Molecular Human Reproduction vol.2 no.10 pp. 807-811, 1996
Internal carbohydrate complexity of the oligosaccharide chains of
recombinant human follicle stimulating hormone (Puregon®,
Org 32489): a comparison with Metrodin and Metrodin-HP
S.D.Harris, C.J.Anobile, J.D.McLoughlin, R.Mitchell, A.Lambert and W.R.Robertson1
department of Medicine, University of Manchester, Hope Hospital, Eccles Old Road, Salford M6 8HD, UK
n
To whom correspondence should be addressed
Introduction
Follicle stimulating hormone (FSH), in common with other
glycoproteins, luteinizing hormone (LH), thyroid stimulating
hormone (TSH) and human chorionic gonadotrophin (HCG),
exhibits heterogeneity mainly in terms of carbohydrate structures, particularly sialic acid, attached to the protein core
(Ulloa-Aguirre et al, 1995). The physiological significance of
these glycoforms remains unclear although they have different
in-vivo half-lives which may influence their bioactivity
(Stockell-Hartree and Renwick, 1992). However, they are
known to vary in relation to the stage of the menstrual cycle
with a relative increase in the proportion of less acidic serum
FSH glycoforms during the late follicular phase and mid-cycle
(Padmanabhan et al, 1988; Wide and Bakos, 1993; Zambrano
et al., 1995), probably related to the high oestrogenic milieu
and increased gonadotrophin-releasing hormone (GnRH) secretion at that time. Other studies on glycoform distributions have
used lectin affinity chromatography (Papandreou et al., 1993;
Creus et al., 1996) and have demonstrated that FSH secreted
during the menopause contains less branched glycoforms (more
simple forms) compared with that secreted during the follicular
phase of the normal menstrual cycle. Our work on human
FSH (Lambert et al., 1995) indicated that the more alkaline
forms have an increased bioactivity and Padmanabhan (1995)
has reported that the different FSH glycoforms may encode
for different biological functions. In addition, 15 FSH forms
isolated from human pituitary extracts possessed a 5-fold range
in specific activity by radioreceptor assay, again indicative of
a functional difference between the glycoforms (Stanton et al.,
© European Society for Human Reproduction and Embryology
1992). In summary, it is now becoming clear that the internal
carbohydrate branching and overall charge of the glycoforms
of FSH are under endocrine control and represent a further
subtle mechanism by which in-vivo bioactivity is modulated.
The production of recombinant human gonadotrophins represents a significant advance in assisted reproduction as they
have advantages over existing urinary preparations, e.g. continual production of unlimited quantities of greater consistency
and purity batch to batch (Howies, 1996; Olivje et al., 1996).
The use of recombinant human FSH in assisted reproduction
programmes has recently been reviewed (Loumaye et al.,
1995). Recent clinical trials with two different recombinant
human FSH preparations (Puregon®; Organon and Gonal F®;
Serono) have demonstrated the efficacy of these preparations
compared to urinary FSH during in-vitro fertilization (IVF)
treatment (Out et al, 1995; Recombinant human FSH Study
Group, 1995). In view of the likelihood that alterations in
carbohydrate structures may affect in-vivo bioactivity, we
recently examined the glycoform distribution of a variety
of FSH preparations currently used in assisted reproduction
programmes and found the charge heterogeneity of recombinant
human FSH (rhFSH, Puregon) glycoforms to be similar to
follicular and luteal phase glycoforms (Lambert et al, 1995).
We have now extended our studies on the physico/chemical
characterization of rhFSH to an investigation of the complexity
of internal carbohydrate branching of this compound using
concanavalin A chromatography, and compared this characterization with two urinary FSH (uFSH) preparations (Metrodin
and Metrodin-HP) currently used in assisted reproduction
programmes.
807
Downloaded from http://molehr.oxfordjournals.org/ at Pennsylvania State University on March 5, 2014
Glycoforms of recombinant human follicle stimulating hormone (rhFSH) (Org 32489, Puregon®) were characterized using concanavalin A lectin affinity chromatography to reveal information about the internal carbohydrate
complexity (extent of carbohydrate side-chain branching) of the preparations. The rhFSH glycoforms were
measured by radioimmunoassay and a two-site immunoradiometric assay and compared with those in two
urinary preparations (Metrodin and Metrodin-HP) used in assisted reproduction programmes and a urinary
FSH international standard 70/45 (uFSH IS 70/45). Similar data were obtained with both assays; rhFSH had
6% complex internal carbohydrate structures compared with 22-27% for Metrodin, Metrodin-HP and uFSH.
