Cope's gray tree frog Hyla chrysoscelis accumulates glycerol during cold acclimation. We hypothesized that, during this process, gray tree frogs adjust renal filtration and/or reabsorption rates to retain accumulated glycerol. During cold acclimation, plasma concentrations of glycerol rose >200-fold, to 51 mmol/l. Although fractional water reabsorption decreased, glomerular filtration rate (GFR) and, consequently, urine flow were <5% of warm levels, and fractional glycerol reabsorption increased. In contrast, dehydrated frogs increased fractional water reabsorption, decreased GFR, and did not accumulate glycerol. We hypothesized that expression of proteins from the aquaporin (AQP)/glyceroporin (GLP) family was associated with changing patterns of water and glycerol movement. We cloned the cDNA for three such proteins, quantified mRNA expression in nine tissues using real-time quantitative PCR, and functionally characterized them using a Xenopus oocyte expression system. HC-1, an AQP1-like water channel conferring low glycerol permeability, is expressed ubiquitously in warm- and cold-acclimated tissues. HC-2, a water channel most similar to AQP2, is primarily expressed in organs of osmoregulation. HC-3, which is most similar to AQP3, is functionally characterized as a GLP, with low permeability to water but high permeability to glycerol. Aspects of expression levels and functional characteristics varied between cold and warm conditions for each of the three AQPs, suggesting a complex pattern of involvement in osmoregulation related to thermal acclimation.
- Xenopus oocyte
- glomerular filtration rate
some organisms inhabiting regions with subfreezing temperatures are intolerant of freezing and avoid ice formation by mechanisms such as supercooling. Others, such as Cope's gray tree frog Hyla chrysoscelis, tolerate actual freezing and implement mechanisms that minimize damage from the formation of ice crystals. Among these mechanisms is the accumulation of solutes that may serve a variety of functions, including cryoprotection (to stabilize protein and/or membrane structure and function) and osmotic regulation (to promote distribution of water between intracellular and extracellular fluids) (34). On freezing, a number of amphibian species liberate glucose to accomplish these physiological objectives. However, frogs of the gray tree frog complex, H. chrysoscelis and its tetraploid sister species Hyla versicolor, are unusual, in that they also accumulate glycerol during cold acclimation before freezing (21). It is likely that this glycerol is synthesized by the liver and eventually released to the circulation and distributed to tissues throughout the body. Glycerol may circulate at elevated concentration (and accumulate at high concentrations in tissues) in cold-acclimated organisms for weeks or months, as long as the animals remain cold (12). Although the function of this glycerol has not been definitively established and other frogs that tolerate freezing do so without glycerol, the presumption is that this solute acts as a cryoprotectant, as described above.
As a small solute, glycerol should be freely filtered in the renal glomeruli. How, then, do the frogs prevent loss of this solute? Two possibilities, not mutually exclusive, arise: 1) the frogs could reduce glomerular filtration and, thereby, minimize filtration of glycerol, or 2) they could enhance its subsequent reabsorption.
Thus it is likely that glycerol transport across cell membranes, whether to exit hepatocytes, to enter other cells as a protective solute, or to be reabsorbed after glomerular filtration, is an important physiological demand during cold acclimation in this group of frogs. At the same time, pathways for water flux must be maintained for water balance during cold acclimation (e.g., renal water reabsorption or, potentially, water redistribution) and for the eventuality of freezing (which likely entails shifting distribution of water between fluid compartments and may occur too quickly for upregulation of water transport pathways). Glycerol and water transport can be accomplished via proteins from the membrane integral protein (MIP) family (37). Some of these proteins [aquaporins (AQPs)] function as selective water channels; others [glyceroporins (GLPs)] additionally transport small organic solutes, such as glycerol (37). Several members of the MIP family have been identified in amphibians (1, 2, 9, 24).
Thus we hypothesized that, during cold acclimation, in anticipation of freezing, gray tree frogs accumulate glycerol and, to retain that solute, adjust rates of renal filtration and/or reabsorption. Moreover, we hypothesized that tissues from gray tree frogs would express AQPs and GLPs and that the pattern of their expression among tissues would relate to roles in water and glycerol transport. To test these hypotheses and predictions, we 1) evaluated renal function in H. chrysoscelis that were acclimated to warm and cold temperatures and 2) identified and characterized AQP/GLPs from gray tree frogs and described patterns of tissue expression for mRNA encoding those proteins.
MATERIALS AND METHODS
Animal Collection and Maintenance
Male H. chrysoscelis were collected from ponds in southwestern Ohio (Butler and Greene Counties). Species identification was based on trill frequency, which differs between this species and H. versicolor. Animals were held in outdoor enclosures through the summer, during which time they were fed crickets several times per week. In the fall, as ambient temperature and day length declined, frogs ceased eating and then were transferred to constant-temperature rooms at Wright State University. Warm-acclimated frogs were returned to a warmer temperature (23°C) and an 8:16-h light-dark cycle. They received food and water ad libitum and soon resumed feeding. Cold-acclimated frogs were moved to covered containers with moistened sphagnum moss as bedding. The temperature was adjusted from 8°C for 4 wk to 5°C for 8 wk and, finally, to 2°C for ≥2 wk. In addition, we tested one additional group for the effects of dehydration: we emptied the bladders of warm-acclimated frogs and then denied access to water until loss of 20% of standard body mass, which required ∼48 h at 23°C. We measured kidney function in six animals under each physiological condition. All housing procedures and experimental protocols for the care and use of H. chrysoscelis were approved by the Institutional Animal Care and Use Committee at Wright State University.
