Elevated osmolytes in rainbow smelt: the effects of urea, glycerol and trimethylamine oxide on muscle contractile properties

Rainbow smelt, Osmerus mordax (Mitchill) are eurythermal fish that encounter extreme cold in winter and moderate summer temperatures (Nellbring, 1989). Overwintering smelts in the North Atlantic endure water temperatures below the freezing point of the body tissues of most animals (e.g. −1.8°C; Raymond, 1995). Smelt not only survive but are also remarkably active under these conditions (Driedzic, 2015). To avoid freezing, these fish display a variety of thermal acclimation responses related to prevention of freezing in sub-zero temperatures (Driedzic and Ewart, 2004).

The muscle function of smelt in winter acclimates to the cold, with apparent changes in myosin and parvalbumin content in the myotomal muscle (Woytanowski and Coughlin, 2013). These changes in the protein composition of the muscle lead to faster contractile properties, such as faster maximum shortening velocity and faster muscle relaxation in cold- versus warm-acclimated smelt. In addition, these cold-acclimated smelt display faster swimming speeds when tested at a common temperature (10°C) against warm-acclimated smelt (Woytanowski and Coughlin, 2013).

Another element to the thermal acclimation response in these smelt is the accumulation of anti-freeze proteins (AFPs) and a variety of osmolytes to prevent freezing in sub-zero temperatures (Driedzic, 2015). When exposed to cold temperatures (both in a laboratory setting and naturally during the winter months), smelt show increased levels of glycerol, urea, trimethylamine N-oxide (TMAO) and AFPs in their plasma, muscle and other tissues (Raymond, 1998; Treberg et al., 2002; Liebscher et al., 2006). Over winter, glycerol content is elevated to >200 µmol kg⁻¹, while urea and TMAO are maintained at 40–50 µmol kg⁻¹ in the plasma, muscle and other tissues (Raymond, 1994; Treberg et al., 2002). Total osmolarity in winter in smelt tissues can reach 1000 mOsm (Raymond, 1992).

The influence of the osmolytes glycerol, TMAO and urea on muscle function was the focus of this work. (AFPs, although vital for survival, do not act as significant osmolytes and were therefore not part of this study.) Little research has examined how these osmolytes affect muscle activity. Some research suggests that urea and TMAO affect the actin–myosin interaction in different ways and balance each other (Kumemoto et al., 2012). TMAO is known to support protein folding, while urea is a protein inhibitor (Yancey, 2005; Yancey and Siebenaller, 2015). Urea appears to enhance freeze survival and preserve muscle function during recovery from winter in wood frogs (Costanzo et al., 2008, 2013). In elasmobranchs, TMAO and urea are routinely at a ratio of approximately 1:2 TMAO:urea, where they are predicted to balance each other relative to protein function (Treberg et al., 2006). In smelt, that ratio is reversed (Raymond, 1994), but no physiological data are available to examine how they influence muscle or other physiological functions. Yancey (2005) suggests that glycerol also plays a stabilizing role in protein function. Glycerol does participate in the cold-acclimation response of Cope's gray tree frog (Zimmerman et al., 2007) and various fish species (Raymond, 1992), but no research has examined how the dramatic increase in glycerol content in winter (>200 mmol kg⁻¹ in smelt muscle) influences force production and power output in muscle.

The goal of the present work was to examine how osmolytes such as urea, TMAO and glycerol affect muscle function. Previous research has not determined how these osmolytes affect muscle function at the tissue level. We hypothesized that the accumulation of osmolytes in the tissues of cold-acclimated fish collectively enhances muscle function in smelt myotomal muscle. Urea, glycerol and TMAO are known to contribute to thermal acclimation by enhancing freeze resistance. We postulated that they also promote muscle function. To test this, we examined muscle from fish acclimated to 10°C. These fish have low endogenous levels of the experimental osmolytes in their tissues. By exposing white muscle preparations to urea, TMAO and glycerol individually and in combination, the effects of these osmolytes could be isolated from other elements of the cold acclimation in rainbow smelt. Measurements of isometric contractile properties, muscle shortening velocity and power output were then collected and compared between the pre-treatment condition and 1 h after the various osmolyte treatments for each individual muscle bundle. This provides a rigorous comparison that compensates for the inter-muscle bundle variation inherent in muscle physiology research. In addition, control treatments were used in which contractile properties were measured initially and after 1 h without the addition of any osmolytes.

