Mechanisms and Regulation of Polar Surface Attachment in Agrobacterium tumefaciens

Summary

Agrobacterium tumefaciens

is a plant pathogen that transfers a segment of its own DNA into host plants to cause Crown Gall disease. The infection process requires intimate contact between the infecting bacteria and the host tissue. A. tumefaciens attaches efficiently to plant tissues and to abiotic surfaces, and can establish complex biofilms at colonization sites. The dominant mode of attachment is via a single pole in contact with the surface. Several different appendages, adhesins and adhesives play roles during attachment, and foster the transition from free-swimming to sessile growth. This polar surface interaction reflects a more fundamental cellular asymmetry in A. tumefaciens that influences and is congruent with its attached lifestyle.

Introduction

Species of Agrobacterium are common soil bacteria, and are members of the Rhizobiaceae family within the Alphaproteobacteria group. Agrobacterium tumefaciens causes Crown Gall, a neoplastic disease of dicotyledonous plants. Most studies of A. tumefaciens have focused on its ability to genetically transform host plants via the interkingdom transfer of T-DNA, a specific replicated segment of the large tumor-inducing, or Ti, plasmid [1]. Virulence is responsive to plant exudates, and is dependent on attachment of the bacterium to plant cells and deployment of a Type IV secretion (T4S) system to deliver the T-DNA and associated proteins [2–4]. Attachment is also a key step to the establishment of A. tumefaciens adherent biofilms on plants and in the soil environment. Although the attachment process remains poorly understood, it is an area of active investigation.

Individual cells transition from an unattached free-swimming stage, through reversible surface interactions towards stable or irreversible attachment (See Ref. [5] for a review; Figure 1). These attached populations may elaborate an extracellular matrix, and mature into a complex, multicellular biofilm. Dispersal releases cells from the biofilm into the planktonic phase. Each stage of biofilm formation, including attachment, is regulated by complex genetic mechanisms and subtle responses to the environment. In this review we focus on recent findings about the mechanisms and regulation of A. tumefaciens attachment and potential linkages of this process with polar development.

Figure 1. Model of multistage biofilm formation process.

Clockwise cycle initiates with attachment of individual cells maturing into a three-dimensionally complex, multicellular biofilm. Eventual dispersal releases a subpopulation of cells from the biofilm into the planktonic phase. Lower panels: Confocal scanning laser micrographs of A. tumefaciens C58 flow cell showing (from right to left) the maturation of a A. tumefaciens biofilm from early, polarly-attached single cells through formation of a mature biofilm (Micrographs courtesy of P.M. Merritt).

Organization and asymmetry of A. tumefaciens cells

Bacterial morphology and extracellular structures greatly impact attachment, when single cells forge interactions with target surfaces [5]. Agrobacteria are typically short, motile rod-shaped cells (averaging 0.8 × 2 um) with several flagella localized at and around a single pole of the cell [6]. As described below, there are several other structures that localize to cell poles (Figure 2). One of these is a polysaccharide-containing adhesive structure, described as the unipolar polysaccharide (UPP; Merritt et al, in preparation). Additionally, immunolocalization and fluorescent tagging of proteins comprising the T4S system of A. tumefaciens also suggest that it localizes predominantly to a single pole [7,8]. An extracellular appendage called the T-pilus, also a component of the T4S system, has been visualized to extend from cell poles [9,10]. The relative locations of these polar structures in relation to each other are unknown. It is a longstanding observation, however, that A. tumefaciens often forms stable associations with plant tissue and abiotic surfaces via a single cellular pole ([11,12]; also see Figures 1 and 3). Furthermore, star-shaped clusters (Figure 3c), often called rosettes, with A. tumefaciens cells bound together through their poles have been reported for many years [13].

Figure 2. Polar structures and polar development homologues.

