Effects of human follicular fluid and high-density lipoproteins on early spermatozoa hyperactivation and cholesterol efflux.

The preovulatory human follicular fluid contains only HDLs as a lipoprotein class with a typically high proportion of preβ HDL. We first examined the role of follicular fluid and HDL subfractions on human spermatozoa capacitation, a process characterized by a hyperactivation of the flagellar movement and a depletion of plasma membrane cholesterol. Whole follicular fluid and isolated HDL, used at constant free cholesterol concentration, were both able to promote an early flagellar hyperactivation. Moreover, incubation of [3H]cholesterol-labeled spermatozoa with follicular fluid induced a rapid cholesterol efflux from spermatozoa that was confirmed by mass measurements of cholesterol transfer. Using isolated HDL, the cholesterol efflux had a similar time course and represented 70% of that mediated by whole follicular fluid. We then analyzed the time course of radioactive labeling of HDL subfractions. In the first minute of incubation, we found that the preβ HDL fraction incorporated the main part of the radioactivity (60%), with the rest being found in α-HDL, but strikingly, the labeling of α-HDL increased with time at the expense of preβ HDL.Thus, our results indicate that HDLs are involved in both spermatozoa hyperactivation and cholesterol effl ux and suggest the role of preβ-HDL particles as fi rst cellular cholesterol acceptors.

velocity > 100 µ/s, linearity <65%, and amplitude of lateral head displacement >7.5 µm ( 26 ). Spermatozoa hyperactivation assays were performed at 37°C, using whole FF and isolated HDL at different concentrations varying from 5-70% (v/v), during incubation times from 1-300 min, depending on experiments. Finally, all the given rates of hyperactivation take into account those measured with B2 medium, free of FF or of isolated HDL.

Spermatozoa cholesterol effl ux assays
Spermatozoa concentration was adjusted to 5 × 10 7 cells/ml by dilution with modifi ed B2 medium and labeled during 1 h incubation at 37°C with [ 3 H] free cholesterol in ethanol (10 µCi/ml). At the end of the labeling period, sperm cells were centrifuged during 5 min at 400 × g . Spermatozoa were resuspended in modifi ed B2 medium and recentrifuged to eliminate unbound labeled cholesterol. After three washes with albumin-free B2 medium and centrifugation, spermatozoa bearing [ 3 H]cholesterol were resuspended in medium at 5 × 10 7 cells/ml and were used for the cholesterol effl ux studies. The sperm cells were able to incorporate an average of 37% of the [ 3 H]cholesterol during the labeling procedure. The specifi c radioactivity range was about 2-3 × 10 5 dpm per nmol of cholesterol. Each effl ux assay was performed on the basis of 1.2 × 10 7 cells/ml incubated with isolated FF HDL or whole human FFs used at the following concentrations: 1, 2.5, 5, 10, 15, 20, 25, and 50% (v/v). For each concentration and sample, we have measured the fractional cholesterol effl ux (FE) at different incubation times, 1, 5, 15, 30, and 60 min. At the end of the effl ux experiment, microcentrifuge tubes from Millipore (Ultrafree-MC centrifugal fi lter units, Durapore membrane, 0.45 µm) were used for fi ltrate collection of the effl ux medium to ensure the absence of contamination by labeled spermatozoa. After washing sperm cells with Tyrode's buffer, the radioactivity associated with cells was counted. Fractional cholesterol effl ux (FE) was calculated as the ratio between the label released to the medium and the total radioactivity recovered in media plus cells. Finally, all the given FEs take into account those measured with medium free of FF or isolated HDL at all the incubation time with cells.

Agarose electrophoresis and passive transfer
Electrophoresis was carried out on 0.75% (w/v) low-medium resolution agarose (1.5 mm thickness) at a constant voltage of 150 V in Barbital buffer 50 mM (pH 8.6) and at 4°C, using GelBond as a support. After effl ux experiments, the samples (12 µl) were added to a 1 cm sample channel and run at 150 V for 1 h 30 min in a S60 electrophoresis tank (Sebia, Issy-les-Moulineaux, France). After electrophoresis, the gel was divided in two parts. Part of the gel was transferred by pressure onto nitrocellulose membrane and was immunoblotted against apoA-I according to Barrans et al. ( 27 ), using as fi rst antibody a mixture of monoclonal antibodies raised against human apoA-I and then 125 I-labeled goat anti-mouse F(ab') 2 as second antibody. Membranes were exposed overnight to a PhosphorImager screen (Molecular Dynamics, France), and the cassettes were imaged by PhosphorImager. Quantifi cation of apoA-I was performed using ImageQuant 1.0, and data were expressed as pixel points determined by the computer and were linearly correlated with the dpm of the 125 I cpm associated with the antigen-antibody complex. From the other part of the gel, about 12 strips of agarose by lane (0.5 × 1 cm) were cut out and the [ 3 H] radioactivity was counted.