Hatchery-bred rainbow smelt parented by anadromous wild adults were obtained courtesy of John Whalen from the Maine Smelt Hatchery at Harmon Brook Farm in Canaan, ME, USA. The fish (N=38) had a mean (±s.d.) body mass of 6.7±4.0 g and total length of 10.8±1.9 cm. They were approximately 1 year old and included male and female fish. The fish were maintained at 10°C in a 600 l re-circulating aquarium system and fed a diet of bloodworms. For comparison experiments, two rainbow trout (Oncorhynchus mykiss, Walbaum) were obtained from the Huntsdale Fish Culture Station in Carlisle, PA, USA, of the Fish and Boat Commission of the Commonwealth of Pennsylvania. The trout, with a mean body mass of 374 g (243 and 505 g) and mean total length of 25.6 cm (26.7 and 32.0 cm), were maintained at 10°C in a 265 l recirculating aquarium system on a diet of pelleted trout food (Trout Grower, Ziegler Brothers). All animal handling and care were approved by the Widener University Institutional Animal Care and Use Committee in accordance with the Guide for the Care and Use of Laboratory Animals of the National Research Council.

For each experiment, a smelt was killed by decapitation with a scalpel. The hypaxial muscle from the pectoral girdle to the pelvic girdle was quickly dissected away from the body and maintained in physiological saline at 4°C (Coughlin and Carroll, 2006). The physiological saline was composed of (mmol l⁻¹): 132 NaCl, 2.6 KCl, 2.7 CaCl₂, 10 imidazole, 10 pyruvate and 1 MgCl₂, pH buffered to 7.7 (Altringham and Johnston, 1990). Individual bundles were then dissected from the hypaxial musculature. The bundles included one myomere with an active muscle length of ∼4 mm and ∼0.5 mm² cross-section. Responsive muscle bundles were tied into one of two muscle mechanics apparatuses composed of a servomotor (Aurora Scientific 318B), force transducer (Aurora Scientific 404A), temperature-controlled chamber of re-circulating physiological saline, platinum stimulating electrodes and PC control via custom-written LabView software and an input/output computer board (National Instruments). Silk thread (6-0) was used to tie the myosepta found on each end of the myotomal bundles to the servomotor and force transducer. Care was taken to minimize the non-contractile elements in the preparation. All muscle mechanics experiments were carried out at 10°C. For a given fish, two bundles were employed if possible, one in each mechanics apparatus. The duration of the experiments (5–6 h) limited the number of bundles that could be studied for each fish. Treatments were allocated randomly until the sample size noted below was achieved.

At the start of muscle mechanics experiments, muscle length and stimulation conditions were optimized for each bundle to maximum tetanic force output. Once conditions were optimized, the bundle was allowed to recover for 1 h and then isometric twitch and tetanus contractions were recorded. For tetanic contractions, time of activation (TA) was defined as the time from 10% to 90% of maximum peak isometric stress, and time of relaxation (TR) was the time from 90% to 10% of peak isometric stress. In addition, twitch time (TW90) was defined as the time from stimulation to 90% recovery (10% of peak isometric stress) in twitch contractions. For experiments focusing on isometric contractile properties, the bundle was then exposed to 50 mmol l⁻¹ TMAO, 50 mmol l⁻¹ urea or 200 mmol l⁻¹ glycerol to mimic the levels observed in overwintering smelt in the wild (Raymond, 1994; Treberg et al., 2002). Two combination conditions were also used: TU, 50 mmol l⁻¹ TMAO and 50 mmol l⁻¹ urea; and TGU, 50 mmol l⁻¹ TMAO, 50 mmol l⁻¹ urea and 200 mmol l⁻¹ glycerol. Concentrated (1 mol l⁻¹) urea, TMAO and/or glycerol in physiological saline were added to the experimental physiological saline to generate the appropriate concentration of osmolyte treatment for each bundle. In addition, a control treatment was also used in which the bundle remained in normal physiological saline for both pre-treatment and post-treatment measurements. After a 1 h exposure to a given osmolyte treatment, the isometric contractile properties of each muscle bundle were again determined. Urea and glycerol have rapid equilibration across the cell membrane (Yancey, 2005), indicating that 1 h would be sufficient to achieve high intracellular concentrations of these osmolytes. Muscle in rainbow smelt (Raymond, 1998) and other vertebrates (Treberg et al., 2006) accumulates TMAO in vivo, suggesting that the membrane is permeable to this osmolyte. Sample size was N=5 for each condition for isometric contractions.

