Putrescine catabolism is a metabolic response to several stresses in Escherichia coli



Genes whose products degrade arginine and ornithine, precursors of putrescine synthesis, are activated by either regulators of the nitrogen-regulated (Ntr) response or σS-RNA polymerase. To determine if dual control regulates a complete putrescine catabolic pathway, we examined expression of patA and patD, which specify the first two enzymes of one putrescine catabolic pathway. Assays of PatA (putrescine transaminase) activity and β-galactosidase from cells with patAlacZ transcriptional and translational fusions indicate dual control of patA transcription and putrescine-stimulated patA translation. Similar assays for PatD indicate that patD transcription required σS-RNA polymerase, and Nac, an Ntr regulator, enhanced the σS-dependent transcription. Since Nac activation via σS-RNA polymerase is without precedent, transcription with purified components was examined and the results confirmed this conclusion. This result indicates that the Ntr regulon can intrude into the σS regulon. Strains lacking both polyamine catabolic pathways have defective responses to oxidative stress, high temperature and a sublethal concentration of an antibiotic. These defects and the σS-dependent expression indicate that polyamine catabolism is a core metabolic response to stress.


The major polyamines in Escherichia coli and many other bacteria are putrescine and spermidine (Tabor and Tabor, 1985; Cohen, 1998). Six pathways synthesize polyamines, and a mutant lacking all six has only recently been isolated. The mutant could grow aerobically, but not anaerobically (Chattopadhyay et al., 2009). The function of polyamines during anaerobic growth is unknown. Similarly, two pathways of polyamine catabolism have been characterized, and recently loss of both has been shown to result in at least one phenotype – failure to grow with putrescine as sole nitrogen source (Schneider and Reitzer, 2012). These multiple pathways have hindered analysis of polyamine function. Despite this difficulty, some functions of polyamines can be ascertained. Polyamines are often associated with enhanced macromolecular function (Cohen, 1998), and not surprisingly polyamine concentrations positively correlate with growth rate (Tweeddale et al., 1998). Putrescine stimulates translation of several genes, including several general transcription factors, which in turn stimulates transcription of several genes. Putrescine affects at least 300 genes (Yoshida et al., 2004; Terui et al., 2009). Spermidine, which is derived from putrescine, can bind to double-stranded mRNA, altering the structure at the Shine–Dalgarno sequence and start codon to facilitate its interaction with the 30S ribosomal subunit (Yoshida et al., 1999). Nonetheless, exogenous spermidine is toxic during some stresses, such as high pH and low temperature (Fukuchi et al., 1995; Yohannes et al., 2005). The toxicity of spermidine implies that modulation of intracellular polyamine concentrations can be important. Several factors can potentially regulate intracellular polyamines, including synthesis, export or catabolism (Glansdorff, 1996).

Gene profiling indicates that nitrogen limitation induces about 30 genes of polyamine metabolism (Zimmer et al., 2000). Nitrogen-limited growth often involves utilization of nitrogen sources other than an ammonium salt, such as putrescine or arginine. In E. coli, nitrogen-limited growth induces the nitrogen-regulated (Ntr) response, which requires σ54 (RpoN), and the sensor kinase-response regulator pair NtrB–NtrC (Zimmer et al., 2000). Loss of these regulators impairs or prevents growth in most nitrogen-limiting media (Reitzer, 2003). These regulators also activate expression of another transcriptional activator, Nac. Loss of Nac results in subtle and minor growth alterations in E. coli (Muse and Bender, 1998). The known functions of Ntr genes are ammonia assimilation and scavenging of nitrogen-containing compounds (Zimmer et al., 2000; Reitzer, 2003). If these were the only functions of the Ntr response, then it might be expected that Ntr genes would degrade abundant nitrogen-containing compounds in the environment. The most abundant amino acids in the soil are serine, glycine, alanine, aspartate and threonine. The most abundant amino acids in the intestine appear to be arginine, glutamate, lysine, tryptophan and tyrosine (Savageau, 1983; Alves and Savageau, 2005). However, the products of Ntr genes degrade arginine, but none of the other abundant amino acids. Instead, Ntr genes in E. coli specify catabolic pathways for pyrimidines, possibly purines, arginine, ornithine, putrescine and γ-aminobutyrate (GABA) (Schneider et al., 1998; 2002; Xi et al., 2000; Kiupakis and Reitzer, 2002; Kim et al., 2010a). It is conceivable that these particular catabolic operons within the Ntr response may have a function in addition to nitrogen acquisition.

The pathways for arginine, ornithine, putrescine and GABA catabolism are related (Fig. 1) (Schneider et al., 1998; 2002; Kiupakis and Reitzer, 2002; Schneider and Reitzer, 2012). Arginine and ornithine are degraded via succinylated intermediates to glutamate (enzymes 11–15, Fig. 1) (Schneider et al., 1998). A possible arginine catabolic pathway via agmatine to putrescine exists, but is not used to degrade arginine as sole nitrogen source (Fig. 1) (Schneider et al., 1998). Putrescine is degraded via GABA to succinate by either the glutamylated putrescine (GP) pathway (enzymes 5–10, Fig. 1) or the transaminase pathway (enzymes 1–4, Fig. 1). The first reaction of the GP pathway is glutamylation of putrescine (Kurihara et al., 2005; 2008). The first reaction of the transaminase pathway is a transaminase-dependent deamination (Schneider and Reitzer, 2012). The GP pathway degrades putrescine as sole carbon source at ≤ 30°, but not at higher temperatures (Kurihara et al., 2008; Schneider and Reitzer, 2012). Either the GP or transaminase pathway can degrade putrescine as sole nitrogen source, even over 30° (Schneider and Reitzer, 2012). Gene profiling suggests that nitrogen limitation controls the arginine catabolic operon (astCADBE), two operons of the transaminase pathway (patA and gabDTP-csiR), but not the operons of the GP pathway (Zimmer et al., 2000; Schneider and Reitzer, 2012).

