- Email: [email protected]
Morphometry of stallion spermatozoa by computer-assisted image analysis
MORPHOMETRY OF STALLION SPERMATOZOA BY COMPUTER-ASSISTED IMAGE ANALYSIS’ B. A. Ball and H. 0. Mohammed Department of Clinical Sciences College of Veterinary Medicine Cornell University Ithaca, NY 14853 Received for publication: December 28, 1994 Accepted: March I, 199: ABSTRACT Metric measurements of stallion spermatozoal heads were determined for live, unfixed spermatozoa and for Feulgen-stained spermatozoa by videomicroscopy and computerized image analysis. Two ejaculates were collected from each of five stallions of normal fertility. Air-dried semen smears were Feulgen-stained, and live, unfixed spermatozoa were examined as wet-mount preparations. For Feulgen-stained spermatozoa, videoimages (x3850) were captured, and sperm heads were detected via image segmentation and particle ‘analysis. For live, unfixed spermatozoa, phase contrast videoimages (x3850) were measured to determine width and length of the sperm head. For Feulgen-stained spermatozoa, there were significant effects (P < 0.001) of stallion and ejaculate on measured parameters of area, circumference, and the length and width of the sperm head. For live, unfixed spermatozoa, there were significant effects of stallion on length and width and of ejaculate on length of the sperm heads. There was a very poor correlation between length and width of sperm heads between Feulgen-stained and live, unfixed spermatozoa. Two indices of sperm shape (oval factor and aspect ratio) were also determined. Both aspect ratio and oval factor were significantly affected by stallion (P < 0.001); however, oval factor was not affected by ejaculate and therefore may represent a less variable determination of sperm head shape across stallions. Overall, length and width of stallion sperm heads were larger (P < 0.01) for live, unfixed spermatozoa than for Feulgen-stained spermatozoa (length: 6.3 f 0.4 vs 5.08 f 0.44; width: 3.08 f 0.34 vs 2.71 f 0.28 pm, respectively). Computerized image analysis may be usetil as a means to objectively measure sperm head dimensions in the stallion and could be usefid in future studies to determine associations with stallion fertility. Kev words:
equine, spermatozoa, image analysis, measurement, sperm head
This research was supported by the Harry M. Zweig Memorial Fund for Equine Research and Thornbrook Farms. The authors thank M. Lillard for technical assistance and P. G. A. Thomas and I. Dobrinski for a review of the manuscript.
Theriogenology 44:367377,1995 0 1995 by Elsevier Science Inc. 655 Avenue of the Americas, New York, NY 10010
0093-691 X/95/$1 0.00 SSDI 0093-691 X(95)001 91 -A
368
Theriogenology INTRODUCTION
Morphology of spermatozoa has been used in the fertility evaluation of stallions (2,3,11,19), and abnormal sperm morphology has been associated with reduced fertility (11). Morphologic assessment, as currently conducted, remains extremely subjective, particularly regarding the size and shape of the sperm head. A moderate negative correlation has been reported between the incidence of abnormal heads and stallion fertility (11). Previously, alterations in sperm head size have been either gross microcephaiy or macrocephaly as determined by subjective assessement (11). In human male fertility assessment, strict, objective morphologic guidelines have been established for evaluation of the size and shape of the sperm head (8,16,22). Based on these guidelines, size and shape of the sperm head appear to be well correlated with IVF results in humans (9,13,16). Measurements can be determined by either eyepiece micrometer or by automated, computer-assisted image analysis systems (7,10,14,16). Automated systems hold the potential advantage of reducing technical variation that is inherent in manual, subjective morphology analysis. However, their use requires careful specimen preparation and staining in order to reduce digitization errors (7). The widespread availability of videoimage analysis on microcomputers provides the opportunity to evaluate more critically morphometric parameters in stallion spermatozoa. The purpose of the work described here was: 1) to examine the length and width of live, unfixed stallion sperm heads and 2) to evaluate the use of semi-automated measurements of the Feulgen-stained stallion sperm head with an image analysis program. MATERIALS AND METHODS Five stallions ranging in age from 5 to 21 y were used in this experiment (Table 1). These stallions were considered to have normal fertility based upon previous breeding history and upon the parameters of sperm number, progressive motility and subjectively assessed spermatozoa1 morphology (19; Table 1). Two ejaculates were collected with an artificial vagina within a 30-d interval (April) from each stallion. Immediately after collection, 10 ul of raw semen were smeared onto a warm (37°C) slide and allowed to air dry. The smears were subsequently stained for videomicroscopy. An additional aliquot of raw semen was transported back to the laboratory in an insulated container and was used for preparation of wet mounts for videomicroscopy. All reagents used in this study were from Fisher Scientific, Rochester, New York unless otherwise indicated. Air-dried semen smears were Feulgen-stained based on a modification of the method of Barth (1). The smears were allowed to dry for a minimum of 1 h prior to staining. Smears were then hydrolyzed with 5 N HCl for 30 min, washed in running water for 5 min, and then treated with Schiffs reagent for 30 min. Slides were rinsed in distilled water for 2 min, allowed to dry, mounted in permount and coverslipped. For measurements of live, unfixed spermatozoa, 1.0 ul of raw semen was placed on a slide and mounted with a 22x22-mm coverslip (12). This thin semen smear served to partially immobilize
Theriogenology
369
