Hypoosmotic stress activates a phosphorylation-dependent oxidative burst. In-gel kinase assays were performed to characterize the protein kinases that could be implicated in osmoregulation and in the activation of the oxidative burst. Hypoosmotic stress activated several kinases among which 50 and 46 kDa proteins displayed mitogen-activated protein kinase (MAP kinase) properties. They phosphorylated myelin basic protein in the absence of calcium, were recognized by antibodies directed against human MAP kinases, and were phosphorylated on tyrosine. Immunoprecipitation with an antibody directed against the tobacco MAP kinase Ntf4 showed that at least one of the activated kinases would be Ntf4-like. Apigenin, a MAP kinase and cyclin-dependent kinase inhibitor which prevents the hypoosmotically induced oxidative burst ( Cazaléet al. 1998 ; Plant Physiol. 116, 659–669), inhibited these kinases in vitro suggesting that they may play a role in the activation of the oxidative burst. Like the oxidative response, activation of the kinases depended on extracellular calcium influx and protein kinases sensitive to staurosporine and 6-DMAP. However, kinase activation did not depend on effluxes through anion channels or on the oxidative burst. Two-dimensional in-gel kinase assays revealed the presence of three protein kinases with an apparent molecular mass of 50 kDa and one of 46 kDa, all four being activated by hypoosmotic stress. The same kinases were also activated by oligogalacturonides and salicylic acid, underlying the importance of these MAP kinases as common components of different signaling pathways triggered by different extracellular stimuli.
Plant osmoregulation processes have been described in stomatal guard cells but are less well known in other cell types. In guard cells, osmoregulation involves water and ion fluxes ( Schroeder & Hedrich 1989), and its regulation depends on phosphorylation and calcium signaling ( Li et al. 1998 ; McAinsh et al. 1997 ; Pei et al. 1997 ). Phosphorylation cascades induced by osmotic stress have been identified in animals, yeasts and bacteria. Upon osmotic stress of bacterial or yeast cells, an osmosensor is activated that induces a histidyl-aspartyl phosphorelay which then activates a mitogen-activated protein (MAP) kinase pathway ( Kültz & Burg 1998 for review). The induction of a MAP kinase pathway by cell swelling was also observed in mammals ( Schliess et al. 1995 ). Upon perception of the signal, a kinase cascade is induced which results in the activation of the MAP kinase by dual phosphorylation on tyrosine and threonine residues. MAP kinases have been shown to act as gene transcription activators translocated to the nucleus ( Ferrigno et al. 1998 ) but some could also act directly in the cytosol on phospholipase A2 or cytoskeleton-associated proteins ( Davis 1993).
Activation of protein kinases has also been shown to occur in response to osmotic stress in algae. In Lamprothamnium, a calcium-dependent protein kinase is activated in response to hypoosmotic treatment ( Yuasa et al. 1997 ). In halotolerant green alga, osmotic stress activated 40 kDa kinases, LAP kinase and HAP kinase (low- and high-pressure activated protein kinases) ( Yuasa & Muto 1996). The LAP kinase phosphorylated MBP and histone in vitro, similarly to MAP kinases, but although its activation was phosphorylation-dependent, it was not recognized by anti-phosphotyrosine antibodies. Hyperosmotic stress also led to a transient activation of a 74 kDa plasma membrane protein kinase, most probably identified as a MAP kinase kinase ( Chitlaru et al. 1997 ). In higher plants, very little information is available regarding osmotically induced kinase activities. A Medicago MAP kinase MMK4 was shown to be activated by drought ( Jonak et al. 1996 ). Three protein kinases of 50, 75 and 80 kDa were activated by hypoosmotic stress on BY2 tobacco cells and depended on upstream phosphorylation and calcium transduction steps ( Takahashi et al. 1997 . Only the 50 kDa was phosphorylated on tyrosine. Together, these data suggest that MAP kinase cascades may be involved in plant osmoregulation.
Hypoosmotic stress has been shown to induce, together with ion fluxes, an oxidative burst both in animals ( Miyahara et al. 1993 ) and plants ( Cazaléet al. 1998 ; Yahraus et al. 1995 ). This increase in active oxygen species production has been previously known as a rapid defence response to various elicitors like pathogenic molecules from bacteria and fungi and plant cell wall-derived molecules ( Wojtaszek 1997; for review). In all models tested, the onset of the oxidative burst depends on protein phosphorylation events but, until now, only the Ser-Thr kinase Pto has been shown to be involved in this activation ( Chandra et al. 1996 ). The Pto kinase is specifically involved in the oxidative burst induced by avrPto elicitation and is not required for responses to non-specific elicitors like oligogalacturonides or for osmotic stress. MAP kinase activation has been evidenced in response to pathogens in various cell types, but its involvement in oxidative burst activation is not clearly established. Ligterink et al. (1997) showed MAP kinase activation following Pep-13 elicitation in parsley. The kinase did not depend on the oxidative burst, even if anion channel involvement was a common transduction step in their activation. In response to Avr9 elicitation of Cf9 tobacco suspension cells and plants, SIPK and WIPK activations were not upstream of the oxidative burst induction, as they depended on diverging transduction steps ( Romeis et al. 1999 ).
