In striated muscle, the structural genes associated with muscle fiber phenotype determination as well as muscle mass accretion are regulated largely by mechanical stimuli. Passive stretch of skeletal muscle stimulates muscle growth/hypertrophy and an increased expression of slow muscle genes. We previously identified Ankyrin repeat-domain protein (Ankrd2) as a novel transcript expressed in fast tibialis anterior muscles after 7 days of passive stretch immobilization in vivo. Here, we test the hypothesis that the expression of Ankrd2 in stretched fast muscle is associated with the stretch-induced expression of slow muscle phenotype rather than the hypertrophic response. Our results show that, in 4- and 7-day stretched tibialis anterior muscle, the expression of Ankrd2 mRNA and protein was significantly upregulated (P > 0.001). However, in fast muscles of kyphoscoliotic mutant mice, which lack the hypertrophic response to overload but have a slower muscle phenotype than wild-type, Ankrd2 expression was significantly upregulated. The distribution pattern of Ankrd2 in fast and slow muscle is also in accord with their slow fiber composition. Furthermore, it was markedly downregulated in denervated rat soleus muscle, which produces a pronounced shift toward the fast muscle phenotype. Using a sensitive proteomics approach (Ciphergen Technology), we observed that Ankrd2 protein was undetectable in soleus after 4 wk of denervation. We suggest that Ankrd2, which is also a titin binding protein, is a stretch-response gene associated with slow muscle function and that it is part of a separate mechanotransduction system to the one that regulates muscle mass.
- kyphoscoliosis mice
mammalian skeletal muscles consist of populations of slow-contracting, fatigue-resistant, oxidative fibers that are adapted for slow repetitive or postural type of contractile activity and also fast-contracting, highly fatigable, glycolytic fibers that are recruited for fast phasic movements (30, 36). These muscle fiber types differ phenotypically in that they express different subsets of myofibrillar protein isoforms as well as different types and levels of metabolic enzymes suited to the type of function they perform. A remarkable property of skeletal muscle fibers is their ability to change phenotype in response to developmental and environmental stimuli (11). Endurance training converts fast muscles toward a slow phenotype (38), whereas strength training causes muscle fiber hypertrophy and a shift to a fast phenotype (37). Increased load and passive stretch also have marked effects on muscle mass and fiber phenotype. In response to immobilization in the stretched position, the tibialis anterior (TA) muscle of the rabbit has been shown to undergo hypertrophy and increase the expression of its slow type 1 myosin heavy chain isoform (42). The hypertrophic response allows the muscle to add more sarcomeres in series and to increase in mass, whereas the expression of slow fibers enables the muscle to adopt a more economical contractile phenotype (41). This inherent ability of skeletal muscle to adapt to increased functional length imposed by passive stretch is related to its ability to “switch on” or “switch off” different genes and to alter the general levels of expression of different subsets of genes (3, 12). The regulation of such phenotypic changes is complex, and our current understanding of the process is fragmentary.
In a search for key stretch-activated genes in muscles we reported the identification of Ankyrin repeat-domain protein (Ankrd2) gene, which in normal mature muscle is expressed preferentially in slow type 1 muscle fibers (34, 33). After a single bout of eccentric contraction Ankrd2 mRNA is upregulated and may therefore be considered as an immediate/early response gene (3). Ankrd2 mRNA encodes a protein, which has 4.5 ankyrin-like repeat motifs and several phosphorylation sites (20). It has both nuclear and cytoplasmic localization and binds to transcription factor YB-1 and to cell cycle signaling molecules such as promyelocytic leukemia protein 1 and p53 (23). Recently, binding studies have also demonstrated an association between Ankrd2 and titin a stretch-sensitive protein implicated in muscle signal transduction (16, 31). Thus we tested the hypothesis that Ankrd2 is a stretch-activated gene associated with slow muscle phenotype rather than hypertrophy.
Here, we show that Ankrd2 is highly expressed in fast skeletal muscle after 4 days of passive stretch in a similar fashion to that observed for the IGF-1 splice variant mechano growth factor (29). This high-level expression is maintained up to 7 days of passive stretch. To determine whether induction of Ankrd2 is part of the muscle's stretch-induced hypertrophic response and/or its stretch-induced expression of a slow muscle phenotype, we studied its distribution in fast and slow skeletal muscles of normal animals and in an experimental rodent model, the kyphoscoliosis mutant mouse, wherein phenotypic gene switching occurs but is not accompanied by hypertrophy (5). We also studied Ankrd2 expression in denervated soleus muscle, which undergoes a shift to fast muscle phenotype. We suggest that Ankrd2 is a stretch-response protein functionally downstream of titin with a role in slow muscle function.
