Distinct subcellular location of the Ca²⁺-binding protein S100A1 differentially modulates Ca²⁺-cycling in ventricular rat cardiomyocytes

S100A1 belongs to a multigenic family of small (M_r 9-13×10³), nonubiquitous Ca²⁺-sensing proteins of the EF-hand type. S100 proteins, which exhibit a cell- and tissue-specific expression pattern, play a major role in intracellular Ca²⁺-homeostasis. In addition, they have been linked to the Ca²⁺-dependent regulation of a variety of intracellular activities such as cell proliferation and differentiation or the dynamics of cytoskeletal constituents (for reviews, see Donato, 2003; Heizmann and Cox, 1998; Zimmer et al., 1995). S100A1 is the most abundant S100 protein in striated muscle and predominantly expressed in the adult heart (Haimoto and Kato, 1988; Kato and Kimura, 1985). Recent gain-of-function experiments have shown that increasing intracellular S100A1 protein levels in the adult heart and skeletal muscle resulted in enhanced Ca²⁺-cycling and contractile function in vitro and in vivo (Most et al., 2001; Most et al., 2003b; Most et al., 2003c; Remppis et al., 2002; Remppis et al., 2004). The identification of S100A1 as a novel intracellular regulator of cardiac and skeletal muscle contractility may provide new impulses in the field of excitation-contraction coupling.

Although S100 proteins apparently lack the classic signaling sequences required for secretion, some S100 proteins seem to reach the extracellular space through secretory pathways that neither involve the common endoplasmic reticulum/Golgi nor the alternative interleukin-1 route (Rammes et al., 1997). From here, S100 proteins may exert an increasing number of downstream functions. For instance, S100B, S100A4 and S100A12 have been shown to promote differentiation of embryonic neurons (Mikkelsen et al., 2001; Novitskaya et al., 2000; Selinfreund et al., 1991). Moreover, S100B that had been secreted by astrocytes was found to be internalized by neurons, where it influenced intracellular Ca²⁺-levels ([Ca²⁺]_i) and cellular excitability (Barger and Van Eldik, 1992; Kubista et al., 1999). Some extracellular activities of S100 proteins such as S100A12 and S100B are apparently mediated through interaction with the multiligand receptor for advanced glycation end products (RAGE) (Hofmann et al., 1999; Huttunen et al., 2000). Indeed, Roth and co-workers have shown that the S100A12/RAGE interaction plays an important role in the pathology of chronic inflammatory diseases such as bowel disease (Foell et al., 2003c), cystic fibrosis (Foell et al., 2003d), rheumatoid arthritis (Foell et al., 2003b) and vascular disorders (Foell et al., 2003a). Importantly, Donato and colleagues have shown that both RAGE-dependent and -independent extracellular effects of S100B are involved in the regulation of cell survival (Huttunen et al., 2000) and muscle development (Sorci et al., 2003; Sorci et al., 2004a; Sorci et al., 2004b). Although there is no evidence for S100A1 secretion from cardiac cells so far, a release into the extracellular space during myocardial damage has been reported (Kiewitz et al., 2000). We have recently shown that extracellularly added S100A1 is endocytosed by neonatal ventricular cardiomyocytes (NVCMs) in a RAGE-independent manner via a Ca²⁺-dependent clathrin-mediated pathway (Most et al., 2003a). Subsequently, internalized S100A1 protected NVCMs from apoptosis via the specific activation of the extracellular signal-regulated kinase (ERK1/2) prosurvival pathway. Thus, similar to other S100 family members, S100A1 may act `outside-in' and its extracellular presence may indicate a function as a paracrine factor in protecting the heart.

Because intracellular S100A1 plays an essential role in cardiac Ca²⁺-handling we examined whether extracellularly added S100A1 has a similar effect on myocardial Ca²⁺-homeostasis. The present study compares the consequences of extracellularly added S100A1 to those of virally transduced S100A1 overexpression on Ca²⁺-cycling in NVCMs in vitro.

Thapsigargin (T-9033), tetracaine (T-7645), monodansylcadaverine (MDC) (M-4008) and caffeine (C-0750) were purchased from Sigma. Inhibitors for protein kinase A (PKA) (myr-PKI, 476485), protein kinase C (PKC) (calphostin-c, 208725) and phospholipase C (PLC) (U-73122, 662035) were obtained from Calbiochem. NCX inhibitor myr-FRCRCF was custom-made by Eurogentec (Belgium).