spermatozoa to prevent problems with accurate image acquisition. Prior to preparation of wetmounted, live spermatozoa, semen was maintained at 37°C. Temperature of slides was also maintained at 37°C by a microscope stage heater (Kocky Mountain Instruments, Fort Collins, CO) during acquisition of videoimages of live, unfixed spermatozoa. Feulgen-stained smears were examined via bright-field microscopy (Olympus BH-2 microscope, Olympus, Inc., Lake Success, NY) with a xl00 oil immersion objective (plan apochromatic; N.A. 1.3) and wetmounts of live, unfixed spermatozoa were examined via phasecontrast microscopy with a xl00 oil immersion objective. For bright-field microscopy, Kohler illumination was established, and a green filter was placed below the stage condenser to enhance contrast of the acquired bright-field and phase-contrast images. A x3.3 photoeyepiece was used along with a CCD camera (CCD 72 camera; Dage-MTI, Inc., Michigan City, IN) to acquire videoimages. The intensity of the halogen bulb and camera black level were standardized across all samples. The autogain feature of the CCD 72 camera was used to provide standard field illumination between samples. Linear grey scale reproductions were produced with a camera gamma setting of 1.O. Videoimages of spermatozoa were digitized (640x480 pixels at 8-bit depth) with a Scion LG3 frame grabber (Scion Corporation, Walker&he, MD) in a Macintosh Quadra 800 microcomputer (Apple Computer Corp., Cupertino, CA). The image analysis system was calibrated against a stage micrometer (0.098 urn per pixel), and the final image magnification on the video display was x3850. Camera aspect ratio was adjusted within software to compensate for differences (< 2%) in vertical and horizontal measurements. Digitized images were analyzed with the NIH Image program (ver. 1.54, Wayne Rasband, National Institute of Mental Health Health). For semi-automated measurements of spermatozoa1 heads, Feulgen-stained smears were scanned by the operator to locate a region of the smear with an adequate number of spermatozoa per field for analysis. Once a region was determined on the slide, sequential, nonoverlapping fields were digitized, and images were stored prior to analysis. The images were analyzed to determine area, perimeter, major ellipse (length), minor ellipse (width), and mean grey scale value. Major and minor ellipses represent the lengths of the axes of the best fitting ellipses. The oval factor (4 n area&rime&) and aspect ratio (widtMength) were also calculated to evaluate head shape (10,18). The integrated density of each sperm head was calculated based upon the product of area and mean grey scale value for that sperm head. Spermatozoal heads were differentiated based upon image segmentation within a predetermined range of grey-scale values and subsequent analysis of spermatozoal heads as particles within a range of 7.7 to 17.3 pm2 (Figure la and b). The range of particle sizes to be included or excluded as spermatozoal heads was determined in preliminary experiments. A total of 300 spermatozoa was analyzed per slide and duplicate slides were analyzed To determine the frequency of digitization errors that resulted in for each ejaculate. misidentification of spermatozoa, an operator reviewed all images to determine the frequency of debris or overlapping spermatozoa that were measured as individual spermatozoa. For wet-mount preparations of live, unfixed spermatozoa, digitized phase-contrast images of spermatozoa were individually measured by an operator with the NIH Image software. Length and width of the spermatozoal head were determined by arbitrary measurement of the greatest width and length of the spermatozoal head. A total of 100 spermatozoa was measured for each ejaculate.
370
Theriogenology
Data were exported from NIH Image to Microsott Excel (ver. 4.0; Microsoft Corp., Redmond, WA) and analyzed in Stat&x (ver. 4.1; Analytical Software, Tallahassee, FL). The design consisted of slides nested within ejaculates and ejaculates nested within stallions. Data were analyzed by the General Analysis of Variance in Statistix. Means were ranked and pairwise comparisons between individual means were computed via Tukey’s HSD at a < 0.01. Correlations between corresponding measurements from Feulgen-stained preparations and raw sperm measurements were also computed in Statistix. For comparisons between Feulgen-stained spermatozoa and live, unfixed spermatozoa, data from Feulgen-stained spermatozoa were arbitrarily truncated at 100 cells in order to provide equal numbers of spermatozoa for comparison by ANOVA and computation of Pearson’s correlation coefficient.
RESULTS For semi-automated analysis of Feulgen-stained spermatozoa, there were significant effects (P < 0.001) of stallion and ejaculate but not slide on area, perimeter, minor axis and integrated density of spermatozoal heads (Tables 2 and 3). In addition, stallion, ejaculate and slide affected (P < 0.01) major axis of spermatozoal heads and aspect ratio (Tables 2 and 3). Stallion, but not ejaculate or slide, significantly affected (P < 0.001) oval factor for spermatozoal heads (Table 3). For length measurements derived from phase-contrast images of live, untixed spermatozoa, there were significant effects of stallion (P < 0.001; Table 4). For corresponding width measurements of live, untixed spermatozoa, there were significant effects (P < 0.001) of both stallion and ejaculate (Table 4). Aspect ratio for live, unfixed spermatozoa was significantly affected (P < 0.001) by both stallion and ejaculate (Table 4). Length and width measurements derived from raw, unfixed spermatozoa were poorly correlated with corresponding major and minor axis determinations derived from Feulgen-stained spermatozoa by semi-automated image analysis (r = 0.03 and 0.08, respectively). Length and width measurements from raw spermatozoa were significantly larger (P < 0.01) than the corresponding measurements of major and minor axis from Feulgen-stained spermatozoa (Figure 2). Aspect ratios derived from Feulgen-stained spermatozoa were significantly larger (P < 0.01) than aspect ratios of live, unfixed spermatozoa (0.53 f 0.07 vs. 0.43 f 0.07, respectively). Aspect ratios derived from the 2 methods were moderately correlated (r = 0.66). The overall frequency of digitization errors (misidentification of spermatozoa) for Feulgenstained spermatozoa was 5.4% (range of 1.7 to 9%). The most common digitization error appeared to be overlapping spermatozoa that were identified and measured as a single spermatozoa (Figure lb). DISCUSSION Measurements of stained sperm heads in the stallion have been reported previously by investigators using either a projected image (21), scanning electron microscopy (3) or automated computerized morphometry (5). In general, values derived from studies that used fixation and/or drying (including the Feulgen stained spermatozoa in the present study) appeared to be lower than
of Feulgen Figure 1. Digital micrograph stained equine spermatozoa (a). Digital micrograph after image segmentation (b) to identify individual spermatozoa (outlined with black line). Note that overlapping spermatozoa are not selected and that sperm heads touching the edge of field are not selected. Bar represents 10 pm.
Live
m
Length
q
Width
Feulgen-stain
Figure 2. Length and width measurements of equine sperm heads determined from either five, unfixed spermatozoa (live) or from Feulgen-stained sperm heads. Data presented as mean + std deviation. Corresponding values differ (P < 0.01) between live and Feulgen-stained spermatozoa.
Theriogenology
372 Table 1.
Age, breed and semen parameters for 2 ejaculates from each of stallions.
Stallion
Age
Breed
1
6
Arabian
2
3
4
5
Table 2.
