DNA supercoiling differences in bacteria result from disparate DNA gyrase activation by polyamines

DNA supercoiling is essential for all living cells because it controls all processes involving DNA. In bacteria, global DNA supercoiling results from the opposing activities of topoisomerase I, which relaxes DNA, and DNA gyrase, which compacts DNA. These enzymes are widely conserved, sharing >91% amino acid identity between the closely related species Escherichia coli and Salmonella enterica serovar Typhimurium. Why, then, do E. coli and Salmonella exhibit different DNA supercoiling when experiencing the same conditions? We now report that this surprising difference reflects disparate activation of their DNA gyrases by the polyamine spermidine and its precursor putrescine. In vitro, Salmonella DNA gyrase activity was sensitive to changes in putrescine concentration within the physiological range, whereas activity of the E. coli enzyme was not. In vivo, putrescine activated the Salmonella DNA gyrase and spermidine the E. coli enzyme. High extracellular Mg2+ decreased DNA supercoiling exclusively in Salmonella by reducing the putrescine concentration. Our results establish the basis for the differences in global DNA supercoiling between E. coli and Salmonella, define a signal transduction pathway regulating DNA supercoiling, and identify potential targets for antibacterial agents.


Introduction
All living cells supercoil their DNA. The resulting changes in DNA structure enable compaction of the DNA so that it fits within the limited space of a bacterial cell or eukaryotic organelle [1]. DNA supercoiling is necessary for transcription in prokaryotes [2,3] and eukaryotes [4], recombination in bacteriophages [5,6] and eukaryotic viruses, including HIV [7], and chromosome segregation [8]. Because some of the enzymes governing bacterial DNA supercoiling are essential, their pharmacological inhibition is an efficient antibacterial strategy [9,10].
Global DNA supercoiling in bacteria results from the opposing activities of topoisomerase I, which relaxes DNA, and DNA gyrase, which compacts DNA. (Although DNA topoisomerase IV also changes DNA supercoiling, its contribution is negligible compared to that of DNA gyrase under physiological conditions [11].) The global DNA supercoiling effects of these enzymes are in addition to the primarily local effects of gene transcription and DNA replication [12] and nucleoid structuring proteins [13].
Formation of negative supercoiled DNA by the purified DNA gyrase from Escherichia coli requires Mg 2+ , adenosine triphosphate (ATP) and spermidine [14], the most abundant polyamine in bacteria [15]. Mg 2+ and ATP are required for strand cleavage and strand passage, respectively [16,17], but the specific role that spermidine plays has remained unknown. By contrast, neither ATP nor polyamines are required for activity of the investigated bacterial topoisomerase I [18], and its eukaryotic equivalent does not even require Mg 2+ [19]. Therefore, topoisomerase I activity is currently considered to be independent of metabolic co-factors [9]. Synthesis of topoisomerase I is controlled by a homeostatic mechanism whereby an increase in negatively supercoiled DNA promotes transcription of the topoisomerase I-encoding topA gene [20]. It is currently unknown whether specific metabolites control the activities of DNA gyrase and topoisomerase I in living cells.
Polyamines are a family of positively charged molecules consisting of aliphatic hydrocarbon chains harboring several amine groups [21]. Polyamines are found in all living organisms except some Archaea (i.e., Methanobacteriales and Halobacteriales [22]). The three most common polyamines are spermidine, putrescine and spermine. The abundance of individual polyamines often differs across organisms. For example, E. coli has high amounts of spermidine and its precursor putrescine but lacks the spermidine-derived spermine [15]. Polyamines have RNA-and DNA-binding capacities [23]. In bacteria, they have been implicated in translation [24], protection against acidic [25] and oxidative [26] stresses, iron scavenging [27], and virulence [28,29].
Salmonella enterica serovar Typhimurium (S. Typhimurium) supercoils its DNA less than E. coli under basal growth conditions [30]. This is surprising because these two bacterial species are closely related [31] and also because their TopA (topoisomerase I) proteins share 96% amino acid identity, and their GyrA and GyrB proteins, which constitute the two subunits of DNA gyrase, share 91% and 97% identity, respectively. Differences in the latter proteins are responsible, in part, for the distinct global DNA supercoiling behaviors of the two species because: (i) identical mutations in gyrA [32] or gyrB [33] have different effects on the two species; (ii) the E. coli DNA gyrase is faster than the S. Typhimurium enzyme in vitro; and (iii) DNA is less negatively supercoiled in an E. coli strain expressing a chimeric GyrA protein consisting of the E. coli GyrA with its 35 C-terminal residues replaced by the corresponding region from the S. Typhimurium GyrA than in an E. coli strain expressing the wild-type E. coli GyrA [30].
Here, we report that in vivo, putrescine and spermidine specifically activate the E. coli and S. Typhimurium DNA gyrases, respectively. We identify an environmental condition that controls DNA supercoiling in S. Typhimurium by decreasing putrescine abundance and establish that this condition does not alter DNA supercoiling in E. coli. Our findings suggest new targets to disrupt DNA supercoiling specifically in bacterial pathogens without harming commensal bacterial species or humans.

