A Study of the Regulatory and Environmental Factors Affecting Trimethylamine Oxide Accumulation in Marine Organisms

Trimethylamine oxide (TMAO) was first described in marine organisms as an osmolyte, involved in the balance of water and solutes. After its discovery, it was found to be part of a subset of osmotic constituents termed counteracting solutes. These solutes exhibit stabilizing properties and can preserve protein functionality against biological and environmental perturbations. TMAO acts as a universal stabilizer, protecting macromolecular structure and function in response to numerous stressors, including urea destabilization, hydrostatic pressure, temperature and salinity. The studies presented in this dissertation address the regulatory and environmental factors affecting TMAO accumulation. Both exogenous and endogenous sources are involved in the maintenance of TMAO. Exogenous TMAO accumulates through absorption from the diet while endogenous TMAO is synthesized from dietary or cellular precursors with the flavincontaining monooxygenase trimethylamine oxidase (TMAoxi). Species without a physiologically relevant synthetic capacity are hypothesized to rely entirely on dietary contributions for accumulation. Chapter 1 examines the necessity of an exogenous TMAO source on long-term maintenance in elasmobranch species with and without the ability for endogenous synthesis. These data show that presence or absence of TMAoxi cannot be used as a proxy to determine the importance of dietary TMAO on prolonged conservation. It seems that all species, regardless of synthesizing potential, rely to an extent on contributions from the diet. Chapter 2 further examines the regulatory factors affecting TMAO. This study provides evidence for endogenous production via an understudied synthetic pathway whereby TMAO is accumulated as a byproduct during lipid storage. The existence of this pathway is supported by a correlation between TMAO content and total lipid in a variety of Hawaiian mid-water fishes. The regulatory role of evolutionary relatedness on accumulation potential is also addressed in this chapter. Phylogenetic independent contrasts (PIC) showed no relationship between phylogeny and TMAO content across 27 species spanning nine orders. This suggests that environmental factors impart a larger influence on TMAO retention than evolutionary history. Chapter 2 goes on to examine TMAO’s role in combatting the environmental stress associated with increasing hydrostatic pressure. TMAO was shown to increase with increasing depth of occurrence across all species of Hawaiian mid-water fishes studied. These data support previous reports of TMAO accumulation as an environmental adaptation to combat the destabilizing effects of elevated hydrostatic pressure. Chapter 3 explores TMAO’s ability to counteract environmental fluctuations in temperature. Previous in vitro studies showed intracellular transport and accumulation of TMAO with increasing temperature in elasmobranch red blood cells. Further, this was shown to suppress the traditional heat shock response of heat shock protein 70 (HSP70) upregulation. However, we saw no increase in plasma or tissue TMAO in response to elevated temperature for two shark species in vivo. Either mechanisms established in vitro are not applicable at the organismal level or additional regulatory factors are limiting TMAO accumulation. Lastly, a brief study examining regulation of TMAO through ontogeny in an elasmobranch species, Squalus acanthias, is presented in the Appendix. Pups of this species exhibit low levels of urea and TMAO, their two primary osmolytes. However, total osmotic pressure is maintained at adult levels. Therefore, a shift in the osmotic milieu occurs sometime between birth and adulthood. These findings are in contrast to those reported for the little skate, Leucoraja erinacea, which expresses adult levels of these osmotic constituents early in development. These data point to divergence in the early osmoregulatory strategies of differing elasmobranch groups. In the enclosed chapters, key objectives regarding the regulatory and environmental factors influencing TMAO are addressed. Specifically, this research examines how contributing sources, evolutionary restrictions and environmental stress affect TMAO accumulation. These studies elucidate TMAO’s multifaceted role in marine organisms and provide insight into the factors regulating its adaptive potential.

stabilizer, protecting macromolecular structure and function in response to numerous stressors, including urea destabilization, hydrostatic pressure, temperature and salinity.
The studies presented in this dissertation address the regulatory and environmental factors affecting TMAO accumulation.
Both exogenous and endogenous sources are involved in the maintenance of TMAO. Exogenous TMAO accumulates through absorption from the diet while endogenous TMAO is synthesized from dietary or cellular precursors with the flavincontaining monooxygenase trimethylamine oxidase (TMAoxi). Species without a physiologically relevant synthetic capacity are hypothesized to rely entirely on dietary contributions for accumulation. Chapter 1 examines the necessity of an exogenous TMAO source on long-term maintenance in elasmobranch species with and without the ability for endogenous synthesis. These data show that presence or absence of TMAoxi cannot be used as a proxy to determine the importance of dietary TMAO on prolonged conservation. It seems that all species, regardless of synthesizing potential, rely to an extent on contributions from the diet.
Chapter 2 further examines the regulatory factors affecting TMAO. This study provides evidence for endogenous production via an understudied synthetic pathway whereby TMAO is accumulated as a byproduct during lipid storage. The existence of this pathway is supported by a correlation between TMAO content and total lipid in a variety of Hawaiian mid-water fishes. The regulatory role of evolutionary relatedness on accumulation potential is also addressed in this chapter. Phylogenetic independent contrasts (PIC) showed no relationship between phylogeny and TMAO content across 27 species spanning nine orders. This suggests that environmental factors impart a larger influence on TMAO retention than evolutionary history.
Chapter 2 goes on to examine TMAO's role in combatting the environmental stress associated with increasing hydrostatic pressure. TMAO was shown to increase with increasing depth of occurrence across all species of Hawaiian mid-water fishes studied. These data support previous reports of TMAO accumulation as an environmental adaptation to combat the destabilizing effects of elevated hydrostatic pressure.
Chapter 3 explores TMAO's ability to counteract environmental fluctuations in temperature. Previous in vitro studies showed intracellular transport and accumulation of TMAO with increasing temperature in elasmobranch red blood cells. Further, this was shown to suppress the traditional heat shock response of heat shock protein 70 (HSP70) upregulation. However, we saw no increase in plasma or tissue TMAO in response to elevated temperature for two shark species in vivo. Either mechanisms established in vitro are not applicable at the organismal level or additional regulatory factors are limiting TMAO accumulation.
Lastly, a brief study examining regulation of TMAO through ontogeny in an elasmobranch species, Squalus acanthias, is presented in the Appendix. Pups of this species exhibit low levels of urea and TMAO, their two primary osmolytes. However, total osmotic pressure is maintained at adult levels. Therefore, a shift in the osmotic milieu occurs sometime between birth and adulthood. These findings are in contrast to those reported for the little skate, Leucoraja erinacea, which expresses adult levels of these osmotic constituents early in development. These data point to divergence in the early osmoregulatory strategies of differing elasmobranch groups.
In the enclosed chapters, key objectives regarding the regulatory and environmental factors influencing TMAO are addressed. Specifically, this research examines how contributing sources, evolutionary restrictions and environmental stress affect TMAO accumulation. These studies elucidate TMAO's multifaceted role in marine organisms and provide insight into the factors regulating its adaptive potential.

