It is becoming clear that stress proteins play a role in various aspects of postischemic myocardial recovery and that the cytoskeleton of cardiac myocytes is an important determinant for cellular survival during ischemia and energy depletion. In the present study, we addressed the question of whether the cytoskeleton-binding stress protein αB-crystallin may be involved in early cellular responses of rat and porcine myocardium to ischemia. Immunostaining and subcellular fractionation revealed a rapid ischemia-induced redistribution of αB-crystallin from a cytosolic pool to intercalated disks and Z lines of the myofibrils. This striking translocation of αB-crystallin from the cytosol to sites of the myofibrillar system that are known to be sensitive to ischemiareperfusion injury was accompanied by a rapid shift of a fraction of αB-crystallin to a more acidic isoelectric point. This shift is caused by αB-crystallin phosphorylation, as identified by its augmentation in the presence of phosphatase inhibitors (vanadate, fluoride) and comigration of the acidic αB-crystallin form with the phosphorylated B1 form of lenticular αB-crystallin. In view of the chaperone-like function of αB-crystallin in conjunction with its high level of constitutive expression in the myocardium (1–2% of soluble protein content), we consider αB-crystallin an excellent candidate to play a role in early aspects of the protection of the myocardial contractile apparatus against ischemia-reperfusion injury.
- heat shock protein
the biochemical mechanismsunderlying ischemic myocardial cell injury and final cell death are still not fully understood, and there is also no generally accepted model to explain why cardiomyocytes exposed to transient sublethal ischemic episodes acquire a significant degree of ischemic tolerance when exposed later to a new ischemic stress, a phenomenon known as ischemic preconditioning (37). Understanding the molecular changes involved in ischemic myocardial cell injury and preconditioning would be an important step in developing therapeutic strategies aimed at increasing the survival time of the ischemic myocardium in patients at risk of infarction.
The rationale of the present study is based on two observations. First, it is becoming clear that the cytoskeleton of cardiac myocytes and other cell types is an important determinant for cellular survival during ischemia and energy depletion. During ischemia and subsequent reperfusion, disruption of the plasma membrane appears to be the ultimate cause of myocardial cell death, and this fatal incident is thought to result from disintegration of the membrane-bound cytoskeleton and subsequent destabilization of the plasmalemmal lipid bilayer (14, 41). Within the first minutes of reperfusion, cytoskeletal damage becomes obvious, such as detachment of actin filaments from the intercalated disks and disruption of I bands at the level of the Z lines (15). This dramatic cytoskeletal damage is probably caused by more subtle changes occurring during the preceding ischemic period.
The second observation relevant to the present study is the finding that a brief period of hyperthermia was associated with enhanced postischemic ventricular recovery in rats (9, 24, 35). Heat shock-induced thermotolerance or tolerance to energy deprivation has been observed in several other tissues and cell types, and this phenomenon is generally believed to be mediated at least in part by a class of proteins termed heat shock or stress proteins (13, 33). Many of these stress proteins have been shown to bind to proteins undergoing denaturation and to prevent them from unfolding and aggregating (49). One stress protein that has been shown to bind to components of the cytoskeleton, i.e., intermediate filaments (vimentin, desmin) and actin filaments, is αB-crystallin (2, 22, 38). αB-crystallin is an abundant protein of the ocular lens (5–10% of total protein content), where it forms soluble oligomers of up to 800 kDa. In the lens, αB-crystallin is believed to be involved in certain aspects important for the maintenance of lenticular transparency by prevention of protein denaturation and cytoskeletal disturbance (6, 51). Because αB-crystallin is abundant in cardiac tissue (4, 5, 12), we addressed the question of whether αB-crystallin may play a role in stabilization of myofibrils at the intercalated disks and Z lines during ischemia. In several cell types, in particular fibroblasts and glial cells, increased synthesis of αB-crystallin was observed in response to heat shock (20, 21, 25, 26) or other kinds of stresses (17, 27, 45). Because αB-crystallin is able to prevent heat-induced denaturation of proteins (e.g., alcohol dehydrogenase; Ref. 19), it has been suggested that heat-induced expression of αB-crystallin may play a role in certain aspects of cellular stress tolerance.
