Rhesus monkeys (Macaca mulatta) were maintained in the Fu-Zhou Primate Research Centre, China. All experimental work was approved by the Animal Ethics Committee at the Institute of Zoology, Chinese Academy of Sciences. Menstrual cycles of female monkeys were monitored for 2–3 cycles before sampling (n = 10). One ovary and uterine tissue were collected 3, 2 and 1 day before (ov-3, ov-2 and ov-1 respectively), and 5, 9, 10, 11, 12, 14 and 15 days after (ov+5, ov+9, ov+10, ov+11, ov+12, ov+14 and ov+15 respectively) ovulation. For pregnant monkeys (n = 5), animals were permitted to mate over a period of 3 days at the anticipated time of ovulation, the second day of mating was designated day 0 of pregnancy. The presence of a conceptus was confirmed by ultrasound diagnosis and uterine tissue collected. At the appropriate time of either the cycle (ov-3 to ov+15) or pregnancy (d22 or d35), the monkeys were sacrificed humanely, the ovaries, uterus, aorta, bladder, small intestine, skeletal muscle, heart, liver and kidney were removed. Appropriately selected wedges of the ovaries and the uterus were fixed in buffered formalin overnight at 4°C, washed in Tris-buffered saline (pH 7.4) (TBS) and processed to paraffin wax blocks. The remaining sections of the ovary and the uterus and all other tissue types were collected in RNA later (Ambion, CA, US) and stored at -80°C for RNA extraction.

RNA was extracted from adult monkey tissue by homogenisation in Trizol reagent (Qiagen Sciences, MD) according to the manufacturer's instructions. All samples were treated with RNase-free DNase (Ambion) and the concentration and purity assessed by spectrophotometry and agarose gel electrophoresis.

Total RNA (1 ug) was reverse transcribed as previously described 9. A 2 μl aliquot of RT product was amplified in a total volume of 25 μl using 2 × PCR-Master mix with added magnesium (Promega) and 10 pmol sense and antisense primers for long and short form HTRA3 and 18S rRNA as outlined in Table 1. Primers for long form HTRA3 were designed to specifically amplify exons 9 and 10, therefore amplifying exons specific to the long form and a region that spans an intron (Fig 1). Primers for short form HTRA3 were designed to amplify a region of exon 7, an exon specific to the short form HTRA3 (Fig 1). PCR was performed using a conventional PCR block cycler (Hybaid PCR Express) with an annealing temperature of 58°C or 53°C for 30 cycles for the long form and the short form of HTRA3 respectively. Cycle numbers in the linear phase were used and 18S (with an annealing temperature of 58°C for 20 cycles) was used as a control. PCR products were analysed by electrophoresis on a 1% (w/v) agarose/TBE gel and bands visualised via ethidium bromide staining. The bands were excised and the identity of each PCR product was also verified by DNA sequencing in initial experiments.

Five-micron serial sections of ovarian and uterine tissue were subjected to standard immunohistochemistry. All ovarian sections underwent immunostaining in a single experiment as did all uterine sections. For each section of tissue, a serial tissue section was used as a negative control. Following antigen retrieval by microwaving the sections for 5 min in 0.1 M citric acid buffer (pH 6.0), the endogenous hydrogen peroxidase activity was quenched by incubating the sections with 3% H₂O₂. For the ovarian tissue, non-specific binding was blocked by pre-incubation with a blocking buffer containing high salt TBS (0.3 M NaCl in 50 mM Tris, pH 7.6), 0.1% Tween 20, 15% rabbit serum and 2% normal human serum for 1 h at room temperature. A similar blocking buffer and method was used for the uterine tissue except 12% rabbit serum and 6% fetal calf serum were used in the buffer. The primary antibody [sheep anti-HTRA3 antibody #153 7 (2 μg/ml for ovarian tissue, 0.5 μg/ml for uterine tissue) or pre-immune sheep IgG as a negative control, at the same concentration as the primary antibody] was incubated in a blocking buffer containing high salt TBS, 0.1% Tween 20, 10% rabbit serum and 2% normal human serum at 37°C for 1 h and washed with high salt TBS plus 0.6% Tween. The secondary antibody [#BA-6000, biotinylated rabbit anti-sheep IgG (1:200, Vector Laboratories, Burlingame, CA)] was applied in the same blocking buffer as the primary antibody for 30 min at room temperature. Positive immunostaining was revealed by incubating the sections with an avidin-biotin-complex conjugated to horseradish peroxidase (DakoCytomation, Botany, Australia) in the dark for 30 min at room temperature, followed by the application of the peroxidase substrate 3,3'-diaminobenzidine (DakoCytomation) leading to a brown precipitate. The sections were counterstained with Harris hematoxylin. Microscopy was performed using an Olympus BH2 microscope fitted with a Fujix HC-2000 high resolution digital camera (Fujix, Tokyo, Japan).