Spermidine and putrescine stimulate DNA supercoiling in E. coli
Because activation of the E. coli DNA gyrase in vitro requires spermidine [14], we wondered whether spermidine plays such a role in vivo. Thus, we examined the DNA supercoiling status of a reporter plasmid in isogenic wild-type and speE E. coli using the classical agarose/chloroquine gel method [34]. The speE gene specifies a protein that converts putrescine into spermidine, and spermidine into spermine [35] (Fig 1). Therefore, a speE null mutant is unable to make spermidine and spermine (Fig 1).
DNA was more negatively supercoiled in wild-type E. coli than in the isogenic speE mutant (Fig 2A), suggesting that spermidine and/or spermine increases negative DNA supercoiling. We ascribe this activity to spermidine because spermine was not detected under conditions in which spermidine and putrescine were (Fig 3), in agreement with published results [15].
The oat and puuA genes specify proteins that convert putrescine into 4-aminobutanoate (Fig 1). Therefore, the oat puuA speE triple mutant is predicted to accumulate putrescine but lack spermidine and spermine. DNA was more negatively supercoiled in the oat puuA speE triple mutant than in the speE single mutant or in wild-type E. coli (Fig 2A), suggesting that putrescine can increase global DNA supercoiling in vivo.
Putrescine is produced from ornithine by either the SpeF or SpeC proteins, depending on the environmental condition, notably extracellular pH [36], or from agmatine by the SpeB protein (Fig 1). The speB speC speF triple mutant displayed the lowest negative DNA supercoiling of the investigated E. coli strains (Fig 2A) as it is unable to make putrescine, spermidine and spermine (Fig 1). Cumulatively, the results in this section indicate that both putrescine and spermidine promote DNA supercoiling in E. coli, and that spermidine plays a dominant role in this process.

Putrescine stimulates DNA supercoiling in S. Typhimurium, but spermidine has a negligible effect
The SpeE proteins of S. Typhimurium and E. coli share 94% amino acid identity, which is higher than the 90% median amino identity between proteins present in the two species [37]. Therefore, inactivation of the speE gene was anticipated to result in analogous phenotypes in S. Typhimurium and E. coli: decreased negative DNA supercoiling. Unexpectedly, DNA was actually more negatively supercoiled in the S. Typhimurium speE mutant than in the wild-type parent ( Fig 2B). This is the exact opposite phenotype observed in E. coli upon speE inactivation (Fig 2A). These results suggest that putrescine is more important for DNA supercoiling in S. Typhimurium than in E. coli.
That putrescine promotes DNA supercoiling in S. Typhimurium is further supported by the phenotypes of mutants in pathways predicted to alter putrescine abundance. For instance, DNA was more negatively supercoiled in the oat speE double mutant than in oat or speE single mutants. In turn, DNA was more negatively supercoiled in the latter two mutants than in wild-type S. Typhimurium (Fig 2B). Deletion of the putrescine exporter-encoding sapABCDF operon [38] increased negative DNA supercoiling more than deletion of the oat and speE genes (Fig 2A). By contrast, DNA was less negatively supercoiled in a speB argE double mutant than in wild-type S. Typhimurium (Fig 2B), likely due to the inability of the mutant to synthesize putrescine and spermidine (Fig 1).
Because the S. Typhimurium speB speC speF triple mutant had a severe growth defect in minimal media, it was not possible to determine its DNA supercoiling status. This is in contrast to the E. coli speB speC speF triple mutant, which grew well enough to measure DNA supercoiling, albeit not as well as wild-type E. coli. Therefore, mutations in homologous genes that compromise putrescine abundance result in different phenotypes in S. Typhimurium and E. coli.
The results in this section indicate that putrescine strongly stimulates DNA supercoiling in S. Typhimurium and that the contribution of spermidine to this process is minimal. Moreover, they raise the possibility that the different basal global DNA supercoiling status of S. Typhimurium and E. coli [39] result from disparate abundances of specific polyamines in the two species, and/or from the response of their respective topoisomerases or DNA gyrases to these polyamines. As stated above, differences in DNA gyrase activity contribute to the distinct basal DNA supercoiling of S. Typhimurium and E. coli [30].  (Table 1). Bold blue arrows denote pathways upregulated during growth in 10 mM Mg 2+ compared to 0.8 mM Mg 2+ . Slim light blue arrows denote pathways downregulated during growth in 10 mM Mg 2+ compared to 0.8 mM Mg 2+ . � : the PuuABCD pathway is present in E. coli but not in S. Typhimurium. https://doi.org/10.1371/journal.pgen.1009085.g001 The concentrations of putrescine, spermidine, and spermine are similar in