Introduction
Trimethylamine oxide (TMAO) is a small compound accumulated as an intracellular osmolyte in a diversity of marine organisms ; see  for reviews). It serves an additional role as a universal protein stabilizer Yancey et al., 2001;, protecting structure and function against a multitude of environmental stressors. TMAO can counteract perturbations in protein function due to salinity Hammerschlag, 2006;, temperature , hydrostatic pressure (Yancey et al., 2002; and the nitrogenous waste compound, urea Yancey, 2001;Zou et al., 2002). TMAO is retained at exceptionally high levels by elasmobranchs (sharks, skates and rays) that accumulate urea as their primary osmolyte . These species maintain a 2:1 ratio of urea to TMAO + other stabilizing osmolytes  to promote macromolecular stability .
There are two general mechanisms used by elasmobranchs to regulate TMAO.
Some possess the flavin-containing monooxygenase, trimethylamine oxidase (TMAoxi), and have the ability to synthesize TMAO from endogenous or dietary precursors such as choline . Species without this enzyme are thought to accumulate TMAO from the diet alone , requiring prey items rich in TMAO to maintain intracellular levels. However, there is little direct evidence for dietary absorption and it is possible that alternate synthetic pathways exist that have yet to be described.
Several previous studies have examined TMAO maintenance during extended starvation. One study showed that spiny dogfish, Squalus acanthias, were unable to synthesize TMAO from radiolabeled precursors and postulated that constant plasma levels were achieved during 20 days of starvation through active reabsorption at the kidneys and release from tissue pools (Goldstein et al., 1967). The winter skate, Leucoraja ocellata, was also found to maintain plasma TMAO over 45 days of starvation . Maintenance in this case was attributed to decreased excretion with no increase in TMAO synthesis found. Treberg and Driedzic (2007) also examined muscle TMAO in the winter skate over 28 days of starvation and speculate that constant levels were due to release from tissue catabolism and subsequent recycling. Another study showed stable plasma TMAO and urea over 56 days of starvation in spiny dogfish. These authors also suggest that the large pools of TMAO stored in various tissues, primarily muscle, supply adequate amounts for plasma maintenance (Kajimura et al., 2008). Another possible explanation for prolonged maintenance was provided by , who suggested that TMAO could be synthesized in the absence of dietary contributions from choline, released during hydrolysis of membrane phospholipids, whether endogenous or dietary.
Although it is possible that each of these processes may contribute to conservation of TMAO, all aforementioned studies examined TMAO flux in elasmobranchs during starvation when nitrogen metabolism is significantly altered . Fasted skates have been shown to suppress TMAO excretion during starvation  and dogfish exhibit decreases in plasma urea and total osmolarity during fasting (Leech et al., 1979). In another study, spiny dogfish were able to preserve urea during long-term starvation, although it was reported to come at the cost of significant protein catabolism. It was estimated that 69.5 g of protein would need to be broken down over 56 days for continued urea synthesis under starvation conditions (Kajimura et al., 2008). Further, urea synthesis is an energy-expensive process requiring 5 moles of ATP per 1 mol urea Lee et al., 2006).
It is thought that spiny dogfish must feed every 5-6 days to sustain nitrogen balance (Kajimura et al., 2006). In fact, the rate of nitrogen loss does not increase after feeding, suggesting these animals are nitrogen limited (Wood et al., , 2007. After a meal, there is a switch from net urea efflux to net intestinal absorption  and plasma urea spikes 20 hours postprandially in spiny dogfish (Kajimura et al., 2006;Kajimura et al., 2008) indicating absorption from dietary constituents or elevated synthesis. Similarly, plasma TMAO also increases 20 hours after feeding in this species (Kajimura et al., 2006; for review).
It has been suggested that elasmobranchs excrete between 4-14% of their whole body TMAO per day (Goldstein and Palatt, 1974).  published a more conservative estimate for the winter skate, Leucoraja ocellata, of less than 1% whole body TMAO lost per day. However, the rate of TMAO loss decreased after one week of starvation, which would have reduced Treberg and Driedzic's estimate. Even if elasmobranchs lose an average of 1% TMAO per day, this would amount to losses greater than 50% of total body stores over 56 days without an endogenous or dietary input. As elasmobranchs are already nitrogen limited , questions of TMAO flux and maintenance are better addressed under less stressful physiological conditions when individuals are actively feeding.
Here, we directly test the effect of diet on TMAO content in three elasmobranch species with differing synthetic capacities. The smoothhound, Mustelus canis, a shark with TMAoxi activity was compared to the spiny dogfish, Squalus acanthias, and little skate, Leucoraja erinacea, two species with negligible synthesis . Individuals were fed a high or low TMAO diet for 56 days to examine the effects of an exogenous source on long-term TMAO maintenance.

Collection and transport
Three elasmobranch species were obtained by otter trawl off the commercial fishing vessel Virginia Marise. Smoothhounds, Mustelus canis (n=10), spiny dogfish, Squalus acanthias (n=13), and little skates, Leucoraja erinacea (n=19), were captured in Narragansett Bay, RI during summers 2013 -2015. Males and females were placed in 150 L insulated coolers provided with chilled, aerated seawater and transported to holding facilities less than an hour away. Individuals were placed in a random fashion into one of two 2.4 m diameter, 2850 L continuous flow circular holding tanks. Sharks were housed up to n=5 and skates up to n=10 per tank.

Feed trials
Seawater temperature was maintained at 17 ± 0.22°C (mean ± SEM) and light on a 12h:12h light/dark cycle for the duration of the experiment. Individuals were acclimated for a minimum of 72 hours before initiation of a feeding trial. After acclimation, animals were fed diets high or low in TMAO content for 56 days. The high TMAO diet was comprised of a mixture of herring (Clupea harengus) and squid (Doryteuthis pealei) fed at 2.5% body weight, twice a week. Rations were chosen in accordance with previous studies Liew et al., 2013). Herring and squid are part of the regular diet consumed in the wild ) and a rich source of TMAO (~50-80 mmol kg -1 , Treberg and Driedzic, 2007;Supplementary Fig. 1). Individuals placed on a low TMAO diet were fed brook trout (Salvelinus fontinalis) at 2.5% body weight, twice a week. Brook trout contain negligible levels of TMAO (<0.5 mmol kg -1 , Supplementary Fig. 1) and constitute a low TMAO diet. Freshwater fish tissue is known to contain lower concentrations of essential vitamins compared to marine tissue (Käkelä et al., 1999); therefore, as brook trout are a freshwater species that would not contribute to the regular diet experienced by these marine elasmobranchs in the wild, a vitamin supplement (SEA TABS for Birds, Turtles, Fish and Sharks by Pacific Research Laboratories) was included in the low TMAO diet to ensure individuals experienced no confounding deficiencies. Three separate feeding trials were conducted and data pooled for analysis. Five spiny dogfish, two smoothhounds and eight little skates were included in trial one which ran for 56 days from 2/20/13 -4/17/13. 11 little skates were included in trial two which ran for 56 days from 9/08/14 -11/03/14. Eight smoothhounds and seven spiny dogfish were included in trial three which ran for 56 days from 7/29/15 -9/23/15. Test species included in each trial were dependent on local availability.

Sampling
Blood samples were taken by the caudal method at time 0 and once monthly using an 18-gauge hypodermic needle. Prior to blood sampling, specimens were anaesthetized with 0.05 g l -1 MS-222 dissolved in a seawater bath. Red blood cells were separated by centrifugation at 10,000 rpm for three minutes and discarded.
Plasma was flash frozen in liquid nitrogen and stored at -80 for later analysis.
Measurements of weight, standard and total lengths were taken on each sampling day.
Sex and spiracle length were also recorded. These parameters were used to assess growth and as an estimate of age to help assess the influence of size, ontogeny or sex-related differences on feeding habits and TMAO content Bockus and Seibel, in prep).
On day 56, individuals were euthanized with MS-222 (0.15 g l -1 ) dissolved in a seawater bath. Length and weight measurements were taken. Plasma samples were obtained by the caudal method as described above. White muscle (~1 g) was harvested from the left dorsolateral epaxial muscle. The whole liver was removed and weighed.
A sample of liver tissue (~1 g) was harvested from the periphery of the major lobe. All samples were duplicated for each individual and immediately flash frozen for later analysis. The presence of maturing embryos was recorded to take into account any gestational influence. All and analyses were conducted in accordance with IACUC #AN12-07-026.

Analytical techniques
Plasma, muscle and liver samples were analyzed for TMAO and urea. Samples were deproteinated and homogenized 1:5 in 5% trichloroacetic acid solution. TMAO content was determined spectrophotometrically with ferrous sulfate and 2% picric acid as described by . Homogenates were further used to assess urea using the diacetyl monoxime method . Total lipid was measured in liver tissue using a 2:1 chloroform to methanol extraction modified for small sample mass (Lee et al., 1996).

Statistical analysis
Where variables scaled with tissue or body weight, individual values were normalized to a common mass before analysis. Means of individuals fed a high or low TMAO diet were compared between time points by two-way-, or two-way-RM, ANOVA followed by a Holm-Sidak post-hoc test. Terminal samples were compared between diets within species using one-way unpaired student's t-test. Significance was set at p<0.05. Statistics and graphs were generated using GraphPad Prism 7.0.

Lipid
Liver lipid content was not different between diets within species (Fig. 1).
Smoothhounds fed a high TMAO diet had a mean lipid content of 36.93% wet wt. ± 9.29 SEM compared to low TMAO diet individuals at 18.67 ± 7.62%. Spiny dogfish fed a high TMAO diet had a mean lipid content of 62.11 ± 11.22% compared to low TMAO diet individuals at 51.96 ± 12.91%. Little skates fed a high TMAO diet had a mean lipid content of 22.38 ± 5.13% compared to low TMAO diet individuals at 23.57 ± 4.26%.
There were no further differences across time points within diets or between diets at an individual time point within species (Fig. 2a-c).

