The Ca2+-activated K+ current of human sperm is mediated by Slo3

Sperm are equipped with a unique set of ion channels that orchestrate fertilization. In mouse sperm, the principal K+ current (IKSper) is carried by the Slo3 channel, which sets the membrane potential (Vm) in a strongly pHi-dependent manner. Here, we show that IKSper in human sperm is activated weakly by pHi and more strongly by Ca2+. Correspondingly, Vm is strongly regulated by Ca2+ and less so by pHi. We find that inhibitors of Slo3 suppress human IKSper, and we identify the Slo3 protein in the flagellum of human sperm. Moreover, heterologously expressed human Slo3, but not mouse Slo3, is activated by Ca2+ rather than by alkaline pHi; current–voltage relations of human Slo3 and human IKSper are similar. We conclude that Slo3 represents the principal K+ channel in human sperm that carries the Ca2+-activated IKSper current. We propose that, in human sperm, the progesterone-evoked Ca2+ influx carried by voltage-gated CatSper channels is limited by Ca2+-controlled hyperpolarization via Slo3. DOI: http://dx.doi.org/10.7554/eLife.01438.001

A picture has emerged that the inventory and control of ion channels in mouse and human sperm are surprisingly different. For example, human but not mouse sperm harbor functional proton Hv1 channels (Lishko et al., 2010); purinergic P2X channels are functional in mouse (Navarro et al., 2011), but not in human sperm (Brenker et al., 2012); the sperm-specific CatSper (cation channel of sperm) Ca 2+ channel is activated by progesterone in humans (Lishko et al., 2011;Strünker et al., 2011), but not in mouse (Lishko et al., 2011). These different channel inventories and different mechanisms of channel activation might reflect adaptations to species-specific challenges encountered by sperm in the female genital tract.
In humans, it is unknown whether Slo3 is functionally expressed in sperm and serves a similar key role for fertilization. Here, we examine the properties of human sperm K + current by patch-clamp recording and also define properties of currents arising from heterologous expression of hSlo3 and its auxiliary subunit hLRRC52 . We find that human I KSper and heterologously expressed human Slo3 currents share similar biophysical properties, pharmacology, and ligand dependence. Furthermore, we identify Slo3 and LRRC52 proteins in human sperm. Remarkably, whereas mouse Slo3 is exclusively controlled by pH i (Schreiber et al., 1998;Zhang et al., 2006a;Yang et al., 2011;Zeng et al., 2011), activation of human Slo3 is regulated by [Ca 2+ ] i and also, more weakly, by cytosolic alkalization. These results show that, between mouse and human sperm, signalling pathways controlling the principal K + channel and, thereby, V m are also distinctively different.

