A nuclear-targeted cameleon demonstrates intranuclear Ca 2+ spiking in Medicago truncatula root hairs in response to rhizobial nodulation factors

Lipo-chitooligosaccharide nodulation factors (NFs) secreted nitrogen-fixing rhizobia trigger Ca 2+ spiking in the cytoplasmic perinuclear region of host legume root hairs. In order to determine whether NFs also elicit Ca 2+ responses within the plant cell nucleus we have made use of a nucleoplasmin-tagged cameleon (NupYC2.1). FRET-based confocal microscopy using this nuclear-specific calcium reporter has revealed sustained and regular Ca 2+ spiking within the nuclear compartment of Medicago truncatula root hairs treated with Sinorhizobium meliloti NFs. Since the activation of Ca 2+ oscillations is blocked in M. truncatula nfp , dmi1 and dmi2 mutants, and unaltered in a dmi3 background, it is likely that intranuclear spiking lies on the established NF-dependent signal transduction pathway leading to cytoplasmic calcium spiking. A semi-automated mathematical procedure has been developed to identify and analyse nuclear Ca 2+ spiking profiles, and has revealed high cell-to-cell variability in terms of both periodicity and spike duration. Time-lapse imaging of the cameleon FRET-based ratio has allowed us to visualize the nuclear spiking variability in situ and to demonstrate the absence of spiking synchrony between adjacent growing root hairs. Finally, spatio-temporal analysis of the asymmetric nuclear spike suggests that the initial rapid increase in Ca 2+ concentration occurs principally in the vicinity of the nuclear envelope. The discovery that rhizobial NF perception leads to the activation of cell-autonomous Ca 2+ oscillations on both sides of the nuclear envelope raises major questions about the respective roles of the cytoplasmic and nuclear compartments in transducing this key endosymbiotic signal. that multiple phospholipid signaling pathways link Nodulation factor perception


