A Flow Cytometric Method for Rapid Determination of Sperm Concentration and Viability in Mammalian and Avian Semen


Department of Clinical Studies, Section for Reproduction, Royal Veterinary and Agricultural University, Dyrlaegevej 68, 1870 Frederiksberg C, Denmark (e-mail: PC@kvl.dk).


ABSTRACT: A new flow cytometric method has been developed to rapidly determine sperm concentration and viability in semen from bulls and boars. Sperm viability was determined on the basis of staining with SYBR-14 and propidium iodide (PI), and this allowed detection of live (membrane-intact) sperm, dying (moribund) sperm, as well as dead cells. Fluorescent microspheres (beads) were used to determine sperm concentration. The use of SYBR-14 at 50 nM and PI at 12 μM in combination with the FACSCount diluent in the counting tubes resulted in a uniform staining after 2.5–5 minutes at room temperature. Reagent staining was reproducible enough to allow subsequent semiautomated analysis of data using Attractors software. In experiment 1, this method was used to analyze semen from boars, rams, rats, rabbits, humans, and turkeys. In experiment 2, Attractors analysis was performed by the FACSCount AF flow cytometer, and sperm concentration determination with this system was compared with results obtained by a spectrophotometer and an electronic cell counter, which is routinely used by bull artificial insemination centers. When compared to microscopic counting in a hemocytometer, the FACSCount AF flow cytometer was two and four times more accurate than the spectrophotometer and the electronic cell counter, respectively. In addition, the FACSCount AF flow cytometer determined both sperm concentration (coefficient of variation 3.3%) and sperm viability (coefficient of variation 0.7%) with high precision.

Accurate and precise determination of sperm concentration is essential for the artificial insemination industry to provide customers with insemination doses containing the desired number of sperm (Foote et al, 1978; Fenton et al, 1990; Woelders, 1991; Evenson et al, 1993; Donoghue et al, 1996). The primary method used for most species is to estimate sperm concentration from a spectrophotometric determination of turbidity of a diluted semen sample (Foote et al, 1978; Woelders, 1991; Donoghue et al, 1996). However, any debris present in the raw semen, such as cytoplasmic droplets, will affect the accuracy of this method (Evenson et al, 1993). The ejaculates from animals such as boars, stallions, and rabbits contain various amounts of gel particles, which make the spectrophotometric assessment particularly inaccurate (Woelders, 1991; Graham, 1994; Hansen et al, 2002).

Electronic cell counters provide a rapid determination of sperm concentration and are easy to use, however, any debris in the same size range as sperm may be included in the count (Evenson et al, 1993). Up to now, numbers in extended or frozen-thawed semen have been difficult to verify because most extenders contain egg yolk particles, fat droplets, or other particulate matter, which can affect measurements with electronic cell counters and spectrophotometers (Parks, 1992; Evenson et al, 1993). Use of hemocytometers for quality control of sperm count in extended semen has been limited because this method is slow, and multiple measurements of each sample are necessary to achieve acceptable precision (Freund and Carol, 1964). Evenson et al (1993) described a flow cytometric method in which beads (fluorescent microspheres) were used to determine sperm number in thawed straws with bull semen. However, this method required a skilled operator for analysis and a laborious preparation of beads. The method has therefore not been widely used.

Assessment of sperm concentration is an important procedure for evaluating effects of drugs on testis function in laboratory animals (Yamamoto et al, 1998). According to World Health Organization (1992) guidelines, assessment of sperm count is an essential part of the diagnostic and prognostic evaluation of male fertility. Several publications over the past decade indicate that human semen quality and sperm count is declining (Carlsen et al, 1992; Auger et al, 1995; Irvine et al, 1996; Aitken, 1999). Although guidelines for hemocytometric determination of sperm concentration have been proposed by the World Health Organization (1992), results are difficult to compare due to variations between laboratories and technicians (Auger et al, 2000). Attempts have been made to use flow cytometry to determine sperm concentration and to improve agreement between different laboratories (Eustache et al, 2001; Tsuji et al, 2002).

