Ammonia excretion and acid–base regulation in the American horseshoe crab, Limulus polyphemus

Amino acid-catabolizing organisms produce toxic nitrogenous waste products that must be eliminated via excretion strategies such as ammonotelism, where ammonia is the dominant excretory product (Wright, 1995). Compared with other nitrogenous waste products, ammonia is energetically beneficial as it can be released as is and does not require additional energy for conversion into its less toxic counterparts, such as urea or uric acid. Aquatic animals (excluding mammals and elasmobranchs) commonly exhibit ammonotelism because of abundant water availability for continuous excretion, preventing toxic build-ups (Larsen et al., 2014).

Aquatic animals may experience excess extracellular ammonia (in this study, ammonia refers to the sum of NH₃ and NH₄⁺) whilst burying or upon emersion, when excretion is impaired (Weihrauch et al., 1999). Increased concentration of circulating ammonia can cause numerous deleterious effects, such as acid–base imbalance (Goldsmith and Hilton, 1992; Wilson and Taylor, 1992), ionoregulatory disruption (Young-Lai et al., 1991) and neurotoxicity (Butterworth, 2002; Marcaida et al., 1992). Several key proteins have been suggested to influence ammonia excretion of invertebrate species, including Na⁺/K⁺-ATPase (NKA), V-type H⁺-ATPase (HAT) and glycosylated Rhesus proteins (Rh-proteins) (Chasiotis et al., 2016; Larsen et al., 2014; Masui et al., 2002; Pitts et al., 2014; Quijada-Rodriguez et al., 2015; Weihrauch et al., 1998, 2012).

Mounting evidence suggests that ammonia excretion and acid–base regulation of several invertebrate species are intricately linked, possibly because of the acidic and basic forms of ammonia (NH₄⁺ and NH₃, respectively) and the sharing of key transporters such as NKA, HAT and Rh-proteins (Fehsenfeld and Weihrauch, 2016a). This notion has been encouraged by investigations concluding that anterior and posterior gills of Carcinus maenas, the green shore crab, have similar capacities for ammonia and H⁺-equivalent excretion (e.g. Fehsenfeld and Weihrauch, 2013). Although extensive studies have focused on ammonia excretory mechanisms of crustaceans and teleost fishes, there has yet to be an equivalent study on chelicerates. This is likely because the vast majority of this subphylum are terrestrial arachnids, which excrete guanine as their dominant nitrogenous waste product because of environmental water constraints (Larsen et al., 2014). Xiphosura, which includes the American horseshoe crab, Limulus polyphemus (Linnaeus 1758), is an exception to the terrestrial chelicerates in terms of lifestyle.

Limulus polyphemus has remained morphologically unchanged for over 200 million years (Avise et al., 1994) and is currently widespread along the east coast of the USA, and Mexico (Shuster, 1979). The horseshoe crab is of economic importance to the biomedical industry as an extract collected from its hemolymph is used for testing bacterial endotoxin contamination in medical products (Novitsky, 1984). Limulus polyphemus occupy the constantly fluctuating estuarine and coastal areas, inhabiting deeper waters as they age into adulthood and returning to sandy beaches to mate, where they remain emersed (Rudloe, 1981; Shuster, 1982). When not mating, much of this benthic animal's time is spent buried in search of prey, such as polychaete worms and bivalve mollusks (Botton, 1984), which are rich in protein. Based on these observed behaviors, L. polyphemus likely experiences natural exposure to both hypercapnia (elevated ambient P_CO2) and high ambient ammonia (HAA). Emersion during mating exposes animals to elevated P_CO2. Buried animals experience minimal water circulation, which could lead to the accumulation of wastes such as CO₂ and ammonia, particularly for those that feed whilst buried (McGaw, 2005; Taylor et al., 1985). Thus, it is expected that L. polyphemus is equipped with effective mechanisms to avoid accumulation of toxic ammonia and compensate for any acid–base disturbances caused by HAA and hypercapnia.

Gills and gill-like structures act as the predominant excretory organs of several aquatic ammonotelic species (Weihrauch et al., 2009; Wright and Wood, 2009), suggesting that the book gills of L. polyphemus play a major role in ammonia excretion. Five pairs of book gills can be found on the ventral side of the horseshoe crab (Fig. 1A) and each book gill comprises over 100 lamellae (Shuster, 1982). An individual lamella consists of two single cell layered half-lamellae or epithelia (ventral and dorsal) separated by hemolymph space and stabilized by pillar cells (Henry et al., 1996). Ultrastructural differences exist within a single lamella, where a thick central mitochondria-rich area (CMRA) exists within the ventral epithelium and is surrounded by thin peripheral mitochondria-poor areas (PMPAs; Henry et al., 1996; Fig. 1C). The CMRA has also been shown to possess extensive membrane infoldings and higher NKA activity compared with the PMPA of the ventral epithelium and the uniformly thin dorsal epithelium (Henry et al., 1996). Initially, such branchial traits may indicate the presence of active ion transport, although its role in ammonia regulation is so far unknown.

