Phosphorylation of S344 in the calmodulin-binding domain negatively affects CCaMK function during bacterial and fungal symbioses

Authors

  • Pratyush Routray,

    1. Graduate Program in Molecular Plant Sciences, Washington State University, Pullman, WA, USA
    2. Department of Horticulture, Washington State University, Pullman, WA, USA
    Current affiliation:
    1. Department of Biochemistry, Cellular and Molecular Biology, University of Tennessee, Knoxville, TN, USA
    Search for more papers by this author
    • These authors contributed equally to this work.
  • J. Benjamin Miller,

    1. Department of Cell & Developmental Biology, John Innes Centre, Norwich, UK
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Liqun Du,

    1. Graduate Program in Molecular Plant Sciences, Washington State University, Pullman, WA, USA
    2. Department of Horticulture, Washington State University, Pullman, WA, USA
    3. College of Life and Environmental Sciences, Hangzhou Normal University, Hangzhou, Zhejiang, China
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Giles Oldroyd,

    1. Department of Cell & Developmental Biology, John Innes Centre, Norwich, UK
    Search for more papers by this author
  • B. W. Poovaiah

    Corresponding author
    1. Department of Horticulture, Washington State University, Pullman, WA, USA
    • Graduate Program in Molecular Plant Sciences, Washington State University, Pullman, WA, USA
    Search for more papers by this author

For correspondence (e-mail poovaiah@wsu.edu).

Summary

Calcium and Ca2+/calmodulin-dependent protein kinase (CCaMK) plays a critical role in the signaling pathway that establishes root nodule symbiosis and arbuscular mycorrhizal symbiosis. Calcium-dependent autophosphorylation is central to the regulation of CCaMK, and this has been shown to promote calmodulin binding. Here, we report a regulatory mechanism of Medicago truncatula CCaMK (MtCCaMK) through autophosphorylation of S344 in the calmodulin-binding/autoinhibitory domain. The phospho-ablative mutation S344A did not have significant effect on its kinase activities, and supports root nodule symbiosis and arbuscular mycorrhizal symbiosis, indicating that phosphorylation at this position is not required for establishment of symbioses. The phospho-mimic mutation S344D show drastically reduced calmodulin-stimulated substrate phosphorylation, and this coincides with a compromised interaction with calmodulin and its interacting partner, IPD3. Functional complementation tests revealed that the S344D mutation blocked root nodule symbiosis and reduced the mycorrhizal association. Furthermore, S344D was shown to suppress the spontaneous nodulation associated with a gain-of-function mutant of MtCCaMK (T271A), revealing that phosphorylation at S344 of MtCCaMK is adequate for shutting down its activity, and is epistatic over previously identified T271 autophosphorylation. These results reveal a mechanism that enables CCaMK to ‘turn off’ its function through autophosphorylation.

Introduction

Oscillations of intracellular Ca2+ concentration have been associated with numerous biological processes in plants, and their role in plant–microbe symbioses are starting to be understood (Veluthambi and Poovaiah, 1984; Poovaiah and Reddy, 1987; Ehrhardt et al., 1996; Oldroyd and Downie, 2008; DeFalco et al., 2010; Oldroyd et al., 2011). In root nodule symbiosis (RNS) and arbuscular mycorrhizal symbiosis (AMS), Ca2+ oscillations in the nucleus and perinuclear region (termed Ca2+ spiking) have been identified as one of the earliest cellular responses after plants perceive the presence of symbionts (Kosuta et al., 2008; Sieberer et al., 2009; Maillet et al., 2011; Harrison, 2012). Genetic studies on model legumes such as Medicago truncatula and Lotus japonicus have identified genes that are involved in RNS and AMS; some of these genes are required for both symbioses, revealing a common symbiotic pathway (Oldroyd and Downie, 2008). Calcium and Ca2+/calmodulin-dependent protein kinase (CCaMK) is a component of the common symbiotic pathway (Levy et al., 2004; Mitra et al., 2004).

CCaMK is a serine/threonine protein kinase with an N–terminal kinase domain, a C–terminal visinin-like domain with three EF hands and a calmodulin (CaM)-binding domain (Patil et al., 1995). The presence of EF hands and a CaM-binding domain in CCaMK strongly suggests that CCaMK is a decoder of Ca2+ spiking during RNS and AMS (Mitra et al., 2004; Gleason et al., 2006). Since its initial cloning from Lilium longiflorum, CCaMK has been identified in various legumes and non-legume plants, and has been documented as a multi-functional kinase with various roles in various parts of the plant (Liu et al., 1998; Pandey and Sopory, 2001; Ma et al., 2012).

