New insights into an old story: Agrobacterium-induced tumour formation in plants by plant transformation

Abstract

Agrobacterium tumefaciens causes tumour formation in plants. Plant signals induce in the bacteria the expression of a range of virulence (Vir) proteins and the formation of a type IV secretion system (T4SS). On attachment to plant cells, a transfer DNA (T-DNA) and Vir proteins are imported into the host cells through the bacterial T4SS. Through interaction with a number of host proteins, the Vir proteins suppress the host innate immune system and support the transfer, nuclear targeting, and integration of T-DNA into host cell chromosomes. Owing to extensive genetic analyses, the bacterial side of the plant–Agrobacterium interaction is well understood. However, progress on the plant side has only been achieved recently, revealing a highly complex molecular choreography under the direction of the Vir proteins that impinge on multiple processes including transport, transcription, and chromosome status of their host cells.

Introduction

Agrobacterium species are known as the only organisms capable of interkingdom gene transfer. This soil-borne Gram-negative bacterium is a broad-host range plant pathogen, which initiates tumour formation on most dicotyledonous and some monocotyledonous species (DeCleene and DeLay, 1976). Such tumours do not require the continuous presence of the bacteria for proliferation (White and Braun, 1942), showing that the plant cells have been transformed genetically. The factors required for tumour formation are encoded on a large tumour-inducing (Ti) plasmid of virulent Agrobacterium strains. The Ti plasmid also serves as a source for the transfer DNA (T-DNA), a DNA region that is imported into plant cells and integrated into the host chromosomal DNA—resulting in genetic manipulation of the host. The expression of T-DNA-encoded bacterial genes in the host cell results in the production of enzymes that catalyse the synthesis of plant hormones, which are responsible for tumour growth and the formation of novel amino-acid–sugar conjugates, termed as opines. As opines can serve as carbon and sometimes nitrogen sources for Agrobacterium to the exclusion of most other microorganisms, they provide a selective advantage for this species (Tempé and Petit, 1982). The capacity for gene transfer into plants has been used to develop Agrobacterium tumefaciens as a vector for genetic manipulation. Engineered DNA segments of interest, which are first cloned into the T-DNA region of ‘disarmed' plasmids, are then introduced into Agrobacterium and subsequently transferred into plants. From these disarmed plasmids, the genes responsible for tumourous growth have been removed, ensuring that the transformed cells can be regenerated into fertile plants that transmit the engineered DNA to their progeny (Hooykaas and Schilperoort, 1992, Newell, 2000). By these means, the host range of Agrobacterium has been extended to include other bacterial species as well as fungi and even some mammalian cells (Lacroix et al, 2006). Under laboratory conditions, normally recalcitrant plants (Ishida et al, 1996; Hiei et al, 1997; Chen et al, 2006), fungi (Bundock et al, 1995; Abuodeh et al, 2000), and even human cells (Kunik et al, 2001; Tzfira et al, 2006) can be transformed by Agrobacterium. Agrobacterium-mediated transformation serves as an important model system for studying host–pathogen recognition and delivery of macromolecules into target cells. The interaction between Agrobacterium and plant cells can be divided into several steps: recognition, virulence (Vir) gene expression, attachment to the host cell, targeting of Vir factors and T-DNA into the host cell, and chromosomal T-DNA integration (Figure 1). On chemical recognition of plant-derived compounds, Agrobacterium Vir gene expression is induced, which is followed by the physical interaction between bacterium and plant cells. A bacterial transfer machinery is subsequently produced and assembled to import the de novo produced T-DNA strand along with a number of Vir factors into the host cell. Once inside the plant cell, the T-DNA is translocated into the nucleus, in which it integrates into the host chromosome. On expression of T-DNA genes, plant cells are re-programmed for tumour growth and production of opines.

Figure 1.

Overview of the Agrobacterium–plant interaction. 1. Plant signals induce 2. VirA/G activation and thereby 3. T-DNA synthesis and vir gene expression in Agrobacterium. 4. Through a bacterial type IV secretion system (T4SS) T-DNA and Vir proteins are transferred into the plant cell to assemble a T-DNA/Vir protein complex. 5. The T-DNA complex is imported into the host cell nucleus in which 6. the T-DNA becomes integrated into the host chromosomes by illegitimate recombination.

Recognition of plant cells as host by Agrobacterium

Agrobacterium strains are widely distributed in the soil. Moreover, most isolates do not contain a Ti plasmid and are capable of living independently of a plant host. Yet, as tumour-produced opines are a specific food source for Agrobacterium, the capacity of Agrobacterium strains to induce such tumours is a clear selective advantage. However, because plant transformation is a complex process and energetically demanding, Vir gene expression must be carefully regulated. The identification of vir genes, which are required for virulence, but lie outside the T-DNA (Klee et al, 1983; Stachel and Nester, 1986), was a major step towards understanding the transformation process. With the exception of virA and virG, the vir genes were found to be essentially silent unless the bacteria are cultured with plant cells (Stachel and Nester, 1986). Although vir gene induction depends on molecules exuded by the plant, attachment to plant cells is necessary for transformation and is mediated by chromosomally encoded Agrobacterium genes (Lippincott and Lippincott, 1969; Douglas et al, 1982). Thus, host recognition by Agrobacterium resulting in transformation is composed of two independent processes: Vir gene activation and attachment to the host cell.

Agrobacterium Vir gene expression

The vir gene activation by plant factors requires two genes, virA and virG (Stachel and Nester, 1986), which are constitutively expressed at a basal level, but can become highly induced in a feed-forwards manner (Winans et al, 1988). The virA and virG genes encode a two-component phospho-relay system in which VirA is a membrane-bound sensor and VirG is the intracellular response regulator (Wolanin et al, 2002). On signal sensing, the histidine kinase VirA activates VirG through transferring its phosphate to a particular aspartate of VirG, thereby activating VirG to function as a transcription factor. Phosphorylated VirG then binds at specific 12 bp DNA sequences of the vir gene promoters (vir boxes), thereby activating transcription (Brencic and Winans, 2005).

