The Nod factor-elicited annexin MtAnn1 is preferentially localised at the nuclear periphery in symbiotically activated root tissues of Medicago truncatula


* For correspondence (fax 33 561285061; e-mail


The Medicago truncatula MtAnn1 gene, encoding a putative annexin, is transcriptionally activated in root tissues in response to rhizobial Nod factors. To gain further insight into MtAnn1 function during the early stages of nodulation, we have examined in detail both spatio-temporal gene expression patterns and MtAnn1 activity and localisation in root tissues. Analysis of transgenic Medicago plants expressing a pMtAnn1-GUS fusion has revealed a novel pattern of transcription in both outer and inner cell layers of the root following either Nod factor-treatment or rhizobial inoculation. The highest gene expression levels were observed in the endodermis and outer cortex. These transgenic plants also revealed that MtAnn1 expression is associated with lateral root development and cell differentiation in the root apex independent of nodulation. By purifying recombinant MtAnn1 we were able to demonstrate that this plant annexin indeed possesses the calcium-dependent binding to acidic phospholipids typical of the annexin family. Antisera against recombinant MtAnn1 were then used to show that tissue-specific localisation of the MtAnn1 protein in Medicago roots matches the pMtAnn1-GUS expression pattern. Finally, both immunolabelling and in vivo studies using MtAnn1-GFP reporter fusions have revealed that MtAnn1 is cytosolic and in particular localises to the nuclear periphery in cortical cells activated during the early stages of nodulation. In the light of our findings, we discuss the possible role of this annexin in root tissues responding to symbiotic rhizobial signals.


Nitrogen-fixing nodules are specialised plant organs, which are formed on roots of leguminous plants in symbiotic association with soil bacteria known as rhizobia. For a functional nodule to be established, rhizobia must be able both to penetrate the plant root (infection process) and at the same time trigger nodule organogenesis in the inner cortex. In the symbiotic association between the model legume Medicago truncatula and its symbiont Sinorhizobium meliloti, bacteria attached to epidermal root hairs induce hair deformation and curling, thereby creating the site for subsequent root entry via cell wall enclosed tubular structures called infection threads. In parallel, rhizobia at the root surface induce mitotic activity in inner cortical cells, which leads to the formation of the nodule primordium. When the inward-growing infection threads reach the developing nodule primordia, bacteria are released from the infection thread and colonise the plant cells (for recent reviews see Gage and Margolin, 2000; Schultze and Kondorosi, 1998).

The capacity of rhizobia to initiate nodulation in the host plant is dependent on the secretion of lipochitooligosaccharide signals, the Nod factors. These molecules play a key role in the specificity of plant-bacteria recognition and in several key steps of nodulation (reviewed in Downie and Walker, 1999). When purified and applied at low concentrations, Nod factors can induce many of the typical host plant responses normally elicited by the bacterial symbiont, such as root hair deformation, cortical cell divisions and the transcriptional activation of certain plant genes known as early nodulin genes (reviewed in Bladergroen and Spaink, 1998; Schultze and Kondorosi, 1998). A number of early nodulin genes transcriptionally activated by both Sinorhizobium and Nod factors have been characterised in M. truncatula. These genes possess specific spatio-temporal expression patterns which make them useful markers for studying various aspects of Nod factor signalling (for example, Catoira et al., 2000; Vernoud et al., 1999).

A few years ago we isolated and characterised a novel Nod factor-induced gene from M. truncatula named MtAnn1, encoding a protein homologous to the annexin family of calcium binding proteins (de Carvalho-Niebel et al., 1998). Annexins are a family of structurally related proteins found in eukaryotic organisms (excepting yeast), with the capacity to bind acidic membrane phospholipids in a calcium-dependent manner. The well-studied vertebrate annexins constitute a large gene family which are involved in a variety of Ca2+-regulated cellular processes (reviewed in Gerke and Moss, 1997; Hawkins et al., 2000). Plant annexins, which represent a unique phylogenetic group within the annexin family, appear to be a less diverse gene family than in vertebrates (Clark et al., 2001; Delmer and Potikha, 1997). Functionally, plant annexins share common biological activities with their animal counterparts, such as the ability to bind to F-actin (Calvert et al., 1996; Hu et al., 2000), to stimulate Ca2+-dependent exocytosis (Carroll et al., 1998), or possessing GTPase activity (Calvert et al., 1996; Lim et al., 1998; Shin and Brown, 1999).