5
Thorough-
bred
21
17
l9
Warmblood
Morgan
Cl-
Ejaculate
Volume (a
Concentration 106hd
Total no. of Spermatowa ww
Motility %
Normal ;orpho””
1
45
151
6.8
80
86
2
40
150
6
65
82
1
100
135
13.5
70
-57
2
45
60
2.7
75
67
1
110
130
14.3
70
52
2
90
122
11
65
49
1
55
194
10.7
80
56
2
40
100
4
75
60
1
100
417
41.7
50
52
2
60
385
23.1
70
50
Mean and standard deviation of metric parameters of Feulgen-stained heads determined by computerized image analysis.
sperm
Stallion
Area (pm2)
Perimeter(pm)
Major ellipse (urn)
Minor ellipse (pm)
1
10.22a (1.43)
12.98a (1.17)
5.01a (0.41)
2.59a (0.25)
2
10.49b (1.67)
12.78b (1.42)
4.75b (0.44)
2.80b (0.27)
3
11.02c (1.34)
13.36~ (1.07)
5.20~ (0.36)
2.70~ (0.25)
4
ll.llc (1.32)
13.19d (1.07)
5.02a (0.36)
2.81b (0.23)
5
11.29d (1.34)
13.69e (1.09)
5.37d (0.36)
2.68~ (0.27)
All
10.82 (1.49;13.7)
13.20 (1.21; 9.2)
5.07 (0.44; 8.7)
2.72 (0.27; 9.8)
Values within columns with diierent superscripts differ (P < 0.01). Expressed as mean data. n = 600 spermatozoa per stallion, For overall, values in parentheses represent standard deviation; CV).
373
Theriogenology Table 3.
Shape parameters and integrated density of Feulgen-stained determined by image analysis.
Stallion
Oval Factor
Aspect Ratio
Integrated density
1
0.76a (0.04)
0.52a (0.06)
2429a (347)
2
0.8lb (0.06)
0.59b (0.06)
2534b (413)
3
0.78~ (0.04)
0.52a (0.05)
2687~ (329)
4
0.8Ob (0.05)
0.56~ (0.06)
2695c .(326)
5
0.76a (0.05)
0.50d (0.06)
2761d (331)
All
0.78 (0.05; 6.4)
0.54 (0.07; 12.4)
2621 (371; 14.2)
sperm heads as
Oval factor expressed as (4 x area/perimetefi. Aspect ratio expressed as (width/length). Integrated density expressed as (area * mean grey scale value). Values within columns with different superscripts differ (P < 0.01). Expressed as means (standard deviation). n = 600 spermatozoa per stallion. For overall, values in parentheses represent standard deviation; CV). Table 4.
Length and width of fresh, live spermatozoa as determined by manual measurements on spermatozoa from 5 stallions:
Stallion
Length (pm)
Width (pm)
Aspect Ratio
1
6.63a (0.43)
3.10a (0.29)
0.39a (0.04)
2
6.03b (0.40)
3.36~ (0.34)
0.46b (0.06)
3
6.27~ (0.29)
2.86b (0.22)
0.43c (0.04)
4
6.16c (0.34)
3.17a
(0.31)
0.46b (0.05)
5
6.40d
2.9lb
(0.24)
0.4ld
(0.05)
3.08 (0.34) 0.43 All 6.30 (0.41) Values within columns with different superscripts differ (p < 0.01). Expressed as means (standard deviation). n = 200 spermatozoa per stallion.
(0.06)
(0.32)
374
Theriogenology
the measurements derived from live, unfixed spermatozoa reported here. A reduction in size of the spermatozoal head with fixation and staining also has been reported for human spermatozoa, with a reduction of approximately 15% in the length and width of the sperm head atIer fixation (12). In the present study, sperm head length and width were 20 and 13% lower, respectively, after Feulgen staining than in live, unfixed spermatozoa. Estimates of sperm head length and width in our study appeared lower than those reported in an earlier study by Davis et al. (5) in which equine spermatozoa were stained with a modified Papanicolaou stain. The acid hydrolosis of spermatozoa during Feulgen staining resulted in loss of sperm membranes, including the acrosome, with a concommitant reduction in the size of the spermatozoal head. In the current study, there was no or a very poor correlation between estimates of sperm head length and width based upon automated image analysis of Feulgen-stained sperm heads compared with measurements made on live, unfixed spermatozoa. Apparently, factors other than the length and width of the sperm nucleus affect corresponding measurements of live, unfked spermatozoa. However, the aspect ratio (WidtMength) was strongly correlated between the two methods. This implies that some aspects of the sperm nucleus are related to the shape of live, unfixed spermatozoa. For Feulgen-stained spermatozoa, there were significant effects of stallion and ejaculate on all metric parameters determined via image analysis. Similar differences between males and between ejaculates have been reported in humans, and, in general, the variablitiy between males is greater than that between ejaculates within males (13). However, in some studies, the variability between ejaculates has been reported to be equivalent or greater than the variability between males (12,lS). Two indices of sperm head shape, oval factor and aspect ratio, were also different between stallions. Although aspect ratio was affected by ejaculate and stallion, oval factor was not significantly affected by ejaculate and therefore may represent a more constant determination of sperm head shape across stallions. There were significant differences between stallions (length and width) as well as between ejaculates (length only) for measurements of the sperm head of live, unfixed spermatozoa. No other reports were found regarding expected variability in head length and width of live, unfixed equine spermatozoa. Based upon our results, it appears that both stallion and ejaculate within stallion affect head dimensions of equine spermatozoa. A variety of staining techniques have been used in computerized, image-based morphometry of spermatozoa. These include Feulgen staining (17), Papanicolaou staining or modifications thereof (5,6,7,10,12,14,18), and Shorr staining (20). In the current study, Feulgen staining of the sperm nucleus facilitated segmentation of digitized image to identity the sperm nucleus. The rate of digitization errors (misidentification of individual sperm heads) in the current study (5.4%) was due almost entirely to overlapping sperm heads that were identified as a single sperm head. Acid hydrolysis of spermatozoa during the Feulgen reaction apparently removed much of the debris present in the ejaculate and also removed much of the sperm midpiece, which is a source of digitization errors in studies with other staining methods (12). Overall, the incidence of sperm digitization errors appeared comparable to those reported in other studies with computerized image analysis of stained spermatozoa (5,6). The relatively wide gate settings for selection of spermatozoal heads may have resulted in a higher proportion of overlapping sperm heads that were incorrectly identified as single sperm heads.
