Gap junctions in Malpighian tubules of Aedes aegypti

Malpighian tubules of insects initiate the formation of urine by secreting fluid into the blind-ended (distal) segment of the tubule. The distal segment of the Malpighian tubule in the yellow fever mosquito, Aedes aegypti,consists of two types of epithelial cells: principal cells and stellate cells,with a relative distribution of 5:1(Satmary and Bradley, 1984; Yu and Beyenbach, 2004). Nevertheless, principal cells make up more than 90% of the tubule mass owing to their large size (Beyenbach,2001; Wu and Beyenbach,2003; Yu and Beyenbach,2004). Principal cells mediate transepithelial secretion of K⁺ and Na⁺ from the hemolymph to the tubule lumen by active transport (Beyenbach,1995; Beyenbach,2001; Beyenbach,2003b). Current–voltage plots of a single principal cell in the intact Malpighian tubule indicate rather low cell input resistances which,compared to the transepithelial resistance, suggest the electrical coupling of 5–6 principal cells (Masia et al.,2000). Moreover, voltage recordings from two or more adjacent principal cells of the same tubule are strikingly similar in amplitude and frequency under control and experimental conditions, which is consistent with electrical coupling.

There are two pathways for the electrical coupling of principal cells. One pathway leads through gap junctions; the other passes through the apical membrane of one cell into the tubule lumen and then across the apical membrane into the neighboring cell. Thus, gap junctions couple neighboring cells directly, whereas the tubule lumen couples them indirectly. Although insects do have gap junctions (Phelan,2005; Phelan et al.,1998), their presence in Malpighian tubules has not been established to this date.

In the present study we ascertain the functional presence of gap junctions between principal cells of Malpighian tubules of the yellow fever mosquito. The gap junctions permit the passage of current from one cell to the next, but they prohibit the passage of the dye Lucifer Yellow. Metabolic inhibition of the tubule, which is known to reduce intracellular ATP concentrations(Wu and Beyenbach, 2003) and to halt transepithelial electrolyte and fluid secretion(Beyenbach and Masia, 2002; Beyenbach et al., 2000b; Pannabecker et al., 1992),increases the gap-junction resistance 66-fold, consistent with gap-junction gating by ATP. Furthermore, PCR studies on cDNA derived from AedesMalpighian tubules reveal the expression of four innexin-like transcripts.

The colony of mosquitoes Aedes aegypti L. was maintained as described previously (Pannabecker et al.,1993) except for feeding larvae with TetraMin™ Tropical Flakes ground manually with mortar and piston. On the day of the experiment a female mosquito (3–7 days post-eclosion) was cold-anesthetized and decapitated. A Malpighian tubule was removed under Ringer solution from its attachment to the gut and transferred to a Lucite™ perfusion chamber containing 1.0 ml Ringer solution. The bottom of the chamber was covered with thinly stretched Parafilm`M' (American National Can, Greenwich, CT, USA), to which Malpighian tubules adhere and stabilize for the impalement of principal cells with microelectrodes. The tubules were viewed from above at ×50 magnification using a stereomicroscope (Wild, Heerbrugg, Switzerland).

Ringer solution contained the following in mmol l^–1: 150.0 NaCl, 25.0 Hepes, 3.4 KCl, 1.8 NaHCO₃, 1.0 MgCl₂, 1.7 CaCl₂ and 5.0 glucose. The pH was adjusted to 7.1 with NaOH. The osmolality of the Ringer solution was approximately 320 mosmol kg^–1 H₂O. Synthetic leucokinin-VIII was a gift from Ron Nachman (USDA, Texas A&M University).

As shown in Fig. 1, three adjacent principal cells of a Malpighian tubule were selected for impalement with conventional microelectrodes. Principal cell 1 typically was 5–10 cells away from the blind (distal) end of the tubule, and principal cells 2 and 3 were downstream towards the open-end of the tubule. Principal cell 1 was impaled with both current and voltage microelectrodes. The latter measured the basolateral membrane voltage (V_bl1) as well asΔ V_bl1 when cell 1 was voltage clamped to a hyperpolarizing voltage of 40 mV for 50 ms. Principal cells 2 and 3 were each impaled with a voltage microelectrode for the measurement of V_bl2 and V_bl3, respectively, and theirΔ V_bl when cell 1 was voltage clamped.

Stellate cells are too small for impalement with microelectrodes. Accordingly, electrical coupling could be studied only between principal cells. Moreover, we selected for study principal cells that made direct physical contact with no obvious signs of stellate cells between them. Nevertheless, we cannot entirely rule out coupling to stellate cells. As will be shown in the Results, the contributions of stellate cells to the axial coupling of principal cells can be considered minor.

Microelectrodes (Omega dot borosilicate glass capillaries, 30-30-1;Frederick Haer & Co., St Bowdoinham, ME, USA) were pulled on a programmable puller (Model P-97; Sutter Instruments, Novato, CA, USA) to yield resistances of 20–30 MΩ when filled with 3 mol l^–1 KCl. The microelectrodes were bridged to the measuring hardware using Ag/AgCl junctions that were prepared by first degreasing the silver wire with alcohol, and then by Cl-plating it in 0.1 mol l^–1 HCl for 20 min at a current of 50 μA. The bath was grounded with the Ag/AgCl junction lodged in a 4% agar bridge of Ringer solution.

The electronic hardware consisted of (1) the Gene Clamp model 500B voltage and patch clamp amplifier, (2) head stage HS-2A gain 10MGU for current injection, and (3) head stage HS-2A gain 1LU for voltage recording (all from Axon Instruments, Sunnyvale, CA, USA). Voltage deflections in principal cells 2 and 3 were recorded using custom-made high-impedance amplifiers (Burr-Brown,10¹¹ Ω). We used Clampfit (pClamp 9) for data analysis (Axon Instruments).

