Lack of correlation between vascular endothelial growth factor ...

6 downloads 40 Views 388KB Size Report
production in human endometrial tissues from different stages of the menstrual cycle and relating these data to endothelial cell proliferation in the same tissues.
Human Reproduction vol.14 no.8 pp.2080–2088, 1999

Lack of correlation between vascular endothelial growth factor production and endothelial cell proliferation in the human endometrium

C.E.Gargett2, F.L.Lederman, T.M.Lau1, N.H.Taylor and P.A.W.Rogers Monash University Department of Obstetrics and Gynecology, Monash Medical Centre, 246 Clayton Road, Clayton, Victoria, 3168 Australia 1Present

address: School of Life Science and Technology, Victoria University of Technology, McKechnie Street, St Albans, Victoria 3021, Australia

2To

whom correspondence should be addressed

The role of vascular endothelial growth factor (VEGF) in endometrial angiogenesis was examined by measuring its production in human endometrial tissues from different stages of the menstrual cycle and relating these data to endothelial cell proliferation in the same tissues. Conditioned medium was collected from explant, and separated glandular epithelial and stromal cells cultured from 24 normal human endometrial biopsies and VEGF measured by enzyme-linked immunosorbent assay (ELISA). Immunohistochemistry was also used to assess VEGF and the percentage of proliferating microvessels in the samples. Wide variation in results between individual endometrial samples at each stage of the menstrual cycle was observed for all parameters measured. There was no significant difference in VEGF secretion by explant, glandular epithelial or stromal cell cultures across the menstrual cycle, or in the percentage of proliferating vessels. VEGF immunostaining in the stroma was elevated during the early proliferative stage (P J 0.03). Epithelial cells secreted more VEGF than stromal cells (1.76 K 0.46 versus 0.46 K 0.06 ng per 105 cells; P J 0.002). There was no correlation between VEGF secreted by cultured explants, epithelial or stromal cells, VEGF immunostaining and the proportion of proliferating microvessels. These results show that the majority of endometrial VEGF is produced by glands, but neither total glandular nor stromal VEGF is correlated with endometrial endothelial cell proliferation. There is still no clear understanding on the regulation of human endometrial angiogenesis. Key words: angiogenesis/endometrium/endothelial cell/human/ VEGF

Introduction Angiogenesis is the development of microvessels from existing blood vessels, a process that normally occurs during fetal development, wound healing and in the reproductive tract of the adult female (Folkman and Shing, 1992; Gordon et al., 2080

1995; Risau, 1997). The critical role of angiogenesis in ovarian and endometrial function was demonstrated when the angiogenesis inhibitor, AGM-1470, substantially retarded growth of both the corpus luteum and the endometrium when administered chronically to non-pregnant, cycling mice (Klauber et al., 1997). In human endometrium, angiogenesis occurs periodically as part of the cyclical growth and shedding which take place during the menstrual cycle. There are three different times during the endometrial cycle when angiogenesis occurs (Rogers and Gargett, 1999): during menstruation for repair of the vascular bed (Markee, 1940); during rapid endometrial growth of the proliferative phase (Rogers et al., 1993); and during the secretory stage when spiral arterioles exhibit significant growth and coiling (Kaiserman-Abramof and Padykula, 1989). While circulating oestrogen and progesterone primarily regulate the overall control of endometrial growth and regression, it appears unlikely that steroids are direct regulators of vascular growth in this tissue. For example, immunohistochemical analysis of endothelial cell proliferation in human endometrium across the cycle shows no clear relationship to hormonal changes (Goodger (MacPherson) and Rogers, 1994). It is clear that vascular endothelial cell growth factor (VEGF) is a major regulator of developmental angiogenesis, as well as of pathological angiogenesis associated with growth of solid tumours and with retinal and synovial disorders (Ferrara and DavisSmyth, 1997). Targeted disruption of even a single VEGF allele resulted in abnormal blood vessel formation and embryonic death in the mouse (Carmeliet et al., 1996; Ferrara et al., 1996). VEGF mRNA is upregulated in most malignant tumours (Ferrara and DavisSmyth, 1997), and tumour growth is inhibited by strategies targeting tumour angiogenesis. Thus, treatment with neutralizing VEGF monoclonal antibodies in vivo, but not in vitro, inhibited tumour cell growth (Kim et al., 1993), as did tumour cell expression of antisense VEGF (Saleh et al., 1996). Direct evidence for VEGF as a mediator of normal angiogenesis was recently demonstrated when injection of a truncated, soluble flt-1 receptor prevented corpus luteum angiogenesis, maturation and progesterone release in a rat model, with secondary failure of endometrial development (Ferrara et al., 1998). VEGF is expressed in a wide range of cells and tissues (Ferrara and DavisSmyth, 1997), including rodent, primate and human endometrium (Gordon et al., 1995; Smith, 1995; Greb et al., 1997; Torry and Torry, 1997). There are five isoforms of the VEGF molecule, which are generated by alternative exon splicing of a single VEGF gene (Ferrara and DavisSmyth, 1997). In the human endometrium, expression of four of these isoforms has been identified, VEGF121, VEGF145, © European Society of Human Reproduction and Embryology

