Loss of mechanical loading induces rapid bone loss resulting from reduced osteoblastogenesis and decreased bone formation. The signaling mechanisms involved in this deleterious effect on skeletal metabolism remain poorly understood. We have previously shown that hindlimb suspension in rats increases osteoblast apoptosis associated with decreased phosphatidylinositol 3-kinase (PI3K) signaling. In this study, we investigated whether transforming growth factor (TGF)-β2 may prevent the altered signaling and osteoblast apoptosis induced by skeletal unloading in vivo. Hindlimb suspension-induced decreased bone volume was associated with reduced α5β1-integrin protein levels and PI3K/Akt signaling in unloaded bone. Continuous administration of TGF-β2 using osmotic minipumps prevented the decreased α5β1-integrin expression and the reduced PI3K/Akt signaling in unloaded bone, resulting in the prevention of osteoblast apoptosis. We also show that TGF-β2 prevented the decreased Bcl-2 levels induced by unloading, which suggests that TGF-β2 targets Bcl-2 via PI3K/Akt to prevent osteoblast apoptosis in unloaded bone. Furthermore, we show that TGF-β2 prevented the decrease in phosphorylated Bad, the inactive form of the proapoptotic protein Bad, induced by unloading. These results identify a protective role for TGF-β2 in osteoblast apoptosis induced by mechanical unloading via the α5β1/PI3K/Akt signaling cascade and downstream Bcl-2 and phospho-Bad survival proteins. We thus propose a novel role for TGF-β2 in protection from unloading-induced apoptosis in vivo.
- phosphatidylinositol 3-kinase
- bone loss
bone loss in immobilized patients is a growing problem in modern countries in which the aging population is increasing (24). Immobilized patients show reduced bone formation and increased bone resorption, resulting in rapid bone loss (62). The mechanisms involved in these effects remain, however, poorly known. Osteopenia induced by hindlimb suspension in rats results mainly from decreased bone formation (10, 46, 48). In this model, the altered bone formation results from reduced recruitment of osteoblast precursors (6, 37, 42) associated with altered osteoblast differentiation and function (2, 9, 25, 42, 66). Osteoblast apoptosis is an important mechanism controlling bone formation and bone mass (31, 44, 50, 65). Recent studies (1, 23, 53) indicate that alteration of osteoblast/osteocyte survival may contribute to the diminished bone formation in unloaded bone. Osteoblast survival is in part dependent on cell attachment, which is mediated by integrin-extracellular matrix interactions (16). Interestingly, skeletal unloading was found to decrease αvβ3 mRNA expression in osteoprogenitor cells, resulting in altered activation of IGF-I signaling and osteoblast proliferation (59). Additionally, we showed that skeletal unloading in rats decreases α5β1-integrin expression and increases osteoblast apoptosis (23), suggesting that unloading alters integrin-mediated signals that control osteoblast survival. In vitro studies (7, 12, 28, 29, 33) have shown that mechanical stress induces integrin-mediated activation of focal adhesion kinase (FAK), ERK p42/44, and phosphatidylinositol 3-kinase (PI3K). In vivo, we recently found that skeletal unloading decreases phosphorylated PI3K (p-PI3K) levels, whereas p-FAK and p-ERK p42/44 levels are not significantly altered, possibly implicating decreased PI3K signaling in osteoblast apoptosis induced by unloading (23). However, the signaling mechanisms acting downstream of PI3K and controlling osteoblast survival are yet to be identified.
TGF-β is an important modulator of bone formation (14, 38). Binding of TGF-β to its cell surface receptor triggers several signaling pathways, resulting in modulation of osteoblastogenesis. Both TGF-β1 and TGF-β2 are expressed in bone, although at variable levels depending on the bone site (30). Bone cells may respond differentially to TGF-β1 and TGF-β2 as a result of different affinity of the ligands for TGF-β receptors (14). The effects of TGF-β on cell death were shown to be highly variable (60). In bone, TGF-β1 was found to decrease activated apoptosis in osteoblasts (11, 15, 27, 35). In contrast, increased TGF-β signaling promotes bone marrow-derived osteoprogenitor cell apoptosis in vivo (8), indicating the complex regulatory effects of TGF-β on bone cell fate. Several signaling pathways can be involved in the control of cell apoptosis by TGF-β (60). In some systems, TGF-β was found to activate PI3K/Akt (63). TGF-β was also found to modulate the expression of α2-, α5-, β1-, and β5-integrins, which are involved in cell attachment in osteoblasts (29, 49, 57, 58). The implication of these signaling mechanisms in the control of osteoblast survival by TGF-β in vivo remains unknown.
