The role of an ancestral hyperpolarization-activated cyclic nucleotide-gated K+ channel in branchial acid-base regulation in the green crab, Carcinus maenas

Potassium channels form a diverse family of ubiquitously expressed transmembrane proteins that are crucial for a number of important physiological functions (Tian et al., 2014). In particular, K⁺ fluxes are essential for the membrane excitability of neurons and muscles (Choe, 2002). In the crustacean gill epithelium, K⁺ channels have been identified to participate in the establishment of an electrochemical gradient over the basolateral membrane mainly generated by a basolateral Na⁺/K⁺-ATPase (Onken et al., 1991; Riestenpatt et al., 1996). Furthermore, apical K⁺ channels seem to co-localize with apical Na⁺/K⁺/2Cl⁻ (NKCC) to allow K⁺ circulation and thereby an uptake of Na⁺ and Cl⁻ in crustaceans (Riestenpatt et al., 1996; Onken et al., 2003). Because of the presence of apical NKCC and apical K⁺ channels, as well as the accompanying basolateral Cl⁻ channel, the moderately leaky gill epithelium of C. maenas has been compared with the thick ascending limb (TAL) of the mammalian kidney (Riestenpatt et al., 1996). The apical K⁺ channel present in the TAL is a member of the inwardly rectifying K⁺ channel family K_IR (Kir1.1) and promotes apical efflux of K⁺ to energize the apical NKCC. Additionally, Kir4.1/5.1 has been identified to be co-localized basolaterally with the Na⁺/K⁺-ATPase in the kidney's distal tubule, providing a basolateral back flow of K⁺ out of the cell into the body fluids to maintain the electrochemical gradient (Hibino et al., 2010).

As a general, relatively unspecific inhibitor of K⁺ currents, apical 15 mmol l⁻¹ Ba²⁺ reduced the short-circuit current (I_SC) and trans-epithelial conductance (G_TE) of split gill lamellae of C. maenas (Riestenpatt, 1995; Riestenpatt et al., 1996). In gills of Neohelice (Chasmagnathus) granulata, basolateral application of 10 mmol l⁻¹ BaCl₂ altered I_SC without affecting the conductance, but had no effect when applied apically even though apical Cs⁺, another unspecific K⁺ channel blocker, resulted in a significant decrease of I_SC (Onken et al., 2003). At these relatively high concentrations, Ba²⁺ is believed to target voltage-gated/delayed-rectifier K⁺ channels (Rudy, 1988; Wollmuth, 1994), while specifically inhibiting Ca²⁺-activated K⁺ channels and K⁺ channels of the K_IR family only in lower concentrations of up to 100 µmol l⁻¹ (Wollmuth, 1994; Franchini et al., 2004). The fact that the K⁺-dependent current in the C. maenas gill could be more potently inhibited by 10 mmol l⁻¹ apical Cs⁺, however, indicates the presence of at least two, a Ba²⁺-sensitive and a Ba²⁺-insensitive K⁺ channel population (Riestenpatt, 1995; Riestenpatt et al., 1996). Furthermore, application of Ba²⁺ in split gills of the Chinese mitten crab, Eriocheir sinensis, resulted in a complex di-phasic response of the short-circuit current I_SC (Onken et al., 1991), also indicating the participation of more than one kind of K⁺ channel.

In addition to their role in osmoregulation, basolateral K⁺ channels also participate in branchial ammonia excretion in C. maenas, with NH₄⁺ substituting for K⁺ because of its similar size and charge (Lignon, 1987; Skou, 1960; Weihrauch et al., 1998). Apical application of Cs⁺, however, did not affect branchial ammonia excretion in this species (Weihrauch et al., 1998). Furthermore, branchial ammonia regulation in C. maenas has been observed to be closely linked to acid-base regulation (Fehsenfeld and Weihrauch, 2013), but the specific role of K⁺ channels in this context is not clear.

Interestingly, a K⁺ channel of the hyperpolarization-activated cyclic nucleotide-gated channel family (HCN2) has recently been shown to be involved in acid-base and ammonia regulation in the collecting duct of the mammalian kidney (Carrisoza-Gaytán et al., 2011). Although to date, HCN2 has been mainly associated with pacemaker activities (DiFrancesco, 1993), Carrisoza-Gaytán and colleagues (2011) showed that this gene is also expressed in the basolateral epithelium of rat kidney cells and is capable of promoting NH₄⁺ transport over the membrane, thus participating in mammalian renal pH homeostasis.

The present study aims to investigate the role of K⁺ channels in branchial acid-base regulation of the osmoregulating green crab C. maenas, and particularly of an HCN-like protein as identified in the transcriptome. By application of two different K⁺ channel inhibitors, Ba²⁺ and the HCN-specific ZD7288, acid-base and ammonia regulatory properties of the gill epithelium will be characterized with respect to K⁺ and NH₄⁺ transport. A potential role for K⁺ channels, and specifically for HCN, will be evaluated.

