Expression of human serum albumin (HSA) mRNA in human granulosa cells: potential correlation of the 95 amino acid long carboxyl terminal of HSA to gonadotrophin surge-attenuating factor

Abstract

INTRODUCTION: According to previous studies, gonadotrophin surge-attenuating factor (GnSAF), which is assumed to be produced in human granulosa cells, has a homology with the carboxyl terminal of the human serum albumin (HSA) protein. In an attempt to validate these findings, whole or partial expression of the HSA gene was studied by RT–PCR analysis in human granulosa cells from women undergoing IVF treatment. METHODS: RT–PCR analysis of HSA RNA transcripts in luteinized granulosa cells was done in order to investigate the possible expression of the HSA gene. To ensure the specificity of PCR products, restriction enzyme and sequence analysis were performed. Western blot analysis was carried out to detect the possible expression of the albumin gene in granulosa cells. RESULTS: RT–PCR analysis and sequencing analysis of cDNA from granulosa cells revealed bands identical with those from the positive control for the amino as well as the carboxyl terminal corresponding to HSA gene at the cytoplasmic level. CONCLUSION: We have demonstrated that human granulosa cells express the carboxyl and amino terminal part of the HSA gene in levels comparable to those found in human hepatocytes. It is suggested that the coding gene for GnSAF may be a result of an alternative expression of HSA gene.

albumin, GnSAF, granulosa cells, ovary

Introduction

Gonadotrophin surge-attenuating factor (GnSAF) is a non‐steroidal ovarian substance, which, in women undergoing ovulation induction, attenuates the endogenous LH surge via a significant reduction of LH response to GnRH (Messinis and Templeton, 1991a). A prevention by GnSAF of the self-priming effect of GnRH on the pituitary has been reported both in women and animals (Messinis and Templeton, 1991b; Koppenaal et al., 1992; de Koning, 1995; de Koning et al., 2001). A line of evidence dating back to the late 1970s supports the existence of such a factor in human as well as in other species (Geiger et al., 1980; Sopelak and Hodgen, 1984; Messinis et al., 1985, 1986). GnSAF is distinct from inhibin, a gonadal protein that suppresses basal secretion of FSH (Ying, 1988; Fowler et al., 1990; Byrne et al., 1996; Mroueh et al., 1996; Pappa et al., 1999). It has been suggested that the production of GnSAF from the ovaries is regulated by FSH (Messinis et al., 1991, 1993, 1994; Fowler and Price, 1997). From a physiological point of view, GnSAF antagonizes the stimulating effect of estradiol on the pituitary and maintains the gonadotrophs in a state of low responsiveness to GnRH during the follicular phase in women and other mammals (Messinis and Templeton, 1991a; Messinis et al., 1998; Fowler et al., 2003). Although in animals this is interpreted as indicating that GnSAF is important for the timing of the LH surge onset (Koppenaal et al., 1992; Danforth and Cheng, 1995), data in women support the assumption that GnSAF controls the amplitude rather than the onset of the surge (Messinis and Templeton 1991a; Messinis et al., 1998; Fowler et al., 2001).

Four attempts have been made to date to purify and characterize bioactive GnSAF molecule(s). In these attempts, Sertoli cell-conditioned medium (Tio et al., 1994), porcine follicular fluid (Danforth and Cheng, 1995), human follicular fluid of women undergoing ovulation induction (Pappa et al., 1999) and human granulosa–luteal cell conditioned medium (Fowler et al., 2002) were used as a source. The proposed sequences refer to protein molecules of varying size from 12.5 to 69 kDa. When human follicular fluid was used, an amino acid sequence of purified GnSAF that showed homology to the carboxyl terminal of human serum albumin (HSA) was proposed (Pappa et al., 1999). Recently, recombinant polypeptides of HSA were produced by employing the expression–secretion system of Pichia pastoris (Tavoulari et al., 2004). The carboxyl terminal 95 peptide of HSA (residues 490–585) produced in recombinant form displayed GnSAF bioactivity by reducing the GnRH-induced LH secretion of primary rat pituitary cultures (Tavoulari et al., 2004). If GnSAF, therefore, is the carboxyl terminal fraction of HSA, it would be expected that the granulosa cells would express the whole or part of the HSA gene. The present study was undertaken to test this hypothesis in human luteinized granulosa cells.