Urea
Plasma urea (mM) values at day 0, 28 and 56 listed in Table 1. Little skates fed a low TMAO diet showed an increase in plasma urea from day 0 at 368.91 ± 11.15 mM to day 28 at 416.97 ± 9.08 mM (two-way RM ANOVA, p=0.01) and day 56 at 433.70 ± 21.17 mM (p=0.002). There were no further differences across time points within diets or between diets at an individual time point within species (Fig. 3a-c).

Urea
Tissue urea (mM) values at day 56 listed in Table 2. Liver urea was lower in low TMAO diet spiny dogfish at 120.48 ± 13.94 mM than high TMAO diet dogfish at 195.72 ± 23.67 mM (one-way unpaired student's t-test, p=0.04). There were no further differences between diets within tissue types for individual species (Fig. 6).

Discussion
Animals in this study readily fed on the high and low TMAO diets. Lipid levels were conserved across groups in all species (Fig. 1), as would be expected in actively feeding individuals not relying on protein or lipid stores to support metabolism. Plasma TMAO fell by day 56 in dogfish fed the low TMAO diet but not in the smoothhound or little skate (Fig. 2a-c). This suggests that spiny dogfish (a nonsynthesizing species) do rely, to an extent, on dietary contributions for TMAO maintenance. When faced with an absence of TMAO in the diet, they may transport plasma TMAO to alternate tissues, such as muscle or liver, that depend on TMAO for preserved cellular function. Given these results, one would also expect to see a drop in plasma TMAO in the little skate that similarly exhibit no TMAO synthetic capacity.
However, TMAO was maintained in the plasma of the little skate regardless of diet.
Skates rely to a lesser extent on TMAO to support osmotic potential. Skates retain plasma TMAO at half the content present in spiny dogfish. Further, they exhibit similar concentrations of urea but higher levels of alternative amino acid osmolytes (King and Goldstein, 1983). It is possible there is a greater discrepancy in the way sharks and skates regulate TMAO than previously thought. TMAO accumulation patterns are different through spiny dogfish and little skate ontogeny, perhaps supporting disparities in the way this molecule is utilized between the two species Bockus and Seibel, submitted).
Surprisingly, muscle TMAO fell in smoothhounds fed a low TMAO diet, the only species in this study to exhibit a TMAO synthetic capacity , but was maintained in spiny dogfish and the little skate (Fig. 5). Feeding provides ample precursors, such as choline , to support endogenous production supposedly minimizing this species reliance on dietary TMAO for longterm maintenance. It is possible that all species, whether they can synthesize TMAO or not, rely on dietary contributions for absorption and retention. Alternatively, species that can readily synthesize TMAO may not express the regulatory pathways for reabsorption and retention. Spiny dogfish and little skates appear to exhibit a greater capacity for TMAO conservation than smoothhounds via such possible mechanisms as reabsorption at the gill and kidney, transfer between tissue pools, and recycling through catabolism, leading to a greater capacity for TMAO retention over time. In accordance with this view, spiny dogfish and little skates maintained muscle TMAO regardless of whether TMAO was present in the diet or not. These data support previous findings showing TMAO maintenance during prolonged starvation in spiny dogfish and the winter skate L. ocellata (Goldstein et al., 1967;Driedzic, 2006, 2007;Kajimura et al., 2008) and likely result from a combination of the proposed retention mechanisms. The possibility of unidentified TMAO synthetic pathways cannot be ruled out as a source contributing to extended preservation.
Urea excretion accounts for 90% or more of the total nitrogen excreted by elasmobranchs Kajimura et al., 2006). Urea conservation depends largely on selective impermeability at the gills Part et al., 1998;Wood et al., 2013); however, there is an unavoidable "leakiness" (Kajimura et al., 2008) and urea stores must be supplemented through exogenous absorption or endogenous synthesis from dietary precursors. Dietary urea may be provided by the low concentrations retained in some marine fishes or by the observed cannibalism of other marine sharks rich in urea . In this study, there was no difference in plasma urea between diets in any species. Although there were subtle, but significant, differences in plasma urea concentration over time in the little skate ( Fig.   3c), all values fell within the range reported previously for wild caught specimens of this species.
There was a decrease in liver urea in spiny dogfish fed a low TMAO diet but not the other two species studied (Fig. 6). Again, although the difference in spiny dogfish was significant, it fell within the previously reported range and is not likely physiologically relevant. There is no reason to expect the low TMAO diet would restrict the urea synthetic pathway (ornithine-urea cycle) or affect retention mechanisms. However, freshwater fishes exhibit lower concentrations of urea, ~1.5mM in brook trout , compared to marine fish with reported concentrations up to ~25mM . Therefore, less urea was available for direct dietary absorption in our low TMAO diet, perhaps explaining the difference.
A number of the given explanations for TMAO and urea regulation are likely involved in the conservation of these molecules. However, the presence or absence of TMAoxi cannot be used as a metric to assess a species reliance on dietary TMAO. Treberg et al. (2005) found TMAoxi activity did not correlate to TMAO content in smelt (Osmerus mordax). Likewise, although FMO activity was present in the winter skate (L. ocellata), there was no evidence of TMA oxidation . This suggests that although TMAoxi has been shown to oxidize TMA to TMAO it may not be the key enzyme regulating TMAO synthesis and retention. In fact, there is some discrepancy in the literature regarding the ability of dogfish and skates to synthesize TMAO (Schlenk and Li-schlenk, 1994; although these concerns were addressed by  that support the conclusion that these species are incapable of physiologically relevant production. However, this disagreement demonstrates the lack of understanding with regard to relevant sources involved in TMAO production. There may be alternate enzymes capable of oxidizing TMAO such as cytochrome P450 monooxygenases  or others that have not been identified. Additionally, TMAO oxidation by gut microflora (Koeth et al., 2013) may diminish a species reliance on dietary contributions from prey tissue.
When accumulation of an organic osmolyte is limited, animals have been shown to replace the osmotic deficit with alternate compounds.  showed an increase in betaine, another methylamine, to offset TMAO losses and maintain osmotic balance in the little skate. Similarly, NaCl replaced urea in the plasma of nitrogen-limited European dogfish . Although these alternatives pose possible short term solutions, increasing NaCl intracellulary has been shown to destabilize protein function (Yancey et al., 1982) and betaine, although a protein stabilizer, is not as effective at preserving protein structure as TMAO . Therefore, the concentration of betaine needed to affect the same degree of stability would require an increase in total osmolarity or decrease in other osmolytes. The vitamin supplement we added to our low TMAO diet did include taurine, an alternative counteracting solute, which may have been accumulated in place of TMAO. However, this would not explain the differences we saw between species and tissue types of individuals fed the low TMAO diet.
This study shows species specific differences in elasmobranchs ability to regulate TMAO without an exogenous source. Further, there was no clear delineation between synthesizing potential and ability to regulate TMAO without a dietary contribution. Therefore, presence or absence of TMAoxi activity cannot be used as a proxy to determine whether a dietary contribution is needed for long-term TMAO maintenance. Diet was also important for urea conservation, although effects seem to differ between tissue type and species. Although alternative compounds may be substituted to maintain osmotic balance when TMAO or urea accumulation are limited, these molecules serve a variety of functions and fluctuations caused by shifts in diet may affect a number of physiological processes in these animals. More evidence is needed to determine the mechanistic regulatory pathways involved in preservation of these nitrogenous compounds, specifically those controlling TMAO No significant differences (one-way unpaired student's t-test, p<0.05) between diets within species.   Significant differences (*) determined between diets within tissue types for individual species (one-way unpaired student's t-test, p<0.05). Käkelä, R., Käkelä, A., Hyvärinen, H., Asikainen, J., 1999. Vitamins A1, A2, and E in minks exposed to polychlorinated biphenyls (Aroclor 1242®) and copper, VIA diet based on freshwater or marine fish. Environ. Toxicol. 18, 2595-2599.

Abstract
Trimethylamine oxide (TMAO) is a common osmolyte and counteracting solute. It is believed to combat the denaturation induced by hydrostatic pressure as some deep-sea animals contain higher TMAO levels than their shallow water counterparts. It has also been proposed that TMAO may accumulate passively during lipid storage resulting in a correlation between lipid content and TMAO levels in some groups. Previous research showed that lipid content decreased with depth in species of Hawaiian fishes presenting a novel test of these competing hypotheses. TMAO ranged from 20.4 to 92.8 mmol/kg. Lipid content ranged from 0.50 to 4.7 % WW. After completing a comprehensive search for depths available in the literature, provided here, we analyzed TMAO and lipid as a function of average, minimum and maximum depth of occurrence for 27 species of fishes from nine orders. We found that TMAO is positively correlated with all measures of habitat depth (hydrostatic pressure) but the relationship is strongest with average depth. We further showed using phylogenetic independent contrasts that this relationship was not influenced by the evolutionary relatedness of these species. Interestingly, we found that lipid content increased with depth, in direct contrast to previous studies. TMAO is thus also positively correlated with lipid content. While we are unable to distinguish between these hypotheses, we show that TMAO is strongly correlated with depth in mid-water fishes.