The signals perceived by VirA are phenols, aldose monosaccharides, low pH, and low phosphate (Palmer et al, 2004; Brencic and Winans, 2005). Phenols are indispensable for vir gene induction, whereas the other signals sensitise VirA to phenol perception, for example sugars allow induction of the VirA/VirG system at much lower phenol concentrations and increase the response several-fold (Shimoda et al, 1990). The identification of phenols, such as acetosyringone, as inducers of vir gene expression was achieved through analysis of root exudates and leaf protoplasts (Stachel et al, 1985). Acetosyringone is now routinely used for enhancing the efficiency of Agrobacterium-mediated plant transformation. The capability of the VirA/VirG system to recognise a diversity of phenols and sugars is a likely explanation for the broad-host range exhibited by Agrobacterium.

Plant entry sites for Agrobacterium

In nature, Agrobacterium attacks mainly wounded tissue (Braun, 1952). A wound site may simply be a portal of entry, but other specific processes specifically occurring at these sites are likely to facilitate transformation: wound-secreted compounds such as phenols and sugars induce vir gene expression. In addition, the latter act as chemotactic attractants of Agrobacterium. Thus, wound-specific features such as high activity of the phenylpropanoid pathway, low pH, and sugars associated with cell wall synthesis/wound repair correlate with enhanced transformation frequency and efficiency (Baron and Zambryski, 1995). Although transformation can also occur in unwounded plants—with Agrobacterium cultures grown in pre-induction medium (Escudero and Hohn, 1997)—it seems that Agrobacterium has optimised the VirA/VirG system to respond to signals from wound sites. Cell division activity at the wound sites is thought to be equally important for transformation (Braun, 1952). However, cells in the root elongation zone were found to be the most highly transformable (Yi et al, 2002). Cells of this non-meristematic zone are not undergoing a normal cell cycle, but endoreduplication.

Host cell attachment by Agrobacterium

As T-DNA and proteins are transferred from A. tumefaciens into plant cells, an intimate association between pathogen and host cell is a prerequisite for transformation. Quantitative-binding assays have revealed a non-specific interaction that is readily removed and a specific interaction (Neff and Binns, 1985). The specific attachment of A. tumefaciens to plant cells is not dependent on the Ti plasmid (Douglas et al, 1982; Neff and Binns, 1985). Instead, it is facilitated by the chromosomally encoded bacterial genes chvA, chvB, and pscA (exoC), which are involved in the synthesis and/or localisation of periplasmic β-1,2 glucan (reviewed in McCullen and Binns, 2006).

Early studies revealed that the exposure of A. tumefaciens cells to soluble pectic plant cell wall fractions decreases both the specific binding of Agrobacterium to plant cells and tumour-induction frequencies (reviewed in Gelvin, 2000), suggesting the presence of as yet elusive Agrobacterium receptor-like components. Possible candidates are BTI-domain proteins that had been isolated from a screen for VirB2-interacting proteins. Owing to its transient increase immediately after Agrobacterium infection and its preferential localisation to the periphery of root cells, a direct contact of BTI1 with the Agrobacterium T-pilus in the initial interaction of Agrobacterium with plant cells has been proposed (Hwang and Gelvin, 2004).

Genomic studies are beginning to provide new insight into possible plant molecules involved in the attachment process. A number of Arabidopsis mutants have been isolated that are recalcitrant to Agrobacterium transformation (rat mutants) (Zhu et al, 2003) and it was shown that Agrobacterium can no longer bind efficiently to some rat mutants. One well-characterised mutant is affected in the gene encoding a cell wall arabinogalactan protein to which bacteria bind poorly (Nam et al, 1999; Zhu et al, 2003). Further analysis of these mutants should help to unravel the recognition process and physical interaction of Agrobacterium and host cells.

Agrobacterium secretion of T-DNA and Vir proteins into plant cells

After vir gene activation and attachment of Agrobacterium to plant cells, a transporter complex formed by VirB proteins and VirD4 enables Vir proteins and T-DNA to cross the inner bacterial membrane, the peptidoglycan layer, and outer membrane, as well as the plant host cell wall and membrane. The VirB complex belongs to the class of type IV secretion systems (T4SS), which are found across a broad range of Gram-negative bacteria and are involved in the conjugative transfer of plasmids between bacteria as well as the translocation of Vir factors from pathogens to host cells during infection (Cascales and Christie, 2003).

The VirB complex is composed of at least 12 proteins: VirB1–11 and VirD4 and is required for virulence. The proteins associate with the cell envelope and form a multi-subunit envelope-spanning structure (Christie et al, 2005). The bacterial factors transported into host cells by the VirB complex include the VirD2-T-DNA, VirE2, VirE3, VirF, and VirD5 (Vergunst et al, 2005). VirD2 nicks the T-DNA at the 25-nucleotide long repeats that border the T-DNA and VirD2, then becomes covalently bound to the 5′-end of the T-DNA. VirD2 seems to be transported with the T-strand into the plant cell, in which it is involved in nuclear import and integration of the T-DNA into the host genome (Gelvin, 2003). VirE2 is a single-stranded DNA-binding protein that can coat the length of the T-strand in vitro (Christie et al, 1988; Citovsky et al, 1992). It likely interacts with T-DNA in the plant cell cytoplasm and also has functions in nuclear import and integration (Gelvin, 2003). Intriguingly, the VirB/D4 complex can not only transport Ti-derived T-DNA, but also the broad-host range plasmid RSF1010 to plants or other Agrobacterium species, showing that conjugative intermediates must also be substrates (Buchanan-Wolloston et al, 1987; Beijersbergen et al, 1992).

Import of Agrobacterium Vir factors into host cells

The A. tumefaciens virB-encoded T4SS transports substrates across the bacterial cell envelope. Certain C-terminal motifs were found to be required for the export of targeted substrates. These export signals mediate the interaction of substrates with the T4SS. The C-termini of VirF, VirE2, and VirE3 are sufficient to mediate transport of fusion proteins to plants (Vergunst et al, 2000). The minimal size of VirF required to direct protein translocation to plants is the C-terminal 10 amino acids (Vergunst et al, 2005), from which the minimal consensus sequence R-X(7)-R-X-R-X-R required for substrate secretion by the VirB complex could be derived (Vergunst et al, 2005).

