Effects of temperature on survival, moulting, and expression of neuropeptide and mTOR signalling genes in juvenile Dungeness crab (Metacarcinus magister)

Crustaceans, like other ecdysozoans, must shed their exoskeleton periodically to develop and grow (reviewed in Aiken and Waddy, 1992; Charmantier-Daures and Vernet, 2004; Skinner, 1985). Two endocrine glands regulate the moult cycle (reviewed in Chang and Mykles, 2011; Webster, 2015). The X-organ/sinus gland complex in the eyestalk ganglia secretes neuropeptides of the crustacean hyperglycaemic hormone (CHH) superfamily, including moult-inhibiting hormone (MIH) and CHH (reviewed in Webster, 2015; Webster et al., 2012). The neuropeptides act on the moulting gland (Y-organ), which synthesizes ecdysteroid moulting hormones. A drop in circulating MIH activates the Y-organ and the animal enters premoult (stages D₀ through D₄; Chang and Mykles, 2011; Skinner, 1985). Haemolymph ecdysteroid concentration increases approximately 100-fold from intermoult (C) to a peak in late premoult (D₂), then drops precipitously shortly before ecdysis and is lowest in postmoult (Thomton et al., 2006; reviewed in Mykles, 2011). In the target tissues, ecdysteroids are converted to their active forms and bind to nuclear receptors. This induces, for example, cuticle formation, partial resorption of the old exoskeleton, claw muscle atrophy and limb regenerate growth (Chang and Mykles, 2011; Mykles, 2011). Moulting is suspended or delayed by adverse environmental conditions, such as crowding or temperature stress, via direct or indirect inhibition of the Y-organ by neuropeptides (Aiken and Waddy, 1976; Anger and Spindler, 1987; Chung and Webster, 2005, Pitts et al., 2017; reviewed in Aiken and Waddy, 1992). Moult suspension occurs only in intermoult and early premoult animals. In mid and late premolt, the Y-organ becomes insensitive to neuropeptide control and the animal proceeds to ecdysis without delay (Anger, 1987; Chang and Mykles, 2011).

The highly conserved mTOR (mechanistic target of rapamycin) signalling pathway regulates cell growth and metabolism by integrating signals of energy, oxygen and nutrient supply, as well as of growth factors in organisms from yeast to mammals (reviewed in Reiling and Sabatini, 2006). It has been implicated in growth and moult regulation in insects (reviewed in Arquier et al., 2010; Nijhout et al., 2014) and crustaceans (Chang and Mykles, 2011; Das et al., 2016). mTOR activity is required for Y-organ activation, as rapamycin inhibits Y-organ ecdysteroid synthesis in vitro and lowers haemolymph ecdysteroid titre in Gecarcinus lateralis induced to moult by eyestalk ablation (Abuhagr et al., 2014b, 2016). Moreover, mTOR signalling genes, such as Gl-mTOR, Gl-Rheb, Gl-AKT and Gl-S6K, are upregulated in the Y-organ of G. lateralis induced to moult by multiple leg autotomy or eyestalk ablation (Abuhagr et al., 2014b, 2016; Das et al., 2018; Shyamal et al., 2018). Rheb (Ras homolog enriched in brain) is a GTP-binding protein that binds to mTOR and other proteins to form the active mTOR Complex 1 (mTORC1; reviewed in Saxton and Sabatini, 2017). Rheb activity is regulated upstream by protein kinase B (PKB, also known as AKT), which prevents inactivation of Rheb by the tuberous sclerosis complex (TSC1/2). A downstream target of mTORC1 is ribosomal protein S6 kinase (S6K), which is an important activator of protein synthesis and growth. mTORC1 can interact with transcription factors and coactivators involved in mitochondrial biogenesis and lipid metabolism and synthesis (e.g. peroxisome proliferator-activated receptors, PPARs; Sengupta et al., 2010). Transcriptomic analysis shows that mTOR activity, either directly or indirectly, affects the mRNA levels of thousands of contigs in the G. lateralis Y-organ (Shyamal et al., 2018). A negative regulator of mTORC1 is the cellular energy sensor AMP-dependent protein kinase (AMPK), which is activated at a low ATP/ADP ratio resulting from excessive exercise, hypoxia or acute thermal stress (Hardie et al., 2006; Reiling and Sabatini, 2006). During acute heat stress, reducing cellular energy consumption and growth by inactivation of mTOR may assure survival (Aramburu et al., 2014; Bandhakavi et al., 2008; Chou et al., 2012; Gibney et al., 2013; Hansen et al., 2007; Jarolim et al., 2013). In agreement with this, AMPK activation has been observed during acute temperature stress in marine ectotherms (Anttila et al., 2013; Frederich et al., 2009; Han et al., 2013). However, there is a paucity of studies addressing chronic, sublethal effects of temperature on mTOR activity. Transcriptional changes of raptor, an adaptor protein in mTORC1 (Clark et al., 2013), PPARs (Windisch et al., 2011) and regulators of the phosphatidylinositol 3-kinase (PI3K)-AKT signalling pathway upstream of mTOR (Windisch et al., 2014) indicate the importance of mTOR signalling for the reorganisation of metabolism with repercussions on growth during long-term thermal exposures.