Glomerular filtration rate (GFR) was measured from the clearance of a single injection of [14C]inulin (32). Frogs were injected intraperitoneally with 90 μl of distilled water containing 2.7 μCi of [14C]inulin (Dupont/New England Nuclear). Blood was subsequently sampled by puncture of an inguinal blood vessel and collected into heparinized capillary tubes. Serial blood samples in preliminary trials indicated that the distribution of inulin in the extracellular fluid, as indicated by the time to reach peak concentration of inulin in the plasma, required up to 2 h in warm-acclimated frogs and up to 12 h in cold-acclimated or dehydrated animals. Thus, although this method is based on a single injection, the relatively slow inulin clearance in ectotherms allows application of standard clearance calculations (32, 46). Kidney function was evaluated in initial blood samples obtained at those equilibration times (2 h after injection in warm hydrated animals and 12 h after injection in others) and two subsequent samples obtained at 2-h (for warm hydrated frogs) to 6-h (for cold and dehydrated frogs) intervals. GFR was calculated from the inulin concentrations in the second and third samples. This protocol ensured that GFR was calculated from samples obtained after the point of full distribution in the extracellular fluid, when inulin concentration falls as a monoexponential curve (7), and an exponential decline in plasma inulin was assumed in the calculations (see below). Blood was centrifuged (5,000 g for 3 min), and hematocrit was recorded for all samples. Erythrocytes were discarded, and plasma inulin concentration was measured with a liquid scintillation analyzer (Tri-Carb 2300 TR). For quantitative collection of bladder urine, at the time of each blood sample collection, a glass tube was inserted into the cloaca, and the urine was drained into preweighed microcentrifuge tubes. We used the urine collected at the time of the third blood sample (i.e., urine that accumulated between the 2nd and 3rd blood samples) to calculate the GFR.
GFR was calculated on the basis of inulin concentrations in two plasma samples and the intervening urine sample as follows: GFR (ml/h) = (UFR × V)/P, where UFR is urine flow rate (ml/h) and V is urine inulin concentration (cpm/μl); P, the average plasma inulin concentration for the clearance interval, was calculated assuming exponential loss from the plasma (32).
Accumulation and Excretion of Glucose and Glycerol
Plasma and urine remaining after inulin analysis were frozen at −80°C. Glucose and glycerol concentrations in the urine (5 μl) and blood (5 μl) were subsequently assayed spectrophotometrically (Spectronic 1001 Plus, Milton Roy) with assay kits (GAHK-20 and F6428, respectively, Sigma) (4). Fractional reabsorption (FR) was determined for glucose and glycerol as follows: FR = [(amount filtered − amount excreted)/amount filtered] × 100, where amount filtered was calculated as GFR × plasma concentration of solute and amount excreted was calculated as UFR × concentration of solute in bladder urine.
In each animal, we acquired three blood samples and three urine samples (initial bladder drainage and samples from the intervals between blood samples). We measured the inulin concentration in all plasma and urine samples. If sample volume permitted, we also measured glucose and glycerol concentrations; however, in some smaller samples, we could measure one but not the other. Analysis of data for animals from which we acquired repeated measures indicated that glucose and glycerol concentrations did not change significantly over those intervals in plasma or urine. We therefore based our clearance calculations on glucose and glycerol concentrations from the second or third plasma or urine sample if only single measures were available or from the average of those two if concentrations were measured on both samples.
In some cold- and warm-acclimated animals, we measured the mass of subsamples of liver and thigh skeletal muscle. Tissue samples (0.15–0.25 g) were dried to constant mass at 80°C and reweighed for determination of water content. Glycerol content of dried tissues was measured according to methods described by Irwin and Lee (12). Briefly, tissues were homogenized in perchloric acid and neutralized with KHCO3, and glycerol was assayed using the glycerol phosphate oxidate procedure (kit F6428, Sigma).
Identification of AQPs
RNA extraction and cDNA synthesis.
Tissues were quickly removed from animals, placed in liquid nitrogen, and stored at −80°C until RNA was isolated. Total RNA was prepared from warm-acclimated (n = 3) and cold-acclimated (n = 2) H. chrysoscelis brain, liver, lung, ventral abdominal skin, kidney, fat body, muscle, bladder, and small intestine (just distal to the stomach, ∼2 cm long) according to the manufacturer's protocol (Tri-Reagent, Molecular Research Center, Cincinnati, OH) and as previously described (16). All tissues yielded RNA of high purity, as determined by the ratio of optical density at 260 nm to optical density at 280 nm of ∼1.8, and high quality, as determined visually by the presence of three intact rRNA bands size fractionated on denaturing agarose gels (16; see supplemental figures in the online version of this article for a more detailed description of RNA isolation methods). For the initial cloning step, RT (Superscript III RT, Invitrogen, Carlsbad, CA) was used to generate oligo(dT)-primed first-strand cDNA from 5 μg of total RNA in a 20-μl reaction according to the manufacturer's directions (Superscript III First-Strand cDNA Synthesis kit, Invitrogen). A −RT was performed for each RNA preparation as a control for DNA contamination of the RNA preparation (not shown). No amplification products were observed when any of the −RT controls were used as templates in subsequent PCRs.
Degenerate PCR for isolation of novel AQP/GLP.
The nucleotide sequence of degenerate primers designed to amplify AQP and GLP homologs from H. chrysoscelis was determined on the basis of the conservation of amino acid identity between previously reported AQP and GLP sequences from other amphibians, including a vasopressin-regulated water channel (GenBank accession no. AAC69694.1) and three others (accession nos. AAC69695.1, AAA67782, and AAC69696.1) from Bufo marinus and water channel proteins from Rana pipiens (accession no. Q06019), Xenopus laevis (accession no. JN0557), and Rana esculenta (accession no. P50501). First-strand cDNA from kidney and bladder from warm-acclimated frogs and from liver of cold-acclimated frogs was PCR amplified using degenerate oligonucleotide primers (Table 1) designed to flank the first NPA motif (…GHVSGAH. …) (forward primer AQP-F1) and the second NPA motif [. …(g)NPARSF(g). …] (reverse primer AQP-R1) conserved in AQP family members. Degenerate oligonucleotide primers were also designed to flank the first NPA motif […VSGAHLN(c). …] (forward primer GLP-F) and the second NPA motif (…NPARDFG. …) (reverse primer GLP-R) conserved in GLP family members. Amplification reactions consisted of 10 pmol of each primer, 0.25 mM each dNTP, 1.5 mM MgCl2, 60 mM Tris·HCl (pH 9.0), 12.5 mM (NH4)2SO4, 0.1 U of Taq DNA polymerase (Invitrogen), and 0.5 μl of first-strand cDNA in a total reaction volume of 20 μl. PCR was performed in an Eppendorf Mastercycler Gradient Thermocycler. PCR products obtained using AQP-F1 and AQP-R1 primers were then subjected to a second PCR using a nested forward primer (AQP-F2; Table 1) and the AQP-R1 reverse primer. −RT and no-template controls were included. PCR products were size fractionated by electrophoresis on a 2% agarose gel in 1× Tris-borate-EDTA at 100 V for 2 h, stained with ethidium bromide, and visualized under UV light. The expected size of the PCR product is ∼380 bp.