Lophotrichous (or circumthecally arranged) flagella populate one pole of the cell. Other polar structures are the T4S machinery (black oval) and the associated T-pilus, and the a unipolar polysaccharide (UPP, grey circle). The relative pole indicated for localization of UPP, T4S and T-pilus was used for ease of depiction, and the actual locations remain to be determined. Homologues of C. crescentus polar development genes in A. tumefaciens are listed within the cell.

Figure 3. Polar surface association, rosettes and UPP localization.

(A) Electron micrographs of A. tumefaciens on root cap cells show polar attachment (Micrograph by Martha Hawes) (B). Confocal scanning laser micrograph of fluorescent (fl) WGA lectin staining of attached GFP-labeled cells on glass surface. UPP is pseudocolored magenta (micrograph courtesy of M.E. Hibbing and P.M. Merritt. (C) Electron micrographic examinations of A. tumefaciens growth in carrot extract media demonstrated the formation of star-like clusters in culture (Figure used with permission from ASM Press [13]). (D) These clusters can also be observed in cultures grown in standard minimal media, and fl-WGA staining (red) shows that these clusters consist of cells bound at the pole by UPP; overlay of fluorescence and bright field micrograph (courtesy of E.R. Morton).

The asymmetric locations of extracellular appendages in A. tumefaciens may also reflect a more fundamental cellular asymmetry. Although generally considered symmetrical rods, more recent observations suggest that during A. tumefaciens cell division, at least under some conditions, the division plane is offset from the midcell, giving rise to daughter cells of uneven sizes [14]. Observations over the years reveal that A. tumefaciens rods may also have one wide end and one narrow end [7,13]. It remains to be determined whether the polar surface features described above consistently localize to a specific pole. Cellular asymmetry in bacteria has been best characterized in the stalked alphaproteobacterium Caulobacter crescentus. C. crescentus has a biphasic life cycle in which non-motile stalked cells, often adhered to surfaces, give rise to motile daughter cells via repeated rounds of asymmetric cell division [15]. Extensive studies on C. crescentus have identified several different key regulators (Table 1) that play roles in defining the flagellated pole and the stalked pole, at different stages of the life cycle [15]. Genome sequencing has revealed the presence of homologues to these polar development genes in a number of Alphaproteobacteria including A. tumefaciens [14]. In A. tumefaciens, reasonably strong homologues exist for the PleC, PleD, DivK, DivJ, PodJ, CckM and CtrA polar development proteins (Table 1). Indeed, mutations in several of these A. tumefaciens developmental homologues result in deficiencies for cell division and polar localization (Kim and Fuqua, in preparation). The specific locations and temporal control of these presumptive polar development regulators in A. tumefaciens cells undergoing division and during polar attachment remain to be determined.

Table 1.

Polar development homologues in A. tumefaciens

Protein name	A. tumefaciens Homologuea	Predicted function and localization (Caulobacter model)b	C. crescentus Homologuec
CtrA	Atu2434	Master transcriptional regulator of biphasic life cycle in C. crescentus. Enriched in stalked cells and at the stalk pole.	CC3035
CckA	Atu1362	Cognate histidine sensor kinase for CtrA. Dynamic polar localization.	CC1078
PleC	Atu0982	Histidine sensor kinase – Required to localize other developmental proteins. Inhibits DivK and PleD. Dynamic polar localization and delocalization.	CC2482
DivJ	Atu1888	Histidine sensor kinase – Localized to pole opposite PleC in developing stalked cell. Activates DivK and PleD.	CC1063
DivK	Atu1296	Response regulator – target of DivJ kinase. Nonphosphorylated DivK is daughter cell. Phospho DivK at stalk pole.	CC2463
DivL	Atu0027	Tyrosine kinase suspected to stimulate CtrA phosphorylation. Dynamic polar localization dependent on DivJ.	CC3484
PleD	Atu1297	Response regulator with GGDEF domain and c-di-GMP synthase activity. Required for flagellar ejection.	CC2462
PodJ	Atu0499	Polar development protein J – membrane protein required to develop holdfast and target correct flagellar placement. Dynamic polar localization and proteolytic processing.	CC2045

