Cx37 and Cx43 Localize to Zona Pellucida in Mouse Ovarian Follicles

Abstract

In the ovarian follicle, granulosa cells adjacent to the oocyte extend processes through the zona pellucida matrix, and these projections establish gap junctions both with the oocyte and with neighboring transzonal projections. The identity of connexins contributing to gap junctions between transzonal projections has not been extensively studied. Here, we examined the expression pattern of Cx37 and Cx43 in mouse zona pellucida using multiple connexin-specific antibodies. Immunofluorescence staining revealed abundant Cx37 and Cx43 puncta within the zona pellucida of both preantral and antral follicles. Cx37 persisted in the zona pellucida of mature follicles up to 5 h after an ovulatory stimulus whereas Cx43 was reduced in the zona pellucida by 3 h after an ovulatory stimulus. We suggest that in addition to its role in oocyte-granulosa cell communication, Cx37 could enable a distinct communication pathway between those granulosa cells that are in direct contact with the oocyte.

Keywords:

• connexin
• Cx37
• Cx43
• gap junction
• ovary
• zona pellucida

INTRODUCTION

Ovarian follicular development and function depends on direct signaling mediated by gap junctions (GJs) in addition to endocrine and paracrine pathways (Eppig 1991; Gilchrist et al. 2004). GJs are clusters of intercellular channels formed by the docking of hemichannels (connexons) present in the membranes of adjacent cells (Evans and Martin 2002; Saez et al. 2003). By linking the cytoplasms of connected cells, GJ channels allow direct intercellular transfer of ions, small metabolites and signaling molecules (<1 kDa) (Bruzzone et al., 1996). Within the ovarian follicle, GJs couple granulosa cells and also provide a direct communication pathway between the oocyte and the surrounding granulosa cells (Kidder and Mhawi 2002). This latter pathway is mediated by oocyte-granulosa cell GJs that occur where threadlike processes, extended by granulosa cells adjacent to the oocyte, contact the oolemma (Amsterdam et al. 1976; Anderson and Albertini 1976; Gilula et al. 1978). To make these junctions, the granulosa cell projections traverse the thick glycoprotein-rich zona pellucida (ZP) layer surrounding the oocyte (Sotelo and Porter 1959; Odor 1960; Bjorkman 1962; Zamboni 1974).

GJ-mediated communication is thought to be crucial for the ovarian follicle in a number of ways (Kidder and Mhawi 2002). First, a gap-junctional network facilitates the transfer of amino acids, nucleotides, and glucose metabolites throughout the follicle and into the oocyte (Moor et al. 1980). This syncytial network for metabolic cooperation is crucial because blood vessels are restricted to the periphery of follicles, making the interior of large follicles essentially avascular. Second, junctional coupling is important not only for oocyte growth and maturation but also for controlling the subsequent meiotic status of the oocyte, as at least some signals that regulate meiotic maturation in oocytes are thought to pass through oocyte-granulosa GJs (Dekel and Beers 1978; Dekel et al. 1981; Fagbohun and Downs 1991; Downs 1995). In the periovulatory period, gonadotropins substantially disrupt GJ-mediated communication between granulosa cells, and the decreased coupling is correlated with the resumption of meiosis by the oocyte and the initiation of ovulation (Dekel et al. 1981; Granot and Dekel 2002; Dekel 2005). Third, GJs between granulosa cells could facilitate the spread of signaling molecules throughout the follicle following hormone reception (Lawrence et al. 1978). In addition, there is evidence that GJ-mediated communication is required for optimal paracrine signaling within the follicle (Gittens et al. 2005). Finally, intercellular coupling could play a role during apoptosis in atretic follicles (Mayerhofer and Garfield 1995; Krysko et al. 2004).

GJ channels are oligomeric structures made up of connexin (Cx) subunits. In the human and mouse genomes, 20 and 21 connexin genes have been identified, respectively (Sohl and Willecke 2004). At least five members of the Cx family are localized to the rodent ovarian follicle: Cx26, Cx32, Cx37, Cx43, and Cx45 (Kidder and Mhawi 2002). Cx26 and Cx32 are detected in theca cells that lie outside the basal lamina encompassing the follicle (Wright et al. 2001). Cx32 has also been detected in cumulus granulosa cells and in the oocyte (Valdimarsson et al. 1993). The most abundant GJ protein in granulosa cells is Cx43 and its expression is elevated as follicle size increases in response to FSH (Beyer et al. 1989; Valdimarsson et al. 1993; Mayerhofer and Garfield 1995; Okuma et al. 1996). In addition to granulosa cells, Cx43 has been noted in oocytes in rat follicles (Granot et al. 2002). C57BL/6 mice lacking Cx43 exhibit follicular developmental arrest at the unilaminar primary follicle stage, indicating a failure of granulosa cells to proliferate (Ackert et al. 2001). Moreover, antisense knockdown of Cx43 expression in bovine oocyte-cumulus cell complexes has demonstrated a role for Cx43 in meiotic maturation of bovine oocytes (Vozzi et al. 2001). Cx45 has also been identified as a component of GJs between granulosa cells, where it colocalizes with Cx43 in some plaques (Okuma et al. 1996; Alcolea et al. 1999; Kidder and Mhawi 2002). Heterocellular GJs between oocytes and granulosa cells contain Cx37, and there is evidence that Cx37 is expressed both by the oocyte and by the adjacent granulosa cells (referred to as cumulus cells in antral follicles) (Simon et al. 1997; Veitch et al. 2004). It is thought that cumulus cells differentially target Cx37 to the ends of processes (transzonal projections) that cross the ZP to form homotypic GJ channels with the oocyte (Veitch et al. 2004). Ablation of Cx37 in mice results in the loss of oocyte-granulosa cell dye-transfer and the failure of follicles to form antral follicles. In addition, oocytes in the follicles of these mice are small and fail to reach meiotic competence, and follicles as a whole undergo premature luteinization (Simon et al. 1997; Carabatsos et al. 2000). These studies indicate that at least two Cxs, Cx43 and Cx37, are necessary for normal folliculogenesis.

