Identification of the role of Rh protein in ammonia excretion of the swimming crab Portunus trituberculatus

Rhesus-like glycoprotein (Rh protein) was first detected in human erythroid cells in 1939 (Levine and Stetson, 1939), and is an analog of the ammonia (NH₃/NH₄⁺) permease (MEP)/ammonia transporter (Amt) family. The Amt family in bacteria and plants and the MEP family in yeast function to promote the uptake of ammonia. In humans, there are at least four kinds of Rh proteins, including Rh30 (RhD/RhCE) and RhAG in erythrocytes, and RhBG and RhCG in other tissues (Peng and Huang, 2006), while Rhag, Rhbg and Rhcg have been identified in non-human vertebrates (Wright and Wood, 2009). Although there are many members belonging to the Rh protein family, ammonia transport capabilities have only been shown for the glycosylated forms of RhAG/Rhag, RhBG/Rhbg and RhCG/Rhcg so far, and the function of non-glycosylated Rh30 protein remains unclear (Nawata et al., 2007). In addition to the vertebrate Rh proteins, increased attention has been paid to the presence of Rh homologs in lower invertebrates. Rh genes have been cloned in a few arthropods, such as Callinectes sapidus (AAM21147), Metacarcinus magister (JF276906), Carcinus maenas (AF364404), Bombyx mori (NM_001043888), Apis mellifera (NM_001040265) and Drosophila melanogaster (AF193812). Furthermore, molecular evidence indicates only one isoform of Rhesus-like protein is present in crabs. Rhesus-like proteins are a group of glycosylated proteins of molecular weight ∼50 (Rh-50 proteins); they have 12 transmembrane segments indicative of a transport function and show limited homology to microbial Amt proteins first noticed by Marini et al. (1997). Although it is believed that Rh proteins can implement transport activity, their physiological substrate remains controversial. Physiological and structural evidence revealed that Rh proteins are gas channels for NH₃ (Soupene et al., 2001, 2002; Khademi et al., 2004; Nawata et al., 2010), which is contrary to the widely held view that Rh proteins act as an active NH₄⁺ concentrator (Ludewig et al., 2001). It may be that Rh proteins act as both, depending on the circumstances and species (Weihrauch et al., 2004). In parallel, Rh proteins are proposed to be gas channels for CO₂, rather than functioning as equivalents of Amt proteins (Li et al., 2007; Wright and Wood, 2009).

Because benthic habitats can increase the possibility of exposure to ammonia, ammonia transportation mechanisms of crabs are considered to be of relevance for all aquatic organisms. Understanding how Portunus trituberculatus transports ammonia is essential for discovering how the ammonia-rich benthic water of crab habitats affects larval development and adult emergence. There are at least four ammonia transportation and metabolic pathways in crustaceans: excess internal ammonia is probably converted to glutamine (Gln) by the coupling of glutamate dehydrogenase (GDH) and glutamine synthetase (GS) (Anderson et al., 2002); urea is generated via the ornithine–urea cycle (OUC) (Chen and Chen, 1997) to reduce ammonia toxicity; a small amount of uric acid has also been detected in crustaceans, such as Marsupenaeus japonicus (Lee and Chen, 2003), Nephrops norvegicusto (Bernasconi and Uglow, 2011) and Homarus americanus (Battison, 2013); and as a major ammonotelic animal, active branchial ammonia excretion mechanisms are present in crabs to prevent unfavorable passive influxes (Weihrauch, 1999). Ammonia makes up around 86% of the total excreted nitrogen in Carcinus maenas (Needham, 1957), and 95% in Palaemonetes varians (Snow and Williams, 1971) and Crangon crangon (Regnault, 1983). Numerous studies have demonstrated the ability of Rh proteins to transport ammonia not only in aquatic (seawater and freshwater) animals but also in mammals, and these Rh proteins exhibit high similarity in protein length, amino acid composition and predicted secondary structure, as is known to mediate transmembrane ammonia transport (Braun et al., 2009; Weihrauch et al., 2009; Weiner and Verlander, 2010). Furthermore, research work in our laboratory (Liu et al., 2014; Ren et al., 2015) has found that the mRNA expression level of Rh protein was significantly upregulated under ammonia exposure in P. trituberculatus, which indicates that Rh protein may participate in ammonia transport. Importantly, knockdown of Rh protein showed a reduction in the rate of ammonia excretion in Aedes aegypti and Caenorhabditis elegans (Durant et al., 2017; Adlimoghaddam et al., 2016). Aquaporins (AQPs) are commonly recognized as a mediator for water transport. In mammals, 13 members of the AQP family have been identified, including AQP0–9 (initial members) and AQP10–12 (new members) (Ishibashi, 2009). The human aquaporins, such as AQP3, AQP7, AQP8, AQP9 and possibly AQP10, are permeable to ammonia (Litman et al., 2009). However, only a few AQPs have been obtained from crustaceans. The AQP of P. trituberculatus is probably AQP1, which shares 88% sequence identity with Callinectes sapidus AQP1. Classification of AQP subtypes in aquatic animals may differ from that in mammals. AQP1 in Penaeus monodon and Danio rerio established that it was permeable to ammonia (Peaydee et al., 2014; Chen et al., 2010). In addition, inhibitor experiments of C. maenas suggest that ammonia is transported from the hemolymph into the cytoplasm of the gill epithelial cells via the basolaterally localized Na⁺/K⁺-ATPase (NKA) and K⁺ channels (Weihrauch et al., 1998). Originally, transport experiments employing amiloride demonstrated a role for an apically localized Na⁺/H⁺-exchanger (NHE) in the ammonia excretion of marine crabs (Weihrauch, 1999; Hunter and Kirschner, 1986). However, Na⁺ and NH₄⁺ conductance across the isolated cuticle of C. maenas can be blocked by apically applied amiloride (Onken and Riestenpatt, 2002; Weihrauch et al., 2002). Importantly, application of S3226 (a specific inhibitor of NHE3) to A. aegypti larvae decreased NH₄⁺ efflux at the anal papillae (Chasiotis et al., 2016). Moreover, it was reported that NHE3 is involved in ammonia excretion, with NH₄⁺ substituting for H⁺ in the proximal tubules of the mammalian kidney (Weiner and Hamm, 2006). Therefore, the possibility of NHE participating in ammonia excretion in crustaceans cannot be excluded. Furthermore, a microtubule-dependent ammonia excretion mechanism with V-type H⁺-ATPase (V-ATPase) and vesicle-associated membrane protein (VAMP, also called synaptobrevin) was proposed in C. maenas (Weihrauch et al., 2002), and it was also verified in P. trituberculatus (Ren et al., 2015). Specifically, V-ATPase probably serves in acidification of intracellular vesicles, while VAMP is an important component of the synaptic vesicle-docking fusion complex (soluble N-ethylmaleimide-sensitive factor attachment protein receptor, SNARE), which is involved in membrane fusion and vesicle exocytosis (Moghaddam et al., 2010). As mentioned above, a complex ammonia detoxification metabolic network exists in crabs (Weihrauch et al., 2004). However, the significance of Rh protein in these ammonia excretion and metabolic mechanisms remains elusive.