VEGF and angiogenesis in human endometrium

VEGF165 and VEGF189 (Charnock-Jones et al., 1993; Torry et al., 1996), although the predominant isoforms expressed are the secreted forms, VEGF121 and VEGF165 (Smith, 1995). Attempts have been made to relate VEGF expression in the human endometrium to stages of the menstrual cycle and, while a generalized picture is emerging, there is still some disagreement (Torry and Torry, 1997; Rogers and Gargett, 1999). From mRNA studies (in-situ hybridization and Northern blots), it appears that VEGF expression in whole endometrium is low during the proliferative phase, increases during the late secretory phase, and reaches a maximum at menses (CharnockJones et al., 1993; Shifren et al., 1996; Graubert et al., 1997). However, there is no clear pattern of VEGF protein expression across the menstrual cycle, although expression is greater in glands than stroma (Li et al., 1994; Shifren et al., 1996; Lau et al., 1999). Most studies of VEGF expression in the human endometrium have been descriptive, and only two had large enough sample numbers for meaningful statistical analysis (Graubert et al., 1997; Lau et al., 1999). The relative ability of glands and stroma to secrete VEGF in vitro has not been determined. Furthermore, no study has examined the relationship between VEGF expression across the menstrual cycle with indices or markers of endometrial angiogenesis. The aims of the present study were to: (i) measure in-vitro production of VEGF by human endometrial tissues taken at four stages of the menstrual cycle; (ii) relate VEGF production by cultured endometrial cells to endothelial cell proliferation in the same tissue samples; and (iii) relate VEGF expression as measured by immunohistochemistry to endothelial cell proliferation in the same tissue, using sufficient sample numbers for each stage of the cycle to allow for meaningful statistical analysis.

Cell isolation and culture Explants (4310 mg each, weighed after blotting to remove all fluid) of finely chopped tissue were cultured in 0.5 ml phenol red-free DMEM/Ham’s F-12 medium containing 0.1% bovine serum albumin (BSA; Gibco BRL) for 24 h at 37°C in a humidified incubator containing 5% CO2 in air for production of conditioned medium (CM). Glandular epithelial and stromal cells were separated from remaining chopped endometrial tissue using a modified method (Marsh et al., 1994). Briefly, tissue was digested with 45 U/ml collagenase type III (Worthington Biochemical Corporation, Freehold, NJ, USA) and 16 µg/ml deoxyribonuclease type I (Boehringer Mannheim GmbH, Mannheim, Germany) in Ca21- and Mg21-free phosphate-buffered saline (PBS; pH 7.4) for 30 min at 37°C and filtered sequentially through 77 µm and 10 µm nylon mesh (Henry Simon & Co., Stockport, UK) filters to collect gland fragments and stromal cells respectively. Erythrocytes were removed from stromal cell preparations by centrifugation over Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden). Gland fragments were recovered from the filters by backwashing. Washed stromal and gland cell preparations were resuspended in DMEM/Ham’s F-12 medium containing 10% charcoal-treated fetal calf serum (FCS; PA Biologicals, Sydney, Australia), gentamicin (50 µg/ml), fungizone (0.5 µg/ml) and 2 mM glutamine (Sigma Chemical Co., St Louis, MO, USA) and seeded in quadruplicate in 24-well plates at 33105 cells/well and 4–103103 glands/well respectively. The purity of the cultures was .95%, confirmed by immunohistochemical staining for cytokeratin using mouse anti-human cytokeratin antibody (Clone MNF116; Dako, Cambridge, UK) and vimentin with mouse anti-vimentin antibody (Clone V9; Zymed, San Francisco, CA, USA), to detect epithelial and stromal cells respectively. Glandular epithelial cells were cultured for 4 days and stromal cells until confluent (2–6 days) at 37°C in 5% CO2, with medium changes every 48 h. For collection of CM, cultures were rinsed and the medium changed to 0.5 ml/well serum-free DMEM/Ham’s F-12 medium containing 0.1% BSA, antibiotics and glutamine, and cultures were incubated for 24 h. The CM was harvested, centrifuged at 7500 g for 1 min and the supernatant collected and stored at –80°C.

Materials and methods Human tissues Endometrial biopsies were obtained from a total of 24 normal ovulating women (aged 20–40 years) undergoing laparoscopy for investigation of infertility not related to uterine factors, or undergoing tubal ligation. Informed consent was obtained from each patient and ethical approval was obtained from the Monash Medical Centre Human Research and Ethics Committee. Samples included in the study were from women who had not taken exogenous hormones in the previous 3 months, who showed a normal pelvic and uterine cavity at laparoscopy or microhysteroscopy, and whose biopsies were histologically normal. The samples were categorized into four groups on the basis of menstrual histories, and by dating of biopsies by an experienced histopathologist using established criteria for the normal menstrual cycle (Noyes et al., 1950). The groups were early proliferative (n 5 5), mid-late proliferative (n 5 8), early-mid secretory (n 5 7), and late secretory (n 5 4). A small portion of each biopsy was fixed for 4 h in 10% phosphatebuffered formalin (pH 7.4), for routine paraffin embedding. Serial sections (5 µm) were used for immunohistochemical analysis and staining with haematoxylin and eosin for dating. The remaining biopsy was collected into phenol red-free HEPES-buffered Dulbecco’s minimum essential medium Ham’s F-12 (DMEM/Ham’s F-12) culture medium (Gibco BRL, Gaithersburg, MD, USA) containing 50 µg/ml gentamicin (Gibco BRL) and 0.5 µg/ml fungizone (Gibco BRL).

