Stimulation of Human Spermatozoa with Progesterone Gradients to Simulate Approach to the Oocyte INDUCTION OF [Ca 2 (cid:1) ] i OSCILLATIONS AND CYCLICAL TRANSITIONS IN FLAGELLAR BEATING* □ S

Progesterone is present at micromolar concentrations in the cumulus matrix, which surrounds mammalian oocytes. Exposure of human spermatozoa to a concentration gradient of progesterone (0–3 (cid:1) M ) to simulate approach to the oocyte induced a slowly developing increase in [Ca 2 (cid:2) ] i upon which, in many cells, slow oscillations were superimposed. [Ca 2 (cid:2) ] i oscillations often started at very low progesterone ( < 10 n M ), and their frequency did not change during the subsequent rise in concentration. Oscillations also occurred, but in a much smaller proportion of cells, in response to stepped application of progesterone (3 (cid:1) M ). When progesterone was removed, [Ca 2 (cid:2) ] i oscillations often persisted or quickly resumed. Superfusion with low-Ca 2 (cid:2) bathing medium (no added Ca 2 (cid:2) ) did not prevent [Ca 2 (cid:2) ] i oscillations, but they could be abolished by addition of EGTA or La 3 (cid:2) . reticulum Ca oscillations. In most washout of progesterone ( a decrease in base-line (cid:1) ] near resting levels and cessation of oscillations. However, in many oscillations then resumed in the absence of progesterone, often preceded by a slow increase in base-line 2 (cid:1) some the of washout was virtually undetectable superfusion in progesterone-induced oscillations were already with (cid:1) sEBSS (sEBSS no added Ca (cid:1) and (cid:1) M progesterone) caused a reduction a slight superfusion

The best characterized agonist of [Ca 2ϩ ] i signaling in mammalian spermatozoa is the zona pellucida, which surrounds the oocyte. The zona pellucida induces elevations in [Ca 2ϩ ] i upon sperm contact, resulting in AR at the zona surface (14,15). The zona pellucida-activated [Ca 2ϩ ] i signal in mouse spermatozoa is initially generated by Ca 2ϩ influx through a voltage-operated Ca 2ϩ channel. There is good evidence for subsequent participation of a low capacity (or fractionally filled) Ca 2ϩ store in this process, but the function of store mobilization is to permit activation of store-operated Ca 2ϩ channels rather than to provide a significant source of Ca 2ϩ (16). Progesterone, which is synthesized by and present in mammalian cumulus (17)(18)(19), also causes elevations in [Ca 2ϩ ] i and is the best characterized agonist of human spermatozoa. As with the response to the zona pellucida, stimulation with progesterone induces a "simple" signal by Ca 2ϩ influx (20,21). This involves at least two influx pathways, but their identity is far from clear (22). Intriguingly, a small proportion of progesterone-stimulated human spermatozoa generate large slow [Ca 2ϩ ] i oscillations, possibly reflecting store mobilization (23).
That responsiveness of human spermatozoa to progesterone is correlated with fertilization success in vitro (24) indicates its biological significance, but the role of progesterone in fertilization is poorly understood. Micromolar doses of progesterone induce AR, but the relevance of this response for fertilization in vivo is disputed (22,25). Progesterone is also reported to induce hyperactivation, a vigorous swimming pattern caused by marked changes in flagellar beating. Hyperactivated motility is adopted by sperm as they gain the ability to fertilize and is characteristic of cells retrieved at the site and time of fertilization (4).
Progesterone is present in the cumulus at micromolar concentrations (17), but human spermatozoa respond to the hormone at nanomolar doses (26,27). Spermatozoa will therefore detect cumulus-derived progesterone, distributed by diffusion and/or ciliary currents, prior to encountering the oocyte zona cumulus. Furthermore, the progesterone stimulus encountered by the spermatozoa is likely to occur as a concentration gradient, which the cell ascends as it approaches the oocyte. Previous studies on the action of progesterone on spermatozoa have used stepped application. To understand better the response of human spermatozoa to the progesterone stimulus that they encounter in vivo, we have exposed human spermatozoa to a rising logarithmic progesterone concentration gradient. Cells stimulated in this way respond with novel and complex changes in [Ca 2ϩ ] i and flagellar activity, generated by an unusual mechanism.
Preparation and Capacitation of Spermatozoa-Donors were recruited at Birmingham Women's Hospital. Highly motile spermatozoa were harvested into sEBSS as described previously (27) and were left to capacitate (to acquire the ability to fertilize, a process that normally occurs during residence in the female tract) for 6 h at 37°C in 5% CO 2 .
