1. Biochemistry and Chemical Biology
  2. Chromosomes and Gene Expression

Free spermidine evokes superoxide radicals that manifest toxicity

  1. Vineet Kumar
  2. Rajesh Kumar Mishra
  3. Debarghya Ghose
  4. Arunima Kalita
  5. Pulkit Dhiman
  6. Anand Prakash
  7. Nirja Thakur
  8. Gopa Mitra
  9. Vinod D Chaudhari
  10. Amit Arora  Is a corresponding author
  11. Dipak Dutta  Is a corresponding author
  1. Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, India
  2. Division of Medicinal Chemistry, CSIR Institute of Microbial Technology, India
  3. Clinical Proteomics Unit, Division of Molecular Medicine, St. John's Research Institute, St John's Medical College, India
https://doi.org/10.7554/eLife.77704
Share this article
Cite this article
  1. Vineet Kumar
  2. Rajesh Kumar Mishra
  3. Debarghya Ghose
  4. Arunima Kalita
  5. Pulkit Dhiman
  6. Anand Prakash
  7. Nirja Thakur
  8. Gopa Mitra
  9. Vinod D Chaudhari
  10. Amit Arora
  11. Dipak Dutta
(2022)
Free spermidine evokes superoxide radicals that manifest toxicity
eLife 11:e77704.
https://doi.org/10.7554/eLife.77704
  2. Download BibTeX
  3. Download .RIS

Abstract

Spermidine and other polyamines alleviate oxidative stress, yet excess spermidine seems toxic to Escherichia coli unless it is neutralized by SpeG, an enzyme for the spermidine N-acetyl transferase function. Thus, wild-type E. coli can tolerate applied exogenous spermidine stress, but ΔspeG strain of E. coli fails to do that. Here, using different reactive oxygen species (ROS) probes and performing electron paramagnetic resonance spectroscopy, we provide evidence that although spermidine mitigates oxidative stress by lowering overall ROS levels, excess of it simultaneously triggers the production of superoxide radicals, thereby causing toxicity in the ΔspeG strain. Furthermore, performing microarray experiment and other biochemical assays, we show that the spermidine-induced superoxide anions affected redox balance and iron homeostasis. Finally, we demonstrate that while RNA-bound spermidine inhibits iron oxidation, free spermidine interacts and oxidizes the iron to evoke superoxide radicals directly. Therefore, we propose that the spermidine-induced superoxide generation is one of the major causes of spermidine toxicity in E. coli.

Editor's evaluation

The authors argue that a polyamine, spermidine, causes the production of reactive oxygen species (ROS) in Escherichia coli by oxidizing Fe2+, but spermidine can also be protective against ROS at lower concentrations when bound to other cellular molecules such as RNA. Thus, spermidine has both protective and antagonistic effects on ROS stress, depending on the cellular concentration.

https://doi.org/10.7554/eLife.77704.sa0

Introduction

Polyamines are ubiquitously present in all life forms. They tweak a diverse array of biological processes, for example, nucleic acid and protein metabolism, ion channel functions, cell growth and differentiation, mitochondrial function, autophagy and aging, protection from oxidative damage, actin polymerization, and perhaps many more (Casero et al., 2018; Gawlitta et al., 1981; Madeo et al., 2018; Michael, 2018; Miller-Fleming et al., 2015; Oriol-Audit, 1978; Pegg, 2016; Pohjanpelto et al., 1981; Tabor and Tabor, 1984; Wallace et al., 2003). The cationic amine groups of polyamines can avidly bind to the negatively charged molecules, such as RNA, DNA, phospholipids, etc. (Igarashi and Kashiwagi, 2000; Miyamoto et al., 1993; Schuber, 1989; Tabor and Tabor, 1984). Polyamines have been demonstrated to protect DNA from reactive oxygen species (ROS) such as singlet oxygen, hydroxyl radical (•OH), or hydrogen peroxide (H2O2) (Balasundaram et al., 1993; Ha et al., 1998a; Ha et al., 1998b; Jung and Kim, 2003; Khan et al., 1992a; Khan et al., 1992b; LØVaas, 1996; Pegg, 2018; Murray Stewart et al., 2018). Indeed, knocking out polyamine biosynthesis enzymes from Escherichia coli and yeast confers toxicity to oxygen, superoxide anion radical (O2-), and H2O2 (Balasundaram et al., 1993; Chattopadhyay et al., 2003; Eisenberg et al., 2009).

Most prokaryotes including E. coli synthesize cadaverine, putrescine, and spermidine, while higher eukaryotes additionally synthesize spermine. E. coli also acquires spermidine and putrescine from the surrounding medium (Igarashi and Kashiwagi, 2000; Miller-Fleming et al., 2015). However, polyamine in excess is toxic to the organisms unless polyamine homeostasis in the cell is operated at the levels of export, synthesis, inactivation, and degradation (Miller-Fleming et al., 2015). Notably, spermine/spermidine N-acetyl transferase (SSAT or SpeG), which inactivates spermidine and spermine, constitutes the most potent polyamine homeostasis component of the cells (Miller-Fleming et al., 2015).

A tremendous volume of work has been dedicated to unravel the biological importance of spermidine and its homeostasis mechanisms. It has also been known for long that spermidine (or spermine) in excess is toxic to the organisms and viruses (Pegg, 2013). It has been proposed that the excess polyamines may affect protein synthesis by binding to acidic sites in macromolecules, such as nucleic acids, proteins, and membrane, and by displacing magnesium from these sites (Limsuwun and Jones, 2000; Pegg, 2013). However, a precise molecular detail of spermidine toxicity is not yet understood. In this study, we decipher a molecular mechanism of spermidine toxicity in bacteria. We find the intertwined relationships among spermidine toxicity, iron metabolism, and O2- radical production in bacteria.

Results

Increased cellular spermidine inhibits overall oxidative stress while apparently evoking less harmful O2- production

To determine the working concentrations of exogenous spermidine that sufficiently inhibits the growth of ΔspeG, but not WT strain, we added various amounts of spermidine in the growth medium. WT cells showed a modest reduction in growth up to 6.4 mM of spermidine concentration (Figure 1—figure supplement 1). On the contrary, ΔspeG strain exhibited a striking decrease in growth when supplemental spermidine level was >3.2 mM (Figure 1—figure supplement 1). Therefore, we chose spermidine concentration ≥3.2 mM for our further experiments. We performed HPLC analyses to show whether elevated spermidine level in the ΔspeG strain caused growth inhibition. Supplementation of 3.2 mM exogenous spermidine in the growth medium increased the intracellular spermidine levels in the ΔspeG strain, while no significant increase was observed in the WT cells (Figure 1—figure supplement 1). The SpeG function apparently converted the excess spermidine to N1- and N8-acetyl-spermidines maintaining the level of spermidine in the WT cells (Miller-Fleming et al., 2015). The spermidine synthase-defective (ΔspeE) strain of E. coli also acquired spermidine at a low level from the LB medium (Figure 1—figure supplement 1).

It is well documented that polyamine spermidine is an anti-ROS agent (Balasundaram et al., 1993; Chattopadhyay et al., 2003; Chattopadhyay et al., 2006; Ha et al., 1998a; Khan et al., 1992a; Khan et al., 1992b; Pegg, 2018; Murray Stewart et al., 2018). However, all the in vivo studies in the past have been conducted under polyamine deficient conditions to show ROS production, thereby implicating the anti-ROS function of polyamines. Thus, assessing ROS levels both in spermidine-enriched and spermidine-deficient conditions are missing. To address this, we incubated E. coli strains with 2',7'-dichlorodihydrofluorescein diacetate (H2DCFDA) and dihydroethidium (DHE) probes, which generate fluorescent compounds reacting with one-electron-oxidizing species. While H2DCFDA is a generic ROS probe that nonspecifically reacts with many ROS, the DHE is somewhat specific to the O2- anions in the system (Chen et al., 2013; Kalyanaraman et al., 2012). The relative mean fluorescence intensity (MFI) of H2DCFDA was increased about 1.5-fold in the spermidine synthase-defective (ΔspeE) strain, while no change in MFI was observed in the ΔspeG strain (Figure 1A). However, spermidine treatment significantly decreased the H2DCFDA fluorescence in WT, ΔspeG, and ΔspeE strains (Figure 1A). Interestingly, despite no apparent increase in the spermidine level in WT cells under spermidine stress (Figure 1—figure supplement 1), a significant decrease in the H2DCFDA fluorescence was observed (Figure 1A). It is possible that the acetylated products of spermidine might have some role in the declined ROS levels causing decreased H2DCFDA fluorescence in spermidine-fed WT cells. Similarly, the relative MFI of DHE probe was increased significantly (1.5-fold) in ΔspeE strain (Figure 1B). These findings are consistent with the observations that spermidine is an anti-ROS agent (Balasundaram et al., 1993; Chattopadhyay et al., 2003; Chattopadhyay et al., 2006; Ha et al., 1998a; Khan et al., 1992a; Khan et al., 1992b; Pegg, 2018; Murray Stewart et al., 2018).

Figure 1 with 1 supplement see all
Download asset Open asset
Spermidine (SPD) stress and intracellular reactive oxygen species (ROS) in Escherichia coli.

(A) The relative mean fluorescence intensity (MFI) values for the 2',7'-dichlorodihydrofluorescein diacetate (H2DCFDA), which is an indicator of •OH radical production, obtained by flow cytometry analyses are plotted. (B) The relative MFI values of dihydroethidium (DHE) probe, which is an indicator of O2- radical production, obtained by flow cytometry analyses are plotted. (C) The absolute H2O2 production for a span of 5 hr from the different E. coli strains are shown. *** are p-values generated comparing with WT value. (D) Zone of inhibitions (ZOIs) surrounding SPD well on the agar plates were shown for the WT and ΔspeG strains of E. coli under aerobic and anaerobic conditions. (E) Serially diluted E. coli cells were spotted on LB-agar plates to show their sensitivity to SPD. (F) Viability of different knockout strains were plotted from the CFU counts in different time intervals after treatment with lethal dose of SPD. ** and *** are p-values generated comparing with the values of ΔspeG and ΔspeGΔsodA, respectively. Error bars in the panels are mean ± SD from the three independent experiments. Whenever mentioned, *** and ** are <0.001 and <0.01, respectively; unpaired t test. See also Figure 1—figure supplement 1, and Figure 1—source data 1, Figure 1—source data 2, Figure 1—source data 3, Figure 1—source data 4.

Figure 1—source data 1

Figure 1A Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig1-data1-v3.xlsx
Figure 1—source data 2

Figure 1B Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig1-data2-v3.xlsx
Figure 1—source data 3

Figure 1C Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig1-data3-v3.xlsx
Figure 1—source data 4

Figure 1F Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig1-data4-v3.xlsx

Surprisingly, the relative MFI of DHE probe was increased significantly (2-fold) in the spermidine-fed ΔspeG as compared to WT strain of E. coli (Figure 1B). Tyron (Tr), an O2- quencher, decreased the MFI of DHE in the spermidine-fed ΔspeG strain (Figure 1B). These observations indicate that although spermidine accumulation in the ΔspeG strain reduces overall ROS levels and oxidative stress (Figure 1A), it may simultaneously evoke less harmful O2- production (Balasundaram et al., 1993; Chattopadhyay et al., 2003; Chattopadhyay et al., 2006; Ha et al., 1998a; Khan et al., 1992a; Khan et al., 1992b; Pegg, 2018; Murray Stewart et al., 2018). In another assay, we determined that the ΔspeE and spermidine-fed ΔspeG strains release substantially low levels of H2O2 compared to the untreated counterpart and WT cells (Figure 1C).