The proportion of simple carbohydrate structures was also different, with rhFSH having 18.5 compared with
4.5-9.3% for Metrodin, Metrodin-HP and uFSH. A linear relationship was observed between the percentage
glycoforms with an isoelectric point (pi) <4 and the log percentage simple forms (logarithmic regression; r =
0.93) indicating a direct relationship between carbohydrate complexity and charge heterogeneity. In summary,
rhFSH contains fewer complex forms and an increased proportion of simple carbohydrate structures in
comparison with Metrodin, Metrodin-HP and IS 70/45.
Key words: concanavalin A/glycoform/immunoreactivity/lectin affinity chromatography/rhFSH
S.D.Harris or al.
Materials and methods
Chemicals
Recombinant human FSH (Org 32489, batch IP 190/0824; 166 IU/
ampoule, standardized against IS 70/45) was a gift from Organon
Laboratories Ltd. The first international standard (IS) of urinary FSH
and LH (LH, 46 IU/ampoule and FSH, 54 IU/ampoule) was obtained
from the National Institute of Biological Standards and Control
(NIBSC), Hampstead, London, UK. Metrodin and Metrodin-HP, both
75 IU FSH/ampoule as determined by in-vivo bioassay using the IS
70/45, were obtained from Serono Laboratories.
Concanavalin A affinity chromatography procedure
Concanavalin A (Con A) interacts with A'-linked oligosaccharide
structures according to their branching properties such that triantennary, tetra-antennary and bisecting oligosaccharides (complex forms)
do not bind to Con A, biantennary and truncated hybrids (intermediate
complexity forms) bind weakly and high mannose and hybrid oligosaccharide (simple forms) firmly bind to Con A (Cummings and
Komfeld, 1982).
Before use, the lectin columns (2.5ml Con A agarose, Sigma
Chemical Co.) were equilibrated with five column volumes of prewash
solution followed by 0.05 M phosphate buffer. Either 1 IU/100 [i\ of
international reference preparation FSH 70/45 (urinary), recombinant
FSH (Org 32489), Metrodin or Metrodin-HP, all prepared in
Dulbecco's modification of Eagle's medium (DMEM) containing
0.1% BSA were loaded onto the column and allowed to interact for
10 min at room temperature. Unbound FSH was collected by passing
5 ml of eluent 1 down the column in 0.5 ml fractions. Weakly and
firmly bound FSH was collected in a similar way by passing 25 ml
of eluent 2 and 13 ml of eluent 3 respectively down the column. A
total of 86 fractions were collected and stored frozen at -40°C for
subsequent FSH immunoassay. Recoveries ranged from 73-113%
(91.7 ± 2.6, mean ± SEM, n = 18). The column was regenerated
using five column volumes of regeneration solutions 1 and 2 in
sequence followed by prewash solution. The column was stored
capped and upright in 0.05 M phosphate buffer at 4°C.
FSH radioimmunoassay
The matrix of the FSH standard IS 70/45 was adjusted to be identical
to that of the fractions (i.e. 0.05 M phosphate buffer, pH 7.4 containing
0.1% BSA). Every 10th fraction was assayed at multiple dilutions in
order to check parallelism. All fractions tested diluted in a parallel
manner. FSH was measured by competitive double-antibody radioimmunoassay using lyophilized [l25I]-labelled FSH and rabbit antihuman pituitary FSH polyclonal antibody, code F87/2 (Chelsea
808
FSH immunoradiometric assay
FSH was measured using a non-competitive in-house immunoradioassay. The standards and quality controls used were those prepared
for the FSH radioassay. Parallelism was assessed in a similar way to
that described for the FSH radioassay and there was no evidence for
non-parallelism in the assay. The antibody label consisted of intact
FSH [125I]-labelled antibody (Serotech Ltd) and the solid phase
second antibody was a polyclonal anti-FSH antibody (SAPU) linked
to dynospheres (Dynal, Liverpool, UK). The cross-reactivity with
LH, HCG and TSH was <0.1%. Quality control samples (0.75, 15
and 30 IU/1) were assayed in duplicate in each assay. Intra-assay and
interassay coefficients of variation were < 9 and <10% over the
range 0.39-50 IU/1.