Cell counts Cells numbers per well were measured after CM collection using the CellTitre96 MTS tetrazolium-based bioassay (Promega, Madison, WI, USA), in which the absorbance of the formazan product is directly proportional to the number of viable cells. Briefly, CM was replaced with 250 µl of standard DMEM/Ham’s F-12/10% charcoal-treated FCS medium, 50 µl 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulphophenyl)-2H-tetrazolium (MTS) solution was added, and the cells incubated for 75 min at 37°C. Two 120 µl aliquots were transferred to wells of a 96-well plate, the absorbance measured at 490 nm using a plate reader (Emax, Molecular Devices, Sunnyvale, CA, USA) and analysed with the associated software, SOFTmax, version 2.32. A standard curve was produced for each assay by seeding serial dilutions of passaged endometrial stromal cells (23104) in 100 µl volumes of DMEM/Ham’s F-12/10% charcoaltreated FCS into wells of a 96-well plate. The cells were allowed to attach for 90 min before addition of 20 µl of MTS reagent, incubated for a further 75 min, and the absorbance measured. Gland cell standard curves using primary cultured glandular epithelial cells gave slopes matching those of stromal cells, indicating that the two cell types metabolize the MTS at a similar rate. Values ranged from 73103 to 673103/well for glandular epithelial cell cultures, and from 203103 to 1003103/well for stromal cell cultures, depending on the degree of confluence. In some cultures, cell numbers were estimated

2081

C.E.Gargett et al.

using the established relationship between cell number/well and level of confluence. Enzyme-linked immunosorbent assay (ELISA) for measurement of VEGF in conditioned medium Purified recombinant mouse VEGF164 (a gift from Dr S.Stacker, Ludwig Institute, Melbourne, Australia) was used to immunize rabbits for production of polyclonal anti-VEGF antibodies by standard methods. The titre and specificity of the anti-VEGF response was evaluated by inhibition of VEGF-mediated proliferation of VEGFR2/ EpoR bioassay cells and Western blotting respectively. The IgG fraction of the VEGF antiserum was obtained by affinity chromatography using a protein A–agarose column (Pharmacia Biotech). Biotinylated anti-VEGF IgG was obtained by conjugation with sulpho-biotin-X-NHS (Calbiochem-Novabiochem GmbH, Bad Soden, Germany) for 1 h at 24°C using a 25 molar excess of biotin to IgG. ELISA were performed in Maxisorb 96-well plates (Nunc, Roskilde, Denmark), which were coated with anti-VEGF antibody (2.5 µg/ml) in 50 mmol/l carbonate buffer, pH 9.6 for 20 h at 4°C and blocked with 3% BSA in PBS (0.5 mol/l NaCl, 0.01 mol/l phosphate) overnight at 4°C. Plates were washed in PBS/Tween (0.1% Tween 20), which was also used for subsequent washing steps between addition of reagents. Standards, rhVEGF165 (PeproTech, Rocky Hill, NJ, USA) (0.16–10 ng/ml) and CM samples, neat and diluted 1:1, were then added and incubated for 2 h at room temperature with gentle mixing. Bound VEGF was detected with biotinylated anti-VEGF (1.25 µg/ml in 3% BSA/PBS; incubation 2 h) and streptavidin–horseradish peroxidase (HRP) (Zymed, diluted 1/10 000; incubation 30 min) and the substrate was 3,39,5,59-tetramethylbenzidine dihydrochloride (Sigma) (0.1 mg/ml in 0.05 M phosphate citrate buffer, pH 5.0 containing 0.006% vol/vol H2O2). After a 7 min incubation, the reaction was terminated with 1 M HCl and absorbance measured at 450 nm using a microplate reader. All CM samples from a given patient were analysed on the one plate. The standard curve was fitted by a fourparameter, non-linear regression-fitting program (SOFTMax version 2.32) and was linear between 0.5 and 10 ng/ml VEGF. Intraand interassay variability were ,11% and ,15% respectively. The technique was sensitive to 0.30 ng/ml. Serially diluted CM samples produced curves parallel with the standard curve. This ELISA detected both mouse VEGF164 and human VEGF165 and VEGF121, but no VEGF was detected when pre-immune rabbit serum was used to coat the plates. Immunohistochemistry Proliferating endothelial cells were detected by a standard double immunohistochemistry protocol using mouse anti-rat antibody to proliferating cell nuclear antigen (PCNA) (clone PC10; Novacastra, Newcastle, UK) and a mouse monoclonal antibody against the cluster determination (CD)34 antigen (clone QBEND/10; Serotec, Oxford, UK), an endothelial cell marker, according to a previously published protocol (Goodger (MacPherson) and Rogers, 1994). All other reagents were from Zymed. A sequential protocol was used with the anti-PCNA as the first primary antibody. Both primary antibodies were applied for 1 h at 37°C, followed by incubation with the same biotinylated secondary antibody (rabbit anti-mouse immunoglobulin). A streptavidin–HRP conjugate with AEC chromogen (red) was used for PCNA staining, and a streptavidin–alkaline phosphatase conjugate with AP-blue chromogen was used for CD34 staining. Endogenous peroxidase was quenched with 3% H2O2 in 50% methanol. Positive and negative control sections were included in each staining run as described (Goodger (MacPherson) and Rogers, 1994). At least 100 microvessel profiles were counted in sequential fields scanned at 3400 magnification from the surface epithelium through the full depth