In tobacco cells, some transduction steps of the hypoosmotically induced oxidative burst have been previously identified. This oxidative response depends on extracellular calcium, anion channel opening and protein phosphorylation event(s) ( Cazaléet al. 1998 ). In the present work, the protein kinases activated by the hypoosmotic stress are investigated and among them, some are identified as MAP kinases. The transduction pathway leading to their activation and their potential involvement in the oxidative burst induction is investigated. Finally, the question of the specificity of the hypoosmotic activation of the protein kinases is tested using other stimuli such as oligogalacturonides and salicylic acid.
Several protein kinases are activated in response to hypoosmotic stress
To investigate the protein kinase activities after hypoosmotic stress, cell suspensions were transferred to an isoosmotic medium or a hypoosmotic medium, and the kinetics of kinase activation were assayed by an in-gel assay using MBP (myelin basic protein) as a substrate and in the presence of calcium ( Fig. 1a). Numerous MBP-phosphorylating activities were observed, but of particular note was the activation of 50 and 46 kDa protein kinases which occurred 2 min after the hypoosmotic stress. This activation peaked at 5 min and lasted for at least 30 min. Induction factors were evaluated to be 2.0 ± 0.2 (n = 4) and 3.5 ± 0.9 (n = 4) fold for the 50 and 46 kDa protein kinases, respectively. Another protein kinase with an apparent molecular mass of 44 kDa was activated after 30 min. The transfer to isoosmotic medium resulted in a slight, delayed activation. Transfer of suspension cells to 600 mOsm hyperosmotic medium also resulted in kinase activation ( Fig. 1b). Two kinases with molecular masses of 50 and 44 kDa were activated 5 min after the hyperosmotic stress. Comparison of the kinase profiles in hyperosmotically and hypoosmotically stressed cells ( Fig. 1b) indicates that the response is different in the two situations.
The kinase activity was assayed in the presence of 1 m m EGTA or 1 m m CaCl2 to determine the calcium dependence of the kinases induced by hypoosmotic stress ( Fig. 1c). Four protein kinases with molecular masses between 56 and 75 kDa were detected only in the presence of calcium; these kinases did not appear to be activated significantly by hypoosmotic stress. In contrast, activation of 50 and 46 kDa kinases by hypoosmotic stress also occurred in the absence of calcium thus demonstrating their calcium independence. In these conditions, induction factors were 3.1 ± 1.3 (n = 4) and 4.4 ± 2.2 (n = 4). Protein kinase activity was tested without a substrate, or with histone or casein as substrates, in the absence of calcium ( Fig. 1d). No hypoosmotic-induced kinase activation was observed in the absence of a substrate or in the presence of casein. Using histone as a substrate, the 50 and 46 kDa kinases, as well as a 75 kDa kinase, were activated by hypoosmotic stress. Therefore, hypoosmotic stress leads to the activation of 50 and 46 kDa calcium-independent protein kinases that phosphorylate both MBP and histone. These kinases were further characterized as potential members of the MAP kinase family. The analysis was performed using an enriched supernatant fraction (see Experimental procedures), and the kinase activity assays were undertaken using MBP as a substrate in the absence of calcium.