We previously showed that TGF-β2 administration can prevent the alteration of osteoblast proliferation and differentiation induced by skeletal unloading (2, 3, 22, 43). Here, we determined whether TGF-β2 administration can prevent osteoblast apoptosis induced by unloading. We also investigated the potential role of α5β1-integrin and PI3K/Akt signaling in the skeletal response to TGF-β2 and analyzed the downstream survival signaling pathways in vivo. We report here that TGF-β2 positively regulates survival signaling pathways to prevent osteoblast apoptosis in unloaded bone via upregulation of α5β1-integrin, PI3K/Akt signaling, and downstream survival proteins.
Animals and hindlimb unloading.
Fifty five adult 4-wk-old Wistar male rats weighing 130 g (CER Janvier) were randomly assigned to three groups. Animals were not suspended (loaded), suspended by the tail (unloaded) for up to 7 days, or suspended and treated with TGF-β2 (kindly provided by Genzyme, Framingham, MA; 2 mg/kg/day) administered by Alzet minipump (Charles River) after approval of the study by the IMASSA Animal Care Committee. The base of the tail was attached via a clip to the top of a specially designed Plexiglass cage (CERMA-Biomeca) to have hindlimbs nonweight bearing. The rats were maintained on a 12:12-h light-dark cycle and were fed a standard chow containing 1% calcium and 0.8% phosphorus (UAR). At 2, 4, and 7 days postunloading, the animals (5 rats per group per time point) were anesthetized and killed (23, 43).
At death, the right tibia metaphysis was processed for histomorphometric analysis. The proximal halves of the tibias were fixed in 10% phosphate-buffered formaldehyde, dehydrated in methanol, and embedded in methylmetacrylate resin without decalcification (23). Undecalcified 5-μm-thick longitudinal sections were prepared using a Leica 2055 microtome (Leica, Paris, France) equipped with a tungsten carbide blade. Sections were stained with Goldner trichrome and histomorphometric indexes were measured under blind conditions (23) using a Leitz Aristoplan microscope (Leica) connected to a Sony DXC-930P color video camera. An automatic image analyzer (Microvisions Instruments, Evry, France) was used for bone volume measurements. Parameters of bone formation such as the osteoid surface (with and without osteoblasts), the osteoblast surface, and the osteoid thickness (51) were measured in the secondary metaphyseal area of the proximal end of the tibia in a standardized zone located 400–600 μm from the growth cartilage and 200–300 μm from the cortices.
Terminal deoxynucleotidyl transferase dUTP-mediated nick-end labeling assay.
The right femurs were fixed in paraformaldehyde, decalcified and embedded in paraffin and thin sections were processed for terminal deoxynucleotidyl transferase dUTP-mediated nick-end labeling (TUNEL) staining according to the manufacturer's recommendations (Roche Diagnostics). Positive staining consisted of sections treated with DNase I (Sigma, St Louis, MO). Negative controls consisted of sections that were not incubated with the substrate. Apoptotic cells were characterized by TUNEL-stained nucleus with condensed chromatin. The number of apoptotic osteoblasts was quantified in the metaphysis in the same area as for osteoblast surface and bone volume and was evaluated as a percentage of the total osteoblast number (23).
Western blot analyses.
