Differential Gene Expression in the Liver of the African Lungfish, Protopterus annectens, after 6 Months of Aestivation in Air or 1 Day of Arousal from 6 Months of Aestivation

The African lungfish, Protopterus annectens, can undergo aestivation during drought. Aestivation has three phases: induction, maintenance and arousal. The objective of this study was to examine the differential gene expression in the liver of P. annectens after 6 months (the maintenance phase) of aestivation as compared with the freshwater control, or after 1 day of arousal from 6 months aestivation as compared with 6 months of aestivation using suppression subtractive hybridization. During the maintenance phase of aestivation, the mRNA expression of argininosuccinate synthetase 1 and carbamoyl phosphate synthetase III were up-regulated, indicating an increase in the ornithine-urea cycle capacity to detoxify ammonia to urea. There was also an increase in the expression of betaine homocysteine-S-transferase 1 which could reduce and prevent the accumulation of hepatic homocysteine. On the other hand, the down-regulation of superoxide dismutase 1 expression could signify a decrease in ROS production during the maintenance phase of aestivation. In addition, the maintenance phase was marked by decreases in expressions of genes related to blood coagulation, complement fixation and iron and copper metabolism, which could be strategies used to prevent thrombosis and to conserve energy. Unlike the maintenance phase of aestivation, there were increases in expressions of genes related to nitrogen, carbohydrate and lipid metabolism and fatty acid transport after 1 day of arousal from 6 months aestivation. There were also up-regulation in expressions of genes that were involved in the electron transport system and ATP synthesis, indicating a greater demand for metabolic energy during arousal. Overall, our results signify the importance of sustaining a low rate of waste production and conservation of energy store during the maintenance phase, and the dependence on internal energy store for repair and structural modification during the arousal phase, of aestivation in the liver of P. annectens.


Introduction
Lungfishes are an archaic group of Sarcopterygian fishes characterized by the possession of a lung opening off the ventral side of the oesophagus. They hold an important position in the evolutionary tree with regard to water-land transition, during which many important physiological and biochemical adaptations occurred (e.g. air-breathing, urea synthesis, redirection of blood flow, heart partitioning). These adaptations supposedly facilitated the migration of fishes to terrestrial environments, leading to the evolution of tetrapods. There are six species of extant lungfishes, four of which (Protopterus aethiopicus, P. amphibius, P. annectens and P. dolloi) are found in Africa. African lungfishes are obligate air-breathers; they typically inhabit fringing weedy areas of lakes and rivers where dissolved oxygen levels are low, daytime temperatures are high, and seasonal drying is common. Without limbs to facilitate locomotion on land, lungfishes would have to passively tolerate desiccation, and aestivation could be the only means for survival under desiccation at high temperature. Aestivation involves corporal torpor at high environmental temperature with absolutely no intake of food and water for an extended period. African lungfishes can aestivate in subterranean mud cocoons for~4 years [1], which could be the longest aestivation period known for vertebrates.
Traditionally, aestivation experiments on African lungfishes were performed either in mud or in cloth bags in the laboratory [2][3][4][5]. Chew et al. [6] were the first to achieve induction of aestivation in P. dolloi in pure mucus cocoons in air inside plastic boxes. Subsequently, it has been confirmed that P. annectens, P. aethiopicus [7][8][9][10][11] and P. amphibius (Y.K.I. and S.F.C, unpublished observation) can also be induced to aestivate in pure mucus cocoons in air. There are three phases of aestivation. During the induction phase in air, the fish detects environmental cues and turn them into some sort of internal signals that would instill the necessary changes at the behavioral, structural, physiological and biochemical levels in preparation of aestivation. It secretes a substantial amount of mucus which turns into a dry cocoon within 6-8 days. Aestivation begins when the fish is completely encased in a dried mucus cocoon, and there is a complete cessation of feeding and locomotor activities. During the maintenance phase, the fish has to preserve the biological structures and sustain a slow rate of waste production to avoid pollution of the internal environment. It can perpetuate to aestivate under such conditions for more than a year. The aestivating lungfish can be aroused from aestivation by the addition of water. Upon arousal, the fish struggles out of the cocoon and swims, albeit sluggishly, to the water surface to gulp air. After arousal, it excretes the accumulated waste products, and feeds for repair and growth. Completion of aestivation occurs only if arousal is successful; if not, the animal have had apparently succumbed to certain factors during the maintenance phase. Feeding begins approximately 7-10 days after arousal, and the fish grow and develop as normal thereafter.
It is apparent that adaptive (physiological, biochemical and molecular) changes in various organs of the aestivating African lungfish would vary during the three phases of aestivation. However, the majority of studies in the past focused only on the maintenance phase, and there is a dearth of information on the induction and arousal phases of aestivation [12]. Loong et al. [13] pioneered in using suppression subtractive hybridization (SSH) polymerase chain reaction (PCR) to identify aestivation-specific gene clusters in the liver of P. annectens after 6 days (induction phase) of aestivation in a mucus cocoon in air (normoxia). They reported up-or down-regulation of several gene clusters which were involved in urea synthesis, prevention of clot formation, activation of the lectin pathway for complement activation, conservation of minerals (e.g. iron and copper) and increased production of hemoglobin beta. Since there were up-and down-regulation of mRNA expressions of genes related to ribosomal proteins and translational elongation factors, there could be simultaneous increases in protein degradation and protein synthesis during 6 days of aestivation, confirming the importance of reconstruction of protein structures in preparation for the maintenance phase of aestivation [13].
The liver is involved in diverse metabolic activities which include detoxification, oxidative defense, urea synthesis, carbohydrate and amino acid metabolism, and iron and copper metabolism. Even during the maintenance phase of aestivation, the liver has to continue functioning to detoxify ammonia to urea; only then, would the aestivating fish be able to mobilize protein and amino acid as an energy source for survival during the aestivation process. Therefore, in this study, we continued to examine the effects of 6 months of aestivation and 1 day arousal from 6 months of aestivation on the up-and down-regulation of genes in the liver of P. annectens using SSH PCR. SSH involves two types of cDNAs: testers (with treatment) and drivers (control). In order to examine differential gene expression in the liver during the maintenance phase (6 months) of aestivation (tester), liver of fish kept in fresh water was used as the driver. Results obtained would indicate changes in gene expression in aestivating fish with reference to non-aestivating fish. However, in order to examine differential gene expression in the liver during the arousal phase (1 day arousal from 6 months of aestivation) of aestivation (tester), liver of fish that had undergone 6 months of aestivation in air were used as driver instead. In this way, results obtained would reveal changes in gene expression in aroused fish with reference to aestivating fish. The zebrafish nomenclature system (see https://wiki.zfin.org/display/general/ZFIN+ Zebrafish+Nomenclature+Guidelines) for genes and proteins of fish origin and the human nomenclature (see http://www.genenames.org/guidelines.html) for genes and proteins of mammalian origin were adopted in this paper. Specifically, for fishes, gene symbols are italicized, all in lower case, and protein designations are the same as the gene symbol, but not italicized with the first letter in upper case.