Microscopic estimation of the percentage of motile sperm in a semen sample is probably the most common method routinely used by the artificial insemination industry as well as by toxicology laboratories and human fertility clinics (Graham, 1994; Vetter et al, 1998; Auger et al, 2000). The microscopic assessment of motility suffers from human bias, which in combination with the evaluation of a relatively low number of sperm, leads to high intra-laboratory and inter-laboratory variability (Auger et al, 2000). The reproducibility of motility estimates has been improved by the use of computer-assisted sperm analysis (CASA), but small changes in assay conditions or instrument settings appear to significantly affect the results (Davis and Katz, 1992; Krause et al, 1993; Holt et al, 1994). Moreover, the sample throughput in most laboratories is too high to allow routine use of this time-consuming method.

Flow cytometry offers the possibility of analyzing a high number of sperm (>10 000) in less than 1 minute, and several hundred papers have been written on sperm measured by flow cytometry (eg, Gledhill et al, 1976; Garner et al, 1986; Morrell, 1991; Parks, 1992; Graham, 1994, 2001). In addition to achieving an objective assessment with high precision, the technique allows examination of several different characteristics of sperm such as plasma membrane integrity (sperm viability; Evenson et al, 1982; Garner et al, 1994, 1995), mitochondrial function (Evenson et al, 1982; Graham et al, 1990; Garner et al, 1997), acrosomal status (Graham et al, 1990; Thomas et al, 1997) and chromatin structure (Evenson et al, 1980). However, practical use of flow cytometry has been limited by the need for a skilled operator and expensive instrumentation. The FACSCount System (BD Biosciences Immunocytometry Systems, San Jose, Calif) is an easy-to-operate clinical flow cytometer for enumeration of lymphocyte populations in blood. The system has improved intra-laboratory and inter-laboratory precision and accuracy in CD4 lymphocyte quantitation (Strauss et al, 1996). Since 1996, the FACSCount instrument, with modifications, has been applied to improve the precision and accuracy in determination of sperm concentration. The protocol assesses sperm concentration with fluorescent microspheres (beads) and sperm viability using a modification of the SYBR-14 and propidium iodide (PI) method reported by Garner et al (1994). The objective of this paper is to describe the protocol and flow cytometric analysis, and to present data to indicate that this system can differentiate live, moribund, and dead sperm in mammalian and avian semen. A validation of the method with regard to determination of sperm concentration in raw bull semen is also presented.

Materials and Methods

Preparation of Counting Tubes

Sealed polyethylene counting tubes were provided by BD Biosciences Immunocytometry Systems. Each counting tube contained 400 μL of FACSCount diluent with a predetermined number of fluorescent microspheres (beads). Prior to use, tubes were vortexed for 5 seconds in an upright position and for 5 seconds inverted. After opening the tubes and adding dyes, tubes were capped and vortexed briefly (2 seconds) in an upright position.

Semen Dilution

Dilution of semen is necessary to achieve a suitable ratio between sperm and beads. The most optimal relationship is a ratio close to 1:1, which implies that the number of sperm or beads analyzed will be close to 5000 if a total of 10 000 events are acquired, provided background is low. Each lot of counting tubes contain a specific number of beads and variation from tube to tube is small (lot CV <2.4%). For most batches, the number of beads per tube was around 100 000. Sperm concentration in raw semen for most species varies significantly between males and ejaculates. For bulls and boars, there is a 10-fold variation from the most dilute ejaculate (respectively, 300 × 106 and 100 × 106 sperm/mL) to the most concentrated ejaculates (3000 × 106 and 1000 × 106 sperm/mL). For practical purposes, it is desirable to use the same dilution procedure for each type of semen. For raw bull semen, a 1:250 dilution was used, whereas raw boar semen was diluted 1:100. For frozen-thawed bull semen (concentration approximately 60–100 × 106 sperm/mL) or insemination doses of boar semen (20 to 50 × 106 sperm/mL), dilutions of, respectively, 1:20 and 1:10 were used. Dilution for bulls and boars were performed in phosphate-buffered saline (PBS; KH2PO4 2.6 mM, Na2HPO4 8.4 mM, NaCl 143.7 mM, pH 7.4) and an EDTA extender (Glucosemonohydrate 302.7 mM, Na3citrate 12.6 mM, NaHCO3 14.3 mM, Na2EDTA, 2H20 10 mM, pH 7.2; Weitze, 1991), respectively.

For other species, dilutions should be performed in a suitable extender for that species, and dilutions should be aimed at ensuring that an ejaculate with average sperm concentration results in a ratio between sperm and beads close to 1:1. Because sperm concentration in raw semen for most species varies significantly (10-fold), it is clear that the ratio between sperm and beads may vary from ejaculate to ejaculate. However, if 15 000 events are acquired per analysis, the smallest population will include at least 1000 events (Table 1).