Activated protein kinases may be identified as MAP kinases
Western blot experiments using an antibody directed against ERK1 and ERK2 (extracellular signal-regulated kinase), two mammalian MAP kinases involved in osmotic regulation ( Schliess et al. 1995 ), detected two major bands with the same electrophoretic mobility as the activated 50 and 46 kDa kinases among several other bands ( Fig. 2a). Other specific animal kinase antibodies (against 14.3.3, ERK2, ERK3, MST1 or p97MAPK) did not recognize proteins with the same apparent molecular weight (data not shown). This suggests that the kinases recognized by the anti-ERK1/ERK2 antibody were of the ERK1 type. The anti-ERK1/ERK2 antibody was able to immunoprecipitate the 50 and 46 kDa kinases activated by hypoosmotic stress ( Fig. 2b). An antibody directed against phosphotyrosine is able to immunoprecipitate the two kinases, with low binding in the presence of the competitor phosphotyrosine ( Fig. 2c). The non-competitor phosphoserine did not impede the immunoprecipitation. This indicates that the two kinases are phosphorylated on tyrosine, like activated MAP kinases. These data strongly suggest that the 50 and 46 kDa kinases are activated MAP kinases. Antibodies directed against two tobacco MAP kinases, Ntf4 and Ntf6, were used to identify the kinases more precisely. Each antibody did not cross-react with the other MAP kinase ( Wilson et al. 1997 ). The activity of the immunoprecipitates could only be assayed in tube as the detection was too low by in-gel kinase assay (data not shown) and then the MBP phosphorylation was visualized. Immunoprecipitation by anti-Ntf4 revealed an increase of MBP phosphorylation after hypoosmotic stress while anti-Ntf6 did not immunoprecipitate any activated kinase activity ( Fig. 2d). Thus, at least one kinase of the Ntf4 family was activated by hypoosmotic stress but the in-tube assay does not allow for discrimination between the 50 and 46 kDa kinases. This last result, based on the use of a plant MAP kinase antibody, strengthened the identification of one activated kinase as a MAP kinase. It also excluded a possible involvement of Ntf6-like protein kinase.
Several 50 kDa MAP kinases are induced by hypoosmotic stress
Most MAP kinases have similar molecular weights. To better distinguish between them, the protein kinase activity was assayed after two-dimensional electrophoresis (2D). In-gel kinase assay on 2D gels with pH 3–10 strips only revealed activated proteins in the pH 4–7 range (data not shown). Therefore, a pH 4–7 separation was used to gain better resolution. In-gel kinase activity on 2D gel revealed three activated protein kinases, one with an apparent molecular mass of 46 kDa, and two of 50 kDa differing in charge ( Fig. 3a). It should be noted that in 2D gels the 46 kDa kinase was as active as the two 50 kDa kinases taken together, contrary to the lower activity of the 46 kDa kinase compared to the 50 kDa kinase activity observed by SDS-PAGE. One possible explanation is that isoelectric focusing separation modified the 50 kDa kinase renaturation. It is also possible that some of the 50 kDa kinases were not renatured or that both of the detected 50 kDa kinases had lost activity.
Protein kinase inhibitors were added during the in-gel kinase activity assay to assess the correspondence between the bi- and mono-dimensional in-gel assays and to try to differentiate the two 50 kDa kinases. Staurosporine, a general kinase inhibitor, prevented the activity of the 50 kDa kinases but allowed the activity of the 46 kDa kinase, while apigenin, an inhibitor of MAP kinases, prevented both protein kinase activities ( Fig. 3a). Similar results were obtained after SDS-PAGE, with 50 kDa kinases fully inhibited by staurosporine and both 50 and 46 kDa kinases highly sensitive to apigenin ( Fig. 3b). Therefore, the kinases detected on 2D gels should correspond to the kinases studied by SDS-PAGE.
Hypoosmotically induced MAP kinases can be activated by molecules inducing defence responses
The question of the specificity of activation of the 50 and 46 kDa kinases was addressed. Cells were treated with oligogalacturonides, which are elicitor molecules, and salicylic acid, a defence response molecule. These molecules are known to activate MAP kinases in addition to defence responses ( Lebrun-Garcia et al. 1998 ; Zhang & Klessig 1997). They allowed the comparison of the specificity of the kinases and the effect of the physical or chemical nature of the signal. The 50 and 46 kDa kinases detected by in-gel kinase assay after SDS-PAGE were activated rapidly by oligogalacturonides and salicylic acid, as previously shown after hypoosmotic stress (data not shown). In 2D kinase assays, it appeared that these signals activated three 50 kDa and one 46 kDa kinase after 5 min ( Fig. 4a). No active MAP kinase was detected in the absence of the signal molecules (data not shown). Two of the 50 kDa proteins and the 46 kDa protein corresponded to the kinases activated by hypoosmotic stress as they had the same isoelectric points and molecular masses. Oligogalacturonides activated the 46 kDa kinase more strongly when compared to the other signals.