Identification of I Ksper in human sperm
We recorded currents from human sperm by the patch-clamp technique . Depolarizing voltage steps from a holding potential of −80 mV evoked outwardly rectifying voltagegated currents ( Figure 1A,B). At pH i 7.3, current amplitudes at −100 mV and 100 mV were −7.5 ± 5 pA and 80 ± 15 pA, respectively (n = 5) (mean ± SD; n = number of experiments) ( Figure 1F). Several controls established that the currents are carried by K + channels and not by Cl − channels or CatSper : lowering the extracellular K + concentration ([K + ] o ) from 150 to 5 mM shifted the reversal potential (V rev ) from 9.2 ± 1.5 mV to −16.5 ± 10 mV (n = 5) ( Figure 1B,C). At low [K + ] o , a decrease of extracellular [Cl − ] o did not change V rev any further ( Figure 1C, B), showing that currents are not carried by Cl − channels. Replacing intracellular K + by Cs + almost completely abolished outward currents at V m ≤ 100 mV ( Figure 1F, D). However, at V m ≥ 100 mV, residual Cs + outward currents persisted. In mouse eLife digest A sperm that has been ejaculated into the female reproductive tract must complete a number of tasks to pass on its genes to the next generation. First it must travel along a meandering route to encounter an egg, before pushing through a jelly-like coating that surrounds the egg and then, finally, fusing with the egg's surface membrane. In order to complete these steps and fertilise the egg, a sperm must undergo a process called 'capacitation'. This process, and a variety of other sperm functions, involves the controlled flux of positive ions into and out of the sperm via specific ion channels that are located in the cell membrane.
The properties of the ion channels that allow protons and calcium ions to move into and out of human sperm are well understood, but less is known about the channels that control the movement of potassium ions. In mice, a channel called Slo3 allows potassium ions to flow out of the sperm and makes the membrane voltage of these cells more negative. Also, in mice, this channel is essential for the sperm to function correctly, and for fertilization. However, in humans, it is unclear if the Slo3 channel is present in sperm and if it performs the same role. Now, Brenker et al. have shown that the flow of potassium ions out of human sperm occurs via the Slo3 channel, and that human Slo3 is responsible for setting the membrane voltage of these cells. However, whereas the mouse Slo3 channel is opened in response to a decrease in the concentration of protons within the sperm (i.e., an increase of the pH inside the cell), human Slo3 is largely controlled by changes in the levels of calcium ions. An increase in the calcium concentration within the cell opens the human Slo3 channel, more than a decrease in the proton concentration does.
Altogether, Brenker et al. identify Slo3 as the principal potassium channel in human sperm and reveal more fundamental differences between human sperm and mouse sperm. Thereby, this work further stresses the need to be cautious about using mice as a model of male fertility in humans.   Slo3 −/− sperm, monovalent outward currents persisting at very positive V m are carried by CatSper . Monovalent mouse and human CatSper current is suppressed by extracellular Ca 2+ (Kirichok et al., 2006;Lishko et al., 2011;Lishko et al., 2012;Zeng et al., 2013). Consistent with CatSper channels conducting the residual Cs + current in human sperm, current amplitudes at 120 mV were progressively suppressed by increasing extracellular Ca 2+ (Figure 1-figure supplement 1E,F).
We conclude that monovalent cation currents at V m ≤ 100 mV are carried by K + -selective channels; we refer to this current as human I KSper, analogous to the principal K + current in mouse sperm. At more positive V m , currents are also carried by CatSper channels.
The principal K + channel in human sperm is controlled by Ca 2+ rather than pH i Mouse I KSper is strongly pH dependent (Navarro et al., 2007;Santi et al., 2010;Zeng et al., 2011;Zhang et al., 2006a). We, therefore, examined the pH sensitivity of I KSper and V m in human sperm. Decreasing pipette pH from 7.3 to 6.2 reduced outward currents by 2.2 ± 1.0-fold (n = 4) ( Figure 1A,D,F). Moreover, intracellular alkalization by NH 4 Cl enhanced outward currents by 1.8 ± 0.9-fold (at 100 mV, n = 4) at pipette pH 6.2 ( Figure 1D-F), but not at pipette pH 7.3 ( Figure 1A,B,F). Thus, human I KSper is less sensitive to pH i than mouse I KSper , which is enhanced about fourfold by increasing pH i from 6.5 to 7.5 (Navarro et al., 2007). In mouse sperm, I KSper sets V m in a strongly pH i -dependent manner (Navarro et al., 2007;Santi et al., 2010;Zeng et al., 2011). Therefore, we tested under current-clamp whether, in human sperm, the control of V m by I KSper is dependent on pH i . At pipette pH 6.2, V m was −23 ± 5 mV (n = 4) ( Figure 2A,C left); raising pipette pH to 7.3 or alkalization by NH 4 Cl hyperpolarized sperm only slightly to −30 ± 5 mV and −31 ± 4 mV, respectively (n = 4) ( Figure 2A,B,C left). At pipette pH 7.3, alkalization by NH 4 Cl did not further change V m ( Figure 2B,C left). Of note, at pH i 6.2 and 7.3, V m was independent of [Cl − ] o and dropped to about 0 mV at high [K + ] o (Figure 2A,B,C left). In conclusion, under the recording conditions used here, V m of human sperm is only modestly pH i sensitive. Given the modest effect of alkalization on human K + current, we tested whether Ca 2+ controls I KSper in human sperm. To compare current-voltage (I-V) relations at low and high [Ca 2+ ] i in the same cell, we studied voltage activation of I KSper before and after rapid photorelease of Ca 2+ from caged Ca 2+ (DMNP-EDTA) ( Figure 3A,B), while monitoring [Ca 2+ ] i with the Ca 2+ indicator Fluo-4. Prior to Ca 2+ release, currents were similar to those recorded without Ca 2+ in the pipette solution ( Figure 3A,B). Photorelease of Ca 2+ altered I KSper in several ways: currents activated more rapidly; at V m ≤ 70 mV, amplitudes were enhanced and at V m ≥ 70 mV, amplitudes saturated or even declined; thereby, the outward rectification was diminished; finally, V rev was shifted to more negative potentials ( Figure 3A,B).

