concentrations were directly
proportional to vitreous concentrations of latent TGF-β2 (r l 0.831 ; P l 0.0002), but inversely
proportional to vitreous concentrations of active TGF-β2 (r l 0.495 ; P l 0.072), which inhibits
hHGF production. A decreased conversion of latent into active TGF-β2 in ocular disorders such
as PDR is likely to result in an increased concentration of hHGF in the vitreous. Thus intraocular
hHGF may be involved in pathological mechanisms causing not only the occurrence, but also the
recurrence, of PDR.
INTRODUCTION
Proliferative diabetic retinopathy (PDR) is a major cause
of adult blindness, but the precise mechanisms causing
the retinal fibrovascular proliferative changes, which may
result in vitreous haemorrhage or tractional retinal
detachment, have not been determined [1]. Numerous
mitogenic and angiogenic factors have been reported to
be involved in PDR, including insulin-like growth
factors, transforming growth factor-β (TGF-β), fibro-
Key words : hepatocyte growth factor, neovascularization, pars plana vitrectomy, proliferative diabetic retinopathy, retinal
detachment, transforming growth factor-β .
#
Abbreviations : hHGF, human hepatocyte growth factor ; PDR, proliferative diabetic retinopathy ; TGF, transforming growth
factor.
Correspondence : Dr Masato Nishimura (e-mail nishim!labmed.kpu-m.ac.jp).
# 2000 The Biochemical Society and the Medical Research Society
9
10
M. Nishimura and others
blast growth factor, tumour necrosis factor and vascular
endothelial growth factor [2–6]. Recently we showed that
mean vitreous concentrations of human hepatocyte
growth factor (hHGF) are higher in subjects with PDR
than in non-diabetic control subjects, non-diabetic subjects with proliferative vitreoretinopathy or diabetic
subjects without PDR [7]. Previous studies indicate that
hHGF is a powerful inducer of angiogenesis [8,9], and
this growth factor may contribute to the genesis of
AIDS-associated Kaposi’s sarcoma, a cytokine-dependent neoplasm characterized by a major component of
neovascularization [10]. These findings suggest that
vitreous hHGF plays a key role in neovascularization in
PDR.
The diffusable factors associated with pathological
changes in the retinal region should be present within the
vitreous [3]. Intraocular hHGF may be involved in
the recurrence of PDR after vitrectomy, as well as in the
original occurrence of PDR. The aim of the present study
was to investigate whether the vitreous hHGF concentration, not only in the original vitreous obtained at
first vitrectomy but also in the artificial vitreous obtained
at repeated vitrectomy, is related to the severity of PDR,
by measuring vitreous hHGF concentrations in PDR
patients undergoing repeated pars plana vitrectomy. We
also investigated the relationship between the vitreous
concentrations of hHGF and TGF-β , which is a pre#
dominant isoform of TGF-β in the posterior segment of
the eye [11,12] and has an inhibitory effect on hHGF
production [13–16].
METHODS
Subjects
Vitreous samples were collected from diabetic patients
with PDR undergoing pars plana vitrectomy, which was
performed in the Ophthalmology Department, Kyoto
Prefectural University of Medicine. No patients enrolled
in the study had severe hepatic or renal disorders which
could affect the serum concentration of hHGF. The
study was approved by the Ethical Committee for
Human Research of Kyoto Prefectural University of
Medicine, and all subjects provided informed consent for
participation.
Changes in vitreous hHGF concentrations in
PDR
In total, 13 diabetic patients with PDR, who had
undergone pars plana vitrectomy at least twice, were
studied [nine females and four males ; age 56p8 years
(meanpS.D.)]. Samples of original or artificial vitreous
were collected at each vitrectomy, and hHGF concentrations were measured in 32 vitreous samples in all
(Table 1). The indications for the vitrectomy were
# 2000 The Biochemical Society and the Medical Research Society
vitreous haemorrhage, vitreo-retinal fibrovascular proliferation or tractional retinal detachment. All subjects
suffered from type II diabetes mellitus, and had been
treated with either insulin or anti-diabetic drugs. The
mean serum haemoglobin A1c value at the time of the
first vitrectomy was 7.5p0.2 %. The detailed characteristics of the participants are described in Table 1.