To identify potential candidates for the hyperpolarization-activated cyclic nucleotide-gated potassium channel family in C. maenas (CmHCN), the partial amino acid sequence for the hyperpolarization-activated ion channel of Cancer borealis (GenBank accession no. AAZ80094.3) was used as assembly scaffold for a recently generated transcriptome for C. maenas (Verbruggen et al., 2015) using the basic local alignment search tool (BLAST) feature in Geneious^© version 7.1 (Biomatters Limited, Auckland, New Zealand). Accordingly, potential candidates for the basic voltage-gated (K_V), inward-rectifying (K_IR), two-pore domain (K_2P) and Ca²⁺-activated (K_Ca) K⁺ channels were identified in the C. maenas transcriptome using the sequences for the potassium voltage-gated channel subfamily H member 6 isoform 1 (accession no. NP_110406.1) and the ATP-dependent inward-rectifying potassium channel KIR4.1 and 5.1 of Homo sapiens (accession nos AAB07046.1 and AAF73240.1), the acid-sensitive two-pore domain K⁺ channel dTASK-7 of Drosophila melanogaster (accession no. CAI72673.1), as well as the calcium-activated potassium channel of C. borealis (accession no. AAZ80093.4).

The best protein model for the gene analysis on all HCN candidates as listed in Table 1 was chosen according to the test function provided in MEGA 6.06 (Tamura et al., 2013). Subsequently, a gene tree was constructed in MEGA 6.06, applying the maximum likelihood method using the gamma-distributed GL model with invariant sites (LG+I+G; Le and Gascuel, 2008). A matrix of pairwise distances estimated using a JTT model (Jones et al., 1992) was used to automatically generate an initial tree(s) for the heuristic search by applying neighbour-joining and BioNJ algorithms and subsequently selecting the topology with the superior log-likelihood value. Evolutionary rate differences among sites were modelled by discrete gamma distribution (six categories). All positions with less than 95% site coverage were eliminated (i.e. less than 5% alignment gaps, missing data and ambiguous bases were allowed at any position). Bootstrap analyses were performed with 1000 replicates.

For gill perfusion experiments, male green crabs, Carcinus maenas (Linnaeus 1758), were caught in traps in the Kiel Fjord (Baltic Sea) in front of the GEOMAR pier (west shore building, 54°19.8′N, 10°9.0′E) at 3–4 m depth in May 2012 (collection permit issued September 2009, file no. V313-72241.121-19). Animals were kept in ca. 250 litre flow-through tanks in a climate chamber at GEOMAR, directly fed by water from the Kiel Fjord (salinity=14.3±0.0 ppt, temperature=17.2±0.4°C, pH_NBS=7.96±0.03, P_CO2=96±7 Pa), until gill perfusions were performed (see below). Green crabs were fed ad libitum with blue mussels, Mytilus edulis, twice per week. Animals were starved for a minimum of 2 days before experiments.

For quantitative real-time PCR and Ussing chamber experiments, male green crabs were caught in Barkley Sound at the opening of the Pipestem Inlet (Vancouver Island, BC, Canada) in summer 2013 under Department of Fisheries and Oceans collection permit XR-235-2013, and transported back to Winnipeg, MB, Canada. They were kept at the animal holding facility of the University of Manitoba in aerated 1200 litre tanks with artificial seawater adjusted to a salinity of 32 ppt at 14°C (Seachem Marine Salt^®) until experimentation. For acclimation to dilute salinity of 11 ppt, eight animals were transferred into a 120 litre tank for a minimum of 7 days and fed ad libitum with squid once a week. Water was exchanged at least every second day and always on the day after feeding. Green crabs were starved for a minimum of 2 days prior to experiments.

For subsequent acclimation of green crabs to high environmental ammonia (HEA) for 7 days, the brackish water of the 120 litre tanks was enriched with 1 mmol l⁻¹ NH₄Cl. Water was exchanged on a daily basis and animals were not fed during HEA acclimation.

Gene expression analysis was performed on tissues obtained from animals held at University of Manitoba. Posterior gill 7 and all other tissue samples of osmoregulating C. maenas (11 ppt) acclimated to control conditions and HEA (1 mmol l⁻¹ NH₄Cl, 7 days) were harvested from animals acclimated as described above, while samples for hypercapnia-acclimated green crabs (400 Pa) were used as described in the recent study by Fehsenfeld and Weihrauch (2013). Tissue samples representing anterior and posterior gills were generated by pooling equal tissue amounts of the respective gills prior to isolating the RNA (gills 4 and 5 for anterior gills, and gills 7, 8 and 9 for posterior gills).