In homogenates of ischemic rat hearts, αB-crystallin was shown to disappear from the soluble cytosolic protein pool (8). More recently, it was shown that αB-crystallin is a component of a 23-kDa band of myofibrils in the ischemic rat heart (1). This band was previously considered to be a fragment of troponin (50). The mechanism responsible for translocation of αB-crystallin from a soluble pool to the myofibrillar fraction during ischemia still remains an open question. In the present study, we show an ischemia-induced phosphorylation of cardiac αB-crystallin that is accompanied by a complete shift of αB-crystallin from the soluble to the insoluble protein fraction. By immunostaining we found that during ischemia αB-crystallin becomes translocated to intercalated disks and myofibrillar Z lines, two cytoskeletal structures known to become destabilized during prolonged ischemia. Translocation of αB-crystallin to the cytoskeleton of ischemic myocardiocytes may be a physiologically important mechanism that helps to delay ischemic cell damage.
All animals in this study were handled in accordance with the “Guiding Principles in the Care and Use of Animals” as approved by The American Physiological Society, and the investigation conformed with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and was approved by the Bioethical Committee of the District of Darmstadt, Germany.
Adult male Wistar rats (8–20 wk old) were decapitated under ether anesthesia and placed in a 37°C warm box. The hearts were removed immediately (within 1 min) or after 5, 15, or 30 min and then transferred to cold (4°C) phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 8.1 mM Na2HPO4, and 1.5 mM KH2PO4, pH 7.4). For each time point, hearts of three to five rats were examined. Tissue samples from the left ventricular myocardium were excised with a scalpel and either frozen for immunohistochemical studies or homogenized for biochemical analysis. Homogenization was performed with a Braun homogenizer (Melsungen, Germany) using a 0.5-ml volume of 20 mM tris(hydroxymethyl)aminomethane ⋅ HCl (pH 7.4) containing a mixture of various protease inhibitors (pepstatin, aprotinin, and leupeptin, 2 μg/ml each; Sigma, Deisenhofen, Germany). The homogenate was further separated into a cytosolic supernatant and a myofibril-containing pellet by centrifugation at 3,000g for 10 min.
Six male landrace-type domestic pigs (34.6 ± 2.7 kg) were premedicated with azaperone (2 mg/kg body wt im) and piritramide (2 mg/kg body wt sc) 30 min before the initiation of anesthesia with metomidate (10 mg/kg body wt). After tracheal intubation, a bolus (25 mg/kg body wt) of α-chloralose was given intravenously. Anesthesia was maintained by a continous intravenous infusion of α-chloralose (25 mg/kg body wt). The animals were ventilated artificially with a pressure-controlled respirator (Stephan Respirator ABV, F. Stephan, Quickborn, Germany) with room air enriched with 2 l/min oxygen. Arterial blood gases were analyzed frequently to guide adjustment of the respirator settings. Additional doses of piritramide (10 mg) were given intravenously every 60 min. Both internal jugular veins were cannulated with polyethylene tubes for administration of saline, piritramide, and α-chloralose. Arterial sheath catheters (7-Fr) were inserted into both common carotid arteries. To measure aortic blood pressure, the left sheath was advanced into the aortic arch and connected with a Statham transducer (P23XL, Statham). The chest was opened by midsternal thoracotomy, and the heart was suspended in a pericardial cradle. A loose ligature was placed halfway around the left anterior descending artery (LAD) and was subsequently tightened to occlude the vessel. Drill biopsies of the ischemic area were obtained at 0 min and 25 min of the ischemic period. Biopsies of the nonischemic myocardium taken 25 min after LAD occlusion served as internal controls. Tissue samples were processed as described for rats.
Antibodies and immunostaining.
