Graft-union development: a delicate process that involves cell-cell communication between scion and stock for local auxin accumulation.

Grafting is an ancient cloning method that has been used widely for thousands of years in agricultural practices. Graft-union development is also an intricate process that involves substantial changes such as organ regeneration and genetic material exchange. However, the molecular mechanisms for graft-union development are still largely unknown. Here, a micrografting method that has been used widely in Arabidopsis was improved to adapt it a smooth procedure to facilitate sample analysis and to allow it to easily be applied to various dicotyledonous plants. The developmental stage of the graft union was characterized based on this method. Histological analysis suggested that the transport activities of vasculature were recovered at 3 days after grafting (dag) and that auxin modulated the vascular reconnection at 2 dag. Microarray data revealed a signal-exchange process between cells of the scion and stock at 1 dag, which re-established the communication network in the graft union. This process was concomitant with the clearing of cell debris, and both processes were initiated by a wound-induced programme. The results demonstrate the feasibility and potential power of investigating various plant developmental processes by this method, and represent a primary and significant step in interpretation of the molecular mechanisms underlying graft-union development.

Introduction

Grafting is an asexual plant propagation technique in agriculture that has been widely used for thousands of years. Woody plants such as apple trees are grafted for the purpose of dwarfing, ease of propagation, and sturdiness, while herbaceous grafting can increase productivity and control damage caused by soil-borne disease and abiotic stress, such as in tomato, cucumber, and melon (Estañ ; Sigüenza ; Zhou ). It has been used widely for decades as a convenient technique in long-distance signalling research and has generated persuasive evidence in various plant processes. For example, the most important discovery using grafting was that FLOWERING LOCUS T (FT) protein was confirmed as the florigen (Corbesier ). The epigenetic state can also be converted, as mobile small RNAs from the scion could direct epigenetic modifications in the genome of stock in Arabidopsis (Molnar ). The findings that morphological and physiological changes could occur with a specific stock–scion combination have also been reported, as grafting altered leaf morphology (Kim ) and improved salinity tolerance (Estañ ) in tomato. In cherry trees, dwarfing rootstock can trigger an early cessation of terminal meristem growth of the scion, conferring up to a 50% reduction in scion size compared with standard rootstocks (Prassinos ). Aside from these applications, graft-union development is an intricate process during which histological and physiological changes, such as organ regeneration, are initiated and progress dramatically, and even the genetic material between different cells of the scion and stock can be exchanged (Stegemann and Bock, 2009). Thus, grafting is not only a useful technique but also an attractive research subject. Although a number of reports have depicted the process with regard to the histological and physiological aspects (Kollmann and Glockmann 1985; Fernandez-Garcia ; Flaishman ), the molecular mechanisms related to the process remain elusive.

In the model organism Arabidopsis thaliana, grafting techniques were developed early in 1993 (Tsukaya ) but were restricted to the grafting of inflorescence stems late in development when most key processes had already been determined. In 2002, Colin Turnbull and colleagues developed a method using 3–4-d-old seedlings, allowing experiments to be conducted on many aspects of long-distance signalling for the first time (Turnbull ; Bainbridge ). This seedling grafting technique has been applied to the study of many plant processes including flowering (Corbesier ), shoot branching (Turnbull ), disease resistance (Xia ; Thatcher ), nutrient allocation (Rus ), post-transcriptional gene silencing (Brosnan ; Molnar ), cytokinin activity (Foo ; Matsumoto-Kitano ), and root-to-shoot hydraulic signalling (Christmann ).

Here, a simple and effective seedling micrografting method that improved on the previous protocol (Turnbull ; Bainbridge ) was developed. This approach reduced the operational work and facilitated sample analysis, yet yielded higher success rates than the previous method. It could also easily be applied to the seedling micrografting methods of other dicotyledonous plants such as tomato, alfalfa, and tobacco. This method was further used to model graft-union development in Arabidopsis. Anatomic and transcriptomic analysis uncovered elaborate stages for graft-union healing processes and highlighted a cell–cell communication process during its early stage.

Materials and methods

Plant materials and grafting procedure

A. thaliana ecotype Col-0 was used as the wild-type (WT) plant. The transgenic lines proDR5:GUS and procycB:GUS expressing GUS under the control of the DR5 (a synthetic auxin response element) and cyclin B (cycB) promoters, respectively, were purchased from NASC (European Arabidopsis Stock Centre, http://arabidopsis.info/). The transgenic line proSUC2:GFP expressing green fluorescent protein (GFP) under the control of the sucrose-proton symporter 2 (SUC2) promoter was a kind gift from Dr Norbert Sauer.

