Pathways of DNA Transfer to Plants from Agrobacterium tumefaciens and Related Bacterial Species

Abstract

Genetic transformation of host plants by Agrobacterium tumefaciens and related species represents a unique model for natural horizontal gene transfer. Almost five decades of studying the molecular interactions between Agrobacterium and its host cells have yielded countless fundamental insights into bacterial and plant biology, even though several steps of the DNA transfer process remain poorly understood. Agrobacterium spp. may utilize different pathways for transferring DNA, which likely reflects the very wide host range of Agrobacterium. Furthermore, closely related bacterial species, such as rhizobia, are able to transfer DNA to host plant cells when they are provided with Agrobacterium DNA transfer machinery and T-DNA. Homologs of Agrobacterium virulence genes are found in many bacterial genomes, but only one non-Agrobacterium bacterial strain, Rhizobium etli CFN42, harbors a complete set of virulence genes and can mediate plant genetic transformation when carrying a T-DNA-containing plasmid.

INTRODUCTION

Agrobacterium tumefaciens is often described as a natural genetic engineer, equipped to horizontally transfer bacterial genes and genetically transform plant cells (45). Indeed, transfer of genetic material from A. tumefaciens and related species to their host plants represents the first known case of active horizontal gene transfer from prokarya to eukarya. The main factors conferring this ability to A. tumefaciens are located on the large Ti (tumor-inducing) plasmid, which contains a region with the vir (virulence) genes encoding most of the proteins required to mediate the DNA transfer and the T-DNA (transferred DNA) itself. The T-DNA sequences naturally transferred by several Agrobacterium spp. contain two types of genes under the control of promoters compatible with expression in eukaryotic cells. The first set of genes (oncogenes) encodes proteins that affect the biosynthesis of or plant cell response to growth regulators (auxins and cytokinins) and induce uncontrolled cellular division, resulting in tissue proliferation and formation of neoplastic growths (crown galls). The second set of genes encodes enzymes involved in the synthesis of small molecules (opines) composed of an amino acid and an organic acid or a carbohydrate, which can be used by Agrobacterium cells as a source of carbon and nitrogen (36). The ability of Agrobacterium to transfer DNA, either for transient expression or stable genetic transformation, is widely used as a tool in research and biotechnology (11). Although the mechanism of T-DNA transfer and integration has been studied extensively from the early 1970s (91), there are still many aspects of the process that are not completely understood. Recent discoveries indicate that there is a potentially important variability in the pathways used by Agrobacterium strains to deliver the T-DNA to the plant genome, which may also reflect adaptation to different hosts. Furthermore, the presence of homologs of the Agrobacterium genes required for virulence in related bacterial species suggests that DNA transfer to eukaryotes might be more widespread among bacteria outside the Agrobacterium genus.

MECHANISM OF AGROBACTERIUM-MEDIATED DNA TRANSFER AND INTEGRATION

A. tumefaciens interactions with host plant cells represent a reference model for the transfer of DNA from bacteria to eukaryotic cell. The molecular mechanism of T-DNA transfer from Agrobacterium to its host plant cell genome has been reviewed in detail in several articles (45, 65); here, we provide a brief account of the current state of knowledge of this system. Several Agrobacterium species are known to genetically transform plants, resulting in distinct plant diseases (36); however, most of the research on the mechanism of DNA transfer has focused on a few strains of A. tumefaciens (i.e., the nopaline C58 and octopine A6 strains), which are presented in this section. For the purpose of this review, we divided the process of transfer of DNA from A. tumefaciens to its host cell genome into four steps (see Figure 1 for an overview).

Figure 1.

Schematic representation of the main steps of T-DNA transfer from Agrobacterium to the plant cell genome. Circled numbers represent the major steps in the pathway. ❶ Plant-derived and environmental signals activate the bacterial virulence system, resulting in the induction of vir (virulence) gene expression and the generation of the single-stranded T-DNA. ❷ The T-DNA covalently attaches to VirD2, and several vir-encoded effector proteins (VirD5, VirE2, VirE3, and VirF) are exported out of the bacterial cell via the VirB/VirD4 T4SS. ❸ T-DNA and effector proteins enter in the plant cell and are targeted into the nucleus. ❹ The T-DNA is processed in the nucleus and integrated into the plant cell chromosomal DNA. Abbreviations: AS, acetosyringone; Chv, chromosomal virulence protein; DSB, double-strand break; dsT-DNA, double-stranded transferred DNA; Exo, exocellular; ssT-DNA, single-stranded transferred DNA; T-DNA, transferred DNA; T4SS, type IV secretion system; Ti plasmid, tumor-inducing plasmid; vir, virulence gene region; Vir, virulence protein; VBP, VirD2 binding protein; VIP1, VirE2 interacting protein 1; VIP2, VirE2 interacting protein 2.

Step 1: Virulence Induction and Generation of Single-Stranded T-DNA

Upon induction by plant-emitted and environmental signals, the expression of Agrobacterium vir genes is activated, resulting in the synthesis of proteins required for DNA transfer and the generation of the single-stranded T-DNA.

Signal recognition and integration.