FSH immunoradioassay/radioassay ratio
The immunoradioassay/radioassay ratios were similar (P >0.05) for
each FSH preparation before separation giving a ratio of 1.19 ± 0.12
(mean ± SEM, n = 9). The glycoform profiles for each FSH
preparation were also similar when assessed by both radioassay and
immunoradioassay.
Buffer interference
Interference of the sugars (10 mM a-MG and 300 mM a-MM) was
assessed in both the immunoradioassay (I) and the radioassay (R). A
series of rhFSH standards (2, 8 and 32 IU/1) was prepared in 0.05 M
phosphate buffer containing 0.1% BSA alone or in combination with
10 mM a-MG or 300 mM a-MM. In four separate experiments
there were no differences between R-rhFSH concentrations found in
phosphate buffer or phosphate buffer containing 10 mM a-MG or
300 mm a-MM suggesting that these sugars do not interfere in the
radioimmunoassay. Similarly, there were no differences between IrhFSH concentrations in phosphate buffer or phosphate buffer containing 10 mm a-MG. However, I-rhFSH concentrations were 31%
(2 IU/1 rhFSH), 32% (8 IU/1 rhFSH) and 50% (32 IU/1 rhFSH) lower
in the presence of 300 mM a-MM. Interference by the sugar ceased
at dilutions of 1:4 and above (i.e. <75 mM a-MM) and for all
subsequent experiments fractions were diluted >1:4 before assay.
Column capacity
The column capacity (overload) was assessed by determining the
recovery of the unbound rhFSH (complex band) passed down a
second fresh column; rhFSH (1 IU) was placed on a Con A column
and 6 ml of phosphate buffer was passed down the column to remove
the unbound FSH fraction. The amount of FSH in this fraction was
then measured by immunoradioassay and the remaining material was
passed down a second fresh column. The recovery from this column
was 85%, indicating that the capacity of the column to bind FSH is
not exceeded when 1 IU of rhFSH is eluted under the conditions
described.
Reproducibility of lectin affinity chromatography procedure
Reproducibility was assessed by performing three identical lectin
runs on a sample of uFSH (IS 70/45) standard. The percentage
Downloaded from http://molehr.oxfordjournals.org/ at Pennsylvania State University on March 5, 2014
Preparation of solutions for concanavalin-A affinity chromatography
Phosphate buffer [0.05 M, pH 7.4, containing 0.1% bovine serum
albumin (BSA, ICN Biochemicals, High Wycombe, Bucks, UK)] was
used to prepare all eluents. Eluent 1 was phosphate buffer only, eluent
2 was phosphate buffer containing 10 mM a-D-methylglucopyranoside
(a-MG; Sigma Chemical Company, St Louis, MO, USA) and eluent
3 was phosphate buffer containing 300 mM a-D-methylmannopyranoside (a-MM; Sigma Chemical Company). Column regeneration
solution 1 was 0.1 M borate (pH 8.5), regeneration solution 2 was
0.1 M sodium acetate/1 M sodium chloride (pH 4.5) and the column
prewash solution consisted of 1 M sodium chloride, 5 mM calcium
chloride dihydrate, 5 mM magnesium chloride and 5 mM manganese
(II) chloride (all chemicals from BDH Chemicals Limited, Poole,
Dorset, UK). The storage and equilibration buffer was phosphate
buffer only.
Reagents, Hammersmith Hospital, London, UK). The crossreactivities of the anti-serum on a weight for weight basis, at 50%
binding using IRP 78/549 were HCG 9.7% and fi-HCG 15%. The
solid phase second antibody consisted of donkey anti-rabbit (dAR)
immunoglobulin (Ig)G antibodies bound to sac-eel (IDS, Boldon,
Tyne and Wear, UK). Quality control samples (0.75, 15 and 30 IU/1)
were assayed in duplicate in each assay. Intra-assay and interassay
coefficients of variation respectively were < 9 and <10% over the
range 0.39-50 IU/1.