For the same set of experimental osmolyte treatments, including the control treatment, the effects of the osmolytes on oscillatory power was determined across a range of oscillation frequencies relevant to swimming in smelt. Oscillatory power output was measured using workloop experiments (Rome et al., 1993; Coughlin, 2000; Woytanowski and Coughlin, 2013). Oscillatory stimulation conditions (phase and duration) were developed in pilot experiments to maximize power output at each frequency across a range of stimulation frequencies (2–8 Hz) that spanned observed tailbeat frequencies in swimming fish (Woytanowski and Coughlin, 2013). A standard ±3% length change was used. The stimulation conditions were not intended to replicate in vivo muscle conditions but to allow comparison of power output before and after treatment with the osmolytes described above. For each bundle, power output was determined in the following frequency order: 2, 3, 4, 5, 2, 6, 7, 8 and 2 Hz. The 2 Hz conditions were repeated throughout the experiment to assess the level of fatigue, and only bundles that maintained at least 90% initial power at 2 Hz were retained. After determining power output in normal physiological saline, the osmolyte treatment was applied, and the power output across the frequency range was measured again after 1 h. For the control treatment of normal physiological saline, power output was measured across the range of oscillatory frequencies twice, 1 h apart. Sample size was N=6 for each treatment, and these bundles were a different set from those used for isometric contractions above.

Isovelocity experiments were used to measure maximum muscle shortening velocity, V_max, in muscle bundles for the TMAO treatment. Force–velocity curves were constructed, and V_max was determined as described previously (Woytanowski and Coughlin, 2013). A series of isovelocity ramps of increasing velocity were imposed on the muscle. Muscle tension was determined during each ramp, generating pairs of force versus velocity points to be plotted as a force–velocity curve (Coughlin et al., 1996). Peak tetanic force was monitored throughout the experiment with each ramp, and a maximal isometric contraction was generated at the end of the series of ramps. Muscle bundles were included if the tension produced remained ≥90% of the initial maximum tetanus. After correcting for passive tension, V_max was found by fitting the Hill muscle model. After determining V_max in normal physiological saline, 50 mmol l⁻¹ TMAO was added to the saline. After 1 h, V_max was determined again. In addition, a control treatment of normal physiological saline was used with the same before and after measurements being made 1 h apart. Sample size was N=5 for TMAO and N=4 for normal physiological saline. These muscle bundles were different from those used above for isometric and oscillatory contractions.

For rainbow trout experiments, isometric contractions were examined in trout white bundles before and after the TMAO treatment. A total of four white muscle bundles were used from each of two fish for a total sample size of N=8. Preparation of muscle bundles and the procedure for isometric contractions was as described above for rainbow smelt.

To assess the success of the osmolyte treatments in altering the muscle ionic environment, glycerol and urea assays were used to determine the content of these two osmolytes in muscle bundles exposed to the glycerol, urea or TGU treatment. Muscle samples were homogenized in a solution containing 250 mmol l⁻¹ sucrose, 100 mmol l⁻¹ KCl, 20 mmol l⁻¹ Tris-base and 5 mmol l⁻¹ EDTA. Muscle urea content (mmol kg⁻¹) was determined using the BioAssay Systems QuantiChrom Urea Assay Kit (DIUR-500) with muscle urea content corrected for dilution in homogenization buffer. Muscle glycerol content (mmol kg⁻¹) was similarly determined using the BioAssay Systems EnzyChrom Glycerol Assay Kit (EGLY-200). A TMAO assay was not available for this work. Samples of saline from the experiments were collected and subsequently analyzed for osmolarity using a Wescor Vapro^® Vapor Pressure Osmometer 5520.

