Physiological and molecular responses of the spiny dogfish shark (Squalus acanthias) to high environmental ammonia: scavenging for nitrogen

The discovery that Rhesus (Rh) glycoproteins are expressed in the gills (Nakada et al., 2007; Nawata et al., 2007) has created a new understanding of the mechanisms of ammonia excretion in teleosts (reviewed by Wright and Wood, 2009; Wright and Wood, 2012; Weihrauch et al., 2009). These channel proteins appear to facilitate the diffusion of NH₃ along P_NH3 (partial pressure of NH₃) gradients (Nawata et al., 2010b) and in the gills they function as part of a metabolon which flexibly couples ammonia excretion to Na⁺ uptake (Tsui et al., 2009; Ito et al., 2013). The theory proposes that the deprotonation of NH₄⁺ as NH₃ enters the apical Rh channel creates a source of protons to fuel Na⁺ uptake via Na⁺/H⁺ exchangers (NHEs) and/or via a coupled Na⁺ channel/V-type H⁺-ATPase system, while at the same time the NH₃ leaving the Rh channel traps protons, thereby creating an external microenvironment in which [H⁺] is less concentrated. This mechanism becomes particularly prominent during chronic exposure to high environmental ammonia (HEA), as evidenced by increased Na⁺ uptake, ammonia excretion against the gradient and increased expression of metabolon components at the gills (Nawata et al., 2007; Zimmer et al., 2010; Sinha et al., 2013). To date, most studies have been on freshwater ammoniotelic teleosts, but there is also evidence that some components of the system are present and respond to HEA in marine ammoniotelic teleosts (Nawata et al., 2010a; Wood and Nawata, 2011), and even in a 100% ureotelic tilapia living in hypertonic Lake Magadi (Wood et al., 2013b).

But what about marine elasmobranchs, which funnel most of their N-waste production into the ornithine–urea cycle (OUC) so as to make urea for osmoregulation (Smith, 1931; Wright, 1995)? These fish appear to be nitrogen limited (Haywood, 1973; Wood et al., 2005; Kajimura et al., 2006; Kajimura et al., 2008), exhibiting elaborate urea-retention mechanisms at the gills (Boylan, 1967; Pärt et al., 1998; Fines et al., 2001; Wood et al., 2013a) and kidney (Schmidt-Nielsen et al., 1972; Morgan et al., 2003a; Morgan et al., 2003b). Nevertheless, there is still substantial urea leakage across the gills, so elasmobranchs are predominantly ureotelic, excreting very little ammonia. Rh proteins are known to be expressed in these fish (Nakada et al., 2010; Anderson et al., 2010; Anderson et al., 2012), but their function has been little studied. Early investigations focused on mechanisms of ammonia excretion at the gills and the possible involvement of branchial Na⁺/NH₄⁺ exchange mechanisms in acid–base balance (Payan and Maetz, 1973; Evans et al., 1979; Evans, 1982; Evans and More, 1988; Claiborne and Evans, 1988). To our knowledge, there has been only one HEA study on a marine elasmobranch (Cameron, 1986), performed long before the involvement of Rh proteins in teleost ammonia excretion was discovered. Cameron (Cameron, 1986) reported that for the same HEA exposure, ammonia built up far more slowly in the plasma of the ray Raja erinacea than in a teleost or a crab, and interpreted this as resulting from low gill permeability to ammonia, analogous to that for urea. This conclusion was supported by a later study which reported that gill ammonia permeability (in the efflux direction) in the dogfish shark was about 22-fold lower than that in standard

teleosts, even more pronounced than the 13-fold lower urea permeability (Wood et al., 1995). Cameron (Cameron, 1986) also speculated that the ray might convert a substantial proportion of the ammonia entry into urea-N through the OUC, but presented no evidence on this point. This idea is in accord with the results of an NH₄Cl infusion experiment on sharks (Wood et al., 1995) as well as several studies on HEA exposure in two ureotelic teleosts, the Magadi tilapia (Wood et al., 1989; Wood et al., 2013a) and the gulf toadfish (Walsh et al., 1990); in all of these, ammonia loading resulted in a substantial rise in urea-N excretion.

The present study examined the responses to prolonged (48 h) HEA exposure in the dogfish shark, Squalus acanthias suckleyi L. Rh cDNAs were cloned, and their mRNA expression tracked in gills, kidney and red blood cells, together with other gill transporters that could be involved directly or indirectly in the metabolon [Na⁺/H⁺ exchanger2 (NHE2), carbonic anhydrase IV (CA-IV), Na⁺/K⁺-ATPase (NKA), V-type H⁺-ATPase]. Ammonia and urea-N fluxes with the environment were measured, together with blood gases, acid–base status, major ions, and plasma ammonia and urea-N concentrations. In light of recent evidence that changes in the transepithelial potential (TEP) across the gills play a significant role in the adaptive response of freshwater teleosts to HEA (Wood and Nawata, 2011; Liew et al., 2013), TEP was also monitored so that both P_NH3 and NH₄⁺ electrochemical gradients across the gills could be calculated.

Our specific hypotheses were that: (i) ammonia influx and build-up in the blood plasma would both occur more slowly relative to previous studies on teleosts; (ii) HEA exposure would induce a marked increase in urea-N excretion; (iii) Rh proteins would be expressed in gills and other tissues, and their mRNA expression would respond to HEA; and (iv) other potential components of the metabolon would respond in a manner coherent with the Rh mRNA response. Some but not all of these hypotheses were confirmed, and our data point to a surprising conclusion – that the gill of the N-limited elasmobranch may be poised to scavenge ammonia-N from the environment.

