Introduction

Studies in fish indicated that environmental osmotic changes trigger sperm motility 1, 2. Interestingly, in most mammalian species examined, the sperm journey from the male to female reproductive tract also experiences a natural osmotic decrease 3, 4, implicating that an osmotic stress response upon ejaculation is evolutionarily conserved for normal sperm function. Indeed, increasing lines of evidence have shown that a physiological hypotonic stress could facilitate acrosome reaction in human sperm through calcium increase 5 and acrosome swelling 6, and is required for mouse sperm motility “start-up” once released from cauda epididymis 7, supporting the notion that a postcopulatory hypotonic stress is beneficial for normal sperm function. However, like a double-edged sword, the hypotonic stress could also cause potential harms to sperm function by inducing unwanted cell swelling 8, 9. To counteract this negative impact, mammalian sperm have acquired mechanisms to drive rapid transmembrane water movement for efficient cell volume regulation 10. Indeed, high water permeability has long been found in mammalian sperm 11, 12, 13, and ion channel-controlled water influx/efflux has been shown to be involved in postcopulatory sperm volume regulation 14, 15, 16. However, the sperm protein(s) (such as plasma membrane channels) responsible for this trait have not been identified. Aquaporins (AQPs), a family of proteins that are believed to be involved in water transport 17, have been expected as potential candidates for sperm volume regulation due to their highly specialized function for water permeability 4, 10, 18. Among the AQPs, AQP7 and 8 are expressed in rodent testis and sperm 19, 20, but genetic deletion of these AQPs in mice did not show obvious defects in the sperm function 21, 22, 23, possibly because of functional compensation by other AQP members. We previously identified that aquaporin3 (AQP3) is expressed in mouse testis 24, suggesting a potential role in male reproductive function. In this study, we demonstrated that AQP3 is a sperm water channel localized on the tail of mouse and human sperm. Aqp3-null sperm showed normal motility activation in response to hypotonicity, but with impaired volume regulation against hypotonic cell swelling, characterized by increased tail deformation due to excessive mechanical membrane stretch. Our data revealed that AQP3 functions as a key fluid regulator responsible for rapid sperm hypotonic osmoadaptation during the male-female reproductive tract transition, thus optimizing postcopulatory sperm behavior. Besides, the defect of the Aqp3-null sperm is clinically related, because similar sperm tail deformation is commonly found in clinical patients with male infertility/subfertility 25.

Results

AQP3 expression in sperm

AQP3 expression in mouse testis and sperm was detected at mRNA and protein levels (Supplementary information, Figure S1), which is in accordance with previous report that AQP3 is expressed in mouse testis 24. We further found that AQP3 is specifically localized to the sperm tail during late spermatogenesis and sperm storage in epididymis (Figure 1A-1C). High-magnification immunofluorescence and immunogold electron microscopic analyses of isolated sperm from cauda epididymis further demonstrated that the localization of AQP3 was confined to the principal piece of the sperm tail membrane (Figure 1D-1G). In addition, immunofluorescence staining of AQP3 in ejaculated human sperm also showed highly specific expression at the principal piece of the sperm tail (Figure 1H), consistent with the localization detected in mouse sperm.

Figure 1
figure 1

AQP3 is expressed in mouse sperm. (A-D) Immunofluorescence staining of AQP3 in mouse testis (A), cauda epididymis (B, C) and isolated sperm (D) from cauda epididymis. Note the intensive green signal at principal piece of sperm tail. (E, F) AQP3 antibody staining in the absence (E) or presence (F) of competing immunogen. Nucleus was counter stained by propidium iodide (PI). (G) Immunogold-labeled electron microscopic detection shows that gold particles are stained at plasma membrane of the principal piece (indicated by arrows). Scale bars: 0.2 μm. (H) Immunofluorescence detection of AQP3 in human sperm.

