Regulation of Angiogenic Activity of Human EndometrialEndothelial Cells in Culture by Ovarian Steroids

UMIT A. KAYISLI, JANELLE LUK, OZLEM GUZELOGLU-KAYISLI, YASEMIN SEVAL,RAMAZAN DEMIR, AND AYDIN ARICI

Department of Obstetrics and Gynecology, Yale University School of Medicine (U.A.K, J.L., O.G.-K., Y.S., A.A.), New Haven,Connecticut 06520-8063; and Departments of Histology and Embryology (U.A.K., Y.S., R.D.) and Medical Biology andGenetics (O.G.-K.), Akdeniz University School of Medicine, Antalya 07070, Turkey

Blood vessel growth and regression in human endometriumare regulated throughout the menstrual cycle. We sought adirect role of ovarian steroids on human endometrial endo-thelial cell (HEEC) proliferation and vascularization. To in-vestigate the HEEC angiogenicity of sex steroids, we devel-oped a reliable method for the isolation of HEEC, whichallowed us to investigate the angiogenic effects of sex steroidsusing immunohistochemistry, immunocytochemistry, West-ern blot, 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxy-phenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt prolifer-ation, and vascular tube formation analyses. We were able toobtain 95–99% pure HEEC with our isolation technique. HEECexpressed predominantly estrogen receptor �, minimally ex-pressed estrogen receptor �, and but did not express proges-

terone (P4) receptors A and B in vivo and in vitro. Estradiol (E2;10�10–10�8 M) and P4 (10�12–10�8 M), alone or in combination,induced HEEC proliferation compared with control valuesafter 48 h of treatment (P < 0.05). Furthermore, after 8 d oftreatment, there were significantly more angiogenic patternsin E2 (10�8 M), P4 (10�10 M), and E2 plus P4 (10�8 and 10�10 M)treatment groups compared with the control group (angio-genic scores, 2.95 � 0.16, 3.26 � 0.16, 3.06 � 0.17, and 1.93 � 0.15,respectively; P < 0.01). In conclusion, our results suggest thatthere are direct effects of E2 and P4 on HEEC and provide anew understanding of the physiological role of sex steroids inthe regulation of endometrial events such as angiogenesis.(J Clin Endocrinol Metab 89: 5794–5802, 2004)

THE ENDOMETRIUM GROWS at a remarkable rate dur-ing much of the menstrual cycle and is capable of

blastocyst implantation, regulation of trophoblast invasion,control of infectious agents, and efficient disposal of bloodand desquamated cellular debris with menstruation. All ofthese changes occur in parallel with the coordinated involve-ment of angiogenic and angiostatic alterations in endome-trial endothelial cells (1, 2).

Angiogenesis is a process of new microvessels emergingfrom existing blood vessels. Physiological angiogenesis takesplace during placental development after the vasculogenicstage and during fetal development. Physiological angio-genesis rarely occurs in postnatal life, except for woundhealing and in tissues of the human female reproductivetract, such as corpus luteum and endometrium, influencedby steroid-dependent growth and/or regression (3–5). How-ever, pathological angiogenesis occurs in many abnormaltissue growths, such as malignancy and endometriosis (6).

Endothelial cell proliferation varies throughout the men-strual cycle independently from the stage of the cycle (5).However, the necessity for new vessel growth and its re-

gression in the human endometrium varies spatially andtemporally in relation to the menstrual cycle phase. Differ-ently from classical angiogenesis (7, 8), endometrial angio-genesis is likely to occur through a process of elongation andexpansion of preexisting blood vessels (9) and may be sub-divided into three distinct stages: during menstruation withthe aim of vascular bed refurbishment, during the prolifer-ative phase with the aim of vascular supply, and during thesecretory phase with the aim of spiral arteriole growth andcoiling (9). Therefore, in contrast to most vascular beds,which keep a persistent structure throughout life, the endo-metrial vascular network grows and regresses during themenstrual cycle and provides a dynamic model for the studyof physiological angiogenesis.

Albeit circulating estrogen and progesterone predomi-nantly regulate the overall control of endometrial growthand regression, the direct and indirect roles of sex steroids inendometrial angiogenesis are less clear. Moreover, there areconflicting reports on the expression of estrogen or proges-terone (P4) receptors (ER and PR, respectively) in humanendometrial endothelium both in vivo and in vitro (10–13).

To date, only a few studies have developed a technique forthe isolation and characterization of endothelial cells of theendometrium (10, 14–16). The limitations of culturing hu-man endometrial endothelial cells (HEEC) for in vitro studieshave restricted our understanding of their physiologicalfunctions as well as their pathological role in vascular en-dothelium-related endometrial diseases such as endometri-osis, abnormal endometrial bleeding, endometrial vascularthrombosis, and endometrial malignancy. Our hypothesiswas set up to understand the direct role of ovarian steroids

Abbreviations: E2, Estradiol; ER, estrogen receptor; FBS, fetal bovineserum; HEEC, human endometrial endothelial cell; MTS, 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt; P4, progesterone; PBS-T, 0.05% Tween 20 inPBS, pH 7.4; PR, progesterone receptor; PRA, progesterone receptor A;PRB, progesterone receptor B; VEGF, vascular endothelial growth factor;vWF, von Willebrand factor.JCEM is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving the en-docrine community.

0021-972X/04/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 89(11):5794–5802Printed in U.S.A. Copyright © 2004 by The Endocrine Society

doi: 10.1210/jc.2003-030820

5794

in endometrial endothelial cell angiogenesis, such as prolif-eration and vascularization. To investigate the endometrialangiogenic response of estrogen and P4 in an in vitro model,we have developed a reliable method for the isolation ofHEEC. This allowed us to investigate the regulation of an-giogenesis by sex steroids.