Figure 1.

Pathways of putrescine, GABA, ornithine and arginine degradation. Enzymes 1–4 constitute the transaminase pathway of putrescine degradation. Enzymes 5–10 define the glutamylated putrescine (GP) pathway of putrescine degradation. Enzymes 11–15 show the pathway of arginine degradation. Ornithine is degraded by AstA-dependent succinylation (not shown), which generates succinyl ornithine (Schneider and Reitzer, 2012). The transaminase and GP pathways are independently regulated (Schneider and Reitzer, 2012). Ntr regulators control the transaminase pathway and the arginine catabolic pathway (enzymes 1–4 and 11–15), but none of the other enzymes (Zimmer et al., 2000; Kiupakis and Reitzer, 2002; Schneider et al., 2002) (this article). The putrescine-binding PuuR repressor regulates the GP pathway (enzymes 5–10) (Schneider and Reitzer, 2012). In other words, putrescine induces. PuuC is a broadly specific dehydrogenase that oxidizes three different aldehydes: ABAL, γ-glutamyl ABAL and SSA (Schneider and Reitzer, 2012). Abbreviations: ABAL, γ-aminobutyraldehyde; GABA, γ-aminobutyric acid; SSA, succinic semialdehyde; αKG, α-ketoglutarate; GLT, glutamate: and GSA, glutamic semialdehyde.

In addition to Ntr regulators, the general stress regulator, σS, controls the operons of GABA and arginine catabolism (Schneider et al., 1998; 2002; Kiupakis and Reitzer, 2002; Metzner et al., 2004). RNA polymerase complexed to σS is considered a master regulator of stress responses (Weber et al., 2005). This form of RNA polymerase affects about 10% of E. coli genes and mediates responses to nutrient depletion, hyperosmolarity, pH downshifts, and high and low temperature (Weber et al., 2005). σS accumulates during stress, and several dozen factors control transcription, translation, mRNA stability and protein stability (Battesti et al., 2011). σS would control a complete putrescine catabolic pathway, if σS also activates patA and patD, which specify the first two enzymes of the transaminase pathway (reactions 1 and 2, Fig. 1), which might suggest a function during stress responses.

This article describes an analysis of patA and patD expression (see reactions 1 and 2, Fig. 1). The results of gene-specific expression studies confirm the gene profiling analysis that suggested Ntr control. In addition, σS-RNA polymerase controls both genes, which means that σS controls a complete putrescine catabolic pathway (reactions 1–4, Fig. 1). Loss of both putrescine catabolic pathways impairs responses to at least three different stresses. We propose that polyamine catabolism is a core metabolic response to several stresses.


Nitrogen source control of patA and patD expression

Our initial goal was to examine regulation of the metabolically related pathways of arginine, putrescine and GABA catabolism. Some Ntr genes not only respond to general nitrogen limitation, but also to other factors. For example, the arginine catabolic operon, astCADBE, not only responds to Ntr control (via NtrC), but also to specific hyperinduction by arginine (via ArgR), and activation by σS (Schneider et al., 1998; Kiupakis and Reitzer, 2002). To determine potential regulatory factors for patA and patD expression, we examined cells grown with a variety of nitrogen sources.

We assessed patA expression by assay of putrescine transaminase activity from crude extracts. PatA is the only putrescine transaminase in E. coli (Schneider and Reitzer, 2012). Putrescine induced at least 10 times better than other nitrogen sources, except for ornithine (which can be decarboxylated to putrescine) (Fig. 2A). These results suggest putrescine-specific control. To provide insight into the mechanism of putrescine-specific control, we examined expression from patA–lacZ transcriptional and translational fusions. For the transcriptional fusion, ammonia-containing media (nitrogen excess) did not induce, alanine and serine induced moderately (< 10% optimal expression), and glutamine, proline, ornithine, arginine, GABA and putrescine induced best (> 30% optimal expression) (Fig. 3A). The lower expression with alanine and serine has been previously observed for other Ntr genes (Schneider et al., 1991; 1998; 2002). For comparison, we examined expression of nac, which is known to be nitrogen regulated, using a nac–lacZ transcriptional fusion. The results were similar to those for the patA fusion (compare Fig. 3A and C). Although there were differences in expression, putrescine did not induce better than other nitrogen sources. In contrast, for the strain with the patA–lacZ translational fusion, expression was higher with putrescine than any other nitrogen source (Fig. 3D). These results suggest Ntr transcriptional control and putrescine-specific translational control.

Figure 2.

Putrescine transaminase (PatA) and NAD-dependent γ-aminobutyraldehyde dehydrogenase (PatD) activities. Cells were grown with glucose as the carbon source, and the indicated nitrogen source. Abbreviations: Orn, ornithine; Putr, putrescine.

A. Putrescine transaminase activity was assayed from a wild type strain (W3110). The transaminase units are μmole h−1 mg protein−1.

B. PatD-dependent γ-aminobutyraldehyde dehydrogenase was assayed from a ΔpuuC strain (BLS90) to eliminate PuuC-dependent dehydrogenase activity. The dehydrogenase units are μmole min−1 mg protein−1.

Figure 3.

β-Galactosidase activity from strains with lacZ fusions in a wild type background. Cells were grown with glucose as the carbon source, and the indicated nitrogen source. Abbreviations: orn, ornithine; putr, putrescine; GABA, γ-aminobutyric acid. All units are μmole min−1 mg protein−1.

