Cyclic Nucleotide-gated Channels on the Flagellum Control Ca²⁺ Entry into Sperm

We will use a nomenclature for CNG channels that describes the subunit type (α, β) and the channel subfamily (1, 2, 3). In vertebrates, three distinct genes encoding α subunits have been identified. The respective gene products have been originally identified either in rod (1) or in cone (2) photoreceptors, or in olfactory sensory neurons (3); alternatively spliced variants are indicated by small letters. A preceding p (for plasmid) indicates the respective cDNA sequence.

Bovine testicular tissue and retinae were frozen immediately in liquid nitrogen at the abattoir. Poly(A)⁺ RNA was isolated using the guanidinium isothiocyanate method (Chirgwin et al., 1979), followed by a cesium-trifluoroacetic acid gradient and oligo(dT)-cellulose chromatography (Aviv and Leder, 1972). First-strand cDNA was synthesized with M-MLV reverse transcriptase (Life Technologies, Eggenstein, Germany) using oligo(dT)₁₇ or random hexamers as primer. Primers were removed by filtration on a Centricon-100 spin column (Amicon Corp., Easton, TX). A random-primed cDNA library was constructed from bovine testis poly(A)⁺ RNA in λZAP II vector (Stratagene, La Jolla, CA).

Degenerate primers corresponding to amino acid (aa) sequences KYMAFFE (aa 882–888) and QMIFD (aa 1083–1087) of the CNG channel β subunit from bovine rods (CNCβ1a; Körschen et al., 1995) were used to amplify a fragment from bovine testis first-strand cDNA. The PCR fragment was used to screen an oligo(dT)-primed testis cDNA library (Weyand et al., 1994). The longest clone thus isolated contained nucleotides 1319–3061 of the final clone pCNCβ1c (see Results, Fig. 2,a). The sequence was identical to the corresponding sequence of CNCβ1a (Körschen et al., 1995). Subsequently, a random-primed cDNA library was screened with a cDNA probe of CNCβ1a (corresponding to nucleotides 1193–1798 of pCNCβ1c). The longest clone thus isolated carried nucleotides 1138–2916 of pCNCβ1c. The combined sequence of the partial clones (pCNCβP) did not harbor the complete coding region (see Fig. 1,a). Because cDNA sequences were identical with the corresponding sequences of CNCβ1a, the 5′ end of testis β cDNA was probed by PCR using primers derived from the CNCβ1a sequence (for positions of primers, see Fig. 1 a). cDNA from bovine retina was used as control for amplification. The initial primer set consisted of primer 1 (GGATGGATTCCAGGCGG; inverse complement of nucleotides 1305–1321 of pCNCβ1c) as the 3′ primer and primer 2 (TGCTCTGCTGCAAGTTCAAA; nucleotides 860–879) or primer 3 (GAACTGCAGGTGGAAGAC; nucleotides 370–387) as 5′ primers. A set of nested primers consisted of primer 4 (GGCGTTTGAACTTGCAGC; inverse complement of nucleotides 866–883) as 3′ primer, and primer 5 (AGCTCATCGACCCTGACG; nucleotides 728–745), primer 6 (GCAACCTCGACAGCCAGC; nucleotides 631–648), or primer 7 (CTCAAGATGCTGTCACCG; nucleotides 547–564) as 5′ primers. For blot hybridization, PCR fragments were hybridized under high-stringency conditions (5× SSC, 5× Denhardt's, 0.1 mg/ml denatured herring testis DNA, 0.1% SDS, 65°C) with ³²P-labeled DNA probes (∼10⁶ cpm/ml). Filters were washed with 1× SSC, 0.1% SDS at 65°C (2 × 30 min).

The PCR analysis demonstrated that short and long β transcripts are expressed in testis. The 5′ ends of the long β cDNAs were isolated by modification of the rapid amplification of cDNA ends (RACE) technique (Frohman et al., 1988). Random-primed cDNA was (dA)-tailed. PCR was carried out with the gene-specific primer R (CTTGGGGCTCTCCTCATCGG, inverse complement of nucleotides 752–771 of pCNCβ1c; for the location of primer R, see Fig. 1 a), the hybrid adapter primer GACTCGAGTCGACATCGA(T)₁₇, and the adapter primer GACTCGAGTCGACATCGA. To enrich the desired PCR product, a second PCR was carried out on the first PCR product using a nested gene-specific primer (TTCCTGAGGCTCCCTGTGG; inverse complement of nucleotides 494–512 of pCNCβ1c), and the adapter primer. PCR fragments >150 bp were cloned into pBluescript SK-vector. Colonies were hybridized with a DNA probe amplified with two primers (GAACTGCAGGTGGAAGAC, nucleotides 370–387 of pCNCβ1c; TTCCTGAGGCTCCCTGTGG, inverse complement of nucleotides 494–512) using cDNA of CNCβ1a as template. We obtained several clones with three distinct 5′ ends: pCNCβ1c (seven clones), pCNCβ1d (two clones), and pCNCβ1e (three clones).