Briefly, total RNA was isolated under RNase-free conditions using TRIZOL (Invitrogen, Carlsbad, CA, USA). Following DNase I (Ambion™, Thermo Fisher Scientific, Pittsburgh, PA, USA) treatment, regular PCR was performed to ensure the absence of DNA traces (RbS-3 primers; forward: 5′ GTCCCTTTTCACCAAGGACA; reverse: 3′ CAAGGCCAAACTCAACA GGTT). Subsequently, verified DNA-free RNA was transcribed with the iScript cDNA synthesis kit (Bio-Rad, Mississauga, ON, Canada). Quantitative real-time PCR (qPCR) was performed on 20 ng cDNA and a final primer concentration of 0.4 µmol l⁻¹ in 15 µl reactions with the SSO FastEvaGreen Supermix (Bio-Rad). The gene-specific primers for CmHCN were designed based on the identified transcriptome contig (forward: 5′ GTCTTCATGCGCATCTTCAA; reverse: 5′-GTTGATTGCCACCCAAGAGT) and resulted in the expected specific 123 bp fragment. The primers were ensured to result in a single signal in the qPCR by adding a melting curve analysis following the regular qPCR. In every qPCR run, a negative control (no template) was included to verify that the assay was not contaminated, as well as one of the standards treated as an unknown sample to ensure homogeneous performance. A standard curve was included in each qPCR run based on a dilution series of known quantities of the respective gel-extracted gene fragment (0.01–100 fg; QIAquick Gel Extraction Kit, Qiagen). Additionally, the housekeeping gene ribosomal RbS-3 was used for internal normalization of hypercapnia-related tissues in accordance with Fehsenfeld and Weihrauch (2013). No housekeeping gene (i.e. RbS-3, arginine kinase) with constant mRNA expression levels could be identified for HEA-acclimated green crabs. Hence, absolute quantification was solely assessed using the included standard curve. In this case, a standard curve based on diluted cDNA samples was performed to ensure comparable amplification efficiency [PCR fragments: 99.4±0.2% (N=3); cDNA templates: 100.5±0.7% (N=3)].

For gill perfusion experiments in Kiel, hemolymph ammonia, hemolymph carbonate system parameters, as well as hemolymph cation concentrations (Na⁺, K⁺ and Ca²⁺; flame photometer analysis performed by T. Stegmann, Institute of Physiology, Christian-Albrechts-University Kiel) of control green crabs were assessed for the adjustment of the perfusion solution as described in Fehsenfeld and Weihrauch (2013) to account for in vivo conditions of the Baltic Sea individuals. Accordingly, the perfusion solution consisted of (in mmol l⁻¹): 340 NaCl, 9 CaCl₂, 7 MgCl₂, 9 KCl, 14 NaHCO₃, 0.13 NH₄Cl, 0.5 glutamine, 0.5 glucose and 0.1 glutathione. The pH (NBS scale) was adjusted to 7.9. Posterior gills (no. 7) were perfused according to the procedure described in Fehsenfeld and Weihrauch (2013), following the technique of Siebers et al. (1985). Each perfusion consisted of a control step, followed by the application of the respective inhibitor and a third wash-out step to ensure the survival of the gill tissue. Each perfusion step allowed for 10 min adjustment of the gill, after which the perfusate was collected for 30 min. To block basolateral K⁺ channels, BaCl₂ was applied basolaterally in the second perfusion step at a concentration of 12 mmol l⁻¹ (Schirmanns and Zeiske, 1994; Riestenpatt, 1995; Riestenpatt et al., 1996; Onken et al., 2003). The inhibitor ZD7288 (4-ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidinium chloride) was applied basolaterally at a concentration of 10 µmol l⁻¹ to specifically block the K⁺ current generated by the identified HCN, according to Carrisoza-Gaytán et al. (2011). Gill ammonia excretion rates were calculated based on the difference in the perfusate compared with the perfusion solution. Perfusate proton concentration was calculated based on the pH, whereas perfusate P_CO2 and HCO₃⁻ were calculated using the equations and constants following Truchot (1976). An increase in perfusate [H⁺], CO₂ and/or [HCO₃⁻] in comparison to the initial perfusion solution was considered equivalent to a decrease in excretion.

Ussing chamber experiments were performed at room temperature in Winnipeg, Canada. Single gill lamellae of posterior gill 7 were split according to Schwarz and Graszynski (1989) and mounted in the EM-CSYS-6 Ussing chamber system (Physiologic Instruments, San Diego, CA, USA) with specific sliders (P2308), providing a circular aperture of 1 mm that resulted in an open epithelial surface area of 8×10⁻³ cm². Silicon grease was used to seal the edges of the preparation to prevent potential edge damage. Preparations were aerated throughout the experiment. Providing six channels and chambers, the EM-CSYS-6 Ussing chamber system together with the automatic clamping device (VCC 600, Physiologic Instruments) allowed for six split gill lamellae to be investigated in parallel in one experimental run.