The purpose of this study was to determine the effects of temperature on survival, moulting, and neuropeptide and mTOR signalling gene expression in the heart, Y-organ and eyestalk ganglia of the Dungeness crab, Metacarcinus magister. Juveniles from Northern California and Oregon experience temperatures ranging between 10 and 25°C (Brown and Terwilliger, 1999; Tasto, 1983). They moult successfully at 21°C, but growth is reduced compared with that at 14°C (Terwilliger and Dumler, 2001). We hypothesize that this pattern is reflected in temperature- and moult cycle-dependent expression of mTOR signalling components and neuropeptides in the tissues involved in moult regulation (eyestalk ganglia and Y-organ) and the heart of juvenile decapod crustaceans. The heart has been the subject of many ecophysiological studies with respect to thermal acclimation and adaptation owing to its central function in oxygen supply (Frederich et al., 2009; Stillman and Tagmount, 2009; Tepolt and Somero, 2014; Wittmann et al., 2012). We transferred naturally moulting juvenile M. magister from ambient conditions (∼15°C) to six temperatures (5, 10, 15, 20, 25 and 30°C) for up to 14 days and monitored survival and moult stage progression. At the three temperatures that allowed moult progression (10, 15 and 20°C), mRNA levels were quantified by real-time RT-PCR of mTOR signalling components in the eyestalk ganglia, Y-organ and heart, and of the neuropeptides Mm-CHH and Mm-MIH in the eyestalk ganglia. There was a decrease in mRNA levels of certain mTOR signalling genes (e.g. Mm-Rheb) with increasing temperature, which may compensate for higher metabolic rates at higher temperature, in order to maintain consistent allocation of cellular energy to protein synthesis. As shown in previous studies of other decapod species, Mm-Rheb mRNA level can serve as a marker for Y-organ ecdysteroidogenic activity.

Juvenile Metacarcinus magister (Dana 1852) were collected at the end of June 2013 from the lower intertidal of Bodega Harbor, CA, USA, using pitfall traps (Grosholz and Ruiz, 1995). Pitfall traps consisted of 20-litre buckets placed in the sediment to ground level. When crabs were collected after one tidal cycle (ca. 24 h), water temperatures in the traps ranged from 14 to 22°C. At the University of California, Davis, Bodega Marine Laboratory (BML), Bodega Bay, CA, USA, they were kept individually under a 12 h:12 h light:dark cycle in a flow-through system in trays with compartments of 7×7×11 cm with a mesh bottom and solid transparent plastic walls and lids at ambient water temperature. Ambient temperature was determined daily in one of the trays and ranged from 11 to 15°C until the experiment started in late August 2013, and usually was similar to BML ambient seawater temperature (data provided by BML; http://boon.ucdavis.edu/data_seawater_temperature.html). Salinity was 33–34 PSU (http://boon.ucdavis.edu/data_seawater_salinity.html). The crabs were cleaned and fed ad libitum with pieces of frozen squid three times per week. Moults were collected five days per week and carapace widths (CW) including the 10th anterolateral spines of the moults were determined to the nearest 0.01 mm using digital callipers. During maintenance at ambient temperature, most crabs moulted twice and grew in CW by approximately 30% with each moult. At the start of the experiment, CW was 42±4 mm and wet mass was 11±3 g for a total of 180 animals (means±s.d.), and moult interval ranged between 35 and 45 days owing to variation in size and ambient temperature.

The recirculating incubation systems at 5, 10, 15, 20, 25 and 30°C each consisted of a water reservoir of 200–500 litres, individual glass jars of 470 ml (91 mm in diameter and 94 mm in height), which each housed one animal, and water tables that collected overflowing water from the jars and drained back into the reservoir tanks. The jars received water from the reservoir using a submersible aquarium pump (Eheim, Deizisau, Germany), a manifold and regular airline tubing at a flow rate of 400 ml min⁻¹ at each outlet, and were covered with pieces of plastic light diffusion panels. All systems were equipped with air pumps and air stones, a temperature regulator (Pentair Aquatic Eco-Systems, Apopka, FL, USA) and up to three 300 W heaters (Eheim Jäger). Water was changed at least once per week to maintain water quality. The crabs were introduced and taken out for dissection in a staggered fashion, so that no more than 20 crabs were kept in each recirculating system at any time of the 14 day incubation. To sample moult stages C to D_2–3, the experiment was started (t₀) with crabs at 12, 19 or 26 days postmoult, and tissues were collected from crabs at 26, 33 or 40 days postmoult at the end of the experiment (t₁₄). Only four sampled crabs moulted during the incubation period. Groups of 10 crabs each at 12 or 26 days postmoult at t₀ were incubated at 5°C, and groups of 10 crabs each at 12, 19 or 26 days postmoult were incubated at 10, 15, 20 or 25°C. An additional group of 10 crabs at 26 days postmoult was incubated at 10°C, and an additional group of 10 crabs at 19 days postmoult was incubated at 25°C. Two groups of 10 crabs at 19 days postmoult were incubated at 30°C, but died within hours, so crabs at 12 or 26 days postmoult were not tested. To minimize thermal shock during transfer, the 5, 25 and 30°C groups were first incubated for 2 h at each of the intermediate temperature steps, i.e. at 10 or 20°C and 25°C. We controlled for the sex ratio and size distribution in each of the temperature and days postmoult groups. We checked for moults and survival five days a week, and cleaned and fed ad libitum with pieces of squid mantle three times per week.

