Bioelectric patterning during oogenesis: stage-specific distribution of membrane potentials, intracellular pH and ion-transport mechanisms in Drosophila ovarian follicles

Bioelectric phenomena have been found to exert influence on various developmental and regenerative processes. Little is known about their possible functions and the cellular mechanisms by which they might act during Drosophila oogenesis. In developing follicles, characteristic extracellular current patterns and membrane-potential changes in oocyte and nurse cells have been observed that partly depend on the exchange of protons, potassium ions and sodium ions. These bioelectric properties have been supposed to be related to various processes during oogenesis, e. g. pH-regulation, osmoregulation, cell communication, cell migration, cell proliferation, cell death, vitellogenesis and follicle growth. Analysing in detail the spatial distribution and activity of the relevant ion-transport mechanisms is expected to elucidate the roles that bioelectric phenomena play during oogenesis.

To obtain an overview of bioelectric patterning along the longitudinal and transversal axes of the developing follicle, the spatial distributions of membrane potentials (Vmem), intracellular pH (pHi) and various membrane-channel proteins were studied systematically using fluorescent indicators, fluorescent inhibitors and antisera. During mid-vitellogenic stages 9 to 10B, characteristic, stage-specific Vmem-patterns in the follicle-cell epithelium as well as anteroposterior pHi-gradients in follicle cells and nurse cells were observed. Corresponding distribution patterns of proton pumps (V-ATPases), voltage-dependent L-type Ca2+-channels, amiloride-sensitive Na+-channels and Na+,H+-exchangers (NHE) and gap-junction proteins (innexin 3) were detected. In particular, six morphologically distinguishable follicle-cell types are characterized on the bioelectric level by differences concerning Vmem and pHi as well as specific compositions of ion channels and carriers. Striking similarities between Vmem-patterns and activity patterns of voltage-dependent Ca2+-channels were found, suggesting a mechanism for transducing bioelectric signals into cellular responses. Moreover, gradients of electrical potential and pH were observed within single cells.

Our data suggest that spatial patterning of Vmem, pHi and specific membrane-channel proteins results in bioelectric signals that are supposed to play important roles during oogenesis, e. g. by influencing spatial coordinates, regulating migration processes or modifying the cytoskeletal organization. Characteristic stage-specific changes of bioelectric activity in specialized cell types are correlated with various developmental processes.

Keywords: Drosophila melanogaster, Oogenesis, Bioelectricity, Cell communication, Pattern formation, Membrane potential, Ion channel, Ion pump, Gap junction, Live-cell imaging

Oogenesis of Drosophila melanogaster provides an excellent model system for studying various aspects of cell and developmental biology, including cell migration, organization of the cytoskeleton, cell and tissue polarization, intra- and intercellular transport processes, signal transduction and determination of spatial coordinates. In several developing and regenerating systems, such cellular processes have been found to be regulated by bioelectric signals like membrane-potential gradients, ionic current patterns or domains of differing intracellular pH [1-4]. In Drosophila ovarian follicles, stage-specific extracellular current patterns [5,6] as well as membrane-potential changes in germ-line cells have been desribed that partly depend on the exchange of protons, potassium ions and sodium ions [7-9]. However, the possible functions of these bioelectric phenomena as well as the cellular mechanisms by which they might act have largely remained elusive.

Drosophila follicles consist of 16 germ-line cells - 15 nurse cells (NC) and one oocyte (Ooc) - that are surrounded by a layer of somatic follicle cells (FC). The germ-line cells form a cytoplasmic continuum, since they are connected via intercellular bridges (ring canals) as well as via gap junctions [10]. The same holds true for the somatic cells [11], whereas germ-line and somatic cells are interconnected only via gap junctions [12]. Besides these communication paths, a variety of intercellular signaling mechanisms play a role during follicle development, since the formation of a mature egg, surrounded by a functional eggshell and containing information for embryonic axes, requires a wealth of specifications as well as patterning and symmetry-breaking steps [13].

According to morphological criteria, Drosophila oogenesis has been divided into 14 stages (S1 to S14). In the FC epithelium, 6 types of cells can be distinguished on the morphological as well as on the molecular level during mid-vitellogenic stages 9 to 10B [14-16] (Figure 1). The terminal FC (tFC) are a population of about 200 cells at the posterior follicle pole. The mainbody FC (mFC) form a broad band anterior to the tFC. The centripetal FC (cFC) reside at the anterior end of the Ooc and migrate in between NC and Ooc during S10B. The stretched FC (sFC) become flattened and cover the NC. A pair of polar cells (PC) resides at the posterior pole of the Ooc. The border cells (BC), a group of 6–10 anterior FC, migrate between the NC from the anterior follicle pole to the anterior pole of the Ooc during S9.