Previous biochemical studies on lily CCaMK provided the necessary knowledge base to understand the mechanism behind the mode of action of this kinase (Takezawa et al., 1996; Ramachandiran et al., 1997; Sathyanarayanan et al., 2000). Under basal conditions when intracellular Ca2+ concentrations are low, CCaMK behaves as an inactive kinase with little or no activity, but becomes activated with increased Ca2+ concentrations. One effect of Ca2+ binding to the EF hands in the visinin-like domain of CCaMK is promotion of autophosphorylation, predominantly at T267 of lily CCaMK, which increases its affinity for CaM (Takezawa et al., 1996; Sathyanarayanan et al., 2000, 2001). Binding of CaM to CCaMK promotes substrate phosphorylation in vitro (Takezawa et al., 1996). Gains of function of CCaMK generated by either mutation of the autophosphorylation site or removal of the regulatory domains support the results of the biochemical studies (Ramachandiran et al., 1997; Gleason et al., 2006; Tirichine et al., 2006). Recently, an in vivo study of L. japonicus CCaMK, in which a loss-of-function mutant was complemented with various CCaMK mutants to understand the in planta activation mechanism of the kinase, also validated the activation model of CCaMK that was based on the above biochemical studies of lily CCaMK (Shimoda et al., 2012).

Ca2+-induced autophosphorylation at T271 in MtCCaMK (or T265 in LjCCaMK) plays a significant but complex role in the function of CCaMK as it promotes CaM-binding, but a mutation at this site creates a gain of function. A recent study on L. japonicus CCaMK showed that T265 in its unphosphorylated state forms a hydrogen bond with the side chain of R317 in the predicted CaM-binding/autoinhibitory region, and this interaction is disrupted by mutation of T265 to alanine or isoleucine (Shimoda et al., 2012). Together with previous biochemical results, this demonstrates that the native T265 (T271 in Medicago) is an ideal residue to maintain the intra-molecular interaction between the autoinhibitory domain and the kinase domain of CCaMK. The Ca2+-stimulated autophosphorylation at this position provides a potential molecular ‘switch’ to regulate the function of CCaMK. Even after detailed studies on activation and function of CCaMK in various signaling pathways, little is known about how the kinase is brought back to its basal state.

Here we report the identification of two autophosphorylation sites, S9 and S344, in MtCCaMK, and the characterization at both biochemical and genetic levels to determine the functional significance of phosphorylation at S344 in the CaM-binding/autoinhibitory region on the kinase activity and its role in plant–microbe interactions. We show that phosphorylation at S344 negatively affects the interaction of CCaMK with CaM, its kinase activity and also its interaction with interacting protein of DMI3 (IPD3). Furthermore, our studies reveal that phosphorylation at S344 blocks RNS and AMS, and spontaneous nodulation activity of T271A. From this study, it is apparent that autophosphorylation at S344 may be one way of shutting down the function of MtCCaMK, a mechanism that returns the activated CCaMK to its basal state.

Results

Calcium-stimulated autophosphorylation of T271A MtCCaMK

Previous studies on lily CCaMK have shown that T267, the equivalent of T271 in MtCCaMK, is a major autophosphorylation site of the kinase (Sathyanarayanan et al., 2001). To determine whether there is any other autophosphorylation site(s) in MtCCaMK in addition to T271, we mutated the T271 of MtCCaMK to alanine, a non-phosphorylatable mutation (T271A), and performed in vitro autophosphorylation of the mutant in the presence and absence of Ca2+ (Figure 1a). It has been shown that the T271A mutant is autophosphorylated in the presence of either EGTA or Ca2+; however, the intensity of the autophosphorylation was increased substantially in the presence of Ca2+. The autophosphorylation of the T271A mutant and its dependence on Ca2+ concentration clearly indicates the presence of other Ca2+-stimulated autophosphorylation site(s) in MtCCaMK. To further understand this autophosphorylation of the T271A mutant, we performed time-dependent autophosphorylation of both wild-type (WT) MtCCaMK and T271A proteins in the presence and absence of Ca2+. The time-dependent autophosphorylation studies of T271A in the presence of either Ca2+ or EGTA showed a continuous increase in autophosphorylation intensity as time increased, following the pattern of the WT MtCCaMK protein (Figure 1b). However, the initiation of autophosphorylation of T271A was delayed compared with the WT, and, in addition, the intensity of autophosphorylation of the the T271A protein decreased significantly. Two conclusions may be drawn from these results. First, although there are other autophosphorylation site(s), T271 is the preferred and major autophosphorylation site of MtCCaMK, and, second, like T271, the other putative autophosphorylation site(s) is/are also phosphorylated in a Ca2+- and time-dependent manner.

Figure 1.

