All mice used in this study were housed in the animal facility of the German Cancer Research Center (Germany). GM mice lines originally received from different sources were bred and expanded in-house whereas wild-type (WT) mice with corresponding genetic backgrounds (BDF, C3H, C57BL/6, CBA, DBA/2, FVB/N, NMRI) were received from Charles River (Germany). 149 sperm donors, 745 oocyte donors, and 112 foster mothers were included in this study.

Individually ventilated caging systems (IVC) and specified pathogen-free (SPF) facilities (barrier with open caging systems) were used as previously described in detail [5]. Ages of the male mice used ranged between 3 and 9 months. Males were housed alone and females were kept in groups of five. Health of the animals was monitored according to the Federation of European Laboratory Animal Science Associations (FELASA) recommendations [16]. Most mice were housed in an IVC facility that was maintained under SPF conditions. During this study no infectious agents listed in the FELASA guidelines were detected. Animal experimentation was performed according to the German Animal Welfare Act and the Cornell Center for Animal Resources and Education. All animal experiments were approved by the Animal Welfare Department of the Competent Authority (Regierungspräsidium Karlsruhe, Germany) and conducted under the surveillance of the intramural Animal Welfare Committee of the German Cancer Research Center.

Spermatozoa were cryopreserved according to method of Ostermeier et al. [17]. In brief, 3 to 9 months old males were sacrificed by cervical dislocation. Spermatozoa were then collected from the epididmides and vasa deferentia. Spermatozoa were allowed to disperse from the tissue for 10 min at 37℃ in cryoprotective media (CPM) [18% (w/v) raffinose (Sigma-Aldrich, USA), 3% (w/v) skim milk (BD Diagnostics, USA), and 477 µM monothioglycerol (Sigma-Aldrich, USA) in distilled water] and were loaded into 0.30 mL French straws (IMV Technologies, France). The straws were sealed with an impulse tong sealer (Polystar 110 GE/150 D; Rische + Herfurth, Germany), placed onto a polystyrene raft floating in liquid nitrogen (LN₂) for at least 10 min, and then stored in LN₂. All GM donor males were subjected to re-genotyping by the working-groups responsible for the individual lines.

Cryopreserved samples were stored in the liquid phase of LN₂. Storage of the straws was performed according to Schwab and Schenkel [22].

For staining, spermatozoa were thawed and pre-incubated for 1 h at 37℃. Spermatozoa were stained with two dyes: Hoechst 33342 stain for all spermatozoa (0.001 mg/mL; Invitrogen, USA) and propidium iodide (PI) (M3181.0010, 0.0001 mg/mL; Genaxxon, Germany) to identify cells with compromised membranes. Both dyes we used can stain spermatozoa of several species [3,13,30,33]. A Hoechst 33342 stock solution (10 mg/mL) was stored at -20℃ in the dark. A PI stock solution (1 mg/mL) was stored at 4℃ in the dark no longer than 6 months. Both dyes were dissolved in distilled water. In another, not successful approach, spermatozoa with intact membranes were stained with fluorescein diacetate (FDA) (0.001 mg/mL; Sigma-Aldrich, USA) and spermatozoa with compromised membranes were stained with ethidium bromide (EtBr) (0.001 mg/mL; Applichem, Germany).

A suspension of spermatozoa (20 µL) and 1 µL of each dye were carefully mixed in a 1.5-mL amber microcentrifuge tube. This solution was incubated at 37℃ for 5 min. Next, 10 µL of the suspension were spotted onto thoroughly cleaned and pre-warmed slides (R. Langenbrinck, Germany). Fluorescence microscopy was carried out with an inverted microscope with appropriate filters (wavelength of the emitted light; Hoechst 33342: 461 nm, PI: 617 nm) by using a fluorescence microscope (Axio Observer Z1; Carl Zeiss AG, Germany).