The metaphyseal area in left femurs was dissected in sterile conditions and used for cell protein extraction (23). Cell proteins in the metaphysis were extracted in lysis buffer consisting of 50 mM Tris-HCl (pH 7.2), 150 mM NaCl, 1% NP-40, 1 mM PMSF, 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 2 mM Na3VO4. Lysates were clarified by centrifugation at 12,000 g for 10 min at 4°C, and the protein content of the supernatants was determined using the DC protein assay (Bio-Rad Laboratories). Equal aliquots (50 μg) of protein lysates were then resolved on 8–16% SDS-PAGE gel (Euromedex) and transferred onto polyvinylidene difluoride membrane (Perkin Elmer). The membranes were incubated overnight with 5% blocking solution and then with anti-α5-integrin (sc-10729, Santa Cruz), anti-β1-integrin (sc-8978, Santa Cruz), anti-ERK (sc-1647, Santa Cruz), anti-phospho-ERK1/2 (sc-7383, Santa Cruz), anti-PI3K (4292, Cell Signaling), anti-phospho-PI3K (sc-12929, Santa Cruz), anti-Akt (9272, Cell Signaling), anti-p-Akt (9271, Cell Signaling), anti-FAK (sc-1688, Santa Cruz), anti-phospho-FAK (3284, Cell Signaling), anti-Bax (sc-7480, Santa Cruz), anti-Bcl-2 (ab7973, AbCam), anti-Bad (9292, Cell Signaling), anti-phospho-Bad (9295, Cell Signaling), or anti-β-actin antibodies (Sigma) diluted 1/500. Membranes were washed and incubated with specific peroxydase-coupled appropriate secondary antibodies, and the signal was visualized by chemiluminiscence detection system (Perbio). Two separate runs were made, and autoradiographs were scanned for quantification and corrected for loading using β-actin.
Data were analyzed by one-factor ANOVA with post hoc testing (Bonferroni) for significance. A minimal level of P < 0.05 was considered significant.
TGF-β2 prevents the altered bone formation in unloaded bone.
As expected, skeletal unloading significantly reduced metaphyseal bone volume in unloaded tibial metaphysis compared with loaded rats at 4 days (28.3 ± 0.6 vs. 35.2 ± 0.8%; P < 0.05) and 7 days (25.3 ± 0.6 vs 32.4 ± 0.3%; P < 0.05; Fig. 1A), confirming our previous results in young rats (23, 43). Quantitative analysis showed that this effect was associated with reduced osteoblast surface, reflecting decreased osteoblast number, and reduced osteoid surface, reflecting decreased bone matrix synthesis, which resulted in decreased trabecular thickness in the tibial metaphysis (Fig. 1B). Treatment with TGF-β2 prevented the decreased osteoblast surface, osteoid surface, and trabecular thickness induced by unloading (Fig. 1B). These results support our previous study (43), indicating that the altered bone formation induced by skeletal unloading can be prevented by TGF-β2 supplementation and that this effect results from a positive effect of TGF-β2 on osteoblastogenesis.
TGF-β2 protects against osteoblast apoptosis in unloaded bone.
We then analyzed whether the beneficial effect of TGF-β2 on osteoblastogenesis may be at least in part mediated by protection against osteoblast apoptosis. TUNEL analysis that identifies apoptotic cells in vivo (23) showed that skeletal unloading induced a twofold increase in the number of osteoblasts undergoing apoptosis in the metaphyseal area compared with loaded bones (Fig. 2). The increased ratio of apoptotic osteoblasts in unloaded rats was due to both an increase in apoptotic cells and a decrease in the number of total osteoblasts. Strikingly, TGF-β2 administration prevented the increased number of TUNEL-stained cells in unloaded bone (Fig. 2). The preventive effect of TGF-β2 occurred as soon as 2 days and was sustained for the 7-day study (Fig. 2). These data show that the rapid increase in osteoblast apoptosis induced by skeletal unloading in the metaphyseal femur can be prevented by TGF-β2, indicating that TGF-β signaling can protect against unloading-induced osteoblast apoptosis in vivo.
TGF-β2 prevents the altered α5β1-integrin expression in unloaded bone.
One important effect of TGF-β in vitro is to regulate cell-matrix interactions through modulation of integrin expression and organization (29, 49, 58). We therefore assessed whether TGF-β2 may act by modulating integrin expression in vivo. We focused our analysis on the changes in α5β1-integrin that we previously found to be involved in osteoblast adherence and survival in vitro (34). As shown in Fig. 3A, skeletal unloading decreased α5-integrin levels in the femur metaphysis compared with loaded femur at 2 days. This effect was attenuated at 4 and 7 days. Additionally, β1-integrin levels were decreased by 40–50% in unloaded bone at 2 and 4 days and by ∼10% at 7 days (Fig. 3A). Strikingly, we found that TGF-β2 administration prevented the decrease in α5- and β1-integrin subunit levels despite unloading (Fig. 3B). These data show that the altered α5β1-integrin expression in metaphyseal bone induced by skeletal unloading can be corrected by TGF-β2 signaling in vivo.