Materials and methods

Primary cells and cell lines

Granulosa cells were obtained from the follicular fluid of women 20–40 years old (n = 80) undergoing ovulation induction for IVF. The patients had given consent and the local ethics committee approved the project. The cells were isolated from the follicular fluid with consecutive washes with Dulbecco’s phosphate-buffered saline (PBS; Biochrom, UK). In detail, the recovery of the granulosa cells from the follicular fluid was performed by initially separating the follicular fluid in 6 ml aliquots in the collection tubes. They were then washed with 20 mmol/l PBS (Biochrom, UK) and allowed to settle for 2–3 min at room temperature and sterile conditions. Subsequently, the supernatant follicular fluid was aspirated and the cells were washed again until the cell pellet was clear from blood contaminations. The granulosa cells were pooled and used for RNA extraction. The total number of granulosa cells recovered each time from follicular fluid was 1–2 × 10⁶ cells.

The established and characterized human hepatic cell line HepG2 was used as a positive control for HSA expression. The cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma, UK) supplemented with l-glutamine (2 mmol/l; Gibco/BRL, UK), 10% (v/v) fetal bovine serum (Gibco/BRL, UK) and penicillin/streptomycin (100 IU/ml; Gibco/BRL, UK) at 37°C in an atmosphere of 5% CO₂ in air with extra humidity. The cells were taken from the culture flask after trypsinization (trypsin/EDTA, Biochrom UK) according to the protocol described by Sambrook et al. (1989).

An alternative positive control for HSA expression was cells derived from quickly frozen liver samples obtained from biopsy, where the patients were found to be healthy or from cadaver donors.

Finally, as negative controls, human mononuclear cells were used and isolated from whole blood on a ficoll histopaque (Sigma, UK) centrifugation, as well as the established human cell line K562. The latter cell line was cultured in RPMI (Sigma, UK) medium supplemented with l-glutamine (2 mmol/l; Gibco/BRL, UK), 10% (v/v), fetal bovine serum (Gibco/BRL, UK) and penicillin/steptomycin (100 IU/ml, Gibco/BRL, UK) at 37°C in an atmosphere of 5% CO₂ in air with extra humidity.

Extraction of cytoplasmic, nuclear, total RNA and mRNA. cDNA synthesis

All types of RNA were extracted from luteinized granulosa cells, peripheral blood leukocytes, HepG2 and K562 cells and human liver tissue. Cytoplasmic and nuclear RNA (n = 20) were extracted from lutein granulosa cells, HepG2, K562 and mononuclear cells by the vanadyl ribonucleoside complex method (Davis et al., 1986). Polyadenylated RNA (n = 15) was obtained using the QuickPrep mRNA purification kit, which routinely gives between 25 and 50% polyadenylated RNA (Invitrogen, UK). Briefly, cells were homogenized in a buffered solution containing a high concentration of guanidinium isothiocyanate, centrifuged, and the supernatant was further fractionated on an oligo(dT) cellulose spin column.

Total cellular RNA (n = 45) was isolated by the guanidinium isothiocyanate (RNAzol, Sigma, UK), subjected to DNase I (Promega, UK) in order to avoid DNA contamination and stored until use at –70°C (Chomczynski and Sacchi, 1987).

Preparation of cDNA

In vitro reverse transcription of 1 µg of RNA to cDNA was performed using Moloney Murine Leukaemia Virus Reverse Transcriptase (RT) (Gibco-BRL, UK) and random hexamers as primers. After an initial denaturation at 65°C for 5 min, cDNA synthesis was performed at 37°C for 60 min.

As a control for the presence of amplifiable RNA, 5 µl of the reverse transcription (RT) cDNA product was amplified by polymerase chain reaction (PCR), using primers specific for the β-actin gene. Amplification consisted of an initial denaturation step of 5 min at 94°C, followed by 30 cycles of denaturation at 94°C/1 min, annealing at 55°C/1 min and extension at 72°C/1 min with a final extension step of 10 min at 72°C.

Both in the RT reaction and the ensuing amplification reactions, recommended measures to prevent cross-contamination of samples were followed (Kwok and Higuchi, 1989). In addition, for each experiment, a control with no template was used to check for the presence of any contamination.