Introduction
Trimethylamine oxide (TMAO) is an important cellular component in a wide range of taxa, from bacteria to humans Treacy et al., 1995). It was first described in marine organisms in ref. Suwa, 1909) as a prominent osmolyte (Cholette and Gagnon, 1973;). Later, it was shown to be a strong counteracting solute,  protecting protein structure  and function  from various environmental perturbants including, hydrostatic pressure (Gillett et al., 1997), urea and ammonia toxicity (Yancey and Somero, 1980;Minana et al., 1996), and temperature stress (Treberg et al., 2005;.

TMAO increases with habitat depth inter-and intraspecifically in benthic
fishes and skates as well as some invertebrate groups (Kelly and Yancey, 1999;Yancey et al., 2001;Yancey et al., 2002;Laxson et al., 2011;Samerotte et al., 2007), suggesting that this molecule is used to combat the increasing stress of hydrostatic pressure. Most recently,  showed a hadal snailfish at 7,000 m with a TMAO content of 386 mmol/kg, almost eight times higher than the average fish in the euphotic zone. These observed correlations with depth have been further supported by evidence that TMAO prevented hydrostatic pressure denaturation in vitro .
However, not all taxa show an increase in TMAO with depth . Some shallow-living squids have TMAO levels that approach that reported for the hadal snailfish. These authors suggest a novel mechanism of TMAO synthesis leading to accumulation as a byproduct of lipid metabolism and storage and that TMAO is not necessarily retained as a specific adaptation to high hydrostatic pressure. This hypothesis was supported by a strong correlation between total lipid content and TMAO in cephalopods as well as anecdotal evidence in a variety of other groups. For example, lipid content is often higher in deep-living and polar species, which may explain the tendency of species in those habitats to accumulate large quantities of TMAO. However, a subsequent study did not find a relationship between mean TMAO and triacylglycerol content in fishes (Samerotte et al., 2007), perhaps due to the differing time courses of accumulation and retention that resulted in differing size-scaling relationships of these two compounds.
Furthermore, an evolutionary relationship has been suggested for TMAO synthetic capacity between elasmobranchs and chimaeras , which may impose inherited limitations on accumulation potential. If phylogeny also plays a role in TMAO accumulation in teleosts, it is possible that depth-related differences are driven by evolutionary history rather than environmental selection or substrate availability. Alternatively, a relationship to phylogeny may coexist and mask environmental trends making analyses between distantly related taxa difficult.
In (1990), Childress et al. examined a population of Hawaiian mid-water fishes that exhibited decreasing lipid content with increasing habitat depth (and hydrostatic pressure). Here, we examine TMAO and lipid content in 27 species of Hawaiian fishes from the same region studied in Childress et al. (1990) to test the competing hypotheses of hydrostatic pressure and lipid content on TMAO accumulation. A stronger relationship to hydrostatic pressure should elicit an increase in TMAO with habitat depth while a decrease with depth may be seen if TMAO is primarily accumulated as a by-product of lipid metabolism. Alternatively, an increase in lipid and TMAO with depth could represent a situation in which fishes accumulate TMAO passively during lipid storage with deeper fishes retaining the molecule for further pressure counteraction.

Collection and Sampling
Fishes were collected aboard the R/V Kilo Moana (University of Hawaii) in June 2012 off the west coast of Oahu in the Hawaiian Islands. Specimens were captured using a modified opening-closing Mother Tucker trawl with 3 m 2 mouth  between depths of 50-2000 meters. Animals were recovered in a 30-l thermally insulated cod end and immediately processed for later analysis.
Individuals from 17 different species were gently blotted dry then flash frozen whole for determination of total lipid content. Additionally, muscle tissue was excised from similar specimens of the same and additional species, for a total of 27 species, and flash frozen for subsequent analysis of TMAO. All samples were collected in accordance with IACUC #AN12-07-026 and stored at -80 ᵒ C until experimentation was conducted. Representatives of each species were preserved in 5% formalin or photographed for later identification using taxonomic and identification references available in the literature.

Analytical Techniques
Total lipid content for whole body was measured using a similar method to the 2:1 chloroform to methanol extraction described by  paired down for small sample mass (Lee et al. 1996). Muscle tissue samples were deproteinated and homogenized in 5x volume 5% trichloroacetic acid (TCA) followed by spectrophotometric determination of TMAO using the ferrous sulphate-EDTA assay . All values represent averages taken from replicate individuals from n = 1 to 12.

Depth analysis
Habitat range was determined according to currently published literature values describing the depth distribution of each species. Average depth is reported as the median of the habitat range, especially in highly migratory species or as average depths specifically reported in the literature. The average depth of a species can be considered as the depth at which the fish spends most of its time (in non-migratory animals) or as a depth that represents the average level of depth stress (e.g. hydrostatic pressure) encountered by the species (migratory species). Minimum depth of occurrence (MDO) is defined as the depth below which 90% of the population of each species can be found . Here, MDOs were taken directly from the literature. Where no MDO was available, the shallowest reported depth for the species was used, substituting 10 m for those reported at the surface. TMAO, lipid and size were further analyzed against capture depth with no correlations found (data not shown).
Due to the limited amount of data available for these fishes, references were taken from studies conducted circumglobally (Supplementary Table 1). For some species, reported depths vary widely between publications; in such cases, the depths chosen for use in this study were based on the most recent and regionally specific data available. Occasionally a species vertical distribution changes with size, where smaller fish are frequently found at shallower depths (Collins et al., 2008). In these instances, reported depths are specific to the size of fish analyzed in this study; therefore, authors should be cautious when reporting these listed depths elsewhere.

Phylogenetic Comparison and Statistical Analysis
TMAO data were subjected to independent contrasts phylogenetic analysis (PIC) to determine if the phenotypic trends seen in this study could be explained by evolutionary relationships among fish species (Felsenstein, 1985;Seibel and Carlini, 2001). The phylogenetic tree used for this analysis was a compilation of trees previously published in the literature (Stiassny et al., 1996;Harold, 1998;Miya and Nishida, 1998;DeVaney, 2008;Davis, 2010;Kenaley, 2010;Denton, 2014). The tree was further rooted in the outgroup Chondrichthyes; however, this group is not included in the analysis as elasmobranch values deviate significantly from all teleost values. All data concerning TMAO, lipid, depth of occurrence and weight were further analyzed using regression analysis to assess whether any statistically significant relationships occurred. Statistics and graphs were generated using GraphPad Prism 6.0 and the phylogenetic tree used for PIC was made with statistical package R. Estimated TMAO and depth values were calculated for all ancestral nodes assuming equal branch lengths (punctuated model) and included in Supplementary Fig. 1. Further, contrast values were calculated for each node, which indicate both TMAO and depth after points have been made independent by accounting for any phylogenetic signal.

Fish collection
We collected 27 species of mid-water fishes from 15 trawls ranging in depth from 50 to 2000 m. The species represent 12 families from 9 orders. The habitat depths of each species (average, minimum and maximum) are listed in Table 1.

TMAO vs. depth
Average TMAO content ranged between 20-93 mmol/kg wet mass (Table 1), which is consistent with values reported for fishes elsewhere .
TMAO content increased linearly with all measures of habitat depth. The relationship was strongest with a species' average depth (r 2 = 0.5309, p < 0.0001; Fig. 1a) but was also significant as a function of MDO (r 2 = 0.5074, p < 0.0001; Fig. 1b) and maximum depth (r 2 = 0.2520, p = 0.0076; Fig. 1c). Separating fishes into non-migrating and vertically migrating species did not strengthen the trend with depth and variance between these groups was not significantly different (data not shown). Additionally, a phylogenetically independent analysis of the data (Phylogenetic Independent Contrasts) also resulted in a significant positive relationship between TMAO and habitat depth ((r 2 = 0.4036, p = 0.0009; Fig. 2), which suggests the trend is independent of any phylogenetic relationships across these 27 species.