MATERIALS AND METHODS
All animal experiments were carried out according to the guidelines of the local Animal Care Committee and complied with the United Kingdom Animal Scientific Procedures Act 1986 or were approved by the ethical committee of the University Medical Centre Nijmegen, The Netherlands, as appropriate. Mice and rats were fed ad libitum using the same chow throughout the course of the experimental period.
First, we studied Ankrd2 expression in stretched TA muscle of 12-wk-old C57/BL10 mice. Passive stretch was achieved by plaster cast immobilization of the right hindlimb in the plantar flexed position, as described previously (20). Three groups of mice were used in these experiments (n = 6). In groups 1 and 2, muscles were stretched for 4 and 7 days, respectively. Group 3, designated “non-stretched controls,” served as a reference group for normal muscle growth. Extensor digitorum longus (EDL) and soleus muscles were also dissected from the control age-/gender-matched group and served as reference muscles for Ankrd2 expression studies during normal growth.
Second, we studied Ankrd2 expression in the fast muscle of kyphoscoliosis (ky) mutant mice. Ky mice carry a null mutation in the muscle-specific ky gene encoding a putative transglutaminase/protease of unknown function, which arose spontaneously in 1970 in the BDL mouse strain (10). The BDL strain is essentially a normal strain of mouse first mentioned by the laboratories of the Agricultural Research Council in Edinburgh, UK. It was developed for studies into the effects of scrapie virus infection. There is no reference to this strain in any of the main strain catalogs. Muscles of ky mice do not show the normal hypertrophic response to stretch/overload in vivo (3). In addition all ky muscles, including fast ones, are smaller than those of normal BDL mice; their muscles show extreme shifts toward a slower phenotype so that, for example, the soleus expresses only slow myosin heavy- and light-chain isoforms (27). Thus the ky mouse provides a unique tool for the characterization of factors involved in the adaptation of skeletal muscle to a slow phenotype. Ky and BDL mice were bred as specified pathogen-free colonies at the Medical Research Council Mouse Genetics Centre, Harwell, UK, and were a gift from Dr. Gonzalo Blanco. At autopsy, individual TA muscles were dissected, flash frozen in liquid nitrogen, and stored at −80°C.
Rat model of muscle denervation.
We studied Ankrd2 expression in the soleus muscle of denervated rats. The denervation procedure has been described previously by Degens et al. (9). Briefly, 6-mo-old male Wistar rats were anesthetized with 4% isoflurane and 2.0 l/min O2. The level of anesthesia was maintained with 1–2% isoflurane and 0.5 l/min O2. After reflex activity had disappeared, the branches of the tibial nerve that innervate the medial and lateral heads of the gastrocnemius muscle and the soleus muscle were transected as close to their entry point to the muscle belly as possible. To prevent reinnervation, the nerve stumps were sutured into the m. biceps femoris. The contralateral muscles served as controls. All procedures were performed under aseptic conditions. After completion of the surgery, the rats received an intraperitoneal injection of rimadyl (0.5 mg/kg) as an analgesic. The rats were killed by an overdose of pentobarbital sodium injected intraperitoneally. Muscles were dissected after 1 (n = 5), 2 (n = 5), and 4 (n = 5) wk of denervation, flash frozen in liquid nitrogen, and stored at −80°C for analysis.
Synthetic peptides H2N-MEGPEAVQRATELC-CONH2 and H2N-RPGSGRETPQPIPAQ-COOH (synthesized by Eurogentec, Belgium) derived from the Ankrd2 amino and carboxyl terminal, respectively, were used to generate anti-Ankrd2 antibodies. Polyclonal antibodies raised by immunization of New Zealand rabbits with keyhole-limpet hemocyanin coupled to the Ankrd2 derived synthetic peptides (Eurogentec, Belgium) were affinity purified, and their specificity for Ankrd2 was confirmed by Western blot analysis using purified mouse His-tagged Ankrd2 fusion protein as antigen and by subsequent immnuoprecipitation and mass spectrometry.