All current and voltage data were digitized with the aid of a computer and the A/D converter DigiData1332x (Axon Instruments). Current and voltage data from principal cell 1 were also displayed on an oscilloscope (Iwatsu, Tokyo,Japan) and on a strip chart recorder (model BD 41; Kipp and Zonen, Crown Graphic, Totnes, Devon, UK).

The transepithelial secretion of electrolytes in Malpighian tubules of Aedes aegypti can be modeled using an electrical circuit consisting of two major transepithelial transport pathways; one is active, and the other is passive (Fig. 2). Na⁺ and K⁺ must take the active transport pathway through principal cells that provide the energy for moving the two cations against their electrochemical potentials into the tubule lumen. The active transport pathway consists of the electromotive forces (E) and the membrane resistances (R) at apical (a) and basolateral (bl)membranes. The passive pathway is located outside principal cells and is represented by the single shunt resistance R_sh(Fig. 2B).

In the past we have modeled the Malpighian tubule as a simple electrical cable with only one axial resistance, i.e. the core resistance R_co as shown in Fig. 3A (Beyenbach,2003b; Beyenbach and Masia,2002; Pannabecker et al.,1992; Pannabecker et al.,1993; Scott et al.,2004; Wu and Beyenbach,2003; Yu and Beyenbach,2002). In the present study, we model the Malpighian tubule as a cable with two parallel axial resistances, the resistance of the tubule lumen(R_lu) and the gap-junction resistance(R_gj) as shown in Fig. 3B.

To obtain values of R_gj, we simplified the equivalent electrical circuit of Fig. 3Bby taking advantage of the known effects of the diuretic hormone leucokinin-VIII on the tubule. Leucokinin-VIII is known to nearly short circuit the transepithelial voltage by decreasing the paracellular shunt resistance (R_sh) from 57.8 Ωcm² to 9.9Ωcm², thereby decreasing the transepithelial voltage(V_t) from 59 mV to 6 mV(Pannabecker et al., 1993). If we assume that both R_sh and V_tcollapse to zero in the presence of leucokinin-VIII, then the `short-circuit assumption' eliminates R_sh from the circuit, thereby placing the tubule lumen at the same ground potential as the peritubular bath. The tubule lumen is thus eliminated as a current path to adjacent principal cells, leaving the gap-junction resistance (R_gj) as the only axial resistance. The value of R_gj is determined as follows.

We begin with the tubule modeled as the double cable in Fig. 4A, where gap junctions and the tubule lumen present two parallel axial resistances along the length of the tubule. The transepithelial short circuit induced by leucokinin eliminates the paracellular shunt resistance (R_sh) and places the resistances of the apical membrane (R_a) and the tubule lumen (R_lu) in parallel to the resistance of the basolateral membrane R_bl(Fig. 4B). Representing these two parallel resistances as the single non-junctional resistance(R_nj) yields a circuit consisting of R_nj and the gap-junction resistance(R_gj) alone. The two resistances R_njand R_gj are determined as follows.

Voltage-clamping cell 1 yields the input resistance(R_input) as the ratio ofΔ V_bl1/ΔI_inject (Figs 1, 4). The input resistance includes cell 1 and all other cells coupled to it, as illustrated in Fig. 4B. The input resistance can also be predicted from the values of R_gj and R_nj determined in the circuit analysis above(Eqn 11, Eqn 12). Good agreement between values of R_input measured directly and values predicted from the circuit analysis would validate the short-circuit assumption needed for reducing the double cable in Fig. 4A to a manageable level for circuit analysis. To predict values of R_input from values of R_gj and R_nj in circuit Fig. 4C, the experimental preparation illustrated in Fig. 4A is redrawn in Fig. 5 from the perspective of current injected into cell 1 flowing symmetrically into cells upstream and downstream from cell 1. The parallel symmetry of all resistances in Fig. 5 allows the reduction of all resistances to a single resistance, which is the predicted input resistance. We have limited the number of coupled cells to 7 in this data reduction because previous studies suggested that 5–6 principal cells are coupled electrically (Masia et al.,2000).

To visualize the coupling of principal cells via gap junctions,one principal cell was injected with Lucifer Yellow to observe whether the dye appears in neighboring principal cells. Microinjecting pipettes were made from borosilicate glass (TW100F-4, WPI, Sarasota, FL, USA) using a horizontal puller (P87, Sutter Instrument Co., Novato, CA, USA) and a two-step pulling protocol. Injection pipettes had an average resistance of 4.33±0.37 MΩ (N=9) when filled with 3 mol l^–1 KCl. For injections of Lucifer Yellow, the pipette was back-filled by capillary action with a 2.5 mmol l^–1 Lucifer Yellow CH dilithium salt (Sigma,St Louis, MO, USA) dissolved in water.

After isolation, the Malpighian tubule was transferred to a perfusion bath pre-coated with 0.125 mg ml^–1 poly-l-lysine(Sigma), which prevents the movement of tubules when a single principal cell is injected with dye. After impaling a principal cell, a hydrostatic pressure of approximately 5200 mmHg was applied to the pipette for 250 ms with the aid of a pneumatic picopump (PV830, WPI). The pressure pulse injected a volume of approximately 0.7 pl, as determined in pre-experiment pipette calibrations. The cytoplasmic volume of a principal cell is about 200 pl. The intracellular Lucifer Yellow was immediately visible when viewed with an inverted microscope(Diaphot, Nikon, Kawasaki, Japan) equipped with a B-3A filter (Chroma Technology, Brattleboro, VT, USA) and a mercury lamp light source (Chiu Technical Corp., Kings Park, NY, USA). Typically, we observed the tubule for more than 1 h for signs of Lucifer Yellow diffusing from the injected principal cells to the adjacent principal cells or stellate cells. To prevent photo bleaching of the dye, we turned on the light source briefly every 10 min.