Next, we allowed WT and ΔspeG strains to grow against the spermidine-diffusing wells on agar plates in aerobic and anaerobic conditions (Figure 1D). A far wider zone of inhibition (ZOI) of growth for ΔspeG strain was observed compared to WT under aerobic condition (Figure 1D), while a narrow ZOI was observed under anaerobic conditions for both strains (Figure 1D). This data further indicates that O2- production in aerobic condition could be a cause of the observed spermidine toxicity.

If spermidine induces O2- production, superoxide dismutase (SOD) genes (e.g., sodA and sodB) would play vital roles. Therefore, the serial dilutions of WT, ΔspeG, ΔsodA, ΔsodB, and corresponding double and triple mutants, viz. ΔspeGΔsodA, ΔspeGΔsodB, ΔsodAΔsodB, and ΔspeGΔsodAΔsodB, were transformed with either empty vector, pDAK1, or pSodA vectors. The ΔspeGΔsodA and ΔspeGΔsodAΔsodB mutants containing empty vector exhibited higher growth defects than ΔspeG strain on LB-agar plate supplemented with spermidine (Figure 1E). However, the cell viability of the double mutants was similar to the ΔspeG strain, while the triple mutant exhibited an accelerated loss of cell viability, in the presence of spermidine (Figure 1F). The multicopy induction of SodA from pSodA plasmid suppressed the growth defect in the ΔspeG and ΔspeGΔsodA strains (Figure 1E). The overexpression of SodA also improved the viability of ΔspeGΔsodA strain (Figure 1F). Note that, unlike ΔspeG strain, the single mutants show growth and viability similar to the WT strain in the presence or absence of spermidine (Figure 1E and Figure 1—figure supplement 1). This data suggests that the absence of SOD enzymes aggravates O2- toxicity in the spermidine-fed ΔspeG strain.

Spermidine stress evokes O2- production in ΔspeG strain

Although the above experiments apparently suggest for the production of O2- anions under spermidine stress, they are not direct and confirmatory in nature, as the ROS probes often reacts with multiple ROS (Kalyanaraman et al., 2012). Spermidine transport is a proton motif force (PMF)-dependent process (Kashiwagi et al., 1986). Therefore, the observed narrower ZOI in the presence of spermidine under anaerobic condition (Figure 1D) could also be due to the low PMF under anaerobic condition. Thus, to determine the relative levels of intracellular O2- species, we performed electron paramagnetic resonance (EPR) using a cell-permeable cyclic hydroxylamine spin-probe, 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine (CMH) (Dikalov et al., 2018). Compared to spin-trap agents, lower level of CMH reacts at a much faster rate with O2- anion, producing highly stable and EPR-sensitive nitroxide radicals (Dikalov et al., 2018). However, peroxynitrite and •OH radicals can also oxidize CMH (Dikalov et al., 2018; Thomas et al., 2015).

In the first set of reactions, the unfed and spermidine-fed ΔspeG cells carrying an empty vector were incubated with CMH. In the second set, portions of the unfed and spermidine-fed ΔspeG cells carrying an empty vector were preincubated with dimethyl thiourea (DMTU) and uric acid (UA), the scavengers for the •OH and peroxynitrite (ONOO-) radicals, respectively, before CMH addition. In the third set, the unfed and spermidine-fed ΔspeG cells harboring pSodA plasmid were incubated with CMH. In the first set, a high level of EPR signals were detected with more signal in the unfed sample than the spermidine-fed one (Figure 2B and C). This data indicates that the overall ROS production is higher in the absence of exogenous spermidine, which is consistent with the notion that the spermidine is an anti-ROS agent (Balasundaram et al., 1993; Chattopadhyay et al., 2003; Chattopadhyay et al., 2006; Ha et al., 1998b; Khan et al., 1992a; Khan et al., 1992b; Pegg, 2018; Murray Stewart et al., 2018). In contrast, EPR signal was higher in the spermidine-fed cells than unfed one in the second set (Figure 2D and E), suggesting that the signals apparently represent CMH oxidation by O2- anions. Finally, the decrease in EPR signals under the multicopy expression of SodA (Figure 2F and G) suggests that the signals in the second set were indeed generated from O2--mediated oxidation of CMH.

Figure 2
Download asset Open asset
Spermidine stress generates O2- radical production in ΔspeG strain.

(A) 1-Hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine (CMH) probe incubated with KDD buffer before electron paramagnetic resonance (EPR) analysis. (B, C, D) EPR spectra ΔspeG strain with the plasmids pDAK1 (empty vector) or pSodA were grown without spermidine and performed EPR adding CMH probe. (E, F, G) ΔspeG strain with the plasmids pDAK1 (empty vector) or pSodA were grown with spermidine and performed EPR adding CMH spin probe, as mentioned in the Materials and methods. See also Figure 2—source data 1.

Figure 2—source data 1

Figure 2 Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig2-data1-v3.xlsx

O2- production under spermidine stress affects cellular redox state

Antioxidant chemicals viz. Tr, sodium pyruvate (SP), and thiourea (TU) scavenge O2-, H2O2, and •OH, respectively (Bleeke et al., 2004; Franco et al., 2007). Whereas, N-acetyl cysteine (NAC) and ascorbate counterbalance oxidative stress replenishing glutathione levels and donating electrons to reducing partners (Nimse and Pal, 2015; Sun, 2010). We show that Tr, NAC, and ascorbate, but not SP and TU, rescued the spermidine-mediated growth inhibition phenotype (Figure 3A). This observation further suggests that the O2- stress-derived redox imbalance could be the route of spermidine toxicity.

Figure 3
Download asset Open asset
O2- radical production affects redox balance in the spermidine-fed ΔspeG strain.

(A) Growth curves show that Tyron (Tr), ascorbate (Asc), and N-acetyl cysteine (NAC) can overcome spermidine (SPD) stress while sodium pyruvate (SP) and thiourea (TU) fail to do so. (B) Growth curves show that ΔspeGΔzwf strain is hypersensitive to SPD in comparison to ΔspeG strain. Complementation of ΔspeGΔzwf strain with pZwf plasmid overcomes this SPD hypersensitivity. (C) CFUs were obtained for different Escherichia coli strains pretreated with SPD for desired time points and plotted to show the reduced viability of ΔspeGΔzwf strain in comparison to theΔspeG strain. (D) Relative levels of NADPt and reduced nicotinamide adenine dinucleotide phosphate (NADPH) were significantly decreased in the ΔspeG strain under SPD stress. (E) Relative levels of GSt, GSH, and GSSG were significantly decreased in the SPD-fed ΔspeG strain. (F) No significant change in the relative total NAD (NADt), NAD+, and NADH levels were recorded. However, NAD+ to NADH ratio was significantly increased in the ΔspeG strain compared to WT cells. No further increase of the ratio was observed by adding SPD in the growth medium of WT and ΔspeG strain. (G) The relative level of ATP was declined in ΔspeG strain and spermidine-fed WT cells in comparison to the unfed WT. SPD supplementation decreased the ATP level further in the SPD-fed ΔspeG strain. Error bars in the panels are mean ± SD from the three independent experiments. Whenever mentioned, the *** and ** denote p-values < 0.001 and < 0.01, respectively; unpaired t test. See also Figure 3—source data 2, Figure 3—source data 3, Figure 3—source data 4, Figure 3—source data 5.

Figure 3—source data 1

Figure 3A Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig3-data1-v3.xlsx
Figure 3—source data 2

Figure 3B Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig3-data2-v3.xls
Figure 3—source data 3

Figure 3C Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig3-data3-v3.xlsx
Figure 3—source data 4

Figure 3D, E and F Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig3-data4-v3.xlsx
Figure 3—source data 5

Figure 3G Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig3-data5-v3.xlsx

The reduced nicotinamide adenine dinucleotide phosphate (NADPH) is a potent reducing agent. NADPH drives glutathione and thioredoxin cycles, thereby producing reduced forms of glutathione (GST), glutaredoxins, and thioredoxins to cope up with oxidative stress. A large fraction of NADPH in E. coli is provided by a glucose-6-phosphate 1-dehydrogenase (Zwf) catalyzed reaction (Olavarría et al., 2012). We show that both the growth and viability of ΔspeGΔzwf double mutant were significantly affected compared to the ΔspeG strain under spermidine stress (Figure 3B and C). Complementing ΔspeGΔzwf with a plasmid, pBAD-zwf, rescues the growth defect and mortality under spermidine stress (Figure 3B and C). We compared the levels of the total NADP (NADPt), total glutathione (GSt), and their oxidized (NADP+ and GSSG) and reduced (NADPH and GSH) species in the WT and ΔspeG strains grown in the absence and presence of spermidine. The relative levels of total and reduced species of NADP and GST were decreased significantly in the spermidine-fed ΔspeG strain (Figure 3D and E). NAD serves as the precursor for NADP production. However, the levels of total (NADt), oxidized (NAD+), and reduced (NADH) did not alter significantly (Figure 3F). Nevertheless, the NAD + to NADH ratio was significantly increased in the ΔspeG strain compared to WT cells (Figure 3F). No significant increase of the ratios was observed by adding spermidine in the growth medium of WT and ΔspeG strain (Figure 3F). In consistence with the increased ratio of NAD + to NADH, the level of ATP was declined in ΔspeG strain compared to the unfed WT (Figure 3G). ATP level was further decreased in the spermidine-fed ΔspeG strain (Figure 3G).

Spermidine blocks the induction of SoxR regulon

To understand the global impact of spermidine toxicity, we performed a microarray experiment on the ΔspeG strain in the presence and absence of spermidine. The genes that were >2-fold downregulated are involved in flagellar biogenesis, acid resistance, hydrogenase function, nitrogen metabolism, electron transport, aromatic and basic amino acid metabolism, etc. (Figure 4A and Supplementary file 1). Interestingly, transcription of the genes encoding chaperones, heat shock, and other stress factors (groL, groS, dnaK, hdeAB, ibpAB, uspAB, etc.) was also downregulated under spermidine stress (Supplementary file 1). On the other hand, among the highly upregulated category, the genes that encode for the ribosome, RNA polymerase, transcription factors, DNA polymerase, and enzymes for the fatty acid biosynthesis and iron-sulfur cluster (isc) biogenesis were prominent (Supplementary file 1 and Figure 4A). These observations indicate that apart from inducing superoxide production (Figures 1 and 2), the excess spermidine could interfere with broad cellular processes, such as protein folding and proteostasis, DNA, RNA and lipid metabolisms, and iron-sulfur cluster biogenesis. Many operons regulated by Fis and IHF were activated or repressed in our microarray indicating that spermidine could activate Fis and IHF regulon (Supplementary file 2). Performing Fisher’s exact test, we show that the differential expression of the Fis-regulated operons was significantly enriched (p-value 0.0023). Corroborating with this finding, we show that ΔspeGΔfis, but not ΔspeGΔihfA strain, generated small colonies upon overnight incubation (Figure 4—figure supplement 1), suggesting that the role of Fis regulator is critical under spermidine stress. Quantitative real-time PCR (RT-qPCR) experiment was performed to validate the microarray data partially (Figure 4—figure supplement 2).

Figure 4 with 2 supplements see all
Download asset Open asset
Spermidine blocks the activation of superoxide defense circuit.

