Trehalose production by PA14 is required for virulence in Arabidopsis
Reasoning that the tough cellulosic walls of plant cells may pose a unique challenge to plant pathogens, we surveyed the fully sequenced and annotated P. aeruginosa
PA14 genome 
to determine whether canonical cell wall degrading enzymes including cellulases, xylanases, and pectinases are encoded in the genome. In susceptible ecotypes (wild accessions) of Arabidopsis, P. aeruginosa
PA14 causes soft-rot symptoms 
, typically caused by pathogens that secrete pectinases and other hydrolytic cell wall degrading enzymes. Moreover, PA14 infection causes extensive degradation of Arabidopsis mesophyll cell walls including the generation of “holes” approximately the diameter of P. aeruginosa
through which the bacteria enter host cells 
. We thus expected that the PA14 genome would encode a variety of cell wall degrading enzymes (CWDEs). However, our survey of the PA14 genome identified only a single, candidate cellulase, identified ambiguously as “cellulase/peptidase” (PA14_36500). Although PA14_36500 was upregulated two and three days post-inoculation in planta
, correlating with the development of disease symptoms (Figure S1A
), a transposon insertion in PA14_36500 (PA14_36500::MAR2xT7
), in-frame deletion of the cellulose/peptidase gene (ΔPA14_36500
), or in-frame deletion of a putative cellulase/peptidase operon (ΔPA14_36480-36520
) did not cause a significant attenuation in virulence in Arabidopsis leaves (Table S1
Because PA14_36500, which encodes the putative cellulose/peptidase, was induced during plant infection and because genes are often functionally clustered on bacterial genomes, we sought to identify genes adjacent to PA14_36500 that are co-regulated with PA14_36500. This led to the identification of a set of 38 genes (42.23 kb region; PA14_36375 to PA14_36830) spanning the cellulase/peptidase gene that is coordinately down-regulated in an mvfR
(multiple virulence factor regulator) mutant grown under various culture conditions 
. Importantly, the quorum sensing-associated transcriptional regulator MvfR is required for maximum PA14 virulence in Arabidopsis 
. Consistent with the in vitro
transcriptional profiling data, cellulase/peptidase PA14_36500 expression was significantly reduced in planta
in an mvfR
mutant (Figure S1B
Besides the putative cellulase/peptidase, the PA14_36375–36830 42.23 kb region encodes putative glucanolytic enzymes (PA14_36590, PA14_36630, PA14_36740) as well as two closely linked predicted operons (http://www.pseudomonas.com
), PA14_36570-36630 consisting of six genes, and PA14_36710-37640 consisting of three genes, referred to hereafter as the “treYZ”
operons, respectively, that encode enzymes involved in two different trehalose biosynthetic pathways (; Table S2
). TreY and TreZ convert maltodextrins into trehalose in a two-step enzymatic reaction 
, whereas TreS catalyzes conversion of maltose into trehalose in a single reaction 
). In addition to treY
(PA14_36605) and treZ
(PA14_36580), the predicted treYZ
operon contains glgA
(PA14_36590), hypothetical gene (PA14_36620) and glgX
encode enzymes with a putative role in α-1,4-linked glucan synthesis (glgA
) and degradation (malQ
), that could serve as precursors for trehalose synthesis. In addition to treS
, the treS
operon contains a predicted α-amylase (PA14_36740), and glgB
(PA14_36710), a predicted α-1,4-branching enzyme ().
Annotation of a 42.23 kb region of the P. aeruginosa PA14 genome encoding 38 genes (PA14_36375–36830) and schematic representation of transposon and deletion mutants used in this study.