A polyclonal antibody specific for αB-crystallin was raised in a rabbit immunized with purified αB-crystallin from the bovine lens. Purification of αB-crystallin included gel filtration (Sepharyl S-300) and subsequent anion exchange chromatography (DEAE-cellulose) of bovine lens homogenates, as described in detail elsewhere (47). For immunostaining, antibodies were affinity purified to their antigen adsorbed to nitrocellulose (10). A second polyclonal antibody specific for αB-crystallin was raised in a mouse. This antibody was applied for double-immunofluorescence studies in combination with a polyclonal rabbit antibody against chicken striated muscle α-actinin (11). Five-micrometer-thick cryostat sections were fixed and permeabilized with acetone. After preincubation of the sections with 2% bovine serum albumin and 1% normal goat serum in PBS for 30 min at room temperature (RT), the affinity-purified primary antibody was applied at 4°C overnight (∼1 μg/ml PBS). After being washed with PBS (3 × 5 min), sections were incubated for 30 min at RT with Cy3-labeled goat anti-rabbit IgG (Dianova, Hamburg, Germany; dilution 1:800 in PBS). After final rinsing for 3 × 5 min with PBS, sections were mounted in 60% glycerol in PBS containing 1.5%n-propylgallate as antifading compound. For double immunofluorescence, sections were incubated overnight at 4°C with a mixture of the polyclonal αB-crystallin antibody (mouse) and the polyclonal α-actinin antibody (rabbit). Final concentration of both antibodies in this mixture was 1:100. As secondary antibodies, a mixture was used containing fluorescein isothiocyanate-labeled anti-mouse IgG (Dianova, final dilution 1:100) and Cy3-labeled goat anti-rabbit IgG (Dianova; final dilution 1:800). Incubation with these secondary antibodies was performed for 30 min at RT. Controls were performed by preabsorption of the polyclonal αB-crystallin antibodies (mouse, rabbit) with an excess of purified bovine lenticular αB-crystallin.
Electrophoresis and immunoblotting.
Samples of myocardial fractions and purified proteins were dissolved in Laemmli sample buffer (29) and subjected to sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) (13% polyacrylamide). Protein contents were quantified according to Heinzel et al. (18). For immunoblotting, proteins were transferred in Kyhse-Andersen transfer buffer (28) to Hybond enhanced chemiluminescence (ECL) membranes (Amersham, Braunschweig, Germany) that were blocked with 5% low-fat milk in PBS for 3 h at RT and incubated overnight at 4°C with the polyclonal αB-crystallin antibody at a dilution of 1:1,500 (in PBS plus 5% low-fat milk). As secondary antibody, horseradish peroxidase-labeled goat anti-rabbit IgG (Bio-Rad, Munich, Germany) was used at a dilution of 1:3,000. Bound immunoglobulins were visualized by the ECL technique (Amersham). Two-dimensional gel electrophoresis (2D-GE) was performed essentially according to the method of O’Farrell (39) using the Mini Protean 2D-system of Bio-Rad. Samples of homogenates of the left ventricular myocardium (rat) or bovine lens were solubilized in 9.8 M urea, 2% Nonidet P-40 (NP-40), 2% ampholytes pH 7–9 (Serva, Heidelberg, Germany) and 100 mM dithiothreitol (DTT). Ten micrograms of total protein in this sample solution were loaded onto the urea-acrylamide tube gels [9 M urea, 1.5% ampholytes pH 3–10 (Serva), 4.5% ampholytes pH 5–9 (Serva), 2% NP-40, 4% (wt/vol) polyacrylamide], covered with 25 μl of overlay solution (8 M urea, 2% ampholytes pH 5–9, 5% NP-40, and 100 mM DTT), and focused to their isoelectric points. Tube gels were then mounted on top of 13% SDS-polyacrylamide minislab gels and subjected to SDS-PAGE and immunoblotting.
αB-crystallin in normal rat myocardium.
The polyclonal antibody raised against bovine lenticular αB-crystallin reacted specifically with lenticular αB-crystallin (21-kDa band) but not with αA-crystallin. This was shown by SDS-PAGE and subsequent immunoblotting of homogenates of the bovine lens and of purified bovine lenticular αB- and αA-crystallin (Fig.1). In homogenates of the nonischemic rat heart, the αB-crystallin antibody reacted selectively with a 21-kDa band (further described as cardiac αB-crystallin) that comigrated with αB-crystallin of bovine lens in immunoblots of both SDS-PAGE and 2D-GE. In the nonischemic rat myocardium, the αB-crystallin spot displayed an isoelectric point of pI ∼7 that migrated to the same position as the lenticular B2 form of αB-crystallin (Fig. 2). The more acidic phosphorylated B1 forms known to occur in the lens (7, 16, 23, 42, 48) were not detected in the nonischemic rat myocardium.