Attempts to clone the 5′ end of the short β cDNA (CNCβ1f) by screening cDNA libraries were not successful. The PCR analysis, however, indicated that the short β cDNA begins in the segment flanked by primers 5 and 6.

A blot with poly(A)⁺ RNA from bovine testis (6 μg) and retina (10 μg) was hybridized under high-stringency conditions (50% formamide, 5× SSC, 5× Denhardt′s, 0.1 mg/ml denatured herring testis DNA, 0.1% SDS, 65°C) with ³²P-labeled riboprobes. Filters were washed with 1× SSC, 0.1% SDS at 65°C (2 × 30 min).

Polyclonal antibody FPc 21K is directed against the NH₂-terminal domain (aa 574–763) of the β′ part of CNCβ1a (Körschen et al., 1995), whereas polyclonal antibody PPc 32K is directed against an epitope (aa 1292–1334) close to the COOH terminus of CNCβ1a. Polyclonal antibody PPc 23 was raised against the COOH-terminal domain (aa 593–706) of the α subunit of the bovine cone CNG channel (CNCα2; Weyand et al., 1994). Polyclonal antibodies were purified from rabbit serum by affinity chromatography on a column consisting of the respective antigen (3–5 mg) coupled to activated CH-Sepharose 4B (Pharmacia Biotech, Inc., Piscataway, NJ). The antibody was eluted with 0.1 M glycine HCl, pH 2.5, immediately neutralized with 1/8 vol 1 M Tris HCl, pH 8.0, and dialyzed against PBS. Purified antibodies were concentrated using a Centricon-100 spin column (Amicon Corp.) and stored at –80°C in the presence of 50% glycerol, 0.2 mg/ml BSA, 0.1% NaN₃. The specificity of the antibodies in immunohistochemistry was tested on cryosections of bovine retina.

For transient expression, cDNAs for the short testis β subunit (CNCβ1f) and one of the long β subunits (CNCβ1c) were cloned into pcDNAI vector (Invitrogen Corp., Carlsbad, CA). COS-1 cells were transfected by a modified calcium phosphate coprecipitation method (Chen and Okayama, 1987). Transfected COS-1 cells, bovine sperm, and testicular tissue were homogenized in 10 mM Hepes, pH 7.5 (NaOH), 10 mM NaCl, 1 mM DTT, 0.1 mM EGTA, 5 μg/ml aprotinin, 5 μg/ml leupeptin, 1 μg/ml pepstatin, 500 μg/ml PEFABLOC, and 10 mM benzamidine (buffer A). The suspension was centrifuged at 100,000 g for 20 min (4°C). The membrane pellet was resuspended in buffer A containing 500 mM NaCl, washed by centrifugation, resuspended in buffer A containing 100 mM NaCl, and 0.8% N-dodecylmaltoside, and left on ice for 15 min. Insoluble material was separated from solubilized membrane proteins by centrifugation (8,000 g, 10 min, 4°C). Membranes of bovine rod outer segments were washed two times with buffer A. Membrane proteins were separated by SDS-PAGE, transferred to Immobilon-P membrane (Millipore Corp., Bedford, MA), and sequentially probed with primary antibodies and secondary HRP-coupled goat anti-rabbit antibody for enhanced chemiluminescence detection (Amersham Corp., Arlington Heights, IL).

Tissue blocks of bovine testes and epididymes were fixed in Bouin's fluid for 24 h at 20°C. Subsequently, blocks were embedded in paraffin; 6-μm sections were cut and mounted on chrome-gelatine precoated slides. Immunoreactivity was tested using a combination of the peroxidase-antiperoxidase (PAP) technique and the avidin-biotin-peroxidase complex (ABC) method (Middendorff et al., 1996). Sections were incubated for 24 h at 4°C with the primary antibodies PPc 23 (0.3–3 μg/ml), FPc 21K (1–10 μg/ml), or PPc 32K (0.06–0.9 μg/ml) diluted in PBS, 0.2% BSA, 0.25% Triton X-100, 0.1% NaN₃. Peroxidase activity was visualized by the nickel glucose oxidase technique (Záborsky and Léránth, 1985) with DAB as chromogen.

Cauda epididymal sperm and ejaculated sperm were fixed on glass slides for 10 min with 4% paraformaldehyde, PBS, and were then washed twice with PBS. Cells were preincubated with PBS, 0.1% BSA, 1% Triton X-100, and 0.1% Tween-20 (30 min), and were then incubated overnight at 4°C with the primary antibody diluted in preincubation solution (PPc 23: 1–3 μg/ml; FPc 21K: 1–10 μg/ml; PPc 32K: 0.3–0.6 μg/ml). Cells were washed three times and incubated for 3 h with secondary goat anti–rabbit antibody coupled to a carboxymethylindocyanine dye (CY3) diluted 1:1,000 (1.5 μg/ml) in PBS, 2% FCS, 0.1% BSA, 1% Triton X-100, 0.1% Tween-20. After washing three times, cells were coverslipped in Moviol and examined by fluorescence microscopy.

