The widespread plant-colonizing bacterial species Pseudomonas syringae detects and exploits an extracellular pool of choline in hosts



The quaternary ammonium compound (QAC) choline is a major component of membrane lipids in eukaryotes and, if available to microbial colonists of plants, could provide benefits for growth and protection from stress. Free choline is found in homogenized plant tissues, but its subcellular location and availability to plant microbes are not known. Whole-cell bacterial bioreporters of the phytopathogen Pseudomonas syringae were constructed that couple a QAC-responsive transcriptional fusion with well-characterized bacterial QAC transporters. These bioreporters demonstrated the presence of abundant free choline compounds released from germinating seeds and seedlings of the bean Phaseolus vulgaris, and a smaller but consistently detectable amount of QACs, probably choline, from leaves. The localization of P. syringae bioreporter cells to the surface and intercellular sites of plant tissues demonstrated the extracellular location of these QAC pools. Moreover, P. syringae mutants that were deficient in the uptake of choline compounds exhibited reduced fitness on leaves, highlighting the importance of extracellular choline to P. syringae on leaves. Our data support a model in which this choline pool is derived from the phospholipid phosphatidylcholine through plant-encoded phospholipases that release choline into the intercellular spaces of plant tissues, such as for membrane lipid recycling. The consequent extracellular release of choline compounds enables their interception and exploitation by plant-associated microbes, and thus provides a selective advantage for microbes such as P. syringae that are adapted to maximally exploit choline.


Microbes are adept at exploiting the chemical environment of their hosts. Quaternary ammonium compounds (QACs) such as choline and glycine betaine are of particular interest at the host–microbe interface due to their potential abundance in plant hosts and their propensity to provide stress protection and nutrients to the plant-associated microbiota. The presence in plants of specific QACs arises, in part, from the prominence of phosphatidylcholine as a major membrane lipid and from the common strategy of accumulating glycine betaine (hereafter called betaine) for protection from drought and salinity stress. Despite the protective benefits of these QACs to plant and bacterial cells, few bacterial species can synthesize QACs de novo, although many, such as the phytopathogen Pseudomonas syringae, have transporters for their uptake (Chen and Beattie, 2007, 2008; Chen et al., 2010).

Pseudomonas syringae is a widespread bacterial species on plants and is comprised of many pathovars that cause plant diseases. For example, P. syringae pv. syringae is a foliar pathogen of bean that can colonize the seeds and leaves of its host species, Phaseolus vulgaris, as well as the leaf surfaces of a wide variety of non-host plant species (Hirano and Upper, 2000). A major stress confronting P. syringae during plant colonization is the fluctuating and often low availability of water (Beattie, 2011; Yu et al., 2013). In general, bacterial cells adapt to low water availability by accumulating small organic osmolytes called compatible solutes. This accumulation may result from de novo synthesis of compatible solutes, which in P. syringae include glutamate, trehalose, and the dipeptide N-acetylglutaminylglutamine amide (NAGGN) (Freeman et al., 2010; Kurz et al., 2010; Li et al., 2013), or through transporter-mediated uptake of compatible solutes or their precursors from the environment. Accumulation by uptake is energetically more favorable than by de novo synthesis (Oren, 2011) and enables a more rapid resumption of bacterial growth following a decrease in water availability. Interestingly, even nanomolar levels of exogenous osmoprotectant compounds favor bacterial growth when tested at low cell densities (Cosquer et al., 1999). The ability to transport QACs may therefore confer a fitness benefit to bacteria during their colonization of plants.

Pseudomonas syringae exploits a narrower range of osmoprotective compounds than many soil-inhabiting organisms, including the free-living Bacillus subtilis and the root-associated Sinorhizobium meliloti. The commonly studied strain P. syringae pv. tomato DC3000 (Pst DC3000) derives osmoprotection from QACs, including choline, betaine, l-carnitine, and acetylcholine, but not from proline, ectoine, and various disaccharides, which are used by B. subtilis and S. meliloti (Gouffi et al., 1999; Pittelkow and Bremer, 2011; Hoffmann et al., 2012). Although betaine serves as the resulting compatible solute after choline uptake (Le Rudulier et al., 1984), Pst DC3000 cells derive better osmoprotection from choline than betaine (Chen and Beattie, 2007), suggesting that Pst DC3000 favors the transport of choline over betaine. We have identified three QAC transporters in Pst DC3000: two osmoregulatory transporters, the BCCT-type transporter BetT (Chen and Beattie, 2008) and the ABC transporter OpuC (Chen and Beattie, 2007), and one catabolism-associated ABC transporter, Cbc (Chen et al., 2010). All three transport choline, whereas only two transport betaine. The low-affinity transporter BetT exhibits a particularly high capacity for choline uptake, enabling Pst DC3000 to import choline rapidly when choline is available at high concentrations, suggesting that Pst DC3000 has evolved to exploit choline-rich habitats.

The extent to which plant-derived QACs are available to the plant microbiota is not known. Free choline and, to a lesser extent, phosphocholine are commonly found in homogenized plant tissues (Strange et al., 1972; McNeil et al., 2001; Zeisel et al., 2003). Radiotracer studies have detected choline only after the appearance of phosphatidylcholine (PC) (Datko and Mudd, 1988b), suggesting that choline is a degradation product of PC, and this is probably due to activity of phospholipase D (PLD) (Welti et al., 2002; Bargmann et al., 2009). Plant genomes encode an unusually diverse group of PLDs, many of which can hydrolyze PC to choline and phosphatidic acid (PA). Whereas the lipid messenger PA has been extensively studied for its roles in signaling (Hong et al., 2010; Testerink and Munnik, 2011), the fate of choline has not been examined.