After finding the evidence for gap junctions in Aedes Malpighian tubules and determining the gap-junction resistance, it was clear that the tubule should be modeled as a double cable. In previous studies we have treated the tubule as a single cable in order to determine electromotive forces (E) and the resistances (R) of transcellular and paracellular transport pathways shown in Fig. 2B(Beyenbach and Masia, 2002; Pannabecker et al., 1992; Pannabecker et al., 1993). It was therefore important to determine values of E and R for the double cable model of the tubule. In brief, we modeled the tubule as a linear series of ten principal cells where each cell was represented by the electrical circuit shown in Fig. 3B. Analysis of this tubule model with the software of Electronics Workbench® 5.12 (National Instruments, Austin, TX, USA) allows current to be injected at any position in the circuit to observe voltage deflections across any two points of the circuit. By simulating our previous in vitro microperfusion experiments, where a known current was injected into the tubule lumen at one end of the tubule(Pannabecker et al., 1992), we fitted data collected from the single cable to the double cable, to obtain new values of E and R (Fig. 3).

To prepare Malpighian tubule cDNA, 175 Malpighian tubules were isolated from 35 female mosquitoes. The tubules were immersed in ice-cold Trizol reagent (Invitrogen, Carlsbad, CA, USA) and then stored at –80°C. Total RNA from the tubules was isolated by homogenization in Trizol reagent(Invitrogen), followed by a phenol:chloroform phase separation and an isopropyl alcohol precipitation(Chomczynski and Sacchi,1987). The resulting RNA was used as a template to synthesize a pool of single-stranded cDNA using Superscript III Reverse Transcriptase(Invitrogen) and a GeneRacer oligo dT primer (Invitrogen). Before use in PCR,the cDNA was diluted tenfold with nuclease-free H₂O (Integrated DNA Technologies, Coralville, IA, USA).

The genome of Drosophila lists eight genes encoding innexins(Stebbings et al., 2002). Using the predicted amino-acid sequences encoded by these genes, we searched the genomic database of Aedes(http://aaegypti.vectorbase.org)with a Basic Local Alignment Search Tool (BLAST). The search yielded six innexin-like genes in Aedes (see Results). We designed oligonucleotide-primer pairs (Table 1) to amplify fragments of the open-reading frames (ORF) for each Aedes innexin using PCR. All of the PCRs were conducted on 0.5 μl of Malpighian tubule cDNA in Platinum PCR Supermix HF (Invitrogen) using the following cycling parameters: one cycle of 94°C for 2 min; 35 cycles of 94°C for 30 s, 50°C for 30 s, and 68°C for 1 min; one cycle of 68°C for 10 min. As a negative control, each PCR was conducted on 0.5μl of Malpighian tubule RNA that was not reverse transcribed, but was diluted to the same degree as the cDNA. To verify PCR results, we also performed all PCRs on 3 μl of a double-stranded cDNA library derived from adult Aedes Malpighian tubules (generously provided by Dr Dimitri Boudko, The Whitney Laboratory, St Augustine, FL, USA). All PCR products were separated via electrophoresis on a 1%-agarose gel and stained with ethidium bromide.

PCR products were either ligated into a pCR 4-TOPO vector (Invitrogen) or purified using a QIAquick PCR purification kit (Qiagen, Valencia, CA, USA). The ligated PCR products were transformed into TOP 10 Chemically Competent E. coli (Invitrogen) according to the manufacturer's protocol. Plasmid DNA from four resulting colonies was purified using a QIAprep Spin Miniprep kit (Qiagen). DNA sequencing of plasmid DNA and purified PCR products was performed in both the 5′ and 3′ directions by the Cornell DNA Sequencing Center (Cornell University, Ithaca, NY, USA).

Each tubule/cell was used as its own control so that the data could be analyzed for the difference between paired samples, control vsexperimental (paired Student's t-test).

We often observe spontaneous voltage oscillations in Malpighian tubules of Aedes aegypti (Beyenbach et al.,2000a; Sawyer and Beyenbach,1985; Williams and Beyenbach,1984; Yu and Beyenbach,2002). The oscillations stem from spontaneous changes in the paracellular shunt Cl^– conductance(Beyenbach et al., 2000a) and have a frequency not unlike those of spontaneous cyclical changes in cytoplasmic Ca²⁺ concentration(Fig. 6). Leucokinin-VIII increases the paracellular Cl^– conductance via the activation of a Ca²⁺ channel in the basolateral membrane of principal cells, thereby increasing the cytoplasmic Ca²⁺concentration (Yu and Beyenbach,2002). Elevated cytoplasmic Ca²⁺ levels obliterate the spontaneous changes in cytoplasmic [Ca²⁺], eliminate spontaneous voltage oscillations, and maintain an increased paracellular Cl^– conductance as long as leucokinin-VIII is present(Beyenbach, 2003a; Yu and Beyenbach, 2001).

Fig. 6 illustrates spontaneous oscillations of the basolateral membrane in three principal cells of a single Malpighian tubule. The three cells shared a base membrane voltage between –55 and –65 mV, and as one cell hyperpolarized spontaneously, the other cells followed with virtually no time delay(Fig. 6). The magnitude of the oscillations was smallest in cell 1 (Fig. 6, red trace) and largest in cell 4(Fig. 6, green trace),suggesting that the pacemaker must be located in cell 4 or beyond. The perfect temporal superimposition of voltage oscillations from different principal cells demonstrates tight electrical coupling among them, which is consistent with the presence of gap junctions.

Principal cells 1, 2 and 3 were impaled with current and voltage electrodes as shown in Figs 1 and 4. Principal cell 1 was then voltage-clamped at a hyperpolarizing voltage of 40 mV for 50 ms and the voltage deflections were recorded in cells 2 and 3(Fig. 7). After obtaining control data, the tubule was treated with 1 μmol l^–1leucokinin-VIII. In the presence of this diuretic peptide, the voltage deflections in principal cells 2 and 3 were less than those under control conditions, in view of (1) the increase in the Cl^–conductance of the paracellular shunt pathway(Yu and Beyenbach, 2001), and(2) the increase in the Ca²⁺ conductance of the basolateral membrane of principal cells (Yu and Beyenbach, 2002).