C-terminal fusions of VirE2 blocked its translocation to host cells (Vergunst et al, 2000). Accordingly, insertion of a FLAG tag at the C-terminus of VirE2, or truncation of the C-terminal 18 amino acids of VirE2, renders the protein non-functional in A. tumefaciens, while not affecting its capability to bind single-stranded DNA (Simone et al, 2001). However, overexpression of such VirE2 C-terminal mutant derivatives in transgenic plants confers susceptibility to transformation by an A. tumefaciens virE2-deficient strain, suggesting that the mutations disrupted a region of amino acids required for translocation, such as a secretion signal.

In an elegant experimental approach, using fusion proteins of VirE2 or VirF to the Cre recombinase, the transport of these proteins in the absence of T-DNA has been studied (Vergunst et al, 2000). The experiments were designed in such a way that the transport of Cre-VirE2 or Cre-VirF fusions into host plant cells results in a recombination event conferring kanamycin resistance to host tissues. Thus, it could be shown that the transport of bacterial factors is dependent on the VirB/VirD4-complex. In addition, VirE3 and the C-terminus of VirD5 were found to mediate substrate targeting into host cells (Schrammeijer et al, 2003; Vergunst et al, 2005). VirE3 may function in the plant to aid nuclear localisation of VirE2 (Lacroix et al, 2005), and VirD5 may function as transcription factor in the plant cells (Schrammeijer et al, 2000).

Host cell entry of Agrobacterium factors

It is presently unclear how the Vir proteins and the T-DNA protein complex traverse the host cell wall and membrane barriers. In T4SS-mediated plasmid transfer, the pilus enables the interaction between donor and recipient, followed by the fusion of outer membranes in a mating junction (Schroder and Lanka, 2005). The mechanism by which the transferred conjugal intermediate traverses the bacterial wall and inner membrane is not known. Even less is known about VirB-mediated transfer across host cell barriers. The enormous host range transformed by Agrobacterium suggests that the specificity of host–pathogen interaction required to breach the host cell wall and membrane barriers may be less important than expected.

Targeting of Agrobacterium T-DNA into the host cell nucleus

Once inside the plant cell, the T-DNA must find its way into the nucleus. Several Agrobacterium Vir proteins, as well as a number of plant proteins, seem to be involved in this process (Figure 2). The proteins VirD2 and VirE2 contain plant-active nuclear localisation signal (NLS) sequences. VirD2, which is covalently linked to the 5′-end of the T-DNA, contains two NLS regions, both of which can direct chimeric proteins to the nucleus. Sterical considerations suggest that the bipartite NLS in the carboxy-terminus of VirD2 might be biologically important for nuclear targeting of the T-DNA complex (Tinland et al, 1995).

Figure 2.

Molecular details of Agrobacterium and host factors involved in T-DNA transfer and host responses. 1. Nuclear import of the T-DNA complex along with other Vir proteins. 2. VIP1 becomes activated by PAMP-triggered MAPK activation, but undergoes proteasomal degradation through VirF. 3. Manipulation of host gene expression by Vir proteins helps 4. T-DNA integration and tumour development as well as suppression of defence responses.

VirE2 protein contains two separate bipartite NLS regions that can target fusion reporter proteins to plant nuclei (Citovsky et al, 1992, 1994). Fluorescently labelled single-stranded DNA coated with VirE2 and microinjected into plant cells localises to the nucleus, whereas naked single-stranded DNA remains in the cytoplasm (Zupan et al, 1996).

In agreement with a function of VirD2 and VirE2 in T-DNA nuclear guidance, deletion of the VirD2 bipartite NLS resulted in almost complete loss of transformation (Rossi et al, 1993), indicating that VirE2 NLS domains cannot compensate for the loss of the VirD2 NLS. Furthermore, VirD2 and VirE2 proteins were shown to be necessary for nuclear targeting of in vitro synthesised T-complexes in permeabilised HeLa cells (Ziemienowicz et al, 1999).

As shown by Shurvinton et al (1992), the C-terminal NLS of VirD2 is essential for virulence, but not for intrabacterial T-strand production. This NLS does not contribute to targeting of the T-DNA to the nucleus (Shurvinton et al, 1992; Rossi et al, 1993; Mysore et al, 1998), and it was suggested that deletion of the VirD2 NLS may alter the structure of the VirD2 protein such that it can still nick the T-DNA border, but may fail to pass through the T4SS or the nuclear pore (Mysore et al, 1998). VirE2 might provide nuclear targeting in the absence of VirD2 NLS. In support of this notion, Gelvin (1998) observed that an A. tumefaciens virE2 virD2ΔNLS double mutant was able to form tumours on VirE2-producing transgenic tobacco, but not on wild-type tobacco, and suggested that the NLS of VirE2 could have a function in directing T-DNA to the nucleus.

Several studies suggest additional functions of VirE2 as transmembrane DNA transporter. VirE2 can insert itself into artificial membranes and form channels. These channels can facilitate the efficient transport of ssDNA through membranes (Dumas et al, 2001; Duckely et al, 2005). Indeed, as shown by biophysical experiments and particle-bombarded tobacco cells transiently expressing VirE2 fusion protein, VirE2 seems to actively pull ssDNA into the host (Grange et al, 2008).

Despite the prominent function of VirE2 in the transformation process, some strains of Agrobacterium rhizogenes lacking this protein can still transfer T-DNA efficiently. This is achieved through GALLS proteins (Hodges et al, 2004, 2006, 2009). Interestingly, despite their dissimilarity to VirE2, GALLS protein restored pathogenicity to virE2 mutant A. tumefaciens (Hodges et al, 2004). The GALLS gene encodes for two proteins: full-length GALLS and a C-terminal domain that initiates at an internal in-frame start codon. Full-length GALLS protein contains domains for ATP binding, nuclear localisation, and type IV secretion (Hodges et al, 2006). In plant cells, interaction of GALLS-FL with VirD2 was observed (Hodges et al, 2009). On the basis of these findings, as well as the nuclear localisation of GALLS-FL and its predicted helicase activity, the authors proposed that GALSS-FL may pull T-strands into the nucleus.