S. Typhimurium and E. coli
We determined that wild-type S. Typhimurium strain 14028s harbors 20 mM putrescine, 4.5 mM spermidine, and no detectable spermine when grown in N-minimal media with glycerol as carbon source and 0.8 mM Mg 2+ (Fig 3), a concentration similar to that of free cytoplasmic Mg 2+ [40]. The abundance of these polyamines is in contrast to those reported for a different S. Typhimurium strain grown in complex LB medium in which putrescine and spermine were detected but spermidine was not [41]. Wild-type E. coli strain MG1655 has polyamine concentrations similar to those of S. Typhimurium (15 mM putrescine, 2.5 mM spermidine, no detectable spermine) when grown in the same medium (Fig 3). The latter concentrations are lower than those reported for the E. coli strain W3110 grown in nutrient broth, albeit in the same order of magnitude [23,42].
The speE oat double mutant had more total putrescine (38 mM) than wild-type S. Typhimurium (20 mM) and no detectable spermidine (Fig 3). This result makes physiological sense given that the speE oat mutant is defective in both putrescine degradation and spermidine synthesis (Fig 1).
Cumulatively, the data in this section argue against the possibility of the disparate DNA supercoiling phenotypes exhibited by S. Typhimurium and E. coli upon speE inactivation resulting from differences in polyamine abundance between the two species (Fig 2). Moreover, they suggest that the differences in basal negative DNA supercoiling between S. Typhimurium and E. coli result from the response of their respective DNA gyrases and/or topoisomerases to particular polyamines.

The S. Typhimurium and E. coli DNA gyrases are differentially activated by individual polyamines
The genetic data presented above argue that negative DNA supercoiling is stimulated by putrescine in S. Typhimurium and by spermidine in E. coli (Fig 2). Thus, we examined the ability of these polyamines to activate the purified DNA gyrases from the two species. To guide the design of an in vitro DNA gyrase activation assay using polyamine concentrations in the physiological range, we considered the affinity constants between polyamines and various cellular components and determined that only~40% of putrescine and~5% of spermidine are free in the cell [23,42] (i.e., not bound to DNA, RNA, ATP or phospholipids). If the proportion of free polyamines is the same in S. Typhimurium and E. coli, then, based on our and spermine were assessed by HPLC in exponentially growing wild-type S Typhimurium strain 14028s or the isogenic speE oat mutant (AAD58) in N-minimal medium with the indicated Mg 2+ concentrations. Stars denote statistical significance compared to wild-type in 800 μM Mg 2+ . ��� : p<0.001; � : p<0.05; ns: not significant (Student's t-test, n = 3). C,D: Same as above but for wild-type E. coli strain MG1655. None of the tested polyamines was detectable in the speB speC speF mutant (AAD95). Spermine was not detected under the investigated conditions (detection limit = 0.2 mM).
The in vitro activities of both the S. Typhimurium and E. coli DNA gyrases were stimulated by low concentrations of both putrescine and spermidine (Fig 4). However, the two enzymes responded differently to an increase in the concentrations of these polyamines. For instance, 11 mM putrescine activated the Salmonella DNA gyrase as much as 0.4 mM spermidine (Fig  4). Both putrescine and spermidine activated the E. coli enzyme more strongly than the S. Typhimurium DNA gyrase, the effect being greater for putrescine than for spermidine (Fig 4). Activation of the E. coli DNA gyrase plateaued at a putrescine concentration between 3.4 mM and 11 mM (from +122% to +152%, i.e. a 1.2-fold increase). This is in contrast to the S. Typhimurium enzyme, the activity of which increased 2.1-fold between these two concentrations (from +26% to +53%).
Putrescine, spermidine, and spermine modified DNA supercoiling in a DNA gyrase-dependent manner because no changes in DNA supercoiling were observed when DNA gyrase was left out of the reaction (S2 Fig), and also because neither putrescine nor spermidine modified the activity of the purified topoisomerase I from E. coli (S2 Fig).
Putrescine inhibited both DNA gyrases at 40 mM (Fig 4). This in vitro result is unlikely to have physiological significance because inactivation of the putrescine-degrading genes speE and oat increased putrescine concentration only two-fold in vivo (Fig 3), and the putrescine concentration would need to increase five-fold to inhibit DNA gyrase activity (i.e., from 8 to 40 mM free putrescine). Likewise, spermine, which is not detected in Salmonella (this work) or E. coli (this work and [15]), inhibited both DNA gyrases at 0.86 mM, but strongly activated them at 0.29 mM (Fig 4). By contrast, spermidine had no inhibitory effect at the highest tested concentration (i.e., 2 mM) (Fig 4).
Cumulatively, the results in this section indicate that both putrescine and spermidine activate the S. Typhimurium and E. coli DNA gyrases at physiological concentrations. However, these enzymes differ in that the S. Typhimurium DNA gyrase is further activated (2.1 fold) by increases in putrescine concentrations that have negligible (1.2 fold) effects on the activity of the E. coli DNA gyrase.