Research article
The Ca 2+ dependence of I KSper was quantified by recording the I-V relation at 2.5-1000 µM Ca 2+ in the pipette ( Figure 3C,D). In the absence of Ca 2+ , the mean current amplitude (V m = 3.5 mV) was 3.5 ± 3 pA; increasing Ca 2+ to 40 µM and 1 mM Ca 2+ increased the amplitude to 14 ± 6 pA and 32 ± 17 pA, respectively (n = 3) ( Figure 3C). For [Ca 2+ ] i > 10 µM, current amplitudes were increased in a concentrationdependent manner, and the normalized I-V relation and V rev were shifted to more negative potentials ( Figure 3D, Figure 3-figure supplement 1C). At V m ≿ 100 mV, Ca 2+ -activated currents levelled off or even declined. Control experiments showed that Ca 2+ activated I KSper , and not Ca 2+ -activated Cl − channels that were identified in human sperm (Orta et al., 2012): with 1 mM [Ca 2+ ] in the pipette, a decrease of [K + ] o from 150 to 5 mM shifted V rev from −5.1 ± 10.3 mV to −52.9 ± 7.8 mV (n = 3) ( Figure  Moreover, Ca 2+ -activated I KSper was enhanced to some extent by alkalization. At pH i 6.2, NH 4 Cl increased the mean current amplitude from 52 ± 14 pA to 89 ± 8 pA (50 mV) and shifted V rev from −33 ± 7 mV to −54 ± 14 mV (n = 3) ( Figure 3G,H). Thus, the enhancement of I KSper upon alkalization was similar in the presence and absence of intracellular Ca 2+ .   Figure 2C right, D). In conclusion, under the conditions used here, V m is set by I Ksper that is controlled strongly by Ca 2+ and only modestly by pH i .

Ca 2+ -activated K + currents exhibit hallmarks of Slo3 channels
The pH sensitivity of I KSper , although modest, is reminiscent of the I KSper current in mouse sperm carried by Slo3 channels, whereas the Ca 2+ sensitivity is reminiscent of the prototypical Ca 2+ -activated K + channel Slo1 (Salkoff et al., 2006). To identify the ion channel underlying human I KSper , we tested several inhibitors of Slo3 and Slo1 channels at high [Ca 2+ ] in the pipette, that is when Ca 2+ -activated K + channels are strongly activated. The non-selective K + channel inhibitors quinidine and clofilium, previously shown to inhibit mouse I KSper (Navarro et al., 2007;Zeng et al., 2011) and heterologous Slo3 (Tang et al., 2010), abolished currents ( Figure 4A,B,E). The inhibition by clofilium was irreversible ( Figure 4B), which is a hallmark of its action on I KSper in mouse sperm (Navarro et al., 2007;Zeng et al., 2011). Moreover, perfusion of sperm with quinidine and clofilium depolarized the cell ( Figure 4F). In contrast, tetraethylammonium (TEA) and iberiotoxin (IBTX), which block Slo1 but not Slo3 (Tang et al., 2010), neither affected I KSper nor V m of human sperm ( Figure 4C-F). Although the Slo3 inhibitors employed are not selective for Slo3, the action of these drugs together with the ineffective Slo1 inhibitors provides critical evidence that I KSper is carried by Slo3, but not by Slo1.
As a further signature of the K + channels, we compared single-channel currents of heterologous hSlo3 with those recorded in human sperm. At −60 mV and high intracellular [Ca 2+ ] i , K + channels openings displayed a single-channel conductance of 60-70 pS ( Figure 4H-J), that is similar to that of hSlo3 (70 pS; Figure 6-figure supplement 1;Zhang et al., 2006b), but not to that of Slo1 (280 pS) (Dworetzky et al., 1994). These results demonstrate that the channel underlying I KSper displays pharmacological and functional properties consistent with hSlo3.