PCR-amplified partial cDNAs products were ligated into the plasmid cloning vector pGEM-T-Easy according to the manufacturer's instruction (Promega, Madison, WI), transformed into Escherichia coli strain JM109, and plated on LB-ampicillin agar in the presence of 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside and isopropyl-1-β,d-thiogalactopyroside (Sigma) for blue-white color selection for the presence of an insert. Plasmid DNA was prepared using Miniprep spin columns (Qiagen, Valencia, CA) and restriction digested with EcoRI to excise the insert from the vector to confirm the presence of an ∼380-bp insert.
Plasmid DNA was sequenced using fluorescent dye chemistry and capillary electrophoretic sequencing at the University of Cincinnati DNA Core Facility (Cincinnati, OH). Sequences obtained were compared with those published in GenBank (National Center for Biotechnology Information) using nucleotide (BLASTN) and amino acid (TBLASTX) comparison. Three distinct partial AQP (HC-1 and HC-2) and GLP (HC-3) cDNAs were identified by sequence analysis. HC-1 was initially amplified from warm kidney, HC-2 from warm bladder, and HC-3 from cold liver.
Isolation of full-length cDNA.
The partial cDNA sequences obtained for HC-1, HC-2, and HC-3 were used to design gene-specific oligonucleotide primers (Table 1) for the rapid amplification of 5′ and 3′ cDNA ends (RACE). PCRs using 5′- and 3′-RACE gene-specific primers for all three partial cDNAs were used to obtain the 5′ and 3′ ends of each cDNA sequence from warm kidney (HC-1), warm bladder (HC-2), and cold liver (HC-3) as directed by the manufacturer (5′-RACE and 3′-RACE kits, Invitrogen).
Full-length cDNA and predicted amino acid sequence for HC-1, HC-2, and HC-3 were submitted to GenBank with the following accession numbers: DQ364243, DQ364244, and DQ364245, respectively. Nucleotide and predicted amino acid sequences were compared with previously published sequences available through GenBank with use of BLASTN and TBLASTX, respectively, to determine percent identity and percent similarities. Discontinuous MegaBlast was used to determine nucleotide similarity with HC-2 because of high divergence in the nucleotide sequence. Multiple sequence alignments with similar sequences were generated using the global sequence alignment program feature of CLUSTALW available through the European Molecular Biology Laboratory-European Bioinformatics Institute (http://www.ebi.ac.uk/clustalw/).
Real-time quantitative PCR.
To examine the relative expression levels of each mRNA transcript in warm- and cold-acclimated tree frog tissues, we used primers for HC-1, HC-2, HC-3, and β-actin in real-time quantitative PCR. Custom oligonucleotide primer pairs specific for HC-1, HC-2, and HC-3 (Table 1) were designed using the Primer Pick-3 Program available through the Whitehead Institute for Biomedical Research website (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi) and synthesized (Integrated DNA Technologies, Coralville, IA). Oligonucleotide primers (Table 1) specific for the housekeeping gene β-actin were designed on the basis of sequence homology between X. laevis cytoplasmic β-actin mRNA (GenBank accession no. AF079161), Hyla japonica β-actin 1 (GenBank accession no. AB092519), and H. japonica β-actin 2 (GenBank accession no. AB092520). HC-1 (206 bp), HC-2 (177 bp), HC-3 (149 bp), and β-actin (119 bp) were amplified from first-strand cDNA via end-point PCR (as described above). PCR products were TA subcloned into pGEM-T-Easy, and sequence was confirmed. Real-time PCR conditions were optimized using a standard dilution curve series (102–108 copies) of the sequence-confirmed plasmid DNA as templates in eight separate sets of reactions for each primer set, each with a negative (no-template) control. Each reaction consisted of 25 μl of iQ SYBR Green Supermix (Bio-Rad, Hercules, CA), 10 pmol of each primer, 1 μl of diluted plasmid DNA, and water up to 50-μl final reaction volume. For each primer set, each set of serial dilutions was subjected to a different annealing temperature (53–63°C) using the gradient feature of the Bio-Rad iCycler iQ. The cycle at threshold (CT values) for each dilution in a series at a specific annealing temperature was then plotted vs. dilution using the iCycler iQ standard curve software. Optimal PCR conditions, chosen from analysis of standard curves, were defined as those yielding PCR efficiency of 95–102% and r2 = 0.99–1.0 (14). Melt curve analysis performed on each primer set showed a single peak at 88–93°C (depending on the primer set), indicating specificity of the reaction (i.e., no nonspecific priming and no primer dimers were present).
Real-time quantitative PCR was applied to RNA from warm- and cold-acclimated tree frog brain, liver, lung, skin, kidney, fat body, muscle, bladder, and gut. We used 2.5 μg of DNase-treated RNA from each tissue as a template for cDNA synthesis (+RT) and 2.5 μg as a negative (−RT) control. First-strand cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad) according to the manufacturer's protocol. HC-1, HC-2, HC-3, and β-actin tissue expression was assessed using the PCR conditions described above by substitution of 1 μl of cDNA (+RT) or 1 μl from the −RT control as the template. No amplification in the −RT controls was observed in any tissue with any primer set. The relative expression level of each transcript was determined using the CT value plotted against the standard dilution curve for each reaction. Each set of reactions was replicated. Relative HC-1, HC-2, and HC-3 expression for each warm- and cold-acclimated tree frog tissue was normalized to β-actin expression. β-Actin expression did not vary between tissues or between warm and cold conditions within the same tissue, inasmuch as similar CT values were obtained with each cDNA synthesized from 2.5 μg of RNA.