Lipid vs. depth and TMAO
Lipid content ranged between 0.5 -4.7% wet weight in these fishes. Lipid values showed a significant increase with increasing average depth (r 2 = 0.2888, p = 0.0261) in the 17 species analyzed for lipid in this study. Additional lipid values taken from the literature (n = 6) strengthened this relationship (r 2 = 0.2496, p = 0.0152;  Fig. 3). Lipid values from the literature were only included for species in this study where lipid was not measured directly. Lipid also significantly increased with MDO but not maximum depth (data not shown). When divided into non-migrating and vertically migrating species, groups did not exhibit significant differences in variance (data not shown). Further, lipid was positively correlated with size in the family Myctophidae (r 2 = 0.4145, p = 0.0009) and negatively correlated with size in the species, Sternoptyx diaphana (r 2 = 0.8834, p = 0.0175; Fig. 4). However, size was not related to any measure of habitat depth for these species (data not shown). TMAO increased linearly with increasing lipid content (r 2 = 0.2744, p = 0.0309) across the 17 fish species analyzed. Adding lipid values from the literature (n = 6) strengthened this relationship (r 2 = 0.4328, p = 0.0006; Fig. 5).

TMAO vs. depth
TMAO increases with a species' habitat depth in a number of different clades including, anemones , crustaceans , Chondrichthyes (Laxson et al., 2011) and teleosts (Kelly and Yancey, 1999;Yancey et al., 2002). Yancey and colleagues hypothesize that these groups have converged on a similar mechanism of using TMAO to counteract the perturbing effects of hydrostatic pressure on protein function. TMAO is able to protect protein function against pressure better than other osmolytes such as betaine, glycine, taurine and myo-inositol.
These compounds do show some stabilizing potential against hydrostatic pressure  but higher concentrations are required to counteract comparable pressures. Additionally, TMAO acts as a universal cytoprotectant and is able to stabilize different types of proteins  as well as protein homologs from distantly related species  against denaturation. Samerotte et al. (2007) found a sigmoidal pattern in the relationship between TMAO and habitat depth in benthic teleost fishes between 0-1400 m and a linear relationship at greater depths to at least 7,000 m . The TMAO values we report fall near those found in the fishes previously examined but increase linearly with depth to 1,200 m. This supports the hypothesis that TMAO is being used to counteract hydrostatic pressure but that the relative accumulation needed for stabilization may be different between groups, ecotypes or locations. Alternatively, extracellular to intracellular volume ratios may be different between the mid-water fishes studied here and the demersal fishes examined previously, which would imply similar intracellular TMAO contents between these groups.
TMAO showed increases when examined against average, minimum and maximum habitat depth of these fishes (Fig. 1a-1c), with the strongest relationship to average depth. MDO is commonly used to relate metabolic rate to depth as metabolic rates in strongly visually-orienting taxa seem to be largely dependent on light and visual predator-prey interactions that are most important at the upper depth limit of the organism Drazen and Seibel, 2007;Seibel and Drazen, 2007).
Conversely, one might expect TMAO to correlate most strongly with the maximum pressure experienced by a species if accumulation is being driven by pressure counteraction. However, it is interesting to note that TMAO accumulation is most tightly coupled to average depth, where fishes may spend the majority of their time.
TMAO fluctuations may be inhibited by time-course restrictions, especially in diel vertical migrators, which could impose limitations on their ability to match TMAO to minimum and maximum depths and explain the strong relationship to average depth. Therefore, it is possible that these fishes are experiencing modest conformational changes to protein structure during their time spent at maximum depth. This has been shown to occur during dormancy  and other circumstances of urea destabilization . These changes could be used to facilitate metabolic suppression and energy conservation during the time spent at daytime depths among vertically migrating species. However, metabolic suppression has only been demonstrated for vertical migrators living in pronounced oxygen minimum zones (Seibel, 2011;Seibel et al., 2014) so further evidence is needed to support this supposition.

Phylogenetic Comparison
The only study to examine the evolutionary history of TMAO synthetic capacity, as described by the activity of trimethylamine oxidase (TMAoxi), found it to be a derived characteristic in elasmobranchs and chimaeras .
Species lacking measurable TMAoxi activity must rely on dietary contributions to accumulate TMAO , potentially placing ecological restrictions on their ability to use TMAO as a counteracting solute. If teleosts were to exhibit a similar phylogenetic pattern, it would suggest differing capacities for TMAO regulation between clades and could influence inherent TMAO concentrations as well as certain species ability to accumulate TMAO. We found no relationship between total TMAO content and evolutionary relatedness ( Supplementary Fig. 1). Instead, when the interrelatedness between data points imposed by evolutionary history was accounted for (contrast values), there was still a significant increase in TMAO with depth ( Fig. 2). Therefore, in these Hawaiian fishes, trends seem to be driven primarily by environmental and ecological variability and not by an innate phylogenetic signal.

Lipid vs. depth
High energy materials, such as protein and lipid, decrease with depth in Southern California fishes and are replaced by less expensive materials such as water which lowers organisms' metabolic demands and allows deep-sea species to reach larger sizes with minimal cost . A similar trend was shown for Hawaiian fishes (Childress et al., 1990) where decreasing lipid levels with depth were attributed to lower metabolic rates. However, many species increase lipid levels with depth as has been shown in copepods (Lawrence, 1976), crustaceans , zooplankton, fish (Reinhardt and Van Vleet, 1986) and cephalopods . We showed increasing lipid with average depth for Hawaiian fishes (Fig. 3). These results are opposite those reported by Childress et al. (1990).
The methods employed for lipid analysis by Childress and colleagues are best for samples of large mass with the size of these fishes averaging less than five grams wet weight. We found the modified protocol for small sample mass to yield more reliable results, perhaps explaining the discrepancy. The three values included in this study do not reflect the overall trend found by Childress of decreasing lipid with depth, most likely due to the large variability found in that study (0.2 to 10% wet weight) and the small number of data points included here. We also found evidence of changing lipid content with size in some species, which may complicate interpretations based on habitat alone (Fig. 4). It is not likely that seasonal variability plays a large role in lipid storage for the warm water fishes studied here (Childress et al., 1990). Alternatively, it is possible that the deeper living species accumulate lipid to sustain them between the intermittent meals experienced in the deep-sea environment or to fuel extensive egg-brooding periods as in the squid, Gonatus onyx (Seibel et al., 2000), and the lophigastrid crustacean, Gnathophausia ingens . The increase in lipid with depth may also be due to replacement of the gas-filled swim bladder with fatty tissue for buoyancy shown to occur in other myctophid species . In such cases, swim bladders are typically filled with wax esters which are derived from metabolic pathways independent of the diacylglycerol ethers and triacylglycerols whose formation leads to accumulation of TMAO precursors (e.g. choline; . All these factors can impart selective pressure on lipid content and it is possible the additional taxa included in the Childress study were experiencing different combinations or levels of selection resulting in the opposite trend with depth.

TMAO vs. lipid
Although TMAO may be used to combat hydrostatic pressure in many organisms, there are species that do not accumulate TMAO with depth: such as some echinoderms, mollusks, polychaetes, and vestimentiferans . These animals seem to accumulate a plethora of alternative osmolytes with potential stabilizing properties including a serine-phosphate compound, other methylamines and polyols (Yancey et al., 2002). Therefore, a number of mechanisms exist whereby fishes may be combatting hydrostatic pressure aside from TMAO accumulation. In fact, TMAO performs a number of roles including osmotic balance, buoyancy regulation, as well as urea and temperature counteraction, all of which may impart competing selection on TMAO content. However, the ability of TMAO to aid in buoyancy is limited in hypoosmoregulating fishes (Gillett et al., 1997) and plays a larger role in invertebrates and elasmobranchs. Further, TMAO regulation may be influenced by diet or passive accumulation during lipid storage .
The latter hypothesis, passive TMAO accumulation during lipid storage, has received little attention. In 2002, a new synthetic pathway for TMAO was proposed whereby phosphatidylcholine, a compound readily available from the diet or the breakdown of cellular membranes, is converted to diacylglycerol or triacylglycerol (TAG) for lipid storage. During this process a choline moiety is cleaved from phosphatidylcholine, which can then be transformed to TMAO . These authors demonstrate a correlation between lipid and TMAO content in cephalopods and discuss the tendency of many organisms from deep and polar environments to accumulate high concentrations of both TMAO and storage lipid.
Although no correlation between total TAG and TMAO was found for 15 species of fish caught in the eastern Pacific (Samerotte et al., 2007), the relationship between the two may be confounded by retention or excretion of TMAO and the active use of storage lipid for metabolic purposes. Cell membrane restructuring during growth or osmotic challenges, for example, may also lead to TMAO precursor availability without the accumulation of storage lipid . Additionally, dietary TMAO may negate the need for endogenous production. However, little information is available regarding turnover or TMAO content in the diet of these fishes making conclusions speculative. The fishes in this study show increasing levels of TMAO with total lipid content (Fig. 5), supporting evidence for the possible existence of a synthetic pathway whereby TMAO is accumulated during lipid storage.