In this study, we aimed to gain basic knowledge of ammonia and acid–base regulatory patterns in the marine chelicerate L. polyphemus and further explore differences in branchial regions (ventral versus dorsal half-lamellae; CMRA versus PMPA) at the tissue and molecular levels to predict the roles of each region in ammonia and acid–base regulation. HAA and hypercapnia are examples of environmental conditions that can affect the animals' nitrogen metabolism and ability to maintain acid–base homeostasis via changes in ammonia excretion, hemolymph carbonate system parameters and mRNA expression levels of genes putatively involved in regulating such physiological factors. Therefore, we applied these treatments to investigate how L. polyphemus responds to such changes in seawater parameters in comparison with previously studied marine arthropods. Based on ultrastructural differences between the various branchial regions, we predict that the CMRA of the ventral epithelium is an important site for ammonia excretion and acid–base regulation, and that this species can tolerate elevated ambient ammonia and CO₂ because of its lifestyle as a burying species that likely encounters such stressors in the wild.

Juvenile American horseshoe crabs, Limulus polyphemus (carapace width=6–8 cm; mass=14–43 g), were acquired from an aquarium store (Reefs2Go, Clearwater, FL, USA); male adults (carapace width=14–15 cm; mass=322–441 g) were captured by the Whitney Laboratory for Marine Bioscience (St Augustine, FL, USA) from the Indian River Lagoon (FL, USA). Animals were maintained at 22°C with a 12 h:12 h light:dark cycle in Animal Holding Facility (University of Manitoba, Winnipeg, MB, Canada), where a maximum of 10 animals were held in 1200 l tanks filled with artificial seawater (32 ppt; SeaChem Marine Salt, Madison, WI, USA) and equipped with external filters and a UV-sterilization system. Horseshoe crabs were fed ad libitum daily with raw shelled shrimp, and a fine sand bed with crushed oyster shells was provided as substrate.

where [CO₂^*] is the concentration of dissolved CO₂, K₀ is the solubility coefficient of CO₂ in seawater and K₁ and K₂ are the first and second dissociation constants for carbonic acid in seawater. The program assumes a pressure of approximately 1 atm and uses this assumption to perform the conversion between partial pressure and fugacity (Lewis and Wallace, 1998).

Seawater parameters of control conditions were as follows: pH of 8.07±0.02, P_CO2 of 64±3 Pa, [HCO₃⁻] of 2.3±0.1 mmol l⁻¹, total alkalinity of 2.7±0.1 mmol kg⁻¹ seawater (SW) and ammonia concentration of 6.0±0.5 µmol l⁻¹. Treatment of HAA was achieved by enriching aquaria with NH₄Cl to reach an average concentration of 996.7±25.5 µmol l⁻¹. A full water change was carried out every 1–2 days with pre-equilibrated HAA seawater to minimize fluctuations in ammonia concentration. For the hypercapnia treatment, the IKS Aquastar (IKS ComputerSysteme GmbH, Karlsbad, Baden–Württemberg, Germany) provided continuous control of CO₂ injection to reach an average P_CO2 of 311±9 Pa, which resulted in a pH of 7.44±0.01, an [HCO₃⁻] of 2.5±0.1 mmol l⁻¹ and a total alkalinity of 2.7±0.1 mmol kg⁻¹ SW. A full water change was conducted every 1–2 days with pre-equilibrated high P_CO2 seawater. Animals were starved for 2 days prior to all experiments and hemolymph collection. All animals were exposed to their respective conditions (control, HAA or high P_CO2) for 7–9 days.

α_CO2 is the solubility coefficient for CO₂ (mmol l⁻¹ Torr⁻¹) and pK₁ is the first dissociation constant of carbonic acid. The remaining hemolymph sample was frozen at −20°C until analyzed for ammonia concentration using the same method as previously described for tank water samples.

All whole-animal ammonia excretion experiments were conducted at 22°C in 32 ppt salinity seawater with an air stone, and animals within containers were covered to reduce stress. Juveniles were placed in individual containers with 300 ml of seawater (SeaChem Marine Salt) and an air stone. Water samples were taken hourly for 2 h. In order to rid the container of residual ammonia from the previous period, seawater was gently siphoned out after each trial and the container was rinsed with fresh seawater. The process was repeated once more prior to starting the next time period. The ammonia excretion rate of control crabs was measured while they were in control (ammonia-free) seawater, then measured again after placing the animals in HAA seawater. Afterwards, the same animals were exposed to HAA for 7–9 days, after which the ammonia excretion rate was determined in both control and HAA seawater. A different set of animals was exposed to hypercapnia for 7–9 days and the ammonia excretion rate was only determined while animals were in high P_CO2 seawater. All samples were frozen at −20°C and later analyzed for ammonia concentration.

Each individual sample consisted of pooled whole lamellae collected from the anterior-most gill pair of three juveniles, for a total of nine sampled animals. Determination of NKA activity and total protein concentration of the samples followed methods described for aquatic frogs (Cruz et al., 2013), with the exception of a higher ouabain concentration (5 mmol l⁻¹) because invertebrates have been shown to exhibit lower ouabain sensitivity than vertebrates (Postel et al., 1998), and a doubling of pyruvate kinase and lactate dehydrogenase concentrations to ensure adequate enzyme availability. The NKA assay protocol was modified after Gibbs and Somero (1989) and McCormick (1993) using a UV-Vis spectrophotometer (Agilent Technologies, Santa Clara, CA, USA) at 20°C. Gill samples (∼90 mg) were homogenized at 9000 rpm in 14 volumes of cold SEID homogenization buffer (pH 7.0) composed of 150 mmol l⁻¹ sucrose, 10 mmol l⁻¹ EDTA, 50 mmol l⁻¹ imidazole and 0.1% (w/v) deoxycholate. Homogenates were centrifuged at 4°C for 1 min at 5000 g and the supernatant of each sample was collected for subsequent enzyme assay and protein assay.