Prior studies of MtAnn1 expression using Northern analysis had revealed that mRNA levels are enhanced in root tissues responding to Nod factors and also during the symbiotic association with S. meliloti, both in inoculated roots and in nodule structures. By in situ hybridisation we previously demonstrated that MtAnn1 mRNA localises to distal zone II of the nodule, compatible with a role for MtAnn1 during early symbiotic processes (de Carvalho-Niebel et al., 1998). In order to further our understanding of the role of MtAnn1 during the early stages of root nodulation, we now present detailed studies of both gene expression and annexin localisation at the root tissue and cellular level. Analysis of transgenic Medicago lines expressing a pMtAnn1-GUS fusion reveal that MtAnn1 is transcribed simultaneously in both inner (primarily endodermal) and outer (cortical) root tissues in response to Nod factors and during the early stages of nodulation. These results have been confirmed by immunolocalisation studies using antisera specific to MtAnn1. Finally, high resolution in vivo studies using MtAnn1-GFP fusions expressed in transformed M. truncatula roots indicate a peri-nuclear localisation for MtAnn1 in S. meliloti-activated cortical cells. Possible roles for this nodulation-associated annexin in Nod factor-elicited signalling and cortical cell differentiation are discussed.


Tissue-specific transcription of the MtAnn1 gene during early stages of the Rhizobium–legume association and in response to purified Nod factors

To study tissue-specific expression of MtAnn1, we generated transgenic Medicago lines (both M. varia and M. truncatula) expressing a transcriptional fusion between a 2.2-kb-long MtAnn1 promoter fragment and the GUS reporter gene (see Experimental procedures). Histochemical localisation of pMtAnn1-GUS expression was initially performed on roots of transgenic Medicago plants harvested 24 and 48 h following S. meliloti inoculation, thus covering the period of infection and early nodule development. Figure 1(a) shows typical GUS activity within the nodulation-susceptible root zone 48 h post-inoculation (hpi). Sectioning of stained tissues revealed that pMtAnn1-GUS expression is present simultaneously both in outer and inner cortical tissues (Figure 1b). The highest expression levels were generally found in the outer cortex and the endodermis in the proximity of protoxylem poles. However, GUS activity was also present in neighbouring inner cortical and pericycle cells. Analysis of early infection events revealed a clear correlation between GUS activity and infection thread progression. As shown in Figure 1c, GUS staining is present in outer cortical cells that are located just below a growing infection thread which has initiated within a root hair. These activated cortical cells are presumably preparing for subsequent penetration by the infection thread. Subsequently, expression of the pMtAnn1-GUS reporter gene is present throughout cortical cell invasion and often in neighbouring cells in the outer cortex (Figure 1d,e). Since low levels of GUS activity were already detectable in the nodulation-susceptible root zone as early as 24 hpi (results not shown), MtAnn1 transcription is presumably activated prior to root hair infection (see below). Finally, it should be noted that weak GUS activity was occasionally associated with infected root hairs in epidermal cells 48 hpi (data not shown).

Figure 1.

Histochemical analysis of pMtAnn1-GUS expression in transgenic Medicago roots.

Localization of GUS activity in transgenic roots in response to S. meliloti(a-e), to purified Nod factors (f,g), and in non-symbiotic conditions (h-k).

(a) Sites of GUS activity in whole root segment close to the root tip, 48 h post-inoculation (hpi) with S. meliloti.

(b) 80 µm-thick transversal sections of agarose-embedded roots 48 hpi. GUS activity is mainly in the outer cortex (oc) and the endodermis (arrowhead).

(c,d,e) 80 µm-longitudinal sections of S. meliloti-inoculated roots (48 hpi). Tissues have been stained both for GUS and β-galactosidase. The latter localises S. meliloti expressing the constitutive lacZ fusion in infection threads (arrows); d and e correspond to different focal planes at the same site showing GUS activity associated with infected and neighbouring cortical cells.

(f) Root segment of transgenic M. varia 48 h after the addition of 10−8 M Nod factors.

(g) 80 µm-thick transverse section of agarose-embedded M. truncatula root 48 h after 10−9 M Nod factor addition. GUS staining is indicated in the outer cortex (oc) and endodermal tissues (arrowhead).

(h-k) Sequential stages of lateral root development in transgenic M. truncatula roots. (h) GUS activity is localised in dividing cells of the lateral root primordium; (i) At the base of an emerging lateral root; (j) The GUS staining pattern of the lateral appears ring-like when viewed from above; (k) GUS activity is present in the subapical region of the root tip referred to as the distal elongation zone (DEZ, Ishikawa and Evans, 1995). Bars = 50 µm.

When roots of transgenic pMtAnn1-GUS plants were treated with purified S. meliloti Nod factors at doses known to elicit symbiotic responses (10-9−10-8m), GUS activity was first detected within 12–24 h following Nod factor addition, with levels increasing significantly over the following 48 h. Whilst histochemical GUS staining also localised to the nodulation-susceptible zone the Nod factor-induced response generally extended over a wider region of the root (compare Figure 1a with f). Most importantly, as for the Smeliloti-induced response, GUS activity localised to both inner (mainly endodermal) and outer (outer cortical) root tissues (Figure 1g), thus confirming that infection is not required for MtAnn1 transcriptional activation during early symbiotic stages. Finally, it should be noted that identical results were obtained with transgenic pMtAnn1-GUS lines of either M. truncatula or M. varia. In conclusion, our findings indicate that MtAnn1 expression occurs simultaneously in both inner and outer root tissues during both pre-infection and infection and that Nod factors alone can elicit the pre-infection expression pattern.