^aA. tumefaciens C58 genome number (http://agro.vbi.vt.edu/public/)

^bSee Brown et al. for a comprehensive review on the functions of these proteins in C. crescentus development

^cC. crescentus gene number (Genbank)

Transitioning to the surface: a role for motility

The first hurdle in attachment and biofilm formation is the transition out of the planktonic phase and onto the surface. Most bacteria exhibit a reversible attachment stage, that allows for surface sampling, and either subsequent detachment or the conversion to a more stable association [5]. Motility clearly has an important influence on surface interactions and biofilm formation among many different bacteria. A. tumefaciens is only recognized to exhibit swimming motility through flagellar locomotion, with no known alternate motility mechanisms [11]. Non-motile and non-chemotactic mutants exhibit modest, but reproducible virulence attenuation on plants [6,16]. Motility is required for efficient surface attachment and biofilm formation by A. tumefaciens [11]. Mutants with unpowered flagella (due to mutations in the mot genes), are also attachment deficient, indicating that active propulsion is required. In flowing environments aflagellate mutants rapidly grew into very dense biofilms, presumably due to lack of migration away from the site of colonization. These findings also suggest that motility increases the frequency of collisions with the surface and that flowing conditions increase the number of passive collisions. Perhaps of equal importance during surface colonization, motility must be repressed as the cell becomes sessile. A recent report has identified a clutch function that disengages flagellar rotation during biofilm formation in Bacillus subtilis, although such a function has not been identified for A. tumefaciens [17].

Adhesin proteins and early attachment

As in other bacterial systems, the initial binding to surfaces is thought to be mediated by molecular adhesins. Several different potential adhesins have been proposed for A. tumefaciens, but many of these have failed to validate (Box 1). During the plant infection process, the Ti plasmid-encoded vir genes are required to direct T-DNA processing, assembly of the T4S system, and DNA transfer. The vir genes are tightly regulated by the presence of plant-released phenolics, and in their absence, are not significantly expressed. The VirB operon encodes the T4S system and also a structure known as the T-pilus [18]. The current view is that the T-pilus does not function directly in T-DNA transfer, but rather by analogy to conjugal pili, may mediate host cell contact [2,19]. The T-pilus is a filament of the VirB2 T-pilin protein, apparently with the VirB5 protein at its tip [20]. By analogy to other T4S systems, VirB5 is speculated to function as an adhesin, perhaps associating with the receptors on the plant cell [19]. The VirB proteins and several other Vir proteins (VirD4 and VirC1) frequently localize to single cell poles [7,8]. Correspondingly, the T-pilus has been visualized to extend from cell poles, and vir-induced A. tumefaciens were observed to bind to Streptomyces filaments via a pilus-like appendage [9,10]. These observations prompt an appealing model in which the T-pilus, localized at a single A. tumefaciens cell pole, promotes intimate contact between the bacterium and the plant cell, and positions the infecting bacterium appropriately to deploy the T4S system. It is also clear however, that agrobacteria which have not been induced for the Vir system, or lack the Ti plasmid entirely (and thus produce no T-pilus or T4S system), also stably attach by their poles to abiotic surfaces and plant tissues (see Figure 1, and Refs [21,22]). Therefore, although the Ti-plasmid VirB proteins of the T4S system may promote stable contact with plants prior to T-DNA transfer, there are functions encoded elsewhere in the genome that can mediate polar attachment.

Box 1. The Att genes: Dispensable for attachment after all?

For many years A. tumefaciens attachment research focused on the Att gene cluster [45]. Att genes, with a wide range of predicted functions, were proposed to mediate attachment and were reported to be required for virulence. The genome sequence of A. tumefaciens C58 [46,47] however revealed the Att cluster to be located on the 500 kb accessory plasmid pAtC58, known to be dispensable for virulence [48]. Analysis using isogenic derivatives found pAtC58 to exert only mild effects on virulence gene expression, with no obvious impact on attachment [49]. Directed mutation of attR, the most extensively studied Att gene, did not result in virulence deficiencies. The role of these genes in the attachment process is therefore unclear, and a comprehensive re-analysis of the att genes may be required to resolve lingering questions.