Conclusions
TMAO was positively correlated with depth in the 27 species of Hawaiian teleost fishes studied here. Additionally, this trend was independent of phylogenetic relatedness suggesting that environment, not evolution, is playing a larger role in driving the relationship. As depth and lipid were positively correlated, it was not possible to definitively rule out either the hydrostatic pressure or the lipid accumulation hypothesis although we provide supportive evidence for both. However, the two hypotheses are not mutually exclusive and it is possible the choline substrate produced during lipid accumulation may be converted to TMAO and actively retained to counteract hydrostatic pressure. Where a MDO has not been reported, the shallowest reliable observation was used.

Introduction
Trimethylamine oxide (TMAO) is a small intracellular osmolyte and cytoprotectant. It acts as a chemical chaperone to conserves protein structure and function Yancey et al., 2001) against a number of destabilizing biotic and abiotic factors including urea , salinity , hydrostatic pressure , and temperature (see  for reviews). The majority of studies examining TMAO's ability to combat thermal fluctuations in vivo have focused on teleost fishes and their response to cold acclimation. Rainbow smelt (Osmerus mordax) increased serum TMAO in response to winter-acclimatization . Similarly, cold-water Pacific herring (Clupea harengus) express elevated serum TMAO compared to their temperate water counterparts . In fact, some of the highest values recorded for teleosts were found in Antarctic fishes, with TMAO increasing total osmolarity and depressing the freezing point . TMAO accumulation has also been linked to thermal acclimation in response to warming temperatures. The presence of TMAO increased RNA melting temperatures  and addition of TMAO to growth medium enhanced heat resistance in Escherichia coli  and spiny dogfish, Squalus acanthias, cells in vitro Kolhatkar et al., 2014).
An alternative response to elevated temperature is accumulation of heat shock proteins (HSPs). Upregulation of the inducible isoforms of HSPs has traditionally been referred to as the "heat shock response" , with preferential accumulation of HSP70 (Luft et al.,1996) a highly conserved and widely studied biomarker. Under a variety of stress conditions, HSP70 expression is induced to counteract protein misfolding (Freeman et al., 1999). HSP70 is an ATP-dependent protein chaperone that functions as part of the proteostasis network to prevent aggregation and promote refolding of misfolded and denatured proteins in collaboration with several co-chaperone partners and other chaperone systems Finka et al., 2015 for reviews). A number of fish species increase hsp70 mRNA transcript levels and HSP70 protein in response to thermal stress Iwama et al., 2004 for review), including tilapia , rainbow trout , brook trout (Lund et al., 2003), gilthead seabream ) and common carp (Sung et al., 2014). Additionally, fish preconditioned to sublethal elevated temperatures accumulate HSP70 resulting in enhanced thermotolerance (Sung et al., 2014).
As TMAO and HSP70 serve similar functions, albeit by different mechanisms, in the thermal stress response, recent attention has focused on potential interactions between these pathways. S. acanthias subjected to hyposmotic stress showed decreasing levels of TMAO in gill tissue and subsequent accumulation of HSP70, while muscle TMAO and HSP70 did not change (MacLellan et al., 2015). The authors suggest that protein stabilization provided by TMAO, when present, may inhibit the HSP70 response. A similar relationship was found in S. acanthias choroid plexus tissue, with addition of TMAO to incubation medium suppressing HSP70 accumulation in response to increased temperature .
Presence of extracellular TMAO also inhibited HSP70 accumulation following heat shock in S. acanthias red blood cells (Kolhatkar et al., 2014). Although these studies provide evidence of cooperation between the two pathways, to our knowledge no work has been done to examine these interactions in response to elevated temperatures in vivo.
Here, TMAO and HSP70 were measured in two elasmobranch species during 72 hours at elevated temperature. Marine elasmobranchs were used as a model system as they retain exceptionally high levels of TMAO (~180mmol kg -1 ) due to their unique osmoregulatory strategy of accumulating urea, a protein denaturant . The smoothhound (Mustelus canis), a species capable of TMAO synthesis, was compared to S. acanthias, which requires dietary TMAO input . This study examines whether TMAO can be accumulated as an alternative to HSP70 during the thermal stress response in these species and whether endogenous TMAO production leads to preferential use of TMAO over HSP70. previous studies Liew et al., 2013). All animals were acclimated for a minimum of 72 hours before initiation of a temperature trial.

Treatment
Control individuals were maintained at 15°C ± 0.86 for 72 hours. Temperature was increased from 15°C to 21°C at a rate of 2°C every two hours in treatment tanks and held constant at 21°C ± 0.78 for 72 hours. At time 0 control specimens were anaesthetized with 0.075 g l -1 MS-222 dissolved in a seawater bath one at a time.
Blood samples were obtained from the caudal vein using an 18-gauge hypodermic needle pretreated with 30 units heparin. Blood was centrifuged at 10,000 rpm for three minutes and the plasma isolated. The sex, weight and length were recorded for each animal. Treated individual were similarly sampled at 2 and 20 hours. Muscle biopsies were also taken from the dorsal epaxial muscle of treated individuals using a 5 mm biopsy punch (Alimed). Tissue samples (20-50 mg) were removed and the procedure completed in less than two minutes. All animals were fed mixed herring and squid at 2.5% body weight 30-48 hours after initiation of a temperature trial to ensure TMAO availability for S. acanthias individuals lacking a synthetic capacity.
At the end of 72-hours, control and treatment animals were euthanized with 0.25 g l -1 MS-222 dissolved in a seawater bath one at a time and subsequently cervically transected. The liver was weighed and blood plasma separated by centrifugation at 10,000 rpm for three minutes. Blood plasma, white and red muscle, and liver were flash frozen in liquid nitrogen and stored in a -80 °C freezer for later analysis. Terminal samples were collected in triplicate.

Analyses
Samples were homogenized 1:5 in 5% trichloroacetic acid (TCA) with a glass homogenizer or mortar and pestle on ice. Supernatant was obtained by centrifuging at 10,000 rpm for five minutes. TMAO was measured in duplicate using the ferrous sulfate/EDTA method described by . Homogenates were run in triplicate for urea using diacetylmonoxime .
Soluble protein was extracted from white muscle tissue by homogenization in a buffer containing 50mM Tris-HCl (pH 7.5), 2% sodium-dodecyl sulfate (SDS), and protease inhibitor cocktail (88666 ThermoFisher Scientific). Following homogenization, sample protein concentrations were determined by the bicinchoninic acid (BCA) assay (BCA-1KT Sigma Aldrich). Samples (15µg) were separated by SDS-PAGE using 4-12% Bis-Tris 10-well acrylamide gradient gels and transferred to a nitrocellulose membrane. Samples were immunoblotted with rabbit polyclonal anti-Hsp70/Hsc70 primary antibodies (Agrisera, AS05 083A) and goat anti-rabbit IgG secondary antibodies (Invitrogen, G-21234) at 1:10000 dilution. Hsp70 bands were detected with SuperSignal West Femto Chemiluminescent substrate (Thermo Scientific 34095). Specificity was not tested as only minor non-specific bands distinguishable from 70 kDa were detected. Images were captured using autoradiography film (Life Technologies) and band intensities quantified through densitometry analysis using ImageJ. Bands were normalized to a S. acanthias control sample run adjacent to each set of samples included in the analysis.

Statistics
Time points within control groups were compared using one-way paired student's t-test. Where sample sizes differed between time points, data were compared using one-way unpaired student's t-test. Time points within treatment groups were compared using one-way RM ANOVA with Holm-Sidak post-hoc test. Tissue comparisons between treatments within species were determined using one-way unpaired student's t-test. Linear regression was performed to assess variable scaling with mass. Significance was set at p<0.05. All analyses and graphs were generated with GraphPad Prism 7.0.  4B).

Urea
There was no significant difference in urea concentration between treatments  (Fig. 5B).