The swimming crab P. trituberculatus (Miers 1876) (Crustacea: Decapoda: Brachyura), a benthic animal, is a crucial aquaculture crab and widely distributed in China with 125,317 tons of aquaculture production and 542,070 tons of capture production in 2016 (Ministry of Agriculture Fisheries and Fisheries Administration, 2017). Ammonia is a major environmental limitation for crabs, and ammonia content may accumulate over time as a result of the decomposition of residual bait and excretion by cultured aquatic animals (Camargo and Alonso, 2006). It is usually present in an ionized (NH₄⁺) and un-ionized (NH₃) state in water. As a weak base (pK_a of 9.3–9.4), it occurs predominately in its protonated form, NH₄⁺, in physiological solutions. However, the higher lipid solubility of NH₃ makes it more diffusible through phospholipid bilayers (Cameron and Heisler, 1983). More importantly, ammonia can be excreted directly with the help of Rh protein, which plays an important role in the detoxification mechanisms of ammonia. In this study, a novel full-length Rh protein cDNA of P. trituberculatus was cloned (abbreviated to PtRh), and the tissue distribution of this gene was analyzed. Based on previous studies in our laboratory, double-stranded RNA (dsRNA)-mediated PtRh RNA interference (RNAi) was employed. Ammonia excretion rate was determined, while changes in ammonia excretion-related gene expression and enzyme activity as well as hemolymph ammonia and Gln levels were analyzed. These results not only provide molecular evidence for the existence of Rh protein but also explore the critical role of Rh protein in ammonia detoxification mechanisms of P. trituberculatus.

Fresh and vibrant P. trituberculatus (mean mass 120±7.8 g) were obtained from Nanshan market (Qingdao, China) and acclimated in tanks (60 cm×40 cm×30 cm) with 60 l aerated seawater (salinity 31‰, pH 8.1, ammonia 0.75 µmol l⁻¹) at 20±0.5°C for 1 week prior to the experiment. Seawater in tanks was aerated continuously using air-stones and the tank was separated into six cubicles with perforated plastic plates, with one crab placed in each cubicle. During the acclimation period, half of the water was renewed in each tank twice daily, and the photoperiod was maintained on a 12 h:12 h light:dark cycle. The crabs were fed twice daily with fresh clams Ruditapes philippinarum by 10% body mass of crabs. All crabs were fasted for 2 days before the experiment to prevent any change in the level of ammonia due to metabolic ammonia production following feeding. All procedures were in accordance with the Guidelines of the Chinese Association for Laboratory Animal Sciences.

Crabs were placed on ice for about 30 min, then gills (both anterior and posterior), hepatopancreas, stomach, heart and muscle were collected using RNase-treated scissors and forceps, and then frozen in liquid nitrogen. Samples used for RNA extraction were lysed with RNAiso Plus reagent (Takara, Dalian, China) in 1.5 ml RNase-free tubes. After full mixing, the lysis samples were centrifuged at 12,000 g for 15 min at 4°C and the supernatant stored at −80°C until processed. Samples used for enzyme activity assays were milled in liquid nitrogen, and 80–100 mg tissue powder was quickly placed into 2 ml centrifuge tubes and stored immediately at −80°C. Hemolymph samples were obtained from the arthrodial membrane at the third walking leg using a sterilized syringe with an equal volume of anti-coagulant (450 mmol l⁻¹ NaCl, 100 mmol l⁻¹ glucose, 30 mmol l⁻¹ trisodium citrate, 26 mmol l⁻¹ citric acid, 10 mmol l⁻¹ EDTA, pH 7.45), modified from the anti-coagulant devised by Söderhäll and Smith (1983). Samples were immediately centrifuged at 700 g for 10 min at 4°C and the supernatant was collected as the serum sample and frozen at −80°C until analysis. The pellet was suspended in 1 ml RNAiso Plus reagent (Takara) and stored at −80°C until total RNA extraction.