Agrobacterium cells can detect several plant-emitted signals and respond by modifying their lifestyle and adjusting the transcription level of their vir genes. The key regulator of vir gene expression is the two-component receptor system encoded by the virA and virG genes (120). The first-identified and major plant-produced molecule involved in vir activation is acetosyringone (AS; 3,5-dimethoxyacetophenone), a phenolic compound often found in plant exudates; AS activates the VirA/VirG two-component system, resulting in the induction of the vir gene expression (17, 118). Although virA and virG are expressed at low levels without induction, they are themselves inducible by AS (141), and VirA/VirG activation results in rapid and strong induction of all the vir genes. Reducing sugar monomers, such as d-glucose and d-galactose, can both increase the sensitivity of the VirA/VirG system to phenolics and elevate the saturating concentration of phenolics for virulence activation (21, 116). These monosaccharides bind to the periplasmic chromosome-encoded protein ChvE, which then interacts with the VirA periplasmic domain to enhance its vir gene–inducing activity (21, 115). VirA is also activated by low pH (between pH 5 and 6) either directly (88) or via its interaction with ChvE (41). In addition, low pH combined with low phosphate concentration activates a different two-component regulatory system (ChvG/ChvI), which results in increased expression of virG (24). The response to low pH relies on the periplasmic ChvG inhibitor ExoR, which is degraded under acidic conditions (52).

Activating vir gene expression.

Upon activation of the VirA/VirG system, VirG is phosphorylated and induces vir gene expression by direct binding to a 10- to 12-bp sequence (vir box) (121). One or several vir boxes are found in promoter regions of each of the vir operons, usually located between 50 and 200 bp upstream of the translation initiation of the first gene of each vir operon. Expression of vir promoters is observed a few hours after the initial induction by phenolics, and it generally reaches a plateau after 12 to 24 hours (141). In addition to several signaling pathways that converge to generate the activated (phosphorylated) VirG, there are other factors affecting vir gene expression. For example, the virC and virD operons are repressed by Ros, a chromosome-encoded transcriptional regulator (30). More recently, small RNAs regulating some of the vir genes were identified (33).

Turning off vir gene expression.

After successful infection of a host plant by Agrobacterium cells, i.e., when the virulence system is no longer needed, the virulence system should be shut off. Indeed, the energy cost of virulence induction is high, and the activation of virulence results in a decrease in the population growth rate (103). Among potential factors negatively affecting vir gene expression, IAA (indole acetic acid) interferes with vir gene induction, probably as a competitive inhibitor of AS binding to VirA (79). The role of IAA, synthesized at high levels during the development of Agrobacterium-induced crown gall tumors, could be to turn off virulence induction as well as to inhibit additional transformation by competing bacterial strains or by the initially infecting strain. In addition, the change of lifestyle of the bacteria between free-living, nonpathogenic bacteria and pathogenic cells attached to the plant cell surface and embedded in the biofilm matrix may also affect the vir gene expression.

Generation of single-stranded T-DNA.

The T-DNA is a segment of the Ti plasmid, delimited by two short (24–25 bp) direct repeat sequences, the left border (LB) and right border (RB) (102, 144). T-DNA is mobilized from the Ti plasmid and transferred into the host cell as a single-stranded DNA (ssDNA) intermediate, termed the T-strand (119). Two essential proteins for T-DNA processing are VirD1 and VirD2. VirD2 is an endonuclease (2, 147), which, in association with the VirD1 DNA topoisomerase (47), mediates the mobilization of the transferable T-DNA from the Ti plasmid via a strand replacement mechanism. Importantly, the T-DNA borders are the only sequences required for recognition by VirD2/VirD1, and, thus, the sequences between these borders may be modified at will. Two other Vir proteins, VirC1 and VirC2, have been shown to increase the number of T-strand molecules, most likely by binding to sequences, termed overdrive, close to the T-DNA borders (34). At the end of the process, VirD2 remains covalently linked to the 5′ end of the T-strand (148), forming an immature T-complex.

Step 2: Export of the T-DNA and Effector Proteins and Cell-to-Cell Interactions

Macromolecules are translocated across the bacterial membranes via a T4SS (type IV secretion system) composed of the 11 proteins encoded by the virB operon and VirD4 by a mechanism closely resembling plasmid transfer during bacterial conjugation. T4SSs are molecular complexes that mediate the transport of proteins and nucleoprotein complexes, usually comprising an ssDNA with a protein molecule at its 5′ end, through the membranes and cell walls of gram-negative bacteria (27).

Targeting of exported macromolecules to type IV secretion system.

Interactions with bacterial factors are likely required to mediate targeting of the exported substrates—i.e., the VirD2–T-strand complex and the effector proteins VirD5, VirE2, VirE3, and VirF—to the VirB/D4 T4SS. Protein export from Agrobacterium may occur independently of DNA export, and it depends on the presence of an arginine-rich export signal found in all exported Vir proteins (133, 134). Several factors have been suggested to mediate targeting of the VirD2–T-strand complex and individual translocated proteins to the T4SS machinery. For example, VirC1 and VirC2 may assist targeting of VirD2 and the T-strand to the bacterial membrane at the cell poles where T4SS is thought to be assembled (7). VirE2 might be recruited to the cell poles via its association with the coupling protein VirD4 (8). Other bacterial factors, VBPs (VirD2 binding proteins), appear important for VirD2 recruitment to T4SS as well as for recruitment of diverse relaxase proteins in other T4SS systems (51). Indeed, VBPs conserved in Agrobacterium spp. can target VirD2 and the associated T-strand to the energizing components of the T4SS, i.e., VirD4, VirB4, and VirB11, and, thus, to the T-DNA export machinery (49, 51). Furthermore, VBPs, which do not interact with the other exported effector proteins, appear to be important for the recruitment of conjugative DNA transfer intermediates to T4SS during conjugation (51).