Protein Expression Studies
Muscle samples were homogenized in RIPA buffer containing 6 M urea, and the total protein was quantified using a Bradford protein assay kit (Bio-Rad). For each sample, 20 or 50 μg of total protein extract were fractionated by NuPage Bis-Tris acrylamide gel electrophoresis with MOPS running buffer. SeeBlue or MagicMark prestained standard (Invitrogen, UK) was used as molecular weight marker. After electrophoresis, the proteins were blotted onto polyvinylidene difluoride membrane and probed with anti-Ankrd2 polyclonal antibodies. For relative quantification, blots were stripped and reprobed with anti-GAPDH antibody (Abcam). Bands were visualized by chemiluminescence using Lumi-Light Plus reagent (Roche Molecular Biochemicals), according to the supplier's instructions, and then exposed to X-ray film. Signal intensities were quantified directly using Imagequant software (Molecular Dynamics).
Proteomic analysis of Ankrd2 in denervated rat muscle.
The absence of Ankrd2 protein in the soleus muscle after 4 wk of denervation was confirmed using a Ciphergen interaction discovery mapping (IDM)-based proteomic approach. Briefly, Ankrd2 antibody was coupled to IDM affinity beads via protein-AG in 50 mM sodium acetate buffer (pH 5.0) at 4°C overnight. This was then used to capture endogenous Ankrd2 in pooled protein extracts (n = 5) from denervated and contralateral soleus muscle, respectively. Bound Ankrd2 protein was eluted in a low pH buffer [50% acetonitrile/0.3% trifluoracetic acid (TFA)] and 2 μl of this protein eluate spotted directly onto a strong anionic exchange (Q10) protein array for analysis by surface-enhanced laser desorption and ionization-time of flight mass spectrometry. This was performed using the ProteinChip Biology System reader (PBS II, Ciphergen Biosystems). Relative peak intensity/profiles were determined as directed in the automated data collection protocol in the manufacturer's software. Matrix-assisted laser desorption and ionization-time of flight mass spectrometry was then used to confirm that the 37 kDa Ciphergen peak corresponded to rat Ankrd2. Briefly, the remaining protein eluate from the anti-Ankrd2-protein-AG IDM bead capture was separated by zoom gel electrophoresis (Invitrogen) over a pH range between 3 and 10 in the first dimension. The protein spot at a molecular mass corresponding to Ankrd2 (37 kDa) was “in-gel” digested with sequencing grade trypsin (Promega) and the tryptic digest extracted with 1% TFA. Digests were desalted using C18 zip tips (Millipore) and eluted in 50% acetonitrile-0.1% TFA solution. This eluate (0.5 μl) mixed with 0.5 μl of α-cyano-4-hydroxy-cinnamic acid saturated in 70% ACN-0.5% TFA solution was then spotted onto a stainless steel mass spectrometry plate and analyzed by matrix-assisted laser desorption and ionization-time of flight mass spectrometry on a Kratos Axima CFR instrument with pulsed extraction operating at 20 kV in reflectron mode. External calibration was performed using a mixture of peptides: des-Arg-bradykinin, angiotensin 1, neurotensin, ACTH-39 clip (Sigma). Subsequently, peptide masses were labeled and used for database searching and protein identification using MASCOT (MatrixScience; http://www.matrixscience.com).
mRNA Expression Studies
Real time quantitative RT-PCR was used to quantify the Ankrd2 and GAPDH mRNA levels in stretched and unstretched skeletal muscle. Total RNA extracted from stretched and unstretched muscle was reverse transcribed with thermoscript reverse transcriptase (Invitrogen) according to the manufacturer's instructions. The oligo-dT-derived cDNA was amplified using a Roche LightCycler. PCR reactions were performed in a 20-μ1 volume capillary using FastStart DNA SYBR Green kit, according to manufacturer's instruction (Roche Diagnostics). The thermal cycling profile consisted of a preincubation step at 95°C for 10 min followed by 40 cycles of 95°C denaturation steps for 10 s, 65°C annealing steps for 5 s, and 72°C extension steps for 20 s. At the end of the PCR, a melting curve analysis was performed by gradually increasing the temperature from 70 to 95°C (0.1°C/s). The level of expression of each mRNA and its estimated crossing points in each sample were determined relative to that of the standard preparation using the LightCycler computer software version 3.5. Standards were prepared by successive dilutions of a fixed concentration of oligo-dT-derived cDNA generated from mouse gastrocnemius muscle. A standard curve was constructed for each PCR run. Primers used for Ankrd2 amplification were AKDF100 (5′-GCTCCTGGGAAGCTGTCCAT-3′) and AKDR575 (5′-CAGTGCATGGCTGTGCAGTC-3′), respectively. The primers used for the GAPDH amplification were GAPDH-F (5′-CCTGGCCAAGGTCATCCATGACAA-3′) and GAPDH-R (5′-GAGGTCCACCACCCTGTTGCTG-3′), respectively.