In order to investigate bioelectric phenomena in more detail, we compared membrane potentials (Vmem) and intracellular pH-values (pHi) of different cell types using a potentiometric probe and a fluorescent pH-indicator. In addition, we localized various membrane-channel proteins, namely vacuolar-type H+-ATPases (V-ATPases), voltage-dependent L-type Ca2+-channels, amiloride-sensitive Na+-channels and Na+,H+-exchangers as well as the gap-junction protein innexin 3.

The spatial distribution of V-ATPases was analysed immunohistochemically in order to detect interrelations between V-ATPase activity and the pHi- and Vmem-patterns, especially in different FC types. V-ATPases are a family of proton pumps composed of multiple subunits [17,18], that are required for the acidification of intracellular compartments [19,20] and are located in the plasma membranes of many epithelia. Active proton translocation by V-ATPases is electrogenic and has influences on intra- and extracellular pH. The proton gradient is used to drive secondary active antiport via different mechanisms, for example Na+,H+-exchangers (NHE), Na+,K+,Cl−-cotransporters and K+,H+-exchangers [21-23]. For ovarian follicles of several insects, including Drosophila, a role of V-ATPases in energizing the plasma membranes has been proposed [24-26]. Moreover, V-ATPases have been found to influence various developmental processes, like cell division, cell migration and differentiation [27,28]. For example, in vertebrate embryos, the activity of V-ATPases and its resulting pHi- and Vmem-patterns play a role during the formation of left-right asymmetries [29].

One possible mechanism for the transduction of bioelectric signals, like Vmem-alterations, into cellular responses is Ca2+-flux through voltage-dependent channels [30]. We, therefore, analysed the spatial distribution of L-type Ca2+-channels using a specific fluorescent inhibitor as well as immunohistochemistry. Members of this channel family are sensitive to Vmem-changes and are characterized through a pharmacological sensitivity to 1,4-dihydropyridines and phenylalkylamines mediated by the α1-subunit of the protein [31,32]. L-type Ca2+-channels appear in three functionally distinct states: resting, or closed, channels, activated, or open, channels, and inactivated channels. In nerve cells, depolarization results in a conformational change from the resting to the activated state and then to rapid inactivation, while repolarization is necessary to return to the resting state [32]. The binding affinity of phenylalkylamines, like verapamil, depends on the channel’s state in the following order: inactivated > > activated > resting [33,34]. Voltage-dependent L-type Ca2+-channels were originally considered to be unique to excitable cells [35], but now they are known to be present in non-excitable cells as well. For example, they play a role in Ca2+-reabsorption in mammalian renal epithelia [36] and in epithelial fluid transport in Drosophila Malpighian tubules [37].

The spatial distribution of amiloride-sensitive NHE and Na+-channels was investigated using fluorescent amiloride. NHE are important for pHi-homeostasis, cell-volume regulation and transepithelial Na+-transport [38]. In insect epithelia, the electroneutral exchange of Na+ against H+ through NHE is driven by the H+-gradient generated by V-ATPases [21,39]. Amiloride-sensitive Na+-channels have been found to play a role during reproductive as well as developmental processes. For example, they help to block polyspermy in Xenopus oocytes [40] and, in Drosophila, the amiloride-sensitive Na+-channel dGNaC1 is expressed in gonads and early embryos. This channel has been proposed to be involved in cytoplasmic transport processes and in water uptake during final maturation of the Ooc [41].

Members of the innexin family are the main gap-junction proteins in invertebrates [42,43], although other proteins, like pannexins [44] and ductins [45-48], have also been observed in gap junctions. In the Drosophila ovary, the mRNAs of innexins 1, 2, 3, 4 and 7 were detected [49] and innexins 1 to 4 have been shown immunohistochemically to be involved in the formation of different types of gap junctions [50]. Since innexins 1, 2 and 4 did not show conspicuous spatial distribution patterns, we only present the distribution of innexin 3 in more detail.

Using fluorescent indicators, inhibitors and antisera, we found corresponding asymmetries in the distribution patterns along the longitudinal and transversal axes of the follicle during S9 to S10B. Characteristic stage-specific changes observed in different cell types are correlated with various cellular and developmental processes.