Vitreous samples
Before intraocular infusion, the vitreous core was cut and
aspirated via the pars plana with a vitreous cutter, and
collected undiluted. After removal of the vitreous, irrigating solution (OPEGUARD-MA : glucose 1.5 mg\ml,
NaCl 6.6 mg\ml, KCl 0.36 mg\ml, CaCl 0.18 mg\
#
ml, MgCl 0.3 mg\ml, NaHCO 2.1 mg\ml, pH 6.7–8.2 ;
#
$
Senju Pharmaceutical Co. Ltd., Osaka, Japan) was used
to filled the intraocular space as the artificial vitreous.
Vitreous samples were spun for 15 min at 13 000 g in a
refrigerated centrifuge at 4 mC to remove particles, and
then stored in aliquots in polypropylene tubes at k80 mC
until assay.
Vitreous concentration of hHGF
Vitreous concentrations of hHGF were measured using a
specific ELISA kit (Otsuka Pharmaceutical Co. Ltd.,
Tokyo, Japan). The intra- and inter-assay variation was
2.9 % and 2.6 % respectively.
Vitreous concentration of TGF-β2
Vitreous concentrations of active and latent TGF-β were
#
measured using a specific-capture ELISA kit (Amersham
Life Science) in samples from 14 subjects with PDR
(eight females and six males, age 54p10 years). The
original vitreous obtained at first vitrectomy was used in
the measurement of vitreous TGF-β concentrations.
#
Because this ELISA kit only detects active TGF-β ,
#
vitreous samples were diluted 1 : 2 (v\v) with the standard
diluent provided in the kit. Half the volume of each
vitreous sample was used for assay of active TGF-β , and
#
the other half was activated by acidification with 150 mM
HCl for 30 min at room temperature, followed by
neutralization with NaOH [17]. Concentrations of latent
TGF-β were determined by subtracting the former
#
(active) TGF-β concentration from the latter (total)
#
TGF-β concentration. Vitreous hHGF concentrations
#
were also measured in these samples, and the correlation
between vitreous concentrations of hHGF and TGF-β
#
was analysed.
Statistical analysis
Data are expressed as meanspS.E.M. The significance of
differences between groups was evaluated by analysis of
variance followed by Duncan’s multiple-range test. The
criterion for statistical significance was P 0.05. Simple
regression analyses were performed to assess the re-
Vitreous hepatocyte growth factor in proliferative diabetic retinopathy
Table 1
Clinical characteristics of 13 study participants who had undergone repeated vitrectomies due to PDR
Abbreviations : F, female ; M, male ; VH, vitreous haemorrhage ; RD, tractional retinal detachment ; P, vitreo-retinal fibrovascular proliferation. Serum concentrations of
haemoglobin A1c are the values measured at the last vitrectomy.
Subject
(years, sex)
Duration of
diabetes
(years)
Serum concn.
of haemoglobin
A1c (%)
Treatment
1 (48, F)
6
7.8
Insulin
2 (49, M)
5
6.5
Insulinjsulphonylurea
3 (54, M)
8
8.1
Insulin
4 (54, F)
6
7.1
Sulphonylurea
5 (54, F)
8
6.9
6 (70, F)
16
8.2
Sulphonylureaj
α-glucosidase inhibitor
Sulphonylurea
7 (56, F)
6
8.4
8 (69, M)
11
6.8
9 (43, F)
3
7.5
10 (54, F)
10
6.6
11 (56, M)
6
7.7
Sulphonylureaj
α-glucosidase inhibitor
Insulinjsulphonylurea
12 (65, F)
8
7.3
Sulphonylurea
13 (57, F)
6
7.9
Insulin
Insulinj
sulphonylurea
Sulphonylureaj
α-glucosidase inhibitor
Insulin
lationship between vitreous hHGF concentration and
other parameters.