INTRODUCTION
A key step in the initiation of the root endosymbiotic Rhizobium-legume association is the perception by the host plant of specific decorated lipo-chitooligosaccharides known as Nod factors (NFs). These bacterial signals activate a number of molecular and cellular responses in root epidermal cells and cortical root tissues which are required for rhizobial infection and/or nodule organogenesis (reviewed in Oldroyd and Downie, 2008). NF perception in Medicago truncatula root hairs is necessary for the reorientation of root hair tip-growth leading to bacterial entrapment (Esseling et al., 2003) as well as the activation of a signal transduction pathway that leads to the transcription of early nodulin genes such as ENOD11 (Charron et al., 2004). A central component of this pathway is the triggering of sustained intracellular Ca 2+ oscillations within the host root hair (reviewed in Oldroyd and Downie, 2006).
Convincing evidence indicates that this NF-specific calcium spiking response is decoded and transduced by a calcium/calmodulin-dependent kinase (CCaMK), encoded by the DMI3 gene (for doesnt make infections; Lévy et al., 2004;Gleason et al., 2006). To date three essential M. truncatula genes have been shown to function upstream of Ca 2+ spiking in the NF signal transduction pathway (Wais et al., 2000;Amor et al., 2003). NFP (for Nod factor perception; Amor et al., 2003) encodes a LysM-receptor-like kinase (LysM-RLK) thought to be directly involved in the perception of NFs. DMI1 (Catoira et al., 2000;Wais et al., 2000) encodes a putative cation channel (Ané et al., 2004) and DMI2 an LRR receptor-like kinase of unknown function (Endre et al., 2002). With the exception of NFP, these genes are also essential for initial root penetration by arbuscular mycorrhizal fungi, suggesting that the segment of the NF signal transduction pathway involving Ca 2+ signaling is conserved between the two endosymbiotic associations (reviewed in Oldroyd and Downie, 2004).
Modulations in the levels and localization of intracellular Ca 2+ in legume root hairs in response to the application of purified NFs have been studied by two approaches. Firstly, dextran-coupled single wavelength calcium-indicator dyes such as Calcium Green or Oregon Green have been microinjected into young growing root hairs (e.g. Wais et al., 2000). Cytoplasmic Ca 2+ responses were then monitored either directly or via ratio-imaging techniques using a second calcium-insensitive dye such as Texas Red (Shaw and Long, 2003). These experiments revealed that NFs in the nanopicomolar concentration range elicit a persistent oscillatory Ca 2+ response in the host cell cytoplasm, which initiates within 10 min after NF addition to the external medium.
High spatial and temporal resolution imaging further showed that the calcium spiking originates in the perinuclear region of the Medicago root hair and propagates through the cytoplasm towards the cell tip (Shaw and Long, 2003). More recently, in order to overcome the many limitations of microinjection, Miwa et al. (2006) have reported the use of a FRET-based cameleon calcium sensor (YC2.1) to monitor cytoplasmic Ca 2+ responses in transgenic M. truncatula roots. These studies revealed that NFs elicit cell-autonomous perinuclear Ca 2+ spiking in root hairs and have provided evidence that the number of consecutive calcium spikes may be critical for regulating ENOD11 gene activation. In the light of these data it was therefore particularly intriguing to discover that DMI3 CCaMK, the presumed Ca 2+ decoder, locates not to the cytoplasm but to the nuclear compartment of M. truncatula root hairs (Smit et al., 2005). This finding rapidly led to the proposal that oscillatory Ca 2+ signaling must be elicited both within the nucleus as well as around it (reviewed in Oldroyd and Downie, 2006;Oldroyd and Downie, 2008). Although direct experimental evidence for the activation of intra-nuclear Ca 2+ spiking in response to NFs is currently lacking, recent research has revealed that animal and plant cell nuclei possess their own calcium stores and associated signaling apparatus and have the potential to generate independent nucleoplasmic Ca 2+ signatures capable of regulating nuclear-specific cellular processes (reviewed in Gomes et al., 2006;Kim et al., 2009;Mazars et al., 2009 were then grown on a semi-solid support in petri dishes, with the roots covered by a gas-permeable membrane as described in Fournier et al. (2008). Confocal microscopy was used to localize and measure the fluorescence of the cameleon reporter in growing root hairs.
The confocal images presented in Figure 1 show that NupYC2.1 specifically localizes to the nucleus of the root hair, whether in elongating (Fig. 1A) or fully-grown hairs ( Fig. 1B). Fluorescence is undetectable in the cytoplasm, even within the cytoplasmdense tip region of elongating root hairs (Fig. 1A). NupYC2.1 fluorescence appears to be homogeneously distributed within the nucleus (with the exception of the nucleolus), and the signal intensity is stable throughout imaging. The high fluorescence level of NupYC2.1 and its excellent signal-to-noise ratio make it possible to perform confocal imaging with reduced laser intensity and fast scanning mode (see Material and Methods). As a result, FRET-based ratio imaging can be performed continuously for up to 40 min with 5 s imaging intervals or for up to 20 min with 1 s imaging intervals without substantial photo-bleaching of the nuclear cameleon or any negative effects on root hair development and growth rate. In conclusion, confocal laser scanning microscopy using the non-invasive cameleon reporter NupYC2.1 allows direct monitoring of specific changes in nuclear Ca 2+ levels in M. truncatula root hairs at a high temporal and spatial resolution.

