Regulation of the Rana sylvatica brevinin-1SY antimicrobial peptide during development and in dorsal and ventral skin in response to freezing, anoxia and dehydration

The skin of frogs, like that of other animals, acts as an important physical and chemical barrier to microbes. Synthesized and stored within the dermal granular glands of the frog's skin, and secreted onto the skin following injury or sympathetic nervous system stimulation (Rollins-Smith and Conlon, 2005), antimicrobial peptides (AMPs) act as effector molecules of the innate immune system. AMPs are small (~10–50 amino acids), amphipathic, positively charged peptides that bind preferentially to microbial membranes (Rinaldi, 2002). Many, but not all, amphibian species studied to date produce many types of AMPs (Conlon, 2011) that can act alone or in synergy to inhibit a range of microbes (Barra and Simmaco, 1995; Schadich et al., 2010).

Although different genera of frogs express common families of AMPs, the peptides from any given species do not share identical amino acid sequences with other species (Conlon et al., 2004). The brevinin family of AMPs are found in Ranidae frogs and share only five invariant residues (Pro³, Ala⁹, Cys¹⁸, Lys²³, Cys²⁴). Unlike other ranids, the skin of the adult wood frog Rana sylvatica LeConte appears to produce a single AMP, the novel brevinin-1SY (Matutte et al., 2000). Amino acid sequencing of the active peptide from the frog skin extracts showed brevinin-1SY to be a 24-amino-acid peptide and characterization of synthesized brevinin-1SY peptides demonstrated antimicrobial activity against Escherichia coli Castellani and Chalmers and Staphylococcus aureus Rosenbach (Matutte et al., 2000). Incidentally, brevinin-1SY could only be isolated from R. sylvatica warmed to 37°C and not from R. sylvatica specimens retrieved from cold ponds (<7°C), suggesting environmental regulation of AMP expression (Matutte et al., 2000).

Rana sylvatica is one of just a few terrestrially hibernating amphibians that can survive whole-body freezing during the winter. As a consequence of freezing, R. sylvatica are also exposed to anoxia and dehydration of their cells. The onset of freezing initiates the synthesis of high concentrations of glucose from hepatic glycogen stores (Storey and Storey, 1992) and is prompted by ice nucleation on the skin. The skin is typically the first tissue to experience freezing and other environmental stresses, yet the consequential impacts on AMP gene and protein expression in this tissue have received very little attention. Given that R. sylvatica may be exposed to a myriad of conditions including anoxia, dehydration and freezing for weeks at a time, the regulation of their innate immune system, and thus their vulnerability to pathogens, during environmental stress was questioned. In the study presented here we investigated the mRNA levels of the brevinin-1SY gene in tadpoles, in adult R. sylvatica dorsal and ventral skin, and the impacts that anoxia, dehydration and freezing stresses have on brevinin-1SY mRNA and protein levels in adult frog skin. Furthermore, we examined whether the antimicrobial activity of extracts from the dorsal and ventral skin of R. sylvatica differed in animals that had undergone anoxia, dehydration or freezing. Lastly, in silico analysis was performed to predict tertiary structure and interaction of brevinin-1SY with a model membrane.

Using primers designed against consensus regions of brevinin-1 sequences from other ranids and the available brevinin-1SY

protein sequence (UniProt entry no. P82871), a 57-nucleotide amplicon encoding the partial sequence of brevinin-1SY was identified (Fig. 1A). The BLASTn search showed that the brevinin-1SY transcript shared the closest homology with the Rana palustris brevinin-1 PLc sequence (AM745088) with 80.7% homology to the corresponding sequence fragment. A nucleotide alignment of brevinin-1SY with other brevinin-1 family members highlights the differing nucleotides (Fig. 1A). The brevinin-1SY amplicon encoded 19 amino acids (Fig. 1B), covering 79% of the full length, 24 amino acid brevinin-1SY peptide (P82871; Fig. 1C). The missing amino acids are the first three at the N-terminus and the last two at the C-terminus (Fig. 1C). A single amino acid substitution occurred between the translated amplicon and the known peptide sequence previously identified (Matutte et al., 2000): an ATG encoding Met, where Ile¹⁶ should be, despite the reverse primer encoding the Ile at the correct position (Fig. 1C). A BLASTp of the brevinin-1SY peptide (P82871) showed that this peptide had the closest homology with the Rana catesbeiana ranatuerin-4 precursor (ACO51651) with 79.2% homology to the corresponding sequence fragment. Alignment of brevinin-1SY with brevinin-1 of other frog species revealed the conservation of the two cysteine residues important for disulfide bonding (Fig. 2). Phylogenetic analysis of brevinin-1SY with other brevinin-1 sequences confirmed the close relationship of brevinin-1SY with R. catesbeiana ranatuerin-4 precursor (ACO51651) as well as brevinin-1R from Pelophylax ridibundus (P86149) and brevinin-1Ea from Pelophylax esculentus (P40835; supplementary material Fig. S1).