Grafting in Arabidopsis was conducted as follows. Plates were tilted using a glass rod of 0.7–1 cm in diameter. Seeds were sown on 1% agar medium containing the macro-nutrients 5 mM KNO3, 2 mM Ca(NO3)2.4H2O, 2 mM MgSO4.7H2O, 2.4 mM KH2PO4, 0.1 mM K2HPO4, Fe.EDTA, 50 μM FeSO4.7H2O, 50 μM EDTA-Na2.2H2O, and the micro-nutrients 0.5 μM KI, 10 μM H3BO3, 10 μM MnSO4.4H2O, 3 μM ZnSO4.7H2O, 0.1 μM Na2MoO4.2H2O, 0.01 μM CuSO4.5H2O and 0.01 μM CoCl2.6H2O, at pH 5.7–5.8. The plates were incubated in the dark at 4 °C for 2–3 d and then placed vertically in a growth room (temperature 22–25 °C; light intensity 6000 lux; photoperiod 16 h light/8 h dark) with the thin side pointing downward and the thick side upward. Grafting was performed using a dissecting microscope at 4 d post-germination. Seedlings were chosen with long straight hypocotyls and the hypocotyl was cut transversely while on the agar. The stock donor was cut near the shoot apical meristem and the scion donor was cut halfway from the base of the hypocotyl (the hypocotyl consequently appeared longer than usual). Forceps were used to keep the seedling stable while cutting the hypocotyl with the razor blade under a dissecting microscope. It was important to make the cut quickly and to push the blade rather than pull it, so that the cut surface was clean and smooth. The scion was lifted up to bring it to the stock, and the stock was carefully picked up to approach the scion and connect them together. Note that it is important to be aware of raising the graft union up away from the agar surface (the oblique surface will provide enough space to make it possible). The scion or stock was carefully and slightly pushed to adjust the relative position of the two parts, and the graft was inspected from all sides of the graft junction to make sure that the two parts connected and supported each other thoroughly. Initially, this process may take some practice to connect the scion and stock together successfully while lifting the union up, but in our experience the operation went smoothly after about 1–2 weeks of practice. The plate was returned to the growth room to the same vertical position with the thin side downward and the thick side upward.

Grafting details for other plant species are illustrated in Supplementary Methods at JXB online.

Pseudo-Schiff propidium iodide (PS-PI) staining

PS-PI staining followed the method of Truernit with some modifications. Samples were fixed in 50% ethanol and 10% acetic acid at 4 °C overnight and then transferred to 80% ethanol and treated in a water bath at 80 °C for 1 min. The tissues were then transferred back to fixative and incubated for 1 h at room temperature. Next, the tissues were rehydrated in an ethanol series (50, 30 and 10%) for 20 min each, and rinsed with distilled water three times. Samples were then incubated in 1% (w/v) periodic acid at room temperature for 40 min and rinsed with water again. The tissues were incubated in PI-Schiff solution (100 mM sodium metabisulphite, 150 mM HCl, 30 μg/ml propidium iodide) for 1 h. The samples were rinsed with water and incubated in chloral hydrate solution (chloral hydrate:glycerol:water, 8:1:2) for 2–3 d. Sample observation was performed with a Zeiss confocal laser scanning microscope to record the images.

β-Glucuronidase (GUS) staining

Collected tissues were fixed in cold 90% acetone at 4 °C for 10 min and rinsed five to six times with staining buffer (50 mM sodium phosphate buffer, 0.1% Triton X-100, 2 mM potassium ferrocyanide, 2 mM potassium ferricyanide, 10 mM EDTA-Na2). The staining buffer was removed from the samples and staining solution containing 2 mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronide (X-Gluc) was added. The samples were incubated at 37 °C. The exact time course for incubation depended on GUS activity in the chosen line (for example, 1 h for DR5/WT and 4 h for WT/DR5). Samples were then dehydrated in a progressive series of ethanol dilutions (10, 30, 50 and 70%) and incubated in 70% ethanol overnight. Next day, the ethanol was removed, chloral hydrate solution was added and the samples were incubated overnight at room temperature. Samples were then placed onto a slide, a coverslip was added and images were obtained as described above.

Acid fuchsin staining

An aqueous solution of 0.1% (w/v) acid fuchsin (Sigma) was introduced into the vascular system of the grafts (SUC2/WT) by submerging the roots in the solution at room temperature (Flaishman ), taking care that the hypocotyl of the stock was above the solution. The cotyledon of the scion should become red in the vascular veins over 1–2 h, indicating that the xylems of the scion and stock have been connected.

Microscopy

Fluorescence was viewed with a confocal laser scanning microscopy (LSM 510, Zeiss). PI and GFP fluorescence were excited with a 543 and 488 nm argon laser, and emission was detected with 565–615 and 500–550 nm band-pass filter combinations, respectively. GUS-stained images were recorded using a Nikon ECLIPSE 80i microscope.

Plasmid construction and plant transformation

A region of approximate 2000 bp upstream of the ATG of heat-shock HSP21 (At4G27670) gene was amplified with the high-fidelity polymerase PrimeSTAR™ (TaKaRa) using the gene-specific primers 5′-CTGCAGCTGACTCTTTGGCAATAG-3′ (forward) and 5′-GGATCCTTGTTTCGAGTATGAGCC-3′ (reverse) and cloned into the vector pCM1300-GUS. Arabidopsis transformation was carried out using the floral dip method. Transgenic plants were obtained by screening successive generations in terms of hygromycin resistance.