Lectin affinity chromatography of rhFSH
% recovery
pI<4.0
. pfSH
40 -
4
S
20
10
50
% simple forms
Table I. Carbohydrate composition of follicle stimulating hormone (FSH)
preparations
Composition (%)
S
9 13 17 II 15 19 33 37 41 45 49 53 57 61 65 69 73 77 81 85
fraction
Figure 1. Distribution of glycoforms of follicle stimulating
hormone (FSH) as measured by radioimmunoassay in four
preparations (recombinant FSH, Metrodin, Metrodin-HP and urinary
FSH IS 70/45) following concanavalin A affinity chromatography.
Each preparation (1 IU) was placed on the lectin column and eluted
with phosphate buffer (fractions 1-10), 10 mM a-MG (fractions
11-60) and 300 mM cc-MM (fractions 61-86). Data are presented
as the percentage amount recovered in each fraction relative to the
total eluted.
I-uFSH (mean ± SEM, three separate experiments) was 24.3 ± 1.8
(complex structures), 68.7 ± 2.3 (intermediate structures) and 7.0 ±
0.6 (simple structures). There were no differences in the proportion
of each form detected by immunoradioimmunoassay and radioimmunoassay (all P >0.05) within each carbohydrate class. The total
recovery of uFSH (mean ± SEM, three separate experiments) was
100.7 ± 4.6 (I) and 101.3 ± 4.4% (R) respectively.
Statistical analysis
Combined data from separate experiments were presented as the mean
± SEM. Statistical significance of combined data was determined by
single sample Student's Mest and differences were significant when
P <0.05.
Results
The distribution of FSH glycoforms as measured by radioimmunoassay, following lectin affinity chromatography of
rhFSH (Org 32489), Metrodin, Metrodin-HP and urinary IS
rhFSH
Metrodin
Metrodin-HP
IS 70/45
Complex
Intermediate
Simple
6.0
22.0
26.0
27.0
75.5
72.5
69.5
63.7
18.5
5.5
4.5
9.3
± 1.0"
± 1.0b
± 2.0b
± 3.2b
±
±
±
±
2.5
2.0
0.5
1.9
±
±
±
±
3.5
2.5
2.5
1.9
The percentage of FSH glycoforms in each carbohydrate class (complex,
intermediate and simple) as assessed by concanavalin A chromatography for
recombinant human FSH (rhFSH), Metrodin, Metrodin-HP and urinary FSH
(uFSH, IS 70/45). Data are mean ± SD, three experiments.
"^Values within columns with different superscripts were statistically
significantly different (P <0.05).
70/45 is shown in Figure 1. The percentages of the carbohydrate
classes (complex, intermediate and simple) for each preparation
are given in Table I. There was a wide range of complexities
in the intermediate class of all four preparations as shown by
the tailing peak (Figure 1), which never fully returned to
baseline. Furthermore, rhFSH contained a relatively high
proportion of simple glycoforms (18.5 ± 3.5%) compared
with the other preparations (<10%). Conversely, it had far
fewer (6 ± 1 %) complex forms (P <0.05) compared with the
other preparations (>21%).
The glycosylation variants separated by internal carbohydrate complexity were compared to those separated on the
basis of charge (by chromatofocusing, Lambert et al., 1995).
A linear relationship was observed (Figure 2) between the
percentage glycoforms with an isoelectric point (pi) < 4
and the log percentage simple forms (logarithmic regression;
r = 0.93).
809
Downloaded from http://molehr.oxfordjournals.org/ at Pennsylvania State University on March 5, 2014
Figure 2. Relationship between the isoelectric point (pi) and
carbohydrate complexity of follicle stimulating hormone (FSH)
preparations: rhFSH = recombinant FSH; Metrodin; Metrodin-HP;
pFSH = pituitary FSH 83/575; uFSH = urinary FSH 70/45;
follicular = follicular phase FSH*. Data are presented as the
percentage isoforms with pi <4.0, separated by chromatofocusing
versus percentage simple forms separated using concanavalin A
affinity chromatography (logarithmic regression, r = 0.93). *Data
adapted from Padmanabhan et al. (1988). Data on pi of the
preparations were taken from Lambert et al. (1995).
S.D.Harris et al.
Discussion
A comparison of the glycosylation variants of the FSH
preparations separated on the basis of charge (Lambert
et al., 1995) with those separated by internal carbohydrate
complexity has revealed a negative linear relationship
between the percentage glycoforms with pi < 4 and the
810
Acknowledgements
This study was funded by the North West Regional Health Authority,
Salford Royal Hospitals NHS Trust and Organon Laboratories Ltd.