Imaging-Aliquots were loaded with Oregon Green BAPTA-1/AM and imaged in a continuously perfused chamber as described previously (27). Stepped (3 M) progesterone stimuli and drugs were applied by addition to the perfusion header. 1 mM La 3ϩ was applied in HEPESbuffered saline containing 150 mM NaCl, 5 mM KCl, 2 mM CaCl 2 ), 5 mM MgCl 2 , 10 mM glucose, and 0.3% bovine serum albumin (pH 7.4). All experiments were carried out at 25°C unless stated otherwise. Data acquisition and storage were controlled by a PC running AQM Orca 2001 (Kinetic Imaging Ltd., Nottingham, United Kingdom).
Generation and Assessment of Gradient Stimuli-Progesterone gradients were generated using a five-chamber gradient maker (0, 3 nM, 30 nM, 300 nM, and 3 M progesterone) connected to the inflow of the imaging chamber. Gradient characteristics were assessed in two ways. To determine the time of initiation of the gradient and to assess smoothness/stability, an FITC gradient was generated using 0, 2, 10, 20, and 30 nM FITC. Fluorescence intensity in the imaging chamber was measured at 10-s intervals. The outflow from the imaging chamber was collected at various time points during a number of the gradient experiments, and progesterone was measured using the ELISA kit following the manufacturer's instructions.
Single Cell Data Processing-Data were processed off-line using AQM Orca 2001 as described previously (27). Raw intensity values from the caudal part of the head of each sperm were imported into Microsoft Excel and normalized to pre-stimulus values. At each time point, the normalized fluorescence intensity values (R) for each cell were compiled to generate an overall average normalized head fluorescence (R tot ).
The latency of response to gradient stimuli was estimated by comparing mean control fluorescence (at least 20 images immediately prior to stimulation) with each mean from a 10-point moving average (increment ϭ 1). The response to the gradient stimulus was considered to have been initiated from the midpoint of the first 10-point sample at which there was a significant (and subsequently maintained) increase in fluorescence (p Ͻ 0.05) using unpaired Student's t test.
Assessment of the Characteristics of [Ca 2ϩ ] i Oscillations-Initial attempts to use Fourier analysis as an objective method to identify oscillating cells and to determine oscillation frequency were unsuccessful due to variability, particularly in the cycle period and in the base line upon which oscillations were superimposed. Traces that clearly incorporated repeated transitions between fluorescence levels (such as in Fig. 2a) generated noisy spectra in which the dominant frequency was often barely detectable, and oscillating cells could not be identified with any degree of reliability. Cells displaying oscillations in [Ca 2ϩ ] i were therefore identified directly from time/fluorescence intensity plots. Only cells with cyclical changes in fluorescence, comprising repeated events of consistent characteristics, were categorized as oscillators. The period of [Ca 2ϩ ] i oscillation in each cell was calculated by taking an average over 3-10 cycles.
Flagellar Activity-Immediately following [Ca 2ϩ ] i imaging, the field of cells was observed under phase-contrast microscopy (2 Hz for 12 min). We then selected cells that were well adhered to the slide but had a freely motile and clearly visible flagellum in the phase contrast images. A scale with 1.5-m graduations was superimposed on the image, normal to the axis of the sperm, at a point on the flagellum (typically not Ͼ10 m beyond the mid-piece) where the position could be detected in every frame. The point where the tail crossed this scale was then recorded for each image in the phase contrast series. The square of each frame/frame increment in tail position was calculated ((change in position in micrometers) 2 ) and plotted as a 20-point moving average.
Labeling with BODIPY FL-X Ryanodine-Capacitated spermatozoa (6 million cells/ml) were labeled with 2.5 M BODIPY FL-X ryanodine for 30 min at 37°C in 5% CO 2 and then transferred to the imaging chamber. An additional 30-min incubation at 37°C in 5% CO 2 was then carried out for the cells to adhere. The chamber was connected to the perfusion apparatus, and 2 ml of medium was perfused through the chamber to remove excess dye.
Assessment of Acrosomal Status-After collection of a series of Oregon Green BAPTA-1/AM images for assessment of [Ca 2ϩ ] i signaling, the chamber was perfused with 100% methanol for 30 s and then washed with sEBSS. The field was bleached by exposure to fluorescent illumination for 30 s. (A control was carried out to confirm that this process did not significantly increase the number of acrosome-reacted cells.) FITC-conjugated P. sativum agglutinin (0.2 mg/ml) was introduced into the chamber and left for 45 min, and then distilled water was perfused through the chamber for 10 min. This procedure was carried out at 25°C. An image of the field of view was captured, and the each cell was assessed for acrosomal status and related to the [Ca 2ϩ ] i signaling activity from the preceding fluorescence image series. Approximately 15% of the cells were not scored for acrosomal status either because they were washed off during the staining stage or because we were not able to assess acrosomal status from the labeling pattern.