Calcium-stimulated autophosphorylation of T271A and schematic diagram of CCaMK showing various autophosphorylation sites. Autophosphorylation of protein was performed at 30°C for 30 min (or as otherwise indicated) as described in Experimental procedures. A 0.4 μg aliquot of protein was used for each reaction. (a) The intensity of autophosphorylation of T271A in the presence of Ca2+ compared to phosphorylation in the presence of EGTA clearly indicates the presence of other Ca2+-stimulated autophosphorylation site(s) on MtCCaMK. (b) Time-dependent autophosphorylation of WT and T271A in the presence or absence of Ca2+ clearly indicates that the onset and intensity of autophosphorylation of T271A are significantly delayed and reduced compared to WT protein in both the presence and absence of Ca2+. The upper panels show the autoradiogram, while the lower panels represent Coomassie blue staining (loading control). (c) Schematic diagram of MtCCaMK showing the various domains and autophosphorylation sites identified so far. The N–terminal part of the protein has a serine/threonine protein kinase domain, the C–terminus consists of three functional Ca2+-binding motifs (otherwise known as EF hands), and the middle part contains a putative CaM-binding domain that overlaps with the autoinhibitory domain of the protein. The presence of both CaM-binding and Ca2+-binding domains clearly suggest its ability to bind to both Ca2+ and Ca2+/CaM. The amino acid sequence of the predicted CaM-binding and autoinhibitory domain is shown. The newly identified autophosphorylation sites S9 and S344, and the previously identified autophosphorylation site (T271), are indicated by black arrows. The inverted triangle shows the approximate position of a 14 bp deletion in the kinase domain of ccamk–1 that leads to a premature stop codon (Levy et al., 2004).

Identification and confirmation of S9 and S344 as additional autophosphorylation sites of MtCCaMK

To identify other autophosphorylation site(s) of MtCCaMK, we analyzed the T271A mutant protein after phosphorylating it for 30 min at 30°C using mass spectrometry (LC–MS/MS). We assumed that as T271 is the major site of autophosphorylation, its presence may hinder the identification of other autophosphorylation site(s). In these mass spectrometric studies, S9 and S344 were found to be autophosphorylation sites of MtCCaMK (Figures S1 and S2). S9 is located near the glycine-rich region of the kinase domain of CCaMK, whereas S344 is present in its CaM-binding/autoinhibitory region (Figure 1c). To further confirm the results of the LC–MS/MS, we mutated individual autophosphorylation sites to non-phosphorylatable amino acid residues with similar structural features (S9A and S344A) in the T271A background, on the assumption that these mutations would block autophosphorylation of CCaMK without affecting its overall structure. The double and triple mutant proteins were autophosphorylated in the presence and absence of Ca2+, and their autophosphorylation intensity was compared with the autophosphorylation intensity of the T271A protein (Figure 2). The double mutant proteins T271A-S9A and T271A-S344A showed reduced autophosphorylation intensity compared to the single T271A mutant. Autophosphorylation of this kinase was almost abolished for the triple mutant T271A-S9A-S344A (Figure 2a). However, further quantification of the autophosphorylation intensity of these mutants revealed that, even in the triple mutant protein, there was some residual phosphorylation, suggesting the presence of other autophosphorylation site(s) (Figure 2b). Both LC–MS/MS and site-directed mutagenesis data confirmed that the newly identified autophosphorylation sites of the MtCCaMK are genuine.

Figure 2.

Confirmation of S9 and S344 as autophosphorylation sites of CCaMK. To confirm the autophosphorylation sites identified by mass spectrometry (S9 and S344), both autophosphorylation sites were mutated to alanine individually or together with the previously identified autophosphorylation site (T271), and the autophosphorylation reaction was performed in the absence or presence of Ca2+ and CaM at 30°C for 10 min as described in Experimental procedures. A 0.4 μg aliquot of protein was used for each reaction. (a) Autoradiograms of autophosphorylated CCaMK mutant proteins. For each test, an autoradiogram (upper panel) and Coomassie blue staining (lower panel) are shown. (b) Quantification of [32P] incorporation into CCaMK mutant in the presence of Ca2+ or EGTA. Values are means ± SE of three individual replicates.

The S344D mutation blocks the MtCCaMK interaction with calmodulin

CaM binding is required for CCaMK to attain its maximal kinase activity and is an essential/indispensable process for normal rhizobial infection during RNS (Takezawa et al., 1996; Shimoda et al., 2012). Autophosphorylation at the CaM-binding domain of mammalian multi-functional Ca2+/CaM-dependent protein kinase II (CaMKII) has been reported as a mechanism to prevent further binding of CaM to activated CaMKII (Colbran and Soderling, 1990). Recently a similar observation was made for LjCCaMK, for which autophosphorylation at the CaM-binding region reduced the CaM-binding affinity (Liao et al., 2012). Therefore, identification of S344 as an in vitro autophosphorylation site prompted us to examine whether phosphorylation at this site also affects CaM binding. To study the effect of autophosphorylation of MtCCaMK at its CaM-binding domain, we mutated the serine at position 344 to alanine (S344A; phospho-ablative mutant) or aspartate (S344D; phospho-mimicking mutant), on the assumption that the S344D mutant would behave like the S344 phosphorylated form whereas autophosphorylation of the S344A protein would be blocked without much effect on the overall structure of the protein. We compared the CaM-binding properties of WT MtCCaMK, S344D and S344A mutants (Figure 3a). The results revealed that, in the presence of Ca2+, the S344A protein binds to CaM with similar or slightly higher affinity compared with WT kinase. In contrast, the S344D mutant showed a greatly reduced CaM-binding affinity, indicating a negative effect of S344 phosphorylation on CaM binding.