2082

of the section. The number of these vessels containing proliferating endothelial cells was also determined and the percentage of microvessels containing proliferating endothelial cells calculated. VEGF protein was detected using a rabbit anti-human polyclonal anti-VEGF antibody (Sc-152; Santa Cruz Biotechnology Inc, Santa Cruz, CA, USA) as previously described (Lau et al., 1999). Briefly, sections were boiled for 10 min in 10 mmol/l sodium citrate buffer, pH 6 for antigen retrieval, endogenous peroxidase quenched, and sections blocked with 10% normal goat serum. The primary antibody (2 µg/ml) was applied for 2 h at 37°C, followed by sequential incubations with biotinylated secondary antibody, streptavidin–HRP conjugate and AEC chromogen. Positive and negative controls were included in every staining run as described (Lau et al., 1999). Several images per section were captured using a digital video camera (Fujix, Fuji, Tokyo, Japan) and the same light intensity settings for all sections. For each captured image, the staining intensity of the entire glandular area and the stromal area was measured by Analytical Imaging Station™ software (AIS, Image Research Inc., Ontario, Canada) and corrected for background staining. The average staining intensity of the glandular and stromal regions of several captured images was then calculated for each sample. Similarly, for each captured image, the total area occupied by glands and by stroma was also recorded and the average proportional area for each sample calculated. A staining intensity index (SI) was then obtained for the glandular and stromal compartments of each sample by multiplying the average staining intensity by the average proportion of stained area. Thus, the SI represents the total glandular and total stromal VEGF production for each sample as measured by immunostaining. Interassay variation was ,19% (Lau et al., 1999). Statistical analysis Analyses were performed using SPSS, version 6.1.3 (SPSS, Australasia, North Sydney, Australia). The Levene homogeneity test (Snedecor and Cochran, 1989) was used to measure variance and the Kolmogorov–Smirnov–Lilliefors/Shapiro–Wilk test was used to test for normality and to determine whether parametric or non-parametric tests were to be used for further analyses. The Kruksal–Wallis one-way analysis of variance (ANOVA) was used to examine for differences between stages of the menstrual cycle and the Mann– Whitney U-test was used to confirm a difference between the early proliferative stage and each of the other stages. The Wilcoxon matched-pairs signed-rank test was used for comparing paired data (glands versus stroma), correlations were performed using least squares regression analysis, and Spearman’s correlation coefficients (RS) were determined. Some of the data were also transformed into natural logarithms and compared by one-way ANOVA. Results were considered statistically significant when P , 0.05.

Results Cultured endometrial cells and tissue secrete VEGF There was wide variation in the amount of VEGF detected in the CM obtained from individual cultured endometrial explants and glandular epithelial cell samples taken from the same stage of the cycle. This variability is illustrated by the wide spread of data shown in the box and whisker plots (Figure 1A and B). In contrast, cultured stromal cells showed little variation between samples within cycle stages (Figure 1C). There was no significant difference between the stages of the cycle and the amount of VEGF secreted by explant (P 5 0.87), glandular epithelial cell (P 5 0.80) and stromal cell (P 5 0.23) cultures.

VEGF and angiogenesis in human endometrium

Table I. Vascular endothelial growth factor (VEGF) production in vitro and in vivo by endometrial glandular epithelium or stromal cells Endometrial component

Glandular epithelium Stromal cells P-value

VEGF Conditioned medium ng/105 cells (n 5 20)

Fixed tissue SI (3102) (n 5 24)

1.77 6 0.46 0.46 6 0.06 0.002

5.77 6 0.60 4.93 6 0.50 0.14

VEGF was measured in conditioned medium from glandular epithelial and stromal cell cultures by enzyme-linked immunosorbent assay (ELISA). The staining intensity index (SI) of VEGF immunostained samples was measured by computer-assisted image analysis of the glandular and stromal compartments. Data are expressed as mean 6 SEM of n samples.

Endometrial VEGF expression during the menstrual cycle VEGF protein in fixed tissue from the same samples as those cultured for production of CM was assessed by immunohistochemistry and image analysis. As with in-vitro production of VEGF, large variation was observed between samples taken from the same stage of the menstrual cycle for both glands and stromal compartments (Figure 2). However, there was a significant difference between the stages of the menstrual cycle for the VEGF SI of the stromal cell compartment (P 5 0.03). The stromal VEGF SI was significantly greater in the early proliferative than for mid-late proliferative (P 5 0.005) and early-mid secretory (P 5 0.01) stages. The staining intensity of VEGF in the glandular compartment was higher in the early proliferative phase compared with the other phases, but this difference was not significant across the menstrual cycle (P 5 0.17) due to the greater variability of immunostaining observed in the glands. Transformation of the glandular data did not affect the statistical significance of the result (F3,19 5 1.30; P 5 0.18; one-way ANOVA). By immunohistochemistry, glandular SI was greater than that for stroma, but the difference was not significantly different (Table I). Figure 3A and B show the stronger VEGF immunostaining of glands compared with the stroma. In some endometrial samples, intense VEGF immunostaining was observed in some cells of the stromal compartment (Figure 3C) and in the majority of samples examined some blood vessels also showed intense reactivity (Figure 3B). No correlation was demonstrated between the VEGF SI for glands or stroma and the amount of VEGF secreted by the respective cultured cells (results not shown). Figure 1. Box and whisker plots of vascular endothelial growth factor (VEGF) production by cultured human endometrium taken at different stages of the menstrual cycle. (A) Endometrial explants; (B) glandular epithelial cells; (C) stromal cells. VEGF (ng/ml) in conditioned medium was measured by ELISA, normalized to wet weight of tissue or 105 cells for explants and cultured cells respectively. Means 6 SEM for (n) samples at each cycle stage are shown above each box plot. For each box plot, solid bar indicates median; upper and lower limits of box, 75th and 25th percentiles; upper and lower bars, maximum and minimum values; s and *, outliers and extremes. ep 5 early proliferative; lp 5 mid-late proliferative; es 5 early-mid secretory; ls 5 late secretory.