VirF has been implicated in the degradation of host cell factors during infection. In the host nucleus, VirF, in concert with the host proteasome machinery, is believed to mediate degradation of the T-DNA complex, thus facilitating the release of the T-DNA and its subsequent choromosomal integration (Schrammeijer et al, 2001; Tzfira et al, 2004a, 2004b).

Functions of Vir proteins in T-DNA integration

Relatively little is known about the precise mechanism of T-DNA integration into the plant genome or the function specific proteins have in this process. The major mode foreign DNA integrates in plants is by illegitimate recombination or non-homologous end-joining; and T-DNA integrates into plant chromosomes by a similar mechanism (Paszkowski et al, 1988; Gheysen et al, 1991; Mayerhofer et al, 1991). An alternative model proposes that the initial invasion of plant DNA by the T-DNA is also of importance for integration (Meza et al, 2002). In addition, analysis of T-DNA-integration sites suggests the involvement of microhomologies in the integration process (Pelczar et al, 2004). Measurements of the relative amounts of transient versus stable expression of reporter genes in Agrobacterium-infected plant cells suggests that most T-DNA is not stably integrated into chromosmal DNA (Nam et al, 1997). Although T-DNA enters the nucleus as a single-stranded molecule, much of the T-DNA likely becomes double stranded, because the conversion to a transcriptionally competent form requires the synthesis of a complementary DNA to the T-strand (Narasimhulu et al, 1996). It is not yet clear whether the T-DNA integrates through single-strand invasion of locally denatured plant DNA (Tinland, 1996), or whether the extrachromosomal double-stranded T-DNA is the substrate for integration. As VirD2 is covalently linked to the T-DNA strand, it likely has some function in the integration process: VirD2 can mediate site-specific cleavage and reversal of it, site-specifically only (Pansegrau et al, 1993). Ligation of T-DNA to plant DNA is rather mediated by host proteins. VirD2 failed to facilitate in vitro ligation-integration reactions in which T-DNA was ligated to a model target sequence, whereas this reaction did take place in the presence of plant extracts (Ziemienowicz et al, 1999). The integration of the 5′ end of the T-strand into plant DNA is generally precise and only a few 5′ nucleotides are usually deleted on T-DNA integration into the plant genome (Tinland et al, 1995). This may result from the protection from exonucleases that VirD2 offers to the capped 5′ T-strand end, as mutations in VirD2 result in imprecise ligation and deletion of the 5′ end of the T-DNA to plant chromosomal DNA (Tinland et al, 1995). Moreover, the ω domain of VirD2, a conserved region outside the NLS, is important for tumourigenesis (Shurvinton et al, 1992; Mysore et al, 1998), but has only minor effects on transient T-DNA expression (Narasimhulu et al, 1996).

The virE2 mutants are extremely attenuated in virulence (Stachel and Nester, 1986). As VirE2 can function as a gated pore for the passage of ssDNA (Dumas et al, 2001; Duckely et al, 2005; Grange et al, 2008), the severe attenuation of virE2 mutant Agrobacterium strains might be explained by a defect in nuclear host transport. However, the integrated T-DNA molecules transferred from virE2 mutant Agrobacterium strains also exhibit extensive deletions corresponding to the 3′ ends of the T-strand (Rossi et al, 1996), suggesting that nucleolytic protection of the T-strand is also a major function of VirE2.

Plant factors and defence responses involved in Agrobacterium tumour formation

Although certainly unintentionally, the host plant actively participates in Agrobacterium transformation. This ‘assistance' occurs at several levels: Vir protein/T-DNA import, dissociation of the Vir/T-DNA complex, T-DNA integration, and re-programming of gene expression for tumour development. A number of host factors that are exploited by Agrobacterium to achieve transformation have been identified. Major progress has been made through yeast-two-hybrid (Y2H) screens for identifying host proteins that interact with Agrobacterium Vir proteins (Ballas and Citovsky, 1997; Deng et al, 1998; Hwang and Gelvin, 2004). Another important progress was achieved through large plant mutant screens (Zhu et al, 2003; Crane and Gelvin, 2007).

Functions of plant proteins in T-DNA nuclear import

VirD2 was reported to interact with several members of the Arabidopsis importin α and cyclophilin families in vitro and in the Y2H system (Koncz et al, 1989; Deng et al, 1998; Bako et al, 2003; Bhattacharjee et al, 2008). Importins α are NLS-binding proteins of the nuclear import machinery that specifically interact with the bipartite NLS region of VirD2 and may thus facilitate its nuclear import (Ballas and Citovsky, 1997). Deng et al (1998) identified an Arabidopsis cyclophilin that interacts in Y2H experiments with a central domain of VirD2. As some cyclophilins have peptidyl-prolyl isomerase activity, the authors speculated that this protein might serve as a chaperonin to hold VirD2 in a transfer-competent conformation during T-strand trafficking through the plant cell. However, this hypothesis is put into question by a recent study in which the cyclophilin-binding domain of VirD2 was found to be dispensable for virulence (van Kregten et al, 2009).

Furthermore, VirD2 is subject to post-translational modification. Nuclear targeting of VirD2 is apparently controlled through phosphorylation of a serine residue close to the bipartite NLS (Tao et al, 2004). Alanine substitution of this residue resulted in the predominantly cytoplasmic localisation of a β-glucuronidase (GUS)-VirD2 NLS fusion protein. Moreover, Tao et al (2004) identified DIG3, a type 2C serine/threonine protein phosphatase, which negatively affects nuclear import of a GUS-VirD2 NLS fusion protein.

