MicroRNAs regulate survival in oxygen-deprived environments

In the winter, the freshwaters of eastern North America become ice-locked, cutting off access to fresh nutrients, and thus rendering waters anoxic (Bradford, 1983), while in the summer, warm and stagnant waters present a hypoxia-stressed environment. As such, aquatic organisms living in these conditions are faced with the year-round challenge of oxygen-poor waters. Among the organisms that have adapted to these conditions is the northern crayfish (Orconectes virilis). Prevalent throughout much of North America, as far north as the coast of Hudson Bay and extending south into the northern USA, this organism displays profound hypoxia and anoxia tolerance. Indeed, it was observed that while most fish species migrate downstream to escape ice-locked waters, O. virilis tolerate these oxygen-poor environments, conditions that were lethal to hibernating leopard frogs at the same location (Schueler and Karstad, 1996) Although the crayfishes' physiological response to anoxia has not been thoroughly studied, comparable species exhibit a drop in overall metabolic rate by approximately 90% when exposed to a similar oxygen-deprived environment (Storey and Storey, 1990). Anoxia tolerance is a complex balance of suppressing non-essential functions while simultaneously upregulating basic survival and protective mechanisms to conserve limited energy supplies until favourable conditions are restored (Seibel et al., 2018; Hochachka and Lutz, 2001).

Anoxia and hypoxia tolerance are two stress-induced survival responses that are characterized by a reduction in net ATP turnover, stabilization of macromolecules for extended preservation and the reprioritization of cellular resources (Storey, 2015). In oxygen-poor environments, cell functions such as vascularization and proliferation are controlled by transcription factors such as hypoxia-inducible factor (HIF)-1 (Ziello et al., 2007). A variety of invertebrates have adapted to extreme hypoxic and anoxic conditions and employ these mechanisms to promote survival (Dawson and Storey, 2011). Under such conditions, these processes have been found to be differentially regulated in a tissue-specific manner in response to the unique metabolic requirements of each organ (Ziello et al., 2007). For instance, anoxia-induced responses have been shown to be tissue specific in various other models of hypoxia and anoxia tolerance, such as turtles, frogs and squid (Mattice et al., 2018; Al-attar et al., 2017; Gerber et al., 2016; Hadj-Moussa et al., 2018; Krivoruchko and Storey, 2015).

The molecular mechanisms that underlie anoxia tolerance adaptations include: post-translational modifications, epigenetic regulation, transcription factor regulation, the accumulation of organic osmolytes, and the post-transcriptional regulation of small regulatory non-coding RNAs called microRNAs (miRNAs) (Szereszewski and Storey, 2018; Wijenayake et al., 2018; Biggar and Storey, 2017; Green and Storey, 2016; Kornfeld et al., 2012; Krivoruchko and Storey, 2010; Muir et al., 2007). miRNAs are highly conserved sequences of non-coding RNA approximately 22 nucleotides long that have been identified as critical regulators of a diverse range of cellular functions through their ability to suppress translation (Lund et al., 2004; Christopher et al., 2016). Indeed, the need for energetically inexpensive regulatory mechanisms to coordinate stress responses has drawn attention to the role of miRNAs: (i) as they can rapidly and reversibly bind and unbind their mRNA targets to exert short-term gene repression, and (ii) as the molecules are short, they are energetically inexpensive and disposable modulators. Moreover, there is large potential for regulation by these short nucleotide sequences, as one miRNA may bind to a variety of mRNA transcripts, and a single transcript may bind many miRNAs (Mazière and Enright, 2007).

To further understand the ubiquitous role of miRNAs within the cell, the stress response of the northern crayfish was examined by quantifying an array of 76 miRNAs using quantitative real-time PCR (qPCR) under normoxic, 2 h acute anoxic and 20 h chronic anoxic conditions. Two tissues that exhibit different overall metabolic responses, the hepatopancreas and tail muscle, were selected for this study (Cowan et al., 2001). These differences were highlighted by the changes in anoxia-responsive miRNAs identified, where hepatopancreas anoxia responses were found to be heavily mediated by miRNAs, while muscle tissues displayed a lesser dependence. Taken together, the differentially expressed miRNAs were found to be involved in facilitating an adaptive response to anoxia tolerance.