The 42.23 kb PA14_36375–36830 region containing 38 genes is highly conserved among several sequenced P. aeruginosa
strains that were examined and the treYZ
operons are conserved among pseudomonads in general (Table S2
We utilized a previously constructed non-redundant PA14 transposon insertion mutant library 
to determine whether particular PA14 genes in the 38-gene region promote pathogenesis in Arabidopsis. Among 16 transposon insertions in 16 different genes that were available in the library, two were significantly attenuated in virulence. These mutants, with insertions in glgA
, exhibited a decrease in virulence of 20 and 16 fold, respectively, as measured by in planta
growth (Table S1
are the first two genes in the treYZ
operon, pointing to an important role for trehalose in the infectious process.
To further investigate whether the trehalose operons and/or other genes in the 38-gene cluster are required for virulence, we constructed an in-frame deletion of the entire 42.23 kb region (referred to hereafter as Δ42
) by homologous recombination. In contrast to insertions in glgA
, which exhibited at most a 20 fold decrease in growth compared to wild-type, the Δ42
mutant exhibited severe attenuation in virulence, affecting growth of PA14 infiltrated into Arabidopsis leaves about 120 fold and preventing the appearance of pathogenic symptoms (). Similar results were obtained with four independently constructed Δ42
mutants (data not shown), demonstrating that the non-pathogenic phenotype was caused by the deletion of the 42.23 kb region. Importantly, the Δ42
mutant does not appear to be slow growing or to be generally deficient in a variety of phenotypes associated with virulence in P. aeruginosa
. The Δ42
deletion mutant was not auxotrophic, grew at the same rate as wild-type PA14 in a variety of minimal and rich media, and had no observable phenotypes with respect to the production of pyocyanin (Figure S3
), motility, or biofilm formation (Table S3
), and similar results were obtained with a second independently-constructed Δ42
mutant (Figure S3
; Table S3
). Because independently-constructed Δ42
mutants exhibited the same phenotypes, one of the Δ42
mutants was chosen for subsequent experiments.
Attenuation of Δ42 and trehalose biosynthetic mutants in Arabidopsis leaves and suppression of attenuation of trehalose mutants with exogenous trehalose.
We next constructed several smaller deletions within the 42 kb region to determine which of the 38 encoded genes are primarily responsible for the severe avirulent phenotype of Δ42: ΔPA14_36375-36560 (sub-region I) contains a deletion of the cellulase/peptidase operon and several adjacent genes, and ΔPA14_36570-36700 (sub-region II) and ΔPA14_36710-36830 (sub-region III) contain deletions of the treYZ and treS genes, respectively, including some neighboring genes ().
Deletion of sub-region I that includes the putative cellulase/peptidase gene had a modest 3.3 fold reduction in virulence. In contrast, deletion of sub-region II that contains the treYZ
operon had a much more significant effect on virulence (28.7 fold decrease in growth; Figure S4
), whereas deletion of sub-region III that contains the treS
operon caused a 5.9 fold decrease in growth (Figure S4
). These experiments suggested that the treYZ
operons play a significant role in PA14 pathogenesis in Arabidopsis.
To corroborate the involvement of the trehalose genes in plant pathogenesis we constructed ΔPA14_36570-36630
) and ΔPA14_36710-36740
) containing deletions of only the two putative operons containing the treYZ
genes, respectively, and ΔPA14_36570-36630
) containing deletions of both of the trehalose biosynthetic operons (). Deleting either the putative treYZ
or the treS
operons () had approximately the same effects as deleting the more extensive corresponding subregions II or III, respectively (Figure S4
), and deleting both trehalose operons resulted in an approximately 50 fold decrease in virulence compared to the approximate 120 fold decrease observed with the Δ42
mutant (). These data show that the treYZ
operons play a key role in pathogenesis in Arabidopsis leaves, but that genes in the 42 kb region in addition to those involved in trehalose biosynthesis also play a role in plant pathogenesis.