To determine the expression level and cellular localization of HTRA3 protein in the non-pregnant ovary, ovaries collected at ov-3, ov-2, ov+5, ov+9, ov+11, ov+12 and ov+14 were examined (n = 1 monkey for each day). Follicles were counted and investigated from all 7 monkeys (primordial follicles, n = 158; primary follicles, n = 119; secondary follicles, n = 78; late secondary follicles, n = 4; antral follicles, n = 5 (antral follicles were observed in monkeys at ov-3 and ov-2), corpora lutea, n = 3 (CLs were observed in monkeys at ov+11, ov+12 and ov+14). Non-pregnant uterine tissues sampled at ov-1 (n = 3), ov+5 (n = 3), ov+10 (n = 3) and ov+15 (n = 2) of the menstrual cycle were also examined. The relative intensity of the immunostaining was scored semiquantitatively by two independent observers; zero representing no staining and four representing maximal staining. Such analysis of immunostaining has previously been reported 1011. The scores for each type of cellular subtype in each different follicle type investigated by each observer were then averaged. Comparisons of staining intensities between cellular subtypes within different follicle types were performed using One-way ANOVA, followed by Tukey's post hoc test using PRISM v4 for Windows (GraphPad Software Inc, San Diego, CA). The staining intensity of the endometrial glandular epithelium of non-pregnant monkeys at different stages of the cycle was scored and analysed in a similar fashion, with the quantitation resulting in the average of the signal. P < 0.05 was considered statistically significant.

The expression of HtrA3 at the maternal-fetal interface was determined by examining implantation sites (including the uterus and the placenta) from d22-d35 pregnant monkeys (n = 5). For these tissues, trophoblast cells were identified by immunostaining for cytokeratin using a mouse anti-human cytokeratin monoclonal antibody (7 μg/ml, #34902, BD Biosciences, San Jose, CA) as the primary and a biotinylated horse anti-mouse IgG (1:200, #BA-200, Vector Laboratories) as the secondary antibody, following antigen retrieval as for immunostaining of HtrA3.

Using RT-PCR, the long and short forms of HTRA3 mRNA were detected in the ovary, aorta, bladder, small intestine, skeletal muscle, heart and uterus but not the liver nor the kidney (Fig 2). The long form HTRA3 mRNA product size was 337 base pairs (bp) while the short form HTRA3 mRNA was 320 base pairs. The highest level of the long form HTRA3 mRNA was in the ovary, heart and uterus (Fig 2). The highest expression of short form HTRA3 mRNA was in the ovary and bladder (Fig 2). 18S was detected in all adult rhesus monkey tissues investigated (Fig 2). The RNA extracted from ovaries at each stage of the cycle was analysed by traditional RT-PCR, however there was no difference in the expression of either the long form HTRA3 or the short form HTRA3 between the different stages (data not shown).