The activation of three 50 kDa kinases by oligogalacturonides and salicylic acid instead of two ( Fig. 3) was potentially an important difference with the hypoosmotic stress activation. The question was addressed as to whether a third 50 kDa kinase could be induced by hypoosmotic stress at a different time point. The kinetics of kinase activation after hypoosmotic stress were then followed in 2D gels ( Fig. 4b). Apart from the 46 kDa activation, the activation of three 50 kDa kinases corresponding to the kinases activated by the chemicals was detected. Two of the three 50 kDa kinases displayed increased activity which then decreased with time, while the third appeared to be activated very early. Longer time courses of up to 20 min indicated a decrease in all activities (data not shown). Slight kinetic variations were observed from one experiment to another and spot 3 could also be visualized after 5 min in some cases. The 50 kDa kinases probably correspond to different protein kinases, with forms 1 and 3 differing in charge by 0.3 pH unit. These results show that the activation of the 50 kDa kinases is not specific to oligogalacturonide and salicylic acid signaling, but is also induced by hypoosmotic stress. Two minutes after the hypoosmotic stress treatment, a second 46 kDa kinase was observed ( Fig. 4b). The two 46 kDa kinases could be the same kinase with different degrees of phosphorylation, leading to slight changes in the isoelectric point.
Transduction pathways leading to the activation of the 50 and 46 kDa kinases and of the oxidative burst may be related
The transduction pathway leading to the activation of the 50 and 46 kDa kinases was investigated. Gadolinium, an inhibitor of stretch-activated channels, inhibited activation in both kinases ( Fig. 5a). This suggests the involvement of these channels in the kinase activation through mechanical perception of volume change at the membrane surface. Use of lanthanide, known to be an inhibitor of calcium channels, or depletion of extracellular calcium by EGTA inhibited the activation of both kinases ( Fig. 5a). Therefore, calcium influx is necessary for kinase activation as it is for the hypoosmotically induced oxidative burst ( Cazaléet al. 1998 ; Fig. 5g).
Anion channel inhibitors, such as A9C and DIDS, have been previously shown to efficiently inhibit the chloride effluxes induced by hypoosmotic stress in tobacco cell suspensions ( Cazaléet al. 1998 ). It is shown here that A9C and DIDS did not inhibit the activation of the 50 and 46 kDa kinases ( Fig. 5b), which thus would not depend on anion fluxes, unlike the oxidative burst ( Cazaléet al. 1998 ; Fig. 5g).
The kinase inhibitors staurosporine and 6-dimethylaminopurine (6-DMAP) when added in vivo inhibited the activation of the kinases ( Fig. 5c). Therefore, the 50 and 46 kDa kinase activation should be dependent on upstream phosphorylations by kinases sensitive to staurosporine and 6-DMAP, thus suggesting the involvement of a kinase cascade in response to hypoosmotic stress (such as for oxidative burst induction). Apigenin, which inhibits oxidative burst ( Fig. 5g), did not act on transduction steps upstream of the MAP kinases ( Fig. 5c) but acted directly on MAP kinase activity ( Fig. 3). These results suggested that MAP kinases may be upstream of the oxidative response. Calyculin and cantharidin, two protein phosphatase inhibitors able to slowly induce an oxidative burst response in tobacco cells ( Mathieu et al. 1996b ), activated the 50 and 46 kDa kinases in the absence of hypoosmotic stress with the same lag time ( Fig. 5d). This result is consistent with the involvement of Ser or Thr phosphorylations, inhibited by protein phosphatase 1 or 2A, prior to the activation of MAP kinases and oxidative burst.
The effect of the oxidative burst on the kinase activation was investigated. First, mechanical stress, another way to induce oxidative burst ( Cazaléet al. 1998 ), could also activate the 50 and 46 kDa kinases ( Fig. 5e). IDP, which prevented the hypoosmotically induced oxidative burst in tobacco ( Cazaléet al. 1998 ; Fig. 5g), did not inhibit the activation of the protein kinases ( Fig. 5f). Therefore, oxidative burst is not upstream of the activated protein kinases. Consistent with the suggestion of MAP kinases upstream of the oxidative burst, the addition of 1 m m H2O2 suppressed the activation of the kinases ( Fig. 5f). This result suggests a feedback control by the hypoosmotically induced active oxygen production. The amount of H2O2 was chosen taking into account the short duration of application and the scavenging by peroxidases and catalases in the extracellular medium.