Collection and maintenance of fish
Protopterus annectens (80-120 g body mass) were imported from Central Africa through a local fish farm in Singapore. They were maintained in plastic aquaria filled with dechlorinated freshwater at pH 7.0 and at 25°C in the laboratory. Water was changed daily. No attempt was made to separate the sexes. Fish were acclimated to laboratory conditions for at least 1 month before experimentation. During the adaptation period, fish were fed with frozen fish meat and food was withheld 96 h prior to experiments.

Ethics Statement
Approval to undertake this study was obtained from the Institutional Animal Care and Use Committee of the National University of Singapore (IACUC 035/09).

Experimental conditions and tissue sampling
Protopterus annectens were induced to aestivate at 27-29°C and 85-90% humidity individually in plastic tanks (L29 cm x W19 cm x H17.5 cm) containing 15 ml of dechlorinated tap water (adjusted to 0.3‰ with seawater) following the procedure of Chew et al. [6]. During the induction phase of aestivation, the experimental fish would secrete plenty of mucus during the first few days, and the mucus would slowly dry up between day 5 and day 7 to form a mucus cocoon. Aestivation was considered to begin when the fish was fully encased in the cocoon and displayed no locomotor activities. Protopterus annectens can be maintained in aestivation for a long period of time and this was regarded as the maintenance phase of aestivation.
Fish maintained in freshwater served as controls. Control fish were killed with an overdose of neutralized MS222 (0.2%) followed with a blow to the head. Aestivating fish were killed on day 186 (6 months; prolonged maintenance phase) or after 1 day arousal from 6 months of aestivation with a blow to the head. The liver was quickly excised and frozen in liquid nitrogen. The frozen samples were kept at-80°C until analysis.
Total RNA and poly (A) mRNA extraction Frozen tissues were homogenized using a polytron homogenizer (Kinematica AG, Lucerne, Switzerland) in 400 μl of chaotropic buffer (4.5 M guanidine thiocyanate, 2% N-lauroylsarcosine, 50 mM EDTA (pH 8.0), 25 mM Tris-HCl (pH 7.5), 0.1 M β-mercaptoethanol, 0.2% antifoam A). Total RNA was extracted from the liver, using the chaotropic extraction protocol described by Whitehead and Crawford [14]. The RNA pellet obtained was rinsed twice with 500 μl of 70% ethanol, and further purified using the Qiagen RNeasy Mini Kit (Qiagen Inc., Valencia, CA, USA). The concentration and purity of the purified RNA were determined using the NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, DE, USA). The RNA quality was determined by visualising the presence of the 18S and 28S ribosomal RNA bands using the BioRad Universal Hood II gel documentation system (BioRad, Hercules, CA, USA) after carrying out electrophoresis of 1 μg of RNA on 1% (w/v) agarose gel in TAE buffer (40 mM Tris-acetate, 1 mM EDTA, pH 8.0) with nucleic acid staining dye GelRed (1:20000, Biotium Inc., Hayward, CA, USA) at 100 V for 30 min. The presence of sharp 28S and 18S bands in the proportion of about 2:1 indicate the integrity of the total RNA.
Poly (A) mRNA was extracted from 200 μg of total RNA using the Oligotek mRNA kit (Qiagen Inc.). The RNA sample (200 μg) was mixed with 15 μl of Oligotex suspension (resin) and was heated at 70°C for 3 min and then cooled at 25°C for 10 min. The Oligotex:mRNA complex was spun at 14,000 xg and the pellet obtained was resuspended in 400 μl of Buffer OW2 (Qiagen Inc.) and then passed through a small spin column by centrifuging at 14,000 xg for 1 min. The column was washed with another 400 μl of Buffer OW2. The resin in the column was resuspended with 50 μl of hot (70°C) Buffer OEB (Qiagen Inc.) and eluted by centrifugation at 14,000 xg for 1 min to obtain the Poly (A) RNA. Another 50 μl of hot (70°C) Buffer OEB was added to the column and the process was repeated to ensure maximal Poly (A) mRNA yield.