(A) (i) Microarray heat map showing various categories of genes (I: Replication and transcription associated genes, II: Iron homeostasis, ROS, multidrug resistance and sugar metabolism genes, III: Ribosomal and ribosome biogenesis-associated genes, IV: Oxidoreductase and ATP synthesis genes, V: Fatty acid metabolism-related genes, VI: Flagellar biogenesis-related genes, VII: Acid resistance and chaperone genes, VIII: Hydrogenase and nitrogen metabolizing genes, IX: Amino acid metabolizing genes; see Supplementary file 1) that were differentially expressed under spermidine stress. (ii) Zoomed in heat map of the category II genes responsible for iron metabolism and reactive oxygen species (ROS) regulation. (iii) Color key represents the expression fold-change (FC) of the genes. (B) The subpanel (i) represents a flow cytometry experiment to demonstrate that spermidine (SPD) stress inhibits menadione (MD)-induced PsoxS-gfpmut2 reporter fluorescence. The subpanel (ii) represents relative mean fluorescence intensities (MFIs) in the presence or absence of SPD and MD calculated from three different flow cytometry experiments. (C, D, E) Western blotting experiments show SodA, KatG, and AhpC levels in the various strains in the presence or absence of SPD: (i) developed blot, (ii) ponceau S-stained counterpart of the same blot, (iii) the bar diagrams represent relative FC of the proteins under SPD stress. The relative FC values were calculated from the band intensity values obtained from three independent blots in comparison to the untreated WT counterparts. Purified 6X His-tagged SodA, KatG, and AhpC proteins were loaded as positive controls. The cellular protein extracts from ΔsodA, ΔkatG, and ΔahpC strains were used for negative controls. Whenever mentioned, the *** and ** denote p-values < 0.001, < 0.01, respectively; unpaired t test.Figure 4—source data 2., Figure 4—source data 3, Figure 4—source data 4, Figure 4—source data 5, Figure 4—source data 6, Figure 4—source data 7, Figure 4—source data 8, Figure 4—source data 9, Figure 4—source data 10, Figure 4—source data 11, Figure 4—source data 12, Figure 4—source data 13 and Figure 4—source data 14.

Figure 4—source data 1

Figure 4B–ii Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data1-v3.xlsx
Figure 4—source data 2

Figure 4C–i Raw unedited image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data2-v3.zip
Figure 4—source data 3

Figure 4C–i Raw uncropped and labeled image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data3-v3.zip
Figure 4—source data 4

Figure 4C–ii Raw full image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data4-v3.zip
Figure 4—source data 5

Figure 4C–ii Raw uncropped and labeled image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data5-v3.zip
Figure 4—source data 6

Figure 4D–i Raw full image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data6-v3.zip
Figure 4—source data 7

Figure 4D–i Raw uncropped and labeled image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data7-v3.zip
Figure 4—source data 8

Figure 4D–ii Raw full image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data8-v3.zip
Figure 4—source data 9

Figure 4D–ii Raw uncropped and labeled image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data9-v3.zip
Figure 4—source data 10

Figure 4E–i Raw full image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data10-v3.zip
Figure 4—source data 11

Figure 4E–i Raw uncropped and labeled image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data11-v3.zip
Figure 4—source data 12

Figure 4E–ii Raw full image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data12-v3.zip
Figure 4—source data 13

Figure 4E–ii Raw uncropped and labeled image.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data13-v3.zip
Figure 4—source data 14

Figure 4C, D and E Fold change values of the western blots.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig4-data14-v3.xlsx

Iron-sulfur center of SoxR senses the levels of cellular O2- or NO (Fujikawa et al., 2017; Hidalgo and Demple, 1994; Kobayashi, 2017; Liochev and Fridovich, 2011; Lo et al., 2012) and triggers transcription of a set of genes, including soxS, sodA, and zwf (Touati, 2000; Wu and Weiss, 1992). Surprisingly, none of the three critical genes was found to be activated in the microarray. RT-qPCR analyses verified the unaltered expression of soxS, sodA, and zwf under spermidine stress (Figure 4—figure supplement 2). Consistently, using ΔspeG harboring pUA66_soxS, a reporter plasmid expressing gfpmut2 from the soxS promoter (PsoxS-gfpmut2), and RKM1 strain containing a chromosomally fused lacZ reporter under sodA promoter (PsodA-lacZ) (Table 1), we did not find any transcriptional activation of soxS and sodA promoters (Figure 4—figure supplement 2). Therefore, we suspected whether spermidine in excess blocks the O2- -mediated activation of SoxR, thereby aggravating O2- toxicity. However, an alternative explanation for this observation would be that the redox cycling drugs, but not O2-, are the efficient activators of SoxR (Gu and Imlay, 2011). Therefore, we used menadione, a redox cycling agent and O2- generator, to observe the PsoxS-gfpmut2 reporter induction and chased it by spermidine in the ΔspeG strain. Spermidine also suppressed the menadione-induced GFP reporter fluorescence (Figure 4B), suggesting that spermidine indeed blocks SoxR-mediated activation of soxS in E. coli. A possible mechanism of spermidine-mediated SoxR inactivation is discussed. Among other ROS-responsive genes, the catalase coding genes (katE and katG) were downregulated (Figure 4A), while no change was observed in the expression of ahpCF genes under spermidine stress (GEO accession #154618). Using pUA66_ahpC and pUA66_katG reporter plasmids (PahpC-gfpmut2 and PkatG-gfpmut2, respectively), we validated these microarray observations (Figure 4—figure supplement 2).

Table 1
The list of strains and plasmids used in this work.
Strains and plasmidsGenotype/featuresReferences
Strains
BW25113Escherichia coli; rrnB3 ΔlacZ4787 hsdR514Δ(araBAD) 567 Δ(rhaBAD)568 rph-1Baba et al., 2006
ΔspeGBW25113, ΔspeG::kanRBaba et al., 2006
ΔsodABW25113, ΔsodA::kanRBaba et al., 2006
ΔsodBBW25113, ΔsodB::kanRBaba et al., 2006
ΔzwfBW25113, ΔsodC::kanRBaba et al., 2006
ΔfisBW25113, Δfis::kanRBaba et al., 2006
ΔihfABW25113, ΔihfA::kanRBaba et al., 2006
ΔiscUBW25113, ΔiscU::kanRBaba et al., 2006
ΔygfZBW25113, ΔygfZ::kanRBaba et al., 2006
ΔsoxSBW25113, ΔsoxS::kanRBaba et al., 2006
ΔmarABW25113, ΔmarA::kanRBaba et al., 2006
ΔmarBBW25113, ΔmarB::kanRBaba et al., 2006
ΔahpCBW25113, ΔahpC::kanRBaba et al., 2006
ΔkatGBW25113, ΔkatG::kanRBaba et al., 2006
ΔspeGΔsodABW25113, ΔspeG, ΔsodA::kanRThis study
ΔspeGΔsodBBW25113, ΔspeG, ΔsodB::kanRThis study
ΔspeGΔsodAΔsodBBW25113, ΔspeG, ΔsodA, ΔsodB::kanRThis study
ΔspeGΔzwfBW25113, ΔspeG, Δzwf::kanRThis study
ΔspeGΔsoxSBW25113, ΔspeG, ΔsoxS::kanRThis study
ΔspeGΔfisBW25113, ΔspeG, Δfis::kanRThis study
ΔspeGΔihfABW25113, ΔspeG, ΔihfA::kanRThis study
ΔspeGΔiscUBW25113, ΔspeG, ΔiscU::kanRThis study
ΔspeGΔygfZBW25113, ΔspeG, ΔygfZ::kanRThis study
ΔspeGΔmarABW25113, ΔspeG, ΔmarA::kanRThis study
ΔspeGΔmarBBW25113, ΔspeG, ΔmarB::kanRThis study
JRG3533MC4100 ф(sodA-lacZ)49, cmRTang et al., 2002
RKM1BW25113, ΔspeG, sodA-lacZ:cmRThis study
Plasmids
pET28a (+)kanR; T7-promoter; IPTG inducibleNovagen
pBAD/Myc-His AampR; pBAD-promoter; Ara inducibleThermoFisher
pDAK1pBAD/Myc-His A; Two NdeI sites were mutated and NcoI site was replaced by NdeILab resource
pZwfzwf cloned in pDAK1 NdeI and HindIII sitesThis study
pSodAsodA cloned in pDAK1 vectorThis study
pET-sodAsodA cloned in pET28a (+) vectorThis study
pET-ahpCahpC cloned in pET28a (+) vectorThis study
pET-katGkatG cloned in pET28a (+) vectorThis study
pSpeGspeG cloned in pDAK1 vectorThis study
pUA66_soxSkanR; soxS promoter cloned upstream of gfpmut2 reporter in pUA66Zaslaver et al., 2006
pUA66_ahpCkanR; ahpC promoter cloned upstream of gfpmut2 reporter in pUA66Zaslaver et al., 2006
pUA66_katGkanR; katG promoter cloned upstream of gfpmut2 reporter in pUA66Zaslaver et al., 2006
Note: kanR, kanamycin resistance; ampR, ampicillin resistance, and cmR, chloramphenicol resistance.

Consistent with the microarray expressions, our western blotting experiments exhibited the unchanged expression of SodA and a decreased expression of KatG in the spermidine-treated ΔspeG strain compared to untreated counterparts (Figure 4C and D). However, SodA level was modestly elevated in the ΔspeG strain, and the spermidine-treated WT strain, in contrast to the untreated WT strain (Figure 4C). Contrary to the microarray data, a profound increase in AhpC level was observed while growing WT or ΔspeG cells in the presence of spermidine, indicating a translational elevation of AhpC level under spermidine stress (Figure 4E). Increased AhpC level indicating the activation of alkyl hydroperoxidase (AhpCF) enzyme could be responsible for the decline in cellular H2O2 level (Figure 1C). Thus, declined H2O2 concentration could be the limiting factor for the cellular •OH radical production under spermidine stress (Figure 1A).

Spermidine affects iron-sulfur cluster biogenesis

O2- has the potential to oxidize the solvent-exposed iron-sulfur clusters of E. coli dehydratases, aconitase, and fumarase enzymes to liberate free Fe2+ (Benov, 2001; Fridovich, 1986; Imlay, 2008). Therefore, supplementation of Fe2+ ions helps to repair the damaged clusters (Gardner and Fridovich, 1992; Imlay, 2008). Consistently, we observed that the declined aconitase activity in the spermidine-stressed ΔspeG strain was rescued by supplemental Fe2+ ion (Figure 5A). Besides, the intracellular level of iron in the ΔspeG strain was decreased more than 3-fold in the presence of spermidine (Figure 5B). Consequently, the supplementation of Fe2+ salt in the LB-agar plate rescued the growth of spermidine-fed ΔspeG strain supports this claim (Figure 5C).

Figure 5 with 1 supplement see all
Download asset Open asset
Spermidine-mediated O2- radical production affects iron metabolism.

(A) The bar diagram represents relative aconitase activity in the Escherichia coli WT and ΔspeG strains in the presence and absence of spermidine (SPD). (B) Intracellular levels of Fe in the E. coli strains determined in the presence or absence of SPD stress were plotted. (C) Spot assay using serially diluted ΔspeG cells demonstrated that Fe2+ can rescue SPD stress. (D) Intracellular levels of Mn levels in the E. coli strains determined in the presence or absence of SPD stress were plotted. (E) Spot assay shows the relative sensitivity of various double mutants, ΔspeGΔygfZ, ΔspeGΔiscU, and ΔspeGΔsoxS strains to SPD. Error bars in the panels are mean ± SD from the three independent experiments. Whenever mentioned, the ***, **, and * denote p-values < 0.001, < 0.01, and < 0.1 respectively; unpaired t test. See also Figure 5—figure supplement 1, and Figure 5—source data 1, Figure 5—source data 2, Figure 5—source data 3.

