For most vertebrates, oxygen is a prerequisite for survival. Although we have previously shown that Drosophila melanogaster is hypoxia tolerant, how this species senses O2 deprivation and how it survives oxygen-limiting conditions are as yet poorly understood. We began to address this question by testing for anoxic responsiveness in Drosophila adult flies following overexpression of existing EP lines. In this screen, we identified Drosophila CG14709 gene as a homolog of the human multidrug resistance protein 4 (MRP4/ABCC4) that is tightly regulated to oxygen. Ubiquitous expression of dMRP4 in adult flies resulted in increased sensitivity to anoxia as they had longer recovery time from anoxic stupor. When exposed to 4% oxygen chronically (throughout its lifespan), constitutive expression of dMRP4 in larvae caused larval lethality due to growth arrest. Mutations of dMRP4 led to a hypersensitive response to acute anoxia in adult flies but had less impact on larval survival under chronic hypoxia compared with dMRP4 overexpression. Selective expression of this gene in neurons, but not in glia or muscles, mirrored the same phenotype as the ubiquitous one. Thus, our data suggest novel roles for MRP in vivo: 1) dMRP4 regulates the sensitivity to acute or chronic O2 deprivation, and 2) dMRP expression in neurons is sufficient to induce the sensitivity to O2 in the whole organism.
- oxygen deprivation
- ATP-binding cassette transporter
- multidrug resistance-associated protein
molecular oxygen is essential for development and growth of multicellular organisms and for the functional integrity of cells and tissues. In order for most prokaryotic and eukaryotic organisms to survive under hypoxic conditions, they must be able to sense and respond to changes in oxygen accordingly (13, 15). Severe or prolonged hypoxia often causes irreversible injury to cells, especially cells in the central nervous system, leading to necrosis or apoptosis. It is therefore not surprising that anoxic damage is a major cause of morbidity and mortality in humans, especially early or late in life. There are, however, some species of animals that are capable of tolerating very low oxygen for extended periods. These animals range from fish (31, 43), to amphibians (3, 10), to reptiles (26, 39, 42), and even to some mammals (4, 17, 18). Despite these observations, a number of questions still remain unanswered. For example, the genetic basis of oxygen responsiveness at an organismal level has not been well understood in previous studies.
Drosophila, as a model system, has been used to address a variety of questions relevant to human physiology, biology, and disease pathogenesis, including hypoxia. Although adult fruit flies normally do not live at extremely low oxygen atmospheres, we have previously demonstrated that Drosophila responds to oxygen changes (15, 16); indeed both Drosophila embryos and, to a less degree, larvae can survive severe hypoxia for many hours and days (9, 45). Adult flies can also survive even after complete deprivation of oxygen for hours without apparent injury (19, 21). Although certain gene products indeed are up- or downregulated in adults during anoxia (28), the molecular mechanisms by which Drosophila protects itself from anoxic injury remain largely unknown.
More recently, there have been several studies attempting to address questions related to mechanisms of hypoxia tolerance in Drosophila. The nitric oxide/protein kinase G signaling has been implicated in hypoxia-induced cell cycle arrest in embryos and behavioral changes in larvae (9, 45), yet it is not clear whether this signaling is also involved in adult response to hypoxia. An RNA editase, adenosine deaminase acting on pre-mRNA (ADAR), has been shown to mediate adult behavioral changes in response to anoxia (19, 27), probably through its editing functions on a number of target RNAs encoding for several ion channels, specifically in the nervous system (32, 37). Furthermore, the Drosophila basic helix-loop-helix-PAS proteins Similar and Tango have been demonstrated to function as homologs of mammalian hypoxia inducible factor (HIF)-1α and HIF-1β subunits, respectively (2, 14, 22). Whereas the HIF response is conserved in Drosophila, the hypoxic response shows distinguishable patterns: it reaches peak levels in the late stages of embryogenesis, then gradually decreases afterwards, and drops probably near baseline in adult (22). In a study using genome-wide microarrays, the increased transcription levels after exposing adult flies to hypoxia are evident but vary over time and according to the severity of the oxygen loss (25). Therefore, whether HIF is essential for the adult fly to sense low oxygen remains unclear.