All concentrations and flux rates of urea have been expressed in units of urea-N (i.e. two N per urea molecule) to allow comparison with ammonia (one N per ammonia molecule).

Under control conditions, dogfish excreted ammonia-N at a very low rate (~25 μmol kg⁻¹ h⁻¹) relative to urea-N (~565 μmol kg⁻¹ h⁻¹). Exposure to 48 h of HEA resulted in a reversal of J_Amm (net ammonia uptake) during the first 36 h and a subsequent re-establishment of net excretion at a rate greater than control values at 36–48 h (Fig. 1A). Urea-N excretion (J_Urea-N) was significantly enhanced by about 90% from 12 to 48 h (Fig. 1B).

Terminal blood samples demonstrated that plasma total ammonia (T_Amm) was significantly elevated from control levels of 42±23 μmol l⁻¹ (N=7) to 1031±111 μmol l⁻¹ (N=12) after 48 h of HEA exposure. However, these plasma T_Amm levels remained significantly lower than simultaneously measured water T_Amm concentrations (1331±113 μmol l⁻¹, N=12) in these same fish. Plasma total urea-nitrogen (T_Urea-N) in the HEA-exposed fish was 641±16 mmol l⁻¹ (N=12), significantly lower than the 788±18 mmol l⁻¹ (N=7) measured in control animals.

Plasma T_Amm rose significantly to levels that closely matched the water T_Amm levels (~1000 μmol l⁻¹) within 2 h of HEA exposure (Fig. 2). From 24 to 48 h, levels in the plasma dropped significantly below those in the water to about 750 μmol l⁻¹, in qualitative agreement with the terminal sampling data of series 1. Nevertheless, plasma T_Amm remained greatly elevated relative to pre-exposure control levels (Fig. 2).

Plasma T_Urea-N remained stable across time with the exception of a significant decrease at 36 h (Table 1), similar to the decrease measured in the sharks exposed to HEA for 48 h in series 1. Cl⁻, Na⁺and Ca²⁺ levels in the plasma were unchanged throughout the 48 h HEA treatment, but Mg²⁺ increased significantly, almost doubling at 36 h (Table 1).

Arterial pH (pHa) and partial pressure of arterial O₂ (Pa_O2) levels (Fig. 3A,D) did not vary significantly across the entire HEA treatment. Partial pressure of arterial CO₂ (Pa_CO2) was significantly elevated at 2 h (Fig. 3B) and arterial bicarbonate concentration ([HCO₃⁻]a) was elevated at 2 and 6 h (Fig. 3C).

Partial pressures of arterial plasma NH₃ (Pa_NH3) and water NH₃ (P_NH3) were calculated from simultaneous measurements of T_Amm and pH in the plasma (Fig. 2, Fig. 3A) and water (Fig. 2), respectively. Water pH remained constant at 7.7–7.8 except at 36 h when it fell significantly to 7.53 (data not shown). Both Pa_NH3 and water P_NH3 increased greatly throughout the period of HEA exposure, and Pa_NH3 remained significantly greater than water P_NH3 throughout the entire 48 h of HEA exposure (Fig. 4). Therefore, the P_NH3 gradient across the gills (ΔP_NH3), which was negative (−11.2 μTorr) under control conditions, became even more negative (up to −263 μTorr at 6 h) from 2 to 36 h of HEA (Table 2), reflecting a tendency for NH₃ to be driven out of the fish. This increase in ΔP_NH3 was driven by the overall increases in T_Amm and the higher pH (by about 0.35 units) in blood versus external seawater.

The Nernst potential for NH₄⁺ (E_NH4⁺) could also be calculated from these measurements. Under control conditions, E_NH4⁺ was negative (−23 mV), suggesting a tendency for NH₄⁺ to be driven out of the fish, but upon exposure to HEA it increased significantly to positive values (~+10 mV) at 24–48 h of HEA exposure, suggesting a tendency for NH₄⁺ to be driven into the fish (Fig. 5). However, this does not take the TEP into account. Indeed, it is the difference between E_NH4⁺ and TEP, and not the E_NH4⁺ alone, which is the true net driving force (F_NH4⁺) for NH₄⁺ across the gills. Therefore, changes in TEP were measured in series 3.

The TEP was ~0 mV under control conditions, but increased significantly to about +30 mV throughout HEA exposure (Fig. 5). Therefore, this large increase in TEP paralleled and essentially negated the rise in E_NH4⁺. As a result, F_NH4⁺ remained negative at about −20 mV throughout the period of HEA, as under control conditions, except at 48 h where it fell to about −10 mV (Table 2). E_NH4⁺ and TEP were measured on different fish so the changes in F_NH4⁺ could not be statistically evaluated. Nevertheless, it is clear that the uptake of ammonia during the first 36 h of HEA exposure (Fig. 1) occurred despite the fact that both the partial pressure gradient for NH₃ (ΔP_NH3) and the electrochemical gradient for NH₄⁺ (F_NH4⁺) remained in the outward direction (i.e. negative; Table 2). Thus, ammonia was taken up from the water against both gradients during HEA exposure.