For the most part, the transport pathway of the VirD2–T-strand complex through bacterial membranes has been deciphered, and it comprises sequential interactions with different protein components of T4SS (23). The lumen size of the VirB2 pilus of T4SS appears sufficient to accommodate passage of ssDNA and partially unfolded proteins (62); indeed, in other bacterial species, relaxases transported through T4SS channels have been reported to unfold during transport (124).

Attachment and close-range cell-to-cell interactions.

Close-range interaction and attachment between bacterial and host cells are thought to be required for the transfer of T-DNA and effector proteins (84). Although under laboratory conditions, bacterial virulence can be induced without interaction with the host cell surface, in nature these two events are likely linked, and the induction of virulence is coincidental with a change in the bacterial cell lifestyle from free bacteria in the rhizosphere to bacteria attached at the surface of the host cell and embedded in a biofilm. Indeed, the same signals that trigger vir gene expression also induce chemotaxis. Specifically, Agrobacterium cells respond to phenolics and sugars secreted by plants by moving toward their source via chemotaxis (50); at low concentration of these molecules, chemotaxis is activated, whereas high concentrations result in virulence activation. Furthermore, in addition to its positive effect on virulence, low phosphate concentration enhances biofilm formation and cellular adhesion (143).

Analogous to other host–plant interactions, such as Rhizobium–legume symbiosis (106), the cellular interaction is believed to occur in two steps. The initial contact between bacteria and eukaryotic cells usually relies on host cell-surface receptors and represents a reversible interaction. The bacterial attachment is then stabilized via the synthesis of cellulose fibrils, and bacterial cells are embedded in a biofilm at the surface of the plant tissue. In the case of Agrobacterium, the precise role of the different factors affecting attachment during the infection process is not completely understood. The Agrobacterium T4SS components, i.e., VirB2 (pilin) and VirB5, exposed to the bacterial cell surface represent good candidates for interaction with potential host cell receptors (9). Four Arabidopsis proteins were identified to interact with VirB2 (59) and shown to affect the efficiency of the T-DNA transfer. However, it is not clear whether these VirB2-interacting proteins are involved in cellular attachment or in other steps of the DNA transfer process, such as signal transmission or passage of the T-DNA through the host-cell membrane. Interestingly, pilin homologs encoded by Agrobacterium, CtpA and PilA, appear to be involved in the early stages of Agrobacterium cell-surface attachment, although their role in virulence remains unknown (138). VirB5 localizes at the tip of the VirB2 pilus (4) and may have a dual function: (a) during T4SS biogenesis, which requires VirB5 expression in the bacterial cell, and (b) outside the bacterial cell (64), although it plays no obvious role in cellular attachment. Exocellular polysaccharides produced by Agrobacterium are important for attachment and biofilm formation. Synthesis and export of cyclic 1,2-β-d-glucan, which relies on proteins encoded by the chvA, chvB, and exoC genes, are involved in attachment and virulence (22, 32); UPPs (unipolar polysaccharides) and cellulose also may play a role in bacterial adhesion and biofilm formation (83, 143). However, plant receptors that, similarly to the plant lectins facilitating Rhizobium–plant cell recognition (53), bind these exopolysaccharides and are involved in Agrobacterium-mediated transformation have not been identified.

Step 3: Entry and Subcellular Sorting of T-DNA and Effector Proteins in the Host Cell

The entry of T-DNA and effector proteins in the host cell cytoplasm, across plant cell plasma membrane, is not completely understood; different hypotheses are presented below. Multiple interactions with host factors mediate the nuclear import of these macromolecules.

Entry of T-DNA and associated proteins into the host cell cytoplasm.

The mechanism by which the VirD2–T-strand complex and effector proteins pass through the host cell wall and plasma membrane is unknown. Although wounding of the plant tissue enhances the Agrobacterium-mediated transformation efficiency, T-DNA transfer from Agrobacterium without wounding the host plant cell has been reported (18). Several mechanisms are possible for entry through the host plasma membrane. First, similar to a mechanism proposed for bacterial conjugation, depolymerization of the VirB2 pilus may bring the bacterial outer membrane and the host cell plasma membrane together, resulting in temporary membrane fusion and allowing the transfer of cargo (20). Second, the VirB2 pilus may act as a needle via a mechanism similar to type III secretion system (T3SS)-mediated effector protein transport (96); in this scenario, macromolecular substrates pass through the pilus, and the pilus interacts with the host membrane or integral membrane proteins to allow the entry of the cargo. So far, however, no interactions with host membrane proteins or bacterial factors able to form a pore in the host membrane have been identified in the Agrobacterium–host plant cell system. Furthermore, Agrobacterium mutants unable to form pili still retain a low-level virulence, demonstrating that T-DNA transfer can occur in the absence of the VirB2 pilus (108). Third, macromolecules could be first exported into the intercellular space and then internalized by the host cell, for example, via an endocytosis-like mechanism, which might involve recognition between the exported macromolecules and a potential host receptor. Indeed, a recent study suggested that VirE2 associates with early endosomes in the host plant cell and that endocytosis inhibitors affected both VirE2 transport and transformation efficiency (76). Thus, endocytosis might be involved in the internalization of VirE2 and potentially other translocated molecules. In addition, VirE2 has been shown to form channels through artificial membranes (35); although formation of VirE2 pores has not been demonstrated in infected plant cells, such pores might mediate transport of other macromolecules through the host cell membranes.