Symmetrical conditions were used to account for only active ion transport, with the following identical crab saline being applied to both the apical and basolateral sides (in mmol l⁻¹): 248 NaCl, 5 CaCl₂, 4 MgCl₂, 5 KCl, 2 NaHCO₃, 5 HEPES and 2 glucose (after Riestenpatt et al., 1996). The pH_NBS was adjusted to 7.9. Two pairs of Ag/AgCl electrodes were connected to the preparation via agar bridges (3% agarose in crab saline to avoid diffusive effects of KCl): one pair served to determine the trans-epithelial potential difference (PD_TE), and the second pair served as current electrodes to short-circuit the PD_TE with the VCC600. The reference electrode was positioned at the basolateral side. Before the tissue was introduced into the experimental setup, the voltage offset between paired electrodes was set to 0.0. It was ensured that the agar bridges did not produce excessive resistance by passing current in this open setup (resulting in ca. 60 µA). Using the fluid resistance compensation function of the clamping device, the potentiometer was adjusted so that the voltage was again nullified while passing current. The resistance of the crab saline (R_S) could be read from the potentiometer knob and was determined as 8.2±0.1 Ω cm⁻². After mounting the six split gill lamellae, the (area-)corrected short-circuit current I_SC (as a measure of ion/charge movements over the gill epithelium; Riestenpatt et al., 1996) was directly recorded simultaneously for every channel using Acquire Analyze software (version 2.2). Voltage pulses (5 mV every 60 s for 1 s) were applied to assess the epithelial conductance {G_TE=1/R_TE, R_TE=[(ΔV/ΔI)−R_S]=[(5 mV/ΔI_SC)−8.2 Ω cm⁻²], where R_TE is the transepithelial resistance and ΔV is the amplitude of the voltage pulse}.

The dose–response curve for ZD7288 was obtained by increasing the concentration basolaterally in a step-wise fashion. As soon as I_SC levelled off for one concentration (ca. 10–15 min) and remained stable for >1 min, the next higher concentration was applied. In case of the dose–response curve for BaCl₂, preceding the basolateral application of each concentration of BaCl₂ was the application of an equivalent concentration of choline chloride apically to account for potential Cl⁻-dependent changes in I_SC. To account for time effects on the I_SC, one split gill lamella per experimental run was kept under control conditions during the full duration of the experiment.

All statistical analyses were performed with PAST3 (Hammer et al., 2001). Data sets were tested for normal distribution (Shapiro–Wilk test) and homogeneity of variances (F-test) to qualify for parametrical testing. If one or both requirements were not met, data were log transformed. If log transformation still did not result in a data set appropriate for parametric testing, non-parametrical tests were applied instead. For comparison of single means, a paired t-test was applied (gill perfusion, mRNA expression changes stressors), whereas Kruskal–Wallis with pairwise Mann–Whitney comparisons was used to compare multiple means (tissue expression). Dose–response curves were evaluated by repeated measures ANOVA with post hoc Tukey's pairwise comparisons after verifying a normal distribution (Shapiro–Wilk test) and homogeneity of variances (Levene's test). All results were considered significant at P<0.05. Graphs were generated using the software Inkscape, version 0.48 (https://inkscape.org/). The dose–response curve for Ba²⁺ was generated by GraphPad Prism version 6.00 for Windows (GraphPad Software, San Diego, CA, USA, www.graphpad.com).

The two different identified contigs (resulting in an open reading frame of >100 amino acids) in the green crab transcriptome resembling HCN, GBXE01087917.1 (326 amino acids) and GBXE01153810.1 (148 amino acids), aligned to different regions of the HCN sequence of C. borealis (positions 291–712 versus 66–213, respectively) and HCN2 of H. sapiens (positions 360–659 versus 180–299, respectively). Only GBXE01087917.1 contained the specific motifs for voltage-gated potassium channels (PLN03192) and the cyclic nucleotide-monophosphate binding domain (cAMP/cNMP/Crp) as determined by the NCBI protein BLAST tool and was therefore chosen for further analysis. The partial 326 amino acid sequence for CmHCN was highly conserved and resembled the human HCN2 with 69% identity and 83% overall conserved domains (Fig. 1). The major binding sites for ZD7288 were identified as Ala-425 and Leu-432, as recently described by Cheng et al. (2007).

For each of the four classes of K⁺ channels (K_V, K_IR, K_2P and K_Ca), a respective contig could be identified in the transcriptome of C. maenas as generated by Verbruggen et al. (2015). A discrete gamma distribution was used to model evolutionary rate differences among sites between all sequences included in the gene tree analysis (six categories; +G, parameter=3.3). The rate variation model allowed for some sites to be evolutionarily invariable ([+I], 1.1% sites). All positions with less than 95% site coverage were eliminated, resulting in a total of 141 positions in the final dataset. For all included invertebrate and vertebrate sequences, the respective K⁺-channel families formed distinct monophyletic groups (Fig. 2). As expected, the HCNs represented a distinct (sub)family of the voltage-gated K⁺ channels. Furthermore, the gene tree analysis revealed that C. maenas grouped with the other arthropod channels in all cases. Although four different HCNs could be found in mammals (only HCN2 included, other data not shown), arthropods seem to express only one HCN. A similar observation was made for the K_IR-related proteins of the arthropods, which seemed to resemble KIR5.1 channels as found in mammals and fish.