Construction of SSH libraries
Two sets of forward (up-regulated genes) and reverse (down-regulated genes) SSH libraries for the liver were generated using the PCR-Select cDNA subtraction kit (Clontech Laboratories, Inc., Mountain View, CA, USA); one set for fish aestivated for 6 months in air (prolonged maintenance phase) with reference to the freshwater control, and the other set for fish that was aroused for 1 day after 6 months of aestivation in air (arousal phase) with reference to 6 months of aestivation in air. Two micrograms of poly (A) mRNA from each condition was used for cDNA synthesis. After the first and second strand synthesis, the double stranded cDNA from both groups were digested with Rsa I. A portion of the digested cDNA was ligated with either Adapter 1 or Adaptor 2R, and the rest was saved for subsequent usage as the driver for hybridization. The forward library was generated from the hybridization between adapterligated cDNA obtained from fish that had undergone 6 months of aestivation in air or fish that were recovered for 1 day (tester) and Rsa I-digested cDNA from the control fish kept in fresh water or fish aestivated for 6 months in air (driver). The reverse library was made the same way, except that the adapter-ligated cDNA from the control in fresh water or 6 months of aestivation served as the tester while the Rsa I-digested cDNA from fish aestivated for 6 months in air or fish that were recovered for 1 day acted as the driver, respectively. The driver cDNA was added in excess to remove common cDNA by hybrid selection, leaving over-expressed and novel tester cDNAs to be recovered and cloned. The PCR amplification of the differentially expressed cDNAs was performed using the Advantage cDNA polymerase mix (Clontech Laboratories, Inc.) and 9902 Applied Biosystems PCR thermal cycler (Life Technologies Corporation, Carlsbad, CA, USA). The primary and secondary PCR amplification of these reciprocal subtractions of cDNA from the control and aestivated fish produced 1 forward and 1 reverse SSH libraries enriched in differentially expressed transcripts.
Differentially expressed cDNAs were cloned using pGEM-T easy vector system kit (Promega Corporation, Madison, WI, USA), transformed into chemically competent JM109 Escherichia coli (Promega Corporation), and plated onto Luria-Bertani (LB) agar with ampicillin, 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) and isopropyl β-D-thiogalactopyranoside (IPTG). Selected white colonies were grown overnight in LB broth with ampicillin. The plasmids were extracted using the resin-based plasmid miniprep kit (Axygen Biosciences, Union City, CA, USA). The plasmids were quantified by the NanoDrop ND-1000 spectrophotometer. Approximately 80-100 ng of plasmid DNA was used in BigDye Terminator v3.1 Cycle Sequencing Kit (Life Technologies Corporation) with 2 μM T7 primers. Excess fluorescent nucleotides and salts were removed from the samples by ethanol precipitation. The dried samples were resuspended in Hi-Di Formamide (Life Technologies Corporation) before loading to the Prism 3130XL sequencer (Life Technologies Corporation). A total of 500 clones for each forward and reverse library were selected for sequencing.
Sequence output was exported as text and edited manually to remove vector sequences using BioEdit Sequence Alignment Editor software version 7.0.9 [15]. BLAST searches were performed using the tBLASTx algorithm [16] and default search conditions. Proteins were considered significant when the E value was <1E-04. The annotated sequences were grouped based on Gene Ontology classification. The sequences were deposited in Genbank EST database and were assigned with accession numbers JZ575382 to JZ575617.