Table 1. . Sperm: bead ratios for raw bull semen with sperm concentrations of 300 to 3000 × 106
Sperm Concentration (109/mL)Number of Sperm Added to Counting Tube* 
Sperm: Bead RatioSpermBeads
  1. *The initial dilution was 1:250, and 50 μL of the diluted sperm was added to a counting tube containing 100 000 beads. The number of sperm and bead events are shown for a file size of 15 000 events.

0.36.0 × 1040.6:156259375
0.51.0 × 1051:175007500
1.02.0 × 1052:110 0005000
1.53.0 × 1053:111 2503750
2.04.0 × 1054:112 0003000
2.55.0 × 1055:112 5002500
3.06.0 × 1056:112 8572143

Accurate and precise pipetting is essential to achieve a correct determination of sperm concentration. Initially, the electronic BioControl pipette (Labsystems, Helsinki, Finland) was used, but in more recent work, an electronic pipette that provides reverse pipetting and dispenses 50 μL has been used (FACSCount pipette, BD Biosciences Immunocytometry Systems).

Fluorescent Staining

SYBR-14 and PI (Molecular Probes, Eugene, Ore) were dissolved in, respectively, anhydrous DMSO (4 μM) or distilled water (1 mM). Aliquots of the two working solutions were mixed 1:1, and 10 μL was added per counting tube, resulting in a concentration of 50 nM SYBR-14 and 12 μM PI. Prepared counting tubes generally were used within 12 hours of preparation. After the initial dilution of the semen (see above), 40 μL of the diluted semen was added to a prepared counting tube. The counting tube was vortexed briefly (2 seconds) and incubated at room temperature. After 4 to 5 minutes, the counting tube was vortexed briefly and analyzed on the flow cytometer. In experiment 2, the dye solution included in the Sperm Counting Reagent (BD Biosciences Immunocytometry Systems) was used. To each counting tube, 20 μL of the dye solution was added, resulting in the same concentrations of SYBR-14 and PI as mentioned above.

Flow Cytometry and Attractor Analysis

Analyses were performed on a modified FACSCount flow cytometer with a fiber-coupled 488-nm external air-cooled laser. This instrument collects two parameters of fluorescence and one parameter of calculated “size” data for each event. Emission signals were separated by a 620 nm short-pass dichroic mirror. The green fluorescence was collected through a 515–545 nm band pass filter and the red fluorescence was collected through a 645 nm long-pass filter. The size parameter value is related to the particle size and is calculated by the instrument using the area and height of an event's red pulse. Data are collected into 256 channels with logarithmic amplification covering four decades. No compensation is used. Subsequent analysis of quantitative data was performed on a Macintosh Quadra 650 computer using Attractors Software (BD Biosciences Immunocytometry Systems). This software uses analysis regions, defined in multiple parameters, to classify the collected data into populations. Each region has an “attraction radius” that allows it to locate a nearby data population (Figure 1A and B). Small variations in the position of individual populations is therefore not problematic. Furthermore, debris in freezing extenders or gel particles in raw semen, which may cause problems in a straight fluorescence analysis, can be gated out using the size parameter (Figure 1C and D). Identification of the moribund sperm (dual stained) in the green-versus-size diagram (Figure 1D) circumvents problems relating to debris in the samples. The resulting numbers for fluorescent microspheres and for live, dead, and moribund sperm can be combined with the dilution rate and the bead number for the counting tube to provide the sperm concentration and the percentage of living cells (viability). With the FACSCount AF System used in experiment 2, the Attractors analysis has been incorporated into the software to automate the data processing. No additional analysis is necessary, and results for sperm concentration and viability are presented on a printout.

Figure 1.

. Attractors analysis of dot-plot with log of green fluorescence (green) versus log of red (red) fluorescence (A, B, C) and relative size versus log of green fluorescence (D). All signals were recorded on a 4-decade logarithmic scale but for green and red fluorescence this is shown as 256 channels in the analyzed cytograms. Small variations in staining intensity can be accommodated by the Attractors Software (BD Biosciences Immunocytometry Systems). With this software, the attractor rings are able to move in order to surround a population in the dot-plot. (A) The data have not been classified. (B) The attractor rings have moved in order to identify live sperm (1, green), dead sperm (2, red), moribund sperm (3, orange), and beads (4, blue). Debris are gray (5) and are not surrounded by an attractor ring. (C) Debris (ie, in frozen-thawed bull semen egg yolk particles or gel particles in raw semen) can make it difficult to identify the moribund sperm correctly. However, only the live and dead populations are classified in the green versus red cytograms. Moribund sperm are classified in the size versus green cytogram (D) as a population between the live sperm (green) and dead sperm (red). Note that the debris are largely unclassified, although they reach into the orange attractor ring in the green versus red cytogram (C).