Human recombinant S100A1 was produced by an isopropylthiogalactosid (IPTG)-driven expression system in Escherichia coli and purified as described previously (Most et al., 2003c). The coupling of S100A1 to tetramethyl-rhodamine (Rh-S100A1) was carried out by Eurogentec.

The S100A1 adenovirus (AdS100A1) was generated by the use of the pAdTrack-CMV/pAdEasy-1 system as recently reported (Most et al., 2001). To facilitate identification of infected cells, AdS100A1 carried the green fluorescent protein reporter gene (GFP) in addition to the human S100A1 cDNA (accession number X58079). Each transgene was independently expressed under the control of a cytomegalovirus promotor sequence. To rule out that the infection procedure in itself had an effect on the amount of S100A1 in the cell, NVCMs were infected with a corresponding adenovirus carrying GFP cDNA alone as a control (Adcontrol). Both replication-deficient adenoviruses were amplified in human embryonic kidney 293 cells (HEK 239), purified and enriched by cesium chloride centrifugation. Activity was tested by plaque assay.

Ventricular cardiomyocytes from 1-2-day-old neonatal hearts (NVCMs) were prepared as published in detail elsewhere (Most et al., 2003a). NVCMs were cultured in Dulbecco's modified Eagle's medium (DMEM; Biochrom) supplemented with penicillin/streptomycin (100 units/ml), L-glutamine (2 mM), and 1% fetal calf serum (FCS Gold; PAA Laboratories GmbH; DFCS) at 37°C in a 95% air/5% CO₂ humidified atmosphere for 2-3 days. Adenoviral infection of NVCMs was carried out in serum-free HEPES-modified medium 199 (M199) with a multiplicity of infection (MOI) of 8 plaque forming units (pfu) per cell. After 4 hours incubation at 37°C in a 95% air/5% CO₂ humidified atmosphere, medium was changed to DFCS. Efficiency of adenoviral gene transfer was monitored 24 hours later by GFP fluorescence. Approximately 95% of NVCMs were infected.

Incubation of NVCMs with recombinant S100A1 (1 μM) was performed in DFCS for 24 hours and mock-treated cells served as control.

Imaging of NVCMs was carried out as described previously (Most et al., 2003a). Briefly, 3-day-old NVCMs grown on glass coverslips were treated with either Rh-S100A1 (1 μM) or S100A1 and GFP adenovirus (MOI 8 pfu/cell) as described above. After 24 hours cells were fixed, permeabilized and labeled with a polyclonal anti-S100A1 (SA 5632, Eurogentec; 1/300), a monoclonal anti-S100A1 (Sigma, 1/200), a monoclonal anti-SERCA2a (ABR) (1/500) or anti-RyR2 (ABR) (1/500) antibodies, followed by the corresponding Cy3-conjugated (Jackson ImmunoResearch Lab) (1/3000), Cy5-conjugated (Jackson ImmunoResearch Lab) (1/400) and ALEXA Fluor 488-conjugated (Alexis; 1/800) secondary antibodies. Polyclonal rabbit anti-SERCA2a antibody (1/1000) was a kind gift from F. Wuytack (Department of Molecular Cell Biology, Leuwen). Confocal images (CLSM) were obtained using a 100× oil objective on a Leica TCS SP laser scanning confocal microscope. Digitized confocal images were processed by Leica software and Adobe Photoshop.

Calibration and measurement of intracellular Ca²⁺-transients in field-stimulated NVCMs using the Ca²⁺-fluorescent dye FURA2-AM was carried out as described in detail by Most and colleagues (Most et al., 2003b). Analysis of steady state Ca²⁺-transients at 1 Hz, 37°C and 2 mM [Ca²⁺]_e was performed with T.I.L.L. Vision software (version 4.01). Baseline data from five consecutive steady-state transients were averaged for analysis of transient amplitude (Ca²⁺ amplitude; [nM]), diastolic Ca²⁺-levels ([nM]) and transient decay (τ; [ms]). The sarcoplasmic reticulum (SR) Ca²⁺-load and sodium-calcium exchanger (NCX) activity were assessed by changes in the transient amplitude ([nM]) and decay (τ; [ms]), respectively, of quiescent NVCMs in modified Tyrode solution (mM: 140 NaCl, 6 KCL, 1 MgCl₂, 2 CaCl₂, 10 glucose, 5 HEPES), subjected to rapid caffeine application (10 mM) after 5 seconds rest. Exchange against sodium- and calcium-free Tyrode solution (choline chloride and EGTA replaced NaCl and CaCl₂, respectively) was used to block NCX current during the decay of the caffeine-induced transient. Myr-FRCRFC (30 μM) was used to inhibit NCX current in field-stimulated NVCMs. Free intracellular Ca²⁺ concentration [Ca²⁺]_i was calculated by the equation of Grynkiewicz et al; [Ca²⁺]_i=K_d×β×((R-R_min)/(R_max-R)). The Ca²⁺-transient amplitude was calculated as the difference between calibrated systolic and diastolic [Ca²⁺]_i. Data sets represent analysis of 50 cells from three different preparations.