Full size image

Aqp3−/− male shows impaired fertility

The prominent expression of AQP3 in the sperm led us to revisit the reproductive phenotype of Aqp3-null male. In the present study, each male mouse with different Aqp3 genotypes was allowed to continuously mate with four wild-type females (each time with one female); after one successful mating (indicated by vaginal plug), the male was allowed to rest for 2 days before another female was placed in. The plugged females were individually caged and their pregnancy results were tracked. By this breeding assay, we found that although the Aqp3-null males could normally plug the females, the pregnancy rate and litter size were markedly lower compared with wild-type or heterozygous males. Nearly half of the wild-type females that mated with the Aqp3-null males showed no signs of pregnancy, and the remaining ones displayed reduced litter size (Table 1). These results clearly demonstrated that Aqp3 deficiency resulted in impaired male fertility, strongly suggesting a crucial role of AQP3 in male reproduction.

Table 1 Reproductive data of different Aqp3 males mate with wild-type females
Full size table

Aqp3−/− males have normal sperm production and their sperm show normal motility activation in response to physiological hypotonicity

Initial examination of major male reproductive organs, including testis and epididymis, revealed no apparent differences between wild-type and Aqp3−/− males in gross morphology and histology, as well as sperm count (Supplementary information, Figure S2A-S2C). As endocrine defects usually result in size and histological changes in reproductive organs, especially in seminal vesicles, these data suggest that endocrine defects are highly unlikely involved in the reduced fertility of Aqp3-null males, and the observed fertility decrease might be due to causes after sperm leave for female reproductive tract. Upon ejaculation, mouse sperm normally experience a physiological osmotic decrease from cauda epididymis (∼415 mOsm) to uterine cavity (∼310 mOsm) 26, which is beneficial for sperm motility activation. To determine the initial sperm motility activation between the wild-type and Aqp3-null sperm, we released cauda epididymal sperm directly into NaCl solution with different osmolarities and the sperm motility was assessed by Computer-Assisted Sperm Analyzer (CASA). As shown in Supplementary information, Figure S3A-S3E, both the genotypes of the sperm displayed lower-motility parameters in a relatively high osmolarity (440 mOsm) similar to the male reproductive tract, while in a relative hypotonic environment (300 mOsm) that mimics female uterine cavity, both showed robust motility activation.

Aqp3-deficient sperm undergo postcopulatory tail deformation due to increased vulnerability to hypotonic uterine environment

In contrast to the seemingly normal initial motility, the Aqp3-null sperm showed increased tail deformation in the uterus after copulation (2 and 8 h), characterized by a typical appearance of hairpin-like bending (Figure 2A and 2B). The observed bending of the sperm is not likely due to structural defects of flagellum as shown by electron microscopic analyses (Supplementary information, Figure S2D). In situ examination of the Aqp3-null sperm by immediate organ fixation also revealed no increased tail bending within cauda epididymis as compared with wild-type males (Figure 2A). These results indicate that the observed in utero sperm tail deformation is a postcopulatory defect that happens after the sperm enter the female reproductive tract.

Figure 2
figure 2

Aqp3−/− sperm showed increased in utero tail bending and exaggerated cell swelling upon physiological hypotonic stress. (A) Tail bending of wild-type and Aqp3−/− sperm in cauda epididymis and in postcopulatory uterus (NS: P > 0.05, **P < 0.0001, t-test). (B) Demonstrative pictures of wild-type and Aqp3−/− sperm within uterus at 2 h after copulation. Scale bars: 50 μm. (C) In vivo osmotic environment of wild-type and Aqp3−/− sperm is comparable (NS: P > 0.05, t-test). (D, E) Examination of the sperm tail bending after releasing cauda epididymal sperm into different osmotic environment. NaCl solution (n = 8 for each data presented, NS: P > 0.05, ***P < 0.0001, **P < 0.005. KO vs WT, t-test; D); HTF media (n = 5-8 for each data presented, NS: P > 0.05, ***P < 0.0001, *P < 0.05. KO vs WT, t-test; E). (F) Flow cytometry recorded forward scatter laser (FSC) distribution of the wild-type and Aqp3−/− sperm within different osmolarities (NS: P > 0.05, **P < 0.005, t-test). Numbers within the bars indicate number of mice used for each assay. All error bars represent s.e.m.