Freshwater Orconectes virilis (Hagen 1870) were caught in the Rideau River in Ottawa, ON, Canada, near Carleton University. Crayfish were acclimated in aerated buckets for 7 days at 15°C and then randomly separated into three containers, four animals per condition: normoxic (control), 2 h anoxia exposed (acute anoxia) or 20 h anoxia exposed (chronic anoxia). Prior to introducing crayfish to their experimental containers, water was aerated by bubbling for 45 min with air for the control condition and with 100% nitrogen gas for both anoxic stress conditions as previously described (Lant and Storey, 2011). Anoxia was defined as ≤1% O₂ per 2 Torr (∼266.6 Pa). The crayfish were killed by decapitation and tail muscle and hepatopancreas tissues were excised rapidly, flash frozen in liquid nitrogen and stored at −80°C until use.

Tail muscle and hepatopancreas tissues from n=4 different animals were weighed (100 mg), powdered in liquid nitrogen and homogenized in 1 ml of Trizol (cat. no. 15596-018; Invitrogen, Carlsbad, CA, USA) using a Polytron PT1200 homogenizer (Kinematica, Lucerne, Switzerland). Aliquots of 200 µl of chloroform were added prior to centrifugation at 10,000 rpm for 15 min at 4°C. The RNA-containing supernatant was extracted and precipitated on ice for 10 min using 500 µl of isopropanol. Samples were spun in a centrifuge at 12,000 rpm for 15 min at room temperature and the supernatant was then discarded. The RNA pellet was washed with 70% ethanol before spinning again at 7500 rpm for 5 min at room temperature. The supernatant was decanted, and pellets were air dried and then resuspended in 50 µl RNase-free water. Purity of the samples was ensured by verifying the absorbance ratio at 260 nm/280 nm using a Take3 micro-volume quantification plate (BioTek, Winooski, VT, USA) and PowerWave HT spectrophotometer (BioTek). Only samples with ratios between 1.8 and 2.2 were used. RNA integrity was verified by running samples on a 1% agarose gel stained with SYBR Green and examining the presence of sharp 28S and 18S ribosomal RNA bands. Samples were stored at −80°C until use.

RNA was prepared for miRNA analysis by polyadenylation, stem-loop primer ligation and reverse transcription, as described by Hadj-Moussa et al. (2016). The EpiBio PolyA tailing kit (cat. no. PAP5104H; Epicentre, Madison, WI, USA) was used for polyadenylation. For each reaction, 3 µg of RNA (1 µg µl⁻¹) was mixed with 1 µl of ATP (10 mmol l⁻¹), 1 μl of 10× E. coli poly A polymerase buffer (0.1 mol l⁻¹ Tris-HCl pH 8.0, 0.25 mol l⁻¹ NaCl and 10 mmol l⁻¹ MgCl₂) and 5 µl of RNase-free water for a final volume of 10 µl. Polyadenylation was performed using an Eppendorf thermocycler (Mississauga, ON, Canada) at 37°C for 30 min and reactions were terminated at 95°C for 5 min; samples were immediately chilled on ice. Specialized miRNA stem-loop primers (see Table S1 for sequence) were ligated by adding 5 µl of 250 pmol l⁻¹ stem-loop adapter primers to the polyadenylated RNA samples and heated in a thermocycler at 95°C for 5 min to denature RNA, incubated at 60°C for 5 min to allow primer annealing and then immediately chilled on ice (Biggar et al., 2014). For reverse transcription, each polyadenylated and stem-loop ligated RNA sample was combined with 1 µl mouse Maloney leukaemia virus (M-MLV) reverse transcriptase (2 U; cat. no. 18080044; ThermoFisher Scientific, Waltham, MA, USA), 1 µl of 25 mmol l⁻¹ deoxynucleotide triphosphate (dNTP) mixture (cat. no. R1121; ThermoFisher Scientific), 2 µl 0.1 mol l⁻¹ dithiothreitol (DTT) and 4 µl 5× first-strand buffer (cat. no. 18080044; ThermoFisher Scientific). Reverse transcription was initiated with 30 min of incubation at 16°C, 30 min at 42°C and 5 min at 85°C on a thermocycler. The cDNA samples were then serially diluted and stored at −20°C until use.