Relative quantitative real-time PCR (qPCR)
In order to validate the changes obtained in the SSH studies, nine genes were selected for the determination of mRNA expression using quantitative real-time PCR (qPCR). These include acyl-CoA desaturase (acd), argininosuccinate synthetase 1 (ass1), betaine-homocysteine Smethyltransferase 1 (bhmt1), ceruloplasmin (cp), carbamoyl phosphate synthetase III (cpsIII), fumarate hydratase (fh), ferritin light chain (ftl), glyceraldehyde-3-phosphate dehydrogenase (gapdh) and superoxide dismutase 1 (sod1). Prior to first strand cDNA synthesis, RNA from the liver of fish kept in fresh water, aestivated for 6 months in air or aroused for 1 day after 6 months of aestivation in air were treated separately with Deoxyribonuclease I (Qiagen Inc.) to remove any contaminating genomic DNA. First strand cDNA was synthesized from 1 μg of total RNA using random hexamer primer and the RevertAid first stand cDNA synthesis kit, following the manufacturer's instruction (Thermo Fisher Scientific Inc). mRNA expressions of the selected genes were quantified using a StepOnePlus Real-Time PCR System (Life Technologies Corporation). Each PCR reaction contained 5 μl of 2x Fast SYBR Green Master Mix (Life Technologies Corporation), a certain aliquot of gene-specific primers (listed in Table 1) and 0.1-2 ng of cDNA in a total volume of 10 μl. Samples were run in triplicate. qPCR reactions were performed with the following cycling conditions: 95°C for 20 s (1 cycle), followed by 40 cycles of 95°C for 3 s and 60°C of 30 s. Data was collected at each elongation step. Each run was followed by a melt curve analysis by increasing the temperature from 60°C to 95°C at 0.3°C increment to confirm the presence of only a single PCR product. In addition, random PCR products were electrophoresed in a 1.8% agarose gel to verify that only one band was present. All the data were normalized to the abundance of β-actin mRNA. The amplification efficiencies for β-actin and all selected genes were between 90-100%. The subsequent application of the 2 -ΔΔCT calculation for relative quantification was validated by confirming that the variation between the amplification efficiencies of the target and reference gene through a 100-fold dilution remained relatively constant [17]. The mean fold-change values were transformed into logarithmic values (log 2 ) to enable valid statistical analysis.

Statistical analysis
Results for qPCR were presented as means ± standard errors of the mean (S.E.M.). Student's ttest was used to evaluate the difference between means. Differences with P<0.05 were regarded as statistically significant.

Results
SSH libraries from liver of P. annectens after 6 months of aestivation (with fresh water control as the driver) Two SSH-generated libraries, forward ( Table 2) and reverse (Table 3), were constructed for genes that were up-and down-regulated, respectively, in the liver of P. annectens which had undergone 6 months of aestivation in air. A total of 98 genes were identified from these SSH libraries, of which 20 genes were up-regulated (Table 2) and 78 genes were down-regulated (Table 3). There were 340 unidentified sequences which could be genes that are yet to be characterized in P. annectens. Ribosomal protein S12 appeared in both forward and reverse Table 1. Primers used for quantitative real-time PCR on acyl-CoA desaturase (acd), argininosuccinate synthetase 1 (ass1), betainehomocysteine S-methyltransferase 1 (bhmt1), ceruloplasmin (cp), carbamoyl-phosphate synthetase III (cpsIII), fumarate hydratase (fh), ferritin light chain (ftl), glyceraldehyde-3-phosphate dehydrogenase (gapdh), superoxide dismutase 1 (sod1) from the liver of Protopterus annectens.

Gene
Primer sequence (5' to 3') subtraction libraries, indicating that it could be false positives or encoding for different isoforms of the same protein.
The forward library indicated the up-regulation of bhmt1 and fh expression levels in the liver of P. annectens after 6 months of aestivation. Certain genes related to nitrogen metabolism such as ass1 and cps III and a number of ribosomal genes that was involved in protein synthesis were also up-regulated ( Table 2).
The reverse library indicated the down-regulation of expression levels of genes related to antioxidative stress (e.g. sod1) and copper transport (e.g. cp) in the liver of P. annectens after 6 Table 2. Known transcripts found in the forward library (up-regulation) obtained by suppression subtractive hybridization PCR from the liver of Protopterus annectens aestivated for 6 months in air with fish kept in fresh water as the reference for comparison.

