The quormone degradation system of Agrobacterium tumefaciens is regulated by starvation signal and stress alarmone (p)ppGpp


  • Hai-Bao Zhang,

    1. Institute of Molecular and Cell Biology, 30 Medical Drive, Singapore 117609.
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Chao Wang,

    1. Institute of Molecular and Cell Biology, 30 Medical Drive, Singapore 117609.
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Lian-Hui Zhang

    Corresponding author
    1. Institute of Molecular and Cell Biology, 30 Medical Drive, Singapore 117609.
    2. Department of Biological Sciences, The National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260.
    Search for more papers by this author


A unique signal degradation system has recently been discovered in Agrobacterium tumefaciens. Upon entering stationary phase, A. tumefaciens terminates quorum sensing-dependent Ti-plasmid conjugation by degradation of acyl homoserine lactone (AHL) quormone via the enzyme AttM (AHL-lactonase). attM, together with attK and attL, constitute one transcriptional unit subjected to the control of a common promoter. AttJ, the other member of the signal degradation system, is an IclR-like negative transcriptional factor, which tightly represses the expression of AttM at the early stage of bacterial growth. In this study, we found that this quormone degradation system is activated by either carbon or nitrogen starvation. Quormone degradation was significantly delayed when bacterial culture was supplemented with extra carbon or nitrogen source in the nutrient-limited minimal medium before the onset of stationary phase. To identify the signalling pathway and regulatory mechanisms that mediate quormone degradation, we constructed a reporter strain A6(attKLM::lacZ) in which the promoterless lacZ was transcriptionally fused to the attKLM promoter. Transposon mutagenesis of strain A6(attKLM::lacZ) led to identification of the relA gene, which encodes the stress alarmone (p)ppGpp synthetase. Tn5 knock-out of relA abolished the stationary phase-dependent expression of attM. We concluded that the A. tumefaciens quormone degradation system is coupled to and regulated by the generic (p)ppGpp stress response machinery.


Quorum sensing is a bacterial community behaviour to sense the change in bacterial population density and co-ordinate different biological functions. Studies in the past few years have greatly advanced our understanding of the molecular mechanism of bacterial quorum sensing (Fuqua et al., 2001; Miller and Bassler, 2001). Bacterial cells in a prequorum population produce a basal level of quorum-sensing signal (quormone), a good example of which is the N-acyl-homoserine lactones (AHLs) identified in many Gram-negative proteobacterial species (Eberhard et al., 1981; Cao and Meighen, 1989; Jones et al., 1993; Zhang et al., 1993; Pearson et al., 1994). AHL quormone accumulates as bacterial cells proliferate. Once a threshold concentration is reached, AHL interacts with its cognate transcription factor, usually a member of the LuxR-type proteins, to form a complex that triggers the expression of target genes (Zhu and Winans, 1999; Qin et al., 2000; Welch et al., 2000). Evidence is now accumulating that bacterial cells could switch off the quorum-sensing machinery by degradation of quormone signals in response to a change in growth (Zhang et al., 2002). Both non-enzymatic and genetically controlled quormone degradation systems have been reported recently (Byers et al., 2002; Zhang et al., 2002). Interestingly, the two groups of AHL-degrading enzymes, i.e. AHL-lactonase and AHL-acylase, were initially identified in a Bacillus sp. isolate and a Variovorax paradoxus species, respectively, that do not produce AHL signals (Dong et al., 2000; Leadbetter and Greenberg, 2000). It was shown that V. paradoxus could use AHL signals as the sole source of energy and nitrogen (Leadbetter and Greenberg, 2000).

Quormone N-3-oxo-octanoyl homoserine lactone (3OC8HSL), originally known as conjugation factor, regulates Ti-plasmid conjugal transfer in A. tumefaciens (Zhang and Kerr, 1991; Piper et al., 1993; Zhang et al., 1993). In combination with the cognate transcription factor TraR, 3OC8HSL positively regulates the expression of genes required for Ti-plasmid transfer (Tra genes) (Piper et al., 1993; Hwang et al., 1994; Zhu and Winans, 1999; Qin et al., 2000). However, the efficiency of Ti-plasmid conjugal transfer decreases rapidly after bacterial cells enter stationary phase (Tempéet al., 1978). Coincidentally, 3OC8HSL concentration also declines rapidly at stationary phase (Zhang et al., 2002). Molecular and biochemical analyses demonstrated that the AHL-lactonase encoded by attM is expressed and degrades 3OC8HSL at stationary phase. The attM gene is negatively regulated by the IclR-type transcription factor AttJ. When attJ was knocked out in A. tumefaciens strain A6 by transposon mutagenesis, AHL-lactonase became constitutively expressed, which resulted in completely deficient phenotypes in 3OC8HSL accumulation (detected on solid agar plates) and Ti-plasmid conjugal transfer (Zhang et al., 2002). It was shown that attM, together with attK and attL that encode putative succinic semi-aldehyde dehydrogenase and alcohol dehydrogenase, respectively, constitute one operon subjected to the control of the common promoter located upstream of attK. In vitro analysis showed that AttJ specifically binds to the PattKLM promoter that directs AHL-lactonase expression (Zhang et al., 2002). However, it is not clear what causes the release of AttJ repression when bacterial cells enter stationary phase.

In the present work, we intended to identify the unknown factors that are involved in the regulation and induction of the quorum-sensing signal degradation system in A. tumefaciens. We showed that AHL quormone degradation in A. tumefaciens was induced by either carbon or nitrogen starvation. We demonstrated that the stress alarmone (p)ppGpp was involved in the regulation of attM expression. Tn5 mutagenesis of the relA gene that encodes the (p)ppGpp synthetase abolished the stationary phase-dependent expression of AHL-lactonase. Furthermore, we investigated the expression patterns of the negative transcription factor AttJ in the relA knock-out background to probe the molecular mechanism of (p)ppGpp in the regulation of quormone signal degradation in A. tumefaciens.