A. patAlacZ transcriptional fusion (from strain BLS9).

B. ydcSlacZ transcriptional fusion (from strain BLS60), which assesses patD expression.

C. naclacZ transcriptional fusion (from strain BLS54).

D. patAlacZ translational fusion (from strain BLS8).

E. ydcSlacZ translational fusion, which assesses patD expression (from strain BLS59).

Similar assays were performed for PatD (NAD-dependent γ-aminobutyraldehyde dehydrogenase). Since PuuC also has γ-aminobutyraldehyde dehydrogenase activity, PatD was assayed from a ΔpuuC mutant. Assays from cells grown with a variety of nitrogen sources indicated general regulation by nitrogen limitation, but not putrescine-specific control (Fig. 2B). PatD from most nitrogen-limited cells (25 μmoles min−1 mg protein−1) was 6000 times more active than PatA (0.25 μmoles h−1 mg protein−1), which suggests that PatA is the limiting reaction of the transaminase pathway. We also examined patD expression from lacZ transcriptional and translational fusions. A lacZ transcriptional fusion of the intergenic region preceding patD showed little activity and no control by nitrogen limitation, which suggests that a promoter does not precede patD (results not shown). Instead, patD appears to be the fifth gene of a ydcSTUV-patD operon based on coexpression of the five genes (Zimmer et al., 2000), and a small intergenic region (22 bases) between ydcV and patD. The ydcSTUV genes are annotated as a putrescine/spermidine ABC transport system (Zimmer et al., 2000). An in-frame deletion of these genes, leaving patD, has no effect on putrescine utilization as sole nitrogen source (Schneider and Reitzer, 2012). This was expected since the major putrescine transporter is PuuP (Kurihara et al., 2009; Schneider and Reitzer, 2012). To study patD expression, we examined expression of the ydcS regulatory region. Expression from the ydcS–lacZ transcriptional and translational fusions (Fig. 3B and E) was essentially identical to control of PatD activity (Fig. 2B). These results confirm that patD is part of the ydcSTUV-patD operon, and show that nitrogen-limited growth activates patD expression, but putrescine does not exert any specific control.

Because PatA is probably the rate-limiting reaction of the transaminase pathway, patA translational control is a significant aspect of regulation. Samsonova et al. noted two potential start codons for patA, which would produce proteins of 429 or 459 residues (Samsonova et al., 2003). They successfully overproduced each protein, and found transaminase activity only with the larger protein. In contrast, we had found that the smaller protein was active, and had exactly the same substrate specificity as that reported for the larger protein (C. Pybus and L. Reitzer, unpubl. obs.). Furthermore, the 30-amino-acid N-terminal extension of the larger protein is not conserved in homologous proteins. We altered the non-canonical upstream start codon, TTG, to TAG (a stop codon) or to ATG (the optimal start codon) in patA–lacZ translational fusions. The TAG codon reduced expression 100-fold, while the ATG codon increased expression 3.7-fold (Table 1). This implies that the more upstream start codon is correct. The optimal ATG at the upstream site did not eliminate putrescine-enhanced expression (Table 1).

Figure 4.

Regulation of patA in a patAlacZ transcriptional fusion. The backgrounds of the strains are indicated: W3110 (wild type background, strain BLS9), rpoS (BLS13), ΔrpoN (BLS21), rpoS ΔrpoN (BLS23) and ΔntrC (BLS43). β-Galactosidase units are μmole min−1 mg protein−1.

A. Cells grown in glucose-ammonia-glutamine (nitrogen excess) medium.

B. Cells grown in glucose-glutamine (nitrogen-limiting) medium.

C. Cells grown in glucose-putrescine (nitrogen-limiting) medium.

D. Sequence of the patA regulatory region. The Shine–Dalgarno and translational start sites are underlined. The distance between patA and aer, the nearest upstream gene, is 507 bases. A consensus σS site is defined as TCTATACTTAA in the −10 region, with the most highly conserved nucleotides underlined, and may not have a recognizable −35 region (Weber et al., 2005). A consensus σ54 promoter has been defined as TGGCACGNNNNTTGC (Barrios et al., 1999). A consensus NtrC site has been suggested to be TGCACCANNNTGGTGCA (Magasanik, 1996).

Table 1. Altering the upstream start codon in patAlacZ translational fusions
Nitrogen source


(wild type)




(optimal start)

  1. The most upstream of two possible start codons is altered. The sequence changed is shown in Fig. 4. The downstream start codon is 90 bases away, and is intact in all constructs. BLS12, BLS113 and BLS114 were grown with glucose as the carbon source and the indicated nitrogen source. Units are nmoles min−1 mg protein−1 for β-galactosidase activity.
Ammonia94 ± 720 ± 0.6543 ± 76
GABA1500 ± 37103 ± 1910600 ± 1100
Putrescine7970 ± 42076 ± 1229400 ± 3100

Transcriptional control of patA expression: σ54, σS, and competition between RNA polymerases

Ntr control requires σ54 (RpoN) complexed with core RNA polymerase (E) and the transcriptional activator NtrC. We examined patA expression in Ntr mutants which lack either σ54 or NtrC. Ntr mutants do not grow in most nitrogen-limited media, except with either glutamine or putrescine as sole nitrogen source. Glutamine can be considered the end-product of the Ntr response, and high intracellular glutamine prevents Ntr gene expression. When glutamine is the nitrogen source, its apparent rapid degradation results in nitrogen limitation. Ntr mutants can utilize putrescine as the sole nitrogen source because the Ntr regulators do not control GP pathway genes. Therefore, we could examine patA expression in Ntr mutants grown in nitrogen-limited medium. Results are shown for the patA–lacZ transcriptional fusion. The patA–lacZ translational fusion gave identical results (not shown).