Theriogenology
375
Feulgen staining of spermatozoa and other.cell types has been used along with image-based cytometry to measure DNA content in these cells (4,15,23). There were differences between stallions in the current study in integrated density of sperm heads which was determined based upon area and grey-scale value of Feulgen-stained specimens. Because the monochromatic illumination was not used in this study and because no background subtraction was used to correct images, no inferences can be drawn relative to differences in DNA content between stallions in this study as would be required for quantitative image cytometry (4). Future studies could potentially use this methodology to examine DNA content as well as sperm head morphometry in stallions. Sperm head morphometry determined by computerized image analysis has been associated with fertility in humans primarily through work in in vitro fertilization (12,14). From these studies, it appears that the degree of variation in sperm head size rather than the mean sperm head size is the better predictor of fertility in humans (12). Williams et al. (21) suggested that stallions that had a CV greater than 6.0 in sperm head length were more likely to have reduced fertility. Their study was based on observations made on 47 stallions in which sperm head length was determined by projecting a microscopic image at a defined magnification. The stallions used in the current study were classi&d as having normal fertility based upon previous breedings to those stallions. Future studies should address the possible relationship between variation in sperm head size and fertility in the stallion to confirm the observation of Williams et al. (21) utilizing available technology for rapid assessment of large numbers of sperm heads based on image analysis. In summary, computerized image analysis of Feulgen-stained equine spermatozoa revealed a significant variation between normal stallions in sperm head size, shape and integrated density. Variation between stallions was generally greater than variation between ejaculates within stallion in the measured parameters. A similar variation in the size of live, unlixed sperm heads was also noted. Sperm head size as determined from Feulgen-stained spermatozoa was smaller than that determined from live, unfixed spermatozoa. The ability to conduct rapid, semi-automated analyses of sperm head shape and size of Feulgen-stained spermatozoa may provide a fairly rapid means to assess these parameters in stallion spermatozoa. Although no associations with fertility could be made in this study, this technique may have future application as a method for evaluation of stallion fertility based upon differences in sperm head size or, more likely, differences in the amount of variation in sperm head size within a stallion. REFERENCES 1. 2. 3. 4.
5.
Barth AD, Oko RJ. Abnormal Morpholgy of Bovine Spermatozoa. Iowa State University Press, Ames, IA, 1989;1-280. Bielanski W, Dudek E, Bittmar A, Kosiniak K. Some characteristics of common abnormal forms of spermatozoa in highly fertile stallions. J Reprod Fertil 1982;Suppl32: 21-26. Bielanski W, Kaczmarski F. Morphology of spermatozoa in semen from stallions of normal fertility.. J Reprod Fertil 1979;27: 39-45. Chieco P, Jonker A Melchiorri C, Vanni G, Van Noorden CJ. A user’s guide for avoiding errors in absorbance image cytometry: a review with original experimental observations. Histochem J 1994;26: 1-19. Davis RO, Gravance CC, Casey PJ. Automated morphometric analysis of stallion spermatozoa. Am J Vet Res 1993;54: 1808-1811.
376
Theriogenology
6.
Davis RO, Bain DE, Siemers RJ, Thai DM, Andrew JB, Gravance CG. Accuracy and precision of the CellForm-Human automated sperm morphometry insttument. Fertil Steril 1992;58: 763-769. Davis RO, Gravance CG. Standardization of specimen preparation, staining, and sampling methods improves automated sperm-head morphometry analysis. Fertil Steril 1993;59: 412-417. Enginsu ME, Dumoulin JCM, Pieters MH, Bras M, Evers JLH, Geraedts JPM. Evaluation of human sperm morphology using strict criteria after Diff-Quik staining: Correlation of morphology with fertilization in vitro. Human Reprod 1991;6: 854-858. Grow DR, Oehninger S, Sehman HJ, Toner JP, Swanson RJ, Kruger TF, Muasher SJ. Sperm morphology as diagnosed by strict criteria: probing the impact of teratozoospermia on fertilization rate and pregnancy outcome in a large in vitro fertilization population. Fertil Steril 1994;62: 559-567. Jagoe JR, Washbrook NP, Hudson EA. Morphometry of spermatozoa using semiautomatic image analysis. J Clin Path01 1986;39: 1347-1352.13. Jasko DJ, Lein DH, Foote RH. Determination of the relationship between sperm morphologic classifications and fertility in stallions: 66 cases (1987-1988). J Am Vet Med Assoc 1990;197: 389-394. Katz DF, Over-street JW, Samuels SJ, Niswander PW, Bloom TD, Lewis EL. Morphometric analysis of spermatozoa in the assessment of human male fertility. J Androl 1986;7: 203-210. Kruger TF, Acosta AA, Siions KF, Swanson RJ, Matta JF, Oehninger S. Predictive value of abnormal sperm morphology in in vitro fertilization. Fertil Steril 1988;49: 112-l 17. Kruger TF, DuToit TC, Franken DR, Acosta AA, Oehninger SC, Menkveld R, Lombard CJ. A new computerized method of reading sperm morphology (strict criteria) is as efficient as technician reading. Fertil Steril 1993;59: 202-209. Makkus AC, van ‘t Hof-Grootenboer AE, Pahlplatz MM, de Wilde PC, Gemmink AJ, Cuypers VM, Vooijs GP. Practical aspects of fixatives in high resolution nuclear image analysis. Cytometry 1994;15: 302-310. Menkveld R, Stander FS, Kotze TJ, Kruger TF, van Zyl JA. The evaluation of morphological characteristics of human spermatozoa according to stricter criteria. Hum Reprod 1990;5: 586-592. Moruzzi JF, Wyrobek AJ, Mayall BH, Gledhill BL. Quantification and classificaiton of human sperm morphology by computer-assisted image analysis. Fertil Steril 1988;50: 142-152. Schrader SM, Turner TW, Simon SD. Longitudinal study of semen quality of unexposed workers: sperm head morphometry. J Androl 199O;ll: 32-39. Voss JL, Pickett SW, Squires EL. Stallion spermatozoal morphology and motility and their relationship to fertility. J Am Vet Med Assoc 1981;178: 287-289. Wang C, Leung A Tsoi WL, Leung J, Ng V, Lee KF, Chart SY. Computer-assisted assessment of human sperm morphology: comparison with visual assessment. Fertil Steril 1991;55: 983-988. Williams WL, Savage A. A study of the head length of equine spermatozoa. Can J Res 1930;3: 327-332. World Health Organization. WHO laboratory manual for the examination of human semen and semen-cervical mucus interaction. Cambridge, University of Cambridge, 1987; l-67.