Figure 3.

In vitro kinase activity of S344 mutants, and their interaction with IPD3 and calmodulin. (a) The effect of autophosphorylation of CCaMK at S344 on its CaM-binding affinity was tested by transferring equal amounts (2 μg) of S344 mutant (S344A and S344D) and WT CCaMK proteins onto a poly(vinylidene difluoride) membrane, and incubating them with horseradish peroxidase (HRP)-conjugated CaM in the presence of Ca2+ (1 mM) as described in Experimental procedures. The top gel shows the CaM-binding affinities of the various CCaMK mutants, and the bottom gel shows Ponceau staining (loading control). In the presence of Ca2+, the CaM-binding affinity of the S344A mutant is similar to or higher that that of the WT CCaMK protein, but significantly reduced for the S344D mutant. (b) The in vitro substrate phosphorylation activity of S344 mutants and WT CCaMK proteins was determined in the absence of Ca2+ (5 mM EGTA), the presence of Ca2+ (0.5 mM CaCl2), or the presence of Ca2+ (0.5 mM CaCl2) and CaM (1 μm) using bovine MBP (2 μg) as the substrate. The data were normalized to the maximal kinase activity of CCaMK in the presence of Ca2+ and CaM, and values are means ± SE of three independent experiments. Substrate phosphorylation by the S344A mutant in the presence and absence of Ca2+ and CaM followed the same pattern as for WT CCaMK, and the phosphorylation activity was maximal in the presence of both Ca2+ and CaM. The S344D mutant protein showed a significant reduction in substrate phosphorylation activity relative to WT CCaMK in the presence of Ca2+ and CaM. (c) The interaction of CCaMK and various mutants (S344A, S344D, T271A and T271A-S344D) with IPD3 was tested by an in vitro yeast two-hybrid assay; the degree of interaction was calculated based on the activity of β–galactosidase and is expressed as a percentage of that for the WT CCaMK and IPD3 interaction, which is set at 100%. Values are means ± SE of three samples from one experiment, and the experiment was repeated twice with similar results.

Phosphorylation of S344 reduced the kinase activity in vitro

Ca2+ and CaM are required for CCaMK activity, with binding of Ca2+/CaM helping the kinase to attain its maximal kinase activity (Takezawa et al., 1996). As the S344D mutation significantly reduced the CaM-binding affinity of MtCCaMK, we hypothesized that it would also have an effect on the substrate phosphorylation activity of CCaMK. To further analyze the significance of S344 autophosphorylation on kinase activity, we studied the substrate phosphorylation activity of S344A and S344D mutant proteins in the presence of both Ca2+ and CaM, and compared it to that of WT kinase (Figure 3b). As bovine myelin basic protein (MBP) has been extensively used as a substrate in previous biochemical studies of lily CCaMK, and recent studies also showed that the substrate phosphorylation activities of CCaMK for both MBP and IPD3 are similar (Liao et al., 2012; Shimoda et al., 2012), we used commercially available MBP as a substrate for these in vitro phosphorylation experiments. Similar to previous studies, we found that WT MtCCaMK displayed a negligible amount of basal activity in the absence of Ca2+ and CaM, and the kinase activity increased several-fold in the presence of both Ca2+ and CaM. A similar pattern was also observed for S344A, with a 30% increase in the kinase activity of the mutant in the presence of Ca2+ and CaM compared to WT. As expected, we did not observe CaM-stimulated substrate phosphorylation for the S344D protein; in fact, we found that the basal kinase activity of the S344D protein in the absence of Ca2+ and CaM increased slightly compared to that of WT.

The S344D mutation significantly reduced the interaction between CCaMK and IPD3

IPD3 or its homolog in L. japonicus, CYCLOPS, is one of the components of the common symbiosis-signaling pathway, and hence is required for both AMS and RNS. Furthermore, IPD3/CYCLOPS has been shown to be phosphorylated by CCaMK in vitro, and physically interacts with the active version of CCaMK but not its inactive mutants (Messinese et al., 2007; Chen et al., 2008; Yano et al., 2008; Horvath et al., 2011; Ovchinnikova et al., 2011). To study the effect of autophosphorylation of MtCCaMK at S344 on its interaction with IPD3, we tested various S344 mutants in yeast cells and compared the results with those for WT MtCCaMK. Our quantitative yeast two-hybrid results showed that the LacZ enzyme activity resulting from interactions between IPD3 and the S344A mutant protein is 44% of that resulting from the IPD3 and WT MtCCaMK interaction. However, the S344D mutation drastically compromised interaction of MtCCaMK with IPD3, reducing the LacZ activity by over 90%. Furthermore, the S344D mutation in the T271A background showed a similar effect (Figure 3c). These results demonstrate that the IPD3 interaction with MtCCaMK is considerably reduced by phosphorylation of S344.

