Abstract
Ovarian follicle counting is a method to assess ovarian toxicity in reproductive toxicity studies in rats. Although ovarian follicle counting has been traditionally performed manually on hematoxylin and eosin (H&E)-stained sections, the use of immunohistochemical methods, including human cytochrome P450 1B1 (CYP1B1) and proliferating cell nuclear antigen (PCNA), have been used to enhance the visibility of the primordial and primary follicles to facilitate manual counting. In this study, serial sections from both ovaries from ten 3-month-old female Sprague Dawley rats were stained using routine H&E and immunohistochemistry for PCNA. Counting of primordial and primary follicles was performed manually using these two stains and by semi-automated image analysis of PCNA-stained slides. Although manual counting of PCNA-stained slides is preferable to manual counting of H&E-stained slides, manual counting involves variability between individual counters. Semi-automated image analysis of PCNA-stained slides yields an accurate and consistent count of these primordial/primary follicles and eliminates variability between individual counters.
Keywords
Introduction
The ovary is considered a major target of xenobiotics that affect female fertility (Matttison and Thomford 1989). Certain toxicants, such as 4-vinylcyclohexene diepoxide, 1,3-butadiene, citral, galactose, cyclophosphamide, and dimethylbenz(a)anthracene (DMBA), have specific effects on primordial and primary follicles, resulting in loss of oocytes and eventual adverse effects on fertility (Bucci et al. 1997; Davis and Heindel 1999; Hoyer and Sipes 1996; Mattison 1983; Morrissey et al. 1990; Springer et al. 1996). For this reason, the US Environmental Protection Agency (EPA), US Food and Drug Administration (FDA), and the Organization for Economic Cooperation and Development (OECD) guidelines for two-generation reproductive studies recommend quantitative evaluation of primordial follicles in the ovary (OECD 2001; US EPA 1998; US FDA 2000). At the present time, regulatory guidelines for general toxicity studies do not require ovarian follicle counting. However, even in these routine general toxicity studies, properly conducted follicle counts can supplement qualitative ovarian assessment to characterize ovarian toxicants and understand their site of action (Regan et al. 2005).
Traditionally, ovarian follicle counts were and are still performed microscopically on hematoxylin and eosin (H&E)-stained sections of ovaries in mice and rats. This task is tedious and prone to errors, because small follicles are difficult to see (Bolon et al. 1997; Bucci et al. 1997; Muskhelishvili et al. 2005). To facilitate the counting procedure, immunohistochemical methods to enhance the visibility of the primordial and primary follicles have been implemented with success using proliferating cell nuclear antigen (PCNA) and cytochrome P450 1B1 (CYP1B1) (Muskhelishvili et al. 2002; Muskhelishvili et al. 2005). It has been reported that oocytes within primordial follicles of formalin-fixed, paraffin-embedded rat ovaries stain strongly with PCNA if heat-induced epitope retrieval (HIER) techniques and a high concentration of primary antibody (1:800) are used (Muskhelishvili et al. 2005). Although PCNA staining increases the speed and accuracy of counting, there is still significant labor and cost associated with manual ovarian follicle counting. As was pointed out by Muskhelishvili et al. (2005), the labor and costs could be further decreased when counts are done using semi-automated image analysis.
In this study, we demonstrate that semi-automated image analysis may be used on PCNA-stained tissue sections to obtain an accurate recording of the numbers of primordial and primary follicles in the rat ovary. Image analysis automates the counting procedure and therefore bypasses problems caused by variability between individual counters.
Material and Methods
Animals
Ten 3-month-old female Sprague Dawley rats were sacrificed by carbon dioxide (CO2) asphyxiation. Ovaries were removed, fixed in 10% neutral buffered formalin for seventy-two hours, and embedded in paraffin. Following the Internationally Harmonized Health Effects Test Guidelines (1998), serial sections (5 μm) were taken from five levels 100 μm apart in the middle third of each ovary. At each consecutive interval, two sections were cut serially. Each specimen was placed on a separate slide, resulting in two sets with nearly identical ovarian sections. One set of sections was stained with H&E, and the other was stained by PCNA immunohistochemistry. This study was approved by the WIL Research Laboratories, Animal Care and Use Committee and conducted in accordance with the guide for the care and use of laboratory animals (ILAR 1996).