Statistical Analysis-Proportions of oscillating cells, oscillation cycle frequency, and rates of AR were compared in Microsoft Excel using paired or unpaired t tests (two-tailed) as appropriate. Percentage data were arcsine-transformed before analysis. Statistical significance was set at p Ͻ 0.05. All data are presented as means Ϯ S.E.

Response to Stepped Progesterone Treatment
We previously described the biphasic [Ca 2ϩ ] i response of progesterone-stimulated human sperm (27,28). Both population and single cell responses comprise a transient [Ca 2ϩ ] i increase followed by a sustained elevation in [Ca 2ϩ ] i (Fig. 1a). The response is dependent upon [Ca 2ϩ ] o , occurs in ϳ98% of viable cells, and is dose-dependent (27).

Response to a Progesterone Gradient
To simulate the progesterone stimulus during approach to the oocyte, we used a logarithmic progesterone gradient (0 -3 M) that started 4 -5 min after commencing recording and developed over 20 min (see "Experimental Procedures") ( Fig.  1b). In contrast to the progesterone step-induced biphasic response, population responses (R tot ) showed a smooth ramped rise in [Ca 2ϩ ] i that was detectable within Ͻ1 min of progesterone introduction and that usually peaked before the end of the progesterone gradient (latency of peak ϭ 17 Ϯ 1 min, n ϭ 7).
[Ca 2ϩ ] i then stabilized or decreased slowly. Examination of single cell responses showed a similar pattern: cells generating a gradual increase in [Ca 2ϩ ] i that often peaked before completion of the progesterone gradient and fell toward the end of the recording, possibly reflecting desensitization of the response (Fig. 1c). Phasic changes in [Ca 2ϩ ] i , superimposed on the raised level of [Ca 2ϩ ] i , were often observed (see below), but initial [Ca 2ϩ ] i transients of the type characteristic of the response to stepped application of progesterone (Fig. 1a) never occurred. Transient responses of human spermatozoa to a stepped progesterone stimulus are well synchronized, with 98% of the cells responding within 20 s of progesterone application (27). In contrast, only 28 Ϯ 3% of the cells generated a significant increase in [Ca 2ϩ ] i (fluorescence exceeding control levels; p Ͻ 0.05) within the first 20 s after initiation of a progesterone gradient, with the response rate rising to 66 Ϯ 1 and 96 Ϯ 0.2% after 60 s and 5 min, respectively (seven experiments, 885 cells).

Gradient Stimulation Induces Oscillations in [Ca 2ϩ ] i
In addition to causing a slow increase in sperm [Ca 2ϩ ] i , progesterone gradients induced slow [Ca 2ϩ ] i oscillations in the caudal part of the head of more than one-third of the cells (34 Ϯ 2%; oscillation period ϭ 4.1 Ϯ 0.3 min; seven experiments, 885 cells) (Fig. 1, d and e). The onset of oscillations varied between cells, but typically occurred within 3-10 min of gradient initiation, with 36 Ϯ 5% of the cells generating the upstroke of the first oscillation within 4 min of gradient initiation (Fig. 1d), at a progesterone concentration of Ͻ10 nM according to gradient calibration. The subsequent increase in progesterone concentration was never reflected in a discernible increase in the frequency of oscillations, despite an increase in agonist concentration of Ն100-fold after oscillations commenced (Fig. 1d). Once established, oscillations continued throughout recording, although [Ca 2ϩ ] i between cycles often decayed, similar to the response in non-oscillating cells (Fig. 1d, pink trace). The amplitude of oscillations was typically a 40 -100% increase in fluorescence, at least as great as the [Ca 2ϩ ] i transient evoked by 3 M progesterone (Fig. 1a), which had an amplitude of 700 nM in fluorometric measurements (27). The kinetics of [Ca 2ϩ ] i oscillations varied between cells, possibly reflecting cell/cell differences in Ca 2ϩ mobilization and/or clearance processes.