Primers (Integrated DNA Technologies, Coralville, IA, USA) were designed based on DNA sequence alignments with 5–7 species including invertebrates such as the red swamp crayfish (Procambarus clarkia), the kuruma prawn (Marsupenaeus japonicus), the western honeybee (Apis mellifera), and vertebrates such as zebrafish (Danio rerio) and mouse (Mus musculus). miRNA targets were selected based on previously studied arrays in P. clarkia (Xu et al., 2014), where 75 were analysed in hepatopancreas and 71 were analysed in tail muscle. miRNA sequences were obtained from miRbase (Release 21, http://www.mirbase.org/). All miRNA-specific forward primer sequences, reference gene primers and the universal reverse primer are listed in Table S3. qPCR reactions were performed using a CFX Connect™ Real-Time PCR Detection System (cat. no. 1855201; Bio-Rad, Hercules, CA, USA), in accordance with MIQE guidelines (Bustin et al., 2009). Each 20 µl PCR reaction contained: 2 µl of diluted cDNA, 10.115 µl of ddH₂O, 4 µl of 1 mol l⁻¹ trehalose (cat. no. TRE222; BioShop Canada Inc., Burlington, ON, Canada), 2 µl qPCR buffer (100 mmol l⁻¹ Tris-HCl pH 8.5, 500 mmol l⁻¹ KCl, 1.5% Triton X-100, 20 mmol l⁻¹ MgCl₂, 2 mmol l⁻¹ dNTPs and 100 nmol l⁻¹ fluorescein), 0.5 µl formamide (cat. no. FOR001; BioShop Canada Inc.), 0.5 µl of 25 mmol l⁻¹ miRNA-specific forward primer, 0.5 µl of 25 mmol l⁻¹ universal reverse primer, 0.1 µl of 100× SYBR green mix diluted in dimethyl sulfoxide (DMSO) (cat. no. S7585; Invitrogen), 0.16 µl of 25 mmol l⁻¹ dNTPs and 0.125 µl of 5 U µl⁻¹ Taq polymerase (cat. no. TAQ001.1; BioShop Canada Inc.). The miRNA amplification and quantification qPCR program used was as follows: an initial denaturation for 3 min at 95°C, then 40 cycles of 15 s at 95°C and 1 min at 60°C. A post-run melt curve analysis was performed to validate primer specificity; any reactions that displayed the amplification of more than one PCR product were rejected from our analyses and study.

ovi-miR-33-5p binding site predictions were made using HIF1α mRNA sequences of four closely related invertebrate species – Litopenaeus vannamei, Palaemon pugio, Macrobrachium nipponense and Scylla paramamosain – with SFold 2.2 STarMir bioinformatics software (http://sfold.wadsworth.org/cgi-bin/starmirtest2.pl). ALG-1 CLIP (worm) was the model used for site predictions.

HIF1α mRNA primers were designed using DNA multi-species sequence alignments; species included: whiteleg shrimp (L. vannamei), daggerblade grass shrimp (P. pugio), oriental river prawn (M. nipponense) and green mud crab (S. paramamosain). See Table S4 for HIF1α and reference gene GAPDH primer sequences. qPCR reactions were carried out as described above (see ‘Relative miRNA quantification’). The qPCR program used was as follows: an initial denaturation for 3 min at 95°C, then 49 cycles of 15 s at 95°C, 20 s at 57°C and 20 s at 72°C. A post-run melt curve analysis was used to validate primer specificity and single-product amplification.

Relative miRNA expression levels were calculated by the comparative ΔΔC_q method. Raw C_q values were linearized (2^−C_q) to standardize against endogenous controls. The miRNA reference genes used were ribosomal 5S RNA in tail muscle and U6 spliceosomal RNA in the hepatopancreas. These targets were designated as controls because of their stable expression in control and stress conditions (Schmittgen and Livak, 2008). Data were analysed as mean relative expression (mean±s.e.m.), with n=4 independent biological replicates at each sampling point and two highly comparable technical replicates for each quantification. miRNA relative expression between control, 2 h anoxia and 20 h anoxia conditions was considered significantly different (P<0.05, P<0.01) when one-way ANOVA and Dunnett's post hoc tests yielded P<0.05 using the RBIOPLOT statistics and graphing R package (Zhang and Storey, 2016).

Relative mRNA transcript abundance was also calculated by the comparative ΔΔC_q method. Raw C_q values were linearized (2^−C_q) to standardize against endogenous controls. GAPDH was the designated reference gene because of its stable expression in control and stress conditions (Schmittgen and Livak, 2008). Data were analysed as mean relative expression (mean±s.e.m.), with n=4 independent biological replicates at each sampling point and two highly comparable technical replicates for each quantification. mRNA abundance between control, 2 h anoxia and 20 h anoxia conditions was considered significantly different (as above) when one-way ANOVA and Dunnett's post hoc tests yielded P<0.05 using the RBIOPLOT statistics and graphing R package (Zhang and Storey, 2016).