TGF-β2 prevents the altered PI3K/akt signaling induced by skeletal unloading.
We next examined whether the effect of TGF-β2 on α5β1-integrin expression can transduce in signaling events in vivo. Integrins are known to induce intracellular signals in part by activating kinases and downstream signaling effectors. We first examined the changes in FAK and ERK p42/p44, which are integrin-dependent intracellular kinases that are activated by mechanical forces in vitro (4, 12, 61). We found that FAK and ERK p42/p44 were unaffected by TGF-β2 supplementation (data not shown). PI3K/Akt signaling is another important signaling pathway involved in osteoblast survival in vitro (5, 20). As shown in Fig. 4, p-PI3K/PI3K levels were decreased in unloaded bone compared with loaded bone at all time points. TGF-β2 administration prevented the decrease in p-PI3K/total PI3K levels induced by unloading (Fig. 4B), indicating that TGF-β2 targets PI3K signaling in vivo. PI3K activates Akt, a serine/threonine kinase that promotes cell survival through inactivation of proapoptotic proteins (17). We therefore examined whether TGF-β2-induced activation of PI3K results in Akt phosphorylation in vivo. We found that p-Akt levels were decreased at 2–7 days of unloading and that treatment with TGF-β2 prevented the decrease in p-Akt levels induced by unloading (Fig. 4B). These results indicate that skeletal unloading alters PI3K/Akt signaling in metaphyseal bone cells and that the altered cell signaling in unloaded bone can be prevented by TGF-β2 in vivo.
TGF-β2 upregulates survival proteins in unloaded bone.
Having shown that TGF-β2 activates PI3K/Akt signaling in vivo, we examined the changes in signaling molecules acting downstream of PI3K/Akt signaling to modulate osteoblast survival. Akt might act to promote survival by regulating the expression or activity of Bcl-2 family members (18). We found that Bax levels were not altered in unloaded metaphyseal bone, whereas Bcl-2 levels were decreased at 4 and 7 days compared with loaded bones (Fig. 5A). Treatment with TGF-β2 prevented the decrease in Bcl-2 levels induced by skeletal unloading. This effect resulted in normalization of the Bax-to-Bcl-2 ratio (Fig. 5B).
Finally, we analyzed whether bone cell survival induced by Akt activation by TGF-β2 in vivo implicates modulation of the proapoptotic Bad, a member of the Bcl-2 family (17). Akt phosphorylation of Bad at Ser-136 leads to inactivation of Bad and to cell survival, whereas dephosphorylation of Bad results in targeting of Bad to mitochondrial membranes to induce cell death (18). We found that skeletal unloading altered total Bad levels and p-Bad levels in metaphyseal bone compared with control bones (Fig. 6A). Quantitative analysis showed that unloading decreased the p-Bad-to-Bad ratio (Fig. 6B), which is of physiological significance. Furthermore, treatment with TGF-β2 prevented the decreased in p-Bad-to-Bad ratio in unloaded bone (Fig. 6B). These results suggest that Bad is involved in unloading-induced apoptosis and that Bad phosphorylation induced by TGF-β2 contributes to prevent the rapid osteoblast apoptosis in unloaded bone. Altogether, our data indicate that skeletal unloading alters PI3K/Akt signaling, resulting in decreased Bcl-2 expression and Bad phosphorylation in vivo. Furthermore, TGF-β2 can correct the deranged PI3K/Akt signaling that targets the survival proteins Bcl-2 and phospho-Bad, resulting in the prevention of osteoblast apoptosis (Fig. 7).
Understanding the mechanisms leading to osteoblast survival is key to the regulation of bone mass. This study shows that TGF-β2 prevents skeletal unloading-induced osteoblast apoptosis by targeting PI3K/Akt signaling and downstream survival proteins. This is the first study to demonstrate the anitapoptotic effect of TGF-β2 on osteoblast apoptosis in vivo via PI3K/Akt signaling and Bcl-2 family proteins in vivo.