Progesterone Step-induced Oscillations
Exposure of capacitated human spermatozoa to a 3 M progesterone step induced slow oscillations in [Ca 2ϩ ] i in Ϸ10% of the cells after the initial transient response (23). In stepped progesterone experiments carried out in parallel with our progesterone gradient experiments (Fig. 2a), 14 Ϯ 2% of the cells displayed oscillations (nine experiments, 1346 cells), less than half as many as with gradient stimuli (p Ͻ 0.01) (Fig. 2b). In five instances, direct comparisons were made between paired aliquots from the same sample, prepared identically before exposure to either a stepped or gradient stimulus. Gradient stimuli were consistently more effective in activating oscillatory [Ca 2ϩ ] i signaling (p Ͻ 0.01) (Fig. 2c). The period of oscillation in step-stimulated cells was typically 2-4 min (mean ϭ 3.4 Ϯ 0.2 min; six experiments, 525 cells), slightly shorter than the gradient-induced oscillation period of 4.1 Ϯ 0.3 min (p Ͻ 0.05). In almost 30% of the oscillating cells activated by a progesterone step, [Ca 2ϩ ] i returned almost to control levels between cycles (Fig. 2a), a pattern seldom observed in gradientstimulated cells.
To confirm that the recorded oscillations were not affected by aliasing, we investigated the effect of image acquisition frequency. The recorded rate, amplitude, and kinetics of the oscillations were independent of the rate at which images were collected (between 0.1 and 0.67 Hz). Fig. 2a shows two cells from an experiment in which the sampling rate was changed from 0.1 to 0.67 Hz after 15 min of recording.

Temperature Sensitivity of [Ca 2ϩ ] i Oscillations
In experiments at 31 and 37°C (four experiments at each temperature; stepped progesterone stimuli), the proportions of cells displaying oscillations (5-15%) and the amplitudes of oscillations (typically a 40 -50% increase in fluorescence) were similar to those observed at 25°C. However, the kinetics (rise and fall) and oscillation rate were markedly temperature-sensitive. The oscillation period at 37°C (0.39 Ϯ 0.03 min; range of 0.2-0.6 min; four experiments, 298 cells) was 10 times shorter than that at 25°C (3.4 Ϯ 0.2 min). At 25°C, the initial progesterone-induced [Ca 2ϩ ] i transient often had an amplitude and kinetics similar to those of subsequent oscillations (Fig. 2a), but at 31 and 37°C, the oscillations were much shorter than the transient (Fig. 2d), indicating that oscillations are not generated by periodic repetitions of the initial transient. This conclusion was supported by the observation that the first [Ca 2ϩ ] i oscillation occasionally activated before decay of the initial [Ca 2ϩ ] i transient and was superimposed upon it (data not shown). Because of the difficulty of applying drugs via the gradient maker and of maintaining stable recording conditions at temperatures above 25°C, all subsequent studies in which saline manipulations and drug application were used for characterization of the processes underlying progesterone-induced oscillations were done using stepped progesterone stimuli at 25°C.

Is the Induction of Oscillations Reversible?
After fertilization, the sperm no longer exists as a separate cell. Transformations of signaling and cell activity induced by stimuli during approach to the oocyte (such as progesterone) could therefore be irreversible. To investigate this possibility, we observed the effect of removing the progesterone stimulus on established [Ca 2ϩ ] i oscillations. Three patterns of response were seen. In 10% of the cells generating [Ca 2ϩ ] i oscillations, the effect of progesterone washout was essentially undetectable (Fig. 3a, Ⅺ). In 40% of the cells, [Ca 2ϩ ] i fell toward pre-stimulus levels but recovered discernibly within 1-2 min, and oscillations restarted (latency ϭ 1.5-6 min after progesterone washout; seven experiments, 792 cells) (Fig. 3a, f). In the remaining 50% of the cells, oscillations ceased and did not restart, but these cells resumed oscillating immediately upon restoration of the progesterone stimulus (data not shown). In most oscillating cells, the only effect was that [Ca 2ϩ ] i between cycles fell to levels at or below that seen before the progesterone stimulus, such that oscillations were clearly enlarged (Fig.  3b). The rate of the rise in [Ca 2ϩ ] i during the upstroke was at least as fast as in normal sEBSS. Upon return to standard sEBSS, oscillations reverted to their previous characteristics. Superfusion of cells with low-Ca 2ϩ sEBSS containing 2 mM EGTA abolished [Ca 2ϩ ] i oscillations (in some cells, a truncated [Ca 2ϩ ] i transient was generated before complete arrest) (Fig.  3c), and they did not resume following removal of EGTA (two experiments, 191 cells). Subsequent readmission of standard sEBSS caused a large [Ca 2ϩ ] i transient (Fig. 3c), but oscillations rarely resumed.