At t₀ and t₁₄, wet mass (g) and CW (mm) of the crabs were determined. Haemolymph samples were collected using needles and syringes (5 µl haemolymph immediately added to 50 µl dd H₂O) and stored at −20°C at both time points. Haemolymph ecdysteroid concentration was quantified with an enzyme-linked immunosorbent assay (ELISA) modified from Kingan (1989) and Tamone et al. (2007) as described in Abuhagr et al. (2014a). At t₀ and t₁₄, the basal endite of one of the third maxillipeds was excised and stored in crab saline and photographed under a microscope within 72 h to determine moult stage. Moult stage was determined after Drach and Tchernigoftzeff (1967) using illustrations of endites from Moriyasu and Mallet (1986), as well as haemolymph ecdysteroid concentrations [10–40 pg µl⁻¹ in intermoult (C), 50–160 pg µl⁻¹ in early premoult (D₀), 200–600 pg µl⁻¹ in mid premoult (D₁), ≥1000 pg µl⁻¹ in D₂, and a drop in ecdysteroids from D₃ onwards]. At t₁₄, 5–18 individuals per treatment group were dissected (total N=93, see Fig. S1), and the pairs of eyestalk ganglia and Y-organs, as well as 3–4 mm cubes of heart tissue, were stored in RNAlater (Life Technologies, Carlsbad, CA, USA) at 4°C overnight and then transferred to −20°C.

Pairs of eyestalk ganglia or Y-organs were hand-homogenized in 100 µl TRIzol reagent (Life Technologies, Carlsbad, CA, USA) using Kontes Pellet Pestle Grinders and 1.5 ml tubes (Kimble Chase, Vineland, NJ, USA). Heart tissues were homogenized with two 5 mm stainless steel balls (Abbott Ball Co., West Hartford, CT, USA) in precooled (−80°C) 2 ml tubes in a TissueLyzer II (Qiagen, Frederick, MD, USA) at 2×1 min at 30 Hz, then suspended in 1 ml of TRIzol reagent. After centrifugation for 15 min at 12,000 g at 4°C, total RNA was isolated from the supernatant. For eyestalk ganglia, the Direct-zol RNA MiniPrep kit with microcolumns (Zymo-Spin IC Column, Zymo Research, Irvine, CA, USA) was used according to the manufacturer's protocol with an in-column DNase I treatment including RNase inhibitor (RiboLock, Thermo Fisher Scientific, Waltham, MA, USA) for 30 min at 37°C. Total RNA of Y-organs and heart was extracted using the phenol-chloroform method as described in Covi et al. (2010). RNA purity and concentrations were determined by measuring absorbance at 260 and 280 mn using a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific). Then, 400 ng of eyestalk ganglia RNA or 1 µg of Y-organ or heart RNA were reverse transcribed in 20 µl reactions using qScript cDNA SuperMix (Quanta Biosciences, Gaithersburg, MD, USA).

Partial cDNAs encoding M. magister Mm-AKT (GenBank accession no. KT285226), Mm-CHH (KY070318), Mm-Rheb (KT315722), Mm-mTOR (KT315723), Mm-S6K (KT367806) and Mm-AMPKα (KT315720) were cloned using nested specific or degenerate primers and RNA ligase mediated rapid amplification of cDNA ends (RLM-RACE). Primers (Table S1; Integrated DNA Technologies, Coralville, IA, USA) were designed to bind to highly conserved regions, as identified by multiple sequence alignment including a suite of pancrustacean and vertebrate species. Sequences were aligned using Clustal X (http://www.clustal.org/clustal2/) and GeneDoc software (http://iubio.bio.indiana.edu/soft/molbio/ibmpc/genedoc-readme.html). Template cDNA from eyestalk ganglia, Y-organ, heart, claw or thoracic muscle was used for initial PCR reactions, which consisted of 1 µl cDNA, 1 µl forward and 1 µl reverse primer (10 µmol l⁻¹), 5 µl 2× PCR mastermix (Thermo Fisher Scientific) and 2 µl nuclease-free water. 3′ RACE was performed with the FirstChoice RLM-RACE kit (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocol, except PCR reactions consisted of 1 µl cDNA, 2 µl gene-specific forward primer and 2 µl reverse primer (10 µmol l⁻¹, from RLM-RACE kit), 25 µl 2× PCR mastermix (Thermo Fisher Scientific) and 20 µl nuclease-free water. The PCR was carried out using a Veriti 96 Well Thermal Cycler (Applied Biosystems, Foster City, CA, USA) with an initial denaturation for 4 min at 96°C, followed by 35 cycles of denaturation for 30 s at 94°C, annealing at the respective melting temperature of the primers for 30 s, and elongation for 30–120 s at 72°C, depending on the expected product length (1000 bp min⁻¹). Final elongation for 7 min occurred at 72°C, followed by a hold at 4°C.

Products were separated on a 1.0% agarose gel in TAE buffer for 90 min at 100 V, stained with ethidium bromide and extracted using the Qiaex II Gel Extraction kit (Qiagen, Valencia, CA, USA). Sequencing was carried out by Davis Sequencing (Davis, CA, USA) using the respective primers that yielded the products.