Table 2 summarizes the results of 14 experiments. Under control conditions the basolateral membrane voltage (V_bl) was –82.2 mV in principal cell 1 (and similar in cells 2 and 3, data not shown). Clamping V_bl1to a hyperpolarizing voltage 40 mV above the resting membrane voltage required the intracellular injection of 168.8 nA. Consequent to the current injection,the ΔV_bl in principal cell 2(ΔV_bl2) hyperpolarized by 20.7 mV, and theΔ V_bl in principal cell 3(ΔV_bl3) hyperpolarized by 11.9 mV(Table 2).

In the presence of leucokinin-VIII, V_bl hyperpolarized significantly from –82.2 mV to –100.4 mV(Table 2). The hyperpolarization is due to the increased coupling of V_blto the apical membrane voltage as the shunt resistance drops to 12% of control values, nearly short-circuiting the transepithelial voltage(Pannabecker et al., 1993). In the presence of leucokinin-VIII, cells 1, 2 and 3 are electrically coupled by primarily gap junctions, which allows measurement of the gap-junction resistance (R_gj, Table 2) and the non-junctional resistance (R_nj, Table 2). Importantly, R_gj was derived from the circuit analysis of voltage deflections described in Eqn 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12.

The input resistance of principal cell 1 (R_input) can be measured directly as the ratio ofΔ V_bl1/ΔI_inject. R_input was 246.5±17.8 kΩ under control conditions and fell significantly (P<0.05) to 194.8±14.3 kΩ in the presence of leucokinin-VIII in the 14 experiments summarized in Table 2. The decrease reflects in part the increased transcellular secretion of Na⁺ and K⁺ in the presence of leucokinin-VIII(Beyenbach, 2003a; Hayes et al., 1989; Pannabecker et al., 1993), and also the activation of Ca²⁺ channels in the basolateral membrane(Yu and Beyenbach, 2002).

The input resistance R_input can be predicted from values of R_gj and R_nj determined by circuit analysis (Eqn 11, Eqn 12; Fig. 5). The predicted R_input is 209.6±15.4 kΩ, which comes close to the measured R_input of 194.8 kΩ in the same tubules. Good agreement between the R_input measured directly and the R_input predicted from the circuit analysis validates the `short-circuit assumption' needed to obtain measures of R_gj.

One principal cell of the Malpighian tubule was microinjected with Lucifer Yellow at time zero (Fig. 8). At the time of injection, the fluorescence of Lucifer Yellow was confined to the injected principal cell. About 5 s after the injection, the fluorescence of Lucifer Yellow started to appear in the tubule lumen and 1 min later the dye exited the tubule lumen at its open end, indicating the secretion of Lucifer Yellow across the apical membrane of the injected principal cell.

Although the fluorescent intensity of Lucifer Yellow faded with time in the injected cell (Fig. 8), no fluorescence was observed in neighboring principal cells or in the stellate cells, not even after 60 min. We made these observations consistently in eight additional tubules.

Since it is known that intracellular calcium and metabolic inhibition can close gap junctions, we tested the effects of the calcium ionophore A23187 and dinitrophenol on the gap-junction resistance (R_gj) and the non-junctional resistance (R_nj). In these experiments we first treated the isolated Malpighian tubule with 1 μmol l^–1 leucokinin-VIII to measure R_gj and R_nj by the circuit analysis shown in Figs 1, 4 and Table 2. After 5 min in the presence of leucokinin-VIII we added 2 μmol l^–1 A23187 to the peritubular medium and took measurements of R_gj and R_nj 5 min later. We then added 0.5 mmol l^–1 dinitrophenol (DNP) to the peritubular Ringer bath (in the presence of leucokinin-VIII and A23187) and measured R_gj and R_nj 2 min later. Table 4 summarizes the results.

The addition of leucokinin-VIII to the peritubular bath significantly(P<0.05) hyperpolarized V_bl1 from –83.5 to –103.3 mV (Table 4). The circuit analysis of voltage deflections in neighboring cells yielded a mean R_gj of 468.0 kΩ and a mean R_nj of 578.5 kΩ in this series of experiments(Table 4). After 5 min in the presence of A23187, neither V_bl1 nor R_gj changed significantly. In contrast, R_nj decreased significantly which is further reflected by the significant decrease in R_input(Table 4).

Upon the addition of 0.5 mmol l^–1 DNP to the peritubular medium, V_bl1 significantly depolarized from –99 mV to –13 mV while R_gj significantly increased from 526 kΩ to 30 762 kΩ (Table 4). In the presence of DNP, R_nj significantly increased from 504 kΩ to 1483 kΩ, and R_inputsignificantly increased from 194 kΩ to 1474 kΩ(Table 4). In summary,significant effects of Ca²⁺ ionophore A23187 were limited to a small reduction in the non-junctional resistance that may reflect the pore-forming action of the ionophore at the basolateral membrane. The effect of metabolic inhibition was profound on every measured variable. It increased R_nj threefold, it dropped V_blsevenfold, increased R_input eightfold, and increased R_gj 60-fold (Table 4).

A BLAST-search for homologues of the known eight Drosophilainnexins in the genome of Aedes identified six innexin-like genes(Table 5). We refer to the innexins in Aedes after their most similar homologues in Drosophila. Among the Aedes innexins, AeInx1, AePassover, AeInx2, and AeInx3 shared over 50%amino-acid identity to their Drosophila homologues, whereas AeInx4 and AeInx7 were less than 50% identical to their Drosophila homologues (Table 5). Homologues of Inx5 and Inx6 from Drosophila were not apparent from BLAST-searches in the Aedes genome.