Using in-gel kinase assays, 50 and 46 kDa protein kinases were shown to be activated by physical stimulation such as hypoosmotic and mechanical stresses in tobacco cell suspensions. These protein kinases displayed MAP kinase properties both enzymatically and immunochemically. They could phosphorylate MBP and histone in the absence of calcium. They were recognized by anti-mammalian MAP kinase antibodies and were shown to be phosphorylated on tyrosine residues like activated MAP kinases. At least one of them was recognized by an antibody directed against the tobacco MAP kinase Ntf4. This result would be in agreement with the activation of Ntf4 by pollen hydration, which can be considered as an osmoregulation response ( Wilson et al. 1997 ). However, as the antibody used is directed against the carboxy terminus of Ntf4, it cannot be excluded that it is also able to recognize a closely related tobacco MAP kinase such as SIPK, which displays a very similar sequence to Ntf4 (92.5% identity between the 68 residues of the immunoreactive Ntf4 peptide and the corresponding peptide from SIPK). In the tobacco cell suspension used, antibodies directed against Ntf3 did not recognize any protein in the range 45–50 kDa (data not shown) and it was observed that antibodies directed against Ntf6 did not immunoprecipitate activated MAP kinases, so none of these MAP kinases appear to be involved in the response to hypoosmotic stress. MAP kinases activated by physical stress such as touch, stirring or drought have been described previously. MMK4, from Medicago, was activated by stirring ( Bögre et al. 1996 ) as well as drought ( Jonak et al. 1996 ), but did not respond to high salt or heat shock. Tobacco SIPK and WIPK could also be activated by stirring ( Romeis et al. 1999 ). It might be expected that SIPK and WIPK could also be activated by hypoosmotic stress. Moreover, the 46 kDa protein kinase was shown here to be insensitive to staurosporine, like WIPK, and the 50 kDa protein kinases were sensitive to staurosporine, like SIPK. On the other hand, two-dimensional assays demonstrated the activation of three different 50 kDa protein kinases by hypoosmotic stress. It can be hypothesized, therefore, that the 46 kDa kinase corresponds to WIPK and one of the three 50 kDa kinases to SIPK. The other two 50 kDa protein kinases could be different from SIPK, even if phosphorylation events, independent of the activating dual phosphorylation, and modifying the isoelectric point cannot be fully excluded. Specific antibodies would be necessary to determine whether SIPK corresponds to any of the three 50 kDa protein kinases. Enrichment by protein purification would allow us to determine the identity of the kinases by microsequencing and/or mass spectrometry. Both antibody engineering and protein purification approaches are currently being developed.
It was also shown that the 50 and 46 kDa protein kinases were sensitive to apigenin, which could be an interesting tool for their identification. Apigenin has been demonstrated here for the first time to be an efficient inhibitor of plant MAP kinases. As it inhibits all of the four protein kinases, it is most probably a general inhibitor of MAP kinases. In contrast to staurosporine and 6-DMAP, apigenin did not prevent the transduction pathway leading to the MAP kinase activation, suggesting that apigenin has a narrow specificity because it did not inhibit upstream protein kinases.
Oligogalacturonides and salicylic acid also activated the same number of MAP kinases with identical apparent molecular masses and isoelectric points compared to those activated by hypoosmotic stress. Considering the previous hypothesis that one of the 50 kDa kinase corresponds to the SIPK and the 46 kDa kinase to the WIPK, it enlarges the role of these two MAP kinases to hypoosmotic and mechanical stresses and would confirm the activation of WIPK by salicylic acid ( Romeis et al. 1999 ). The role of WIPK therefore seems to be wider. It was previously described in gene-for-gene responses ( Romeis et al. 1999 ; Zhang & Klessig 1998b) and in response to wounding ( Seo et al. 1995 ; Seo et al. 1999 ).
The question of the relationship between salicylic acid and oxidative burst appears to be particularly interesting. On one hand, salicylic acid was reported to induce activated oxygen production ( Chen et al. 1993 ; Kawano et al. 1998 ) and activate SIPK and WIPK ( Romeis et al. 1999 ; Zhang & Klessig 1997). On the other hand, it has been shown here that hypoosmotic stress activated several protein kinases which may participate in the oxidative burst induction. The same protein kinases are inducible by salicylic acid and it is tempting to speculate that activation of oxidative burst by salicylic acid may proceed via MAP kinases.