Experimental Design

Experiment 1: FACSCount Cytograms

Semen was obtained from boars, rams, rats, rabbits, humans. and turkeys. Samples from two males were processed and analyzed for each species.

Boar. Diluted semen samples were supplied by Hatting KS (Ringsted, Denmark). One ejaculate per boar was collected by gloved hand technique and filtered to remove the gel. After microscopic assessment of sperm motility and determination of sperm concentration using a Corning 254 spectrophotometer (Sherwood Scientific Ltd, Manchester, United Kingdom), the ejaculates were diluted with EDTA extender to an approximate concentration of 30 × 106 sperm/mL. One semen sample per ejaculate was transported to the laboratory at approximately 18°C.

Ram. Two 0.5-mL straws with frozen semen were used for each ram. Straws were thawed in a water bath at 37°C for 30 seconds. The two straws were mixed and diluted to approximately 30 × 106 sperm/mL after assessment of sperm concentration using a Bürker-Türk hemocytometer (VWR International, Albertslund, Denmark).

Rat. Epididymal sperm samples from two mature rats were supplied by Scantox (Ll. Skensved, Denmark). The rats were anaesthetized with carbon dioxide and were killed by cervical dislocation. A 2-cm section of each vas deferens was removed and cut into sections of 0.5 cm. The sections were placed in 5 mL of PBS buffer at 37°C for 20 minutes and were vortexed every 5 minutes during incubation. After incubation, the tissue segments were removed and sperm concentration was assessed using a Bürker-Türk hemocytometer. Samples were subsequently diluted to 5 × 106 sperm/mL.

Rabbit. One ejaculate from each of two males was collected with an artificial vagina (IMV, Cedex, France). Samples from each ejaculate was diluted to approximately 30 × 106 sperm/mL in PBS buffer after determination of sperm concentration with a Bürker-Türk hemocytometer.

Human. Two donors provided one ejaculate each, which were processed according to World Health Organization (1992) protocols. Sperm concentration was determined with Bürker-Türk hemocytometer and a sample from each ejaculate was diluted to approximately 5 × 106 sperm/mL in PBS buffer.

Turkey. Diluted semen samples from two males were provided by Moorgut Kartzfehn (Bösel, Germany). A Bürker-Türk hemocytometer was used for determination of sperm concentration, and samples were diluted to approximately 30 × 106 sperm/mL using PBS.

Prior to staining, semen samples from boars, rams, rabbits, and turkeys were diluted in PBS buffer to 5 × 106 sperm/mL. For each sample, a 40-μL subsample was pipetted into a counting tube that was prepared with dyes as described above. Staining was performed at 20°C and after, 2.5, 5, 10, and 15 minutes, aliquots of 80 μL were decanted into 35 × 7 mm glass tubes (Bie & Berntsen A/S, Rødovre, Denmark) and analyzed on the FACSCount flow cytometer.

Experiment 2: Determination of Sperm Concentration in Bull Semen

During 5 experimental days, 50 ejaculates were collected from 28 bulls. A sample of 1 mL was taken from each ejaculate for further analysis by spectrophotometer, electronic cell counter, flow cytometer, and microscopic determination of sperm concentration using a Bürker-Türk hemocytometer.

Spectrophotometer. A 20-μL subsample of raw semen was diluted 1:200 in a cuvette using the Z069 Cavro autodilutor (IMV, Cedex, France). The cuvette was subsequently placed in the L'Aiglon spectrophotometer and measured. After the first measurement, the cuvette was covered by foil and inverted gently three times and remeasured. Two subsamples were processed per ejaculate, resulting in four measurements.

Electronic cell counter. For the Sysmex F-820 hemotology counter, a two-step dilution (1:50 000) was performed with the Sysmex AD-270 autodilutor (Sysmex, GmbH, Hamburg, Germany). Determination of cell count was obtained using the aperture for red blood cell determination and 0.25 mL of the dilution was counted. Two diluted subsamples were made per ejaculate, and each dilution was measured twice.