7.
9.
10. 11.
12.
13. 14.
15.
16.
17.
18. 19. 20.
21. 22.
Theriogenology 23.
377
Zhu J, Ban&t CL, Lippes J, Pacey AA, Cooke ID. The sequential effects of human cervical mucus, oviductal fluid, and follicular fluid on sperm function. Fertil Steril 1994;61: 1129-l 135.
equine, spermatozoa, image analysis, measurement, sperm head
This research was supported by the Harry M. Zweig Memorial Fund for Equine Research and Thornbrook Farms. The authors thank M. Lillard for technical assistance and P. G. A. Thomas and I. Dobrinski for a review of the manuscript.
Theriogenology 44:367377,1995 0 1995 by Elsevier Science Inc. 655 Avenue of the Americas, New York, NY 10010
0093-691 X/95/$1 0.00 SSDI 0093-691 X(95)001 91 -A
368
Theriogenology INTRODUCTION
Morphology of spermatozoa has been used in the fertility evaluation of stallions (2,3,11,19), and abnormal sperm morphology has been associated with reduced fertility (11). Morphologic assessment, as currently conducted, remains extremely subjective, particularly regarding the size and shape of the sperm head. A moderate negative correlation has been reported between the incidence of abnormal heads and stallion fertility (11). Previously, alterations in sperm head size have been either gross microcephaiy or macrocephaly as determined by subjective assessement (11). In human male fertility assessment, strict, objective morphologic guidelines have been established for evaluation of the size and shape of the sperm head (8,16,22). Based on these guidelines, size and shape of the sperm head appear to be well correlated with IVF results in humans (9,13,16). Measurements can be determined by either eyepiece micrometer or by automated, computer-assisted image analysis systems (7,10,14,16). Automated systems hold the potential advantage of reducing technical variation that is inherent in manual, subjective morphology analysis. However, their use requires careful specimen preparation and staining in order to reduce digitization errors (7). The widespread availability of videoimage analysis on microcomputers provides the opportunity to evaluate more critically morphometric parameters in stallion spermatozoa. The purpose of the work described here was: 1) to examine the length and width of live, unfixed stallion sperm heads and 2) to evaluate the use of semi-automated measurements of the Feulgen-stained stallion sperm head with an image analysis program. MATERIALS AND METHODS Five stallions ranging in age from 5 to 21 y were used in this experiment (Table 1). These stallions were considered to have normal fertility based upon previous breeding history and upon the parameters of sperm number, progressive motility and subjectively assessed spermatozoa1 morphology (19; Table 1). Two ejaculates were collected with an artificial vagina within a 30-d interval (April) from each stallion. Immediately after collection, 10 ul of raw semen were smeared onto a warm (37°C) slide and allowed to air dry. The smears were subsequently stained for videomicroscopy. An additional aliquot of raw semen was transported back to the laboratory in an insulated container and was used for preparation of wet mounts for videomicroscopy. All reagents used in this study were from Fisher Scientific, Rochester, New York unless otherwise indicated. Air-dried semen smears were Feulgen-stained based on a modification of the method of Barth (1). The smears were allowed to dry for a minimum of 1 h prior to staining. Smears were then hydrolyzed with 5 N HCl for 30 min, washed in running water for 5 min, and then treated with Schiffs reagent for 30 min. Slides were rinsed in distilled water for 2 min, allowed to dry, mounted in permount and coverslipped. For measurements of live, unfixed spermatozoa, 1.0 ul of raw semen was placed on a slide and mounted with a 22x22-mm coverslip (12). This thin semen smear served to partially immobilize
Theriogenology
369
spermatozoa to prevent problems with accurate image acquisition. Prior to preparation of wetmounted, live spermatozoa, semen was maintained at 37°C. Temperature of slides was also maintained at 37°C by a microscope stage heater (Kocky Mountain Instruments, Fort Collins, CO) during acquisition of videoimages of live, unfixed spermatozoa. Feulgen-stained smears were examined via bright-field microscopy (Olympus BH-2 microscope, Olympus, Inc., Lake Success, NY) with a xl00 oil immersion objective (plan apochromatic; N.A. 1.3) and wetmounts of live, unfixed spermatozoa were examined via phasecontrast microscopy with a xl00 oil immersion objective. For bright-field microscopy, Kohler illumination was established, and a green filter was placed below the stage condenser to enhance contrast of the acquired bright-field and phase-contrast images. A x3.3 photoeyepiece was used along with a CCD camera (CCD 72 camera; Dage-MTI, Inc., Michigan City, IN) to acquire videoimages. The intensity of the halogen bulb and camera black level were standardized across all samples. The autogain feature of the CCD 72 camera was used to provide standard field illumination between samples. Linear grey scale reproductions were produced with a camera gamma setting of 1.O. Videoimages of spermatozoa were digitized (640x480 pixels at 8-bit depth) with a Scion LG3 frame grabber (Scion Corporation, Walker&he, MD) in a Macintosh Quadra 800 microcomputer (Apple Computer Corp., Cupertino, CA). The image analysis system was calibrated against a stage micrometer (0.098 urn per pixel), and the final image magnification on the video display was x3850. Camera aspect ratio was adjusted within software to compensate for differences (< 2%) in vertical and horizontal measurements. Digitized images were analyzed with the NIH Image program (ver. 1.54, Wayne Rasband, National Institute of Mental Health Health). For semi-automated measurements of spermatozoa1 heads, Feulgen-stained smears were scanned by the operator to locate a region of the smear with an adequate number of spermatozoa per field for analysis. Once a region was determined on the slide, sequential, nonoverlapping fields were digitized, and images were stored prior to analysis. The images were analyzed to determine area, perimeter, major ellipse (length), minor ellipse (width), and mean grey scale value. Major and minor ellipses represent the lengths of the axes of the best fitting ellipses. The oval factor (4 n area&rime&) and aspect ratio (widtMength) were also calculated to evaluate head shape (10,18). The integrated density of each sperm head was calculated based upon the product of area and mean grey scale value for that sperm head. Spermatozoal heads were differentiated based upon image segmentation within a predetermined range of grey-scale values and subsequent analysis of spermatozoal heads as particles within a range of 7.7 to 17.3 pm2 (Figure la and b). The range of particle sizes to be included or excluded as spermatozoal heads was determined in preliminary experiments. A total of 300 spermatozoa was analyzed per slide and duplicate slides were analyzed To determine the frequency of digitization errors that resulted in for each ejaculate. misidentification of spermatozoa, an operator reviewed all images to determine the frequency of debris or overlapping spermatozoa that were measured as individual spermatozoa. For wet-mount preparations of live, unfixed spermatozoa, digitized phase-contrast images of spermatozoa were individually measured by an operator with the NIH Image software. Length and width of the spermatozoal head were determined by arbitrary measurement of the greatest width and length of the spermatozoal head. A total of 100 spermatozoa was measured for each ejaculate.