Further evidence suggesting an important role for trehalose biosynthesis in plant pathogenesis was obtained by measuring the levels of trehalose synthesized in vitro
by PA14 wild-type and trehalose biosynthetic mutants. While wild-type PA14 synthesized readily detectable levels of trehalose, there was approximately 50% less trehalose in the ΔtreS
mutant, and there were undetectable levels of trehalose in the glgA
, and Δ42
mutants (). These data show that the treYZ
operons encode enzymes involved in trehalose biosynthesis. These data also suggest that treS
operon may be dependent on treYZ
for trehalose production, as reported previously 
. When we compared the levels of trehalose synthesized in vitro
() and the extent of growth of the various strains in Arabidopsis leaves (; Table S1
), we found an excellent positive correlation coefficient (R2
Trehalose levels in the trehalose mutants.
Importantly, we found that co-infiltration of the PA14 trehalose mutants and pure trehalose essentially completely suppressed the avirulent phenotypes of the ΔtreYZ, ΔtreS, and ΔtreYZΔtreS mutants and mostly suppressed the phenotype of the Δ42 mutant (). However, 0.25 mg/ml trehalose also rescued the Δ42 mutant almost as well as 2.5 mg/ml, and 0.025 mg/ml trehalose partially suppressed the growth defect of the Δ42 mutant (). These data indicated a requirement for trehalose for PA14 virulence in planta, potentially at physiologically relevant concentrations.
In summary, the data in this section shows that the ΔtreYZ, ΔtreS, and ΔtreYZΔtreS mutants are less virulent in planta, that they either synthesize undetectable (ΔtreYZ and ΔtreYZΔtreS) or reduced (ΔtreS) levels of trehalose, that their level of virulence positively correlates with the level of trehalose they synthesize, and that their reduced virulence phenotype can be suppressed by exogenous trehalose. These data demonstrate that the virulence deficient phenotypes of the ΔtreYZ, ΔtreS, and ΔtreYZΔtreS mutants are a consequence of the inability of these strains to synthesize trehalose, thereby correlating the genotype of these mutants with their avirulent phenotypes.
Specific Arabidopsis cell wall mutants suppress the phenotype of PA14 trehalose mutants
Since a major difference between plant and animals cells is the plant cellulosic cell wall, we reasoned that trehalose may function in a process that involves the plant cell wall. Because PA14 infection in Arabidopsis leaves causes extensive degradation of mesophyll cell walls 
, we first investigated the possibility that trehalose enhances the activity of cell wall degrading enzymes (CWDEs). We tested whether trehalose enhanced the activity of a variety of commercial CWDEs to hydrolyze partially purified Arabidopsis cell walls in vitro
to generate reducing sugars, which were measured using the Somogyi-Nelson assay 
. However, we were not able to conclusively demonstrate that trehalose enhanced the activity of the CWDEs tested (data not shown).
We next reasoned that if trehalose interacts with the plant cell wall, specific Arabidopsis cell wall mutants might suppress the phenotype of the ΔtreYZΔtreS
mutant. We tested the growth of wild-type PA14 and the ΔtreYZΔtreS
mutant in several Arabidopsis cell wall mutants involved in xyloglucan (mur2-1
), arabinose (mur4-1
), or cellulose (mur10-2
) synthesis. Remarkably, the ΔtreYZΔtreS
mutant grew to the same titer as wild-type PA14 in an xxt1/xxt2
double mutant that completely lacks xyloglucan in its cell walls and in a mur4-1
mutant that has decreased levels of arabinose in pectins, xylans, and xyloglucans 
(). Similar results were obtained with the Δ42
mutant; i.e., the Arabidopsis xxt1/xxt2
mutant completely suppressed and the mur4-1
mutant mostly suppressed the avirulent phenotype of the Δ42
mutant (Figure S6
The in planta growth defect of the PA14 ΔtreYZΔtreS mutant in Arabidopsis is suppressed by cell wall mutations.