The S344D mutation of MtCCaMK negatively affects symbioses

As the S344D mutation impaired the interaction between MtCCaMK and CaM, as well as between MtCCaMK and IPD3, and reduced the kinase activity of MtCCaMK (Figure 3), it is reasonable to speculate that this autophosphorylation site may also influence symbioses. In order to study the significance of S344 phosphorylation on RNS and AMS, we transformed the loss-of-function ccamk–1 mutant with S344A and S344D mutants using Agrobacterium rhizogenes-mediated transformation. Nodulation was assessed 28 days after inoculation with Sinorhizobium meliloti carrying the hemA::LacZ reporter gene. The S344A mutant was able to complement ccamk–1 plants, and ccamk–1 plants transformed with the S344A mutant produced normal-sized pink nodules, a general indication of normal nodulation (Figure S3a). However, S344A-complemented plants produced 30% fewer root nodules per plant (Student's t test, = 0.0172; Figure 4a). This was comparable to the results for T271A-complemented ccamk–1 plants, with no statistical difference in nodule number between plants complemented with T271A and S344A mutants (Student's t test, = 0.1032). The decreased nodulation efficiency of S344A may be a non-specific effect resulting from substitution of a residue that supports the optimal functionality of CCaMK. Similar effects were also observed for T271A in MtCCaMK and T265I in LjCCaMK, which also showed sub-optimal levels of rhizobium-induced nodulation (Gleason et al., 2006; Tirichine et al., 2006). This is supported by the observation that both the S344A and T271A proteins show decreased interaction with IPD3 (Figure 3c). As T271A has been generally accepted as a gain-of-function mutant that is able to rescue RNS of the ccamk–1 mutant, it is reasonable to conclude that the S344A mutant is able to complement ccamk–1 in supporting the development of RNS (Figure 4a). To further support this, nodule cross-sections were analyzed and these showed the presence of rhizobia in nodules on S344A-resuced ccamk–1 mutants (Figure S3b,c). In contrast to S344A, the phospho-mimicking mutant S344D completely failed to complement ccamk–1 in terms of nodulation (Figure 4a and Figure S3b). These results indicate that phosphorylation of S344 is not required for establishment of RNS but plays a negative role in the function of CCaMK during RNS.

Figure 4.

The S344D mutant negatively affects symbiosis and spontaneous nodulation. The loss-of-function ccamk–1 mutants were transformed with various S344 mutants.(a) WT CCaMK and T271A mutants were inoculated with Sinorhizobium meliloti 1021. The roots were examined for formation of mature nodules 28 days after rhizobial inoculation, and the number of nodules per plant for each construct was quantified. (b) Roots of ccamk–1 plants transformed with various S344 mutants and WT CCaMK were harvested 8 weeks after inoculation with Rhizophagus irregularis, and the percentage of mycorrhizal colonization for various constructs was quantified. (c) The S344A mutant complemented ccamk–1 for both RNS and AMS, but the S344D mutant did not. To study the effect of the S344D mutation on spontaneous nodulation induced by the T271A mutation, various single and double mutants were grown in the absence of rhizobia. Spontaneous nodules were observed for the T271A mutant and the T271A-S344A double mutant, but not in ccamk–1 plants transformed with the T271A-S344D double mutant. Scale bar = 500 μm.

To analyze the effect of S344 phosphorylation on AMS, hairy roots transformed with various S344 mutants were inoculated with Rhizophagus irregularis. While supporting AMS at lower levels than that of ccamk–1 plants complemented with WT CCaMK (Student's t test, = 0.0027), ccamk–1 plants complemented with the S344A and T271A-S344A mutants supported AMS to levels that are statistically the same as for T271A-complemented ccamk–1 (Student's t test  0.2037). In contrast, the S344D mutant only supported fungal colonization at reduced levels compared to colonization on either WT- or T271A-complemented lines (Figure 4b and Figure S4). These results suggest that, similar to its role in RNS, autophosphorylation of MtCCaMK at S344 also suppresses AMS.