Flow cytometry. The FACSCount AF system was used in combination with BD Sperm Counting Reagent, which includes a solution of SYBR-14 and PI in anhydrous DMSO. To each counting tube, 20 μL of dye solution was added and resulted in concentration of 50 nM SYBR-14 and 12 μM PI. Two samples from each ejaculate were diluted 1:250 in PBS buffer using the Sysmex AD-270 autodilutor. From each dilution, 50 μL was transferred to two counting tubes using an electronic FACSCount pipette. Samples were collected with a threshold on green fluorescence. Sperm concentrations were automatically calculated by the flow cytometer software with user-provided dilution information.

Bürker-Türk hemocytometer. Depending on the determination of sperm concentration by the spectrophotometer, samples for microscopic counting were diluted from 1:100 (dilute ejaculates) to 1:750 (concentrated ejaculates). The purpose was to ensure that approximately 150 sperm were counted per 10 fields in the hemocytometer. Dilutions were prepared in polypropylene tubes of 2, 7, 13, or 30 mL (Sarsted, Nümbrecht, Germany). PBS buffer was used, and to immobilize sperm, 10 μL of 4% (w/v) formaldehyde was added. Dilutions were performed using a Dispensette III (Bie & Berntsen, Rødovre, Denmark) and an electronic FACSCount pipette (BD Biosciences Immunocytometry Systems).

From the diluted samples, two separate aliquots of approximately 7 μL were withdrawn with a FinnPipette (Labsystems, Helsinki, Finland) to fill the two counting areas of the Bürker-Türk hemocytometer. Filling the Bürker-Türk hemocytometer was always performed after the coverslip was applied and fixed by the two clips, and after Newton rings had been observed. After approximately 1 minute, the counting was carried out using phase-contrast microscopy and 200× magnification. In each side of the hemocytometer, 10 squares of 0.2 × 0.2 mm2 were counted. This procedure was performed by 3 technicians in random order for each ejaculate. Each operator recorded the time point for the counting procedure in order to be able to identify any difference between operators or a time-dependent effect on the results. To avoid bias, all dilutions of raw semen were performed by one person who did not participate in the counting procedure.

Statistical Analysis

Univariate analyses of hemocytometer counts, flow cytometry, electronic cell counter, and spectrophotometer results were based on variance component models with random effect of sample and replicate. Results were log-transformed to achieve approximate normality and variance homogeneity. The models had a fixed effect of time and technician (hemocytometer) as well as replicate and interactions thereof. Nonsignificant effects were removed before calculation of the variance components and coefficients of variation. Regression analysis with hemocytometer counts as an independent variable and with flow cytometry, electronic cell counter, or spectrophotometer as dependent variables were performed with standard least-squares regression and using a measurement error model. All regression analyses were carried out at the ejaculate level using the estimated values for the ejaculate from the univariate models. The measurement error analysis was based on an assumption of known ratio between the variances of independent variables (Fuller, 1987). The ratio was estimated from the replicate and corrected for the number of replicates. A sensitivity analysis on the value of the ratio was performed.


Experiment 1: FACSCount Cytograms

Dot-plot cytograms obtained with sperm from boars, rams, rats, rabbits, humans, and turkeys after staining for 5 minutes at 20°C are shown in Figure 2. In general, staining of raw semen samples for 2.5 minutes resulted in good separation between the different populations. However, for human semen, the separation between debris and live sperm was slightly better after staining for 5 minutes (Figure 2E). For frozen-thawed semen from rams (Figure 2B), an intermediate population of dying (moribund) cells was noted at all time points. Samples from boars (Figure 2A) and rabbits (Figure 2D) stained rapidly, but some samples contained a significant amount of debris (Figure 2D). Particularly good separation between the live and dead populations was noted for turkey sperm (Figure 2F). However, threshold on green fluorescence had to be reduced from channel 64 to channel 40 to include all dead sperm in the analysis.

Figure 2.