In this paper we investigate whether QACs including choline are available to P. syringae at the host–microbe interface. We present evidence that free choline and/or its ester phosphocholine are available to P. syringae during colonization of bean and soybean tissues. Since P. syringae does not penetrate plant cells, these results provide evidence that extracellular QAC pools are present in these plants. Moreover, we show that plant-derived choline is critical to the in planta fitness of P. syringae, indicating that this common plant-associated species has adapted to exploit this pool.


As a species, Pseudomonas syringae is particularly well adapted to exploit choline and other QACs for osmoprotection and nutrition

Pseudomonas syringae pv. tomato DC3000 is the only bacterium thus far known to derive better osmoprotection from choline than from betaine. We evaluated if this unusual trait is conserved in other pseudomonads and found that the plant-associated strains P. syringae pv. syringae B728a and P. syringae pv. phaseolicola 1448A, but not the opportunistic human pathogen Pseudomonas aeruginosa PAO1 or the soil microbe Pseudomonas putida KT2440, showed superior osmoprotection from choline (Figure 1a–e and Figure S1 in Supporting Information). We also evaluated the impact of these QACs on the expression of the water stress-responsive transcriptional fusion proU-gfp (Axtell and Beattie, 2002) in B728a following an osmotic upshift. Choline suppressed the osmoinduction of proU-gfp much more than did betaine (Figure 1f), consistent with its greater osmoprotective effect. In contrast, proline showed little suppression of proU-gfp induction, consistent with its lack of osmoprotection (Figure S1) and reduced uptake under high osmolarity (Chen and Beattie, 2007).

Figure 1.

The superior osmoprotection of choline over betaine is specific to Pseudomonas syringae strains based on growth and attenuation of osmoinduction.

(a)–(e) Growth of P. syringae (Pst, Psp, Pss), P. aeruginosa (Pa), and P. putida (Pp) strains in MinA medium amended with 0.4 m NaCl and choline or betaine at the indicated concentrations. Growth was measured as the optical density at 600 nm (OD600) at 8 h (Pst, Pss, and Pa) or 12 h (Psp and Pp) after osmotic upshift. The OD600 values were significantly higher in the presence of choline than betaine for Pst and Psp at all concentrations ≥50 μm and at 200 and 1000 μm for Pss, but were significantly lower at 200 and 500 μm for Pa and did not differ for Pp (< 0.05 in Student's t-tests). (f) The effect of choline, betaine, and proline on expression of proU-gfp following an osmotic upshift from 0 to 0.5 m NaCl. The fold-change values are shown. The fold-change values differed significantly for each of the treatments at 12 h after the upshift (< 0.05 in a Fisher's least significant difference test). Values are means ± SE (a–e, = 3; f, = 4).

Genes for the uptake and catabolism of choline and other QACs are highly conserved among all sequenced P. syringae strains and thus are in the core P. syringae genome (Lindeberg et al., 2008; Chen et al., 2010; Winsor et al., 2011). A survey of plant-associated P. syringae strains from various hosts indicated conservation of their ability to grow on choline, betaine and acetylcholine (Table S2). However, the catabolism of l-carnitine, which is more widely distributed in animal than plant tissues (Zeisel et al., 2003; Bourdin et al., 2007), was limited to approximately half of the strains, with this deficiency probably being due to the absence of carnitine uptake and catabolic genes in those strains (Chen et al., 2010).

Quaternary ammonium compound-specific Pseudomonas syringae bioreporters were developed for probing plant-derived choline and other QACs

To investigate the extent to which P. syringae cells are able to access choline and other QACs, we designed P. syringae-based bioreporters that respond specifically to these compounds. Their design involved two components: a transcriptional fusion that responded to QACs as a group and the manipulation of uptake specificity by exploiting the substrate preferences of the QAC transporters (Chen and Beattie, 2007, 2008; Chen et al., 2010). A transcriptional fusion that contained the promoter of the Cbc transporter-encoding operon, cbcXWV, on plasmid pCbc-GUS was only marginally induced by choline when expressed in B728a (Figure 2a,b); despite that, this operon was highly induced by choline in P. aeruginosa strain PA14 (Malek et al., 2011). In contrast, fusions containing the promoter of the glyA-soxBDAG operon on pSox-GUS and a chimera of the cbc and sox promoters, designated the QAC promoter, on pQAC-GUS were highly induced by choline (Figure 2a,b). The gly-sox operon encodes enzymes required for the degradation of sarcosine (Chlumsky et al., 1995; Wargo et al., 2008), a downstream product of betaine catabolism. The gly-sox operon was highly expressed in B728a cells recovered from the surface and intercellular spaces of bean leaves (Li et al., 2013; Yu et al., 2013).

Figure 2.

Pseudomonas syringae-based bioreporters were constructed that report on choline and other quaternary ammonium compounds (QACs).

(a) B728a genomic region showing the cbc and sox promoters and a chimera of the two, called QAC, which were cloned into pFAJ1701. Photographs show the GUS activity of the cbc-uidA, sox-uidA and QAC-uidA fusions in B728a cells grown with choline and X-gluc.

(b) Expression of these fusions in B728a following a 2-h exposure to glucose or choline in MinA medium. The GUS activity was measured using the fluorimetric substrate 4-methylumbelliferyl glucuronide (MUG), with subsequent conversion of fluorescence to nmol of 4-methylumbelliferone (MU) based on a standard curve. Exposure to choline resulted in higher GUS activity than exposure to glucose or no inducer for each of constructs except pFAJ1701 [< 0.05 in Fisher's least significant difference (LSD) tests].