Despite these major observations of the past few decades, little is known about how animals sense changes in oxygen tension and what signaling pathways mediate their responses. It is also not clear whether animals can distinguish between acute hypoxia, chronic hypoxia, or anoxia and whether the same molecular and physiological mechanisms of oxygen sensing apply to each condition. In this work, we identify a Drosophila protein related to a human multidrug resistance protein 4 (MRP4/ABCC4) using an overexpression screen and show that it plays a novel and important role in hypoxia regulation.
MATERIALS AND METHODS
Drosophila strains and genetics.
Drosophila stocks were maintained on standard cornmeal-molasses-yeast media at room temperature. All experimental crosses were done at 25°C. Gal4 lines were obtained from the Bloomington Stock Center. Enhancer and promoter (EP) lines (34, 35) were obtained from the Bloomington Stock Center and the European Drosophila Research Center. Males from EP lines were crossed to da-Gal4 or actin-Gal4 virgin females in pilot experiments, and then only da-Gal4 was used in the entire screen. Other Gal4 lines that have been used in this work were: elav-Gal4, drl-Gal4, Ddc-Gal4, D42-Gal4, repo-Gal4, 24B-Gal4, dpp-Gal4, engrail-Gal4, ptc-Gal4, and GMR-Gal4.
To generate excision of EP lines, both EP(3)3177 and EP(3)3221 were mobilized by crossing to a strain containing the Δ2–3 transposase. Excision lines were identified by loss of the expression of mini-white gene. We determined whether we have excision alleles by anoxia testing. Those that behaved like wild-type yw strain in response to anoxia were considered to be revertants, whereas those that behaved like the original homozygous EP(3)3177 flies were considered to be new alleles of dMRP4. The suspected alleles were then selected, and the levels of dMRP4 transcript were determined by semiquantitative RT-PCR.
EP screen and anoxia assay.
Individual EP lines were crossed to five virgins of da-Gal4 for 3 days, and then parents were discarded. The anoxia assay was performed essentially as described previously (19, 21). In general, 10–15 progeny of the correct genotype were scored for each EP line in the preliminary screen, and a candidate line was selected only if its average recovery time was ±60 s of control. We used both da-Gal4/+ and yw strains as controls. Both controls displayed essentially similar recovery time ∼6 min. We then tested ∼30 adult flies for each candidate line in the second round. In the second round, a candidate line was chosen only if it had an average recovery time ± 90 s of control and if most of tested flies for the same genotype showed similar recovery time as one another. Consistent candidates were further subjected to more tests with more number of flies. All data were analyzed by Microsoft Excel. Sequence information surrounding EP insertions were obtained from the Berkeley Drosophila Genome Project.
Chronic hypoxia assay.
Flies with different genotypes were allowed to lay eggs on the apple-juice plates for 3 h, and the newly hatched first instar larvae were transferred to vials containing standard food with density of 30 larvae per vial. Vials containing larvae with different genotypes were divided into two groups of duplicates or triplicates, and one group was kept at room temperature as normoxia control, whereas the other was placed into a chamber filled with 4% oxygen, which was monitored by an OM200 oxygen analyzer. After 20 days, numbers of pupae were counted on each vial. Similar experiments were carried out independently three or four times, and the statistic columns in Fig. 4 were pooled from at least three separate experiments.
Total RNA was isolated from whole flies or pupae using RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. cDNA synthesis from total RNA (3 μg) was performed with 0.5 μg oligo-dT and 50 ng random primers using SuperScript III first-strand synthesis system (Invitrogen, Carlsbad, CA) in a 20-μl volume. We amplified 1 μl of cDNA products with 2.5 units of HotStarTaq Plus DNA polymerase (Qiagen) in the presence of 10 pmol of each specific primer for CG14709 (dMRP4, 5′-GCTGAGTAGAACCGCTTTGG-3′ and 5′-GCACAGCATCTCATTGAAGC-3′) and CG6790 (5′-CCAGGGCTACATCAAGGTGT-3′ and 5′-GTCACCCCGAGTGGAATGTG-3′), or 10 pmol of each CG14709 primers together with 5 pmol of each Actin5C (5′-GTGGATACTCCTCCCGACAC-3′ and 5′-GCAGCAACTTCTTCGTCACA-3′) primers that were used as internal control. Amplification reactions were conducted for a cycle of 5 min at 95°C followed by 30 s at 94°C, 1 min at 56°C, and 1 min at 72°C for 26 cycles (Figs. 1, C and D, and 4B) or 28 cycles (Fig. 3C) in a 50-μl volume. The conditions were chosen so that none of the RNAs analyzed reached a plateau at the end of the amplifications. We also tested different sets of primers used in each reaction to make sure they did not compete with each other.