. Dot-plot cytograms printed by the FACSCount flow cytometer, all with log of green (y-axis) fluorescence versus log of red fluorescence (x-axis). Threshold was on green channel 64 except for turkey (channel 40, E). All dot-plots were obtained after a staining period of 5 minutes at 20°C. (A). Boar semen with a large population of live sperm (1), few dead sperm (2), few moribund sperm (3), beads (4), and debris (5). (B) Frozen-thawed ram sperm containing a significant amount of dead cells (2) as well as moribund sperm (3). (C) Rat epididymal sperm showing two small discrete populations of live (1) and dead (2) cells. (D) Rabbit semen contains gel particles that may be responsible for the intensive background (5). (E) A typical human sample with a lengthy population of live cells (1) and dead cells (2). (F) Dead turkey sperm has a relatively low green fluorescence intensity and threshold on green has been decreased to channel 40 in this dot-plot. Live sperm (1) and dead sperm (2) are easily separated.

Staining patterns did not change significantly when incubation was increased from 5 to 15 minutes, but it was noted that the percentage of viable sperm declined. For samples from boars, rats, rabbits, humans, and turkeys, the decline in percentage of viable cells was −0.2% to −0.4%/min. For the frozen-thawed samples of ram semen, the decline in percentage of viable cells was −0.7%/min.

Experiment 2: Determination of Sperm Concentration in Bull Semen

For the Bürker-Türk hemocytometer results, analysis of variance components showed that 96.0% of the total variation was caused by the ejaculates, whereas 0.1% of the variation was because of technician (P < .018) and 3.9% of the variation was random. Coefficient of variation for the hemocytometer results was 9.2%.Variation caused by ejaculates accounted for 98.3% of the variation in results with the L'Aiglon spectrophotometer and the remaining 1.7% of the variation was random. The coefficient of variation was 5.0%. For the Sysmex F-820 electronic cell counter, 97.4% of the variation was caused by the ejaculates, 2.6% of the variation was random, and the coefficient of variation was 6.4% for this method. The coefficient of variation for the FACSCount AF flow cytometer was 3.3% for determination of sperm concentration and 0.7% for the assessment of sperm viability. Respectively. 99.5% (concentration) and 99.2% (viability) of the variations were due to variation between ejaculates.

Regression lines obtained from the measurement error models are shown in Figure 3A (L'Aiglon spectrophotometer), Figure 3B (Sysmex F-820 electronic cell counter) and Figure 3C (FACSCount AF flow cytometer). The figures clearly show that the points are closest to the regression line for the FACSCount AF flow cytometer (Figure 3C) and are most scattered for the Sysmex F-820 electronic cell counter. The scattering of the points around the regression lines is the error in the measurement model (Table 2) and were 23 × 106 sperm/mL (FACSCount AF flow cytometer), 46 × 106 sperm/mL (L'Aiglon spectrophotometer), and 102 × 106 sperm/mL (Sysmex F-820 electronic cell counter). In other words, the accuracy of the FACSCount AF for determination of sperm concentration is approximately two times higher than for the L'Aiglon spectrophotometer and four times higher than for the Sysmex F-820 electronic cell counter. Using a simple least-squares regression analysis, the corresponding coefficients of determination (R2) were 0.99 (FACSCount AF), 0.96 (L'Aiglon), and 0.89 (Sysmex F-820). Intercepts with the Y-axis were not significantly different from zero for any of the three methods. The slopes of the regression lines (Table 2) were 1.05 (SE = 0.02; FACSCount AF), 1.05 (SE = 0.05; Sysmex F-820), and 1.12 (SE = 0.03; L'Aiglon). For the FACSCount AF flow cytometer and the L'Aiglon spectrophotometer, the slopes were significantly (P < .05) different from 1.0.

Figure 3.

. (A) The regression line obtained from the measurement error model for the L'Aiglon spectrophotometer versus counting of the sperm concentration in a Bürker-Türk hemocytometer. (B) The regression line obtained from the measurement error model for the Sysmex F-820 electronic cell counter versus counting the sperm concentration in a Bürker-Türk hemocytometer. (C) The regression line obtained from the measurement error model for the FACSCount AF flow cytometer versus counting the sperm concentration in a Bürker-Türk hemocytometer.

Table 2. . Regression analyses in the measurement error model for three flow cytometry systems compared to the Bürker-Türk hemocytometer
ComparisonEstimateSEEstimateSEError (SD)
L'Aiglon vs. Bürker-Türk21411.120.0346
Sysmex F-820 vs. Bürker-Türk−11671.050.05102
FACSCount AF vs. Bürker-Türk25231.050.0223


Since 1996, the aim of our work has been to develop a rapid, precise, and accurate method for determining sperm concentration, and which also assesses sperm viability simultaneously with high precision. The funding for this work was supplied by Danish cattle and pig breeders, and consequently, most of the effort has focused on these two species.