Nuclear import.

Before integration can occur, the T-DNA, as well as translocated effector proteins with a nuclear function, must be imported into the nucleus. Efficient nuclear import via simple diffusion is unlikely for large molecules such as T-strands. Genetic transformation of plant cells using protocols that do not involve Agrobacterium implies that nuclear import of foreign DNA can occur without exogenous effector proteins, most likely using cellular DNA-binding proteins that facilitate import; such transformation techniques are considered less efficient than the Agrobacterium-mediated transformation, although it is difficult to compare efficiency between such different methods. The nuclear import step can be circumvented altogether if the transformation process occurs during cell division, when the nuclear envelope is disrupted (135). However, transient expression of T-DNA, which obviously requires its nuclear import, occurs efficiently in nondividing cells following agroinfiltration (146). Thus, active nuclear import most likely is involved in most cases of Agrobacterium-mediated genetic transformation. Generally, bacterial proteins interacting with the T-DNA are presumed to mediate its nuclear uptake via the importin alpha-mediated import pathway. First, VirD2, attached to the 5′ end of the T-DNA, interacts directly with importin alpha via its NLS sequences and is targeted to the host cell nucleus (10). VirE2, an ssDNA-binding protein, is also thought to interact with the T-DNA after its entry in the host cell cytoplasm, forming the mature T-complex (28, 44). Although the VirE2–T-strand complex has not been directly shown to form in living cells, a significant amount of data suggests that such formation occurs. First, the T-DNA integrated in absence of VirE2 displays increased truncations, suggesting that VirE2 associates with and protects the T-strand against degradation (107). VirE2 then has a strong affinity for ssDNA in vitro (26, 28), producing helical ssDNA–VirE2 filaments with well-defined structure (1). Initially, several studies demonstrated that VirE2 tagged with different markers was, at least partially, targeted to the nucleus in plant cells (29, 78, 151). Other studies showed that fusion of VirE2 with fluorescent proteins remained largely cytoplasmic (73, 113). Because of its strong homopolymerization, VirE2 tends to form aggregates when expressed ectopically in plant cells, which hinders assessment of its localization; it is also possible that only a fraction of VirE2 is directed to the nucleus but that this fraction is sufficient for VirE2 functionality in the T-strand import process. VirE2 was shown to interact with several plant proteins likely to affect its intracellular distribution and/or function: VIP1 (VirE2 interacting protein 1) (126), VIP2 (VirE2 interacting protein 2) (5), importins alpha (14), and core histones (69, 81). Furthermore, VirE2 also interacts with VirE3 (71, 77), and this interaction likely assists accumulation of VirE2 at the sites of entry into the host cell (77) and/or subsequent nuclear import of VirE2 (71). Both VirD2 and VirE2 have been shown to mediate nuclear import of short segments of ssDNA (152). Potentially, these two proteins participate in T-DNA nuclear import; VirD2, alone or with the help of VirE2, targets the T-strand to the nuclear pores, whereas VirE2 packages the T-strand and mediates its movement through host cell cytoplasm (152) and through the nuclear pore. It has also been suggested that the T-strand and its associated proteins could interact with the host cell cytoskeleton and endoplasmic reticulum during its transport to the nucleus (145).

Step 4: T-DNA Integration in the Host Chromosomal DNA

The mechanism of T-DNA integration into the host genome remains largely obscure (46). We first present the main known facts about the integration process and then incorporate them into potential integration pathways. Two main approaches have been used to characterize T-DNA integration: (a) analysis of the locations of the integrated T-DNA and its patterns of integration and (b) studies of plant and bacterial factors that may affect integration. In the first approach, early studies showed that integrated T-DNAs were preferentially located in transcriptionally active chromatin (3); however, these studies relied on analyses of transgenic plants regenerated under antibiotic selection. This made it virtually impossible to detect integration into heterochromatin, which does not support the expression of the antibiotic-resistant reporter. Indeed, a completely different result was obtained in studies performed without selection, which showed that T-DNA integrated randomly in all regions of chromatin (63), although a local bias might occur toward specific epigenetic markers (114). Nucleosomal histones have been suggested to be involved in the targeting of the T-DNA complex to the host chromatin by allowing interaction between the T-complex and the host chromatin before integration. First, histone H2A was found to be important for T-DNA integration, as an Arabidopsis mutant in this gene displayed lower transformation efficiency (90) and, later, other histones were shown to increase T-DNA integration (122). Interaction between VIP1 and different histones was also demonstrated (81), and VirD2 was found to interact with histones (142).