A biochemical approach has led to the identification of a VirD2-interacting kinase, CAK2M (Bako et al, 2003). VirD2 interacts with and is phosphorylated by CAK2M in vivo; CAK2M might correspond to the kinase that regulates nuclear import of VirD2 (Gelvin, 2000). Another substrate of CAK2M is RNA polymerase II large subunit (RNApolII CTD), a factor that is responsible for recruiting TATA-box-binding proteins (TBP) to actively transcribed regions. Comparative sequence analysis of insert junctions and target sites suggested a preferential integration of the T-DNA into promoters of transcribed chromatin domains (Koncz et al, 1989; Mayerhofer et al, 1991; Brunaud et al, 2002; Szabados et al, 2002). Moreover, experiments in Agrobacterium-transformed Arabidopsis cells revealed an association of VirD2 with TBP, one of the most conserved nuclear proteins in eukaryotic cells (Nikolov et al, 1992) in transformed Arabidopsis cells (Bako et al, 2003). A hypothetical scenario suggested by the authors is that TBP or CAK2M may target VirD2 to the CTD, thereby controlling T-DNA integration. However, an alternative explanation for TBP–VirD2 interaction might have to be found. A genome-wide analysis of T-DNA-integration sites in Arabidopsis performed under non-selective conditions does not support the earlier concept of preferential T-DNA integration into transcriptionally active regions. Instead, Kim et al (2007) found that T-DNA integration occurs rather randomly and that the earlier reported enrichment of such integration sites in gene-rich or transcriptionally active regions of chromatin is due to the selection pressure applied for recovery of T-DNA insertions. However, the results leading to non-selected T-DNA-integration events could not be verified, as sequence analysis was presented only and transgenic plants could not be recovered because of the special experimental design. O'Malley et al (2007) developed a novel PCR-based method for high throughput sequencing the T-DNA/genomic DNA junction of 150 000 T-DNA insertional mutants. The analysis of this library should provide detailed information on the sequence requirements for T-DNA integration at a large scale.

Two VirE2-interacting proteins, VIP1 and VIP2, have been isolated and characterised. Most data on VIP1 originate from experiments in tobacco cells (Tzfira et al, 2001, 2002). Elevated levels of VIP1 enhance Agrobacterium susceptibility and transformation efficiency, as shown by larger calli formed in infected VIP1 overexpressing plants (Tzfira et al, 2002). Further interaction experiments led to the suggestion that VIP1 functions as a bridge between VirE2 and the plant importin α-1 (IMPa-1), thereby facilitating nuclear import of VirE2 and its associated T-DNA (Tzfira et al, 2002).

Nuclear transport of the T-DNA complex

VIP1 functions in the shuttling of the T-DNA complex into the host cell nucleus, but this function can be partially complemented for by Agrobacterium VirE3, which, similar to VIP1, was found to be capable of binding to VirE2 and IMPa-1 (Lacroix et al, 2005). In vitro, VIP1 forms ternary complexes with VirE2 and ssDNA (Tzfira et al, 2001). Analysis of an Arabidopsis vip1-1 mutant, which produces a truncated VIP1 protein, revealed that the C-terminal region of the protein is required for stable transformation, but dispensable for transient transformation (Li et al, 2005a). VIP1 is a mobile protein, which undergoes cytoplasmic-nuclear trafficking in a stress-dependent manner (Djamei et al, 2007). The other VirE2-interacting protein, VIP2, shows nuclear localisation (Tian et al, 2004). The most characteristic feature of VIP2 is an NOT domain (negative on TATA-less), rendering VIP2 a putative transcriptional repressor protein (Anand et al, 2007). Virus-induced gene silencing of VIP2 in Nicotiana benthamiana and characterisation of the Arabidopsis vip2 mutant revealed that VIP2 is required for stable, but not for transient, transformation. As shown by Y2H and bimolecular fluorescence complementation (BiFC) studies, VIP2 does not only interact with VirE2, but also with VIP1 (Anand et al, 2007). Microarray analysis revealed a major impairment of the transcriptional response of vip2 to Agrobacterium in comparison with wild-type plants. Moreover, as many as 52 histone/histone-associated genes are constitutively repressed in vip2 mutants (Anand et al, 2007). Together, these observations prompted the authors to suggest that the recalcitrancy of vip2 mutants to Agrobacterium infection and the decreased transformation efficiency are due to impaired Agrobacterium-reponsive gene induction and constitutive histone gene repression.

Data on some aspects of VirE2 are still somewhat controversial. Although Ballas and Citovsky (1997) reported specific interaction of VirE2 with Arabidopsis importin IMPa-1 and nuclear localisation of VirE2, other studies showed a predominant cytoplasmic VirE2 localisation (Bhattacharjee et al, 2008; Grange et al, 2008) and interaction of VirE2 with several importin isoforms in planta (Bhattacharjee et al, 2008). Moreover, impa-4, but not other importin mutants, is recalcitrant to transformation. This deficiency can be overcome through ectopic overexpression of heterologous importin isoforms (Bhattacharjee et al, 2008).

A possible explanation for the differences in reported VirE2 localisation might be found in the stress-dependent subcellular translocation of the VirE2-interacting protein VIP1. Stress-triggered phosphorylation of VIP1 mediates VIP1 nuclear localisation and virulence presumably by directing VirE2 to the nucleus (Djamei et al, 2007). Thus, under stress conditions such as those occurring in cell bombardment, phosphorylated VIP1 may pull the otherwise cytoplasmic VirE2 into the nucleus.

VIP1 interacts with VirE2 and IMPa-1. Tzfira et al (2002) found interaction between VIP1 and IMPa-1, but not between VirE2 with IMPa-1. They, therefore, suggested that VIP1 may serve as an adaptor molecule to facilitate the import of VirE2-bound T-strands into the nucleus. Bhattacharjee et al (2008), however, found that VirE2 can directly bind all tested importin to isoforms (IMPa-1, -2, -3, -4, -7, and -9) in Y2H and BiFC interaction studies. Combining own results and those of Tzfira et al (2002), Bhattacharjee et al (2008) proposed that VirE2 may use several cellular mechanisms for nuclear import, thereby creating additional opportunities for T-complex entry into the nucleus. This explanation seems plausible, particularly when considering that MPK3, the VIP1-nuclear-targeting kinase, is only transiently (5–15 min) activated on Agrobacterium contact (Djamei et al, 2007). VIP1-independent nuclear translocation of VirE2 may, therefore, be particularly relevant for securing T-DNA-complex entry over a prolonged period. Moreover, as VIP1, irrespective of its phosphorylation status, can bind IMPa-1 (Djamei and Pitzschke, unpublished) and VirE2 (Lacroix et al, 2008), it may still assist VirE2 nucelar translocation indirectly by guiding VirE2 to the nuclear periphery. Direct VirE2–IMPa-4 interaction would then accomplish nuclear VirE2 import. This assumption is in accordance with the results from Lee et al (2008). Particle bombardment in onion cells revealed VirE2–IMPa-1 protein complexes around the nucleus, but VirE2–IMPa-4 complexes exclusively within the nucleus.