Initial work focused on the selection of a dye combination for use with either the 543 nm standard laser of the FACSCount or the 488 nm laser of a modified FACSCount. For the 488 nm laser, several of the SYTO-related dyes resolved live from dead sperm when combined with PI, but the published combination of SYBR-14 and PI worked best with incubations of 2.5 to 20 minutes. Dye combinations for the 543 nm laser proved more problematic. Although several SYTO dyes resolved live and dead sperm when paired with either PI or ethidium homodimer II (Molecular Probes), background noise for all tended to contaminate populations of interest. Consequently, further system development was focussed on the 488 nm laser.

The efficiency with which DNA binding dye pairs discriminate live cells from dead depends on a number of factors, including membrane permeability, spectral responses, and quantum yield. The SYBR-14 and PI combination has proven particularly effective for analysis of sperm viability (Garner et al, 1994). Propidium iodide is an impermeant DNA-binding dye, showing essentially no cellular uptake unless the membrane is compromised. SYBR-14 is a permeant DNA-binding dye that stains sperm with intact membranes as well as moribund or dead sperm. As with other dyes related to the SYTO dye series (Molecular Probes), the lipophilic character of the dye allows it to cross membranes; the rate of uptake depending on dye permeability. SYBR-14 demonstrates a rapid uptake as well as a high quantum yield (QY) with 488 nm excitation (QY = 0.65, data from Molecular Probes), which reflects the percentage of light actually emitted as fluorescence. Sperm stained with both PI and SYBR-14 show loss of green emission because the close proximity of the dyes in the DNA causes fluorescence resonance energy transfer from SYBR-14 to PI (Garner and Johnson, 1995; Vetter et al, 1998).

Garner et al (1994) observed that the percentage of live sperm decreased over time when stained with 100 nM SYBR-14. Garner et al (1994) used a 10-minute staining time, but suggested that the time interval between staining and flow cytometric analysis should be kept to a minimum. In order to decrease staining time, we initially investigated the effect of different media and pH and observed that the FACSCount diluent (used in the counting tubes) resulted in an effective staining with 50 nM SYBR-14 after 2.5 minutes of incubation at 20°C. This work was performed with raw, diluted, and frozen-thawed semen from bulls. In experiment 1, staining with 50 nM SYBR-14 and 12 μM PI was tested on semen from boars, rams, rats, rabbits, humans, and turkeys. For all species, good separation of the different populations was achieved on subsequent analysis on a modified FACSCount. Garner and Johnson (1995) used 100 nM SYBR-14 in combination with a HEPES dilution medium containing bovine serum albumin and observed good staining after 15 minutes of incubation at 36°C. In experiment 1, the percentage of viable sperm decreased −0.2% to −0.4% per minute for boar, rat, rabbit, human, and turkey sperm. With raw bull semen, the decline in sperm viability has been estimated to be −0.1% to −0.2% per minute (unpublished data). For frozen-thawed ram semen, an intermediate population of moribund (dying) sperm was noted, and the percentage of viable sperm decreased −0.7% per minute.

An intermediate population of moribund sperm has also been observed for frozen-thawed semen from other species such as bull and stallion, and the decrease in the percentage of viable sperm in general is higher than for raw samples. It is not known whether the decrease in the percentage of viable sperm is related to toxicity of SYBR-14 (or PI) or is caused by a slow uptake of PI in sperm with compromised membranes. Toxicity of several dyes used for staining of live sperm have been systematically examined (Downing et al, 1991), but such data are presently not available for SYBR-14. A short staining time (ie, 5 minutes) is recommended in order to minimize a time-dependent effect on the determination of sperm viability.