Microarray analysis and quantitative RT-PCR

WT/WT grafts at 22–26 h after grafting were collected for microarray, with intact WT seedlings and ungrafted scions and stocks used as controls. For sampling, all three types of sample—grafts (group A: whole grafts), ungrafted scions and stocks (group B: mixture of ungrafted scions and stocks), and intact seedlings (group C: whole intact seedlings)—were collected in liquid N2 at the same time of 22–26 h after grafting. For group B, scions and stocks were cut at the same time during grafting. For group C, intact seedlings were also reserved during grafting.

Total RNA from frozen samples of groups A, B, and C were extracted using Trizol reagent (Invitrogen) and purified with an RNeasy® Mini kit (Qiagen) according to the manufacturer’s instructions. Total RNA concentration was assessed by spectrophotometry (absorbance at 260 nm), and the purity and integrity of the RNA were determined by the absorbance ratio at 260/280 nm and visualization after electrophoresis. An Arabidopsis thaliana 4 × 44K oligomicroarray (Agilent Technologies) representing 43 803 (version 4) probes was used, and the RNA labelling and microarray hybridization were carried out according to the manufacturer’s technical manual (Biochip Corp.). Groups A, B, and C also had three independent biological replicates each, and group C was applied as a technical replicate.

Hybridized microarrays were scanned with a DNA microarray scanner (G2565BA, Agilent Technologies) and features were extracted using the Feature Extraction software 10.7 (Agilent Technologies) using default protocols and settings (scan resolution=5 μm, PMT 100%, 10%). Data pre-processing and differential expression analysis of the gene expression data were done in R (version 2.10.0; Bioconductor, http://bioconductor.org/). Data were normalized between arrays using the quantile method. Normalized expression data was subjected to log2 transformation. For differential expression analysis, Student’s t-test assumed that the variance of the two classes not being the same was applied. First, data were analysed between groups A and C with a P value ≤0.01, q-value ≤0.05, and fold change ≥2. Differential probes were subsequently compared between A and B with P value ≤0.01 and q-value ≤0.05 (see Supplementary Fig. S4). Fold change was not applied in the comparison of groups A and B due to the high similarity of expression profiles between the two groups (see Supplementary Fig. S5). Probe annotation was according to Agilent’s web instructions (https://earray.chem.agilent.com/). Gene-enrichment tests for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) categories were performed in R program with the fisher.test function.

Quantitative RT-PCR was carried out as described by Mu .

Results

Improvement of the micrografting method in Arabidopsis

For a single successful graft, a good union between the scion and stock is the most important key point. In the previous protocol (Turnbull ; Bainbridge ), 0.3 mm diameter silicon tubing was used as a collar to support the graft and hold the stock and scion together during graft healing. The collar greatly increased the success rate but also interfered with the observation of the graft union. There are some reported methods that do not use such collars, but they do not have satisfactory success rates (Turnbull ; Bainbridge ).

Following the ‘good union’ principle, the protocol was improved in terms of enhancing the connection between the scion and stock. The cut surface of the scion and the stock cannot be placed closely together due to their different morphology—expanded cotyledons will raise the hypocotyl of the scion and result in a poor connection between the scion and stock. Moreover, the water on the medium surface will infiltrate the graft union and consequently interfere with the union healing. An oblique medium surface was found to overcome these problems through trial and error. In the first step, a glass rod was used to prop up one side of the plate before pouring the liquid medium into it (Fig. 1A). The agar solidified as an inclined plane at approximately 15° relative to the plate and formed an oblique surface that was thick at one side and thin at the other. Seeds were sown on this oblique surface, and after stratification at 4°C for 2–3 d, the plate was placed vertically in the growth room with the thin side downward and the thick side upward. On the oblique surface (Fig. 1B, reversed triangle), enough space (Fig. 1B, dotted line) was provided for the cut surfaces to connect the scion and stock thoroughly and completely (Fig. 1B, arrowhead). Therefore, there were strong mutual-effect forces on the joint when the plate was placed vertically in the growth room (Fig. 1B, white arrows), supporting a tight connection between the scion and stock. As it was easy to cut adventitious roots at their first emergence and to get rid of grafts that fail to form a good union immediately after grafting, because the graft union could easily be inspected from all sides (Fig. 2A–L), the success rate could be greatly improved. An 80% success rate was achieved in Arabidopsis, which was higher than the previous method used (no more than 70%; we performed grafting experiments according to Turnbull , and Bainbridge , and got a success rate of approximately 60%).

Fig. 1.

Illustration of the grafting procedure. (A) Illustration of how the oblique solid medium is established. Under axenic conditions, the plate was propped up by a glass rod at an angle of about a 15° and liquid medium was poured into the plate. (B) Side view of a WT/WT (scion/stock) graft. The image shows a side view of the graft that was observed parallel to the medium surface. The arrowhead indicates the graft union, suggesting that the two graft parts contact each other across the whole of the cut surface with no gaps. Images were taken immediately after grafting. The reversed triangle suggests that, when the plate is positioned vertically in the growth room, the scion and stock will support each other, and consequently mutual-effect forces will form under the force of gravity in the joint (white arrows). The dotted line shows the interspace under the graft union. Bar, 200 μm .

Fig. 2.