CmHCN was observed to be ubiquitously expressed in all analyzed tissues. Although the cerebral ganglion exhibited the highest mRNA expression levels (ca. 10-fold compared with the other tissues), posterior gills still exhibited significant levels of the transcript (Fig. 3).

The excretion of ammonia, CO₂ and HCO₃⁻ as well as the concentration of H⁺ in the perfusate were significantly affected by both Ba²⁺ and the HCN-specific ZD7288 in posterior gill 7 of osmoregulating green crabs when applied basolaterally (Fig. 4). The degree of inhibition, however, varied with both inhibitors. Whereas branchial ammonia excretion decreased to a greater extent with the application of Ba²⁺ (ca. 70% versus 40% with ZD7288; Fig. 4A), excretion of acid equivalents (as equivalent to accumulation of protons in the hemolymph) decreased more drastically under the influence of ZD7288 (1.3-fold; Fig. 4B). The effect of Ba²⁺ on perfusate proton concentration was small, yet significant. In regard to CO₂ excretion, the application of both inhibitors resulted in a significant, ca. 1.1-fold increase of P_CO2 in the perfusate (equivalent to a 1.1-fold decrease in excretion; Fig. 4C). The most pronounced difference between Ba²⁺ and ZD7288 was observed in relation to HCO₃⁻: whereas ZD7288 resulted in a small, but significant increase of 5% in the perfusate, Ba²⁺ resulted in the opposite response, namely, a 5% increase in perfusate [HCO₃⁻] (reduced excretion; Fig. 4D).

Split gill lamellae of osmoregulating green crabs employed in the present study generated an outside positive PD_TE of 4.5±0.4 mV (n=17) and an inward negative I_SC of −445±28 µA cm⁻² (n=17) in the Ussing chamber setup, indicating a net flux of negative charge from the apical to the basolateral side. The corresponding conductance G_TE accounted for 35.0±2.5 mS cm⁻² (n=17). Although G_TE exhibited similar values in all preparations, I_SC was more variable, with values between −196 and −634 µA cm⁻². Without influence of any inhibitor and only accounting for changes over time, the I_SC remained relatively stable over the experimental time period and decreased on average by 1.7±0.4% per 10 min (n=5). The maximal overall decrease was observed in one of the runs during the initial 20 min and accounted for 5% of the initial I_SC.

When applied basolaterally on split gill lamellae (gill 7) in Ussing chamber experiments, both inhibitors (Ba²⁺ and ZD7288) reduced the initial short-circuit current in a dose-dependent manner, as can be seen in the exemplary time curves (Figs 5, 6) as well as the dose–response curves (Fig. 7). The conductance remained relatively stable and only seemed to decrease slightly with higher concentrations of BaCl₂.

For the dose–response curves, all measurements for I_SC have been corrected for the observed maximum time effect of −5% per 20 min. In the case of Ba²⁺, the basolateral inhibition of I_SC followed typical Michaelis–Menten kinetics and could be fitted by the Hill equation with ΔI_SCmax=−346.9 µA cm⁻² (73% inhibition) and IC₅₀=0.76 mmol l⁻¹. The application of ZD7288 resulted in an almost linear decrease of the I_SC and hence was not fitted by a typical dose–response curve. The maximum inhibition of I_SC at 10 µmol l⁻¹ (plateauing after 1 µmol l⁻¹) accounted for 30% of the initial I_SC (ΔI_SCmax=−120.2 µA cm⁻²) and the basolateral application of ZD7288 resulted in an IC₅₀ of 6×10⁻⁷ mmol l⁻¹.

mRNA levels of CmHCN in posterior gill 7 were significantly downregulated in green crabs exposed to 7 days of either high environmental P_CO2 (hypercapnia, 400 Pa; ca. 50% reduction) or elevated environmental ammonia (1 mmol l⁻¹ NH₄Cl HEA, ca. 80% reduction; Fig. 8).

In studies to date, the presence of K⁺ channels in the gill epithelium of decapod crustaceans has only been verified by ion flux and short-circuit current studies on split gill lamellae applying the inhibitors Ba²⁺ and Cs⁺ (Onken et al., 1991, 2003; Riestenpatt et al., 1996; Weihrauch et al., 1998, 1999). Because of its similar radius in comparison to K⁺, Ba²⁺ is able to enter the channel pore and bind to the selection filter motif (SF4) domain of the respective K⁺ channels, thereby competing with K⁺ and blocking its binding site more potently because of its dual positive charge (Rossi et al., 2013). As Fig. 2 shows, representatives for K_Ca, K_2P, K_IR, K_V and HCN can be found throughout the animal kingdom, and, based on transcriptome analysis, also in the green crab C. maenas. However, to our knowledge, the specific nature of K⁺-promoting membrane channels in the crustacean gill has not been investigated to date.