Materials and MethodsTissue collection

Endometrial tissues were obtained from human uteri after hysterec-tomy and from endometrial biopsies, which were conducted for benigndiseases other than endometrial disease. Informed consent in writingwas obtained from each patient before surgery; consent forms and pro-tocols were approved by the human investigation committee of YaleUniversity. The mean age of the patients was 38.3 yr (range, 28–44 yr).The patients underwent surgery for leiomyomata (n � 4) and voluntarysterilization by tubal ligation (n � 4). The day of the menstrual cycle wasestablished from the patient’s menstrual history and was verified byhistological examination of the endometrium (four of them were frommidproliferative, two were late proliferative, and two were from earlysecretory). The tissues were placed in Hanks’ balanced salt solution andtransported to the laboratory for endometrial endothelial cell isolationand long-term culture. Each experimental setup was repeated on at leastthree occasions using cells obtained from different patients.

Isolation and culture of human endometrial stromal andglandular cells

Endometrial stromal and glandular cells were separated and main-tained in monolayer culture as described previously with minor mod-ifications (17). Briefly, endometrial tissue was minced with a sterilestainless surgical blade and digested by incubation of tissue minces inHanks’ balanced salt solution (Sigma-Aldrich Corp., St. Louis, MO) thatcontained HEPES (25 mmol), penicillin (200 U/ml), streptomycin (200mg/ml), collagenase H (1 mg/ml, 15 U/mg; Roche, Mannheim, Ger-many), and deoxyribonuclease (0.1 mg/ml, 1500 U/mg; Roche) for45–60 min at 37 C with agitation every 5 min using a 20-ml syringe.Collagenase H is an enzyme mixture that is used for the disaggregationof tissues and the isolation of endothelial cells from various sources (18).The dispersed endometrial cells were separated by filtration through awire sieve (73-�m diameter pore; Sigma-Aldrich Corp.). The endome-trial glands (largely undispersed) were retained by the sieve, whereasthe dispersed stromal and endothelial cells passed through the sieve intothe filtrate.

The stromal cells were plated in Ham’s F-12/DMEM (1:1 vol/vol;Sigma-Aldrich Corp.) and fetal bovine serum (FBS; 10%, vol/vol; In-vitrogen Life Technologies, Inc., Gaithersburg, MD). Cells were platedin plastic flasks (75 cm2; Falcon, BD Biosciences Franklin Lakes, NJ),maintained at 37 C in a humidified atmosphere (5% CO2 in air), andallowed to attach to the flask. On the following day of the culture, themedium was changed to remove unattached cells, dead cells, anderythrocytes.

Preparation of microbead-conjugated anti-CD105-coatedpetri dishes

CD105 (endoglin) is a glycoprotein, and its expression is highlyrestricted to endothelium in all tissues except bone marrow (19). Amicrobead-conjugated anti-CD105 (20 �l/ml; Miltenyi Biotech, Auburn,CA) solution was prepared in 50 mm Tris-Cl, pH 9.5. After adding theanti-CD105 microbead solution to petri dishes, the outside bottom sur-face of the petri dishes was enforced with a magnet to enhance andstabilize the antibody binding to dishes. Petri dishes (60-mm diameter;Falcon) were then incubated with the Tris-Cl solution (3 ml for each petridish) for 2 h at 37 C. Thereafter, the solution was removed, and each dishwas rinsed three times with 0.15 m NaCl. Dishes were then incubated

with BSA [0.1% (w/v) in PBS] for 30 min at room temperature. The BSAsolution was discarded before adding the cells.

Isolation of endometrial endothelial cells

On the second day of the endometrial stromal/endothelial cell culture(�60–80% confluence), cells were washed with PBS harvested by stan-dard methods of trypsinization and were centrifuged at 1800 rpm for 5min. The supernatant was discarded, and pellet was suspended in 2 ml(/107 cells) cell dilution buffer (PBS, pH 7.2, supplemented with 0.5%BSA and 2 mm EDTA). Thereafter, cells were filtered through a 30-�mpore size nylon mesh (Miltenyi Biotech) to prevent cells from clumping.After applying the filtered cell suspension to the dishes, the dishes wereplaced on a shaker, and CD105-positive cells were allowed to attach toantibody-covered beads with gentle agitation for 7–10 min at 8–10 C.Afterward, unattached cells were discarded, and dishes were rinsedseveral times with cell dilution buffer. Finally, the remaining attachedcells were cultured with EGM MV-Microvascular endothelial cell me-dium, supplemented with SingleQuots containing growth factors, cy-tokines, and endothelial growth supplements (Bulletkits, Cambrex-Clonetics, Baltimore, MD). The medium was replaced every 2 d. TheHEEC were grown to 80–90% confluence in a 37 C in a 95% air/5% CO2incubator. Cells were harvested using EDTA, trypsin-EDTA, and trypsininhibitor solutions (Cambrex-Clonetics). Cells were then split 1:4 forpassaging to 60-mm culture dishes (Falcon) or to four-well chamberslides (Falcon). Immunocytochemistry of endothelial cells, glandularcells, and leukocyte-specific markers was carried out in the secondpassage for cellular characterization. Second passage HEEC were alsocultured in 12-well culture plates coated with or without growth factor-reduced Matrigel (BD Biosciences, Bedford, MA), and their appearanceswere recorded microscopically every 24 h for morphology while grow-ing to confluence.

When cells reached 70–80% confluence, culture medium wasswitched for 24 h to a medium that contained 2% charcoal-stripped,steroid-depleted FBS. All experiments were then carried out in a phenol-red free medium containing 2% charcoal-stripped, steroid-depleted FBS.