To assess protein expression of S100A1 (Sigma) (S-2407, 1/500), sarcoplasmic reticulum Ca²⁺-ATPase (SERCA2a; Biomol SA209, 1/5000), sodium-calcium exchanger (NCX) (ABR) (MA3-926, 1/5000) and cardiac actin (Progen Ac1-20.4.2, 1/10000), western blotting from whole-cell homogenates of NVCMs was carried out as previously described (Most et al., 2003b). Briefly, cells were washed three times with EGTA/EDTA buffer (PBS, pH 7.4, containing 2 mM EGTA/EDTA), scraped off the dish and immediately homogenized in EGTA/EDTA buffer. Homogenates were centrifuged for 10 minutes at 15,000 g and the supernatant was used for biochemical analysis. S100A1 protein was concentrated by sequential size exlusion centrifugation columns (Amicon) (Microcon YM-100 and -3).

Data are presented as mean±s.e.m. Unpaired two-tail Student's t-test and a two-way repeated ANOVA analysis was performed to test for differences between groups. A value of P<0.05 was accepted as statistically significant.

To examine the effects of an increased S100A1 protein level on the cycling of cytosolic Ca²⁺ in neonatal rat ventricular cardiomyocytes (NVCMs), we employed two different procedures (Fig. 1): the NVCMs were either transduced by means of an adenoviral S100A1-expression construct (AdS100A1 cells, Fig. 1A) or incubated for 24 hours with DFCS containing 1 μM recombinant human S100A1 protein (Fig. 1C). As illustrated in Fig. 1, overexpressed S100A1 mainly displayed a fine granular network-like pattern. In addition, small amounts were detected in the nucleus. This pattern corresponds to the distribution of endogenous S100A1 observed in control cells (Fig. 1B). Western blotting using an antibody that reacts with both human and rat S100A1 showed that adenoviral-mediated S100A1 gene delivery to NVCMs resulted in approximately threefold higher amounts of total S100A1 protein compared with the level of endogenous S100A1 in control cells infected with an adenoviral construct lacking the S100A1 gene (Fig. 1A). Equivalent protein loadings were confirmed by comparable cardiac actin levels.

As an alternative approach to achieve an increase of the intracellular S100A1 level in NVCMs, we incubated cells for 24 hours in medium containing 1 μM rhodamine-coupled recombinant human S100A1 (Rh-S100A1). The CSLM image shown in Fig. 1C reveals an intracellular perinuclear accumulation of Rh-S100A1. This finding is consistent with our recently published data, which shows that following endocytosis via a Ca²⁺-dependent calthrin-mediated process, the extracellularly added S100A1 is routed to the endosomal compartment (Most et al., 2003a). The uptake of S100A1 caused an almost fourfold increase of intracellular S100A1 in NVCMs compared with mock-treated control cells (Fig. 1C). To directly correlate the distribution of exogenous S100A1 resulting from the two different procedures, we incubated AdS100A1 cells for 1 hour with Rh-S100A1 protein (1 μM) 24 hours after adenoviral infection (Fig. 1D-F). As in the individually treated cells, the internalized recombinant human Rh-S100A1 (red, Fig. 1E) displayed a vesicular-like endosomal localization in AdS100A1 cells. When immunolabeled with a polyclonal antibody that specifically recognizes the hinge region of human S100A1, both S100A1 populations were recognized in the S100A1-treated AdS100A1 cells (blue, Fig. 1D). Superimposing the CSLM image of total exogenous S100A1 with the internalized Rh-S100A1 (Fig. 1F) confirmed that the localization of internalized Rh-S100A1 (violet) was clearly distinct from the fine granular network-like pattern of overexpressed human S100A1 (blue).