The MtAnn1 gene is also expressed during sequential stages of lateral root development

Histochemical staining of transgenic Medicago plants also revealed that the pMtAnn1-GUS gene fusion is expressed in root tissues associated with lateral root development and cell differentiation in the root apex. As shown in Figure 1h, GUS activity localised specifically to sites of lateral root initiation. Later, during lateral root emergence (Figure 1i), reporter gene activity is highest at the base of the emerging root, with a ring-like pattern when viewed from above (Figure 1j). GUS activity is also present in the subapical region of the root tip, in the so-called DEZ zone (Distal Elongation Zone, Ishikawa and Evans, 1995), associated with recently differentiated cortical cells (Figure 1k). The presence of MtAnn1 transcription in roots independent of nodulation is consistent both with our previous Northern hybridisation studies (de Carvalho-Niebel et al., 1998), and also with the Western immunoblotting studies described later in this article. These results suggest that both symbiotic and non-symbiotic MtAnn1 expression are associated with developmental programmes specific to cell differentiation in cortical root cell layers.

Recombinant MtAnn1 possesses calcium-dependent phospholipid binding activity

In order to further our understanding of the role of the MtAnn1 gene in the context of root nodulation, it was essential to focus on the biological activity and intracellular localisation of the putative annexin encoded by this gene. One of the principle characteristics of annexins is their capacity to bind to acidic phospholipids in a calcium-dependent manner. To investigate the lipid-binding properties of MtAnn1, we first constructed a GST-MtAnn1 fusion protein using a full length MtAnn1 cDNA, and then purified recombinant MtAnn1 protein after expression in Escherichia coli (see Experimental procedures). Following incubation with phospholipid vesicles in the presence or absence of Ca2+, the proportion of recombinant MtAnn1 protein bound to the liposome pellet was determined by comparing the pellet and supernatant fractions after separation on Coomassie stained SDS-PAGE. Figure 2(a) shows that, in the presence of calcium, efficient MtAnn1 protein binding and precipitation only occurred with liposomes containing the acidic phospholipid phosphatidylserine (PS) and not with liposomes containing exclusively phosphatidylcholine (PC). Furthermore, recombinant MtAnn1 protein was not present in the pellet fractions when incubated with either calcium (Figure 2a) or PS-containing vesicles alone (data not shown), indicating that both are required for protein binding. As a further control, Figure 2b shows that, under identical conditions, purified GST protein was unable to bind to liposomes whether in the presence or absence of calcium. In conclusion, these experiments demonstrate that MtAnn1 possesses calcium-dependent phospholipid binding activity, and therefore behaves as a typical annexin.

Figure 2.

Calcium-dependent phospholipid binding and tissue-specific distribution of the MtAnn1 annexin.

(a,b) Recombinant MtAnn1 (a) and GST (b) proteins (5 µg) were incubated in buffer-containing 1 mm EGTA (EGTA), or 1 mm CaCl2 (Ca2+), or in the presence of either 1 mm CaCl2 and liposomes composed of PC (Ca2+ + PC), or 1 mm CaCl2 and liposomes composed of PC/PS (Ca2+ + PC/PS). Following centrifugation, equal proportions of supernatants (S) and liposome-containing pellets (P) were analysed by 10% SDS-PAGE after Coomassie staining. M corresponds to low molecular weight markers from AmershamPharmacia Biotech.

(c,d) Western blot analysis of purified recombinant MtAnn1 protein (0.1 µg) (MtAnn1) and total protein extracts from M. truncatula leaves (Le), nitrogen-starved control roots (R0), roots harvested 2 days (R2d) and 4 days (R4d) after S. meliloti inoculation, and 7-day-old isolated root nodules (N7d). 20 µg total protein from leaves and 10 µg total protein from root and nodule tissues were loaded per lane. Anti-maize annexin and anti-MtAnn1 antisera were used at the respective dilutions of 10 000 and 50 000.

Antiserum to MtAnn1 specifically recognises a nodulation-enhanced 33 kDa protein

Having established that recombinant MtAnn1 possesses bona fide annexin activity, antiserum was then raised against the purified protein. As shown by Western blot analysis in Figure 2c, the MtAnn1 polyclonal recognises a single protein band in M. truncatula root and nodule tissues, migrating with the same size as the purified recombinant protein. Despite a predicted molecular weight of 35 kDa, both the native and the recombinant MtAnn1 proteins migrate slightly faster (33 Kda), as previously reported for other plant annexins (Battey et al., 1996). In control experiments, bands were not detected when blots were hybridised with similar dilutions of MtAnn1 pre-immune serum (data not shown). As expected from previous gene expression studies (de Carvalho-Niebel et al., 1998), the 33 kDa protein, presumably corresponding to MtAnn1, is both absent in leaf tissue and present at significantly higher levels in nodules.