(c) Expression of QAC-uidA in B728a, BD, and BT following a 3-h exposure to the indicated compounds (1 mm). The GUS activity was measured using MUG and the fluorescence values were normalized by the OD600. Comparisons among the strains exposed to a single compound indicated significantly higher values for B728a(pQAC-GUS) than the other strains in response to carnitine, betaine, and choline-O-sulfate (COS), and higher values for B728a(pQAC-GUS) and BD(pQAC-GUS) than for BT(pQAC-GUS) in response to choline, acetylcholine (Acho) and phosphocholine (Pcho) (< 0.05 in Fisher's LSD tests). Values are means ± SE (= 3). DMG, dimethylglycine.

The bioreporter B728a(pQAC-GUS) responded not only to choline but also to betaine, phosphocholine, acetylcholine, choline-O-sulfate, l-carnitine, and the betaine degradation products dimethylglycine and sarcosine, but not to succinate, glucose, or proline (Figure 2c), or to the amino acids glycine, alanine, l-arginine, aspartate, glutamine, glutamate, proline, l-serine, or l-threonine. We introduced the QAC-uidA fusion into isogenic B728a derivatives that differed in the number of QAC-specific transporters. B728a is similar to Pst DC3000 in encoding the three choline transporters BetT, OpuC, and Cbc (Figure 3a). Strain BD lacks OpuC and Cbc but retains BetT, which imports choline and acetylcholine under both low and high osmotic conditions (Chen and Beattie, 2008). Strain BT lacks all three transporters and thus was deficient in the uptake of [14C]betaine, [14C]carnitine, and [14C]choline (Figure 3b). As expected, choline- and acetylcholine-induced QAC-uidA expression was lost in BT(pQAC-GUS) but was maintained in BD(pQAC-GUS) (Figures 2c and 3c). Similarly, phosphocholine-induced QAC-uidA expression was lost in BT(pQAC-GUS) but was maintained in BD(pQAC-GUS) (Figure 2c), presumably due to the release of choline from phosphocholine by the periplasmic phosphocholine phosphatase (Massimelli et al. 2005). Both dimethylglycine and sarcosine induced similarly high QAC-uidA expression in all three reporter strains (Figure 2c), suggesting the presence of alternative transporters for these compounds. The lack of detectable accumulation of sarcosine or dimethylglycine in plant tissues (Goyer et al., 2004) indicates a low probability of interference when using these bioreporters in planta. Lastly, the QAC-uidA fusion was induced approximately 15-fold in B728a by 8-h exposure to choline concentrations as low as 0.1 μm (Figure S2), indicating that the fusion was highly sensitive to choline.

Figure 3.

The specificity of the bioreporters was altered by expressing QAC-uidA in P. syringae derivatives differing in quaternary ammonium compound (QAC) transport.

(a) Diagram of the three known QAC transporters in B728a, their substrates, and their presence in B728a, BD, and BT.

(b) The [methyl-14C]QAC uptake activity of B728a, BD, and BT when measured in MinA medium, designated low-salt (black, grey, and white bars, in nmol mg−1 protein), and MinA amended with 0.5 m NaCl, designated high-salt (red bar, in disintegrations min−1). Comparisons among the strains provided with a given substrate indicated that B728a transported significantly more choline, betaine, and carnitine than the other strains under the low-salt conditions, significantly more choline than the other strains under the high-salt conditions, and BD transported significantly more choline than BT under the high-salt conditions [< 0.05 in Fisher's least significant difference (LSD) tests]. Values are means ± SE (= 2–4).

(c) Effect of glucose, choline, and the non-QAC sarcosine on QAC-uidA expression in B728a, BD and BT. The GUS activity was measured using MUG, and the fluorescence of cells in amended cultures was expressed relative to their fluorescence in unamended cultures. Comparisons among the strains exposed indicated no differences when not exposed to a compound or were exposed to glucose, but significant differences among each of the three strains when exposed to choline, and higher values for BT(pQAC-GUS) than for the other strains when exposed to sarcosine (< 0.05 in Fisher's LSD tests). Each value is the mean fold-change ± SE (= 3).

Quaternary ammonium compounds including choline or its esters were released from bean seeds and seedlings

We deployed these bioreporters to germinating bean seeds to probe for the presence of choline and other QACs. As a seedborne pathogen, B728a propagates rapidly in association with germinating seeds of its bean host plant prior to the colonization of cotyledons and leaves (Hirano et al., 1999). After inoculation of germinating seeds on minimal medium agar plates containing the substrate X-gluc, the bioreporter cells multiplied on the plate surface around the seeds. After 24 h, the blue signal associated with the presence of B728a(pQAC-GUS), but not BT(pQAC-GUS), indicated that one or more QACs were liberated from the germinating seeds (Figure 4a). The similarity of the blue signals of BT(pNpt-GUS) and B728a(pNpt-GUS), which each constitutively express uidA from the nptII promoter, indicate that the lack of GUS activity in BT(pQAC-GUS) is due to the lack of QAC-uidA induction and not a lack of bacteria. To evaluate the relative expression of QAC-uidA in these bioreporter strains, bacterial cells were introduced onto imbibed bean seeds and incubated for 17 h before measuring their GUS activity. Among the strains expressing QAC-uidA, only BT(pQAC-GUS) failed to show detectable GUS activity (Figure 4c). The GUS activity of BD(pQAC-GUS) (Figure 4c), which has only the choline-specific BetT transporter, indicates that choline or its esters acetylcholine or phosphocholine are among the QACs, or are the sole QACs, released from germinating bean seeds. Moreover, given the low affinity of BetT for choline (Figure S2) (Chen and Beattie, 2008), the GUS activity of BD(pQAC-GUS) suggests the presence of a relatively large extracellular choline pool.