The identified CmHCN, as described in the present study, thus provides the first (partial) sequence-based characterization of a potential K⁺/NH₄⁺ channel in a decapod crustacean and documents its involvement in branchial acid-base and ammonia regulation.

Importantly, Fig. 1 shows clearly that CmHCN exhibits the respective binding sites for ZD7288, Ala-425 and Leu-432, as described by Cheng et al. (2007). Even though the latter differs from the mammalian Ile-432, the authors of that study showed that substitution of Ile with Leu did not shift the IC₅₀ significantly. Accordingly, this provided the basis for the functional studies applying ZD7288 in gill perfusion and Ussing chamber experiments.

As seen by the sequence and gene tree analysis, the level of conservation of the HCN sequence between arthropods and vertebrates in general, and C. maenas and H. sapiens specifically, is significant. A similar level of conservation between invertebrate and vertebrate proteins is found in other important membrane transporters such as the α-subunit of the Na⁺/K⁺-ATPase, one of the most important epithelial proteins (C. maenas versus Homo sapiens, 79% identity and 89% overall conserved domains, NCBI). Consequently, HCN can be hypothesized to inherit a significant role in the course of evolution regarding epithelial ion transport.

In collecting ducts of rat renal cortex and medulla – structures that are responsible for both ammonia and acid-base regulation in mammals (Weiner and Verlander, 2013) – HCN2 has recently been observed to transport not only K⁺, but also NH₄⁺ (Carrisoza-Gaytán et al., 2011). As the major ion, ammonia and acid-base regulatory organ (Larsen et al., 2014), the crustacean gill resembles many of the transport processes that are observed in the mammalian kidney (Riestenpatt et al., 1996) and it is therefore not surprising that, in addition to in the cerebral ganglion, significant mRNA expression levels of HCN can be found in the osmoregulatory posterior gills of C. maenas. Interestingly, the difference of CmHCN mRNA levels between anterior and posterior gills (ca. fivefold) mirrors what is seen for Na⁺/K⁺-ATPase (NKA) activity in these tissues (ca. 2 versus 11 µmol P_i mg protein⁻¹ h⁻¹), emphasizing the postulated close relationship between basolateral NKA and K⁺ channels in mammalian (Hibino et al., 2010) as well as invertebrate epithelia (Riestenpatt et al., 1996).

Furthermore, the detected downregulation of the mammalian renal HCN2 mRNA in response to chronic metabolic acidosis in rats (Carrisoza-Gaytán et al., 2011) was also observed in the present study in the gills of C. maenas in response to the metabolic acidosis resulting from acclimation to high environmental ammonia (unpublished data). In addition, a role for CmHCN in general acid-base regulation is indicated by its significant downregulation as observed in the present study in response to a respiratory acidosis caused by exposure to hypercapnia (Fehsenfeld and Weihrauch, 2013). Even though HCN homologs have been cloned from several invertebrate species and their hyperpolarization-activated current characteristics have been observed to closely resemble mammalian HCNs (Biel et al., 2009), their potential function has so far only been associated with chemo-sensitivity in the fruit fly Drosophila melanogaster (Chen and Wang, 2012) and signal transduction in olfactory neurons of the spiny lobster Panulirus argus (Gisselmann et al., 2005).

Therefore, the present study is the first to deliver evidence for a function of a conserved (ancestral) HCN as a key player in acid-base and ammonia regulation in an invertebrate.

The results of the gill perfusion experiments clearly indicated at least two different K⁺/NH₄⁺-promoting structures in the branchial, basolateral epithelial membranes of C. maenas. These include at least one Ba²⁺-sensitive K⁺ channel, as well as the identified ZD7288-sensitive member of the hyperpolarization-activated cyclic nucleotide-gated K⁺-channel family, CmHCN. The findings of the present study therefore strengthen results of earlier studies on split gill lamellae of C. maenas (Riestenpatt, 1995) and E. sinensis (Onken et al., 1991) that indicated that different populations of K⁺ channels promote transcellular K⁺ excretion because of the observed incomplete and complex di-phasic response of the I_SC when Ba²⁺ was applied to the apical and basolateral membranes, respectively.

Moreover, the participation of the respective K⁺ channels in branchial acid-base and ammonia excretion became obvious by the observed decrease of ammonia and CO₂ excretion as well as the accumulation of protons in the perfusate with the basolaterally applied inhibitors Ba²⁺ and ZD7288 in posterior gill 7. Therefore, these results also support findings of a recent study that claim the ultimate linkage between acid-base and ammonia regulatory processes in the gill epithelium of this decapod crustacean by using the same transporter inventory (Fehsenfeld and Weihrauch, 2013). Interestingly, HCO₃⁻ excretion by the posterior gill is induced by ZD7288 but decreased with Ba²⁺, thus verifying that the observed differences in excretion rates of ammonia, acid equivalents and CO₂ are not solely based on differences in the effective concentration of each inhibitor, but by the inhibitors targeting different K⁺/NH₄⁺-promoting structures.