We identified three Rh glycoprotein cDNA sequences in the dogfish; a partial (1117 bp) Rhag sequence (GenBank accession no. KJ960196) and full-length sequences of Rhbg (GenBank accession no. KJ960197; encoding a 460 amino acid protein) and Rhp2 (GenBank accession no. KJ960198; encoding a 476 amino acid protein). PCR analyses revealed that the both Rhbg and Rhp2 transcripts are ubiquitously expressed (Fig. 6). Protein BLAST analyses revealed that the partial Rhag sequence is most similar (81%) to the holocephalan elephant fish (Callorhinchis milii) Rhag sequence (GenBank accession no. XP007901455). The Rhbg sequence is 70% identical to the Rhbg sequences of the amphibian Xenopus laevis (GenBank accession no. NP001087152), the primitive actinopterygian gar (Lepisosteus oculatus) (GenBank accession no. XP006641869) and the teleost midshipman (Porichthys notatus) (GenBank accession no. A6A93879), and has 12 predicted transmembrane regions (Fig. 7). The Rhbg EST of the elasmobranch skate Leucoraja erinacea (GenBank accession no. FL670153) was not large enough for meaningful sequence comparison to the Squalus acanthias Rhbg sequence, and the same was true for the Rhbg fragment identified in the holocephalan ratfish Hydrolagus colliei (GenBank accession no. HQ852228) by Anderson et al. (Anderson et al., 2012). Rhp2 is highly conserved (95% sequence identity) in the dogfish and another elasmobranch, the houndshark (Triakis scyllium) (GenBank accession no. AB2877007) (Fig. 8).

In the gill, expression of Rhbg (Fig. 9E) and H⁺-ATPase (Fig. 9B) was significantly downregulated after 48 h of HEA by ~5- and 2-fold, respectively. Carbonic anhydrase IV (CA-IV; Fig. 9A), Na⁺/H⁺ exchanger 2 (NHE2; Fig. 9C) and Rhp2 (Fig. 9F) expression levels remained unchanged while Na⁺/K⁺-ATPase (NKA; Fig. 9D) expression was significantly elevated by almost 4-fold at 24 h, but fell back to control levels at 48 h.

In erythrocytes, Rhag mRNA levels decreased significantly by about 2.5- to 3-fold after 24–48 h of HEA (Fig. 10A). In the kidney, Rhbg and Rhp2 mRNA levels were stable during HEA while urea transporter (UT) levels increased significantly by about 5-fold at 24 h (Fig. 10B).

This is the first detailed investigation of the responses of a marine elasmobranch to prolonged HEA exposure. As such, it has confirmed some of our original hypotheses, disproven others, and also provided some completely unexpected findings. In contrast to our first hypothesis, ammonia influx and build-up in the blood plasma of the dogfish were not markedly slower than in comparable previous studies on teleosts, despite the fact that the efflux permeability to ammonia is very low under control conditions. This relates to the unexpected finding that the effective gill permeability to ammonia appears to be strongly rectified, such that permeability in the uptake direction is much greater than permeability in the efflux direction. Thus, ammonia uptake is favoured even in the presence of outwardly directed partial pressure (ΔP_NH3) and electrochemical gradients (F_NH4⁺), such that the gill appears to be set up to scavenge ammonia when it is elevated in the environment. However, our second hypothesis was confirmed: ammonia uptake and the associated rise in plasma T_Amm resulted in a marked elevation of urea-N excretion, probably reflecting increased urea-N synthesis by the OUC. In accord with our third prediction, two Rh proteins (Rhbg and Rhp2) were expressed at the mRNA level in the gill and other tissues and a third (Rhag) was expressed in the red blood cells. Furthermore, branchial Rhbg and erythrocytic Rhag responded to HEA exposure, but with a downregulation rather than the expected upregulation seen in most previous studies on teleosts. Finally, other components of the metabolon involved in the coupling of ammonia excretion and Na⁺ uptake in freshwater fish gills did respond at the mRNA level to HEA exposure, but not in a manner comparable with previous findings on teleosts. Clearly, the responses of this marine elasmobranch to HEA are fundamentally different from those in freshwater teleosts, and also appear to differ from the limited data available to date on seawater teleosts.

Within 2 h of acute HEA exposure, plasma T_Amm in the dogfish shark rose to levels equal to water T_Amm (Fig. 2). This was very different from the response seen in the only previous HEA study on a marine elasmobranch, the little skate (R. erinacea), where plasma T_Amm rose very slowly, reaching only about 50% of seawater T_Amm levels by 11 h (Cameron, 1986); Cameron attributed this pattern to low gill permeability to ammonia. In comparison to previous HEA studies on a variety of freshwater and marine teleosts (Cameron and Heisler, 1983; Cameron, 1986; Claiborne and Evans, 1988; Wilson and Taylor, 1992; Wilson et al., 1994; Nawata et al., 2007; Nawata et al., 2010a; Zimmer et al., 2010; Wood and Nawata, 2011; Liew et al., 2013), the rate of plasma T_Amm rise in the dogfish was comparable or faster, but the subsequent pattern was different. In the teleosts, plasma T_Amm never reached external water T_Amm levels, but rather was maintained at a plateau or slow decline significantly below the latter. In some of these teleost studies, flux rates of ammonia (J_Amm) and/or the P_NH3 gradients across the gills (ΔP_NH3) were also measured, showing that net ammonia excretion was re-established relatively rapidly (within 3–12 h) despite the fact that ΔP_NH3 remained reversed, suggesting active ammonia excretion across the gills. In only one study (Wood and Nawata, 2011) was the electrochemical gradient for NH₄⁺ (F_NH4⁺) across the gills also measured, but the conclusion was the same: J_Amm was quickly re-established despite the fact that both F_NH4⁺ and ΔP_NH3 remained reversed during HEA exposure. In contrast, the dogfish shark continued to take up ammonia for 36 h and net outward J_Amm was only re-established at 36–48 h (Fig. 1A), yet both ΔP_NH3 and F_NH4⁺ were not reversed and remained negative (i.e. outwardly directed) throughout the entire HEA period (Table 2). Cardiovascular parameters were not measured in this study, so it is possible that some of the temporal changes and/or differences from teleosts could reflect responses in gill blood flow rate or perfusion pattern.