Images of the RT-PCR ethidium bromide-stained agarose gels were acquired with a charge-coupled device camera (Bio-Rad), and quantification of the bands was carried out by ImageQuant version 5.2 (Molecular Dynamics). Means and SD of all experiments performed were calculated after normalization to Actin5C.
Overexpression of CG14709 renders adult flies anoxia sensitive.
To identify candidate genes that have an impact on hypoxia tolerance or susceptibility, we made use of a gain-of-function screen (34, 35). We reasoned that if a protein is ubiquitously overexpressed throughout development and regulates the response to hypoxia, we should be able to observe differences between the overexpressing and the wild-type fly in terms of responses to O2 deprivation. We therefore screened adult flies carrying a variety of existing EP lines for behavioral changes, i.e., response to anoxia, using established methods in our laboratory (19). An EP line is a DNA element containing binding sites for the transcription factor GAL4 and a basal promoter activity to direct the expression of adjacent genes. Thus, EP lines are activated only in the presence of GAL4, hence determining the temporal and special control of EP gene expression.
Adult flies lose coordination, fall to the bottom of jar, and become motionless soon after being exposed to 0% O2 or anoxia. However, they completely recover after reoxygenation. We first determined the recovery time for the control group and then identified those that had significant variance from control mean recovery time and considered these as candidates of interest. We chose the Gal4 line da-Gal4 because it directs expression in almost all cells, from embryo to adult fly. Adult flies bearing da-Gal4 alone had an average recovery time (arousal latency) of 344 s after a 5-min period of anoxia, a period that we used for control throughout the screen. Only flies that produced a mean recovery time ≥90 s more than control were considered. This threshold was chosen because it was equal to 2–3 SD over the mean recovery time. By these criteria, 10 lines (from 1,600 EP lines screened) were selected as candidates for more rounds of repeated tests. Among those candidates, four EP lines, EP(3)3177, EP(3)3221, EP(3)3655, and EP(3)0430, were identified as giving very consistent results in this screen. Flies overexpressing these EPs using da-Gal4 displayed a profound increase in recovery time [mean ± SE: 553.5 ± 11.8 s for EP(3)3177, 455 ± 9.5 s for EP(3)3221, 454 ± 11.4 s for EP(3)3655, and 447 ± 15.7 s for EP(3)0430, respectively] after anoxia compared with control (mean ± SE: 344 ± 5.1 s for da/+; Fig. 1B). Similar behavioral responses were observed when the EP lines were crossed to actin-Gal4, a Gal4 driver that is also ubiquitously expressed throughout development (data not shown). Thus, we believe that the anoxia-sensitive responses observed here resulted from overexpression of these EPs with different Gal4 drivers and that the overexpression of these genes induced an interesting phenotype.
Subsequently, we found that four EP lines were all targeted to drive the same downstream transcript, and, in particular, both EP(3)3177 and EP(3)3221 were inserted at 88 bp, EP(3)3655 at 47 bp, and EP(3)0430 at 228 bp from the transcription start site of the predicted gene CG14709, respectively (Fig. 1A). The fact that four different EP lines rendered adult flies anoxia sensitive convinced us that CG14709 is responsible for this phenotype. In addition, we found that there was another EP insertion nearby CG14709. EP(3)3328 was inserted at 79 bp upstream of CG14709 but in an opposite orientation of the DNA strand. Hence, CG14709 expression should not be activated by EP(3)3228 with any Gal4 drivers and could serve as an internal control for CG14709. Indeed, in the presence of da-Gal4, flies bearing EP(3)3328 showed a normal response to anoxia (mean ± SE: 367 ± 7.0 s). To ascertain that CG14709 was upregulated in the EP line in the presence of Gal4, we used semiquantitative RT-PCR and show that CG14709 RNA expression was substantially increased in these EP flies induced by da-Gal4 (Fig. 1C). We thus concluded that CG14709 represented a gene whose upregulation conferred a more sensitive phenotype to O2 deprivation in adult flies.