370
Theriogenology
Data were exported from NIH Image to Microsott Excel (ver. 4.0; Microsoft Corp., Redmond, WA) and analyzed in Stat&x (ver. 4.1; Analytical Software, Tallahassee, FL). The design consisted of slides nested within ejaculates and ejaculates nested within stallions. Data were analyzed by the General Analysis of Variance in Statistix. Means were ranked and pairwise comparisons between individual means were computed via Tukey’s HSD at a < 0.01. Correlations between corresponding measurements from Feulgen-stained preparations and raw sperm measurements were also computed in Statistix. For comparisons between Feulgen-stained spermatozoa and live, unfixed spermatozoa, data from Feulgen-stained spermatozoa were arbitrarily truncated at 100 cells in order to provide equal numbers of spermatozoa for comparison by ANOVA and computation of Pearson’s correlation coefficient.
RESULTS For semi-automated analysis of Feulgen-stained spermatozoa, there were significant effects (P < 0.001) of stallion and ejaculate but not slide on area, perimeter, minor axis and integrated density of spermatozoal heads (Tables 2 and 3). In addition, stallion, ejaculate and slide affected (P < 0.01) major axis of spermatozoal heads and aspect ratio (Tables 2 and 3). Stallion, but not ejaculate or slide, significantly affected (P < 0.001) oval factor for spermatozoal heads (Table 3). For length measurements derived from phase-contrast images of live, untixed spermatozoa, there were significant effects of stallion (P < 0.001; Table 4). For corresponding width measurements of live, untixed spermatozoa, there were significant effects (P < 0.001) of both stallion and ejaculate (Table 4). Aspect ratio for live, unfixed spermatozoa was significantly affected (P < 0.001) by both stallion and ejaculate (Table 4). Length and width measurements derived from raw, unfixed spermatozoa were poorly correlated with corresponding major and minor axis determinations derived from Feulgen-stained spermatozoa by semi-automated image analysis (r = 0.03 and 0.08, respectively). Length and width measurements from raw spermatozoa were significantly larger (P < 0.01) than the corresponding measurements of major and minor axis from Feulgen-stained spermatozoa (Figure 2). Aspect ratios derived from Feulgen-stained spermatozoa were significantly larger (P < 0.01) than aspect ratios of live, unfixed spermatozoa (0.53 f 0.07 vs. 0.43 f 0.07, respectively). Aspect ratios derived from the 2 methods were moderately correlated (r = 0.66). The overall frequency of digitization errors (misidentification of spermatozoa) for Feulgenstained spermatozoa was 5.4% (range of 1.7 to 9%). The most common digitization error appeared to be overlapping spermatozoa that were identified and measured as a single spermatozoa (Figure lb). DISCUSSION Measurements of stained sperm heads in the stallion have been reported previously by investigators using either a projected image (21), scanning electron microscopy (3) or automated computerized morphometry (5). In general, values derived from studies that used fixation and/or drying (including the Feulgen stained spermatozoa in the present study) appeared to be lower than
of Feulgen Figure 1. Digital micrograph stained equine spermatozoa (a). Digital micrograph after image segmentation (b) to identify individual spermatozoa (outlined with black line). Note that overlapping spermatozoa are not selected and that sperm heads touching the edge of field are not selected. Bar represents 10 pm.
Live
m
Length
q
Width
Feulgen-stain
Figure 2. Length and width measurements of equine sperm heads determined from either five, unfixed spermatozoa (live) or from Feulgen-stained sperm heads. Data presented as mean + std deviation. Corresponding values differ (P < 0.01) between live and Feulgen-stained spermatozoa.
Theriogenology
372 Table 1.
Age, breed and semen parameters for 2 ejaculates from each of stallions.
Stallion
Age
Breed
1
6
Arabian
2
3
4
5
Table 2.
5
Thorough-
bred
21
17
l9
Warmblood
Morgan
Cl-
Ejaculate
Volume (a
Concentration 106hd
Total no. of Spermatowa ww
Motility %
Normal ;orpho””
1
45
151
6.8
80
86
2
40
150
6
65
82
1
100
135
13.5
70
-57
2
45
60
2.7
75
67
1
110
130
14.3
70
52
2
90
122
11
65
49
1
55
194
10.7
80
56
2
40
100
4
75
60
1
100
417
41.7
50
52
2
60
385
23.1
70
50
Mean and standard deviation of metric parameters of Feulgen-stained heads determined by computerized image analysis.
sperm
Stallion
Area (pm2)
Perimeter(pm)
Major ellipse (urn)
Minor ellipse (pm)
1
10.22a (1.43)
12.98a (1.17)
5.01a (0.41)
2.59a (0.25)
2
10.49b (1.67)
12.78b (1.42)
4.75b (0.44)
2.80b (0.27)
3
11.02c (1.34)
13.36~ (1.07)
5.20~ (0.36)
2.70~ (0.25)
4
ll.llc (1.32)
13.19d (1.07)
5.02a (0.36)
2.81b (0.23)
5
11.29d (1.34)
13.69e (1.09)
5.37d (0.36)
2.68~ (0.27)
All
10.82 (1.49;13.7)
13.20 (1.21; 9.2)
5.07 (0.44; 8.7)
2.72 (0.27; 9.8)
Values within columns with diierent superscripts differ (P < 0.01). Expressed as mean data. n = 600 spermatozoa per stallion, For overall, values in parentheses represent standard deviation; CV).