Total RNA was extracted from 100 mg gill powder using RNAiso Plus reagent (Takara), according to the manufacturer's instructions. RNA quantity, purity and integrity were examined by both native RNA electrophoresis on 1.0% agarose gel and UV absorbance ratio at 260 nm and 280 nm (Multiskan Go 1510, Thermo Scientific, Vantaa, Finland). In order to obtain the full-length cDNA for Rh protein, 5′ and 3′ RACE was carried out on P. trituberculatus cDNA using the SMART™ RACE cDNA amplification kit (Clontech, Mountain View, CA, USA). Gene-specific primers Rh-3′GSP and Rh-5′GSP were designed based on the PtRh partial sequence (Table S1). Amplification was conducted on cDNA using 1 μmol l⁻¹ of universal primer coupled with 1 μmol l⁻¹ of gene-specific primer. The cDNA preparation methods and RACE amplification conditions were as suggested by the manufacturer. The touchdown PCR profile was as follows: conducted at 95°C for 3 min followed by two cycles of 95°C for 40 s, 62°C for 50 s and 72°C for 2 min, two cycles of 95°C for 40 s, 60°C for 50 s and 72°C for 2 min, two cycles of 95°C for 40 s, 58°C for 50 s and 72°C for 2 min and 30 cycles of 95°C for 40 s, 56°C for 50 s and 72°C for 2 min; total of 36 cycles.

The amino acid sequence, protein molecular mass and isoelectric point (pI) of the PtRh gene were predicted using DNAstar software. The homology search was performed with the BLAST program (GenBank; https://blast.ncbi.nlm.nih.gov/Blast.cgi). The PFAM domains and transmembrane region were predicted by SMART (http://smart.embl-heidelberg.de/). Secondary structure of the PtRh protein was analyzed using GOR IV Method (https://npsa-prabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_gor4.html). The 3D protein structure model was generated by SWISS-MODEL Workspace (https://swissmodel.expasy.org/interactive). Multiple protein sequence alignments were performed with ClustalW by the MegAlign program (DNASTAR, Inc.). The mature peptides of PtRh were used to construct phylogenetic trees by the neighbor-joining method using MEGA 6.0 (Arizona State University, USA). Aligned sequences were bootstrapped 10,000 trials by Seqboot.

The dsRNA sequences for PtRh and green fluorescent protein (GFP; as a non-specific dsRNA control) were prepared with T7 RiboMAX Express RNAi Kit (Promega, Madison, WI, USA) (Table S1). Briefly, partial sequences of PtRh and GFP (from a pEGFP-C1 vector, Clontech) were generated and amplified by PCR as linearized DNA templates. Then, the T7 promoter sequence was incorporated into these linearized DNA templates by performing PCR with the following specific primer sets: dsRh(T7)-F/dsRh-F and dsRh(T7)-R/dsRh-R; dsGFP(T7)-F/dsGFP-F and dsGFP(T7)-R/dsGFP-R. Finally, the T7 RiboMAX Express RNAi Kit was used to synthesize the dsRNAs according to the manufacturer's instructions. The corresponding single-stranded RNAs were mixed and annealed by incubation at 70°C for 20 min to become dsRNA, followed by slow cooling to room temperature for 30 min. After digestion with DNase and RNase, dsRNA was extracted and purified by isoamyl alcohol extraction. The integrity of the dsRNA was checked on an agarose gel, and the concentration of dsRNA was checked with a UV spectrophotometer (Multiskan Go 1510, Thermo Scientific). The final dsRNA products were stored at −80°C before being used in the subsequent in vivo experiments. Crab saline [440 mmol l⁻¹ NaCl, 11 mmol l⁻¹ KCl, 13.3 mmol l⁻¹ CaCl₂, 26 mmol l⁻¹ MgCl₂, 26 mmol l⁻¹ Na₂SO₄ and 10 mmol l⁻¹ Hepes (Solarbio, Beijing, China), pH 7.4 with NaOH; Duan and Cooke, 1999] and dsGFP were injected as controls. For in vivo RNAi, each crab was injected via the arthrodial membrane of the third walking leg using a 1 ml syringe with 50 µg (dissolved in 100 µl crab saline) of PtRh or GFP dsRNA after a pre-experiment with different injection doses. The pre-experiment also showed that PtRh RNAi had no significant effect on the expression of related genes (in addition to PtRh) in normal seawater, which may be caused by a lack of external ammonia pressure.