Ca 2ϩ Influx and Generation of Oscillations
Effect of La 3ϩ -250 M La 3ϩ strongly attenuates the Ca 2ϩ influx induced by progesterone in human spermatozoa (20). To prevent precipitation, 1 mM La 3ϩ was applied in bicarbonatefree HEPES-buffered medium. Bicarbonate-free medium caused a fall in [Ca 2ϩ ] i and cessation of oscillation in ϳ30% of the oscillating cells, despite the continued presence of progesterone (data not shown). Application of La 3ϩ then reduced [Ca 2ϩ ] i to levels at or below the pre-stimulus concentration and caused arrest of oscillations in all cells (six experiments, 578 cells) (Fig. 3d). In ϳ30% of the cells, there was a slow rise in [Ca 2ϩ ] i on which slow [Ca 2ϩ ] i ripples or occasional [Ca 2ϩ ] i transients were sometimes superimposed (Fig. 3d).
The effects of progesterone washout, extracellular EGTA, and blockade of Ca 2ϩ influx by La 3ϩ suggest that sustained progesterone-induced Ca 2ϩ influx contributes significantly to the generation and organization of the oscillations. 10 M SKF-96365 (which inhibits store-operated channels in sea urchin sperm (30)) failed to inhibit oscillations in cells previously stimulated with progesterone (two experiments, 187 cells) (data not shown). An alternative phasic Ca 2ϩ entry pathway is through arachidonate-regulated calcium channels, which are activated by arachidonic acid generated by receptor-activated phospholipase (31). The phospholipase inhibitor aristolochic acid (250 M) had no effect on the oscillations in progesteronestimulated cells (three experiments, 376 cells) (data not shown).

Ca 2ϩ Stores and Generation of Oscillations
Effects of Inhibition of Ca 2ϩ Store ATPases-In somatic cells, it is usually possible to arrest store-mediated [Ca 2ϩ ] i oscillations by inhibition of sarcoplasmic/endoplasmic reticulum Ca 2ϩ -ATPases using the inhibitors thapsigargin and cyclopiazonic acid. Thapsigargin at doses between 100 nM (a saturating but specific inhibitory dose in somatic cells (32)) and 1 M had no effect on established oscillations in spermatozoa (four experiments, 410 cells) (see Supplemental Fig. 1a), and it also failed to prevent induction of oscillations by progesterone (two experiments, 338 cells) (data not shown). In contrast, thapsigargin from the same stock (100 nM to 1 M) caused store emptying and immediate arrest of [Ca 2ϩ ] i oscillations in primary rat osteoblasts. 2 10 M cyclopiazonic acid had no effect on oscillations (see Supplemental Fig. 1b).
Involvement of Phospholipase C/Inositol Trisphosphate Receptors-Treatment of human spermatozoa with progesterone is reported to induce synthesis of inositol trisphosphate (33) iments, 387 cells) (Fig. 4, a-c). Caffeine (5 mM) caused 75 Ϯ 6% of the oscillating cells to arrest at near base-line [Ca 2ϩ ] i , but after a delay of 3-6 min, [Ca 2ϩ ] i rose in Ͼ70% of the cells, and many resumed or commenced well defined oscillations (three experiments, 298 cells) (Fig. 4d). Oscillations slowed and became less regular following removal of caffeine (Fig. 4d).
Because the pharmacological experiments implicated RyRs in the generation of the progesterone-initiated [Ca 2ϩ ] i oscillations, we used BODIPY FL-X ryanodine (a membrane-permeant fluorescein-tagged derivative) (36,37) to identify RyRs (three experiments, 721 cells). 45% of the cells became stained with BODIPY FL-X ryanodine, with the majority being labeled at the caudal part of the head and the head/mid-piece junction (38% of the cells) (Fig. 4e, colored arrows). There was also some labeling in the acrosomal area (15% of the cells including those labeled in both areas) (Fig. 4e, white arrow). Comparison with phase contrast images of the same cells showed that very heavy labeling of the mid-piece (Ͻ5% of the cells) was often associated with large cytoplasmic droplets (Fig. 4e, yellow arrows). The presence of RyRs was clearly not a result of excessive cytoplasmic retention since most of the labeled cells appeared anatomically normal (Fig. 4e, blue arrows), and some cells possessing cytoplasmic droplets showed labeling behind the nucleus, but not in the droplet itself (Fig. 4e, green arrow).
Tetracaine (25 M to 2 mM), a potent inhibitor of Ca 2ϩ release through RyRs, inhibited oscillations in a dose-dependent manner (six experiments, 633 cells) (Fig. 5a) and usually caused [Ca 2ϩ ] i to fall below resting levels (Fig. 5b). In Ͻ10% of the oscillating cells, small oscillations persisted in the presence of tetracaine (Fig. 5b, Ⅺ). Upon removal of tetracaine, oscillations restarted in 26% of the cells in which they had previously arrested (two experiments, 142 cells) (Fig. 5c).