RESULTS
Vitreous hHGF concentrations in subjects
with PDR
The mean hHGF concentration in vitreous samples from
PDR patients was 5.12p0.41 ng\ml (n l 32 measurements). No difference was found between the mean
hHGF concentrations in the original (5.63p0.72 ng\ml ;
n l 13) and artificial (4.76p2.15 ng\ml ; n l 19) vitreous. Vitreous hHGF concentrations were measured
Time from
previous
vitrectomy
(days)
Causes of
vitrectomy
Vitreous hHGF
concn. (ng/ml)
First vitrectomy
7
First vitrectomy
245
3
73
160
First vitrectomy
98
86
First vitrectomy
15
First vitrectomy
43
First vitrectomy
7
First vitrectomy
8
First vitrectomy
15
First vitrectomy
28
First vitrectomy
35
First vitrectomy
17
35
First vitrectomy
17
First vitrectomy
9
37
VHjPjRD
VHjPjRD
VHjPjRD
P
VHjP
VHjP
VHjPjRD
VHjPjRD
VH
VH
VHjPjRD
VH
VHjPjRD
PjRD
PjRD
VH
VH
VH
VHjPjRD
VHjP
VHjPjRD
VH
PjRD
VHjP
VH
VH
VH
PjRD
VH
PjRD
VHjP
VHjP
4.43
10.05
4.23
5.23
4.53
2.33
8.37
4.15
3.21
4.05
9.97
4.45
6.13
5.11
5.07
4.85
2.45
4.71
10.42
3.88
8.77
1.82
4.35
1.87
2.21
5.12
3.13
5.23
4.37
5.81
8.20
5.18
five times in one patient (no. 2), three times in three
patients (nos. 3, 11 and 13) and twice in each of the other
nine patients (Table 1). The interval between successive
measurements of vitreous hHGF in any one subject
ranged from 3 days to 8 months. Vitreous hHGF
concentrations ranged from 1.82 ng\ml to 10.42 ng\ml
(Table 1). In some patients, the vitreous hHGF
concentration showed a rapid recovery after vitrectomy (patient no. 2, 4.53 ng\ml at 3 days after
vitrectomy ; patient no. 1, 10.05 ng\ml at 7 days
after vitrectomy ; patient no. 6, 4.85 ng\ml at 7
days after vitrectomy ; patient no. 7, 4.71 ng\ml at
8 days after vitrectomy ; patient no. 13, 8.20 ng\ml
at 9 days after vitrectomy). Because hHGF concen# 2000 The Biochemical Society and the Medical Research Society
11
12
M. Nishimura and others
Figure 1 Vitreous hHGF concentrations and PDR lesions in
the original vitreous obtained at first vitrectomy (A) and in
the artificial vitreous obtained at repeated vitrectomy (B) in
13 subjects with PDR
Figure 2 Correlations between vitreous concentrations of
hHGF and of active TGF-β2 (A) or latent TGF-β2 (B) in 14
subjects with PDR
Panel (B) includes hHGF concentrations measured in the artificial vitreous obtained
at different vitrectomies in the same patients (see Table 1 ; patient nos. 2, 3, 11
and 13). Abbreviations : VH, vitreous haemorrhage ; P, fibrovascular proliferative
changes in the retina and vitreous ; RD, tractional retinal detachment. Significance
of differences : *P 0.05.
subgroup with vitreous haemorrhage, vitreo-retinal
fibrovascular proliferation and tractional retinal
detachment (6.87p1.05 ng\ml ; n l 7) than in that with
vitreous haemorrhage alone (2.33p0.12 ng\ml ; n l 2)
(Figure 1A). For the artificial vitreous, the mean vitreous
hHGF concentration was higher (P 0.05) in the
subgroup with vitreous haemorrhage, fibrovascular
proliferation and tractional retinal detachment
(9.12p0.84 ng\ml ; n l 2) than in those with vitreous
haemorrhage alone (3.97p0.35 ng\ml ; n l 9) or with
vitreous haemorrhage plus fibrovascular proliferation
(4.33p0.93 ng\ml ; n l 6). However, the number of
patients with vitreous haemorrhage, fibrovascular proliferation and tractional retinal detachment was lower in
the artificial vitreous group than in the original vitreous
group (Figure 1).
trations in the artificial vitreous were not related to the
time from the previous vitrectomy (r l 0.059, P l 0.812,
n l 19), there seemed to be no correlation between the
hHGF concentration in the artificial vitreous and the
time between vitrectomies.