Grafts in Arabidopsis and tomato. (A–L) Time-lapse images demonstrating the phenotypic development of a graft union of a WT/WT combination in Arabidopsis. Arrowheads indicate the graft union. Images were ordered as dag (days after grafting). The graft union was examined on all sides: (A), (C), (E), (G), (I), and (K) show the front view, which was observed perpendicular to the medium surface, while (B), (D), (F), (H), (J), and (L) show the side view, which was observed parallel to the medium surface. (M–T) Graft union development in tomato. White arrowheads indicate the graft union. Tomato grafts recover faster than other species studied. At 3 dag, the scion and stock were connected completely. Bars, 200 μm.

The improved seedling micrografting method can be applied directly to other dicotyledonous plants

The method was applied to other dicotyledonous plant species. The first problem was whether the medium used for Arabidopsis would be suitable for the growth of other dicots. All plants studied, including tomato (Lycopersicon esculentum cv. Jiufeng NO.2), tobacco (Nicotiana tabacum), and alfalfa (Medicago sativa), exhibited favourable development and could be used for grafting in the germinating experiment, which included the same nutrient components of the medium and growth conditions as used in Arabidopsis. The operational procedure varied slightly according to the species of plant, as different plants have different shapes (see Supplementary Methods at JXB online). Eventually, the final success rate was about 80% in tobacco (Supplementary Fig. S1 at JXB online). An approximately 70% success rate was achieved in tomato due to the fast healing of its graft union (Fig. 2M–T). The success rate of alfalfa was 30–50% because of its irregular shape and slow recovery rate (Supplementary Fig. S2 at JXB online).

Functional recovery of vasculature at 3 d after grafting (dag) in Arabidopsis

The series of histological changes that took place during graft-union healing was examined by modified PS-PI staining (Truernit ) and procycB:GUS expression in Arabidopsis. For the purpose of homogenization, all grafts were examined under a dissecting microscope at 1 dag before sampling. The grafts that were connected tightly and with no obvious gaps were fixed and used for subsequent treatments. At 1 dag, there were no apparent changes were recorded (Fig. 3A, D, J). In the cycB/cycB (scion/stock) union, there was no GUS staining observed in the graft union at 1 dag, suggesting the absence of cell division at this time point (Fig. 3G). The first evidence of cell division became apparent at 2 dag in both WT/WT and cycB/cycB grafts (Fig. 3B, E, H, K). By 3 dag, substantial cell division occurred on both the scion and stock at the graft interface (Fig. 3I, L). A united vessel was also detected in the graft union (Fig. 3C, F, I, arrow), implicating functional recovery of the vasculature.

Fig. 3.

Anatomical analysis of graft union development in Arabidopsis. (A–C) PI stained WT/WT grafts. Panels are projections from confocal z-stacks. The approximate time points of sampling of the grafts are indicated. The box in (A) indicates the graft union. A complete union resulted in a poor staining at 3 dag. (D–E) Differential interference contrast (DIC) images of the same graft shown in (A–C). The dotted line in the DIC images shows the graft union. Arrows indicate cell divisions and the united vessels at 2 and 3 dag, respectively. Note that the good union of the 3 dag graft caused poor staining for PI permeation. (G–I) GUS staning of cycB/cycB grafts. GUS activity in the graft union is indicated by arrows at 2 dag. At 3 dag, the arrow indicates the united vessels, while the arrowhead indicates deep staining showing an adventitious root. (J–L) Sections of grafts in Arabidopsis. Samples were embedded in Technovit 7100 (Kulzer) and 5μm sections were cut. Arrows indicate the graft unions. SC, Scion; ST, stock. Bars, 50 μm.

To determine experimentally whether the transport activity of vascular tissue was restored, grafting with the combination of SUC2/WT was conducted. The proSUC2:GFP transgenic line has been used as a marker for detection of source-to-sink transition (Imlau ; Oparka ). Under the control of the AtSUC2 promoter, soluble GFP was expressed specifically in phloem companion cells, in which GFP was diffused into the phloem and passively transported with the photoassimilates in the direction of source to sink. Thus, detection of GFP fluorescence in the WT stock root of SUC2/WT grafts could be indicative of the recovery of phloem transport activity. Consistent with the histological results, GFP fluorescence was observed in WT stock root, and its distribution pattern was the same as in SUC2 intact seedling roots (Fig. 4A–C). SUC2/WT grafts in which GFP was detected in the stock root were subsequently stained with acid fuchsin to explore the recovery of xylem activity. The dye transport was inspected in cotyledons, shown as a red leaf vein (Fig. 4D–F). The results showed that, as in WT intact seedlings (Fig. 4F), the SUC2 scion was stained red in the leaf vein (Fig. 4D, E), suggesting the functional transport activity of the xylem in the graft. Taken together, these results demonstrated functional recovery of the graft vasculature at 3 dag.

Fig. 4.

Functional analysis of vascular transport activity in SUC2/WT grafts. (A–C) GFP fluorescence detected in WT stock roots. Images are the merges of GFP and DIC. WT and SUC2 represent the intact seedlings, respectively. Bar, 100μm. (D–F) Acid fuchsin transport experiment. (D) and E) represent the graft, and (F) represents a WT intact seedling. The roots were submerged in fuchsin solution, as shown in (D), with a higher magnification shown in (D). Arrows indicate the red leaf veins stained by fuchsin.