The enzyme reaction buffer (pH 7.5) included 10 mmol l⁻¹ NaCl, 5 mmol l⁻¹ MgSO₄, 50 mmol l⁻¹ imidazole, 3 mmol l⁻¹ ATP disodium salt, 2 mmol l⁻¹ phosphoenolpyruvate and 0.2 mmol l⁻¹ NADH sodium salt. Pyruvate kinase was also added at 10 IU ml⁻¹ as well as 8 IU ml⁻¹ of lactate dehydrogenase. Either 10 mmol l⁻¹ KCl or 10 mmol l⁻¹ NH₄Cl was provided as substrate, and NKA activity was calculated as the difference in the rate of change in absorbance at 340 nm in the absence (total ATPase activity) and presence of 5 mmol l⁻¹ ouabain. Each sample was measured in triplicate and NKA activity was calculated using the NADH extinction coefficient of 6.2 mmol l⁻¹ cm⁻¹. The protein concentration of each sample was determined using the Pierce BCA protein assay kit (Fisher Scientific) with bovine serum albumin used as standard.

Ussing chambers (EM-CSYS-6, Physiologic Instruments, San Diego, CA, USA) were used to measure ammonia movement across split gill lamellae of adult L. polyphemus when an ammonia gradient of ∼300 µmol l⁻¹ (basolateral) to 0 µmol l⁻¹ (apical), which is similar to what was observed in intact animals used in this study, was applied (see Fig. 5A,C). For HAA-treated juvenile animals, an inwardly directed ammonia gradient of ∼1000 µmol l⁻¹ (apical) to ∼300 µmol l⁻¹ (basolateral; as measured in hemolymph ammonia of control animals) was used (see Fig. 5B,D). Lamellae were collected from the left anterior-most gill and immediately placed in chilled seawater. The thick chitinous edge of an individual lamella was held with fine forceps under a dissecting microscope, while a 21 gauge needle was inserted between the ventral and dorsal epithelia to separate the two half-lamellae. Another pair of fine forceps was inserted into the partially split area of the lamella and then used to tease apart the two lamellar halves. The thick chitinous edge along each lamella was trimmed to ensure a tight seal within the tissue holders. In an intact animal, the dorsal half-lamella faces the body and the ventral half-lamella faces away from the body (see detailed diagram in Henry et al., 1996). A total of six dorsal and six ventral half-lamellae were collected from control animals, and nine dorsal and eight ventral half-lamellae were collected from HAA-treated animals.

Each half-lamella was individually mounted on a custom-made tissue holder (aperture: 0.25 cm² for control animals, 0.16 cm² for HAA animals), ensuring approximately equal surface area of the central and peripheral regions for all epithelia. All chambers were temperature controlled to 22°C and half-chambers were filled with either 4 ml (control tissues) or 3 ml (HAA-treated tissues) of their respective solution: artificial seawater for the apical side and physiological saline for the basolateral side. The apical solution (pH 8.1) consisted of 430 mmol l⁻¹ NaCl, 10 mmol l⁻¹ CaCl₂, 37 mmol l⁻¹ MgCl₂, 10 mmol l⁻¹ KCl and 2 mmol l⁻¹ NaHCO₃. For HAA-treated tissues, 1 mmol l⁻¹ NH₄Cl was also added to the apical solution. The composition of the physiological saline followed that of earlier studies (Robertson, 1970; Smith et al., 2002) and the measured [HCO₃⁻] of two non-acclimated animals prior to any experiment. Physiological saline (pH 7.6) consisted of 450 mmol l⁻¹ NaCl, 11.8 mmol l⁻¹ KCl, 9.9 mmol l⁻¹ CaCl₂, 31.9 mmol l⁻¹ MgCl₂, 14.1 mmol l⁻¹ MgSO₄, 4.4 mmol l⁻¹ NaHCO₃, 3.2 mmol l⁻¹ glucose and 0.3 mmol l⁻¹ NH₄Cl. Tissues were allowed to equilibrate in the artificial solutions for 20 min; during this time, no parameters were measured. After incubation, chamber solutions were replaced with fresh solutions and allowed to incubate for either 2 h (control tissues) or 3 h (HAA-treated tissues), then a sample from each solution was collected. Each sample was weighed to estimate its volume and frozen at −20°C until ammonia analysis.

where C₂ and C₁ are the final and initial ammonia concentrations (µmol l⁻¹) of the bathing solution with an observed ammonia loss (apical or basolateral), V is the volume of sample collected from the Ussing chamber (l), SA is the surface area of the gill epithelium (cm²) and t is the sampling period (h).

where J_amm,TE is the transepithelial ammonia efflux rate of control tissues (220.28±63.89 nmol cm⁻² h⁻¹), SA is the epithelial surface area (0.25 cm²) and ΔC is the ammonia concentration gradient between the initial apical and basolateral solutions of control tissues (305.94 µmol l⁻¹ higher in the basolateral side). Once determined, the diffusion coefficient was used in conjunction with SA (0.16 cm²) and ΔC for HAA tissues (728.40 µmol l⁻¹ higher in the apical side) to obtain the predicted influx rate of 819.45 nmol cm⁻² h⁻¹ across the ion-leaky ventral epithelium.