Effect of TMB-8 -To further investigate the participation of stored Ca 2ϩ in the ability of human spermatozoa to generate [Ca 2ϩ ] i oscillations, we investigated the effect of TMB-8, an inhibitor of the release of stored Ca 2ϩ that blocks caffeineinduced Ca 2ϩ mobilization (38). TMB-8 (200 and 300 M) arrested oscillations in all cells, with [Ca 2ϩ ] i stabilizing at the level occurring between oscillation cycles (six experiments, 551 cells) (Fig. 5d).
Effect of 2,4-Dinitrophenol-Mitochondria accumulate Ca 2ϩ in the matrix compartment and can contribute to cellular Ca 2ϩ buffering. To confirm that mitochondria were not responsible for the generation of [Ca 2ϩ ] i oscillations, we applied the uncou- 100 experiments). During analysis of images from oscillating cells, we noticed similar effects, with head movement increasing markedly during oscillation peaks (Fig. 6a, Supplemental  Fig. 5 and Supplemental video). This effect occurred at 25, 31, and 37°C. Cyclical patterns of movement occurred only in cells showing [Ca 2ϩ ] i oscillations (always occurring in tight association with increased [Ca 2ϩ ] i peaks), and temporary arrest of [Ca 2ϩ ] i oscillation (upon progesterone washout) followed by resumption (e.g. Fig. 3a, f) caused associated arrest and resumption of movement. No such pattern was ever seen in cells that did not generate [Ca 2ϩ ] i oscillations. These movements were often asymmetrical ( Fig. 6a and Supplemental Material video). Increasing the rate of image acquisition from 0.1 to 0.67 Hz during recording confirmed that these movements and associated [Ca 2ϩ ] i oscillations were not sampling/aliasing artifacts. We therefore carried out experiments in which we collected fluorescence images and then observed the same cells under phase-contrast microscopy to assess flagellar movement (lateral excursion of the proximal flagellum) over a time scale appropriate for detecting oscillation-regulated changes. Using frame-by-frame analysis (see "Experimental Procedures") ( Fig.  6b), we detected clear long-term patterns of flagellar activity. In cells showing [Ca 2ϩ ] i oscillations, induced by stepped or gradient progesterone stimuli, we always observed cycles of flagellar activity with kinetics corresponding closely to those of the [Ca 2ϩ ] i oscillations in that cell (correlation coefficient for period of oscillation/motility cycles ϭ 0.94; seven experiments) (Fig. 6, c and d). Cells with a smooth [Ca 2ϩ ] i signal following stimulation with progesterone did not show cyclical flagellar activity (Fig. 6e), and cells with an irregular sustained phase of the progesterone response showed irregular variation in flagellar activity. We examined in detail those cells in which the flagellum was in focus for most or all of its length. During periods assessed as high flagellar activity, more marked bending occurred in the proximal flagellum, and lateral excursion was greatly increased in the distal portion. Sampling at several points during a series of phase contrast images showed that quantification of flagellar activity faithfully reflected repeated transition between patterns of flagellar beating (Fig. 6f).
To reinforce the conclusion that cycling of flagellar activity was not an artifact caused by undersampling, we attempted to visualize the arc of flagellar bending during collection of long exposure fluorescence images (800 -1000 ms, sufficient to capture several complete beat cycles). In a small number of cells, the magnitude of excursion of the proximal flagellum was clearly visible as a "V" marking the extremes of the arc described during the beat cycle. Fig. 6g shows a plot of fluorescence intensity and flagellar arc obtained from the same series of long exposures, showing that flagellar activity oscillates in synchrony with [Ca 2ϩ ] i oscillations.

Do Oscillations Induce the Acrosome Reaction?