Thus, while ammonia efflux permeability of the dogfish gill is low (Wood et al., 1995), and the influx permeability of the skate gill may also be low (Cameron, 1986), the influx permeability of the dogfish gill during HEA exposure certainly is not. Indeed, ammonia is taken up for 36 h under circumstances where this should not occur passively (i.e. both ΔP_NH3 and F_NH4⁺ are outwardly directed; Table 2). We speculate that active ammonia uptake across the gills (i.e. ‘ammonia scavenging’) may occur when environmental ammonia levels are sufficiently high, and that this is a manifestation of a system which is normally used for active ammonia retention under control conditions. This system would be adaptive for an animal that needs to continuously make urea and that appears to be nitrogen limited in nature (Haywood, 1973; Wood et al., 2005; Kajimura et al., 2006; Kajimura et al., 2008). In this respect, sharks, which encounter and ingest pelagic prey only at irregular intervals, may differ from skates, which can continuously browse on benthic invertebrates. While this idea of active ammonia uptake by a fish, and therefore rectified branchial ammonia permeability, is unusual, there is a precedent in the urea-N retention mechanism at elasmobranch gills (Boylan, 1967; Wood et al., 1995; Pärt et al., 1998; Wood et al., 2013a). Urea-N retention appears to involve an active transporter, and Fines et al. (Fines et al., 2001) have presented evidence in Squalus acanthias for a Na⁺ versus urea exchanger powered by the electrochemical Na⁺ gradient across the basolateral membrane of gill cells, which would scavenge urea as it diffuses out. A comparable mechanism for ammonia scavenging such as a basolateral Na⁺ versus NH₄⁺ exchanger (NHE?) would minimize ammonia excretion under normal circumstances, thereby explaining the low efflux permeability yet high influx permeability, and would be recruited to actively take up ammonia from HEA water. An alternative or additional possibility would be active apical NH₄⁺ uptake via substitution on the K⁺ site of H⁺/K⁺-ATPase, which has been identified in elasmobranch gills (Choe et al., 2004). NH₄⁺ is well known to be able to substitute for K⁺ in a variety of transport systems, including members of this family of transport ATPases (Codina et al., 1999; Cougnon et al., 1999; Crambert, 2014).

How relevant is the present ammonia exposure scenario to real world conditions? The water T_Amm level used in our tests (1000 μmol l⁻¹) was chosen for comparison to the many previous HEA studies on teleosts using a similar level, as cited above. Water quality objectives for marine waters set by regulatory authorities are typically about half this value (reviewed by Ip et al., 2001). In the open oceans, ammonia concentrations are in the low micromolar range, but in coastal waters and estuaries may rise to the concentration used here as a result of anthropogenic input (Eddy, 2005), and these inputs are steadily rising as a result of non-point source runoff from agriculture (Howarth et al., 2002) and intensive aquaculture operations (Tovar et al., 2000). Furthermore, sharks appear to be attracted to ammonia (Mathewson and Hodgson, 1972; Gilbert, 1977). Thus, while the present HEA exposure may represent an extreme, it is not unrealistic, and elasmobranchs may well exhibit ammonia scavenging at lower water T_Amm levels.

Under control conditions, TEP was close to 0 mV in S. acanthias, in agreement with previous measurements in marine elasmobranchs (Bentley et al., 1976; Evans, 1982; Evans et al., 1982; Evans and More, 1988). A large and persistent rise (approximately +30 mV) in TEP during HEA exposure paralleled the simultaneous rise in E_NH4⁺, such that F_NH4⁺ remained virtually unchanged (Fig. 5, Table 2). This phenomenon of a more positive TEP (inside) during HEA has now been seen in three freshwater teleosts [trout, goldfish and carp (Wood and Nawata, 2011; Liew et al., 2013)], but curiously did not occur in seawater trout (Wood and Nawata, 2011). However, the rise in TEP in the dogfish in seawater was greater than that in any of these studies. Potential mechanisms behind the phenomenon have been discussed (Wright and Wood, 2012) but the proximate cause remains unknown. Had this not occurred, F_NH4⁺ would have favoured NH₄⁺ uptake to an even greater extent during HEA. Therefore, it may be interpreted as a mechanism to prevent NH₄⁺ uptake in teleosts and to modulate it in this elasmobranch so that plasma T_Amm levels do not overwhelm the capacity of the OUC to consume ammonia.