Figure 5—source data 1

Figure 5A Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig5-data1-v3.xlsx
Figure 5—source data 2

Figure 5B Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig5-data2-v3.xlsx
Figure 5—source data 3

Figure 5D Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig5-data3-v3.xlsx

The iron scarcity was also reflected in the gene expression pattern of IscR regulon (Figure 4A). IscR forms a functional holoenzyme with the iron-sulfur cluster. The de-repression of iron-sulfur cluster biogenesis operon (iscRSUA-hscBA-fdx-iscX) in the microarray (Figure 4A) signifies the presence of non-functional apo-IscR under the scarcity of cellular Fe2+ ion (Esquilin-Lebron et al., 2021; Schwartz et al., 2001). Besides, apo-IscR and apo-Fur activate and derepress the alternative iron-sulfur cluster assembly system (sufABCDSE), respectively (Esquilin-Lebron et al., 2021; Outten et al., 2004). Interestingly, no genes of the suf operon were found to be upregulated under spermidine stress. Instead, a 3-fold downregulation of sufA was observed (Figure 4A). Since suf operon is also positively regulated by OxyR (Esquilin-Lebron et al., 2021) but spermidine stress declined the cellular H2O2 levels (Figure 1C), we suggest that the combined action of apo-IscR, apo-Fur, and inactivated form of OxyR kept suf operon expression indifferent under spermidine stress. Spermidine also activated rsxA and rsxB (Figure 4A), which encode the critical components of the iron-sulfur cluster reducing system of SoxR (Koo et al., 2003).

The level of manganese, an antioxidant metal that determines sodA activity, is usually increased under iron scarcity (Kaur et al., 2017; Kaur et al., 2014; Martin et al., 2015; Waters et al., 2011). However, a modest decrease in the level of cellular manganese under spermidine stress was observed (Figure 5D). The low level of manganese could slow down the rate of dismutation of O2- anion compromising SodA function, thereby elevating the O2- anion levels in the spermidine-treated cells. Finally, we spotted the serially diluted cultures of E. coli strains to show that the deletion of two individual genes (iscU and ygfZ), which are involved in the iron-sulfur cluster biogenesis (Waller et al., 2010), affects the growth of the spermidine-treated ΔspeG strain (Figure 5E). Interestingly, the ΔspeGΔsoxS strain was more sensitive to spermidine than the ΔspeG strain (Figure 5E), indicating that the basal level of soxS expression has some potential to ameliorate O2- under spermidine stress. Although marA and marB genes were expressed at the highest level in the spermidine-stressed ΔspeG strain (Figure 4A), ΔspeGΔmarA and ΔspeGΔmarB strains did not show any difference in growth compared to ΔspeG strain under spermidine stress (Figure 5E). Note that, unlike ΔspeG strain, the single mutants, viz. ΔmarA, ΔmarB, ΔygfZ,ΔiscU, andΔsoxS, grow similarly to the WT strain in the presence or absence of spermidine (Figure 5E and Figure 5—figure supplement 1).

Free spermidine interacts and oxidizes Fe2+ ion to generate superoxide radicals in vitro

To probe whether spermidine directly interacts with iron, we performed isothermal titration calorimetry (ITC) using Fe3+ (ferric citrate) and Fe2+ (ferrous ammonium sulfate) ions. Titration of spermidine with Fe3+ generated exothermic peaks indicating a standard binding reaction with a stoichiometry (N) of 0.711 (Figure 6A). On the other hand, titration of spermidine with Fe2+ in two different isothermal conditions produced consistent and complex patterns (Figure 6B and C). To explain it, we divided the pattern into two halves. In the first half, Fe2+ injections to spermidine generated alternate exothermic and endothermic peaks till the ratio of spermidine to Fe2+ reaches about 1:1.3 (Figure 6B and C). In the second half of the profile, after the ratio of spermidine to Fe2+ crosses 1:1.3, no endothermic peaks were observed, and a gradual shortening of exothermic peaks was generated, leading to saturation (Figure 6B and C). From the first half of pattern, we suspected Fe2+ interaction with spermidine also involves some other reactions, such as oxidation of the Fe2+ to generate Fe3+ and O2-, Fe3+ release, and subsequent Fe3+ binding to spermidine.

Figure 6
Download asset Open asset
Spermidine oxidizes Fe2+ generating O2- radical in aerobic condition.

(A) Isothermal titration calorimetry (ITC) data demonstrates the interaction of spermidine with Fe3+. (B) and (C) ITC data shows the interaction of spermidine with Fe2+ ion at 4°C and 25°C, respectively. (D) 100 µM spermidine was incubated with different concentrations of Fe2+ followed by estimation of Fe2+ levels by bipyridyl chelator. The color formation was recorded at 522 nm and plotted them along with standard curve. The panel depicts that the incubations of 100 µM spermidine with 100, 200, and 300 µM of Fe2+ in the anaerobic condition do not lead to the loss of Fe2+ ions detected by bipyridyl chelator. However, when 100 µM spermidine was incubated with the different concentrations of Fe2+ (25–350 µM) in the aerobic condition, the bipyridyl-mediated color formation was observed when Fe2+ level was between above 125 µM and 150 µM (i.e., till spermidine to Fe2+ ratio reaches approximately 1.3). The mean values from the three independent experiments were plotted. SD is negligible and is not shown for clarity. (E) Nitro blue tetrazolium (NBT) assay was performed to determine that spermidine and Fe2+ interaction yields O2- radical. The colorimetry at 575 nm suggests that 100 µM of spermidine interacts with approximately 125 µM of Fe2+ (ratio 1:1.3) to generate saturated color. Error bars in the panel are mean ± SD from the three independent experiments. *** denotes p-value < 0.001; unpaired t test. (F) Model to show final coordination complex formation. An Fe2+ interacts with two spermidine molecules forming hexadentate coordination complex. This interaction oxidizes Fe2+ liberating one electron to reduce oxygen molecule. Finally, two spermidine coordinates one Fe3+ with an octahedral geometry. (G) The curves represent the Escherichia coli total RNA inhibits iron oxidation. Spermidine further reduces the RNA-mediated iron oxidation at concentration 10 µM but higher concentrations of spermidine increase the iron oxidation despite the presence of RNA. The mean values are derived from the three independent experiments and plotted. SD is negligible and is not shown for clarity. See also Figure 6—source data 1, Figure 6—source data 2, Figure 6—source data 3.

Figure 6—source data 1

Figure 6D Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig6-data1-v3.xlsx
Figure 6—source data 2

Figure 6E Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig6-data2-v3.xlsx
Figure 6—source data 3

Figure 6G Raw data.

https://cdn.elifesciences.org/articles/77704/elife-77704-fig6-data3-v3.xlsx

To test whether Fe2+ was oxidized in the presence of spermidine to liberate Fe3+, we titrated spermidine by increasing amounts of Fe2+ iron followed by assessing the level of Fe2+ by using bipyridyl chelator. Chelation of Fe2+ ions by bipyridyl generates pink color indicating Fe2+ levels. No color formation was observed till the ratio of spermidine to Fe2+ reaches 1:1.3 (Figure 6D), a number that exactly matches with the ratio of spermidine to Fe2+ in the first half of ITC experiments (Figure 6B and C). The color formation starts appearing when the ratio crosses 1:1.3 (Figure 6D), suggesting that 1 molecule of spermidine (or 10 molecules) exactly oxidizes 1.3 molecules (or 13 molecules) of Fe2+. The colorimetric values overlap with the standard curve when reactions were under anoxic condition, indicating Fe2+ was not oxidized (Figure 6D). We used nitro blue tetrazolium (NBT) dye to check whether the loss of one electron from Fe2+ generates O2- anion under spermidine stress. An increased NBT absorption at 575 nm till the ratio of spermidine to Fe2+ reaches 1:1.3 confirms that 1 molecule (or 10 molecule) of spermidine interacts with 1.3 molecules (or 13 molecules) of Fe2+ generating 1.3 molecules (or 13 molecules) O2- anion radical (Figure 6E). From the stoichiometry of 0.711 (which is close to 0.5) (Figure 6A), we postulate that two spermidine and one Fe3+ together could form a hexadentate coordination complex with an octahedral geometry (Figure 6F). It appears that when spermidine molecules engaged to form a hexadentate coordination complex with Fe2+, the former helps oxidizing latter to form Fe3+ in sufficient concentrations. Fe3+ finally forms coordination complex with spermidine (Figure 6F). It may be noted that the binding of spermidine and Fe3+ is entirely enthalpy-driven, as indicated by a large negative ΔH. The negative entropy (ΔS) value presumably results from the ordering of spermidine from an extended conformation to a compact and rigid one after metal chelation (Figure 6A).

The cellular spermidine barely exists as a ‘free’ species; rather, majority of them remain ‘bound’ with RNA, DNA, nucleotides, and phospholipids (Igarashi and Kashiwagi, 2000; Miyamoto et al., 1993; Schuber, 1989; Tabor and Tabor, 1984). It has been reported that these phosphate-containing biomolecules have the inherent property to inhibit iron oxidation blocking O2- production (LØVaas, 1996; Tadolini, 1988a; Tadolini, 1988b). The bound spermidine further enhances the inhibitory effects of these biomolecules toward iron oxidation. Consistent with the report (Tadolini, 1988b), we noticed that 1 μg of RNA inhibited the oxidation of 200 µM Fe2+. The presence of 10 μM spermidine further decreased iron oxidation (Figure 6G). However, increasing the concentrations of spermidine (50, 100, and 200 μM) accelerated iron oxidation gradually (Figure 6G). This data clearly indicates that cell maintains a level of cellular spermidine that may remain optimally bound with the biomolecules inhibiting O2- generation. However, when homeostasis fails due to speG deletion, excess spermidine accumulates that can remain in a ‘free’ form inducing O2- radical toxicity.

Discussion

Our study presented in this paper answers why spermidine homeostasis is intriguingly fine-tuned in bacteria. We provide clear-cut evidence that excess spermidine, which remains as a free species (Figure 6G), stimulates the production of toxic levels of O2- radicals in E. coli (Figures 1 and 2). O2- anion thus generated affects cellular redox balance (Figure 3) and damages iron-sulfur clusters of the proteins (Figures 4 and 5). Since spermidine directly interacts with Fe2+ (Figure 6), it may abstract iron from some of the iron-sulfur clusters, thereby inactivating some of the proteins. On the other hand, when spermidine level is at optimum, most of it remain as bound form with the biomolecules, thereby slows down iron oxidation and subsequent O2- production (LØVaas, 1996; Tadolini, 1988a; Tadolini, 1988b; Figure 6G). Thus, spermidine deficiency would enhance the rate of iron oxidation (Figure 6G), leading to ROS production (Figure 1A and B). This is why spermidine is a double-edged sword where in excess, it provokes O2- anion production, and in scarcity, it leads to higher ROS levels.