Both inhibitors in the present study were applied basolaterally based on the fact that: (1) apical CsCl (10 mmol l⁻¹) had been shown to have no substantial effect on ammonia excretion in C. maenas (Weihrauch et al., 1998) even though it significantly reduced I_SC (Riestenpatt et al., 1996), and (2) the novel candidate gene CmHCN seemed to resemble the HCN2 as identified in basolateral membranes of the vertebrate kidney (Carrisoza-Gaytán et al., 2011) as discussed above. In the case of ZD7288, however, it cannot be excluded that it targeted the respective structure also in the apical membrane because of its strong lipophilic character and subsequent diffusion into the cell (Gasparini and DiFrancesco, 1997; Harris and Constanti, 1995). A predominantly intracellular presence and action of ZD7288 would also explain why the effect of this inhibitor as observed in gill perfusion experiments was not reversed in the wash step.

Further characterization of the effect of both inhibitors on overall epithelial ion movements and the resulting trans-epithelial current was achieved in Ussing chamber experiments on split gill lamellae. The observed values for general characteristics of split gill 7 were comparable to values in the literature. Even though slightly higher outside positive potential differences were measured in posterior gill 7 in green crabs acclimated to 10 ppt in gill perfusion experiments (7.6±0.2 mV; Siebers et al., 1985), split gills seemed to generate a slightly lower potential difference of 6.6±1.3 mV (Onken and Siebers, 1992), which is even more comparable to the value observed in the present preparation (4.4±0.4 mV). Posterior gills of 10 ppt-acclimated C. maenas were also observed to generate an inward negative I_SC between −240 and −375 mA cm⁻², as well as a conductance of 40–45 mS cm⁻² (Onken and Siebers, 1992; Riestenpatt et al., 1996), values that were also reached in the present study, validating the used technique.

The application of the inhibitors revealed the very different sensitivity of potential K⁺ channels to Ba²⁺ and ZD7288. Under symmetrical conditions, the inward negative I_SC generated by the split gill epithelium of C. maenas is mainly carried by an electrogenic inward-directed Cl⁻ current dependent on the negative membrane potential over the basolateral membrane that is generated by the basolateral Na⁺/K⁺-ATPase and basolateral and apical K⁺ channels (Riestenpatt et al., 1996). Increasing extracellular Cl⁻ concentrations by the addition of BaCl₂ would lead to the weakening of the driving force for Cl⁻ uptake and hence result in a passive decrease of I_SC. Hence, when not corrected for by addition of an equivalent concentration of Cl⁻ in the apical bath, no saturation of the effect of BaCl₂ on the I_SC was observed, and eventually, I_SC switched to inward positive values (>20 mmol l⁻¹ BaCl₂, data not shown). Consequently, in the present study, preceding every increase in basolateral BaCl₂ concentration, choline chloride was added to the external bath to correct for this effect. As could be expected by an increased apical [Cl⁻], I_SC became transitionally more negative. As a result, however, the effect of Ba²⁺ on the I_SC followed distinct Michaelis–Menten kinetics and the unbiased IC₅₀ could be determined.

The IC₅₀ for Ba²⁺ (0.76 mmol l⁻¹) in the split gill lamellae of green crab as determined in the present study was very similar to what was observed in the basolateral midgut epithelium of the tobacco hornworm, Manduca sexta (0.50 mmol l⁻¹; Zeiske et al., 1986), as well as the apical epithelium of frog skin (0.11 mmol l⁻¹; De Wolf and Van Driessche, 1986) and the basolateral epithelium of toad urinary bladder (1 mmol l⁻¹; Van Driessche and Erlij, 1988). At these relatively high concentrations, however, Ba²⁺ is believed to target delayed rectifier K⁺ channels and K_V channels rather than K_IR and K_Ca channels (Rudy, 1988; Franchini et al., 2004). Hence, the obtained results are not conclusive in regard to the nature of the affected K⁺ channel(s), taking into account that potentially all of the identified K⁺ channels (K_v, K_2P, K_Ca, K_IR and HCN) might be present in the branchial epithelium and targeted by Ba²⁺. At the concentration of 100 µmol l⁻¹ that can still be considered specific for K_IR and K_Ca channels (Franchini et al., 2004; Hassinen et al., 2015), the observed inhibition of I_SC in the posterior gill 7 of osmoregulating C. maenas accounted for ca. 10% of the initial I_SC, indicating that a potential involvement of branchial K_IR and/or K_Ca channels cannot be excluded, but if present, only accounts for a minor fraction of the K⁺ current. Furthermore, it cannot be excluded that Ba²⁺ also targets the HCN, as has been observed for high millimolar concentrations in the rat kidney (Carrisoza-Gaytán et al., 2011). According to the results of the present study, however, only ca. 30% of the measured K⁺-related current can be assigned to the identified CmHCN in the branchial gill epithelium as demonstrated by the application of ZD7288. Consequently, the maximum observed effect of Ba²⁺ (73% inhibition of I_SC) remains partly unexplained for and additional experiments are necessary to further characterize the Ba²⁺-sensitivity of the K⁺ current in the branchial epithelium of C. maenas.