In accord with our prediction, HEA exposure induced a marked increase in J_Urea-N (Fig. 1B). Similar responses have been seen in two ureotelic teleosts which express the OUC (see Introduction) and in an earlier study in which S. acanthias were infused with NH₄Cl (Wood et al., 1995). This response became significant at 12–48 h of HEA exposure and probably represents an increase of urea production rate by the OUC. It may be the reason why plasma T_Amm levels were subsequently reduced significantly below water T_Amm concentrations (Fig. 2). This response was also suggested, but not measured, by Cameron (Cameron, 1986) in his HEA studies on R. erinacea as an explanation (alternative or additional to low gill ammonia permeability) for the blunted rise in plasma T_Amm seen in that species. The very high affinity of shark glutamine synthetase for ammonia, together with its intramitochondrial location in association with carbamoyl phosphate synthetase III (Shankar and Anderson, 1985) would allow it to act as an ammonia trap, funnelling this valuable nitrogen source directly into the OUC for urea-N synthesis.

The amount of ‘extra’ urea-N excreted over the 48 h period (Fig. 1B) was about 2.5-fold greater than the amount of ammonia-N taken up (Fig. 1A). Why does the shark not save this urea-N if it is so valuable? One answer may be that its storage anywhere in the body would have unfavourable osmotic effects, raising osmolality substantially above environmental levels, and inducing unwanted water influxes. An alternative possibility is that not all of this ‘extra’ urea-N originated from new synthesis, but rather from a decline in body stores. The terminal samples of series 1 indicated a drop of 147 mmol l⁻¹ of plasma T_Urea-N, whereas in series 2 the drop was a more modest 43 mmol l⁻¹ (Table 1). If distributed throughout the whole body water pool, even this smaller decrease would be almost large enough to explain the 2.5-fold discrepancy. Overall, we conclude that plasma T_Amm and therefore intramitochondrial T_Amm levels during HEA are reset to levels where urea-N synthesis is over-stimulated, resulting in extra urea-N being released to the environment, presumably through a downregulation of the retention mechanisms at the gills and kidney. A similar ‘over-release’ of urea-N by both these routes was seen in dogfish infused with NH₄Cl (Wood et al., 1995). The observed increase in UT levels in the kidney (Fig. 10A) may have contributed to this response.

Three Rh cDNAs (Rhbg, Rhp2, Rhag) were cloned. Rhbg and Rhp2 mRNA were expressed in every tissue examined – red blood cells, brain, gill, intestine, kidney, liver, muscle, rectal gland, cardiac stomach and pyloric stomach (Fig. 6) – and Rhag was expressed in red blood cells. In elasmobranchs, there are only two previous reports of Rh proteins – Rhbg in the intestine, rectal gland and kidney (the only tissues examined) of the little skate, Leucoraja erinacea (Anderson et al., 2010), and Rhp2 in the kidney of the houndshark, T. scyllium (Nakada et al., 2010). In contrast to the present findings, Nakada et al. (Nakada et al., 2010) reported that Rhp2 was present only in the kidney, but this conclusion was based on northern blot screening, which is less sensitive than the qPCR employed here.

As branchial Rhp2 expression did not respond to HEA exposure (Fig. 9F), it is difficult to draw any conclusions on its function. However, the widespread tissue distribution of Rhp2 suggests that it could be involved in processes other than ammonia reabsorption from the urine in the kidney tubules suggested by Nakada et al. (Nakada et al., 2010). Based on analogy to other fish (Wright and Wood, 2009; Wright and Wood, 2012; Weihrauch et al., 2009), dogfish Rhbg is probably a basolateral ammonia transporter in the gills and elsewhere. Rhbg was expressed at high levels in the gut of the skate, and Anderson et al. (Anderson et al., 2010) speculated that it is involved in taking up ammonia generated in the chyme for urea-N synthesis by the intestine. While we were unsuccessful in finding a sequence for Rhcg (the putative apical ammonia transporter), subsequent searches of the Squalus acanthias transcriptome suggest that it is present (A. Clifford and G. Goss, personal communication). Rh proteins are bidirectional transporters (e.g. Nawata et al., 2010b). Thus, in the shark, their major roles may not be in ammonia excretion, but rather in ammonia scavenging at the gills and gut, transport through the bloodstream to internal sites of urea-N synthesis (liver, muscle and intestine), and transcellular or intracellular transport at these sites to intramitochondrial glutamine synthetase (Shankar and Anderson, 1985) for conversion into urea-N by the OUC.

Apart from NKA (Fig. 9D), components of the metabolon that flexibly couple branchial ammonia excretion to Na⁺ uptake in teleost fish (see Introduction) did not exhibit the expected upregulation during HEA exposure in the shark (Fig. 9), in contrast to findings in the former (Nawata et al., 2007; Zimmer et al., 2010; Wood and Nawata, 2011; Wood et al., 2013b; Sinha et al., 2013). The transient NKA upregulation at 24 h in the shark (Fig. 9D) could be associated with maintaining the Na⁺ electrochemical gradient for ammonia scavenging by a basolateral Na⁺ versus NH₄⁺ exchanger, as speculated earlier.

