Human Immunoglobulins Inhibit Thrombin-induced Ca Movements and Nitric Oxide Production in Endothelial Cells*

Binding of natural antibodies to endothelial cell plays an important role in hyperacute xenograft rejection between discordant species. Human intravenous immunoglobulins (IVIg) delay this hyperacute rejection, but their mechanisms of action on endothelial cells have to be defined. Here we demonstrate that IVIg dose-depend-ently prevent thrombin from eliciting cytosolic Ca 2 (cid:49) movements and nitric oxide (NO) production in aortic endothelial cells from guinea pig. The Ca 2 (cid:49) response to thrombin was similarly affected by IVIg whether they were removed or not from the incubation medium before stimulation. Pretreatment by rat natural antibodies also suppress the thrombin-induced Ca 2 (cid:49) peak corre-sponding to Ca 2 (cid:49) release from intracellular stores but stimulate the subsequent sustained increase in [Ca 2 (cid:49) ] i and the release of NO. The action of human intravenous immunoglobulins seems to be selective for the thrombin receptor because they do not

Vascular endothelial cells respond to various neurohumoral and physical stimuli by increasing cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] i ) and releasing substances such as nitric oxide (NO), 1 prostacyclin, and endothelins (1,2). Thrombin, histamine, bradykinin, and ATP, which are released or locally produced in the blood during vascular injury, produce a well characterized biphasic rise of [Ca 2ϩ ] i in endothelial cells by activating surface receptors coupled to phospholipases C (3)(4)(5)(6)(7). Bradykinin simultaneously increases [Ca 2ϩ ] i and the NO secretion (8), whereas histamine and ATP activate the formation of cGMP, a marker for NO production (7,9,10). Nitric oxide plays a critical role in the maintenance of vascular homeostasis by promoting vasodilatation and inhibiting platelet aggregation (11,12).
Vascular dysfunction leading to organ failure is an important limitation to successful transplantation. Binding of natural xenophile antibodies, activation of the complement cascade, and endothelial cell activation are the main mechanisms leading to hyperacute rejection of discordant xenografts (13)(14)(15)(16). Human polyclonal immunoglobulins for intravenous use (IVIg) can delay xenograft rejection in a guinea pig to rat model (17) as well as in a pig to baboon model (18). To the best of our knowledge, no study has investigated the potential contribution of natural xenoreactive antibodies on the cell mechanisms leading to endothelial cell activation.
The present study was undertaken to investigate the action of IVIg and rat natural antibodies (IgG) on cytosolic Ca 2ϩ movements and NO production elicited by thrombin in aortic endothelial cells from guinea pig. It provided evidence for the selective interaction of IVIg with thrombin receptors or signal transduction pathways, resulting in inhibition of thrombininduced [Ca 2ϩ ] i increases and NO production. It also demonstrated that rat IgG suppress the first phase of the Ca 2ϩ response to thrombin without alteration of NO production.
Preparation of Immunoglobulins-Lyophilized IVIg were provided by Baxter (Lessines, Belgium). The absence of IVIg degradation was controlled by electrophoresis on a SDS-polyacrylamide gel, and then IVIg were concentrated to 120 mg/ml. Rat IgG were isolated from rat Lewis decomplemented serum (20). Briefly, sera were passed over a protein G-Sepharose column. IgG were eluted with 0.2 M glycine-HCl, pH 2.8. IgG was more than 95% pure as determined by electrophoresis on a SDS-polyacrylamide gel and by enzyme-linked immunosorbent assay using anti-rat Fcg (Jackson Immunoresearch). These IgG preparations were concentrated to 10 mg/ml.