373
Theriogenology Table 3.
Shape parameters and integrated density of Feulgen-stained determined by image analysis.
Stallion
Oval Factor
Aspect Ratio
Integrated density
1
0.76a (0.04)
0.52a (0.06)
2429a (347)
2
0.8lb (0.06)
0.59b (0.06)
2534b (413)
3
0.78~ (0.04)
0.52a (0.05)
2687~ (329)
4
0.8Ob (0.05)
0.56~ (0.06)
2695c .(326)
5
0.76a (0.05)
0.50d (0.06)
2761d (331)
All
0.78 (0.05; 6.4)
0.54 (0.07; 12.4)
2621 (371; 14.2)
sperm heads as
Oval factor expressed as (4 x area/perimetefi. Aspect ratio expressed as (width/length). Integrated density expressed as (area * mean grey scale value). Values within columns with different superscripts differ (P < 0.01). Expressed as means (standard deviation). n = 600 spermatozoa per stallion. For overall, values in parentheses represent standard deviation; CV). Table 4.
Length and width of fresh, live spermatozoa as determined by manual measurements on spermatozoa from 5 stallions:
Stallion
Length (pm)
Width (pm)
Aspect Ratio
1
6.63a (0.43)
3.10a (0.29)
0.39a (0.04)
2
6.03b (0.40)
3.36~ (0.34)
0.46b (0.06)
3
6.27~ (0.29)
2.86b (0.22)
0.43c (0.04)
4
6.16c (0.34)
3.17a
(0.31)
0.46b (0.05)
5
6.40d
2.9lb
(0.24)
0.4ld
(0.05)
3.08 (0.34) 0.43 All 6.30 (0.41) Values within columns with different superscripts differ (p < 0.01). Expressed as means (standard deviation). n = 200 spermatozoa per stallion.
(0.06)
(0.32)
374
Theriogenology
the measurements derived from live, unfixed spermatozoa reported here. A reduction in size of the spermatozoal head with fixation and staining also has been reported for human spermatozoa, with a reduction of approximately 15% in the length and width of the sperm head atIer fixation (12). In the present study, sperm head length and width were 20 and 13% lower, respectively, after Feulgen staining than in live, unfixed spermatozoa. Estimates of sperm head length and width in our study appeared lower than those reported in an earlier study by Davis et al. (5) in which equine spermatozoa were stained with a modified Papanicolaou stain. The acid hydrolosis of spermatozoa during Feulgen staining resulted in loss of sperm membranes, including the acrosome, with a concommitant reduction in the size of the spermatozoal head. In the current study, there was no or a very poor correlation between estimates of sperm head length and width based upon automated image analysis of Feulgen-stained sperm heads compared with measurements made on live, unfixed spermatozoa. Apparently, factors other than the length and width of the sperm nucleus affect corresponding measurements of live, unfked spermatozoa. However, the aspect ratio (WidtMength) was strongly correlated between the two methods. This implies that some aspects of the sperm nucleus are related to the shape of live, unfixed spermatozoa. For Feulgen-stained spermatozoa, there were significant effects of stallion and ejaculate on all metric parameters determined via image analysis. Similar differences between males and between ejaculates have been reported in humans, and, in general, the variablitiy between males is greater than that between ejaculates within males (13). However, in some studies, the variability between ejaculates has been reported to be equivalent or greater than the variability between males (12,lS). Two indices of sperm head shape, oval factor and aspect ratio, were also different between stallions. Although aspect ratio was affected by ejaculate and stallion, oval factor was not significantly affected by ejaculate and therefore may represent a more constant determination of sperm head shape across stallions. There were significant differences between stallions (length and width) as well as between ejaculates (length only) for measurements of the sperm head of live, unfixed spermatozoa. No other reports were found regarding expected variability in head length and width of live, unfixed equine spermatozoa. Based upon our results, it appears that both stallion and ejaculate within stallion affect head dimensions of equine spermatozoa. A variety of staining techniques have been used in computerized, image-based morphometry of spermatozoa. These include Feulgen staining (17), Papanicolaou staining or modifications thereof (5,6,7,10,12,14,18), and Shorr staining (20). In the current study, Feulgen staining of the sperm nucleus facilitated segmentation of digitized image to identity the sperm nucleus. The rate of digitization errors (misidentification of individual sperm heads) in the current study (5.4%) was due almost entirely to overlapping sperm heads that were identified as a single sperm head. Acid hydrolysis of spermatozoa during the Feulgen reaction apparently removed much of the debris present in the ejaculate and also removed much of the sperm midpiece, which is a source of digitization errors in studies with other staining methods (12). Overall, the incidence of sperm digitization errors appeared comparable to those reported in other studies with computerized image analysis of stained spermatozoa (5,6). The relatively wide gate settings for selection of spermatozoal heads may have resulted in a higher proportion of overlapping sperm heads that were incorrectly identified as single sperm heads.