In addition to the presence of large GJs between granulosa cell bodies, very small GJs between neighboring cumulus cell transzonal projections have also been described. Gilula et al. first noted by electron microscopy the presence of numerous GJs between the thin cumulus cell processes within the ZP (Gilula et al. 1978). These results were later supported by the findings of Larsen et al., who additionally suggested that cumulus cell processes may self interact through multiple junctional connections (Larsen et al. 1987). Until recently, however, the Cx composition of these transzonal projection GJs has not been closely examined. Identifying the contributing Cxs is an important issue because spatially restricted expression of a Cx to the transzonal projection GJs might allow for a selective communication compartment that could coordinate the specialized functions of cumulus cells in direct contact with the oocyte. In this study, we have used newly generated antibodies against Cx37 to reexamine the expression of this Cx in the ovarian follicle. Cx37 was considered a likely candidate to be localized to GJs between transzonal projections, because prior studies indicate that it is similarly targeted to the ends of transzonal projections that terminate at the oocyte surface. In support of this hypothesis, Cx37 was found to be extensively localized to the ZP region of the follicle in addition to its previously described localization at the oocyte surface. Moreover, we found that Cx43 is also present in the ZP. To address the possibility that Cx37 and Cx43 in cumulus cells might be differentially regulated during the periovulatory period, we examined Cx37 and Cx43 expression in mature follicles at different time points following intraperitoneal injection of an ovulatory stimulus. While this work was in progress, Teilmann reported on the localization of Cx37 and Cx43 in the ZP of mouse ovary (Teilmann 2005). We discuss significant differences in the immunolocalization results obtained in the two studies and suggest a possible explanation for some of the differences.

MATERIALS AND METHODS

GST-Cx Fusion Proteins

GST-Cx37 and GST-Cx40 plasmids were provided by David Paul (Harvard Medical School) and contain coding sequence for amino acids 229–333 and 231–331 of rat Cx37 and rat Cx40, respectively, fused to the glutathione-S-transferase (GST) coding sequence (Gabriels and Paul 1998). GST-Cx43 plasmid was the gift of Alan Lau (University of Hawaii) and contains sequence encoding amino acids 236–382 of rat Cx43 fused to GST (Loo et al. 1995). GST-Cx40 and GST-Cx43 were purified from lysates of induced BL21 bacteria (Stratagene, La Jolla, CA) using Glutathione Sepharose (Amersham Pharmacia Biotech, Pistcataway, NJ). GST-Cx37 was isolated from bacterial lysates by an inclusion body preparation using CelLytic B reagent (Sigma, St. Louis, MO). The protein was further purified by preparative SDS-PAGE and copper sulfate negative staining (Lee et al. 1987). Electroeluted protein was dialyzed in phosphate buffered saline (PBS) and concentrated to ∼1 mg/ml with 10,000 MWCO centrifugal filter devices (Millipore, Bedford, MA).

Cx37 Antibody Production

Gel-purified GST-Cx37 was injected into two rabbits (18263 and 18264) at Pocono Rabbit Farm and Laboratory, Inc. (Canadensis, PA) according to the company's standard protocol for fusion protein antigens. Crude sera were affinity purified using a two-step protocol. First, GST-reactive antibodies were removed by passing the sera three times over a column containing GST coupled to CNBr-activated Sepharose 4B. Second, the partially purified serum was passed three times over a column containing GST-Cx37. The column was washed extensively with PBS and bound antibody was eluted with 100 mM glycine, pH 2.5. The eluted fractions were neutralized by addition of 1M Tris, pH 9.5, and pooled fractions were concentrated with 30,000 MWCO Millipore centrifugal filters to ∼1.5 mg/ml. During the concentration step, the buffer was exchanged for PBS.

Animals

Cx37^−/− mice (C57BL/6 strain background) were genotyped as previously described (Simon et al. 1997; Simon and McWhorter 2002). Wild-type mice were also C57BL/6 background unless otherwise noted. 129S1/SvImJ, BALB/cByJ, and F1(C57BL/ 6Jx CBA/JM) strains were obtained from The Jackson Laboratory (Bar Harbor, ME). ICR (CD-1) strain mice were purchased from Harlan (Indianapolis, IN). Rats were Harlan Sprague Dawley strain.