However, Table I shows that cultured glandular epithelial cells secreted significantly more VEGF in CM than stromal cells (P 5 0.002).

Endothelial cell proliferation in the endometrium The percentage of PCNA-positive microvessels was highly variable between individual samples within each of the cycle stages (Figure 4). No significant difference in the percentage of proliferating vessels across the menstrual cycle was observed (P 5 0.09), although there was a suggestion that endothelial proliferation was greater during the proliferative stages. Transformation normalized the data, but did not influence the significance of the result (F3,20 5 2.43; P 5 0.10; oneway ANOVA). No correlation was demonstrated between the proportion of vessels containing proliferating endothelial cells and the in-vitro production of VEGF by cultured explants (RS 5 –0.35, P 5 0.17), glandular epithelial cells (RS 5 2083

C.E.Gargett et al.

Figure 2. Box and whisker plots of VEGF immunostained human endometrium taken at various stages of the menstrual cycle. Staining indices were measured by computer-assisted image analysis in (A) glands and (B) stromal compartments. Means 6 SEM for (n) samples at each cycle stage are shown above each box plot. Key to box plots and abbreviations as for Figure 1.

–0.11, P 5 0.62) and stromal cells (RS 5 –0.06, P 5 0.78) (Figure 5). Neither was there a correlation between the percentage of proliferating vessels and VEGF immunostaining indices (SI) for the glandular (RS 5 –0.06, P 5 0.79) or stromal (RS 5 0.17, P 5 0.42) compartments (Figure 6). Discussion In the first study to examine the relationship between VEGF production and endothelial cell proliferation in human endometrium, we report a lack of statistical correlation between in-vitro or in-vivo VEGF production by human endometrial cells or tissue and endothelial cell proliferation. In contrast, a descriptive comparison between VEGF immunolocalization and endothelial cell proliferation in human ovary demonstrated a close relationship (Yamamoto et al., 1997). The lack of correlation between overall VEGF production and endothelial cell proliferation in human endometrium suggests one or more of the following: (i) that the majority of human endometrial VEGF has a non-angiogenic role; (ii) that a minor component 2084

of endometrial VEGF, possibly located in discrete foci, may regulate blood vessel growth in a paracrine/autocrine manner; (iii) that VEGF is not the primary regulator of endothelial cell proliferation in human endometrium; (iv) that VEGF secretion and endothelial cell proliferation are not related temporally; or (v) that the receptors for VEGF are regulated and control endothelial cell proliferation in human endometrium. Alternatively, between-subject variability for the parameters measured may actually mask a relationship between VEGF production and endothelial cell proliferation in human endometrium. However, it is noteworthy that those endometrial samples with very high percentages of proliferating vessels failed to show correspondingly high amounts of VEGF production by either glandular or stromal components in vivo or by cultured cells (Figures 5 and 6). Conversely, samples producing the highest amounts of VEGF in vivo or in vitro had modest or low numbers of PCNA-positive vessels. This evidence supports our conclusion that there is no relationship between overall endometrial VEGF and endothelial cell proliferation. However, this study is limited in that it does not allow for a comparison of VEGF production and endothelial cell proliferation between the basalis and functionalis layers of the endometrium because endometrial biopsies, rather than hysterectomy samples, were used. We have previously reported lower endothelial cell proliferation indices in the basalis compared with the functionalis of human endometrium (Rogers et al., 1998). It is well known that VEGF has functions other than inducing endothelial cell proliferation. In the endometrium, VEGF may mediate increases in vascular permeability (Dvorak et al., 1995; Torry et al., 1996), phenotypic changes observed in microvessel endothelial cells (Roberts et al., 1992), or provide trophic signals to the implanting embryo (Hornung et al., 1998). VEGF may act in a paracrine/autocrine manner and therefore its secretion would be localized. The observation of small, discrete patches of intense VEGF immunoreactivity associated with some blood vessels and also in a population of stromal cells in most of the samples examined supports this hypothesis (Figure 3). These foci of VEGF immunoreactivity were not included in the calculation of the SI, since their quantification requires methods other than described in this study, nor could they be examined in cultured endometrial cells. Localized VEGF immunostaining in perivascular stromal cells and endothelial cells has previously been observed in endometrium (Li et al., 1994), while focal endothelial cell proliferation has been reported for pathological angiogenesis (Walsh et al., 1998). The source of focally produced VEGF in human endometrium may be pericytes (Nomura et al., 1995; Reynolds and Redmer, 1998), vascular smooth muscle cells (Gu and Adair, 1997; Abramovitch et al., 1998) in the spiral arterioles, endothelial cells (Namiki et al., 1995; Nomura et al., 1995), or adjacent stromal cells. It is likely that endometrial angiogenesis is regulated by a complex interaction between angiogenesis promoters and inhibitors (Folkman and Shing, 1992). In this study, a correlation between endothelial cell proliferation and bulk changes in glandular and stromal VEGF was not observed, suggesting that factors other than VEGF may also be involved. Synergism between VEGF and its receptors, VEGFR-1 (flt-1) and