To further evaluate the specificity of the MtAnn1 antiserum, Western blot analysis was then performed using an antiserum raised against maize annexins (kindly provided by Nicholas H. Battey, University of Reading, UK). This antiserum is capable of recognising annexins from other plant species (Blackbourn et al., 1991, 1992). As shown in Figure 2d, the maize annexin antiserum recognises two protein bands migrating at approximately 35 and 33 kDa positions in M. truncatula tissues. Interestingly, the 35 kDa band is present in leaf, root and nodule tissues, whereas the 33 kDa band shows a similar expression profile to the single band recognised by the MtAnn1 antiserum. Taken together, these data imply that the MtAnn1 antiserum is specific for this nodulation-associated annexin with 33 kDa mobility. Furthermore, the clear correlation between variations in MtAnn1 transcript levels throughout nodulation (de Carvalho-Niebel et al., 1998) and MtAnn1 protein levels suggests that expression is primarily controlled at the transcriptional level.

The MtAnn1 antiserum immunolabels cortical tissues during early stages of nodulation

Having shown that the MtAnn1 antiserum specifically recognises the nodulation-associated annexin, we performed immunolocalisation studies on sections of S. meliloti-inoculated roots to assess both the tissue and intracellular localisation of MtAnn1. As expected from the transgenic plant/pMtAnn1-GUS reporter experiments described earlier, MtAnn1 antiserum labelling localised to the same nodulation-susceptible zone of the root. Figure 3(a) shows clearly that antiserum labelling is strong in both outer cortical cells and in endodermal tissues, again in line with the data for MtAnn1 transcriptional activity. Lower levels of immunolabelling are observed in internal cell layers of the root cortex and in certain root hairs cells in the nodule primordia-containing region. MtAnn1 antiserum also labels cells at the base of developing lateral roots (data not shown). Thus, a good correlation exists between the tissue-specific distribution of the MtAnn1 protein based on immunolocalisation and the transcriptional activity of the MtAnn1 gene determined via the GUS reporter strategy.

Figure 3.

Cellular localisation of the MtAnn1 annexin.

(a,b) Immunolocalization of the MtAnn1 annexin protein in S. meliloti-inoculated (48 hpi) root sections (16 µm). Sections were labelled with the antirabbit Alexa 488 secondary antibodies (400 × dilution) following incubation with primary anti-MtAnn1 antiserum (3000 × dilution). (a) Strong labelling is found in the outer cortex (white arrow) and endodermal (black arrow) root tissues. (b) A section through outer cortical cells showing labelling associated with the nuclear periphery (arrow). (Wide field epi-fluorescence microscopy).

(c-e) Transient expression of a MtAnn1-GFP fusion and the control GFP proteins in onion epidermal cells. (c,d) The MtAnn1-GFP fusion is cytosolic and localises primarily to the nuclear periphery. (e) Uniform fluorescence in the nucleus of an onion cell expressing the control 35S-GFP construct. (Wide field epi-fluorescence microscopy).

(f-j) Confocal microscopy of transgenic M. truncatula roots expressing pMtAnn1-MtAnn1-GFP and pMtAnn1-GFP constructs following S. meliloti inoculation (48 hpi). (f) S. meliloti-activated outer cortical cells expressing the pMtAnn1-MtAnn1-GFP construct are highly fluorescent, while adjacent cells are not labelled. (g-i) The MtAnn1-GFP fusion protein localises to the cytoplasm and the nuclear periphery, either as a fine network (g, arrows) or as intense fluorescence concentrated around the nucleus (h and i, arrows). (j) An activated cortical cell expressing the control pMtAnn1-GFP construct showing uniform fluorescence within the nucleus. Bars = 20 µm.

MtAnn1-GFP fusion protein preferentially localises to the nuclear periphery in S. meliloti-activated cortical cells

When examining the pattern of immunolabelling at high magnification in activated cortical cells of S. meliloti-inoculated roots (48 h), we occasionally observed preferential immunolocalisation around the centrally positioned nucleus (Figure 3b). To examine the subcellular localisation of MtAnn1 in more detail, we adopted a complementary approach based on the use of a recombinant GFP fusion reporter (see Experimental procedures). However, prior to studying in vivo localisation in Mtruncatula roots we initially placed the MtAnn1-GFP fusion under the transcriptional control of the 35S promoter and introduced the transgene into onion epidermal cells using particle bombardment. Onion cells expressing the MtAnn1-GFP fusion protein showed fluorescence in the cytoplasm and most strikingly associated with the nuclear periphery (Figure 1c). The distribution of the MtAnn1-GFP fusion is non-uniform, and can be seen associated to specific regions both in the cytoplasm and around the nuclear periphery (Figure 1d). In contrast, onion cells expressing a control p35S-GFP construct revealed a fluorescent signal both in the cytoplasm and within the nucleus (Figure 3e). This is consistent with passive transport of the GFP protein into the nucleus (reviewed in Hanson and Köhler, 2001). Confocal laser scanning microscopy analysis confirmed that the MtAnn1-GFP fusion protein is excluded from the nucleus (data not shown).