A model of brevinin-1SY protein structure was generated using the QUARK server and the full brevinin-1SY amino acid sequence (P82871). The protein model was protonated and underwent energy minimization in molecular operating environment (MOE) using the AMBER force field and with water as a solvent (Fig. 3A). Brevinin-1SY was predicted to have an alpha helical secondary structure (Fig. 3A). The predicted secondary structure and solvent accessibility of brevinin-1SY were also generated using the QUARK server (Fig. 3B). In general, there was low solvent accessibility of all residues in brevinin-1SY (Fig. 3B). Using the EMBOSS pepwheel tool, brevinin-1SY was determined to have 66% hydrophobic residues, resulting in a hydrophobic and hydrophilic face of the peptide (Fig. 3C). Brevinin-1SY was predicted to have a charge of +3.

The brevinin-1SY mRNA was amplified at all stages tested during tadpole growth and metamorphosis. However, gene expression increased strongly in the later stages of development; expression levels significantly increased in Gosner stages 36–41 (by 2.87-fold), stages 42–43 (by 4.5-fold) and stages 44–45 (by 6.22-fold; Fig. 4A,B), compared with stages 14–20 (Gosner, 1960). These three Gosner stage brackets were also significantly different from each other, with stages 42–43 being 1.57-fold higher than stages 36–41, and Gosner stages 44–45 being 1.38-fold higher than stages 42–43 (Fig. 4).

Primers for brevinin-1SY were used in RT-PCR to assess the relative mRNA expression levels in wood frog tissues. Dorsal skin from control animals was found to have 9.78-fold (±0.66) higher levels of brevinin-1SY transcripts compared with ventral skin (1±0.13) from control animals (data not shown), and is in accordance with a greater proportion of dermal glands in the dorsal versus ventral skin (Gammill et al., 2012). Next, the relative mRNA levels in adult ventral skin, dorsal skin, lung, stomach, small intestine and large intestine tissues in control conditions and in response to freezing, dehydration and anoxia were assessed. The brevinin-1SY mRNA level of each tissue from control animals was set to a reference value of 1 and the mRNA levels in tissues from stressed animals expressed as a fold change. In response to 24 h freezing, brevinin-1SY mRNA expression levels decreased significantly in dorsal skin, to 0.6-fold of control values, and large intestine to 0.67-fold of control values (Fig. 5A,B). Anoxia for 24 h strongly increased brevinin-1SY mRNA transcript levels in ventral skin by 5.23-fold, and significantly decreased in the small intestine to 0.7-fold and in the large intestine to 0.69-fold compared with controls (Fig. 5B). During 40% dehydration, the expression of brevinin-1SY increased significantly in dorsal skin by 2.39-fold, in ventral skin by 3.29-fold and in lung by 1.57-fold (Fig. 5B). No changes were observed in the mRNA levels of brevinin-1SY in stomach tissue under any experimental stress (Fig. 5B).

Dorsal skin extracts from frozen, anoxic and dehydrated animals possessed higher levels of brevinin-1SY protein compared with that of dorsal skin extracts from control animals (Fig. 6A,B), although not statistically different (P=0.0502). The levels of brevinin-1SY in dorsal skin were variable amongst frogs within each group, but on average, dorsal skin from stressed animals appeared to have approximately fivefold higher levels of brevinin-1SY than that of control animals (Fig. 6B). Although we were able to obtain data on protein levels of brevinin-1SY in the dorsal skin by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), we were unable to obtain equivalent data from the ventral skin extracts, probably because of the low proportion of dermal glands containing AMPs in the ventral skin (Gammill et al., 2012). Protein identification by mass spectrometry (LC-MS/MS) confirmed that the band of interest from dorsal skin was indeed brevinin-1SY. Three unique peptide sequences were identified: FLPVVAGLAAK, LPVVAGLAAK and VLPSIICAVTK. These sequences provided 92% coverage of the brevinin-1SY peptide (P82871). Unlike that observed in the PCR amplicon, no amino acid substitutions were identified at the Ile¹⁶ position.

Dorsal skin extracts of R. sylvatica produced no significant inhibition of S. cerevisiae, B. cinerea or R. stolonifer growth (P>0.05; data not shown). However, various extracts from dorsal skin significantly inhibited the growth of E. coli, B. subtilis and P. sulcatum when compared with the blank (Fig. 7A). Indeed, extracts from stressed samples (anoxic, dehydrated and frozen) most strongly inhibited E. coli. The non-stressed control also significantly inhibited E. coli growth, albeit to a lesser extent. Extracts of all treatments equally inhibited the growth of B. subtilis when compared with the blank (Fig. 7A). P. sulcatum growth was significantly inhibited by the extracts from the dehydrated and frozen frogs, whereas extracts from the anoxic and non-stressed control were not significantly different when compared with the blank (Fig. 7A).