Autophosphorylation of MtCCaMK at T271 is considered to be a critical regulator of the function of CCaMK, and its mutation to alanine (T271A) creates a deregulated kinase that is able to complement ccamk–1 for both RNS and AMS. It also results in spontaneous nodulation in the absence of rhizobia. The negative effect on CCaMK function provides an opportunity to study how autophosphorylation at S344 affects the rhizobia-induced and spontaneous nodulation phenotypes in the T271A background. To this end, we transformed ccamk–1 plants with double mutants in which S344 is either mutated to alanine or aspartic acid together with the T271A mutation (T271A-S344A and T271A-S344D), and compared their complementing ability with that of the T271A mutant. While both the S344A and T271A mutants are able to complement the RNS-supporting capacity of ccamk–1, the functions of both are sub-optimal compared to WT CCaMK (Gleason et al., 2006). As expected, an additive effect of the T271A and S344A mutations on CCaMK function was detected. The number of nodules on T271A-S344A-complemented plants was almost half that of T271A-complemented plants (Student's t test, = 0.0053, Figure 4a). These complemented plants produced normal pink nodules that are comparable to nodules produced in T271A-complemented plants (Figure S3a). Cross-sections of the nodules confirmed normal rhizobial infection (Figure S3b,c). On the other hand, T271A-S344D did not complement ccamk–1 (Figure 4a and Figure S3b). Both T271A and T271A-S344A were able to produce spontaneous nodules at similar levels, but T271A-S344D did not (Figure 4c and Table S1). Our data suggests that phosphorylation at S344 completely turns off the RNS-supporting function of CCaMK, even in the T271A gain-of-function background.

Discussion

Autophosphorylation of a kinase is critical for its regulation, either through positive or negative effects. By mass spectrometry analysis, we identified and characterized a critical autophosphorylation site of MtCCaMK, S344, in its CaM-binding/autoinhibitory region. Although the time course of autophosphorylation clearly suggests that the previously identified T271 is the major autophosphorylation site of MtCCaMK and most likely the first site to be phosphorylated in the presence of Ca2+ (Figure 1b), the autophosphorylation of S344 in MtCCaMK can be clearly verified in vitro using site mutants of MtCCaMK (Figure 2a).

The mutation of S344 to phospho-mimicking or non-phosphorylatable forms produced different biochemical and physiological results. CaM binding to MtCCaMK is abolished for S344D (Figure 3a); in addition, in vitro phosphorylation studies of the kinase showed that the S344D mutant shows drastically reduced substrate phosphorylation activity compared to WT and S344A (Figure 3b). Consistent with biochemical studies, the S344D mutant blocked RNS and failed to rescue AMS formation in ccamk-1 (Figure 4a,b and Figures S3 and S4). Mutation of S344 to aspartic acid in addition to the T271A mutation showed a similar phenotype to that of S344D single mutant, indicating a negative epistasis effect of S344 phosphorylation over T271 autophosphorylation (Figure 4a,b and Figures S3 and S4). Furthermore, the S344D mutation suppressed spontaneous nodulation resulting from the T271A mutation (Figure 4c). This dominant negative activity of S344D over T271A clearly indicates that autophosphorylation in the CaM-binding domain counteracts the effect of the T271A mutation and returns CCaMK to its inactive state.

A study on Lotus japonicus CCaMK showed that S337 (equivalent to S343 in MtCCaMK) is an autophosphorylation site (Liao et al., 2012). At the biochemical level, phosphorylation at S337 of L. japonicus and phosphorylation at S344 of M. truncatula, both in the CaM-binding domain, negatively regulate the function of CCaMK. Consistent with the negative regulation at the biochemical level (Figure 3), our research revealed that autophosphorylation of S344 in MtCCaMK negatively affects the function of CCaMK and blocks establishment of both RNS and AMS (Figure 4). In contrast to the autophosphorylation at S344, autophosphorylation of S337 of LjCCaMK was reported to be required for establishment of RNS and AMS, a sign of positive regulation of symbioses through this autophosphorylation, and yet the phospho-mimicking mutation at S337 of LjCCaMK was also reported to block RNS and AMS (Liao et al., 2012). The different physiological responses resulting from decreased kinase activities through autophosphorylation at these two autophosphorylation sites in the CaM-binding domain imply that complex and delicate regulatory mechanism(s) are involved in fine-tuning the actions of CCaMK during bacterial and fungal symbioses.

Multiple sequence alignments of CCaMK of various plants showed that the S344 autophosphorylation sites are conserved across legumes, non-legumes, monocots and dicot plants, with the exception of the rice homolog OsCCaMK (Figure S5). OsCCaMK has a cysteine corresponding to S344 of MtCCaMK. It has been shown that complementation of the L. japonicus CCaMK loss-of-function mutant, ccamk–3, with OsCCaMK not only restores both RNS and AMS, but also produces spontaneous nodules in the absence of Nod factor (Banba et al., 2008). In another study, OsCCaMK-complemented Medicago CCaMK loss-of-function ccamk–1 mutants were only able to initiate nodule organogenesis but failed to support bacterial infection (Godfroy et al., 2006). Cysteine and serine are almost identical in structure, except that the hydroxyl group in serine is replaced with a sulfhydryl group in cysteine. The sulfhydryl group in cysteine cannot be phosphorylated, but it may form disulfide bonds with other sulfhydryl groups. The lack of the corresponding S344 autophosphorylation site in OsCCaMK is probably the major reason why OsCCaMK confers this deregulated kinase activity and spontaneous nodulation phenotype in a legume ccamk mutant. However, we are unable to exclude the possibility that these phenotypes may stem from the potential association of OsCCaMK with other sulfhydryl group-containing molecules/proteins through its cysteine residue at this critical position corresponding to S344 in MtCCaMK.