As illustrated by the Ca²⁺-transients in Fig. 2, the overexpression of S100A1 in AdS100A1-transduced NVCMs gave rise to an enhanced cytosolic Ca²⁺-cycling compared with control cells. The overall enhanced Ca²⁺-cycling was reflected by an increased amplitude of Ca²⁺-transients in AdS100A1 cells (Fig. 2B). The larger amplitude was predominantly due to an S100A1-induced rise of systolic [Ca²⁺]_i as indicated by the higher peak in AdS100A1-transduced NVCMs (Fig. 2A, bar). Concomitantly, S100A1 overexpression also decreased diastolic [Ca²⁺]_i (Fig. 2C), which was associated with a reduction in the Ca²⁺-transient decay-constant τ (Fig. 2D).

Similar to S100A1 overexpression in AdS100A1 cells, uptake of S100A1 also had an effect on the Ca²⁺-transients recorded in S100A1-treated cells (Fig. 2A'). Accordingly, they exhibited an increased amplitude (Fig. 2B'), indicating that the rise in intracellular S100A1 caused an increased cytosolic Ca²⁺-turnover in S100A1-treated NVCMs. However, in contrast to Ca²⁺-transients of cells with adenoviral-mediated S100A1 overexpression, the amplitude of S100A1-treated cells was primarily increased by a marked reduction of diastolic [Ca²⁺]_i, and to a lesser extent by elevated systolic [Ca²⁺]_i (Fig. 2A', bar and Fig. 2C'). For both, overexpression and uptake from the extracellular environment, increased S100A1 levels accelerated diastolic Ca²⁺-elimination as indicated by a decrease in the Ca²⁺-transient decay-constant τ (Fig. 2D and D').

To test whether binding of extracellular S100A1 to NVCMs was sufficient to induce an improved cytosolic Ca²⁺-cycling or whether endocytosis was required, NVCMs were treated with recombinant S100A1 in the presence of monodansylcadaverine (MDC, 50 μM), which was shown to effectively prevent S100A1 uptake (Most et al., 2003a). As illustrated by the Ca²⁺-transients, the presence of MDC precluded changes in the amplitude (Fig. 2B') and diastolic [Ca²⁺]_i (Fig. 2C') as well as in the decay of the Ca²⁺-transient (Fig. 2D').

Together these results show that an augmented intracellular S100A1 protein level in NVCMs increases cytosolic Ca²⁺-turnover. More specifically, we observed a differential effect of adenoviral-mediated S100A1 overexpression versus increased S100A1 following endocytosis: higher systolic Ca²⁺-peaks were responsible for the increased amplitude of Ca²⁺-transients in NVCMs overexpressing S100A1, whereas a marked decrease in diastolic [Ca²⁺]_i largely accounted for the increased amplitude of Ca²⁺-transients in cells after endoyctosis of extracellular S100A1.

We next examined whether the differential effect of overexpressed and internalized S100A1 on Ca²⁺-turnover was associated with the distinct subcellular location of S100A1. Because S100A1 protein has previously been shown to co-immunoprecipitate with cardiac SR Ca²⁺-regulatory proteins in human myocardium (Kiewitz et al., 2003), we examined the distribution of the SR Ca²⁺-ATPase (SERCA2a) and its potential association with S100A1 in NVCMs with an increased level of S100A1 compared with control cells by immunolabeling with anti-SERCA2a antibodies (Fig. 3,). To rule out that the adenoviral transduction procedure by itself caused a translocation of SERCA2a, we first established that its distribution in cells infected with a control virus was similar to that in uninfected NVCMs expressing only endogenous S100A1 (data not shown). Subsequently, the distribution of S100A1 in control cells (blue, Fig. 3A,C), in AdS100A1 NVCMs (blue, Fig. 3A',C') and in S100A1-treated cells (red, Fig. 3A”,C”) was compared to the location of immunolabeled SERCA2a (Fig. 3B,B',B”). Merging the corresponding confocal images clearly showed that overexpressed S100A1 and endogenous S100A1 partially colocalized with the SR Ca²⁺-pump (violet, Fig. 3C,C'). Conversely, as illustrated in Fig. 3A” to C”, internalized Rh-S100A1 protein (red) did not colocalize with the reticular distribution of SERCA2a (green).