Despite lower mRNA and protein levels of brevinin-1SY in the ventral skin, antimicrobial activity of ventral skin extracts was observed. Ventral skin extracts of R. sylvatica did not inhibit the growth of S. cerevisiae, B. subtilis or R. stolonifer (P>0.05; data not shown). Conversely, the growth of E. coli, B. cinerea and P. sulcatum was inhibited by various ventral skin extracts (Fig. 7B). Frozen skin extracts showed the highest inhibition of E. coli growth. Skin extracts from the anoxic, dehydrated or the non-stressed control frogs also significantly inhibited E. coli growth when compared with the blank (Fig. 7B). For B. cinerea, skin extracts from anoxic and frozen animals provided the highest inhibition. The dehydration treatment also resulted in inhibition of B. cinerea, albeit to a lesser extent than anoxia. The non-stressed control skin extracts were not significantly inhibitory when compared with the blank (Fig. 7B). When assayed against P. sulcatum, only the frozen stress ventral skin extract provided significant inhibition of mycelial growth (Fig. 7B).

Interaction between brevinin-1SY and a cell membrane were modeled in an attempt to provide insight into how brevinin-1SY potentially binds to membranes and causes microbial growth inhibition. The model predicts the positively charged face of brevinin-1SY to interact with the extracellular surface of the membrane and insert into the membrane (Fig. 8).

Environmental impacts on amphibian disease are thought to contribute to unprecedented declines of amphibians around the world, yet the dynamic relationship between immunity and environment is not well understood (Bosch et al., 2007; Fisher, 2007). Environmental conditions can affect the microbial flora present, thus exposing frogs to altered types and patterns of pathogens (Castro et al., 2010; Sheik et al., 2011). Alternatively, environmental conditions, such as temperature, can influence the frog immune system (Raffel et al., 2006; Rohr and Raffel, 2010). The plasticity of R. sylvatica to a number of environmental stressors (freezing, anoxia, dehydration) provides a natural model to analyze the impacts of the environment on the immune system. Using R. sylvatica, this study examined the regulation of brevinin-1SY during development and in the skin of wood frog in response to anoxia, dehydration and freezing and presents evidence that brevinin-1SY is regulated in adult frogs in response to environmental stress.

Brevinin-1 family members show conservation and divergence in their sequence and structure, yet all act as important host defense molecules. Elucidation of the brevinin-1SY nucleotide sequence and alignment of the protein sequence to other available brevinin-1 sequences showed the five conserved residues characteristic of the brevinin-1 family. Although the core residues, overall alpha helical structure and amphipathic nature of the peptide are conserved, the protein alignment displays the divergence between the brevinin-1 family members found in ranids, the diversity of which may be linked to the pathogens each species is exposed to. In this study we observed a single amino acid substitution that occurred between the identified amplicon and the known brevinin-1SY peptide sequence (Matutte et al., 2000): an ATG encoding Met, where Ile¹⁶ should be. However, when mass spectrometry was performed on the brevinin-1SY bands, no corresponding amino acid substitution was detected. It is likely that the discrepancy in the mRNA and protein sequence could be an individual variation in brevinin-1SY, could possibly represent an additional brevinin-1SY or could simply be due to an error introduced by the Taq polymerase used for amplification. Currently, brevinin-1SY is the only described AMP of adult R. sylvatica (Matutte et al., 2000), in contrast to the multiple AMPs in other ranids. Recently, a possible temporin-1SY was identified in skin secretions of R. sylvatica metamorphs (Groner et al., 2013). However, it is not known whether this proposed temporin-1SY is expressed in adult R. sylvatica, and it was not observed in the previous study of skin from adult R. sylvatica (Matutte et al., 2000). Therefore, brevinin-1SY appears to be the only described AMP produced in the skin of adult R. sylvatica, and its regulation at a transcriptional or post-transcriptional level during development, would be paramount for maintaining an effective innate immune response to pathogens during environmental stress.

Metamorphosis is a key event in the development of the amphibian immune system (Rollins-Smith et al., 1997; Ohnuma et al., 2010) (reviewed in Rollins-Smith, 1998). During metamorphosis, R. sylvatica tadpoles must adapt from an aquatic to a mainly terrestrial environment, concomitant with a changing microbial environment. In general, previous studies have shown AMP mRNA expression to be undetectable before the onset of metamorphosis; in Xenopus laevis, AMP protein was observed approximately two stages after mRNA expression (Vanable, 1964; Reilly et al., 1994; Ohnuma et al., 2006). Although low levels of brevinin-1SY transcripts were found in all analyzed Gosner stages of R. sylvatica, strong increases in transcript levels occurred only at the end stages of metamorphosis (stages 36–45). The timing of increased brevinin-1SY mRNA levels are consistent with the emergence of AMP expression as seen in other anurans (Clark et al., 1994; Reilly et al., 1994; Ohnuma et al., 2006), probably induced by thyroid hormones (T₃) and corresponding to development of mature mucosal and dermal glands in the adult skin tissue (Ohmura and Wakahara, 1998). The basal levels of brevinin-1SY gene expression observed before adult skin growth, as reported in other anurans (Wabnitz et al., 1998; Ohnuma et al., 2010), may be associated with brevinin-1SY expression in the internal organs, as brevinin-1SY mRNA was detected in all tissues of adult R. sylvatica examined. The marked coordination between AMP levels and frog readiness to leave the aquatic environment suggests that AMPs are a requirement for dealing with terrestrial pathogens.