Immunohistochemistry

Ovaries were fixed by perfusion through the left ventricle with either 2% or 4% paraformaldehyde in PBS. Aortas were fixed with 1–2% paraformaldehyde. Excised ovaries were further fixed by immersion for 1 h on ice then washed with PBS and tris-buffered saline. Fixed specimens were soaked overnight in 0.5 M sucrose, 1 mM EDTA and then frozen in Tissue-Tek OCT medium. Then, 8 μm cryosections were collected on Superfrost Plus slides (VWR) and stored at −75°C. Sections were post-fixed in acetone at −20°C for 5 min. After drying, sections were covered with PBS for 10 min and then blocked with PBS containing 4% fish skin gelatin, 1% normal goat serum (or donkey serum), and 0.1% triton X-100 for 30–60 min. 18264 antibody was diluted 1/600 or 1/1200 in blocking buffer and incubated on the sections for 1–2 h at room temperature (RT). Commercial Cx37 antibodies, directed against short peptide sequences from the cytoplasmic tail region of mouse Cx37, were diluted as follows: Alpha Diagnostic International (ADI, San Antonio, TX) Cx37A11-A antibody (1/250); Zymed C-term Cx37 antibody (Invitrogen, Carlsbad, CA) (1/200). Cx43 serum obtained from David Paul (Harvard Medical School) is directed against amino acids 252–271 of rat Cx43 (Beyer et al. 1989). The serum was affinity purified and used at a dilution of 1/300. Cx43 antibody (C6219) obtained from Sigma Chemical Co. (St. Louis, MO) is directed against amino acids 363–382 of rat Cx43 and was used at a dilution of 1/1000. After primary antibody, slides were washed in PBS containing 0.1% triton X-100 and then incubated with the secondary antibody (Rhodamine RedX-conjugated F(ab′)₂ fragments of donkey anti-rabbit IgG, Jackson ImmunoResearch Labs, West Grove, PA) for 30 min at RT at a dilution of 1/200. (Note: CY3-conjugated secondary antibodies from several sources produced a significant punctate background signal in oocytes on frozen sections, presumably because the CY3 fluorophore has affinity for an oocyte component). After washing the sections, coverslips were mounted with Mowiol 40–88 medium containing 1,4-diazobicyclo-(2,2,2)-octane and fluorescence viewed with an Olympus BX51 microscope and 20x UPlanFl, 40x UPlanFl, or 60x PlanApo objectives. Images were captured with a Sensys 1401 CCD camera (Photometrics) and ImagePro Plus 4.5 software (Media Cybernetics).

Blocking experiments to test antibody staining specificity were done by preincubating the primary antibody (1 μl of antibody in 50 μl of PBS) with an excess of fusion protein or peptide for 1 h at 37°C and then overnight at 4°C. 10 μg of GST-Cx37 or GST-Cx43 or 10 μg of the ADI blocking peptide (Cx37A11-P) was added to the diluted antibody. As a control, antibodies were preincubated with the same volume of PBS or with 15 μg of GST. Tubes were centrifuged for 15 min at 4°C and blocking buffer was added to bring the antibody to its final dilution for immunostaining.

Antigen Retrieval

Paraffin sections of ovary fixed by perfusion and immersion with 4% paraformaldehyde were deparaffinized and the rehydrated sections were gently boiled in either 1 mM EDTA, pH 8.0 or 10 mM citrate, pH 3.0 for 10 min. Antigen retrieval with citrate buffer was pH-dependent, as antigen recovery was poor with 10 mM citrate, pH 6.0. The slides were cooled for 20 min at RT and then washed in PBS. Sections were blocked as described for frozen sections and 18264 antibody was diluted 1/600. The secondary antibody was the same as with frozen sections, or alternatively, CY3-conjugated Donkey anti-rabbit IgG (Jackson ImmunoResearch). Unlike with frozen sections, CY3-conjugated secondary antibody alone did not produce a punctate background in oocytes of paraffin sections subjected to heat-induced antigen retrieval. In a few experiments, antigen retrieval was achieved by digesting sections with 0.01% trypsin at 37°C for 4 min, without a boiling step, and Rhodamine RedX-conjugated secondary antibody was used.

Western Blotting

Membrane fractions of mouse aortic endothelium were prepared as previously described (Simon and McWhorter 2003). The equivalent of ∼2.5 mouse aortas were loaded on each lane. Western blotting was performed with nitrocellulose and 18264 antibody was used a dilution of 1/5000. Horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (Pierce, Rockford, IL) was diluted 1/80,000. SuperSignal West Dura substrate (Pierce) was applied for chemiluminescence detection and images captured with a Kodak Imagestation 2000 system.

Hormone Injections

Four to eight week-old female mice were injected intraperitoneally in the morning with 5 units of pregnant mare serum gonadotropin (PMSG) (G4877, Sigma), which acts similarly to endogenous follicle stimulating hormone (FSH). Forty-eight h later, the mice were injected with 5 units of human chorionic gonadotropin (hCG) (CG10, Sigma), which serves as an ovulatory stimulus similar to endogenous luteinizing hormone (LH). Ovaries were collected 3 h, 5 h, or 8 h after hCG injection and fixed by immersion in 4% paraformaldehyde in PBS for 1 h at 4°C. Ovulated ovum-cumulus complexes were collected by removing and fixing the ampulla region of the oviduct 12.5 h after hCG injection. Specimens were then processed for cryosectioning and immunostaining.