Figure 4.

Quaternary ammonium compounds (QACs) including choline are available to P. syringae cells during colonization of germinating bean seeds and seedlings.

(a), (b) Qualitative GUS activity as reflected by a blue color of the bioreporter strains B728a(pQAC-GUS), B728a(pNpt-GUS), BT(pQAC-GUS), and BT(pNpt-GUS) after (a) growing on germinating bean seeds on media amended with X-gluc for 24 h or (b) recovery from seedlings after seed inoculation and 2 days of growth.

(c), (d) The GUS activity of the bioreporter strains after (c) incubation of bioreporter cells on seeds for 24 h or (d) inoculation of seedlings following transfer to a Petri dish and incubation for 24 h. The GUS activity was measured by adding X-gluc to the recovered cells and measuring the absorbance at 650 nm. Values are means ± SE (= 6). Values indicated by the same letter do not differ significantly (< 0.05 in Fisher's least significant difference tests applied separately to strains carrying each bioreporter plasmid).

To further investigate the composition of the QACs released by the germinating seeds, we introduced pQAC-GUS into B728a derivatives that expressed distinct components of the Cbc transporter. The substrate specificity of this transporter is dictated by recognition between the core transporter CbcWV and each of the substrate-binding proteins (SBPs) BetX, CaiX, and CbcX (Figure 5a), and also CosX, which is a choline-O-sulfate (COS) SBP (Psyr_0028) (Figure 5b). The strain BT lacks betT and the opuC and cbcXWV operons, but expresses betX, caiX, and cosX, which are located in separate regions of the chromosome (Chen et al., 2010). We have previously constructed BT derivatives lacking these SBP genes, namely BTbetX::tet, BTcaiX::tet, and BTcosX::tet (Table S1). Introduction of plasmids expressing either the core transporter, pNCbcWV, the complete cbc operon, pNCbcXWV, or an empty vector, pN, into these BT derivatives allowed us to manipulate transport specificity, as demonstrated with strains containing pQAC-GUS (Figure 5b). Each of the nine resulting strains was evaluated for its response to individual QACs (Figure 5b). As expected, following the introduction of pN (vector 1), BTbetX::tet was unable to transport QACs due to the lack of the core transporter (Figure 5b). In contrast, following the introduction of pNCbcWV (vector 2), BTbetX::tet could import l-carnitine, its analog acetylcarnitine, and COS due to the chromosomal caiX and cosX loci (Figure 5b). Similarly, following the introduction of CbcX and the core transporter on pNCbcXWV (vector 3), BTbetX::tet extended the range of QACs that induced QAC-uidA to include the CbcX substrates choline and betaine (Figure 5a,b). Induction of the latter by acetylcholine, which is not a substrate of CbcX (Chen et al., 2010), suggests that acetylcholine is hydrolyzed into choline in the periplasm of B728a. The introduction of these pN-based vectors into BTcaiX::tet and BTcosX::tet resulted in expression patterns that confirmed the role of BetX in betaine uptake, CaiX in l-carnitine and acetylcarnitine uptake, CosX in COS uptake, and CbcX in the uptake of betaine, choline, and presumably choline resulting from acetylcholine hydrolysis.

Figure 5.

The quaternary ammonium compounds (QACs) released from bean seeds were characterized using bioreporters with enhanced QAC specificity due to manipulation of the Cbc transporter components.

(a) Diagram of the Cbc transporter system and the substrate specificity of each substrate-binding protein, BetX, CbcX, CaiX, and CosX. COS, choline-O-sulfate, GB, glycine betaine.

(b) Expression of QAC-uidA in nine bioreporter strains, each of which is a BT derivative lacking betX, caiX, or cosX and containing pN (vector 1), pNCbcWV (vector 2), or pNCbcXWV (vector 3). Expression in response to individual compounds was evaluated colorimetrically for cultured cells and cells on germinating seeds (photographs), and was quantified using 4-methylumbelliferyl glucuronide for cells recovered from the seeds. Values indicated by the same letter do not differ significantly (< 0.05 in a Fisher's least significant difference test).

Each of the nine bioreporter strains was also introduced onto imbibed bean seeds. Bioreporter strains with the pNCbcXWV vector produced intense blue signals on the imbibed bean seeds and significantly higher GUS activity than the corresponding pN-expressing strains (Figure 5b). The BTbetX::tet strain with the pNCbcWV vector showed similar GUS activity to the corresponding pN-expressing strain, indicating that l-carnitine, acetylcarnitine, and COS were not major components of the QACs released by the germinating seeds. In contrast, the BTcaiX::tet strain with the pNCbcWV vector showed significantly greater GUS activity than the corresponding pN-expressing strain, with a similar trend exhibited by the BTcosX::tet strains, suggesting that betaine is among the QACs released by the germinating seeds. The large differences between the pNCbcXWV- and pNCbcWV-containing derivatives of each strain, however, provide strong evidence that choline or its esters comprise a major component of the QAC pools released by germinating seeds.

We obtained similar results with bean seedlings as with germinating seeds. Here, the bean seeds were inoculated and planted in a non-sterile potting mix, in contrast to the sterile culture medium above, and the seedlings were grown for 4–5 days, removed from the potting mix, and flooded with the substrate X-gluc. Again, among the strains, B728a(pQAC-GUS) exhibited much higher GUS activity than BT(pQAC-GUS) (Figure 4b). The bean seedlings were also grown for 3 days in water in test tubes, and the bioreporter strains were introduced and incubated for 24 h before measuring the GUS activity (Figure 4d). The greater GUS activity of BD(pQAC-GUS) than BT(pQAC-GUS) indicated that choline and/or its esters are among the QACs released from seedlings.