According to the ammonia-N safety concentration of benthic seawater in aquaculture ponds measured by Chen et al. (1988) and Zhong et al. (1997), and the ammonia concentration in shrimp and crab polyculture ponds (∼74.8 µmol l⁻¹ in Rizhao, China; L.S., L.P., H.W. and X.Z., unpublished), which is more in line with the conditions of the crab habitat, 93.5 µmol l⁻¹ ammonia concentration was chosen for the experiment. Ammonia concentration was prepared by adding 18.7×10⁴ µmol l⁻¹ NH₄Cl stock solution into the seawater. Other environmental conditions were similar to those of the acclimation period. In addition, the pH of the ammonia treatment groups changed slightly, which was considered negligible. Animals were randomly classified into three groups with three replicates. Each replicate containing five crabs at any sampling time, and there was no death throughout the experiment. Gills and hemolymph were collected at 0, 3, 6, 12, 24 and 48 h post-injection and simulated benthic seawater (93.5 µmol l⁻¹ ammonia) challenge. The methods for gill and hemolymph sampling are detailed above in ‘Tissue preparation’. Quantitative real-time PCR (qPCR) and reverse-transcription PCR (RT-PCR) were used to detect the efficiency of the RNAi at 24 h post-injection. All molecular reactions were carried out in triplicate using independently extracted RNA samples of gills. Hemolymph sample was used to determine the nitrogen content of compounds.

Ammonia excretion rates of intact crabs were determined according to the method of Martin et al. (2011) and Cruz et al. (2013). Crabs were weighed immediately before the onset of each excretion experiment and randomly classified into three groups with three replicates. Individual crabs were transferred into separate aquaria after crab saline, GFP or PtRh dsRNA injection, each holding 4 l of aerated seawater that was enriched with 93.5 µmol l⁻¹ NH₄Cl. During each experiment, animals were kept in the same environmental conditions as above. Water samples (10 ml) were taken from the containers prior to the addition of the animal to establish background ammonia levels. Measurements for excretion rates started 5 min after transfer, as preliminary experiments showed an enhanced ammonia excretion immediately after the crabs were transferred into the tanks, which may be due to imposed stress during handling. Crabs were kept for 24 h in the containers before tank water samples were collected. Ammonia concentration in the water samples was measured by hypobromite oxidation (State Oceanic Administration, 2007). Whole-animal ammonia excretion rates (J) were calculated as follows: J=[(C_t2−C_t1)×V]/(t×FM), where C_t2 is the ammonia concentration in the container at the end of the measuring period (µmol l⁻¹); C_t1 is the ammonia concentration in the container 5 min after the crabs were transferred (µmol l⁻¹); V is the volume of the tank water (liters); t is the entire sampling period (h); and FM is the fresh mass of the experimental crab (g).

Hemolymph ammonia content was determined according to the instructions of the hemolymph ammonia assay kit (no. A086, Nanjing Jiancheng Bioengineering Institute, China). Hemocyanin was precipitated and the enzyme activity was disrupted with a protein precipitant to prevent the production of free ammonia, while most of the color-interfering substances were removed. Ammonia was prepared in protein-free filtrate by the Berthelot reaction, and then the hemolymph ammonia content was measured by comparison with the standard solution. In order to measure Gln content, an equal volume of acetonitrile (HPLC grade) was pipetted into the serum sample and then the sample was shaken and subsequently centrifuged at 3000 g for 10 min at 4°C and the supernatant was retained. Gln content was assayed by HPLC (LC-20A, Shimadzu, Kyoto, Japan) equipped with a Zorbax Eclipse XDB-C18 column (4.6×250 mm, Agilent, Santa Clara, CA, USA) at 35°C (Liang et al., 2003).

To measure the activity of GDH and GS, 100 mg gill powder was homogenized in 500 µl of ice-cold 25 mmol l⁻¹ Tris buffer (pH 7.5) containing 1 mmol l⁻¹ dithiothreitol (DTT), 1 mmol l⁻¹ Na₂EDTA and 0.2 mmol l⁻¹ phenylmethylsulfonyl fluoride (PMSF) at 24,000 rpm for 60 s at 4°C. The homogenate was then centrifuged at 10,000 g for 15 min and the supernatant was used for the assays. The activity of GDH was determined according to the method of Ellis and Goldberg (1972) and expressed as μmol NADH utilized min⁻¹ mg⁻¹ protein by monitoring activity before and after 5 min of 37°C incubation at 340 nm. GS activity was assayed and expressed as μmol γ-glutamylhydroxamate formed min⁻¹ mg⁻¹ protein by the method of Shankar and Anderson (1985) at 500 nm. The activities of NKA α-subunit and V-ATPase subunit B were evaluated according to the methodology of Zare and Greenaway (1998). A 100 mg sample of gill powder was homogenized with 4 ml buffer [250 mmol l⁻¹ sorbitol (Solarbio), 6 mmol l⁻¹ EDTA, 25 mmol l⁻¹ Tris-acetate buffer, 0.1 mmol l⁻¹ DTT, 0.2 mmol l⁻¹ PMSF, 100 U ml⁻¹ aprotinin (Sigma), pH 7.4] on ice. The homogenate was centrifuged at 10,000 g for 30 min at 4°C. The difference in measured activity between tubes following addition of solution A (6 mmol l⁻¹ MgCl₂, 100 mmol l⁻¹ NaCl, 10 mmol l⁻¹ KCl and 25 mmol l⁻¹ Tris-acetate, pH 7.4) and solution B [addition of 5 mmol l⁻¹ ouabain (Sigma) to inhibit NKA] was taken to represent the activity of NKA. To measure the activity of V-ATPase, after adding assay buffer (34.8 mmol l⁻¹ Hepes, 173.9 mmol l⁻¹ KCl, 6.95 mmol l⁻¹ MgCl₂ and 0.01 mmol l⁻¹ orthovanadate, pH 7.4), the difference in measured activity between tubes following addition solution A (added DMSO) and solution B [added 10 µmol l⁻¹ bafilomycin A1 (Sigma) to inhibit V-ATPase] was taken to represent the activity of V-ATPase. The supernatants after reactions were assayed for inorganic phosphate (Pi) following the ammonium molybdate ascorbic acid method. The specific activity of the ATPase was calculated as μmol Pi released mg⁻¹ gill protein h⁻¹. Total protein was determined using bovine serum albumin (BSA, Solarbio) as the standard according to the method of Bradford (1976).