Experiments were performed in the absence of primary (and in the case of tissue sections also of secondary or tertiary) antibodies. For control of specificity, primary antibodies were preadsorbed with an 80-fold molar excess of the respective immunogenic or an unrelated antigen.

Cryoconserved ejaculated bovine sperm were thawed (45 s, 37°C), diluted 10-fold with 200 mM Tris, 55 mM glucose, 65 mM citric acid monohydrate, pH 7.2 (NaOH), and washed twice by centrifugation (830 g, 4 min, room temperature). Sperm were resuspended in buffer B (140 mM NaCl, 4.6 mM KCl, 2 mM CaCl₂, 10 mM glucose, 10 mM Hepes, pH 7.4 (NaOH), checked for motility (∼60% progressive motile cells), and then loaded in the dark for 30 min with the cell-permeant acetomethoxy ester of Fluo-3 (4.4 × 10⁻⁶ M in the presence of 0.01% Pluronic F-127; Molecular Probes, Inc., Eugene, OR). Sperm were centrifuged, washed (buffer B; 830 g, 4 min, room temperature), resuspended in buffer B, and then put on coverslips. After 5 min, buffer B was carefully aspirated, and sperm were incubated (10 min) with the respective incubation solution containing the membrane-permeable caged cyclic nucleotide analogue (see below). Incubation solutions were: buffer B, buffer B with additives (see Results), or buffer B with 500 μM EGTA instead of 2 mM CaCl₂.

Ca²⁺-dependent changes in fluorescence intensity of Fluo-3 were detected with a confocal laser scanning microscope LSM 410 invert (Zeiss GmbH, Jena, Germany) using an oil immersion lens (100/ 1.3). Fluo-3 was excited with an argon–krypton laser (488 nm; Melles Griot, Carlsbad, CA). Fluorescence was measured at wavelengths >515 nm. For beam-splitting, a dichroic mirror FT 510 and a cutoff filter LP 515 in front of the detector unit were used.

For flash photolysis experiments, lyophilized 4,5-dimethoxy-2-nitrobenzyl (DMNB) 8-pCPT-cGMP (axial isomer), (7-methoxy-coumarin-4-yl)methyl (MCM) 8-Br-cGMP (axial isomer), and MCM 8-Br-cAMP (axial isomer) were dissolved in DMSO and then diluted in the respective incubation solution to the desired final concentration. The final DMSO concentration (maximally 2%) had no detectable effect on the fluorescence intensity of sperm. Synthesis and photochemical properties of the caged compounds will be described elsewhere. Due to its limited solubility in aqueous solutions, DMNB 8-pCPT-cGMP was used at concentrations ≤50 μM.

DMNB 8-pCPT-cGMP was photolyzed with a continuous argon-ion laser (364 nm; Spectra-Physics, Darmstadt, Germany) that allowed local UV irradiation of sperm. Therefore, UV-induced bleaching of Fluo-3 is restricted to the illuminated region, and thus does not compromise determination of the Ca²⁺-induced increase of fluorescence in nonilluminated regions. Either the proximal or the distal part of the flagellum was irradiated eight times by 8-ms UV flashes at intervals of 1 s. The change in fluorescence of Fluo-3 was measured 1 s after the last UV flash in various regions of the sperm (see legend to Fig. 7).

The MCM-esters of 8-Br-cGMP and 8-Br-cAMP have a very low absorbance at 364 nm, and therefore were photolyzed with an N₂ pulsed- laser MSG 800 (337 nm; 500 ps; Lasertechnik, Berlin, Germany). The UV light was guided onto the coverslip with fiber optics that allowed illumination of entire sperm. The sperm were UV-irradiated 10× by 500-ps flashes in intervals of 0.5 s. The fluorescence of Fluo-3 was measured 1 s after the last UV flash. To determine the decrease in fluorescence caused by bleaching the dye, the fluorescence in different regions of sperm was measured with the same protocol before and after UV irradiation without caged cyclic nucleotides. The mean difference in fluorescence, ΔF, determined from measurements of 12 different sperm, was subtracted from the increase in fluorescence measured in the presence of the caged compounds.