Common transduction steps have been found between the hypoosmotically induced oxidative burst ( Cazaléet al. 1998 ) and 50 and 46 kDa protein kinase activation. Both depend on extracellular calcium and upstream phosphorylation events involving kinases sensitive to staurosporine and 6-DMAP. The inhibition of phosphatases 1 and 2A induced both responses which also suggests common deactivation steps. Bromophenacyl bromide and quinacrine, which prevent phospholipase A2 activity, inhibited both activations (data not shown and M.A. Rouet-Mayer, personal communication). A similar involvement of phospholipase A2 in oxidative burst and protein kinase induction was also reported, in the case of Avr9 elicitation of Cf9 tobacco cells ( Romeis et al. 1999 ). The oxidative burst is unlikely to be upstream of the protein kinase activation which is insensitive to the burst inhibitor IDP. In addition, hydrogen peroxide inhibited the kinase activation. Apigenin prevents the oxidative burst without inhibiting the staurosporine and 6-DMAP sensitive protein kinases involved in the oxidative burst induction. Inhibition of the oxidative burst by apigenin was not due to its hydrogen peroxide scavenger properties because, in the performed assay, scopoletin had more affinity for hydrogen peroxide than apigenin ( Cazaléet al. 1998 ). Thus, the 50 and 46 kDa MAP kinases whose activity is prevented by apigenin could be usptream of the oxidative burst. The time course activation of oxidative burst, detected after 2 min ( Cazaléet al. 1998 ), is consistent with the rapid activation of the MAP kinases ( Figs 1a and 4b). Nevertheless, diverging transduction steps were found between the kinase activation and the oxidative burst induction. Although the oxidative burst depended on anion channel opening, the kinase activation was not affected by the anion channel inhibitors A9C and DIDS. The inhibitory effect of H2O2 observed here on protein kinase activation, and the partial inhibition of the anion effluxes by the oxidative burst ( Cazaléet al. 1998 ), suggest the involvement of feedback control mechanisms.
The hypoosmotic response transduction pathway described here on tobacco cells differs from those induced in response to other stimuli. The MAP kinase which is activated in response to elicitation by Pep-13 in parsley depended on anion channel opening ( Ligterink et al. 1997 ) whilst, in hypoosmotic response, protein kinase activation was shown to be independent of it. In the Avr9 elicitation of Cf9 tobacco, the MAP kinase activation could be inhibited by PD98059, an inhibitor of MAP kinase kinases, with an IC50 of 50 μm, whilst this inhibitor did not impede the oxidative burst ( Romeis et al. 1999 ). Calyculin activated the oxidative burst without activating the MAP kinases ( Romeis et al. 1999 ). Lebrun-Garcia et al. (1998) also found in tobacco that cantharidin and calyculin did not activate MAP kinases. These data strongly suggest that the MAP kinases activated on a defence response pathway are independent of the oxidative burst induction. In hypoosmotically induced tobacco cells, PD 98059 at 100 μm and 250 μm did not affect MAP kinase activation or oxidative burst (data not shown) and the two phosphatase inhibitors (calyculin and cantharidin) activated both MAP kinases and oxidative burst (herein and Mathieu et al. 1996b ). These data are in agreement with protein kinase inhibition by apigenin, suggesting that in response to hypoosmotic stress the MAP kinase could be upstream of the oxidative burst. The different effects of PD98059, depending on the experimental system used, might indicate that the specificity of the response to a particular signal occurs at the level of the MAP kinase kinase.
Chemical and physical stimuli are probably recognized by different receptors. Although receptors to some elicitors have been identified ( Nennstiel et al. 1998 ), the receptor for physical stress has not yet been identified in plants, while in osmoregulation of bacteria and yeast it has been identified as a histidine kinase ( Kültz & Burg 1998). The plant ethylene receptor ETR1, which shows histidine kinase properties, is not required in response to mechanical stimulation ( Johnson et al. 1998 ). Therefore, even if common transduction steps can be found in response to both kinds of stimuli, the initial steps would still present peculiarities, as shown here, and the MAP kinases could be at the crossroad of the transduction pathways.
The different signals could also cause cytoplasmic acidification ( Mathieu et al. 1996a ) which then induces MAP kinase activation ( Tena & Renaudin 1998). NPPB, an anion channel inhibitor which prevented the chloride effluxes and the oxidative burst ( Cazaléet al. 1998 ), activated the same kinases in the absence of hypoosmotic stress, as indicated by mono- and bi-dimensional gel analysis (data not shown). NPPB can induce cytosolic acidification ( Brown & Dudley 1996). MAP kinase activation would then be expected to induce the oxidative burst if they were involved in its activation, but intense cytoplasmic acidification can also inhibit the oxidative burst (Y. Mathieu, personal communication).
The general question on how the specificity of the physiological response to a given signal is established remains unanswered. The two-dimensional assay analysis of the kinetic of the hypoosmotically activated kinases suggests that the kinetic behaviour, the intensity of activation, as well as the ratio of activation between the different kinases could be a means for discriminating signal responses. Results from the study of the specificity of calcium signaling systems ( McAinsh & Hetherington 1998) should be of great help in giving keys for decoding the protein kinase language.
Tobacco cells (Nicotiana tabacum cv Xanthi) were cultured in B5 Gamborg medium with 1 μm 2,4-D and 60 n m kinetin in constant light. Suspension cells were used after 4 days subculturing with 60–100 mg FW ml–1 cell density.