Roche LightCycler 480 and LightCycler Fast Start DNA Master PLUS SYBR Green I reaction mix (Roche Applied Science, Penzberg, Germany) were used to quantify absolute mRNA levels of Mm-AKT, Mm-Rheb, Mm-mTOR, Mm-S6K, Mm-AMPKα, Mm-CHH, Mm-MIH (GenBank accession no. AF031493; Umphrey et al., 1998) and Mm-RbS3 (ribosomal protein S3; JF276909.1; Martin et al., 2011; previously used as reference gene). Reactions were combined on 384-well plates and consisted of 1 µl cDNA template, standard, no reverse transcriptase control or negative control (nuclease-free water), 0.5 µl (10 µmol l⁻¹) gene-specific forward and reverse primers (Table 1), 5 µl 2× SYBR Green master mix and 3 µl nuclease-free water. PCR conditions were 3 min at 95°C for initial denaturation, then 45 cycles of 30 s at 95°C, 30 s at the primer annealing temperatures (Table 1) and 20 s at 72°C for elongation, followed by 7 min at 72°C for final extension. Melting curves and crossing point (C_p) values were determined using Roche LightCycler 480 Software version 1.5.0 SP4. Absolute mRNA levels (copies µg⁻¹ RNA) were calculated from the respective standard curves ranging from 10⁻¹⁰ to 10⁻¹⁸ g µl⁻¹ of purified cDNA product of each gene (see Table 1 for primers; see ‘PCR cloning’ above for PCR conditions and product purification). Data are presented as log₂(fold expression) relative to the mean absolute mRNA level at stage C at 15°C (Table 2).

Statistical analyses were carried out using GraphPad Prism Version 7.0b (GraphPad Software, La Jolla, CA, USA). The Mantel–Cox test was used to test effects of temperature on survival. Moult progression is presented only for animals in D₀ at t₀ as t₁₄/t₀ ratios of ecdysteroid concentration and as progression from D₀ to D₁ or D_2–3. Ecdysteroid concentrations and their ratios were log₁₀ transformed to fulfil the prerequisite of equal variances. Because moult stage of the individuals at a given day postmoult was variable at both t₀ and t₁₄ at all temperatures (Fig. S1), qPCR data were grouped by moult stage rather than by days postmoult. qPCR data from 5°C-incubated animals were excluded from the analysis, as no data or not enough data were available from animals in mid and late premoult (D₁ and D_2–3). Factorial changes in absolute mRNA levels relative to the mean absolute mRNA levels at stage C at 15°C were log₂ transformed. Outliers were detected using the ROUT test, and removed. Subsequently, one-way or two-way ANOVA and the post hoc Tukey multiple comparisons test were used to test for significant effects of temperature and moult stage at the P<0.05 level. Pearson correlation coefficients (r) and goodness of fit of linear regressions (R²) were determined for relationships between log₁₀-transformed ecdysteroid concentrations and Mm-Rheb mRNA levels. An ANCOVA was used to compare the slopes and y-intercepts of the regression lines.

Temperature affected survival (P<0.0001). At 30°C, all animals died within 24 h (Fig. 1). At 25°C, animals started dying after 3 days, with only 48% of the animals surviving after 7 days and 0% surviving after 14 days. Survival was 100% at 5 and 15°C, and 98% and 97% at 10 and 20°C, respectively, over the 14-day incubation period.

Moult progression increased with temperature. Among the group of animals in D₀ at t₀, individuals tested at 5°C did not progress in the moult cycle within the observation period, whereas the crabs incubated at 10, 15 or 20°C did so (Fig. 2). Most animals at 10, 15 or 20°C exhibited t₁₄/t₀ ecdysteroid ratios greater than 1, resulting in significantly higher means than the animals at 5°C (F=0.54, P<0.0001; Fig. 2A). Consequently, all the D₀ animals incubated at 5°C remained in this stage (Fig. 2B), while at 10°C, 43% remained in D₀ and 57% progressed to D₁. At 15°C, almost half of the animals (46%) advanced to D_2–3, while 23% remained in D₀ and 31% progressed to D₁. At 20°C, 27% of the animals remained in D₀, while 36% progressed to D₁ and 36% progressed to D_2–3 by 14 days.

At t₁₄, haemolymph ecdysteroid levels of all the individuals incubated at 10, 15 or 20°C varied by moult stage (P<0.0001), but not by temperature (Fig. 3). Ecdysteroid concentrations increased significantly from moult stage to moult stage in each temperature group and were approximately 100-fold higher in stage D₂ compared with stage C. In intermoult, ecdysteroid titres varied from 19±3 pg µl⁻¹ at 20°C to 28±5 pg µl⁻¹ at 10°C. In stage D₀, values increased to 147±32 pg µl⁻¹ at 10°C, 169±34 pg µl⁻¹ at 15°C and 146±25 pg µl⁻¹ at 20°C. In D₁, titres ranged from 383±68 pg µl⁻¹ at 15°C to 420±65 pg µl⁻¹ at 20°C. Maximal concentrations at stage D₂ ranged from 1506±318 pg µl⁻¹ at 20°C to 2123±403 pg µl⁻¹ at 15°C.

mRNA levels in eyestalk ganglia were significantly affected by temperature, with a general trend of being higher at 10°C and/or lower at 20°C (Fig. 4, see Table 2 for results of two-way ANOVA). Mm-MIH, Mm-AKT and Mm-S6K were also significantly affected by moult stage. Only for Mm-MIH was a significant interaction observed between temperature and moult stage. Mm-RbS3 and Mm-mTOR were not included in the analysis because a reliable quantification was not possible owing to low mRNA levels and residual genomic DNA.