In experiment 2, sperm concentration was determined in 50 ejaculates from bulls using a spectrophotometer, an electronic cell counter, and the FACSCount AF flow cytometer. The regression lines for the different methods against microscopic counting of sperm concentration using the Bürker-Türk hemocytometer (Figure 3A, B, and C) clearly shows that the points around the line are least scattered for the FACSCount AF flow cytometer. The error estimated by the measurement model was estimated to 23 × 106 sperm/mL for the FACSCount AF flow cytometer (Table 2). The estimates for the L'Aiglon spectrophotometer and the Sysmex F-820 electronic cell counter were, respectively, two times (46 × 106 sperm/mL) and four times (102 × 106 sperm/mL) higher. In addition, determinations made by FACSCount AF was more repeatable (CV = 3.3%) in comparison to the L'Aiglon spectrophotometer (CV = 5.0%) and the Sysmex F-820 electronic cell counter (CV = 6.4%). The present results are in agreement with results obtained for boar semen (Hansen et al, 2002). In both studies, the estimated slopes for the regression against the Bürker-Türk hemocytometer have been slightly above 1.0. This is not regarded as a particular problem relating to the FACSCount AF system because slopes for the spectrophotometer and electronic cell counter also are above 1.0. For the species tested in experiment 1, the flow cytometric results for sperm concentration appeared to agree with results obtained with the hemocytometer. However, the number of samples included in experiment 1 is insufficient to draw a definitive conclusion for these species. Besides the data presented for rams, rats, rabbits, humans, and turkeys in this paper, we also analyzed semen samples from goats, stallions, dogs, and fowls using the same protocol.

Assessment of sperm concentration using flow cytometry has recently been published for humans (Eustache et al, 2001; Tsuji et al, 2002). Evenson et al (1993) and Parks (1992) suggested the use of flow cytometry for validation of sperm numbers in straws with bull semen, and Yamamoto et al (1998) used flow cytometry for counting sperm in epididymal fluid from rats. Both Yamamoto et al (1998) and Tsuji et al (2002) used flow cytometers that estimate the volume of sample analyzed. Parks (1992), Evenson et al (1993), and Eustache et al (2001) used assays based on beads that were added to samples prior to analyses. The number of beads in these studies was estimated from electronic cell counters or hemocytometric counting. Fluorescent microspheres (beads) have previously been used for quantification of leucocyte numbers (Stewart and Steinkamp, 1982). Strauss et al (1996) used counting tubes (BD Biosciences Immunocytometry Systems) that contain a specific number of beads per counting tube. This minimizes laborious preparation of reagent prior to analysis and was shown to result in high agreement between results obtained between different laboratories (Straus et al, 1996). The counting tubes used in the present experiments and by Hansen et al (2002) are easy to use and require only vortexing, opening, and addition of dyes prior to use. Results in experiment 2 show that this method is more accurate and more precise than other methods for assessment of sperm concentration in bull semen. Because live, dead, and dying sperm are stained, sperm concentration can be determined equally well in semen with debris such as gel particles (Hansen et al, 2002). The protocol described by Yamamoto et al (1998) used scattering of light to estimate the number of viable (PI unstained) sperm, whereas other protocols use permeabilization and staining with PI to obtain a determination of sperm concentration (Parks, 1992; Evenson et al, 1993; Eustache et al, 2001). Evenson et al (1993) also used staining with acridine orange for assessment of the sperm chromatin structure (SCSA) and determined sperm concentration simultaneously. Although the SCSA method provides valuable information regarding the sperm DNA, assessment of sperm membrane integrity (viability) is probably easier to implement for routine analyses. Results from experiment 2 show that sperm viability can be determined with high precision (CV = 0.7%). The correlation between sperm viability and fertility after insemination have recently been investigated for bulls (in preparation) and boars (in preparation). Both studies showed that a significant variation in fertility occurred for different ejaculates obtained from the same male. The ability to select the best ejaculates for insemination depends first of all on high precision in the semen analysis (Matson, 1997). Simultaneous determination of sperm concentration and viability results in a more precise assessment than a combination of methods that include more variation from handling the semen sample as well as variation from pipetting and the analysis itself.

In conclusion, the data presented here describe a new protocol for simultaneous determination of sperm concentration and viability. Both parameters are determined with high precision, and sperm concentration was determined with higher accuracy than with other methods available for raw bull semen as well as boar semen (Hansen et al, 2002). Data presented for rabbit, rat, human, and turkey sperm indicate that this method may be very useful for a variety of mammalian and avian species.


We thank BD Biosciences Immunocytometry Systems for making flow cytometers available for these experiments and for technical support. We are grateful to Dr Duane L. Garner for suggestions provided during the review of this manuscript.


  1. Supported by Danish artificial insemination societies for cattle and pigs, and by The Danish Directorate for Development (grants 93S-2465-Å97–0705 and 93S-2465-Å00–01120).