Auxin canalizes the pathway to direct the reconnection of vascular tissue between the scion and stock

From the histological experiments, a vascular differentiation between 2 and 3 dag could be inferred. Auxin significantly contributes to the pattern formation of vascular tissue (Sauer ; Scarpella ; Caño-Delgado ), so the distribution of GUS staining of proDR5:GUS was investigated during graft-union development.

In ungrafted DR5 scions, there was no detection of GUS staining close to the graft union at 1–3 dag. GUS activity was observed mainly in the vasculature of the hypocotyls, which is the site for adventitious root initiation (Supplementary Fig. S3 at JXB online). For DR5/WT scions, GUS staining was absent from the graft union and was seen mainly in the middle of hypocotyls at 1 dag (Fig. 5A). A deep staining was observed in vascular tissue adjacent to the graft union at 2 dag (Fig. 5B, arrow). PI staining demonstrated that cell division was initiated at 2 dag (Fig. 3B), and no GUS staining of the cycB/cycB union was recorded before 2 dag in the graft union (Fig. 3G). In DR5/WT grafts, GUS staining could also be detected at 2 dag (Fig. 5B). These results demonstrated that auxin initiated cell division and differentiation at 2 dag, as well as its function in organ formation (Benková ). At 3 dag, GUS staining could be observed across the graft union (Fig. 5C). The significance of auxin was further emphasized by WT/DR5 grafts. In contrast to the extensive distribution in the scion, accumulation of auxin appeared as a concentrated pattern that was distributed exclusively in a group of cells within the graft union (Fig. 5D–F). This specific pattern was detected as early as 2 dag, approximate 40 h after grafting (Fig. 5D), and was maintained throughout the healing process until the reconnection of vasculature (Fig. 5D–F). Furthermore, the tracheary elements were detected to differentiate towards the auxin accumulation site (Fig. 5E, arrows) and eventually reconnected the vessels at the site (Fig. 5F, arrow). Taking these into account, a central organizer role of auxin acting as a director for reconnection of the vascular tissue in the graft union was concluded.

Fig. 5.

DR5 grafts show a pivotal role for auxin in vascular reconnection. (A–C) DR5/WT grafts. At 1 dag (A), no obvious GUS activity was detected closed to the graft union, but by 2 dag, there was a deep staining adjacent to the union (white arrow; B). Vessel union was apparent by 3 dag (white arrow; C). Arrowheads indicate adventitious root initiation. Bars, 100 μm. (D–F) WT/DR5 grafts. The earliest detection of GUS staining was at 40 h after grafting (D). The arrows in (E) indicate the vessel elements that differentiated towards the accumulation site at 62 h. The arrow in (F) represents the united vessels at the auxin accumulation site. Dotted lines show the graft union and the scions and stocks were as indicated. Bars, 50 μm.

Transcriptional analysis demonstrates that cell–cell communication accompanies clearing of cell debris under the control of a wound-induced programme at 1 dag

Comparing the GUS distribution of ungrafted DR5 scion (Supplementary Fig. S3, 2 dag) with DR5/WT grafts (Fig. 5B, arrow), it was apparent that some unknown signals were being transported from the stock to the scion to induce the auxin distribution changes. These results suggested that a communication network had been established between the scion and stock at 2 dag. A cell–cell communication process that represents signal exchange between cells of scion and stock is obviously compulsory for subsequent processes in graft-union development. In addition, no obvious changes could be detected by anatomic analysis (Fig. 3A, D, G, J) at 1 dag, implicating a potential signal-exchange process. To confirm this hypothesis and to characterize the molecular identity of the graft-healing processes on a genome-wide scale, microarray analysis was performed to investigate the transcriptome changes at 1 dag in WT/WT grafts.

Samples were grouped as follow: group A comprised grafts, group B was ungrafted scions and stocks, and group C was intact seedlings. Microarray data were first compared between groups A and C, and differentially expressed probe sets were then tested based on groups A and B to obtain graft-specific probes (Supplementary Fig. S4 at JXB online). Differential probes were annotated for a total of 306 genes, including 120 upregulated and 186 downregulated genes according to the fold change in group A relative to group C (Supplementary Fig. S5 and Table S2, and Supplementary File 2 at JXB online). Several genes showing expression-level changes were confirmed by real-time PCR (Fig. 6A). Furthermore, the promoter region of HSP21 (At4G27670) was fused upstream of the GUS coding sequence to control enzyme activity, and GUS staining also suggested a graft-specific pattern (Fig. 6B–E).

Fig. 6.

Corroboration of microarray data. (A) Quantitative RT-PCR analysis. Expression levels were normalized relative to those of tubulin, and the levels in grafts are shown relative to the level in intact seedlings, which was set to 1.0. Error bars represent standard deviation from three independent experiments. (B–E) GUS staining for HSP21 promoter activity during graft-union development. The results for the graft of proHSP21/WT are shown in (B), while (C) shows the ungrafted proHSP21 scion. Arrows indicate that grafting raised the GUS activity in the proHSP21/WT graft. For intact seedlings, GUS staining could not be detected in hypocotyls (data not shown). The fold changes in HSP21 were 49.35 and 7.98 for group A relative to group C and group A relative to group B, respectively (Supplementary Table S2). A graft of WT/proHSP21 is shown in (D), while (E) represents ungrafted proHSP21 stock. Arrows in (D) indicate GUS staining close to the graft union; this staining was absent in the ungrafted stock (E). Bars, 100 μm (B); 50μm (D).