Potential candidates for chromosomally encoded adhesin proteins have been identified in the rhizobia, and are called Raps (rhizobial adhesin proteins) and rhicadhesin [23,24]. Rhicadhesin is a small Ca⁺⁺-binding protein that was enriched from the surface of Rhizobium leguminosarum cells grown under Ca⁺⁺-limitation and blocked rhizobial attachment to peas [24]. Rhicadhesin-like activity has been reported for several rhizobia and for A. tumefaciens [25,26]. Rap proteins are also small Ca⁺⁺-binding adhesins that localize to a single cellular pole. Sequences for several Rap proteins were obtained, and the genes that encode them identified [27], but their relationship to rhicadhesin remains unclear. No agrobacterial genomes encode readily identifiable Rap sequences, as is also true for several other rhizobia. Although rhicadhesin activity was reported in A. tumefaciens [28], there is as yet no genetic or biochemical support for the function of such an adhesin.

Polysaccharide adhesins and irreversible attachment

A significant fraction of bacteria that sample a surface will transition from reversible testing of the surface to secure, irreversible binding [29]. For A. tumefaciens, elaboration of cellulose fibrils upon interaction with plants has been linked to initiation of irreversible attachment, because cellulose mutants are readily dislodged [30] [31]. On abiotic surfaces, however, Cel⁻ mutants are fully competent for surface attachment, suggesting differential roles for the polysaccharide that depend upon the surface that is colonized (Xu et al., unpublished data). A cellulose biosynthetic protein in Escherichia coli, BcsQ, was recently reported to localize to a single cellular pole and electron micrographs revealed extrusion of cellulose fibrils from a pole [32]. Polar synthesis of cellulose has not yet been reported for agrobacteria, but if this were found to be the case, it might contribute to polar attachment under certain conditions.

An exciting recent discovery regarding surface attachment is that polarly attached A. tumefaciens cells stain with fluorescent wheat germ agglutinin (WGA) at the pole which contacts the surface (Figure 3b; and Merritt et al. in preparation). Very few unattached cells bind lectin in suspension, although suspended multicellular rosettes stain at the site of polar contact (Figure 3d). WGA specifically binds N-acetyl glucosamine (GlcNAc) residues in polysaccharides, and hence we describe the structure visualized at the site of cell-to-surface contact as the unipolar polysaccharide (UPP). The UPP bears facile similarity to the Caulobacter holdfast, which also labels with this lectin, and functions as a strong adhesive [33]. As with the holdfast, the A. tumefaciens UPP is likely comprised of sugars besides GlcNAc, but in neither case have these other sugars been defined. In C. crescentus, the holdfast is anchored to the cell via a set of holdfast attachment (Hfa) proteins [34], but no Hfa homologues have been identified in agrobacterial genome sequences.

In C. crescentus there is a discrete cluster of genes required for holdfast biosynthesis (hfs genes), and these clearly comprise a Wzy-dependent polysaccharide biosynthetic pathway, one of the two primary pathways by which these polymers are synthesized [35,36]. A screen for biofilm deficient A. tumefaciens mutants revealed an attachment mutant disrupted in a gene homologous to the C. crescentus hfsE gene, encoding the predicted WbaP component of the holdfast biosynthesis pathway (Merritt et al., in preparation). Five additional genes with predicted polysaccharide biosynthetic functions are linked to this hfsE homologue, and one of these is also homologous to HfsD, the predicted Wza outer membrane export porin. Deletion of this gene cluster completely abolishes UPP production and adherence to abiotic surfaces as well as plant tissues. The UPP is observed almost exclusively on cells attached by their poles, on surfaces or to each other. It is therefore tempting to speculate that extrusion of the UPP is induced by conditions at or physical contact with the surface. However, there is as yet no direct evidence regarding the regulation of UPP elaboration or the genes that apparently encode its synthesis. In addition, the integration of UPP localization and activity in A. tumefaciens, with that of motility, cellulose biosynthesis, rhicadhesin activity, and virulence functions including the T-pilus remains to be investigated. Localized polysaccharide adhesives may however be more common than previously appreciated. For example, a unipolar surface polysaccharide, found to be a glucomannan, was reported for the pea symbiont Rhizobium leguminosarum, and mediates polar attachment specificity through binding a pea lectin [37].