Full size image

The observed intrauterine tail bending in the Aqp3−/− sperm is markedly similar to previous reports that exposing normal sperm to drastic osmotic decreases (exposed to 150 mOsm PBS or distilled water) resulted in excessive cell swelling 7, 8. Indeed, upon ejaculation, mouse sperm normally experienced a physiological osmotic decrease during the male-female reproductive tract transition (from ∼415 to ∼310 mOsm) 26, which is tolerable for most wild-type sperm to maintain normal morphology. Given that Aqp3 deficiency results in the sperm tail bending after entering uterine environment, we hypothesized that the increased tail bending of the Aqp3−/− sperm could be due to abnormal drastic osmotic change upon ejaculation or intrinsic defects in response to physiological hypotonic stress.

By measuring the osmolarity of cauda epididymal fluid and uterine contents, we ruled out the first hypothesis, as the in vivo osmotic environments are comparable between wild-type and Aqp3-null mice (Figure 2C), which suggested that the abnormal tail bending of the Aqp3-null sperm is likely due to intrinsic defects, resulting in impaired resistance to hypotonic stress. To further define the impact of osmotic changes on the sperm tail bending, we performed an osmotic gradient challenge experiment on both the wild-type and Aqp3-null sperm. By directly releasing cauda epididymal sperm into different osmotic media (NaCl solution or HTF media), we found that at an osmolarity similar to cauda epididymis (440 mOsm), most sperm kept a straight form for both the genotypes. While at a lower osmolarity mimicking uterine environment (300 mOsm), the Aqp3-null sperm showed increased tail bending as compared with the wild-type sperm (Figure 2D and 2E). The wild-type sperm showed highly increased tail bending only at a much lower osmotic environment (150 mOsm) (Figure 2D and 2E).

Tail deformation in Aqp3−/− sperm is caused by mechanical membrane stretch, followed by exaggerated cell swelling

To confirm the hypothesis that the observed the sperm tail bending is caused by hypotonicity-induced cell swelling, the sperm volume changes were monitored using flow cytometric analyses, as reflected by the changing pattern in laser forward scatter (FSC), a sensitive method to simultaneously and objectively assess the sperm cell volume changes 27. By comparing the pattern of FSC, it was shown that while both the wild-type and Aqp3-null sperm showed similar pattern at an osmolarity mimicking cauda epididymis (440 mOsm), the Aqp3-null sperm showed markedly increases in larger-size cell population in an osmotic environment mimicking uterine cavity (300 mOsm) (Figure 2F), indicating a sign of exaggerated cell swelling. These results clearly demonstrated that the sperm without AQP3 failed to control cell volume expansion under a physiological hypo-osmotic stress, which was responsible for the observed tail deformation.

By using time-lapse imaging recording, we further demonstrated the transforming process of how the abnormal cell expansion caused the sperm tail bending. As shown in Figure 3A and 3B, and Supplementary information, Movies S1 and S2, compared with the mild swelling of the wild-type sperm, the Aqp3-null sperm showed progressive volume expansion beginning at the site of cytoplasmic droplet (CD), then the increasingly expanded CD stretched the membrane, forcing the sperm tail to angulate and finally form a hairpin-like structure. In some swelling sperm cells, the membrane failed to stretch to such an extent and finally ruptured and the tail “straight out” (Figure 3C and Supplementary information, Movie S3). Scanning and transmission electronic microscopic image analyses further revealed that the hairpin portion of the bending sperm tail was encapsulated within the stretched membrane, forming a state with increased intracellular volume (Figure 3D). Due to the hypotonicity-induced tail deformation, although sperm from both genotypes showed similar initial motility parameters after epididymal release (Supplementary information, Figure S3A-S3E), a large number of Aqp3-null sperm showed low/no motility with tail bending 2 h after copulation (recovered from uterus) as compared with the wild-type sperm (Supplementary information, Movie S4 and S5), suggesting that the impaired sperm motility is a secondary effect following the sperm tail deformation.