VEGF and angiogenesis in human endometrium

Figure 3. VEGF immunostained human endometrium showing (A) strong reactivity in the basal region of glandular epithelial cells and weaker stromal reactivity in a sample taken during the early secretory phase of the menstrual cycle (magnification 3400); (B) higher-power view (magnification 31000) of the same sample showing focal VEGF in the endothelium of a capillary (arrow); and (C) focal VEGF in single stromal cells from a late proliferative endometrium. (D) Negative control. All scale bars 5 50 µm.

VEGFR-2 (KDR), and basic fibroblast growth factor (bFGF), all of which are expressed in human endometrium (Ferriani et al., 1993), may have a role in endometrial angiogenesis. In addition, changing levels of endogenous angiogenesis inhibitors across the menstrual cycle, such as thrombospondin-1 (IruelaArispe et al., 1996), may influence endothelial cell proliferation. VEGF secretion and endometrial angiogenesis detected by PCNA immunoreactivity may occur sequentially due to transient or pulsatile VEGF secretion. VEGF, which acts during G0/G1 may no longer be detectable when endothelial cells become PCNA-positive during late G1 and S phase. Samples shown as outliers or extremes in Figures 1, 2 and 4 may represent transient peaks of VEGF secretion or angiogenic activity. Both hypoxia and oestrogen can rapidly upregulate VEGF within an hour, with return to baseline values within 24 hours of removal of the stimulus (Cullinan-Bove and Koos, 1993; Gu and Adair, 1997; Yasuda et al., 1998). A second finding in the present study was that in vivo, VEGF was significantly elevated during the early proliferative phase (Figure 2) of the menstrual cycle. However, no relationship was observed between in-vitro VEGF production by cultured endometrial explant, glandular epithelial or stromal cells and the cycle stage when the tissue was collected (Figure 1). Two other studies, using large sample numbers, have also demonstrated a trend to higher stromal VEGF concentrations during the proliferative stages (Li et al., 1994; Lau et al., 1999), although one study of unknown sample number demonstrated increasing VEGF mRNA and protein from early proliferative

Figure 4. Box and whisker plots of the percentage of proliferating microvessels in human endometrium taken at various stages of the menstrual cycle. Sections were double-immunostained for PCNA and CD34 to mark proliferating endothelial cells and blood vessels respectively, and the percentage of vessels containing PCNApositive endothelial cells of at least 100 vessels determined by counting at 3400 magnification. Means 6 SEM for (n) samples at each cycle stage are shown above each box plot. Key to box plots and abbreviations as for Figure 1.

to late secretory endometrium (Shifren et al., 1996). Inconsistencies in patterns of VEGF expression in glands and stroma between the various stages of the menstrual cycle may relate to the descriptive nature of some of these studies, as well as the low sample numbers used, considering the degree of between-subject variability. Only two studies were sufficiently large or quantitative to report on variability between individual 2085

C.E.Gargett et al.

Figure 6. The VEGF staining index (SI) for (A) glands or (B) stromal compartments of normal human endometrium was not correlated with the percentage of vessels containing proliferating endothelial cells.

Figure 5. VEGF production by (A) explants, (B) glandular epithelial or (C) stromal cell cultures from normal human endometrium was not correlated with the percentage of vessels containing proliferating endothelial cells.

subjects (Torry et al., 1996; Lau et al., 1999). In this study, we have used moderate numbers of samples to allow comparison between four main stages of the menstrual cycle, although menstrual samples were not examined because there was insufficient tissue for culture and immunohistochemical studies. Elevated VEGF mRNA observed in secretory and menstrual stages (Charnock-Jones et al., 1993; Shifren et al., 1996; Torry et al., 1996; Graubert et al., 1997) probably reflects glandular expression, since considerably more VEGF has been reported in glandular epithelial compared with stromal cells (Table I) 2086

(Li et al., 1994; Shifren et al., 1996; Torry et al., 1996; Lau et al., 1999). Since the bulk of endometrial VEGF is glandular, and 80% of this VEGF is secreted from the luminal surface (Hornung et al., 1998), it is unlikely that most of the VEGF produced in the glands has a role in endometrial angiogenesis. A third observation of the present study was the large degree of variation between subjects for each of the parameters measured. Our laboratory has consistently reported high variability for endothelial cell proliferation indices, and has recently demonstrated that 84% of this variability was attributable to between-sample variation, while only 16% was due to sampling variation within the same uterus (Rogers et al., 1998). Difficulties in obtaining consistency between staining runs may also contribute to the particularly high variability observed in VEGF staining indices, although captured images were always analysed relative to the same positive control section. Nevertheless, immunoreactive VEGF was significantly elevated during the early proliferative phase. In summary, our results suggest that ovarian steroids, which