PCNA Immunohistochemistry
For immunohistochemical staining of ovarian sections for PCNA, deparaffinized tissue sections were placed in tris-buffered saline (pH 6.0) and placed in a commercial decloaker (Pascal) at 125°C at 18–22 psi for five minutes. Endogenous peroxidase was blocked using 0.3% H2O2 in distilled water for fifteen minutes at room temperature. Nonspecific staining was blocked using 10% normal goat serum for thirty minutes. Slides were then incubated with M0879 anti-PCNA antibody (Dako Corporation, Carpinteria, CA, USA) at 1:400 dilution (1.3 μg/mL). Sections were then incubated with biotinylated goat anti-mouse IgM μ-chain specific antibody (Jackson Laboratories, West Grove, PA, USA) at a dilution of 1:400 for thirty minutes at room temperature. PCNA was visualized by staining with strepavidin-biotin conjugated horse-radish peroxidase (Zymed Laboratories, Carlsbad, CA, USA) at a 1:30 dilution for thirty minutes. Peroxidase was developed by diaminobenzidine for six minutes at room temperature. Slides were counterstained with hematoxylin for three minutes, dehydrated through a graded series of ethanol, placed into xylene, and coverslipped. For negative control, phosphate-buffered saline replaced the primary antibody.
Follicle Counts
All sections were examined by routine light microscopy. Primordial and primary follicles were counted using a 20X objective. Primordial follicles were those follicles having an oocyte surrounded by one or more flattened pregranulosa cells at the periphery of the follicle and no cuboidal cells. Primary follicles were those follicles having a central oocyte surrounded by either a mixture of flattened pre-granulosa cells and plump cuboidal granulosa cells or a single layer of cuboidal granulosa cells (Oktay et al. 1995; Peters et al. 1978). Growing follicles, including transitional and pre-antral follicles containing a zona pellucida and/or multiple layers of granulosa cells were not counted. Only primordial and primary follicles with intact oocytes were counted. Counts were performed independently by two veterinary pathologists certified by the American College of Veterinary Pathologists and two trained technicians. Each counter (observer) was instructed to spend equivalent amounts of time scoring either H&E- or PCNA-stained sections.
Image Analysis
Images of ovarian sections were captured with a digital camera (Olympus DP70) attached to a compound light microscope (Olympus BX51) using a 20X objective. Five to fifteen images per ovarian section were systematically taken (in rows from left to right, top to bottom). Captured images covered the entirety of each ovarian section, with minimal tissue omission. Images of each section were opened in Image Pro software (Media Cybernetics, Bethesda, MD, USA). Adjusting histogram intensity is important in helping to improve image quality (signal-to-noise ratio). The brightness (intensity) in each digital image was equalized to reduce variation between images. The “Best Fit” command was used to instruct ImagePro to optimize the values for each image. The results were achieved by stretching the histogram to accomplish the best possible contrast distribution of pixel values in the image. “Best Fit” assigns the bottom 3% of the values to the shadow point (0) and the top 3% of the values to the highlight point (255). The rest of the values were distributed evenly across the scale. An unsharp filter was used to create ideal separation of the follicles from the background. The unsharp filter is a sharpening operator that derives its name from the fact that it enhances edges (and other high-frequency components in an image) via a procedure that subtracts an unsharp, or smoothed, version of an image from the original image. Regions of granulosa and corpora luteal cells were “masked” out from the image to exclude them from subsequent analysis. Once the mask had been created, intensity values for each image were saved and later used to identify follicles of interest. Immediately prior to running Image Pro on the digital photographs, the corpora lutea were manually excluded using a freeform Area of Interest (AOI) tool. The software searched for oocytes within primordial and primary follicles based on preset ranges for the size, roundness, and color of the central oocyte. An overlay of the count was then applied to the original image to apply a display reference for the counted objects. The image analysis macro was then run on the series of digital photographs to arrive at a number or count of primordial and primary follicles for each ovary.
Statistical Analysis
The t test was used to evaluate statistical differences with a significance level of p < .01 (Microsoft Office Excel 2003, Tools, Data Analysis, t test: Two Samples Assuming Equal Variances). Variability was calculated where the coefficient of variation (CoV) is equal to the standard deviation (STD) divided by the mean (X̄):
Variability is expressed as a percentage by multiplying the variability by 100:
For interobserver variability, ten ovarian follicle counts were compared between observers. To evaluate variability between automated and manual counting methods, semi-automated derived PCNA follicle counts were compared with average manual H&E and PCNA follicle counts for ten animals.
Results
Results of these experiments demonstrate highly specific PCNA immunostaining for small follicles (primordial, early primary, and primary), growing follicles (transitional and pre-antral), and large antral follicles. Compared to basic H&E staining, PCNA immunostaining improved the visual detection of small follicles (Figures 1, 2, and 3). PCNA staining was less intense in proliferating granulosa, thecal, and corpora luteal cells. Squamous pregranulosa cells surrounding the oocyte of primordial and early primary follicles were not stained by PCNA (Figure 4). Primordial or primary follicles that did not contain stained oocytes were not evaluated in these experiments. There was no significant PCNA staining detected for negative controls. Results in these experiments show that the semi-automated image analysis method (using Image Pro software) allowed the selective identification of primordial and primary follicles and selectively excluded growing and antral follicles (Figures 5 and 6).