Nod factors trigger nuclear Ca 2+ -spiking in M. truncatula root hairs
To address the question of whether NFs are able to elicit Ca 2+ signaling responses within the root hair nucleus, we performed experiments using A. rhizogenes- observed sustained Ca 2+ spiking within the nucleus for more than 95% of the root hairs examined. As shown in Figure 2, this intranuclear Ca 2+ response initiated after a variable time delay (average 6 min) and continued over the entire 30 min observation period. Our experiments revealed considerable cell-to-cell variability in the nuclear Ca 2+ spiking, and this is well illustrated by the examples of low ( Fig. 2A) and high ( Fig.   2B) frequency oscillatory profiles recorded for two growing root hairs on the same root. The extent of this cell-to-cell variability in terms of the spike-periodicity and spike-duration is analysed in more detail in a later section. For approximately 50% of the root hair nuclei examined, the sustained Ca 2+ spiking was preceded by a short burst of very high frequency spiking (Fig. 2B), which comprised from 3 to 6 spikes and lasted for less than 2 min. There did not appear to be any correlation between the presence of this initial rapid spiking and the profile of the subsequent sustained spiking. A broad Ca 2+ transient was also occasionally observed (approx. 25% of nuclei) preceding the sustained Ca 2+ oscillations ( Fig. 2A).
Cytoplasmic Ca 2+ spikes generally have asymmetric profiles, resulting from the initial very rapid release of calcium from internal stores such as the ER, followed by the much slower pumping of calcium back into the store (Oldroyd and Downie, 2004).
The nuclear calcium spike elicited by NFs is also asymmetric, and this is clearly visible in the case of the low frequency, long-duration spiking profile shown in Fig.   2A. In order to analyse the nuclear spike anatomy in more detail we recorded the onset of nuclear Ca 2+ spiking using 1 sec instead of 5 sec imaging intervals. Figure   2C, D shows the 1 sec resolution spiking profiles for two root hairs which have similar sustained spiking frequencies. In the case of Figure 2C the main spiking is preceded by a short high-frequency spiking sequence. The higher temporal resolution reveals that, during spiking, the nuclear Ca 2+ concentration reaches a maximum level within only a few seconds (irrespective of the spike frequency). In conclusion, rhizobial NFs trigger sustained, regular and asymmetric Ca 2+ oscillations within the root hair nucleus.   Figure 3 shows that nfp, dmi1 and dmi2 mutants are all defective for the NF-elicited nuclear Ca 2+ responses described above. In contrast, growing root hairs of the dmi3 mutant responded with sustained nuclear calcium spiking (Fig. 3).
Furthermore, as for the WT, approximately 50% (n = 18) of the dmi3-1 nuclei showed an initial high-frequency spiking sequence. In conclusion, since NF-elicited nuclear Ca 2+ spiking is dependent upon the identical genes as the cytoplasmic response, it is likely that the same signal transduction pathway is responsible for triggering calcium signaling in both cellular compartments.

Variability and cell-autonomy of nuclear Ca 2+ spiking in root hairs
As illustrated in Figure 2, the nuclear Ca 2+ spiking profiles elicited in root hairs can vary significantly both in terms of the initial transient responses and the subsequent patterns of sustained spiking. These differences are not a reflection of the developmental stage of the root hair, since we selected only actively growing hairs with their characteristic polarized cytoarchitecture (Fig. 1A). In order to investigate the extent of this spiking variability and also whether the NF concentration can influence the nuclear Ca 2+ spiking response, we performed parallel experiments on more than 70 root hairs treated with either 10 -9 or 10 -11 M NF. 10 -11 M was chosen because this was the lowest NF concentration for which we could still observe clear spiking responses for over 90% of the growing root hairs. In order to analyse the Ca 2+ spiking profiles in detail we developed a mathematical algorithm capable of automatically identifying spikes and measuring their duration (see Materials and Methods and Suppl. Protocol S1). A histogram representing the distribution of the average calcium spiking periodicities for individual root hairs is presented in Figure 4.
Although spiking periodicities range from below 30 sec to above 200 sec, the majority of root hairs (approx. 90%) display spiking periodicities which lie between 50 and 150 sec. The spike duration can also be highly variable, but the majority of spikes last for between 15 and 40 sec (data not shown). Lower spiking frequencies generally correlated with longer spike durations as illustrated in Figure 2A, B. These experiments also revealed that the lag time between NF addition and the initiation of the sustained spiking varied considerably between 3.5 and 12 min. Although the data presented in Figure 4 initially suggested that there may be differences in the distribution of nuclear spiking frequencies as a function of the NF concentration, statistical analysis was unable to identify significant differences due to the high variability between root hairs and between individual plants. Finally, it should be noted that whilst the early broad calcium transient was totally absent in nuclei of 10 -11 M NF-treated root hairs, the short-duration high frequency response was still observed in over 50% of the root hairs examined.
In order to visualize the cell-to-cell variability in nuclear calcium spiking in situ we created time-lapse movies of the relative changes in the intranuclear YFP:CFP ratios in adjacent growing root hairs throughout several spiking cycles. Supplemental Movie S1 illustrates the Ca 2+ spiking responses over time for a total of 6 root hair nuclei following treatment with 10 -9 M NF. This representation clearly shows the cell-to-cell variability in nuclear spiking between adjacent root hairs and the absence of spiking synchrony. Taken together, these data illustrate the cell-autonomous nature of NFelicited calcium spiking within the nuclear compartment, as well as the extent of cellto-cell variability in terms of the time-lag prior to induction, the spiking frequency and the spike duration, even amongst neighbouring root hairs. Visualizing Ca 2+ oscillations in the form of a time-lapse movie can also provide valuable information about the spatio-temporal localization of this secondary messenger within the nucleus throughout the different phases of the spike. This is illustrated in Supplemental Movie S2, which shows both the distribution and intensity of the relative changes in the nuclear NupYC2.1 FRET-signal for a single NF-treated root hair over a 10 min period with sampling at 5 sec intervals. In addition to the expected very rapid build-up in Ca 2+ concentration within the nucleus, this time-lapse provides a clear indication that the FRET-signal ratio increases preferentially at the periphery of the nuclear compartment during this initial phase of the oscillation.