The slide containing the spermatozoa suspension was covered with a pre-warmed 18 × 18 mm² coverslip and placed onto an inverted microscope (TissueFAXSi, Tissue Gnostics, Austria). This technique covered reproducibly the whole volume, no solution or cells were lost, and air-bubbles never appeared under the coverslip. To evaluate motility, a video sequence was taken using Pixelink Capture OEM software and a Pixelink 622 camera (Pixelink, Canada). Dark field illumination was used to acquire 200 frames (1 msec each) at ×5 magnifications. For the membrane integrity assay, 25 fields of view were automatically acquired (with a ×40 objective lens) with TissueFAXSi software controlling the camera (PCO270xs, PCO, Germany), filters, and automatic stage movement (Märzhäuser, Germany). Typical exposure times were 110 ± 30 msec (DAPI filter) and 270 ± 20 msec (Texas Red filter). The entire process including fine adjustment of camera settings did not take longer than 5 to 7 min, and all measurements were taken for the same sample.

Fluorescence microscopy images were processed with TissueQuest software (TissueGnostics, Austria). First, all spermatozoa were identified by Hoechst 33342 stain. Cell density was expressed in fluorescence-activated cell sorting-like scattergrams where each dot represented one cell. On the y-axis Hoechst 33342 mean staining intensity of every cell was plotted vs. the identified cell size on the x axis (Hoechst 33342 stained area). In this scattergram spermatozoa were identified according to their size, doublets and small debris were excluded from further analysis. All positively identified spermatozoa were then displayed in a second scattergram as Hoechst 33342 mean intensity vs. PI mean intensity. Cut-off values for distinguishing PI-positive from PI-negative cells were set according to the negative controls. Images of cells with PI-staining intensities near the cut-off value (as shown in the scattergram) were evaluated visually to verify and fine-tune the position of the cut-off for each sample individually if necessary. Cells stained exclusively by Hoechst 33342 (those with intact plasma membranes) fell under the cut-off line while double-stained cells (ones without an intact plasma membrane) fell above the cut-off line. As images with known size were analyzed, the resulting quadrant statistics of the second scattergram showed density of spermatozoa (events/mm²), total number of spermatozoa and percentage of living/dead spermatozoa.

To calculate the concentration of spermatozoa in cells/µL we used the values from the scattergrams and calculated the volume as following: 10 µL of the homogenous spermatozoa solution were spotted onto a slide and covered with an 18 × 18 mm² coverslip. Taking the volume of the staining solution and the dilution factor into account, 1 mm² corresponded to 0.28 µL or 3.56 mm² represented 1 µL. This procedure was confirmed by counting spermatozoa in a Neubauer improved counting chamber (Carl Roth, Germany).

IVF was performed as previously described by Ostermeier et al. [17]. Donor females (3~4 weeks old; 5 donors/IVF) with a genetic background according to the individual sperm donor (see Tables 1 and 2) were superovulated as previously described [14,15,22]. These animals were sacrificed 12 to 14 h after the administration of human chorionic gonadotropin (50 IU/mL; Sigma-Aldrich, USA) and the oocytes were removed from the swollen ampullae. Human tubal fluid (HTF) was used as IVF culture medium (Millipore MR-070-D; EMD Millipore, USA). The medium-drop was overlaid with mineral oil (Sigma-Aldrich, USA) as described in [22]. The IVF-dishes (Becton Dickinson, USA) contained one 500-µL HTF drop. Sperm samples were thawed in at 37℃ (in a water bath) for about 10 min. The spermatozoa in CPM were pushed out of the French straw into the HTF drop and were incubated for 1 h at 37℃. Co-incubation of oocytes and spermatozoa in HTF was lasting 5~6 h, the resulting zygotes were washed and incubated overnight in a 200-µL drop of potassium simplex optimization medium (KSOM) (Millipore MR-020P-5F; EMD Millipore, USA). Next, the proportion of two-cell embryos was calculated. These embryos were subsequently used for embryo transfer, cryopreservation, or continued culturing in KSOM.

The success of an IVF was determined by counting two-cell embryos out of all oocytes used for this IVF. However, the two-cell embryos were afterwards used for embryo-transfer, in vitro culture or cryopreservation.