Immunocytochemistry and immunohistochemistry

HEEC were grown to preconfluence on four-chamber slides. Cham-ber slides were fixed in cold methanol (�20 C) for 10 min. After severalwashings in distilled water and then three times for 10 min each timein PBS (pH 7.4), endogenous peroxidase activity was quenched by 3%H2O2 (0.6 ml H2O2 and 5.4 ml methanol) for 10 min and rinsed in 0.05%Tween 20 in PBS, pH 7.4 (PBS-T). Slides were then incubated with5% normal horse serum or 5% normal goat serum for 30 min andsubsequently with mouse monoclonal antibodies against CD31 (1:100;Novocastra Laboratories Ltd., Newcastle, UK), CD146 (P1H12 antibody;5 �g/ml; Chemicon International, Temecula, CA), ER� (1:80; NovocastraLaboratories Ltd.), ER� (developed against recombinant protein encod-ing amino acids 1–153 of human ER� expressed in Escherichia coli usedas immunogen; 3 �g/ml; GeneTex, Inc., San Antonio, TX) (20), PR (readyto use; Lab Vision Corp., Fremont, CA; and DakoCytomation, Carpin-teria, CA), and rabbit polyclonal von Willebrand factor (vWF) antibody(1:1000; Sigma-Aldrich Corp.) for 60 min at room temperature. In neg-ative control slides, equivalent isotype antibodies or normal rabbit IgGwere used instead of primary antibodies. After several rinses in PBS,biotinylated secondary antibodies (horse antimouse IgG or goat anti-rabbit IgG, Vector Laboratories, Inc., Burlingame, CA) were applied for30 min. After several PBS rinses, culture slides were incubated withstreptavidin-peroxidase complex for 30 min (Vector Laboratories). Sub-sequently, slides were rinsed several times in PBS and then incubatedwith 3-amino-9-ethyl-carbazole (BioGenex, San Ramon, CA) for 10 min.Slides were lightly counterstained with hematoxylin before permanentmounting.

Double immunohistochemistry staining

After the initial immunostaining of endometrial tissue sections for PR,the same sections were incubated with anti-CD31 antibody for 1 h atroom temperature for double immunostaining using the streptavidin-alkaline phosphatase technique. After several rinses in PBS, horse bio-tinylated antimouse IgG was applied to sections for 30 min. After several

Kayisli et al. • Angiogenicity of Sex Steroids on HEEC J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 5795

PBS rinses, tissue sections were incubated with streptavidin-alkalinephosphatase complex for 30 min (Lab Vision Corp.). Subsequently,slides were rinsed several times in PBS, then incubated with Fast Red(Vector Laboratories, Inc.) for 10 min. Slides were lightly counterstainedwith hematoxylin before permanent mounting.

Western blot analysis

Total protein from second and third passage HEEC was extracted ina lysis buffer [50 mm HEPES (pH 7.4), 150 mm NaCl, 10% glycerol, 1%Triton X-100, 1.5 mm MgCl2-6H2O, 1 mm EGTA, 100 mm NaF, 10 mmsodium pyrophosphate and protease inhibitors, 1 mm Na3VO4, 10mg/ml leupeptin, 10 mg/ml aprotinin, and 4 mm phenylmethylsulfo-nylfluoride]. The protein concentration was determined by a detergent-compatible protein assay (Bio-Rad Laboratories, Hercules, CA). Samples(20 �g) were loaded, electrophoretically separated by sodium dodecylsulfate-polyacrylamide gel using 7.5% Tris-HCl Ready Gels (Bio-RadLaboratories), and electroblotted onto Hybond ECL nitrocellulose mem-brane (Amersham Biosciences, Little Chalfont, UK). The membrane wasblocked with 5% nonfat dry milk in PBS-T buffer for 1 h to reducenonspecific binding. Then the membrane was incubated for 1 h withmonoclonal antibodies against ER� (Novocastra Laboratories Ltd. andGeneTex, Inc.), ER� (GeneTex, Inc.), and PR (Neomarkers, Fremont,CA). After several rinses with PBS-T for 15 min once and 10 min twice,the membrane was incubated for 1 h with peroxidase-labeled secondaryantibodies (Vector Laboratories, Inc.), diluted at 1:10,000, and subse-quently washed with PBS-T for 15 min once and 10 min twice. Theimmunoblot was developed using a chemiluminescent kit (NEN, Bos-ton, MA).

Cell proliferation assay

HEEC (2 � 104/well) were seeded in 96-well culture plates. HEECproliferation in 96-well culture plates was determined by a colorimetricassay using the CellTiter 96 AQueous Non-Radioactive Cell Prolifera-tion Assay kit (Promega Corp., Madison, WI). This kit determines thenumber of viable cells. The CellTiter 96 AQueous Assay is composed ofsolutions of a novel tetrazolium compound [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, in-ner salt (MTS)] and an electron coupling reagent (phenazine methosul-fate). MTS is bioreduced by cells into a formazan product that is soluble

in tissue culture medium. The absorbance of the formazan at 490 nm canbe measured directly in 96-well assay plates without additional pro-cessing. The conversion of MTS into aqueous, soluble formazan wasaccomplished by dehydrogenase enzymes found in metabolically activecells. The quantity of formazan product, as measured by the amount of490 nm absorbance, was directly proportional to the number of livingcells in culture. The first column of each 96-well plate did not containany cells and was used as a blank. Consecutive columns were treatedwith various concentrations of estradiol (E2), P4, E2 combined with P4,and phenol red-free DMEM containing 2% FBS, alone as the controlgroup. Cell proliferation was evaluated after 24 and 72 h of treatment.Four hours before the end of each experiment, MTS solution was addedto all wells (10 �l/100 �l medium/well), and plates were incubated at37 C. At the end of the incubation period, plates were read with amultiwell plate reader (Thermomax, Molecular Devices Corp., MenloPark, CA). Data were expressed in OD units. Experiments were con-ducted with replicates of 12 wells/treatment condition. Similar exper-iments were conducted on at least three different occasions with cellsprepared from four different endometrial tissues.