As adults, wood frogs are subjected to a range of environmental conditions each acting as a stress to which the frog must adapt. Low temperatures have been shown to impair various aspects of the immune system of frogs (Maniero and Carey, 1997; Raffel et al., 2006); however, little is known about the mechanisms that control the interaction between environmental stress and the innate immune system. Environmental temperature extremes can challenge viability, limit energy availability and trigger metabolic rate depression to support survival during these periods of bioenergetic constraint (Storey and Storey, 2004). Rana sylvatica undergoes whole body freezing as an adaptation to cold during the winter (Storey and Storey, 2004). Previous studies revealed that skin from wood frogs collected from cold ponds immediately after winter hibernation lacked detectable concentrations of brevinin-1SY in the skin (Matutte et al., 2000), suggesting environmental regulation of brevinin-1SY. Results in this study showed that brevinin-1SY mRNA levels in dorsal, but not ventral, skin decreased in comparison with the skin from non-stressed R. sylvatica following 24 h freezing. However, the relative protein levels of brevinin-1SY in the dorsal skin tended to be higher compared with controls, and the antimicrobial capacity of dorsal and ventral skin extracts from 24 h frozen animals showed enhanced antimicrobial activity towards certain microbes than that of the corresponding controls. Therefore, although transcription of brevinin-1SY may be suppressed during freezing as a result of energetic constraint, protein translation does not appear to be suppressed in the skin of the wood frog, at least during the short (24 h) freezing time frame examined in this study. In a natural setting where the wood frog overwinters and undergoes long bouts of freezing or multiple freeze-thaw cycles, the decrease in brevinin-1SY mRNA levels observed with freezing may eventually lead to a lack of brevinin-1SY protein in the skin following winter hibernation, as previously reported (Matutte et al., 2000).

Along with freezing, R. sylvatica experience cellular dehydration (Storey and Storey, 2004). The skin of amphibians is highly permeable, and regulates water uptake, salt balance (Greenwald, 1971; Shoemaker and Nagy, 1977) and gas exchange during normal and hypoxic conditions (West and Burggren, 1984; Boutilier et al., 1986; Pinder and Burggren, 1986). Owing to their highly permeable skin, amphibians have a high susceptibility to evaporative water loss. In the wood frog, up to 50–60% of total body water loss can be tolerated and is important during freezing when extracellular ice build-up and evaporative water loss cause extensive cellular dehydration (Churchill and Storey, 1993). In fact, freeze tolerance has been suggested to have arisen from a pre-existing mechanism used to defend against water stress (Churchill and Storey, 1994b; Churchill and Storey, 1995). In wood frogs, dehydration alone provokes a hyperglycemia response, comparable to the build-up of cryoprotectants during freezing (Churchill and Storey, 1993; Churchill and Storey, 1994a). Similarly, the build-up of urea to protect against colligative water loss is thought to be important in both dehydration and freezing (Costanzo and Lee, 2005; Muir et al., 2007). Parallel molecular responses have also been observed between wood frog dehydration and freezing, including reduced metabolism, activated liver glycogen phosphorylase and elevated PKAc and liver second messenger cAMP levels (Pinder et al., 1992; Churchill and Storey, 1994a; Holden and Storey, 1997). Despite these similarities, brevinin-1SY mRNA patterns differed between dehydration and freezing stresses in wood frogs. Whereas brevinin-1SY transcript levels were significantly decreased in dorsal skin during freezing, loss of 40% of total body water caused significant increases in brevinin-1SY mRNA levels in the ventral and dorsal skin. Furthermore, the increase in brevinin-1SY mRNA in the dorsal skin of dehydrated animals corresponds to an increased trend in brevinin-1SY protein levels and total antimicrobial activity. Although the transcriptional regulation of brevinin-1SY under freezing and dehydration conditions differs, the functional outcome of an increased trend in brevinin-1SY protein levels and enhanced total antimicrobial activity is similar, suggesting underlying commonalities in the response between freezing and dehydration.

Along with freezing and dehydration, anoxia of R. sylvatica cells occurs as the result of cessation of heart beat and blood circulation (Storey and Storey, 2004). Even in the face of high water loss, the skin remains important for cutaneous CO₂ excretion (Boutilier et al., 1979; Burggren and Vitalis, 2005). During anoxia, breathing rate has been shown to progressively decrease until it eventually ceases in Rana pipiens and as lymph movement is dependent on lung ventilation in frogs, the physiological responses to anoxia may inadvertently lower immunity (Winmill et al., 2005; Hedrick et al., 2007) as the lymphatics, as well as the blood, carry immune cells. With a decreased internal immune function, the protection of the skin by AMPs may become increasingly important. This might account for the rise in ventral and dorsal skin brevinin-1SY mRNA levels observed in R.sylvatica during 24 h anoxia. In conjunction, an increase in brevinin-1SY protein levels in the dorsal skin and total antimicrobial activity in the dorsal and ventral skin supports this hypothesis of enhancing the skin innate immune response. Alternatively, under hypoxic conditions hypoxia-inducible factor (HIF) is produced and stabilized, and has recently been shown to have a role in innate immune activation (Nizet and Johnson, 2009). In fact, HIF is implicated in acting to promote the NF-kB pathway (Nizet and Johnson, 2009), a pathway known for controlling the transcriptional activation of antimicrobial peptides in vertebrates (Simmaco et al., 1998).