Generation of HC Xenopus oocyte expression constructs.
Functional characteristics of the AQP/GLP molecules were evaluated by expression in oocytes from X. laevis. Custom oligonucleotide primer pairs flanking the full coding sequence (cds) of each AQP (HC-1 cds F and R, HC-2 cds F and R, and HC-3 cds F and R; Table 1) were designed using the Primer Pick-3 program. HC-1, HC-2, and HC-3 full-length coding sequences were PCR amplified in an Eppendorf Mastercycler Gradient Thermocycler using 1 μl of first-strand kidney cDNA in a total reaction volume of 20 μl. PCR products were TA subcloned into the pGEM-T-Easy vector (Promega). Clones with inserts in the forward transcriptional orientation from the T7 promoter were digested with AatII and SacII to remove a 28-bp fragment upstream of the ATG start site in each vector. A Xenopus β-globin 5′-untranslated region (UTR) was PCR amplified from a Xenopus oocyte expression vector containing human AQP1 (hAQP1; American Type Culture Collection, Rockville, MD) using a forward primer flanked on the 5′ end with a manufactured AatII site (Xenopus β-globin F) and a reverse primer with a manufactured SacII site at the 5′ end (Xenopus β-globin R; Table 1). The Xenopus β-globin 5′-UTR was subcloned into the pGEM-T-Easy vector and, subsequently, digested with AatII and SacII to excise a 95-bp fragment from the vector. This fragment was ligated into the AatII/SacII sites of the HC clones, upstream of the ATG start site in each HC sequence. The DNA sequence for each construct was confirmed by direct sequencing of plasmid DNA (University of Cincinnati DNA Core Facility). Linearized DNA from each HC plasmid was then used as a template for in vitro transcription in the presence of a 7-methylguanosine cap using T7 polymerase (mMESSAGE mMACHINE, Ambion, Austin TX). The hAQP1 Xenopus oocyte expression vector was used as a positive control. The hAQP1 vector was linearized with SmaI and transcribed in vitro using T3 polymerase (mMESSAGE mMACHINE). The transcripts were purified using MEGA CLEAR columns (Ambion) and eluted in RNase-free water. The cRNA concentration was determined by spectrophotometry. After agarose gel electrophoresis, 1 μg of the RNA was stained with ethidium bromide and visualized under UV light. The presence of a single band (1,105, 1,119, and 1,147 nt for HC-1, HC-2, and HC-3, respectively) was used to confirm the synthesis of full-length cRNA for each clone.
Expression of HC-1, HC-2, and HC-3 in Xenopus oocytes and measurement of water permeability.
Stage V and VI X. laevis oocytes were prepared as described previously (45). Oocytes were injected with 50 ng of HC-1, HC-2, HC-3, or hAQP1 cRNA or 50 nl of water (negative control) and incubated at 18°C for 72 h in modified Barth's buffer. Oocytes were transferred to a cell culture dish containing 67 mosmol/kgH2O diluted Barth's buffer at 23°C. Oocyte size was measured every 30 s until the oocytes burst or for up to 5 min. To determine mercury sensitivity, oocytes were preincubated in 0.3 mM HgCl2 for 5 min before hypotonic challenge. To determine the effect of cold temperature on water permeability, oocytes were preincubated at 10°C for 10 min before hypotonic challenge, which was also carried out at 10°C. Oocyte areas were measured on a Nikon Eclipse TE 2000-S microscope with a Cool Snap ES camera and Metamorph version 6.1r4 software. ImageJ 1.34s software was used to convert area to volume, and osmotic water permeability (Pf) was determined as described previously (45).
Glycerol permeability of HC-1, HC-2, and HC-3 in Xenopus oocyte expression system.
HC-1, HC-2, HC-3, and hAQP1 cRNAs and water (negative control) were injected into oocytes, which were incubated for 72 h as described above. After the 72-h incubation, oocytes were incubated for 3 min at room temperature in 200 mosM modified Barth's buffer containing 1 μCi/ml [14C]glycerol and 1 mM unlabeled glycerol. Oocytes were washed four times with ice-cold modified Barth's buffer (200 mosM) containing 10 mM unlabeled glycerol and then placed in 200 μl of 10% SDS and incubated for 24 h at room temperature. Scintillation fluid was added to the lysis solution, and 14C activity of individual oocytes was measured with a liquid scintillation analyzer (model 1900TR, Packard, Meriden, CT). Mercury-sensitive glycerol uptake was determined by preincubation of the oocytes in 0.3 mM HgCl2 for 5 min before glycerol addition. Oocytes with and without mercury were tested at 23°C and 10°C. Relative uptake of [14C]glycerol was calculated as described previously (24).
Values are means ± SE. Differences between mean values of physiological functions were tested for statistical significance using Statistica for Windows. ANOVA was followed by a Newman-Keuls test for pair-wise comparisons. The threshold of significance was set at α < 0.05. Pair-wise comparisons of Pf values between each HC clone and controls were evaluated using two-tailed Student's t-tests with equal variance. A two-factor univariate ANOVA was used to evaluate the influences of temperature and mercury and their interaction on Pf and glycerol uptake for each HC clone. In each case, P < 0.05 was considered significant.
Kidney Function and Glycerol Accumulation
Warm-acclimated (control) condition.
Standard mass of hydrated, warm-acclimated tree frogs was not significantly different from that of cold-acclimated animals (7.9 ± 0.6 and 7.3 ± 0.7 g, respectively), but standard mass of dehydrated frogs decreased (6.1 ± 0.4 g at initiation of kidney function measurement). Hematocrit averaged 0.32 in these animals. Glycerol and glucose concentrations were 0.18 and 0.78 mmol/l, respectively, in plasma (Table 2) and 0.17 and 0.14 mmol/l, respectively, in urine.