PCR amplification of HSA cDNA sequences

Ten microlitres of the RT reaction was amplified by PCR using primers specific for the HSA gene. PCR was carried out in a final volume of 100 µl, with 50 pmol of each primer, 200 µmol/l each of dNTP, 1 IU of Taq DNA Polymerase (Gibco-BRL, UK) in PCR buffer (20 mmol/l Tris–HCl, pH 8.4, 50 mmol/l KCl, 1.5–2.0 mmol/l MgCl₂ depending on primer set).

Primers used for amplification were designed based on HSA gene sequence (GeneBanK M12523; Minghetti et al., 1986) in order to cover almost the whole HSA gene (Table I, Figure 1).

Table I.

Oligonucleotides used for the analysis of HSA mRNA transcripts

Primer sequence	Corresponding fragment	Expected size (bp)
F: 5′-TCAGCTCTGGAAGTCGATG-3′	D	196
R: 5′-AAGCTGCGAAATCATCCATAA-3′	(exons 12–13)
F: 5′-CCAAGTGTCAACTCCAACTCT-3′	C	413
R: 5′-AAGCTGCGAAATCATCCATAA-3′	(exons 11–13)
F: 5′-CTCGATGAACTTCGGGATGAA-3′	B	348
R: 5′-CTCATCATTTTCCACTTCGGCA-3′	(exons 6–8)
F: 5′-GTAATCGGTTGGCAGCCAATG-3′	A	230
R: 5′-CACTCTTGTGTGCATCTCG-3′	(exons 1–2)
F: 5′-GTAATCGGTTGGCAGCCAATG-3′	E	356
R: 5′-GGTAGAAGTGATTTGTCAC-3′	(exons 1–4)
F: 5′-TCTCAGATGCACACAAGAGTG-3′ R: 5′-CATCGAGCTTTGGCAACAG-3′	H (exons 2–6)	543
F: 5′-CTCGATGAACTTCGGGATGAA-3′	G	674
R: 5′-GCATTCTGGAATTTGTACTC-3′	(exons 6–10)
F: 5′-GAATATGCAAGAAGGCATCC-3′	F	221
R: 5′-GCATTCTGGAATTTGTACTC-3′	(exons 9–10)

F = forward primer; R = reverse primer.

Figure 1.

Primers used for HSA gene amplification.

Amplification for each set of primers used a pre-denaturation step at 94°C for 5 min, followed by denaturation at 94°C for 1 min, annealing varying from 53 to 62°C depending on primer set for 1 min and extension at 72°C for 1 min and final extension at 72°C for 10 min.

In each experiment, cDNA from human leukocytes and K562 cells were used as negative controls, while HepG2 cells and liver tissue were used as positive controls. The leukocytes are considered the only possible cell contaminants in the human follicular fluid.

Analysis of PCR products

To ensure the specific character of the PCR products, we performed restriction enzyme analysis and sequencing analysis. PCR products (20–30 µl) spanning the region of exon 11 and 12 as well as the carboxyl terminal of HSA gene (exon 12–13) were digested with the enzymes DdeI and AvaII.

Determination of the specific PCR products was also performed using direct sequencing. For the sequencing analysis, PCR-specific bands were isolated and purified from 3% low melting point agarose gel and sequenced by the dideoxy-chain termination method (Sanger et al., 1977) using the modified version of T7 DNA polymerase (Sequenase 2.0; US Biochemicals, Cleveland, Ohio, USA).

Semi-quantification of RNA

Amplification was performed as described above for HSA cDNA sequences corresponding to the regions from the initiation site to the start of exon 2 and from the beginning of exon 12 to the end of exon 13. In parallel, β-actin mRNA transcripts were amplified from both granulosa and HepG2 cells.

Each PCR product (20 µl) was analysed using electrophoresis on 3% low melting agarose gel (Invitrogen, UK), stained with ethidium bromide and photographed. An image captured with Polaroid camera was analysed for optical density with Hewlett Packard ScanJet 3300C (Image Analysis System). Each sample was tested at least twice.

The amount of cDNA present in each sample corresponding to the different regions of HSA gene was determined by identification of the ratio of the intensity of the band corresponding to each sample to that of β-actin (Table II).

Table II.