3OC8HSL quormone signal degradation is switched on at stationary phase

We have reported previously that both wild-type octopine-type A. tumefaciens strain A6 and conjugation constitutive strain K588 displayed a similar growth phase-dependent expression pattern of the attM-encoded AHL-lactonase (Zhang et al., 2002). K588, with pTiB6S3Trac in the genetic background of C58C1, is constitutive in 3OC8HSL production, and the quormone is detectable in liquid culture, whereas strain A6 failed to produce detectable 3OC8HSL in liquid culture even in the presence of octopine inducer. K588 was therefore used in this part of the study for the detection of 3OC8HSL. The relationship between the growth of K588 and 3OC8HSL accumulation was analysed carefully. In the defined minimal medium, the bacterial cells grew exponentially for about 18 h at 28°C with shaking (200 r.p.m.) and then entered stationary phase (Fig. 1A). Similarly, accumulation of 3OC8HSL in the liquid culture was maximized at 18 h and then declined sharply (Fig. 1A). Western blotting analysis showed that AttM expression was significantly induced upon bacterial culture entering stationary phase, but the enzyme was hardly detectable at exponential growth phase (Fig. 1B). This pattern parallels the enzymatic activity profile of AttM (Zhang et al., 2002), indicating that it is the elevated expression of AttM that contributes to 3OC8HSL signal degradation at stationary phase.

Figure 1.

Activation of A. tumefaciens quorum-sensing signal (3OC8HSL) degradation system when bacterial cells enter stationary growth phase.
A. Time course of population density (open circles) and 3OC8HSL accumulation (filled squares) in bacterial culture of A. tumefaciens strain K588.
B. Western blotting detection of AttJ and AttM expression at different growth stages of K588.

AttJ is a negative transcription factor involved in A. tumefaciens quormone degradation, and transposon insertion within the attJ coding region resulted in constitutive expression of AHL-lactonase encoded by attM (Zhang et al., 2002). The Western blot analysis demonstrated that AttJ is expressed at a constitutive level throughout the growth phases, contrasting with the growth phase-dependent expression pattern of AttM (Fig. 1B). This expression pattern precluded the possibility that the induced expression of AttM at stationary phase is the result of a changed expression pattern of AttJ, suggesting that there might be a factor(s) that releases AttJ repression of attM and hence switches on the quormone degradation system at stationary phase.

Quormone 3OC8HSL degradation is induced by starvation

Upon entering stationary growth phase, bacteria encounter a range of environmental and physiological stresses, and an obvious one is nutrient deprivation (Kolter et al., 1993). When growing in the defined BM minimal medium containing 0.2% mannitol and 0.2% ammonia sulphate as sole carbon and sole nitrogen sources, respectively, the bacterial population density maximized at 20 h after inoculation (OD600 = 1.63) and then declined slowly (Fig. 2A). To determine whether carbon source is the limiting factor in BM minimal medium, we added an additional 0.2% of mannitol to the bacterial culture at 18 h after inoculation. We found that the bacterial population density continued to increase and maximized at 28 h (OD600 = 2.05) (Fig. 2A). We then analysed the concentration of 3OC8HSL in bacterial cultures with and without supplementation of mannitol. As shown in Fig. 2B, the addition of supplementary carbon source sustained the 3OC8HSL accumulation until 24–28 h after inoculation, whereas in the unsupplemented control culture, degradation of quorum-sensing signal started at 18 h after inoculation. We also analysed the AHL-lactonase (AttM) enzyme activity in the bacterial culture supplemented with mannitol. The results in Fig. 2C show that AttM enzyme activity in the culture with mannitol supplementation was maintained at a very low level until the stationary phase occurred at 28 h, suggesting that the accumulation of 3OC8HSL results from the delayed expression of AttM. We also tested glucose and sucrose separately as the supplementary carbon source and obtained similar results (data not shown). The data indicate that carbon source starvation is a signal that triggers 3OC8HSL degradation at stationary phase.

Figure 2.

Effect of supplementary carbon source on 3OC8HSL degradation at stationary phase. A. tumefaciens strain K588 culture grown in carbon source-limited minimal medium (0.2% mannitol and 0.2% ammonia sulphate) was supplemented with mannitol to double its concentration at late log phase (18 h after inoculation).
A. Bacterial population densities (OD600) after the addition of supplementary carbon source (filled circles) and the control without supplementation (open circles).
B. 3OC8HSL accumulation in bacterial cultures with carbon source supplemented (filled squares) and the control without supplementation (open squares).
C. AHL-lactonase activities in bacterial cultures with carbon source supplemented (filled triangles) and the control without supplementation (open triangles).

To test whether nitrogen source starvation could also activate 3OC8HSL signal degradation, we prepared the nitrogen-limiting but carbon-sufficient BM minimal medium containing 0.05% ammonia sulphate and 0.6% mannitol. We monitored the bacterial population density and 3OC8HSL accumulation after supplementation of ammonia sulphate or ammonia chloride (0.2% final concentration) at the onset of stationary phase (18 h after inoculation). Similar to carbon source supplementation (Fig. 2), the bacterial population density in the supplemented medium continued to increase until 28 h, whereas in the control culture without a supplementary nitrogen source, bacterial density maximized at 20 h, then declined slowly as a result of bacterial cell lysis (data not shown). Accordingly, 3OC8HSL degradation was also delayed when nitrogen source was supplemented, because of the postponed expression of AHL-lactonase (data not shown).