Total RNA was extracted from 80–100 mg of tissue powder using RNAiso Plus reagent (Takara). After analysis of quality and quantity, cDNA was synthesized based on 1 μg of total RNA using PrimeScript RT reagent kit with gDNA Eraser (Takara). The mRNA expression of the PtRh gene in various tissues including gill, hepatopancreas, heart, muscle, hemocytes and stomach was determined by SYBR Green qPCR on a PikoReal 96 Real-Time PCR System (Thermo Scientific). In addition, the mRNA levels of PtRh, K⁺ channel, NHE, VAMP and AQP were evaluated in gills post-dsRNA injection, using sequences obtained from our transcriptome library of P. trituberculatus (SRP018007). Primers used for qPCR (Table S1) were designed by Primer Premier 5.0 software after verification and synthesized by BGI (Beijing, China). β-Actin and RpL8 (ribosomal protein L8) were selected as candidate housekeeping genes. The stability of β-actin and RpL8 genes was evaluated using the BestKeeper method (Pfaffl et al., 2004). During the experiment, RpL8 was found to have lower variation than β-actin and indeed was equally expressed in all investigated tissues and in the knockdown mutant; therefore, it was selected as the housekeeping gene for the rest of the analysis (Tables S2 and S3). PCR amplification was carried out in triplicate on a 96-well rotor in a 10 μl system containing 5 μl of 2× SYBR Premix Ex Taq, 1 μl of cDNA template, 0.2 μl of 10 μmol l⁻¹ forward and reverse primer and 3.6 μl of sterile water. The qPCR program was executed at 95°C for 30 s followed by 40 cycles of 95°C for 10 s, 57°C for 20 s and 72°C for 30 s. Melt curve analysis (60–95°C) was performed on the PCR products at the end of each run to ensure that only one PCR product was amplified and detected. A control lacking cDNA template was included in qPCR analysis to determine the specificity of target cDNA amplification. The relative expression ratio (R) was calculated using the equation: R=(E_target)^{ΔCP_target(control−sample)}/(E_ref)^{ΔCP_ref(control−sample)} (Pfaffl, 2001). PCR efficiency (E) was determined by running standard curves with 10-fold serial dilutions of cDNA templates, and calculated according to E=10^(−1/slope) (Rasmussen, 2001).

All data are presented as means±s.e.m. A one-way analysis of variance (ANOVA) was utilized to examine the differences between control and treated groups with SPSS version 17.0 (SPSS Inc.). Significant differences were considered at P<0.05. Tukey's test was used to identify differences between treatments when significant differences were found.

The full-length cDNA of PtRh protein was obtained by RACE. The complete sequence of PtRh cDNA was 1855 bp, containing a 5′ untranslated region (UTR) of 184 bp and an open reading frame (ORF) of 1437 bp, followed by a 3′ UTR of 234 bp. The ORF encoded a protein of 478 amino acid residues with a calculated molecular mass of 52.05 kDa and pI of 5.99. The cDNA sequence of PtRh was submitted to GenBank (accession no. KJ126844).

Application of SMART analysis showed that PtRh is a transmembrane protein with an ammonium transporter domain (57–454 amino acids), which consists of a duplication of two structural repeats of five helices each plus one extra C-terminal helix. The Rh protein has no signal peptide (N-terminal signal peptide) and it has been defined as a channel that spans the membrane 12 times (TMHMM v2.0 program). There were three N-glycosylation sites in PtRh protein, and two N-glycosylation sites in M. magister and C. maenas Rh protein, respectively (NetNGlyc 1.0 Server). Thirteen phosphorylation sites were discovered in PtRh protein, including seven serine sites, two threonine sites and four tyrosine sites; M. magister Rh protein contains seven serine sites, three threonine sites and four tyrosine sites and C. maenas Rh protein has eight serine sites, four threonine sites and four tyrosine sites (NetPhos 2.0 Server). The PtRh sequence contained a putative nuclear localization sequence (NLS) at K²⁷³HKK²⁷⁶ (Fig. 1A). Analysis of the amino acid sequences with GOR IV revealed the secondary structure consists of an alpha helix, extended strand and random coil, accounting for 37.24%, 19.04% and 43.72%, respectively. Analysis of PtRh using the space structure prediction software SWISS-MODEL with the crystal structure of the human Rhesus glycoprotein RhCG (PDB ID: 3hd6) as the template showed a sequence similarity of 51.56%. The space structure of PtRh is composed of an alpha helix and random coil, presenting a homotrimer compared with RhCG (Fig. 1B).