Skeletal unloading is know to induce age-specific changes in bone metabolism (10, 48). Young rats undergo a transient suppression of skeletal growth, whereas older animals actually lose bone (10, 48) as a result of decreased bone formation. One mechanism that may contribute to the altered bone formation induced by unloading is osteoblast apoptosis, which reduces the life span of osteoblasts (44). Consistent with our previous study (23), we show here that skeletal unloading increases the number of apoptotic osteoblasts in young growing animals. Although the percentage of apoptotic osteoblasts was low, the cumulative number of apoptotic osteoblasts may contribute to reduce the number of functional osteoblasts in vivo (44). The resulting effect on bone formation is likely to be amplified due to the high rate of bone formation in these young animals. We previously found that unloading-induced osteoblast apoptosis in this model is glucocorticoid independent (23), suggesting that local mechanisms cause cell apoptosis. One possible mechanism by which unloading may induce apoptosis may be a decreased growth factor expression and/or signaling in bone. Notably, we and others (22, 59, 66) have shown that skeletal unloading is associated with the altered signaling of IGF-I and TGF-β, two major factors regulating osteoblastogenesis. Here we tested the hypothesis that TGF-β2 administration may prevent osteoblast apoptosis in unloaded bone. The dose of TGF-β2 was chosen on the basis of our previous studies (43) showing that it was effective in preventing the alteration of bone formation in unloaded rats, whereas this dose had no effect on bone formation in normally loaded animals. We show that TGF-β2 administration was effective to prevent cell death in unloaded bone; however, it remains unknown whether osteoblast apoptosis was caused by altered TGF-β signaling or upstream events induced by unloading. Nevertheless, these data indicate that, in addition to prevent the deranged osteoblast recruitment and differentiation induced by unloading (2, 43), TGF-β2 administration can provide protection from osteoblast apoptosis in young adult rats. Given the different response to unloading in young rats, in which there is a reduced bone growth and a transient decrease in bone formation, and older rats that lose bone, the actual response to TGF-β administration may differ with age. Whether the protective effect of TGF-β2 also occurs in older animals remains to be determined.
Having demonstrated the protective effect of TGF-β2 on osteoblast apoptosis in unloaded bone, we determined the signaling mechanisms that are implicated in this effect. We hypothesized a role for α5β1-integrin, since its expression is decreased in unloaded bone (23). We show here that TGF-β2 prevented the decrease in α5β1-integrin expression induced by unloading, which provides a mechanism by which TGF-β2 may promote cell adhesion in vivo. This in vivo finding is supported by in vitro studies indicating that α5β1-integrin is positively modulated by TGF-β1 in nonskeletal cells (29, 57) and osteoblasts (49). Although other integrins may be involved in osteoblast attachment and survival (26), our previous data (34) showed that α5β1-integrin plays an important role in osteoblast attachment and cell survival. Integrins are known to be associated with cytoskeletal proteins at focal points of adhesion (16) and we showed that TGF-β enhances the polymerization and synthesis of cytoskeletal proteins, resulting in osteoblast spreading (41). We thus suggest that TGF-β2 may provide protection from unloading-induced apoptosis in part by acting on the adhesive property of osteoblasts through increased α5β1-integrin-mediated cell adherence.
We then determined the survival signaling pathways that are activated by TGF-β2 in unloaded bone. Mechanical stress is known to act through integrin-cytoskeleton interactions (52, 54, 64) to trigger intracellular signals controlling cell death or survival (33). We previously found that skeletal unloading does not markedly alter FAK or ERK1/2 signaling that controls cell survival (17, 28, 32) but rather reduces PI3K signaling in unloaded bone (23). Interestingly, α5β1-integrin can provide protection from apoptosis through PI3K/Akt signaling (40) and TGF-β1 was reported to phosphorylate Akt in a PI3K-dependent manner to trigger cell survival (60). We therefore hypothesized that TGF-β2 may act by modulating PI3K/Akt survival signaling in unloaded bone. Our finding that Akt signaling was reduced by skeletal unloading and that TGF-β2 prevented the altered PI3K/Akt signaling in unloaded bone indicates that the observed anitapoptotic effect of TGF-β2 is linked to PI3K/Akt activation in vivo. Although the overall changes in PI3K/Akt signaling were not striking in unloaded bone, it has to be emphasized that osteoblasts represent only a fraction of cells extracted from the bone metaphysis. Nevertheless, it is striking that TGF-β2 prevented the alteration of PI3K/Akt signaling in unloaded bone. We thus suggest that the protective effect of TGF-β2 on osteoblast apoptosis in unloaded bone involves mechanisms, including increased α5β1-integrin and PI3K/Akt signaling, independently of FAK or ERK1/2 activation. This does not rule out the possibility that other mechanisms may be involved in the altered PI3K activity induced by unloading. Interestingly, anabolic factors for bone such as FGF2 and Wnt3a were recently shown to trigger osteoblast survival through PI3K signaling in vitro (5, 20). The present in vivo findings further support an important role of PI3K/Akt in the control of osteoblast survival and thereby osteoblastogenesis.