Loss of either σ54 or NtrC reduced β-galactosidase about twofold with either glutamine (Fig. 4B) or putrescine as the nitrogen source (Fig. 4C). This is consistent with Ntr control, but the residual expression suggests that a second form of RNA polymerase transcribes patA. Loss of σS had little effect with glutamine or putrescine as the nitrogen source, while loss of both σS and σ54 eliminated patA expression (Fig. 4B and C). For cells grown in nitrogen excess (glucose-ammonia-glutamine) medium, loss of σ54 increased expression fourfold, and loss of σS eliminated expression (Fig. 4A). We conclude that either Eσ54 or EσS can transcribe patA, and which dominates depends on the growth medium.

Two results suggest that EσS and Eσ54 may compete for binding: loss of σS increased patA expression about 30% in a nitrogen-limited medium (Fig. 4B), and loss of σ54 increased expression fourfold in nitrogen-excess medium (Fig. 4A). Competition for binding suggests that the transcription start sites are likely to be near each other. The transcription start sites for patA in rpoS and rpoN mutants were shown to be identical (Fig. 4D). Likely binding sites for Eσ54 and EσS are just upstream from the common transcription start site (Fig. 4D). Potential NtrC sites are centred at about −128, −172 and −194, which is not unusual for such sites (Reitzer and Schneider, 2001). We conclude that the two forms of RNA polymerase compete for binding.

Transcriptional control of patD expression: σS and Nac

Gene profiling analysis suggests that Ntr control of patD expression requires Nac (Zimmer et al., 2000). Nac is under Ntr control and activates Eσ70-dependent promoters. Loss of Nac or NtrC reduced ydcS–lacZ expression at least threefold for nitrogen-limited cells (Fig. 5A). We tested whether EσS is required for residual expression, and unexpectedly observed that loss of σS eliminated expression (Fig. 5A). We compared ydcS expression with that of a known EσS-dependent gene, katG, which specifies a catalase. Expression of ydcS (β-galactosidase activity) and katG (catalase activity) varied in parallel during the growth cycle (Fig. 5B), which provides further evidence for EσS-dependent control.

Figure 5.

Regulation of patD expression.

A. Expression of patD is assessed by β-galactosidase activity in strains with a ydcSlacZ transcriptional fusion in different backgrounds: W3110 (wild type background, strain BLS60), rpoS (BLS116), Δnac (BLS118) and ΔntrC (BLS120). Cells were grown in glucose-glutamine (nitrogen-limiting) medium, which is inducing. The units are nmole min−1 mg protein−1.

B. Comparison of patDlacZ expression (β-galactosidase activity, filled circles) and catalase activity (open circles) from BLS60. Cells were grown in LB medium. The units are μmole min−1 mg protein−1. The numbers are not normalized. The activities are coincidently almost identical for the two activities.

C. Potential σS and Nac sites in the ydcSTUV-patD regulatory region. The +1 co-ordinate is the A of the initiation codon. The stop codon of the upstream ydcR gene is TAG at co-ordinates −247 to −245. A consensus σS site is defined as TCTATACTTAA in the −10 region, with the most highly conserved nucleotides underlined, and may not have a recognizable −35 region (Weber et al., 2005). A consensus Nac site has been proposed to be ATAA-N5-TNGTAT (Frisch and Bender, 2010). Two potential sites (underlined) can be identified for both proteins in the ydcR-ydcS intergenic region. The allowable distances between the two sites are not known. The transcription initiation site was not determined.

These results suggest that EσS is required for expression, and that Nac can enhance this expression. There is no precedent for Nac interacting with EσS. Therefore, to test this possibility, we reconstituted transcription with purified components. Eσ70 did not transcribe from the ydcS promoter, although it transcribed well from the promoter for RNA1 from the ColE1-based plasmid (Fig. 6A). EσS did transcribe from the ydcS and RNA1 promoters (Fig. 6A). Nac enhanced EσS-dependent ydcS transcription about twofold (Fig. 6B). Potential EσS and Nac sites for the ydcS promoter are shown in Fig. 5C.

Figure 6.

Transcription with purified components. Multiple-round transcription using plasmid pydcS as the template and E, Eσ70 or EσS.

A. The bottom part shows RNA from the ydcS promoter and the plasmid RNAI promoter with the indicated form and concentration of RNA polymerase. The upper part quantifies the transcript intensity from the lower left panel.

B. Quantification of EσS-dependent transcription from the ydcS promoter, with and without Nac.

Defective putrescine catabolism and stress responses

Polyamines have been implicated in control of transcription factors, such as σS and σE (Yoshida et al., 2001; 2004; Terui et al., 2009). Furthermore, at least one stress, the transition to anaerobic growth, induces the GP pathway (Fig. 1) (Partridge et al., 2006). Therefore, we examined the growth of mutants with defective polyamine catabolism during exposure to 1 mM hydrogen peroxide (oxidative stress), high temperature (45°), and a sublethal concentration of the antibiotic kanamycin (Table 2). A ΔpatA or ΔpuuA strain (which lacks the first enzymes of the transaminase and GP pathway respectively) grew as well as a wild type strain. In contrast, a ΔpatA ΔpuuA double mutant, which lacks both putrescine catabolic pathways, prevented growth during oxidative stress, and impaired growth at 45° or with a sublethal kanamycin concentration. A patA plasmid in the double mutant restored or improved growth for all three stresses, whereas a puuA plasmid restored or improved growth for two of the stresses. (Both of these plasmids poorly express their respective product – results not shown.) These phenotypes were only seen for cells grown in minimal medium, but not for cells grown in broth (results not shown). We also examined growth in different nitrogen-limited media. The single and double mutants grew as well as wild type with alanine, aspartate or arginine as sole nitrogen source. These results suggest that putrescine catabolism is important during several stresses, but not necessarily during nitrogen-limited growth.