One short and three long cDNAs encoding β subunits of CNG channels have been identified (see Materials and Methods). The three cDNAs (pCNCβ1c [3095 bp], pCNCβ1d [3102 bp], and pCNCβ1e [3302 bp]) encoding the long β subunits differ only in their 5′ends (see Fig. 1,a and Fig. 2). A cRNA probe (for location of probe A, see Fig. 1,a) specific for the 5′ terminal region of the long β subunit cDNAs, hybridized to a transcript of ∼3.3 kb in testis poly(A)⁺ RNA (Fig. 1,b, lane T, left), and to ∼7.4 kb and ∼4.4 kb transcripts in retinal poly(A)⁺ RNA (Fig. 1,b, lane R, left). The retinal ∼7.4-kb transcript codes for the β subunit of the CNG channel from rod photoreceptors (CNCβ1a), and the ∼4.4-kb transcript for a glutamic acid– rich protein (GARP; Körschen et al., 1995). The length of the cDNAs (3095–3302 bp) is consistent with the idea that the long β subunits are encoded by the ∼3.3-kb transcript(s). For pCNCβ1c and pCNCβ1e, the translational initiation sites were assigned to the first ATG triplet downstream of a nonsense codon in the same reading frame. pCNCβ1e harbors a second ATG triplet downstream of the first triplet (see Fig. 2 c). In contrast to the first ATG triplet, the second ATG triplet is flanked by a reasonably well-conserved Kozak sequence (GCATCATGG), and therefore represents a better translational initiation site than the first triplet. Clone pCNCβ1d does not contain a nonsense codon upstream of a putative translational initiation site; therefore, the initiation site is not certain. The deduced aa sequence of CNCβ1c consists of 938 aa with a calculated molecular mass of ∼104.5; pCNCβ1e either encodes a polypeptide of 952 aa (∼105.9 kD), or if the second ATG triplet is used, a polypeptide of 941 aa (∼104.8 kD). Biel et al. (1996) have identified a β subunit (CNG4c) from bovine testis by cloning cDNA that differs from CNCβ1c by four aa exchanges and an additional residue.

A cRNA probe derived from the β′ part of the rod β subunit (for location of probe B see Fig. 1,a) in addition to the ∼3.3 kb transcript recognized a transcript of ∼2.4 kb in testis poly (A)⁺ RNA (Fig. 1,b, lane T, right) and, as expected, the ∼7.4-kb transcript in retinal poly(A)⁺ RNA (Fig. 1 b, lane R, right). The hybridization signal resulting from the ∼2.4-kb transcript was much stronger than that from the ∼3.3-kb transcript, suggesting that at least two transcripts of different size and abundance are expressed in testis.

The 5′ end of the ∼2.4-kb transcript was obtained by PCR. We used a set of primer pairs specific for the NH₂-terminal region of the β′ part of CNCβ1a to amplify 5′- located sequences (for positions of primers, see Materials and Methods and Fig. 1,a). Using primer pairs 1/2 and 1/3, fragments were amplified from testis cDNA that matched in size and sequence the corresponding fragments amplified from retinal cDNA. Amplification with primer pair 1/2 was significantly stronger than with primer pair 1/3 (data not shown). This result indicated that the 5′ end of an abundant testis transcript is located between primers 2 and 3, and that a larger transcript is much less abundant as demonstrated above by the Northern blot (Fig. 1,b, lane T, right). Further PCR experiments with nested primer pairs 4/5, 4/6, and 4/7 (Fig. 1 c) indicated that the short abundant transcript begins in the segment flanked by primers 5 and 6. The size of the cDNA ranging from this segment to the poly(dA) tail of the partial clone pCNCβP (∼2.4 kbp) agrees well with the size of the short ∼2.4-kb transcript, which suggests that the segment flanked by primers 5 and 6 is part of the 5′ nontranslated region of the ∼2.4-kb transcript. However, this segment is part of the open reading frame in the cDNAs of the long testis β subunits. Perhaps the ∼2.4-kb transcript is produced by use of an alternative promotor, as has been recently proposed for a short β transcript from human retina (Ardell et al., 1996).

The cloned cDNA of the short β subunit variant (CNCβ1f) is similar to a cDNA isolated from a human retina library (hrCNC2a; Chen et al., 1993; Ardell et al., 1996). Coexpression of hrCNC2a with the α subunit from human rod photoreceptors gives rise to functional heterooligomeric channels (Chen et al., 1993). We therefore assigned the translational initiation site of the short β subunit from testis to the corresponding triplet in hrCNC2a (see second boxed ATG in Fig. 2 a). The deduced amino acid sequence predicts a relative molecular mass of ∼74.3.

The relationship between the β subunits from rod photoreceptor (CNCβ1a) and testis (CNCβ1c–f) is illustrated in Fig. 1,a and Fig. 2. CNCβ1a consists of a GARP part and a β′ part (Körschen et al., 1995). While the β′ part is conserved, only a small COOH-terminal region of the GARP part is left in the long variants CNCβ1c–e. The GARP part and some of the NH₂-terminal region of the β′ part (Fig. 1,a) are missing in CNCβ1f. The different NH₂-terminal ends of CNCβ1c–e are highlighted in Fig. 2.