When we analyzed the CG14709 genomic organization, we noticed that a neighboring gene CG6790 with unknown function overlaps in the opposite orientation and shares 354 nt of 3′-untranslated region sequence with CG14709. An RNA interference effect could take place when CG14709 is overexpressed. Therefore, the anoxia phenotypes could potentially be due to reduced CG6790 transcript level. We examined this possibility by RT-PCR and found that the overexpression of CG14709 did not alter the expression levels of CG6790 transcript (Fig. 1D). Thus, the anoxia phenotype was not related the gene CG6790.
CG14079 encodes a protein that belongs to the MRP subfamily, a member of the ABC transporter superfamily. Previous phylogenetic analyses have revealed that CG14079 may be an ortholog of MRP4/ABCC4 (8, 40), which is thought to transport a diverse array of physiologically related substrates (8, 36, 41). Whereas the human MRP4/ABCC4 has been overexpressed in insect cells to study its role in cyclic nucleotide transport (33, 44), the biological roles of Drosophila MRP4/ABCC4 homologs are yet to be determined. Therefore, we decided to further explore the CG14709 gene function, with a main focus on anoxia/hypoxia-related phenotypes at the whole animal level. We named this Drosophila anoxia-sensitive gene dMRP4.
Anoxia-sensitive phenotype is neuron dependent.
Neurons are extremely sensitive to oxygen deprivation. In mammals, hypoxia can often lead to neuronal cell death. In Drosophila, dADAR mutant flies display a prolonged recovery behavior after oxygen deprivation (19) and an age-dependent neurodegeneration phenotype (27). It has been found that the recovery behavior of adult flies during and after anoxia is highly correlated with physiological activity, such as evoked potentials (21). Since the da-Gal4 is expressed in all cells, including neurons, we wanted to test whether adult flies overexpressing dMRP4 only in neurons can mimic its ubiquitous overexpression phenotype. We therefore crossed EP(3)3177 flies to elav-Gal4, a driver that is expressed in all postmitotic neurons from embryos to adult flies (24). We found that elav>EP(3)3177 adult flies recovered after oxygen deprivation at a prolonged latency, very similar to what we observed in da>EP(3)3177 flies (Fig. 2). A comparable result was also obtained in elav>EP(3)3221 flies (mean ± SE: 478 ± 22 s). Thus, overexpression of dMRP4 only in neurons is capable of reproducing the adult anoxia-sensitive phenotype that is caused by ubiquitous overexpression.
This finding was further strengthened when other neuron-specific drivers were tested. drl-Gal4 is expressed primarily in the mushroom body and central body complexes (30). Ddc-Gal4 drives expression in dopamine and serotonin neurons (23). Both drivers, when expressed with dMRP4, resulted in a significantly prolonged recovery time after oxygen deprivation (Fig. 2). Interestingly, when dMRP4 overexpression was induced in glial cells with repo-Gal4 (1), flies behaved as controls. Furthermore, targeted expression of dMRP4 in cells other than neurons with 24B-Gal4, a driver that is expressed in developing muscles (5), had no significant effect on recovery time (Fig. 2), further suggesting that neurons may be the main targets for dMRP4 to function in response to anoxia. However, our results also suggest that probably only a specific set of neurons plays a major role in the anoxia-sensitive phenotype.
Mutations in dMRP4 are also sensitive to oxygen deprivation.
Since several EP lines were inserted very close to the transcriptional start site of dMRP4, the insertions themselves could interfere with dMRP4 expression, resulting in partial loss-of-function mutations. Because overexpression of dMRP4 led to anoxic sensitivity, we argued that dMRP4 mutant flies would be relatively resistant to anoxia. However, we found that homozygous EP(3)3177 adult flies recovered slowly in response to anoxia (Fig. 3A). Different assays using EP lines that resulted from allelic combinations for dMRP4 also showed prolonged latency to anoxia (Fig. 3A). Removal of the P-element from the EP(3)3177 chromosome restored the recovery time to control levels (i.e., line ex11 or line ex31 in Fig. 3B), indicating that the P-element insertion is the primary cause of the anoxia-sensitive phenotype in the absence of a Gal4 driver. Some excision lines also showed anoxia-sensitive behaviors when combined with the original EP(3)3177 chromosome. Both ex661/EP(3)3177 and ex621/EP(3)3177 flies took a longer time to recover from anoxia treatment (Fig. 3B). In these cases, the dMRP4 transcript was found to be downregulated in these anoxia-sensitive flies (Fig. 3C), suggesting that the some of dMRP4-related EP lines as well as excision lines, such as ex661 and ex621, are potential mutant alleles for dMRP4.