Quaternary ammonium compounds including choline or its esters are present and available to Pseudomonas syringae on the surface and in the apoplast of leaves

Plant leaves typically contain much less phosphatidylcholine than seeds (Li-Beisson et al., 2013), and therefore may release less choline than the germinating seeds. We probed for the presence of choline and other QACs in mature leaves by infiltrating bioreporter cells into bean leaves, incubating for 4 days, and visually assessing GUS activity based on an X-gluc overlay followed by chlorophyll removal (Figures 6a and S3a). B728a(pQAC-GUS) and BD(pQAC-GUS) showed detectably higher signals than BT(pQAC-GUS) (Figure 6a), suggesting the availability of choline and/or its esters in the leaf apoplast. The infiltrated leaves were homogenized 4 days post-inoculation (dpi) and the GUS activity was measured using a fluorimetric assay. B728a(pQAC-GUS) exhibited significantly higher GUS activity than BT(pQAC-GUS) (Figure 6b), and although the GUS activity was also higher for BD(pQAC-GUS) than for BT(pQAC-GUS), this difference was not significant, possibly due to the high leaf-to-leaf variability in QAC-uidA expression in BD(pQAC-GUS) (Figure 6b). When plants were inoculated and incubated until disease lesions appeared, the B728a(pQAC-GUS) cells recovered from the leaves showed visibly greater GUS activity than the BT(pQAC-GUS) cells (Figure S3c), again demonstrating the availability of an extracellular pool of QACs such as choline in the apoplast of host bean leaves.

Figure 6.

Quaternary ammonium compounds are available to Pseudomonas syringae cells during colonization of bean leaves.

(a) Qualitative GUS activity of the bioreporter strains in bean leaves after infiltration and growth for 4 days. The pNpt-GUS signal reflected B728a distribution based on constitutive nptII-uidA expression.

(b) The GUS activity of the bioreporter strains after infiltration and growth in bean leaves for 4 days. The GUS activity was measured using 4-methylumbelliferyl glucuronide (MUG) and was normalized to cell number. Values are means ± SE (= 3). Asterisk indicates a significant difference between the BT and the B728a strains expressing QAC-uidA (< 0.05 in a Student's t-test).

(c) The GUS activity and population sizes of the bioreporter strains after spray inoculation onto soybean leaves and incubation for 3 days. The GUS activity was measured using MUG; population sizes were expressed as the log of the colony-forming units (CFU) per leaf. Values are means ± SE (= 6). Asterisks indicate significant differences between BD or BT and B728a expressing QAC-uidA (< 0.05 in a Student's t-test).

To test the availability of QACs to P. syringae on the surface of leaves, we used a non-host plant species because P. syringae growth is restricted primarily to exterior sites on such species (Sabaratnam and Beattie, 2003). We sprayed the bioreporter cells onto leaves of soybean (Glycine max) and incubated the plants under conditions of 75–90% relative humidity for 3 days. Although the three bioreporter strains established similar population sizes on the leaves, B728a(pQAC-GUS) but not BD(pQAC-GUS) exhibited significantly higher GUS activity than BT(pQAC-GUS) (Figure 6c), demonstrating that QACs are available to these bioreporter bacteria on leaf surfaces. Taken together, the high variability in QAC-uidA expression in BD(pQAC-GUS) cells recovered from the apoplast, and the lack of expression in cells recovered from the leaf surface, are consistent with the presence of choline and/or its esters in leaves, but at much lower concentrations than in seeds or seedlings, because the only choline transporter in strain BD, BetT, requires high choline concentrations for uptake whereas the other choline transporters in B728a, OpuC and Cbc, function best at only low choline concentrations (Figure S2).

An extracellular plant choline pool contributes to the fitness of Pseudomonas syringae under natural conditions

To evaluate the impact of QACs on P. syringae plant colonization, we compared the competitive fitness of BT with B728a on host and non-host plant species. For comparison, we included strain BN, which is a B728a derivative that is deficient in the synthesis of the compatible solute NAGGN (Li et al., 2013) but retains all three QAC transporters. In a controlled environment, BN was similar to B728a in its colonization of host bean plants since, after inoculation of an approximately 1:1 mixture with B728a, it comprised approximately 50% of the cells recovered from leaves at 17 dpi (Figure 7a); the lack of an effect of NAGGN deficiency on the epiphytic fitness of B728a was reported previously (Kurz et al., 2010). In contrast, the BT strain exhibited a moderate reduction in fitness as it decreased from 61% of the population at inoculation to 34% at 17 dpi. This reduction was much greater on plants under field conditions where the environmental stress conditions were greater (Hirano and Upper, 2000; Sabaratnam and Beattie, 2003). Strain BT comprised only 17% of the P. syringae cells recovered from leaves at 20 dpi as compared with 63% immediately following inoculation, and only 25% after 20 days, even when BT comprised 77% immediately following inoculation (Figure 7b). On the non-host soybean, where P. syringae established lower population densities than on host plants (Figure S4), BT exhibited a dramatic reduction from 44% at inoculation to 7% at 24 dpi (Figure 7c). Interestingly, the difference between the BT and BN strains at 24 dpi was particularly large on older leaves, 8% versus 45% for the BT and BN strains, respectively (Figure S5). The importance of choline to P. syringae was observed in an additional field experiment in which strain BD, which expresses the BetT transporter, consistently out-performed strain BT when applied as single-strain inoculum onto beans (Figure 7d). Similar fitness comparisons with isogenic Pst DC3000 derivatives also showed a greater fitness reduction for the choline transporter-deficient strain DT than for DD, which could transport choline (Figures S4 and S6). These results further support our conclusions that plant-derived choline is readily available and is specifically required for maximal P. syringae fitness under natural conditions.