In wood frogs, freezing, dehydration or anoxia result in metabolic rate depression and ultimately may trigger stress signals that activate the innate immune response through danger-associated molecular patterns (DAMPs). DAMPs have been shown to consist of a number of host proteins that signal host stress and result in immunoregulation (Denk et al., 2012; Hirsiger et al., 2012). In the wood frog, it appears that environmental stress (freezing, anoxia or dehydration) results in upregulation of brevinin-1SY protein levels, which might correspond to the overall enhanced antimicrobial activity of the dorsal skin extracts to a variety of pathogens (this study). Although the trigger(s) of brevinin-1SY production in the skin of R. sylvatica in response to environmental stress are unknown, it is possible that brevinin-1SY gene transcription could be upregulated in conjunction with skin wound healing mechanisms. Previous studies have shown that skin wound healing, such as regeneration of the integument following unnatural ecdysis due to dehydration, is associated with an induction of AMPs (Nakazawa et al., 2003; Bardan et al., 2004; Ohnuma et al., 2006). Additionally, the increased production of fibrinogen, an acute phase protein involved in wound healing, in the liver of R. sylvatica in response to freezing and dehydration (Cai and Storey, 1997) suggests that there is systemic activation of innate immunity possibly as a result of the production or release of DAMPs in response to environmental stress. Therefore, the production of brevinin-1SY in the skin may result from the systemic activation of the innate immune system.

In other animals, stress such as hypoxia results in glucocorticoid secretion and has been linked to immunosuppressive and anti-inflammatory effects (Sapolsky et al., 2000; Wang et al., 2012). Additionally, studies in frogs have shown that administration of glucocorticoids reduces de novo synthesis of AMPs (Simmaco et al., 1997). The increased synthesis of brevinin-1SY mRNA in response to environmental stress in the wood frog suggests that environmental stress influences glucocorticoid levels and lead to the upregulation of AMP mRNA transcripts. However, the levels of glucocorticoids in wood frogs during stress (freezing, anoxia or dehydration) have not been measured. Thus, we cannot directly link glucocorticoid levels (sustained or short-term levels) to increased brevinin-1SY mRNA levels in the skin of wood frog skin exposed to anoxia or dehydration. Further experiments are necessary to examine this hypothesis.

Once produced, AMPs act as direct antagonistic compounds to pathogens, but can also activate other facets of the immune system if microbes bypass AMPs and enter the host. AMPs can recruit additional immune cells, induce their activation and trigger cytokine production (reviewed in Haney and Hancock, 2013). Although it has not been shown in frogs that AMPs are capable of activating innate immune cells, frog AMPs have been shown to activate mammalian phagocytes (Chen et al., 2004) and it is likely that this dual function of AMPs (antimicrobial and pro-inflammatory) are conserved across vertebrates. It is interesting that brevinin-1SY protein production tends to be enhanced in the dorsal skin of animals from all environmental stresses tested, although not significantly different from that of control animals. Because frog AMPs may act to activate the immune response, future studies should examine whether other innate immune molecules in the skin of R. sylvatica are regulated in response to environmental stress to gain insight into the regulation of innate immunity at the level of the skin.

In conclusion, our study provides evidence that brevinin-1SY is regulated during development and in response to environmental stress. The regulation of brevinin-1SY in ventral and dorsal skin during environmental stress may have direct implications for antimicrobial activity of the skin towards a broad spectrum of pathogens and thus survival of wood frogs. Although freeze-responsive genes in wood frogs can also be categorized as responsive to either the anoxia (representing freeze-induced ischemia) or dehydration (representing freeze-induced cell volume reduction) component, the present results show that brevinin-1SY responds differently to each stress at a transcriptional level and leads to different functional outcomes in terms of antimicrobial activity. These data suggest that production of brevinin-1SY is an environmentally regulated, physiological change to promote whole animal survival. As AMPs act as barriers to pathogen entry as well as immune cell activation, the regulation of brevinin-1SY in wood frog skin would be crucial for host defense against pathogens.

Male wood frogs (5–9 g) and eggs were collected from spring breeding ponds in the Ottawa region. Tadpoles were raised in an aquarium and fed boiled endive fragments and goldfish food ad libitum daily. Their developmental stages were determined using the Gosner 46-stage developmental chronology (Gosner, 1960) based on observable physiological characteristics: Gosner stages 14–20 (tadpoles within eggs), 21–25 (free swimming tadpole), 26–30 (development of the back limb bud), 31–35 (extension of the limb bud into a jointed leg), 36–41 (toe differentiation), 42–43 (development of front legs) and 44–45 (adsorbed tail bud). Tadpoles were progressively moved to other containers to avoid crowding.