Mm-Rheb was upregulated up to 2.2-fold between 15 and 10°C in all moult stages (P ranged from <0.0001 in C to 0.0033 in D_2–3, Tukey post hoc test; Fig. 4A), downregulated in stage C by 0.6-fold between 20 and 15°C (P=0.0036) and downregulated by 0.3-fold between 10 and 20°C (P<0.0001). In the other moult stages, the differences between 15 and 20°C were not significant, but the differences between 10 and 20°C were all highly significant (0.3- to 0.4-fold, P<0.0001). There were no significant differences in Mm-S6K levels in post hoc comparisons of biological interest (Fig. 4B). Mm-AMPKα levels differed in stage C and D₁ between 10 and 20°C (both by 0.6-fold, P<0.0001, Tukey test; Fig. 4C) and in stage D₁ also between 15 and 20°C (0.6-fold, P=0.0069). Mm-AKT differed between 15 and 20°C in stages C (P=0.0042; Fig. 4D) and D₁ (P=0.0002), and between 10 and 20°C in stages C to D₁ (P<0.0001, P=0.0005, P<0.0001, 0.6-fold).

Mm-MIH levels decreased in stage C animals 0.5-fold between 10 and 20°C (P=0.0007, Tukey test; Fig. 4E). In stage D₁, Mm-MIH levels at 20°C were lower than at 10 and 15°C (P<0.0001 and P=0.0019, 0.3-fold). At 20°C, Mm-MIH was upregulated by a factor of 2.5 from stages D₁ to D_2–3 (P=0.0008). Although the trends in Mm-CHH levels were similar to those in Mm-MIH levels, the differences were smaller as no upregulation occurred upon exposure to 10°C (Fig. 4F). In stage D₁, Mm-CHH levels at 20°C were lower than at 10°C (P=0.0067, Tukey test) and at 15°C (P=0.0057), both by 0.5-fold.

In the Y-organ, Mm-Rheb, Mm-AKT, Mm-AMPKα, Mm-mTOR and Mm-S6K were significantly affected by moult stage, and Mm-Rheb, Mm-AKT, Mm-mTOR, Mm-S6K and Mm-RbS3 were significantly affected by temperature (ANOVA; Fig. 5, Table 2). There was no significant interaction between moult stage and temperature in any of the genes. When affected by temperature, mRNA levels decreased with increasing temperature. For Mm-Rheb (Fig. 5A), Mm-S6K (Fig. 5B) and Mm-AMPKα (Fig. 5C), there was a trend for an increase with progressing moult stages. For Mm-AKT (Fig. 5D) and Mm-mTOR (Fig. 5E), there was a trend for higher mRNA levels in stages D₀ and D₁ compared with stages C and D_2–3.

Mm-Rheb mRNA levels were lower at 20°C compared with 10°C in stages C and D₀ by 0.4-fold (P<0.0001 and P=0.0050, Tukey post hoc test; Fig. 5A), and in stage C at 15°C compared with 20°C (P=0.0111, 0.5-fold). At 10°C, mRNA levels were elevated by a factor of 1.8 in stage D₁ compared with stage C (P=0.0389). At 20°C, Mm-Rheb levels increased from stage C to D₁ and D_2–3 (2.4- and 3.0-fold, P=0.0028 and P<0.0001). Mm-AKT levels were lower at 20°C compared with 10°C in stage C (0.6-fold, P=0.0435, Tukey test; Fig. 5D). There were no significant differences in Mm-S6K (Fig. 5B), Mm-AMPKα (Fig. 5C), Mm-mTOR (Fig. 5E) or Mm-RbS3 (Fig. 5F) mRNA levels in post hoc comparisons of biological interest.

In the heart, the mRNA levels of all genes except Mm-RbS3 were significantly affected by temperature, and the mRNA levels of Mm-Rheb, Mm-AKT and Mm-S6K also varied with moult stage (ANOVA; Fig. 6, Table 2). There was a significant interaction between moult stage and temperature in Mm-mTOR mRNA levels.

In all moult stages, Mm-Rheb mRNA levels at 15 and 20°C were lower than at 10°C (P<0.0001 to P=0.0285, Tukey test; Fig. 6A). Factorial change ranged among the moult stages from 0.3 to 0.4 between 10 and 15°C and from 0.2 to 0.3 between 10 and 20°C. The latter were the largest fold changes found by this study; in other words, a 5.3-fold increase at a temperature decrease from 20 to 10°C in stage D₁. Although not identified by the post hoc test, it is noteworthy that there was a trend for reduced Mm-Rheb mRNA levels with progression of the moult cycle by up to 0.5 at 20°C. The mRNA levels of the kinases Mm-S6K (Fig. 6B), Mm-AMPKα (Fig. 6C), Mm-AKT (Fig. 6D) and Mm-mTOR (Fig. 6E) varied little and exhibited similar patterns. There were differences only in Mm-S6K between 10 and 15°C in stage C (P=0.0360, 0.6-fold, Tukey test; Fig. 6B) and in Mm-mTOR between 10 and 20°C in stage C (P=0.0018, 0.6-fold; Fig. 6E).

In eyestalk ganglia, the correlation between log₁₀ ecdysteroids and Mm-Rheb copies µg⁻¹ RNA varied with temperature (Fig. 7A, see Table S2 for statistics). There was a negative correlation at 10°C (P=0.0420), no correlation at 15°C and a positive correlation at 20°C (P=0.0159). Consequently, there were differences in the slopes of the regression lines (F=4.758, P=0.0118). Given that the slopes differed greatly, it was not possible to determine whether the elevations differed significantly. Goodness of linear fits (R²) varied between 0.00646 (15°C) and 0.2367 (20°C).

In the Y-organ, a positive correlation of the two variables was highly significant at all temperatures (P between 0.0081 and <0.0001; Fig. 7B, Table S2). The slopes of the lines did not differ significantly (F=0.1279, P=0.8802), but the difference between the elevations was highly significant (F=23.78, P<0.0001), which indicates an inverse relationship between Mm-Rheb mRNA levels and temperature. R² ranged from 0.2322 (10°C) to 0.7008 (20°C).