Table 2. The phenotypes of putrescine catabolic mutants during stress
Strain genotypePlasmid genotypeNo stressH2O245°Sublethal kanamycin
  1. Cells were grown overnight in glucose-ammonia minimal medium at 37°, diluted 100-fold into fresh medium, grown for 2 h, and then treated for 2 h with 1 mM H2O2, at 45°, or with 5 μg ml−1 kanamycin. Cells were diluted and spotted on LB plates. Each plus represents approximately an order of magnitude difference in survival, i.e. the difference between + and +++ is about two orders of magnitude. – represents no survival. ND, not determined.
Wild type++++++++++++
ΔpatA ΔpuuA++++++
ΔpatA ΔpuuAEmpty ASKA+++NDND
ΔpatA ΔpuuApatA++++++++++
ΔpatA ΔpuuApuuA+++++++++

We also examined the effect of elevated putrescine catabolism in two different ways. First, patA on a plasmid should increase transaminase pathway activity, since PatA appears to be the limiting reaction (Fig. 2). Plasmid-borne patA impaired growth on several growth-limiting nitrogen sources, but had no effect on other stresses (not shown). Second, a ΔpuuR strain has elevated expression of the GP pathway (Schneider and Reitzer, 2012). A ΔpuuR strain exhibited defective growth during oxidative stress, 45°, and with sublethal kanamycin, but had normal growth during nitrogen limitation. These results suggest that enhanced putrescine catabolism might be detrimental. However, this conclusion is complicated by potentially lower α-ketoglutarate (a substrate in the transaminase pathway), lower glutamate and ATP (substrates in the GP pathway), or higher H2O2 (a product of the GP pathway).


Control of putrescine catabolism

We showed that either Eσ54 or EσS controls patA and patD. Similar dual regulation controls the astCADBE and gabDTP-csiR operons, whose products degrade arginine, ornithine and GABA (Schneider et al., 1998; 2002; Kiupakis and Reitzer, 2002; Metzner et al., 2004). In other words, two different regulatory circuits, σS and Ntr regulators, control the catabolism of putrescine and the putrescine prescursors, arginine and ornithine. Despite the same overall outcome by these regulatory systems, a diversity of mechanisms mediate this regulation (Fig. 7).

Figure 7.

The relative locations of the σS and σ54 promoters in the astCADBE, patA, ydcSTUV-patD and csiD-lhgO-gabDTP-csiR operons, and the gabDTP-csiR suboperon. Transcription with purified components has verified these assignments for the ast operon (both promoters), the ydc operon and the Nac-regulated promoter for gabD. Transcription start sites have been determined for the ast promoters, the patA promoters and the two promoters preceding gabD. The two promoters preceding astC and gabD are separated by 5 and 278 bases respectively. The two patA promoters have the same transcription start site.

At least two different post-transcriptional mechanisms control the activity of PatA, which is the rate-limiting reaction of the transaminase pathway. First, putrescine-specific translational control regulates patA. This translational control will be functional regardless of the mechanism of transcriptional control, since transcription initiates from the same nucleotide for both EσS and Eσ54. Putrescine is known to stimulate the binding of mRNA with non-optimal spacing binding to ribosomes (Yoshida et al., 1999; 2004; Terui et al., 2009), and such a mechanism may regulate patA translation. Second, PatA degradation may be regulated. A leucyl/phenylalanyl-tRNA–protein transferase can add a leucyl residue to the N-terminus of PatA, which promotes ClpS-mediated degradation (Ninnis et al., 2009). PatA may be less rapidly degraded, if stress exposes more proteins to modification, and PatA becomes only weakly modified.

The defective stress responses in mutants lacking both putrescine catabolic pathways raises an apparent contradiction that suggests an additional regulatory mechanism. Polyamine content correlates with growth rate. How can the higher polyamines associated with rapid growth fail to induce, but the lower polyamines during slower growth and stress induce? This apparent contradiction can be partially reconciled if an additional putrescine-independent factor stimulates activity of the putrescine catabolic pathways. For the transaminase pathway, this factor may be PatA stability (discussed above). Another possibility is that putrescine is sensed differently during stress. For example, it appears that putrescine binds only weakly to the PuuR repressor. Putrescine (100 mM) reduces binding of purified PuuR to DNA, but by only 50% (Nemoto et al., 2012). Furthermore, deletion of PuuR increases expression from the puuA and puuD promoters 15-fold, but putrescine induces only fourfold (for cells grown in glucose-ammonia minimal medium) (Schneider and Reitzer, 2012). Perhaps an unknown modification or an altered intracellular milieu during various stresses (e.g. altered ion or metal concentrations) enhances the binding of putrescine to PuuR.

Overlap of the EσS and Ntr regulons

Two different gene-profiling analyses show that EσS controls only a subset of Ntr genes. The first analysis showed that EσS controlled 19 of 75 previously identified Ntr genes (Zimmer et al., 2000; Dong et al., 2011). The operons identified were astCADBE, chaBC, gabDTPC, nupC, ydcSTUV-patD and yeaGH. It was suggested that EσS and Eσ54 had antagonistic effects on these genes, that is, if EσS enhances expression, then Eσ54 diminishes expression and vice versa, at least for one growth condition (Dong et al., 2011). While our results are consistent with this observation, especially for patA expression, we suggest that the apparent antagonism is less important than the activation by different stresses acting via different RNA polymerases. A second genome-profiling showed that three different stresses (nutrient depletion, osmotic stress and acid stress) induced a core set of 140 σS-dependent genes (Weber et al., 2005). These core operons include patA, csiD-lhgO-gabDTP-csiR and ydcSTUV-patD, whose products include all the enzymes of the transaminase pathway of putrescine catabolism. These two studies show that several Ntr genes are part of a core response to a variety of stresses. The net effect of the σS-regulated subset of Ntr genes is presumably to modulate intracellular polyamines by controlling the concentrations of arginine and ornithine (precursors of putrescine synthesis), putrescine and derivatives (e.g. spermidine), and GABA (a product of putrescine catabolism). This implies that an important function of this subset of Ntr genes may not be simply nitrogen assimilation and scavenging.