Unlike many integrating viruses, Agrobacterium does not encode a dedicated integrase among its effector proteins. Although early studies suggested that VirD2 might act as an integrase (98, 123); integration was later shown to be mediated by host factors (153). Yet it cannot be excluded that VirD2 or another Agrobacterium translocated effector protein facilitates T-DNA integration by interacting with the host factors that directly mediate integration. Furthermore, the analysis of integration in various host species, particularly with different yeast mutants, has shown that the integration of T-DNA relies mostly on the host cell pathways. Several studies have suggested a role for the host cell DNA repair pathways in T-DNA integration, and double-strand breaks (DSBs) were shown to represent preferred sites for T-DNA integration (25, 110, 125). Measuring T-DNA integration rates in Arabidopsis mutants in different genes encoding DNA repair proteins yielded inconclusive results (see below). Using a combination of these two approaches, it was recently reported that an Arabidopsis mutant in the DNA polymerase theta was deficient in T-DNA integration, suggesting an integration mechanism based on microhomologies (132). DNA polymerase theta was first identified as a suppressor of genome instability, and it is known for its role in genomic DNA ligation in the microhomology-mediated end-joining [MMEJ; or alternative end-joining (alt-EJ)] DSB repair pathway (16). However, this mechanism does not explain the integration of double-stranded T-DNA and recombination between several T-DNAs in different orientation, suggesting that several concurrent integration pathways may underlie transformation events (46).

Potential pathways for T-DNA integration.

T-DNA enters the nucleus as a segment of ssDNA; it may either be converted to a double-stranded DNA (dsDNA) before integration, most likely into a DSB in the genomic DNA, or anneal partially to the host genomic DNA via microhomologies before synthesis of its second strand and ligation. There is direct proof that T-DNA can integrate into DSBs as a dsDNA; by introducing a rare cutting dsDNA endonuclease site in both the T-DNA and the host genome and transiently expressing this enzyme, precise reconstruction of the original restriction site at junctions between T-DNA and host DNA was observed (25, 125). Interestingly, it has been shown that the formation of circularized T-DNA (T-circles) occurs after T-DNA transfer into the plant cell (117), although there is no indication that these T-circles act as a substrate for integration. The observation of microhomologies at the junction of some integration sites suggests that the second pathway is also possible, and recent implication of DNA polymerase theta in T-DNA integration (132) shows involvement of this pathway in integration.

Experiments using T-DNA transfer into yeast (Saccharomyces cerevisiae)—this model host allows the use of numerous viable mutants in different DSB repair pathways—demonstrated that the integration pathway depends mostly on the host mechanisms. Taking advantage of the ability of yeast cells to support DNA integration via either homologous recombination (HR) or nonhomologous end-joining (NHEJ), depending on the presence of homologous sequences in the target genome and the T-DNA, T-DNA integration was assessed in mutants impaired in these pathways. Disruption of Rad52 or Rad51 resulted in integration only via NHEJ, whereas in the absence of Ku70 or Mre11 expression, only integration via HR was observed (130, 131). In plants, HR occurs only at extremely low rates (48, 85), and NHEJ is believed to be the main pathway for foreign DNA integration. However, studies using Arabidopsis mutants in the NHEJ pathways yielded conflicting results. AtLig4 and AtKu80 were found to be dispensable for T-DNA integration in one study (40) but were required in two other studies (39, 75). More recently, a systematic survey of Arabidopsis mutants impaired in different genes involved in the known pathways of NHEJ reported that T-DNA integration efficiency was not reduced in any of these lines, and it was even increased in some of the mutants (100). In rice, however, reduced rates of overall integration were observed in plant lines with downregulated Ku70, Ku80, and Lig4 (93). Because of high levels of redundancy between DNA repair pathways, it is difficult to prove their specific involvement in T-DNA integration. Yet when several NHEJ pathways were mutated in Arabidopsis, the resulting viable plants supported only very low levels of T-DNA integration (89). The involvement of DSB repair pathways in T-DNA integration also appears to be complex and may vary at different time points during the infection process. For example, targeting of the incoming T-DNA to open DSBs may be achieved in a less efficient repair pathway, but subsequent ligation of the T-DNA into the DSB may require efficient DSB repair. Furthermore, the host NHEJ machinery may be manipulated by Agrobacterium effector proteins; for example, VirE2 interacted with XRCC4, a component of the NHEJ pathway, and potentially prevented DSB repair, allowing the T-DNA to be targeted to an available DSB site (129). Other host nuclear proteins, such as the transcriptional regulator VIP2 (VirE2 interacting protein 2), might play a role in T-DNA integration (5). Finally, histone post-translational modification (specifically, methylation) was shown to affect T-DNA integration (60).

VARIABILITY OF THE MOLECULAR PATHWAY FOR AGROBACTERIUM INFECTION

Different strains and species of Agrobacterium use different pathways for the transfer of DNA to different eukaryotic organisms. Besides the wide range of plant species that serve as hosts to Agrobacterium spp. in nature (31), under artificial conditions this range extends further to species from all the clades of the plant kingdom (92) as well as to non-plant cells, such as yeast, other fungi, and animal cultured cells (70 and references therein). This possibility to transfer DNA to virtually all eukaryotic cell types (70) reflects Agrobacterium adaptability beyond plant-specific factors.