Extraction and analysis of neutral lipids by gas-liquid chromatography
After cholesterol effl ux assays, lipids of each sample of FF and spermatozoa were extracted according to Bligh and Dyer proto-HDL from the FF where they are relatively abundant ( 14,23 ). However, whether the HDL particles of the FF can take part in spermatozoa hyperactivation as well as in the cholesterol effl ux has not been yet investigated. Hence, in this study, we confi rmed that HDL particles of the FF are involved in both processes and deciphered the contribution of each HDL subfraction in the cholesterol effl ux. Finally, we revealed the distinctive role of pre ␤ HDL particles.

Reagents
FFs and plasma were obtained from consent and informed patients undergoing an in vitro fertilization program as previously described by Parinaud et al. ( 24 ). Human serum albumin, aprotinin, and leupeptin were obtained from Sigma (St. Louis, MO

Human spermatozoa collection
Human sperm ejaculates were obtained by masturbation after 2 to 4 days of sexual abstinence, from consent and informed normozoospermic men, and were allowed to liquefy for about 20 min at 37°C. Briefl y, sperm was applied onto a three-layer Percoll gradient (90, 80, and 60%, v/v) and centrifuged at 300 × g during 20 min. The 90% Percoll fraction was kept and washed with Tyrode's buffer prior to centrifugation at 400 × g during 5 min. Then, the pellet was resuspended in modifi ed Menezo B2 medium without albumin, cholesterol, and phenol red ( 25 ) and maintained at 37°C under 5% CO 2 , at a fi nal concentration of 1.5 × 10 7 spermatozoa/ml for motility studies and 5 × 10 7 spermatozoa/ml for the [ 3 H]cholesterol labeling procedure.

Sperm kinematics
Spermatozoa were adjusted to 1.5 × 10 7 spermatozoa/ml, and 20 µL of each sample was put into a in 2-well microcell of 20 µm maintained at 37°C (BICEF, L'Aigle, France). Motility parameters were assessed at 37°C using a Hamilton Thorn Motility Analyzer, version 10.8 (Hamilton Thorn Research, Beverly, MA). Machine parameters were a frame rate of 25/s, minimum contrast 8, minimum size 6, low/high size gates 0.5/1.7, low/high intensity gates 0.4/1.7, nonmotile head size 9, nonmotile intensity 200, and a path velocity >5 µm/s to be counted as motile. The variables measured included the percent motile, percent progressive motility, straight line velocity (VSL), mean path velocity (VAP), curvilinear velocity (VCL), mean linearity, mean amplitude of lateral head displacement (ALH), and percent of motile sperm with a path velocity >25µm/s defi ned as the rapid motility. Hyperactivated motility was defi ned as follows: curvilinear control) followed by a second peak that appeared later, after 300 min of incubation.
We then compared FF and HDL effects on the fi rst wave of spermatozoa hyperactivation ( Fig. 2 ). FF induced a rapid process that reached a peak at 5 min (16 ± 5%) followed by a slight decrease and a plateau up to 30 min (13 ± 3%). Of interest, when HDL isolated from FF were used at the same free cholesterol concentration than that of the 20%-FF dilution (i.e., 13.6 µmol/l of HDL-free cholesterol), level and kinetics of spermatozoa hyperactivation were in the same range (14 ± 2% at 5 min). We noticed that FF with HDL free cholesterol concentration lower than 3.4 mol/l was unable to promote the spermatozoa hyperactivation (data not shown).