In conclusion, the immunolocalization studies revealed that, depending on the experimental procedure used to increase S100A1 levels, S100A1 was routed to a different cellular compartment. Consequently, an association of S100A1 with the SR was only observed in AdS100A1-transduced NVCMs.

Because overexpressed S100A1 was associated with SERCA2a at the SR, we tested the influence of SERCA2a antagonists on the efficiency of Ca²⁺-cycling in AdS100A1- as well as in S100A1-treated NVCMs. Because Ca²⁺-cycling in infected control cells and mock-treated NVCMs was indistinguishable, the two groups were pooled and are hereafter referred to as control NVCMs. As documented in Fig. 4A, a reduced amplitude in Ca²⁺-transients of control NVCMs indicated that the SERCA2a-mediated Ca²⁺-uptake into the SR was inhibited by thapsigargin (1 μM). Incubation of AdS100A1-cells with this SERCA2a inhibitor completely abolished the S1000A1-induced gain in the Ca²⁺-transient amplitude (Fig. 4A). The Ca²⁺-transients in thapsigargin-treated NVCMs did not significantly differ with respect to the diastolic Ca²⁺-load or the decay rate between AdS100A1-transduced and control NVCMs: thapsigargin inhibition of SERCA2a equally raised diastolic [Ca²⁺]_i (Fig. 4B) and prolonged SR resequestration (Fig. 4C). By contrast, the enhancement of the Ca²⁺-transient amplitude in S100A1-treated NVCMs was not affected by thapsigargin (Fig. 4A). Moreover, inhibition of SERCA2a had virtually no effect on the decreased diastolic [Ca²⁺]_i (Fig. 4B) and accelerated cytosolic Ca²⁺-elimination (Fig. 4C) in S100A1-treated NVCMs. Like thapsigargin, inhibition of SR Ca²⁺-release by tetracaine (100 μM), a blocker of the SR Ca²⁺-release channel/ryanodine receptor (RyR2), decreased the amplitude of the Ca²⁺-transient in control NVCMs and completely abolished the gain in the Ca²⁺-transient amplitude in S1000A1-overexpressing NVCMs (Fig. 4D). However, tetracaine had no effect on the augmented Ca²⁺-transient amplitude in S100A1-treated NVCMs (Fig. 4D). Consistent with these findings, immunofluorescence studies indicated that internalized S100A1 did not colocalize with SERCA2a (Fig. 3) or RyR2 (data not shown).

Because the data suggested that the increased efficiency of Ca²⁺-turnover in S100A1-overexpressing but not in S100A1-treated NVCMs depended on SR Ca²⁺-fluxes, we next investigated the SR Ca²⁺-load in both groups. To address this issue, AdS100A1-infected and S100A1-treated NVCMs were exposed to caffeine (10 mM), and the amplitude of caffeine-releasable SR Ca²⁺ served as an indirect measure of the SR Ca²⁺-content. As depicted in Fig. 5A, S100A1 overexpression caused an increase in the caffeine-induced Ca²⁺-transient amplitude while internalized S100A1 apparently reduced the SR Ca²⁺-load compared with the control.

Cytosolic Ca²⁺-turnover in NVCMs not only invokes the SR, but is simultaneously regulated by sarcolemmal Ca²⁺-fluxes. The sodium-calcium exchanger (NCX) is a major regulator of sarcolemmal Ca²⁺-fluxes. Analysis of the decay-constant τ of the caffeine-induced Ca²⁺-transients, which served as a measure for NCX-mediated sarcolemmal Ca²⁺-extrusion, revealed unchanged kinetics in AdS100A1-transduced NVCMs compared with cells infected with a control construct (Fig. 5B). Thus, it appears that overexpressed S100A1 does not affect NCX activity. Conversely, in S100A1-treated NVCMs τ of the caffeine-induced Ca²⁺-transients was decreased compared with mock-treated NVCMs (control, Fig. 5). This finding indicated that the increased S100A1 level in the endosomal compartment could be responsible for the accelerated decay of the caffeine-mediated rise in cytosolic [Ca²⁺]_i.

Because the NCX forward-mode mainly accounts for trans-sarcolemmal Ca²⁺-efflux, the influence of NCX inhibition in S100A1-treated cells was studied. Selective NCX inhibition was achieved by rapidly replenishing the medium with sodium/calcium-free medium (Bassani et al., 1994; Despa et al., 2002). As documented by the caffeine-induced Ca²⁺-transients in Fig. 5C, this regimen effectively abrogated the S100A1-mediated acceleration of the Ca²⁺-transient decay.