In vitro assay of steroid-driven angiogenic capacity andvascular tube formation

Third passage HEEC (1 � 104) was cultured on ECMatrix (a 96-wellplate format of angiogenesis assay kit, Chemicon International) andcollagen I (collagen-coated 12-well culture dishes). Short-term (4–48 h)and long-term (3–8 d) angiogenesis assays were performed in the pres-ence or absence of sex steroids. Tube formation is a multistep time-dependent process involving cell adhesion, migration, differentiation,and growth. This in vitro angiogenesis assay kit provides a convenientsystem for evaluation of tube formation by endothelial cells (21, 22).ECMatrix consists of laminin, collagen type IV, heparan sulfate proteo-glycans, entactin, and nidogen. It also contains various growth factors(TGF-� and fibroblast growth factor) and proteolytic enzymes (plas-minogen, tissue plasminogen activator, and matrix metalloproteinases).The short-term angiogenesis assay was performed as described in themanufacturer’s instructions (Chemicon International). A long-term an-giogenesis assay was performed on collagen I-coated, 12-well plates.HEEC (0.5 � 105/well) were seeded in 12-well culture plates. Overall,in both short- and long-term angiogenesis assays, the vascular tube

FIG. 1. Morphological organization ofHEEC in culture. Second passages ofHEEC were cultured to confluence onplastic or gelatin-coated culture dishes.HEEC were examined and imaged atdifferent confluence levels by phasecontrast microscopy. Morphological ap-pearances of HEEC at 70–80% (A) and100% (B) confluence are seen. Morpho-logical appearances of cells grown ongelatin-coated culture dishes at 70–80%(C) and 100% (D) confluence are alsoseen.

5796 J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 Kayisli et al. • Angiogenicity of Sex Steroids on HEEC

formation process was numerically scored as: 1, cells beginning to alignwith each other; 2, visible capillary tubes; 3, sprouting of secondarycapillary tubes; 4, closed polygons of capillaries beginning to form; and5, complex mesh-like capillary structures.

Statistical analysis

Cell proliferation, Western blot, and angiogenic scores were normallydistributed as assessed by Kolmogorov-Smirnov test. ANOVA and posthoc Tukey test for pairwise comparisons were used in statistical analysis.P � 0.05 was considered significant. Statistical calculations were per-formed using SigmaStat for Windows (version 2.0, Jandel ScientificCorp., San Rafael, CA).

ResultsCharacterization of HEEC

HEEC were isolated by selection of CD105-positive cellsusing petri dishes coated with magnetic bead-conjugatedanti-CD105 as described above. Morphological analysispointed out that the HEEC formed a monolayer on non-coated culture plates (Fig. 1, A and B). Morphologically,subconfluent monolayer stage of endothelial cells had manycytoplasmic extensions that extend to adjacent cells (Fig. 1).Moreover, the culture on gelatin-coated plates resulted information of networks and tube-like structures in some areas,which were observed by phase-contrast microscopy (Fig. 1,C and D).

After two purification processes, isolated cells werecharacterized as 95–98% pure endothelial cells on the ac-count of endothelial markers, vWF, CD31, and CD146expression (Fig. 2). CD31 was localized mainly in theplasma membrane. In contrast, CD146 was observed

mostly membranous, but also weakly cytoplasmic. Inter-estingly, CD146 immunocytochemistry also revealed astrong expression in cell-cell contact areas (Fig. 2D, inset).Furthermore, cells showed no immunoreactivity whenstained with CD45 antibody, a common marker for leu-kocytes (data not shown). As an epithelial cell marker,cytokeratin antibody did not reveal any immunoreactivityin cultured HEEC. Only few cells were negative for en-dothelial markers (1–5%). On the other hand, after six orseven passages, most of the cells lost the morphologicalappearance of endothelial cells.

In vivo and in vitro sex steroid receptor expression of HEEC

By immunocytochemistry, HEEC in culture showed astrong immunoreactivity for ER� (Fig. 3A), but a very weakimmunoreactivity for ER� (Fig. 3B). Mostly nuclear, but alsocytoplasmic, immunoreactivity was observed for ER�. HEECalso revealed a weak immunoreactivity for PR (Fig. 3C).Similarly, in vivo analysis of HEEC demonstrated a clearimmunoreactivity for ER� (Fig. 3D). However, ER� immu-noreactivity was absent (Fig. 3E). No PR immunoreactivitywas detected in endothelial cells (Fig. 3F), which was con-firmed by double immunostaining for CD31 and PR (Fig. 3F,inset). Confirming the immunocytochemistry and immuno-histochemistry results, Western blot analysis revealed thatHEEC have a clear 56-kDa band for ER�, whereas no bandfor ER� was detected (Fig. 4). At the same time, although noband was detected around the expected 94 and 116 kDa sizes

FIG. 2. Characterization of HEEC in culture. HEECwere grown in four-well chamber slides, and immu-nocytochemistry was performed. The vWF immuno-reactivity of cells cultured on noncoated (A) andgelatin-coated slides (B) is shown. Membranous CD31(C) and CD146 (D) immunoreactivities are also shown.The immunoreactivity for CD146 is increased in cell-cell contact areas (inset D).

Kayisli et al. • Angiogenicity of Sex Steroids on HEEC J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 5797

for PR A and B (PRA and PRB, respectively), a nonspecificband was observed around 50 kDa (Fig. 4)

Effect of sex steroids on HEEC proliferationConcentration-dependent proliferative effects of E2

(10�12–10�8 m) and P4 (10�12–10�8 m) on HEEC were as-sessed using the MTS colorimetric cell proliferation assay.HEEC were treated with ovarian steroids for 24–48 h. Theproliferative effect of E2 was significant at 10�10–10�8 mconcentration after 48 h of treatment (P � 0.05; Fig. 5A). Theproliferative effect of progesterone was significant at 10�12–10�8 m after 48 h of treatment compared with control (P �0.05; Fig. 5A). When combined with E2, P4 induced additionalHEEC proliferation (P � 0.05; Fig. 5A). Moreover, when thecells treated with E2 combined with ICI 182,780 and proges-terone combined with RU486, they did not reveal a signifi-

cant increase in cell proliferation compared with control(Fig. 5B).