Stock solutions of 100 m m A9C, 100 m m DIDS, 0.5 m m staurosporine, 250 m m 6-DMAP, 500 m m apigenin and 20 m m IDP chloride were prepared in DMSO. Stock solutions of 500 m m gadolinium chloride, 250 m m lanthanide nitrate, 100 μm calyculin and 100 m m cantharidin were prepared in water.
Suspension cell treatments
The osmotic stress was applied as described previously ( Cazaléet al. 1998 ). Briefly, cells were washed and equilibrated for 2 h in an ion-poor medium which is isoosmotic to the culture medium (160 mOsm) and contained 10 m m Mes-Tris pH 5.2, 1 m m CaSO4 and 150 m m sucrose. Afterwards, extracellular medium was replaced either by the same volume of hypoosmotic medium (10 m m Mes-Tris pH 5.2, 1 m m CaSO4, sucrose-free, 15 mOsm), or hyperosmotic medium (10 m m Mes-Tris pH 5.2, 1 m m CaSO4 and 500 m m sucrose, 640 mOsm), or fresh isoosmotic medium. Hyperosmotic medium corresponds to a plasmolysis-inducing medium for tobacco cells. The final osmotic strengths of extracellular media after transfer of the cells were about 40 and 600 mOsm for hypo- and hyperosmotic conditions, respectively. Osmolarity was monitored using a freezing point osmometer (Roebling, Berlin, Germany) on 100 μl aliquots.
The mechanical stress was applied as described previously ( Cazaléet al. 1998 ). Briefly, a cross-section rod was maintained for 10 or 15 min in open vials containing aliquots of cell suspensions previously equilibrated for 2 h in isoosmotic medium.
For treatments with 15 μg ml–1 oligogalacturonides, 500 μm salicylic acid, 200 n m calyculin or 100 μm cantharidin, these chemicals were added to the 2 h isoosmotic-equilibrated cells.
To stop treatments, suspension cells were filtered, frozen in liquid nitrogen and stored at –80°C until use.
Preparation of protein extracts
Cells were ground in liquid nitrogen and homogenized at 4°C in extraction buffer (100 m m Hepes pH 7.5, 5 m m EDTA, 5 m m EGTA, 2 m m ortho-vanadate, 10 m m NaF, 20 m mβ-glycérophosphate, 5 m m DTT, 1 m m phenylmethylsulfonylfluoride (PMSF), 5 μg ml–1 leupeptin, 5 μg ml–1 antipain). Homogenate was centrifugated either at 16 000 g ( Fig. 1) or 100 000 gfor enriched supernatant ( Figs 2–7). Part of the 16 000 g supernatant proteins was pelleted after 100 000 g centrifugation, allowing a better specific activity of the studied kinases. The 16 000 g supernatant was diluted 5 : 3 in 2× SDS-PAGE sample buffer (1× = 2% SDS, 0.08 m Tris–HCl pH 6.8, 10% glycerol (w/v), 1 mβ-mercaptoethanol). Proteins of the 100 000 g supernatant were precipitated in 10% (w/v) TCA solution containing 20 m m NaF, washed twice with 80% (v/v) cold acetone and resuspended either in SDS-PAGE sample buffer or in isoelectric focusing (IEF) sample buffer containing 9 m urea, 2% (w/v) CHAPS, 0.4% (v/v) Triton X-100, 15 m m DTT, 1% (v/v) Pharmalytes pH 3–10 (Pharmacia), 10 m m NaF and 8 m m PMSF. Protein concentration was determined by the Bradford method with BSA as standard ( Bradford 1976), modified according to Ramagli & Rodriguez (1985) for protein quantification in IEF extracts.
SDS-PAGE in-gel kinase assay
Extracts containing 10–20 μg of proteins were electrophoresed on 10% SDS-polyacrylamide gels embedded with 0.2 mg ml–1 myelin basic protein (MBP), 0.5 mg ml–1 casein or histone as substrates for the kinases. The gels were then treated for washing and renaturating steps as described by Zhang et al. (1998) . For the activity, the gels were pre-incubated for 30 min at room temperature in activity buffer (40 m m Hepes pH 7.5, 2 m m DTT, 20 m m MgCl2 with 1 m m EGTA or 1 m m CaCl2) then for 1 h in 5 ml of the same buffer added with 25 μm cold ATP and 1.8 MBq 33P-ATP per gel. The gel was then washed extensively in 5% TCA (w/v) and 1% disodium-pyrophosphate (w/v) solution. For the gel treatment with kinase inhibitors, the activity buffer was completed with 500 μm apigenin or 1.5 μm staurosporine during pre-incubation and incubation steps. The protein kinase activity was revealed on the dried gels by PhosphorImager (Molecular Dynamics, Evry, France).