Group and Gene
Gene symbol P. annectens accession no.  months of aestivation. The mRNA expression levels of some genes involved in complement activation, blood coagulation and iron transport were also down-regulated (Table 3). Relative quantification of mRNA expression levels of selected genes were performed using qPCR to verify the up-or down-regulated of selected genes. In agreement with the SSH results Differential Gene Expression in the Liver of the African Lungfish of the forward library, there were significant increases in the mRNA expression levels of bhmt1, fh, ass1, cps III in the liver of P. annectens after 6 months of aestivation (Fig. 1A-D). In addition, there were significant decreases in the mRNA expression levels of sod1 and cp in corroboration of the SSH results ( Fig. 1E and F).
SSH libraries from liver of P. annectens after 1 day of arousal from 6 months of aestivation (with 6 months of aestivation as the driver) Similarly, forward (Table 4) and reverse (Table 5) libraries were constructed to reflect the genes that were up-and down-regulated, respectively, in the liver of P. annectens after 1 day of arousal from 6 months of aestivation. A total of 143 genes were identified from these subtraction libraries, in which 76 genes were up-regulated (Table 4) and 67 genes were down-regulated ( Table 5). Out of these 1000 sequences obtained, 391 were unidentified and they could again be genes that are yet to be characterized in P. annectens. Fructose-bisphosphate aldolase B (aldob) and some genes related to ribosomal proteins appeared in both forward and reverse subtraction libraries, indicating that they could be false positives or encoding for different isoforms of the same protein.
As revealed by the forward library (up-regulation), the mRNA expressions of aldob and ass1, related to carbohydrate and nitrogen metabolism, respectively, were up-regulated in the liver of P. annectens after 1 day of arousal from 6 months of aestivation. Some genes involved in lipid metabolism (acd, desaturase 2, fatty acid binding protein and stearoyl-CoA desaturase), ATP synthesis and iron metabolism (ftl, ferritin middle subunit and transferrin-a) were also up-regulated ( Table 4).
The reverse library (down-regulation) revealed that the down-regulation of expression levels of certain genes related to carbohydrate metabolism (aldob and plasma alpha-L-fucosidase precursor putative) in the liver of P. annectens after 1 day of arousal from 6 months of aestivation. The mRNA expression levels of some genes related to protein synthesis, signaling and iron metabolism (alpha globin chain, ferritin heavy chain and transferrin) were also down-regulated (Table 5).
In support of the SSH results, there were significant increases in the mRNA expression levels of acd, ftl and gapdh in the liver of P. annectens after 1 day of arousal from 6 months of aestivation as confirmed by qPCR (Fig. 1G-I).

Discussion
Maintenance phase: up-regulation of ornithine-urea cycle (OUC) capacity African lungfishes are ureogenic and they possess a full complement of OUC enzymes including CpsIII in their livers [8,18,19]. During the maintenance phase of aestivation, ammonia released through amino acid catabolism must be detoxified because its excretion would have been completely impeded during desiccation [12]. By synthesizing and accumulating urea, which is less toxic, P. annectens can carry out protein catabolism for a longer period without being intoxicated by ammonia [12]. Therefore, there is a need to increase the urea-synthesizing capacity during the maintenance phase of aestivation. Indeed, there were increases in mRNA expression levels of OUC enzymes, particularly ass1 and cpsIII, in the liver of P. annectens after 6 months of aestivation (Table 1). There was also a significant increase in the expression level of fh. Fh catalyzes the reversible conversion between fumarate and malate and is believed to play an important role in the tricarboxylic acid cycle [20]. It can also be involved in nitrogen metabolism as it could regulate the fumarate levels produced by the OUC [20]. BHMT is a cytosolic zinc metalloprotein belonging to the family of methyltransferases [21]. It catalyzes the transfer of a methyl group to homocysteine to form methionine [22], and contributes to~50% of methionine synthesis in liver [23]. In human, defects in methionine and cysteine metabolism in the liver lead to increased homocysteine concentration in the plasma, i.e. hyperhomocysteinemia, which is associated with vascular diseases [24,25], birth defects such as spina bifida [26], and neurodegenerative diseases such as Alzheimer's disease [27]. When accumulated abnormally in tissues and organs, homocysteine can produce multiple deleterious changes simultaneously [28], leading to multi-organ failure involving the brain, kidney, heart, vascular system and/or musculoskeletal system [29][30][31][32]. Hence, it is highly probable that bhmt1/Bhmt1 expressions were up-regulated in the liver of P. annectens to reduce the hepatic homocysteine concentration during the maintenance phase of aestivation as suggested by Ong et al. [33].
Maintenance phase: down-regulation of genes related to blood coagulation As the heart rate of African lungfish, P. aethiopicus, drops from 22-30 beats min -1 before aestivation to 12-17 beats min -1 by the end of 1-1.5 months in the mud [34], it is probable that a severe decrease in the rate of blood flow would have occurred. Thus, any mechanism that can prevent the formation of a thrombosis when the fish is inactive during aestivation would be of considerable survival value. Indeed, several genes related to blood coagulation, which included fibrinogen (7 clones), apolipoprotein H (8 clones) and serine proteinase inhibitor clade C (antithrombin) member 1 (serpinc1; 3 clones) were down-regulated in the liver of fish after 6 months of aestivation (Table 3) and this could signify a decrease in the tendency of blood clot formation.