Human sperm express Slo3 and LRRC52
Next, we tested for the presence of Slo1 and Slo3 proteins in human sperm. Using targeted protein mass-spectrometry (MS) ( Figure 5A, Supplementary file 1), we identified in purified human sperm 5 and 3 proteotypic peptides corresponding to Slo3 and its auxiliary subunit LRRC52, respectively. As positive controls, we identified the following proteins known to be expressed in human sperm: CatSper, the Ca 2+ -ATPase PMCA4, the proton channel Hv1, Na + /K + -ATPase α4, and IZUMO. However, we did not detect Slo1. Similar results were obtained by shot-gun proteomics (Wang et al., 2013), which identifies Slo3 and other known components of human sperm, but not Slo1.
Moreover, we expressed hSlo3 heterologously with a hemagglutinin(HA)-tag in chinese hamster ovary (CHO) cells ( Figure 5-figure supplement 1). In Western blots of hSlo3-transfected, but not of wild type cells, both anti-HA and anti-hSlo3 antibodies labelled polypeptides with an apparent molecular weight (M w ) of about 120 kDa ( Figure 5B). The predicted M w of Slo3 is 130 kDa. In Western blots of human sperm, the anti-Slo3 antibody labelled polypeptides of ∼125 kDa and ∼80 kDa ( Figure 5B); in both polypeptide bands, we confirmed by MS that the bands recognized by the antibody contained hSlo3. The 80 kDa polypeptide might be a product of the cleaved Slo3 channel. Post-translational cleavage has been reported also for other ion channels in ciliary structures (Molday et al., 1991;Bönigk et al., 2009). Finally, the anti-hSlo3 antibody stained the flagellum ( Figure 5C), consistent with the localization of I KSper in mouse sperm (Navarro et al., 2007). Together, MS, Western blot-analysis, and immunocytochemistry show that human sperm express Slo3.

Heterologous human Slo3 is activated by Ca 2+
Considering that mouse Slo3 is insensitive to Ca 2+ (Schreiber et al., 1998), the Ca 2+ activation of human I KSper is remarkable. Therefore, we studied human Slo3 co-expressed with hLRRC52 in Xenopus oocytes. First, we investigated the pH i sensitivity of hSlo3 in inside-out patches with step depolarizations between −60 and 260 mV from a holding potential of −140 mV. In the absence of Ca 2+ , currents were only modestly activated at pH i 7 ( Figure 6A, upper left), but enhanced at pH i 8 ( Figure 6A, upper right). Similar to previous observations , an increase of pH i from 7 to either pH i 8 or 9 increased currents by 1.9 ± 0.4-fold and 2.2 ± 0.5-fold, respectively (200 mV; n = 4).
At pH i 7 and pH i 8, 60 μM Ca 2+ enhanced both outward currents and inward tail currents ( Figure 6A, middle traces). At 300 μM Ca 2+ , inward tail currents were further enhanced but outward currents were reduced at potentials ≥ 200 mV ( Figure 6A, bottom traces), indicating a block of outward currents by Ca 2+ . Conductance-voltage (G-V) curves generated from tail currents illustrate the activation of hSlo3 at pH i 7 and pH i 8 in the presence of 0, 60, and 300 μM Ca 2+ ( Figure 6B,C). At pH i 7, raising Ca 2+ from 0 to 300 μM enhanced hSlo3 conductance by 6.6 ± 1.7-fold (200 mV), in contrast to the only 2.2-fold increase evoked by raising pH i from 7 to 9. Thus, Ca 2+ activates hSlo3 much more effectively than alkaline pH i . Surprisingly, the amplitudes of Ca 2+ -activated tail currents, whether at 60 or 300 µM Ca 2+ , were similar at pH i 7 and pH i 8 (compare Figures 6B and 6C), suggesting that, at elevated [Ca 2+ ] i , Slo3 is rather insensitive to changes in pH i > 7. At pH i 8, we also examined the action of a broader range of [Ca 2+ ] i . For [Ca 2+ ] i ≥ 10 µM, hSlo3 tail currents increased over at least three orders of magnitude of [Ca 2+ ] ( Figure 6D). Finally, Slo3 single-channel openings in patches held at −60 mV (pH i 8) were also markedly increased by 60 µM and 300 μM Ca 2+ (Figure 6-figure supplement 1), indicating that Ca 2+ enhances Slo3 activation also at physiological V m .
Although the ranges of pH i and Ca 2+ concentrations that affect I KSper and hSlo3 expressed in oocytes are similar, compared to I KSper , the I-V relation of hSlo3 currents appears shifted to more positive potentials (compare Figure 1B,E and 3B with Figure 6B,C). This difference might be due to the non-mammalian expression system, the excised-patch conditions, differences in ionic composition of solutions, or a combination of all of them. Therefore, we recorded whole-cell currents evoked by voltage steps in CHO cells that co-expressed hSlo3 and hLRRC52, using conditions similar to those used for sperm recordings. hSlo3 currents in CHO cells were also modestly enhanced by alkalization ( Figure 6-figure supplement 2), and Ca 2+ shifted the I-V relation to more negative V m ( Figure 6E, Figure 6-figure supplement 2). In the absence and presence of Ca 2+ , the I-V relations of hSlo3 and I KSper were similar ( Figure 6F).
Taken together, the heterologous expression demonstrates that human Slo3 is a Ca 2+ -activated rather than a strictly alkaline-activated K + channel. These results strengthen our conclusion that Slo3 underlies the voltage-, Ca 2+ -, and alkaline-activated I Ksper current in human sperm.