Functions of plant proteins in T-DNA integration

As T-DNA must interact with chromatin to integrate into plant chromosomal DNA, it is likely that altering chromatin conformation will affect T-DNA integration. Forward and reverse genetic approaches have been carried out to determine which chromatin proteins are important for transformation (Zhu et al, 2003; Crane and Gelvin, 2007). In this way, mutants were identified in or near various histone genes, histone acetyltransferase genes, histone deacetylase genes as rat mutants. Moreover, 340 stable Arabidopsis RNAi mutant lines were screened for rat phenotypes. These lines comprised 109 chromatin genes of 15 gene families, including bromodomain and chromodomain proteins, chromatin remodelling complexes, DNA methyltransferases, global transcription factors, histone acetyltransferases, histone deacetylases, histone H1, methyl-binding-domain proteins, MAR-binding filament-like proteins, nucleosome assembly factors, and SET-domain proteins. Silencing of 24 chromatin genes reproducibly resulted in some level of decreased transformation susceptibility. As T-DNA integrates into the plant genome by illegitimate recombination (Mayerhofer et al, 1991), plants deficient in DNA repair and recombination may be deficient in T-DNA integration. Such DNA metabolism mutants are likely to be hypersensitive to DNA-damaging agents such as UV and radiation and DNA-damaging drugs such as bleomycin. Sonti et al (1995) investigated a number of radiation-sensitive Arabidopsis mutants for transient and stable Agrobacterium-mediated transformation. Among these, uvh1 and rad5 mutants seemed to be resistant to stable, but not to transient transformation, as assessed by formation of kanamycin-resistant calli. However, an in-depth analysis by Nam et al (1998) confirmed stable transformation deficiency only for rad5, but not uvh1 mutants. Although the latter did form less calli on kanamycin selection medium, tumour growth on non-selective medium as well as stable phosphinotricin resistance were similar to wild type. Furthermore, results from these authors also suggest RAD5 to be involved in some step before T-DNA integration, such as T-DNA transfer or nuclear targeting (Nam et al, 1998).

In addition to rad5, some radiation-sensitive Arabidopsis ecotypes are also recalcitrant to Agrobacterium-mediated transformation. By examining almost 40 Arabidopsis ecotypes for susceptibility to root transformation by Agrobacterium, ecotype UE-1 was found to be both slightly radiation-hypersensitive and transformation-deficient (Nam et al, 1997). Further testing of the rat mutants revealed that 5 of the initial 21 rat mutants were integration deficient, as indicated by high transient, but low stable transformation efficiency. One of these mutants, rat5, contains an insertion in the 3′ untranslated region of a histone H2A gene (Mysore et al, 2000). Although highly recalcitrant to stable transformation by root inoculation, rat5 is efficiently transformed by flower vacuum infiltration. These results suggest that some factor(s) required for efficient transformation are present in the female gametophyte, but absent in root somatic tissue. As rat5 plants can be complemented to transformation proficiency with the wild-type RAT5 histone H2A gene, the rat5 mutant is haplo-insufficient (dosage-dependent). The function of histones in T-DNA transformation is further emphasised in a recent study by Tenea et al (2009), who tested the effect of overexpression of several histones on Arabidopsis transformation and transgene expression. After transfection, transgene DNA was found to accumulate more rapidly in histone HTA1-overexpressing plants. The authors proposed that enhanced Agrobacterium-mediated transformation through histone overexpression is due to the protection of incoming transgene DNA during the initial stages of transformation. The main mechanism by which histones confer susceptibility seems to be conserved, as overexpression of Arabidopsis HTA1 can not only enhance transformation efficiency in Arabidopsis (Yi et al, 2002, 2006), but also in rice (Zheng et al, 2009).

Mutations in fas1 and fas2, encoding two subunits of the chromatin assembly factor CAF1, show greatly increased frequencies of homologous recombination and T-DNA integration (Endo et al, 2006). Studies on the Arabidopsis protein KU80 further stress the active participation of the host's repair machinery in T-DNA integration. KU80, an important protein in the non-homologous end-joining complex (Jeggo et al, 1999), directly binds to double-stranded T-DNA intermediates (Li et al, 2005b), which are rapidly converted from T-strands early in the infection process and are essential intermediates of T-DNA integration (reviewed in Tzfira et al, 2004a, 2004b). Ku80 mutants are defective in T-DNA integration, but not in transient T-DNA expression, whereas KU80 overexpression results in increased susceptibility to Agrobacterium infection and increased resistance to DNA-damaging agents (Li et al, 2005b). The function of KU80 during the transformation of germ-line cells, however, is still debatable. Ku80 has been reported to be both required (Friesner and Britt, 2003) and dispensable (Gallego et al, 2003) for T-DNA integration.

Agrobacterium and the plant innate immune response

Plant factors involved in Agrobacterium perception

Agrobacterium-attacked plants do not simply come to terms with their fate. Similar to other pathogens, Agrobacterium is sensed as an invader and triggers the ‘innate immune response', characterised by the expression of defence genes and accumulation of reactive oxygen species (reviewed in Pitzschke et al, 2009a). This reaction is achieved through the perception of pathogen-associated molecular patterns (PAMPs) by specific receptors. Although several PAMPs have been isolated, only few receptors are yet identified. Putative plant receptors for Agrobacterium include a vitronectin-like protein (Wagner and Matthysse, 1992), a rhicadhesin-binding protein (Swart et al, 1994), and several VirB2-interacting proteins (Hwang and Gelvin, 2004; reviewed in Citovsky et al, 2007). A recent in-depth study by Clauce-Coupel et al (2008) on vitronectins strongly indicates that this group of proteins is unlikely to act as receptors for site-specific Agrobacterium attachment.