Our biochemical and genetic studies have shown that CCaMK modulates its kinase activity by autophosphorylation at various positions, and that autophosphorylation of S344 negatively affects CCaMK activity. Our assumption is that, at the start of Ca2+ spiking, the inactivated CCaMK is likely to autophosphorylate at T271, which increases its affinity for CaM and helps in attaining maximum kinase activity. After a prolonged exposure to Ca2+ spiking, CCaMK inhibits its own kinase activity by phosphorylation of S344 in the CaM-binding domain, which blocks further binding of CaM. The autophosphorylation site present in the CaM-binding domain (S344) appears to be associated with negative regulation of MtCCaMK, and blocks the activated form of CCaMK.

A similar mode of action is also observed for other CaM-binding kinases. Animal CaMKII, a well-studied CaM-binding protein kinase, was shown to have several autophosphorylation sites in its CaM-binding region (Hanson and Schulman, 1992). Moreover, CaMKII's mode of action is highly regulated by its various autophosphorylation sites. Autophosphorylation of CaMKII at its CaM-binding region reduces the sensitivity of the kinase to further Ca2+/CaM stimulation, thereby slowly decreasing its kinase activity (Colbran and Soderling, 1990; Hanson and Schulman, 1992). Similarly, negative autophosphorylation has been shown for DRP1: a pro-apoptotic Ca2+/CaM-regulated serine/threonine kinase for which autophosphorylation at S308 within the CaM-binding region reduces its affinity for CaM and its ability to form homodimers (Shani et al., 2001).

Although CCaMK has structural and biochemical dissimilarities with CaMKII, the presence of autophosphorylation at the CaM-binding region of CCaMK, its negative effect on CaM-binding and kinase activity, and, more importantly, its epistatic effect on phosphorylation of T271 (the primary autophosphorylation site) clearly suggest that CCaMK follows a similar mode of inactivation to that of CaMKII.

Experimental Procedures

Expression and purification of CCaMK

The MtCCaMK coding sequence was into cloned into the bacterial expression vector pET28B (Novagen, http://www.emdmillipore.com), and transformed into Escherichia coli strain BL21 (DE3)/pLysS. The protein was purified using Ni–NTA agarose (Qiagen, http://www.qiagen.com) according to the manufacturer's instructions, dialyzed against buffer containing 40 mm Tris/HCl (pH 7.6), 1 mm dithiothreitol, 1 mm EDTA and 10% ethylene glycol, quantified using Bradford reagent (Bio–Rad, http://www.bio-rad.com/) with BSA as the standard, and stored at −80°C in 15% glycerol.

Site-directed mutagenesis

Site-directed mutagenesis was performed using a QuikChange™ site-directed mutagenesis kit (Stratagene, http://www.genomics.agilent.com) according to the manufacturer's instructions, and was confirmed by sequencing.

Autophosphorylation assay

Autophosphorylation assays were performed at 30°C in 50 mm HEPES (pH 7.5) buffer, containing 10 mm of magnesium acetate, 1 mm dithiothreitol, 200 μm ATP and 5 μCi [γ–32P] ATP, in the presence of either 5 mm EGTA or 0.5 mm CaCl2 (Sathyanarayanan et al., 2001) for 30 min unless otherwise stated.

Protein kinase assay

An in vitro kinase assay was performed using 0.4 μg CCaMK protein per assay in a total 10 μl reaction for 30 min at 30°C in the presence of 50 mm HEPES (pH 7.5) containing 10 mm magnesium acetate, 1 mm dithiothreitol, 200 μm ATP and 5 μCi [γ–32P] ATP with either 5 mm EGTA or 0.5 mm CaCl2 and/or 1 μm bovine brain CaM (Sigma-Aldrich, http://www.sigmaaldrich.com) as described previously (Takezawa et al., 1996). Bovine myelin basic protein (MBP, 2 μg; Sigma-Aldrich) was used as substrate.

Mass spectrometry analysis

Autophosphorylation of CCaMK was performed in the presence of 0.5 mm CaCl2, and run on 10% SDS–PAGE. The desired band was cut out and sent to the Proteomic Facility of the W.M. Keck Foundation (Yale University, New Haven, CT) for mass spectrometry analysis. The phosphopeptide enrichment, LC–MS/MS analysis and identification of phosphorylated sites were performed as described by Kiraly et al. (2011).