The GFR of warm, hydrated tree frogs was 226 ± 107 μl/h (Table 3). UFR was 85 ± 37 μl/h (Table 3), indicating a fractional water reabsorption of 62%. When these values of kidney function and glycerol concentrations were combined, the fractional reabsorption of glycerol was 64%, leaving a net urinary glycerol excretion rate of 15 nmol/h.
In warm-acclimated frogs, glycerol could not be detected in liver or muscle. Tissue water content was 67 ± 1% (n = 4) in liver and 73 ± 1% in muscle (n = 4).
Cold-acclimated gray tree frogs accumulated significant glycerol: the circulating concentration averaged 51 mmol/l. The circulating concentration of glucose rose to a lesser extent (2.9 mmol/l). GFR in cold-acclimated animals averaged 6 ± 0.4 μl/h, <3% of the value in warm-acclimated frogs and one-sixth that in dehydrated frogs (Table 3). Similar to the rate of filtration, UFR was also markedly reduced (4 ± 0.7 μl/h) in the cold-acclimated frogs. The urine contained 14 mmol/l glycerol, resulting in a urinary glycerol excretion rate of 52 nmol/h, more than three times that in warm-acclimated animals (Tables 2 and 3). The fractional tubular water reabsorption under these circumstances was diminished to 33% compared with warm-acclimated frogs, whereas fractional tubular reabsorption of glycerol increased to 82% compared with 64% in warm-acclimated animals.
Tissue glycerol concentration increased in the cold-acclimated frogs to 187 ± 18 and 131 ± 22 μmol/g dry mass in liver and muscle, respectively (n = 4). Tissue water was 66 ± 1% in liver and 71 ± 1% in muscle and was unchanged compared with warm-acclimated frogs. Inulin distribution time was longer in cold animals, consistent with an expanded extracellular fluid (7). Moreover, attempts at blood sampling from cold-acclimated animals yielded samples with one of two characteristics: few erythrocytes (hematocrit <5%) or “blood-rich” samples. This too suggests an expanded lymphatic (extracellular, extravascular) space in cold-acclimated animals. The hematocrit measured from blood-rich samples did not differ from that in warm-acclimated animals (Table 2).
Response to Dehydration
Gray tree frogs dehydrated by 20% of standard body mass in ∼48 h. This loss of body water did not induce a significant increase in plasma concentrations of glycerol or glucose (0.80 and 0.39 mmol/l, respectively) or urinary glycerol (0.26 mmol/l), although urinary glucose concentration was significantly elevated (0.7 mmol/l) compared with warm-acclimated frogs. GFR also changed significantly (83.6% reduction to 37 ± 16 μl/h). Similarly, UFR was reduced to 2 μl/h, reflecting an increase in fractional tubular water reabsorption from 62% in warm-acclimated frogs to 95% in cold-acclimated frogs (Table 3).
Isolation of two novel AQPs in H. chrysoscelis.
Zardoya and Villalba (48) identifed a number of sequences that are highly conserved among AQPs, specifically those surrounding the two highly conserved NPA motifs, the first of which occurs in the intracellular B loop and the second within the extracelluar E loop. Other regions typically differ in particular paralogs, with the strongest divergence observed in the NH2 and COOH termini of the protein. Two regions of ∼100% amino acid identity common to seven anurans were identified using published AQP sequences from anurans available through GenBank. Degenerate oligonucleotide primers designed to the first conserved region (GHVSGAHLNPAVT) and to the second conserved region (MNPARSF) amplified partial cDNAs for two orthologous AQPs, HC-1 and HC-2, and full-length cDNA sequences were obtained for each AQP using RACE.
HC-1, initially cloned from kidney of warm-acclimated frogs, is predicted to encode a 272-amino acid protein (see supplemental Fig. 1S in online version of this article). HC-1 has 98% amino acid identity and 97% nucleotide identity with the water channel AQP-h1 previously isolated from H. japonica (36). It also has a high proportion of amino acid identity with B. marinus AQP-t1, Xenopus tropicalis AQP1, and hAQP1 (Table 3; see supplemental Fig. 1S). HC-2 was initially cloned from urinary bladder of warm-acclimated frogs and encodes a 280-amino acid protein (see supplemental Fig. 2S). No apparent similarities between HC-2 and other known AQP family members were observed at the nucleotide level with a BLASTN search of GenBank, although a discontinuous MegaBlast search revealed small regions of conserved nucleotide sequence (Table 4). The largest contiguous region with a relatively high degree of identity was with mouse AQP2 (69% identity over a span of 493 nt). No stretch of 20 contiguous nucleotides in the HC-2 sequence was found to match ≥18 contiguous nucleotides in H. japonica AQP2-h2, although 60% amino acid identity and 75% amino acid similarity were observed between the proteins encoded by the two genes. Additionally, single short regions of high nucleotide identity were identified with AQP-t4 of B. marinus (extending from nt 504–583 of HC-2), with an AQP from X. tropicalis (80% identity to HC-2 nt 492–592), and with hAQP2 (77% identity to HC-2 nt 540–623; Table 4). Although divergent at the nucleotide level, HC-2 exhibits strong amino acid identity and similarity with AQP2 orthologs, with the longest contiguous stretches of identity spanning the highly conserved NPA motifs (Table 3; see supplemental Fig. 2S). These data suggest that HC-2 is similar to AQP2, although it is encoded by a highly divergent nucleotide sequence.
Isolation of a GLP in H. chrysoscelis.
Because thermal acclimation in H. chrysoscelis entails glycerol transport, we attempted to identify members of the MIP family that might facilitate glycerol transport (GLPs). Sequence alignments of previously sequenced GLPs identified two highly conserved regions, again surrounding the two canonical NPA motifs. Starting with primers designed to those conserved regions, we were able to identify one putative GLP, designated HC-3. HC-3 encodes a 292-amino acid protein (see supplemental Fig. 3S). Sequence comparison revealed striking amino acid identity (82–88%) and similarity (90–94%) with anuran and mammalian AQP3 (Table 4). Similarity extends to the nucleotide level, with 79–81% sequence identity among the AQP3 family members.