Vascular tube-like formations and in vitro angiogeniccapacity by HEEC

We assessed the effect of E2 and P4 on HEEC’s angiogeniccapacity and formation of capillary-like tubular structures invitro. Angiogenic pattern and vascular tube formation wasnumerically scored as: 1, cells beginning to align with eachother; 2, differentiation to visible capillary-like structures; 3,sprouting of secondary capillary tubes; 4, closed polygons ofcapillaries beginning to form; and 5, complex mesh-like cap-illary structures. HEECs were cultured on ECMatrix for 4–48h (short-term) and were scored every 8 h. In control cells,angiogenic pattern was generally scored 1 or 2, and 3. At 48 hthere was no significant difference in P4-treated (10�10 m)

FIG. 3. Detection of estrogen and P4 receptors inHEEC in culture and endometrial tissue. Immunocy-tochemistry was performed on HEEC for ER� (A),ER� (B), and PR (C) in four-well chamber slides.Similarly, in vivo expressions of ER� (D; from lateproliferative phase), ER� (E; late proliferative), andPR (F; from early secretory phase) in HEEC wereevaluated in endometrial tissue sections using stan-dard immunohistochemistry and PR-CD31 double-staining immunohistochemistry (inset F; from mid-secretory phase). Representative negative controlsare presented in the C and E insets. Arrows and ar-rowheads show arterial and capillary endothelialcells, respectively, in E. F and inset F, PR-negative(arrowheads) cells in arterial and capillary endothe-lial cells, respectively. Arrows in F and inset F, PR-positive stromal and glandular cells.

5798 J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 Kayisli et al. • Angiogenicity of Sex Steroids on HEEC

cells compared with control cells. In contrast, E2 (10�8 m)treatment alone or combined with P4, induced angiogenicscores of 2–3, a trend for a higher scores than vehicle (control)and P4-treated cells (P � 0.12; Fig. 6).

Using collagen type I-coated 12-well culture dishes pre-pared for long-term (3–8 d) analysis, angiogenic patternswere scored every 24 h, beginning at 72 h of treatment. Ond 5–8 of treatment, E2 (10�8 m), P4 (10�10 m), and E2 with P4(10�8 and 10�10 m, respectively) treated cells formed signif-icantly more angiogenic patterns compared with control cells(angiogenic scores, 2.95 � 0.16, 3.26 � 0.16, 3.06 � 0.17, and1.93 � 0.15 respectively; P � 0.01; Fig. 7).

Discussion

Endothelial cells of different organs have their tissue-specific protein expressions. Also, noteworthy dissimilaritiesare present among arterial, venous, and capillary endothe-lium. Moreover, it has been reported that even capillaryendothelium of normal and tumoral tissues of the same or-gan have different molecular expression models (23). Thesefindings suggest that new approaches focusing on endothe-lial cells may provide additional understanding of patho-

FIG. 4. Detection of ER�, ER�, PRA, and PRB in HEEC in culture byWestern blot analysis. No band for ER� around 65 kDa was seen, anda clear band around 56 kDa for ER� was observed. In contrast, nospecific bands were observed for either PRA or PRB at 94 and 116 kDa.Each vertical line represents protein extracted from different cellcultures.

FIG. 5. Cell proliferation analysis of HEEC in response toE2 and P4. Cells were treated with E2, P4, E2 combinedwith P4 (A) at different concentrations, E2 combined withICI 182,780, and P4 combined with RU486 (B) for 48 h. E,E2; P, P4; E�P, E2 combined with P4; E�ICI, E2 combinedwith ICI 182,780; P�RU, P4 combined with RU486. Val-ues are expressed as the mean � SEM of 12 replicates foreach group. Each experiment was repeated on four occa-sions in cells obtained from four different patients withsimilar results. A representative graph from a single ex-periment is presented. *, Significant increase in cell pro-liferation compared with controls (P � 0.05).

Kayisli et al. • Angiogenicity of Sex Steroids on HEEC J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 5799

physiological mechanisms of various diseases. The impor-tance of endothelial cells in human endometrial physiologyand pathology is becoming more relevant.

Endometrial endothelial cells are the least studied cells ofthe human uterus. Most of the previous studies have focusedon endometrial stromal, epithelial, myometrial cells and/orleukocytes. In the present study we evaluated the direct andfunctional effects of ovarian steroids on human endometrialvascular endothelial cells.

Using a novel isolation technique of HEEC by anti-CD105-microbead technique that we are presenting, we have obtaineda quite pure population of endothelial cells confirmed by mark-ers of endothelial cells such as CD31, CD146, and vWF. Addi-tionally, morphological appearance (confluent monolayer withcobblestone-like tightly packed endothelial cells) and emer-gence of capillary-like tube structures on gelatin-coated platesconfirmed the endothelial aspect of isolated cells.

The antibody P1H12 used in this study binds specifically toCD146. In the blood vessel and bone marrow it reacts only withendothelial cells and has been used to detect endothelial cells.

It positively stains normal, primary endothelial cells in culture.Anti-CD146 positively stains endothelial cells of all blood ves-sels in frozen sections of human tissues, such as skin, intestine,

FIG. 6. In vitro angiogenic capacity of HEEC. Cells were grown onECM matrix-coated, 96-well plates and treated with E2 (10�8 M), P4(10�10 M), and E2 combined with P4 (10�8 and 10�10 M) for 48 h. Theangiogenic score was evaluated in five steps as described in Materialsand Methods. Compared with the control group (A), no significantdifference was observed in E2-treated (B), P4-treated (C), or E2- plusP4-treated (D) groups.