Regulation of attachment

There is little known regarding how the process of bacterial attachment is controlled. In many systems, motility must be decelerated, relevant exopolysaccharides synthesized, and metabolism extensively modified during the transition to the sessile phase [38]. Certain conditions, such as phosphorus limitation (Danhorn et al. 2004), promote attachment and subsequent biofilm formation of A. tumefaciens, but the mechanism(s) by which this occurs remains obscure. The synthesis and turnover of the cellular second messenger cyclic digaunosine monophosphate (c-di-GMP) has emerged as a major regulator of the free-swimming to sessile transition in many different bacteria [39]. A. tumefaciens encodes many presumptive c-di-GMP synthases and phospodiesterases that might degrade this signal, and the cellulose synthase is allosterically responsive to c-di-GMP [40], but there is as yet no direct evidence for control of attachment or biofilm formation by these systems.

An unusual regulatory protein called ExoR, homologous to the well-studied ExoR protein of S. meliloti [41], profoundly impacts A. tumefaciens attachment, and exoR mutants do not effectively colonize abiotic surfaces or plant tissues [42]. ExoR has a series of tetratricopeptide repeat (TPR or Sel1) domains that mediate diverse protein-protein interactions [43], and is predicted to be a secreted protein. In S. meliloti ExoR is secreted to the periplasm [44]. ExoR mutants are hypermucoid, reflecting negative control of the exopolysaccharide succinoglycan (SCG). However, in A. tumefaciens SCG is not required for biofilm formation, and the attachment defect of the exoR mutant is largely independent of SCG overproduction (Tomlinson et al. submitted). A. tumefaciens exoR mutants are also aflagellate and nonmotile, perhaps contributing to, but not solely responsible for, poor attachment. DNA microarray analysis suggests that ExoR impacts the expression of many different A. tumefaciens genes (Tomlinson and Fuqua, unpublished). ExoR has been demonstrated to require interaction with the ExoS-ChvI two-component regulatory system in S. meliloti [44]. Although A. tumefaciens has the homologous ChvG-ChvI regulatory system, genetic evidence indicates that in A. tumefaciens ExoR functions through a different mechanism. ExoR regulatory partners are actively being sought in A. tumefaciens.

Conclusions

A. tumefaciens cells adopt a polar orientation on biotic and abiotic surfaces preceding pathogenesis and biofilm formation. The apparent asymmetry of the A. tumefaciens cell is well suited to this mode of attachment and is surely related to, or possibly even an outcome of, this sessile growth strategy. The similarity between A. tumefaciens and C. crescentus in these respects is striking, and provides a useful comparative framework. Flagellar-based motility, protein adhesins and polysaccharides all are likely to have roles in the planktonic-to-sessile transition, and in stable physical association of A. tumefaciens with surfaces. The localization and activity of several of these functions at the cell poles, such as with the UPP polysaccharide-containing structure and the T4S system, likely facilitate the eventual polar orientation. It is not yet clear whether the mechanisms identified to this point represent a single integrated attachment pathway, or whether there are multiple routes to the attached state, that may be distinct for certain surfaces and conditions, including during pathogenesis and benign associations. The temporal and spatial coordination of surface association is very poorly understood at this juncture, and will be a major focus of future research.