HSP70
There was no significant difference in HSP70 between 15°C and 21°C white muscle in M. canis ( Fig. 6A and C) or between either M. canis group and S. acanthias 15°C control. In contrast, 21°C S. acanthias exhibited an almost 3 fold higher relative HSP70 concentration than 15°C individuals at 72 hours ( Fig. 6B and D, p=0.0008).

Discussion
Both S. acanthias and M. canis are acutely temperature sensitive with seasonal thermal fluctuations regulating migration patterns both in -offshore and latitudinally . Adult S. acanthias in the northwest Atlantic express a thermal preference of ~10-11°C (McMillan and Morse, 1999; with catches occurring at temperatures as low as 6°C . This population of S.
acanthias was found off the coast of South Carolina within a range of 10.5 -29.1°C and exhibited a mean catch temperature of 15.45°C. S. acanthias appeared in this area when temperatures dropped to 13°C and departed when temperatures reached 19°C .
Catch rates for the northwest Atlantic population of M. canis decline significantly at temperatures above 21°C . M. canis was shown to occur within a narrower thermal range of 12.2 -24.5°C with a mean of 17.72°C.
Temperature dictated migration in this region with M. canis arriving when temperatures dropped to 18°C and disappearing when temperatures rose above 19°C . The inner waters of estuaries serve as nurseries for this species , and likely represent the upper end of their recognized thermal range.
These upper temperatures do not accurately describe the preferred habitat of adults that are found deeper and further offshore -up to depths greater than 300m .
Both species co-occurred in this study.  (Kolhatkar et al., 2014). At the organismal level, we find no evidence of plasma or tissue TMAO accumulation in 21°C treated individuals relative to 15°C controls in these two shark species (Fig. 1, 2, 4). Although TMAO increased over time in M. canis 21°C white muscle (Fig. 2), treated individuals did not exhibit higher TMAO than 15°C controls and thus this increase cannot be directly attributed to a temperature effect. If these sharks do induce TMAO in response to elevated temperature, 72 hours may not have been long enough for the increase to become apparent; although elasmobranch red blood cells in culture were able to accumulate TMAO from the external medium within two hours of encountering thermal stress (Kolhatkar et al., 2014). The dynamics regulating these interactions may be more complex at the organismal level or the +11°C employed in the Kolhatkar study may have elicited a stronger cellular response than the +6°C administered here.
Further, Kolhatkar targeted an elevated temperature of 24°C compared to our 21°C, a condition deviating more significantly from this species thermal optimum and possibly resulting in a higher level of treatment stress.
Elasmobranchs with no endogenous synthetic capacity for TMAO are dependent on dietary contributions for maintenance and accumulation . Although food was offered during the experiment, consumption was suppressed in some heat stressed individuals (personal observation), which would have limited availability of dietary TMAO for accumulation. This may have contributed to the lack of increase at 21°C in nonsynthesizing S. acanthias (Fig. 4A). In fact, the repressed feed response at elevated temperature is problematic as recent evidence suggests all elasmobranchs depend on dietary contributions for TMAO maintenance regardless of synthetic capacity (Bockus and Seibel, in prep). This could further explain the lack of accumulation with elevated temperature in either species (Fig. 4A-B). However, plasma TMAO spikes 20 hours postprandially  providing ample time for assimilation in individuals that did feed.
Another possibility is that TMAO accumulation is not initiated as a thermal protective mechanism at the whole organism level, with preferential regulation of other cytoprotective pathways. Accumulation may be restricted by the high levels of urea found in elasmobranch tissue (Fig. 5A-B). Urea is retained in elasmobranchs as a major osmolyte and their primary form of nitrogenous waste . Here, urea averaged ~400 mM in plasma and muscle of these species, levels well established in the literature for marine elasmobranchs (Kempton, 1953;. Regulation of urea and TMAO is tightly coupled as the two act in an additive capacity for optimal protein stabilization (Mello and Barrick, 2003) and are generally found in a 2:1 ratio of urea to TMAO + other stabilizing osmolytes . The high concentration of urea in elasmobranch cells may diminish their capacity to adjust TMAO to combat alternate stressors due to co-regulation of these compounds. In comparison, ammonotelic teleost fishes have the ability to synthesize urea, particularly during early development (LeMoine and Walsh, 2013), but retain negligible levels . These species may serve as better models for future studies examining TMAO regulatory processes in response to environmental perturbations.
Like TMAO, HSP70 is a known component of the stress response that acts on diverse protein substrates to protect them from inactivation. Elevated HSP70 levels elicited by increasing temperatures have also been shown to provide additional protection against subsequent stresses such as exposure to environmental pollutants  and ammonia (Sung et al., 2014). Surprisingly, no increase in HSP70 was observed in M. canis at 21°C compared to 15°C. However, we observed significantly higher levels of HSP70 in white muscle of S. acanthias at 21°C compared to 15°C (Fig. 6). Although S. acanthias may transiently experience temperatures in the mid to upper 20s in the wild, the present study shows they are temperature stressed at 21°C, further supported by previous catch data and adults preference for cooler, deeper waters . As TMAO was not shown to increase in either species (hypothetically with the ability to combat the HSP70 response), the question remains as to why HSP70 increased in S. acanthias but not M.

canis.
M. canis is not likely more temperature tolerant than S. acanthias, based on this species' narrower thermal range. Therefore, the lack of HSP70 accumulation in M.
canis at 21°C is surprising. As has been shown to occur in Antarctic ice fishes , constitutively high expression of inducible HSP70 in M. canis may impart partial protection against a rise in temperature.
Although 15°C M. canis HSP70 levels were not significantly higher than 15°C S. acanthias, relative expression was around 1.0 compared to 0.5 respectively (Fig. 6).
Further, the methods used in this study did not differentiate between constitutive and inducible isoforms, which would provide a clearer understanding of innate differences between these species.
Aside from the regulation of HSPs, M. canis may combat thermal stress using different mechanisms altogether. A number of other osmolytes are categorized as "counteracting" solutes  and have the ability to protect cellular function against a variety of stressors, including thermal fluctuations. These include certain methylamines, carbohydrates and amino acids; although TMAO is one of the most effective stabilizers, lowering the K m of NADH under elevated pressure more than betaine, myo-inositol or glycine . Organisms express a wide array of these compounds  with preferential accumulation changing by clade. Some cephalopods and molluscs retain betaine as their primary osmolyte, while crustaceans and decapods accumulate glycine, and euphausids and fishes express high levels of TMAO . However, significant levels of TMAO and other osmolytes, such as taurine, are also retained in many of these organisms.
Accumulation of other protein stabilizers in response to thermal stress and their cooperation with HSP70 is an understudied alternative in need of further investigation.
Here, we show no increase in plasma or tissue TMAO in response to thermal stress.   and upregulation can occur in a matter of minutes. However, evidence suggests changes in TMAO may take 2 to 20 hours depending on availability of substrates for synthesis vs. absorption from the diet Kolhatkar et al., 2014). TMAO regulation may be further limited by ecological availability of prey items and coupling to urea in ureosmotic organisms. Much remains to be done to elucidate the role of alternate counteracting solutes in the thermal stress response and how these cytoprotectants work in concert in elasmobranchs. Squalus acanthias and 15°C control (n=8) and 21°C treated (n=8) Mustelus canis.
Data presented as means ± SEM. No significant differences found between treatments within species (one-way unpaired student's t-test) or between time points within treatments (one-way RM ANOVA or one-way paired student's t-test, p<0.05).  control. However, 21°C treated S. acanthias was higher than S. acanthias 15°C control (two-tailed student's t-test, p=0.0008).