Influence of chronic hypoxia on larval survival.
To test whether dMRP4 plays a role in chronic hypoxia, we exposed newly hatched larvae to 4% O2 for 20–30 days. Under this condition, no animals could develop into adulthood, and, instead, they died during late pupal stages. We evaluated the role of dMRP4 in chronic hypoxia by determining the percentage of pupae that formed. Chronic hypoxia showed no effects on developmental progression of larvae, as both yw and da/+ controls had nearly the same percentage of pupae when exposed to 21% or 4% O2. However, when dMRP4 was overexpressed ubiquitously with da-Gal4 or only in neurons using elav-Gal4, no pupae could form under hypoxia (Fig. 4A). Similar results were obtained with EP(3)3221 or EP(3)3655 driven by da-Gal4 (data not shown). Overexpression of dMRP4 in the mesoderm derivates using 24B-Gal4 showed no effect on larval development under hypoxia (Fig. 4, A and B). Thus these results suggest that appropriate regulation of dMRP4 expression in neurons may be critical for larvae to survive under prolonged hypoxia.
How animals sense and respond to oxygen deprivation has been a challenge for decades. For instance, adult flies can recover from anoxia for up to 240 min without visible injury (21), but the genetic basis for such anoxia tolerance is largely unsolved. In previous studies, we found that there is a nonlinear relationship between anoxia time and time to recovery from anoxic stupor in adult Drosophila; hence we have previously suggested that anoxia tolerance is a sum of a number of processes rather than a single mechanism (21). To explore these processes, we have started to dissect the genetic basis of anoxia tolerance by an overexpression screen in adult Drosophila. Although the overexpression and hence a gain-of-function phenotype do not define the normal function of a gene during development, our initial goal in this study was to identify anoxia-related genes.
In this work we show that a Drosophila homolog of human MRP4, dMRP4, plays an important role in response to hypoxia. When this gene product is overexpressed, adult flies profoundly delay their recovery time from anoxia. Larvae cease their development and die before pupation during chronic hypoxia (exposed to 4% O2 during development). We also show, by targeted expression of dMRP4 in neurons, that neurons alone are sufficient to induce the sensitivity to low O2. The role that dMRP4 plays in sensing oxygen is not clear at this stage. Since neuromuscular transmission and muscle evoked responses have been tightly correlated with the recovery behavior of adult flies during and after anoxia (21), it was interesting to know the role of muscle in anoxia tolerance, in addition to neurons. We addressed this question by targeted expression of dMRP4 specifically in muscles under control of 24B-Gal4. We found that overexpression of dMRP4 in muscles only did not alter the behavioral response to anoxia in adult flies (Fig. 2) and had no effect on larval survival when exposed to 4% O2 (Fig. 4, A and B). Whereas these results did not completely rule out the potential muscle contribution to anoxia tolerance, we propose that, based on our results with targeted expression, the dMRP4 activity in neurons mainly accounts for sensing the oxygen changes by Drosophila. In support of this notion, targeted expression of dMRP4 only in larval imaginal discs did not change adult behavior in response to anoxia (data not shown).