This finding suggests that endocytosed S100A1 decreases diastolic [Ca²⁺]_i through enhanced sarcolemmal Ca²⁺-extrusion, possibly involving an increased NCX forward-mode activity.

To further explore the influence of endocytosed S100A1 protein on NCX activity, we studied the impact of myristylated (myr)-FRCRCFa, a specific NCX blocker on diastolic [Ca²⁺]_i in S100A1-treated field-stimulated NVCMs. Myr-FRCRCFa is a novel, cell-permeable peptide inhibitor of NCX that alters neither L-type Ca²⁺-channel/dihydropyridine receptor (LCC) nor Na⁺- or K⁺-channel activity, and appears to exhibit a higher degree of selectivity compared with other widely used NCX-inhibitors such as KB-R7943 or SEA0400 (Convery et al., 1998; Hobai et al., 1997; Khananshvili et al., 1995; Reuter et al., 2002). Administration of 30 μM myr-FRCRCFa caused a large rise in diastolic [Ca²⁺]_i in control cells and abrogated the S100A1-mediated decrease in diastolic [Ca²⁺]_i -treated cells (Fig. 6A).

Given that endocytosed S100A1 activates the PLC-PKC-p44/42 pathway (Most et al., 2003a) and PKC is known to stimulate NCX function (Iwamoto et al., 1996), we investigated whether triggering a corresponding pathway may be involved in increasing NCX activity in S100A1-treated NVCMs. Incubation with U73122 (5 μM), an inhibitor of PLC, resulted in a significant increase in diastolic [Ca²⁺]_i in S100A1-treated NVCMs (Fig. 6B). Similar results were obtained after addition of calphostin-c (1 μM), a PKC-inhibitor (Fig. 6C), indicating that enhanced cyotosolic [Ca²⁺]_i removal by endocytosed S100A1 might involve PLC and PKC. Notably, PD98095 (10 μM) as well as myr-PKI (5 μM), specific inhibitors of the mitogen-activated protein kinase kinase MEK1 and protein kinase A (PKA), respectively, failed to inhibit the S100A1-mediated decrease in diastolic [Ca²⁺]_i (data not shown).

Because increased S100A1 levels in NVCMs impinged on Ca²⁺-regulatory proteins we assessed the abundance of NCX and SERCA protein in S100A1-treated and AdS100A1-transduced NVCMs. As shown by representative immunoblots in Fig. 7, neither endocytosis nor overexpression of S100A1 altered the levels of NCX and SERCA2a compared with control cells.

These findings suggest that the changes in Ca²⁺-cycling seen in response to an increase of intracellular S100A1 are based on the altered activity of Ca²⁺-regulatory proteins rather than their amounts.

S100 Ca²⁺-binding proteins have been associated with a variety of intracellular Ca²⁺-mediated processes (Heizmann and Cox, 1998). Binding of Ca²⁺ induces a conformational change in the threedimensional structure and enables these proteins to interact with target proteins, thereby transducing the intracellular Ca²⁺-signal (Wang et al., 2001). Recent findings involving S100A1 gene transfer into ventricular cardiomyocytes have shown that S100A1 overexpression enhances cytosolic Ca²⁺-cycling and contractile properties both in vitro and in vivo (Most et al., 2001; Most et al., 2003b; Most et al., 2003c; Remppis et al., 2002; Remppis et al., 2004). Consistently, S100A1-deficient mice display an impaired cardiac contractility in response to hemodynamic stress (Du et al., 2002). These studies promote S100A1 as a novel key regulator of cardiac function. In addition, recent studies from our laboratory provide evidence that S100A1 may also act on ventricular cardiomyocytes through an unrelated mechanism: uptake of S100A1 from the extracellular environment protects cells from apoptosis induced by reactive oxygen species or 2-deoxyglucose (Most et al., 2003a). This downstream effect apparently relies on RAGE-indepedent endocytosis of extracellular S100A1 and involves the activation of the PLC-PKC-ERK1/2 prosurvival pathway. Several studies support the notion that extracellular S100 proteins may have distinct functions in cellular physiology and pathology. For example, S100A4 was reported to stimulate the outgrowth of neurite cells (Novitskaya et al., 2000), whereas S100B has been shown to be involved in the regulation of cell differentiation and survival (Huttunen et al., 2000; Sorci et al., 2003; Sorci et al., 2004a; Sorci et al., 2004b). Another example is S100A12, which has been implicated as an important pro-inflammatory regulator in numerous chronic inflammatory diseases (Foell et al., 2003a; Foell et al., 2003b; Foell et al., 2003c; Foell et al., 2003d). Notably, release of S100A1 into the extracellular space in reponse to ischemic cadiac injury has been reported (Kiewitz et al., 2000), but it is unclear whether extracellular S100A1 might affect Ca²⁺-homeostasis in ventricular cardiomyocytes. In the present study we addressed this issue by comparing the effects of S100A1 added to the extracellular environment to the effects of adenoviral-mediated S100A1 overexpression on cardiac Ca²⁺-cycling in vitro.