Binding of Human Immunoglobulins-Endothelial cells were detached with 0.1% trypsin-EDTA solution, centrifuged at 450 ϫ g for 10 min at 4°C, and the pellet was resuspended in PBS (Life Technologies, Cergy Pontoise, France) containing 0.01% NaN 3 . Cells (5.10 5 cells/ tubes) were then incubated for 2 h at 4°C with 100 l of different IVIg concentrations and washed twice with PBS-NaN 3 . The binding of human IgG anti-guinea pig endothelial cells was detected by staining the cells with ␤-phycoerythrin-conjugated goat anti-human IgG (HϩL) (Jackson Immunoresearch). The fluorescence of labeled cells was analyzed by FACScan (Becton Dickinson). The nonspecific binding was assessed by performing similar experiments in the presence of equivalent quantities of human albumin. Only alive cells as defined by propidium iodure staining exclusion were considered. In the range of IVIg concentrations studied, the presence of antibodies did not alter cell viability estimated as the percentage of cells excluding trypan blue or propidium iodure (Ͼ95%). Specific binding of IVIg to endothelial cells was obtained after subtraction of nonspecific binding from the total.
Determinations of Cytosolic Ca 2ϩ Concentration-Cytosolic free Ca 2ϩ concentrations were determined as described previously (21). Briefly, cells cultured on coverslips were transferred into buffer A (in mM: NaCl 136, KCl 5, NaH 2 PO 4 2, MgSO 4 0.4, NaHCO 3 4, glucose 8, Hepes 25, pH 7.4 at 37°C) containing 20% fetal calf serum, a coktail of amino acids, and 2 mM glutamine. They were first equilibrated in buffer A for 30 min at 37°C and then loaded with 2 M fura-2-tetrakis (acetoxymethylester) for 25 min. De-esterification was completed in fresh buffer A containing 20% of fetal calf serum for 25 min at 37°C. To record fluorescence signals, cells were washed twice with buffer A, and the coverslip was inserted in a special quartz-suprasil cuvette thermostated at 37°C (22). Fura-2 was excited at 340 and 380 nm, and emission was collected at 505 nm on a spectrofluorimeter SPEX CMIII (ISA-Jobin-Yvon, Longjumeau, France). Each measurement was systematically corrected for autofluorescence of unloaded cells incubated or not with different concentrations of human immunoglobulins.
[Ca 2ϩ ] i was calculated from the 340/380 ratios of fluorescence intensities using the equation of Grynkiewicz et al. (23) and the previously determined K d and calibration parameters (21).
The cells were incubated in PBS medium complemented with 5 mM glucose and 0.5 mM MgCl 2 (buffer B). The use of a micromanipulator (Biologic Instruments, Claix, France) attached to the stage of an inverted microscope (Zeiss, Germany) allowed to set the probe 10 -15 m above the cell surface, all the apparatus being enclosed in a Faraday's chamber. At the end of each experiment, internal calibration of the electrochemical sensor was performed by the addition of NO standard solutions as detailed previously (25). The sensitivity of electrodes varied from 0.5 to 2.6 nM NO/pA (1.6 Ϯ 0.4 nM NO/pA, n ϭ 9). Under our experimental conditions, the detection limit was 2-5 nM NO.
Identification of NO as responsible for the amperometric signal induced by 2 units/ml thrombin was checked by preincubating the cells for 5 min at 37°C with 10 Ϫ3 M L-arginine or 10 Ϫ3 M N G -monomethyl-Larginine monoacetate. Thrombin-induced NO production was enhanced from 46 Ϯ 15 up to 94 Ϯ 10 nM by L-Arg pretreatment (n ϭ 3) and decreased to 6.7 Ϯ 3.3 nM in N G -monomethyl-L-arginine monoacetatetreated cells (n ϭ 3).