The Gram-negative bacterium Escherichia coli (strain DH5α; Invitrogen, Burlington, ON, Canada) and the Gram-positive bacterium Bacillus subtilis (Ehrenberg) Cohn (strain ATCC 23857; American Type Culture Collection, Manassas, VA, USA) were maintained on tryptic soy agar (TSA; Becton Dickinson, Sparks, MD, USA). The yeast Saccharomyces cerevisiae Meyen ex E.C. Hansen (strain BY4742; Invitrogen), the filamentous fungi Botrytis cinerea Pers. (Ascomycota) and Rhizopus stolonifer (Ehrenb.: Fr.) Vuill. (Zygomycota), and the oomycete Pythium sulcatum R.G. Pratt & J.E. Mitch were maintained on potato dextrose agar (PDA; Becton Dickinson). Botrytis cinerea and R. stolonifer were obtained from the Laboratoire de diagnostic en phytoprotection (MAPAQ, Québec, QC, Canada). Pythium sulcatum was isolated from an infected carrot root and is available at the Canadian Collection of Fungal Cultures (Agriculture and Agri-Food Canada, Ottawa, ON, Canada).

Nucleotide and protein sequences were obtained from NCBI (http://www.ncbi.nlm.nih.gov/) for use in analysis. Protein analysis was completed using the CLC Protein Workbench 6 software (CLC bio, Aarhus N, Denmark).

Input of the brevinin-1SY protein sequence into SWISS MODEL yielded no suitable templates for modeling of brevinin-1SY. Therefore, the amino acid sequence for R. sylvatica brevinin-1SY (P82871) was submitted to the QUARK server (Xu and Zhang, 2012) for protein prediction studies. The EMBOSS pepwheel tool (http://www.tcdb.org/progs/?tool=pepwheel) was used to examine amphipathicity of brevinin-1SY. The PPM web server (opm.phar.umich.edu/server.php) was used to model the potential interaction of brevinin-1SY with a membrane. The model of brevinin-1SY interaction with a membrane was visualized using Molecular Operating Environment (MOE; www.chemcomp.com/MOE-Molecular_Operating_Environment.htm).

All materials and solutions were treated with 0.1% v/v diethylpyrocarbonate (DEPC; BioShop, Burlington, ON, Canada) and autoclaved prior to use. Total RNA was extracted from tissues using TRIzol™ reagent (Invitrogen) according to the manufacturer's instructions. Briefly, tissue samples were homogenized in 1 ml TRIzol using a Polytron homogenizer. To each sample 200 μl chloroform was added, mixed and centrifuged at 12,000 g for 15 min at 4°C. The aqueous phase was collected, 500 μl of isopropanol were added, mixed and centrifuged at 5400 g for 15 min at 4°C to precipitate RNA. The supernatant was discarded and the RNA pellet washed with 1 ml of 70% ethanol followed by centrifugation at 5400 g for 5 min and the supernatant aspirated. The pellet was air-dried for 10 min before being resuspended in 20–50 μl DEPC-treated water. The RNA concentration for each sample was determined on a GeneQuant Pro spectrophotometer (Pharmacia, Markham, ON, Canada) at 260 nm. RNA purity was assessed using a ratio of absorbance at 260/280 nm, whereas RNA quality was examined by observing the integrity of 18S and 28S rRNA (rRNA) bands on native agarose gel electrophoresis with ethidium bromide staining. For studies using tadpoles, two tadpoles were used for each isolation to obtain sufficient RNA for use in cDNA synthesis, and four or five independent batches of total RNA were prepared from the tadpoles for each time point examined (N=4 or 5), and later converted into four or five independent batches of cDNA for use in RT-PCR. For experiments involving adult frog tissues, the tissue from one animal was used per RNA isolation and four or five independent total RNA isolations performed using individual tissues from four or five frogs (N=4 or 5) for each tissue and stress (e.g. four or five dorsal skins from each of control, frozen, anoxic and dehydrated frogs were used, for a total of 16–20 frogs). Dorsal skin samples were taken from the back of the frog by cutting a roughly oval shape of skin spanning from the shoulders to the hind legs. Similarly, ventral skin samples were taken from the underside of the frog and generally spanned from the shoulders to the hind legs. In this way, the majority of skin, both dorsal and ventral, could be obtained from the frog.

For first strand cDNA synthesis, 3 μg of total RNA were diluted using DEPC treated water to achieve a total volume of 10 μl and 1 μl of oligo(dT) (200 ng μl⁻¹; 5′-TTTTTTTTTTTTTTTTTTTTTV-3′; V=A or G or C; Sigma Genosys) was added to the sample. The samples were incubated at 65°C for 5 min in a thermocycler (Mastercycler Eppendorf) and then chilled on ice for 5 min. Four microliters of 5× first strand buffer (Invitrogen), 2 μl 100 mmol l⁻¹ dithiothreitol (Invitrogen), 1 μl 10 mmol l⁻¹ dNTPs (Bio Basic), and 1 μl Superscript II reverse transcriptase (Invitrogen) were added to each sample. Samples were incubated at 42°C for 60 min in an Eppendorf thermocycler (Mississauga, ON, Canada) and stored at 4°C until use.