Semi-quantification of PCR products corresponding to different fragments of HSA gene in granulosa and HepG2 cells in relevance to β-actin mRNA transcripts

	Fragment
	A	B	C	D
Granulosa cells	0.20	0.05	0.10	0.27
HepG2 cells	0.30	0.60	0.50	0.47

Fragments A and D (corresponding to regions of HSA gene as described in Materials and methods) in granulosa cells are expressed in comparable amounts to that of HepG2 cells, while for the fragments B and C comparison of the expression between the granulosa and the HepG2 cells showed almost undetectable expression of these parts of the HSA gene in granulosa cells.

Northern blot analysis

Total RNA (20 µg) was separated by electrophoresis at 4 V/cm through a 1% agarose/0.66 mol/l formaldehyde gel and transferred to N-Hybond membranes (Amersham Ltd, UK) by capillary elution (Sambrook et al., 1989). Probes were labelled by the random priming method (Feinberg and Vogelstein, 1984) using 20–50 ng DNA or cDNA and 50 µCi [³²P]dCTP per reaction. The HSA probe was prepared from the heterologous expression system Pichia pastoris (Invitrogen, UK), where the full length HSA cDNA molecule was subcloned; the human glyceraldyhyde-3-phosophatase dehydrogenase (GAPDH) cDNA (in T7, 207 bp HindII insert) was used as a control.

Western blot analysis

Western blot analysis was performed as described earlier (Sambrook et al., 1989). In short, granulosa cells were treated using lysing buffer (Biorad, UK) and the lysates were boiled in sample buffer for 10 min. Equal amounts of protein (40 µg) were separated by 8–13% sodium dodecyl sulphate–polyacrylamide gel and transferred onto nitrocellulose membranes. The blots were blocked using 5% non-fat dry milk in PBS plus 0.05% Tween 20 and exposed to rabbit anti-HSA (Ab-cam, UK) polyclonal primary antibody overnight at 4°C, followed by 1 h incubation at room temperature with goat anti-rabbit IgG conjugated to horseradish peroxidase (HRP). Detection was carried out using the chemiluminescence (ECL) reaction (Pharmacia, UK). To avoid contamination with tissue proteins, human liver tissue as a positive control was excluded and only HepG2 cells were used. For the same reasons, K562 cells and not white blood cells were used as a negative control.

Results

Reverse transcriptase was used to generate cDNA from total, nuclear, cytoplasmic and mRNA from granulosa, HepG2, K562 and mononuclear blood cells. The cDNA were amplified using primers corresponding to each fragment of HSA gene and the full length HSA mRNA transcript (Figure 2). Full-length HSA mRNA transcripts were undetectable in granulosa cells. RNA transcripts corresponding to HSA exon I and II as well as those for exons 12 and 13 (fragments A and D, respectively) were detected both in messenger as well as in cytoplasmic RNA. Primers used for the remaining part of HSA gene showed no detectable band, neither to the cytoplasmic nor to the mRNA.

Figure 2.

PCR was performed using specific primers (see materials and methods) for fragments A, B, C, D of the HSA gene. Samples were from total (T), cytoplasmic (Cy) and nulear (N) RNA. (I) Total RNA extracts from granulosa cells (1, 4, 7, 10), HepG2 cells (2, 5, 8, 11), white blood cells (3, 6, 9, 12). (II) Cytoplasmic and nuclear RNA extracts from granulosa cells (1, 2, 5, 6), HepG2 cells (3, 7), negative control (4, 8). M: marker φX 174 DNA/HaeIII.

To further understand the underlying mechanism of the differential expression of HSA gene in granulosa cells, we studied HSA RNA processing for removal of intervening sequences (IVS, introns) by examining cytoplasmic and nuclear RNA. As seen in Figure 1 and Table I, the primers were designed to produce PCR products corresponding to fragment A, extending from the beginning of exon 1 to the first base of exon 2 (to detect transcripts with IVS-1), to fragment F from exon 9 to the end of exon 10 (to detect IVS-9) and to fragment D from exon 12 to the end of exon 13. The primer sets resulted in the PCR products corresponding to fragments A and D as shown in Figure 2. The resulting products corresponded to the expected size without the presence of introns. Fragment F was not detected in the cytoplasm, as only fragments A and D were expressed by granulosa cells both in the nucleus and in the cytoplasm. Cytoplasmic and nuclear RNA from granulosa cells showed one band when amplified with primers from exons 1 and 2 as well as from exons 12 and 13 corresponding to the correctly spliced (containing no HSA, IVS-1 or IVS-12) transcripts. To examine whether the differential expression of HSA mRNA in granulosa cells was associated with aberrant transcription, we studied the transcription initiation sites of the HSA gene. Initiation of this gene has been reported to be variable (Mignietti et al., 1986). We extracted mRNA and cytoplasmic RNA from granulosa and HepG2 cells and used them for RT–PCR analyses using the appropriate primers for the initiation sites (see Materials and methods). Correctly initiated HSA transcripts were detected both in HepG2 and granulosa cells. Sequencing analysis and digestion of the corresponding PCR products with appropriate restriction enzymes (Figure 3) confirmed the identity of each fragment. RT–RCR analyses of K562 and mononuclear blood cells gave no detectable HSA message.