The model illustrated in Fig. 8 summarizes the data presented in this paper. First, we show that recombinant S100A1 added to the extracellular environment is internalized and subsequently modulates cytosolic Ca²⁺-cycling in ventricular cardiomyocytes (Fig. 8B). Second, compared with the effects of adenoviral-mediated S100A1 overexpression on Ca²⁺-turnover (Fig. 8C), internalized S100A1 modulates Ca²⁺-homeostasis in a different manner, which is likely to involve the NCX. The immunofluorescence studies suggest that this mechanistic difference is related to a distinct subcellular location of internalized versus overexpressed S100A1. Both experimental regimens, endocytosis of extracellularly added S100A1 and intracellular S100A1 overexpression led to a three-to-fourfold increase of intracellular levels of the Ca²⁺-sensor protein. Because the increase in intracellular S100A1 is comparable, we can largely rule out that the total amount of S100A1 is responsible for the differential effects on cytosolic Ca²⁺-turnover. S100A1 uptake was found to enhance the cytosolic Ca²⁺-transient amplitude by a marked decrease in diastolic [Ca²⁺]_i, while systolic [Ca²⁺]_i were mostly unchanged. By contrast, overexpression of S100A1 enhanced the Ca²⁺-transient amplitude mainly by a marked increase in systolic [Ca²⁺]_i in combination with a slight decrease in diastolic [Ca²⁺]_i. The downstream effect of extracellularly added S100A1 on cardiac Ca²⁺-handling together with the previously demonstrated cardioprotective effects argue that S100A1 could act as paracrine cardiac factor which also acts on other cardiac cells, such as fibroblasts, smooth muscle and endothelial cells.

Along with a differential Ca²⁺-handling, we also observed a different location of S100A1 that was taken up from the extracellular environment compared with overexpressed S100A1. The latter is associated with the SR since it colocalized with the SR Ca²⁺-regulatory proteins SERCA2a (Fig. 8C) and RyR2 (data not shown). These findings are in line with recently published data showing Ca²⁺-dependent co-immunoprecipitation of endogenous S100A1 protein with RyR2 and SERCA2a in murine and human myocardium (Kiewitz et al., 2003; Most et al., 2003b). Like Kiewitz et al., who were able to show a colocalization of endogenous S100A1 and SERCA2a in neonatal cardiomyocytes from mouse (Kiewitz et al., 2003), our data reveal a colocalization in rat neonatal cardiomyocytes. Consistently, when RyR2 and SERCA2a activities were inhibited in S100A1-overexpressing cells, the amplitude of Ca²⁺-transients display a marked decrease; also, diastolic [Ca²⁺]_i are noticeably increased and the decay of Ca²⁺-transients is decelerated (Fig. 8C). By contrast, inhibitors of SERCA2a or RyR2 neither affected the S100A1-induced increase in the Ca²⁺-transient amplitude nor lowered diastolic [Ca²⁺]_i in cardiomyocytes following uptake of extracellular S100A1. Thus, enhanced Ca²⁺-cycling in cardiomyocytes overexpressing S100A1 is presumably based on SR Ca²⁺-cycling. These findings are consistent with previous studies showing that overexpressed S100A1 increased both SR Ca²⁺-uptake and Ca²⁺-load, as well as Ca²⁺-release from the SR (Most et al., 2001; Most et al., 2003b; Remppis et al., 2002). Because the levels of SERCA2a were unaltered, we conclude that overexpressed S100A1 enhances SR Ca²⁺-load by an increased SERCA2a activity rather than protein abundance.