Mouse Slo3 is not activated by intracellular Ca 2+
Although mouse Slo3 is insensitive to intracellular Ca 2+ (Schreiber et al., 1998), we wondered whether the auxiliary subunit LRRC52  might confer Ca 2+ sensitivity on Slo3 channels. Therefore, we co-expressed mSlo3 with mLRRC52. Raising pH i from 7 to 8 strongly enhanced mSlo3 steady-state ( Figure 7A) and tail ( Figure 7B) currents, similar to previous results (Schreiber et al., 1998;Zhang et al., 2006a;Yang et al., 2011). At pH i 8, Ca 2+ (60 or 300 µM) did not enhance currents ( Figure 7A,B). Instead, outward mSlo3 currents were strongly suppressed by Ca 2+ ( Figure 7A) and even tail-current amplitudes were reduced (Figure 7B). A G-V plot derived from tail currents confirms that Ca 2+ does not activate mSlo3, but inhibits mSlo3 following activation by positive potentials   Figure 7C). Currents carried by mSlo3 co-expressed with hLRRC52 were also not activated, but suppressed by Ca 2+ (Figure 7-figure supplement 1). These results exclude the possibility that hLRRC52 confers Ca 2+ sensitivity on hSlo3. Suppression of mSlo3 tail currents by Ca 2+ (Figure 7C,D) reflects persistent occupancy of the mSlo3 pore by Ca 2+ following repolarization. We note that the voltagedependent suppression of currents by Ca 2+ is much more pronounced for mSlo3 compared to hSlo3. Ca 2+ occludes the pore of mSlo3 with about 10-fold higher affinity than that of hSlo3 (Figure 7-figure  supplement 2); therefore, inhibition of hSlo3 by Ca 2+ occurs only at very positive, non-physiological V m .

Research article
Together, our results show that regulation of mSlo3 and hSlo3 channels by cytosolic ligands is distinctively different, paralleling the differential regulation of I Ksper in mouse and human sperm by pH i and Ca 2+ .

Progesterone stimulates Ca 2+ levels sufficient to activate Slo3
In human sperm, the female sex hormone progesterone directly activates CatSper (Lishko et al., 2011;Strünker et al., 2011;Brenker et al., 2012;Smith et al., 2013). Progesterone-evoked Ca 2+ influx via CatSper has been implicated in sperm chemotaxis, hyperactivation, and acrosomal exocytosis (Blackmore et al., 1990;Publicover et al., 2007;Publicover et al., 2008). We examined whether stimulation of human sperm by progesterone enhances Ca 2+ levels sufficient to activate Slo3 channels. Sperm were loaded with Ca 2+ indicators of different Ca 2+ affinity as surrogates for high-to low-affinity Ca 2+ -binding sites ( Figure 8A,C). The progesterone-evoked transient Ca 2+ response was faithfully tracked by high-(K D = 0.35 µM), moderate-(K D = 9.7 µM), and low-affinity Ca 2+ indicators (K D = 90 µM) ( Figure 8A). The Ca 2+ ionophore ionomycin evoked a sustained fluorescence increase reflecting the indicator response to near saturating, millimolar [Ca 2+ ] i ( Figure 8A). For indicators with a K D value ≤ 2.3 µM, the amplitudes of progesterone-and ionomycin-induced Ca 2+ signals were similar ( Figure 8A,B), suggesting that these high-affinity indicators become saturated with Ca 2+ during the progesterone-evoked response. For indicators with K D values ≥ 9.7 µM, the amplitude ratio of progesterone-evoked/ionomycinevoked Ca 2+ signals decreased with increasing K D values. However, even for the low-affinity indicator Fluo-5N, which reports [Ca 2+ ] i changes in a concentration range of about 9-900 µM, the amplitude ratio was as large as 0.25. Thus, considering the dynamic range of indicators ( Figure 8C), our results indicate that Ca 2+ levels reached during a physiological Ca 2+ response are sufficient to activate Slo3.