Drosophila melanogaster wild-type Oregon R flies were reared at about 20°C on standard food with additional fresh yeast. Individual 2–3 days old females were killed by crushing the thorax with tweezers without previous etherization or chilling. The ovaries were dissected with tweezers, and single follicles of S9 to S10B were isolated by pulling at the anterior tip of an ovariole. For immunohistochemistry, dissection was carried out in Drosophila PBS [51], while for in-vitro experiments with fluorescent inhibitors and indicators R-14 medium [51] was used, which ensures optimal culture conditions during live-cell imaging of follicles [52].

To localize V-ATPases, we used two antisera: (1) a rabbit antiserum (Anti-ductin; AB5496, Chemicon International, USA) raised against a highly conserved region of the 16 kDa-protein ductin, which forms subunit c of V-ATPases and is also part of gap junctions [48], and (2), as a control, a guinea-pig antiserum (Anti-V-ATPase) raised against an N-terminal region of subunit a of the Manduca sexta V-ATPase, kindly provided by B. Walz and O. Baumann (Potsdam, Germany). For the localization of innexin 3, we used a guinea-pig antiserum (Anti-Inx3 [50]) raised against the C-terminus of innexin 3 from Drosophila, kindly provided by R. Bauer and M. Hoch (Bonn, Germany). In control experiments, L-type Ca2+-channels were localized using a rabbit antiserum (Anti-Cavα1; Anti-Cavpanα1 subunit, Almone Labs, Israel) raised against an intracellular epitope of subunit α1.

Follicles were fixed for 30 minutes at 4°C in 4% formaldehyde dissolved in PBS, washed in PBS and blocked for 1 hour at 20°C with 2% BSA/0,1% Triton X-100 in PBS. Thereafter, the follicles were incubated overnight at 4°C in PBS containing 0,5% BSA/0,1% Triton X-100 and the respective antiserum (Anti-ductin diluted 1:100, Anti-V-ATPase diluted 1:1000, Anti-Inx3 diluted 1:20, Anti-Cavα1 diluted 1:100, controls without antiserum).

After washing 6 times for 10 minutes, the follicles were either treated with goat-anti-rabbit-Cy3 (Jackson, USA; diluted 1:2000) or with donkey-anti-guinea-pig-FP488 (FluoProbes/Interchim, France; diluted 1:100) for 1 hour in PBS containing 0,5% BSA/0,1% Triton X-100. Washing was repeated 6 times and the nuclei were stained with 0,2 μg/ml DAPI (Sigma, Germany) in PBS for 3 minutes. Thereafter, the follicles were mounted in Fluoromount G (Interchim) and viewed, by using ×20 or ×40 objectives and the appropriate filter sets, either on a Zeiss Axiovert 200 wide-field fluorescence microscope (WFM), equipped with a Hamamatsu Orca ER camera, or on a Zeiss AxioImager.M2 structured-illumination microscope (SIM), equipped with a Zeiss ApoTome and a Zeiss AxioCamMRm camera.

Staining of living follicles with fluorescent inhibitors was carried out in R-14 medium. Stock solutions were prepared in DMSO, and the follicles were incubated either in 2 μM amiloride-FL (BODIPY-FL amiloride; Molecular Probes/Thermo Fisher Scientific, USA) or in 2 μM verapamil-FL (BODIPY-FL verapamil, hydrochloride; Molecular Probes) for 15 minutes. Control follicles were preincubated before labeling with the unlabeled inhibitors (Sigma, Germany; stock solutions in ethanol) for 15 minutes using 10 μM amiloride or 100 μM verapamil-HCl (or they were labeled with Anti-Cavα1). Thereafter, the follicles were mounted in R-14 medium and viewed immediately as described above.

In order to analyse Vmem-patterns, we used the fluorescent potentiometric probe DiBAC (bis-(1,3-dibutylbarbituric acid) trimethine oxonol, DiBAC4(3); Molecular Probes). The anionic dye enters cells and binds to intracellular membranes and proteins in a Vmem-dependent manner: depolarization leads to an accumulation and to an increase in fluorescence intensity. Relative fluorescence differences between cells of comparable size were stated, i. e. stronger fluorescence: more depolarized vs. weaker fluorescence: more hyperpolarized. Living follicles were incubated for 15 minutes in R-14 medium containing 1–3 μM DiBAC (dissolved in 70% ethanol). Thereafter, they were mounted in R-14 medium and viewed immediately as described above.