Screening and cloning of the genes involved in the regulation of AttM expression in A. tumefaciens

To identify the genetic components that control growth phase-dependent and starvation-induced attM expression, we constructed a reporter strain A6(attKLM::lacZ), in which the promoterless lacZ gene was transcriptionally fused to the attM promoter in order to monitor attM expression (Experimental procedures). Strain A6 was selected for genetic analysis because it is a wild-type strain showing the same pattern of AttM expression as K588. As shown in the next section (Fig. 4A), the expression pattern of β-galactosidase of strain A6(attKLM::lacZ) is similar to that of AttM enzyme in strain A6 (Zhang et al., 2002), suggesting that the lacZ expression in A6(attKLM::lacZ) represents the expression pattern of attM. Using this reporter strain, we then resorted to a Tn5 mutagenesis approach to identify the mutants showing altered attM expression patterns. From around 20 000 Tn5 insertion mutants of A6(attKLM::lacZ), five mutants were obtained showing decreased levels of attM expression on BM agar plates containing Xgal. Sequencing analysis of one randomly selected mutant found that Tn5 insertion disrupts the relA/spoT homologue (relAatu6) of Escherichia coli. The exact insertion site of Tn5 was located at the 1998th basepair of the 2232 bp coding sequence of relAatu6 (GenBank accession no. AY444344). The relAatu6 gene encodes a peptide of 744 amino acids, with a predicted molecular mass of 83 804 Da. The predicted peptide contains 264 hydrophobic and 149 polar amino acid residues, with an isoelectric point at 6.19.

Figure 4.

Analysis of attM transcription and AHL-lactonase (AttM) activity in relA knock-out strains.
A. The growth curves (open symbols) and the β-galactosidase activities (filled symbols) of strains A6(attKLM::lacZ) (squares), A6(attKLM::lacZ, relA::Tn5) (circles) and A6(attKLM::lacZ, relA::Tn5, pLA-relA) (triangles).
B. RT-PCR analyses of relA transcriptional pattern in A6 (lanes 1 and 4), A6(relA::Kanr) (lanes 2 and 5) and A6(relA::Kanr, pLA-relA) (lanes 3 and 6). In lanes 1–3, the RNA samples were prepared from the cell cultures 12 h after inoculation (exponential phase, OD600≈ 0.85), while the samples used in lanes 4–6 were from the cultures 22 h after inoculation (stationary phase, OD600≈ 1.57).
C. AHL-lactonase (AttM) activities of strains A6(relA::Kanr), A6(relA::Kanr, pLA-relA) and A6.

Sequence comparison by blast ( showed that the relAatu6 gene of octopine-type A. tumefaciens A6 is similar but not identical to the sole counterpart of the nopaline-type strain C58 (relAatu58) with 91% nucleotide and 98% amino acid identity (Goodner et al., 2001; Wood et al., 2001) respectively. Genome sequencing revealed that A. tumefaciens C58 contains one circular and one linear chromosome, as well as two plasmids, i.e. pAT and pTi (Goodner et al., 2001; Wood et al., 2001). Whereas attJ and attM are carried by pAT, relAatu58 is located in the circular chromosome. RelAatu6 shows 31% and 36% polypeptide identity, respectively, to the RelA and SpoT of E. coli. Both RelA and SpoT catalyse (p)ppGpp synthesis, except that SpoT is a bifunctional enzyme with (p)ppGpp hydrolase activity (Heinemeyer et al., 1978; Xiao et al., 1991).

RelAatu6 is responsible for (p)ppGpp synthesis in A. tumefaciens

Guanosine tetraphosphate and guanosine pentaphosphate [(p)ppGpp] are known to function as alarmone signals to trigger the stringent stress response of bacterial cells and co-ordinate entry into stationary phase (Chatterji and Ojha, 2001). To assess the role of RelAatu6 of A. tumefaciens in (p)ppGpp biosynthesis and metabolism, we did a domain comparison using the smart program (Schultz et al., 1998; Letunic et al., 2002) with several RelA/SpoT enzymes with function in (p)ppGpp metabolism that have been experimentally proved. These homologues include RelAeco and SpoTeco from Gram-negative E. coli, RelAbsu from Gram-positive model organism Bacillus subtilis and Sj-RSH from eukaryotic plant Suaeda japonica. Figure 3A shows that these enzymes share only a conserved RelA/SpoT domain, which is known to be essential for (p)ppGpp biosynthesis (Gentry and Cashel, 1996). Similar to SpoTeco of E. coli and RelAbsu of B. subtilis, the RelA homologues of A. tumefaciens strains A6 and C58 also contain three other conserved domains, i.e. HD, TGS and ACT (Fig. 3A). The HD, designated because of its conserved catalytic residues histidine (H) and aspartate (D), was found in a superfamily of enzymes with known or predicted phosphohydrolase activity (Aravind and Koonin, 1998). In agreement with its sole function as a (p)ppGpp synthetase, the typical HD motif was not found in RelAeco of E. coli because the catalytically important residues H and D were replaced by phenylalanine (F) and proline (P) (Fig. 3A and B). The RelAatu6 and RelAatu58 of A. tumefaciens appear to maintain an intact HD domain with conserved HD catalytic residues similar to the SpoTeco of E. coli (Fig. 3B). The TGS domain is probably a nucleotide-binding region, and the ACT domain could be a regulatory domain (Schreiber et al., 1991). These in silico analyses suggested that RelAatu6 is likely to be a (p)ppGpp synthetase but might also have phosphohydrolase activity.

Figure 3.