The most intensively investigated PAMP is flagellin, a highly conserved bacterial protein. In Arabidopsis, it is perceived by the receptor protein FLS2, a leucine-rich repeat receptor such as kinase (LRR-RLK). On perception of flagellin or its derived highly conserved 22 amino-acid peptide, flg22, FLS2 becomes activated and initiates a phospho-relay-based signal transduction through the MAPK cascade MEKK1-MKK1/2-MPK4 (Qiu et al, 2008; Pitzschke et al, 2009a, 2009b). Subsequently, the MPK4-activated transcription factor WRKY33 contributes to the defence-related transcriptional re-programming (Petersen et al, 2008).

The flagellin proteins of Agrobacterium and of symbiotic bacteria (rhizobia) are distinct from those of most other microbes in that they are not recognised and do not trigger a defence response, implying that other PAMP-receptor pairs are responsible for recognition of these organisms. Indeed, a prominent Agrobacterium PAMP, the elongation factor EF-Tu, has been identified. Although highly conserved in all prokaryotes, Agrobacterium EF-Tu is fully active as an elicitor (Kunze et al, 2004). Interestingly, despite their chemical dissimilarity, flg22 and EF-Tu share several characteristics. Both PAMPs inhibit seedling growth and activate a common set of signalling events and defence responses, while acting with no apparent synergy (Zipfel et al, 2006). These responses include MAPK activation, alkanisation of the medium, and an oxidative burst. Moreover, a microarray analysis of the response of Arabidopsis to flg22 and the EF-Tu-derived peptide elf18 revealed a clear correlation of differential gene expression, whereas no apparent flg22 or elf18-specific subsets of genes were identified. A surprisingly high number of RLK-encoding genes (100 of 610) were found to be rapidly induced by these PAMPs (Zipfel et al, 2006).

A targeted reverse genetics approach has led to the identification of a receptor kinase essential for EF-Tu perception, EFR1. EFR1, similar to FLS2, is a member of the LRR-RLK protein family. The important function of LRR-RLKs in plant-microbe sensing has already become evident for root nodule symbiosis. Lotus japonicus mutants affected in the LRR-RLK SYMRK (symbiosis receptor kinase) fail to engage in symbiosis (Stracke et al, 2002). Likewise, Arabidopsis mutants lacking fls2 show no response to flg22 treatment (Asai et al, 2002). The efr mutants are insensitive to elf18, while showing a normal flg22 response (Zipfel et al, 2006) and an otherwise normal phenotype. The striking feature of efr mutants is their high susceptibility to Agrobacterium infection, as shown by the enhanced expression of a T-DNA-harboured reporter gene (GUS) on transient transformation of seedlings. On pre-treatment with flg22, an increase of receptor-binding sites for EF-Tu was observed. Co-injection of flg22 resulted in abolishment of GUS expression in wild type and efr mutants, whereas co-injection of elf18 only abolished GUS expression in wild type, but not in efr. N. benthamiana, an easily transformed plant species, does not have an EF-Tu recognition system. However, transgenic N. benthamiana plants expressing Arabidopsis EFR are capable of inducing elf18-triggered defence responses. Together, these observations lead to the conclusion that EFR-mediated EF-Tu perception restricts Agrobacterium-mediated transformation (Zipfel et al, 2006).

A recent study sheds light on the molecular mechanism that links EFR receptor activation to intracellular signal transduction and stresses differences between flg22- and elf18-triggered limitation of Agrobacterium transformation (Ishikawa, 2009). Arabidopsis mutants affected in the G-protein b-subunit (agb1-2) show significantly reduced ROS production on flg22 or elf18 treatment, whereas stress-triggered MAPK activation (analysed by immunoblotting with an antibody recognising active MPK3 and MPK6) is apparently not affected. Moreover, these mutants are impaired in the elf18- but not in flg22-triggered immunity against Agrobacterium. Therefore, a function of AGB1 as positive regulator integrating flagellin and EF-Tu perception into ROS production, and specifically in EF-Tu signalling to limit Agrobacterium transformation, has been proposed.

Gene expression re-programming in response to Agrobacterium

Agrobacterium attack leads to a major re-programming of gene expression in plants. Already in the pre-microarray-era, large-scale expression analyses (cDNA-AFLP) have revealed that many Agrobacterium-induced genes are related to plant defence and to general stress responses (Ditt et al, 2001). Using different Agrobacterium strains, Veena et al (2003) could show that the transfer of T-DNA and Vir proteins can modulate the expression of plant genes in tobacco cell culture. The authors concluded that T-DNA and Vir protein transfer acts as suppressors of the defence response. Later, it was found that an attachment-deficient Agrobacterium mutant hyper-induced defence-related genes in Ageratum conyzoides cell culture. Interestingly, also non-pathogenic Escherichia coli triggered such hyper-induction (Ditt et al, 2006), and it was concluded that Agrobacterium can dampen plant defence in an attachment-dependent manner (Ditt et al, 2006). The authors also observed a strong variability in transformation efficiency between individual experimental series, and further analyses revealed that enhanced basal defence gene expression correlates with resistance to Agrobacterium transformation. These findings correlate with the observation that co-injection of plants with Agrobacterium plus flg22 or elf18 elicitors abolishes transformation (Zipfel et al, 2006). One factor that may explain the reported negative correlation between transformation efficiency and stress status may be salicylic acid (SA). This plant hormone is an important signal in regulating the plant response to pathogens. It accumulates in local and systemic tissues of stress-exposed plants and induces expression of pathogenesis-related genes. A study by Yuan et al (2007) now implicates SA in the repression of the Vir regulon, the attenuation of the function of the VirA kinase as well as in the degradation of an Agrobacterium quormone. Accordingly, plant mutants overproducing SA were recalcitrant to tumour formation. These finding are further supported by Anand et al (2008). In summary, from the bacterial point of view, for maximal transformation efficiency, a minimally stressed state of host cells seems desirable. Reciprocally, plants may evade Agrobacterium infection if their defence system is in an alerted state.