The low expression level of the large transcripts in testis raises the question as to what extent CNCβ1c–e are expressed in sperm or precursor cells. Translation into the respective polypeptides was examined by Western blot analysis using two different antibodies (Fig. 3). Antibody FPc 21K was raised against that NH₂-terminal region of the long β variants that is lacking in the short β form. This antibody should only recognize the long β subunits. Antibody PPc 32K was directed against an epitope in the common COOH-terminal region, and therefore should recognize all forms of β subunits. FPc 21K proved to be considerably more sensitive to β polypeptides than PPc 32K; in membranes of rod outer segments, FPc 21K stained the 240 kD β subunit and a less abundant ∼105 kD isoform (see legend to Fig. 3) more intensely than did PPc 32K, although FPc 21K was used at lower concentrations (Fig. 3, compare ROS lanes). FPc 21K also recognized the heterologously expressed long β subunit, CNCβ1c, much better than did PPc 32K (Fig. 3, compare βℓ lanes). The heterologously expressed short testis β subunit (CNCβ1f) is not recognized by FPc 21K (Fig. 3, left, lane βs).

Western blot analysis of membrane proteins of cauda epididymal sperm and testis did not provide evidence for a polypeptide recognized by FPc 21K (Fig. 3, left, lanes S and T). This failure to detect the long β subunits may either reflect a low expression level, as suggested by the Northern blot (Fig. 1 b), or expression in a small subset of testicular cells (see below).

In testicular membranes, the common antibody PPc 32K recognized a weak band of ∼130 kD (Fig. 3, right, lane T) that was not detected by the more sensitive antibody FPc 21K. Therefore, we interpret the labeling as an unspecific cross-reaction of PPc 32K. In contrast, in membranes of cauda epididymal sperm (Fig. 3, right, lane S) and ejaculated sperm (not shown), antibody PPc 32K intensely labeled a polypeptide of ∼80 kD. The M_r is similar both to the calculated M_r and the apparent M_r of the heterologously expressed CNCβ1f (Fig. 3, right, lane βs). The occurrence of two CNCβ1f bands of ∼76 and ∼92 kD in transfected COS-1 cells may reflect heterogeneity due to posttranslational modifications. These results suggest, but do not prove, that the prominent ∼2.4 kb transcript (see Fig. 1 b) encodes the short form of the β subunit.

The expression pattern of α and β subunits was examined by immunohistochemistry on testicular sections (Fig. 4) and epididymal sections (Fig. 5) using both the α subunit– specific antibody PPc 23 and the β subunit–specific antibodies FPc 21K and PPc 32K. PPc 23 and PPc 32K stained sperm and precursor cells. The staining pattern for both antibodies varied among individual seminiferous tubules depending on the stage of spermatogenesis (Fig. 4, a and b). Antibody PPc 23 stained flagella of sperm (arrowheads) and granules of late spermatids (arrows, Fig. 4, a and c). Spermatogonia, spermatocytes, early spermatids, Sertoli cells, and intertubular cells were not labeled. The β subunit immunoreactivity obtained by PPc 32K was already detectable in spermatocytes (Fig. 4, b and d, open arrows). PPc 32K stained late spermatids (Fig. 4, b and d, arrows) and sperm flagella (Fig. 4 b, arrowhead) intensely. Spermatogonia, Sertoli cells, and intertubular cells were not labeled. These results suggest that β subunits are expressed earlier in development than are α subunits. The more sensitive antibody FPc 21K did not label testicular or epididymal sections (not shown). These results together with the Western blot analysis demonstrate that the expression level of the long β variants is in fact very low, and that the short β subunit is the sole physiologically relevant β variant in spermatogenic cells.

Both PPc 23 (anti-α; Fig. 5, a and c) and PPc 32K (anti-β; Fig. 5, b and d) intensely stained sperm inside the epididymal ducts. At a higher magnification, long threadlike structures were resolved that most likely represent flagella (Fig. 5,c). The specificity of the testicular and epididymal staining was ascertained by preincubating the primary antibody with either the respective immunogenic or with an unrelated antigen (not shown), as well as by omitting the primary antibody (Fig. 4,e and Fig. 5 e). Under these control conditions, no specific staining was detected.