Statistical analysis was performed using SigmaStat. Experiments were done in triplicate, and signal intensities within the linear range were converted to numerical values and normalized to that obtained for GAPDH. Comparison between groups was performed using the Student's t-test. Data are mean values. Standard deviation from means are presented as bars. Significance was accepted at the levels of P < 0.05 at a power of >0.8.
Expression of Ankrd2 Protein in Response to Stretch
Densitometric quantitation of Western blotting was performed to assess the Ankrd2 protein level after 4 and 7 days of passive stretch. Figure 1, A and B, shows that there is a marked increase in the expression of Ankrd2 as early as 4 days of passive stretch compared with control TA (P = 0.0057). By 7 days of passive stretch, Ankrd2 levels decreased slightly, but the expression remained higher than in the control TA muscles (P = 0.002). Figure 1C shows that the increase in Ankrd2 protein expression correlates with an upregulation of Ankrd2 mRNA after 4 and 7 days of stretch.
Distribution of Ankrd2 Protein in Fast and Slow Mice Muscle
We have shown previously by semiquantitative RT-PCR that Ankrd2 mRNA was expressed at very high levels within slow-twitch skeletal muscles such as soleus (20). To further elucidate the role of Ankrd2, it was important to determine the relative Ankrd2 protein levels in fast and slow muscles from normal C57/Bl10 mice. Figure 2 shows the distribution of Ankrd2 in TA, EDL, and soleus muscles of 12-wk-old mice. The fast and slow myosin heavy chain composition in these muscles is well characterized (1, 2). Our results show that the slow postural soleus muscle, which is known to express relatively high levels of type 1 myosin heavy chain isoform, contained significantly more Ankrd2 protein than did fast TA and EDL muscles, which express relatively low levels of type 1 myosin heavy chain isoform. Interestingly, our anti-Ankrd2 antibody also recognized trace amounts of a 55-kDa protein in all skeletal muscles studied. We observed that, although the main 37-kDa Ankrd2 protein was undetectable in cardiac muscle, the 55-kDa protein was highly expressed (Fig. 2A). A 55-kDa band has also been detected in human cardiac muscle with an anti-Arpp polyclonal antibody raised against a human Ankrd2 protein. By open reading frame translation of the human Ankrd2 cDNA, it was determined that this 55-kDa protein may be due to translation initiation from an inframe ATG site further upstream of the human Ankrd2 gene (39). Our analysis of this predicted N-terminal extension sequence using the Simple Modular Architecture Research Tool (http://smart.embl.de/) identified a potential signal cleavage motif in this region. Because only one transcript is detected in Northern blot analysis of Ankrd2 (20), one possibility is that this is a precursor form of Ankrd2. However, by amino acid sequence comparison, we also observed that the fusion protein used to raise the anti-Arpp polyclonal antibody (39) encompasses the highly conserved C-terminal Ankrd2 sequence from which our anti Ankrd2 polyclonal antibody was derived. Hence, we cannot rule out the possibility that this 55-kDa protein product is consequent of antibody cross reactivity. At present, the functional significance of this 55-kDa isoform remains unclear.
Ankrd2 levels in control fast muscles were very low, but mechanical stretch induced a marked increase in Ankrd2 expression (Fig. 3). However, the stretch-induced increase in Ankrd2 expression in fast TA muscle was still considerably less than the endogenous Ankrd2 present in slow soleus muscle, possibly reflecting the small magnitude of slow phenotypic shifts associated with passive stretch during this relatively short period (unpublished observations).