The Ntr control of patD expression is unusual: Nac acts through EσS. If Nac activates other σS-dependent genes, then nitrogen limitation might induce other stress response genes. This may be an important function of the Ntr response, and of Nac in particular, and would suggest that the Ntr and σS regulons are not completely independent. This may explain Ntr genes that have no obvious relationship to nitrogen metabolism, including chaB (a cation transport regulator) and yeaG/prkA (a serine-threonine protein kinase) (Zimmer et al., 2000). To exclude false positives, this possibility of dual regulation requires verification.

The complex relationship between polyamines and stress

Polyamine reduction during stress has been previously observed. Putrescine is exported during osmotic shock, and spermidine is acetylated during heat shock, cold shock, and exposures to ethanol and alkaline pH (Carper et al., 1991; Kashiwagi et al., 1992; Schiller et al., 2000). We suggest that these mechanisms operate in conjunction with putrescine catabolism to reduce intracellular polyamines during stress.

Although polyamine reduction is beneficial for some stresses, polyamine elevation promotes growth and survival in acidic environments (Yohannes et al., 2005). Part of this effect may result from putrescine stimulation of RpoS and RpoE synthesis (Yoshida et al., 2001; 2004; Terui et al., 2009). Two different mechanisms may increase polyamines during some stresses. First, the arginine, putrescine and GABA catabolic operons are poorly expressed at low pH (Yohannes et al., 2005), which should help maintain elevated polyamines. This may involve additional regulators than those considered in this study. One regulator is undoubtedly ArcA, which not only represses all the genes of the transaminase pathway (Evans et al., 2011), but also genes of the GP pathway (Partridge et al., 2006). In addition to repressing the putrescine catabolic pathway, acidic conditions induce decarboxylases for ornithine, arginine and lysine, which generate polyamines (putrescine or cadaverine). One mechanism of induction is via ArcA, which induces ornithine decarboxylase (Evans et al., 2011).

We have provided strong evidence for a linkage between some stresses and polyamine catabolism. In contrast, induction of amino acid decarboxylases and repression of polyamine catabolic pathways, which should favour polyamine accumulation, enhances survival in acidic environments. Considering the complexity of polyamine function, it is not surprising that the relation between polyamines and stress is also complex. However, the precise function of the polyamines may differ for each stress. Analysis of polyamine function has been difficult in part because of the multiple pathways. Six pathways synthesize polyamines, and loss of all six was required to observe significant phenotypes (Chattopadhyay et al., 2009). Two different pathways degrade putrescine, and loss of both is required to observe several phenotypes (Schneider and Reitzer, 2012). The phenotypes of these recently isolated mutants should allow the genetic analysis of polyamine function during different stresses.

Experimental procedures

Strains and plasmids

All strains are derivatives of E. coli K-12 strain W3110. Table 3 lists the strains and plasmids used in this study. Strains with deletions were constructed and verified as described (Datsenko and Wanner, 2000). Table 3 describes the extents of the deletions. Mutations in strains CP2, CP14 and CP98 were initially made in BW25113 and moved by P1 transduction into W3110 as described (Miller, 1992).