Fractional cholesterol effl ux from spermatozoa to isolated HDL compared with whole FF
We tested the effects of FF, used at total cholesterol concentrations ranging from 0.0076-0.38 mmol/l (i.e., 1-50% v/v of FF in B2 medium), on the spermatozoa cholesterol effl ux during 30 min of incubation at 37°C. As shown in Fig. 3A , the fractional cholesterol effl ux increased along with FF cholesterol concentrations up to 0.2 mmol/l and tended to saturate above this concentration. These data were linearized according to the Lineweaver-Burk plot ( Fig. 3A , inset), and a mean value of 19% of cell 3 H cholesterol was estimated for the maximal effl ux, while 0.023 mmol/l of FF cholesterol was required to promote a 50% cholesterol effl ux. Of interest, the free cholesterol/phospholipid molar ratio of FF increased from 0.124 to 0.138 after 30 min incubation with spermatozoa, whereas the spermatozoa free cholesterol/phospholipid molar ratio decreased from 0.53-0.43 in 30 min, representing a cholesterol depletion of 19% (data not shown), which is in good agreement with the radioactive assay.
We showed above that FF-derived HDLs were able to promote sperm cell early hyperactivation at the same extent as whole FF ( Fig. 2 ). Therefore, we compared the capacity of these HDLs to promote cellular cholesterol effl ux, up to 30 min, to that of whole FF (both used at a similar free cholesterol HDL concentration of 13.6 µmol/l). As shown in Fig. 3B , the fractional cholesterol col after an acidifi cation of the aqueous phase by formic acid. Neutral lipids were then quantifi ed by a gas-liquid chromatographic method that we previously reported ( 28 ).

Analytical methods
Free and total cholesterol were also measured by cholesterol esterase/cholesterol oxydase techniques. Measurement of proteins was carried out according to Lowry procedure using BSA as a standard. Lipid extraction was performed as described above. Phospholipids were measured according to their phosphorus content ( 29 ) following lipid extraction. ApoA-I was measured by a rate nephelometry method using the IMMAGE® immunochemistry system (Beckman Coulter, Villepinte, France), which is widely spread ( 30 ).

Effects of FF or HDL on kinetics of spermatozoa hyperactivation
We fi rst evaluated the effects of varying FF concentrations (5,10,20,30,40, and 50%, v/v in modifi ed B2 medium) on spermatozoa hyperactivation process after 5 min of incubation at 37°C ( Fig. 1A ). The maximal rate of hyperactivation was obtained with 20% of FF. Then, spermatozoa (1.5 × 10 7 cells/ml) were incubated at 37°C with this FF dilution, and the hyperactivation process was analyzed as a function of the incubation time. This experiment revealed a biphasic kinetics ( Fig. 1B ) with a rapid increase of the hyperactivation in the fi rst 5 min (21% versus 6% for  rapid appearance of radioactivity in the medium, and about 79 ± 2% was recovered in both ␣ and pre ␤ electrophoretic zones. The remaining radioactivity was found at the gel origin and did not contain apoA-I. After 1 min of incubation, labeled cholesterol incorporated in the pre ␤ HDL represented 60% of the radioactivity associated with whole HDL compared with 40% in the ␣ -HDL ( Table 1 ). With time, the rate of [ 3 H]cholesterol increased in the ␣ -HDL at the expense of pre ␤ HDL until 30 min where ␣ -HDL retained 72 ± 3% of the radiolabeling versus 28 ± 3% for pre ␤ HDL ( P < 0.001). To get a more precise insight of the contribution of each HDL subfraction in cholesterol effl ux, we analyzed the free cholesterol/apoA-I ratio during the incubation of labeled spermatozoa with FF. As shown in Fig. 4 , pre ␤ HDLs were early saturated with [ 3 H] cholesterol and kept this high level of labeling during the whole incubation, whereas ␣ -HDLs were gradually labeled. To slow down the kinetics of cholesterol exchange, we decreased the number of radiolabeled spermatozoa from 1.5 × 10 7 to 0.5 × 10 7 . After 5 min of incubation with FF (as described in Table 1), the radioactivity associated with the pre ␤ HDL fraction was signifi cantly higher and reached 56 ± 6% (data not shown). Thus, a slower kinetics leads to an accumulation of labeled cholesterol in pre ␤ HDL. These results strongly suggest that FF pre ␤ HDLs behave as the fi rst acceptors of spermatozoa cholesterol with a further transfer to ␣ -HDL. However, we cannot completely exclude that the latter is able to directly remove a minor fraction of cholesterol from spermatozoa.