The total pool of αB-crystallin in the nonischemic rat myocardium is present in the soluble cytoplasmic compartment that remains in the 3,000 g and 100,000g supernatants. The amount of αB-crystallin in the rat myocardium was determined by integrated densitometry of immunolabeled bands using purified αB-crystallin as protein standard (Fig. 3). On the basis of this assay, αB-crystallin was found to make up ∼0.6% of total cellular protein and ∼1.5% of the soluble protein fraction.
Subcellular distribution of αB-crystallin in the rat myocardium.
Cryosections of the rat myocardium were fixed and permeabilized with acetone and incubated with affinity-purified immunoglobulins specific for αB-crystallin using the indirect immunofluorescence technique. The following staining pattern was observed in the nonischemic myocardium (Fig.4 a): faint staining of some myofibrillar Z lines and moderate degree of staining of intercalated disks. Fifteen minutes after the onset of global ischemia, striking changes of this immunostaining pattern were observed, characterized by strong fluorescence of virtually all myofibrillar Z lines and significant increase of the staining intensity of intercalated disks (Fig.4 b). The translocation of αB-crystallin to Z lines and intercalated disks was further confirmed by double-labeling experiments with a rabbit polyclonal antibody to α-actinin (marker protein for Z lines and intercalated disks) and a mouse polyclonal αB-crystallin antibody, revealing precise codistribution of both proteins (Fig. 4,c andd). These ischemia-induced changes in the distribution of αB-crystallin were already visible after 5 min and were maximally developed after 30 min of ischemia. Absorption of the αB-crystallin antibody with an excess of purified bovine lenticular αB-crystallin completely abolished immunostaining of tissue sections (not shown).
The striking subcellular shift of αB-crystallin during ischemia to myofibrils and intercalated disks was also reflected by sedimentation experiments in which samples of the left ventricular myocardium were homogenized and then centrifuged at 3,000g for 10 min. Supernatants and the myofibril-containing pellets were subjected to SDS-PAGE and immunoblotting (Fig. 5). In the nonischemic myocardium, almost the entire cellular pool of αB-crystallin was found in the supernatant. As early as 5 min after the onset of global ischemia, a fraction of αB-crystallin appeared in the pellet. This shift of αB-crystallin from the supernatant to the pellet was completed 30 min after the onset of ischemia, when the entire amount of αB-crystallin occurred in the pellet and was absent from the supernatant. Immunostaining of samples of the pellet spread on glass slides showed that αB-crystallin was mainly bound to myofibrils in a Z-line-like pattern (not shown).
Ischemia-induced phosphorylation of cardiac αB-crystallin in the rat myocardium.
Tissue samples of the left ventricular myocardium were subjected to 2D-GE and analyzed for αB-crystallin by immunoblotting after different periods of global ischemia (5, 15, and 30 min) (Fig.6). As stated above, in the nonischemic myocardium αB-crystallin migrated as a single protein spot that corresponded to the B2 form of lenticular αB-crystallin. As early as 5 min after the onset of ischemia, a more acidic spot of αB-crystallin was formed that comigrated with the major B1 form of lenticular αB-crystallin (Fig. 6). Comigration was shown by adding purified lenticular αB-crystallin to the myocardial samples (Fig.7). During prolonged ischemia, the B1 form of cardiac αB-crystallin increased considerably and reached its maximal value after 15–30 min of global ischemia, when the size and density of the B1 spot made up ∼15% of the total signal for αB-crystallin (5 min: 8.2 ± 0.6%; 15 min: 16.6 ± 3.4%; 30 min: 16.2 ± 4.3%;n = 4). Because the lenticular B1 forms of αB-crystallin have been shown to be phosphorylated, it is likely that the B1 form of cardiac αB-crystallin observed in the ischemic myocardium is phosphorylated as well. A further proof for the ischemia-induced phosphorylation of αB-crystallin was obtained by phosphatase inhibitors (fluoride, vanadate) added to homogenate of the nonischemic myocardium incubated in vitro for 30 min at 37°C (Fig. 8). As in global ischemia in situ, these in vitro conditions also gave rise to the formation of the B1 form of αB-crystallin in a time-dependent way. In the absence of phosphatase inhibitors, the size of the B1 spot was significantly smaller than in the presence of these inhibitors.
αB-crystallin in normal and ischemic porcine myocardium.