Essential and Optional Virulence-Associated Genes

Agrobacterium’s virulence factors fall into three main categories: (a) the core factors absolutely essential for T-DNA transfer, i.e., the two-component regulatory system (VirA, VirG), the T-strand processing machinery (VirD1, VirD2), and T4SS (VirB1-VirB11, VirD4); (b) the important but not absolutely essential factors, such that in their absence the T-DNA transfer occurs only at very low efficiency, i.e., VirE2 effector, VirC1, and VirC2; and (c) the nonessential factors that likely play a role in determination of host range and/or in further facilitating infection, for example, in competition with other microorganisms, i.e., the effector proteins VirD3, VirD5, VirE3, and VirF as well as some bacterial strains containing additional Vir proteins that fall into this nonessential category, such as VirH, VirJ, VirK, VirL, and VirM. Several proteins, usually termed Chv, encoded by the bacterial chromosome also play an important role in Agrobacterium interactions with plant cells; they are involved in different steps of infection, such as virulence activation (e.g., ChvE, ChvG, ChvI, ChvH) or cellular adhesion and biofilm formation (e.g., ChvB, ChvA, ExoC). Finally, in addition to the Ti plasmid, some Agrobacterium strains carry a large At plasmid; its function appears not essential for DNA transfer, although it encodes factors with activities related to survival in the competitive rhizosphere environment, such as quorum-sensing mechanisms that regulate plasmid exchange in bacterial populations (104). Although all species and strains of Agrobacterium share a common general mechanism for T-DNA transfer, there is a certain degree of variability between them, which translates into differences in the bacterial virulence factors and affects the outcome of infection. As described above, the transfer of T-DNA by Agrobacterium relies on a core of essential factors. Presumably, the function of these proteins is conserved between different virulent Agrobacterium strains, although they may interact with different host factors. Nonessential genes were defined based on virulence of the corresponding mutants in highly susceptible hosts, such as tobacco or kalanchoe (57). Thus, although they are not absolutely required for transformation of these plants, they may be necessary to infect other plant species or they may provide a competitive advantage to achieve successful infection in nature.

Variability in Inducers and Repressors of the vir Genes

The outcome of the interactions between a specific Agrobacterium strain and its specific host plant is also affected by the signal molecules emitted by the host and by the response of the bacteria to these signals. Perception of inducing signals varies between different Agrobacterium strains, which may reflect adaptation to specific hosts. A wide variety of phenolic compounds, related to AS, can activate vir gene expression (87), including glycoside derivatives (61). Genetic studies identified the protein able to recognize these phenolic compounds as VirA via swapping virA genes between different strains of Agrobacterium, thereby modifying the range of recognized phenolic molecules (74). That different Agrobacterium strains show variable responses to different phenolic compounds may confer onto each strain a specific inducibility corresponding to the phenolics emitted by the strain’s specific host species. This sensing of phenolics by VirA in different Agrobacterium strains may also be affected by specific monosaccharides that are sensed by the chromosomal virulence protein ChvE (101).

Other plant-produced molecules also affect Agrobacterium’s virulence, likely contributing to the variability of T-DNA transfer efficiencies in different plant species or tissues. Among the signal molecules emitted by plants in response to biotic or abiotic stresses, salicylic acid (SA) inhibits vir gene expression, probably by attenuating the VirA protein kinase activity (149). Tobacco or Arabidopsis plants overproducing SA or treated with exogenous SA displayed increased resistance to Agrobacterium, whereas plants deficient in SA accumulation were more sensitive (6, 149). Furthermore, ethylene might also inhibit Agrobacterium’s virulence, although its direct effect on vir gene expression has not been demonstrated (95). Some plant species emit chemicals that inhibit Agrobacterium virulence, most likely contributing to the variability of susceptibility of different species to Agrobacterium. For example, DIMBOA [2,4-dihydroxy-7-methoxy-2H-1,4-benzoxazine-3(4H)-one] and MDIBOA (2-hydroxy-4,7-dimethoxybenzoxazin-3-one), two chemical compounds found in maize homogenates, are strong inhibitors of Agrobacterium AS-induced virulence and growth (109).

VirE2 versus GALLS

VirE2 is not absolutely essential for Agrobacterium-mediated transformation, although transformation efficiency of virE2 mutants is very low and results in a greater proportion of truncated T-DNA (107). Interestingly, some strains of Agrobacterium rhizogenes do not contain the virE2 gene but carry the GALLS gene instead, which complements virE2 in a virE2-deficient strain of Agrobacterium, although its mode of action appears to be different from that of VirE2 (54). GALLS encodes two proteins, corresponding to the full-length coding sequence or to its C-terminal part, that interact with VirD2 and are targeted to the host cell nucleus (55). Their exact function in the transformation process remains unknown.

Interaction of VirE2 with VIP1 and its Homologs

The interaction between VirE2 and VIP1 was first discovered via yeast-two-hybrid screening (126). Using transgenic tobacco overexpressing VIP1 from Arabidopsis (AtVIP1), it was shown that VIP1 overexpression increased the transformation rate and that VIP1 likely facilitated the nuclear import of VirE2 and thus of the T-complex (126, 127). However, other studies using Arabidopsis mutants reached a different conclusion, i.e., that VIP1 was not required for Agrobacterium-mediated T-DNA transfer (113). Recently, VirE2 proteins from four different Agrobacterium strains were all shown to interact with AtVIP1 and one or more of its close Arabidopsis homologs; interestingly, binding efficiency for the different AtVIP1 homologs was different among the different VirE2 proteins (137). This interaction between VirE2 and several AtVIP1 homologs was confirmed in a more recent paper (72). This study also showed that plants expressing a dominant negative mutant of AtVIP1, e.g., AtVIP1 fused to an SRDX transcriptional repression domain, did not affect transformation; however, SRDX inhibits only the function of AtVIP1 as transcriptional activator rather than as a VirE2 binding partner. An Arabidopsis mutant with three disrupted AtVIP1 homologs showed a modest reduction in transformation efficiency (72), suggesting that functional redundancy might mask the role of these proteins in transformation.