The staining patterns of epididymal sperm obtained by anti-α and anti-β antibodies differed from each other (Fig. 5, c and d), suggesting that the surface distribution of α and β subunits is different. To test this hypothesis, the spatial expression pattern of subunits was studied by immunofluorescence microscopy on isolated sperm. The α subunit–specific antibody (PPc 23) stained almost the entire flagellum of both cauda epididymal sperm (data not shown) and ejaculated sperm (Fig. 6, a and c), whereas the β subunit–specific antibody (PPc 32K) stained the proximal part of the principal piece, but not the midpiece and the distal part of the flagellum (Fig. 6, b and d). Neither PPc 23 nor the PPc 32K stained the head. As expected from the Western blot analysis (Fig. 3), FPc 21K did not stain sperm, even at concentrations that were 15× higher than those of PPc 32K (not shown). The differential expression pattern of α and β subunits along the flagellum was confirmed using the PAP/ABC method (not shown). Primary antibodies preincubated with the respective immunogenic antigen gave no staining (Fig. 6, e and f), as well as the secondary antibody alone (not shown). From these results we conclude that the α subunit is distributed along the entire flagellum, whereas the β subunit is restricted to the principal piece, suggesting that homo- and heterooligomeric forms of the CNG channel coexist in mature sperm.

Owing to their substantial Ca²⁺ permeability, CNG channels are prime candidates for mediating Ca²⁺ entry into sperm. We investigated Ca²⁺ entry into sperm using confocal laser scanning microscopy (CLSM) in conjunction with the Ca²⁺ indicator Fluo-3 and novel caged cyclic nucleotides. The combination of high-resolution CLSM with fast and selective liberation of cyclic nucleotide analogues made it possible to demonstrate for the first time CNG channel–mediated Ca²⁺ influx into sperm.

Sperm loaded with Fluo-3 did not fluoresce uniformly at rest (Fig. 7 A). The Fluo-3 fluorescence was stronger in the acrosomal region and in some local areas of the midpiece where mitochondria are located. This finding suggests that the basal [Ca²⁺]_i is higher in the acrosomal vesicle and in mitochondria than in other cell compartments, consistent with the idea that these organelles serve as intracellular Ca²⁺ stores (Irvine and Aitken, 1986; Meizel and Turner, 1993).

Photolysis of both 4,5-dimethoxy-2-nitrobenzyl (DMNB) 8-Br-cGMP (Hagen et al., 1996) and DMNB 8-pCPT-cGMP by UV light evoked a Ca²⁺ influx into sperm in 88% of the cells (n = 209/237). Because 8-pCPT-derivatives of cyclic nucleotides cross membranes more readily than do 8-Br-derivatives (Butt et al., 1992), DMNB 8-pCPT-cGMP was used to investigate the Ca²⁺ influx in more detail by illuminating either the proximal or the distal region of the principal piece (see Materials and Methods and Fig. 7). In the absence of caged 8-pCPT-cGMP, no change in fluorescence was observed in response to a UV flash (Fig. 8, UV). This control experiment showed that the increase of [Ca²⁺]_i did not arise from UV-induced damage of the plasma membrane, but was due to liberation of 8-pCPT-cGMP from the caged compound. In the presence of 2 mM extracellular Ca²⁺, photoreleased 8-pCPT-cGMP evoked an increase of [Ca²⁺]_i in the acrosomal (A) and postacrosomal (PA) region of the head, in the midpiece (MP), and in the principal piece (PP) of the flagellum (Fig. 8, Ca²⁺). No increase of [Ca²⁺]_i was detected when the extracellular solution contained no Ca²⁺ and 500 μM EGTA (Fig. 8, 0 Ca²⁺). These results demonstrate that the [Ca²⁺]_i increase was caused by Ca²⁺ influx from outside rather than by a release from intracellular Ca²⁺ stores.

A characteristic property of CNG channels is their blockage by extracellular Mg²⁺ (Weyand et al., 1994; Frings et al., 1995). The 8-pCPT-cGMP–induced Ca²⁺ influx into sperm is almost entirely suppressed by 15 mM extracellular Mg²⁺ (Fig. 8, Mg²⁺). Such a [Mg²⁺] is not expected to block Ca²⁺ currents through voltage-activated Ca²⁺ channels (McDonald et al., 1994). Furthermore, sperm preincubated with high concentrations of blockers for voltage-activated Ca²⁺ channels (25 μM d-cis-diltiazem; 2.5 μM verapamil) did respond with an increase of [Ca²⁺]_i to UV light (Fig. 8, Dil, Ver). High concentration (1 μM) of staurosporine, an inhibitor of cyclic nucleotide-dependent kinases, did not suppress the 8-pCPT-cGMP–induced increase of [Ca²⁺]_i (Fig. 8, Sta). These results rule out the possibility that 8-pCPT-cGMP activates a Ca²⁺ conductance indirectly by means of a protein kinase G–mediated phosphorylation. In conclusion, we interpret these results to indicate that the Ca²⁺ influx is due to activation of CNG channels.