Table 3. Bacterial strains and plasmids
Strain or plasmidRelevant genotype or phenotypeSource or reference
  1. aThe extent of the deletion in the protein is indicated. The first two numbers show the codons deleted. The last number shows the total number of amino acid residues in the wild type gene.
  2. bThe extent of the promoter region is indicated for the fusions. Co-ordinate +1 is the first nucleotide of the start codon.
W3110lacL8 lacIqLaboratory stock
CP2W3110 ΔpatA (Δ1–488/488)aSchneider and Reitzer (2012)
CP14W3110 ΔpuuE::cat (Δ2–391/421)aSchneider and Reitzer (2012)
CP19W3110 ΔpatA ΔpuuE::catP1 CP14 into CP2, Catr
CP98W3110 ΔntrC::cat (Δ14–456/470)aThis work
AK3W3110 rpoS::tetKiupakis and Reitzer (2002)
AK23W3110 ΔrpoN (Δ3–477/477)aKiupakis and Reitzer (2002)
HK9W3110 Δnac::cat (Δ1–305/305)aSchneider et al. (2002)
BLS6W3110 attλ::[‘lacZ Kanr] translational fusion controlpBLS17 into W3110
BLS7W3110 attλ::[‘lacZ Kanr] transcriptional fusion controlpBLS18 into W3110
BLS8W3110 attλ::[Φ(patA–lacZ) Kanr] translational fusionpBLS19 into W3110
BLS9W3110 attλ::[Φ(patA–lacZ) Kanr] transcriptional fusionpBLS20 into W3110
BLS10W3110 rpoS::tet attλ::[‘lacZ Kanr] translational fusion controlP1 AK3 into BLS6, Tetr
BLS11W3110 rpoS::tet attλ::[‘lacZ Kanr] transcriptional fusion controlP1 AK3 into BLS7, Tetr
BLS12W3110 rpoS::tet attλ::[Φ(patA–lacZ) Kanr] translational fusionP1 AK3 into BLS8, Tetr
BLS13W3110 rpoS::tet attλ::[Φ(patA–lacZ) Kanr] transcriptional fusionP1 AK3 into BLS9, Tetr
BLS18W3110 ΔrpoN attλ::[‘lacZ Kanr] translational fusion controlpBLS17 into AK23
BLS19W3110 ΔrpoN attλ::[‘lacZ Kanr] transcriptional fusion controlpBLS18 into AK23
BLS20W3110 ΔrpoN attλ::[Φ(patA–lacZ) Kanr] translational fusionpBLS19 into AK23
BLS21W3110 ΔrpoN attλ::[Φ(patA–lacZ) Kanr] transcriptional fusionpBLS20 into AK23
BLS22W3110 rpoS::tet ΔrpoN attλ::[Φ(patA–lacZ) Kanr] translational fusionP1 AK3 into BLS20, Tetr
BLS23W3110 rpoS::tet ΔrpoN attλ::[Φ(patA–lacZ) Kanr] transcriptional fusionP1 AK3 into BLS21, Tetr
BLS42W3110 ΔntrC::cat attλ::[Φ(patA–lacZ) Kanr] translational fusionP1 CP98 into BLS8, Catr
BLS43W3110 ΔntrC::cat attλ::[Φ(patA–lacZ) Kanr] transcriptional fusionYang et al. (2004)
BLS54W3110 attλ::[Φnac–lacZ Kanr] transcriptional fusionYang et al. (2004)
BLS59W3110 attλ::[Φ(ydcS–lacZ) Kanr] translational fusionpBLS27 into W3110
BLS60W3110 attλ::[Φ(ydcS–lacZ) Kanr] transcriptional fusionpBLS28 into W3110
BLS84W3110 ΔpuuR (Δ6–181/185)aSchneider and Reitzer (2012)
BLS88W3110 ΔpatA ΔpuuA (Δ6–462/472)aSchneider and Reitzer (2012)
BLS90W3110 ΔpuuC (Δ5–485/495)aSchneider and Reitzer (2012)
BLS113W3110 attλ::[Φ(patA–lacZ) Kanr] translational fusion TAG start siteThis work
BLS114W3110 attλ::[Φ(patA–lacZ) Kanr] translational fusion ATG start siteThis work
BLS115W3110 ΔrpoS::tet attλ::[Φ(ydcS–lacZ) Kanr] translational fusionP1 AK3 into BLS59, Tetr
BLS116W3110 ΔrpoS::tet attλ::[Φ(ydcS–lacZ) Kanr] transcriptional fusionP1 AK3 into BLS60, Tetr
BLS117W3110 Δnac::kan attλ::[Φ(ydcS–lacZ) Kanr] translational fusionP1 HK9 into BLS59, Kanr
BLS118W3110 Δnac::kan attλ::[Φ(ydcS–lacZ) Kanr] transcriptional fusionP1 HK9 into BLS60, Kanr
BLS119W3110 ΔntrC::cat attλ::[Φ(ydcS–lacZ) Kanr] translational fusionP1 CP98 into BLS59, Kanr
BLS120W3110 ΔntrC::cat attλ::[Φ(ydcS–lacZ) Kanr] transcriptional fusionP1 CP98 into BLS59, Kanr
pBLS1Φ(lacUV5p-ntrC) AmprSchneider et al. (1991)
pBLS17lacZ translational fusion vectorKim et al. (2010b)
pBLS18lacZ transcriptional fusion vectorSchneider and Reitzer (2012)
pBLS19Φ(patA–lacZ) Kanr; patA translational fusion (−856 to +471)bThis work
pBLS20Φ(patA–lacZ) Kanr; patA transcriptional fusion (−856 to +471)bThis work
pBLS22Φ(nac–lacZ) Kanr; nac transcriptional fusion (−287 to −5)bThis work
pBLS27Φ(ydcS–lacZ) Kanr; patD translational fusion (−354 to +27)bThis work
pBLS28Φ(ydcSlacZ) Kanr; patD transcriptional fusion (−354 to +27)bThis work
pBLS29pBLS19 with TAG start codonThis work
pBLS30pBLS19 with ATG start codonThis work
pydcSpIn vitro transcription templateThis work

The plasmid pBLS17 was used to construct lacZ translational fusions (Kim et al., 2010b). pBLS18, a derivative of pBLS17, was used to construct lacZ transcriptional fusions (Kim et al., 2010b). Strains containing the lacZ fusions were integrated into the lambda attachment site of W3110 using pINTS as described (Haldimann and Wanner, 2001). To construct the patA–lacZ, nac–lacZ and ydcS–lacZ fusions, a PCR product containing the promoter region was inserted into either plasmid pBLS17 or pBLS18. The extent of the promoter region is described in the plasmid section of Table 3. The TTG start codon of the translational patA–lacZ fusion was changed using a PCR primer containing ATG or TAG instead of TTG for PCR of the promoter region. The forward primer for the patA ATG translational fusion (pBLS30) was ACGCGTGCCTCCGGAGCATATTTATGAACAGG. The forward primer for the patA TAG translational fusion (pBLS29) was ACGCGTGCCTCCGGAGCATATTTTAGAACAGG. The reverse primer for these two constructions was CCACGGTAAATCACTTGGCGCGGATCCTG.

Media and growth conditions

The minimal medium for growth contained W salts (Rothstein et al., 1980) with 0.4% glucose as carbon source and 0.2% nitrogen source. Starter cultures (typically 24–48 h incubation) were grown in 4 ml of minimal medium with the appropriate nitrogen source, pelleted, washed twice with 150 mM NaCl, and inoculated into cultures for growth studies or assays. All cultures for growth studies and assays were grown at 30°C and aerated at 220 r.p.m. Luria–Bertani (LB) broth and agar plates were supplemented with 100 μg ml−1 ampicillin, 10 μg ml−1 kanamycin, 12.5 μg ml−1 tetracycline or 25 μg ml−1 chloramphenicol where appropriate.