FIG. 7. In vitro angiogenic capacity of HEEC. Cells were grown oncollagen I-coated, 12-well plates and treated with E2 (10�8 M), P4 (10�10

M), and E2 combined with P4 (10�8 and 10�10 M) for 3–8 d. Starting fromthe fourth day of treatment, distinct differences were observed in an-giogenic scoring. The angiogenic score was evaluated in five steps asdescribed in Materials and Methods, and results are shown in the graph.Compared with the control group (A), a significant difference was ob-served in E2-treated (B), P4-treated (C), and E2- plus P4-treated (D)groups. Values are expressed as the mean � SEM of 12 replicates for eachgroup. Each experiment was repeated on three occasions in cells ob-tained from three different patients. Representative pictures and graphfrom a single experiment are presented. *, Significant increase in theangiogenic score of HEEC compared with controls (P � 0.05). Inset Brepresents a higher magnification of cell structure on collagen I, showingcapillary-tube like formation.

5800 J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 Kayisli et al. • Angiogenicity of Sex Steroids on HEEC

ovary, tonsil, and uterus. CD146 is a membrane glycoproteinthat functions as an adhesion molecule involved in heterophiliccell-cell interactions. Its expression has been demonstrated in avery limited number of normal cell types and some malignantneoplasms (23). Although the biological role of CD146 in nor-mal tissue and malignant tumors remains unclear, CD146 hasbeen suggested to play an important role in tumor progressionor suppression in a tissue-specific manner. In human endome-trium, only vascular endothelial cells are immunoreactive forCD146 antibody (24, 25). Positive expression of CD146 in HEECsupports the high purity of the endothelial cell isolation. More-over, increased CD146 immunoreactivity in cell-cell contactareas of HEEC supports a role for CD146 in endometrialangiogenesis.

Although the isolation of HEEC was previously achieved(10, 14–16), none of these reports assessed the direct andfunctional effect of sex steroids on HEEC’s tube formation.One study described the presence of steroid receptors andproliferative and antiproliferative effects of E2 and P4 onHEEC, respectively (10). Thus, to our knowledge, the presentstudy is the first using HEEC to evaluate vascular tube for-mation in the presence of ovarian steroids.

Previously, several in vivo studies were carried out inendometrial tissues to evaluate the expression of PR in hu-man vascular endothelium (11, 26–29). Although some ofthese studies were able to show PR expression, others did notobserve PR expression in HEEC. Our in vitro and in vivoresults showing the absence of PR in the endothelium sup-port later studies. Moreover, immunocytochemistry analysisrevealing a weak immunoreactivity for PR in our study ismost likely due to a nonspecific binding of the antibody,because no specific band for either PRA or PRB was detectedin the Western blot analysis, except for a nonspecific bandaround 50 kDa. However, this band may also be an alter-native form of PRA or PRB (30).

Similar to PR expression, the results of previous in vivostudies reporting the expression of ER� and ER� in HEEC areconflicting (11, 12, 13, 31). Although some studies reportedthe presence of ER� and ER� (13), many other studies re-ported the absence of ER�, but the presence of ER�, in en-dometrial vascular endothelium (11, 12, 31). Supportingthese later studies, the present study confirmed the absenceof or very weak ER� signal and a strong ER� signal in HEEC.Therefore, we conclude that HEEC express ER in vitro, andER� is the main ER in these cells, suggesting the possibilityfor a direct receptor-dependent response to estrogen.

Iruela-Arispe et al. (10) previously reported that althoughestrogen stimulates HEEC proliferation, P4 inhibits it. Al-though we have similar results for E2, in our study P4 has aproliferative effect, rather than an antiproliferative effect, inHEEC in culture. It is possible that culture conditions maycause these contradictory results, because Iruela-Arispe et al.(10) carried out their experiments in the presence of fibroblastgrowth factor-2 and vascular endothelial growth factor(VEGF), whereas we performed our experiments in the pres-ence of 2% FBS. Interestingly, HEEC seem to be highly re-sponsive to P4 even at low concentrations (10�12 m) com-pared with E2. Although our study has demonstrated aproliferative effect of P4 on HEEC, this effect might not bespecific only for these cells, but may also be observed in other

endothelial cells. Treatment with E2 and with E2 combinedwith P4 was shown to have proliferative effects, whereastreatment with P4 alone had an antiproliferative effect in anin vivo model in mice (32). In contrast, chronic treatment ofcycling rhesus monkeys with ZK 137 316, a P4 antagonist,resulted in a clear dose-dependent reduction in the thicknessof the endometrium, the abundance of glands, and endome-trial atrophy. In the same study, after treatment with 0.1 mgZK 137 316, spiral arteries and small veins were observed asmore atrophied and with degenerative hyalinization com-pared with the control (33). Moreover, Johannisson et al. (34)reported that after administration of the P4 antagonist RU486in secretory phase, necrosis was observed in endometrialcapillary structures. However, the mechanism of contracep-tive progestin-associated breakthrough bleeding and bloodvessel rupture is not well understood (35). None of thesestudies reported whether the effect of P4 is due to a directeffect on vasculature or because of a decreased production ofVEGF by stromal and/or epithelial cells. However, our re-sults suggest the possibility of a direct effect of P4 and itsantagonist RU486 on HEEC.