Although the heterologously expressed α subunit from testis is ∼200-fold more sensitive to cGMP than to cAMP, in the native channel consisting of α and β subunits, the ligand selectivity may be changed. Therefore, we investigated whether cGMP and cAMP differ in their efficacy to evoke a Ca²⁺ influx into sperm. We used two novel caged compounds—MCM 8-Br-cGMP and MCM 8-Br-cAMP— that differ only marginally in their aqueous solubilities (≤100 μM) and their photolytic quantum yields (Hagen et al., manuscript in preparation). Therefore, differences between these compounds in their efficacy to evoke a Ca²⁺ influx must be attributed to a difference in their apparent ligand affinity for the CNG channel.

Fig. 9 shows the increase of fluorescence intensity in sperm at various concentrations of either MCM 8-Br-cGMP or MCM 8-Br-cAMP. Over a large concentration range of MCM 8-Br-cAMP (10⁻⁶–10⁻⁴ M), a small progressive increase of [Ca²⁺]_i was detected after liberating 8-Br-cAMP. In contrast, photolysis of MCM 8-Br-cGMP induced much larger increases of [Ca²⁺]_i than did MCM 8-Br-cAMP at the respective concentrations. For example, at 100 μM, photolysis of MCM 8-Br-cGMP produced a six- to sevenfold increase of fluorescence intensity in the principal piece (PP), whereas photolysis of MCM 8-Br-cAMP increased the fluorescence intensity only ∼1.8-fold. Although these experiments do not allow an estimate of the apparent ligand sensitivity of the native channel, the higher efficacy of 8-Br-cGMP to increase [Ca²⁺]_i compared with 8-Br-cAMP agrees well with the high selectivity of the heterologously expressed α subunit for cGMP compared with cAMP. These findings suggest that native CNG channels represent the target of a cGMP-signaling pathway that controls Ca²⁺ entry into sperm.

We have provided evidence that CNG channels are located on the flagellum and serve as a Ca²⁺ entry pathway in sperm. The dissimilar expression of α and β subunits along the flagellum suggests that homo- and heterooligomeric channels coexist in vivo. The physiological implications of these findings are addressed in the following discussion.

A short (∼2.4 kb) and several long (∼3.3 kb) less abundant transcripts of the β subunit are expressed in testis. Western blotting and immunocytochemistry failed to detect long β subunits in sperm as well as in testicular precursor cells; therefore, these subunit species must be expressed at a rather low level if at all.

The 80-kD β subunit (CNCβ1f) expressed in sperm is probably encoded by the ∼2.4-kb transcript. Although we were unable to identify unequivocally the 5′ nontranslated region of the ∼2.4-kb transcript, both the short β clone and a similar β clone isolated from a human retinal library (Chen et al., 1993; Ardell et al., 1996) produce functional polypeptides in a cell line (Chen et al., 1993; J. Weiner and F. Müller, unpublished data). Whether the ∼2.4-kb transcript is generated by use of an alternative promotor as recently proposed for the short β transcript from human retina (Ardell et al., 1996), is not known. However, genes that are expressed in testis and other tissues often give rise to testis-specific transcripts that are generated by using testis-specific promotors, altered polyadenylation, and alternative exon splicing (Erikson, 1990).

CNCβ1f is lacking the entire GARP part and a significant NH₂-terminal portion of the β′ part of the rod β subunit (CNCβ1a). An unconventional CaM-binding site that exists in this NH₂-terminal region of CNCβ1a (see Fig. 1 a) controls ligand sensitivity (Grunwald et al., 1998; Weitz et al., 1998). The decrease of the apparent ligand sensitivity by Ca²⁺/CaM in rod photoreceptors has been proposed to represent a feedback mechanism that terminates the light response and participates in the recovery process after stimulation (for review see Molday, 1996). The NH₂-terminal CaM-binding site is absent in CNCβ1f. Therefore, potential modulation of the sperm CNG channel by Ca²⁺/CaM cannot occur through the β subunit. The sperm α subunit contains in its NH₂-terminal region a segment that is highly homologous to a conventional CaM-binding site that is also present in the α subunit of olfactory CNG channels. However, no modulation by Ca²⁺/CaM was observed for human and bovine CNCα2 channels (Yu et al., 1996; Bönigk et al., 1996; for another review see Biel et al., 1996). Gordon and coworkers (1995) provided some evidence that an unknown Ca²⁺-dependent factor, in addition to or instead of CaM, may control CNG channel activity in rod photoreceptors. An unknown factor may also control activtity of the sperm CNG channel. Expression of CNCβ1f in sperm may be physiologically important. Sperm contain sizeable amounts of CaM, and Ca²⁺/CaM-dependent activation of axonemal proteins play a crucial role in regulating motility (for review see Tash, 1990). Lack of CaM-binding sites in CNCβ1f may reflect a specific cellular adjustment to preclude a Ca²⁺/CaM-dependent regulation of CNG channel activity in sperm, and perhaps expression of a short β form ensures that the channel is specifically modulated by its cognate modulatory factor.