RESULTS

Antibodies against Cx37 were generated in rabbits by immunizing with a GST-Cx37 fusion protein containing the last 105 amino acids of rat Cx37 (cytoplasmic tail region) (Gabriels and Paul 1998). Crude sera was affinity purified and one of the antibodies, referred to as 18264 antibody, was used in the majority of subsequent experiments. The specificity of the Cx37 antibodies was initially tested by western blotting and immunostaining aortic specimens from wild-type and Cx37^−/− mice (Figure 1) (Simon et al. 1997). The 18264 antibody recognized a 37 kD band from wild-type aortic endothelial membrane preparations but not from Cx37^−/− samples (Figure 1A). In addition, the antibody reacted with GST-Cx37 on western blots but did not react with GST-Cx40, GST-Cx43, or GST alone (Figure 1B). Immunofluorescence staining showed that the 18264 antibody strongly labeled vascular endothelium and weakly labeled vascular smooth muscle cells of mouse aorta, as previously observed with other Cx37 antibodies (Figure 1C) (Simon and McWhorter 2003). Vascular staining was absent from immunostained sections of Cx37^−/− aorta (Figure 1C). Since vascular smooth muscle cells also express Cx43, the absence of smooth muscle cell staining in Cx37^−/− sections indicates that 18264 antibody does not crossreact with Cx43 in tissue sections. Furthermore, 18264 antibody did not label Purkinje fibers in the bundle of His portion of the cardiac conduction system, where Purkinje fibers are known to express both Cx45 and Cx40 (not shown) (Coppen et al. 1998). These results indicate that 18264 antibody specifically reacts with Cx37 and does not cross-react with three other Cxs tested, including two (Cx43 and Cx45) that are expressed in the ovarian follicle.

Figure 1 Characterization of a new Cx37 antibody (18264) raised against a GST-Cx37 fusion protein. (A) On western blots of aortic endothelial membranes, affinity purified 18264 antibody detected a 37 kD protein from wild-type (WT) but not Cx37^−/− aorta (Ao). The positive control lane contained GST-Cx37 fusion protein. Size markers are indicated in kD (B) 18264 reacted with GST-Cx37 fusion protein but not GST, GST-Cx43, or GST-Cx40. For each protein, three amounts were loaded in adjacent lanes (30 ng, 10 ng, and 5 ng from left to right). (C) 18264 antibody strongly labeled vascular endothelium and weakly labeled vascular smooth muscle cells in frozen sections of wild-type mouse aorta, but not in sections of Cx37^−/− aorta.

The 18264 antibody was used to examine the expression pattern of Cx37 in C57BL/6 strain mouse ovarian follicles. In particular, we wanted to test the hypothesis that Cx37 is targeted not only to oocyte-granulosa cell GJs, as previously reported, but is also localized to GJs that are formed between cumulus cell processes that traverse the ZP (Gilula et al. 1978). Frozen sections of mouse ovary were immunolabeled with 18264 antibody and fluorescently labeled secondary antibodies. Close examination of the labeled sections revealed the presence of numerous fine fluorescent puncta within the ZP, as well as the expected signal at the oocyte cell surface and ooplasm (Figure 2A Figure 2B Figure 2C). The distribution of puncta was relatively uniform throughout the ZP. Labeling was not observed in the cell bodies of granulosa cells that immediately contacted the ZP nor was there labeling of other cells of the cumulus oophorous or of mural granulosa cells. Labeling of the ZP and oocyte was not observed with the secondary antibody alone (Figure 2D Figure 2E Figure 2F) or when 18264 preimmune serum was substituted for 18264 immune serum (Figure 2G Figure 2H Figure 2I). Vascular endothelium of blood vessels in the ovary sections was strongly labeled by 18264 antibody, providing an internal positive control for Cx37 detection (Figure 6B).

Figure 2 18264 Cx37 antibody labels numerous small puncta within the zona pellucida of mouse ovarian follicles. (A–C) Frozen sections of mouse ovary were immunostained with 18264 antibody, resulting in punctate labeling in the ZP (z) in addition to labeling of the oocyte (o). The ZP is the glycoprotein-rich matrix between the oocyte and the cumulus cells (cc). Labeling was absent when sections were incubated with secondary antibody only (D–E) and when 18264 preimmune serum was substituted for immune serum (G–I). A, D, G are phase contrast images. B, C, E, F, H, I are epifluorescence images. White boxes outline regions that are shown at higher magnification in C, F, I. Scales bars represent 20 μm in A, B, D, E, G, H and 10 μm in C, F, I.

Because our results differed significantly from previous reports, we performed a number of additional experiments to confirm that the abundant 18264 signal within the ZP was the result of specific Cx37 labeling (Simon et al. 1997; Wright et al. 2001; Veitch et al. 2004; Teilmann 2005). First, we compared the immunofluorescence patterns obtained with sections of wild-type and Cx37^−/− follicles (Figure 3A Figure 3B Figure 3C Figure 3D Figure 3D Figure 3F). In follicles lacking Cx37, there was no detectable signal within the ZP (Figure 3D Figure 3E Figure 3F). Within the ooplasm of Cx37^−/− oocytes, there typically remained some large spots of fluorescence of varying intensity, but the normal signal was otherwise absent, including the fine puncta usually present at or near the oocyte surface. Second, we tested the ability of GST-Cx37 to block 18264 immunostaining competitively (Figure 3G Figure 3H Figure 3I). Preincubating 18264 antibody with an excess of GST-Cx37 eliminated labeling of both ZP and oocyte (Figure 3G Figure 3H Figure 3I), whereas labeling of these regions was not blocked when the antibody was preincubated with GST alone (not shown). Third, we compared the immunostaining patterns of 18264 antibody with two commercially available antibodies raised against short peptide sequences from the cytoplasmic tail region of Cx37 (ADI and Zymed antibodies) (Figure 4A Figure 4B Figure 4C Figure 4G Figure 4H Figure 4I). Both peptide antibodies gave a similar immunolabeling pattern, i.e., numerous small spots of fluorescence in the ZP as well as oocyte labeling. Staining with the ADI Cx37 antibody was eliminated when the antibody was preincubated with its cognate blocking peptide (Figure 4D Figure 4E Figure 4F), but the peptide did not block labeling generated with the Zymed antibody and also did not completely block labeling with 18264 antibody (not shown). These results indicate that the ADI and Zymed antibodies recognize different peptide sequences in the cytoplasmic tail region of Cx37. Thus, multiple antibodies recognizing different peptide sequences of Cx37 all resulted in abundant punctate labeling in the ZP, leaving little doubt that the substantial signal observed in the ZP with the Cx37 antibodies is the result of specific Cx37 labeling.