A kinetic study on the response of tobacco BY2 cell cultures to various Agrobacterium strains, including T-DNA and Vir protein transfer-incompetent strains, helped to dissect the stress versus transformation efficiency ambivalence and allowed to distinguish general transcriptional responses resulting from attachment or proximity of Agrobacterium near BY2 cells from transformation-specific responses induced by the transfer of T-DNA and/or Vir proteins into plant cells (Veena et al, 2003). Agrobacterium triggers an early (3–12 h) induction of stress genes, whose expression is subsequently repressed by transfer-competent strains. Concomitant with the decline of this general defence, T-DNA and Vir protein transfer occurs. In contrast, Vir transfer-deficient strains trigger a second wave (after 24 h) of defence gene expression and fail to transfer T-DNA. Through their cDNA macroarray, Veena Jiang et al (2003) also revealed that several histone genes were more strongly induced by transfer-competent strains than transfer-deficient strains. Together with the observed constitutive repression of numerous histone genes in the vip2 mutant, which are affected in stable, but not in transient transformation (Anand et al, 2007), these findings emphasise the importance of histones in Agrobacterium transformation and indicate that an elevated pool of histones is required for facilitating T-DNA integration into the host genome.

The mechanisms of transcriptional re-programming of Agrobacterium-infected tissue and the extent to which bacterial factors contribute to this re-programming are still mostly unknown. On the basis of the findings that VirE3 has trans-activating activity in yeast, localises to plant nuclei, and that VirE3 can bind pBrp, a general plant-specific transcription factor, a function of VirE3 as potential plant transcriptional activator mediating the expression of tumour development-specific genes was proposed (Garcia-Rodriguez et al, 2006).

Defence versus transformation—how to evade the innate immune response

From the above sections, it becomes apparent that a major barrier for Agrobacterium infection is the defence response triggered in the host cell. Once plant defence can be blocked, reduced, or circumvented, much higher transformation efficiencies can be achieved. Even more, in at least one aspect, Agrobacterium have learnt to turn the tables in that they can even benefit from being recognised as a pathogen. Similar to numerous other microbes, Agrobacterium triggers the activation of MAPKs, primarily MPK3, MPK4, and MPK6 (Djamei et al, 2007). Y2H experiments, in vivo interaction, and kinase assays have shown that VIP1, the VirE2-interacting protein 1, interacts with and becomes specifically phosphorylated by MPK3 (Djamei et al, 2007). Stress-triggered phosphorylation of VIP1 results in the rapid translocation of this protein from the cytoplasm to the nucleus. In contrast to VirD2, which passes into the host nucleus on its own, VirE2 needs the assistance of both importin α (Bhattacharjee et al, 2008, see above) and the host protein VIP1. Agrobacterium thus does not only abuse VIP1 as such for delivering the VirE2/T-DNA into the host cell nucleus, but also actively manipulates the subcellular localisation of this plant protein. As this was not enough, once nuclear transfer of the T-DNA complex has been accomplished, T-DNA integration into the plant genome is likely to be achieved through abuse of yet another manipulation of the host cell machinery. Tzfira et al (2004a, 2004b) suggest that nuclear VirF protein may mediate the degradation of the T-DNA complex through the plant proteasome, resulting in the release of the T-DNA, which subsequently can integrate into the host genome, the final step of stable transformation.

As not only Agrobacterium, but also flg22, triggers MPK3 activation and rapid nuclear accumulation of VIP1 (Djamei et al, 2007), the question arises which function the stress-dependent cytoplasmic-nuclear translocation of VIP1 has in the plant response to other stresses. We have recently shown that VIP1 is a functional bZIP transcription factor, which binds to a novel DNA regulatory motif, VIP1 response element (VRE) (Pitzschke et al, 2009c). VREs are overrepresented in the promoter regions of stress-responsive genes. VIP1 binds in vivo to the promoter of a stress-responsive transcription factor, MYB44, under conditions that activate the MPK3 pathway. The mpk3 mutants are impaired in stress-triggered induction of VIP1 target genes, whereas VIP1 overexpressing plants show constitutively elevated transcript levels. From these observations, a function of VIP1 as a mediator of MPK3-mediated stress gene modulation was derived. How does the function of VIP1 as activator of the defence responses correlate with the finding that VIP1-overexpression results in enhanced transformation efficiency (Tzfira et al, 2002)? A plausible explanation is that the higher transformation rate is achieved through a more efficient nuclear transfer of the T-DNA complex, as more VirE2-T-DNA molecules can be ‘piggy-packed' by VIP1. The degradation of the T-DNA complex and VIP1 through the action of VirF and the plant proteasome serves both the release of the T-DNA and the prevention of VIP1-induced defence gene expression. According to this model, not VirF, VirD2, or the fidelity of the plant proteasomal machinery, but VIP1 phosphorylation/localisation and the presumably directly correlated nuclear import of VirE2/T-DNA might be the limiting factors in transformation.

Owing to its dual function as a stress-responsive transcription factor on the one hand and as T-DNA/VirE2 shuttle on the other hand, VIP1 may be one of the factors that ‘tips the scales', that is it decides between successful transformation and failure of transformation because of elevated basal stress levels. As MPK3 is very sensitive to numerous stresses, its target protein, VIP1, can potentially be phosphorylated, translocate to the nucleus and induce stress gene expression equally easily. The battle between Agrobacterium and plant, therefore, is to ‘compete' for one of the two VIP1 functions.

Conclusions/summary

The transformation of plants by Agrobacterium is a complex process that involves multiple steps and the concerted action of both microbial and host factors. As evidenced by a still increasing number of reports on novel plant–Agrobacterium protein interactions, much is still to be learnt to understand the entire transformation process in detail. The main recent progress includes the characterisation of Vir-interacting proteins in vivo, the function of histones in stable integration and the elucidation of surprisingly smart strategies used by Agrobacterium to circumvent or even abuse the plant defence system. Clearly, the in-depth study and molecular analysis of the plant–Agrobacterium interaction will not only add to our understanding of pathogen strategies for host infection, but harbour also the potential to render plants transformable that are otherwise recalcitrant to Agrobacterium transformation and to speed up the progress in generating stress-resistant plants.