Figure 7.

The uptake of quaternary ammonium compounds including choline contributes to fitness of Pseudomonas syringae during colonization of bean and soybean leaves under controlled and natural conditions.

(a)–(c) The percentage of BT and BN in the P. syringae populations recovered from bean or soybean leaves following inoculation of mixtures of B728a and BT or BN, including distinct mixture ratios indicated as Inoc1 and Inoc2. Plants were grown (a) in a growth chamber or (b, c) under field conditions. Values are means ± SE (a, = 6–12; b, c, = 12–24). Asterisks indicate significant differences between the percentage BT mutant at the indicated time to its percentage at 0 days post-inoculation (dpi) (< 0.05, Student's t-test of arcsine(square root)-transformed data); BN did not show significant differences in similar comparisons.

(d) The population sizes of B728a, BD, and BT following single-strain inoculation onto bean leaves under natural conditions; the population sizes were normalized to the population size immediately following inoculation. Values are means ± SE (= 12). Asterisks indicate significant differences between the BT and B728a populations at the indicated times (< 0.05 in a Student's t-test); BD did not show significant differences in similar comparisons.


Our previous discovery that P. syringae is adapted to import particularly large amounts of choline (Chen and Beattie, 2008) led to the prediction that its plant habitats are rich in free choline. Here we confirm this prediction by demonstrating that pools of choline and other QACs are present in bean plant tissues; moreover, their availability to P. syringae indicates that they are localized outside of the plant cells, because P. syringae localizes to the intercellular spaces and surfaces of plant tissues (Boureau et al., 2002; Godfrey et al., 2010). We found particularly large choline pools on germinating seeds and seedlings based on the activation of the bacterial bioreporters. We also found choline pools in and on leaves that were sufficiently large to enhance the fitness of P. syringae during leaf colonization. Collectively, these findings have important implications for the plant microbiota and for plant metabolic processes involved in generating and distributing free choline.

Whereas plant-derived QACs may provide nutritional and osmoprotective benefits to any of the microbes on plant surfaces, P. syringae appears particularly well adapted to exploit these compounds. Here, we showed that P. syringae strains are generally capable of sustained growth via the catabolism of choline, betaine, and other QACs, and thus they are among only a small number of prokaryotes characterized to date with this capability. We also showed that when water availability is limited, the ability to derive highly effective protection from choline is a shared feature of P. syringae strains and is distinct from pseudomonads adapted to other habitats. The evolution of high-capacity choline importers may, in fact, be a critical adaptation enabling P. syringae to maximally grow and survive in and on plants.

Free choline and phosphocholine, and in some species betaine, are easily detected in homogenized plant samples, whereas acetylcholine, sarcosine, and dimethylglycine are present at only low or undetectable levels (Gout et al., 1990; McNeil et al., 2001; Zeisel et al., 2003; Goyer et al., 2004). Sensitive methods for quantifying and localizing choline and other QACs are lacking. In this work we have created P. syringae strains that function as effective bioreporters for QACs. They show sensitivity to concentrations as low as 0.1 μm, at least for choline, specificity derived from the creative use of the components of an unusual prokaryotic QAC transporter, and the ability to respond to a broad range of QACs. Ultimately, we generated a set of transporters that when used in concert can not only demonstrate the presence of extracellular QACs in plant tissues but also begin to identify which QACs are present.

Despite the important roles of plant-derived choline to human and animal nutrition (Zeisel et al., 2003), its biogenesis has not been fully elucidated in plants, largely due to the complexity of the pathways involved. Choline synthesis has been partially characterized in soybean (Dewey et al., 1994; Keogh et al., 2009), where it is produced following synthesis of PC from an ethanolamine precursor via a methylation pathway (Figure S3), and subsequent hydrolysis of PC by one or more phospholipase Ds (PLDs) (Zien et al., 2001; Bargmann et al., 2009). The soybean genome encodes 18 putative PLDs (Zhao et al. 2012), many of which can hydrolyze PC and release choline after being activated by biological and environmental cues (Bargmann and Munnik, 2006). Plants also produce phospholipase Cs (PLCs), which release phosphocholine, although few plant PLCs are PC-specific; those that do are best known for their up-regulation during phosphate deprivation (Pejchar et al., 2010; Wimalasekera et al., 2010). We propose that the GUS activity of the P. syringae bioreporter cells in leaves is due primarily to the availability of choline from PLD-mediated hydrolysis of PC, with a second possible source being PLC-meditated hydrolysis of PC with subsequent conversion of the phosphocholine to choline (Figure S7). Although infiltration of bacterial bioreporter cells into plant leaves may have increased the release of choline, as evidenced by the increased expression of plant PLDs upon bacterial infection (de Torres Zabela et al., 2002), or the infiltration process may have caused wounding that activated PLDs (Bargmann et al., 2009), our ability to detect choline or its derivatives on the surfaces of intact soybean leaves indicates that these QACs are released extracellularly even in the absence of bacterial perturbation.

We cannot rule out the possibility that the choline present in the apoplast fluids is excreted directly from the cytoplasm of plant cells, such as through the action of mechanosensitive channels like those that are widespread in bacteria (Booth and Blount, 2012). For example, when bacterial cells are subjected to an osmotic downshift, intracellular compatible solutes such as betaine are rapidly excreted into the environment to establish a new osmotic balance. Such types of channels, however, have not yet been found in higher plants (Veley et al., 2012).