An increase in the concentration of extracellular Mg 2+ decreases global DNA supercoiling in S. Typhimurium but not in E. coli
Mg 2+ is the most abundant divalent cation in living cells [43]. It is necessary for DNA gyrase activity in vitro [14] and competes with polyamines for binding to negatively charged molecules such as DNA [44]. These properties raise the possibility of a regulatory connection existing among the Mg 2+ concentration, polyamine content, and DNA supercoiling. To test this possibility, we explored whether changes in the Mg 2+ concentration in the growth media modify global DNA supercoiling in S. Typhimurium and E. coli.
DNA supercoiling was maximal in wild-type S. Typhimurium experiencing~1 mM Mg 2+ (Fig 5A), which corresponds to the concentration of free Mg 2+ in living cells [40]. DNA became less supercoiled both when the Mg 2+ concentration in the media decreased to 10 μM Mg 2+ and when it increased above 2 mM Mg 2+ (Fig 5A). By contrast, DNA supercoiling in E. coli was insensitive to changes in the Mg 2+ concentration in the media between 10 μM and 10

PLOS GENETICS
In vivo activation of DNA gyrase by polyamines mM (Fig 5A). These results raise the question: How does an increase in Mg 2+ promote a decrease in DNA supercoiling in S. Typhimurium?

Growth in high Mg 2+ reduces putrescine abundance in S. Typhimurium
Both putrescine (Fig 4) and ATP [45] are required for in vitro activity of the S. Typhimurium DNA gyrase. Thus, we considered the possibility that growth in high Mg 2+ decreases DNA supercoiling in Salmonella by reducing the concentration of putrescine and/or ATP. Such a reduction would not take place in E. coli because growth in high Mg 2+ does not alter DNA supercoiling in this species (Fig 5A).
We determined that the putrescine concentration was 25% higher in S. Typhimurium during growth in 0.8 mM than in 10 mM Mg 2+ (Fig 3). By contrast, the putrescine concentration did not change in E. coli experiencing these Mg 2+ concentrations (Fig 3). Putrescine supplementation to the growth medium overcame the decrease in DNA supercoiling provoked by high extracytoplasmic Mg 2+ in S. Typhimurium (Fig 5B). Growth at the high Mg 2+ concentration resulted in a 1.5-fold decrease in ATP concentration (S3 Fig), which is too small to account for the observed changes in DNA supercoiling [46] (i.e., the decrease in ATP concentration would have to be >5 fold). These results indicate that a growth condition that reduces putrescine abundance also reduces global negative DNA supercoiling in S. Typhimurium. How, then, does growth in high Mg 2+ decrease putrescine abundance?