Figure 3.

Restriction enzyme analysis of PCR products from granulosa cells. PCR products (20–30 µl) spanning the region of exons 11 and 12 (fragment C) as well as the carboxyl terminal of HSA gene (exons 12 and 13, fragment D) were digested with the enzymes DdeI and AvaII. (I) Enzyme DdeI: analysis of fragment D of the HSA gene (1), original size 196 bp, fragmented in 110 and 86 bp after restriction reaction of the PCR product and fragment C (2), original size 413 bp, fragmented in 303 and 110 bp after restriction reaction of the PCR product. (II) Enzyme AvaI: analysis of fragment C of the HSA gene (1), original size 413 bp, fragmented in 296 and 117 bp indicating the correct sequence in the analysed PCR products. M: Marker φX 174 DNA/HaeIII.

To quantify HSA transcripts in granulosa cells, semi-quantitative RT–PCR was performed (Vanden Heuvel et al., 1993) corresponding to fragments A, B, C and D of the HSA gene. In both granulosa and HepG2 cells, significant amounts of fragments A and D were found and the relative intensities of both regions were comparable (Table II), while the expression of fragments B and C was almost undetectable in the granulosa cells.

To further study the expression of HSA gene and the putative GnSAF gene sequence in granulosa cells, total RNA was extracted from granulosa and HepG2 cells and used for northern blot analysis. Correctly sized HSA mRNA transcripts, corresponding to the full transcript of HSA gene, were detected in HepG2 cells (Figure 4). The lack of detection of HSA mRNA in granulosa cells suggest that HSA is possibly expressed in limited amounts by granulosa cells, not detectable by this method.

Figure 4.

Northern blot: total RNA (20 µg) from hepatic tissue (1), HepG2 cells (2), granulosa cells (3) and white blood cells (4) was separated by electrophoresis and the expression of HSA was analysed (I). GAPDH expression is shown as an internal standard (II).

In western blot analyses, the protein extracts of human luteinized granulosa cells and the HepG2 cell line gave detectable bands of the HSA at the expected size of 65 kDa but no other bands were determined (Figure 5).

Figure 5.

Western blot analysis; 40 µg of protein extracts from granulosa cells (1, 2, 3), HepG2 (4) and K562 cells (5) were separated by SDS–PAGE gel and transferred onto nitrocellulose membranes which were exposed to rabbit anti-HSA polyclonal primary antibody (see Materials and methods). The expected size for the HSA protein is 65 kDa. For granulosa cell lanes and HepG2 lanes, only the HSA protein was detected, while K562 cells were used as a negative control.

Discussion

This is the first attempt to investigate the molecular basis of GnSAF that can lead to the better understanding of this factor and its physiological role in the human menstrual cycle.

Our results showed expression of the HSA mRNA in the human luteinized granulosa cells. More specifically, we detected HSA mRNA transcripts extending from the initiation site to the beginning of exon 2 and from exon 12 to the end of exon 13. The above mRNA transcripts were detected both in the nucleus and the cytoplasm indicating the correct processing of these two regions of HSA gene. Moreover, no other mRNA transcripts coding the remaining exons of the HSA gene were detected in the cytoplasm. The absence of the transcribed regions of HSA gene extending from exon 2 to the beginning of exon 11, in contrast to the regions coding the N and C terminus of HSA gene, proposed the differential expression of HSA gene in granulosa cells. These results support previous findings that GnSAF is the carboxyl terminal of HSA (Pappa et al., 1999). In addition, in recent findings regarding recombinant polypeptides of HSA, the carboxyl terminal 95 peptide of HSA as expressed from P. pastoris reduced the GnRH-induced LH secretion of rat pituitary cells by 50–82%, therefore showing GnSAF bioactivity (Tavoulari et al., 2004). In contrast, when tested on the same bioassay, the full-length HSA residues were inactive (Tavoulari et al., 2004).