Primordial follicles (Fig 3A) showed staining in the oocyte, however the squamous granulosa cells surrounding the oocyte were negative for HTRA3 protein. A similar pattern was seen in the primary follicles (Fig 3B arrowhead). The newly forming thecal cells surrounding the granulosa cells were negative for HTRA3 (Fig 3B arrow). Expression of HTRA3 protein began to increase in the secondary follicles with the granulosa cells (Fig 3C arrowhead) showing moderate staining for HTRA3 whilst the thecal cells showed no staining (Fig 3C arrow). When the intensity of staining in the granulosa cells was scored, there was no significant difference between primordial and primary and secondary follicles (Fig 4A and 4B). HTRA3 protein expression in the granulosa cells significantly increased (P < 0.05) from primordial and primary follicles to the cumulus and mural granulosa cells of the late secondary follicles (Fig 3D and Fig 4A and 4B). The thecal cells remained negative. A significant increase in the intensity of granulosa cell staining for HTRA3 protein was observed in the antral follicles (Fig 3E and 4A and 4B). HTRA3 protein in both the cumulus and the mural granulosa cells was significantly higher in antral follicles than in primordial and primary follicles (P < 0.01), and secondary follicles (P < 0.05) (Fig 4A and 4B). The cumulus granulosa cells in the antral follicles (Fig 4A) showed more intense staining for HTRA3 compared to the mural granulosa cells (Fig 4B). The thecal cells of antral follicles were negative for HtrA3 protein. In the CL the granulosa-lutein cells (Fig 3F arrows) showed significantly (P < 0.01) higher expression of HTRA3 protein than the theca-lutein cells (Fig 3F asterisk and 4D).

High expression of HTRA3 protein was observed in the oocytes of all follicles. This level of expression was not significantly different between follicle types (Fig 3A-E and Fig 4C).

In the non-pregnant monkey endometrium, HTRA3 was detected primarily in the glandular epithelium at all examined stages of the cycle (Fig 5A–G,). Within each sample, however, not all glands were stained with equal intensity, rather zonal variations in staining were observed. In general, glands close to the myometrium were stained stronger than those close to the uterine lumen (Fig 5A and 5D). To take this difference within each sample into consideration when comparing different samples, we arbitrarily divided the glands on each section into two groups (outer and inner) according to their location and scored separately. The outer glands were defined as those located in the upper two-thirds of the endometrium near the lumen known as the functionalis and the inner glands in the lower third of the endometrium near the myometrium known as the basalis. In all sections examined, the basalis glands on average, stained more strongly than the functionalis glands (Fig 5B and 5C, F and 5G, and Fig 6). When different stages of the cycle were compared, the two groups of glands showed a very similar trend of immunostaining: lower intensity at ov-1 (Fig 5B–C, Fig 6), higher at ov+5 and maximal at ov+10 (Fig 5F–G, Fig 6), but then decreasing again at ov+15 (Fig 6). Statistical analysis, however, did not reveal these differences to be significant.

The earliest implantation site available was on day 22 and the latest on day 35 of pregnancy. HTRA3 staining pattern was the same between these days and an implantation site on day 35 is shown (Fig 7A). Both maternal decidual cells (Fig 7A and 7C) and glands (Fig 7D) were strongly stained for HtrA3 during pregnancy. In contrast, trophoblast cells, which were positively identified by their staining for cytokeratin (Fig 7B), showed weak staining for HTRA3 depending on the subtypes (Fig 7A). Trophoblast shell was completely negative for HTRA3 (Fig 7C), so were most of the cells in the trophoblast cell columns in the anchoring villi (Fig 7E). On the other hand, the syncytial trophoblast on the exterior of the anchoring (Fig 7E) and the floating (Fig 7F) villi was positive. Some but not all villous cytotrophoblast was positively stained for HTRA3, the intensities were generally weaker than in the adjacent syncytiotrophoblasts (Fig 7F).

A few blood vessels in the decidua close to the trophoblast shell were surrounded by a thick rim of large number of extravillous cytotrophoblasts (EVTs) which stained strongly for cytokeratin (Fig 7G, asterisks). The wall of such blood vessels could no longer be clearly defined, and it was apparent that some trophoblast cells had replaced the endothelium (Fig 7G, arrow) and some were inside the vessel (Fig 7G, arrowhead). When serial sections were immunostained for HTRA3, the EVTs outside the blood vessel were negative (Fig 7H, asterisk), whereas those lining or inside the vessel (Fig 7H, arrow and arrowhead respectively) were weakly positive. Except those EVTs that were closely associated with blood vessels, no interstitial EVTs distributed away from blood vessels was identified on our sections.