In silico and biochemical analysis of RelAatu6.
A. Domain comparison of RelAatu6 with RelA/SpoT homologues. The RelA homologues from A. tumefaciens A6 (RelAatu6, GenBank accession no. AY444344); A. tumefaciensC58 (RelAatu58, GenBank accession no. NP531727); E. coli (RelAeco, GenBank accession no. J04039); B . subtilis (RelAbsu, GenBank accession no. U86377); S. japonica (Sj-RSH; GenBank accession no. AB079577); and SpoT from E. coli (SpoTeco, GenBank accession no.M24503) were analysed separately using the smart program and aligned manually. The designated name of each domain is indicated at the top of RelAatu6. The open rectangle represents no typical HD domain found.
B. Alignment of HD domains of the prokaryotic and eukaryotic homologues. ‘*’, identical residues, ‘:’, conserved substitutions; ‘dot’, semi-conserved substitutions. The conserved histidine (H) and aspartate (D) were boxed.
C. Thin-layer chromatography analysis of (p)ppGpp production. The samples in lines 1–6 were prepared from E. coliCF3120[(p)ppGpp production-positive control] (Cashel, 1994), E. coliCF1693[(p)ppGpp-free negative control] (Cashel, 1994), A. tumefaciens A6, A6(attKLM::lacZ), A6(attKLM::lacZ, relA::Tn5) and A6(attKLM::lacZ, relA::Tn5, pLA-relA) respectively.

Using two E. coli strains that are positive and defective, respectively, in (p)ppGpp biosynthesis as experimental controls (Table 1) (Cashel, 1994), we determined the intracellular (p)ppGpp accumulation in several A. tumefaciens strains. As shown in Fig. 3C, (p)ppGpp was detected in the wild-type A6 and A6(attKLM::lacZ) under carbon starvation conditions. When relAatu6 was knocked out, no visible (p)ppGpp was observed. We then tested whether the wild-type relAatu6 cloned from A. tumefaciens strain A6 could complement the mutant A6(attKLM::lacZ, relA::Tn5). The construct pLA-relA was generated by placing the relAatu6 gene under the control of the Ptac promoter in vector pLAFR3. Figure 3C shows that the (p)ppGpp-deficient phenotype was fully restored when the construct pLA-relA was used to complement the relAatu6 mutant. These data demonstrate that RelAatu6 is responsible for (p)ppGpp synthesis in A. tumefaciens, albeit that it has not been confirmed whether RelAatu6 could also catalyse (p)ppGpp degradation.

Table 1. . Bacterial strains and plasmids used in this study.
Strain or plasmidRelevant characteristicsaSource or reference
  • a.

    Trac, constitutive in 3OC8HSL production and Ti-plasmid conjugal transfer; Ampr, ampicillin resistant; Kanr, kanamycin resistant; Tetr, tetracycline resistant; Gmr, gentamicin resistant; Chlr, chloramphenicol resistant.

A. tumefaciens
NT1(traR, tra::lacZ749C58C1RS harbouring pJM749 and pSVB33, AHL bioassay indicator strain Piper et al. (1993)
A6Wild-type octopine strain containing pTiA6A. Kerr
A6 (attKLM::lacZ) attKLM transcriptionally fused with Tn3HoHo1-lacZ, AmprThis study
A6 (attKLM::lacZ, relA::Tn5 relA inserted by Tn5 at the attKLM::lacZ genetic background of strain A6, Ampr, KanrThis study
A6 (attKLM::lacZ, pLA-relA relA overexpressed at the attKLM::lacZ genetic background of strain A6, Ampr, TetrThis study
A6 (relA::Kanr) relA disrupted by Kanr fragment in strain A6, KanrThis study
A6 (attJ::lacZ) attJ transcriptionally fused to Tn3HoHo1-lacZ, AmprThis study
A6 (attJ::lacZ, relA:: Kanr relA inserted by Kanr fragment at the attJ::lacZ genetic background of strain A6, Kanr AmprThis study
K588 C58C1 harbouring octopine-type Ti plasmid pTiB6S3TracA. Kerr
E. coli
DH5α supE44ΔlacU169(Φ80lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1Laboratory collection
HB101(pRK2013) Thr leu thi recA hsdR hsdM pro, KanrLaboratory collection
HB101(pHoHo1, pSSHe) Thr leu thi recA hsdR hsdM pro, Tn3HoHo1 maintaining strain, Ampr, Chlr Stachel et al. (1985)
BW020767(pRL27)Harbouring Tn5 for A. tumefaciens mutagenesis Larsen et al. (2002)
S17-1pro res mod+ integrated copy of RP4, mob+A. Kerr
CF1693 ΔrelA251ΔspoT207 Cashel (1994)
CF3120 PALS10(Ptac::relA), Ampr Cashel (1994)
pLAFR3IncP, broad-host-range cosmid vector, Tetr Staskawicz et al. (1987)
pPH1JIIncP, used to force homogenotes, Gmr Beringer et al. (1978)
pLA-A6N4A cosmid clone containing attJ and attKLM gene clusters in pLAFR3, TetrThis study
pLA-A6N4-1Derivative of pLA-A6N4, attJ transcriptionally fused with Tn3HoHo1-lacZ inside its coding sequence, Tetr, AmprThis study
pLA-A6N4-2Derivative of pLA-A6N4, attKLM transcriptionally fused with Tn3HoHo1-lacZ, Tetr, AmprThis study
pLA-relApLAFR3 harbouring the full-length relA gene from A6, TetrThis study
pLA-relA:: KanrDerivative of pLA-relA, relA gene disrupted by Kanr fragment, Kanr, TetrThis study
pRL-relA::Tn5A plasmid rescued from strain A6(attKLM::lacZ)(relA::Tn5) by self-ligation, KanrThis study