It is particularly notable that branchial Rhbg expression was progressively downregulated during HEA exposure (Fig. 9E), and the same trends were seen in branchial H⁺-ATPase (Fig. 9B) and erythrocytic Rhag expression (Fig. 10A). This pattern of downregulation of Rhbg and Rhag has been seen previously in only one teleost exposed to HEA, the marine pufferfish (Takifugu rubripes), but in that study gill H⁺-ATPase was upregulated (Nawata et al., 2010a). In the pufferfish study, the Rh responses were interpreted as a mechanism to reduce ammonia loading, and the same may be true in the shark. As with the rise in TEP (Fig. 5), had these changes not occurred, ammonia loading might have overwhelmed the capacity of the OUC to consume ammonia, and it might have been impossible to reduce plasma T_Amm concentration during the later phases of HEA exposure (Fig. 2). With respect to the H⁺-ATPase response, this enzyme does not appear to be apically located, in contrast to teleosts. Instead, all current models for the elasmobranch gill indicate that H⁺-ATPase functions basolaterally in those ionocytes, which excrete base (Piermarini and Evans, 2001; Piermarini et al., 2002; Tresguerres et al., 2005; Reilly et al., 2011), so that H⁺ ions are returned to the blood when HCO₃⁻ is excreted apically at times of blood alkalosis (Tresguerres et al., 2005; Tresguerres et al., 2006; Tresguerres et al., 2007; Roa et al., 2014). In the present study, there was a slight tendency towards alkalosis during HEA exposure, with a significant rise in plasma [HCO₃⁻] (Fig. 3), so downregulation of H⁺-ATPase expression would not be expected. We speculate therefore that H⁺-ATPase may have an additional role, that of diffusion trapping of NH₃ at the extracellular side of the basolateral membrane. In future studies, it will be of interest to see whether H⁺-ATPase and Rhbg are co-localized in the base-excreting ionocytes.

We have presented a speculative model in which the gill of this nitrogen-limited predator is poised not only to prevent nitrogen loss but also to scavenge ammonia-N from the environment for the purpose of enhancing urea-N synthesis. However, ammonia uptake against gradients is modulated to prevent the regulatory capacities of the animal being overwhelmed. In future studies, many aspects of this model can be evaluated, such as the concentration-dependent kinetics of ammonia uptake, the identification and localization of the transporters involved, and the signals that control the system. For example, are the molecular responses identified in the present study a response to internal or external ammonia elevation? A recent study of ventilatory control in S. acanthias found that the hyperventilatory response to HEA was mediated by internal rather than external ammonia levels (Zhang et al., 2014).

Experiments were performed at Bamfield Marine Sciences Centre (BMSC), BC, Canada, in June and July. Pacific spiny dogfish (S. acanthias suckleyi, 0.7–2.9 kg) were obtained by trawl netting in nearby Barkley Sound, BC, Canada, in May and June, under a licence from the Department of Fisheries and Oceans, Canada. [Note Ebert et al. (Ebert et al., 2010) have recently proposed that these north-east Pacific dogfish may be a separate species (Squalus suckleyi) rather than a subspecies of Squalus acanthias.] Only male fish were retained for purposes of population conservation, as females were invariably pregnant. At BMSC they were held in a large 151,000 l circular tank served with running seawater at the experimental temperature (11–12°C), salinity (30±2‰) and pH (7.90±0.15). During this period, the dogfish were fed every 3–4 days with a 3% ration of dead hake (Merluccius productus), but removed to smaller 1500 l tanks in batches of 6–10 for 1–2 weeks, fasting under the same conditions, prior to experimentation. All experimental procedures conformed to the guidelines of the Canada Council for Animal Care, and were approved by BMSC and McMaster University Animal Care Committees.

All experiments were performed in covered, polyurethane-coated wooden fish boxes (105 cm length×16.5 cm width×25 cm depth) fitted with perimeter aeration, as described by Wood et al. (Wood et al., 1995). The boxes were served with a flow-through of 1 l min⁻¹, but could be closed for flux measurements, at which time they were bathed in an external running seawater bath to maintain temperature (11–12°C). The fish were placed in the boxes for 24 h before experiments were started.

In series 2 and 3, the dogfish were fitted with indwelling caudal artery catheters as described by De Boeck et al. (De Boeck et al., 2001). These facilitated repetitive blood sampling and measurements of TEP without disturbance. For cannulation, the fish were anaesthetized in MS-222 (0.15 g l⁻¹; Syndel Laboratories, Qualicum, BC, Canada), weighed, and artificially ventilated on an operating table. The vertebral column was exposed by a lateral 5 cm incision through the muscle of the caudal peduncle, and the catheter (Clay-Adams PE50, Becton Dickinson, Franklin Lakes, NJ, USA) was implanted into the artery through a small hole in the haemal canal. The catheter was filled with dogfish saline (see Wood et al., 2013a) containing 50 i.u. ml⁻¹ lithium heparin (Sigma-Aldrich, St Louis, MO, USA). A PE160 sleeve, heat-flared at both end, was cemented over the PE50 tubing at its exit from the incision, and firmly anchored with several silk sutures so as to secure the catheter in place. The wound was then dusted with powdered oxytetracycline (Sigma-Aldrich) to avoid infection, and tightly closed with silk ligatures. After surgery, the dogfish were revived in anaesthetic-free water and transferred to the wooden fish boxes for 24 h to recover.

For HEA exposures, additions from a 1 mol l⁻¹ stock solution of NH₄HCO₃ (pH 7.90) were made, sufficient to raise the total ammonia concentration to nominally 1 mmol l⁻¹.