Functional prediction based on sequence analysis.
AQPs and GLPs are subfamilies of the larger MIP family of proteins, which are functionally characterized on the basis of water and glycerol transport properties. Previous workers (10, 18) have identified five amino acid positions (P1–P5) that differ consistently between the AQP and GLP subfamilies and are likely to affect channel selectivity to water and solutes, including glycerol. Examination of the amino acids at these five positions indicates that HC-1 and HC-2 fall within the functional class of AQPs, whereas HC-3 is consistent with GLP function (Table 5). Thus, on the basis of sequence similarities (Table 3; see supplemental Figs. 1S, 2S, and 3S) and motif analysis (Table 5), HC-1 and HC-2 are likely to function as AQPs, whereas HC-3 is likely to function in water and/or glycerol transport as a GLP.
Pf of HC-1, HC-2, and HC-3 in Xenopus oocytes.
On the basis of the sequence similarities between HC-1, HC-2, and HC-3 and members of the AQP/GLP family, each clone was evaluated for its ability to facilitate osmotically driven transmembrane water movement. Oocytes injected with hAQP1, HC-1, HC-2, and HC-3 showed time-dependent osmotic swelling at room temperature and in the cold compared with the water-injected oocytes, which showed no swelling (Figs. 1A and 2A). At room temperature, hAQP1, HC-1, and HC-2 burst within 120–240 s after exposure to hyposmotic (67 mosmol/kgH2O) buffer (Fig. 1A), whereas only hAQP1 burst within the 5-min time frame under cold conditions (Fig. 2A). For all three HC clones as well as hAQP1, Pf was significantly higher (P < 0.01) than for water-injected oocytes at 23°C and 10°C. At the warmer temperature, HC-1 showed the highest Pf (121.25 ± 11.69 × 10−4 cm/s, 24-fold higher than water-injected oocytes); Pf of HC-2 and HC-3 (80.12 ± 8.25 × 10−4 and 25.01 ± 1.72 × 10−4 cm/s, respectively) were 16 and 5 times greater, respectively, than Pf of water-injected controls (Fig. 1B). Preincubation with 0.3 mM HgCl2 reduced the average Pf of HC-1 by ∼54% (P < 0.01; Fig. 1B). At 10°C, HC-1 induced the highest Pf (104.11 ± 9.00 × 10−4 cm/s, P < 0.01), 16 times that of water-injected controls (Fig. 2B). Pf of HC-1 was reduced by 56% (P < 0.01) after preincubation with mercury (Fig. 2B). Similarly, under cold conditions, Pf of HC-2 and HC-3 (71.13 ± 7.30 × 10−4 and 10.66 ± 1.16 × 10−4 cm/s, respectively) were significantly greater (P < 0.01 and P < 0.05, respectively) than Pf of water-injected controls (Fig. 2B). Interestingly, the hyposmotic Pf of HC-2 was significantly decreased by ∼42% (P < 0.01) in the presence of mercury at 10°C but not at 23°C (Fig. 2B). Mercury had no effect on HC-3 at either temperature (Figs. 1B and 2B). Although the mean Pf for all three HC clones was lower at 10°C than at 23°C, this difference was statistically significant only for HC-3, the Pf of which was ∼58% lower in the cold (P < 0.01).
Glycerol permeability of HC-1, HC-2, and HC-3.
To address whether HC-1, HC-2, and/or HC-3 function as a GLP in addition to an AQP, glycerol permeability of Xenopus oocytes injected with cRNA for each HC clone as [14C]glycerol uptake relative to water-injected controls was determined at 23°C and 10°C in the presence and absence of mercury. HC-1 expression significantly enhanced glycerol uptake at 23°C, which was inhibited by mercury, whereas no statistically significant glycerol uptake was seen at 10°C (Table 6). HC-3 was more permeable to glycerol than was HC-1: relative glycerol uptake of 2.37 ± 0.18 (P < 0.01) for HC-3 at 23°C compared with 1.48 ± 0.11 for HC-1 (Table 6). Preincubation with mercury inhibited glycerol permeability of HC-3-injected oocytes at 10°C (P < 0.01) but not at 23°C, suggesting that mercury inhibition of HC-3 is temperature sensitive. No enhancement of glycerol permeability was seen with HC-2 under any condition.
Tissue expression of HC-1, HC-2, and HC-3.
To determine tissue-specific expression patterns of HC-1, HC-2, and HC-3 mRNA, cDNAs synthesized from warm- and cold-acclimated tree frog brain, liver, lung, skin, kidney, fat body, muscle, bladder, and gut were used as templates in real-time quantitative PCR using gene-specific primers. Expression levels of HC-1, HC-2, and HC-3 (relative to β-actin expression) were determined for each tissue (Fig. 3). HC-1 mRNA is present in all tissues tested under warm- and cold-acclimated conditions (Fig. 3A). Low levels of HC-1 expression were detected in fat body from warm-acclimated frogs and in skin and fat body from cold-acclimated frogs. HC-1 expression appears to vary with thermal acclimation in certain tissues. HC-1 expression is ∼5 times more abundant in liver from cold- than from warm-acclimated frogs, ∼3 times more abundant in kidney from warm-acclimated frogs, and ∼24 times more abundant in brains from warm-acclimated frogs. In contrast to HC-1, HC-2 mRNA expression is highly tissue specific; it is also subject to tissue-specific thermal regulation. Whereas HC-2 expression is present in skin from cold-acclimated frogs, it is not detected in skin from warm-acclimated frogs (Fig. 3B). HC-2 also is expressed in kidney (1.5 times as abundant in warm- as in cold-acclimated frogs) and in bladder (low levels in warm- and cold-acclimated animals). Similar to HC-1, HC-3 transcript is present in all nine tissues under warm and cold conditions (Fig. 3C). However, HC-3 is expressed at very low levels in muscle and fat body from warm-acclimated frogs and in fat body from cold-acclimated frogs. Interestingly, HC-3 mRNA appears to be more abundantly expressed in muscle (∼3,500-fold), liver (∼5-fold), lung (∼3.8-fold), and bladder (∼3-fold) from cold-acclimated frogs than from warm-acclimated frogs, whereas expression was highest (∼28-fold) in skin from warm-acclimated animals. Taken together, these data indicate that a subset of AQP/GLP expressed in H. chrysoscelis exhibits tissue specificity that is subject to thermal regulation.