The relAatu6 gene is involved in the regulation of attM transcription

To assess the impact of RelAatu6 on attM expression, we quantitatively determined the dynamic change in β-galactosidase activity in strains with and without a functional RelA. Null mutation of RelAatu6 did not affect the normal growth of the bacteria (Fig. 4A), but has a significant impact on attM expression pattern. Figure 4A shows that attM expression in the relA mutant A6(attKLM::lacZ, relA::Tn5) remained flat at a basal level over the entire growth phases, in sharp contrast to the clear stationary phase-dependent expression pattern of its parental strain A6(attKLM::lacZ). We also tested whether the construct pLA-relA, which expresses RelAatu6 constitutively, could restore attM expression. As shown in Fig. 4A, β-galactosidase activity of the transformant A6(attKLM::lacZ, relA::Tn5, pLA-relA) parallels that of the parental strain A6(attKLM::lacZ). Reverse transcription (RT)-PCR results showed that the Ptac promoter directed constitutional transcription of relAatu6 in A. tumefaciens, whereas transcriptional expression of relAatu6 in the wild-type strain A6 occurred only at the stationary phase (Fig. 4B). The enhanced expression level of β-galactosidase in strain A6(attKLM::lacZ, relA::Tn5, pLA-relA) over its parental strain A6(attKLM::lacZ) probably results from the combined influence of relA constitutive expression and the copy number effect of relA carried by vector pLAFR3, which is known to have four or five copies per cell.

We also analysed the effect of (p)ppGpp synthase on AttM translation by generating a relA disruption mutant in the wild-type background of strain A6, designated A6(relA::Kanr), using the marker replacement approach (Experimental procedures). Subsequent analysis showed that the AHL-lactonase activity of A6(relA::Kanr) was kept at a basal level, significantly lower than that of the wild-type strain A6 in the stationary phase (Fig. 4C). As expected, when the construct pLA-relA, which allows constitutive expression of RelAatu6, was introduced into mutant A6(relA::Kanr), the wild-type expression pattern of AHL-lactonase was restored (Fig. 4C). The above experiments demonstrate that the stationary phase-inducible expression of attM requires a functional RelA.

(p)ppGpp promotes AttM expression by counteracting AttJ repression

The above data clearly established that AttJ, as a transcriptional repressor, suppresses AttM expression, whereas (p)ppGpp promotes AttM expression, in particular at the stationary phase. To investigate the molecular mechanism and evaluate the relative strength of AttJ and (p)ppGpp in the regulation of AttM expression, we determined the AHL-lactonase activity of the AttJ null mutant A6(attJ::lacZ), the AttJ and RelA double mutant A6(attJ::lacZ; relA::Kanr) and A6(attJ::lacZ; pLA-relA), which overexpresses RelA in the AttJ null mutation background. The results show that all the mutants that lack a functional AttJ displayed a comparable growth pattern (Fig. 5A) and expressed AHL-lactonase constitutively at high levels (Fig. 5B), regardless whether or not there was a functional RelA. No significant difference in AHL-lactonase activity was observed among the relA knock-out strain A6(attJ::lacZ; relA::Kanr), the relA overexpression strain A6(attJ::lacZ; pLA-relA) and the strain A6(attJ::lacZ) in which the wild-type relA was untouched at the AttJ null mutation background (Fig. 5B), indicating that (p)ppGpp is not required for attM expression in the absence of a functional AttJ repressor. It is also noted that the maximum expression level of AttM in wild-type A6 (<350 pmol mg−1 h−1; Fig. 4C) was slightly lower than that of the attJ knock-out strains (>400; Fig. 5B) at stationary phase. These data suggest that AttJ is the sole transcriptional repressor that (p)ppGpp has to counteract at stationary phase, albeit the detailed mechanism requires further investigation.

Figure 5.

Analysis of the AHL-lactonase (AttM) activity in attJ knock-out strains with and without RelA.
A. The growth curves of strains A6(attJ::lacZ) (open circles), A6(attJ::lacZ, relA::Kanr) (open triangles) and A6(attJ::lacZ, pLA-relA) (open squares).
B. AHL-lactonase (AttM) activities of the corresponding strains.
C. Immunodetection of AttJ in bacterial cultures of strains A6(attKLM::lacZ), A6(attKLM::lacZ, relA::Tn5) and A6(attKLM::lacZ, pLA-relA) at different time points.

(p)ppGpp does not affect AttJ expression

The above result prompted us to check the possibility that (p)ppGpp may regulate AttM biosynthesis by regulating AttJ expression. It is known that accumulation of (p)ppGpp could not only upregulate but also lead to downregulation of a range of genes (Chatterji and Ojha, 2001). Using Western blot, we determined the AttJ protein levels in strain A6(attKLM::lacZ) and its two derivatives, A6(attKLM::lacZ, relA::Tn5) and A6(attKLM::lacZ, pLA-relA), in which relAatu6 was either disrupted or constitutively expressed. As shown in Fig. 5C, no significant difference in AttJ expression was observed at the translational level during the whole growth phases, indicating that (p)ppGpp does not affect AttJ expression.


In this study, using physiological and genetic approaches, we explored the potential environmental and genetic signals that trigger the quormone signal degradation system in A. tumefaciens. We demonstrated that the quorum-sensing signal degradation system in A. tumefaciens is governed by growth phase (Fig. 1), induced by starvation signals (Fig. 2) and mediated by RelAatu6, which synthesizes alarmone (p)ppGpp (Figs 3 and 4). Supplementation of either carbon or nitrogen source postpones the bacterial cells from entry into stationary phase and, accordingly, initiation of quormone signal degradation (Fig. 2). The intracellular alarmone (p)ppGpp appears to play a critical role in the activation of the quormone signal degradation system in A. tumefaciens. Null mutation of relAatu6, which encodes a (p)ppGpp synthase (Fig. 3), abolished growth phase-dependent AttM expression (Fig. 4).