Growth in high Mg 2+ reduces the mRNA amounts of genes that increase putrescine abundance
We carried out an RNA-seq experiment to identify genes differentially expressed at Mg 2+ concentrations resulting in high versus low global negative DNA supercoiling (0.8 mM and 10 mM, respectively) in wild-type S. Typhimurium. Growth in 10 mM Mg 2+ changed the mRNA abundance of multiple genes that encode proteins predicted to lower putrescine amounts (Fig  1; values for each gene presented in Table 1). For example, the mRNA abundance of the speA, speB, and speC genes was lower in 10 mM Mg 2+ than in 0.8 mM Mg 2+ ( Table 1). speA and speB specify enzymes that participate in the synthesis of putrescine from arginine (Fig 1); and speC mediates putrescine synthesis from ornithine (Fig 1).
The mRNA abundance of the putrescine-degrading oat gene and of the arg genes mediating the conversion of ornithine into arginine (Fig 1) was higher in 10 mM Mg 2+ than in 0.8 mM Mg 2+ . The most strongly upregulated gene in 10 mM relative to 0.8 mM Mg 2+ was argI, which specifies the rate-limiting enzyme in the conversion of ornithine into arginine (Fig 1). Thus, the transcriptional response to high Mg 2+ favors expression of genes that promote putrescine degradation and deplete the putrescine precursor ornithine. However, similar agmatinase activity mediated by the SpeB protein was observed in crude extracts prepared from S. Typhimurium grown at 10 mM and 0.8 mM Mg 2+ (S4 Fig), and ornithine decarboxylase activity mediated by the SpeC and SpeF proteins was undetectable (S4 Fig). Therefore, it is presently unclear which of the genes whose transcription changes in 10 mM vs. 0.8 mM Mg 2+ ( Table 1) is responsible for the resulting change in putrescine abundance (Fig 3).
The mRNA abundance of the putrescine importer yeeF gene [47] was lower in 10 mM relative to 0.8 mM Mg 2+ . Taken together with the data presented above, this result indicates that high Mg 2+ reduces the putrescine concentration in S. Typhimurium by altering the expression of genes predicted to decrease putrescine import and synthesis, and to increase putrescine degradation.
Finally, expression of gyrA (but not gyrB), which is also significantly regulated by Mg 2+ in the growth media though not to the extent of argI, speE or oat, may contribute to the observed changes in global negative DNA supercoiling. Therefore, the coordinated response to Mg 2+ (rather than individual enzymatic activities), appears to be responsible for the observed variations in polyamine amounts (Fig 3) and the resulting changes in global negative DNA supercoiling. The effects of Mg 2+ point to specific gene regulation by Mg 2+ rather than effects on DNA supercoiling-sensitive promoters because none of the genes differentially expressed at 10 mM vs. 0.8 mM Mg 2+ displayed sensitivity to DNA supercoiling when 11 different DNA supercoiling-altering conditions were used to examine their expression behavior (see S5 Fig  and GEO entry GSE137586).

Discussion
We have now established that different polyamines activate the DNA gyrases from S. Typhimurium and E. coli. The different response of these DNA gyrases to the polyamines putrescine and spermidine dictates the distinct basal global DNA supercoiling set point of the two species. We determined that environmental Mg 2+ regulates DNA supercoiling in S. Typhimurium, but not in E. coli, by altering putrescine abundance. Our findings reveal a novel physiological role for polyamines, which play critical functions in all domains of life [48]. Moreover, they help explain why DNA is less negatively supercoiled in S. Typhimurium than in E. coli when these bacteria experience the same growth condition. And they suggest potential targets for antibacterial agents to inhibit the essential process of DNA supercoiling.