Discussion
Here, we show that the biophysical and pharmacological properties of human I KSper conform with the properties of hSlo3, but not with those of hSlo1 or other members of the Slo family. First, hSlo3 and I KSper are modestly sensitive to pH i ≤ 7.0 and are more strongly activated by Ca 2+ . Second, the I-V relations of hSlo3 and I KSper are similar, in the absence and presence of intracellular Ca 2+ . Third, several Slo3, but not Slo1 inhibitors block I KSper . Fourth, we identify the Slo3 protein and its auxiliary subunit LRRC52 in human sperm. Finally, both human I KSper  and heterologously expressed hSlo3 ( Figure 9A,C) are inhibited by progesterone; progesterone inhibits human I KSper and hSlo3 with constants of half-maximal inhibition (K i ) of 7.5 µM  and 17.5 ± 2 µM (n = 3), respectively. While this manuscript was under review, Mannowetz et al. (2013) reported that the prototypical Ca 2+ -activated member of the Slo channel family, Slo1, is the principal K + channel in human sperm. Several observations strongly argue against this conclusion. First, Slo1 is inhibited rather than activated at alkaline pH (Avdonin et al., 2003). Second, specific inhibitors of Slo1 did not inhibit I KSper . Third, human Slo1 is largely insensitive to progesterone ( Figure 9B,C). Fourth, we and others (Wang et al., 2013) are unable to identify the Slo1 protein in human sperm. Fifth, if Slo1 carried the current of about 125-150 pA in human sperm (recorded at 100 mV in symmetrical K + and saturating Ca 2+ ) ( Figure 3A,C), this would correspond to the opening of about 5-6 BK channels. Under such conditions, discrete opening and closing transitions of single BK channels would be readily visible; not only at 100 mV, but even more so at voltages < 100 mV. Thus, recordings of I KSper current in human sperm are consistent with lower conductance openings characteristic of Slo3, but not Slo1. Finally, the most obvious discrepancy between the two reports concerns the pharmacology. Mannowetz et al. show that K + currents in human sperm are abolished by the Slo1 inhibitors IBTX, charybdotoxin (CTX), and paxilline. Although we do not know the reason for this discrepancy, there are differences in experimental conditions. We tested the inhibitors at 1 mM extracellular [Ca 2+ ] to prevent monovalent CatSper currents and at 1 mM intracellular [Ca 2+ ] to strongly activate I KSper . Mannowetz et al. tested the drugs at 100 µM extracellular [Ca 2+ ] and in the absence of intracellular Ca 2+ . Under these conditions, sizeable CatSper currents are recorded (Figure 1-figure supplement 1E,F), but activation of Slo1 would be minimal.
What might be the function of Slo3 in sperm? It has been suggested that the I KSper -mediated hyperpolarization reinforces Ca 2+ influx via CatSper by increasing the electrical driving force (Clapham, 2013). Alternatively, we propose that the hyperpolarization serves as a negative feedback that decreases the open probability of CatSper and, thereby, curtails rather than enhances Ca 2+ influx. Why did Slo3 switch in human sperm from a strictly pH-sensitive to a pH-and Ca 2+ -sensitive K + channel? In mouse, Slo3 and CatSper are both voltage-and strongly alkaline-activated. In human but not in mouse sperm, CatSper is directly activated by progesterone and prostaglandins (Lishko et al., 2011;Strünker et al., 2011;Brenker et al., 2012;Smith et al., 2013). Thus, human CatSper mediates a ligand-rather than an alkaline-activated Ca 2+ influx. Moreover, the pH sensitivity of both Slo3 and CatSper is considerably lower in human compared to mouse sperm, suggesting that Ca 2+ activation of Slo3 may have evolved in concert with ligand activation of CatSper. This co-evolution might ensure that ligand-evoked Ca 2+ influx via CatSper is coupled to the Ca 2+ -controlled hyperpolarization via Slo3. Thus, by curtailing Ca 2+ influx via CatSper, I Ksper may serve a similar role in both mouse and human sperm despite its differential regulation by intracellular ligands. The control of V m by Ca 2+ and pH i and the interplay of CatSper and Slo3 deserve further study in intact, freely moving human sperm with non-invasive and kinetic techniques, for example using voltage-sensitive dyes.
Our results suggest that during a progesterone response, global Ca 2+ levels can reach concentrations > 10 µM, sufficient to activate Slo3. CatSper and Slo3 are both located in the principal piece. Moreover, progesterone-induced Ca 2+ signals originate in the flagellum and propagate in a tail-to-head direction (Servin-Vences et al., 2012). Due to the miniscule flagellar volume (about 2.5 fl), opening a few CatSper channels, each conducting several thousand Ca 2+ ions per second, would increase local flagellar [Ca 2+ ] to levels that should readily exceed global [Ca 2+ ] i . The potential interplay of Slo3 and CatSper in sperm is reminiscent of the interplay in neurons between Ca 2+ -activated K + channels and voltage-gated Ca 2+ channels (Ca v ) (Prakriya and Lingle, 2000). In neurons, Slo1 and Ca v channels interact to form local microdomains of Ca 2+ signalling near the plasma membrane (Berkefeld et al., 2006). In microdomains, [Ca 2+ ] can rise to levels ≿ 100 µM that are readily sensed by Ca 2+ -activated K + channels (Rizzuto and Pozzan, 2006). It needs to be shown whether Slo3 and CatSper are organized in similar microdomains.
Channels of the Slo family have been studied as models for allosteric regulation of gating by ligands (Magleby, 2003;Lingle, 2007). Slo channels and a large number of bacterial channels/transporters harbor a homologous octameric intracellular domain, dubbed the gating ring, that provides a template for ligand regulation of the pore domain (Lingle, 2007). The gating-ring motif has evolved for regulation of transmembrane ion flux by nucleotides, Ca 2+ , H + , Na + , Cl − , and probably other cytosolic ligands (Salkoff et al., 2006). Among different Slo isoforms, gating rings share structural similarities (Wu et al., 2010;Leonetti et al., 2012;Yuan et al., 2012). However, the conformational changes that couple binding to gating remain incompletely defined and the location of ligand-binding sites varies markedly (Salkoff et al., 2006). Our study adds an interesting twist, demonstrating that, even among Slo3 orthologues, the gating-ring motif has been exploited for regulation by different ligands -primarily Ca 2+ in hSlo3 and primarily H + in mSlo3. Alignment of human Slo1 sequence with human Slo3 and other mammalian Slo3 sequences illustrates that, although the membrane-associated S0-S6 domain retains considerable identity (Figure 9-figure supplement 1A), there is extensive lack of identity in many segments of the ligand-sensing cytosolic domain (Figure 9-figure supplement 1B). Given that positions of ligandsensing determinants in regulator of K + conductance (RCK)-containing proteins vary markedly within the RCK domain structures, it is not surprising that simple examination of the Slo3 sequence alignments does not reveal obvious determinants of either pH sensitivity in mSlo3 or Ca 2+− sensitivity in hSlo3.
At first glance, the fact that Slo3 orthologues differ in their ligand-dependence seems highly unusual. Yet, it is well-established that proteins essential for fertilization are rapidly evolving; orthologues often display a low degree of sequence similarity (Swanson and Vacquier, 2002;Cai and Clapham, 2008). For a set of mammalian species (human, bovine, mouse, dog, and opossum), the amino-acid identities (excluding a non-specific linker in the cytosolic domain) among Slo1 orthologues is typically > 92.6% (mSlo1 vs hSlo1: 99.5%). Slo3 orthologues exhibit identities in the range of 57-75% (mSlo3 vs hSlo3: 70.4%). Considering that the amino-acid identity between the pH-regulated mSlo3 and Ca 2+ -regulated mSlo1 is 45.7%, the different ligand dependence between hSlo3 and mSlo3 is not so surprising.