Many microorganisms are known to use (p)ppGpp in global regulation and co-ordination of the stringent response upon entry into stationary phase. Intracellular accumulation of (p)ppGpp leads to rapid inhibition of synthesis of stable RNAs, ribosomes and proteins and, ultimately, to the arrest of cell growth (Cashel et al., 1996). The (p)ppGpp alarmone is also involved in positive regulation of the stationary phase sigma factor RpoS (Gentry et al., 1993), antibiotic production (Hoyt and Jones, 1999), quorum sensing (van Delden et al., 2001) and virulence (Hammer and Swanson, 1999; Taylor et al., 2002; Haralalka et al., 2003). This study showed for the first time that the quorum-sensing signal (quormone) degradation system in A. tumefaciens is also positively activated by the (p)ppGpp-mediated stringent response mechanism.

The available evidence suggests that (p)ppGpp might exert its positive regulatory effect through its direct inhibition of stable RNA (rRNAs, tRNAs) transcription. In E. coli, RNA polymerase (RNAP) is composed of a core enzyme (E) and one of the seven sigma factors (σ). Downregulation of synthesis of stable RNAs is believed to occur through the effect of binding of (p)ppGpp at the interface of the β- and β′-subunits of RNAP (Chatterji et al., 1998). Binding of (p)ppGpp destabilizes Eσ70–promoter complexes and reduces the ability of σ70 to compete with the stress-related sigma factors such as σ32 for the core enzyme E (Jishage et al., 2002). Moreover, downregulation of stable RNA transcription could substantially increase the pool of free RNAP, which stimulates those stress-inducible promoters that are poor in their ability to recruit RNAP (Zhou and Jin, 1998). This proposed mechanism of regulatory control, also known as a passive model (Barker et al., 2001a), matches well with the starvation-inducible and (p)ppGpp-dependent in vivo expression patterns of a range of amino acid synthetases as well as the genes dependent on the stationary phase sigma factor σS (Gentry et al., 1993; Kvint et al., 2000; Barker et al., 2001a,b). Our results show that, in A. tumefaciens, the quormone degradation system couples to the (p)ppGpp stringent response regulatory system through AttJ, an IclR-type transcriptional repressor. Mutation of attJ unplugs the connection between AHL expression and the (p)ppGpp regulatory circuit (Fig. 5). Therefore, in wild-type A. tumefaciens, (p)ppGpp might somehow diminish the repression of the negative transcription factor AttJ and, hence, result in upregulation of attM transcription upon entry into stationary phase (Fig. 4). In general, the findings may be explained using the above passive model, but coupling of a negative regulator to the (p)ppGpp regulation system has clearly added a new dimension to the above model. The attM promoter appears to be a σ70-dependent promoter with non-identical but conserved −10 (TCTAAT, consensus TATAAT) and −35 (TTGGCC, consensus TTGACA) sequences (Zhang et al., 2002). It is likely that AttJ binding at the early stage of bacterial growth prevents RNAP from accessing the promoter until reaching stationary phase. As expression of AttJ is constitutive throughout the whole growth stages (Figs 1B and 5C), the stationary phase-dependent expression pattern of AttM (Figs 1B and 4C) could result from either enhanced affinity (or availability) of RNAP to compete the attM promoter (PattKLM) against AttJ repressor or decreased stability of the AttJ–PattKLM repressor–promoter complex at stationary phase. At this stage, we cannot rule out either of these two possibilities.

The alarmone (p)ppGpp has been widely known as a stringent response signal, which accumulates when microorganisms sense nutrient shortage and mediates the expression of a range of stress-responsive genes (Chatterji and Ojha, 2001). One of the direct consequences of nutrient deprivation is the rapid increase in free tRNA, also known as uncharged tRNA or deacylated tRNA. Evidence suggests that the presence of ribosome and abundant free tRNA is essential for the biosynthesis of (p)ppGpp (Haseltine and Bock, 1973). The importance of ribosome and free tRNA has been confirmed recently using an in vitro (p)ppGpp production system derived from E. coli, which showed that RelA-mediated (p)ppGpp production is dependent on mRNA, a free tRNA at the A site of ribosome, and the presence of ribosomal protein L11 (Wendrich et al., 2002). These findings might explain why the A. tumefaciens strain expressing relA constitutively still displayed a clear stationary phase-inducible expression pattern of AttM (Fig. 4), as free tRNAs accumulate only when nutrient shortage occurs.

The biological functions of the putative succinic semi-aldehyde dehydrogenase and alcohol dehydrogenase, encoded by attK and attL, respectively, which share the same operon with attM, have not yet been reported. Given that the A. tumefaciens quormone degradation system is activated by either carbon or nitrogen starvation, it appears reasonable to speculate that the bacterium may be able to metabolize the AttM-degraded product further, and that AttK and AttL may be involved in that process. However, the A. tumefaciens wild-type strains and the E. coli expressing AttK and AttL failed to grow on the medium using N-3-oxo-octanoyl homoserine, the degradation product of AttM, as sole carbon or nitrogen source (C. Wang, unpublished data). Therefore, unlike the previously identified V. paradoxus (Leadbetter and Greenberg, 2000), A. tumefaciens might not be able to reuse AHL signal as the sole energy or nitrogen source, even under starvation conditions. Rather than as a means of carbon and nitrogen source recycling, A. tumefaciens might use this sophisticated signal degradation system for the sole purpose of terminating quorum sensing-dependent conjugation in response to energy and nutritional stress, thus enabling cells to progress from one biological process to a new phase of physiological activities. The quormone signal 3OC8HSL of A. tumefaciens regulates Ti-plasmid conjugal transfer, which requires the expression of dozens of genes for the synthesis of a new strand of Ti-plasmid DNA and to transport it from donor cells to recipient cells (Alt-Mörbe et al., 1996). It is rational that A. tumefaciens evolves such a sensitive quormone degradation system to terminate the quorum sensing-dependent and energy-consuming conjugation process in response to starvation stress. Beyond any doubt, the energy and nutritional gain from termination of the conjugation process would be multimagnitudes higher than what could be obtained by recycling the 3OC8HSL signal, which is accumulated at the low end of the micromolar range (Fig. 1A).