Electrophysiological parameters (transepithelial potential difference, PD_TE; transepithelial conductance, G_TE) of the ventral and dorsal half-lamellae were obtained by mounting half-lamellae in a custom-made Ussing chamber. Here, 0.18 cm² of gill surface was superfused from each side (apical and basolateral) with physiological saline by gravitational flow at a rate of 12.5 ml min⁻¹. In this series of experiments, only the central region of either the ventral (N=5) or dorsal epithelia (N=6) was used. Gill lamellae were collected from the right anterior-most gills and split by the same methods as those used in ammonia transport measurements. Electrodes (Ag/AgCl) to measure voltage were connected via agar bridges (3% agarose in 3 mol l⁻¹ KCl) to each bath. Another set of Ag/AgCl electrodes were directly inserted into each half-chamber to pass current across the tissue for voltage clamping. All electrodes were connected to a VCC-600 voltage/current clamp (Physiologic Instruments, San Diego, CA, USA) and data were collected with a Lab Trax 4/16 (World Precision Instruments, Sarasota, FL, USA). To measure PD_TE, the tissue was observed in open circuit, and to obtain G_TE, the voltage was clamped to 0 mV (short-circuit). Every 30 s, a current was passed across the tissue that clamped the voltage to +1 mV and then to −1 mV for 2 s. The positive and negative currents to reach ±1 mV were identical, indicating that the tissue had no rectifying properties. The conductance of the half-lamella was then calculated according to Ohm's law from the measured currents and the voltage pulses of 1 mV.

Tissue samples used in qPCR consisted of two main groups: (1) branchial CMRA and PMPA, coxal gland and brain tissue samples; and (2) the ventral and dorsal branchial epithelia from control, HAA-treated and hypercapnia-treated animals. Tissue samples were collected from adult male horseshoe crabs. Animals were anesthetized by placing them on ice for 30 min, then a ∼15×4 cm hole was drilled along the dorsal carapace, stretching from the lateral compound eyes to the midpoint of the opisthosoma, using a Dremel rotary tool with a cutting wheel attachment (Robert Bosch Tool Corporation, Mt Prospect, IL, USA). Animals were killed by removing the brain and dorsal nerve cord.

All tissue samples were collected in Ambion RNAlater Stabilization Solution (Fisher Scientific). Total RNA was isolated in TRIzol (Fisher Scientific), homogenized and treated with DNase (Invitrogen DNase I Amplification Grade, Fisher Scientific). RNA samples were tested for DNA contamination by PCR employing the L. polyphemus specific primers EF-1α F/R (Table 1). DNA-free RNA samples were reverse transcribed into cDNA by qScript cDNA synthesis kit (Quanta BioSciences, VWR, Radnor, PA, USA).

Primers were designed based on published sequences of the respective gene and the L. polyphemus genome (GenBank accession no.: AZTN00000000.1), and tested for the presence of a single amplicon of the correct product size using EconoTaq DNA Polymerase (VWR). PCR products were imaged with gel electrophoresis, purified with E.Z.N.A. Gel Extraction Kit (Omega Bio-Tek, VWR) and sequenced at Robarts Research Institute (London, ON, Canada). Products were confirmed to code for target genes by searching the sequence on GenBank using the BLAST algorithm and ensuring a similar match to previously published sequences as well as an exact match to the genome.

During qPCR, 10 µl assays were performed with SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories, Mississauga, ON, Canada) in a MiniOpticon real-time thermocycler (Bio-Rad Laboratories). A single PCR product was verified through melt curve analysis. The results were log transformed and threshold adjusted to optimize for reaction efficiency (88–100%) and correlation (>0.99). Relative transcript abundance was determined according to the Pfaffl method (Pfaffl, 2004), employing RNA Polymerase II (Pol2) as the internal standard. Pol2 did not show a significant difference in transcript abundance in any tissues and/or treatments (data not shown).

All data sets are presented as means±s.e.m. and, for all statistical tests, a P-value of <0.05 was considered significant; changes with P-values between 0.05 and 0.1 are accompanied with the actual P-value and/or noted as having insufficient statistical power. All data sets were analyzed for outliers (Grubbs’ test), normality (Shapiro–Wilk test) and homogeneity of variance (Levene's test) to determine whether the data set was parametric. Parametric two-sample data sets were tested with either Student's t-test or paired t-test, whereas parametric data sets with more than two means were analyzed using one-way ANOVA with post hoc Tukey's pairwise comparisons. If a data set failed the Shapiro–Wilk and Levene's tests, the data were log transformed and the tests were repeated. Two-sample data sets that were still non-parametric were treated with the Mann–Whitney U-test, and Kruskal–Wallis test with post hoc Mann–Whitney pairwise comparisons were performed on data sets with more than two means. All statistical analyses were carried out using the Paleontological Statistics (PAST) software (https://folk.uio.no/ohammer/past/; Hammer et al., 2001).

For verification of the involvement of NKA, the activity of this enzyme isolated from the gills was measured, using either K⁺ or NH₄⁺ as the substrate. With K⁺, NKA activity of whole-gill homogenate was 17.0±0.9 nmol ADP min⁻¹ mg⁻¹ protein. When K⁺ was replaced by NH₄⁺ in this assay, the activity was significantly lowered (∼21%) to 13.4±0.7 nmol ADP min⁻¹ mg⁻¹ protein (N=3 pooled samples from a total of nine individuals, data not shown).