Figure 3
figure 3

Sperm tail bending is caused by mechanical membrane stretch beginning at cytoplasmic droplet. (A-C) Images of wild-type (A) and Aqp3−/− (B, C) sperm from cauda epididymis that were directly released into 300 mOsm NaCl solution. For the Aqp3−/− sperm, time-lapse imaging reveals gradual process of sperm tail bending forced by membrane expansion. Numbers in the picture represent time sequential. Note the swelling state of cytoplasmic droplet (indicated by arrows) in each genotype. Membrane rupture could be clearly observed in 13th photo of (C). Scale bars: 5 μm. (D) Scanning (SEM) and transmission (TEM) electronic microscopy reveal surface and inner appearance of sperm after releasing into 300 mOsm NaCl solution. Note: the bent Aqp3−/− sperm shows expanded intracellular space compared with the wild-type sperm. Scale bars (TEM): 0.5 μm.

Full size image

Aqp3−/− sperm showed compromised in vivo fertilization due to impaired migration into oviduct

To examine the ability of the Aqp3-null sperm to fertilize eggs, we performed in vitro and in vivo fertilization tests in wild-type and Aqp3−/− males. Interestingly, under a standard protocol of in vitro fertilization, sperm of both the genotypes showed similar ability in fertilizing wild-type eggs (Figure 4A). However, in vivo fertilization test revealed a significant decrease in fertilization rate in the Aqp3−/− male (Figure 4B and 4C). The discrepancy between in vitro and in vivo fertilization rates led to the conclusion that the Aqp3-null sperm showed no intrinsic problem in fertilization once they could reach the eggs in culture media, while the decreased in vivo fertilization rate might be due to decreased number of sperm reaching the eggs in the oviduct. Under physiological condition, before reaching the eggs, sperm must successfully migrate through the uterine-oviduct junction, where most of the sperm were blocked out and stayed in the uterine cavity. By checking the number of sperm that arrived in the egg/cumulus complex after time-defined copulation 28, we found a significant decrease in sperm/egg ratio in the Aqp3-null male (Figure 4D and 4E), supporting the idea that the impaired in vivo fertilization in the Aqp3−/− male is due to reduced sperm number that migrated to oviduct. Moreover, in vitro sperm transwell migration assay demonstrated that the Aqp3-null sperm showed decreased ability in passing through the mesh pores of membrane filter that mimicked the barrier of uterine-oviduct junction (Figure 4F-4H and Supplementary information, Figure S4), which further justified our hypothesis. The scenario of AQP3 on sperm osmoregulation and migration was illustrated in Figure 5.

Figure 4
figure 4

Aqp3−/− male shows reduced in vivo fertilization due to impaired sperm migration into oviduct. (A, B) In vitro (A) and in vivo (B) fertilization tests for wild-type and Aqp3−/− sperm (NS: P > 0.05, **P < 0.01, t-test). (C) Representative pictures of embryos flushed from wild-type female oviduct (day 2) after mating with wild-type and Aqp3−/− male. Each picture shows embryos collected from two females. (D) Aqp3−/− sperm shows reduced number in reaching egg-cumulus complex in vivo (**P < 0.01, t-test). (E) Demonstrative figures of egg-cumulus complex (upper) and arrived sperm as indicated by red arrow (lower). Scale bars: 50 μm. (F) Actual size of different pores (3, 5 and 8 μm diameter) in filter membranes (10 μm thick) used for in vitro sperm migration assay. (G) Illustration of in vitro sperm migration assay through filters with different pores and calculation of migration rate. (H) Migration rate for wild-type and Aqp3−/− sperm (released from cauda epididymis) through different filters (NS: P > 0.05, **P < 0.01, KO vs WT, t-test). Numbers within or above the bars indicate number of mice used for each assay. All error bars represent s.e.m.