Interestingly, similar to the expression pattern of AHL-lactonase, transcriptional expression of relAatu6 also occurs only at stationary phase (Fig. 4B). The next challenge will be to understand how relAatu6 transcription is regulated, and how (p)ppGpp counteracts AttJ repression and activates attM expression.

Experimental procedures

Bacterial strains, plasmids and growth conditions

The bacterial strains and plasmids used in this study are listed in Table 1. A. tumefaciens strains were grown at 28°C in LB (contains 10 g of Bacto-tryptone, 5 g of yeast extract and 10 g of NaCl per litre, pH 7.0) or in BM minimal medium (basic minimal nutrient with 0.2% mannitol added as sole carbon source and 0.2% ammonia sulphate as sole nitrogen source unless otherwise indicated) (Zhang and Kerr, 1991). E. coli strains were grown at 37°C in LB medium. Antibiotics were added at the following concentrations when required: kanamycin, 50 µg ml−1; tetracycline, 5 µg ml−1 (A. tumefaciens) or 10 µg ml−1 (E. coli); rifampicin, 50 µg ml−1; ampicillin, 100 µg ml−1; spectinomycin, 100 µg ml−1; gentamicin, 50 µg ml−1; chloramphenicol, 35 µg ml−1. Xgal (Promega) was included in the medium at a final concentration of 50 µg ml−1 for the detection of β-galactosidase activity.

DNA manipulation and plasmid construction

Plasmids were purified using the Plasmid Mini-prepare kit as recommended by the manufacturer (Qiagen). PCR product purification and DNA recovery from agarose gel were carried out with a QIAquick PCR purification kit and a QIAquick gel extraction kit (Qiagen) respectively. For the construction of pLA-RelA, the relAatu6 gene from A. tumefaciens A6 was amplified with forward primer 5′-CGGGATCCCATAATCTGA TCCCTATTGA-3′ (BamHI site underlined) and reverse primer 5′-AACTGCAGAATGCCTCAATCGAAGGCAC-3′ (PstI site underlined), then ligated into vector pLAFR3 after BamHI and PstI digestion. DNA sequencing was performed with the ABI PrismTM dRhodamine terminator cycle sequencing ready reaction kit (Perkin-Elmer). Sequence analysis was performed using dnastar programs, and the sequence comparison using the blast program ( Southern hybridization was conducted using the DIG DNA labelling and detection kit according to the manufacturer's protocol (Roche Molecular Biochemicals).

Cosmid library construction of A. tumefaciens strain A6

The cosmid library of A. tumefaciens strain A6 was constructed using the cosmid vector pLAFR3. Total DNA of A6 was partially digested with BamHI, then the 20–30 kb fragments were purified from low-melting-point agarose gel as described previously (Dong et al., 2000). After ligation of the purified DNA fragments with the BamHI-digested vector pLAFR3, the ligation mixture was packaged with GigapackTM III XL packaging extract (Stratagene) and then transfected to E. coli DH5α. About 2000 colonies grown on tetracycline-supplemented medium were selected and kept as the genomic library of strain A6.

Generation of attJ::lacZ and attKLM::lacZ reporter gene fusions

The reporter strains A6(attJ::lacZ) and A6(attKLM::lacZ), in which promoterless lacZ was transcriptionally fused to attJ and attKLM, respectively, were generated using transposon Tn3HoHo1-lacZ as described previously (Stachel et al., 1985). Briefly, the target cosmid pLA-A6N4, which contains the attJ–attKLM region, was introduced into Tn3HoHo1-lacZ harbouring strain E. coli HB101(pHoHo1, pSSHe) by heat shock transformation. Then, all three plasmids from E. coli HB101(pHoHo1, pSSHe, pLA-A6N4) were transferred into mobile strain E. coli S17-1 by triparental mating in the presence of a helper strain E. coli HB101(pRK2013). The pLA-A6N4::Tn3HoHo1-lacZ insertions were obtained by mating S17-1(pHoHo1, pSSHe, pLA-A6N4) with A. tumefaciens strain A6, in which pHoHo1 and pSSHe cannot be maintained. The resultant mutant A6(pLA-A6N4::Tn3HoHo1-lacZ) grown on selective medium was screened by PCR to determine the site and orientation of Tn3HoHo1-lacZ insertion, using primers specific to attJ, attKLM and Tn3 respectively. The exact insertion sites of Tn3HoHo1-lacZ were determined by sequencing the PCR products. Two desired clones with Tn3HoHo1-lacZ transcriptionally fused with attJ and attKLM, respectively, were identified. In clone pLA-A6N4-1, Tn3HoHo1-lacZ was located at the 47th basepair of the 828 bp attJ encoding region and, in the other, pLA-A6N4-2, Tn3HoHo1-lacZ was located at the 491st basepair of the 3473 bp attKLM operon (Zhang et al., 2002). The two plasmids were homogenotized into A6 genome as described previously (Hwang et al., 1995) with pPH1JI as the eviction plasmid (Beringer et al., 1978). The resultant strains, designated A6(attJ::lacZ) and A6(attKLM::lacZ), respectively, were then used to assay attJ and attKLM transcription by determination of β-galactosidase activity.