To better resolve the relationships between Aedes and Drosophila innexins, we constructed a neighbor-joining phylogenetic tree (Fig. 9) using the predicted amino-acid sequences encoded by the genes. For Inx1, Passover, Inx2 and Inx3, the tree demonstrates that: (1) the Aedes genes cluster into discrete branches with their respective Drosophila homologues and (2) the branch length from a node (e.g. arrow in Fig. 9) to each homologue is short (see legend of Fig. 9 for explanation). For Inx7, the tree demonstrates that the Aedes gene clusters with its Drosophila homologue, but that the length between the node of the branch and the two homologues is long. For Inx4, the tree demonstrates that the Aedes gene does not cluster discretely with its Drosophila homologue, but it appears within a larger branch that includes Inx4, Inx5 and Inx6 of Drosophila. Among these three Drosophila innexins, the cumulative branch length between AeInx4 and DrInx4 is the shortest. The insights gained in the phylogenetic tree are consistent with the results of the BLAST search(Table 5). The implications of these results are analyzed further in the Discussion.

In Aedes Malpighian tubule cDNA we detected partial PCR products near the expected sizes for AeInx1, AeInx2, AeInx3 and AeInx7 cDNAs (Fig. 10, lanes a). In contrast, we did not detect a PCR product for AePassover, and the PCR product for AeInx4 was extremely weak (Fig. 10, lanes a). DNA sequencing verified that (1) the identities of the amplified transcripts were as expected and (2) the PCR products from primers that bracketed introns were not derived from genomic DNA contamination. When total RNA was used as a template for PCR in lieu of cDNA, no products were amplified for any of the innexins (Fig. 10,lanes b). Both the positive and negative results for each gene were verified using (1) a combination of at least two additional, independent primer pairs(data not shown) and (2) cDNA from adult Aedes Malpighian tubules generated by the Boudko laboratory (data not shown). The above results indicate that Aedes Malpighian tubules primarily express mRNA transcripts for four innexins.

We have used two independent electrical approaches to measure the gap-junction resistance between principal cells of Malpighian tubules of the yellow fever mosquito. In the first method we modeled the Malpighian tubule as a double electrical cable (Fig. 3B, Fig. 4) and used circuit analysis to determine the gap-junction resistance. In the second method we used cable analysis of the voltage deflections along principal cells when one principal cell was voltage clamped. Importantly, circuit analysis and cable analysis were performed in the same tubule experiments. The circuit analysis yielded an average gap-junction resistance of 431 kΩ(Table 2), and the cable analysis yielded a gap-junction resistance of 327 kΩ(Table 3). Although the two measurements are significantly different, they both document electrical coupling between principal cells.

A short-coming of the cable analysis is the reliance on the length constant, which is very sensitive to curve-fitting the data. Accordingly, we consider the circuit analysis to deliver the more accurate measurement of the gap-junction resistance, because this analysis yields values of both gap-junctional and non-junctional resistances from which the input resistance can be predicted, i.e. 210 kΩ. The input resistance measured directly as the ratio of ΔV_bl1/I_inject, is 195 kΩ (Table 2). The good agreement between the predicted and measured values confirms the validity of the circuit analysis.

Our measurement of the gap-junction resistance by circuit analysis depends on two assumptions: the current distribution assumption and the short-circuit assumption.

The current distribution assumption states that the current injected into principal cell 1 for voltage clamping splits equally in both directions along the tubule (Figs 4, 5). Indeed, in the course of our study we have found that the decay of ΔV_bl from cell to cell was quantitatively similar in both directions (data not shown). Furthermore, the good agreement between the measured and predicted input resistances supports the symmetrical distribution of injected current, since the predicted value depends on the symmetrical circuit shown in Fig. 5, where no more than seven principal cells are needed for predicting the input resistance. In our experiments we have at least ten cells with at least five cells on either side of principal cell 1.

The short-circuit assumption states that the Malpighian tubule is analyzed at true short circuit. Specifically, we assume that leucokinin completely reduces the paracellular shunt resistance to zero, thereby short-circuiting the epithelium and leaving gap junctions as the major if not exclusive pathway for significant electrical coupling between cells. However, in reality,leucokinin reduces the paracellular shunt resistance nearly sixfold and causes the transepithelial voltage to drop from 59 mV to 6 mV(Pannabecker et al., 1993),which falls short of ideal short-circuit conditions. Thus it is relevant to evaluate the effect of the paracellular shunt resistance on the gap-junctional resistance using circuit analysis (Table 6).

The analysis shows that as the shunt resistance changes 500-fold, the gap-junction resistance changes by only 2%(Table 6). Thus, the effect of the shunt resistance on the gap-junction resistance is negligible, and our assumption, that leucokinin completely short circuits the epithelium,introduces negligible error to our gap-junction resistance measurements.

In previous studies we have modeled the Malpighian tubule as a single cable that consists of a single axial resistance, i.e. the core resistance(R_co) and a radial resistance, the transepithelial resistance (R_t) as shown in Fig. 11A. Our finding of gap junctions in the present study calls for a double cable model of the tubule with two axial resistances, the lumen resistance and the gap-junction resistance (Fig. 11B). Accordingly, the electrical parameters determined in the single cable model should be corrected for the double cable model. The new set of electrical parameters were determined by fitting previous data of voltage and resistances(Pannabecker et al., 1992) to the new equivalent circuit model that includes the gap-junction resistance.

The first revision decreases the resistance of the basolateral membrane(R_bl) from 24.1 kΩcm in the single cable to 22.0 kΩcm in the double cable, and it increases the resistance of the apical membrane (R_a) from 11.4 kΩcm to 18.0 kΩcm(Fig. 11). Since the fractional membrane resistance of the basolateral membrane(f_Rbl) is the ratio of the basolateral membrane resistance and the transcellular resistance, R_bl/(R_bl+R_a), we now find that 55% of the transcellular resistance resides at the basolateral membrane, instead of 68% as determined previously in the single cable model. It follows that 45% of the transcellular resistance resides at the apical membrane, instead of 32% as determined by the previous model.