Spatio-temporal analysis of NF-elicited nuclear Ca
To improve the temporal resolution we then imaged nuclei at 1 sec intervals for up to 15 min and analyzed 40 individual spikes corresponding to the sustained spiking response from 6 different nuclei. A qualitative frame-by-frame analysis of the relative changes in the YFP:CFP ratio in NF-treated nuclei for a single spike is shown in Figure 5 and for eight consecutive spikes in Supplemental Figure S1. These images confirm that the initial very rapid Ca 2+ increase occurs primarily in the vicinity of the nuclear envelope. Maximum ratio changes were reached within a few seconds inside the nucleus (e.g. Fig. 5, frames 7 to 9), although it should be underlined that these steep ratio changes are not uniformly distributed throughout the nucleus (e.g. Fig. 5, frame 7). Following the peak, which only lasts for several seconds, the relatively lengthy return to resting levels appears to initiate within the nuclear core region before reaching the nuclear periphery. In conclusion, spatio-temporal analysis of nuclear spiking provides evidence that the Ca 2+ increase initiates predominately at the periphery of the nucleus.

Rhizobial Nod factors trigger sustained nuclear Ca 2+ spiking in root hairs
Since its initial discovery over a decade ago by Ehrhardt et al. (1996), NF-elicited cytoplasmic Ca 2+ spiking in legume root hairs has been studied in considerable detail and integrated into a complex and still poorly understood signal transduction pathway leading from NF perception to specific gene expression (reviewed in Oldroyd and Downie, 2006  less than 2 min. Although this has not been referred to previously in the context of cytoplasmic spiking, a similar high-frequency spiking sequence can often be identified in published profiles prior to the initiation of the regular and persistent perinuclear oscillations in legume root hairs (e.g., Wais et al., 2002;Shaw and Long, 2003). The significance of this brief oscillatory sequence is currently unclear, although it is obviously not essential for the initiation of the sustained spiking. The same is true for the early broad transient that is occasionally observed in root hair nuclei after treatment with 10 -9 M NF, but absent following 10 -11 M treatment. Finally, time-lapse visualization of NF-elicited Ca 2+ -spiking in nuclei of adjacent growing root hairs clearly shows that these responses are non-synchronous. The cell-autonomous nature of the calcium signaling response to NFs has also been noted for perinuclear spiking in M. truncatula (Miwa et al., 2006).

What is the origin of NF-dependent nuclear Ca 2+ spiking?
The nucleus is a functionally distinct compartment of the eukaryotic cell that is The spatio-temporal imaging of the NupYC2.1 FRET signal throughout several spiking cycles (Fig. 5 and Suppl. Fig. S1) suggests that the rapid increase in nuclear Ca 2+ levels initiates predominantly at the nuclear periphery. This is consistent with the well-established mechanism in animal cells involving Ca 2+ release into the nucleus from the lumen of the NE via the transient opening of Ca 2+ channels located on the inner face of the envelope (Gomes et al., 2006). The spatio-temporal imaging also shows that YFP:CFP ratio changes during a nuclear Ca 2+ spike are not homogeneously distributed throughout the nucleus. Although this could be due in part to the unlabeled nucleolus ( Fig. 1), it is also possible that part of the Ca 2+ release may occur via nuclear grooves and invaginations (Collings et al., 2000) or from intranuclear stores analogous to the nucleoplasmic reticulum of animal cells (Lee et al., 2006). Because Ca 2+ release inside the nucleus is an extremely rapid process with maxima being reached within seconds, significantly higher resolution imaging will now be needed in order to perform more detailed spatio-temporal analysis of NFelicited intranuclear Ca 2+ spiking.