Blast and ClustalW analyses indicated that the deduced amino acid sequence of PtRh showed significant similarity to sequences reported for other Rh proteins. PtRh presented 92.18% identity with C. sapidus partial sequence (AAM21147), 82.85% identity with M. magister Rh sequence (AEA41159), 81.80% with that of C. maenas (AAK50057) and 51.61% with that of Branchiostoma floridae (BAJ10274). Phylogenetic analysis showed that PtRh was clustered together with other crab Rh-like proteins, whereas PtRh was most dissimilar to the AMTs and MEPs (about 14% identity) (Fig. 1C). RhAG/ag, RhBG/bg and RhCG/cg were grouped on other branches, and away from Rh30. Curiously, Erpobdella obscura Rh-like protein (AJO26542) was associated with Rh30.

The results of qPCR showed that expression of PtRh was detected in all six analyzed tissues (gill, hepatopancreas, heart, muscle, hemocytes and stomach) of P. trituberculatus (Fig. 2). Compared with the PtRh expression in heart, the highest levels of PtRh mRNA relative expression were noted in gill (13.7-fold increase). Stomach, muscle, hemocytes and hepatopancreas showed comparable levels of PtRh mRNA expression (2.5-, 2.2-, 1.7- and 1.6-fold) to those in the heart.

To study the effect of silencing PtRh, RNAi was used to suppress its expression in healthy crabs. As shown in Fig. 3, the mRNA level of PtRh was effectively suppressed to 37.1% at 24 h post-injection without ammonia challenge, and there was no significant difference between the two control groups.

In the present study, the gill GDH and GS activity and hemolymph Gln content were changed in the two control groups during the experiment. Compared with the initial level, crabs in the control groups displayed visible changes in gill GDH and GS activity, reaching a peak at 24 and 12 h, respectively. The Gln level in the hemolymph of control groups was higher than that at 0 h, reaching a maximum at 12 h. In PtRh RNAi group, GDH activity increased remarkably after 6 h and a peak 1.90-fold increase versus controls was noted at 24 h (Fig. 4A). There was a similar trend with GS activity and Gln content changes, which became increasingly obvious at 12–48 h post-PtRh dsRNA injection. The highest level of GS activity occurred at 24 h (1.5-fold higher than control) and a peak in Gln content was observed at 24 h (1.3-fold higher than that in the control groups; Fig. 4B,C).

In intact animals, ammonia excretion rates were calculated as 0.2183±0.0046 µmol g⁻¹ FM h⁻¹ in the saline injection group and 0.2117±0.0049 µmol g⁻¹ FM h⁻¹ in the GFP dsRNA injection group (n=5). The ammonia excretion rate in the PtRh dsRNA injection group (0.1356±0.0061 µmol g⁻¹ FM h⁻¹) was significantly lower than that in the two control groups (n=5, P<0.05). The ammonia content in hemolymph of control groups was increased throughout the experimental period and reached a peak value at 24 h (1.6-fold increase) in contrast to that at 0 h. Hemolymph ammonia content in PtRh-silenced crabs was significantly higher than that in the control groups at 6–24 h. The peak value occurred at 24 h (1.2-fold increase versus control) and recovered to the control level at 48 h (Fig. 5A).

mRNA expression levels of PtRh, AQP, K⁺ channel and NHE, and the activity of NKA were investigated in the PtRh-silenced crabs. PtRh expression was significantly knocked down at 6–24 h by PtRh dsRNA in simulated benthic seawater, with the lowest value at 24 h (Fig. 5B). The absence of Rh protein significantly upregulated the expression levels of AQP from 12 to 48 h, reaching a maximum at 24 h (3.0-fold increase versus control) (Fig. 5C). K⁺ channel mRNA expression was dramatically upregulated at 6–48 h and reached a peak at 12 h (5.5-fold increase versus control) (Fig. 5D). A significant decrease in NHE mRNA expression was detected at 6–12 h, after which levels increased markedly again (Fig. 5E). NKA activity in control groups increased over time compared with that at 0 h, and reached a maximum at 24 h (2.3-fold increase). In addition, the effect of PtRh RNAi on NKA activity was discernible from 6 to 48 h, especially at 24 h, where a notable 4.3-fold increase was observed (Fig. 5F).

To demonstrate the existence of a novel ammonia excretion mechanism centered on vesicle transport and exocytosis in P. trituberculatus, V-ATPase activity and mRNA expression of VAMP were investigated. As shown in Fig. 6A, V-ATPase activity in control groups was increased during the entire experiment, and the highest level occurred at 12 h (1.5-fold increase compared with level at 0 h). A significant increase in V-ATPase activity appeared at 6–48 h post-PtRh dsRNA injection and reached a peak value at 24 h (2.8-fold increase). The mRNA expression of VAMP, which shared a similar trend with V-ATPase activity, was markedly upregulated in the PtRh RNAi group from 6 to 48 h and reached a maximum at 24 h (2.2-fold increase) (Fig. 6B).