Winter Acclimation: Solutes and Water
H. chrysoscelis prepares for winter freezing by increasing liver glycogen stores and circulating glycerol (12). On the basis of the time course of glycogen depletion and glycerol accumulation, it is thought that the glycerol derives from hepatic synthesis during freezing mobilization (35), although neither the source (liver or, perhaps, muscle or adipose tissue) nor the metabolic synthetic path has been conclusively determined for anticipatory glycerol accumulation during cold acclimation. Cues to initiate this process may include changes in temperature and photoperiod or metabolic shifts associated with cessation of feeding; the lack of glycerol accumulation in dehydrated frogs suggests that desiccation is not a primary trigger.
In previous studies of cold-acclimated gray tree frogs (H. versicolor and H. chrysoscelis), the circulating concentration of glycerol has ranged from negligible to >100 mmol/l (12). These disparate reports likely reflect different acclimation regimens, including adequacy of food before acclimation (12). The concentrations of glycerol measured in the present study, ∼50 mmol/l, are in the middle of the range of previous reports. Similarly, frogs in our study accumulated glycerol in tissues (liver and skeletal muscle) to an extent similar to that previously reported (12).
In contrast to glycerol, which accumulates in gray tree frogs during cold acclimation, in anticipation of freezing, glucose mobilization in freeze-tolerant anurans is stimulated primarily by freezing (19, 21). Consistent with this finding, we detected just a small rise in plasma glucose in cold-acclimated H. chrysoscelis.
During freezing, water shifts from intracellular to extracellular fluid and potentially even out of the blood as ice forms (12, 21). In the cold, there was an apparent increase in lymph accumulation, but no change in hematocrit or tissue water content, suggesting potentially a combined contribution of reduced renal filtration and water absorbed from external sources.
GFR and UFR in amphibians are labile. In response to dehydration, both of these functions consistently decrease (31, 32). Our results were consistent with this, including an 84% drop in GFR and a >97% drop in UFR. The significantly greater proportional drop in UFR than in GFR indicates a contribution of enhanced water reabsorption, which could have occurred in the renal tubules or the urinary bladder, both of which are responsive to vasotocin, the amphibian antidiuretic hormone (ADH) (31, 38).
The method we employed to evaluate kidney function (sampling bladder urine at intervals) minimizes disturbances to normal function, e.g., from anesthesia or catheterization. However, it is possible that ureteral urine is modified by the bladder epithelium, which can potentially transport water and electrolytes (33) and glucose (3). We assumed that the tissue is inert with respect to inulin, as have others for inulin or creatinine (27–29, 32), and so measures of GFR would be unaffected by bladder transport. However, because the capacity for glycerol transport has not been evaluated, our measures of urinary glycerol excretion and UFR may include an as yet unquantified component of bladder transport.
There are few data on the response of amphibian kidney function to cold. In Rana clamitans kept in 20–25°C water and then transferred to 2–5°C water, GFR decreased but tubular reabsorption decreased as well. The former response predominated, and the result was a reduction in urine flow (28). This pattern occurred also in gray tree frogs, leading to a marked reduction in urine flow.
Urine is potentially an avenue of significant loss of circulating cryoprotectant. For example, wood frogs liberate glucose on freezing. During postfreeze thaws, when renal function resumes, that glucose is filtered, and some is lost in the urine. As a result, repeated cycles of freezing and thawing result in progressive loss of glucose and reduction of glycogen stores (20, 22). Because gray tree frogs may circulate glycerol for many weeks of cold acclimation, they also potentially risk losing that valuable solute. Reducing GFR and UFR should help retain glycerol, and reduced tubular urine flow likely facilitates glycerol reabsorption. Nevertheless, the high concentration of filtered glycerol still results in measurable urinary glycerol. It is possible that some of this glycerol could be reabsorbed postrenally by the urinary bladder (23). In the wild, when animals might remain in one spot for prolonged intervals, excreted urine might even be modified by cutaneous absorption. The capacities of the bladder and skin for transport of glycerol are not known.
AQPs belong to a family of MIPs that occur in all organisms (15, 48). Vertebrates express two main functional classes of these proteins. AQPs proper (AQPs) transport only water, whereas GLPs transport water and organic compounds such as glycerol (5, 40). The relative selectivity for water vs. glycerol may rest with the polarity of a few amino acids in the constriction of the AQP channel (6, 10, 37, 39).
The number and diversity of AQP isoforms increase along with organismal complexity (5). Mammals express ∼12 AQPs and amphibians may as well. In mammals, each AQP has a specific pattern of tissue distribution: AQP1 is present in most tissues, whereas most other mammalian AQPs are expressed more selectively (15).
Even before AQPs were understood as such, studies of toad urinary bladder revealed insertion into the apical membrane of “particle aggregates” or “aggrephores” under conditions of enhanced water reabsorption (43). Since then, several proteins of the AQP family have been sequenced from anurans (17, 24, 30, 41, 44). In H. japonica, the species most closely related to H. chrysoscelis for which AQPs have been studied (9, 28), three AQP genes have been identified. One of these (AQP-h1) is widely distributed among tissues and closely related to mammalian AQP1, whereas expression of AQP-h2 and AQP-h3 is restricted to osmoregulatory organs and may be regulated by ADH.