Mass spectrometry
After purification by a 'swim-up' procedure, human sperm were lysed by several 'freeze/thaw' cycles and sonification steps in buffer containing (in mM): 10 HEPES pH 7.5, 2 EGTA, 1 DTT, protease inhibitor cocktails (Roche Applied Science and Sigma, Mannheim, Germany), and DNaseI (AppliChem, Darmstadt, Germany). Membranes were sedimented by centrifugation (100,000×g, 30 min, 4°C) and membrane proteins were processed by in-gel digestion, in-solution digestion, or a FASP protocol. Peptides were subjected to 1D-ESI-LC-MS/MS using a nanoAcquity UltraPerformance LC System (Waters, Milford, Massachusetts, USA) coupled to an LTQ Orbitrap Velos or Elite instrument (Thermo, Waltham, Massachusetts, USA). The resulting tandem MS data were searched using the Sequest algorithms embedded in Proteome Discoverer 1.2 (Thermo) against a SwissProt/UniProtKB human protein sequence database (including 56,582 entries). The mass tolerance for precursor ions was set to ≤10 ppm; the mass tolerance for fragment ions was set to ≤ 1 amu. For search result filtering, the false discovery rate (FDR) was set to < 1% and only peptides with search result rank 1 were accepted for identification. For targeted mass spectrometry, the instrument control software used a list of theoretical tryptic peptide masses for the proteins of interest for subsequent CID fragmentation, that is once a mass from the list was detected in the orbitrap full scan, it was preferred over all other co-eluting masses, independent of its signal intensity.