Two-dimensional in-gel kinase assay
For two-dimensional electrophoresis, 50 μg of proteins were loaded at the cathodic side of a 7 cm ready-made Immobiline Drystrip pH 4–7 (Pharmacia) previously rehydrated with a solution containing 8 m urea, 0.5% (w/v) CHAPS, 15 m m DTT and 0.8% (v/v) Pharmalytes pH 3–10 (Pharmacia). IEF was completed overnight with a total running of 22 kVh, at 15°C. Gel strips were equilibrated for 10 min in the classical 0.1% SDS Tris-glycine SDS-PAGE running buffer and placed on top of a 10% SDS-polyacrylamide gel containing 0.2 mg ml–1 MBP. For this second-dimensional electrophoresis, the mini BioRad apparatus was used. The in-gel kinase activity was then detected in the presence of 1 m m EGTA, as described above.
SDS-PAGE separated proteins were electroblotted onto Immobilon membrane with a buffer containing 25 m m Tris, 192 m m glycine and 20% (v/v) ethanol. The blot was blocked with 10% (w/v) defatted milk in Tris buffer salt TBS (10 m m Tris–HCl pH 7.5, 154 m m NaCl, 0.1% (v/v) Tween 20), probed with human 1 : 2000 polyclonal MAP kinase (ERK1, ERK2) antibody (Sigma BioSiences) and incubated in alkaline-phosphatase-conjugated anti-rabbit IgG (BioRad) in TBS. Antigenic polypeptides were visualized using the NBT/BCIP kit (BioRad) in Tris-acetate buffer.
Immunoprecipitation assays were performed with four different antibodies. Two of them were raised against two tobacco MAP kinases, Ntf4 and Ntf6. Antibodies (AbC4 and AbP6) were raised, respectively, against the carboxy terminus of Ntf4 ( Wilson et al. 1997 ) and against a Ntf6 peptide ( Calderini et al. 1998 ). The two others were a monoclonal anti-phosphotyrosine (PT-66, Sigma) and the human anti-MAP kinase (ERK1, ERK2) used previously for immunoblotting and coupled to agarose (Sigma).
The 100 000 g protein extract (100 μg) was incubated with 20 μl of AbC4 or 10 μl of AbP6 in immunoprecipitation buffer (25 m m Hepes pH 7.5, 5 m m EGTA, 5 m m EDTA, 5 m m DTT, 60 m mβ-glycerophosphate, 0.1 m m ortho-vanadate, 10 m m NaF, 1 m m PMSF, 5 μg ml–1 leupeptin and antipain, 150 m m NaCl, 0.5% (v/v) Triton X-100, 0.5% NP40) for 2 h. Then 30 μl of 50% Protein A-Sepharose CL4B (Sigma) was added and incubation continued for another 1 h. The immunoprecipitate was washed three times in immunoprecipitation buffer and twice in kinase buffer (30 m m Hepes pH 7.5, 15 m m MgCl2, 5 m m EGTA, 1 m m DTT, 0.1 m m ortho-vanadate, 5 μg ml–1 leupeptin and antipain). Kinase activity was assayed for 30 min at room temperature by the addition of 10 μl kinase buffer containing 0.4 mg ml–1 MBP, 20 μm unlabeled ATP, 0.074 MBq 33P-ATP to the pelleted immunocomplex. Reaction was stopped by the addition of SDS-PAGE sample buffer. After 3 min at 95°C, samples were run on a 15% SDS-polyacrylamide gel. The phosphorylated MBP was visualized by PhosphorImager on the dried gel.
Fifty micrograms or 100 μg of 100 000 g protein extract was incubated with, respectively, 50 μl anti-ERK1,ERK2-agarose (Sigma) or 5 μl anti-phosphotyrosine (PT-66, Sigma) in immunoprecipitation buffer, with the following modifications for anti-ERK1,ERK2-agarose, 75 m m NaCl, 0.1% (v/v) NP40 and no Triton X-100. Phosphotyrosine and phosphoserine were used at 10 m m for competition with anti-phosphotyrosine. After 2 h, 62.5 μl anti-mouse IgG-agarose (Sigma) was added to anti-phosphotyrosine immunoprecipitation and incubation continued for another 1 h. Immunoprecipitates were washed three times with immunoprecipitation buffer, resuspended in SDS-PAGE sample buffer and boiled for 3 min at 95°C. Supernatant fractions were electrophoresed on SDS-polyacrylamide gels embedded with MBP for in-gel kinase assay as described previously.
One representative datum among at least two independent experiments was illustrated.