To examine the intracellular distribution of the MtAnn1-GFP fusion in Medicago root cells during the early stages of nodulation, we generated transgenic M. truncatula composite plants using a recently developed Agrobacterium rhizogenes-mediated transformation procedure (Boisson-Dernier et al., 2001). In these experiments the GFP fusion was placed under the control of the MtAnn1 promoter itself to ensure correct tissue-specific expression (see Experimental procedures). Transgenic M. truncatula roots transformed with the pMtAnn1-MtAnn1-GFP construct (and a pMtAnn1-GFP control construct) were spot-inoculated with S. meliloti and then observed using confocal laser scanning microscopy 2–3 days post-inoculation. For both constructs, fluorescence labelling was specifically observed in S. meliloti-activated root outer cortical cells which have large numbers of cytoplasmic strands, but not in adjacent non-activated cells (Figure 1f), in line with the promoter specificity shown earlier with the GUS reporter. In outer cortical cells expressing the pMtAnn1-MtAnn1-GFP construct, as previously found for the onion epidermal cells expressing 35S-MtAnn1-GFP, fluorescence localised to the cytoplasm with a predominant signal in the peri-nuclear region. In these cells we observed two distinct fluorescent patterns. Either an extended fine network surrounding the nuclei, which are often positioned centrally within the cell (Figure 3g), or an intense fluorescent signal concentrated around the nucleus positioned at the cell periphery (Figure 3h,i). These specific patterns were never observed in control roots expressing pMtAnn1-GFP, where fluorescence was both cytosolic and nuclear, comparable to that found for 35S-GFP expression in onion cells (compare Figure 3e,g). Unfortunately, studies with the confocal microscope were limited to the outer root cell layers, and could not be extended to inner tissues such as the endodermis. Taken together, our results demonstrate that the MtAnn1 annexin accumulates specifically at the nuclear periphery in activated cortical cells elicited in response to S. meliloti inoculation.


The M. truncatula MtAnn1 gene, encoding a calcium-binding protein of the annexin family, is transcriptionally activated during root nodulation and in response to rhizobial Nod factors (de Carvalho-Niebel et al., 1998). In this paper we show that the encoded MtAnn1 protein possesses the calcium-dependent binding to acidic phospholipids typical of the annexin family. We also report the detailed spatio-temporal expression pattern of MtAnn1 and the cellular localisation of the encoded protein during the early stages of the Sinorhizobium-Medicago association. Transgenic Medicago lines expressing a pMtAnn1-GUS gene fusion have revealed a novel expression profile for this annexin gene with simultaneous transcriptional activation in both outer (cortical) and inner (mainly endodermal) cell layers of the root prior to and during bacterial infection. Significantly, Nod factors alone are sufficient to elicit this expression profile. Whilst gene expression in the outer cortex associated with pre-infection and infection events has already been described for several early nodulin genes, the presence of the bacterium is absolutely required for the activation of gene expression (Journet et al., 2001; Mathis et al., 1998; Scheres et al., 1990). In the case of the endodermis, Bauchrowitz et al. (1996) reported transcriptional activation for the Mtlec3 lectin gene, but here again only following rhizobial inoculation. Thus, MtAnn1 is the first pre-infection marker for Nod factor-induced gene expression in outer cortical and endodermal cell layers.

The MtAnn1 gene expression studies reported in this article argue strongly that the previously ignored root endodermis plays an active role in Nod factor signalling leading to nodule organogenesis. The endodermis is the root cell layer which forms the interface between the cortex and the pericycle layer of the central cylinder (reviewed by Schreiber et al., 1999). The chemical composition of the endodermal cell wall varies during root differentiation, and directly determines the properties of this important cell layer. In the immature root zone which responds to rhizobia and their Nod factor signals, the endodermis is presumably at an early stage of development with cell walls essentially composed of casparian strips, thus providing an effective selective barrier to the transport of solutes between the cortex and the central cylinder. Interestingly, casparian bands are also found in the outermost cortical cell layer (hypodermis) of most angiosperm plants (for review see Hose et al., 2001), suggesting that outer cortical and endodermal cell layers might share common functional properties.