Maintenance phase: down-regulation of sod1
SOD is an antioxidant enzyme that catalyzes the dismutation of two O 2 •to H 2 O 2 , and therefore plays a central role in antioxidation. An adaptive response against oxidative stress is often marked by the increased production of intracellular antioxidant enzymes such as SOD, catalase, glutathione peroxidase and glutathione reductase to protect the macromolecules from the stress-induced damage. It was suggested that up-regulation of intracellular antioxidant enzymes during aestivation and hibernation protects against stress-related cellular injury [35,36]. However, the down-regulation in the mRNA expression of sod1 in the liver of P. annectens after 6 months of aestivation (Table 3) suggests that other antioxidant enzymes such as Bhmt1, glutathione-S-transferase, glutathione reductase, glutathione peroxidase or catalase may be involved and their activities would be sufficient to counteract the oxidative stress. Also, these results could be indicative of a decrease in ROS production during the maintenance phase of aestivation due to a slower metabolic rate, including the rate of nitrogen metabolism.      Maintenance phase: down-regulation of genes related to complement fixation The complement system mediates a chain reaction of proteolysis and assembly of protein complexes that results in the elimination of invading microorganisms [37,38]. Three activation pathways (the classical, lectin and alternative pathways) and a lytic pathway regulate these events. Protopterus annectens utilizes lectin pathway for protection against pathogens during the induction phase of aestivation [13]. However, our results showed that many genes related to complement fixation appeared in the reverse library. These included the complement C3 precursor alpha chain (11 clones), complement component 4 binding protein alpha (3 clones) and CD46 antigen complement regulatory protein (2 clones), and seven others (Table 3). Hence, P. annectens might down-regulate the classical complement fixation pathway during the maintenance phase of aestivation, possibly because of three reasons. Firstly, the dried mucus cocoon was already well formed, which conferred the aestivating lungfish a certain degree of protection against external pathogens. Secondly, tissue reconstruction would have subsided after the induction phase, and there could be minimal tissue inflammation during the prolonged maintenance phase. Thirdly, it was important to conserve the limited energy resources, and it would be energetically demanding to sustain the increased expression of genes involved in complement fixation during the maintenance phase of aestivation. Maintenance phase: down-regulation of warm-temperature-acclimationrelated 65 kDa protein and hemopexin The plasma glycoprotein warm-temperature-acclimation-related protein (Wap65) was first identified in the goldfish Carassius auratus [39] and the cDNA showed a homology of 31% to rat hemopexin, a serum glycoprotein that transports heme to liver parenchymal cells [40]. Hemopexins in mammals are mainly synthesized in liver and are responsible for the transportation of heme resulting from hemolysis to the liver. Therefore, the down-regulation of the wap65 and hemopexin in the liver of P. annectens (Table 3) suggested that hemolysis might be suppressed during the maintenance phase of aestivation. There are also indications that the Wap65 can be involved in immune responses in the Channel catfish Ictalurus punctatus [41]. Hence, its down-regulation suggested that a decrease in immune response might have occurred in the liver of P. annectens during the maintenance phase of aestivation.
Maintenance phase: down-regulation of genes related to iron metabolism Iron is involved in many cellular metabolic pathways and enzymatic reactions, but it is toxic when in excess [42][43][44]. Transferrin is one of the major serum proteins, which is synthesized mainly in liver and plays a crucial role in iron metabolism. Under normal conditions, most of the iron in the plasma is bound to transferrin, and iron-transferrin complexes enter the cells via a transferrin receptor-mediated endocytic pathway. Transferrin also has a close relationship with the immune system. It binds to iron, creating an environment with low levels of iron, where few microorganisms can survive and prosper [45]. On the other hand, ferritin is the main iron storage protein in both eukaryotes and prokaryotes; it keeps iron in a soluble and non-toxic form [43,46,47]. Also, up-regulation of ferritin has been observed in oxidative stress [48] and inflammatory conditions in human [49][50][51]. Transferrin and ferritin mRNA expression levels are up-regulated in P. annectens during the induction phase of aestivation [13], probably due to oxidative stress and inflammation arisen through tissue reconstruction, and/or a high turnover rate of free and bound iron resulting from increased production of certain types of hemoglobins or hemoglobin in general. By contrast, our results indicated that there could be a decrease in the capacity of iron metabolism and transport in P. annectens during the maintenance phase of aestivation as transferrin (14 clones) and hemopexin (3 clones) appeared in the reverse library. This correlated well with the aestivation process as a prolonged torpor state would theoretically lead to a lower rate of ROS production, and stabilized expression of hemoglobin genes.
Maintenance phase: down-regulation of genes related to copper metabolism Ceruloplasmin (CP) is crucial in the oxidation of Fe 2+ to Fe 3+ , which enables the binding of iron to transferrin, facilitating the mobilization of iron in the body. It also represents a tightly bound pool of copper that accounts for >90% of the total plasma copper in most species [52,53]. CP synthesis and/or secretion can be altered by inflammation, hormones, and copper. Plasma concentrations of acute-phase globulins, including CP, increase with tissue injury, localized acute inflammation, and chronic inflammatory diseases [54]. The mRNA expression level of cp was up-regulated in the liver of P. annectens during the induction phase of aestivation [13]. However, our results revealed that 6 months of aestivation led to a down-regulation of cp mRNA expression in the liver of P. annectens. This suggested that tissue degradation or inflammation may be limited during the maintenance phase of aestivation due to a profound decrease in metabolic activity. Consequently, there was no longer a need to up-regulate expression level of cp.
Maintenance phase: up-or down-regulation of protein synthesis?
Twelve genes related to protein synthesis, transport and folding appeared in the reverse library of lungfish undergoing 6 months of aestivation in air ( Table 3). The down-regulation of genes related to protein synthesis such as eukaryotic translation initiation factors and other ribosomal proteins is a consistent phenomenon in metabolic rate reduction. Suppression of protein synthesis during aestivation would help the animal to conserve energy and enhance its survival. However, 10 types of ribosomal proteins appeared in the forward library indicating up-regulation of mRNA expressions of these genes in the liver of P. annectens after 6 months of aestivation (Table 2). Taken altogether, these results indicate that the capacity of protein synthesis was not suppressed completely during the prolonged phase of aestivation. This could be an important strategy since the aestivating lungfish would have to maintain the protein synthesis machinery in preparation for arousal from aestivation when water becomes available.