Internalized S100A1, however, does not colocalize with SR Ca²⁺-regulatory proteins (Fig. 3) and was also not detected at the sarcolemmal membrane of cardiomyocytes (Most et al., 2003a). Instead, it displays a vesicular-like perinuclear distribution that has recently been identified as part of the endosomal compartment (Most et al., 2003a). Consistent with the distinct location of S100A1, Ca²⁺-handling by S100A1-treated cells appears to be independent of SR Ca²⁺-regulatory proteins (Fig. 8B). However, since S100A1 uptake decreased the SR Ca²⁺-levels, another mechanism must be affected that accounts for an enhanced cytosolic Ca²⁺-elimination. Accordingly, we measured an accelerated decay of the caffeine-induced Ca²⁺-transient amplitude, which indicates that the enhanced cytosolic Ca²⁺-elimination is likely to involve a trans-sarcolemmal Ca²⁺-efflux via NCX. In addition, inhibition of NCX by switching to sodium/calcium-free medium abolished this effect, which supports the notion that internalized S100A1 stimulates cytosolic Ca²⁺-extrusion through the NCX forward-mode rather than through the slow Ca²⁺-removal systems such as the plasma membrane Ca²⁺-ATPase (PMCA) and the mitochondrial uniporter (Bassani et al., 1994). Similar effects were seen in S100A1-treated field-stimulated cardiomyocytes in the presence of the specific NCX inhibitor myr-FRCRCFa. The inhibitory cell-permeable hexapeptide abolished the S100A1-mediated decrease in diastolic [Ca²⁺]_i. On the basis of these results, we propose a mechanism where endocytosed S100A1 decreases diastolic [Ca²⁺]_i by activating sarcolemmal Ca²⁺-extrusion via NCX.

Because endocytosed S100A1 was not present at the sarcolemmal membrane, the question arose, how an increased S100A1 protein level confined to the endosomal compartment could trigger NCX activity. However, it has been shown that PKC and PLC play a role in the regulation of NCX in neonatal cardiomyocytes (Iwamoto et al., 1996). Moreover, we have shown that internalized S100A1 activates the ERK1/2 signaling pathway through activation of PLC and PKC (Most et al., 2003a). These findings and our Ca²⁺-transient measurements in the presence of specific inhibitors of PLC and PKC prompted us to suggest that a S100A1-mediated activation of NCX forward mode might occur through the PLC and PKC signaling pathway.

In conclusion, our data indicates that the enhanced Ca²⁺-cycling observed in S100A1-overexpressing cardiomyocytes predominantly involves SR Ca²⁺-fluxes (Fig. 8C), whereas endocytosed S100A1 apparently alters intracellular Ca²⁺-turnover through sarcolemmal modulation (Fig. 8B). According to the model, accumulation of internalized S100A1 in the endosomal compartment somehow triggers endosome-associated PLC and PKC, which then activate NCX to increase sarcolemmal Ca²⁺-efflux.

This study provides the first insight into the mechanisms through which S100A1 differentially modulates sarcolemmal and sarcoplasmic Ca²⁺-handling in ventricular cardiomyocytes depending on the subcellular location of this protein. Endosomal S100A1 protein appears to modulate Ca²⁺-cycling through an enhanced NCX forward-mode activity, whereas cytosolic S100A1 increases Ca²⁺-cycling by enhancing SR Ca²⁺-fluxes. The latter is likely to underlie the positive inotropic and lusitropic effects seen in S100A1-overexpressing myocardium (Most et al., 2001; Most et al., 2003b; Remppis et al., 2002; Remppis et al., 2004). In addition, we have very recently shown that cardiac adenoviral S100A1 gene delivery rescues failing myocardium in vitro and in vivo (Most et al., 2004), which warrants further studies on the impact of decreased diastolic [Ca²⁺]_i and SR Ca²⁺-load resulting from S100A1 endocytosis on cardiac function in vivo.

This work was supported in part by grants from the University of Heidelberg to P.M. (Forschungsförderung 93/2002 and 61/2003), the Deutsche Forschungsgemeinschaft (DFG) to P.M. (Mo 1066/1-1) and T.W. (We 2366/3-1), from the M. E. Müller Foundation of Switzerland and the Kanton of Basel Stadt (to M.B., U.A. and C.-A. S.) and the National Institute of Health to W.J.K. (R01-HL59533 and R01-HL56205).