Polyamines remain protonated at physiological pH, yet they are able to coordinate several positively charged metal ions, such as Ni2+, Co2+, Cu2+, and Zn2+, possibly via charge neutralization by counterions that reduces the Coulombic repulsion between spermidine and the metals (LØVaas, 1996). Similar charge neutralization of the nitrogen atoms of spermidine likely allows coordinate covalent bonds with Fe3+ (Figure 6F). About 10 spermidine molecules oxidize Fe2+ to generate 13 Fe3+ cations and equivalent numbers of O2- radicals (Figure 6B and C). When sufficient concentration of Fe3+ is generated, two spermidine molecules coordinate one Fe3+ to form a hexadentate complex with an octahedral geometry (Figure 6F). We substantiated this in vitro spermidine-mediated iron oxidation and subsequent O2- radical production phenomena (Figure 6), showing that cells are highly toxic to the spermidine under aerobic condition but not under anaerobic condition (Figure 1D).

Usually, abundant O2- level leads to general ROS including H2O2 production. However, despite elevated O2- production, spermidine lowers overall ROS levels in ΔspeG strain (Figures 1C, A, 2). The declined H2O2 level could be attributed to the slower rate of O2- anion dismutation due to the failure of sodA activation (Figure 4—figure supplement 2) and the activation of alkyl hydroperoxidase (AhpCF) that neutralizes H2O2, represented by AhpC overexpression (Figure 4E). A low level of cellular manganese (Figure 5D) could also limit SodA activity. Besides, the activation of IscR regulon (Figure 4A), the low cellular iron content (Figure 5B), and the rejuvenation of cell growth by Fe2+ supplementation (Figure 5C) indicate that the spermidine presumably lowers the Fe2+/Fe3+ ratio in ΔspeG strain. Thus, the decreased level of Fe2+ and H2O2 (Figures 5B and 1C) could potentially diminish cellular •OH radical production in the spermidine-fed cells. We have summarized all these observations and hypotheses in the schematic Figure 7.

Figure 7
Download asset Open asset
Flowchart explaining the reactive oxygen species (ROS) generation under spermidine stress.

The model describes that the spermidine administration in the cell interacts with free iron and oxygen to generate O2- radical, increasing Fe3+/Fe2+ ratio. Spermidine also blocks O2- radical-mediated activation of SoxRS that upregulates zwf and sodA. Consequently, reduced nicotinamide adenine dinucleotide phosphate (NADPH) production and dismutation of O2- radical to H2O2 were not accelerated, leading to redox imbalance and O2- -mediated damage to the iron-sulfur clusters, respectively. Additionally, spermidine translationally upregulated alkyl hydroperoxidase (AhpCF) that lowers the level of H2O2. Declined cellular Fe2+ and H2O2 levels weaken Fenton reaction to produce •OH radical.

Interestingly, spermidine stimulates O2- production but SoxR function remained indifferent in the ΔspeG cells (Figure 4B). This observation is consistent with the previous finding that redox cycling drugs, but not O2-, are the efficient activators of SoxR function (Gu and Imlay, 2011). Even spermidine blocked SoxR expression by menadione, a redox cycling drug (Figure 4B). These two observations implicate that free spermidine being an iron chelator (Figure 6A, B and C) might affect SoxR maturation by interfering its iron-sulfur cluster formation. As a result, apo-SoxR remained unreactive to the superoxide or redox cycling drugs, and thereby failed to activate SoxR regulon genes. Since spermidine ubiquitously interacts with DNA and modulates gene expression in many ways (Igarashi and Kashiwagi, 2000; Jung and Kim, 2003; Miyamoto et al., 1993), another possibility could be that excess of it might occlude SoxR-binding to the soxS and sodA promoter regions to activate them. Alternatively, blockage of SoxR activation could result from spermidine-mediated activation of rsxA and rsxB (Figure 4A), which encode the critical components of the iron-sulfur cluster reducing system of SoxR (Koo et al., 2003), to keep SoxR inactive. Nevertheless, a detailed biochemical study on this aspect is needed to understand the mechanism.

Our study in E. coli observed quite a few biochemical aspects which might explain how the horizontal acquisition of speG gene could confer a pathogenic advantage to the Staphylococcus aureus USA 300 strain (Eisenberg et al., 2009). S. aureus, a Gram-positive commensal living on human skin, often causes severe disease upon access to deeper tissues. Since most of the iron in mammals exists intracellularly, the extracellular pathogen, S. aureus faces hardship and competes with the host for the available iron (Hammer and Skaar, 2011). As spermidine declines cellular iron content and interferes with iron metabolism (Figure 4), it is thus possible that S. aureus does not synthesize spermidine (Joshi, 2012). Furthermore, the acquisition of speG gene by S. aureus USA300 (Joshi, 2012) could allow it to inactivate host-originated spermidine/spermine, thereby to maintain cellular iron content. Corroborating to our findings, a recent observation has pointed out that spermine stress upregulates iron homeostasis genes, indicating that spermine toxicity has a specific connection with iron depletion in the speG-negative S. aureus strain, Mu50 (MRSA) (Yao and Lu, 2014). Besides, spermine-mediated iron depletion may be responsible for the synergistic effect of spermine with the antibiotics against S. aureus (Kwon and Lu, 2007). Nevertheless, a thorough in vivo host-pathogen interaction study may unravel a specific link between spermine/spermidine and iron depletion in S. aureus.

Materials and methods

Bacterial strains, plasmids, proteins, and chemicals

Request a detailed protocol

Bacterial strains and plasmids used in this study are listed in Table 1. BW25113 strain of E. coli was used as WT in this study. Oligonucleotides were purchased from IDT. Bacterial broths and agar media were purchased from BD Difco. The knockout strains of E. coli were procured from the KEIO library (Baba et al., 2006), verified by PCR, freshly transduced into the WT background by P1 phage, and sequenced to confirm the deletion. The double and triple knockout mutants were generated following the standard procedure described by Datsenko and Wanner, 2000. E. coli strain JRG3533 was a generous gift from Dr Rachna Chaba, IISER Mohali, India. RKM1 strain was constructed by P1 transduction of sodA-lacZ:CmR genotype of JRG3533 to BW25113ΔsoxS strain.

The plasmids, pUA66_soxS, pUA66_ahpC, pUA66_katG, were the gifts from Dr Csaba Pal, Biological Research Centre of the Hungarian Academy of Sciences (Zaslaver et al., 2006). pBAD-zwf was a generous gift from Dr CC. Vasquez, Universidad de Santiago de Chile (Sandoval et al., 2011). sodA, katG, ahpC, and speG genes were PCR-amplified by DG12-DG13, RM7-RM8, DG9-DG10, and RK3-RK4 primer pairs (Supplementary file 3), respectively. The PCR products were double-digested at the primer-specific unique restriction sites and inserted into identically digested pET28a (+) plasmid vector so that the 6X His-tagged SodA, KatG, and AhpC proteins are being produced (Table 1). The protein expression vectors, pET-sodA, pET-ahpC, pET-katG, were transformed to BL21 (DE3) cells, and expressions were induced by 0.4 mM IPTG. The overexpressed proteins were purified using Ni-NTA beads. The purified proteins were used to raise rabbit polyclonal antibodies following the standard procedure. sodA and speG were additionally subcloned in pDAK1, a derivative of pBAD/Myc-His A vector to get pSodA, and pSpeG multicopy expressions for complementation assays (Table 1). We also PCR-amplified zwf using RK55-RK56 primer pairs and cloned in the pDAK1 vector to get pZwf vector for complementation assays.

Growth, viability, spermidine sensitivity, and complementation assays

Request a detailed protocol

An automated BioscreenC growth analyzer (Oy growth curves Ab Ltd) was used to generate growth curves mentioned in the Results. For this purpose, overnight cultures of different strains were diluted in fresh LB medium and grown in the presence and absence of 3.2–6.4 mM of spermidine. Ten mM of each of the ROS quenchers (TU, Tr, SP, ascorbate, and NAC) were used wherever mentioned. For growth assay of E. coli strains on the LB-agar supplemented with or without spermidine were performed by spotting serially diluted overnight cultures and growing them at 37°C. For viability assays, serially diluted E. coli strains were spread on LB-agar surface supplemented with 6.4 mM spermidine. We determined the viability under spermidine stress from the number of colonies grown. ZOIs, which appeared following overnight growth of the strains in the presence of 6.4 mM spermidine in the wells on agar plates, were determined both in aerobic and in anaerobic conditions. The anaerobic condition was created in an anaerobic Petri dish jar using AnaeroGas Pack 3.5 l pouches. For complementation experiments, the pSodA, pSpeG, and pZwf plasmids were transformed into ∆speG∆zwf and ∆speG∆sodA strains, respectively, and growth assays were performed in the presence of spermidine. Since the leaky expressions were sufficient to rescue growth defects, induction with arabinose was avoided for this purpose.

The reporter plasmids, pUA66_soxS, pUA66_ahpC, pUA66_katG, were transformed into ∆speG strain. The transformed cells were grown in the presence or absence of 3.2 mM spermidine. Wherever mentioned, 25 µM menadione was used as a positive control for O2- generation. The cell pellets were washed twice with PBS and dissolved in 500 µl phosphate buffer saline (PBS). Flow cytometry was done using the Fl1 laser for 0.05 million cells using FACSVerse (BD Biosciences). The MFI values from three biological replicates have been calculated.

Determining relative ROS levels in the cells

Request a detailed protocol

H2DCFDA (10 µM) and DHE (2.5 µM) were used to measure cellular •OH and O2- anion, respectively. The cells were grown in the presence or absence of 3.2 mM spermidine. Cells were harvested, washed with PBS, and an equal mass of cell pellets was incubated with DHE or H2DCFDA probes for an hour. The data were acquired using BD accuri Fl3 laser (for DHE) and Fl1 laser (for H2DCFDA) for 0.05 million cells. The MFI values of triplicate experiments were calculated. For H2O2 detection, the E. coli cells were grown in the presence or absence of 3.2 mM of spermidine for 4 hr. Cells were harvested and washed with 1× M9 minimal media. The equal mass of cells (2.5 mg each) suspended in 6 ml M9 minimal media were incubated for different time points to allow H2O2 liberation. The relative H2O2 liberation was measured by a Fluorimetric Hydrogen Peroxide Assay kit (Sigma Aldrich).

EPR spectroscopy

Request a detailed protocol

The protocol was adopted from Thomas et al., 2015, with some modifications. The ΔspeG strain harboring pDAK1 empty vector or pSodA was grown in the presence or absence of 3.2 mM spermidine for 2 hr and then 0.001% arabinose was added and further grown for 2 more hours; 100 mg cell pellets were quickly resuspended in 700 µl of KDD buffer, pH 7.4 (99 mM NaCl, 4.69 mM KCl, 2.5 mM CaCl2, 1.2 mM MgSO4, 25 mM NaHCO3, 1.03 mM KH2PO4, 5.6 mM D-glucose, 20 mM HEPES, 5 µM DETC, and 25 µM deferoxamine); 100 µl cell suspensions were preincubated with or without 20 mM DMTU and 200 µM UA for 5 min; and 500 µM of CMH spin probe (Enzo Life Sciences) were added and incubated for 30 min at 37°C. EPR spectra were acquired using a Bruker EMX MicroX EPR spectrometer with the following settings: center field, 3438 G, sweep width, 500 G; microwave frequency, 9.45 GHz; microwave power, 8.04 mW; modulation frequency 100 kHz; modulation amplitude, 5.64 G; conversion time, 40 ms; time constant, 40.96 ms; receiver gain, 1120; data points 1024; number of X-Scans, 5.