Significant genes were classified according to GO and KEGG (Tables 1–Tables 4) to explore their functional significance. Among the GO categories, 15 and eight categories were significantly over-represented for up- and downregulated genes, respectively, based on a gene-enrichment test (Tables 1 and 2). In downregulated genes, the cell growth (GO:0016049) category was present, indicating a logical consequence for regeneration after wounding (Table 2). The upregulated genes (Table 1) were more noteworthy than the downregulated ones. Lyase activity (GO:0016829), hydrolase activity (GO:0016787) and oxidoreductase activity (GO:0016491) demonstrated that cells were sweeping away the cell debris caused by wounding. A set of stimulus-response categories was significantly over-represented, suggesting that a wound-induced programme took part in the graft-union developmental processes. This was further confirmed by over-representation of the biosynthesis of ethylene and jasmonic acid in the KEGG plant hormone biosynthesis pathway (Table 3, Fig. 7), as these phytohormones are typical responses involved in wounding (Schilmiller and Howe, 2005). More importantly, the endomembrane system (GO:0012505) was calculated as having the lowest P value and q-value for both up- and downregulated genes (Tables 1 and 2), suggesting a crucial contribution to graft-union development. The endomembrane system consists of serial membranous organelles essential for various aspects of plant development and signal transduction, and is an important passageway for cells to communicate with the environment (Surpin and Raikhel, 2004). In planta, the endomembrane system offers many regulated activities for reproductive processes, which involve substantial signal exchanges between pollen and the pistil (Cheung and Wu, 2008; Kumar and McClure, 2010). For grafts, there is no existing route for communication between the scion and stock cells, which consequently rely on signal transduction across the plasma membrane. In accordance with this, most of the five protein kinases out of the 120 upregulated genes participated in acceptance of extracellular signals. Two of them, At4G04490 and At1G70250, had receptor-like protein kinase activity (Table 5). CIPK5 (At5G10930) and CIPK14 (At5G01820) belong to the calcineurin B-like (CBL)-interacting protein kinase (CIPK) family (Table 5), which represents a critical signalling system in response to a wide range of environmental stimuli (Weinl and Kudla, 2009). Taken together, the microarray data revealed that a cell–cell communication process accompanied the clearing of cell debris during the early stage of the graft-healing process, and that both were initiated by a wound-induced programme.

Table 1.

Significant categories for upregulated genes in GO

GO ID	Name	Hits	Total	P valuea	q-valueb
GO:0003674	Molecular function
GO:0016491	Oxidoreductase activity	14	1420	0.0012	0.0046
GO:0016787	Hydrolase activity	22	2933	0.0018	0.0060
GO:0016829	Lyase activity	6	333	0.0020	0.0060
GO:0005575	Cellular component
GO:0012505	Endomembrane system	40	4046	<1.0E–4	<1.0E–4
GO:0008150	Biological process
GO:0006955	Immune response	7	312	2.0E–4	0.0013
GO:0044238	Primary metabolic process	44	7802	0.0029	0.0080
GO:0007568	Aging	3	74	0.0032	0.0088
GO:0006950	Response to stress	20	1955	1.0E–4	5.0E–4
GO:0006955	Immune response	7	312	2.0E–4	0.0013
GO:0009605	Response to external stimulus	6	332	0.0019	0.0060
GO:0009607	Response to biotic stimulus	8	583	0.0020	0.0060
GO:0009628	Response to abiotic stimulus	15	1156	<1.0E–4	4.0E–4
GO:0009719	Response to endogenous stimulus	12	832	1.0E–4	7.0E–4
GO:0042221	Response to chemical stimulus	17	1746	4.0E–4	0.0018
GO:0051707	Response to other organism	8	539	0.0012	0.0046

P ≤0.01.

q ≤0.01.

Table 4.

Significant pathways for down-regulated genes in KEGG

Pathway name	Hits	Total	P valuea	q-valueb
Biosynthesis of phenylpropanoids	5	247	0.0139	0.0139
Methane metabolism	5	83	1.0E–4	2.0E–4
Phenylalanine metabolism	5	81	1.0E–4	2.0E–4
Phenylpropanoid biosynthesis	6	105	<1.0E–4	2.0E–4
Starch and sucrose metabolism	3	106	0.0234	0.0195
Zeatin biosynthesis	2	22	0.0079	0.0099

P ≤0.05.

q ≤0.05.

Table 2.