Figure 3 The punctate labeling pattern in the zona pellucida produced with 18264 Cx37 antibody is due to the presence of Cx37. 18264 antibody labeling of frozen sections of wild-type (WT) follicles (A–C) results in punctate labeling in the ZP, but this labeling is absent in Cx37^−/− follicles (D–F). Oocyte labeling was also absent in Cx37^−/− follicles except for some large internal spots of variable intensity. (G–I) Labeling with the 18264 antibody was blocked by preincubating the antibody with GST-Cx37 fusion protein but staining was not blocked by GST alone (not shown). White boxes outline regions that are shown at higher magnification in C, F, I. Phase contrast images are shown in A, D, G. Epifluorescence images are shown in B, C, E, F, H, I. Scales bars represent 20 μm in A, B, D, E, G, H and 10 μm in C, F, I.

Figure 4 Two additional Cx37 antibodies directed against different short peptide sequences from the cytoplasmic tail of Cx37 also result in punctate labeling in the zona pellucida. (A–C) Frozen sections of mouse ovary immunostained with a Cx37 antibody from Alpha Diagnostic International (ADI) exhibited punctate labeling in the ZP. (D–F) Labeling with the ADI Cx37 antibody was eliminated by preincubating the antibody with the blocking peptide. (G–I) A Cx37 antibody from Zymed Laboratories, Inc. also labeled the ZP in a punctate manner. White boxes outline regions that are shown at higher magnification in C, F, I. Phase contrast images are shown in A, D, G. Epifluorescence images are shown in B, C, E, F, H, I. Scales bars represent 20 μm in A, B, D, E, G, H and 10 μm in C, F, I.

We next considered the possibility that the expression pattern of Cx37 in mouse ovarian follicles is mouse strain-specific, which might explain discrepancies in immunostaining results. Sections of ovaries collected from two other inbred mouse strains (129/Sv and BALB/c), an outbred strain (ICR), and a F1 hybrid strain (C57BL/6 × CBA) were immunostained with 18264 antibody. Similar to the C57BL/6 results described above, abundant punctate staining was observed in the ZP as well as in the oocyte with these four other mouse strains (not shown). In addition, we also observed labeling in the ZP when sections of rat ovary were immunostained. Thus, targeting of Cx37 to the ZP does not appear to be mouse strain-specific or restricted only to mice.

Differences in sample preparation could account for varying abilities to detect Cx37 in the ZP. In the above experiments, we used frozen sections of ovary, whereas a recent study used paraffin sections treated for antigen retrieval, with differing results (Teilmann 2005). To address this issue, 18264 immunostaining was performed with paraffin sections of ovary and compared to previous results obtained with frozen sections (Figure 5). Paraffin sections were subjected to antigen retrieval by boiling slides in either 1 mM EDTA (pH 6) or 10 mM citrate (pH 3) for 10 min. The immunofluorescence staining pattern using either of these methods differed substantially from the frozen section results. First, there was a much more consistent ring-like concentration of fluorescence at or near the surface of the oocyte compared with frozen sections and relatively less signal in the ooplasm (Figure 5A Figure 5B). Second, in most cases it was very difficult to determine if a Cx37 signal was in the ZP. A substantial compression of the ZP occurred following paraffin embedding and antigen retrieval, making it very difficult to distinguish if a peripheral fluorescent signal was on the surface of the oocyte or in the compressed ZP. In some follicles with less compression, DIC microscopy revealed that the ZP was largely removed or damaged by the procedure. In a few follicles where the ZP was slightly less compressed and still intact, we detected a distinct Cx37 signal in the ZP, but the signal intensity was typically weaker than with frozen sections (Figure 5C, Figure 5D). Similar results were obtained using the Zymed and ADI Cx37 antibodies or when a digestion step with 0.01% trypsin was substituted for heat-induced antigen retrieval (not shown). With all three antibodies, Cx37 detection in paraffin sections required an antigen retrieval step.

Figure 5 Detection of Cx37 in the zona pellucida is more difficult in paraffin sections treated for antigen retrieval than in untreated frozen sections. A paraffin section of mouse ovary was boiled in 10 mM EDTA pH 8 to effect antigen retrieval and then immunostained with 18264 antibody. (A, B) A more consistent ring-like concentration of staining at or near the surface of the oocyte is observed compared with immunostained frozen sections. It is difficult to determine if there is staining in the ZP because the zona is compressed following paraffin embedding and antigen retrieval. There is relatively less staining in the ooplasm compared with frozen sections. (C, D) Higher magnification images of the same follicle shown in A and B. Some Cx37 immunostaining can be discerned in the ZP of this particular follicle (arrow), but the signal was weaker than in frozen sections. In other follicles, ZP Cx37 staining was very difficult to detect. White boxes outline the region shown at higher magnification in C, D. Phase contrast images are shown in A, C. Epifluorescence images are shown in B, D. Scales bars represent 10 μm in A, B and 5 μm in C, D.