We would also like to thank Paul Gibson for technical assistance.
References
Cummings, R.D. and Komfeld, S. (1982) Fractionalion of asparagine-linked
oligosaccharides by serial lectin-agarose affinity chromatography. J. Biol.
Chem., 257, 11235-11240.
Creus, S., Pellizzari, EL, Cigorraga, S.B. and Campo, S. (1996) FSH isoforms:
bio- and immuno-activities in post-menopausal and normal menstruating
women. Clin. Endocrinol, 44, 181-189.
Hard, K., Mekking, A., Damm, J.B.L. et al. (1990) Isolation and structure
determination of the intact sialylated N-linked carbohydrate chains of
recombinant human follitropin expressed in Chinese hamster ovary cells.
Eur. J. Biochem., 193, 263-271.
Howies, CM. (1996) Genetic engineering of human FSH (Gonal-F) Hum.
Reprod. Update, 2, 172-191.
Lambert, A., Rodgers, M., Mitchell, R. et al. (1995) In-vitro biopotency and
glycoform distribution of recombinant human FSH (ORG32489), Metrodin
and Metrodin-HP. Hum. Reprod., 10, 1928-1935.
de Leeuw, R., Mulders, J, Voortman, G. et al. (19%) Structure-function
relationship of recombinant follicle stimulating hormone (Puregon). Mol.
Hum. Reprod., 2, 361-369.
Loumaye, E., Campbell, R. and Salat-Baroux, J. (1995) Human folliclestimulating hormone produced by recombinant DNA technology: a review
for clinicians. Hum. Reprod. Update, 1, 188-199.
Olijve, W., de Boer, W., Mulders, J.W.M. and van Wezenbeek, P.M.G.F.
(1996) Molecular biology and biochemistry of human recombinant follicle
stimulating hormone (Puregon). Mol. Hum. Reprod., 2, 371-382.
Out, HJ., Mannaerts, B.MJ.L., Driessen, S.G.A.J. and Coelingh Bennink,
HJ.T. (1995) A prospective, randomized, assessor-blind, multicentre study
comparing recombinant and urinary follicle stimulating hormone (Puregon
versus Metrodin) in in-vitro fertilization. Hum. Reprod,, 10, 2534-2540.
Padmanabhan, V. (1995) Neuroendocrine control and physiologic relevance
of FSH heterogeneity. [Abstr. no. S3] /. Reprod. FeniL, 15 (Abstr. series).
Padmanabhan, V., Lang, L.L., Sonsteio, J. et al. (1988) Modulation of folliclestimulating hormone bioactivity and isoform distribution by estrogenic
Downloaded from http://molehr.oxfordjournals.org/ at Pennsylvania State University on March 5, 2014
Puregon (rhFSH) has proven to be more efficacious in large
scale clinical trials (Out et al, 1995) producing a higher
number of oocytes and slightly increased pregnancy rates
compared with uFSH during standard IVF treatment.
Although current therapeutic strategies are aimed at controlled
multifollicular development, a different therapy regimen
might involve a more physiological approach with the aim
of developing one or two quality follicles and oocytes.
Towards this end, we have been characterizing rhFSH
(Puregon) to understand further the relationship between
glycoform distribution and bioactivity (Lambert et al, 1995)
and its similarity to FSH glycoforms which occur naturally
during follicular development.
In this study we have compared the structural properties
of Puregon with purified uFSH preparations used in assisted
reproduction. We found that the internal carbohydrate
structure of rhFSH contains far fewer complex forms i.e.