VEGF and angiogenesis in human endometrium

control the characteristic growth cycle of human endometrium, are not the main regulators of the majority of endometrial VEGF production, or of angiogenesis. While oestrogen and progesterone have been shown to increase VEGF protein secretion in cultured stromal cells, these effects have been modest (Shifren et al., 1996; Huang et al., 1998). It is uncertain whether the effect of oestrogen in these studies is direct or indirect, since the VEGF gene does not contain oestrogen or progesterone response elements in its promoter region, although there are several half-palindromes for oestrogen (Tischer et al., 1991; Cullinan-Bove and Koos, 1993). It is possible that other factors such as hypoxia or locally produced cytokines also play a role in regulating angiogenesis during the menstrual cycle. Our results have demonstrated no obvious correlation between overall VEGF production and endothelial cell proliferation for individual endometrial samples, despite the observed increases in stromal VEGF immunostaining and percentage of proliferating vessels (though not statistically significant) during the early proliferative stage, when angiogenesis undoubtedly occurs. Other approaches are still required to provide further understanding of the regulation of endometrial angiogenesis.

Acknowledgements We thank Dr Mark Lawrence from Monash Medical Centre, Dr Bev Vollenhoven and Prof. David Healy from Monash University for obtaining patient biopsies; Dr Beatrice Susil from the Department of Anatomical Pathology, Monash Medical Centre for dating the biopsies, and Dr Steven Stacker from the Angiogenesis Laboratory, Ludwig Institute for Cancer Research, Melbourne, for assessing the titre and specificity of the anti-VEGF antibody. This study was supported by Grant No. 960041 from the Australian National Health & Medical Research Council.

References Abramovitch, R., Neeman, M., Reich, R. et al. (1998) Intercellular communication between vascular smooth muscle and endothelial cells mediated by heparin-binding epidermal growth factor-like growth factor and vascular endothelial growth factor. FEBS Lett., 425, 441–447. Carmeliet, P., Ferreira, V., Breier, G. et al. (1996) Abnormal blood vessel development and lethality in embryos lacking a single VEGF allele. Nature, 380, 435–439. Charnock-Jones, D.S., Sharkey, A.M., Rajput-Williams, J. et al. (1993) Identification and localisation of alternately spliced mRNAs for vascular endothelial factor in human uterus and estrogen regulation in endometrial carcinoma cell lines. Biol. Reprod., 48, 1120–1128. Cullinan-Bove, K. and Koos, R.D. (1993) Vascular endothelial growth factor/ vascular permeability factor expression in the rat uterus: rapid stimulation by estrogen correlates with estrogen-induced increases in uterine capillary permeability and growth. Endocrinology, 133, 829–837. Dvorak, H.F., Brown, L.F., Detmar, M. et al. (1995) Vascular permeability factor/vascular endothelial growth factor, microvascular hyperpermeability, and angiogenesis. Am. J. Pathol., 146, 1029–1039. Ferrara, N. and DavisSmyth, T. (1997) The biology of vascular endothelial growth factor. Endocr. Rev., 18, 4–25. Ferrara, N., Carver-Moore, K., Chen, H. et al. (1996) Heterozygous embryonic lethality induced by targeted inactivation of the VEGF gene. Nature, 380, 439–442. Ferrara, N., Chen, H., DavisSmyth, T. et al. (1998) Vascular endothelial growth factor is essential for corpus luteum angiogenesis. Nature Med., 4, 336–340. Ferriani, R.A., Charnock-Jones, D.S., Prentice, A. et al. (1993) Immunohistochemical localization of acidic and basic fibroblast growth factors in normal human endometrium and endometriosis and the detection of their mRNA by polymerase chain reaction. Hum. Reprod., 8, 11–16.