RESULTS
In aortic endothelial cells from guinea pig, thrombin induced a biphasic increase in [Ca 2ϩ ] i : a transient peak occurring 15-20 s after stimulation followed by a long lasting plateau (Fig. 1A). NO production paralleled the rise of [Ca 2ϩ ] i . NO concentration reached a maximum within 15-20 s after thrombin addition (32 Ϯ 6 nM, n ϭ 9) and remained above basal level for a few minutes (Fig. 1A). Incubation of cells for 5 min at 37°C with 10 mg/ml IVIg or rat IgG followed by a washing to remove unbound antibodies from the incubation medium did not change basal [Ca 2ϩ ] i values (Fig. 1, B and C) but altered the Ca 2ϩ response to thrombin. This IVIg pretreatment prevented thrombin from significantly increasing [Ca 2ϩ ] i and producing NO (Fig. 1B). Rat IgG pretreatment also suppressed the [Ca 2ϩ ] i peak elicited by thrombin but increased the amplitude of the sustained [Ca 2ϩ ] i rise from 56 Ϯ 13 to 138 Ϯ 27 nM (n ϭ 5 and 12) and the NO production from 32 Ϯ 4 to 52 Ϯ 4 nM (n ϭ 5) (Fig. 1C).
For [Ca 2ϩ ] i measurements only, we could compare the response to thrombin in cells pretreated by 10 mg/ml IVIg whether they were removed or not from the incubation medium before stimulation. Under these two conditions, the amplitude of [Ca 2ϩ ] i rise at peak was reduced by 85% on average and that at plateau by 70%. This indicates that the IVIg effects we saw were mainly due to their binding on endothelial cells and not to their direct interactions with thrombin in the extracellular medium. This is confirmed by the observation that the dosedependent inhibitions of IVIg on thrombin-induced Ca 2ϩ move-

IgG Modulate Thrombin-induced [Ca 2ϩ ] i and NO Changes 26474
ments and NO production (Fig. 2, A and B) paralleled their specific binding to guinea pig endothelial cells (Fig. 2C). Both maximal inhibitory effects and binding were reached at approximately 10 mg/ml.
To investigate the possible interaction of IVIg with other membrane receptors or with Ca 2ϩ pathways activated by Ca 2ϩ store depletion, endothelial cells were stimulated either by ATP, endothelin-1, or thapsigargin, a specific inhibitor of endoplasmic reticulum Ca 2ϩ pumps. [Ca 2ϩ ] i transiently increased after 0.5 mM ATP addition, returned to basal values, and then slowly rose to a plateau (Fig. 3, upper panels, trace a). Preincubation of cells with 10 mg/ml IVIg suppressed the first phase of ATP-induced Ca 2ϩ signal but modified neither the time course nor the amplitude of the late Ca 2ϩ response (Fig. 3,  upper panels, trace b). The Ca 2ϩ movements elicited by endothelin-1 and thapsigargin were not altered by pretreatment with IVIg (Fig. 3, middle and lower panels). Concerning NO release, results similar to those obtained for Ca 2ϩ movements were observed except for ATP. Both endothelin-1 and thapsigargin elicited NO synthesis (Fig. 4). This production of NO was similar in cells pretreated or not with 10 mg/ml IVIg. In contrast, ATP, which increased [Ca 2ϩ ] i , did not produce any detectable NO at the cell surface (Fig. 4, upper panel). DISCUSSION The vascular endothelium is more than a physical and selectively permeable barrier between blood and surrounding tissues; it maintains an anticoagulant environment, prevents blood cells adhesion, and secretes vasoactive substances (2). Activation of the endothelial cells have been recently shown to be a major mechanism at play in the genesis of hyperacute xenograft rejection (26). The present study demonstrates that natural antibodies alter the metabolism of Ca 2ϩ ions, one of the most important second messengers for cell activation. Human and rat natural antibodies did not modify per se the cytosolic Ca 2ϩ level in aortic endothelial cells from guinea pig but did modulate the Ca 2ϩ movements elicited by thrombin, a protease generated at sites of vascular injury. The observation that these effects still appeared after removal of antibodies from the incubation medium suggests interactions with specific targets on endothelial cells rather than immunological interferences with thrombin itself.