There are a number of mechanisms by which activation of Akt signaling may inhibit cell death (17). In vitro studies have identified Bcl-2 family members, caspase-9, and the Forkhead family of transcription factors as direct Akt targets (17). FoxOs are Akt targets (13) that control the expression and transactivation of the proapoptotic protein Bim (21). Recent in vitro studies (36, 55) indicate that TGF-β modulates cell apoptosis by upregulating the expression of Bim. On the other hand, the anitapoptotic effect of TGF-β was linked to increased expression of Bcl-XL, a prosurvival Bcl-2 family member (39). However, we found no consistent changes in the expression of FoxO-1, -3, -4; Bim; or Bcl-XL protein levels in response to unloading or TGF-β2 in vivo (data not shown), suggesting that the effects of unloading and TGF-β occurred independently of Bim or Bcl-XL regulation. In contrast, we found that Bcl-2, an important prosurvival protein (56), was downregulated at 4 and 7 days of unloading and that TGF-β2 prevented the decreased Bcl-2 levels in these experiments. This suggests that TGF-β2 may promote osteoblast survival in vivo by a mechanism implicating, at least in part, the survival protein Bcl-2 via α5β1-integrin and Akt activation (47, 67). Another mechanism that may mediate the anitapoptotic effect of TGF-β2 in vivo via PI3K/Akt signaling is the proapoptotic protein Bad, a Bcl-2 family member that plays an important role in Akt survival signaling (17). Dephosphorylation of Bad results in its targeting to mitochondria where it can inactivate prosurvival Bcl-2 family members. Phosphorylation of Bad by Akt at Ser-136 renders it unable to bind to and inactivate the prosurvival protein Bcl-XL, causing its inability to induce cell death (17). We found that unloading affected Bad levels and decreased Ser-136-phospho-Bad levels, resulting in a decreased p-Bad-to-Bad ratio, suggesting that the decreased p-Bad may contribute, at least in part, to the rapid osteoblast apoptosis in unloaded bone. Although detectable in vivo, the changes in p-Bad were not striking most probably because of the low number of osteoblasts compared with other cells present in the metaphysis. Thus, other mechanisms, yet to be identified, may contribute to the rapid increase in osteoblast apoptosis induced by unloading. Our finding that TGF-β2 increased the p-Bad-to-Bad ratio in unloaded bone suggests that TGF-β2-induced Akt activation led to inactivate Bad, thus preventing interaction of Bad with its targets at the mitochondrial level to cause cell death. These results suggest that p-Bad is part of the intracellular anitapoptotic machinery induced by TGF-β2 in unloaded bone. We thus propose that TGF-β2 can act to reduce unloading-induced osteoblast apoptosis in part through upregulation of α5β1-integrin and PI3K/Akt signaling, resulting in increased expression of the survival proteins Bcl-2 and p-Bad that act at the mitochondrial level to reduce cell death (Fig. 7).
In summary, our results identify for the first time a protective role of TGF-β2 in osteoblast apoptosis induced by mechanical unloading in vivo. We also identified PI3K/Akt signaling and the survival proteins Bcl-2 and p-Bad/Bad as important components of the survival signaling cascade induced by TGF-β2 in vivo. Taken together, our data provide a novel in vivo mechanism by which TGF-β2 targets α5β1-integrin, PI3K/Akt signaling, Bcl-2, and p-Bad/Bad to prevent osteoblast apoptosis in vivo.
This work was supported in part by grants from Institut National de la Santé et de la Recherche Médicale (Paris, France) and the Centre National d'Etudes Spatiales (Paris, France).
We thank S. Renault, C. André, X. Butigieg, and L. Bégot (Departement de Physiologie Intégrée, IMASSA) for excellent technical assistance.
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 © 2008 by American Physiological Society