Cx37 expression was examined in frozen sections of ovarian follicles at different developmental stages (Figure 6). In small primary follicles, punctate staining was observed at the interface between oocyte and granulosa cells, but it was not possible to determine if any of the signal was in ZP because the glycoprotein matrix is just beginning to be synthesized at this stage (Figure 6A Figure 6B Figure 6C). In multilaminar secondary follicles (preantral follicles), a prominent punctate signal was observed in the ZP (Figure 6D Figure 6E Figure 6F). These follicles also tended to have a more concentrated signal at or near the surface of the oocyte. Cx37 localization to the ZP was maintained in antral follicles (Figure 6G Figure 6H Figure 6I).

Figure 6 Cx37 is localized to the zona pellucida in preantral and antral follicles. Frozen sections of unilaminar primary follicles (A–C), multilaminar secondary (preantral) follicles (D–F), and antral follicles (G–I) were immunostained with 18264 antibody. In the primary follicles, Cx37 was present at the interface between the oocyte and granulosa cells, but the ZP was not yet evident. The very bright staining in panel B is due to blood vessel labeling. Punctate Cx37 labeling was detected in the ZP of preantral and antral follicles. White boxes outline regions that are shown at higher magnification in C, F, I. Phase contrast images are shown in A, D, G. Epifluorescence images are shown in B, C, E, F, H, I. Scales bars represent 10 μm in A, B, F, I; 2 μm in C; and 50 μm in D, E, G, H.

We also wanted to determine the effect that an ovulatory stimulus has on Cx37 localization at the ZP. Cx43 expression is known to be prominently down-regulated in granulosa cells in the hours shortly before ovulation, a process that has been suggested to be important for meiotic resumption of the oocyte (Dekel et al. 1981; Larsen et al. 1986; Granot and Dekel 2002). We hypothesized that Cx37-containing GJs between transzonal projections could allow for continued coupling between those cumulus cells that are in direct contact with the oocyte during the periovulatory period, when Cx43 is globally down-regulated throughout the follicle. To test this idea, mice were injected with PMSG to stimulate follicle development and 48 h later injected with hCG as an ovulatory stimulus. Ovaries were collected 3 h, 5 h, or 8 h after the hCG injection and processed for cryosectioning. In addition, ovulated ovum-cumulus cell complexes were collected 12.5 h after hCG injection by harvesting and sectioning the ampulla region of the oviduct. At 3 h and 5 h after hCG injection, Cx37 was still detected in the ZP of mature antral follicles (Figure 7A Figure 7B Figure 7C Figure 7D Figure 7E Figure 7F). By 8 h after hCG injection, however, cumulus expansion was evident and there was reduced Cx37 signal in the zona material (Figure 7G–I). In addition, there was less localization of Cx37 at the oocyte surface and a marked increase in large aggregates of Cx37 in the ooplasm. At 12.5 h after hCG injection, ovulated ovum-cumulus cell complexes exhibited very little detectable Cx37 in the remaining ZP and there was a large amount of internally located Cx37 in the ovum (not shown). Thus, Cx37 localization to the ZP is maintained at least up to 5 h after an ovulatory stimulus, but by 8 h, this localization is significantly disrupted.

Figure 7 Cx37 localization to the zona pellucida of mature follicles is maintained up to 5 h after an ovulatory stimulus. Frozen sections of mature follicles from mice injected with PMSG and hCG were immunostained with 18264 antibody. (A–C) Cx37 labeling pattern 3 h after hCG injection. (D–F) 5 h after hCG injection. (G–I) 8 h after hCG injection. Punctate labeling in the ZP was maintained at 3 h and 5 h after hCG injection. By 8 h after hCG injection, cumulus expansion was evident, Cx37 in the ZP was greatly reduced, and there was considerable ooplasmic Cx37 staining. White boxes outline regions that are shown at higher magnification in C, F, I. Phase contrast images are shown in A, D, G. Epifluorescence images are shown in B, C, E, F, H, I. Inset images in A, D, G show low power views of the immunostained follicles to illustrate that they were mature antral follicles. Scales bars represent 20 μm in A, B, D, E, G, H and 10 μm in C, F, I.

Follicle sections were also immunostained with an affinity-purified Cx43 antibody to test if granulosa cells adjacent to the oocyte also target Cx43 to the ZP (Figure 8). Extremely intense Cx43 labeling was observed between granulosa cell bodies, as expected from previous studies (Figure 8A Figure 8B Figure 8C). The high intensity of the fluorescent signal in the granulosa cell bodies, especially in larger follicles, tended to obscure possible Cx43 signal in the ZP. With close examination, however, we detected a significant punctate signal throughout the radius of the zona, although this signal was less distinct than observed for Cx37. In addition to the zona signal, there was typically a concentrated area of Cx43 staining at or near the side of granulosa cells that contacted the ZP. Preincubating the Cx43 antibody with an excess of GST-Cx43 blocked the ZP staining as well as cell body labeling, suggesting that the zona staining was specific (Figure 8D Figure 8E Figure 8F). A similar ZP signal was obtained using a second Cx43 antibody (Sigma) directed against a different Cx43 peptide sequence (not shown).

Figure 8 Cx43 is also detected in the zona pellucida of mouse follicles. (A–C) Frozen sections were immunostained with an affinity purified Cx43 antibody (DP, David Paul, Harvard Medical School). In addition to intense labeling in the cell bodies of granulosa cells, faint but significant punctate labeling was detected in the ZP. (D–F) Labeling with the DP Cx43 antibody was blocked by preincubating the antibody with GST-Cx43 fusion protein. White boxes outline regions that are shown at higher magnification in C, F. Phase contrast images are shown in A, D. Epifluorescence images are shown in B, C, E, F. Scales bars represent 20 μm in A, B, D, E and 10 μm in C, F.