Significant categories for down-regulated genes in GO

GO ID	Name	Hits	Total	P valuea	q-valueb
GO:0003674	Molecular function	166	29448	0.9753	0.6903
GO:0016787	Hydrolase activity	32	2933	5.0E–4	0.0039
GO:0005199	Structural constituent of cell wall	4	39	1.0E–4	0.0011
GO:0046906	Tetrapyrrole binding	8	320	7.0E–4	0.0048
GO:0004601	Peroxidase activity	5	109	6.0E–4	0.0040
GO:0005575	Cellular component	165	28657	0.8255	0.6903
GO:0012505	Endomembrane system	53	4046	<1.0E–4	<1.0E–4
GO:0030312	External encapsulating structure	17	553	<1.0E–4	<1.0E–4
GO:0008150	Biological process	166	28723	0.7888	0.6903
GO:0016049	Cell growth	9	229	<1.0E–4	2.0E–4
GO:0065008	Regulation of biological quality	11	610	0.0011	0.0061

P ≤0.01.

q ≤0.01.

Table 3.

Significant pathways for up-regulated genes in KEGG

Pathway name	Hits	Total	P valuea	q-valueb
α-Linolenic acid metabolism	3	28	2.0E–4	3.0E–4
Biosynthesis of plant hormones	4	304	0.0267	0.0055
Flavone and flavonol biosynthesis	1	4	0.0181	0.0043
Metabolic pathways	12	1257	0.0024	0.0017
Pentose and glucuronate interconversions	2	32	0.0069	0.0027
Starch and sucrose metabolism	3	106	0.0075	0.0027
Taurine and hypotaurine metabolism	1	4	0.0181	0.0043

P ≤0.05.

q ≤0.05.

Fig. 7.

Significant categories of jasmonic acid and ethylene biosynthesis pathways in KEGG. The jasmonic acid and ethylene biosynthesis pathways are adapted from the KEGG website. Bold arrows indicate the steps of enzymatic catalysis that were differential expressed in the microarray data. (This figure is available in colour at JXB online.)

Table 5.

Differential expressed kinases in upregulated genes

Gene ID	Name	Familya	Fold change AC	Fold change AB
AT4G04490	CRK36	Cysteine-rich receptor-like protein kinase	3.27	0.72
AT5G48740		Leucine-rich repeat protein kinase family protein	2.79	1.76
AT1G70250		Transmembrane receptor protein serine/threonine kinase activity	2.5	1.27
AT5G10930	CIPK5	CBL-interacting protein kinase 5	2.39	1.63
AT5G01820	CIPK14	CBL-interacting protein kinase 14	2.27	2.07

Gene information was from TAIR10 (The Arabidopsis Information Resource, http://www.arabidopsis.org/).

Model refinement for graft-union developmental stages

Summarizing the results, a concise model was proposed to illustrate graft-union development in general (Fig. 8). At the beginning, wounding elicited the programmes that stimulated the undamaged cells to eliminate cell debris at the cut surface. Accompanying this clearing process was signal exchange between intimate contacting cells of the scion and stock at the graft interface. This reconstructed the communication network in the graft union to give rise to an integrated background for subsequent events. Once the communication was reconstructed, auxin accumulated, contributing in particular to vascular reconnection. Finally, the cells differentiated into vascular tissues with the auxin in control to reconnect the vasculature between scion and stock. At this point, the key event in the graft-healing process was accomplished, indicating a successful grafting.

Fig. 8.

Model refinement for graft-union developmental stages. This model describes the causal steps for graft-union development, which can also be used for exploration of union development based on traditional grafting methods. In this model, comparing grafts and ungrafted scions and stocks, it is apparent that the process of signal exchange between scion and stock takes a pivotal position in the graft-healing process, because it is a prerequisite for the subsequent events. (This figure is available in colour at JXB online.)

Discussion

Applications of seedling micrografting towards a research platform for diverse aspects of plant biology

In recent decades, grafting has been used to study the long-distance signalling between the shoot apical meristem and root apical meristem, which is essential for plant development and adaptation to environmental change (Turnbull ; Xia ; Estañ ; Rus ; Sigüenza ; Brosnan ; Christmann ; Foo ; Matsumoto-Kitano ; Thatcher ; Zhou ; Molnar ). In this report, an improved seedling micrografting process was characterized and validated in plants. In our long-term experience with this practice, this method is satisfactory in sample analysis and has a high success rate, and also saves labour, time, and reagents. At the same time, there is a high degree of application similarity in other dicotyledonous plants. There are reasons to believe that this method would function better in long-distance signalling research than current methods.

In graft-union development, substantial changes occur, such as cell differentiation and organ regeneration, and even genetic material can be exchanged between the scion and stock at the graft site (Stegemann and Bock 2009). Compared with traditional grafting methods used previously (Kollmann and Glockmann 1985; Napoli, 1996; Fernandez-Garcia ; Estañ ; Sigüenza ; Flaishman ; Stegemann and Bock 2009; Zhou ), this seedling micrografting method is better suited for sample analysis. The simple procedure and high success rate will ensure that researchers can generate abundant and homogeneous grafts within a short period of time to fulfil the requirements of an experiment. A similar micrografting method was reported recently in Nicotiana attenuata plants (Fragoso ), but its use of agar blocks would obviously interfere with observation of the union. Using our improved protocol, the graft union could be inspected easily and flexibly, which will facilitate investigations based on numerous scion/stock combinations with various experimental techniques to interpret the mechanisms underlying the healing process.