In tail muscle tissue, 71 miRNAs were quantified under control (normoxic), acute 2 h anoxic and chronic 20 h anoxic conditions, and a total of four miRNAs were differentially expressed between the three conditions (Fig. 1). In the 2 h anoxic group, ovi-miR-81a-3p was downregulated and ovi-miR-3853-3p was upregulated. In the 20 h anoxic condition, two miRNAs were upregulated: ovi-miR-2788-3p and ovi-miR-2812. For a complete list of miRNAs quantified (mean relative expression and s.e.m.) in muscle tissue, refer to Table S1.

In the hepatopancreas, an array of 75 miRNA targets were quantified under normoxic and 2 h and 20 h anoxic conditions, of which 21 showed significantly decreased expression under acute anoxia compared with the normoxic condition: ovi-bantam-3p, ovi-let-7-3p, ovi-miR-100-5p, ovi-miR-125-5p, ovi-miR-12-5p, ovi-miR-133-3p, ovi-miR-1-3p, ovi-miR-190-5p, ovi-miR-235-3p, ovi-miR-263a-5p, ovi-miR-276-5p, ovi-miR-276a-3p, ovi-miR-279-3p, ovi-miR-283-5p, ovi-miR-2944c-3p, ovi-miR-305-5p, ovi-miR-307-3p, ovi- miR-317-3p, ovi-miR-33-5p, ovi-miR-750a-3p and ovi-miR-8-3p (Fig. 2). Of these 21 targets, ovi-bantam-3p, ovi-miR-133-3p, ovi-miR-1-3p, ovi-miR-263a-5p, ovi-miR-276-5p, ovi-miR-2944c-3p, ovi-miR-307-3p and ovi-miR-33-5p remained downregulated after 20 h anoxic exposure, while ovi-miR-375-3p became downregulated in addition to these after 20 h anoxic exposure. For a complete list of miRNAs quantified (mean relative expression and s.e.m.) in hepatopancreas tissue, refer to Table S2.

miRNAs in hepatopancreas that were significantly downregulated in O. virilis under 2 h anoxic exposure were analysed using the DIANA-microT-CDS algorithm (DIANA-miRPath v.3; Vlachos et al., 2015) for the identification of KEGG pathways that are targeted by these miRNAs. Using Drosophila melanogaster as the reference species, DIANA-miRPath identified 12 of the 21 differentially expressed miRNA homologues, miR-1-3p, let-7-3p, miR-8-3p, miR-87-3p, miR-283-5p, miR-317-3p, miR-12-5p, miR-279-3p, miR-133-3p, miR-305-5p, miR-190-5p and bantam-3p, as regulators of the Hippo signalling pathway, targeting 17 genes within this pathway (P<0.001). In the hepatopancreas under 20 h anoxia, five out of nine downregulated miRNAs were found to target 12 genes within the Hippo pathway (P<0.001). Other pathways identified as potentially regulated (P<0.01) by these miRNAs include: mucin type O-glycan biosynthesis, glycerophospholipid metabolism, Jak-STAT signalling and MAPK signalling. In tail muscle, because of an absence of D. melanogaster homologues of the differentially regulated miRNAs, target enrichment analysis was not possible.

miR-33-5p was predicted to bind to four mRNA transcript variants of HIF1α in closely related invertebrate species. Seedless-site binding was predicted in the 3′-UTR and the CDS of each transcript variant, as well as the 5′-UTR of P. pugio, M. nipponense and S. paramamosain. Fig. 3 illustrates a predicted binding site for miR-33-5p with the 3′-UTR of M. nipponense HIF1α mRNA transcript variant.

The relative transcript abundance of HIF1α was quantified in hepatopancreas under normoxic, and 2 h and 20 h anoxic conditions using qPCR. Transcripts levels were significantly (P<0.05) more abundant during 2 h anoxia exposure with a 1.51±0.03-fold increase, while levels were unchanged during 20 h anoxia exposure compared with control.