β-Galactosidase and GFP reporter assays

Request a detailed protocol

For the β-galactosidase assay, the RKM1 strain was grown in the presence or absence of 3.2 mM of spermidine. The cell pellets were washed twice with Z-buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, and 1 mM MgSO4) and diluted to OD600 ~0.5. Promoter activity was measured by monitoring β-galactosidase expression from single-copy sodA-lacZ transcriptional fusion; 100 µl of 4 mg/ml ONPG was used as a substrate, which was cleaved by β-galactosidase to produce yellow-colored O-nitrophenol. Colorimetric detection of this compound was done at 420 nm.

The reporter plasmids, pUA66_soxS, pUA66_ahpC, pUA66_katG, containing GFP-mut2 reporters, were used to determine the promoter activities of soxS, ahpC, and katG genes in the presence or absence of 3.2 mM spermidine. Flow cytometry was done using the FL1 laser for 0.05 million cells using FACSVerse (BD Biosciences) or BD Accuri C6 Plus Flow Cytometer (BD Biosciences) machine.

Western blotting experiments

Request a detailed protocol

Overnight culture of E. coli strains was inoculated in fresh LB medium in 1:100 dilution and grown for 1.5 hr at 37°C. Next, 3.2 mM of spermidine were added, wherever required and allowed to grow again at 37°C for 2.5 hr. Cells were harvested and lysed with B-PER bacterial protein extraction reagent (Thermo Scientific). The total protein level was checked by the Bradford assay kit (Bio-Rad); 40 µg of total cellular proteins from the individual samples were subjected to SDS-PAGE. The proteins were transferred to a nitrocellulose membrane and stained with Ponceau S to visualize protein resolution and equal loading in the PAGE. Western blotting was performed using polyclonal rabbit primary antibodies and HRP-conjugated secondary antibodies. The blots were developed by Immobilon Forte Western HRP substrate (Millipore).

Estimating cellular spermidine levels

Request a detailed protocol

Cells were grown in presence or absence of spermidine for 4 hr. The cells were washed with 1 M NaCl at 37°C for 10 min; 500 nmol of hexane-diamine (internal standard) was added and the pellets were resuspended in 750 µl of 10% perchloric acid. The cells were lysed by freeze-thawing using liquid nitrogen, and 800 µl of saturated sodium carbonate and 800 µl of 10 mg/ml of dansyl chloride were added to the supernatants. The dansylation was carried out at 60°C for 3 hr in dark. The reaction was stopped using 400 µl of 100 mg/ml proline and kept at 60°C for 30 min; 400 µl toluene was added to each sample and mixed thoroughly. The organic layer was collected and dried using a speed vac; 2 ml 80% acetonitrile was added and sonicated to dissolve the dry samples. The samples were then passed through 0.22 µm filter and injected to HPLC system (Agilent 1260 Infinity II) attached with a reversed-phase C-18 column (Agilent ZORBAX Eclipse Plus C18 of dimension 4.6 × 100 mm, 3.5 µm). Acetonitrile gradient (0–100%) with 0.8 ml/min flow rate was used for all samples. A PDA detector was used to monitor the elution peaks. The corresponding mass of individual peaks were detected using either a single quadrupole Agilent MSD using the ESI source or a separate Agilent LC-MS/MS equipment. Pure spermidine and hexane-diamine were also dansylated and determined their 100% tri- or di-dansylation. The dansylated spermidine was also used to generate a standard curve. The peak areas of spermidine (mAu*s) were normalized with the average peak area of internal standards. The absolute amounts of spermidine were calculated from the standard curve.

Isothermal titration calorimetry

Request a detailed protocol

A MicroCal VP-ITC calorimeter, MicroCal Inc, was used for calorimetric measurements to probe the interaction of spermidine with Fe2+ and Fe3+ species. In order to achieve this, 100 µM of spermidine solution was prepared in 20 mM sodium acetate buffer (pH 5.5) and put into the sample cell. The ligands, 2.1 mM of FeCl3 or ferrous ammonium sulfate, were also dissolved in the identical sodium acetate buffer. The titrations involved 30 injections of individual ligands (5 µl per shot) at 300 s intervals into the sample cell containing 1.8 ml of 100 µM spermidine. The titration cell was kept at some specific temperature and stirred continuously at 286 rpm. The heat of dilution of ligand in the buffer alone was subtracted from the titration data. The data were analyzed using Origin 5.0 software.

2,2′-Bipyridyl and NBT assays

Request a detailed protocol

2,2′-Bipyridyl chelates Fe2+ producing color that absorbs at 522 nm (A522). The standard curve for 0–350 µM of Fe2+ ion was generated simply by recording A522 in the presence of 2,2′-bipyridyl. Dissolved oxygen of medium and headspace oxygen was replaced by flushing N2 gas in the medium for 5 min to create an anoxic condition as described (Stieglmeier et al., 2009). To check whether spermidine acts as a catalyst for Fe2+ to Fe3+ oxidation, we performed 2,2′-bipyridyl assay probing leftover Fe2+ after the reaction. For this assay, 100 µM of spermidine was incubated with increasing concentrations (25–350 µM) of ferrous ammonium sulfate for 10 min at room temperature (RT); 900 µl of the reaction products were mixed with 90 µl 4 M sodium acetate buffer (pH 4.75) and 90 µl bipyridyl (0.5% in 0.1 N HCl). The color formation was recorded at 522 nm (A522) using UV-1800 Shimandzu UV-spectrophotometer. In another experiment, the assay was performed in anoxic condition using rubber-capped sealed glass vials containing anoxic reactants and needle-syringe-mediated mixing of the reagents. Here, three different concentrations (100, 200, and 300 µM) of ferrous ammonium sulfate were reacted with 100 µM of spermidine for 10 min followed by spectrophotometry at A522. The standard curve for 0–350 µM of Fe2+ ion was generated simply by recording A522 of the mixture of 900 µl ferrous ammonium sulfate, 90 µl sodium acetate buffer, and 90 µl bipyridyl solutions.

Iron oxidation in the presence of RNA and spermidine was performed as described (Tadolini, 1988b). One µg RNA and increasing concentrations of spermidine (10–200 µM) were used in 5 mM MOPS buffer, pH 7.4. The oxidation was started adding 200 µM FeCl2. The reactions were stopped at desired time point by adding a stop solution (1:1 4 M sodium acetate:4 M glacial acetic acid) followed by 2,2’-bipyridyl to detect Fe2+ levels.

We used NBT dye to probe whether spermidine-stimulated Fe2+ to Fe3+ oxidation liberates O2- anion in vitro. For this assay, different concentrations of Fe2+ were incubated with 100 µM of spermidine for 2 min; 100 µl of NBT (5 mg/ml) was added to the mixture and incubated at RT for another 5 min. The absorbance was recorded at 575 nm using UV-1800 Shimandzu UV-spectrophotometer.

RT-qPCR

Request a detailed protocol

Bacterial mRNAs were isolated by TRIzol reagent and the Qiagen bacterial RNA isolation Kit. DNase I treatment was done to remove residual DNA contaminant, and the integrity of the mRNA was checked on a 1% agarose gel. The RNA concentration was determined by a Nano-drop spectrophotometer (Thermo Scientific) and by a UV-1800 Shimandzu UV-spectrophotometer; 200 ng of RNA samples, primer pairs (Supplementary file 3), and GoTaq 1-Step RT-qPCR System (Promega) were used for RT-qPCR. Reaction mixture without template were included as negative controls. At least three independent experiments were conducted for the determination of cycle threshold (CT) values. Fold expression change between spermidine-fed and unfed samples was calculated by the ΔΔCT method. The values were normalized to the level of betB mRNA that was expressed constitutively as observed in the microarray.

Other biochemical assays

Request a detailed protocol

The relative levels of cellular NAD+/NADH and NADP/NADPH were measured using MAK037 and MAK038 kits (Sigma), respectively. ATP Bioluminescence assay Kit CLS II (Roche) were used to determine cellular ATP levels. The glutathione assay was performed, as described (Rahman et al., 2006). Cells were grown in the presence or absence of 3.2 mM spermidine for 4 hr. The PBS-washed cell pellets were kept on the ice.

For NAD+/NADH and NADP/NADPH assays, 30 mg of cell pellets were dissolved in 400 µl of extraction buffer supplemented with 50 µg/ml of lysozyme and sonicated. The supernatants were collected and passed through 10 kDa spin columns; 10 µl of 0.1 N HCl or 0.1 N NaOH were added slowly for NAD+ or NADH levels, respectively. On the other hand, 10 µl of 0.1 N NaOH or 0.1 N HCl were added slowly for NADP or NADPH levels, respectively. The samples were incubated at 60°C for 50 min; 50 µl of samples were mixed with the kit-specific 98 µl cycling buffer, and 2 µl cycling enzyme mix, and incubated at RT for 1 hr. Then, 10 µl of NADH or NADPH developer substrates were added in dark. A450 were recorded and the colorimetric values were directly used to calculate the relative levels of the individual species.

For ATP estimation, 30 mg cell pellets were resuspended in 100 mM Tris-HCl (pH 7.75), 4 mM EDTA, and then incubated in boiling water for 2 min. The supernatants were collected and kept on ice; 50 µl of the supernatants and 50 µl of luciferase reagent were mixed taken in 96-well, flat-bottom black microwell plate. Luminescence was measured using BIOTEK plate reader and the values were directly used to represent relative ATP levels.

For GSt assay, 20 mg E. coli cell pellets were resuspended in 5% sulfosalicylic acid and boiled at 95°C for 5 min; 100 µl supernatant was mixed with 700 µl KPE buffer, 0.6 mM DTNB, and 0.3 units of glutathione reductase; 0.2 mM of β-NADPH was added finally. To estimate oxidized form of glutathione (GSSG) only, the cell extracts were pretreated with 10 mM 2-vinylpyridine for 1 hr so that GSH were cross-linked with it. The excess 2-vinylpyridine was neutralized with tri-ethanolamine. The reactions were carried out for 30 min and A412 was recorded and the values were directly used to calculate the relative levels of each species. Aconitase assay was performed as per the protocol described (Gardner and Fridovich, 1992). Metal contents were determined by ICP-MS analyses at Punjab Biotechnology Incubator, Mohali, India. The metal concentration in the cell was determined as parts per billion (mg/kg) of E. coli cell pellets.

Microarray experiments and interpretation

Request a detailed protocol

The saturated overnight culture of ΔspeG strain was inoculated in the fresh LB medium and grown for 1.5 hr. After that 3.8 mM spermidine was added to one of the flasks, and the cultures were grown further for 2.5 hr. The cell pellets were harvested and washed with PBS, and dissolved in RLT buffer. The microarray was done from Genotypic Technology, Bangalore. The microarray had three probes for each gene on average.

RNA extraction and RNA quality control for microarray

Request a detailed protocol

E. coli cell pellet was resuspended in 300 µl of 5 mg/ml lysozyme and incubated at RT for 30 min. Isolation of RNA from E. coli was carried out using Qiagen RNeasy mini kit (Cat # 74106) as per manufacturer’s guidelines. A separate DNase treatment of the isolated total RNA was performed. The purity of the RNA was assessed using the Nanodrop Spectrophotometer (Thermo Scientific; ND-1000), and the integrity of the RNA was analyzed on the Bioanalyzer (Agilent 2100). We considered RNA to be of good quality based on the 260/280 values (Nanodrop), rRNA 28S/18S ratios, and RNA integrity number (RIN) (Bioanalyzer).