Numerous angiogenic and angiostatic factors have beenidentified in the human endometrium. Most of these studieshave focused on VEGF (3–5). Although other researchershave performed HEEC isolation successfully, to our knowl-edge this is the first study assessing the effects of sex steroidson vascularization of HEEC in culture. Koolwijk et al. (16)reported that HEEC begin forming capillary-like tubeswithin 3 d when cultured over a fibrin matrix in the presenceof 20% human serum. Moreover, increased angiogenic ca-pacity was observed when HEEC were cultured on rat col-lagen type I. Similarly, we observed a better capillary-likestructure formation on collagen I compared with Matrigel.Because the highest proliferative responses to E2 and P4 incell proliferation assay were obtained at 10�8 and 10�10 m,respectively, we used these concentrations for the analysis ofvascularization in vitro.

In the presence of sex steroids, HEEC exhibit an enhancedcapacity to form tube-like structures and sprouts on gelatinand collagen I. The higher angiogenic score of HEEC prob-ably reflects the direct effect of E2, because HEEC express ERin culture. However, we do not yet know how P4 may achieveits proliferative and angiogenic effects because of the absenceof PR in HEEC. One possibility is that a direct, but non-genomic, effect of P4 in HEEC may explain these results (30).Previous studies reported several examples of P4-induced,non-PR-dependent, nongenomic effects, including the initi-ation of meiosis in amphibian oocytes, the inhibition ofcholesterol synthesis, the stimulation of human acrosomereaction, a rapid increase in the intracellular calcium con-centration, tyrosine phosphorylation of proteins, the activa-tion of protein kinase C and of extracellular signal-regulatedkinases, and the binding to oxytocin receptor (36–42). Incontrast, it seems that a long incubation period (5–8 d) isneeded to reveal the effect of sex steroids on vascular-liketube formation assay. We also observed spontaneous tubeformation in the control group, which may be a response ofHEEC to growth factors in FBS.

In conclusion, HEEC display increased proliferative andangiogenic activities in response to ovarian steroids. Our

Kayisli et al. • Angiogenicity of Sex Steroids on HEEC J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 5801

results suggest direct effects of E2 and P4 on HEEC function,providing additional understanding of physiological role ofthe endothelium in endometrial function. Additional studiesare needed to investigate the roles of epithelium, stroma,trophoblast, and leukocytes on endometrial angiogenesis inthe presence and absence of sex steroids in the physiologicaland pathological events of the endometrium.

Acknowledgments

Received May 12, 2004. Accepted July 21, 2004.Address all correspondence and requests for reprints to: Dr. Aydin

Arici, Section of Reproductive Endocrinology, Department of Obstetricsand Gynecology, Yale University School of Medicine, New Haven, Con-necticut 06520-8063. E-mail: aydin.arici@yale.edu.

This study is a part of Ph.D. thesis of U.A.K. and was supported bya training grant from Akdeniz University Scientific Research Unit.

References

1. Carr BR 1996 The normal menstrual cycle. In: Carr BR, Blackwell RE, eds. Thetextbook of reproductive medicine. Chap 12. New York: Plenum Press; 233–243

2. Kayisli UA, Mahutte NG, Arici A 2002 Uterine chemokines in reproductivephysiology and pathology. Am J Reprod Immunol 47:213–221

3. Torry DS, Torry RJ 1997 Angiogenesis and the expression of vascular endo-thelial growth factor in endometrium and placenta. Am J Reprod Immunol37:21–29

4. Smith SK 1998 Angiogenesis, vascular endothelial growth factor and theendometrium. Hum Reprod Update 4:509–519

5. Gargett CE, Rogers PA 2001 Human endometrial angiogenesis. Reproduction121:181–186

6. Hanahan D, Folkman J 1996 Patterns and emerging mechanisms of the an-giogenic switch during tumorigenesis. Cell 86:353–364

7. Folkman J, Shing Y 1992 Angiogenesis. J Biol Chem 267:10931–109348. Hanahan D 1997 Signaling vascular morphogenesis and maintenance. Science

277:48–509. Rogers PAW, Gargett CE 1999 Endometrial angiogenesis. Angiogenesis 2:

287–29410. Iruela-Arispe ML, Rodriquez-Manzaneque JC, Abu-Jawdeh G 1999 Endo-

metrial endothelial cells express estrogen and progesterone receptors andexhibit a tissue specific response to angiogenic growth factors. Microcircula-tion 6:127–140

11. Kohnen G, Campbell S, Jeffers MD, Cameron IT 2000 Spatially regulateddifferentiation of endometrial vascular smooth muscle cells. Hum Reprod15:284–292

12. Critchley HO, Brenner RM, Henderson TA, Williams K, Nayak NR, SlaydenOD, Millar MR, Saunders PT 2001 Estrogen receptor �, but not estrogenreceptor �, is present in the vascular endothelium of the human and nonhumanprimate endometrium. J Clin Endocrinol Metab 86:1370–1378

13. Lecce G, Meduri G, Ancelin M, Bergeron C, Perrot-Applanat M 2001 Presenceof estrogen receptor � in the human endometrium through the cycle: expres-sion in glandular, stromal, and vascular cells. J Clin Endocrinol Metab 86:1379–1386

14. Schatz F, Soderland C, Hendricks-Munoz KD, Gerrets RP, Lockwood CJ2000 Human endometrial endothelial cells: isolation, characterization, andinflammatory-mediated expression of tissue factor and type 1 plasminogenactivator inhibitor. Biol Reprod 62:691–697

15. Nikitenko LL, MacKenzie IZ, Rees MC, Bicknell R 2000 Adrenomedullin isan autocrine regulator of endothelial growth in human endometrium. MolHum Reprod 6:811–819

16. Koolwijk P, Kapiteijn K, Molenaar B, Spronsen EV, Vecht BVD, Helmer-horst FM, Van-Hinsbergh VWM 2001 Enhanced angiogenic capacity andurokinase-type plasminogen activator expression by endothelial cells isolatedfrom human endometrium. J Clin Endocrinol Metab 86:3359–3367