The present data might be explained by an alternative splicing of HSA gene, as the mRNA precursors enter the cytoplasm, or an alternative expression of the HSA gene in granulosa cells possibly through a different promoter. The exact mechanism has not yet been investigated. The possibility that the same gene plays different roles in different tissues cannot be excluded as several genes are transcribed and translated differently depending on the tissue (Boissel et al., 1998; Strausberg et al., 2002). It has also been estimated that ≥59% of all human genes utilize alternative RNA processing to generate multiple mRNA products, which can have differences in the composition of exons (Venter et al., 2001). The inclusion or exclusion of exonic sequences enhances generation of additional protein isoforms that can differ in structure and functional properties. It is known that the human albumin gene has an alternative TATA box upstream from its normal one (Minghetti et al., 1986; Urano et al., 1986); and that it is possible to trigger this in an alternative tissue by different signals, in this case the granulosa cells by FSH.

Our data concerning the protein level suggest that GnSAF is initially synthesized as a precursor HSA molecule, and finally, by alternative splicing, the mature molecule is produced. Western blot analysis displayed expression of the whole HSA protein by granulosa cells in amounts similar to that shown by HepG2 cells, while no other bands corresponding to the carboxyl terminal of HSA were detected (12.5 kDa); the lack of detected amounts at protein level is difficult to explain. One possibility, however, is the non-specificity of the commercially available antibody. A second possibility is the cell system used. Although one could argue about the reliability of this system, granulosa cells are considered the main source of the putative GnSAF. Certainly, during the normal menstrual cycle, the role of GnSAF is restricted to the follicular phase, i.e. before the mid-cycle LH surge (Messinis et al., 1998), since higher concentrations of this factor have been found in smaller follicles than in follicles in the pre-ovulatory stage (Fowler et al., 1994, 2001). Therefore, the use of granulosa cells from small growing follicles might be more appropriate. However, in women undergoing ovulation induction, due to asynchronous follicle maturation and the continuous stimulation by FSH, high amounts of GnSAF are maintained throughout the whole ovulation induction process (Messinis et al., 1998). This, together with the limited number of natural cycles in IVF programmes and the fact that during the harvest of granulosa cells a certain number of these cells is lost, led us to employ the use of all follicles, regardless of size, in an attempt to collect sufficient material for the molecular techniques. Whether small follicles could express more fragment D transcripts needs further experimentation.

Today there are databases with gene expression profiles of human tissues and a study concerning one of them (http://expression.gnf.org/cgi-bin/index.cgi) for HSA expression in human tissues reported highest expression in liver and fetal liver, and less in the pancreas and the corpus callusum, while no expression of HSA in the mammalian ovary has been reported (Shamay et al., 2005). Therefore, consistent with our results, it is possible that only a variant of this protein is expressed in this tissue. In addition there was a recent report of an albumin-like protein expressed in the ovary of the Indian vespertilionid bat, Scotophilus heathi (Chanda et al., 2004). This protein was found to be produced mainly by the granulosa cells of the ovary at the recrudescence and the pre-ovulatory periods, coinciding with two peaks of follicular development and steroidogenesis. This 66 kDa protein showed that it shares a 70% homology with HSA (Chanda et al., 2004). All these reports, reflected at the mRNA level, support our observation that two regions of the albumin gene, coding for the N-terminal and the carboxyl terminal of the albumin protein, are transcribed by the granulosa cells in the human ovary. One could speculate as to what the contribution of the ovary and liver to circulating fragment D (GnSAF) might be. However, under physiological conditions, the liver produces only the whole albumin protein and not a part of the HSA gene (Minghetti et al., 1986), and therefore its contribution to circulating GnSAF would be negligible.

In conclusion, we suggest that the coding gene for GnSAF may be a result of an alternative expression of the HSA gene. New studies are required to clarify further these issues.

Acknowledgements

This study was supported by a research grant from the Greek Ministry for Research and Technology programme (PENED-99) (code no 1438).

Author notes

1Department of Obstetrics and Gynaecology and 2Department of Biology, University of Thessalia, Larissa, Greece