Subgroups of patients were defined by the degree of
PDR, based on the presence of vitreous haemorrhage,
vitreo-retinal fibrovascular proliferation and tractional
retinal detachment ; vitreous hHGF concentrations were
compared between these subgroups for both the original
vitreous (Figure 1A) and the artificial vitreous (Figure
1B). Separate analyses were carried out because the
differences in vitreous composition between the original
and artificial vitreous could have major effects on the
distribution, synthesis and breakdown of hHGF
molecules. For the original vitreous, the mean vitreous
hHGF concentration was higher (P 0.05) in the
# 2000 The Biochemical Society and the Medical Research Society
Vitreous TGF-β2 and hHGF concentrations
Vitreous concentrations of TGF-β were 0.28p0.05 ng\
#
ml for the active form (n l 14) and 2.61p0.24 ng\ml for
the latent form (n l 14). Vitreous hHGF concentrations
were directly proportional to vitreous concentrations of
total TGF-β (r l 0.780, P l 0.001, n l 14) and of latent
#
Vitreous hepatocyte growth factor in proliferative diabetic retinopathy
TGF-β (Figure 2B), but inversely proportional to
#
vitreous concentrations of active TGF-β (Figure 2A), in
#
subjects with PDR.
DISCUSSION
In the present study, changes in the vitreous hHGF
concentration were not related to the time between
vitrectomies, but appeared to be associated with the degree of PDR. Mean hHGF concentrations in either the
original vitreous or the artificial vitreous after repeated
vitrectomy were higher in the vitreous of subjects with
severe PDR, which showed vitreous haemorrhage, fibrovascular proliferative changes and tractional retinal detachment, than in the vitreous of subjects not suffering
from these conditions. These findings suggest that vitreous hHGF is involved in the pathological mechanism
causing severe PDR.
The mean vitreous hHGF concentrations seen in the
present study were approx. 24-fold higher than the reported mean serum hHGF concentration of 0.213
p0.025 ng\ml in subjects with PDR [18]. The concentrations of hHGF found in the vitreous of patients
with PDR are within the range shown previously to cause
endothelial cell proliferation ; concentrations of hHGF
ranging from 1 to 10 ng\ml are sufficient to stimulate
DNA synthesis in human aortic endothelial cells [19],
and hHGF at concentrations between 1 and 5 ng\ml
elicits proliferation of human endothelial cells in a dosedependent manner [8]. Therefore hHGF in the vitreous is
likely to play a significant role in retinal neovascularization in PDR.
Concentrations of vitreous hHGF that are higher than
serum concentrations indicate that vitreous hHGF does
not represent leakage from the serum to the vitreous, but
rather that hHGF is produced endogenously in the
human eye. In the present study, the vitreous was completely removed and a fresh irrigating solution was
used to fill up the intraocular space at each pars plana
vitrectomy ; therefore the concentrations of vitreous
hHGF obtained at repeated vitrectomy are not thought
to be affected by the vitreous hHGF concentration at the
previous vitrectomy. In addition, vitreous hHGF concentrations showed rapid recovery after vitrectomy in some
patients. Thus vitreous hHGF is likely to be produced
endogenously in the eye, although further investigation is
required to determine the mechanism of the intraocular
production of hHGF.