Agrobacterium tumefaciens transformation and Tn5 transposon mutagenesis

Genetic transformation of A. tumefaciens strain A6 was carried out with the triparental mating method. Briefly, fresh single colonies of E. coli donor strain containing the plasmid of interest, A. tumefaciens recipient strain and helper strain E. coli HB101(pRK2013) were mixed together on an LB agar plate and incubated overnight at 28°C. Then, the bacterial mixture was resuspended in liquid minimal medium and spread onto selective minimal agar medium containing antibiotics. A. tumefaciens colonies grown on selection media were further confirmed by PCR with specific primers.

For Tn5 transposon mutagenesis, the overnight cultures of E. coli BW020767(pRL27) and A6(attKLM::lacZ) were mixed together, plated on an LB agar plate and grown for 5 h at 28°C. Cells were then scraped off the plates, resuspended in sterile water and plated on minimal medium supplemented with kanamycin (100 µg ml−1) and Xgal (50 µg ml−1) agar plates. After incubation for 3–5 days at 28°C, the potential mutant colonies showing obvious lighter or deeper blue colour than the parent strain were selected. The phenotypes of these potential mutants were confirmed by restreaking on Xgal plates three times. The disrupted gene was identified by sequencing the Tn5-flanking fragments.

Generation of the relA mutants in strains A6 and A6(attJ::lacZ)

To knock out the relAatu6 gene in wild-type A6 and the attJ mutant A6(attJ::lacZ) genetic background, relAatu6 carried by construct pLA-relA was mutated by a mini-transposon containing a kanamycin resistance (Kanr) marker using the Template Generation System II kit (Finnzymes). A clone designated pLA-relA::Kanr, in which the Kanr fragment was inserted at the 1208th nucleotide of the 2232 bp relA coding sequence was identified by PCR and sequencing analysis. This clone was then introduced into A6 and A6(attJ::lacZ) by electroporation. The relA::Kanr fragment was then homogenotized into the chromosome to replace the wild-type relA gene with pPH1JI as the eviction plasmid as described above. The resulting strains were designated A6(relA::Kanr) and A6(attJ::lacZ, relA::Kanr) respectively.

Detection of (p)ppGpp accumulation in A. tumefaciens

Measurement of intracellular (p)ppGpp accumulation in A. tumefaciens under carbon source starvation was conducted according to the method of Cashel (1994). Briefly, the overnight cell cultures of E. coli and Agrobacterium species were grown in a phosphate-limited (0.4 mM) MOPS medium containing 0.025% (w/v) dextrose to an initial OD600 of 0.05, and [32P]-H3PO4 was added to a final concentration of 100 µCi ml−1. The bacteria were allowed to grow to an OD600 of 0.2. An aliquot (30 µl) of the cell culture was mixed with an equal volume of 14 M formic acid. The mixture was frozen and thawed, and the procedure was repeated three times. Each sample of 5 µl of cellular extracts was spotted and separated on a PEI-cellulose thin layer chro-matography (TLC)  plate  using  1.5 M  potassium  phosphate  (pH 3.4) as the running solvent. (p)ppGpp was visualized by autoradiography.

β-Galactosidase assay

Quantitative analysis of β-galactosidase activity in bacterial culture of A. tumefaciens strains was conducted according to the method described previously (Stachel et al., 1985). The β-galactosidase activity was measured and expressed as units per 109 cfu.

AHL-lactonase activity assay and quantitative determination of 3OC8HSL

AHL-lactonase activity was analysed as described previously (Zhang et al., 2002). Soluble protein samples were prepared in phosphate buffer (67 mM, pH 7.4) from fresh culture of A. tumefaciens strains. An equal volume (100 µl) of soluble protein samples (1.5 mg ml−1) was mixed with 3OC8HSL (final concentration 5 µM) and incubated at 28°C. An aliquot of reaction mixture was collected at the time indicated and boiled for 2 min to stop the reaction. The concentration of 3OC8HSL in the bacterial culture supernatant or the reaction mixture was determined quantitatively as described previously (Dong et al., 2000).

RNA isolation and RT-PCR

Total RNAs were extracted from bacterial cells cultivated in minimal medium at various time points using an RNeasy® mini kit (Qiagen), and RT-PCR was conducted following the protocol of one-step strategy (Qiagen). An aliquot of 1 µg of total RNAs was used as template for RT-PCR to amplify the relAatu6 coding region using forward primer 5′-CTATATCTCG CATCCGCTG-3′ and reverse primer 5′-ACGATAGTCGTTCT GCTTC-3′. A 1.3 kb fragment of 16S rRNA was also amplified in each RT-PCR as an internal control using forward primer 5′-TGACGAGTGGCGGACGGGTG-3′ and reverse primer 5′-ATGCAGTTCCCAGGTTGAGC-3′.

Protein electrophoresis and Western blotting analysis

Protein samples were separated by electrophoresis in 12% polyacrylamide gels containing 0.1% SDS (SDS-PAGE). Proteins were visualized by staining with Coomassie brilliant blue. For Western blotting analysis, proteins were transferred from polyacrylamide gel to polyvinylidene difluoride (PVDF) membranes using a Mini Trans-Blot electrophoretic transfer cell (Bio-Rad). Immunoblot was carried out according to the standard protocol. AttJ and AttM were recognized by rabbit anti-AttJ (Zhang et al., 2002) and anti-AiiA (Dong et al., 2001) respectively. Secondary antibody AP-conjugated goat against rabbit IgG and AP-conjugated substrate kit (Bio-Rad) were used for protein immunodetection.


We thank Dr Michael Cashel of the National Institute of Child Health and Human Development, National Institutes of Health, USA, and Dr Williams W. Metcalf of the University of Illinois at Urbana-Champaign, USA, for providing E. coli strains CF1693, CF3120 and BW020767(pRL27) respectively. Funding was from the Agency of Science, Technology and Research (A*Star), Singapore.