The second revision concerns the value of the paracellular shunt resistance R_sh, which increases from 16.8 kΩcm in the single cable model to 23.2 kΩcm in the double cable model(Fig. 11). Among other revisions of the equivalent circuit are the change of the electromotive force of the basolateral membrane (E_bl) from 17.5 mV to–8.1 mV, and the increase of the electromotive force of the apical membrane (E_a) from 146.1 mV to 151.4 mV(Fig. 11). E_bl and E_a summarize all electromotive forces operating, respectively, at the basolateral and apical membrane of principal cells. E_bl includes primarily the electrochemical potentials of ions, Na⁺, K⁺,Mg²⁺, Ca²⁺, Cl^– and HCO₃^– across the basolateral membrane. Therefore,the reversal of E_bl from 17.5 mV to –8.1 mV is an improved estimate of E_bl without identifying the more dominant electrochemical potential. In contrast to the basolateral membrane, E_a derives largely from the V-type H⁺ ATPase inhabiting the apical membrane. Here the revised value of E_a (151.4 mV) suggests a stronger electromotive force of the V-type H⁺ ATPase than previously assumed.

The double cable model also resolves the discrepancy between the diameter of the tubule lumen measured optically and the electrical diameter of the core when the tubule is modeled as a single cable. Measured through a microscope,the lumen of the Aedes Malpighian tubule has a diameter between 10 and 15 μm. However, the diameter calculated from the core resistance (22.9 MΩcm^–1) of the single cable model is considerably larger, 22 μm (Fig. 11A). In the presence of dinitrophenol, which shuts down ionic traffic through apical and basolateral membranes (Wu and Beyenbach, 2003), the core resistance increases to 32.8 MΩcm^–1, reducing the core diameter to 17 μm. Thus,blocking ion transport across the apical membrane reduces the core diameter to values near the optical diameter of the tubule lumen. The double cable confirms this conclusion as follows.

The product of the single cell-to-cell gap-junction resistance (431 kΩ; Table 2) and the average number of principal cells per cm tubule (125 cells cm^–1) yields 53.9 MΩcm^–1 as the gap-junction resistance normalized to a Malpighian tubule of 1 cm length(Fig. 11B). The core resistance in the presence of dinitrophenol yields the resistance of the tubule lumen, 32.8 MΩcm^–1. Combining these two axial resistances yields a core resistance of 20.4 MΩcm^–1that approximates the core resistance of 22.9 MΩcm^–1measured in the single cable model. The good agreement of these axial resistances not only confirms the validity of the short-circuit assumption,but it also confirms the electrical diameter of the core calculated in the presence of dinitrophenol as the actual lumen diameter. Moreover, the electrical diameter represents an average diameter of the tubule lumen, which is more accurate than the optical diameter, especially in view of the elaborate luminal brush border.

The good agreement between the calculated (20.4 MΩcm^–1) and measured (22.9 MΩcm^–1) core resistances suggests that gap junctions of stellate cells contribute little to 53.9 MΩcm^–1, the gap-junction resistance of a tubule 1 cm long. Unfortunately, stellate cells are too thin and spongy for intracellular electrical recordings. The sponginess derives from deep invaginations of the basolateral membrane that leaves only 2–3 μm of cytoplasm before the microelectrode penetrates the apical membrane to arrive in the tubule lumen.

Comparison of the gap-junction resistance of 53.9 MΩcm^–1 and the lumen resistance of 32.8 MΩcm^–1 is instructive. The lumen resistance stems from the saline occupying a 1 cm length of tubule lumen, without any barrier(Fig. 11B). Since the gap-junction resistance is only 1.6 times greater than the lumen resistance,there must be a substantial number of gap-junction channels coupling one cell to the next.

The gap-junction resistance between one principal cell and the next is 431 kΩ, which is equivalent to a gap-junction conductance of 2.3 μS(Table 2). The single channel conductances of invertebrate gap junctions range from 100 pS in earthworm giant axons (Brink and Fan,1989) to 248 pS in epidermal cells of the flour beetle(Churchill and Caveney, 1993)and to 375 pS in a mosquito cell line derived from Aedes albopictus(Bukauskas and Weingart,1994). If the single channel conductance in the mosquito cell line is similar to that in Malpighian tubules of Aedes aegypti, then there must be approximately 6190 gap junctions present in a single principal cell. The estimate is not unreasonable in view of as many as 10 000, the number of gap junctions per μm² in a single gap junction plaque of the heart or liver (Unwin and Zampighi,1980).

The failure of Lucifer Yellow to move from the injected principal cell into neighboring cells (Fig. 8)calls for an explanation since these cells display substantial electrical coupling (Tables 2, 3, 4). Lucifer Yellow has a molecular mass of 457 Da. Since insect gap junctions are thought to permit the passage of hydrophilic molecules up to 1200 Da(Simpson et al., 1977), the dye would be expected to pass freely from one principal cell to another. However, permeation through gap junctions is also influenced by charge. Lucifer Yellow is a divalent anion. Hence, negative fixed charges in the gap junction pore may thwart its passage through the gap junction in Aedes Malpighian tubules (Brink and Dewey, 1980; Veenstra,1996).

In both vertebrate and invertebrate cells, reports of electrical coupling,but not dye coupling, are not uncommon. Micromeres of the starfish(Tupper and Saunders, 1972),early embryonic cells of the amphibian(Slack and Palmer, 1969),embryonic cells of the killifish (Bennett et al., 1972), cells in the developing insect epidermis(Warner and Lawrence, 1982)and cells of mouse blastocysts (Lo and Gilula, 1979), all exhibit electrical coupling but do not allow the passage of Lucifer Yellow. The permeability of gap junctions to Lucifer Yellow depends upon the proteins that compose the gap junctions. Vertebrate gap junctions composed of connexin 43 (Cx43) permit the passage of both Lucifer Yellow and current, whereas those composed of Cx45 only allow the passage of current (Martinez et al.,2002; Steinberg et al.,1994). Developmental changes involving the downregulation of Cx43 and Cx26 and the upregulation of Cx31 and Cx31.1 result in a dramatic reduction of the transfer of Lucifer Yellow between cells without affecting their electrical coupling (Brissette et al., 1994).