triantennary, tetraantennary and bisecting oligosaccharides (6
versus 22-27% for the three urinary preparations) and an
increased proportion of simple forms i.e. high mannose and
hybrid oligosaccharides (18.5 versus 5-9%). Interestingly, in
a previous study (Hard et al. 1990), rhFSH (Puregon) and
urinary hFSH were found to have a very similar monosaccharide composition with only small variations in the amounts
of GlcNAc and N-acetyl neuraminic acid. Further they
reported a much higher proportion of complex forms (>25%)
to be present in Puregon than indicated by our results (6%),
although this discrepancy may relate to the very different
technical procedures employed in the two studies (enzymatic
release of the oligosaccharides followed by separation by
gel permeation chromatography and fractionation by a
combination of FPLC and HPLC (Hard et al., 1990)
compared with concanavalin A chromatography as used in
the present study. Endogenous FSH in the late (day 13)
follicular phase contains a relatively high proportion (45%)
of complex carbohydrate glycoforms and no simple forms
(Creus et al, 1996). Likewise, Papandreou et al. (1993)
found that >80% of FSH from their normal control group
(five women in the follicular phase, day of cycle not
reported) existed as complex forms. The difference between
the glycoform composition in natural serum FSH and the
non-recombinant preparations examined in the present study
undoubtedly relates to the source of that FSH, i.e urine
from post-menopausal women. With regard to the rhFSH,
which least resembled FSH in the follicular phase, the
explanation presumably lies in the decreased ability of the
Chinese hamster ovary cells to synthesize complex, branched
carbohydrate chains. Whatever the explanation the marked
difference between the carbohydrate complexity of rhFSH
and endogenous follicular FSH seems to be unimportant as
Puregon is clearly highly bioactive in vivo.
log percentage simple forms. This may be expected since
pi of the glycoforms reflects the sialic acid termini of the
carbohydrate side chains with increased branching producing
more sialic acid termini and therefore a lower pi. Further,
de Leeuw et al. (1996) working with purified rhFSH
(Puregon) glycoforms of differing pi found that those with
a low pi have a high sialic acid/galactose ratio and are rich
in tri- and tetra-antennary AMinked carbohydrate chains
compared to those with a higher pi. These more acidic
glycoforms have had an increased in-vivo half-life since
sialylation protects the hormone from rapid clearance from
the circulation (Stockell-Hartree and Renwick, 1992).
To summarize, rhFSH is more basic (Lambert et al,
1995) and has less carbohydrate complexity than all
the other preparations (Metrodin, Metrodin-HP, IS 70/45)
examined. Although its glycoform distribution is similar to
follicular and luteal phase plasma forms with respect to
charge it is dissimilar with respect to carbohydrate complexity
although this does not appear to seriously influence its invivo bioactivity as this preparation has proven to be highly
efficacious in clinical trials. However, if different glycoforms
do encode for changed bioactivity and differential function,
it should still be the long-term aim to develop rhFSH
species which allow for a more physiological approach to
assisted reproduction.
Lectin affinity chromatography of rhFSH
Downloaded from http://molehr.oxfordjournals.org/ at Pennsylvania State University on March 5, 2014
steroids in normal women and in gonadal dysgenesis. J. Clin. Endocrinol.
Metab., 67, 465^173.
Papandreou, MJ., Asteria, C , Penerson, K. et al. (1993) Concanavalin A
affinity chromatography of human serum gonadotrophins: evidence for
changes in carbohydrate structure in different clinical conditions. J. Clin.
Endocrinol Melab., 76, 1008-1013.
Recombinant human FSH study group (1995) Clinical assessment of
recombinant human follicle-stimulating hormone in stimulating ovarian
follicular development before in vitro fertilization. Fertil. Steril., 63, 77-85.
Stanton, P.G., Robertson, D.M., Burgon, P.G. et al. (1992) Isolation and
physicochemical characterization of human follicle-stimulating hormone
isoforms. Endocrinology, 130, 2820-2832.
Stockell-Hartree, A. and Renwick, A.G.C. (1992) Molecular structures of
glycoprotein hormones and functions of their carbohydrate components.
Biochem. J., 287, 665-679.
Ulloa-Aguirre, A., Midgley, A.R. Jr., Beitins, I.Z. and Padmanabhan, V.
(1995) Follicle-stimulating isohormones: characterization and physiological
relevance. Endocr. Rev., 16, 765-787.
Wide, L. and Bakos, O. (1993) More basic forms of both human folliclestimulating hormone and luteinizing hormone at mid-cycle compared with
the follicular and luteal phase. J. Clin. Endocrinol. Metab., 76, 885-889.
Zambrano, E., Obvares, A., Pablo Mendez, J. et al. (1995) Dynamics of basal
and gonadotrophin-releasing-hormone-releasable serum follicle stimulating
hormone charge isoform distribution throughout the menstrual cycle. J. Clin.
Endocrinol. Metab., 80, 1647-1656.
Received on July 1, 1996; accepted on September 5, 1996
811