The release of choline into the intercellular spaces has interesting implications for plant physiology. The fact that PLDs are often intracellular (Fan et al., 1999) led to the current model that PLDs access PC only from the cytosol and therefore release choline into the cytosol (Nakamura et al., 2009). However, recent evidence indicates that some PLDs are located in the mesophyll cell walls of healthy plant leaves (McGee et al., 2003), and purified peanut PLD was able to hydrolyze PC after being infiltrated into the leaf apoplast (Andersson et al., 2006), supporting the potential for PLD-catalyzed hydrolysis of PC located at the outer leaflet of the plasma membrane. The release of choline into the intercellular spaces would enable the translocation of choline to other plant cells, where it could be taken up and converted to PC via the energetically favorable nucleotide pathway (Figure S7). This pathway is dependent on the high activity of choline kinase inside the plant cell cytosol for the conversion of choline to phosphocholine (Gout et al., 1990). If choline is transported intercellularly, then choline procurement by P. syringae could essentially be siphoning the choline away from a continuing cycle of membrane lipid turnover. Release of choline into the apoplast would enable its interception by plant pathogens and its availability to epiphytic colonists following diffusion to the leaf surface (Schönherr, 2006). In seed tissues, the production of choline during the hydrolysis of a large amount of stored PC probably represses the expression of enzymes required for the methylation pathway (Datko and Mudd, 1988a,b), making the nucleotide pathway the sole pathway for PC synthesis during the initial phases of soybean germination (Dykes et al., 1976) when total PLD activity is particularly high (Ryu et al., 1996). These findings are consistent with the detection of a large free choline pool in germinating seeds of bean (Figures 4 and 5) as well as soybean (Figure S8).

The reduced competitiveness of the QAC transporter-deficient P. syringae strains on host and non-host plants (Figures 7, S4, and S6) demonstrated a clear fitness benefit of QAC uptake to P. syringae. This benefit could be due to catabolism for nutrition or osmoprotection in planta. Such protection from low water availability would be critical to the fitness of P. syringae in planta, given the dominance of water limitation among the environmental stresses encountered in the phyllosphere (Yu et al., 2013). We found that, in addition to their superior ability to scavenge choline from the environment, P. syringae cells exhibit chemotaxis to choline and other QACs (Figure S9), a trait that could help promote the movement of P. syringae cells to plant-derived choline pools which are probably not evenly distributed due to the heterogeneous distribution of the precursor PC (Horn et al., 2012). During foliar infection, P. syringae may simply intercept the QACs available in the apoplast, but the finding that they increase PLD and PC-specific PLC activities after infiltration (de Torres Zabela et al., 2002) suggests that P. syringae may actively increase choline or phosphocholine availability. A final potential benefit of choline to P. syringae is as a precursor for the biosynthesis of PC, as the pseudomonads sequenced to date share a conserved PC biosynthetic pathway (Wilderman et al., 2002). Interestingly, this pathway requires exogenous choline (Aktas et al., 2011), suggesting an unexpected metabolic dependency on the host plant. The physiological and ecological roles for PC, however, have not yet been identified in P. syringae. A model showing these potential fitness benefits to P. syringae is shown in Figure S7.

Experimental Procedures

Bacterial strains, plant material, and growth conditions

The bacterial strains and plasmid used in this study are described in Table S1. Bacterial strains were maintained at 28°C in King's B medium (KB) before subculturing into MinA containing 20 mm succinate (MinAS) or the low-osmoticum medium ½-21C containing glucose (10 mm) or pyruvate (20 mm), as specified. All media have been described previously (Chen and Beattie, 2007). Antibiotics were used in the following concentrations (μg ml−1): kanamycin (Km), 50; rifampin (Rif), 100; and tetracycline (Tet), 20. Plant cultivars used included bean (P. vulgaris cultivar Bush Blue Lake 274) (Burpee, and soybean (G. max cultivar Williams 82) (obtained from M. Battacharyya, Iowa State University, USA). Unless otherwise stated, plants were grown in Sunshine Mix #1 (LC1) in a growth chamber at 24°C, 70% relative humidity, and 10-h photoperiod (maximum light 980 μmol m−2 sec−1).

Construction of bioreporters

The transcriptional fusion plasmids pCbc-GUS and pSox-GUS were constructed by amplifying the putative promoters of cbcXWV and glyA-sox of B728a by PCR using the previously described primer sets BcbcF1/BcbcR1 and BcbcF2/BcbcR2 (Chen et al., 2010), respectively, and ligating them individually into the HpaI site of vector pFAJ1701 (Dombrecht et al., 2001) to obtain pCbc-GUS and pSox-GUS, or together through splice–overlap–extension PCR before ligation into pFAJ1701 to obtain pQAC-GUS. The constitutive pNpt-GUS was similarly constructed by introducing the nptII promoter (Chen et al., 2010) into the HpaI site of pFAJ1701.

Strain construction

For the construction of the BN mutant, a DNA fragment encompassing ggnAB (Psyr_3747-Psyr_3748) and approximately 1-kb flanking regions was amplified using the primers BNAGF (TGCGCCTTCAATTCGAAACACCTG) and BNAGR (ATGGCCTGGGAAATGAAACGCAAG). The PCR product was then ligated into the SmaI site of pTOK2T. This was used as a template to replace ggnAB with a kanamycin cassette from pKD13 using the primers BNAGH1(CTTTCTCAGTCACCTCCCCAACGGAAACCCGGTAAGGACTTGTGTAGGCTGGAGCTGCTTCG) and BNAGH2 (CAAAGGCGGATGTGCCGTTTGCAGCCGTGCCC-GCAGCTGCGAATTCCGGGGATCCGTCGACC) and to construct a deletion mutant, as described previously (Chen et al., 2010).