Cx43 immunostaining was also examined in sections from the hormone-injected mice described earlier (Figure 9). At 3 h, 5 h, and 8 h after hCG injection, Cx43 was still detected in cumulus cell bodies but a significant amount of the signal appeared to be internalized. Within the ZP, there was a reduction in the Cx43 signal, even at the 3 h time point, compared with noninjected control (Figure 9C Figure 9F). These results suggest that, following an ovulatory stimulus, Cx37 expression at the ZP is maintained for at least a couple hours longer than that of Cx43.

Figure 9 Cx43 labeling in the zona pellucida is decreased by 3 h after an ovulatory stimulus. Frozen sections of mature follicles from mice injected with PMSG and hCG were immunostained with DP Cx43 antibody. (A–C) Cx43 labeling pattern 3 h after hCG injection. (D–F) 5 h after hCG injection. Cx43 labeling in the ZP was reduced at the 3 h and 5 h time points compared with noninjected controls. Cx43 staining was still present in the cumulus cell bodies, but there was an increase in internalized signal. White boxes outline regions that are shown at higher magnification in C, F. Phase contrast images are shown in A, D. Epifluorescence images are shown in B, C, E, F. Inset images in A, D are low power views of the immunostained follicles to illustrate that they were mature antral follicles. Scales bars represent 20 μm in A, B, D, E and 10 μm in C, F.

DISCUSSION

In this study, we examined the expression pattern of Cx37 in the mouse ovarian follicle, using a newly generated antibody specific for Cx37. Our results indicate that Cx37 is not only located at oocyte-granulosa cell GJs, as previously reported, but is also targeted to the ZP region of preantral as well as antral follicles. Several lines of evidence argue strongly that the punctate staining we observed in the ZP with 18264 antibody is specific Cx37 labeling, including the absence of staining in Cx37^−/− follicles and the fact that staining was blocked by preincubating the antibody with GST-Cx37. In addition, multiple antibodies against different Cx37 epitopes all gave the same result. We suggest that Cx37 is localized to small GJs known to be present between granulosa cell transzonal projections (Gilula et al. 1978). Immunoelectron microscopy studies using Cx37 antibodies will be required to definitively conclude that Cx37 is targeted to the transzonal projection GJs. In addition, we also detected a consistent Cx43 signal in the ZP.

Cx37 localization to the ZP to the extent we describe has not been previously reported. In a recent study of 4-week-old mouse ovary, a consistent presence of small Cx37 puncta in the ZP was noted in paraffin sections treated for antigen retrieval, but the number of Cx37 puncta in the zona was a minor portion of the follicle staining (Teilmann 2005). It was suggested that the scarce Cx37 signal within the ZP represented connexons being transported to sites of oocyte contact (Teilmann 2005). Other studies have not reported the presence of Cx37 in the ZP (Simon et al. 1997; Wright et al. 2001; Veitch et al. 2004). It is likely that sensitivity issues and sample preparation differences can explain why Cx37 has not previously been reported in the ZP to the extent that we now describe. The 18264 antibody we have generated, as well as the ADI and Zymed antibodies, provide a sufficiently strong signal to detect easily Cx37 in the ZP. In addition, we found that ZP-targeted Cx37 is much more readily detected in frozen sections than in paraffin sections. In our hands, paraffin embedding and antigen retrieval resulted in a substantial compression of the ZP that made determinations of Cx37 localization to this region problematic. Even in those paraffin-embedded follicles where Cx37 could be detected in the ZP, the signal was weaker than in frozen sections. We suggest that antigen retrieval for Cx37 may not be very efficient in the glycoprotein-rich ZP. The efficiency of antigen retrieval depends, in part, on the degree of cross-linking that occurs during fixation and the local microenvironment can affect cross-linking as well as other antigen masking processes.

In addition to ZP localization, we detected Cx37 on the surface of oocytes and in the ooplasm of both preantral and antral follicles, consistent with our earlier study (Simon et al. 1997). These data are in agreement with those of Veitch et al. and Teilmann, but stand in contrast to the results of Wright et al., who detected Cx37 on oocytes of preantral follicles but not antral follicles (Wright et al. 2001; Veitch et al. 2004; Teilmann 2005). Our results also do not support the findings of Wright et al. that Cx37 is widely distributed in GJs between granulosa cell bodies of antral follicles (Wright et al. 2001).

The uniform distribution of Cx37 puncta throughout the radius of the ZP and the absence of detectable Cx37 in granulosa cell bodies suggests that Cx37 in the ZP is located at GJs between transzonal cumulus cell projections rather than in oocyte-granulosa cell GJs or in transport vesicles. Oocyte microvilli do not generally extend more than halfway across the ZP, at most, so it is unlikely that the puncta we observed resulted from labeling of GJs between granulosa cell projections and oocyte microvilli (Odor 1960). Moreover, granulosa cell processes preferentially form GJs with the oocyte at the oolemma rather than at the tips of microvilli (Anderson and Albertini 1976; Gilula et al. 1978). It also seems unlikely that the ZP Cx37 signal resides in transport vesicles en route to oocyte-cumulus cell GJs at the oocyte surface. Given the high density of fluorescent puncta observed in the ZP, we would have expected to see a detectable vesicular or Golgi Cx37 signal within the cell bodies of the granulosa cells that extend these transzonal projections if there was such a high level of continuous transport occurring in the transzonal processes.