Correlations between different parameters were calculated using SAS software (ver. 9.2; SAS, USA) [6]. Differences between two groups of data were analyzed by Spearman's rank-sum test [12]. p-values < 0.05 were considered significant while p-values < 0.001 were highly significant. The influence of membrane integrity (i.e., % of PI-negative spermatozoa in a sample) on fertility (IVF success = % two-cell embryos among all oocytes used for IVF or expanded for the successful recovery of a mouse line) was measured with cross-classified tables using Fisher's exact test [7] with the prerequisite that at least 30 embryos sufficient for two embryo transfers were generated. The coefficient of variation (CV) was defined as the ratio of standard deviation to the mean. CV < 1 indicated low variance while CV > 1 corresponded to high variance [10].

To determine the number of viable spermatozoa in a sample, we first used an additive strategy involving staining with FDA (specific for viable cells) and EtBr (for dead cells). However, this approach did not produce satisfactory results since spermatozoa do not have enough cytoplasm for these types of staining and develop auto-fluorescence at the excitation wavelength suitable for FDA (450~490 nm, data not shown). Consequently, we chose a subtractive approach in which spermatozoa were treated with two dyes: one that stained all cells (Hoechst 33342, Fig. 1A) and another specific for cells with a compromised membrane (PI, Fig. 1B). This allowed us to determine the number of viable and dead spermatozoa.

Twenty-five fields of view were examined in our fluorescence microscopy analysis. Signals from both dyes in each field of view were analyzed. Merged images were used to identify spermatozoa with intact (blue) or compromised membranes (simultaneous blue and red staining, Fig. 1C). To exclude false-positive signals (e.g., doublets, small debris, or due to contamination) only signals of appropriate size were included into these analyses, the cut-off values for discriminating PI-negative from PI-positive spermatozoa were verified manually in each case (Fig. 2A). Scattergrams allowed us to calculate the amount of PI-negative spermatozoa [% of lower right in Fig. 2B]. As described above, the scattergram allowed us to calculate the concentration of spermatozoa, too. To measure sperm motility, 200 frames (1 msec) were acquired as a video sequence. All video sequences were taken at the same level (Fig. 3). For analyses, the total number of cells and the number of moving spermatozoa were counted, and the ratio of moving spermatozoa was calculated.

The parameters of spermatozoa determined microscopically were compared with the outcome of IVF by using the same sample. Concentration, motility, and membrane integrity data were correlated with two-cell-embryos received from IVF by Spearman's range-sum correlation. Fig. 4A shows the correlation between spermatozoa concentration and IVF outcomes (p = 0.587). Fig. 4B illustrates the correlation between spermatozoa motility and IVF outcomes (p = 0.5733), and Fig. 4C shows the correlation between PI-negative spermatozoa and the IVF results (p = 0.0001). Our results demonstrated that only membrane integrity was significantly correlated with the outcome of IVF.

The production of two-cell embryos was considered as successful IVF outcome. These embryos were afterwards used for additional purposes. Following over night culture of 108 randomly selected IVF samples, 89.91% of two-cell embryos developed to the eight-cell stage without significant two-cell arrests. Others were cryopreserved or transferred into pseudo-pregnant foster mothers to assess the capacity to recover a GM line in a qualitative approach. GM litters resulting from embryo transfer were identified by genotyping (data not shown).

As noted above, positive IVF outcomes were most dependent upon membrane integrity of the spermatozoa. Membrane integrity was further evaluated to determine whether this factor is a reliable marker for sperm viability. The quality of frozen/thawed samples might vary from mouse to mouse. To verify that one out of several samples from the same donor mouse can reliably represent all samples from this donor, 40 samples prepared from eight donor males were analyzed. As demonstrated in detail in Table 1, different samples of the same donor were prepared, spotted onto different cover-slides and were analyzed microscopically. Only a low variance (CV < 1) was detected in all cases. To omit the consequences of a possible in-homogeneity of the material within a slide, four different sets of fields of views (25 fields of view/set) on the same slide were examined. As shown in Table 2, data for eight mice (out of these two different GM lines) with different genetic backgrounds were proven to be homogeneous (CV < 1).