The molecular basis for the DNA supercoiling differences between S. Typhimurium and E. coli
The activity of the purified DNA gyrases from S. Typhimurium and E. coli is enhanced or inhibited by individual polyamines depending on their concentrations (Fig 4). This result explains the surprising finding that the S. Typhimurium speE mutant, which is unable to make spermidine but accumulates its precursor putrescine (Fig 1), is more supercoiled than wild-type S. Typhimurium (Fig 2B), but that the equivalent speE mutation in E. coli decreases DNA supercoiling below the values of wild-type E. coli (Fig 2A). Although putrescine activates the E. coli DNA gyrase at lower concentrations than the S. Typhimurium DNA gyrase, further increases in putrescine concentration stimulate S. Typhimurium DNA gyrase activity but not that of E. coli (Fig 4). Therefore, the increased putrescine amounts resulting from speE inactivation in E. coli have little effect on its DNA gyrase because the enzyme is already maximally activated (Fig 6).
By contrast, the putrescine accumulation resulting from speE inactivation in S. Typhimurium furthers DNA supercoiling because activation of its DNA gyrase requires higher putrescine amounts than does activation of the E. coli DNA gyrase (Fig 3). That conditions resulting in putrescine accumulation hyperactivate DNA gyrase in S. Typhimurium is further supported by the increased DNA supercoiling exhibited by the sapABCDF mutant (Fig 2), which is defective in putrescine export [38]. The different DNA supercoiling phenotypes resulting from inactivation of the speE gene in S. Typhimurium and E. coli are reminiscent of the different behaviors exhibited by these species harboring equivalent mutations in the gyrB, seqA, and mukB genes [33] DNA gyrase is a tetramer consisting of two GyrA and two GyrB proteins [9]. The behavior of organisms with chimeric DNA gyrases originating from S. Typhimurium and E. coli, and the clustering of amino acid differences between the two GyrA proteins in the C-terminus, implicate this region in the species-specific DNA supercoiling behavior. Although the E. coli DNA gyrase is faster in vitro than the S. Typhimurium DNA gyrase in the presence of spermidine, a S. Typhimurium strain bearing the E. coli GyrA protein exhibited decreased DNA supercoiling compared to wild-type S. Typhimurium [30]. Our findings raise the possibility of putrescine and spermidine interacting with the C-terminal region of GyrA to activate the S. Typhimurium and E. coli DNA gyrases, respectively. Under this hypothesis, loss of supercoiling in the S. Typhimurium strain with the E. coli GyrA may reflect disrupted activation by polyamines.
Growth in >2 mM Mg 2+ reduces DNA supercoiling in S. Typhimurium (Fig 5A) by promoting expression of genes specifying products that reduce putrescine accumulation ( Table 1). Because the Mg 2+ effect on DNA supercoiling is independent of the ATP control of DNA gyrase activity [46], the uncovered mechanism allows S. Typhimurium to modify DNA supercoiling while avoiding disruption of energy-dependent cellular processes. Furthermore, it enables S. Typhimurium to alter DNA supercoiling in response to different, and a priori independent, metabolic signals-ATP and putrescine.
S. Typhimurium experiences a decrease in global negative DNA supercoiling inside mammalian cells [49]. A mildly acidic pH was previously proposed to play a role based on an in vivo synergistic effect of the DNA gyrase inhibitor novobiocin and acidic pH on DNA relaxation [50]. However, this observation can be explained by a mildly acidic pH increasing permeability to novobiocin [51]. Changes in the abundance of putrescine and spermidine, or in a metabolite yet to be identified, may contribute to the reported decrease in global DNA supercoiling. E. coli differs from S. Typhimurium in that it does not alter its basal DNA supercoiling in response to an increase in Mg 2+ concentration in its surroundings (Fig 5A). Under the investigated conditions, the levels of negative DNA supercoiling in E. coli are similar to those in S. Typhimurium (0.5 RSU compared to 0-1 RSU, respectively). This argues against the C-terminus of GyrA being responsible for the reported lower DNA supercoiling in S. Typhimurium vs. E. coli [30]. However, the latter experiments were carried out using both a different S. Typhimurium strain (LT2-derived versus 14028s in the current study) and growth conditions (LB at 30˚C versus defined media at 37˚C in the current study), which raises the possibility of the differences in DNA supercoiling between S. Typhimurium and E. coli resulting from the signals they respond to and the magnitude of those responses. The DNA gyrase subunits of S. Typhimurium and E. coli are >90% identical, and this is also the case for the SpeE protein. Therefore, these findings alongside others [30,52,53] demonstrate how phenotypic differences between closely related bacterial species can result from allelic differences in conserved genes and in their regulation, rather than from the presence of a gene(s) in one species and its absence from another.

Inhibiting DNA supercoiling by manipulating polyamine abundance
The vast majority of antibacterial agents that inhibit DNA supercoiling do so by targeting DNA gyrase [54]. Unfortunately, many of these agents are toxic to humans [55], and resistance to these agents is spreading quickly [56]. Furthermore, despite continued efforts to discover compounds that alter DNA gyrase activity [57], there has been no clinical success since the year 2000.
The identification of putrescine and spermidine as key regulators of DNA gyrase in vivo and in vitro raises the possibility of altering DNA supercoiling by targeting the pathways governing polyamine abundance (as opposed to DNA gyrase). That polyamine biosynthesis is essential in several pathogens (reviewed in [48]) supports this possibility. In S. Typhimurium, there is a direct correlation between DNA supercoiling and expression of the invasion gene invA [58], and DNA supercoiling decreases when S. Typhimurium is inside macrophages [59], promoting expression of genes required for intramacrophage survival [59]. Moreover, mutants defective in putrescine synthesis or export exhibit aberrant DNA supercoiling ( Fig 2B) and are attenuated for virulence [28,29].
Spermine inhibited DNA gyrase at submicromolar concentrations (Fig 4). This raises the possibility of organisms using spermine as an endogenous DNA gyrase inhibitor. It also suggests using a spermine scaffold to design novel DNA gyrase inhibitors.
Finally, humans lack some of the putrescine biosynthetic enzymes utilized by bacteria ( S6  Fig) as well as putrescine transporters, making it less likely that an agent targeting putrescine abundance in bacteria would inhibit human proteins carrying out analogous functions. That pathogenic S. Typhimurium and commensal E. coli differ in their control of DNA supercoiling suggests that an antimicrobial agent may target pathogen-specific pathways while sparing those operating in commensal organisms.