We ruled out the possibility that the Arabidopsis cell wall mutants suppress the avirulent phenotype of the PA14 trehalose mutants simply because they are generally more susceptible to pathogen attack. As shown in , the cell wall mutants did not exhibit enhanced susceptibility to the P. syringae
strain DC3000, a well-studied bona fide
Arabidopsis pathogen. The Arabidopsis cell wall mutants were also not more susceptible to a DC3000 hrcC
mutant (), which is greatly impaired in virulence, or to the bean pathogen P. syringae
strain 3121 (), which is not normally pathogenic in Arabidopsis. Consistent with these data, we also showed that the xxt1/xxt2
mutant, which exhibits the most severe cell wall defect of the Arabidopsis mutants tested, mounts a normal defense response when challenged with the flagellin peptide flg22 (). Flg22 elicits so-called “pattern triggered immunity” in Arabidopsis. When Arabidopsis leaves are pre-infiltrated with flg22, flg22 exerts a protective effect against subsequent infection with P. syringae
. As shown in , flg22 elicits the same level of protection against P. syringae
DC3000 in xxt1xxt2
plants as in wild-type plants.
Arabidopsis cell wall mutants are not more susceptible to virulent or non-pathogenic P. syringae strains and the xxt1/xxt2 mutant mounts an effective innate immune response.
Ammonium and nitrate but not glucose or sucrose suppress the phenotype of trehalose mutants
As described in the Introduction
, because bacterial plant pathogens primarily replicate in the intercellular spaces in a leaf, they need to acquire nutrients from plant mesophyll cells. We therefore tested whether trehalose may be involved in the acquisition of a variety of nutrient sources including carbon, nitrogen, sulfur and phosphorous. If this were the case, we reasoned that co-infiltration of particular nutrients with the ΔtreYZΔtreS
or the Δ42
mutant would suppress their non-pathogenic phenotypes.
Co-infiltration of the ΔtreYZΔtreS
double mutant with glucose () or co-infiltration of the Δ42
mutant with glucose or sucrose (Figure S7A
) did not rescue the attenuated phenotype in the Arabidopsis leaf assay. These experiments showed that the ΔtreYZΔtreS
or the Δ42
mutant is not limited by carbon. The fact that trehalose but not glucose or sucrose suppressed the phenotype of the Δ42
mutant also shows that the putative cellulase/peptidase and other hypothetical glucanolytic enzymes encoded in the 38 gene region deleted in the Δ42
mutant do not play a critical role in supplying a carbon source to PA14.
The in planta growth defect of the ΔtreYZΔtreS double mutant is suppressed by ammonium or nitrate ions.
We also entertained the possibility that PA14 could accumulate trehalose as a storage sugar, analogous to glycogen or starch, and then hydrolyze trehalose using the enzyme trehalase (PA14_33450, treA
) and utilize the resulting glucose as a carbon source, thereby promoting virulence. We ruled out this possibility, however, by showing that co-infiltration of a double Δ42treA::MAR2xT7
mutant (which cannot metabolize trehalose) with trehalose rescued the non-pathogenic phenotype similarly as co-infiltration of the Δ42
mutant with trehalose (Figure S7B
). We also confirmed that the Δ42treA::MAR2xT7
cannot metabolize trehalose and utilize it as a carbon source (see Materials and Methods
Finally, we tested various salts to determine whether they would suppress the phenotypes of the ΔtreYZΔtreS
() or the Δ42
mutant (Figure S8
). Interestingly, ammonium and nitrate ions almost completely suppressed the lack of growth phenotype of the ΔtreYZΔtreS
() or the Δ42
mutant (Figure S8
), whereas sulfates and phosphates did not have a significant effect.
The data in this section suggest that trehalose enhances access to nitrogen sources during an Arabidopsis infection. An alternative model is that ammonium nitrate (as well as trehalose) suppresses the avirulent phenotype of the PA14 trehalose mutants by suppressing the plant defense response. To test this possibility, we tested whether infiltration of leaves with trehalose or ammonium nitrate resulted in enhanced susceptibility to P. syringae DC3000 (), the DC3000 hrcC mutant (), or P. syringae pv. phaseolicola strain 3121 (); however, neither trehalose nor ammonium nitrate increased the susceptibility to any of these strains. Moreover, infiltration of trehalose or ammonium nitrate did not block the ability of flg22 to elicit protection against infection by P. syringae DC3000 ().