Repetitive [Ca 2ϩ ] i spiking might increase the efficacy of progesterone in stimulating AR. To investigate this possibility, we assessed cells for acrosomal status after first imaging their [Ca 2ϩ ] i response to stimulation with progesterone. In five experiments, cells were stimulated with a 3 M stepped progesterone stimulus, imaged for 30 min at 25°C, and then assessed for acrosomal status (see "Experimental Procedures"). Progesterone increased the frequency of AR from 13 Ϯ 3% in controls (n ϭ 3; imaged and assessed for AR without a progesterone stimulus) to 25 Ϯ 2% after progesterone stimulation (p Ͻ 0.025; t test of arcsine-transformed data). Progesterone-stimulated cells were then sorted into those that oscillated after the initial [Ca 2ϩ ] i transient and those that did not, and occurrence of AR was assessed. Fewer of the oscillating cells underwent AR (15 Ϯ 3%) compared with the cells showing the simple biphasic response (29 Ϯ 2%) (p Ͻ 0.05, paired t test of arcsine-transformed data; five experiments, 522 cells). A similar analysis of two progesterone gradient experiments confirmed that oscillations did not induce AR, but no significant inhibitory effect was detected. DISCUSSION A logarithmic progesterone gradient, simulating the stimulus encountered by the sperm during approach to the egg, induced a response unlike any described previously in human or other mammalian sperm. Cells responded with a [Ca 2ϩ ] i ramp that (in more than one-third of the cells) triggered slow [Ca 2ϩ ] i oscillations that did not encode stimulus strength in their frequency. An initial large [Ca 2ϩ ] i transient, which has been characteristic of all previous reports of the action of progesterone applied as a bolus, was never seen. Stimulation of cells with a 3 M progesterone step, as reported previously (23), sometimes initiated similar oscillation of [Ca 2ϩ ] i (after the initial [Ca 2ϩ ] i transient) but much less frequently. The occurrence of oscillations only in some cells (up to 45%, mean of 34%) is not surprising since human spermatozoa show marked heterogeneity in their functional attributes (39). Furthermore, the calculated rates of occurrence reflect a rigid definition of oscillations (see "Experimental Procedures"). Many cells generated an irregular variation of [Ca 2ϩ ] i after stimulation with progesterone, which could be converted to regular oscillation by ryanodine (Fig. 4a).
[Ca 2ϩ ] i oscillations in spermatozoa have been reported previously, but the progesterone-induced response reported here is very different. Suarez et al. (40) detected rapid oscillation of [Ca 2ϩ ] i , in synchrony with flagellar beat (3.5 Hz), in the tail of hamster sperm, an effect driven by Ca 2ϩ influx. Wood et al. (41) observed rapid [Ca 2ϩ ] i transients (duration ϭ 0.2-1 s; period ϭ 0.5-5 s; dose-dependent) in speract-stimulated sea urchin spermatozoa, apparently generated by a mechanism involving change in membrane potential and Ni 2ϩ -sensitive Ca 2ϩ influx through voltage-operated Ca 2ϩ channels. In contrast, progesterone-induced oscillations in human spermatozoa are doseindependent, large, and relatively slow (period ϭ 2-5 min at 25°C and 0.2-0.6 min at 37°C) and resemble those that, in somatic cells, are almost always generated by mobilization of intracellular stored Ca 2ϩ . Meizel et al. (42) reported very small [Ca 2ϩ ] i ripples in human spermatozoa (duration and period ϭ 10 -20 s) superimposed on the transient induced by stepped application of 3.2 M progesterone, which may be equivalent to the [Ca 2ϩ ] i "noise" observed in some cells in which regular oscillations did not occur.
Mechanisms Underlying [Ca 2ϩ ] i Oscillations in Human Spermatozoa-The [Ca 2ϩ ] i oscillations described here could be generated either by periodic regulated influx of extracellular Ca 2ϩ or by cyclical emptying/refilling of an intracellular Ca 2ϩ store (as in somatic cells) (34). Blockade of membrane Ca 2ϩ channels with 1 mM La 3ϩ or use of EGTA-buffered saline (to nullify or reverse the inward Ca 2ϩ gradient) both arrested the oscillations. However, under both conditions, a few cells generated occasional [Ca 2ϩ ] i transients after abolition of Ca 2ϩ influx, presumably by mobilization of stored Ca 2ϩ . More strikingly, oscillations persisted without any reduction in the rate of the rise or peak amplitude in low-Ca 2ϩ sEBSS (in which [Ca 2ϩ ] o was Ͻ5 M). In contrast, reduction of [Ca 2ϩ ] o to 500 M abolishes oscillations in sea urchin spermatozoa (41). Established oscillations were arrested by TMB-8 and by tetracaine, inhibitors of the mobilization of Ca 2ϩ stores. All of these observations are compatible with a role for cyclical store emptying/ refilling in the generation of [Ca 2ϩ ] i oscillations in human spermatozoa, but clearly there remains a dependence on Ca 2ϩ influx to maintain the minimum [Ca 2ϩ ] i necessary for store refilling, as is often the case in somatic cell [Ca 2ϩ ] i oscillations (29,43). We have shown that store-operated and arachidonateregulated calcium channels are unlikely to fulfill this role. Valinomycin and nifedipine (23) and nicardipine and verapamil 3 do not affect [Ca 2ϩ ] i oscillations in human sperm, indicating that voltage-operated Ca 2ϩ channels do not participate significantly in their generation. Thus, although extracellular Ca 2ϩ clearly contributes to the generation of [Ca 2ϩ ] i oscillations, the details of this contribution are far from clear. [Ca 2ϩ ] i oscillations were highly resistant to thapsigargin, suggesting that sarcoplasmic/endoplasmic reticulum Ca 2ϩ -ATPases did not contribute significantly to store refilling. The occurrence of thapsigargin-insensitive Ca 2ϩ stores, dependent upon secretory pathway Ca 2ϩ -ATPases, is well documented (44), and expression of these pumps in spermatozoa must be investigated.