In humans, there are three known TGF-β isoforms
(TGF-β , -β and -β ), which are localized to both the
" #
$
anterior and posterior segments of the eye [17,20–22]. Of
the three isoforms, TGF-β is the predominant form
#
in the posterior segment of the eye [11,12]. TGF-β is
known to inhibit HGF production [13–16]. In the
present study, the vitreous concentration of hHGF in
PDR patients was directly proportional to the concentrations of both total and latent TGF-β , but inversely
#
proportional to the concentration of active TGF-β .
#
These findings suggest that a decrease in the conversion
of latent TGF-β into active TGF-β may be involved in
#
#
increasing the concentration of vitreous hHGF in PDR
patients. TGF-β is presumed to be involved in the
#
pathological mechanisms of PDR, such as synthesis of
extracellular matrix, enhancement of fibrosis and contraction of collagen gels [12,23,24], and hHGF is likely
to be involved in the neovascularization of PDR, as described above. Thus the relationship between intraocular
TGF-β and hHGF may be important in the pathogenesis
#
of PDR.
In the present study, high concentrations of hHGF
were found in the original vitreous and also in the
artificial vitreous after vitrectomy of patients with severe
PDR (with fibrovascular proliferation, retinal detachment and vitreous haemorrhage). Increased vitreous
hHGF is presumed to play a role not only in the
occurrence of PDR, but also in its recurrence after
vitrectomy. Decreased conversion of latent into active
TGF-β is likely to be involved in the increased intra#
ocular production of hHGF in PDR, although direct
evidence for this has not been obtained in the present
study. Clarifying the precise role of intraocular hHGF
will be of significance for understanding the complex
pathological mechanisms causing PDR.
ACKNOWLEDGMENTS
This study was supported in part by a Grant-in-Aid for
Scientific Research (B) from the Ministry of Education,
Science and Culture of Japan, and by the Charitable Trust
Clinical Pathology Research Foundation of Japan.
REFERENCES
1 Davis, M. D. (1992) Diabetic retinopathy : a clinical
overview. Diabetes Care 15, 1844–1874
2 Klagsbrun, M. and Folkman, J. (1990) Angiogenesis.
Peptide growth factors and their receptors II. Handb. Exp.
Pharmacol. 95, 549–586
3 McAvoy, J. W. and Chamberlain, C. G. (1990) Growth
factors in the eye. Prog. Growth Factor Res. 2, 29–43
4 Meyer-Schwickerath, R., Pfeiffer, A., Blum, W. F. et al.
(1993) Vitreous levels of the insulin-like growth factors I
and II, and the insulin-like growth factor binding proteins
2 and 3, increase in neovascular eye disease. J. Clin. Invest.
92, 2620–2625
5 Adamis, A. P., Miller, J. W., Bernal, M.-T. et al. (1994)
Increased vascular endothelial growth factor levels in the
vitreous of eyes with proliferative diabetic retinopathy.
Am. J. Ophthalmol. 118, 445–450
6 Aiello, L. P., Avery, R. L., Arrigg, P. G. et al. (1994)
Vascular endothelial growth factor in ocular fluid of
patients with diabetic retinopathy and other retinal
disorders. N. Engl. J. Med. 331, 1480–1487
# 2000 The Biochemical Society and the Medical Research Society
13
14
M. Nishimura and others
7 Nishimura, M., Ikeda, T., Ushiyama, M., Nanbu, A.,
Kinoshita, S. and Yoshimura, M. (1999) Increased vitreous
concentrations of human hepatocyte growth factor in
prolilferative diabetic retinopathy. J. Clin. Endocrinol.
Metab. 84, 659–662
8 Bussolino, F., Di Renzo, M. F., Ziche, M. et al. (1992)
Hepatocyte growth factor is a potent angiogenic factor
which stimulates endothelial cell motility and growth.