Arousal phase: up-regulation of ass1 expression and amino acid metabolism After 1 day of arousal from 6 months of aestivation, ass1 still appeared in the forward library ( Table 4), indicating that there was a further increase in the mRNA expression of ass1 in the liver. Since cpsIII and fh could not be found in the reverse library (Table 5), and their mRNA expressions were already up-regulated during the maintenance phase of aestivation, it can be deduced that their increased mRNA expressions were sustained into the arousal phase. Upon arousal, the fish has to reconstruct cells and tissues that have been modified during the induction phase and repair damages that have occurred during the maintenance phase of aestivation. Such structural changes would require increased syntheses of certain proteins, and since refeeding would not occur until 7-10 days after arousal, it would imply the mobilization of amino acids of endogenous origin [12]. Both substrate and energy are needed for repair and regeneration. Our results indicate that endogenous amino acids could serve such purposes during arousal. Indeed, there could be increases in the capacity of protein turnover, the electron transport system, lipid biosynthesis and iron metabolism in P. annectens after 1 day of arousal from 6 months of aestivation. The energy that supports these activities could be derived from increased amino acid (and perhaps also carbohydrate) catabolism during this period. The ammonia released through increased amino acid catabolism had to be detoxified to urea through the hepatic OUC. Therefore, it can be understood why there were significant increases in the urea-synthesizing capacity upon arousal from aestivation.
Besides being involved in urea synthesis, arginine produced by Ass also acts as a substrate for nitric oxide (NO) production in the liver, where NO is involved in liver regeneration [55] and protection of the liver from ischaemia-reperfusion injury [56]. Indeed, Chng et al [57] had shown that the arginine and NOx concentrations decreased and increased, respectively, in the liver of P. annectens after 6 months of aestivation and after 3 days of arousal from aestivation, supporting the proposition that arginine synthesized through Ass could be used for increased NO production, especially during arousal.
Arousal phase: up-regulation of carbohydrate metabolism?
Compared with the maintenance phase, 1 day of arousal led to increases in mRNA expressions of gapdh and aldob, and a decrease in the expression of another isoform of aldob. Although Gapdh does not catalyse a flux generating step (unlike hexokinase, glycogen phosphorylase, and pyruvate kinase) or act as a regulatory enzyme (unlike phosphofructokinase) in the glycolytic pathway, it involves an oxidation-reduction reaction, and our results could indicate a tendency towards an up-regulation of carbohydrate metabolism in the liver of P. annectens during the arousal phase of aestivation. Frick et al. [58] reported that P. dolloi conserved the glycogen pool during the maintenance phase of aestivation. Naturally, the fish becomes more active after arousal, and there could be an increase in the utilization of glycogen store for energy production during this period before feeding is resumed.
Arousal phase: up-regulation of genes involved in lipid metabolism and fatty acid transport Fatty acid binding proteins (FABPs) are intracellular carriers that transport fatty acids through cytoplasm, linking sites of fatty acid import/export (plasma membrane), internal storage (lipid droplets), and oxidation (mitochondria) [59]. Stearoyl-CoA desaturase is a lipogenic enzyme that catalyzes the synthesis of monounsaturated fatty acids [60]. Acyl-CoA desaturase is the terminal component of the liver microsomal stearoyl-CoA desaturase system that utilizes O 2 and electrons from reduced cytochrome b5 to catalyze the insertion of a double bond into a spectrum of fatty acyl-CoA substrates including palmitoyl-CoA and stearoyl-CoA. The up-regulation of mRNA expressions of fabps (4 clones), stearoyl-CoA desaturase (1 clone), desaturase (5 clones) and acyl-CoA desaturase (11 clones) ( Table 4) indicate that there could be an increase in fatty acid synthesis and lipid metabolism in the liver of P. annectens after 1 day of arousal. Tissue regeneration would be an important activity during arousal, and cell proliferation requires increased lipid metabolism to generate biomembranes. It is probable that the energy required to sustain these activities was derived from amino acid catabolism.
Arousal phase: up-regulation of electron transport system and ATP synthesis?
Conservation of energy is a key feature during the maintenance phase of aestivation to sustain life in adverse environmental condition. Arousal from aestivation marks an increase in the demand for ATP. Indeed, after 1 day of arousal, there were increases in mRNA expressions of ndufa2 (5 clones), cytochrome c oxidase subunit IV isoform 2 (2 clones) and two different types of ATP synthase (mitochondrial F o and F 1 complex; 2 clones each) ( Table 4), indicating that mitochondria became more active. It would be essential to maintain mitochondrial redox balance when activities of oxidation-reduction reactions increased in the mitochondrial matrix. The increase in mRNA expression of 3-hydroxybutyrate dehydrogenase type 1 (5 clones) suggested that mitochondrial activities might not be fully supported by an adequate supply of oxygen, and mitochondrial redox balance might have been maintained transiently through hydroxybutyrate formation during this initial phase of arousal.
Arousal phase: up-or down-regulation of iron metabolism and transport There could be two reasons for the increases in transferrin and ferritin expressions in the liver of P. annectens during arousal. Firstly, it could be a response to increased oxidative stress and inflammation. After arousal, the lungfish would immediately swim to the surface to breathe air. A rapid increase in O 2 metabolism would lead to increased generation of reactive oxygen species, as the rate of superoxide generation at the mitochondrial level is known to be correlated positively with oxygen tension [61,62]. Furthermore, animals experiencing transient metabolic depression followed by restoration of normal O 2 uptake also experience oxidative stress; examples consist of hibernating mammals, anoxia-tolerant turtles, freeze-tolerant frogs and molluscs [35,63,64]. Secondly, it could be due to an increase in the turnover of free and bound iron as a result of the increase in synthesis of certain type of hemoglobins and/or hemoglobin in general. Delaney et al. [65] reported that 4 electrophoretically distinct types of hemoglobins (fraction I, II, III and IV) were present in P. aethiopicus, and there were increases in the amounts of types II and IV hemoglobins during the maintenance phase of aestivation. Hence, it is logical to deduce that changes in hemoglobin types during the induction phase of aestivation must be reverted back to normal during arousal, which could be one of the reasons that led to the up-regulation in mRNA expressions of transferrin and ferritin in the liver of P. annectens.
Arousal phase: up-regulation of glutathione S-transferase (gst) GSTs are a major group of detoxification proteins involved in protecting against various reactive chemicals, including chemical carcinogens, secondary metabolites during oxidative stress, and chemotherapeutic agents [66]. They catalyze the reaction of glutathione with electrophilic centers of organic compounds [67]. These glutathione-conjugated compounds are rendered more water-soluble and more readily excreted. Besides, some GSTs have secondary catalytic activities including steroid isomerisation [68] and a selenium-independent peroxidase activity with organic hydroperoxides [69]. The alpha class GST (GSTa) may also function as intracellular transporters of various hydrophobic compounds (which are not substrates of GSTs) like bilirubin, heme, thyroid hormones, bile salts and steroids [70]. The increase in mRNA expression of gst in the liver of P. annectens after 1 day of arousal ( Table 4) is indicative of a possible increase in secondary metabolites of oxidative stress and/or transport of heme in the liver. Similarly, increases in activity of Gst have been observed in aestivating snails and snails aroused from aestivation [71].