Full size image
Figure 5
figure 5

Schematic figure summarizing the scenario of how AQP3 deficiency impairs male fertility. Upon ejaculation, sperm stored in male reproductive tract enter uterine cavity where they experience a natural osmotic decrease. The physiological hypotonic stress initiates sperm motility, but also poses potential harms to sperm function by inducing sperm cell swelling. Normally, only a small portion of sperm undergoes tail deformation because of the hypotonic stress. However, in Aqp3−/− mice, the Aqp3−/− sperm showed impaired resistance and increased vulnerability to hypotonic-induced cell swelling, resulting in a large portion of sperm showing tail bending after entering uterine environment. The increased tail deformation in the Aqp3−/− sperm led to decreased sperm passage through uterine-oviduct junction and decreased chance for meeting and fertilizing eggs, resulting in impaired male fertility.

Full size image

Discussion

In summary, our data provided direct evidence suggesting that AQP3 is a key fluid regulator essential for sperm regulatory volume decrease (RVD) upon hypotonicity-induced cell swelling. It was revealed that while sperm utilize postcopulatory hypotonicity for motility activation, AQP3 serves to protect sperm from swelling-induced mechanical membrane stretch, thus optimizing the “trade-off” between sperm motility and cell swelling upon physiological hypotonicity. To our knowledge, this is the first identified sperm protein responsible for efficient postcopulatory sperm osmoadaptation, thereby influencing male fertility. However, the Aqp3−/− male is not completely infertile, suggesting that the physiological events of sperm cell regulation should involve other members of AQP family or other membrane channels with compensatory functions. Indeed, RVD process involves active ion influx/efflux of various kinds 29, and many of these ion fluxes have been shown present in sperm. Therefore, further identifying specific ion channels in sperm will be undoubtedly important for a deeper understanding of sperm physiology, especially during the process of postcopulation.

The exact mechanisms by which AQP3 regulates sperm water outlet and the involvement of RVD are still not clear, one possibility is that the spatial arrangement of multiple AQPs (AQP3, 7 and 8) along the specialized architecture of the sperm tail would provide optimal sperm function, the previously found AQP7 and 8 in sperm might serve as major water influx, while AQP3 as the major efflux pathway. Also, it is more likely that AQP3 might serve as a part of the osmosensing system responsible for detecting early events in cell swelling (as a mechanosensor) and conveying signal to the subsequent RVD process, as firstly revealed by mathematical models 30. Indeed, recent years have witnessed growing evidence supporting a critical role of AQPs (AQP1, 2, 3 and 5) during RVD process in various cell systems 31, 32, 33, 34, suggesting that AQPs might have alternative regulatory functions in membrane fluid transport, not just inert pores simply increasing the osmotic permeabilities. As AQP3 is an aquaglyceroporin with permeability to both water and glycerol 17, there is also an open but intriguing possibility that AQP3 might have unique plastic molecular structure that responds to changing cell environment, or AQP3 could facilitate glycerol transport in sperm cell as a potential energy substrate, which is an interesting issue that warrants further investigation. More radically, given the recent contention that an ion pump or channel is evolutionally changeable or even coexisted in one molecule with hybrid behavior 35 and the most recent unexpected finding that the formate transporter shows an AQP-like channel structure 36, 37, it is encouraging to propose that an AQP might have the potential to function as an active water transporter (combined with other molecules) during processes such as sperm osmoregulation. Nonetheless, previous reports, coupled with present data on AQP3-mediated sperm cell volume regulation, would provide future directions for a new round of AQP research toward more fundamental roles as general regulators of living cells, possibly with unexpected novel mechanisms.