CaM-binding assays

Arabidopsis CaM 2 conjugated with horseradish peroxidase was used for the CaM-binding assay. Purified WT and mutant CCaMK proteins were separated by SDS–PAGE and electrophoretically transferred onto poly(vinylidene difluoride) membrane (Millipore, http://www.millipore.com/) (Takezawa et al., 1996). The membrane was subsequently incubated with binding buffer (10 mm Tris/HCl, pH 7.5, 150 mm NaCl and 5% w/v non-fat dry milk) containing horseradish peroxidase-conjugated CaM with 1 mm CaCl2 for 1 h at room temperature. The membrane was then washed with binding buffer without Ca2+ and CaM three times for 10 min each. The bound CaM signal was detected using a BM chemiluminescence Western blotting kit (mouse/rabbit; Roche Applied Science, https://www.roche-applied-science.com) according to the manufacturer's instructions.

Yeast two-hybrid assay

The yeast two-hybrid assay was performed using the Hybrid ZAP 2.1 system (Stratagene). Full-length IPD3 was inserted into the prey vector pAD–Gal42.1; MtCCaMK and its mutated versions were inserted into pBD–Gal4Cam. After sequence confirmation, yeast strain YRG2 was transformed with both bait and prey constructs, and positive transformants selected on selective drop-out medium lacking leucine and tryptophan were used for β–galactosidase quantitative assays as described previously (Du and Poovaiah, 2004). Interaction tests were performed using independent colonies for each of the constructs. To maintain ideal dispersion of measurements, the maximum and minimum readings were discarded, and the values presented are the means ± SE of three samples.

Complementation

Full-length CCaMK and its mutants were cloned into a pK7FWG2–R binary vector (Smit et al., 2005) modified to remove the GFP fusion tag and to also contain a 1 kb native promoter instead of the 35S promoter (Gleason et al., 2006). Wild-type and ccamk–1 (dmi3–1) mutant seedlings were transformed with Agrobacterium rhizogenes strain AR1193 carrying these binary vectors to generate transgenic hairy roots (Boisson-Dernier et al., 2001). After 3 weeks, the plants were screened for transgenic roots on the basis of dsRED fluorescence and used for subsequent experiments.

For nodulation experiments, seedlings with transgenic roots were inoculated with Sinorhizobium meliloti carrying the hemA::LacZ reporter gene, and grown in a 1:1 mix of sterilized sand/terragreen (Oil-Dri Co.) supplemented with nitrogen-deprived nutrient medium (buffered nodulation medium) under a 16 h light/8 h dark cycle at 22°C for 28 days. The nodules and roots were stained for β–galactosidase activity as previously described (Radutoiu et al., 2003). For cross-sections, 21-day-old nodules were excised, fixed with 2.5% glutaraldehyde, dehydrated in an ethanol series (30, 50, 70, 90 and 100% ethanol, each for 30 min) and embedded in Technovit–7100 (Kulzer Histo-Technik, http://heraeus-kulzer-technik.de/) according to the manufacturer's instructions. Fixed nodules were sectioned into 10 μm thin layers using a Leica (http://www.leica.com/) EM UC7 microtome with a glass knife, and stained with toluidine blue. Nodule sections were analyzed using a Nikon (http://www.nikon.com/) Eclipse 800 microscope, and representative images were obtained using a Pixera (http://www.pixera.com/) Pro ES600 camera.

For arbuscular mycorrhizal colonization experiments, seedlings with transgenic roots were grown in a 1:1 mix of sterilized sand/terragreen (Oil-Dri Co.) containing Rhizophagus irregularis spores (Endorize, http://www.agron.co.il/en/Home.aspx) under a 16 h light/8 h dark cycle at 22°C. After 8 weeks, the roots were harvested and cleared with 10% KOH at 95°C for 15 min. The fungal structures were stained with 5% black ink and 5% acetic acid at 95°C for 5 min, and quantified using the gridline intersect method (Saito et al., 2007).

Spontaneous nodulation was performed in sterile growth pouches (Mega International, http://www.mega-international.com/) supplemented with buffered nodulation medium, and scored after 6 weeks (Gleason et al., 2006).

Acknowledgements

This project was primarily supported by the US National Science Foundation (grant number 1021344). We also acknowledge the support of the Washington State University Agricultural Research Center, the Washington State University BIOAg Program, the National Science Foundation of China (grant number U1130304), and the Gatsby Charitable Foundation in the form of a studentship to J.B.M. We wish to thank Gerhard Munsky (Washington Sate University) for his help in analyzing the LC–MS/MS data, and Andrew Davis (John Innes Centre) for help with photography.

Ancillary