To identify the animal's main site of ammonia excretion, the mRNA transcript abundance of putatively key genes in this process was assessed in the coxal gland, the CMRA and PMPA of the gill (Fig. 1C), and brain tissue (Fig. 2). Although both Rh-protein isoforms exhibited the highest abundance in gill tissues, significant differential patterns emerged between the two isoforms: Rhesus-protein 1 (LpRh-1) expression in the CMRA was 20-fold higher than that in the PMPA, whereas Rhesus-protein 2 (LpRh-2) showed over 2-fold higher transcript level in the PMPA than in the CMRA. In contrast to Rh-protein expression patterns, NKA abundance was uniform across the CMRA, PMPA and coxal gland, but was significantly lower in the brain. Hyperpolarization-activated cyclic nucleotide-gated K⁺ channel (HCN) expression in the CMRA did not significantly differ from that in the PMPA, but expression in the brain was significantly higher than that in other investigated tissues.

High transcript levels of Rh-proteins indicated the gills' central role in ammonia excretion; therefore, only the branchial regions were analyzed for mRNA transcript levels of the following genes. Cytoplasmic carbonic anhydrase (CA-2) expression level relative to Pol2 was significantly higher (11-fold) in the CMRA (175.7±18.2; N=5) than in the PMPA (15.9±3.8; N=6). No significant difference was observed for HAT, for which expression levels relative to Pol2 were 75.7±10.4 in the CMRA (N=6) and 60.8±9.1 in the PMPA (N=6).

As a rough comparison of overall abundance of each gene relative to Pol2, the relative mRNA transcript levels in the CMRA of the gill were as follows: NKA 188.7±26.8, CA-2 175.7±18.2, HAT 75.7± 10.4, LpRh-1 56.0±9.7, LpRh-2 1.0±0.2 and HCN 0.6±0.2.

To clarify the ammonia transport function of the book gills in L. polyphemus, a preparation of the split gill lamella was employed to further characterize both the ventral and dorsal half-lamellae. Mounting each half-lamella in an Ussing chamber allowed the observation of lamellar properties relevant to ion transport, including G_TE, PD_TE and transepithelial ammonia fluxes. A noticeable difference in G_TE of the two half-lamellae was observed, where the dorsal half-lamella had significantly lower G_TE (0.20±0.04 mS cm⁻²; N=6) than the ventral half-lamella (145.40±33.95 mS cm⁻²; N=5) (Fig. 3B). The PD_TE was zero for both the ventral (N=5) and the dorsal (N=6) half-lamellae.

When an ammonia gradient mimicking in vivo conditions (no ammonia on apical side, ∼300 µmol l⁻¹ ammonia on basolateral side) was applied over the half-lamella, transepithelial ammonia efflux (J_amm,TE) at a rate of 220.28±63.89 nmol cm⁻² h⁻¹ was observed across the ventral epithelium, as calculated by the loss of ammonia in the basolateral solution, but no net efflux was detected across the dorsal epithelium (Fig. 6A). The ventral half-lamella generated metabolic ammonia at a rate of 83.54±40.80 nmol cm⁻² h⁻¹ released towards the apical side, but it is important to note the possibility of ammonia release towards the basolateral side and/or tissue decomposition. Interestingly, despite the low conductance and lack of transepithelial ammonia flux for the central region of the dorsal half-lamella, this tissue still produced metabolic ammonia. Of this, 98.50±18.30 nmol cm⁻² h⁻¹ was released towards the apical side and 34.03±17.73 nmol cm⁻² h⁻¹ towards the basolateral side (Fig. 6B).

Differences between branchial regions were also observed at the molecular level. Compared with the dorsal half-lamella, the ventral half-lamella exhibited significantly (at least 2-fold) higher relative gene expression of LpRh-1, CA-2 and HCN, but the opposite pattern was detected for LpRh-2 (Fig. 3A).

Juvenile and adult L. polyphemus were exposed to 997 µmol l⁻¹ NH₄Cl to observe the effects of short-term (7–9 days) HAA treatment on the animals' hemolymph parameters, ammonia excretion rates and mRNA transcript levels of genes related to ammonia homeostasis. Hemolymph pH averaged 7.59±0.04 in control adults and slightly decreased to 7.47±0.05 when exposed to HAA (P=0.09; Table 2). Hemolymph P_CO2 (164.1±24.3 Pa) and [HCO₃⁻] (2.47±0.41 mmol l⁻¹) of control adults significantly increased to 346.9±51.8 Pa and 3.95±0.34 mmol l⁻¹, respectively, following HAA treatment. No difference was found in hemolymph ammonia concentration between control animals (320.8±36.9 µmol l⁻¹) and HAA-exposed animals (334.0±28.3 µmol l⁻¹).

Control juveniles excreted ammonia at a rate of 298.0±31.7 nmol g⁻¹ fresh mass h⁻¹ when placed in ammonia-free seawater, but exhibited significantly pronounced uptake of ammonia at a rate of 2393.3±924.4 nmol g⁻¹ fresh mass h⁻¹ immediately after being placed in HAA seawater (1000 µmol l⁻¹ NH₄Cl; Fig. 4A). However, following a 7–9 day treatment of 997 µmol l⁻¹ NH₄Cl seawater (average tank ammonia concentration during the treatment period), animals excreted ammonia at a rate of 331.6±55.7 nmol g⁻¹ fresh mass h⁻¹ in HAA seawater (Fig. 4B), similar to the excretion rate of control animals in ammonia-free seawater. When reintroduced to ammonia-free seawater, HAA-treated animals excreted ammonia at a rate of 733.0±99.6 nmol g⁻¹ fresh mass h⁻¹, which was significantly higher than the excretion rate of control animals in ammonia-free seawater.