Another interesting issue in this study is about the functional observation of the sperm CD, for mammalian sperm, the physiological functions of the CD remain enigmatic 38, although its structure is clear and it has been successfully utilized for sperm cell patch claming in analyzing sperm membrane ion flow 39, 40. Our data clearly showed that when sperm face physiological hypotonic stress (from ∼420 to ∼300 mOsm), the CD site is the most sensitive and vulnerable portion that first swells and from which causes the subsequent sperm tail deformation. And even in wild-type mice, there are nearly 20% of sperm undergoing this kind of tail deformation during postcopulation. In this regard, what is the physiological function of the CD? Why sperm keep it as a weakness point during transit from the male to female reproductive tract? As optimal fertilization happens in a physiological hypotonic environment (300-340 mOsm) compared with the male reproductive tract 41 and the CD is a transient cellular structure with its function in fertilization still debatable 38, our data would support a novel hypothesis that the temporary presence of the CD in postcopulatory sperm might serve as a guardian in optimizing male fertility, by at one hand sensing external hypotonic changes to convey subsequent signals that stimulate sperm motility, while on the other hand filtering out (by inducing tail deformation) unqualified sperm that could not resist the natural hypotonic stress, thus guaranteeing that fertilization happens in optimal osmotic environment with optimal sperm. Notably, recent evidence has shown that sperm tail coiling was commonly found in ejaculates of infertile patients, suggesting defects in cell volume regulation 25. As AQP3 is also intensely localized in human sperm, the information obtained from present study in mice should provide valuable insights to further understand the causes of infertility/subfertility in relevant clinical patients.

Materials and Methods

Mice and breeding assay

AQP3 knockout mice were generated as described previously 24. All the CD1 female mice used in this study were purchased from Vital River Lab Animal Technology Co., Ltd (Beijing, China). Males of different genotypes (8-9 weeks) were used for breeding assay. Each male mouse was caged with a wild-type CD1 female (7-8 weeks) and vaginal plug was checked every morning. Once a vaginal plug was identified (day 1 postcoitus), the male was allowed to rest for 2 days, after which another female was placed in the cage for another round of mating. The plugged female was separated and single caged, and the pregnancy results were recorded. If a female generated no pups until day 22 postcoitus, it was deemed as not pregnant after sacrificed for confirmation. Each male underwent four cycles of the above breeding assay.

RNA extraction and RT-PCR

RNA extraction, reverse transcription and PCR were performed as previously described 42. The primers used in this study were provided in Supplementary information, Table S1.

Immunofluorescence on tissue sections and isolated sperm

Frozen sections (10 μm) were fixed in 4% paraformaldehyde (PFA) for 10 min at room temperature. Mouse sperms (from cauda epididymis) and human sperms (ejaculated sperm donated by volunteers with proven fertility) were washed in PBS, fixed in 4% PFA for 30 min at room temperature and air dried onto poly-L-lysine-treated slides. AQP3 polyclonal antibody (AB3276, Millipore) was used at 1:500 for mouse and 1:200 for human. For immunogen-competing assay, the competing peptide was mixed with antibody at 1:1 concentration before incubation.

Scanning, transmission and immunogold electronic microscopy

The scanning and transmission electron microscopy followed standard protocols performed at the Laboratory of Electron Microscopy, Peking University Health Science Center. For the immunogold electron microscopy, the sperm was fixed in 4% PFA, infused in 10% gelatin, dehydrated in sucrose and then frozen in liquid nitrogen. Cryosections (50 nm) were prepared (Leica ULTRACUT UCT/Leica EMFCS) and incubated at room temperature with 0.1% cold water fish skin gelatin (CWFS gelatin, AURION) and 5% BSA for 30 min, then incubated with AQP3 polyclonal antibody (AB3276, Millipore; 1:50) overnight. After extensive washing in PBS, cryosections were then incubated with colloidal gold-conjugated secondary antibody (15 nm gold particle, Goat-anti-Rabbit, AURION). Sections were then stained with uranyl acetate and methylcellulose, and observed by transmission electronic microscopy.