DISCUSSION
Free cholesterol depletion from plasma membrane appears to be one of the essential mechanisms for sperm cell capacitation ( 31,32 ). The fact that FF is in vitro the best capaciting agent ( 8 ) can be explained, at least in part, by the presence of cholesterol acceptors, such as HDL and albumin, that we and others found out ( 10,13,33 ). Moreover, we evidenced that HDL is the single class of lipoprotein during the preovulatory period ( 13 ) and that pre ␤ HDLs represent about 18% of total HDL of the FF ( 14 ). This subfraction plays a key role in the cellular cholesterol effl ux since it promotes the early phases of this process ( 16,18 ). effl ux to HDL increased from 3-15%, between 3 and 30 min, whereas it reached 22% with whole FF. Thus, one can assume that HDLs contribute to about 70% of the fractional cholesterol effl ux mediated by whole FF.

Distribution of the radiolabeled cholesterol from spermatozoa to the FF HDL subfractions
Spermatozoa were radiolabeled with [ 3 H]cholesterol and incubated with whole FFs during 1, 5, 15, and 30 min at 37°C. At the end of each effl ux period, an aliquot of the effl ux medium was submitted to agarose electrophoresis to assess the [ 3 H]cholesterol distribution in FF HDL subfractions. Following migration on a nondenaturating agarose gel electrophoresis to separate the different HDL subfractions, apoA-I was immunoblotted. We evidenced the presence of pre ␤ HDL and ␣ -HDL on the apoA-I immunoblots. After quantitation by PhosphorImager, the proportion of apoA-I in pre ␤ 1 HDL after 1 and 30 min of incubation of FF with spermatozoa was not signifi cantly different (27 ± 2% and 30 ± 1% of total HDL apoA-I, n = 4), respectively. When FF was incubated for 1 min with spermatozoa labeled with [ 3 H]cholesterol, there was a  Spermatozoa cholesterol effl ux experiments were carried out with whole FF (n = 4) used at 10% (v/v) during incubation time varying from 1-30 min. An aliquot of each effl ux medium was submitted to agarose gel electrophoresis. Strips of agarose were cut out, and associated radioactivity was counted. Results were expressed as the mean ± SE of the ratio between the labeled cholesterol incorporated in pre ␤ or ␣ -HDL and total HDL associated radioactivity.
Moreover, as we recovered upon electrophoresis only 80% of radiolabeled cholesterol in ␣ -HDL and pre ␤ HDL, we cannot exclude other effl ux mediators. Interestingly, we have previously observed apoA-IV in FF in signifi cantly higher amounts (corrected per mg of apoA-I) than in serum ( 34 ). Considering the role of apoA-IV or apoA-IV-and apoAI-containing lipoprotein particles in promoting cholesterol effl ux from various cell lines (35)(36)(37), apoA-IV may also act in modulating spermatozoa cholesterol effl ux to FF. Moreover, expression of apoJ has been recently reported in the rat reproductive tract ( 38 ). Since apoJ is able to promote cholesterol effl ux from mouse macrophage foam cells ( 39 ), it may be involved in spermatozoa capacitation, but this hypothesis has not been yet explored in humans. Finally, a part of the non-apoA-I radioactivity may correspond to prostasomes that remained bound to spermatozoa during the preparation and were released in the presence of FF (S. M. Hamdi and X. Collet, unpublished observations). Indeed, prostasomes are small vesicles of prostate origin that exhibit an unusual membrane composition with a very high cholesterol-to-phospholipid ratio.
Whole FF and isolated HDLs were able to promote spermatozoa hyperactivation with comparable kinetics and extent, which allowed us to conclude that this process was mediated by HDL. As cholesterol effl ux and hyperactivation were both early events stimulated by HDL, this may refl ect their close dependence. Moreover, it has been shown that the action of HDLs, as cholesterol acceptors, is coupled to an adenylyl cyclase/cAMP/protein kinase A signaling that supports spermatozoa capacitation ( 32 ). In this setting, an interesting hypothesis to explore is that the FF pre ␤ HDL subfraction is an early activator of this signaling pathway. Another issue remains to be solved is the identity of transporters that mediate cholesterol effl ux from spermatozoa membrane to FF pre ␤ HDL. A recent report showed that three members of the ABC transporter superfamily, ABCA1, ABCA7, and ABCG1, are involved in cholesterol transport from mouse spermatozoa to apoA-I ( 40 ). Further studies are needed to confi rm that these transporters are also implicated in the cholesterol effl ux from human spermatozoa. Finally, our study raises the question of the fertility of female patients with HDL deficiency or hypoalphalipoproteinemia. Unfortunately, this question has not been systematically addressed. In the three familial disorders in which patients exhibit very low plasmatic HDL levels (apoA-I defi ciency, LCAT defi ciency, and Tangier disease), the most often reported clinical features are premature and severe atherosclerosis and neuropathy but no fertility abnormalities ( 41 ). Since the reproductive function is very robust, one can hypothesize that a compensatory mechanism substitutes for HDL in cholesterol effl ux from spermatozoa. However, owing to the very low incidence of these disorders, as well as ethical issues, it will be diffi cult to decipher the interaction with spermatozoa of FFs from HDL-defi cient women.
Whatever the mechanism is by which HDLs promote spermatozoa hyperactivation, we have shown that they stimulate this process from the fi rst minute of incubation with sperm as well as an early cholesterol effl ux, mainly In this study, we have shown the early effects of whole FF and FF-derived HDL on the spermatozoa cholesterol effl ux as well as on their hyperactivation. Indeed, FF (used at 10%) was able to promote cholesterol effl ux from sperm cells, increasing with time up to 14% after 1 h. Using 20% FF, a 22% cholesterol effl ux was reached after 30 min of incubation. The K m value for cholesterol effl ux was estimated at 3.2% FF, which corresponds to an average value of 24 µM total cholesterol concentration, which is 3-fold less than the value estimated with hepatoma cells Fu5AH (64 µM) ( 34 ). Thus, spermatozoa represent one of the best models of cellular cholesterol effl ux. Of interest, this effl ux could be already measured after the fi rst minute of incubation with cells, which represented about 4% of celllabeled cholesterol. With respect to this early event, and taking into account the role of plasma pre ␤ HDL as initial cellular cholesterol acceptors, we tested the early incorporation of labeled cholesterol into the FF HDL subfractions, following effl ux experiments with whole FF. When FF (10%, v/v) was incubated for 1 min at 37°C with [ 3 H]cholesterol labeled spermatozoa, a major fraction of the radioactivity (60%) was recovered in the pre ␤ HDL fraction, compared with 40% in the ␣ -HDL ( Table 1 ). Pre ␤ HDLs thus appeared as fi rst acceptors for spermatozoa cholesterol. These pre ␤ HDLs were always highly enriched in label (relative to their apoA-I content) compared with the major part of ␣ -HDL ( Fig. 3 ). These results are in good agreement with those obtained by Castro and Fielding ( 16 ) using the fi broblast effl ux model. Since the rate of labeled cholesterol incorporated in the ␣ -HDL increased with time at the expense of that in pre ␤ HDL and since the proportion of apoA-I in pre ␤ HDL remains constant during incubation time, we can thus hypothesize that pre ␤ HDLs act as a shuttle of cholesterol molecules between cells and ␣ -HDL as previously reported in plasma. However, we cannot exclude that ␣ -HDLs are able to directly capture a minor part of cholesterol from spermatozoa. The effl ux capacity of isolated HDLs compared with that of whole FF, used at constant free cholesterol-HDL concentration, revealed that HDL contributed to 70% of the whole process. Part of these differences might arise from pre ␤ HDL loss during ultracentrifugal isolation of HDL. Fig. 4. Evolution of free cholesterol/apoA-I ratio of HDL subfractions during the incubation of labeled spermatozoa with FF (n = 4). The ratio was calculated using the data from Table 1 and according to FF apoA-I content, apoA-I repartition among HDL subfractions, and specifi c radioactivity of [ through the pre ␤ HDL subfraction. Of great interest, a recent report raised the possibility that HDLs are involved in human oocyte health ( 42 ). Thus, by ascribing a new role of HDL in the fertilization process, our study confi rms their signifi cance in human reproduction and opens new avenues for infertility as well as preventing conception research.
X.C. was a recipient of a "Contrat d'Interface" from CHU of Toulouse. The authors thank Dr. Pascale Benlian (Paris, France) for helpful discussions and Guy Cholez for HDL studies.