Involvement of the Ubiquitin–Proteasome System: Roles of VirF

The VirF effector from the A. tumefaciens A6 octopine-type strain (A6-VirF) was shown to contain an F-box domain and bind several Arabidopsis ASK proteins, the Skp1 homologs that function in the SCF pathway for proteasomal degradation (111). The F-box protein activity of A6-VirF was demonstrated in yeast and in plant cells, and one of its potential substrates and interacting partners was identified; specifically, VirF was shown to bind VIP1 and destabilize via the ubiquitin–proteasome system (UPS) both VIP1 and its associated VirE2 (128). More recently, several other Arabidopsis interactors of VirF were identified (42, 43), one of which, VFP4, a transcriptional regulator of genes involved in stress or defense response, was targeted by VirF for proteasomal degradation (42). In addition, VirF itself is destabilized via UPS, and this destabilization is prevented by VirF interaction with another effector, VirD5 (82). Historically, the C58 nopaline-type strain of A. tumefaciens has not been considered to encode an active VirF (86), but more recent data suggested that the C58–VirF can bind Arabidopsis ASK proteins via its predicted F-box domain, which suggests that it is a bona fide F-box protein (66). Unlike A6-VirF, C58-VirF did not interact with VIP1, suggesting that it has a different set of targets. Most bacterial species able to transfer DNA (e.g., A. rhizogenes, Agrobacterium vitis, and R. etli) encode a VirF homolog with potentially different target specificities, likely contributing to the host range specificity of these bacteria. Also, an Arabidopsis F-box protein VBF was shown to substitute to A6-VirF in targeting VIP1 for degradation, potentially explaining the dispensability of VirF for infection of this plant species (150). Interestingly, VBF was among the plant genes whose expression is activated by VirE3, which has transcriptional regulator activity in plants (94).

T-DNA TRANSFER BY NON-AGROBACTERIUM BACTERIAL SPECIES

The most common pathogenic Agrobacterium species include A. tumefaciens, A. rhizogenes, and A. vitis. The taxonomy of Rhizobiaceae is still subject to revision, and we follow the commonly used classification in which Agrobacterium species are named according to their pathogenic interactions with plants (37). Their pathogenicity relies on Ti plasmids or, in the case of A. rhizogenes, on Ri plasmids, which are highly diverse and present a mosaic structure (97). Furthermore, because these plasmids can be transmitted by conjugation within or even between these bacterial species, other non-Agrobacterium species could gain the Ti-plasmid features that allow transfer DNA to eukaryotic hosts.

Agrobacterium T-DNA Transfer Machinery in Related Bacterial Species

Introducing the Agrobacterium vir region and T-DNA into several species of plant-associated bacteria—pathogenic, symbiotic, and nitrogen-fixing—related to Agrobacterium spp. has conferred onto these bacteria the ability to genetically transform plants. So far, all such bacterial species have belonged to two families of the Rhizobiales order, Rhizobiaceae and Phyllobacteriaceae. For example, conjugative transfer of the Ti plasmid from a virulent Agrobacterium strain to Rhizobium trifolii resulted in virulent bacterial cells able to induce crown gall formation in several plant species (56). Several other bacterial species became capable of transforming plants after they have received two plasmids: a helper plasmid, carrying the Agrobacterium vir region, and a binary plasmid with an engineered T-DNA. Using this strategy, Arabidopsis, tobacco, and rice were transformed by three different bacterial species, Rhizobium leguminosarum, R. trifolii, and Phyllobacterium myrsinacearum (19), whereas Sinorhizobium meliloti, Rhizobium sp. NGR234, and Mesorhizobium loti were used to transform potato (140). Similarly, Ensifer adherens (syn. Sinorhizobium adherens), harboring a cointegrated plasmid and carrying the vir region from Agrobacterium and an engineered T-DNA, was able to transform potato and rice plants (139, 154). Thus, all these bacterial species likely possess the chromosomally encoded function required for transformation but not the vir gene functionalities. Attempts to transform plants or other eukaryotes by introducing plasmids carrying a vir region and a T-DNA into bacteria outside of the Rhizobiales order, such as Escherichia coli, have been unsuccessful (80, 99). Agrobacterium spp. are facultative pathogens (12) with a possible transition between pathogenic and nonpathogenic lifestyles. Horizontal gene transfer between bacterial species via the exchange of plasmids by conjugation is well documented (13); thus, vir regioncarrying plasmids may be shared among a pool of related bacteria in the rhizosphere, thereby conferring pathogenicity to the recipient cells.