For the analysis of pHi-patterns, we used the fluorescent pH-indicator CFDA (5-carboxyfluorescein diacetate, acetoxymethyl ester, 5-CFDA, AM; Molecular Probes). The anionic dye enters cells and reports pHi by fluorescence-intensity differences, since protonation leads to fluorescence loss. Relative fluorescence differences between cells of comparable size were stated, i.e. stronger fluorescence: more alkaline vs. weaker fluorescence: more acidic. Living follicles were incubated for 15 minutes in R-14 medium containing 4 μM CFDA (dissolved in DMSO), and then mounted and viewed as described above.

To facilitate interpretation and comparability of the patterns obtained with different staining methods, we analysed median optical sections through the follicles. Representative grey-scale images were transferred into pseudocolour images using ImageJ Fire-LUT (NIH, USA). For immunostaining as well as fluorescent inhibitors, brighter colours represent higher concentrations of the respective membrane-channel protein. For DiBAC, brighter colours refer to more depolarized Vmem, and for CFDA, brighter colours refer to more alkaline pHi. Each experiment was repeated at least three times.

Using the potentiometric probe DiBAC, we revealed stage specific Vmem-patterns during S9 to S10B by comparing fluorescence intensities in different follicle regions. Stronger intensities refer to more depolarized Vmem, weaker intensities to more hyperpolarized Vmem. In S9 (Figure 2A) we found a complex Vmem-pattern in the FC epithelium: mainbody FC (mFC) are characterized by a hyperpolarized Vmem in relation to the neighboring terminal FC (tFC, posterior), centripetal FC (cFC, anterior) and stretched FC (sFC, further anterior). In each mFC, an apicobasal gradient was observed, the apical region being more hyperpolarized than the basal region. The strongest depolarization was found in migrating border cells (BC) and in posterior polar cells (PC). In the oocyte (Ooc), an anteroposterior gradient was observed, the anterior region being more hyperpolarized than the posterior region. During S10B (and S11, data not shown) we found a transversal Vmem-pattern, where one side of the FC epithelium (including tFC, mFC and cFC) was depolarized in relation to the opposite side. In most follicles, the depolarized side could be identified as the ventral side (25 vs. 3 follicles) according to the location of the Ooc nucleus (DIC-microscopy). S10A showed characteristics of both S9 and S10B: mFC are hyperpolarized in relation to cFC and tFC, but one side of the FC epithelium is already more depolarized than the other (Figure 2B and C).

Using the pH-indicator CFDA, we analysed pHi-patterns in the developing follicle. A more acidic pHi is indicated by weaker, a more alkaline pHi by stronger fluorescence intensity. During S9 to S10B, we found stage-specific pHi-patterns in the NC and in the FC (Figure 3). In the NC cluster, remarkable anteroposterior and dorsoventral gradients emerged, where each NC is characterized through its own pHi: the NC adjacent to the anterodorsal region of the Ooc (location of the nucleus) has the most acidic pHi, and the anterior-most NC has the most alkaline pHi. Both gradients develop in S9 and become fully established during S10. In the FC epithelium, we found greater variability: most examined follicles showed an anteroposterior pHi-gradient, the sFC, cFC and mFC being most acidic and the tFC being most alkaline. Some follicles, especially in S9, also showed a transversal pHi-gradient in the FC epithelium, with one side being more acidic than the other (data not shown). In addition, in S9 the migrating BC are characterized through acidic pHi, and the posterior region of the Ooc is more alkaline than the anterior region (Figure 3A).

By indirect immunofluorescence microscopy using two different antisera (Anti-ductin and Anti-V-ATPase) we investigated the localization of V-ATPases during mid-vitellogenic stages. Since the 16 kDa-protein ductin is known to be part of both V-ATPases and gap junctions [48], we used Anti-V-ATPase as a control to decide, whether the observed distribution of ductin results from the distribution of V-ATPases or gap junctions. Both antisera recognized cytoplasmic as well as membranous antigens. Membrane labeling with Anti-ductin was either punctate or continuous (Figure 4), while with Anti-V-ATPase it was always continuous (data not shown). Punctate membrane labeling is presumed to originate from ductin as part of gap junctions, whereas continuous membrane labeling is presumed to originate from ductin as part of V-ATPases [26]. We found V-ATPases in lateral and in apical FC membranes, in NC membranes and in the oolemma, while in the FC epithelium, characteristic asymmetries in the spatial distribution were observed with both antisera (for Anti-V-ATPase, data not shown). Accordingly, V-ATPases (not ductin-containing gap junctions, which show uniform punctate distribution) are enriched in FC on one side of the follicle, including tFC, mFC, cFC and sFC, whereas FC on the opposite side contain considerably less. In most follicles, the enriched side could be identified as the ventral side (13 vs. 1 follicles) according to the location of the Ooc nucleus (DIC-microscopy). Also the membranes of PC showed accumulation of V-ATPases (Figure 4). Control follicles showed no specific staining (data not shown).