The simultaneous activation of MtAnn1 in the outer cortex and in inner root tissues suggests that MtAnn1 function may be related to similar symbiotic responses in these different cell layers. Timmers et al. (1999) reported that microtubular (MT) changes occur in both outer and inner root tissues (inner cortex, endodermis and pericycle) during early symbiotic steps of the Sinorhizobium-Medicago interaction. MT rearrangements in the outer cortex are related to infection thread growth and in particular to the formation of the cytoplasmic bridges known as pre-infection threads (PIT), which are formed in outer cortical cells in the activated zone around the infected root hair (Timmers et al., 1999). Since purified Nod factors alone are not sufficient to elicit PIT formation in Medicago (Timmers et al., 1999), MtAnn1 function presumably precedes PIT formation. In these activated outer cortical cells, differentiation occurs via reactivation of the cell cycle machinery prior to MT rearrangements and PIT formation (Timmers et al., 1999; Yang et al., 1994). In mammalian cells, a correlation has been established between annexin gene expression and cell differentiation, and furthermore both animal and plant annexin gene expression can be regulated during the cell cycle, suggesting a potential role for these proteins during cell division (reviewed by Hawkins et al., 2000; Proust et al., 1999). In addition, the submembranous localisation of vertebrate annexins at cytoskeleton-anchoring sites, as well as their capacity to interact with actin or cytoskeleton-binding proteins, indicate a possible role for certain annexins in cytoskeletal rearrangements (Gerke and Moss, 1997; Hawkins et al., 2000). Plant annexins interacting with filamentous actin have also been reported (Calvert et al., 1996; Hu et al., 2000). Thus MtAnn1 may play a role in cell cycle reactivation and/or cytoskeleton rearrangements during the early stages of nodulation. Such a role would indeed be compatible with the ‘non-symbiotic’ expression profile of MtAnn1 in roots, since transcriptional activity is specifically associated with dividing or recently differentiated cells both in the root apex and in lateral root primordia (Figure 1). An analysis of MtAnn1 expression in the M. truncatula mutant HCL (Catoira et al., 2001), in which cortical cell activation appears normal but PIT formation is absent and correlated with abnormal MT rearrangements, should help to define the relationship between annexin gene expression, cytoskeleton rearrangements and cell cycle reactivation during early nodulation.

Immunolabelling experiments and studies of the in vivo distribution of MtAnn1-GFP fusions expressed in transformed M. truncatula roots reveal a cytoplasmic and peri-nuclear localisation of the MtAnn1 annexin in S. meliloti-activated cortical cells. Studies of the subcellular distribution of plant annexins using immunolabelling approaches have revealed a variety of intracellular localisations including submembranous cell junctions, vacuole membranes and nuclei (Clark et al., 1998; Proust et al., 1999; Seals and Randall, 1997). The alfalfa annexin AnnMs2 (Kovács et al., 1998), which shares only 35% sequence identity with MtAnn1, is reported to preferentially localise to the nucleolus, although antiserum also labels the cytoplasm and the nuclei periphery. Compared to AnnMs2, MtAnn1 annexin is excluded from nuclei, and is principally peri-nuclear. Although the precise peri-nuclear localisation of MtAnn1 is as yet unknown, interactions with nuclear-associated components such as the endoplasmic reticulum (ER) or the cytoskeleton are likely to be involved. MtAnn1 could be specifically associated either with the nuclear envelope ER domain, or possibly with ER-associated cytoskeletal elements (reviewed by Staehelin, 1997). We also observed that the in vivo peri-nuclear distribution of the MtAnn1-GFP fusion protein can vary between neighbouring activated cortical cells, visualised either in the form of a network or as intense fluorescence around the nuclei. This suggests that the intracellular distribution of the MtAnn1 annexin is actively regulated, as indeed has been reported for other annexins, including those from plants (Clark et al., 2000; Thonat et al., 1997). It is also possible that the particular peri-nuclear localisation of the MtAnn1-GFP protein reflects changes in nuclear envelope organisation occurring during cell cycle progression. In dividing cells, the nuclear envelope breaks down reversibly in prophase and reassembles during mitosis, leading to dispersion of all major nuclear envelope components (reviewed in Collas and Courvalin, 2000). Since S. meliloti-activated outer cortical cells in Medicago roots re-enter the cell cycle and in certain cases divide (Timmers et al., 1999), peri-nuclear MtAnn1 distribution could reflect whether the cortical cells are dividing or non-dividing.

In conclusion, the results presented in this paper argue that the MtAnn1 annexin is involved in the early stages of Nod factor signalling in root tissues, and probably plays a role in the activation of cortical cells prior to pre-infection thread formation. In vivo co-localisation of the MtAnn1-GFP reporter protein together with ER, cytoskeleton and cell cycle markers in Medicago roots will now be required to determine whether MtAnn1 function is indeed related to cell division.