Microarray labeling

Request a detailed protocol

The sample labeling was performed using Quick-Amp Labeling Kit, One Color (Agilent Technologies, Part Number: 5190-0442); 500 ng of each sample were denatured along with WT primer with a T7 polymerase promoter. The cDNA master mix was added to the denatured RNA sample and incubated at 40°C for 2 hr for double-stranded cDNA synthesis. Synthesized double-stranded cDNA was used as a template for cRNA generation. cRNA was generated by in vitro transcription, and the cyanine-3-CTP (Cy3-CTP) dye incorporated during this step and incubated at 40°C for 2.30 hr. The Cy3-CTP labeled cRNA sample was purified using the Qiagen RNeasy column (Qiagen, Cat # 74106). The concentration of cRNA and dye incorporation was determined using Nanodrop-1000.

Microarray hybridization and scanning

Request a detailed protocol

About 4 µg of labeled Cy-3-CTP cRNA was fragmented at 60°C for 30 min, and the reaction was stopped by adding 2× GE HI-RPM hybridization buffer (Agilent Technologies, In situ Hybridization kit, Part Number: 5190-0404). The hybridization was carried out in Agilent’s Surehyb Chambers at 65°C for 16 hr. The hybridized slides were washed using Gene Expression Wash Buffer 1 (Agilent Technologies, Part Number: 5188-5325) and Gene Expression Wash Buffer 2 (Agilent Technologies, Part Number: 5188-5326) and were scanned using Agilent Scanner (Agilent Technologies, Part Number: G2600D). Data extraction from the images was done using Feature Extraction Software Version 11.5.1.1 of Agilent.

Microarray data analysis

Request a detailed protocol

Microarray data analysis was undertaken by in-house coded R Script (https://cran.r-project.org/). Processing of raw data into expression profiles was achieved by utilizing the packages limma and affy. Probe intensities were converted into expression measures by standard procedures. Briefly, the design-sets depicting the ‘control/test’ arrays were carefully generated by reading the raw data from MA image files. Background correction was done by the method ‘normexp’. This data was quantile normalized (between arrays depending on the design set), and within-array replicates were averaged. Processed data were categorized into major functional categories and tabulated. The detailed microarray array discussed in this manuscript have been deposited in GEO with accession number GSE154618.

Data availability

Microarray data is available in the GEO server. GEO accession Number GSE154618 has been provided in the material and method section. Source files for the following Figures were provided as a zip folder: Figure 1A, 1B, 1C, 1F Figure 2 Figure 3A, 3B, 3C, 3D, 3E, 3F, 3G Figure 4B (ii), 4C, 4D, 4E Figure 5A, 5B, 5D Figure 6D, 6E, 6G Figure 1-figure supplement 1C.

The following data sets were generated
    1. Dutta D
    2. Kumar V
    3. Mishra R
    4. Arora A
    (2021) NCBI Gene Expression Omnibus
    ID GSE154618. The global transcriptomic profile in the spermidine-stressed E. coli.
    https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE154618

References

    1. Gardner PR
    2. Fridovich I
    (1992)
    Inactivation-reactivation of aconitase in Escherichia coli: A sensitive measure of superoxide radical
    The Journal of Biological Chemistry 267:8757–8763.

Decision letter

  1. Joseph T Wade
    Reviewing Editor; New York State Department of Health, United States
  2. Gisela Storz
    Senior Editor; National Institute of Child Health and Human Development, United States
  3. Joseph T Wade
    Reviewer; New York State Department of Health, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Free spermidine evokes superoxide radicals that manifest toxicity" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Joseph T Wade as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Senior Editor.

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided to reject the paper.

The reviewers raised four major concerns. First, the probes used to determine the type of ROS stress are not specific enough to confidently draw the conclusion that spermidine causes O2- production, as opposed to another ROS. Hence, we would want to see a more direct test of the specific ROS produced. Electron paramagnetic resonance would be a suitable approach for this experiment. Second, several of the experiments use deletion mutant strains but lack a control where the deletion has been complemented, together with a corresponding empty-vector control. Third, the reviewers felt that the S. aureus experiments are peripheral to the main theme of the paper, and should be removed. Fourth, the reviewers were concerned that intracellular spermidine levels for different strains/growth conditions were inferred but not directly measured, particularly for the experiments in Figure 1 with speE and speG mutants and with spermidine supplementation. The use of LB as a growth medium is a concern in this regard, since LB contains spermidine. We would want to see either measurement of intracellular spermidine levels for wild-type and mutant strains grown under the conditions used in the corresponding experiments, or references to prior work where spermidine levels have been measured under the same growth conditions for the same strains. We believe these concerns are addressable, but would require a considerable amount of work, which would take more than two months. Hence, we are rejecting the paper, but we would consider a revised version if you are able to address the four major concerns.

Reviewer #1:

The work presented in this study provides important insight into the role of spermidine in bacterial cells. The data can be supported by the addition of a few simple controls, and a more detailed explanation of some of the methods, but generally speaking I am convinced by the conclusion that spermidine causes production of ROS by oxidizing Fe2+ in the cell. The protective role of other biomolecules such as RNA is more speculative, but nonetheless intriguing, and consistent with the genetic data. I feel that the S. aureus data add little to the paper. A big problem with the S. aureus experiments is that they compare two genetically distinct strains rather than comparing wild-type and mutant derivatives of otherwise genetically identical strains. Nonetheless, the S. aureus data are not needed, since the E. coli work stands on its own.

1. The conclusion from the data in Figure 1 seems to be that both high and low spermidine levels, relative to wild-type cells grown without exogenous spermidine, increase ROS levels. This is consistent with the overall conclusion that there is a "sweet spot" of spermidine levels needed to minimize ROS. This conclusion should be clearly stated in the Results to give context, especially since some of the data appear at first glance to be contradictory.

2. The data in Figure 2 involve comparisons of two strains of S. aureus, one of which has speG. The data are interpreted in a way that attributes all the phenotypic differences between the strains to the presence/absence of speG, but presumably there are many other genetic differences between these strains. The wording in this section of the paper should be softened to reflect the possibility that there could be other reasons for the phenotypic differences. Alternatively, the authors should make mutant strains that introduce or delete speG from the two strains, and test phenotypes in those genetic backgrounds.

3. Perhaps I missed it, but I couldn't see how the authors determined ("estimated", in their words) levels of NADP or glutathione, values for which are reported in Figure 3. The authors do mention a kit-based assay for measuring NAD levels, but there is insufficient description of this.

4. Line 195. There are lots of genes regulated by Fis or IHF. Is there a statistically significant enrichment of these genes among those differentially expressed, based on the microarray experiment?

5. Figure 4C. The authors gloss over the observation that SodA levels are similar in wt, untreated cells and spermidine-treated speG mutant cells. This is an unexpected result that warrants some discussion.

6. Figure 4 and associated text. This feels like a list of genes that went up or down, with little explanation of the expected result based on the prior observations in the paper. Hence, the significance of these observations is lost on the reader.

Figure 1C and 1E. The graphs would be easier to view if the lines were color-coded, especially for Figure 1E.

Line 143: "Therefore, spermidine stress would likely evoke O2- radicals in the S. aureus RN4220 but not in USA300 strain". This statement is too strong, since it is based on data from E. coli.

Line 163. "Confirms" is too strong a word to use here without measuring intracellular levels of the relevant molecules with/without each of the treatments. I recommend softening the wording to "is consistent with".

Figure 3 and associated description in the Results. The authors should make it clear at the start of this section and in the figure title that these experiments were done with E. coli.

Figure 4A. The authors should add a description of each of the numbered gene categories.

Figure 4C. The quantification graph is too small to easily interpret.

Figure 4G. The authors should plot data for Mn on a separate y-axis scale.

Figure 4G. This panel includes data for E. coli and data for S. aureus, which is very confusing. The authors should use a separate panel for data from each species.

Line 335. Should read "spermidine is highly toxic to cells".

Reviewer #2:

The authors have investigated why the polyamine spermidine is toxic to the bacterium Escherichia coli when intracellular spermidine is present in excess. To obtain cells where spermidine might be present in excess, they sought to create a gene deletion strain for the speG gene that encodes spermidine N-acetyltransferase, an enzyme that N-acetylates spermidine when spermidine levels are in excess, thereby neutralizing whatever function of spermidine is deleterious to cell growth. In principle, a speG gene deletion strain of E. coli is likely to have higher intracellular levels of spermidine. The authors also grew the speG strain in media contain high levels of added spermidine, which would be expected to further increase intracellular spermidine levels. A large number of physiological experiments were performed by the authors to demonstrate whether excess spermidine affects oxidative stress, and they concluded that it caused the generation of superoxide. Other analyses were performed to assess the affect of spermidine on iron oxidation. The global impact of spermidine toxicity was assessed by performing a transcriptional microarray experiment where the speG gene deletion strain was grown with and without 3.8 mM added spermidine in the growth medium, and the authors concluded that spermidine affects iron-sulfur cluster biogenesis.

Unfortunately (and inexplicably), the authors at no point measured spermidine or N-acetylspermidine levels in the cells that they were working with. The gene deletion regions for spermidine N-acetyltransferase (speG) and for spermidine synthase (speE) were transferred from KEIO collection strains into the authors' model E. coli wildtype strain by phage P1 transduction but were not confirmed by PCR after transduction, or by measuring spermidine or N-acetylspermidine levels. Thus, there is no proof that spermidine levels are altered in the speG or spermidine synthase (speE) gene deletion strains. The gene deletion strains were not complemented by the wildtype genes to show that any growth effects were not due to off-target damage cause by the P1 transduction. Furthermore, the supposed gene deletion strains were grown in LB medium, which contains spermidine, instead of in chemically defined M9 medium that does not contain spermidine. Because the spermidine content of the gene deletion strains is unknown, one can have very little confidence in any result that is supposedly based on whether cells contain excess spermidine or not. Although the microarray data looked at the effect of {plus minus} 3.8 mM spermidine on gene expression in the speG gene deletion strain, without knowledge of the intracellular spermidine content, any transcriptional effect could be due to indirect effects of spermidine on the cell wall, the outer membrane, on pH, and osmotic effects. In conclusion, the authors have performed a large number experiments where any conclusions on outcomes are in doubt due to the complete lack of knowledge of the content of intracellular spermidine, the metabolite that is supposedly responsible for the observed effects. Since, presumably, the authors do not have the data for intracellular spermidine content in their experiments, it is difficult to see how the data obtained can be used to make physiological conclusions, other than relating the observed effects to the presence or absence of added spermidine in the growth medium.

It is difficult to see how the data from this study can be salvaged. The observed effects cannot be attributed to excess spermidine or the complete absence of spermidine, since the level of spermidine in cells was never measured. Furthermore, the purported spermidine synthase (speE) deletion strain was grown in LB medium, which is a rich medium containing spermidine. Growth of the supposedly spermidine deficient speE deletion strain in LB medium will result in spermidine uptake from the LB medium, such that the supposedly depleted spermidine will be replaced by spermidine from the LB medium. The authors should have used chemically defined M9 medium for any spermidine-related experiment. Not only should spermidine levels and N-acetylspermidine levels have been measured in the speG deletion strain, but the levels of these metabolites should have been measured over the growth curve because N-acetylspermidine is accumulated more in stationary phase.

It may be possible to express the physiological changes observed in this study to the presence or absence of added spermidine to the growth medium, but the changes observed cannot be attributed to the levels of intracellular spermidine, since they were never measured.