Split gill lamella experiments (Fig. 6A) revealed that ventral epithelium from HAA-treated animals exhibited an ammonia influx of 311.38±58.77 nmol cm⁻² h⁻¹ to the basolateral side, which is significantly different from the net efflux of 220.28±63.89 nmol cm⁻² h⁻¹ observed in the control epithelium. Although ammonia entered the branchial lamellae via the ventral epithelium, ammonia was excreted against its P_NH3 and NH₄⁺ gradients via the dorsal epithelium at a rate of 61.89±24.47 nmol cm⁻² h⁻¹, whereas, in contrast, the epithelium of control animals showed no transepithelial transport. Of the metabolically produced ammonia, 37.41±80.68 nmol cm⁻² h⁻¹ was released towards the basolateral side for the ventral epithelium and 75.85±49.57 nmol cm⁻² h⁻¹ was released towards the apical side for the dorsal epithelium; these values were not significantly different from those in control tissues (Fig. 6B).

HAA treatment also induced significant changes in branchial gene expression levels. Following HAA treatment, the relative mRNA transcript levels of CA-2 and HCN in the ventral half-lamella nearly doubled, accompanied by a 1.4-fold increase in HAT expression (Fig. 7A). In the dorsal half-lamella, HAA treatment roughly doubled and tripled the transcript levels of CA-2 and LpRh-1, respectively (Fig. 7B).

A 7–9 day exposure to elevated P_CO2 (311 Pa) induced significant increases of nearly 2-fold in hemolymph P_CO2 (301.7±33.7 Pa) and 2-fold in [HCO₃⁻] (5.32±0.20 mmol l⁻¹; Table 2). However, there was no change in hemolymph pH in high P_CO2-treated animals (pH 7.62±0.05). Simultaneously, both hemolymph ammonia concentration (183.8±19.6 µmol l⁻¹; Table 2) and whole-animal ammonia excretion rate (170.2±40.3 nmol g⁻¹ fresh mass h⁻¹; N=5) of hypercapnia-treated animals significantly decreased by 43% compared with those of control animals. Out of all investigated genes, hypercapnia treatment only significantly elevated the relative gene expression level of HAT in the ventral half-lamella (Fig. 8A).

In this study, we provide evidence that the exposed and well-ventilated book gills of L. polyphemus excrete ammonia (Fig. 6A,B). As outlined in the Introduction, horseshoe crab book gills exhibit distinct regions: a CMRA and a PMPA (Fig. 1C). This is remarkable in itself as these morphological differences have so far only been found in the gills of osmoregulating crustaceans such as Carcinus maenas (Goodman and Cavey, 1990), but not in the gills of osmoconforming invertebrates. The presence of branchial CMRA in osmoconforming L. polyphemus suggests a non-osmoregulatory yet energy-requiring function, and is likely related to acid–base regulation and/or ammonia excretion. A specialized function of CMRA in acid–base regulation/ammonia excretion is supported by significantly higher gene expression levels of CA-2, HCN and LpRh-1 in the CMRA than in the PMPA (Fig. 2A,D), with all three genes known to be involved in either of these related processes in other invertebrates (Adlimoghaddam et al., 2015; Fehsenfeld and Weihrauch, 2013, 2016a,b; Martin et al., 2011; Quijada-Rodriguez et al., 2015; Weihrauch et al., 2012). Although mRNA transcript levels for NKA were not particularly high in the CMRA compared with levels in the PMPA and coxal gland, the branchial NKA could accept NH₄⁺ as substrate; in addition, Henry et al. (1996) confirmed high specific activities of this enzyme, indicating post-transcriptional regulation of the pump.

Both LpRh-1 and LpRh-2 displayed particularly high expression in branchial tissues compared with that in the coxal gland and brain (Fig. 2A,B), a pattern previously observed for Rhcg1 and Rhcg2 in fish (Nawata et al., 2007), as well as the marine Dungeness crab, Metacarcinus magister, Rh-protein (Martin et al., 2011). In contrast, the Rhbg isoform in fish shows a rather uniform expression among tissues (Nawata et al., 2007). Interestingly, although LpRh-1 showed high transcript levels in the CMRA, LpRh-2 mRNA abundance was higher in the PMPA, indicating distinct functions of the two isoforms. However, significant quantitative PCR results should be interpreted as an indication of the protein's importance rather than a direct assumption of actual protein abundance, because mRNA expression does not always correlate with protein tissue abundance or enzyme activity, as post-translational processes could affect the final abundance and/or activity of the protein.

In addition to differences between the CMRA and PMPA, flux experiments on split half-lamellae revealed fundamental differences in ammonia transport rates and G_TE between the ventral and dorsal half-lamellae (Figs 3B and 6A). Considering that L. polyphemus is a marine species, the dorsal half-lamella exhibited, very atypically, low conductance compared with, for instance, the G_TE measured in non-ion transporting exopodites of the brackish water-acclimated isopod Idotea baltica (G_TE≈14 mS cm^–2; Postel et al., 2000) or epithelia from freshwater organisms, such as the gill epithelia of red crab Dilocarcinus pagei (G_TE≈4 mS cm^–2; Onken and McNamara, 2002) and Chinese mitten crabs, Eriocheir sinensis (G_TE≈4 mS cm^–2; Weihrauch et al., 1999). It is notable that while this ion-tight half-lamella did not promote transepithelial ammonia transport, metabolic ammonia was still released towards the environment, indicating an existing pathway for gaseous NH₃. Metabolic ammonia could also have formed as a result of tissue decomposition.