Recently, Molnar et al. (2010) reported that a substantial proportion of endogenous small RNAs had moved across the graft union, and these mobile small RNAs directed DNA methylation in the genome of the recipient cells. DNA methylation differences can be stably inherited across generations in the absence of extensive DNA sequence polymorphisms, and these epigenetic variations influence the plant phenotype impressively and even affect some quantitative traits, such as plant height and yield (Hauben ; Johannes ; Reinders ). These findings open up a new avenue for testing the epigenetic impacts on plant development and heredity by creating desirable combinations of shoot and root phenotypes or genotypes with grafting. Bringing the full power of the seedling micrografting method into play will be critical for rapid progress in understanding plant epigenetics, and for developing new and better ways for plant improvement.

Hormone functions in the union process

The microarray data showed that ethylene and jasmonic acid synthesis were elevated. This early activation of ethylene and jasmonic acid is consistent with findings made in the tissue-reunion process in partially incised inflorescence stems of Arabidopsis (Asahina ). In contrast, gibberellin is required for cell division during tissue reunion of cucumber and tomato hypocotyls (Asahina ). Although we used hypocotyls of Arabidopsis for grafting, the hormones involved in the healing process are similar to those of inflorescence stems rather than the hypocotyls in cucumber and tomato.

The histological data showed that cell division commenced after auxin accumulation (Figs 3 and 5). The precedence of ethylene and jasmonic acid elevation to auxin accumulation remind us of the importance of ethylene and jasmonic acid during the signal-exchange process between scions and stocks. Unlike partially incised inflorescence stems, signal exchange is indispensable for subsequent events during graft-union healing. It is conceivable that similar mechanisms activated by ethylene and jasmonic acid after wounding may play a part in the early responses of both grafted hypocotyls and injured stems in Arabidopsis, but the subsequent signal transduction is different between these two processes.

Similarity of expression profiling between groups A and B

The microarray data showed a high similarity of expression profiling between sample groups A and B (Supplementary Fig. S5). This was a reflection of the high similarity between groups these two groups. The only difference between groups A and B was grafting, which may be less influential at such an early developmental stage (1 dag). This high similarity also demonstrated that plants utilized the wound-induced programme to accomplish communication between the scion and stock. Activation of the ethylene and jasmonic acid biosynthesis pathway (Fig. 7) and the significant stimulus-response categories (Table 1) confirmed this assumption, and were also a reflection of the high similarity of expression profiling between groups A and B.

What is the signal exchanged between scion and stock?

In the model developed in this report, the development of seedling grafts involved several stages in Arabidopsis. However, the necrotic layer and callus reported previously in inflorescence stem grafts (Flaishman ) were not observed. This was probably because of the high potential for division and differentiation of the young seedling cells. The reconnection of vasculature between the scion and stock is essential for a successful graft, and the results indicated a director role for auxin in vascular differentiation and reconnection. A well-established cell–cell communication network is obviously essential in exercising the appropriate functions of auxin, and this was corroborated by the results of the transcriptome analysis. The subsequent question of the molecular nature of the signals exchanged between scion and stock is becoming clearer.

In the differentially expressed genes, SWEET15 (AT5G13170, also named senescence-associated protein SAG29) and SWEET9 (AT2G39060) were recently identified as a new class of sugar transporters (Chen ), and SWEET15 is located at the plasma membrane (Seo ). CIPK14 (AT5G01820) is regulated by sucrose (Chikano ). Sugars, as the energy source for plant growth and development, are also fundamental in coordinating metabolic fluxes in response to the changing environment (Ramon ). It is conceivable that SWEET15 and SWEET9 perceive extracellular sucrose from the scion and transduce them into cells of the stock at the graft interface, in which CIPK14 incorporates the signals and transmits them to appropriate downstream effectors, thereby initiating cell–cell communication between cells of the scion and stock.

Oligosaccharides released from plant cell walls are powerful signalling molecules capable of conveying information to elicit multiple changes in plant defence responses (John ). In the clearing of cell debris, oligosaccharides would be generated from the hydrolysis of cell walls and perceived by specific receptors, which will activate the signal cascade and consequently achieve cell–cell communication.

Plasmodesmata are important channels for signal transduction between plant cells in various plant processes, such as pollen mother-cell development (Wang ). Secondary plasmodesmata will be substantially generated during graft-union development (Kollmann and Glockmann 1985). The expression levels of plasmodesmata-located protein PDLP1A (At5g43980) and At2g41870 were elevated in the microarray data. PDLP1A targets to the plasmodesmata and participates in plasmodesmal trafficking (Thomas ). At2g41870 belongs to the remorin family, which clusters at plasmodesmata and is involved in potato virus X cell–cell movement, suggesting functionality in macromolecular trafficking through plasmodesmata (Raffaele ). These results implicate that plasmodesmata may contribute to cell–cell communication in graft-union development.

These hypotheses need to be pursued further in future experiments.

Supplementary data

Supplementary data are available at JXB online.

. Grafts in tobacco.

. Grafts in alfalfa.

. GUS staining in DR5 scions.

. Procedure of microarray data analysis.

. Heatmap for differential expressed genes.

. Primers used in quantitative RT-PCR.

. The 306 graft-specific genes.

. Operational procedures for grafting in tobacco, tomato, and alfalfa.

. This Excel document lists the 306 significantly altered genes.