Osmoprotection and uptake assays

Bacterial osmoprotection and uptake assays were performed in MinA or ½-21C as previously described for low- and high-salt conditions (Chen and Beattie, 2007; Chen et al., 2010).

Green fluorescent protein expression assays

Cells were grown at 30°C in ½-21C medium containing glucose to an optical density at 600 nm of 0.2 and were mixed in equal volumes with ½-21C medium containing glucose and NaCl (final concentration, 0.5 m), with or without choline, betaine, or proline (final concentrations, 1 mm). These cultures were incubated at 30°C until sampling at 0.5, 1, 2, 4, or 12 h, when an aliquot was spiked with chloramphenicol (50 μg ml−1) to inhibit further expression and stored at −20°C. Thawed samples were subjected to flow cytometric analysis as described previously (Axtell and Beattie, 2002).

β-Glucuronidase assays

An in vitro histochemical assay for measuring GUS activity was modified from a previously described method (Xiao et al., 1992). Cells were grown at 25°C in KB for 22 h and subcultured in MinAS containing Rif and Tet for 20 h, and 5-μl aliquots were placed on MinA agar amended with X-gluc (20 μg ml−1) and a candidate inducer (10 mm). In vitro fluorimetric GUS assays were performed with cells grown in MinAS containing Tet until late log phase. A 0.1-ml aliquot was inoculated into 1 ml of MinA containing pyruvate (20 mm), with or without a candidate inducer compound (10 mm). After incubating at 28°C for 2 h, a 0.1-ml aliquot was mixed with an equal volume of extraction buffer (0.1 m Na2HPO4 pH 7, 0.0014% β-mercaptoethanol, 0.02 m Na2EDTA pH 8, 0.002% Sarkosyl, 0.002% Triton X-100) amended with 4-methylumbelliferyl glucuronide (MUG) (final concentration, 1 mm). The reaction mixture was incubated at 37°C for 10 or 30 min before being stopped with 1 ml of 0.2 m Na2CO3, and the fluorescence was measured using a fluorometer (Quantech fluorometer, ThermoScientific,

The GUS activity of reporter cells on germinating seeds was measured using seeds that were surface sterilized by immersion in 10% bleach for 10 min and washed five times with sterile water. Five seeds were placed on ½-21C medium with pyruvate, Tet, and 0.5% agar, with (for histochemical analysis) or without (for fluorimetric analysis) X-gluc (20 μg ml−1). Each seed was inoculated with 10 μl of bioreporter cells [108 colony-forming units (CFU) ml−1]. The seeds were incubated at 28°C in the dark and scored periodically for the appearance of a blue color indicating GUS activity. Fluorimetric GUS assays were conducted 24 h after inoculation by collecting bioreporter cells in 5 ml of PBS (10 mm phosphate buffer, pH 7, with 0.01% Silwet), and after vigorous mixing to suspend the bacteria, 0.4 ml was mixed with 0.3 ml extraction buffer containing MUG (final concentration, 1 mm), and incubated at 37°C with shaking for 10, 20, or 60 min before being stopped with 1 ml of 0.2 m Na2CO3 and measured for fluorescence.

The GUS activity of reporter cells on bean seedlings was evaluated only by histochemical analysis. A pool of 8–10 seeds was mixed with 1.8 ml of a bioreporter suspension (108 CFU ml−1) and incubated for 10 min; seeds were planted in non-sterile Sunshine mix #1 and incubated at 28°C in a growth chamber. Upon emergence, the seedlings were placed in a Petri dish, flooded with extraction buffer containing X-gluc (20 μg ml−1), incubated at 37°C, and inspected periodically for color development.

The GUS activity of leaf-associated bioreporter cells was evaluated both histochemically and fluorimetrically. For bean leaves, individual leaves of plants were infiltrated with bioreporter cells (3 × 104–3 × 105 CFU ml−1) and the plants were incubated inside a plastic tent at 25°C with a 12-h photoperiod. The infiltrated leaves were collected after 4 days. A histochemical GUS assay was performed by placing each leaf in a Petri dish, overlaying 1 ml of X-gluc (20 μg ml−1) on the leaf, and incubating at 37°C for 1 h with occasional gentle rotation to ensure even staining. Each leaf was washed with increasing concentrations of ethanol (30–80%) or bleach (10%) to remove chlorophyll. A fluorimetric GUS assay was performed by homogenizing a leaf in 2 ml of ice-cold PBS and then proceeding as for the fluorimetric GUS assay for the seeds, described above, with a 60-min incubation. Bacteria were enumerated on KB containing Rif and Tet. For soybeans, leaves were sprayed with bioreporter cells (approximately 2 × 106 CFU ml−1) and plants were covered with a plastic bag to help maintain high humidity (75–90%) and incubated at 23°C for 3 days. Bioreporter cells were recovered from three leaves by washing with 5 ml of PBS and subjected to the fluorimetric GUS assay, as described above, and viable plate count for CFU determination.

Evaluation of bacterial growth on leaves in field plots

Bean and soybean plants were grown at the Horticultural Research Farm of Iowa State University near Gilbert, IA, USA, in July 2008 and September 2011. Plant inoculation and sampling were performed as described previously (Freeman et al., 2010), but with the inclusion of single-strain inocula for the B728a, BD, and BT strains.


We kindly thank Brian Freeman, Kelly Peterson, Yun Li, and Xiang Wang for technical assistance, Michael Millican for synthesizing choline-O-sulfate, Kylie Hrbek for assistance in the graphic illustration, Xianan Liu for reviewing the manuscript, and Erhard Bremer for helpful discussions. This study was funded by National Science Foundation grant MCB-0920156.