Human IVIg prevented both the biphasic Ca 2ϩ response and the NO production, whereas the rat natural IgG suppressed the first phase of the Ca 2ϩ response but increased the sustained [Ca 2ϩ ] i elevation and the NO concentration ([NO]). These distinct actions on Ca 2ϩ pathways and NO production suggest the presence of different binding sites for IVIg and rat IgG on endothelial cell membrane. This is supported by the lack of effect of IVIg pretreatment on rat IgG binding to guinea pig endothelial cells (results not shown). Our results also underline the functional difference between rat IgG and human IVIg. The binding of antidonor antibodies to blood vessels of the xenograft occurs early in the rejection process (16), whereas purified human immunoglobulins reduce the cytotoxic activity of recipient serum in different discordant xenograft models (17,18). From the present observations it may be proposed that binding of rat IgG to guinea pig endothelial cells induce cell contraction by amplifying the sustained Ca 2ϩ response to thrombin. In contrast, the binding of IVIg to guinea pig endothelial cells may delay the cell activation and retraction by reducing the thrombin-induced Ca 2ϩ movements. Thrombin has indeed been demonstrated to cause a marked retraction of confluent endothelial cells coincident with the sustained phase of [Ca 2ϩ ] i response and paralleled by the formation of gaps in F-actin distribution (4). The retraction of endothelial cells leads to disruption of the cell monolayer thereby altering the vascular function.
In vascular endothelial cells, the temporal and causal relationships between rise in [Ca 2ϩ ] i and NO production have mostly been investigated with indirect methods. The production of cGMP, considered to be the cytosolic effector of NO, appears to be directly related to the cytosolic Ca 2ϩ level (7,27,28). This was explained by the Ca 2ϩ /calmodulin dependence of  we detected here were in the same range as those previously reported with the Clark type electrode (30) but 10-fold lower than those obtained by Malinski's group in response to supraphysiological bradykinin concentration (8).
A parallel between [Ca 2ϩ ] i rises; cGMP formation and nitrite/nitrate production has been clearly demonstrated for thapsigargin and for an agonist of nonisopeptide-selective endothelin B receptor (31,32), whereas the results obtained with ATP are less obvious. The question of whether the ATP-induced early cGMP accumulation was dependent on or independent of the NOS activation remains an open question because discordant effects of NOS inhibitors on nitrite and cGMP accumulation have been reported (7,32).
The present study showed that the inhibitory effect of IVIg on Ca 2ϩ movements and NO production depended on the substance stimulating the guinea pig endothelial cells. No effect was observed in cells treated with thapsigargin or with high ET-1 concentrations, although both stimuli increased [Ca 2ϩ ] i and NO production. Thapsigargin-elicited Ca 2ϩ movements have been demonstrated to be phospholipase C-independent (6), and ET-1-induced activation of the constitutive NOS appears to be dependent on Ca 2ϩ /calmodulin and protein tyrosine kinase (33). This suggests that interactions of IVIg with Ca 2ϩ pathways depend on phospholipase C activation. This proposal agrees with the suppression by IVIg of the first phase of Ca 2ϩ response to thrombin and ATP. This first step has been clearly demonstrated to be mediated by phospholipase C (for review see Ref. 34). The inhibitory action of human IVIg and rat IgG on this first phase of Ca 2ϩ signals could therefore be mediated at least in part through the interactions of natural antibodies with phospholipase C. Furthermore, our results also indicated that the action of human IVIg and rat IgG on NO production is more likely related to their effects on the sustained phase of agonist-induced [Ca 2ϩ ] i elevations. This agrees with a recent study demonstrating a clear relationship between nitrite production and sustained [Ca 2ϩ ] i elevation (32). The inhibition of the initial Ca 2ϩ response to agonists seems to be a common feature of natural antibodies, whereas their action on the sustained [Ca 2ϩ ] i rise and NO production appears to reveal their specificity.
The present study indicates that natural antibodies are capable of modulating Ca 2ϩ homeostasis and NO production in stimulated endothelial cells. Their mechanisms of action remains to be better defined. Future studies must be directed toward the action of natural antibodies on the metabolism of intracellular messengers associated to phospholipase C activation and on the production of vasoactive factors of endothelial origin.