Enzyme assays

All specific activities are the results from at least three independent determinations. The errors are reported as the standard error of the mean.

For β-galactosidase assays, cultures (10 ml) were harvested in late exponential phase (95–110 Klett units, #42 filter; OD600 of ∼ 0.6), pelleted at 4°C, washed twice with cold 150 mM NaCl and frozen. Thawed cell pellets were resuspended in 1 ml of Z buffer (Miller, 1972) and sonicated on ice in three 5 s bursts. After centrifugation at 4°C, the supernatants were assayed as described (Miller, 1972). Protein concentrations were determined as described (Lowry et al., 1951). Specific activity units are described in the figure legends.

For assay of putrescine aminotransferase activity, strains were grown in the appropriate minimal media, harvested in late exponential phase, centrifuged, washed twice with 150 mM NaCl, and frozen at −80°C. Cell pellets were resuspended in 1 ml of resuspension buffer (0.1 M potassium phosphate, pH 7.5, 1 mM DTT) and sonicated in three five-second bursts. The cells were centrifuged at low speed and the supernatant was assayed. Putrescine aminotransferase was assayed as described (Albrecht and Vogel, 1964), with the following modifications. The 0.5 ml assay mix contained 0.1 M Tris HCl buffer pH 8.0, 15 mM α-ketoglutarate, 25 mM putrescine, 1 mM pyridoxal-5′-phosphate, and 10 mM EDTA. The mixtures were incubated for 30 min at 37°C. Protein concentration was determined as described (Bradford, 1976). The molar extinction coefficient of the quinazolinium product used for calculations was 1860 M−1 cm−1 (Yamada, 1971).

The assay for γ-aminobutyraldehyde dehydrogenase was performed as described (Schneider and Reitzer, 2012). Crude extracts were prepared like those for putrescine transaminase, and then ultracentrifuged for 90 min at 120 000 g to remove NADH-oxidizing activity. The assay solution contained glycine buffer (pH 9.5), 0.28 mM NAD and 0.5 mM γ-aminobutyraldehyde (freshly made and added last). The A340 for NADH was monitored, and the reaction performed at 37°C.

Catalase assays were conducted as described (Visick and Clarke, 1997). Cell pellets were resuspended in 5 mM potassium phosphate pH 7.0, 5 mM EDTA and 10% glycerol and then lysed by sonication. After addition of H2O2, the absorbance at 240 nm was followed.

Mapping patA transcription start sites

Total RNA from cultures was isolated with RNAprotect and the QIAGEN RNeasy minikits according to the manufacturer's instructions. The rpoN strain AK23 was grown overnight at 30° in glycerol ammonia minimal media. The rpoS strain AK3 was grown on glucose minimal media with 0.2% arginine and 0.2% aspartate as nitrogen sources then harvested during exponential growth. Two to four micrograms of total RNA was used to synthesize cDNA from the 5′ end of the patA mRNA with the gene-specific primer GSP1-patA (CGGTTGTTTCGCAAGTTGAT). The reaction was performed with the SuperScript II RT at 42°C for 30 min. A homopolymeric A-tail was added to the cDNA with the terminal deoxynucleotidyl transferase. The RACE products were synthesized using the abridged anchor primer GGCCACGCGTCGACTAGTACTTTTTTTTTTTTTTTTTTTT and the gene-specific primer GSP2-patA (GCCCCACGTTGAAAATTCCA). A second nested PCR amplifications was performed using the abridged universal anchor primer (GGCCACGCGTCGACTAGTAC) and GSP5-patA (CAGGCAGTCGATAAACTCCT). PCR products were cloned into pUC18 and sequenced by Macrogen, Korea.

Transcription with purified components

Plasmid pydcSp was constructed by cloning a PCR product made using primers F-ydcSp (GGATTACCTTTCCACGCGCC) and R-ydcSp CGCCCCTGAGAAAATTTATTC into the SmaI site of pSA508 (Choy and Adhya, 1992). The transcription reaction was performed in a 10 μl reaction mixture containing 25 mM Tris pH 7.9, 1 mM MgCl, 70 μM β-mercaptoethanol, 0.1 μg ml−1 bovine serum albumin, 20 nM DNA template and 100 mM KCl. The concentrations of the following proteins are indicated in the appropriate figure legends: E. coli RNA σ70-holoenzyme and E. coli RNA core polymerase (both from Epicenter Technologies), RpoS [purified as described (Kiupakis and Reitzer, 2002)], and Nac from Klebsiella pneumoniae (generously provided by Robert Bender). DNA and RNA polymerase were pre-incubated at 37°C and the reactions initiated by the addition of rNTPs to final concentrations of 500 μM each ATP, GTP and CTP and 40 μM UTP (plus 10 μCi [α-32P]-UTP). The mixture was incubated for 10 min at 37°C. The reaction was stopped by the addition of 10 μl of stop solution (95% formamide, 25 mM EDTA, 0.05% bromophenol blue and xylene cyanol). Samples were resolved on a 7 M urea, 8% polyacrylamide gel, visualized by autoradiography and quantified by phosphor-imaging with a Bio-Rad Molecular Dynamics-Molecular Imaging Screen K.


Grants MCB-0323931 from the National Science Foundation and GM085536 from the National Institute of Health supported this work. We thank Christine Pybus for construction of some strains and Robert Bender for purified Nac.