The ubiquitous role of miRNAs as regulators of cellular functions is widely accepted but their role in the stress response in naturally anoxia-tolerant organisms has not been deeply explored. Although exact mechanisms have yet to be deciphered, recent research has linked differential miRNA expression to various stress responses, including the anoxia response in adaptive invertebrates (Hadj-Moussa et al., 2018; Biggar et al., 2012; Lyons et al., 2015; Morin et al., 2008). In this study, 76 miRNAs were examined, and their potential involvement in coordinating the unique suite of adaptations required to survive and overcome anoxia-associated challenges was investigated. Both a tissue-specific and a stress-specific response was observed, where acute anoxia exposures and hepatopancreas tissues displayed more pronounced differential miRNA changes, while fewer changes were observed during both anoxia exposures in tail muscle tissue (Figs 1 and 2).

One central aspect of the stress response is the reprioritization of limited resources (Storey, 2015). Fig. 1 demonstrates this phenomenon in revealing few significant changes in tail muscle tissue miRNAs in response to acute and chronic anoxia. The hepatopancreas, a more metabolically active and vital organ, demonstrates considerable miRNA modulation (Fig. 2), suggesting that muscle tissue does not heavily rely on miRNA-mediated cytoprotective mechanisms to facilitate anoxia tolerance. Additionally, miRNA expression signatures differed between acute and chronic anoxia in both tissues. As the animal acclimated to chronic stress conditions, fewer significant changes were observed (Fig. 2), probably because the acute response to insult includes a myriad of behavioural and physiological changes, such as increased ventilation, shunting of haemolymph to the anterior portion of the animal to promote oxygen delivery to sensitive tissues and, later in the response, decreased heart rate and increased stroke volume (Reiber, 1995). Oxygen-deprived environments have been shown to induce a preparatory response to oxidative stress, stimulating the upregulation of antioxidant defence mechanisms in a diverse array of species (reviewed by Moreira et al., 2017). This may account, in part, for the differential stress response observed in the hepatopancreas (Fig. 2).

The differentially expressed hepatopancreas miRNAs were predicted to target stress-responsive pathways involved in maintaining homeostasis in response to changing oxygen concentrations. One such pathway that was predicted to be significantly affected by the downregulation of these miRNAs during acute anoxia exposures in hepatopancreas was the Hippo signalling pathway. Hippo signalling is a highly conserved regulatory pathway of organ growth, cell proliferation and apoptosis that has recently been implicated in the hypoxia stress response (Ma et al., 2015; Reddy and Irvine, 2008). Moreover, the Hippo signalling promotes angiogenesis and proliferation and restricts tissue growth, while decreased signalling of this pathway promotes transcription of hypoxia-responsive genes, in part through the stabilization of HIF1α (Meng et al., 2016; Morin et al., 2005). Hippo signalling has been researched extensively in both mice and fruit flies, where the core components of the pathway are conserved between vertebrates and invertebrates, including the kinases Hippo (Hpo) and Warts (Wts), and cofactors Salvador (Sav) and Yorkie (Yki) (Halder and Johnson, 2011; Huang et al., 2005; Harvey et al., 2003; Pantalacci et al., 2003; Udan et al., 2003; Wu et al., 2003). Fig. 4 illustrates the transcriptional regulation occurring through this kinase cascade, in which Hpo activates Wts through phosphorylation, which in turn phosphorylates and thus inactivates Yki, resulting in the inhibition of transcription of cell-cycle and cell death regulators (Huang et al., 2005). miR-133-3p was predicted to regulate Wts expression as in Fig. 4 and, as indicated in Fig. 2, this miRNA was downregulated in anoxic hepatopancreas tissue. This suggests the increased activity of Wts and the subsequent inhibition of Yki. It has been shown that overexpression of Yki results in defective apoptosis, cell growth and proliferation (Huang et al., 2005). Additionally, the same phenotype has been observed in systems with overexpression of the miRNA bantam, which was shown to be downregulated in hepatopancreas tissues under anoxic conditions (Figs 1 and 2) (Nolo et al., 2006). The gene encoding the bantam miRNA is activated by Yki, further suggesting that cell growth and proliferation are arrested in the hepatopancreas during oxygen stress (Nolo et al., 2006; Thompson and Cohen, 2006). The predicted overexpression of Yki via miRNA modulation suggests that cell growth and proliferation are simultaneously arrested to promote energy conservation in the hepatopancreas of the northern crayfish under acute anoxic stress (Zhou et al., 2015).