Preparation of recombinant constructs
EST clone BC028701 was obtained from Open Biosystems and verified by sequencing. Based on analysis of the NCBI slo3 gene (kcnu1, NM_001031836), 62 base pairs (bp) present around the S10 region in BC028701 correspond to an intron left over from incomplete mRNA splicing. The intron sequences were removed by site-directed mutagenesis. In addition, there are at least three polymorphic sites present in the EST clone corresponding to amino-acid positions 192, 739, and 768 of hSlo3. The three sites were changed to match those of the human genomic sequence (NCBI reference sequence: NC_000008.10; Chromosome 8). This results in 192W, 739R, and 768W. The full length coding sequence of hSlo3 was subcloned into the oocyte expression vector pXMX (see details in Tang et al., 2010). The hSlo3 sequence used here is identical to that used in another study . An hLRRC52 (NM_001005214.3) clone was generated from two HEK genomic DNA fragments of 622 bp and 320 bp which correspond to hLRRC52 exons 1 and 2, respectively. The two fragments were amplified via over-lapping PCR and subcloned into the pXMX vector. For the mCherry-tagged hLRRC52, seven glycines were added as a spacer between the carboxy-terminus of hLRRC52 and the amino-terminus of mCherry; the hLRRC52-mCherry construct was also subcloned into pXMX vector. All cDNA clones were verified by sequencing. For expression in oocytes, cRNA was synthesized by SP6 polymerase after the cDNA template was linearized with the restriction enzyme MluI.
For expression in CHO cells, the full length coding sequence of hSlo3 was amplified from human testis cDNA. A perfect Kozak consensus sequence preceding the start codon and a sequence coding for a carboxy-terminal hemagglutinin tag (HA-tag) were added. The coding sequence harbored an arginine at the polymorphic site at position 768. The DNA was subcloned into a pcDNA3.1(+) vector (Invitrogen); the sequence coding for the neomycin resistance gene was replaced by the coding sequence for either citrine or EGFP. The hLRRC52-mCherry construct described above was also subcloned into the pcDNA3.1(+) vector for expression in CHO cells. All cDNA clones were verified by sequencing.

Protein preparation and western blot analysis
Wild type CHO cells and cells expressing the human Slo3 channel were lysed in a buffer containing 10 mM Tris/HCl, pH 7.5, 140 mM NaCl, 1 mM EDTA, 1% Igepal CA-630, 0.5% sodium deoxycholate, 0.1% SDS, and mammalian protease inhibitor cocktail (mPIC, Sigma) and incubated on ice for 30 min. The suspension (total lysate) was centrifuged for 5 min at 10,000×g (4°C) and the supernatant was used for WB analysis. Human sperm (5 × 10 6 ) were resuspended in 2x SDS sample buffer containing β-mercaptoethanol. All samples were heated for 5 min at 95°C and separated by 4-12% SDS-polyacrylamide gel electrophoresis. For WB analysis, proteins were transferred onto PVDF membranes, probed with antibodies, and analysed using the LAS-3000 System (Fujifilm).
Author contributions CB, TS, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article; ZY, AM, FAE, CT, AP, WB, CJL, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article; XX-M, Drafting or revising the article, Contributed unpublished essential data or reagents; UBK, Analysis and interpretation of data, Drafting or revising the article

Additional files
Supplementary files • Supplementary file 1. Indicators of merit for the mass spectrometric results.