Since 2000, knowledge regarding the role of Rh glycoproteins in ammonia transport has increased hugely. Crustaceans are almost entirely ammonotelic animals. Therefore, it is of great interest to obtain information on the mechanisms whereby Rh proteins enable ammonia-specific transport. In the present study, a full-length cDNA sequence of PtRh was successfully cloned from P. trituberculatus. The sequence length of PtRh is 478 amino acids, and it has 12 membrane-spanning segments due to an extended amino terminus. An ammonium transporter domain (57–454 amino acids) of Rh protein was annotated by GO:0015696 from SMART analysis. However, when we searched for this information in the GO library, it indicated a function in cation transport and nitrogen compound transport. Rh protein exists as a homotrimer, and each monomer has a hydrophobic central pore through which ammonia transport occurs. In the middle of this central pore is a twin-His site that can transport NH₃, but not NH₄⁺, H₂O, Na⁺ or K⁺ (Weiner and Verlander, 2010). Furthermore, the transport properties of Rh protein may be related to its subtypes and distribution. Caner et al. (2015) reported that RhAG and Rhbg transport NH₄⁺ (and possibly NH₃) on the basolateral membrane, and that Rhcg predominantly transports NH₃ on the apical membrane. The X-ray crystallographic structure of human RhCG (Gruswitz et al., 2010) and physiological studies of fish Rh protein (Perry et al., 2010) indicated that Rh protein functions as a NH₃ channel. Rh protein appears to lack the extracellular vestibule that is present in Amt orthologs, and this absence may result in a decrease in the ammonia affinity of Rh protein compared with Amt protein (Lupo et al., 2007). In addition, the PtRh amino acid sequence showed higher degrees of identity (81.80–92.18%) with C. maenas, M. magister and C. sapidus (partial), clustered together and grouped as a clade with crab Rh proteins. Hitherto, only two full-length sequences of Rh proteins have been reported in crustaceans. A phylogenetic analysis grouped PtRh into crab Rh proteins of the Rh-like protein branch. The phylogenetic tree also showed that PtRh was far from Rh30, and did not belong to RhAG/ag, RhBG/bg or RhCG/cg. PtRh showed very low identity (about 14%) to members of the AMT and MEP family, indicating ammonia transporters are highly species specific. Moreover, E. obscura Rh-like protein (close to Rh30) may not have ammonia transport function.

Expression profiles of PtRh showed that it was broadly expressed in several tissues, indicating that PtRh can be synthesized under non-stimulated conditions. The highest expression of PtRh was observed in the gills, consistent with the previous findings for M. magister Rh protein (Martin et al., 2011) and O. mykiss Rhbg and Rhcg (Nawata et al., 2007). However, the location of Rh proteins in epithelial cells has not been identified, and it is known that Rh proteins transport ammonia in a bi-directional manner (Mayer et al., 2006). Study of PtRh expression in crab tissues is necessary for a full understanding of the tissue-specific regulation of ammonia excretion. This information may be critical for exploring the physiological function of Rh protein.

How do aquatic animals handle excess ambient or internal ammonia? There are numerous strategies, such as glutamine synthesis, urea and uric acid generation (Wright, 1995) as well as direct excretion (Weihrauch et al., 2002). In addition to absolute ammonia excretion, amino acids are the most important nitrogenous waste as they account for 10% of the total excreted nitrogen (Delaunay, 1931). Most amino acids are transaminated to form glutamate firstly, and then transported as glutamine in the hemolymph, and eventually converted to urea for excretion. In crustaceans, urea and uric acid are part of the nitrogenous waste but they are usually excreted in small amounts (below 10%; Corner et al., 1976).

In our previous study, the obvious increase of mRNA expression of Rh protein in 18.7 µmol l⁻¹ and 93.5 µmol l⁻¹ ammonia-treated groups suggests that Rh protein may play an important role in the maintenance or regulation of ammonia content in P. trituberculatus (Liu et al., 2014; Ren et al., 2015). For further study, RNAi-mediated knockdown experiments were performed to evaluate the essential role of PtRh in the ammonia metabolism and excretion mechanism. Notably, both GDH and GS activity in gills increased significantly in the present study, similar to previous reports that the activities of GDH and GS were higher in response to elevated ammonia (Murray et al., 2003; Essex-Fraser et al., 2005; Wright et al., 2007). Additionally, a significant increase of Gln content in hemolymph was detected, which indicated that converting ammonia to Gln may be one of the physiological adaptation mechanisms to cope with an elevated endogenous ammonia level in P. trituberculatus. The ammonia content in hemolymph also increased, indicating that after Rh protein knockdown, the ammonia excretion pathways were not sufficient for ammonia excretion, leading to the synthesis of Gln; thus, we suggest that Rh protein is the core of the ammonia excretion mechanism.