This series (N=8–12) examined the impact of HEA exposure on ammonia (J_Amm) and urea-N (J_Urea-N) flux rates in non-cannulated dogfish. Fluxes were measured for a 12 h pre-exposure control period, and then over successive 12 h periods of HEA. For flux measurements, flow-through to the box was stopped and the water level set to a mark representing 40 l. Water samples (10 ml) for total ammonia-N (T_Amm) and urea-N (T_Urea-N) measurements were taken at the start and end of each flux period, up to and including 48 h, and were stored at −20°C. The water in the boxes was renewed every 12 h. After the 48 h water sample, the fish were rapidly anaesthetized by addition of a concentrated solution of neutralized MS-222 (pH 7.9) so as to bring the water concentration to 0.25 g l⁻¹. A terminal blood sample was drawn by caudal puncture into a 5 ml heparinized syringe. Plasma was immediately separated by centrifugation (1 min, 5000 g) and flash-frozen in liquid N₂, then stored at −80°C for later measurement of plasma urea-N and plasma T_Amm. An additional water sample was drawn simultaneously for measurement of T_Amm in water. This was to determine the plasma-to-water concentration gradient for T_Amm. A control group (N=7) was sampled in the same manner. Some of the fish were also sampled for molecular end points (see series 4). All fish were weighed.

This series (N=6) examined the impact of HEA exposure on arterial plasma ammonia parameters (T_Amm, Pa_NH3) relative to the same parameters in the external water, so as to allow precise calculation of trans-branchial P_NH3 and electrochemical NH₄⁺ gradients. Additionally, the blood sampling facilitated the measurement of arterial blood respiratory gases and acid–base status, and plasma urea-N and major electrolyte concentrations. The fish were exposed to HEA in the same manner as in series 1, with renewal of the HEA water at 12 h intervals. Arterial blood samples (600 and 400 μl) were drawn into two ice-cold gas-tight glass syringes (Hamilton, Reno, NV, USA) under pre-exposure control conditions (0 h), and then at 2, 6, 12, 24, 36 and 48 h of HEA exposure. Water samples were drawn simultaneously for pH measurements and then stored at −20°C for later analyses of T_Amm and P_NH3. The smaller blood sample was analysed for pHa and Pa_O2. The larger blood sample was immediately centrifuged (1 min, 5000 g); the plasma was decanted, then divided into three separate aliquots. One was analysed immediately for plasma total CO₂ concentration; the other two were flash-frozen in liquid N₂ and stored at −80°C for subsequent analyses of (i) plasma T_Amm and T_Urea-N, and (ii) plasma electrolytes. Red blood cells separated by centrifugation were re-suspended in saline, combined with blood recovered from the electrodes, and re-infused together with sufficient saline to restore blood volume.

This series (N=6) examined the impact of HEA exposure on branchial TEP in dogfish fitted with arterial catheters. TEP was measured under pre-exposure control conditions (0 h), and then at 2, 6, 12, 24, 36 and 48 h of HEA exposure.

Non-cannulated dogfish were sampled for molecular end points at 0 h (pre-exposure control), 24 h and 48 h of exposure to HEA (N=6 at each time) as in series 1. At sampling, the fish was anaesthetized with 0.25 g l⁻¹ MS-222 (pH 7.9), a terminal blood sample was drawn by caudal puncture or via the caudal artery catheter, and 1000 i.u. lithium heparin (in 1 ml saline) was injected. The red blood cells were immediately separated by centrifugation (1 min, 5000 g), and then preserved in liquid N₂, taking care to avoid contamination with white blood cells. The fish was then artificially ventilated on an operating table with HEA or control water, as appropriate, and the ventral aorta was exposed by dissection and cannulated with PE160 tubing, taking care to avoid air bubbles. Perfusion with an ice-cold 500 mmol l⁻¹ NaCl solution containing 10 i.u. ml^−l of lithium heparin was then commenced, using a Model 1100 cardiac pump (Harvard Apparatus, Holliston, MA, USA) set to a working mode of 1/3 systole, 2/3 diastole, a heart rate of 30 beats min⁻¹, and a stroke volume adjusted so that the output was ~20 ml kg⁻¹ min⁻¹. Once the gills and tissues were cleared of blood (2–4 min), samples of gill, kidney, brain, intestine, liver, muscle, rectal gland, cardiac stomach (1) and pyloric stomach (2) were removed and stored in RNAlater (Sigma-Aldrich).

Pa_O2, pHa and water pH were measured using Radiometer electrodes (Copenhagen, Denmark) kept at the experimental temperature with water jackets. Outputs of the electrodes (E5036 for P_O2; GK2401C for pH) were displayed on Radiometer pHM 71 and pHM 72 acid–base analysers. True plasma total CO₂ was measured by a Model 965 CO₂ analyser (Corning, Midland, MI, USA). Pa_CO2 and [HCO₃⁻]a were calculated using the solubility of carbon dioxide (α_CO2), the apparent pK (pK_app) for dogfish plasma, and rearrangements of the Henderson–Hasselbalch equation according to Boutilier et al. (Boutilier et al., 1984).

TEP was recorded using 3 mol l⁻¹ KCl–agar bridges connected via Ag/AgCl electrodes (WPI, Sarasota, FL, USA) to a high impedance electrometer (Radiometer pHM 84 meter). The reference bridge was placed in the water in the fish chamber, and the measurement bridge was connected to the arterial blood catheter. TEP values (mV) were expressed relative to the water side as 0 mV after correction for junction potential.