Our observations on translocation and phosphorylation of αB-crystallin during global (supravital) ischemia of the rat heart were supplemented with an in vivo model of myocardial infarction of the pig heart. Biopsies taken at 0 and 25 min after occlusion of the LAD were examined by immunoblotting, 2D-GE, and immunostaining. In the nonischemic myocardium (taken before LAD occlusion as well as 25 min after occlusion but outside the ischemic area), the bulk of αB-crystallin was present in the soluble fraction. However, 25 min after the onset of ischemia a complete shift of αB-crystallin to the insoluble myofibrillar fraction was observed (Fig.9 a). Analysis of the biopsies by 2D-GE confirmed the results obtained with the ischemic rat myocardium in that the shift of αB-crystallin from the cytosolic to the myofibrillar fraction was accompanied by the appearance of the major phosphorylated B1 form of αB-crystallin (Fig.9, b andc). In addition, the ischemic myocardium contained small amounts of the minor phosphorylated B1 form known to occur in the lens but not found in the rat myocardium. Immunostaining fully confirmed the results obtained with global ischemia of the rat heart (not shown). First results obtained with reperfusion experiments show that the translocation of αB-crystallin to the myofibrils is fully reversible under conditions in which the ischemic period (up to 25 min) does not cause irreversible myocardial lesions.
Findings obtained with supravital ischemia of the rat heart and in vivo ischemia of circumscribed areas of the porcine myocardium (infarction model) were virtually identical. Therefore, the results obtained with both animal models are discussed collectively.
In both the rat and porcine myocardium, we found an ischemia-induced redistribution of αB-crystallin from a cytosolic localization to intercalated disks and Z lines of myofibrils. This striking subcellular translocation from cytosol to myofibrils was shown by both immunostaining and cosedimentation of αB-crystallin with the myofibril-containing pellet of the ischemic myocardium. Although in the nonischemic myocardium the entire pool of αB-crystallin remained in the 3,000g supernatant, a shift of a fraction of αB-crystallin to the pellet already was detected a few minutes after the onset of ischemia, and during the following 25–30 min the total amount of αB-crystallin was found in the pellet and was absent from the supernatant. In cultured adult rat cardiomyocytes, localization of αB-crystallin at myofibrillar Z lines has been described to occur already under control conditions (34). However, in view of the present observation, it should be emphasized that isolation of cardiomyocytes and subsequent culturing must be considered stress conditions that are probably strong enough to induce translocation of αB-crystallin to the myofibrils.
Preferential binding of αB-crystallin during ischemia at Z lines and intercalated disks is of particular interest in view of the observation that these sites of myofibrillar actin filament anchorage and crosslinking are particularly sensitive to ischemia-induced alterations (14, 15, 44). Because αB-crystallin can prevent denaturation of proteins (chaperone-like function; Ref. 19), it is tempting to assume that αB-crystallin translocated to intercalated disks and Z lines might assist certain cytoskeletal components of these structures to maintain their correctly folded state and to prevent them from unfolding and loss of function. By such a mechanism, αB-crystallin could play an important role in stabilizing Z lines and intercalated disks during the adverse conditions of prolonged ischemia. Support for this view comes from in vitro studies showing that αB-crystallin binds to actin and desmin filaments and to the desmin-related vimentin filaments (2, 38). Furthermore, it was shown that the affinity of αB-crystallin to desmin and actin filaments is higher at low pH (2). Because ischemia is associated with an intracellular drop in pH (40) and, furthermore, because actin and desmin are components of intercalated disks and Z lines, these proteins are potential candidates for αB-crystallin binding at these structures. However, actin filaments alone may not serve as a primary binding site for αB-crystallin under ischemia, as indicated by the absence of I-band staining under these conditions. If actin is involved in Z-line (intercalated disk) binding of αB-crystallin during ischemia, the formation of ternary complexes between αB-crystallin, actin, and further components of these myofibrillar structures must be postulated. On the other hand, binding of αB-crystallin to desmin filaments would be sufficient to explain why αB-crystallin becomes localized during ischemia at Z lines and intercalated disks. However, in contrast to actin and α-actinin, desmin is not a component of the Z lines themselves but rather is restricted to their periphery (32). Therefore, the precise colocalization of α-actinin and αB-crystallin during ischemia would be difficult to explain by an exclusive binding of αB-crystallin to desmin.