In the heart, log₁₀ ecdysteroids and Mm-Rheb levels were negatively correlated (10°C: P=0.0498; 20°C: P=0.0004) or not correlated (15°C; Fig. 7C, Table S2). There was no significant difference between the slopes of the fitted lines (F=1.182, P=0.3136), but there was a significant difference between the elevations of the lines (F=51.66, P<0.0001), indicating higher Mm-Rheb mRNA levels at 10°C. R² varied between 0.0855 (15°C) and 0.491 (20°C).

Summer water temperatures in the intertidal of embayments and estuaries of Northern California and Oregon vary between 10 and 25°C (Brown and Terwilliger, 1999; Tasto, 1983), and we observed a maximum of 22°C when collecting juveniles from Bodega Harbor during summer 2013. Larval and juvenile M. magister from different locations along the North Pacific coast show high survival and decreasing moult interval from 5 to 18°C, and stagnating moulting frequency and growth at higher temperatures (Brugman, 1972; Kondzela and Shirley, 1993; Sulkin and McKeen, 1989; Sulkin et al., 1996; Terwilliger and Dumler, 2001). This is consistent with our results on temperature-dependent survival and moult progression (Figs 1 and 2). Oxygen consumption rate increases up to 3-fold in M. magister between 10 and 20°C, resulting in extraordinarily high Q₁₀ values (Brown and Terwilliger, 1999; Gutermuth and Armstrong, 1989; McLean and Todgham, 2015; Prentice and Schneider, 1979). At 20°C, metabolic maintenance costs may become so high that energy partitioned for growth is reduced (Pörtner and Knust, 2007; Pörtner, 2010; Sokolova et al., 2012; Terwilliger and Dumler, 2001). Thus, higher metabolic rate does not translate into faster moult progression, and the crabs seem to have reached or surpassed a maximum rate for completing premoult processes. Juvenile survival is only compromised at substantially longer exposures to this temperature than in the present study and at higher temperatures (Brugman, 1972; Kondzela and Shirley, 1993; Sulkin and McKeen, 1989; Sulkin et al., 1996; Terwilliger and Dumler, 2001).

In the temperature range 10 to 20°C, haemolymph ecdysteroid levels throughout the moult cycle of the juveniles were similar to those of adult M. magister (Fig. 3; Thomton et al., 2006). In the lobster Homarus americanus, temperature-dependent regulation of ecdysteroid titers is necessary to prevent premature (i.e. lethal) moults (Aiken and Waddy, 1975; Chang and Bruce, 1980). In the copepod Calanus pacificus, ecdysteroid titers decrease with lower temperatures (Johnson, 2003). It is possible that low temporal resolution precluded detection of temperature-dependent differences in ecdysteroid titre in our study. Alternatively, similar ecdysteroid concentrations at 10, 15 and 20°C in juvenile M. magister may result from the active regulation of both biosynthesis and degradation/elimination of moulting hormone (Mykles, 2011).

In most decapods, the neuropeptide MIH appears to be regulated post-transcriptionally over the normal moult cycle, as MIH mRNA levels in the eyestalk ganglia can remain elevated during premoult (Chung and Webster, 2003, 2005; Covi et al., 2012; Pitts and Mykles, 2017; Techa et al., 2015; Techa and Chung, 2015). Genetic manipulation of Rheb and other mTOR components in insects highlights the importance of this pathway for protein synthesis (Danielsen et al., 2016; Hall et al., 2007; Parthasarathy and Palli, 2011). In Drosophila melanogaster, inhibition of mTORC1 in neurosecretory brain cells that produce prothoracicotropic hormone, which stimulates ecdysteroid production in the prothoracic gland, did not alter the timing of pupariation (Layalle et al., 2008). Thus, mTORC1 does not seem to be necessary for the control of neurohormone expression, synthesis, and/or secretion in insects under normal conditions. However, environmental stressors or conditions may inhibit moulting by upregulating MIH expression, thus maintaining high circulating MIH levels that prevent Y-organ activation. For example, some G. lateralis individuals fail to moult in response to multiple leg autotomy, indicating the animals are in a blocked or chronically inhibited condition (Pitts et al., 2017). The eyestalk ganglia of blocked animals exhibit 200- to 400-fold higher mRNA levels of Gl-MIH, Gl-CHH and three of four mTOR components, including Gl-Rheb, than the eyestalk ganglia from control animals (Pitts et al., 2017). These data suggest that mTOR signalling enhances neuropeptide production in eyestalk ganglia of animals with an activated Y-organ to prevent or delay moulting under adverse conditions.