Bacterial strains, plasmids and growth conditions
Strains, plasmids and primers are described in S1 Table. Mutations were created by λred recombination [60]. Plasmid pSIM6 was used to supply λred in all cases. After PCR verification of the strains, mutations were moved into a clean genetic background using P22-mediated transduction in the case of S. Typhimurium [61] and P1-mediated transduction in the case of E. coli.
Strain AAD181 was built by P22 transduction using a lysate prepared in strain AAD35 to infect strain AAD58.
For strain AAD212, single argE::kan and speB::cat mutations were built in separate strains by λred recombination using pKD4 and pKD3 as templates, respectively, and primer pairs 16698/16699 and 16702/16703, respectively. The speB::cat strain (AAD62) was then transformed by plasmid pCP20 to remove the cat cassette and yield a ΔspeB strain (AAD65). Finally, the ΔspeB strain was transduced with a P22 lysate prepared in strain AAD61 (argE::kan) to yield strain AAD212.
Strains AAD85, AAD95 and AAD96 were sequenced alongside the parental MG1655, and Snippy (https://github.com/tseemann/snippy) was used to detect mutations. AAD96 has a mutation that specifies a WaaO protein with the Asp36Glu substitution. The four strains were otherwise isogenic except for the intended mutations. Data are available in the Sequence Read Archive (SRA) under accession numbers SRR12282603, SRR12282604, SRR12282605, SRR12282606.
All strains were grown in HH800 minimal medium at 37˚C except otherwise indicated. HH800 is N-minimal medium [62] supplemented with 0.8 mM MgCl 2 , 0.1% casamino acids, 0.27% glycerol, and adjusted to pH 7.7. HH was the same medium, but with 10 mM MgCl 2 , and was used as a high magnesium concentration to investigate regulation of polyamine abundance by Mg 2+ . Antibiotics were used at the following concentrations: ampicillin, 50μg/mL; kanamycin, 50μg/mL; chloramphenicol, 25μg/mL. E. coli strains were further supplemented with 1μg/mL biotin and 1μg/mL thiamine when grown in minimal medium.

Measurement of DNA supercoiling on agarose/chloroquine gels
Overnight precultures of strains bearing plasmid pJV were washed once in water, then diluted into fresh HH800 medium to a starting OD 600 of 0.05. Cells were grown until mid-exponential phase (OD 600 = 0.7 ± 0.1 for wild-type, and adjusted accordingly for mutants with altered growth), then plasmids were immediately extracted using a QIAGen Plasmid Mini kit. 800 ng of purified plasmid for each sample was then loaded on a Tris-Borate-EDTA, 0.8% agarose, 2.5 μg/mL chloroquine gel. Gels were run overnight at 1.3 V/cm, washed in water for at least 4 h, then stained using EZ-vision (VWR) and imaged with an ImageQuant LAS400 (GE healthcare). The intensity of each band was quantified with ImageJ. The linking number (Lk) of the top band was arbitrarily set to 0, then the band immediately below had Lk = 1, the next one Lk = 2. . . The intensity-weighted average Lk was calculated for each lane. The measured DNA supercoiling for S. Typhimurium was normalized across experiments to the supercoiling exhibited by the wild-type strain 14028s following growth in HH (= HH800 + 10 mM MgCl 2 ), defined as 0 Relative Supercoiling Units (RSU), and the supercoiling in WT in HH800, defined as 1 RSU.

Considerations about quantification of DNA supercoiling
This work concerns in vivo DNA supercoiling, which is overall negative. Therefore, relaxed DNA (i.e. supercoiling closer to 0) is referred to as "low supercoiling" and has low RSU values. By contrast, supercoiled DNA (i.e., strongly negative DNA supercoiling) is referred to "high supercoiling" and has high values in RSU units. Because apparent linking numbers on agarose/chloroquine gels often vary from gel to gel, we use a relative scale instead (RSUs), which allows easy comparison of different gels thanks to a common set of controls. Under the investigated conditions, one linking number difference out of 7 or 8 visible bands typically converts into 2 RSUs. As we focused on quantitative variations in DNA supercoiling rather than absolute values, we did not attempt to distinguish positive and negative DNA supercoiling on gels.