Adlimoghaddam et al. (2016) reported that compared with that in wild-type C. elegans, ammonia excretion rate was lower by approximately 82% in the Rh-like ammonia transporter RHR-2 knockout mutant strain, which is consistent with AeAmt1 and AeRh50 knockdown in A. aegypti (Chasiotis et al., 2016; Durant et al., 2017). In this study, the ammonia excretion rate was also reduced after PtRh RNAi in P. trituberculatus. Similar to the increase in hemolymph ammonia content post-PtRh dsRNA injection, Chasiotis et al. (2016) found that knockdown of AeAmt1 resulted in an increase in hemolymph NH₄⁺ levels. An increase in ammonia levels in the hemolymph is consistent with a decrease in ammonia excretion in the gills and therefore supports the importance of PtRh in ammonia transport. Nevertheless, the crabs were able to maintain hemolymph ammonia content well below ambient ammonia levels, suggesting that, in addition to Rh protein, there are other ways in which they cope with elevated ammonia. Currently, there have been many reports of Rh proteins that could cooperate with several ammonia transporters to transport ammonia through gills, such as AQP, K⁺ channels, NKA and NHE (Peaydee et al., 2014; Weihrauch et al., 1998; Martin et al., 2011). In the present study, the increase in mRNA expression level of AQP suggests that AQP serves as a NH₃ channel, synergized with Rh protein to facilitate ammonia excretion. A similar report appeared for zebrafish Danio rerio, where exposure to high external ammonia resulted in significantly higher AQP1a1 mRNA and protein abundance in Rhcg1-knockdown larvae (Talbot et al., 2015). Theoretical considerations argued that NH₄⁺ in the hemolymph may substitute for K⁺ via a K⁺ channel and NKA, and transport NH₄⁺ across the basolateral membrane into the epithelial cells (Choe et al., 2000; Wood et al., 2013). The K⁺ channel we investigated was classified as a delayed rectifier potassium channel or a potassium voltage-gated channel (K_V), which was a Ba²⁺-sensitive K⁺ channel. A significant increase in K⁺ channel mRNA expression was detected in the PtRh dsRNA injection group, suggesting that the K⁺ channel was also involved in ammonia excretion, which is consistent with previous reports in C. maenas (Weihrauch et al., 1998; Fehsenfeld and Weihrauch, 2016). In the current study, NKA activity was markedly elevated post-PtRh dsRNA injection in P. trituberculatus gills, especially at 24 h, indicating that NKA participated in mediating the transport of NH₄⁺ from hemolymph into gill epithelial cells. Previously, involvement of NKA in ammonia excretion has been confirmed in crustaceans, such as C. maenas (Weihrauch et al., 1998), Cancer pagurus (Weihrauch 1999) and Callinectes danae (Masui et al., 2002). Hunter and Kirschner (1986) reported a substantial reduction in ammonia excretion of approximately 32% and 56%, respectively, in Cancer antenarius and Petrolisthes cinctipes exposed to amiloride. However, amiloride should be used with caution in the study of NHE Because of the sensitivity of the cuticle to amiloride. In addition, Chasiotis et al. (2016) proposed NHE3 played a role in ammonia excretion of A. aegypti. NH₄⁺ can substitute for H⁺ in NHE on the apical membrane (Randall et al., 1999; Edwards et al., 2002). The observed downregulation of NHE in the initial period was consistent with the results in Martin et al. (2011) and Ren et al. (2015) and can be attributed to the reduction in endogenous ammonia efflux to increase intracellular osmotic pressure and to prevent the entry of external ammonia. Owing to the huge endogenous ammonia pressure, the transcript level of NHE was sharply upregulated after 24 h to facilitate ammonia excretion. However, the involvement of NHE in branchial ammonia excretion requires further study. These results suggested that AQP, K⁺ channels, NKA, NHE and Rh protein can regulate ammonia excretion in a coordinated manner across the gills of P. trituberculatus.

In C. maenas and Scylla paramamosain, V-ATPase is distributed throughout the cytoplasm of gill epithelial cells rather than being localized on the apical membrane specifically, indicating that H⁺-ATPase may be associated with cytoplasmic vesicles (Weihrauch et al., 2001; Tsai and Lin, 2007). V-ATPase was thought to acidify NH₃ to NH₄⁺ in intracellular vesicles, and it was further suggested that the NH₄⁺-laden vesicles were then transported along the microtubule network to the apical membrane, where NH₄⁺ was released by exocytosis (Adlimoghaddam et al., 2015). The increase in V-ATPase activity provided evidence for the presence of a vesicular ammonia-trapping mechanism. Together with V-ATPase, VAMP is required to actively excrete ammonia in vesicles for exocytotic release from the gills (Trimble et al., 1988). It has been reported that the vesicle fusion process requires the interaction of three SNARE proteins: syntaxin, a synaptosome-associated protein of molecular mass 25 kDa (SNAP-25) and VAMP (Prashad and Charlton, 2014). For crustaceans, a VAMP sequence was identified in C. maenas (AY035549) and Procambarus clarkii (KF773142). A significant upregulation of VAMP was observed in the present study, which was in agreement with our hypothesis that the exocytotic ammonia excretion mechanism was enhanced if Rh protein was silenced by dsRNA. The possibility of an exocytotic ammonia excretion mechanism should be considered in P. trituberculatus. In this situation, toxic ammonia was trapped in intracellular vesicles rather than diffusing through the entire cytoplasm, which could reduce the damage. Based on the results of this study, we propose a working model of ammonia excretion by gills of P. trituberculatus (see Fig. 7). These studies highlight the multiple pathways for ammonia excretion and detoxification present in P. trituberculatus, which could explain how P. trituberculatus inhabit the ammonia-rich benthic water and how to adapt to the deterioration of water quality in ponds.

In summary, we have isolated and characterized cDNA of Rh protein from P. trituberculatus for the first time, and analyzed its mRNA expression in different tissues. PtRh gene silencing reduced the ammonia excretion rate and affected the expression of ammonia excretion-related genes and enzyme activity in simulated benthic seawater. In addition, absence of PtRh led to a significant increase of the ammonia and Gln content in hemolymph, indicating that the Rh protein plays a central role in the ammonia excretion mechanism. Yet, virtually nothing is known about the localization of the ammonia transporters in crustaceans and the neuroendocrine signaling mechanisms that regulate these transporters. They may turn out to be a fertile ground for further research.

We thank the staff of the Laboratory of Environmental Physiology of Aquatic Animal for their help with sampling and taking care of the crabs.