The ventral half-lamella is responsible for transbranchial ammonia excretion, and exhibited high G_TE as expected for the gill of a marine arthropod (Fig. 3B; Freire et al., 2008; Weihrauch et al., 1999). The measured ammonia efflux in an outwardly directed gradient (∼300:0 μmol l⁻¹) was comparable with fluxes measured under the same applied gradient in the perfused gill of the green crab C. maenas (Weihrauch et al., 1998), where the flux was ∼202 nmol g⁻¹ fresh mass h⁻¹. The importance of the ventral half-lamella in ammonia transport and acid–base regulation was also evident from the high transcript levels of LpRh-1 and HCN (Fig. 3A; Carrisoza-Gaytán et al., 2011; Fehsenfeld and Weihrauch, 2016b), and of CA-2, one of the general key players in transepithelial CO₂ transport and ammonia excretion (Fehsenfeld and Weihrauch, 2016a; Gilmour, 2012; Wright and Wood, 2009).

Book gill lamellae are positioned in a fashion that may cause excreted waste products of a lamella's ventral epithelium to be released on to the dorsal epithelium of adjacent lamellae. This waste may accumulate when the branchial tissue is at rest, and thus is not ‘flapping’ (Vosatka, 1970); backflow of waste could become an issue, particularly in a tissue that is ‘ion-leaky’, rendering waste excretory processes a semi-futile cycle. Functionally different half-lamellae within the same gill could minimize backflow diffusion and mitigate wasteful excretion. This is perhaps especially beneficial during periods of impaired waste excretion. For example, during air exposure, the thin gill lamellae are likely difficult to keep apart in the absence of water. Therefore, the low conductive properties of the dorsal epithelium can serve as a barrier to ionic waste products such as NH₄⁺. Evidence for this is provided by the observed ammonia flux patterns of tissues from HAA-treated animals, but understanding the way in which water flows over the lamellae of the book gills would be useful information to further such a hypothesis.

When foraging or in hiding, horseshoe crabs often bury themselves within the sediment (Shuster, 1982). Under such conditions of reduced branchial ventilation and limited water exchange, the concentration of ammonia in the immediate vicinity of the animal may substantially increase from both the animal's own metabolism and the reduced bacterial nitrification due to hypoxia (Weihrauch, 1999; Widdicombe et al., 2016). Similar potential impairment in ammonia excretion can also occur when sexually mature adults gather on beaches to mate, where subsequent air exposure of upwards of 4 h (Shuster and Botton, 1985) may slow or halt the excretion of such water-soluble waste products until the animals return to sea. These natural challenges may help explain why horseshoe crabs seemingly cope rather well with impairments in ammonia excretion.

Although a massive ammonia influx occurred after an acute exposure to HAA, excretion rates recovered to control values when animals were exposed to the new conditions for 7–9 days (Fig. 4). The lack of change in hemolymph ammonia levels following a prolonged HAA exposure (Table 2) indicates the animals' capacity to excrete this toxic waste product under the new conditions of a ∼3-fold inwardly directed gradient. A physiological state of enhanced excretion was also evident through a doubling of ammonia excretion rates when animals were placed in ammonia-free seawater, although hemolymph parameters indicated a slight respiratory acidosis (Table 2).

Split gill lamella experiments showed that in the ventral half-lamella of HAA-treated animals, an ammonia influx of 311.38±58.77 nmol cm⁻² h⁻¹ (Fig. 6A) occurred likely as a result of the tissue's high ion conductance. However, the actual measured influx rate was well below the predicted level of 819.45 nmol cm⁻² h⁻¹, as determined through a rearrangement of Fick's laws of diffusion (see Eqn 8), across the ion-leaky ventral epithelium, indicating that the ventral epithelium partially counteracted the ammonia influx (Fig. 6A). This ability may be facilitated by elevated mRNA transcript levels of HCN, CA-2 and HAT (Fig. 7A), a pump that generates a transmembrane H⁺ gradient and often works in liaison with the H⁺-providing cytoplasmic CA-2, which can create a partial pressure gradient for NH₃ as it continues to react with H⁺ to form NH₄⁺ on one side of the membrane. NH₄⁺ can be trapped within vesicles either for an exocytosis-mediated excretion, as seen in the green crab C. maenas (Weihrauch et al., 2002), or for apical ammonia trapping when apically localized in epithelial cells, which is often observed in animals living in low-buffered freshwater habitats and is less likely to occur in the highly buffered seawater environment (Larsen et al., 2014).

Although ammonia influx can occur via the ventral epithelium in HAA, ammonia was excreted against its P_NH3 and NH₄⁺ gradients via its dorsal counterpart. Dorsal half-lamellae from HAA-treated animals were capable of maintaining near-control basolateral ammonia levels (∼300 µmol l⁻¹) against a 3-fold inwardly directed ammonia gradient and even possessed the capacity for some degree of apically directed, possibly active, NH₄⁺ excretion (Fig. 6A). This is likely facilitated by the dorsal epithelium's ion-tight properties prohibiting NH₄⁺ movement, as well as an increase in LpRh-1 and CA-2 mRNA abundance (3-fold and 2-fold, respectively; Fig. 7B). This may indicate a difference in the role of LpRh-1 in the two half-lamellae, given that in control animals LpRh-1 expression was higher in the ventral epithelium than in its dorsal counterpart (Fig. 3A), yet its expression only responded to HAA acclimation in the dorsal epithelium (Fig. 7B).