There are multiple inputs regulating HIF, but the role of hypoxic stress is of great interest; hypoxia was found to deactivate this signalling cascade through ubiquitination and protein destabilization (Ma et al., 2015; Nakayama et al., 2004). Consequently, YAP, the vertebrate orthologue of Yki, is stimulated by this process and is translocated to the nucleus to form a stable complex with HIF1α (Ma et al., 2015). HIF1 plays a critical role in the hypoxia response through transcriptional regulation of a variety of processes including metabolic adaptation, vascular remodelling, angiogenesis and cell survival (Nakayama et al., 2004; Bendinelli et al., 2013; Pugh and Ratcliffe, 2003; Semenza, 1998). Additionally, HIF1α is targeted by miRNAs that were found to be differentially regulated during anoxia (Fig. 2). The HIF1 heterodimer is composed of alpha and beta subunits, the beta subunits being constitutively expressed and the alpha subunits having a short half-life in normoxic conditions, resulting in saturated expression of the inactive protein under standard physiological conditions (Nakayama et al., 2004; Pugh and Ratcliffe, 2003). Under normoxic conditions, rapid turnover of HIF1α occurs through the ubiquitin–proteasome protein degradation pathway (Kallio et al., 1999; Salceda and Caro, 1997). HIF1α stability is determined in large part by cellular oxygen concentration but also by the YAP/Yki protein, which forms a stable complex with the α-subunit (Nakayama et al., 2004; Huang et al., 1998). In particular, and as reflected in Fig. 2, three differentially expressed miRNAs quantified in this study, miR-33-5p, miR-125-5p and miR-190-5, were shown to directly or indirectly target the alpha subunit of the HIF1 master regulator of oxygen stress (De Lella Ezcurra et al., 2016; Li et al., 2016; Zhou et al., 2015).

There is strong evidence to suggest that HIF1α is a direct target of miR-33-5p and, as such, is inhibited by miR-33-5p expression (Zhou et al., 2015). Results from binding site analysis indicated that ovi-miR-33-5p has potential seedless-site binding in the 3′-UTR of invertebrate HIF1α mRNA (Fig. 4). It was recently proposed that seedless-site binding of miRNAs may account for as much as 40% of total miRNA–mRNA binding in some instances (Loeb et al., 2012). Fig. 2 shows decreased abundance of miR-33-5p in the hepatopancreas in response to acute and chronic anoxia, suggesting an active role for HIF1α in the sustained regulation of anoxia in this tissue. One recent study found that decreased levels of miR-33-5p resulted in increased proliferation, and conversely overexpression of this miRNA resulted in decreased proliferation of certain cell lines (Zhou et al., 2015). miRNAs have been studied for their vast role in the simultaneous arrest of cell proliferation and apoptosis in metabolically depressed animals (Biggar and Storey, 2011). We hypothesize that this mechanism, in conjunction with other elements of the Hippo signalling pathway, may be activated in anoxic crayfish to regulate proliferation and inhibit apoptosis in critical organs. To avoid fatal ATP depletion, animals under reduced metabolic states must arrest energetically expensive processes such as proliferation and growth (Biggar and Storey, 2011). It was also speculated that other animal models of anoxia tolerance exhibit regulation of apoptosis and proliferation under stress via miRNA control (Morin et al., 2008; Biggar et al., 2009). Energy metabolism and cell survival are among the responses regulated by HIF1; therefore, it follows that this transcription factor would be regulated during anoxia, as was previously observed in studies on metabolically depressed animals (Morin et al., 2005; Maistrovski et al., 2012).

Interestingly, miR-190-5, which has the opposite effect on HIF1α, was shown to decrease significantly only under acute anoxia in the hepatopancreas (Fig. 2). The oxygen-dependent degradation of HIF1α occurs through prolyl hydroxylation by catalytic enzymes that use oxygen as a co-substrate under normoxic conditions (Jaakkola et al., 2001). This mechanism is further regulated by miRNAs and, specifically, miR-190 has been identified as targeting the prolyl hydroxylase fatiga, an essential negative regulator of hypoxic responses, and was found to be overexpressed in response to hypoxia (De Lella Ezcurra et al., 2016). One explanation proposed for the unexpected decreased abundance of this positive HIF1α regulator is that while the study by De Lella Ezcurra et al. (2016) examined the response to hypoxia (11%, 8% and 5% oxygen), the present study examined the response to anoxia (≤1% O₂ per 2 Torr). Furthermore, because the fatiga gene is inactivated by anoxia, the additional negative regulation by miR-190 may not serve an energetic benefit to the cell if HIF1α will not be degraded under anoxia, regardless of the abundance of the prolyl hydroxylase enzyme.