Manual counts of PCNA-stained sections consistently produced higher follicle counts compared to lower manual counts for H&E-stained sections (Table 1). After all manual follicle counts were completed, the counts were averaged, and the results are presented in Table 2. There were no significant statistical differences between individual manual follicle counts for interval-cut serial sections stained with PCNA or H&E (Table 2). However, there was significant difference between total manual follicle counts for ten animals for H&E- and PCNA-stained sections (p <.01) (Table 1; total counts over ten animals). This substantial difference represents a considerable improvement in the ability to visually detect small follicles using PCNA immunostaining.
Total manual counts also show high interobserver variability for both H&E-stained slide counts (34%) and PCNA-stained slide counts (26%) (Table 2). Importantly, these results show an 8% decrease in manual interobserver variability with PCNA staining, and remarkably, variability was virtually eliminated when PCNA-immunostained slides were evaluated using Image Pro image analysis software.
Follicle counts derived by PCNA staining in combination with image analysis produced numbers that closely matched those obtained by PCNA staining with manual counting (Table 1). Variability between PCNA image counting and average manual PCNA counting was very low, 0.6% (Table 2). The results in this report demonstrate PCNA immunostaining in combination with Image Pro produced accurate and reproducible numbers for small follicle counting.
Discussion
The results of our experiments provide convincing evidence that manual counting of primordial and primary follicles using PCNA-immunostained ovarian sections is preferable to manual counts derived by basic H&E staining. By using PCNA staining techniques, fewer small follicles are missed, and there is less variability between different individual observers. The results described in this report are compatible with those reported by Muskhelishvili et al. (2005).
Routine and highly specific PCNA staining of ovarian follicles in formalin-fixed, paraffin-embedded rat ovaries requires heat-induced epitope retrieval. A commercial decloaker was used to provide the heat-induced antigen retrieval step. A decloaker is an electric pressure cooker used for antigen retrieval in immunohistochemistry laboratories. The combination of high pressure and heat unmasks surface antigens from formalin-fixed tissue specimens. In some reports, high levels of primary PCNA antibodies were required for ovarian follicle staining (Muskhelishvili et al. 2005). We found that the manufacturer’s (Dako Corporation, Carpinteria, CA, USA) recommended level of PCNA antibodies (1:400 dilution or 1.3 μg/mL) is sufficient for the successful immunohistochemical follicle staining achieved in our laboratory.
PCNA is an auxiliary protein of DNA polymerase-δ enzymes necessary for DNA synthesis, and it is used as a standard marker in proliferating cells (Wood and Shivji 1997). PCNA has also been shown to be involved in DNA repair (Celis and Madsen 1986; Shivji et al. 1992; Wood and Shivji 1997). Since high levels of PCNA are found in oocytes, this protein serves as a useful immunohistochemical marker for identifying primordial and primary follicles.
PCNA staining and counting techniques have not been successful with primordial mouse follicle oocytes in our laboratory. After attempting different antigen retrieval procedures, mouse oocytes failed to stain for PCNA. Kerr et al. (2006) also reported that primordial follicle oocytes in the mouse are negative for PCNA using microwave antigen retrieval steps, which was the antigen retrieval procedure used by Muskhelishvili et al. (2005) for detection of PCNA in the rat. Further studies would be needed to determine the usefulness of PCNA immunostaining and PCNA image analysis for ovarian follicle counting in mice and other animal species.
When using semi-automated image analysis for counting primordial or primary follicles for FDA-regulated industries, the digital images are considered “raw data” under the Department of Health and Human Services Good Laboratory Practices (21 CFR Part 58, Section 58.3[k]). The quality and integrity of these images therefore must be assured by operating a quality assurance program, implementing appropriate standard operating procedures, monitoring proper image labeling, and using a calibrated and validated computer system (Tuomari et al. 2007).
Our results provide evidence that semi-automated image analysis of PCNA-stained ovarian sections provide accurate counts of primordial and early primary follicles. Our results also show that follicle counts derived by semi-automated software analyses are similar to those obtained by manual counting. Automated image analysis provides an important advantage by significantly reducing manual counting variability. Based on the results presented in this report, we conclude that automated image analysis is preferable to manual counting, particularly when considerable observer bias is possible. For this reason, we believe image analysis is preferable to manual counting, especially if manual counting reveals a questionable or subtle test-article–related effect on the number of small follicles.