Using the fluorescent inihibitor verapamil-FL, we analysed the distribution of L-type Ca2+-channels. Since binding of the inhibitor depends on the channel’s conformational state, we used immunohistochemistry as a control to discriminate between distribution and activity patterns. While Anti-Cavα1 revealed uniform distributions of L-type Ca2+-channels along the longitudinal and transversal follicle axes during mid-vitellogenic stages (data not shown), verapamil-FL showed asymmetric staining patterns (Figure 5), that are supposed to represent differences in the conformational state of these voltage-dependent channels. Verapamil binds with higher affinity to channels being exposed to depolarized Vmem, which means that they are activated, or open (and inactivated). The distribution patterns of verapamil-FL-labeled channels were very similar to the Vmem-patterns described above (Figure 2). Regions having more depolarized Vmem contained higher concentrations of supposed activated (and inactivated) channels. In S9 and S10A, these regions are the tFC, cFC and sFC as well as the BC and PC, while in the mFC the concentration of supposed resting, or closed, channels was higher.

In each mFC, verapamil-FL revealed an apicobasal gradient of activated (and inactivated) Ca2+-channels, with higher concentration on the depolarized basal side (Figure 5A and B). In S9 we found a higher concentration of such channels also in the posterior region of the Ooc, which is depolarized compared to the anterior region (Figure 2A). In S10B (Figure 5C) an asymmetric distribution along the transversal follicle axis was observed where one side contained a higher concentration of activated (and inactivated) Ca2+-channels than the other side. In control follicles preincubated with unlabeled verapamil the gradual staining disappeared (data not shown).

Using immunohistochemistry, we have previousley analysed in ovarian follicles the localisation of the gap-junction proteins innexin 1 to 4 [50]. In the present study, we show that innexin 3 has a non-uniform distribution along the longitudinal follicle axis between S9 and S10B. As described earlier, Anti-Inx3 recognizes antigens in NC membranes and, especially, in lateral FC membranes, showing either continuous or punctate labeling [50]. Interestingly, innexin 3 is present in considerably higher amounts in the lateral membranes of cFC than of the other FC (Figure 6). Thus, cFC are especially interconnected via innexin-3-containing gap junctions (in addition to other innexin- and ductin-containing gap junctions [46,50]). Control follicles showed no specific staining (data not shown).

To investigate the distributions of Na+-channels and Na+,H+-exchangers (NHE), we used the fluorescent inhibitor amiloride-FL. Since the inhibitor binds to both ion-transport mechanisms with high affinity, they cannot be discriminated with this method. Amiloride-FL mainly recognized these ion transporters in a non-uniform pattern in the NC cytoplasm (Figure 7), and, due to this intense cytoplasmic labeling, it was difficult to detect membrane labeling. During S9 to S10B, most follicles showed an anteroposterior gradient of NHE and Na+-channels, with the highest concentration in the anterior-most NC. Control follicles preincubated with unlabeled amiloride showed no specific staining (data not shown).

The activities of ion-transport mechanisms regulate intracellular ion composition, pHi and Vmem. Coordinated interactions of different ion channels, ion pumps and gap junctions result in bioelectric phenomena, like pH-gradients, voltage gradients and ion fluxes within single cells or tissues. It is known that bioelectricity can act as regulator in various developmental and regenerative processes, e. g. proliferation, differentiation, polarisation and migration [1,3].

The composition and activity of ion-transport mechanisms may vary in space and time leading to stage-specific changes in current flow and voltage gradients. Therefore, we analysed, during the course of Drosophila oogenesis, the spatial distributions of different membrane channels, Vmem and pHi. Especially during mid-vitellogenesis (S9 to S10B), we found characteristic stage-specific distributions that are supposed to provide bioelectric signals for epithelial patterning and morphogenesis.