The permeability of gap junctions to Lucifer Yellow may also be regulated by heteromerization and/or post-translational modification of connexins. Coexpression of Cx45 with Cx43 in vertebrate HeLa cells reduces the permeability of the gap junctions to Lucifer Yellow and results in a unique electrical conductance not seen before the coexpression(Martinez et al., 2002). When gap junctions composed of Cx40 are exposed to cAMP their permeability to Lucifer Yellow is enhanced (van Rijen et al., 2000). In contrast, the permeability of gap junctions made of Cx43 to Lucifer Yellow is decreased by exposure to protein kinase C(Bao et al., 2004).

Although the above examples are from vertebrate gap junctions composed of connexins, similar functional and regulatory properties may also apply to invertebrate gap junctions made of innexins. For example, disruption of the gene encoding the innexin `passover' prevents dye coupling in the giant fiber system of Drosophila (Phelan et al., 1996). Since the transcript for `passover' is not expressed in Aedes Malpighian tubules (Fig. 10), the absence of this innexin may explain the failure of Lucifer Yellow to pass between principal cells in our study.

In addition to the above explanations, previous investigators have suggested that a failure to observe dye coupling between insect cells can be attributed to (1) the use of an unphysiological saline(Bohrmann and Haas-Assenbaum,1993), and (2) impalement damage that allows Ca²⁺ to leak into the injected cell and close dye-permeable gap junctions(Lang and Walz, 1999). The first artifact can be ruled out in the present study, because isolated Malpighian tubules of Aedes aegypti bathed in the saline that we used in the present study secrete fluid for hours(Beyenbach and Dantzler, 1990),which would not be expected with use of an unphysiological solution. The second artifact can also be ruled out, because after impaling a principal cell for the injection of Lucifer Yellow, the dye arrives first in the tubule lumen and later in the fluid exiting the open end of the tubule, which confirms that the cell was not damaged to the point of stopping secretory transport(Fig. 8).

Although it is known that cytoplasmic Ca²⁺ can close gap junctions, the Ca²⁺ concentration required to do so ranges from nanomolar to millimolar (Peracchia,2004). If the gap junctions in Aedes Malpighian tubules are closed by a rise in intracellular Ca²⁺, then a cytoplasmic Ca²⁺ concentration higher than that achieved in the presence of the Ca²⁺ ionophore A23187 and a peritubular Ca²⁺concentration of 1.7 mmol l^–1 is necessary. A23187 significantly decreased the input resistance from 215 kΩ to 194 kΩ(Table 4), which importantly stems from the decrease in the non-junctional resistance but not from the gap-junctional resistance. Accordingly, the ionophore induced changes at the level of basolateral and/or apical cell membranes of principal cells but not gap junctions. These observations indicate that gap junctions in Aedes Malpighian tubules are relatively insensitive to changes in cytoplasmic Ca²⁺ concentrations.

In contrast, the metabolic inhibition of the tubule with dinitrophenol had profound effects on all measured electrophysiological variables(Table 4). What stands out is the dramatic effect on the gap-junction resistance, which increased from 0.526 MΩ to 30.7 MΩ. The 60-fold increase reflects complete uncoupling,i.e. the complete closure of gap junctions. Supporting this conclusion is the increase of the non-junctional resistance to 1483 kΩ, which is strikingly close to the input resistance of 1474 kΩ in the presence of dinitrophenol (Table 4). The non-junctional resistance reflects largely the resistances of the basolateral and apical membranes of the principal cell. When that resistance equals the input resistance of the cell, there may be no other conductive pathways out of the cell, which documents the complete closure of gap junctions during metabolic inhibition. Thus, dinitrophenol closes gap junctions in insect Malpighian tubules. It also closes gap junctions in vertebrate tissues(Deleze and Herve, 1986).

From previous studies we know that dinitrophenol or cyanide completely inhibits transepithelial electrolyte and fluid secretion in AedesMalpighian tubules (Pannabecker et al.,1992). Specifically, dinitrophenol causes (1) intracellular ATP concentrations to drop from 0.9 mmol l^–1 to 0.08 mmol l^–1 within 2 min, (2) membrane and transepithelial voltages to decrease towards zero, and (3) the transepithelial electrical resistance and the cell input resistance to rise to maximum values as pathways for electrolytes across the basolateral and apical membranes shut down(Beyenbach and Masia, 2002; Pannabecker et al., 1992; Wu and Beyenbach, 2003). The present study shows that next to the shutdown of traffic across cell membranes, traffic through gap junctions also shuts down. Whether this shut-down reflects the regulation of gap junctions by intracellular ATP concentration remains to be determined, but ATP regulation is known for vertebrate gap junctions (Sugiura et al.,1990). The direct effect of dinitrophenol on gap junctions is unlikely because the large increase in the input resistance in the presence of dinitrophenol (Table 4) can be duplicated with cyanide, another inhibitor of ATP synthesis(Wu and Beyenbach, 2003).

The molecular basis of gap junctions in invertebrates has been a mystery for decades. Connexin proteins were first identified in the 1970s as the structural components of gap junctions in vertebrates(Goodenough, 1974). For more than 20 years, the term `connexin' became synonymous with gap junctions,whereas efforts to find the homologous connexin in invertebrates were unsuccessful (Phelan et al.,1998). It was only in 1998 that the gene family encoding invertebrate gap-junction proteins, innexins, was first confirmed in Drosophila (Phelan et al.,1998). To date, eight innexin genes have been found in the fruit fly, and 25 innexin genes in C. elegans(Phelan, 2005).