The muscle physiology variables reported in this work are expressed relative to pre-treatment values. For instance, all isometric variables (tetanic and twitch contraction force production, as well as TA, TR and TW90) are expressed as a ratio calculated as the value for each variable measured after osmolyte treatment divided by that measured 1 h earlier in normal physiological saline. A ratio of 1 indicates the osmolyte treatment had no effect, while ratios >1 or <1 indicate that the treatment affected the variable. The ratios were log transformed and analyzed using single-factor ANOVA and a post hoc Tukey multiple comparison test. The results of multiple comparisons for significant ANOVA are shown in the figures; significant differences are indicated for treatments using different letters (X and Y). The independent variable was osmolyte treatment with six levels: urea, glycerol, TMAO, TU=TMAO+urea, TGU=TMAO+glycerol+urea, and control. The control condition involved measurements made in normal physiological saline both before and after the 1 h period. A similar approach was used for oscillatory power output except that the frequency of oscillation was an additional independent variable so that a two-factor ANOVA was employed. Maximum power output by each bundle was also examined as a ratio relative to pre-treatment, log transformed and analyzed using single-factor ANOVA and a post hoc Tukey test. For V_max measurements of rainbow smelt muscle treated with TMAO or normal physiological saline, a t-test was used. For the rainbow trout experiments employing treatment only with TMAO, each isometric contractile variable was again expressed as relative to a pre-treatment measurement. For the rainbow trout data, a single-sample t-test was used to evaluate whether the mean ratio differed from 1. For t-tests, determination of P was based on a two-tailed distribution. No data were excluded from this study. For graphs, mean±s.e.m. values are shown. Data were not included from a pilot study that involved different incubation times following osmolyte treatment.

The physiological saline used in this study varied in osmolarity from 328 mOsm for normal physiological saline to 372 mOsm for the urea treatment, 590 mOsm for the glycerol treatment and 653 mOsm for the combined TGU treatment (Fig. 1A). The actual accumulation of osmolytes in muscle differed from the levels found in the experimental saline. Urea and glycerol both accumulated in muscle samples at a higher concentration than in the experimental osmolyte treatments that included those osmolytes (Fig. 1B,C). For instance, the urea content of white muscle samples exposed to 50 mmol l⁻¹ urea averaged ∼90–100 mmol kg⁻¹ of muscle. Similarly, the glycerol content of muscle bundles exposed to 200 mmol l⁻¹ glycerol was 340–349 mmol kg⁻¹ of muscle.

Osmolyte treatment affected force production in varying ways. Tetanic force production increased after TMAO treatment (Fig. 2A) but decreased after treatment with urea (Fig. 2B) and glycerol. The effect of osmolyte treatment on both maximum tetanic and maximum twitch force production was significant (ANOVA, P=0.003 for tetanic contraction, P=0.004 for twitch contraction; Fig. 2C,D). For both tetanic and twitch contractions, post hoc multiple comparisons (Tukey test, P<0.05) revealed that TMAO and the TU combination treatment differed significantly from the glycerol treatment in terms of force production. While muscle bundles exposed to TMAO or TMAO in combination with urea showed an increase in force production with treatment, those exposed to glycerol showed a substantial drop in force production (Fig. 2C,D). Osmolyte treatment did not affect TA and TR of tetanic contractions (ANOVA, P>0.05 for TA and TR; Fig. 3A,B), but did significantly alter TW90 (ANOVA, P=0.018; Fig. 3C). A post hoc multiple comparison (Tukey) test (P<0.05) revealed differences between TMAO and glycerol treatments. While muscle bundles exposed to TMAO showed little change in TW90 with treatment while those exposed to glycerol showed significantly longer twitch times.