Plasma T_Amm was measured enzymatically (l-glutamate dehydrogenase method) (Mondzac et al., 1965) using an ammonia reagent kit (Product No. 85446, Raichem Life Sciences, Mumbai, India) on the first thaw of frozen plasma. Water T_Amm was measured colorimetrically by the salicylate–hypochlorite method (Verdouw et al., 1978); the two methods were cross-validated in seawater. The colorimetric assay, while economical and very accurate, cannot be used in plasma because of protein interference. Plasma and water ammonia partial pressures (P_NH3) and ammonium ion concentrations ([NH₄⁺]) were calculated from the respective T_Amm and pH measurements, again using the Henderson–Hasselbalch relationship, as detailed by Wright and Wood (Wright and Wood, 1985). In the absence of data for elasmobranch plasma, constants (α_NH3, pK_amm) were taken from the study of Cameron and Heisler (Cameron and Heisler, 1983) on trout, with appropriate adjustments for ionic strength. A sensitivity analysis demonstrated that the adjustments used did not affect the conclusions of this study.

The P_NH3 gradient across the gills (ΔP_NH3) was calculated as the difference between simultaneous measurements of water P_NH3 and arterial blood plasma Pa_NH3. A positive value of ΔP_NH3 will tend to drive NH₃ into the fish, while a negative value will drive NH₃ out of the fish. The Nernst potential for NH₄⁺ (E_NH4⁺) was calculated in the standard fashion where [NH₄⁺]_in is the concentration of NH₄⁺ in the arterial blood plasma and [NH₄⁺]_out is the concentration of NH₄⁺ in the outside water. The true electrochemical potential or net driving force (F_NH4⁺) for NH₄⁺ across the gills was calculated according to Kirschner (Kirschner, 1970) as the difference between E_NH4⁺ and TEP. A positive value of F_NH4⁺ will tend to drive NH₄⁺ into the fish, while a negative value will drive NH₄⁺ out of the fish. Equations for ΔP_NH3, E_NH4⁺ and F_NH4⁺ have been published previously (Wood and Nawata, 2011).

Plasma and water T_Urea-N concentrations were measured colorimetrically by the diacetyl monoxime method (Rahmatullah and Boyde, 1980). Plasma Na⁺, K⁺, Ca²⁺ and Mg²⁺ concentrations were determined by flame atomic absorption spectroscopy (Varian SpectrAA-220FS, Mulgrave, Australia) and Cl⁻ was measured by coulometric titration (Radiometer CMT-10 chloridometer).

Flux rates of total ammonia (J_Amm) and urea-N (J_Urea-N) were calculated from changes of concentration in the closed chambers, factored by fish mass, volume and time. A negative J_Amm indicates a net excretion, and a positive J_Amm a net uptake of ammonia into the fish.

Total RNA was extracted from dogfish tissues with Trizol (Invitrogen, Burlington, ON, Canada) and quantified spectrophotometrically (Nanodrop ND-1000; Nanodrop Technologies, Wilmington, DE, USA). First strand cDNA was synthesized from 1 μg of DNaseI-treated total RNA with Superscript II reverse transcriptase (Invitrogen).

cDNA sequences of Rhag, Rhbg and Rhp2 were obtained by 5′- and 3′-rapid amplification of cDNA ends (Smart RACE cDNA amplification kit, BD Bioscience Clontech, Mississauga, ON, Canada) using erythrocyte and gill cDNA with cloning primers (Table 3) and the protocol described previously (Nawata et al., 2007).

Hydropathy profile and N-linked glycosylation site predictions were made using TMHMM (Krogh et al., 2001; Sonnhammer et al., 1998) and ScanProsite (de Castro et al., 2006), respectively.

Tissue distribution of Rhbg and Rhp2 was determined by PCR with the primer sets listed in Table 3 at 35 cycles for Rhbg and Rhp2 and 25 cycles for β-actin. Products were electrophoresed on 1% agarose gels stained with ethidium bromide.

qPCR was performed using the above synthesized cDNA to determine mRNA levels of CA-IV, Rhbg, Rhp2, H⁺-ATPase (V-type, B subunit), NHE2, NKA (α-subunit), UT in the gill, Rhag in the erythrocytes, and Rhbg, Rhp2 and UT in the kidney. All primer sets used (Table 3) were validated and products sequenced beforehand. qPCR reactions were performed on a Mx3000P QPCR System (Stratagene, Cedar Creek, TX, USA) using Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen) under conditions detailed previously (Nawata et al., 2007) with melt-curve analysis and no-template controls run in tandem. Values were extrapolated from standard curves and were normalized to ng total RNA.

Data are expressed as means ± 1 s.e.m. (N), where N is the number of fish. In series 1, 2 and 3, data were analysed by repeated measures ANOVA, with appropriate transformations where necessary to pass normality and variance tests, followed by a Holm–Šidák or Dunnett's multiple comparison test to detect specific differences relative to the pre-exposure control value. Paired Student's t-tests were used to evaluate differences between simultaneous measurements in blood plasma versus external water. In series 4, groups were independent, so regular ANOVA followed by the LSD test was used to detect values that were significantly different. All tests were two-tailed and a significance level of 0.05 was used throughout.

We thank Bruce Cameron, Research Co-ordinator at Bamfield Marine Sciences Centre, for invaluable logistical assistance, Andrew Esbaugh for initial cDNA cloning, Linda Diao for excellent technical help, Martin Tresguerres, Alex Clifford and Greg Goss for advice, and two anonymous reviewers whose comments improved the manuscript.