Introduction
Elasmobranchs (sharks, skates and rays) retain a suite of molecules, including sugars, polyols, free amino acids, methylamine compounds and urea, to roughly match the osmotic strength of their tissues to that of seawater (~1,000 mOsm), an osmoregulatory strategy referred to as osmoconformation. These constituents differ in their interactive properties within the cell, and have been categorized as perturbing, compatible (having a minimal macromolecular effect) and counteracting solutes, which stabilize proteins that may be otherwise denatured by cellular or external stressors .
Unlike other marine animals, most elasmobranchs studied to date retain urea, a strong destabilizing agent, as their primary osmolyte. Urea concentrations may reach 400 mM, and account for almost half an individual's osmotic pressure; values far above the threshold needed to impart protein denaturation. To offset urea's perturbing effects, elasmobranchs also accumulate large concentrations of the counteracting solute trimethylamine oxide (TMAO). TMAO content is tightly correlated to accumulation of urea and organisms generally exhibit a ratio of 2:1 Urea:TMAO + other methylamines , making TMAO the second most common intracellular osmolyte in these organisms.
Three distinct modes of reproduction -oviparity, ovoviparity and viviparityare present in elasmobranchs and development of the osmoregulatory system in the embryo is hypothesized to depend on the composition of the medium surrounding the egg and, consequently, on the reproductive strategy employed by the species . The enzymes necessary for urea synthesis are functional early in both shark and skate ontogeny  and plasma values reported for late-term embryos suggest a fully developed osmoregulatory capacity . However, plasma osmolytes may not represent the osmotic condition of the entire individual. Osmolyte values in embryonic tissue have not, with few exceptions , been previously studied and only one of these conducted in an ovoviviparous species .
In this study, we present osmolyte values, including urea and TMAO, for the muscle tissue and yolk of late term embryos and neonates (less than 24 hours old, hereafter combined and referred to as pups) as well as adults of the ovoviviparous spiny dogfish, Squalus acanthias. We sought to determine how the intracellular osmotic state changes through ontogeny and how it compares to literature values of elasmobranch species differing in their reproductive strategies.

Sample collection
Adult Adults were held up to two months and fed a mixed diet of herring and squid twice weekly at 2.5% body weight. Neonate dogfish (younger than 24 hours) were berthed in captivity and late term embryos dissected live from in utero. As no significant variability was found between pups collected by these various methods all data have been combined. Only late term (>30 cm, yolk sac still present) or fully developed (no yolk sac) individuals were used in this study. Due to collection methodology, it was not possible to discriminate between the two early ontogenetic life stages during analysis.
All animals were euthanized with MS-222 (0.15 g l -1 seawater) in accordance with IACUC #AN13-05-020. White muscle was excised from the dorsal epaxial of adult (n = 11) and pup (n = 37) spiny dogfish and immediately flash frozen in liquid nitrogen for later analysis. When yolk sacs were present, they were also removed and similarly frozen.

Laboratory analyses
Tissue was homogenized 1:5 (wet weight:volume) in 5% saturated trichloroacetic acid. Homogenates were then assayed for TMAO by the ferrous-sulfate method whereby concentration is determined spectrophotometrically by a colorimetric reaction between reduced TMAO and 2% picric acid .
TMA is generally very low in marine elasmobranchs (< 2 mmol kg -1 ) and no corrections for endogenous TMA have been made. Homogenates were further analyzed for urea, again a colorimetric reaction that spectrophotometrically determines concentration against known standards . Fresh muscle tissue was homogenized 1:1 in deionized water and total osmolarity (mOsm) calculated using the freezing point of the solution as determined by the automatic osmometer micro-osmette model 5004 (Precision Systems Inc.). Group means ± s.e.m.
were compared with two-way Student's t-tests and statistics and graphs generated using GraphPad Prism 7.0. Student's t-test for osmolarity was run with Welch's correction for unequal variance.

Results and Discussion
Adult spiny dogfish, S. acanthias, exhibited an average tissue TMAO content of 154.84 ± 7.43 mmol kg -1 wet wt. In contrast, pups had an average TMAO content of only 42.13 ± 2.98 mmol kg -1 , more than 3.5 times lower than the adult average (p < 0.0001) and slightly lower than the whole embryo content of 67.8 mmol kg -1 reported by . Urea followed a similar trend with adult values averaging 463.70 ± 28.00 mmol kg -1 wet wt. and pup concentrations less than half that (220.95 ± 15.21 mmol kg -1 ; p < 0.0001, Fig. 1). Yolk TMAO and urea were 58.41 ± 2.99 mmol kg -1 and 178.13 ± 10.34 mmol kg -1 respectively, similar to pup tissue values.
The discrepancy between adult and pup values of the two primary osmolytes reported here in Squalus acanthias is in contrast to the situation described for skates.  showed little skate, Raja erinacea, embryos and adults contain similar concentrations of both osmolytes as early as four months after oviposition and well before the embryo hatches from the egg case at nine months. Big skate, Raja binoculata, embryos were also shown to exhibit urea at similar concentrations to adults with no change in urea or TMAO through development . The species specific differences between groups may be attributed to alternate modes of reproduction in the ovoviviparous spiny dogfish compared to the oviparous skates.
The majority of literature in this area has concluded that early osmoregulatory ability in elasmobranchs can best be ascribed to reproductive strategy .
The embryos of oviparous species develop inside a highly permeable egg case deposited on the sea floor, subjecting the developing embryo to full strength seawater and requiring a working osmotic system soon after fertilization . Ovoviviparous species on the other hand begin in egg cases, which hatch inside the mother where the embryos finish developing in utero. The uterine environment during the first part of development bathes the embryo in a solution similar to the composition of maternal plasma establishing low salt and urea gradients and relaxing pressure on the osmoregulatory needs of the early individual.
During the second part of development, the uterine fluid undergoes a compositional change and the embryo is surrounded by a medium with characteristic similarities to seawater . At this time the individual must be able to fully osmoregulate in order to maintain water and solute balance .
Therefore, ovoviviparous species have a delayed requirement for osmoregulatory development based on ionic and osmotic gradients established by the surrounding medium. Lastly, viviparous species complete development inside a placental environment with fluid resembling the composition of maternal plasma, greatly reducing the need for embryonic ionic and osmotic regulation (see  for review).
These differences in embryonic environment readily explain why the oviparous little skate would exhibit early osmoregulatory development. It may also explain why near-term spiny dogfish display a fully operational osmoregulatory system . In accordance with this view, we found the tissue osmotic strength of the spiny dogfish pup to be 943.3 mOsm and closely match that of adults at 1,009 mOsm (p = 0.3112, Fig. 2) and the expected value of seawater around 1,000 mOsm. Therefore, although the two most important osmolytes, urea and TMAO, were found at concentrations significantly lower than expected, total osmolarity is enough to maintain osmotic balance against seawater. These data suggest a large shift in tissue osmolyte composition between late term / early life pups and adult spiny dogfish, during which urea and TMAO values progressively increase, while some other osmolytes or salts decrease, to adult levels.
Low pup TMAO contents may be explained by limited availability of the compound itself. Spiny dogfish do not exhibit the enzyme responsible for endogenous TMAO synthesis  and rely to an extent on absorption from the diet Bockus and Seibel, in prep); possibly limiting accumulation until the pup has eaten its first meal. This limitation may also facilitate the concomitant observation of low urea in these pups due to a limited counteracting capacity. As mentioned above, optimal protein stabilization is achieved by a 2:1 ratio of urea to TMAO + other methylamines and previous studies have shown R. erinacea embryos to maintain this ratio early in embryogenesis . If TMAO is limiting, perhaps accumulation of urea is restricted until the cell is able to protect against its destabilizing properties via TMAO (or other counteracting solute) accumulation.  also showed R. erinacea embryos to rely more heavily on TMAO than other counteracting solutes during embryonic development, further supporting the possibility that dietary regulation of TMAO may be limiting urea accumulation.
We show that pups contain significantly lower tissue urea and TMAO than adults but exhibit a similar total osmotic pressure. These findings imply a large shift in the cellular osmolyte constituents present between birth and adulthood, during which urea and TMAO increase significantly and take over as primary osmolytes; a shift that may impose additional restrictions on this group. For example, low levels of TMAO, which is the most effective counteracting solute present in the tissue, due to possible accumulation restrictions in early ontogenetic stages may render pups particularly susceptible to environmental stress. Our findings also raise an important question regarding which solute (or solutes), if not urea and TMAO, act as the primary osmolyte in pup spiny dogfish. Additionally, this data is in contrast to the oviparous species previously studied, which retain a similar osmolyte milieu through development and into adulthood. This suggests divergence in the osmotic mechanisms employed by the early life individuals of different elasmobranch groups and confirms the need to differentiate developmental osmoregulatory data based on reproductive strategy. and urea concentrations (right Y-axis) were compared in the white muscle of adult and pup spiny dogfish (S. acanthias). All values reported as mean±s.e.m. and analyzed with two-way Student's t-tests. Adult dogfish (n = 9) exhibited 3.5 fold higher TMAO (mmol kg -1 wet wt.) than dogfish pups (n = 28) and urea concentrations (mmol kg -1 ) in adults (n = 11) were twice those seen in similar pups (n = 37). TMAO was significantly lower in the white muscle of S. acanthias pups (p < 0.0001) than the white muscle of adults, with urea also accumulating at significantly lower concentrations (p < 0.0001). Yolk TMAO (n = 10) and urea (n = 12) values also shown.