Upregulation of Ankrd2 in Fast Muscle of ky Mice
Earlier experiments have shown that the muscles of ky mice are unable to hypertrophy when overloaded and that both their fast and slow muscles exhibit a much slower muscle phenotype than the counterpart muscles in normal BDL mice (5, 27). We therefore studied the expression of Ankrd2 in TA muscles of ky and control BDL mice. We chose to analyze the fast TA muscle, because BDL control mice express only trace amounts of Ankrd2 detectable by Western blotting in this muscle (see Fig. 2). Therefore, any difference in Ankrd2 expression between TA muscle from control and ky mice would be readily observed. Indeed, as demonstrated in Fig. 4, Ankrd2 levels were significantly higher in ky TA muscles than in the TA muscles of BDL control mice (P < 0.01), corresponding with the shift toward a slower muscle phenotype consequent to the ky mutation.
Expression of Ankrd2 in Denervated Rat Soleus Muscle
Previous studies have shown that Arpp, the human homolog of Ankrd2, is upregulated in 4-wk-denervated fast-twitch gastrocnemius muscle of rats (39). In this study, we analyzed Ankrd2 expression in denervated slow-twitch soleus muscle. Western blot analysis (Fig. 5A) demonstrates that, in soleus muscle, Ankrd2 was markedly downregulated 1 wk after denervation and was barely detectable 2–4 wk after denervation. Interestingly, we observed that the 55-kDa protein seen in Fig. 2 has an opposite pattern of expression in denervated muscle compared with the 37-kDa Ankrd2 protein. At present, the functional significance of this 55-kDa isoform remains unclear. Because the Ankrd2 protein was barely detectable in rat soleus muscle after 2 wk of denervation by conventional Western blot analysis, we chose a more sensitive proteomics approach to study Ankrd2 expression in 4-wk-denervated rat soleus muscle. Figure 5B shows that, even with the Ciphergen IDM bead-based proteomics approach, the 37-kDa peak corresponding to Ankrd2 in the contralateral control muscle was not detectable in soleus muscle after 4 wk of denervation. These data confirm the specificity of our Ankrd2 antibody to the 37-kDa Ankrd2 in rat soleus muscle and also confirms the reduction of Ankrd2 after denervation.
The findings of this study show that Ankrd2 protein was present at very high levels in slow-contracting soleus muscles of normal adult C57/Bl10 mice, whereas in fast muscles such as EDL and TA, which express very low levels of slow type 1 myosin heavy chain isoform (1, 2), Ankrd2 was almost undetectable. This is consistent with previous observations that Ankrd2 is associated with the slow muscle fibers (35, 39). Further supporting this link is our observation that, in soleus muscle, the slow to fast fiber transition that occurs after denervation (17, 35) is accompanied by a rapid downregulation of Ankrd2 after 1 wk of denervation to undetectable levels after 4 wk of denervation. Given that the stretch by immobilization procedure results in hypertrophy and an increased expression of slow isoforms of muscle genes (12), our data suggest that Ankrd2 upregulation in stretched TA muscle is part of its stretch-induced increase in expression of slow muscle genes.
Our studies on Ankrd2 expression in ky mice provide additional support for a role in slow muscle function rather than in the development of hypertrophy. In normal mice, stretch-induced hypertrophy involves activation of growth factors such as mechano growth factor, resulting in rapid muscle growth, whereas in dystrophic mdx mice functional overload of muscle does not activate mechano growth factor (13, 14) and even has a damaging effect on their muscles (7). Interestingly, mutant ky mice also suffer from a dystrophy characterized by bilateral neonatal degeneration of slow postural muscles over the first few weeks of life. Although these dystrophic muscles completely regenerate by 6 wk of age and show no further degeneration (6), neither fast nor slow muscles of ky mice hypertrophy in response to stretch/functional overload in vivo (5). Ky muscles are also substantially smaller and qualitatively weaker than those of normal BDL mice (28). For example, the mass of the soleus and EDL muscle in ky mice is ∼25–50%, respectively, of the counterpart muscle in age-matched BDL controls. However, ky body weights are ∼80% of normal, suggesting that, as a consequence of this abnormal muscle mass-to-body weight ratio, Ky (14) mouse muscles are in a constant state of overload (unpublished data). In the absence of an ability to undergo adaptive hypertrophy, ky muscles appear to compensate by shifting biochemically toward a slower, less fatigable muscle phenotype. Consistent with this is our observation that Ankrd2 expression is also higher in the fast TA muscle of ky mutant mice than in the TA muscle of BDL mice, supporting the hypothesis that induction and sustained expression of the 37-kDa Ankrd2 protein is correlated with a slow muscle phenotype.