miR-125-5p is known to target HIF1α indirectly, where decreased levels of this miRNA result in the inhibition of proliferation and activation of hepatic stellate cells (Li et al., 2016). This miRNA binds to factor inhibiting HIF1 (FIH-1), which in turn accumulates and negatively regulates HIF1α transcriptional activity (Li et al., 2016; Mahon et al., 2001). The complex repression of HIF1α by FIH-1 and its co-repressor VHL has been characterized in normoxic conditions; however, these novel findings suggest that, under anoxia, a different mechanism of expression is engaged (Mahon et al., 2001). Because miR-125-5p levels were shown to decrease in the hepatopancreas under acute anoxia (Fig. 2), FIH-1 levels were expected to increase, and thus increase the negative regulation of HIF1α.

Transcript levels of HIF1α have previously been shown to increase in response to low-oxygen stress in an anoxia-tolerant insect (Morin et al., 2005). The activation of HIF1α in another invertebrate suggests the upregulation of hif1a we measured during acute anoxia in crayfish hepatopancreas is a potentially conserved response for the adaptation to anoxia-associated challenges. Previous studies have observed that HIF1α transcript abundance peaked between 1 and 2 h exposure but decreased to near-basal levels after 8 h exposure before finally increasing again following 16 h anoxia exposure (Wang et al., 1995). This agrees with the present findings of HIF1α mRNA abundance over the anoxia time course. These findings also reinforce the hypothesis that the HIF1 mechanism of regulating anoxia-responsive genes is highly conserved throughout the animal kingdom and is under miRNA regulation (Morin et al., 2005; Maistrovski et al., 2012; Gorr, 2004). In accordance with the findings of Morin et al. (2005), the increased transcript levels of hif1a also suggest that the transcription factor activation is promoted both through the increased transcription of the hif1α gene and simultaneously, the inhibition of HIF1α degradation.

We also found ovi-miR-133-5p to be significantly downregulated in the hepatopancreas, emphasizing that anti-apoptotic mechanisms are activated in this vital organ under anoxia. Indeed, studies have shown that miR-133 promotes hypoxia-induced cell death through the suppression of anti-apoptotic proteins while promoting the activation of pro-apoptotic proteins (Li et al., 2015). Also downregulated under anoxia stress in the hepatopancreas are miR-263, miR-276 and miR-317, all of which are known to be upregulated at various stages of development among invertebrates (Surridge et al., 2011). This suggests that these miRNAs are downregulated in the hepatopancreas to suppress cell proliferation and growth while the animal is in an anoxia-induced metabolically depressed state.

In contrast to the hepatopancreas, tail muscle may experience a smaller effect of anoxia insult as muscle tissue is less oxygen sensitive compared with hepatopancreas and therefore has a lesser effect on the survivorship of the animal (Reiber, 1995). This is a potential explanation for the total of four differentially expressed miRNAs during anoxia in tail muscle, three of which exhibited upregulation and one of which exhibited downregulation (Fig. 1). The role of these four miRNAs has not yet been characterized and there are currently no known homologues in D. melanogaster. However, miR-2788 has been previously identified in an invertebrate, exhibiting upregulation during development, so a similar suppression in proliferation and cell growth to that proposed in hepatopancreas is also proposed in tail muscle (Surridge et al., 2011). Presently, there is a lack of functional targeting information available for these miRNAs and this is probably due to their crustacean-specific functions that have yet to be explored deeply.

Overall, this study elucidates the central and newfound role of miRNAs as modulators of gene translation during anoxia exposure. Indeed, these findings highlight the tissue-specific nature of miRNA targeting, where pronounced responses were observed in the hepatopancreas under acute anoxic stress. This suggests that more metabolically active tissues such as the hepatopancreas are more dependent on miRNA-mediated anoxia responses than the metabolically inactive muscle, and that a rapid response during acute anoxia exposure is required to provide immediate protection against anoxia insult. Key to the hepatopancreas response is the regulation of Hippo signalling, which modulates proliferation and apoptotic pathways, and the HIF1 transcription factor, which was found to be upregulated during acute anoxia and was also targeted by a variety of miRNAs itself. These novel findings in O. virilis have major implications for how the anoxia response modulates – and is modulated by – miRNAs in aquatic invertebrates.

We thank Jan Storey for her scientific insight and editorial review of the manuscript.