A significant finding of this study concerns the cGMP-specific Ca²⁺ influx into sperm through CNG channels. Numerous studies have emphasized the importance of cAMP as an internal messenger in vertebrate sperm, whereas cGMP has been considered an orphan molecule. For example, a rise in cAMP has been proposed to promote phosphorylation of flagellar proteins by protein kinase A, thereby regulating sperm motility (Tash, 1990). The cGMP levels in vertebrate sperm are low (Gray et al., 1976), and neither a membrane-bound nor a soluble guanylyl cyclase have been convincingly identified. The presence of a CNG channel that is considerably more sensitive to cGMP than to cAMP strongly argues for a physiological function of cGMP as signaling molecule in vertebrate sperm. Furthermore, exclusive localization of the CNG channel in the flagellum favors a role in the regulation of motility rather than in the control of acrosomal exocytosis. However, high concentrations of cGMP (or cAMP) may evoke an elevation of [Ca²⁺]_i that is propagated to the head. This increase of [Ca²⁺]_i could eventually induce acrosomal exocytosis.

The sperm surface is divided into distinct domains, each characterized by a specific protein inventary (Myles et al., 1981; Cowan and Myles, 1993). The segmental distribution of α and β subunits along the flagellum nicely fits the idea of a sectorial organization of the flagellum. In particular, the areas of subunit expression coincide with the segmentation of flagellum based on morphological criteria. While the midpiece and the distal part of the principal piece harbor only the α polypeptide, the proximal part of the principal piece harbors α and β polypetides. These results suggest that CNG channels coexist as homo- and heterooligomers in vivo, although we can not rigorously preclude that additional as yet unknown subunits participate in the formation of heterooligomeric complexes.

The dissimilar distribution of α and β polypeptides along the flagellum may have important physiological implications. The pore region located between transmembrane segments S5 and S6 of all α subunits comprises a negatively charged glutamate or aspartate residue that is crucially important for ionic selectivity, gating, and channel blockage by extracellular Ca²⁺(Root and MacKinnon, 1993; Eismann et al., 1994; Sesti et al., 1995). This residue is replaced by a glycine residue in the pore region of β subunits. As expected, both Ca²⁺ blockage and ionic selectivity of heterooligomeric rod channels (consisting of α and β subunits) are different from those of homooligomeric α subunits (Körschen et al., 1995). Similarly, the homooligomeric α subunit cloned from testis (CNCα2) and heterooligomeric channels (CNCα2 + CNCβ1c) differ markedly in their blockage by extracellular Ca²⁺ (J. Weiner and R. Seifert, unpublished results). It is plausible that the β subunit also modifies Ca²⁺ permeability, although that has not yet been experimentally demonstrated. If so, a regional expression of CNG channel subtypes with different Ca²⁺ permeability is expected to create Ca²⁺ microdomains. Tuning [Ca²⁺]_i along the flagellum may provide a molecular basis for regulating sperm motility. In fact, the behavioral response of sperm induced by chemoattractive factors depends on external Ca²⁺, as has been shown for a variety of invertebrates (for review see Cosson, 1990). Moreover, oscillating changes in [Ca²⁺]_i occur in hamster sperm during hyperactive motility (Suarez et al., 1993).

This study provides the first evidence that motility of sperm may be regulated by Ca²⁺ entry through CNG channels. It will be crucially important for future work to examine whether uneven distribution of CNG channel subtypes in fact gives rise to a spatiotemporal pattern of flagellar [Ca²⁺]. Moreover, it needs to be shown that this pattern provides the molecular basis for modulating sperm swimming behavior.

We thank Drs. A. Baumann, W. Bönigk, J.E. Brown, E. Eismann, S. Frings, and H.G. Körschen for critical reading of the manuscript, H.G. Körschen for the gift of antibodies, and S. Gotzes for testing antibodies on retinal sections. We thank M. Bruns, B. Gentsch, M. Jetten, J. Lossmann, and B. Oczko for technical assistance, and A. Eckert for preparing the manuscript. We express our gratitude to Dr. K.-H. Wrobel (Regensburg) and Dr. M. Davidoff (Hamburg) for helpful comments on the immunohistochemistry of bovine testis, and we acknowledge a gift of ejaculated bull sperm by Dr. A. Görlach (Kleve).

This work was supported by grants from the Deutsche Forschungsgemeinschaft to I. Weyand and V. Hagen and the Ministerium für Wissenschaft und Forschung des Landes Nordrhein-Westfalen to U.B. Kaupp.

aa

amino acid residue

CNC

cyclic nucleotide-gated channel

CNG

cyclic nucleotide-gated

DMNB

4,5-dimethoxy-2-nitrobenzyl

GARP

glutamic acid-rich protein

MCM

(7-methoxy-coumarin-4-yl)methyl