The related small heat shock protein HSP 27 provides another example of a stress protein that acts on the cytoskeleton, particularly on the actin filament system (30, 36). However, the protective action of HSP 27 on actin filaments during ischemia appears to be exerted mainly by a stimulation of actin polymerization rather than by protection of actin filaments from denaturation (3). Whether αB-crystallin may also control the state of polymerization of actin filaments remains to be determined. Regarding intermediate filaments, recent in vitro studies showed that αB-crystallin is capable of inhibiting the assembly of vimentin and glial fibrillar acidic protein (38). Both proteins belong to the type III class of intermediate filaments as does desmin, the intermediate filament protein of cardiomyocytes. Thus it is possible that αB-crystallin may affect the state of assembly of cardiac desmin and that such a function may have certain implications for ischemic cardioprotection.
The signaling mechanism that causes ischemia-induced translocation of αB-crystallin to myofibrils is still unknown. However, analysis of αB-crystallin by 2D-GE during various periods after the onset of ischemia revealed the occurrence of a more acidic form of αB-crystallin. The time course of the formation of acidic αB-crystallin was more or less identical to the time course determined for the disappearance of αB-crystallin from the cytosolic pool. For two reasons, we assume that the shift of a fraction of αB-crystallin to a more acidic isoelectric point is caused by a phosphorylation event. First, acidic cardiac αB-crystallin comigrated with the B1 form of lenticular αB-crystallin that has been shown to result from phosphorylation of αB-crystallin (7, 48). Second, there is an increase of the fraction of acidic myocardial αB-crystallin in the presence of the phosphatase inhibitors vanadate and fluoride. Interestingly, only a small ∼15% fraction of αB-crystallin became phosphorylated during ischemia. Because αB-crystallin has been shown to form oligomers [up to 800 kDa in the lens (43, 46, 51) and 300–400 kDa in the heart (4)], it is possible that in cardiac myocytes phosphorylation of αB-crystallin may play a role in the regulation of oligomer formation. Unfortunately, the oligomeric state of αB-crystallin in ischemia could not be determined because of its insolubility under these conditions. With respect to HSP 27, phosphorylation has been shown to cause a dramatic reduction of the size of the HSP 27 oligomers (31). This, in turn, was associated with an increase in the amount of filamentous actin (3). Likewise, phosphorylation of αB-crystallin may also give rise to smaller (or larger) oligomers that might have a higher binding affinity for components of Z lines and intercalated disks than the oligomer species of αB-crystallin present under nonischemic conditions. Thus the phosphorylation of only a small fraction of αB-crystallin could have a significant effect on either the formation of oligomers or conformational changes of preexisting oligomers that may render them more competent for binding to myofibrillar components.
On the other hand, we must consider the possibility that ischemia-induced phosphorylation of αB-crystallin is not causally involved in its translocation to the myofibrils. If this is the case, we must postulate posttranslational modifications of components of Z lines and intercalated disks or binding of still-unknown adaptor molecules to these structures to explain why the entire pool of αB-crystallin becomes attached to the myofibrils during ischemia.
Heat shock exposure has been shown to reduce infarct size and to improve postischemic ventricular recovery (9, 24, 35). Heat shock proteins are generally assumed to be involved in this process, but the delay required for their synthesis (lag time of at least 1 h) makes it difficult to imagine how heat shock proteins can contribute to the short-term phenomenon of ischemic preconditioning, in which short, repetitive periods of ischemia induce a significant degree of tolerance against a subsequent longer ischemic period (37). In contrast, αB-crystallin would provide a good candidate for involvement in early steps of ischemic preconditioning, mainly because of its constitutive abundance in cardiomyocytes (1–2% of the soluble protein content) and its rapid ischemia-induced translocation to the contractile apparatus.
In summary, the present study provides evidence for a rapid ischemia-induced phosphorylation of αB-crystallin that accompanies its translocation from the cytosol to myofibrillar Z lines and intercalated disks. The functional importance of phosphorylation and translocation to myofibrils still remains to be determined. We assume that because of its well-known chaperone-like function αB-crystallin is an excellent candidate to play a role in early aspects of ischemic cardioprotection.
Address for reprint requests: D. Drenckhahn, Institute of Anatomy, Julius-Maximilians-Univ., Koellikerstr. 6, D-97070 Würzburg, Germany.
- Copyright © 1998 the American Physiological Society