Experimental procedures

Generation of transgenic Medicago plants expressing a MtAnn1 promoter-GUS gene fusion and histochemical GUS staining

A MtAnn1 genomic clone was used as a template for PCR amplification (using Vent DNA polymerase) of a 2.2-kb DNA fragment lying upstream of the MtAnn1 ATG start codon. We used the M13 5′ primer and the 3′ primer Mtoli1 (5′-GGGTAGCCATGGTTCAATATGTATTGTTTATAGTGC-3′) to create a NcoI site overlapping the start codon. The amplified PCR product was subcloned into the pBSGUS vector (Vernoud et al., 1999) to generate a 2.2-kb MtAnn1 promoter-GUS transcriptional fusion, which was subsequently cloned into the pLP100 binary vector (Szabados et al., 1995). The binary vector was introduced into A. tumefaciens according to the procedure described by Holsters et al. (1978). A. tumefaciens-mediated transformation and regeneration of M. sativa ssp. varia A2 transgenic plants was performed essentially as described in Bauchrowitz et al. (1996). Transgenic M. truncatula cv. Jemalong plants were generated according to Chabaud et al. ( Plant growth conditions for in vitro propagation of transgenic lines and for production of transgenic seeds is described in Vernoud et al. (1999).

Histochemical GUS staining was performed essentially as previously described (Vernoud et al., 1999). Stained samples were briefly cleared with sodium hypochlorite (2 min), rinsed with water and observed with a stereomicroscope (Leica Microsystems, Wetzler, Germany) and/or an Axiophot light microscope (Carl Zeiss, Oberkochen, Germany). Double staining for both GUS and β-galactosidase activities after inoculation with a RCR2011 S. meliloti strain carrying a constitutive hemA-lacZ fusion (Ardourel et al., 1994) was performed according to Bauchrowitz et al. (1996). For microtome sectioning roots were prepared as described in Bauchrowitz et al. (1996).

Sinorhizobium meliloti inoculation and Nod factor treatment

Wild type or transgenic Medicago plants were grown under aeroponic conditions or in plastic growth pouches (Mega International, Minneapolis, USA) prior to inoculation with the S. meliloti strain RCR2011 (Ardourel et al., 1994), as described in Vernoud et al. (1999). Nod factor treatment (10-8-10-10m) was carried out with pouch-grown plants as described in Vernoud et al. (1999).

Recombinant MtAnn1 protein expression and antiserum production

Recombinant MtAnn1 was purified using the glutathione-S-transferase (GST) gene fusion system (AmershamPharmacia Biotech, Orsay, France). The MtAnn1 coding sequence was PCR-amplified as described above using the 5′ PCR primer ANN-Bam (5′-ACATATGGATCCATGGCTACCCTTTCTG-3′) and the 3′ PCR primer ANN-Sal (5′-GAGAAGGGGTCGACTCATTCCTCTTTCC-3′), thereby creating the BamHI and SalI sites necessary for cloning into the PGEX 6P-3 vector (AmershamPharmacia) and fusion to the GST sequence. The GST-MtAnn1 fusion protein was expressed in the E. coli strain BL21 following the AmershamPharmacia manual, except that bacterial growth and the induction of protein expression with 1 mm isopropyl-β-thiogalactopyranoside (IPTG) was carried out at 20°C in order to obtain a soluble GST-MtAnn1 protein. Glutathione-sepharose purification and prescission protease cleavage of the fusion protein was performed following the manufacturer's protocol.

Antiserum was raised in rabbit using the purified recombinant MtAnn1 protein by Eurogentec Bel S.A. (Seraing, Belgium). After collecting the pre-immune serum, 4 immunisations were carried out with 100 µg MtAnn1 protein over an 8-week period. Antiserum titre was determined by testing the immunoreactivity of serially diluted antiserum.

Phospholipid binding assay

Liposomes containing either l-α-phosphatidylcholine (PC) (Sigma-Aldrich Chemie, Germany) alone or l-α-phosphatidylcholine and l-α-phosphatidyl-l-serine (PS) (Sigma-Aldrich) in a molar ratio of 1 : 2, were prepared as follows: stock solutions of lipids in chloroform/methanol were mixed, dried under nitrogen gas and resuspended (400 µg ml−1) in EGTA buffer (Boustead et al., 1989). The lipid suspension was then sonicated to obtain a clear suspension and stored at 4°C for 24 h before use. Phospholipid binding assays were carried out by co-sedimentation as described by Boustead et al. (1989). 5 µg of GST or recombinant MtAnn1 were incubated in the presence of 150 µg of lipid vesicles in EGTA or calcium-containing buffers. After centrifugation and precipitation, equal amounts of pellet and supernatant fractions were analysed by Coomassie-stained SDS-PAGE.