The functional difference between the half-lamellae could be useful in minimizing ammonia influx during periods of elevated ambient ammonia. However, despite the dorsal epithelium's capacity to excrete ammonia against an inwardly directed gradient, it did not adequately counteract the influx observed in the ventral epithelium. Perhaps another organ, such as the coxal gland, contributes to such processes under HAA stress. Unlike the gills, the coxal gland is not directly exposed to the environment and its potentially high ammonia levels; in addition, the gland has levels of NKA comparable to those of the gills (Fig. 2C) and the activity of NKA has been observed to surpass that of the central gill regions (Henry et al., 1996; Towle et al., 1982), indicating the tissue's potential for active NH₄⁺ transport. Previous studies have also indicated that the urine of L. polyphemus is acidic (pH 6.85) and rich in ammonia (ca. 650 μmol l⁻¹) compared with the animals' native hemolymph status (Mangum et al., 1976), which is indicative of an ammonia excretory organ. Further investigation is required to determine the functional role of the coxal gland in ammonia excretory processes and how it responds to HAA.

Horseshoe crabs may encounter elevated environmental P_CO2 in their natural environments when buried, as a result of poor water circulation (McGaw, 2005; Taylor et al., 1985), or when emersed, as occurs during reproduction (Rudloe, 1980), which could affect acid–base status. When animals were exposed to hypercapnia for 7–9 days, the hemolymph carbonate system adjusted to a new steady-state level, characterized by elevated P_CO2 levels and [HCO₃⁻] while pH remained unaffected (Table 2). An identical response was observed in the marine Dungeness crab, M. magister, under comparable conditions (Hans et al., 2014), with the predicted purpose of ensuring continued CO₂ diffusion out of the body along its partial pressure gradient (Melzner et al., 2009). Increased hemolymph [HCO₃⁻] may have occurred as a means of counteracting a resulting acidification of the body fluid pH due to an increased CO₂ load, which is typically observed in fish and crustaceans (Appelhans et al., 2012; Hayashi et al., 2004; Spicer et al., 2007).

In both L. polyphemus and M. magister, hemolymph ammonia levels decreased after high P_CO2 treatment (Table 2), which was associated with partial metabolic depression in M. magister (Hans et al., 2014) but requires verification in L. polyphemus. Moreover, metabolic depression in response to elevated ambient P_CO2 and the subsequent respiratory acidosis are usually associated with an uncompensated extracellular pH shift (Pörtner et al., 2004). However, much like the Dungeness crab, the predicted respiratory acidosis was fully compensated in L. polyphemus. The observed upregulation of HAT in the transporting ventral half-lamella (Fig. 8A) might reflect the necessary physiological changes required to keep the hemolymph pH constant. Altogether, this indicates that at least L. polyphemus in the adult stage has sufficient plasticity to withstand short-term exposure to increased ambient P_CO2 levels and decreased pH regimes, which reflect the predicted future ocean scenario for the year 2300 (Caldeira and Wickett, 2005).

Despite similarities in external morphology and the presence of branchial CMRA in both juveniles and adults, it is possible that there are physiological differences between the age groups used in this study as they inhabit different water depths (Rudloe, 1981; Shuster, 1982). However, the two age groups exhibited similar patterns in ammonia regulatory response to hypercapnia, where reduced ammonia excretion by juveniles and a decrease in hemolymph ammonia concentration of adults occurred at the same scale (43%; Table 2). Both age groups also showed efficient counteractive ammonia excretory capacities when challenged with HAA, as seen by the lack of change in ammonia excretion and hemolymph ammonia concentration (Fig. 4B; Table 2), which were tested on juveniles and adults, respectively. Regardless, future studies comparing the juvenile and adult response to environmental stresses would be beneficial in determining any physiological differences between the two age groups.

This study confirmed regional differences in L. polyphemus book gills through the use of split gill experiments with Ussing chambers and qPCR. Although the ventral epithelium excreted ammonia under normal conditions, the dorsal epithelium could be a barrier against inwardly directed gradients of ionic waste products such as NH₄⁺, as shown by active ammonia flux towards the apical side in tissues from HAA-treated animals, as well as elevated LpRh-1 and CA-2 mRNA expression. However, another pathway of ammonia excretion was suggested because of the animal's ability to maintain normal whole-animal ammonia excretion levels and hemolymph concentrations during HAA treatment, possibly involving the coxal gland. Current literature on ammonia regulation would benefit from further studies on the role of the coxal gland, its link with acid–base regulation and the mechanisms behind such processes.

We would like to express our deepest gratitude to Dr Barbara Batelle (Whitney Laboratory for Marine Bioscience) for providing us with adult specimens and for her expertise in tissue dissection. We would like to thank the Animal Holding Facility staff (University of Manitoba) for caring for the research animals, as well as Michael J. Gaudry for his assistance in confirming the cytoplasmic carbonic anhydrase sequence through a RAxML gene tree and creating custom tissue holders for Ussing chambers. The majority of the data in this publication are part of a thesis (Stephanie Hans, Acid–base regulation and ammonia excretion in the American horseshoe crab, Limulus polyphemus, MSc thesis, University of Manitoba, 2016. URI: http://hdl.handle.net/1993/31775).