Osmolyte treatment altered muscle work and power output under oscillatory conditions. When a length change was introduced along with stimulation, TMAO led to more work per oscillatory cycle as evidenced by the larger area of the workloop after treatment compared with that pre-treatment (Fig. 4A). Urea led to a reduced work output per cycle (Fig. 4B). Across a range of oscillatory frequencies, osmolyte treatment affected power output (two-factor ANOVA, P<0.001 for treatment and frequency), although there was no interaction between the two independent variables (two-factor ANOVA, P>0.99). In general, urea reduced power output across all frequencies compared with the other treatments and the control (Fig. 5A). At lower frequencies (2–5 Hz), TMAO led to the highest power output. The control treatment demonstrated the repeatability of the approach, as power output after the control treatment (fresh saline) was statistically identical to pre-treatment power output across the range of oscillatory frequencies (Fig. 5A).

Normalized power output under control conditions revealed an optimal oscillatory frequency of 5 Hz, with lower power at higher and lower frequencies (Fig. 5B). Muscle bundles treated with TMAO, urea and glycerol showed the same pattern of increasing power output from 2 to 5 Hz. Above 5 Hz, urea and glycerol led to significantly depressed power production relative to control and TMAO-treated muscle bundles (Fig. 5B). At 8 Hz, normalized power output for muscle bundles was significantly affected by osmolyte treatment (ANOVA, P=0.015), with a multiple comparison indicating that both TMAO and control treatments generated significantly higher power at 8 Hz than glycerol-treated bundles (Tukey test, P<0.05). At the lower end of the oscillatory frequency range, TGU treatment led to greater power output when normalized to maximum output. For instance, TGU-treated muscle bundles had a lower optimal frequency of oscillation (4 Hz) compared with the other treatments (Fig. 5B). Treatment had a significant effect on normalized power output at 2 and 3 Hz oscillatory frequencies (ANOVA, P<0.001 for both), with TGU treatment generating significantly higher normalized power than the other treatments (Tukey test, P<0.05 for TGU versus TMAO, urea, glycerol and control at 2 Hz, P<0.05 for TGU versus TMAO, urea and control at 3 Hz). However, TGU treatment, which represents the ‘winter condition’ of rainbow smelt, was reversible. Normalized power output of a single bundle showed the same leftward shift after treatment with TGU followed by a rightward shift after the return to normal physiological saline (Fig. 6).

Maximum power output was significantly affected by osmolyte treatment (ANOVA, P=0.022; Fig. 7). Maximum power output was highest in the presence of TMAO and decreased under TGU treatment. A post hoc multiple comparison (Tukey) test (P<0.05) revealed that the maximum power output of TMAO-treated and control bundles was significantly higher than that of TGU-treated bundles (Fig. 7).

TMAO treatment did not affect V_max (Fig. 8). Muscle bundles treated with TMAO and those given a control treatment of normal physiological saline both showed a slight increase in V_max after treatment but did not differ from each other. Isometric contractions by trout white muscle bundles showed no effect of TMAO on tetanic and twitch force production, TA, TR and TW90 (Fig. 9).

The osmolytes tested in this study affected muscle function in rainbow smelt at physiological concentrations in experiments carried out at 10°C. TMAO was associated with the highest force production by muscle in isometric contractions and the highest level of muscle power output during oscillatory activity, although in both cases it was not significantly higher than control. However, glycerol and urea both led to decreased muscle isometric force production and reduced muscle power output during oscillatory activity. Urea led to lower power output across a range of oscillatory activity (Fig. 5), while glycerol led to the most substantial drop in force output in isometric contractions (Fig. 2). Although activation and relaxation times from tetanic and maximal isometric contractions were not affected by treatment, the total duration of submaximal twitch contractions was (Fig. 3). Glycerol led to significantly longer twitch contractions than the other osmolyte treatments. The combined treatment of muscle with TMAO, urea and glycerol (TGU) led to little change in force production or twitch time in isometric contractions. However, TGU treatment did alter the relationship of oscillatory frequency to power output by smelt muscle. TGU-exposed muscle had a lower oscillatory frequency and produced relatively more power across a range of lower oscillatory frequencies compared with the control treatment or the other osmolyte treatments (Fig. 5B).