The induction of Ankrd2 by stretch, its suppression by denervation, and its upregulation in ky mouse muscles suggests that Ankrd2 activation is linked to mechanical systems that sense stretch/load. Recent studies demonstrating the ability of Ankrd2 to bind and interact with proteins that sense and respond to stretch, such as titin, titin-associated proteins, myopallidin, and calpain protease p94, as well as the z-line protein telethonin (23, 33), support this notion. Titin is a giant protein that straddles the entire length of a half sarcomere from M line to Z disc in skeletal muscle (22, 26). Due to its high passive elastic-recoil speed (25) and its elastic N2A stretch-sensing region within the muscle I band, it has been identified as being responsible for a large proportion of the passive tension in skeletal muscles (40). Ankrd2 and its closest homolog cardiac ankyrin repeat-containing protein (CARP) have been localized ultrastructurally within titin's N2A domain (31). Mechanical strain/stretch imposed on the muscle leads to large conformational changes in titin that may range from straightening to complete unfolding of its N2A and Ig-like domains (15, 32). This stretch-induced titin unfolding has been shown to induce CARP expression and to increase the density of CARP bound to titin in cardiomyocytes in vitro (31). Because titin also has a signal transduction role in muscle (16), we suggest that Ankrd2, being the skeletal muscle homolog of CARP, is a component of a mechanotransduction system with titin at the center, where titin determines muscle fiber passive tension and plays an intracellular signaling role.
The possible role of Ankrd2 in this mechanotransduction system is intriguing. Fast to slow phenotypic adaptation in skeletal muscle involves the repression of a fast subset of myosin genes as well as upregulation of a slow subset of myosin genes. For example, 4 days of passive stretch resulted in decreased expression of the fast-type 2X myosin in rabbit TA muscle (29), whereas chronic electrical stimulation resulted in the downregulation of the fast myosin heavy chain as well as an upregulation of the type 1 myosin heavy chain (18). Conversely, denervation has been shown to repress the expression of slow type 1 myosin heavy chain mRNA soleus muscle (17). These observations demonstrate that transcriptional regulatory mechanisms play a key role in skeletal muscle adaptation, and in this context it is interesting to note that Ankrd2 has both nuclear and cytoplasmic localization sequences (23). Furthermore, Ankrd2 mRNA is present in the developing myotome of mouse embryos from as early as E9.5, and it is expressed during myoblast fusion and maturation in vitro (20). These expression patterns precede that of the sarcomeric proteins and strongly imply a role for Ankrd2 in myogenesis.
Structural analysis of Ankrd2 shows no obvious DNA binding motif. However, recent in vitro binding studies show that Ankrd2 can interact with transcription factors such as YB-1, p53, and promyelocytic leukemia protein in human myoblasts in vitro. Furthermore, Ankrd2 has also been shown to enhance p53's activation of the p21WAFI/CIPI promoter during early myogenesis (23). YB-1 and p53 have been shown to influence transcription through complex and opposing regulation of a number of genes, including the fas gene (24). Promyelocytic leukemia protein on the other hand can act as a transcriptional activator after binding to, for example, activator protein-1, CBP, rapidly adapting receptor-α, and p53 or a repressor when binding to Sp1, pRb, and Mad (21, 43). Although these in vitro interactions may not reflect the true binding partners of Ankrd2 in adult muscle in vivo, demonstrate the potential of Ankrd2 to influence transcription both positively and negatively through interaction with different transcription factors. Consistent with this is the observation that Ankrd2's closest homolog CARP, which also has cytoplasmic and nuclear localization sequences, is a component of a transcriptional repressor complex that can suppress cardiac specific gene expression (19).
In summary, our studies support a role for Ankrd2 in the slow muscle phenotype. Unraveling the functional role of Ankrd2 in slow muscle should prove to be important in understanding how sensing of mechanical signals leads to coordinated differential gene expression and hence phenotypic adaptation in skeletal muscle.
This project was funded by grants to G. R. Coulton from the Biotechnology and Biological Sciences Research Council and to G. Goldspink from the Wellcome Trust and University College London Biomedica.
We thank Dr. Gonzalo Blanco for donating ky mutant and BDL mice.
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 © 2005 the American Physiological Society