Sperm analysis in utero and in cauda epididymis

Male and female mice were caged together to induce copulation, and vaginal plug was checked every hour. After finding of the vaginal plug, the female was separated and sacrificed 2 or 8 h later; the uterine content was recovered and smeared onto a glass slide for sperm observation under microscope. To check the in situ status of sperm morphology before ejaculation, the cauda epididymis and vas deferens were undergone immediate immersive fixation in 5% glutaraldehyde 43. After 1 h fixation, sperms were punctured/squeezed from the cauda epididymis and vas deferens for morphological examination under light microscope.

Osmolarity measurement

Measurement of osmolarity in uterine content (2 and 8 h after copulation) and cauda epididymis fluid (coupled with vas deferens fluid) was performed on a Vapor pressure osmometer (VAPRO model 5520, Wescor Inc., Utah, USA) as described previously 26.

Sperm osmotic gradient challenge assay

Sperm analysis in different osmotic gradient was performed by directly releasing cauda epididymis sperm into NaCl solution or HTF media (Millipore) with different osmolarities (440, 300, 220 and 150 mOsm). After 15 min incubation at 37 °C, sperm morphology was analyzed under phase contrast microscope.

Sperm volume analyses

The sperm volume changes were monitored, as reflected in the changes of FSC pattern using a Flow Cytometer (BD FACSAria) as described previously 27. In brief, the cauda epididymis was punctuated by a needle and sperms were released directly into NaCl solution with different osmolarity at 37 °C. After 15 min incubation, a total of 10 000 cells was analyzed and recorded for each sample. Data was then exported by Flowjo5.7.2 software for image presentation and statistics.

Time-lapse microscopic analyses

Cauda epididymal sperms were released and exposed to NaCl solution (300 mOsm, 37 °C), within 3 min after releasing, 10 μl supernatant was pipetted onto a glass slide held by a Slide Warmer at 37 °C (#720230, Hamilton Thorne Research, Beverly, MA). The time-lapse image of sperm was captured and recorded by NIS-BR software under NIKON microscope (ECLIPSE 80i).

Sperm motility analyses

For mouse sperm motility analysis, a CASA system (Version.12 CEROS, Hamilton Thorne Research) was used with the following settings: for cell detection: minimal contrast, 50; minimal cell size, 4 pixels; and 60 frames were acquired at a frame rate of 60 Hz. At least 200 tracks were measured for each specimen at 37 °C with a Slide Warmer (#720230, Hamilton Thorne Research).

In vitro and in vivo fertilization analyses

In vitro fertilization was performed under standard protocols as previously described 44. For in vivo fertilization rate, wild-type CD1 females were mated with wild-type or Aqp3−/− males, the fertilization rate was examined by counting two-cell rate flushed from the oviduct at 15:00-16:00 of day 2 postcoitus (the morning finding of vaginal plug was designated as day 1 postcoitus).

In vivo examination of sperm entry into the egg-cumulus complex

In vivo examination of sperm entry into the egg-cumulus complex after time-defined copulation was performed according to protocols previously described 28. In brief, males were caged with wild-type CD1 females for a 3 h period (9-11 h after hCG), vaginal plugs were checked every hour, oviducts were isolated 2.5 h after copulation and the numbers of sperm within the egg/cumulus complex were examined.

In vitro sperm migration assay

In vitro sperm migration assay was modified based on previous report 45. Transwell plate with different pores in the polycarbonate membrane (3 μm, cat#3415; 5 μm, cat#3421 or 8 μm, cat#3422; Corning, Transwell®) was used for sperm migration assay. Cauda epididymal sperm were initially placed in the lower chamber, after 30 min incubation, upper and lower chamber sperm densities were counted by Flow Cytometer (BD FACSAria).

Statistical analysis

Statistical analyses were performed with SPSS 11.5. The significance of pregnancy rate between Aqp3−/− and wild-type or heterozygous males was tested by rank-sum test. Other statistical analyses between Aqp3−/− and wild-type were done using unpaired two-tailed t-test.