Our results, as well as previous studies, suggest that there is differential targeting of Cxs in granulosa cells adjacent to the oocyte (see model in Figure 10). These cells appear to specifically target Cx37 to their transzonal processes to make GJs with the oocyte and presumably to make junctions with neighboring transzonal processes, while excluding Cx37 from GJs between the cumulus cell bodies (Veitch et al. 2004). On the other hand, in the same cells Cx43 is targeted both to transzonal projections and to the GJs between cumulus cell bodies. It will be important to determine if specific amino acid sequence differences between the Cx37 and Cx43 proteins are crucial for differential targeting of these proteins to distinct subcellular locations in cumulus cells.

Figure 10 Model of Cx37 and Cx43 localization in cumulus granulosa cells of mouse ovarian follicles. Note that other Cxs expressed in the follicle have been excluded from the drawing for simplicity. Cx37 (white rectangles) is localized to transzonal projections extended by cumulus cells, but is excluded from GJs between cumulus cell bodies. Cx37 is localized both to GJs between the transzonal projections in the ZP and to oocyte-cumulus cell GJs, but is excluded from GJs between cumulus cell bodies. Transzonal projections may also originate from cumulus cells outside of the corona radiata cells (not shown). Cx43 (black rectangles) is targeted to GJs between cumulus cell bodies as well as GJs between transzonal projections. Cx37 and Cx43 could be present in the same GJ plaques or different GJ plaques within the ZP. The model suggests that Cx37 localization to GJs between transzonal projections could provide for a distinct communication pathway between cumulus cells that are in direct contact with the oocyte.

The high density of Cx37-containing plaques we observed in the ZP by immunofluorescence is consistent with previous data from electron microscopy studies, although quantitative data is lacking. Gilula et al. described these GJs as being frequently present between adjacent cumulus cell processes (Gilula et al. 1978). Larsen et al. similarly noted the presence of a gap junctional network interconnecting cumulus cell processes within the ZP Larsen et al. 1987). In the latter study, it was estimated that there were about 900 cumulus cell projections within each ZP of a rat follicle. The transzonal projections originate largely from the corona radiata cells directly abutting the ZP, but can also arise from cumulus cells slightly more removed from the zona (Zamboni 1974). Each transzonal projection could potentially form multiple junctional contacts with projections from neighboring cells, or even interact with itself through multiple junctional connections (Larsen et al. 1987). Finally, the small size of the fluorescent puncta we noted is consistent with the very small size of the GJs observed in the ZP by electron microscopy (Gilula et al. 1978; Larsen et al. 1987).

Cx43 staining in the ZP was very faint compared with the intense Cx43 signal that is found between cumulus cell bodies, which may explain why its presence in the zona has been overlooked until recently. Our results are somewhat different from those of Teilmann, who recently also detected Cx43 in the ZP, but mainly in the outer third of this region (Teilmann 2005). Whether or not Cx37 and Cx43 colocalize in the ZP is not known at present. If so, Cx37 and Cx43 could potentially form heteromeric channels with distinct functional properties (Beyer et al. 2001).

Cumulus granulosa cells and mural granulosa cells have distinct functional phenotypes, including differences in hyaluronic acid production in response to FSH, steroid synthesis, growth factor and steroid receptor expression, gonadotropin receptor levels, cell proliferation rates, and ability to interact with the oocyte (Eppig et al. 1997; Gilchrist et al. 2004). A separate gap junctional network, formed between transzonal projections containing Cx37, may be one way for cumulus cells to differentiate themselves functionally from other granulosa cells of the follicle (see model in Figure 10).

Cx37 and Cx43 expressed in cumulus granulosa cells could also be differentially regulated and this might be important during follicular development or during the periovulatory period. Down-regulation of Cx43 by LH results in a dramatic loss of cumulus cell GJs in the preovulatory period (Gilula et al. 1978; Larsen et al. 1986; Granot and Dekel 1994; Granot and Dekel 1997; Granot and Dekel 2002; Kalma et al. 2003). Larsen et al. demonstrated that the net area of GJ membrane between rat cumulus cell bodies is reduced by 15-fold within 2–3 h after an ovulatory stimulus (Larsen et al. 1986; Larsen et al. 1987). GJs between cumulus cell transzonal projections, however, persisted for up to 5–6 h (Larsen et al. 1987). Our results indicate that Cx37 remains in the ZP at least 5 h after an ovulatory stimulus is delivered, consistent with a role for Cx37 in GJs between transzonal projections. It is possible, therefore, that Cx37 localization to transzonal GJs provides a mechanism for continued cumulus cell coupling in a short window of time when Cx43 is globally down-regulated throughout the follicle in response to an ovulatory stimulus, but before ovulation occurs.

In summary, using several antibodies and applying stringent criteria for immunostaining specificity, we detected punctate signals for Cx37 and Cx43 in the ZP of mouse follicles and suggest that these Cxs are localized to GJs between cumulus cell transzonal projections. Thus, in addition to its role in oocyte-cumulus cell communication, Cx37 could provide for a distinct communication pathway between those cumulus cells which are in direct contact with the oocyte via GJs that are spatially segregated from those occurring at cumulus cell bodies.

ACKNOWLEDGMENTS

We thank David Paul for Cx43 antisera, GST-Cx37 plasmid, and GST-Cx40 plasmid and Alan Lau for GST-Cx43 plasmid. This work was supported by a grant from the National Institutes of Health (HL64232 to A.M.S.).