J. Cell Biol. 119, 629–641
9 Grant, D. S., Kleinman, H. K., Goldberg, I. D. et al. (1993)
Scatter factor induces blood vessel formation in vivo. Cell
Biol. 90, 1937–1941
10 Naidu, Y. M., Rosen, E. M., Zitnick, R. et al. (1994) Role
of scatter factor in the pathogenesis of AIDS-related
Kaposi sarcoma. Proc. Natl. Acad. Sci. U.S.A. 91,
5281–5285
11 Pfeffer, B. A., Flanders, K. C., Guerub, C. J., Danielpour,
D. and Anderson, D. H. (1990) Transforming growth
factor beta 2 is the predominant isoform in the posterior
pole. Invest. Ophthalmol. Vis. Sci. 31, 1449 (abstract)
12 Connor, T. B., Roberts, A. B., Sporn, M. B. et al. (1989)
Correlation of fibrosis and transforming growth factor-β
type 2 in the eye. J. Clin. Invest. 83, 1661–1666
13 Ghoda, E., Matsunaga, T., Kataoka, H. and Yamamoto, I.
(1992) TGF-β is a potent inhibitor of hepatocyte growth
factor secretion by human fibroblasts. Cell Biol. Int. Rep.
16, 917–926
14 Liu, Y., Michalopoulos, G. and Zarnegar, R. (1994)
Structural and functional characterization of the mouse
hepatocyte growth factor gene promotor. J. Biol. Chem.
269, 4152–4160
15 Matsumoto, K., Tashiro, K., Tajima, H., Okazaki, H. and
Nakamura, T. (1992) Negative regulation of hepatocyte
growth factor gene expression in human lung fibroblasts
and leukemic cells by transforming growth factor-β and
glucocorticoids. J. Biol. Chem. 267, 24917–24920
16 Okajima, A., Miyazawa, K. and Kitamura, N. (1993)
Characterization of the promotor region of the rat
17
18
19
20
21
22
23
24
hepatocyte-growth-factor\scatter-factor gene. Eur. J.
Biochem. 213, 113–119
Lutty, G., Ikeda, K., Chandler, C. and McLeod, D. S.
(1991) Immunohistochemical localization of transforming
growth factor-β in human photoreceptors. Curr. Eye Res.
10, 61–74
Nishimura, M., Nakano, K., Ushiyama, M. et al. (1998)
Increased serum concentrations of human hepatocyte
growth factor in proliferative diabetic retinopathy. J. Clin.
Endocrinol. Metab. 83, 195–198
Nakamura, Y., Morishita, R., Higaki, J. et al. (1996)
Hepatocyte growth factor is a novel member of the
endothelium-specific growth factors : additive stimulatory
effect of hepatocyte growth factor with basic fibroblast
growth factor but not with vascular endothelial growth
factor. J. Hypertens. 14, 1967–1972
Pasquale, L., Dorma-Pease, M. E., Lutty, G. A., Quigley,
H. A. and Jampel, H. D. (1993) Immunolocalization of
TGF-β , TGF-β and TGF-β in the anterior segment of
" eye. Invest.
#
$
the human
Ophthalmol.
Vis. Sci. 34, 23–30
Tanihara, H., Yoshida, M., Matsumoto, M. and
Yoshimura, N. (1993) Identification of transforming
growth factor-β expressed in cultured human retinal
pigment epithelial cells. Invest. Ophthalmol. Vis. Sci. 34,
413–419
Lutty, G. A., Merges, C., Threlkeld, A. B., Crine, S. and
Mcleod, D. S. (1993) Heterogeneity in localization of
Invest. Ophthalmol. Vis. Sci. 34, 477–487
Roberts, A. B., Sporn, M. B., Assoian, R. K. et al. (1986)
Transforming growth factor type β : rapid induction of
fibrosis and angiogenesis in vivo and stimulation of
collagen formation in vitro. Proc. Natl. Acad. Sci. U.S.A.
83, 4167–4171
Ignotz, R. A. and Massague, J. (1986) Transforming
growth factor-β stimulates the expression of fibronectin
and collagen and their incorporation into the extracellular
matrix. J. Biol. Chem. 261, 4337–4345
Received 10 May 1999/6 July 1999; accepted 10 September 1999
# 2000 The Biochemical Society and the Medical Research Society