In H. chrysoscelis, we also identified one AQP (HC-1) that is a likely homolog of AQP1. Indeed, it has nearly perfect identity to AQP-h1 at the amino acid level and is similarly expressed in all tissues tested in warm- and cold-acclimated frogs. Functionally, HC-1 shows high Pf at 10°C and 23°C and is inhibited by ∼55% in the presence of mercury. These functional characteristics, high Pf and mercury sensitivity, are similar to those of other AQP1 homologs. HC-1 was slightly permeable to glycerol at 23°C but not at 10°C. Interestingly, the level of mRNA expression (more abundant in the brain and kidney of warm frogs and in the liver of cold frogs) is also sensitive to temperature. The physiological role of these changes in vivo is unknown.
The second AQP that we identified from H. chrysoscelis, HC-2, is most like a member of the AQP2 family, which includes the vasotocin-regulated AQPs from H. japonica. Although the nucleotide sequence of HC-2 is highly divergent from genes for other AQP2-like proteins, the amino acid similarity is much greater (78–83%). Similar to other AQP2-like proteins, HC-2 enhanced osmotic water movement in the Xenopus oocyte expression system. Moreover, consistent with its membership in the HC-2 family, we detected HC-2 mRNA primarily in organs of osmoregulation (skin, bladder, and kidney). HC-2 retained a high Pf in warm and cold conditions. However, expression of HC-2 mRNA varied with thermal acclimation: whereas HC-2 was expressed in skin from cold-acclimated frogs, it was absent in skin from warm-acclimated frogs. Amphibian freeze tolerance may well have evolved from more general responses to dehydration. Thus, because HC-2 is related to other ADH-responsive AQPs from the AQP2 group, it is tempting to speculate that shifts in expression (e.g., upregulation in the cold) reflect strategies anticipatory of freezing. This awaits confirmation.
HC-3 was identified using primers designed against conserved regions of several GLP-type proteins. Motif analysis of HC-3 identified GLP-distinctive amino acids at key locations, indicating that HC-3 is a likely member of the GLP branch of the MIP family. The studies of water permeability and glycerol uptake support that conclusion from sequence homology. Of the three HC clones characterized, HC-3 was the weakest water channel at 23°C and was even weaker (>50% decrease in Pf) at 10°C. A decrease in water permeability at colder temperatures has been described previously for another AQP (13), although the mechanism by which this occurs is not known. However, it is noteworthy that HC-3 expression in tree frogs increases in many tissues after cold acclimation. It may be that an increase in abundance is necessary to compensate for a decrease in water channel function.
Glycerol uptake experiments identified HC-3 as a strong glycerol transporter at 10°C and 23°C. At the sequence level, HC-3 cDNA is most similar to anuran and mammalian AQP3, expressed in mammals in the basolateral membrane of renal collecting ducts and elsewhere (26). Taken together, these data indicate that HC-3 is likely a member of the GLP branch of the MIP family. AQP3 is one of several GLPs expressed by mammals. In mice, the GLP AQP7 is required for glycerol exit from adipocytes (8, 11, 25). However, there is little or no evidence that other GLPs play an important physiological role in glycerol transport. Gray tree frogs, with their naturally high circulating concentrations of glycerol, may provide a model for evaluating that function.
We found that the distribution of HC-3 expression among nine tissues of H. chrysoscelis is subject to variation associated with thermal acclimation. Notably, more HC-3 transcript was detected in liver, lung, gut, brain, muscle, and bladder isolated from cold- than from warm-acclimated tree frogs. As discussed above, glycerol transport in liver and/or muscle may participate in synthesis, distribution, and accumulation of this metabolite during cold acclimation in H. chrysoscelis. The upregulation of HC-3 in bladder and downregulation in skin from cold-acclimated tree frogs suggest a potential role for HC-3 in water/glycerol conservation in anticipation of freezing. Thus the findings that HC-3 facilitates glycerol and water transport and is distributed dynamically in tissues that are likely to participate in cold acclimation support our hypothesis that proteins from the MIP family participate importantly in seasonal osmoregulation.
Clearly, gray tree frogs mobilize water and glycerol across cell membranes during cold acclimation and subsequent freezing: glycerol circulates for distribution to the organs, freezing induces cellular dehydration, and kidneys reabsorb water and glycerol in the cold. From a broader perspective, this study combines physiological measures of glycerol handling under changing thermal conditions, while it identifies key molecules proposed to facilitate glycerol transport and a mechanism by which a multicellular organism responds to thermal cues for cold acclimation and potential preparation for freezing.
The data presented here demonstrate that gray tree frogs alter glycerol handling via systemic (renal) and cellular mechanisms in response to the demands of seasonal shifts in temperature. We studied only the changes that occur in anticipation of freezing. In the eventuality of actual freezing, these responses are further accentuated, as renal function becomes quiescent and ice formation induces further shifts of body water. It is intriguing to speculate that AQP/GLPs, including those we have identified, may provide the route for transmembrane water and glycerol movement during these shifts in physiological function. Although correlative expression data are intriguing, they do not define HC-1, HC-2, or HC-3 function in these tissues, nor do they indicate a differential role in osmoregulatory responses during thermal acclimation. The mere presence or abundance of a transcript in a particular tissue does not indicate its importance in a physiological process, and the function of that gene in different tissues may vary. However, our data demonstrate that a subset of tissues show altered AQP/GLP expression under warm- vs. cold-acclimated conditions, suggesting that AQP/GLPs may be subject to previously undescribed mechanisms of thermally induced gene regulation. Moreover, we demonstrate that all three HC clones are functional water channels and that the degree to which HC-3 functions as a water channel is temperature dependent. This study provides the critical first step required for understanding the physiology of water and glycerol handling in response to cold acclimation in H. chrysoscelis and the potential role of three novel AQP/GLP members in this process.
This research was supported in part by National Science Foundation Research Grant IOB-0517301 to D. L. Goldstein and C. M. Krane.
We thank Drs. Richard E. Lee, Jr., and Jon P. Costanzo and Jim O'Boyle for assistance with collecting and maintaining frogs, Dr. Y. Yang for instruction in the oocyte expression system, Dr. Sudhindra Gadagkar for assistance with the statistical analysis, Alison Staton for technical assistance, and Dr. Dan E. Krane for review of the manuscript.
↵* Both authors contributed equally to this work.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society