SDS-PAGE and Western analysis

S. meliloti-inoculated roots or isolated nodules of aeroponically grown M. truncatula plants were harvested, frozen in liquid nitrogen and stored at − 70°C before crude protein extraction, as described in Blackbourn et al. (1991). 10 µg (for root and nodule tissues) and 20 µg (for leaf tissues) of total proteins were separated by SDS-PAGE 10% gels, with Mr protein standards (Low Molecular Weight, AmershamPharmacia Biotech, Orsay, France; Prestained Broad Range Marker, New England Biolabs, Beverly, MA, USA). Gels were either Coomassie stained (Phast Gel Blue R, AmershamPharmacia) or electrophoretically transferred to nitrocellulose membranes (Hybond C Extra, AmershamPharmacia), which were subsequently blocked as described in Blackbourn et al. (1991). Blots were probed with pre-immune and MtAnn1 antisera at a 1 : 50.000 dilution, and with pre-immune and maize annexin antisera (a kind gift of N. Battey) at a 1 : 1000 dilution. Peroxidase-conjugated antirabbit IgG (AmershamPharmacia) was used as second antibody, and proteins were visualised by chemiluminescence (ECL, AmershamPharmacia), according to the manufacturer's instructions.

Indirect immunofluorescence

Root segments (approximately 0.5 cm) were fixed in 4% paraformaldehyde in 0.1 m sodium phosphate buffer (pH 7.2) for 90 min at room temperature. After rinsing and dehydration, the specimens were embedded in low melting point wax as described in Vitha et al. (1997). 16 µm-thick sections were placed on poly l-lysine coated slides and de-waxed prior to immunolocalisation. Sections were incubated with rabbit pre-immune or anti-MtAnn1 antisera at a 1 : 3000 dilution in 0.2% gelatin, 1× PBS buffer, pH 7.4 (overnight at 4°C). After rinsing in PBS, sections were incubated with the antirabbit Alexa 488 antibody conjugate (Molecular Probes, Leiden, The Netherlands) for 2 h at room temperature at a 1 : 400 dilution, rinsed again and treated with 0.5% Evans blue (Sigma) in PBS buffer for 5 min to quench autofluorescence. Samples were mounted as described in Timmers et al. (1999) and observed by fluorescence microscopy (Zeiss, Axiophot 2).

Localisation of the MtAnn1-GFP fusion protein in onion and M. truncatula cells

The complete MtAnn1 coding region was amplified by PCR using the 5′ primer ANN-Bam (see above) and the 3′ primer (5′-GAGAAGGGAAAGGATCCTTCCTCTTTCC-3′), thereby creating BamHI sites at both 5′ and 3′ ends and replacing the termination codon by a glycine residue. This PCR fragment was cloned upstream of the soluble smRS-GFP sequence (Davis and Vierstra, 1998), generating a MtAnn1-GFP fusion, which was subsequently cloned downstream of both the 35S and MtAnn1 promoters, to generate, respectively, p35S-MtAnn1-GFP (in the pBin 121 binary vector) and pMtAnn1-MtAnn1-GFP fusions (in the pLP100 vector). The smRS-GFP sequence was also cloned downstream of the 35S and the MtAnn1 promoters to create the control fusions p35S-GFP and pMtAnn1-GFP. Plasmid DNA from the p35S-MtAnn1-GFP and the p35S-GFP constructs were transiently expressed in onion epidermal cells using the Bio-Rad PDS-1000/He Biolistic Particle Delivery system. Microcarrier preparation (1.6 µm gold particles) and bombardment were carried out at a pressure of 1100 lb in−2. The bombarded onions were placed on water-humid filters in Petri dishes sealed with Parafilm and incubated for 24 h at 25°C in the dark. Pealed epidermal cells were observed with a CCD camera Micromax-1300Y-HS (Princeton Instruments, Evry, France) and by confocal laser scanning microscopy (Zeiss, LSM 410 Invert).

The pMtAnn1-MtAnn1-GFP and the pMtAnn1-GFP constructs were mobilised into the A. rhizogenes strain ARqua1 and used to generate transgenic M. truncatula roots essentially as described in Boisson-Dernier et al. (2001). Composite M. truncatula plants grown on pouch-paper (Mega International) on top of agar plates were spot inoculated with an S. meliloti suspension (5 × 105 bacteria per ml), and observed with a CCD camera and by fluorescence microscopy as described above.


We are grateful to Fabienne Maillet for generously providing purified S. meliloti Nod factors, Laurence Dupou-Cezanne for providing the protocol for liposome vesicle preparation and Nicholas Battey for sending us the anti-maize annexin antiserum. We would also like to acknowledge Marie-Christine Auriac and Alain Jauneau for their helpful advice concerning certain microscopy techniques and Aurélien Boisson-Dernier for critical reading of the manuscript.