HPLC measurement of total polyamines
Strains were cultured as described above for the determination of DNA supercoiling. Five OD 600 units were pelleted and washed three times with PBS. Cells were then resuspended in 200 μL lysis buffer (100 mM MOPS, 50 mM NaCl, 20 mM MgCl 2 , pH 8.0) and plated to obtain Colony Forming Units (CFUs).
Cells were frozen in liquid nitrogen, then thawed at 37˚C three times. 60 μL trichloroacetic acid 40% was added and the mixture was incubated on ice for 5 min. The supernatant was submitted for derivatization.
Derivatization was performed by adding 250 μL NaOH 2M, 10 μL benzoyl chloride and incubating at room temperature for 20 min. The reaction was stopped by adding 400 μL NaCl, then polyamines were extracted using 300 μL diethyl ether. The ether was evaporated at 37˚C and polyamines were dissolved in methanol/water 45%/55% (v/v).
For the standard curve, a mixture of 20 mM putrescine, 10 mM spermidine and 10 mM spermine was made. This mixture was serially diluted in 4-fold increments. 50 μL of the polyamine mix was added to 200 μL lysis buffer, then derivatized as before.
HPLC was performed on a Waters Acquity UPLC H Class with PDA detector (Waters, Milford, MA). The column was a Waters Acquity UPLC BEH C18, 1.7 μm (Waters 186005604), kept at 60˚C throughout the experiment. Flow rate was 0.3 mL/min. Solvent was 45% methanol/55% water (v/v). Detection was performed in UV at 254 nm.
Cell volume was measured in a separate experiment by phase microscopy, and estimated to be 4 μm 3 for the tested strains and conditions. Intracellular polyamine concentrations were then calculated from the measured concentrations by HPLC, the CFUs and the cell volume.

RNA-seq analysis
Overnight precultures of strains bearing plasmid pJV were washed once in water, then diluted into fresh HH800 medium supplemented or not with 10 mM Mg 2+ to a starting OD 600 of 0.05. Cells were grown until OD 600 = 0.8 ± 0.1 (similarly to supercoiling measurements) and RNA was extracted using a QIAGen RNeasy Mini kit. In addition to the included DNase treatment in the kit, DNA was further eliminated by treatment by turbo DNase (Ambion). RNA was finally re-purified using a QIAGen RNeasy Mini kit.
Library construction and sequencing was performed by the Yale Center for Genome Analysis. rRNA was depleted using a RiboZero kit (Illumina). cDNA synthesis was performed by adding A bases to the 3' end of fragments, followed by oligodT priming. The 4 samples were barcoded and multiplexed into a single flow cell. Sequencing was performed by on an HS4000 (Illumina), Hiseq 75x2 paired-end, unstranded.

Assay of ornithine decarboxylase and agmatinase activities
Cells were grown until mid-exponential phase as for DNA supercoiling measurements. Then, cells were washed twice in PBS, and sonicated on a Bioruptor sonicator (Diagenode, NJ, USA) with the following settings: High intensity, 30s ON/30s OFF, 15 min. The supernatant was used to determine activities and protein concentrations.
Protein concentrations were determined using the Bradford colorimetric assay.

Examining the activities of topoisomerase I and DNA gyrase
Wild-type S. Typhimurium strain14028s transformed with pJV was grown in HH800 + 10 mM MgCl 2 + 25 μg/mL novobiocin until OD 600 = 1, then pJV was extracted from 400 mL of culture using a QIAGen Plasmid Maxi kit. This yielded a moderately relaxed pJV. For topoisomerase I assays, 800 ng of purified pJV was incubated with 0.5 units of DNA topoisomerase I (NEB M0301) in Cutsmart buffer (NEB B7204) at 37˚C for 2 h, then inactivated at 65˚C for 20 min.
For gyrase assays pJV was first further relaxed by topoisomerase I as for the topoisomerase I assay. Then, for E. coli gyrase assays, 800 ng of relaxed pJV was incubated with 2 units of DNA gyrase (TopoGen TG-2000G-3) at 37˚C for 5 h, then inactivated at 65˚C for 20 min. For S. Typhimurium gyrase assays, 800 ng of relaxed pJV was incubated with 0.62 pmol GyrA and 0.62 pmol GyrB (both purified from SL1344, which has 100% identity with 14028s gyrase subunits, gifted by Lesley Mitchenall and Anthony Maxwell, John Innes Centre, Norwich, UK, [32]) at 37˚C for 90 min, then inactivated at 65˚C for 20 min. The reaction buffer was as follows: Tris pH 7.5, 35 mM; KCl, 24 mM; MgCl 2 , 4 mM; DTT, 2 mM; ATP, 1 mM; glycerol, 6.5% (v/v); Bovine Serum Albumin, 0.1 mg/mL. In all cases, the whole reaction was loaded onto a Tris-Borate-EDTA, 0.8% agarose gel, 0.4 μg/mL chloroquine gel. Gels were migrated, stained, imaged and quantified as for in vivo DNA supercoiling gels. DNA gyrase activity was defined as the increase in average Lk, per hour, compared to the initial topoisomer distribution. DNA gyrase activity in the absence of any polyamine was used as a baseline (i.e. 0% relative variation).