Arousal phase: increase in protein turnover
Based on the variety of genes related to protein synthesis, transport and folding in the forward and reverse library, it can be concluded that there was a high rate of protein turnover in the liver of lungfish after 1 day of arousal. It would appear that the machinery (e.g. ribosomal protein L12, L17 and L19) involved in the maintenance of protein structure during the maintenance phase (Table 4) was different from that (e.g. eIF4E-binding protein, eukaryotic translation elongation factor alpha 1 and elongation factor-1, delta b) involved in the regeneration of protein structure during the arousal phase (Table 5).

Conclusion
Six months of aestivation led to changes in gene expression related to nitrogen metabolism, oxidative defense, blood coagulation, complement fixation, iron and copper metabolism, and protein synthesis in liver of P. annectens. These results indicate that sustaining a low rate of waste production and conservation of energy store were essential to the maintenance phase of aestivation. On the other hand, there were changes in gene expression related to nitrogen metabolism, lipid metabolism, fatty acid transport, electron transport system, and ATP synthesis in liver of P. annectens after 1 day of arousal from 6 months of aestivation. It would appear that the freshly aroused fish depended on internal energy store for repair and structural modification. Overall, our results indicate that aestivation cannot be regarded as the result of a general depression of metabolism only, but it involves the complex interplay between up-regulation and down-regulation of diverse cellular activities. Hence, efforts should be made in the future to identify and differentiate molecular, biochemical and physiological phenomena in African lungfishes incidental to each of the three phases (induction, maintenance and arousal) of aestivation.

Author Contributions
Conceived and designed the experiments: YKI SFC. Performed the experiments: KCH. Analyzed the data: KCH SFC YKI. Contributed reagents/materials/analysis tools: WPW. Wrote the paper: SFC KCH YKI. Took care of the animals: WPW.