In contrast to the high effective concentration of Ba²⁺, the unusually low IC₅₀ of ZD7288 is astonishing. To verify the effect of ZD7288 and exclude a simply time-dependent shift of the I_SC, the dose–response curve has been corrected for the maximal time-dependent shift observed in split gill preparations without the application of inhibitors. Hence, the analysis represents a conservative approach and the observed effect of 30% inhibition on the I_SC might slightly underestimate the actual inhibition (ca. 5%). The IC₅₀ for ZD7288 of 6×10⁻⁷ mmol l⁻¹ in the present study is, to our knowledge, 500-fold smaller than the next smallest IC₅₀ for ZD7288 determined in sinoatrial node cells isolated from the guinea pig heart (3×10⁻⁴ mmol l⁻¹; BoSmith et al., 1993) and can be interpreted as an extremely high affinity of this reagent for CmHCN and/or a comparably low abundancy of the protein in the branchial epithelium. The fact that CmHCN mRNA expression levels in posterior gills of C. maenas are similarly high compared with those observed for the Na⁺/K⁺-ATPase (Fehsenfeld and Weihrauch, 2013) and a pronounced effect of 10 µmol l⁻¹ was observed in isolated gills perfusion experiments, however, rather supports the high binding capacity of this inhibitor. Generally, similar to Ba²⁺, the mode of action of ZD7288 is voltage-dependent and related to the ambient K⁺ concentration (Cheng et al., 2007), and might therefore differ substantially from other tissues and organisms, explaining the observed differences in IC₅₀ values for ZD7288, but also for Ba²⁺.

Furthermore, the potency of ZD7288 was shown to depend on the respective HCN isoform in mammalian tissues and blocked the I_F (‘funny’) current in the sinoatrial node more effectively (IC₅₀=3×10⁻⁴ mmol l⁻¹; BoSmith et al., 1993) than the I_h (‘hyperpolarization’-activated) current in neurons (IC₅₀=2-41×10⁻³ mmol l⁻¹; Shin et al., 2001). Hence, even though highly conserved on the protein level, the potentially single HCN in C. maenas might exhibit different features regarding ZD7288 binding that make it more sensitive for the block [i.e. shifts in hydrophobicity (Cheng et al., 2007) and epigenetic factors altering methyl-dependent polarization, as observed for Ba²⁺ (Rossi et al., 2013)].

Even though the present study clearly showed that the involvement of HCN in branchial K⁺/NH₄⁺ transport accounted for ca. 30% of the K⁺-generated current, the full extent of the Ba²⁺-related block cannot be explained at this point. The analysis of the C. maenas transcriptome and subsequent gene tree analysis revealed a potential role for all major classes of K⁺ channels in the generation of (branchial) K⁺ currents in the green crab. After verifying the mRNA and/or protein expression pattern for each of the newly identified candidate genes in the green crab, further research also needs to be done to fully characterize the K⁺ current in the branchial epithelium.

In conclusion, the present study strongly indicates the involvement of a basolateral HCN-like potassium channel in branchial acid-base and ammonia regulation in the decapod crustacean C. maenas. Besides potential other basolateral Ba²⁺-sensitive K⁺ channel(s), gill perfusion experiments applying the HCN-specific inhibitor ZD7288 show that this novel candidate gene is involved in NH₄⁺ transport over the basolateral gill epithelial membrane. Interestingly, similar features (i.e. the potential dual transport capabilities of K⁺ and NH₄⁺ and decreased mRNA expression changes in response to metabolic acidosis) have been observed for the HCN2 expressed in the mammalian kidney, indicating a conserved function of this gene for basic homeostatic regulation throughout the animal kingdom. The high level of conservation of the amino acid sequence between invertebrate and vertebrate HCNs strengthens the importance of this gene as a potential key player in acid-base and ammonia regulation in vertebrates as well as invertebrates. However, as this only accounts for ca. 30% of the K⁺-related trans-epithelial short-circuit current, further research should be carried out to elucidate the nature of the complete K⁺-dependent current in the gill epithelium of C. maenas.

The authors thank Drs Greg G. Goss, Chris M. Wood and Tamzin Blewett, as well as the staff of the Bamfield Marine Sciences Center, for the help in collecting the green crabs. Additional thanks go to the Animal Holding Facility of the University of Manitoba for taking care of the crabs. Thanks go also to Dr Frank Melzner for providing supervision, space and laboratory equipment for gill perfusion experiments on C. maenas in Kiel, Germany, and to Michael Gaudry, who assisted with the transcriptome analysis.