17. Arici A, Head JR, MacDonald PC, Casey ML 1993 Regulation of interleukin-8gene expression in human endometrial cells in culture. Mol Cell Endocrinol94:195–204

18. Balconi G, Dejana E 1986 Cultivation of endothelial cells: limitations andperspectives. Med Biol 64:231–245

19. Gougos A, Letarte M 1990 Primary structure of endoglin, an RGD-containingglycoprotein of human endothelial cells. J Biol Chem 265:8361–8364

20. Kubota Y, Kleinman HK, Martin GR, Lawley TJ 1988 Role of laminin andbasement membrane in the morphological differentiation of human endothe-lial cells into capillary-like structures. J Cell Biol 107:1589–1598

21. Grant DS, Tashiro K, Segui-Real B, Yamada Y, Martin GR, Kleinman HK1989 Two different laminin domains mediate the differentiation of humanendothelial cells into capillary-like structures in vitro. Cell 58:933–943

22. Skiliris GP, Carder PJ, Lansdown MRJ, Speirs V 2001 Immunohistochemicaldetection of ER-� in breast cancer: towards more detailed receptor profiling?Br J Cancer 84:1095–1098

23. St Croix B, Rago C, Velculescu V, Traverso G, Romans KE, Montgomery E,Lal A, Riggins GJ, Lengauer C, Vogelstein B, Kinzler KW 2000 Genes ex-pressed in human tumor endothelium. Science 289:1197–1202

24. Shih I, Wang T, Wu T, Kurman RJ, Gearhart JD 1998 Expression of Mel-CAMin implantation site intermediate trophoblastic cell line, IST-1, limits its mi-gration on uterine smooth muscle cells. J Cell Sci 111:2655–2664

25. Shih IM 1999 The role of CD146 (Mel-CAM) in biology and pathology. J Pathol189:4–11

26. Colburn P, Buonassisi V 1978 Estrogen-binding sites in endothelial cell cul-tures. Science 201:817–819

27. Ingegno MD, Money SR, Thelmo W, Greene GL, Davidian M, Jaffe BM,Pertschuk LP 1988 Progesterone receptors in the human heart and greatvessels. Lab Invest 59:353–356

28. Vazquez F, Rodriguez-Manzaneque JC, Lydon JP, Edwards DP, O’MalleyBW, Iruela-Arispe ML 1999 Progesterone regulates proliferation of endothe-lial cells. J Biol Chem 274:2185–2192

29. Bamberger AM, Milde-Langosch K, Loning T, Bamberger CM 2001 Theglucocorticoid receptor is specifically expressed in the stromal compartmentof the human endometrium. J Clin Endocrinol Metab 86:5071–5074

30. Zhu Y, Bond J, Thomas P 2003 Identification, classification, and partial char-acterization of genes in humans and other vertebrates homologous to a fishmembrane progestin receptor. Proc Natl Acad Sci USA 100:2237–2242

31. Critchley HO, Henderson TA, Kelly RW, Scobie GS, Evans LR, Groome NP,Saunders PT 2002 Wild-type estrogen receptor (ER�1) and the splice variant(ER�cx/�2) are both expressed within the human endometrium throughoutthe normal menstrual cycle. J Clin Endocrinol Metab 87:5265–5273

32. Heryanto B, Rogers PA 2002 Regulation of endometrial endothelial cell pro-liferation by oestrogen and progesterone in the ovariectomized mouse. Re-production 123:107–113

33. Slayden OD, Zelinski-Wooten MB, Chwalisz K, Stouffer RL, Brenner RM1998 Chronic treatment of cycling rhesus monkeys with low doses of theantiprogestin ZK 137 316: morphometric assessment of the uterus and oviduct.Hum Reprod 13:269–227

34. Johannisson E, Oberholzer M, Swahn ML, Bygdeman M 1989 Vascularchanges in the human endometrium following the administration of the pro-gesterone antagonist RU 486. Contraception 39:103–117

35. Brenner RM, Slayden OD, Critchley HO 2002 Anti-proliferative effects ofprogesterone antagonists in the primate endometrium: a potential role for theandrogen receptor. Reproduction 124:167–172

36. Metherall JE, Li H, Waugh K 1996 Role of multidrug resistance P-glycopro-teins in cholesterol biosynthesis. J Biol Chem 271:2634–2640

37. Wasserman WJ, Pinto LH, O’Connor CM, Smith LD 1980 Progesterone in-duces a rapid increase in [Ca2�] in Xenopus laevis oocytes. Proc Natl Acad SciUSA 77:1534–1536

38. Baldi E, Luconi M, Bonaccorsi L, Maggi M, Francavilla S, Gabriele A, Prop-erzi G, Forti G 1999 Nongenomic progesterone receptor on human sperma-tozoa: biochemical aspects and clinical implications. Steroids 64:143–148

39. Bonaccorsi L, Krausz C, Pecchioli P, Forti G, Baldi E 1998 Progesterone-stimulated intracellular calcium increase in human spermatozoa is proteinkinase C-independent. Mol Hum Reprod 4:259–268

40. Luconi M, Krausz C, Barni T, Vannelli GB, Forti G, Baldi E 1998 Progesteronestimulates p42 extracellular signal-regulated kinase (p42erk) in human sper-matozoa. Mol Hum Reprod 4:251–258

41. Grazzini E, Guillon G, Mouillac B, Zingg HH 1998 Inhibition of oxytocinreceptor function by direct binding of progesterone. Nature 392:509–512

42. Burger K, Fahrenholz F, Gimpl G 1999 Non-genomic effects of progesteroneon the signaling function of G protein-coupled receptors. FEBS Lett 464:25–29

JCEM is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving theendocrine community.

5802 J Clin Endocrinol Metab, November 2004, 89(11):5794–5802 Kayisli et al. • Angiogenicity of Sex Steroids on HEEC