Trehalose or ammonium nitrate does not suppress the Arabidopsis flg22-mediated defense response and does not make Arabidopsis more susceptible to non-pathogenic P. syringae strains.
Trehalose does not appear to function as a stress-response molecule either in vivo or in vitro
Trehalose is well-studied as a so-called compatible solute, which is defined as a molecule that functions as an osmolyte and helps an organism survive osmotic stress. We therefore tested whether other di- and trisaccharide compatible solutes would suppress the avirulent phenotype of the Δ42
mutant. Indeed, as shown in Figure S9
, both maltose and maltotriose functioned similarly to trehalose in allowing the Δ42
mutant to grow in planta
, albeit somewhat less efficiently than did trehalose.
Given these results, we next considered the hypothesis that trehalose enhances the virulence of PA14 by ameliorating a variety of environmental stresses 
. However, the Δ42
mutant was not more susceptible than wild-type PA14 to osmotic stress in response to 0.5 M NaCl (). As a positive control for the osmotic stress experiment, we constructed an in-frame deletion of a predicted (http://www.pseudomonas.com
) three-gene operon (PA14_19350-19370) responsible for the synthesis of a major organic osmoprotectant in P. aeruginosa
-acetylglutaminylglutamine amide (NAGGN) 
. As expected, the ΔPA14_19350-19370
) was more susceptible to 0.5 M NaCl than wild-type PA14 or the Δ42
The Δ42 mutant is not more susceptible to osmotic or oxidative stress.
We further tested whether trehalose functions to protect PA14 from osmotic stress in vitro
by comparing its ability to enhance growth in minimal medium supplemented with 0.5 M NaCl compared to the well-studied osmoprotectant molecule betaine 
. In vitro
, betaine rescued the growth of PA14, Δ42
, and the ΔNAGGN
mutant in 0.5 M NaCl whereas trehalose had no effect (Figure S10A
). We also tested whether betaine would rescue the Δ42
mutant for in planta
growth, similarly to trehalose. However, as shown in , betaine had no significant effect in rescuing Δ42
growth in planta
, showing that the ability of trehalose to rescue Δ42 in planta
is not likely due to the fact that it is functioning to protect Δ42
from osmotic stress. In contrast to Δ42
, the ΔNAGGN
mutant, which is very susceptible to osmotic stress in vitro
, had no significant impairment in growth in planta
). These data show that the Δ42
mutant is not highly susceptible to osmotic stress and that trehalose does not play a major role as an osmoprotectant in PA14.
As an alternative to functioning as an osmoprotectant, we investigated whether trehalose protects PA14 from reactive oxygen-mediated stress generated as a consequence of the plant innate immune response. However, we found no significant difference between the Δ42
mutant and wild-type PA14 with respect to tolerance to paraquat or hydrogen peroxide (, respectively). Because a P. aeruginosa zwf
mutant has been reported to be hyper-sensitive to paraquat-mediated killing 
, we also tested a PA14 zwf::MAR2xT7
as a positive control for determining the sensitivity of PA14 and Δ42
to paraquat. As shown in , the zwf
mutant exhibited enhanced susceptibility to paraquat in vitro
, but did not exhibit an impaired growth phenotype in planta
). These data show that it is unlikely that trehalose functions to protect PA14 from oxidative stress.
In addition to oxidative and osmotic stress, we also tested whether the Δ42
mutant is susceptible to pH or temperature stress, displayed a defect in biofilm formation under osmotic stress, or was deficient in the generation of persister cells in the presence of antibiotics. However, wild-type PA14 and the Δ42
mutant were indistinguishable in all of these tests (Figure S10B