The possible role for dMRP4 in neurons is also supported by studies of neuromuscular gain-of-function phenotypes in Drosophila. Kraut et al. (20) used elav-Gal4 to drive the expression of EP lines to screen for genes controlling motor axon guidance and synaptogenesis in Drosophila larvae. Overexpression of EP(3)3221 in larvae caused phenotypes such as nerve pathfinding defects and excess/ectopic synapses onto muscles, whereas overexpression of EP(3)0430 caused reduced/abnormal synapses phenotypes (see Table S. 1 in Supplementary materials in Ref. 20). Because both EP(3)3221 and EP(3)0430 are inserted 5′ to dMRP4, Kraut et al.'s work thus provides functional evidence that dMRP4 may play an important role in nervous system development in wild-type Drosophila. Interestingly, although EP(3)3177 has not been identified as a candidate line, like EP(3)3221, in Kraut et al.'s screen, it displayed the strongest anoxia phenotype in our screen compared with EP(3)3221 and other EP lines (Fig. 1). We believe that the quantitative variations in anoxia responses between EP(3)3177 and EP(3)3221 are due mainly to unknown genetic background in EP(3)3177 chromosomes, which somehow enhanced the anoxia phenotype seen in flies overexpressing EP(3)3177. Indeed, after several rounds of backcross with wild-type chromosomes later during our experiments, this line showed a recovery time (mean ± SE: 465 ± 9.3 s, n = 97) that is similar to other EP lines in the same group. Furthermore, we have noticed that overexpression of both EP(3)3177 and EP(3)3221 in neurons with elav-Gal4 resulted in strong lethality in adult flies (Fig. 4, see 21% O2 control group, survival rate 88% for da-Gal4 driver and 77% for elav-Gal4 driver, respectively) and that the lethality was dosage dependent because it affected males more severely than females (data not shown), presumably due, at least in part, to the consequence of the nerve pathfinding defects and excess/ectopic synapses in larvae.
Our screen was designed to identify anoxia-related genes based on adult behavioral responsiveness. Because anoxia represents a rather extreme oxygen environment (complete oxygen deprivation), one interesting question is whether responsive genes under acute anoxia could also be responsive genes under less severe oxygen conditions. Given that overexpression of dMRP4 gene caused adult flies to have prolonged recovery time under anoxia and resulted in growth arrest during larval development when subjected to a less severe O2 condition, the dMRP4 gene may be such a responsive gene. Thus, our overexpression screen has provided us with genes that not only respond to acute anoxia in adults but also to chronic hypoxia during early development. Our data also suggest that appropriate regulation of dMRP4 expression is critical for the Drosophila to live under low O2 environment.
The predicted dMRP4 transcript is 7004 nt in length, which encodes a transmembrane protein of 1,307 amino acids and is a member of the MRP group within the ABCC subfamily. MRPs represent the largest group in the ABCC subfamily, which consists of seven members. The human ABCC subfamily belongs to the ATP-binding cassette superfamily that is important in the transport of a diverse set of substrates across membranes. Based on both amino acids sequence similarities and phylogenic analyses, there are 12 ABCC genes present in all mammals and 12 ABCC genes in Drosophila, respectively (8, 40). Although the human ABCC genes have been implicated in ion transport, cell surface receptors, and toxin secretion activities (7), they have not been reported in hypoxic responsiveness thus far. Therefore, our findings expand our knowledge of unexplored roles of human ABCC genes in an in vivo model animal. Interestingly, in Drosophila, most MRP members appear to cluster with the human MRP4 (38, 40), raising the possibility of functional redundancy. Whether other Drosophila orthologs of human MRP4 also participate in anoxia/hypoxia-related physiological process remains to be determined.
Although the physiological functions of dMRP4 in Drosophila are not fully understood, a possible role for dMRP4 in oxidative stress response has recently been reported (29). Reactive oxygen species (ROS) are believed to be byproducts from O2-consuming enzymes or are produced in response to toxic reagents. High amounts of ROS can cause damage to proteins, DNA, and lipids (11). ROS have also been implicated in many human diseases including neurodegeneration (12). Adult flies overexpressing dMRP4 are found to be sensitive to exogenous ROS such as H2O2 and paraquat (29), raising an intriguing question as to whether there is a correlation between mechanisms that protect against low and high O2 levels. However, we have previously shown that, although dADAR mutant flies are hypersensitive to anoxia, they are resistant and not sensitive to ROS or paraquat (6). Therefore, the relationship between anoxia/hypoxia sensitivity and ROS as well as hyperoxia is not clear at present and deserves in depth investigations into mechanisms of survival and susceptibility.
This work was supported by National Institutes of Health Grants RO1NS-037756 and PO1HD-032573 (to G. G. Haddad).
We thank Ying Lu-Bo and Nuny Morgan for excellent technical assistance during EP screen, and the Bloomington Stock Center and the European Drosophila Research Center for Drosophila strains.
Address for reprint requests and other correspondence: G. G. Haddad, Dept. of Pediatrics, Univ. of California, San Diego, 9500 Gilman Dr., 0735, La Jolla, CA 92093-0735 (e-mail:).
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
- Copyright © 2007 the American Physiological Society