Prior to expression studies, all cDNA samples were examined for potential genomic contamination using the α-tubulin primer set, designed to span an intron. In all cDNA samples, the presence of an ~1100 bp amplicon indicative of genomic contamination was not observed. For expression studies, to ensure that the amplified products had not reached saturation, initial studies tested serial dilutions of cDNA (10⁻¹ to 10⁻³) to identify the dilution (typically 10⁻²) that was non-saturating yet still showed visible bands for brevinin-1SY and α-tubulin for quantification purposes. The amino acid sequence of brevinin-1SY (Matutte et al., 2000) was converted into a hypothetical nucleotide sequence and then compared with the nucleotide sequences of other brevinin-1 genes to determine the codon variants most typical from other Ranidae frogs. Primers for brevinin-1SY were designed from the consensus amino acid sequence. The primer sequences were as follows: (1) brevinin-1SY forward 5′-GAGCCAGATGAABGGATGT-3′ (B=G/T/C), reverse 5′-TTTGGTTACTGCACAAATCA-3′; (2) α-tubulin forward 5′-AAGGAAGATGCTGCCAATAA-3′, reverse 5′-GGTCACATTTCACCATCTG-3′.

PCR was performed by combining 5 μl of a cDNA dilution directly with 20 μl of a prepared PCR master mix: 1.25 μl of primer mixture (0.03 nmol μl⁻¹), 15 μl of DEPC-treated water, 0.75 μl 10× PCR buffer (Invitrogen), 1.5 μl 50 mmol l⁻¹ MgCl₂, 0.5 μl 10 mmol l⁻¹ dNTPs and 1 μl of Taq polymerase (Invitrogen). Thermocycling conditions were 7 min at 95°C, 33–40 cycles of 94°C for 1 min, 62°C (brevinin-1SY) or 54°C (α-tubulin) for 1 min, and 72°C for 1.5 min, and a final elongation of 72°C for 10 min. Xylene blue loading dye (3 μl) was added to each PCR reaction and 12.5 μl of each sample were loaded into 2.5% (brevinin-1SY) or 1.0% (α-tubulin) agarose gels, which were electrophoresed in 1× TAE buffer (2 mol l⁻¹ Tris base, 1.1 ml acetic acid l⁻¹, 1 mmol l⁻¹ EDTA, pH 8.5) and the gels were then stained using ethidium bromide (0.01 g l⁻¹) to visualize nucleic acids. PCR-amplified products were visualized on the stained agarose gels using a Chemi-Genius BioImaging system (Syngene, Frederick, MD, USA) and the band densities quantified using the Gene Tools software. The intensity of the brevinin-1SY PCR products were normalized against the corresponding band intensities of α-tubulin amplified from the same cDNA sample to correct for any minor variations in sample loading. In the case of the tadpole data, the normalized mRNA levels in the Gosner stage 14–20 samples were set to a reference value of 1, and each of the other Gosner stages reported as a fold change compared with this reference group, to generate the relative mRNA transcript levels reported. For tissues from adult frogs, the normalized brevinin-1SY mRNA levels from the tissue of control frogs was set to a reference value of 1, to which the mRNA levels of brevinin-1SY in the frozen, anoxia and dehydrated groups were compared and expressed as a fold change from the corresponding reference group. Values are presented as means ± s.e.m. derived from N=4 or 5 different animals for each time point or treatment group.

The brevinin-1SY and α-tubulin PCR products were excised from the agarose gels and gel purified using the freeze/squeeze method before sequencing. Briefly, gel slices were frozen in liquid nitrogen for ~5 min, thawed, and transferred to 0.5 ml Eppendorf tubes that had been punctured in the bottom, plugged with glass wool, and fitted inside 1.5 ml tubes. Samples were centrifuged at 13,845 g for 5 min and eluents were subsequently transferred to new tubes, 0.1 vol of 3 mol l⁻¹ sodium acetate and 3 vol of 90% ethanol added, mixed and centrifuged at 13,845 g for 15 min. Supernatants were discarded and 0.5 ml 70% ethanol was used to wash the DNA pellet before allowing the pellet to air-dry. Pellets were dissolved in 40 μl of DEPC-treated water and sent for sequencing at Bio Basic (Markham, ON, Canada). Sequences were confirmed as encoding the correct genes by sequence comparison in BLASTN.