Our study provides important insights into the role of mTOR signalling in the eyestalk ganglia of juvenile M. magister. In the thermal range that allows moult progression, transcriptional regulation of Mm-Rheb may contribute to the control of neuropeptide synthesis in response to temperature rather than over the moult cycle. The significant interaction term in Mm-MIH mRNA levels indicates that moulting is regulated differently at different temperatures. At temperatures that juvenile M. magister tolerate over prolonged periods (10 and 15°C; Kondzela and Shirley, 1993), Mm-MIH was constitutively expressed over the moult cycle, as expected. A decrease of MIH expression in premoult was reported for Callinectes sapidus and in Scylla paramamosain (Huang et al., 2015; Lee et al., 1998) and observed in our experimental animals at 20°C. Transcription was downregulated at 20°C also in the other genes, but to a lesser extent. Unlike at lower temperatures, at 20°C, M. magister juveniles control MIH transcriptionally, yet with a time lag. Mm-MIH mRNA levels were reduced by about half from stages C to D₁ (Fig. 4E), which is when the committed Y-organ becomes least sensitive to MIH through the end of premoult (Chang and Mykles, 2011). In stage D_2–3, Mm-MIH recovered by a factor of 2.5, which would allow the eyestalk ganglia to resume a greater rate of MIH synthesis in preparation for MIH release during postmoult and intermoult. Consistent with this, Techa and Chung (2015) found a time lag between transcription and peptide storage in C. sapidus eyestalk ganglia in premoult. Owing to the variable correlation coefficients between ecdysteroid concentration and Mm-Rheb expression in eyestalk ganglia, a (simple) feedback mechanism of ecdysteroid on neuropeptide production (Techa and Chung, 2015) via mTOR seems unlikely. However, transcriptional compensation of mTOR components and neuropeptides may counteract temperature-dependent effects on neuropeptide synthesis, as hypothesized for moderate thermal stress.

mTOR-dependent protein synthesis is required for increased ecdysteroid synthesis and secretion in the activated Y-organ. Transcriptional regulation of this pathway takes place not only during moult induction in G. lateralis (Abuhagr et al., 2014b, 2016; Das et al., 2018; Shyamal et al., 2018), but also during naturally occurring moults in M. magister. However, the expression patterns differ between species (no change to up to 10-fold change), and differing degrees of post-transcriptional regulation are likely (Abuhagr et al., 2014b; 2016). In M. magister, Mm-Rheb increased up to 3-fold from intermoult to late premoult, while gene expression changes of the protein kinases mTOR, AMPK, AKT and S6K were smaller, yet significant (Fig. 5). The positive correlations between Mm-Rheb mRNA levels and ecdysteroid titres at all three temperatures support the role of this GTPase in the regulation of the moult cycle through the mTOR pathway (Fig. 7B). As mTORC1 is a direct target of Rheb (Long et al., 2005), Mm-Rheb may serve as a proxy for mTOR activity as a function of temperature and moult stage in the Y-organ of M. magister. Adjusting Mm-Rheb in the Y-organ to higher levels in the cold and lower levels in the warmth may play an important role in coping with moderate thermal stress.

AMPK was the only protein kinase in the Y-organ not significantly affected by temperature. Keeping a certain level of Mm-AMPKα, especially in mid and late premoult, would maintain regulatory capacity and, at least to some extent, may help in the downregulation of metabolic cost by phospho-AMPK at 20°C. Partial inhibition of protein synthesis in the warmth may contribute to similar ecdysteroid levels across temperatures. In addition, taking into account the expression of the activating mTOR components, a 1.9-fold increase in Mm-AMPKα from intermoult to late premoult at 20°C may counterbalance the 3.0- and 1.6-fold increases of Mm-Rheb and Mm-S6K, respectively. At 10°C, by contrast, a relatively small increase in Mm-AMPKα (1.6-fold) opposes larger changes of Mm-Rheb (1.9-fold), Mm-S6K (2.1-fold) and Mm-mTOR (1.9-fold) throughout the moult cycle. This may promote protein synthesis and thereby ecdysteroid production at thermally reduced kinetic energy. Posttranscriptional regulation of the kinases, e.g. through phosphorylation by upstream kinases, may play an important role in the regulation of the mTOR pathway in the Y-organ, and needs to be studied in the future.

Tissue- and fibre-specific regulation of protein turnover through the mTOR pathway contributes to regulating muscle growth (Bodine et al., 2001; Goodman et al., 2011; Saxton and Sabatini, 2017), cardiac homeostasis and compensatory cardiomyocyte growth in mammals (reviewed in Sciarretta et al., 2014). In crustaceans, muscle protein synthesis and ribosomal activity are higher in premoult than in postmoult and intermoult, and are regulated by ecdysteroid and other factors (El Haj et al., 1996; reviewed in Mykles and Medler, 2015). mTOR signalling components, especially Rheb, are upregulated in atrophic claw muscle of G. lateralis during premoult after moult induction (MacLea et al., 2012) and in naturally moulting Carcinus maenas (Cosenza, 2016). Thus, mTOR activation may relate to increased protein turnover during claw muscle atrophy, which facilitates the passage of the tissue through the narrow joint (Covi et al., 2010; MacLea et al., 2012; reviewed in Mykles and Medler, 2015). There is a significant positive correlation between haemolymph ecdysteroid concentrations and Rheb expression in claw muscle, but no correlation in thoracic muscle (Cosenza, 2016; MacLea et al., 2012). We found no correlation (at 15°C) or negative correlations (at 10 and 20°C) between Mm-Rheb in the heart and ecdysteroid titres of M. magister juveniles (Fig. 7C). Consistent with this, actin and myosin protein content decrease from intermoult to premoult at constant soluble protein and water content in abdominal muscle of juvenile shrimp (de Oliveira Cesar et al., 2006). mRNA levels of muscle proteins exhibit a similar pattern over the moult cycle in Portunus pelagicus (Kuballa et al., 2011). Also, heart sarcoplasmic reticulum Ca²⁺-ATPase (SERCA) mRNA expression declines in late premoult crayfish (Chen et al., 2002). This points to a lower degree of heart muscle remodelling than in skeletal muscle in premoult, and an inverse relationship, if any, between ecdysteroid and heart muscle growth (Qian et al., 2014). In intermoult, low levels of ecdysteroid may contribute to cardiac muscle growth via mTOR.