Folkman, J. and Shing, Y. (1992) Angiogenesis. J. Biol. Chem., 267, 10931–10934. Goodger (MacPherson), A.M. and Rogers, P.A.W. (1994) Endometrial endothelial cell proliferation during the menstrual cycle. Hum. Reprod., 9, 399–405. Gordon, J.D., Shifren, J.L., Foulk, R.A. et al. (1995) Angiogenesis in the human female reproductive tract. Obstet. Gynecol. Survey, 50, 688–697. Graubert, M.D., Mortola, J.F., Kessel, B. et al. (1997) Upregulation of vascular endothelial growth factor in the menstrual phase of the normal endometrium may be mediated by hypoxia. J. Soc. Gynecol. Invest., 40, 80A. Greb, R.R., Heikinheimo, O., Williams, R.F. et al. (1997) Vascular endothelial growth factor in primate endometrium is regulated by oestrogen-receptor and progesterone-receptor ligands in vivo. Hum. Reprod, 12, 1280–1292. Gu, J.W. and Adair, T.H. (1997) Hypoxia-induced expression of VEGF is reversible in myocardial vascular smooth muscle cells. Am. J. Physiol., 273, H628–H633. Hornung, D., Lebovic, D.I., Shifren, J.L. et al. (1998) Vectorial secretion of vascular endothelial growth factor by polarized human endometrial epithelial cells. Fertil. Steril., 69, 909–915. Huang, J.C., Liu, D.Y. and Dawood, M.Y. (1998) The expression of vascular endothelial growth factor isoforms in cultured human endometrial stromal cells and its regulation by 17β-oestradiol. Mol. Hum. Reprod., 4, 603–607. IruelaArispe, M.L., Porter, P., Bornstein, P. et al. (1996) Thrombospondin-1, an inhibitor of angiogenesis, is regulated by progesterone in the human endometrium. J. Clin. Invest., 97, 403–412. Kaiserman-Abramof, I.R. and Padykula, H.A. (1989) Angiogenesis in the postovulatory primate endometrium: the coiled arteriolar system. Anat. Rec., 224, 479–489. Kim, K.J., Li, B., Winer, J. et al. (1993) Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumor growth in vivo. Nature, 362, 841–844. Klauber, N., Rohan, R.M., Flynn, E. et al. (1997) Critical components of the female reproductive pathway are suppressed by the angiogenesis inhibitor AGM-1470. Nature Med., 3, 443–446. Lau, T.M., Affandi, B. and Rogers, P.A.W. (1999) The effects of levonorgestrel implants on vascular endothelial growth factor expression. Mol. Hum. Reprod., 5, 57–63. Li, X.F., Gregory, J. and Ahmed, A. (1994) Immunolocalisation of vascular endothelial growth factor in human endometrium. Growth Factors, 11, 277–282. Markee, J.E. (1940) Menstruation in intraocular endometrial transplants in the rhesus monkey. Contrib. Embryol., 177, 219–308. Marsh, M.M., Hampton, A.L., Riley, S.C. et al. (1994) Production and characterization of endothelin released by human endometrial epithelial cells in culture. J. Clin. Endocrinol. Metab., 79, 1625–1631. Namiki, A., Brogi, E., Kearney, M. et al. (1995) Hypoxia induces vascular endothelial growth factor in cultured endothelial cells. J. Biol. Chem., 270, 31189–31195. Nomura, M., Yamagishi, S., Harada, S. et al. (1995) Possible participation of autocrine and paracrine vascular endothelial growth factors in hypoxiainduced proliferation of endothelial cells and pericytes. J. Biol. Chem., 270, 28316–28324. Noyes, R.W., Hertig, A.T. and Rock, J. (1950) Dating the endometrial biopsy. Fertil. Steril., 1, 3–25. Reynolds, L.P. and Redmer, D.A. (1998) Expression of the angiogenic factors, basic fibroblast growth factor and vascular endothelial growth factor, in the ovary. J. Anim. Sci., 76, 1671–1681. Risau, W. (1997) Mechanisms of angiogenesis. Nature, 368, 671–674. Roberts, D.K., Parmley, T.H., Walker, N.J. et al. (1992) Ultrastructure of the microvasculature in the human endometrium throughout the normal menstrual cycle. Am. J. Obstet. Gynecol., 166, 1393–1406. Rogers, P.A.W. and Gargett, C.E. (1999) Endometrial angiogenesis. Angiogenesis, 3, (in press). Rogers, P.A.W., Au, C.L. and Affandi, B. (1993) Endometrial microvascular density during the normal menstrual cycle and following exposure to longterm levonorgestrel. Hum. Reprod., 8, 1396–1404. Rogers, P.A.W., Lederman, F. and Taylor, N. (1998) Endometrial microvascular growth in normal and dysfunctional states. Hum. Reprod Update, 4, 503–508. Saleh, M., Stacker, S.A. and Wilks, A.F. (1996) Inhibition of growth of C6 glioma cells in vivo by expression of antisense vascular endothelial growth factor sequence. Cancer Res., 56, 393–401.

2087

C.E.Gargett et al. Shifren, J.L., Tseng, J.F., Zaloudek, C.J. et al. (1996) Ovarian steroid regulation of vascular endothelial growth factor in the human endometrium: implications for angiogenesis during the menstrual cycle and in the pathogenesis of endometriosis. J. Clin. Endocrinol. Metab., 81, 3112–3118. Smith, S.K. (1995) Angiogenic growth factor expression in the uterus. Hum. Reprod. Update, 1, 162–172. Snedecor, G.W. and Cochran, W.G. (1989) Statistical Methods. 8th edn. Iowa State University Press, Ames, Iowa. Tischer, E., Mitchell, R., Hartman, T. et al. (1991) The human gene for vascular endothelial growth factor. Multiple protein forms are encoded through alternative exon splicing. J. Biol. Chem., 266, 11947–11954. Torry, D.S. and Torry, R.J. (1997) Angiogenesis and the expression of vascular endothelial growth factor in endometrium and placenta. Am. J. Reprod. Immunol., 37, 21–29. Torry, D.S., Holt, V.J., Keenan, J.A. et al. (1996) Vascular endothelial growth factor expression in cycling human endometrium. Fertil. Steril., 66, 72–80. Walsh, D.A., Wade, M., Mapp, P.I. et al. (1998) Focally regulated endothelial proliferation and cell death in human synovium. Am. J. Pathol., 152, 691–702. Yamamoto, S., Konishi, I., Tsuruta, Y. et al. (1997) Expression of vascular endothelial growth factor (VEGF) during folliculogenesis and corpus luteum formation in the human ovary. Gynecol. Endocrinol., 11, 371–381. Yasuda, Y., Masuda, S., Chikuma, M. et al. (1998) Estrogen-dependent production of erythropoietin in uterus and its implication in uterine angiogenesis. J. Biol. Chem., 273, 25381–25387. Received on December 21, 1998; accepted on April 30, 1999

2088