Theriogenology
375
Feulgen staining of spermatozoa and other.cell types has been used along with image-based cytometry to measure DNA content in these cells (4,15,23). There were differences between stallions in the current study in integrated density of sperm heads which was determined based upon area and grey-scale value of Feulgen-stained specimens. Because the monochromatic illumination was not used in this study and because no background subtraction was used to correct images, no inferences can be drawn relative to differences in DNA content between stallions in this study as would be required for quantitative image cytometry (4). Future studies could potentially use this methodology to examine DNA content as well as sperm head morphometry in stallions. Sperm head morphometry determined by computerized image analysis has been associated with fertility in humans primarily through work in in vitro fertilization (12,14). From these studies, it appears that the degree of variation in sperm head size rather than the mean sperm head size is the better predictor of fertility in humans (12). Williams et al. (21) suggested that stallions that had a CV greater than 6.0 in sperm head length were more likely to have reduced fertility. Their study was based on observations made on 47 stallions in which sperm head length was determined by projecting a microscopic image at a defined magnification. The stallions used in the current study were classi&d as having normal fertility based upon previous breedings to those stallions. Future studies should address the possible relationship between variation in sperm head size and fertility in the stallion to confirm the observation of Williams et al. (21) utilizing available technology for rapid assessment of large numbers of sperm heads based on image analysis. In summary, computerized image analysis of Feulgen-stained equine spermatozoa revealed a significant variation between normal stallions in sperm head size, shape and integrated density. Variation between stallions was generally greater than variation between ejaculates within stallion in the measured parameters. A similar variation in the size of live, unlixed sperm heads was also noted. Sperm head size as determined from Feulgen-stained spermatozoa was smaller than that determined from live, unfixed spermatozoa. The ability to conduct rapid, semi-automated analyses of sperm head shape and size of Feulgen-stained spermatozoa may provide a fairly rapid means to assess these parameters in stallion spermatozoa. Although no associations with fertility could be made in this study, this technique may have future application as a method for evaluation of stallion fertility based upon differences in sperm head size or, more likely, differences in the amount of variation in sperm head size within a stallion. REFERENCES 1. 2. 3. 4.
5.
Barth AD, Oko RJ. Abnormal Morpholgy of Bovine Spermatozoa. Iowa State University Press, Ames, IA, 1989;1-280. Bielanski W, Dudek E, Bittmar A, Kosiniak K. Some characteristics of common abnormal forms of spermatozoa in highly fertile stallions. J Reprod Fertil 1982;Suppl32: 21-26. Bielanski W, Kaczmarski F. Morphology of spermatozoa in semen from stallions of normal fertility.. J Reprod Fertil 1979;27: 39-45. Chieco P, Jonker A Melchiorri C, Vanni G, Van Noorden CJ. A user’s guide for avoiding errors in absorbance image cytometry: a review with original experimental observations. Histochem J 1994;26: 1-19. Davis RO, Gravance CC, Casey PJ. Automated morphometric analysis of stallion spermatozoa. Am J Vet Res 1993;54: 1808-1811.
376
Theriogenology
6.
Davis RO, Bain DE, Siemers RJ, Thai DM, Andrew JB, Gravance CG. Accuracy and precision of the CellForm-Human automated sperm morphometry insttument. Fertil Steril 1992;58: 763-769. Davis RO, Gravance CG. Standardization of specimen preparation, staining, and sampling methods improves automated sperm-head morphometry analysis. Fertil Steril 1993;59: 412-417. Enginsu ME, Dumoulin JCM, Pieters MH, Bras M, Evers JLH, Geraedts JPM. Evaluation of human sperm morphology using strict criteria after Diff-Quik staining: Correlation of morphology with fertilization in vitro. Human Reprod 1991;6: 854-858. Grow DR, Oehninger S, Sehman HJ, Toner JP, Swanson RJ, Kruger TF, Muasher SJ. Sperm morphology as diagnosed by strict criteria: probing the impact of teratozoospermia on fertilization rate and pregnancy outcome in a large in vitro fertilization population. Fertil Steril 1994;62: 559-567. Jagoe JR, Washbrook NP, Hudson EA. Morphometry of spermatozoa using semiautomatic image analysis. J Clin Path01 1986;39: 1347-1352.13. Jasko DJ, Lein DH, Foote RH. Determination of the relationship between sperm morphologic classifications and fertility in stallions: 66 cases (1987-1988). J Am Vet Med Assoc 1990;197: 389-394. Katz DF, Over-street JW, Samuels SJ, Niswander PW, Bloom TD, Lewis EL. Morphometric analysis of spermatozoa in the assessment of human male fertility. J Androl 1986;7: 203-210. Kruger TF, Acosta AA, Siions KF, Swanson RJ, Matta JF, Oehninger S. Predictive value of abnormal sperm morphology in in vitro fertilization. Fertil Steril 1988;49: 112-l 17. Kruger TF, DuToit TC, Franken DR, Acosta AA, Oehninger SC, Menkveld R, Lombard CJ. A new computerized method of reading sperm morphology (strict criteria) is as efficient as technician reading. Fertil Steril 1993;59: 202-209. Makkus AC, van ‘t Hof-Grootenboer AE, Pahlplatz MM, de Wilde PC, Gemmink AJ, Cuypers VM, Vooijs GP. Practical aspects of fixatives in high resolution nuclear image analysis. Cytometry 1994;15: 302-310. Menkveld R, Stander FS, Kotze TJ, Kruger TF, van Zyl JA. The evaluation of morphological characteristics of human spermatozoa according to stricter criteria. Hum Reprod 1990;5: 586-592. Moruzzi JF, Wyrobek AJ, Mayall BH, Gledhill BL. Quantification and classificaiton of human sperm morphology by computer-assisted image analysis. Fertil Steril 1988;50: 142-152. Schrader SM, Turner TW, Simon SD. Longitudinal study of semen quality of unexposed workers: sperm head morphometry. J Androl 199O;ll: 32-39. Voss JL, Pickett SW, Squires EL. Stallion spermatozoal morphology and motility and their relationship to fertility. J Am Vet Med Assoc 1981;178: 287-289. Wang C, Leung A Tsoi WL, Leung J, Ng V, Lee KF, Chart SY. Computer-assisted assessment of human sperm morphology: comparison with visual assessment. Fertil Steril 1991;55: 983-988. Williams WL, Savage A. A study of the head length of equine spermatozoa. Can J Res 1930;3: 327-332. World Health Organization. WHO laboratory manual for the examination of human semen and semen-cervical mucus interaction. Cambridge, University of Cambridge, 1987; l-67.
7.
9.
10. 11.
12.
13. 14.
15.
16.
17.
18. 19. 20.
21. 22.
Theriogenology 23.
377
Zhu J, Ban&t CL, Lippes J, Pacey AA, Cooke ID. The sequential effects of human cervical mucus, oviductal fluid, and follicular fluid on sperm function. Fertil Steril 1994;61: 1129-l 135.