More than 30 years ago, Yancey and Somero (1979, 1980) suggested that methylamines, such as TMAO, counteract the destabilizing effects of urea on protein structure and enzyme function. Their focus was the relatively high urea content of elasmobranchs (0.4 mol l⁻¹) involved in the osmoregulation of shark, skates and rays, as well as some other fishes. In a test of the ‘counteraction’ hypothesis, Baskakov et al. (1998) examined rabbit muscle lactate dehydrogenase (LDH) activity in the presence and absence of TMAO and urea. The K_m for the LDH conversion of pyruvate into lactate was increased in the presence of urea and decreased in the presence of TMAO. The K_m was unchanged when both urea and TMAO were present in a 2:1 ratio of urea to TMAO, similar to findings in elasmobranchs (Baskakov et al., 1998). More recently, Kumemoto et al. (2012) reported that both urea and TMAO affected rabbit myosin activity when examined in solution in an in vitro actin–myosin motility assay. Both osmolytes reduced the sliding velocity of the actin–myosin interaction. However, the effect was reduced when both osmolytes were present at the specific concentrations of 0.6 mol l⁻¹ urea and 0.2 mol l⁻¹ TMAO (Kumemoto et al., 2012).

Yancey and Siebenaller (2015) reviewed the roles of TMAO, urea and glycerol, along with several other osmolytes, in cytoprotection. They reported that the counteracting effects of TMAO and urea are related to the differing effects of the two osmolytes on peptide backbone exposure to water. Urea promotes the unfolding of the secondary structure of proteins, while TMAO enhances folding of proteins and stabilizes secondary structure (Yancey, 2005). TMAO acts to stabilize protein structure by forming complexes with water that promote the exclusion of water from the protein backbone (Zou et al., 2002; Canchi and García, 2013). Similar to TMAO, glycerol is predicted to act as a stabilizer of protein structure (Yancey and Siebenaller, 2015). These authors note that while the negative effects of urea in destabilizing protein structure and reducing measures of enzyme function are understood, it is also important to recognize that stabilization is not by definition beneficial (Yancey and Siebenaller, 2015). TMAO has a negative impact on actin–myosin sliding velocity (Kumemoto et al., 2013) and LDH activity is reduced in the presence of glycerol and TMAO (Fields et al., 2001; Yancey and Siebenaller, 2015). TMAO and urea also have counteracting impacts on actin polymerization (Rosin et al., 2015), suggesting that these osmolytes may have very broad influences on the biology of smelt and other organisms that employ them for osmoregulation and cytoprotection.

This study is the first to examine how these osmolytes affect muscle function at the tissue level. By manipulating the ionic environment of muscle bundles that contained very low endogenous levels of these osmolytes (Fig. 1), we could test the hypothesis that methylamines need to be present in combination with increased levels of urea to maintain enzyme function and, in this case, muscle contractile properties. The results indicate that the mixture of TMAO, urea and glycerol (TGU) at rainbow smelt physiological concentrations leads to muscle force production and twitch times that are identical to control values. TMAO alone led to greater force and unchanged twitch times, while glycerol and urea alone each led to reduced force and relatively longer twitch times. However, TGU treatment did alter power output, with greater power production at lower oscillatory frequencies. Interestingly, TMAO treatment alone did not alter muscle shortening velocity as one might predict based on the work of Kumemoto et al. (2012). Compared with the rabbit muscle used in that study, rainbow smelt muscle appears to be adapted to TMAO by minimizing the ‘negative’ effects of TMAO on muscle shortening velocity. Additional evidence that rainbow smelt are adapted to elevated TMAO levels comes from the rainbow trout experiments. Rainbow trout, which do not evidently use TMAO as an osmolyte, do not show ‘positive’ effects of TMAO exposure such as increased force production (Fig. 9). In contrast, rainbow smelt do show ‘positive’ effects of TMAO on muscle force production and power output, suggesting that the response of smelt muscle proteins to the stabilizing effects of TMAO is not simply a property of all myosins, for instance, but reflects a difference between trout and smelt.