Phenotypic changes of the heart are mirrored in changes in mRNA expression of genes involved in muscle growth both during warm and cold acclimation in fish (e.g. Keen et al., 2016). Furthermore, cardiac transcriptomes vary after heat shock according to seasonal and latitudinal acclimatization in porcelain crabs (Stillman and Tagmount, 2009). In M. magister juveniles, cardiac expression patterns of mTOR components point to a general cold compensation of the mTOR signalling pathway, as expression of activating kinases (Mm-AKT, Mm-mTOR and Mm-S6K) and the inhibitory Mm-AMPK differ to the same extent (Fig. 6). The temperature effect was most pronounced in Mm-Rheb with the largest expression differences found in this study (Fig. 6A). That Mm-Rheb was not downregulated during warm exposure is clearly evident from overlapping 15 and 20°C regression lines when correlation to ecdysteroid concentration was tested (Fig. 7C). This indicates incomplete warm compensation of protein turnover as both protein synthesis and degradation may increase with temperature (Storch et al., 2003, 2005; Todgham et al., 2017; Whiteley et al., 1997). In agreement with this, cold acclimation of cardiac performance was observed in adult M. magister (Prentice and Schneider, 1979), whereas there was no evidence for seasonal (warm) acclimatization in juveniles (McLean and Todgham, 2015). Increased AMPK expression or activation may be a transient phenomenon during acute heat stress, helping to rapidly reduce energy consumption (Frederich et al., 2009; Jost et al., 2012; Han et al., 2013). Our data do not point to mTOR inhibition, and thus reduced growth, of the heart at 20°C. A sustained level of mTOR activation may aid to maintain cardiac muscle capacity (Sciarretta et al., 2014) at chronically increased heart rates expected owing to the lack of seasonal acclimatization (McLean and Todgham, 2015) at the cost of reduced body growth (Kondzela and Shirley, 1993; Terwilliger and Dumler, 2001). Body growth can only take place in the thermal range in which energy supply from aerobic metabolism exceeds maintenance costs (Pörtner and Knust, 2007; Pörtner, 2010; Sokolova et al., 2012). The thermal range that allows moulting and growth is relevant for the biogeography of a species. Knowledge of the underlying mechanisms of thermal tolerance can help to explain and predict past and future changes in distribution and biodiversity owing to climate change with impacts on fisheries (García Molinos et al., 2016; Green et al., 2014; Holsman et al., 2003; Moloney et al., 1994; Quinn, 2017; Toft et al., 2014).

Consistent with previous findings, Y-organ Rheb mRNA levels were positively correlated with ecdysteroid titres, which indicates that Rheb is central to regulating ecdysteroidogenesis through the mTOR pathway (Chang and Mykles, 2011; Das et al., 2018; Shyamal et al., 2018). Expression patterns of the GTPase Mm-Rheb may relate to temperature-dependent mTOR activation and protein synthesis during the moult cycle of juvenile M. magister. In all three organs, Mm-Rheb expression correlated most strongly with ecdysteroid titres at 20°C, which may indicate the tightest regulation of Mm-Rheb (and, vice versa, ecdysteroid production in Y-organs) compared with lower temperatures. This may point towards a need to control metabolic demand more stringently at higher temperatures.

 Fig. 8 summarizes the effects of temperature on mRNA levels in the eyestalk ganglia, Y-organ and heart of juvenile M. magister. In intermoult, MIH inhibits mTOR signalling in the Y-organ, resulting in low ecdysteroid synthesis and secretion (Chang and Mykles, 2011). Low levels of ecdysteroid may foster growth of the heart. Temperature-dependent expression of mTOR signalling genes in the eyestalk ganglia and Y-organ and Mm-MIH in the eyestalk ganglia contributes to temperature compensation of moult control between 10 and 20°C. However, thermal compensation in the heart may be incomplete. The limited capacity for acclimation or acclimatization to 20°C in juvenile M. magister was apparent, as indicated in the gene expression patterns in the heart after 14 days. Lack of downregulation of cardiac Mm-Rheb and other mTOR components at 20°C may relate to higher protein turnover and rising maintenance costs, which may result in similar moult stage progressions at 15 and 20°C. Insufficient warm compensation of heart function may contribute to the negative effects of high temperature on long-term growth and survival. This is consistent with previous studies of juvenile M. magister: successful moulting occurs up to about 20°C and reduced body growth occurs at temperatures above about 15°C (Kondzela and Shirley, 1993; Terwilliger and Dumler, 2001). Future studies should quantify phosphorylation of the protein kinases (e.g. mTOR, AMPK, AKT and S6K) to determine temperature-dependent activation or inhibition of mTOR signalling throughout the moult cycle. Furthermore, studying the mechanisms involved in the crosstalk between tissues will enable us to gain new insights into how moulting and growth are regulated in response to environmental stressors.

We thank A. Abuhagr, K. Cosenza, N. Pitts and M. (Mudron) Hines for technical advice on molecular techniques; J. He and H. Jesberger for cloning Mm-mTOR and Mm-CHH, respectively; K. Holle for analysing heart tissue; S. Chang for instruction and assistance on ecdysteroid ELISA; S. Nimitkul for advice on dissections; E. Grosholz for the use of pit traps; Bodega Marine Laboratory staff J. Newman and K. Menard for assistance with experimental setup; E. Ernst for assistance with animal care and experimental setup; and D. Storch for helpful comments on the manuscript.