We obtained no evidence that inositol trisphosphate receptors participate in the generation of [Ca 2ϩ ] i oscillations in human sperm, but pharmacological manipulations that affect RyRs were clearly effective. Tetracaine inhibited oscillations, and caffeine and ryanodine caused oscillations to become larger and more defined. Ryanodine also influenced oscillation frequency in a characteristic dose-dependent manner. The high doses of ryanodine that were required probably reflect (at least in part) slow permeation of ryanodine into the cells, but it is also possible that the RyR in sperm shows unusually low ryanodine sensitivity. RyRs were not detected in bovine spermatozoa (45), but RyR-3 has been detected in mouse sperm, and RyR-1 has been detected in immature cells (46). The ability of a temporary progesterone stimulus to switch cells to a prolonged (apparently irreversible) pattern of [Ca 2ϩ ] i oscillations (Fig. 3a) is consistent with a model in which Ca 2ϩ -induced Ca 2ϩ release, mediated by RyRs, is sufficient for repetitive store mobilization leading to [Ca 2ϩ ] i oscillation in human sperm.
During spermiogenesis in many species, including humans (47, 48) a membrane complex termed the redundant nuclear envelope is formed in the neck region of the sperm. This structure is believed to function as a Ca 2ϩ store in bovine spermatozoa and to regulate flagellar beating (see below) (12,45).
[Ca 2ϩ ] i oscillations and binding of BODIPY FL-X ryanodine in human spermatozoa both appear to occur primarily in the area of the caudal head/mid-piece junction. Since oscillations were not dependent upon mitochondrial Ca 2ϩ uptake (insensitivity to high doses of 2,4-dinitrophenol), it is possible that the redundant nuclear envelope also functions as a Ca 2ϩ store in human spermatozoa and contributes to the oscillations reported here.
Functional Significance of [Ca 2ϩ ] i Oscillations-The [Ca 2ϩ ] i oscillations do not induce AR (and apparently suppress AR induced by progesterone), so the progesterone stimulus encountered in vivo may not induce significant levels of AR before the cells encounter the zona pellucida. However, in accordance with previous findings that flagellar beat mode is regulated primarily by changes in [Ca 2ϩ ] i (4,49), the [Ca 2ϩ ] i oscillations were synchronized with movements of the sperm head driven by enhancement of flagellar activity during the periods of high [Ca 2ϩ ] i . The increased flagellar bend was less extravagant than hyperactivation as described in human spermatozoa (e.g. Ref. 50), although this may well reflect the restriction caused by cell attachment and/or selection of cells in which visualization of the flagellum was possible. Similar cyclical changes in flagellar beat occur in activated rabbit spermatozoa, with a cycle period of several minutes (51). Mortimer and Swan (52) reported that human spermatozoa reversibly switch between non-hyperactivated and hyperactivated motility. The mean duration of completed periods of hyperactivation observed in that study was 2 s, which is shorter than the typical duration of [Ca 2ϩ ] i oscillations at 37°C (Fig. 2d), but because the maximum length of sperm track analyzed was 8.9 s, prolonged transitions would not have been observed.
Once established, progesterone-induced [Ca 2ϩ ] i oscillations did not change with the strength of the progesterone stimulus, a consequence of the unusual mechanism of their generation, so we consider that cyclical regulation of flagellar activity is likely to be their primary function. Enhancement of motility is vital for penetration of the zona (4). Its failure leads to failure to fertilize in CatSper2-null mice (53). Switching between pushing (low amplitude flagellar bend) and "rocking"/cutting movements of the head is believed to facilitate progress through the zona (54,55), and we consider this to be a likely role for the repetitive [Ca 2ϩ ] i -controlled switching of motility. The relationship of the Ca 2ϩ -mobilizing mechanisms responsible for oscillations to the sperm-specific channels CatSper1 and CatSper2, both of which are required for hyperactivation (53,56), must be elucidated.
In summary, application of progesterone gradients to human spermatozoa initiates a novel response, comprising a [Ca 2ϩ ] i ramp with superimposed slow (dose-independent) oscillations. These oscillations, which are generated by a mechanism that is most unusual compared with that typically described in somatic cells, do not induce AR but modulate flagellar beat, an effect potentially of great importance in penetration of the egg vestments prior to fertilization.