Glycerol and urea, both of which are cryoprotectant molecules (Raymond, 1992; Zimmerman et al., 2007; Costanzo et al., 2015), had significantly negative impacts on muscle force and power production. In smelt, the impact of urea and glycerol appears to be mitigated by the counteracting osmolyte TMAO. Although urea and glycerol each affect protein structure and muscle function in distinct ways, TMAO appears to balance the effects of both of them. While urea is thought to destabilize protein structure, glycerol may generate excessive stabilization. Additional studies of the biochemical impact of glycerol on protein structure and function are warranted. The balance between TMAO and other osmolytes has been observed in a variety of fishes and elasmobranchs (Yancey, 2005; Treberg et al., 2006). The results presented here suggest that species that express urea or glycerol without counteracting osmolytes will demonstrate impaired muscle function. For instance, Cope's gray tree frog accumulates glycerol in the weeks to months prior to a winter freeze (Zimmerman et al., 2007). This leads to the predication that mobility in these animals may be impaired by elevated glycerol. Wood frogs, Rana sylvatica, elevate urea as a cryoprotectant in the months leading to winter (Costanzo et al., 2013, 2015). Interestingly, wood frogs also significantly elevate an additional unidentified osmolyte in winter (Costanzo et al., 2015). From a muscle physiology perspective, this unidentified osmolyte would likely be a counteracting osmolyte that would balance the destabilizing effects of urea.

Interest in the impact of osmolytes such as TMAO and urea on muscle function has increased following recent evidence of their role in human biology. Elevated levels of TMAO have been linked to a greater risk of atherosclerosis (Wang et al., 2011). This has led to substantial research on biochemical pathways that produce TMAO within the human body and the impact of TMAO on human health (Loscalzo, 2011). Additional physiological functions that are impacted by elevated levels of osmolytes have recently been identified. For instance, elevated TMAO is linked to type-2 diabetes (Seeliger et al., 2013). Another osmolyte, urea, is elevated in humans under certain conditions; for instance, extreme endurance events such as ironman triathlons (Knechtle et al., 2010a,b). Recent research has looked at increasing glycerol content in both humans and animals for hyperhydration. Glycerol taken as a supplement has been proposed prior to and for recovery from exercise (van Rosendal et al., 2009). Glycerol treatment has been explored as a way to promote hydration and maintain health in cattle during transport (Parker et al., 2007). Will glycerol under these conditions affect muscle function? Questions remain about all of these osmolytes and their impact on physiological function.

Rainbow smelt are an excellent subject for the study of osmolytes and muscle function. These fish naturally express several osmolytes in winter but have very low levels of urea, TMAO and glycerol in summer (Raymond, 1998; Treberg et al., 2002). The results presented here suggest that their muscle displays adaptations that enhance muscle function in the presence of elevated osmolytes such as TMAO. Smelt also display an impressive thermal acclimation response in terms of swimming performance and muscle function (Woytanowski and Coughlin, 2013). In addition, these fish can be obtained from a hatchery from wild-run adults, a sustainable source of research animals. The present work represents the first attempt to examine how muscle function is influenced by the seasonal increase in osmolyte content. The TMAO, urea and glycerol found in rainbow smelt in winter appear to act as counteracting solutes, balancing stabilizing and destabilizing influences on muscle proteins, permitting these fish to remain active and feeding while their plasma and tissue osmolyte concentration is greater than 1000 mOsm, reportedly the highest of any vertebrate (Raymond, 1992). This study opens the door to examining the effects of osmolyte concentration across a range of temperatures that include the sub-zero temperatures experienced by smelt in mid-winter.

Thank you to Jay Kanaparthi for laboratory support and Sheree Harden and Dalia Lemus for animal care support. The manuscript was substantially improved based on the comments of two anonymous reviewers.