Rhizobium etli CFN42 Contains Functional vir Genes

Although DNA transfer to plants can be achieved using different bacterial species provided with Agrobacterium’s vir region, these bacterial species do not encode an endogenous complete and functional DNA transfer machinery, making Agrobacterium the only species with natural genetic transformation capability. This notion has been altered by the observations that the Rhizobium etli CFN42 strain, a symbiotic nitrogen-fixing bacterium associated with host plants such as beans, contains in its p42a plasmid a complete and functional vir region and is able to mediate DNA transfer and stable integration into the plant genome, albeit with a low efficiency, when a vector carrying a T-DNA with a reporter gene is provided (67). The R. etli CFN42 strain with mutated virG or virE2 and R. leguminosarum, a very similar bacterial strain that does not contain close homologs of the Agrobacterium vir genes, were incapable of T-DNA transfer. The vir regions of R. etli and Agrobacterium share extensive similarity yet exhibit two significant differences: The R. etli virB2 gene is not a part of the virB operon but constitutes a separated operon with its own promoter, and there are two virF operons in R. etli. Analysis of vir gene expression in R. etli showed a pattern of expression close to that observed in Agrobacterium, notably induction by AS, except for the virB2 gene that was expressed constitutively at low levels and was not induced by AS (136). Interestingly, the R. etli p42a plasmid was shown to be exchanged between R. etli and related species, including Agrobacterium spp. (15). Although R. etli has evolved to encode and preserve the functional vir machinery, it remains unknown whether this species also contains endogenous T-DNA-like sequences that could be transferred to the plant hosts.

CONCLUSIONS

There is a wide diversity in the pathways underlying each step of DNA transfer from pathogenic Agrobacterium and related species. This unique capability relies on a core of essential bacterial factors and on their interactions with different host cell factors. In addition, many other bacterium-encoded proteins represent facultative virulence factors that are not essential for DNA transfer to model plants highly susceptible to Agrobacterium. Rather, these nonessential factors may be required for infection of specific hosts as well as for achieving maximally successful infection in the competitive rhizosphere environment. The virulence genes are mostly located on a large plasmid, transmissible between bacterial cells by conjugation, but functions encoded by the bacterial genome are also important for efficient T-DNA transfer under natural conditions. The diversity of pathways, as well as the large array of bacterial factors presumed to facilitate and optimize infection, likely confers to Agrobacterium spp. their seemingly unlimited range of host cells under natural or experimental conditions.

Our present knowledge of the Agrobacterium-mediated T-DNA transfer raises an interesting question: What constitutes the minimal T-DNA transfer machinery? Among the essential vir-encoded proteins, most represent pathways common to many bacterial species. For example, the virB and virD operons encode a DNA transfer machinery similar to those involved in plasmid transfer by conjugation, and the VirA/VirG sensors regulating expression of vir genes are representative of the widespread bacterial two-component regulatory systems. These common pathways may allow easy addition of the genetic transformation capability by other bacterial species via acquisition of the functional vir region. Such gene transfer possibly occurs within natural bacterial populations, and it would render pathogenic those bacterial species that are not normally considered as such, e.g., rhizobia, as they are usually engaged in symbiotic relationships with their host plants.

Interestingly, by introducing an artificial transferable DNA into several human pathogens, DNA transfer to cultured human cells was achieved under laboratory conditions (38, 112) in a T4SS-dependent manner. Analyses of complete eukaryotic genome sequences, which are becoming increasingly available, have shown that they contain a significant number of sequences originating from prokaryotes and resulting from horizontal gene transfer (58, 68). It makes biological sense that at least some of these sequences have been acquired from bacteria via a mechanism similar to the Agrobacterium-mediated T-DNA transfer. In some cases, bacterial sequences present in genomes from several plant species of the Nicotiana and Linaria genera, as well as sweet potato, can be traced back to their Agrobacterium-like donor bacteria that share homologies with today’s Agrobacterium pathogenic species (reviewed in 105).

UNANSWERED QUESTIONS

Some of the fundamental questions about the mechanism of the Agrobacterium-mediated T-DNA transfer remain unanswered. How do the T-DNA and its associated proteins pass through the host cell plasma membrane? What is the exact role of VirE2, and its interacting plant proteins, in packaging the T-DNA and facilitating T-DNA subcellular transport and fate within the host cell? How are the multiple pathways for T-DNA integration into the host cell genome regulated? Besides plant genetic transformation by Agrobacterium spp., the question is whether other natural cases of DNA transfer, via a similar mechanism, from different bacterial species to their eukaryotic hosts exist or have existed in past evolutionary times and whether such events may have contributed to the gene flux from bacteria to eukaryotes. From a biotechnological viewpoint, a better understanding of DNA transfer mechanisms will help expand our toolbox for Agrobacterium-mediated transformation, for example, for improvement of the genetic transformation of recalcitrant plant species or non-plant eukaryotic cells or for better control of the integration sites and integration patterns within the target genome. In this respect, the ultimate feat would be using synthetic biology to refactor the entire Ti plasmid (and, potentially, even the bacterial chromosome) to eliminate all pathogenic and transformation-unrelated (e.g., bacterial conjugation) abilities and to include nonbacterial (e.g., plant) genes known to facilitate transformation and/or transgene expression by refactoring them for optimal prokaryotic expression and export.

Ti plasmid: tumor-inducing plasmid

vir: virulence

T-DNA: transferred DNA

AS: acetosyringone

IAA: indole acetic acid

LB: left border

RB: right border

ssDNA: single-stranded DNA

T4SS: type IV secretion system

VBPs: VirD2 binding proteins

UPPs: unipolar polysaccharides

T3SS: type III secretion system

VIP1: VirE2 interacting protein 1

VIP2: VirE2 interacting protein 2

DSBs: double-strand breaks

MMEJ: microhomology-mediated end-joining

dsDNA: double-stranded DNA

HR: homologous recombination

NHEJ: nonhomologous end-joining

UPS: ubiquitin–proteasome system