Skins from six different frogs for each treatment (N=6 for control, freezing, anoxia and dehydration treatments, a total of 24 different animals) were weighed and individually crushed using a mortar and pestle in the presence of liquid nitrogen. Samples were homogenized in glass homogenizers in a 1:10 w/v ratio of skin tissue to ethanol/0.1% trifluoroacetic acid (TFA) solution (i.e. 10 mg of tissue was homogenized in 100 μl ethanol/0.1% TFA solution). The resultant mixture was transferred to glass tubes fitted with caps. Ethanol/0.1% TFA was used as the blank. Homogenates were mixed for 3 h on an orbital shaker at 4°C and 120 rpm. Following mixing, samples were centrifuged at 5300 g for 20 min at 4°C and the supernatants transferred to fresh tubes. The supernatants were evaporated using a nitrogen drier. Pellets were re-suspended in a 1:5 initial sample w/v of sterile ultra pure water (>18 MΩ cm; i.e. if the initial tissue weight was 10 mg, then the pellet was resuspended in 50 μl ultra pure water). This method of normalization was chosen to provide a representation of what was occurring at the whole animal level. Prior to analysis, samples were centrifuged in a microcentrifuge at 17,000 g for 1 min and the aqueous supernatants were transferred to fresh tubes. Samples were prepared in a laminar flood hood to prevent contamination, and immediately used for antimicrobial assays as described below.

Following antimicrobial assays, only sufficient quantities of dorsal or ventral skin extracts from three frogs for each treatment were available for protein quantification. For gel electrophoresis, a 1:1 (v/v) ratio of 2× SDS-PAGE sample buffer (dorsal skin) or a 1:5 (v/v) ratio of 6× SDS-PAGE sample buffer (ventral skin) was added to a small aliquot of the samples, mixed and boiled for 5 min. Equal volumes of peptide extracts in SDS-PAGE loading buffer were separated on 15% Tris-tricine gels by electrophoresis for 30 min at 30 V followed by 50–55 min at 150 V. Dorsal and ventral skin peptide extracts were separated on 1 mm and 1.5 mm gels, respectively. Silver staining of gels was performed at room temperature with rocking as follows: overnight in fixative (50% v/v ethanol, 12% v/v acetic acid, 0.05% formalin); 20 min in 20% ethanol; 2 min in sensitizing solution (0.02% w/v sodium thiosulphate); 2× 1 min in ddH₂O; 20 min in silver stain solution (0.6% w/v silver nitrate, 0.076% v/v formalin); 2× 1 min in double deionized water (ddH₂O); 5–15 min in developing solution (6% w/v sodium carbonate, 0.04% w/v sodium thiosulphate, 0.05% v/v formalin). The reaction was stopped with 12% acetic acid. Gels were visualized with the ChemiGenius Bio-Imaging system using GeneSnap software, and densitometry analysis was performed using GeneTools software (Syngene). To normalize data for slight variations in protein loading, the density of the brevinin-1SY band was normalized against the intensity of all other bands in the corresponding gel lane. The mean ratio of brevinin-1SY protein to total protein for the skin extracts of control animals was used as a reference and set to an reference value of 1.

Following gel electrophoresis and silver staining, the band in the 2–3 kDa range believed to be brevinin-1SY was excised for use in protein identification. Silver-stained gel slices, in 1.5 ml Eppendorf tubes, were destained by the addition of equal volumes of a 4 g l⁻¹ potassium ferricyanide solution and a 0.4 g l⁻¹ sodium thiosulphate solution. Tubes were agitated intermittently to facilitate destaining of the gel slice. Once destaining was complete, the destaining solution was removed and a solution of 100 mmol l⁻¹ ammonium chloride, followed by ddH₂O was added to remove the yellow residual stain from the gel prior to in gel-tryptic digestion and protein identification by liquid chromatography–tandem mass spectrometry (LC-MS/MS) at the Proteomics Platform of the Québec Genomics Center at the University of Laval (Québec, Canada).

E coli and B. subtilis cells were individually transferred to tryptic soy broth (TSB; Becton Dickinson) and S. cerevisiae was transferred to potato dextrose broth (PDB; Becton Dickinson) using a sterile inoculation loop. Cells were cultured overnight at 28°C and adjusted to 10⁸ cells per ml for the two bacteria, and 10⁷ cells per ml for S. cerevisiae using a hemocytometer. Each cell suspension (100 μl each) was spread on Petri dishes containing TSA (for E. coli and B. subtilis) or PDA (for S. cerevisiae). Sterile paper disks (0.6 mm diameter) were transferred to the inoculated media. Each disk contained 25 μl of dorsal or ventral skin extracts from one of the treatments (blank, non-stressed control, anoxia, dehydration, freezing). For filamentous microorganisms (B. cinerea, R. stolonifer and P. sulcatum), a 0.5 cm agar plug containing actively growing mycelium was placed in the center of a PDA dish. Disks containing 25 μl extracts were transferred to each dish at a distance of 3 cm from the mycelial plug. All dishes were incubated at 23°C in the dark for 72 h. Following incubation, inhibition zones between the microbial growth and the extracts were measured (referred to as the inhibition radius). The experiment was conducted as a randomized complete block design with six repetitions using skin extracts generated from six separate frogs.

Statistical testing of normalized band intensities for RT-PCR data used one-way ANOVA and a post hoc test (Student–Newman–Keuls). Protein data were analyzed using a Kruskal–Wallis test. For antimicrobial assays, an ANOVA was performed. When significant (P<0.05), means were separated according for Fisher's protected least significant difference (LSD) test (α-level=0.05).