Integrating nuclear Ca 2+ spiking into the NF signal transduction pathway
Genetic approaches have identified several key NF signal transduction components upstream of Ca 2+ spiking which are associated with either the nuclear envelope or the nucleoplasm. DMI1 encodes a putative cation channel which localizes to the nuclear periphery . Although DMI1 does not appear to be a Ca 2+ channel, experiments performed in yeast suggest that DMI1 may be involved in regulating Ca 2+ release (Peiter et al., 2007). It has been proposed that this trans-membrane protein is a K + channel located in the inner membrane of the NE capable of opening voltage-gated Ca 2+ channels, and that DMI1 might be the target of a secondary messenger generated following NF perception (Oldroyd and Downie, 2008 , 2006). Although the essential role of these two nucleoporins in activating Ca 2+ spiking remains unclear, it is possible that the nuclear pore complex is required for transporting ions or molecules such as secondary messengers into or out of the nuclear compartment. Alternatively it has been proposed that NUP133/85 may function together to localize inner nuclear membrane proteins such as ion channels (Oldroyd and Downie, 2008). The same authors have suggested that a cytoplasmically-generated secondary messenger such as inositol 1,4,5-trisphosphate (InsP 3 ) might be the key molecule transported into the nucleus via nuclear pores which then targets the DMI1 channel and thus triggers Ca 2+ influx.
However, it should also be borne in mind that the animal cell nucleus possesses its  et al., 2006). If this is also the case for plant nuclei, then it is possible that specific receptors located on the outer membrane of the NE may be involved in perceiving and transducing cytoplasmic signals generated following initial NF perception at the plasma membrane.

Gomes
Once persistent intracellular Ca 2+ spiking has been initiated, this signal needs to be recognised and transduced into specific cellular responses. There is good evidence that the key Ca 2+ -decoding protein during NF signal transduction is the DMI3 CCaMK. This protein can bind calcium both directly and in a complex with calmodulin, and it is thought that this dual binding confers the capacity to recognize an oscillatory Ca 2+ signal (Oldroyd and Downie, 2004). activation and maintenance of Ca 2+ signaling, whether in the nuclear or cytoplasmic compartments. Evidence from animal cell studies suggests that the nuclear Ca 2+ response may initially precede its cytoplasmic counterpart, although differences in timing are generally only of the order of several seconds (Echevarria et al., 2003;Leite et al., 2003). Future challenges will therefore include the development of cameleon variants with different FRET-pairs targeted to both the nucleus and the cytoplasm in order to study NF-dependent activation and potential synchronization of Ca 2+ spiking in these two sub-cellular compartments.

Plant material and Agrobacterium rhizogenes-mediated root transformation
In this study, we have used the wild-type Medicago truncatula genotype Jemalong and with a 12-h photoperiod for the first week of transformation, followed by growth at 25°C with a 16-h photoperiod. fluorescence intensities (Miyawaki et al., 1997;1999)  image series of relative ratio changes were adjusted for contrast and brightness, and pseudo-coloured in green by using ImageJ.

Mathematical-statistical tools for analysing intra-nuclear Ca 2+ responses
The NupYC2. After testing several mathematical time-dependent functions, we discovered that the asymmetric Ca 2+ peak could be successfully modeled by a third-order dynamic system defined by f (t)=t 2 exp(−t/T) where T stands for a time-constant. The pattern is then defined by a+bf((t −τ)/σ) where a is the bias (local mean value of the pattern), b is the amplitude of ratio changes depending on the peak signal power, τ is the time delay between ratio changes and σ is a scale factor. To discriminate between genuine spiking, non-FRET related spikes and background noise, we used a pattern recognition algorithm which computes the likelihood for all the parameters (values of a , b , σ and τ ). For a detailed description of this algorithm see Supplemental Protocol S1. Note that, since the additive noise is assumed to be white and Gaussian, this is equivalent to minimizing the quadratic error between the peak signal and the pattern. of residues was verified by the Kolmogorov-Smirmov test (p value > 0.05). The effect of the NF concentration on the periodicity and duration of the spikes was tested by nested factor ANOVA (plant factor nested in the NF concentration).

SUPPLEMENTAL MATERIAL
Supplemental Figure S1. High spatio-temporal resolution of oscillating intranuclear