In the present study we identify six innexin-like genes in Aedes(Table 5). The neighbor-joining tree (Fig. 9) shows that most of the innexins in Aedes are still closely related to their Drosophila homologues despite the 250 million years of evolution that separate the mosquito from the fruit fly(Severson et al., 2004). The strong evolutionary conservation of innexins confirms the critical roles that gap junctions are now known to play in diverse functions of cells and tissues.

The neighbor-joining tree (Fig. 9) also shows that AeInx7 and AeInx4 are the least conserved with their Drosophila homologues. In particular, AeInx4 is the most divergent as it only loosely clusters with DrInx4, as well as DrInx5 and DrInx6. Two interpretations from this branching pattern are that: (1) the Inx4gene has evolved the fastest since the last common ancestor of both Aedes and Drosophila; and (2) Inx5 and Inx6 are unique to Drosophila, possibly evolving from gene duplications of Inx4. The latter interpretation is consistent with that of Stebbings et al. (Stebbings et al., 2002) who proposed that certain topological features of DrInx5 and DrInx6 are likely to have evolved recently.

Our RT-PCR studies indicate that transcripts for four of the Aedesinnexins are expressed in Malpighian tubules: AeInx1, AeInx2, AeInx3 and AeInx7(Fig. 10). Thus, one or more of these innexins may mediate the electrical coupling we observe between principal cells. Based on previous findings in Drosophila, we hypothesize that AeInx7 may be the dominant gap-junction transcript expressed in Aedes Malpighian tubules. For example, a tissue-specific`transcriptome' analysis of adult Drosophila found detectable expression of Inx2 in Malpighian tubules, but Inx7 was enriched 11-fold in Malpighian tubules compared to expression levels in the whole fly(Chintapalli et al., 2007). Furthermore, developmental studies on the expression of innexins in Drosophila indicate that Inx7 is exclusively expressed in the Malpighian tubules and midgut-progenitor cells from stage 11 onwards(Stebbings et al., 2002). Future studies aimed at quantifying the relative abundance of transcripts for AeInx1, AeInx2, AeInx3 and AeInx7 in Aedes Malpighian tubules are of obvious importance.

Our finding that Aedes Malpighian tubules express more than one innexin provides the possibility that principal cells form heteromeric gap junctions in which a single hemi-channel consists of different innexins. Such heteromerization occurs when Inx2 and Inx3 from Drosophila are coexpressed heterologously in paired Xenopus oocytes(Stebbings et al., 2000). Heteromeric gap junctions may provide a spectrum of permselectivity from ions to second messengers and increase the potential for regulating communications through gap junctions (Cottrell and Burt,2005; Zampighi et al.,2005).

To verify heteromerization of innexins in Malpighian tubules it will be necessary to localize each innexin isoform in the tubule using in situ hybridization and immunohistochemistry. The localizations would also determine if the four innexins that we have identified indeed localize in Malpighian tubules proper, and not in the tracheal tubes associated with Malpighian tubules, as was recently shown for an aquaporin channel in Aedes (Duchesne et al.,2003; Pietrantonio et al.,2000). The localization of innexins would also settle the question about the presence of gap junctions in stellate cells and identify the innexins that form them.

In conclusion, our study establishes the presence of gap junctions in Malpighian tubules of insects that may underlie the nearly perfect electrical coupling of membrane voltages and their oscillations. The inability of Lucifer Yellow to pass through gap junctions may reflect geometric and/or electrical barriers in the gap channel pore. Compared to known single gap-junction conductances, our measurement of the gap-junction resistance suggests that several thousand gap junctions join one principal cell of the tubule to the next. Calcium does not appear to regulate gap junctions in Malpighian tubules of the yellow fever mosquito, but metabolic inhibition with dinitrophenol uncouples gap junctions completely. Whether the shutdown of gap junctions is due to decreased intracellular ATP concentrations and/or reduced membrane voltages is unknown. Our initial molecular studies identified six innexins in the genome of Aedes aegypti, of which four are expressed in Malpighian tubule cDNA. Our measurement of the gap-junction resistance is the first in an intact Malpighian tubule under physiological conditions. The measurement advances the electrical model from a single cable to a double cable with the promise of increased refinement in the measurement and interpretation of electrophysiological data from the tubule.

LIST OF SYMBOLS AND ABBREVIATIONS

 
• DNP

  dinitrophenol

 
• E

  electromotive force

 
• I_gj

  gap-junction current

 
• I_inject

  current injected

 
• I_nj

  non-junctional current

 
• l

  distance

 
• ORF

  open-reading frame

 
• R_a

  apical resistance

 
• r_a

  length-specific axial cell resistance

 
• R_bl

  basolateral resistance

 
• R_co

  core resistance

 
• R_gj

  gap-junction resistance

 
• R_input

  input resistance

 
• R_lu

  tubule lumen resistance

 
• R_nj

  non-junctional resistance

 
• r_r

  length-specific radial cell resistance

 
• R_sh

  single shunt resistance

 
• V_bl

  basolateral membrane voltage

 
• V_t

  transepithelial voltage

 
• λ

  cell length constant

This work was part of a doctoral dissertation submitted by XingHe Weng to the Graduate School of Cornell University. The research would not have been possible without the support from the National Science Foundation(IOB-0542797). We thank Richard Veenstra (SUNY Upstate Medical University,Syracuse, NY, USA) for generous counsel during the years of this study, and we acknowledge Isabel Rodriguez-Barraquer for her exploratory electrophysiological work on gap junctions in mosquito Malpighian tubules. We are grateful to Dimitri Boudko (Whitney Laboratory, St. Augustine, FL, USA)for providing the adult Aedes Malpighian tubule cDNA library that we used as an independent control in our molecular studies.