Reviewer #3:

In the manuscript by Kumar et al., the authors proposed that sperimidine (SPD) triggers the production of superoxide radicals and claimed that this production is the unique molecular mechanism of toxicity in bacteria (line 81). These conclusions are based on the used of ROS probes. The results are interesting, provocative and conceptually new. However, clarifications, controls and new experiences are needed to support their conclusions.

My main concern is that the authors consider the ROS probes as specific for each compound (HO{degree sign}, O2{degree sign}-). The interconnection between ROS in the cells and the absence of great specificity of this type of tools should not make it possible to affirm the dependence of one type of ROS. Other approach would be necessary. Detecting very transient ROS, such as superoxide radicals, is challenging in the study of oxidative stress in vivo and an electron paramagnetic resonance (EPR) approach is considered ideal since it is unique for detecting free radicals.

The anaerobic experiments which support the superoxide conclusion need to be mitigated as polyamine uptake has been reported to be PMF dependant (Kashiwagi et al. J. Bact 1986). Therefore, a possible other explanation will be that in anaerobic condition PMF is slow down, polyamine uptake decrease and bacteria are more resistant. This point needs to be integrating in the text.

How the ROS probes response to SPD in anaerobic condition, what happens in the absence of oxygen if nitrate is present ?

The absence of SoxR activation in presence of spermidine is difficult to understand. The authors could used the Imlay study (Gu et al. 2011 , DOI: 10.1111/j.1365-2958.2010.07520.x) in which it is show that SoxR is directly activated by redox cycling drugs rather than by superoxide.

Figure 1 A and B : add control (wt+SPD ; wt+SPD +/- tiron).

Figure 1C : speE mutant is expected to have less SPD and therefore more H2O2, how the authors explain the opposite result?

Figure 1E : add control (WT+pSodA ; speG+pSodA) and all the strains with empty vector

Sup1 : C : does the sod(A or B) simple mutant are more sensitive to SPD at higher concentration than 4.5mM?

Figure 3 B and C : add control (empty vector).

Figure 4A : microarray result, the authors could add and comment the data on the suf operon (Fe/S cluster biosynthesis during stress condition, ROS and iron limitation).

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Free spermidine evokes superoxide radicals that manifest toxicity" for further consideration by eLife. Your revised article has been evaluated by Gisela Storz (Senior Editor) and a Reviewing Editor.

The manuscript has been improved, and the reviewers are satisfied that you have addressed the major concerns from the previous round of review. Nevertheless, there are a few small issues that still need to be addressed, as detailed in the two reviews below:

Reviewer #1:

The authors have responded positively to the reviewers' comments. I have only a few comments:

1. Figure 1A. MFI levels for H2DCFDA are reduced when spermidine is added to wild-type cells, but the intracellular concentration of spermidine does not detectably increase in these cells (Figure 1 - Figure Supplement 1C). Thus, the data don't support the conclusion that ROS detected by H2DCFDA is reduced as a result of increased intracellular spermidine levels.

2. Line 124-5, "...although spermidine accumulation in the ΔspeG strain reduces overall ROS levels and oxidative stress...". This statement is misleading. Intracellular levels of spermidine are higher in the ΔspeG strain when spermidine is added exogenously, and these cells do exhibit reduced H2DCFDA fluorescence, but so do wild-type cells, where the intracellular spermidine levels do not increase when spermidine is added exogenously.

3. The authors should show their HPLC traces used for quantifying intracellular spermidine levels. These data should include the control data using purified spermidine, and should indicate the quantified area of the HPLC trace.

4. Lines 140-1, "The double and triple mutants containing empty vector exhibited higher growth defects than ΔspeG strain on LB-agar plate supplemented with spermidine (Figure 1E)." I am not convinced by this conclusion; the sodA deletion has a very small effect when combined with the speG deletion, and the sodB deletion has no effect. However, what is clear is that overexpression of sodA rescues the growth defect of a speG mutant grown with exogenous spermidine. The authors don't mention this important result.

5. Figure 4 - Figure Supplement 1. The authors should repeat this restreak so that individual colonies for each strain can be more easily compared.

6. It seems likely that Fis-regulated and IHF-regulated genes are enriched in the set of genes identified as being differentially expressed in the microarray analysis. However, the authors have not done a statistical test to address this. A Fisher's exact test would suffice for this.

Reviewer #2:

I'm very happy to see that some of my suggestions from the first review were implemented by the authors.

Attached are a few more comments.

1) Claiming that the production of superoxyde is the unique molecular mechanism of toxicity is always too strong.

lines 35-36: "Therefore, we propose that the spermidine-induced superoxide radicals cause spermidine toxicity in E. coli"

line 84: "we decipher a unique molecular mechanism of spermidine toxicity in bacteria"

whereas in the rebuttal letter, the authors are more cautious, which is more appropriate: "superoxide generation is one of the major causes of spermidine toxicity" (just before reviewer#2 recommendations for the authors).

2) The comments on S. aureus in the introduction (l.73-77) are not any more necessary

3) Fig.1E: all the strains carrying plasmids (pDAK1 or pSodA), the plasmids are missing for ∆speG ∆sodB and ∆speG ∆sodA ∆sodB: is it a mistake? I hope so because otherwise, it lacks this control. Moreover, it is not indicated the presence of antibiotic to maintain plasmids in the agar plate +/- SPD (methods line 448 and legend).

4) lines 140, 146 and 297: the use of single, double or triple mutants is not clear, please rephrase

5) lines 168 to 176: errors on the panels of EPR experiments in the text:

(2B and 2C) instead of (2B and 2E)

(2D and 2E) instead of (2C and 2F)

(2F and 2G) instead of (2D and 2G)

6) Line 362 : "This is why spermidine is a double-edged sword where in excess, it provokes O2- anion production, and in scarcity, it leads to general ROS production."

Lower SPD level does not lead to ROS production but to a higher ROS level, please rephrase.

https://doi.org/10.7554/eLife.77704.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

The reviewers raised four major concerns. First, the probes used to determine the type of ROS stress are not specific enough to confidently draw the conclusion that spermidine causes O2- production, as opposed to another ROS. Hence, we would want to see a more direct test of the specific ROS produced. Electron paramagnetic resonance would be a suitable approach for this experiment. Second, several of the experiments use deletion mutant strains but lack a control where the deletion has been complemented, together with a corresponding empty-vector control. Third, the reviewers felt that the S. aureus experiments are peripheral to the main theme of the paper, and should be removed. Fourth, the reviewers were concerned that intracellular spermidine levels for different strains/growth conditions were inferred but not directly measured, particularly for the experiments in Figure 1 with speE and speG mutants and with spermidine supplementation. The use of LB as a growth medium is a concern in this regard, since LB contains spermidine. We would want to see either measurement of intracellular spermidine levels for wild-type and mutant strains grown under the conditions used in the corresponding experiments, or references to prior work where spermidine levels have been measured under the same growth conditions for the same strains. We believe these concerns are addressable, but would require a considerable amount of work, which would take more than two months. Hence, we are rejecting the paper, but we would consider a revised version if you are able to address the four major concerns.

We addressed all four issues, listed above, before resubmitting the manuscript:

1. EPR experiments were performed to confirm spermidine-mediated O2-production in ∆speG cells.

2. Now we have shown phenotypes with deletion and complementation along with proper empty vector controls.

3. S. aureus data has been removed.

4. HPLC-coupled MS analyses was done to show the levels of spermidine in different E. coli cells.

Lastly, the editor was rightly pointed out that the addressing the above concerns would take more than two months. Due to COVID situation and other inconveniences, we had to take four months to address the points carefully.

The response to the reviewer’s specific comments are as follows:

Reviewer #1:

The work presented in this study provides important insight into the role of spermidine in bacterial cells. The data can be supported by the addition of a few simple controls, and a more detailed explanation of some of the methods, but generally speaking I am convinced by the conclusion that spermidine causes production of ROS by oxidizing Fe2+ in the cell. The protective role of other biomolecules such as RNA is more speculative, but nonetheless intriguing, and consistent with the genetic data. I feel that the S. aureus data add little to the paper. A big problem with the S. aureus experiments is that they compare two genetically distinct strains rather than comparing wild-type and mutant derivatives of otherwise genetically identical strains. Nonetheless, the S. aureus data are not needed, since the E. coli work stands on its own.

S. aureus data is removed, as suggested.

1. The conclusion from the data in Figure 1 seems to be that both high and low spermidine levels, relative to wild-type cells grown without exogenous spermidine, increase ROS levels. This is consistent with the overall conclusion that there is a "sweet spot" of spermidine levels needed to minimize ROS. This conclusion should be clearly stated in the Results to give context, especially since some of the data appear at first glance to be contradictory.

Now we have rewritten our results to mention the point clearly. See line numbers 102-104, 116-119, 122-127 in the main text.

2. The data in Figure 2 involve comparisons of two strains of S. aureus, one of which has speG. The data are interpreted in a way that attributes all the phenotypic differences between the strains to the presence/absence of speG, but presumably there are many other genetic differences between these strains. The wording in this section of the paper should be softened to reflect the possibility that there could be other reasons for the phenotypic differences. Alternatively, the authors should make mutant strains that introduce or delete speG from the two strains, and test phenotypes in those genetic backgrounds.

S. aureus data is removed, as suggested.

3. Perhaps I missed it, but I couldn't see how the authors determined ("estimated", in their words) levels of NADP or glutathione, values for which are reported in Figure 3. The authors do mention a kit-based assay for measuring NAD levels, but there is insufficient description of this.

The measurements are not absolute levels but a relative analysis. We removed the word estimated “and coined “compared the levels” in the text (Line number 193). We elaborated it in the methodology section (Lines 577-607)

4. Line 195. There are lots of genes regulated by Fis or IHF. Is there a statistically significant enrichment of these genes among those differentially expressed, based on the microarray experiment?

Thanks to the reviewer for this comment. As per the current EcoCyc entry, there are 137 and 106 transcriptional units in E. coli that are regulated by Fis and IHF, respectively. In our microarray, we find 45 Fis-regulated genes, 13 IHF regulated genes, and 4 Fis+IHF regulated genes have altered expression. We think that these numbers are significant especially the numbers of the Fis-regulated genes. Accordingly, ∆speG∆fis mutant, but not the ∆speG∆ihfA mutant, grows slowly than the ∆speG mutant. We just mention this number of genes in the text (Line 219).

5. Figure 4C. The authors gloss over the observation that SodA levels are similar in wt, untreated cells and spermidine-treated speG mutant cells. This is an unexpected result that warrants some discussion.

I think reviewer missed the texts where we expressed our concerns for this unexpected result saying spermidine may directly or indirectly inhibit superoxide-mediated induction of SoxR in the old text. However, further we modified our text in the light of two alternative explanations: Late Prof. Fridovich and many initial work had opined that superoxide can directly activate SoxR. On the other hand, Prof. Imlay group has shown that SoxR is efficiently activated by redox cycling drugs, but not by superoxide. We observed superoxide generation and no activation of SoxR which fits well with the second school of thought. However, we have also used a redox cycling drug, menadione, which activated SoxR greatly, and spermidine deactivates this menadione-induced SoxR activation. This data is pointing towards a bigger and generalized concept that spermidine may keep SoxR in apo-SoxR form by interfering its iron-sulfur cluster biogenesis. As a result, be it superoxide or redox-cycling drug, they cannot activate SoxR! We mentioned these new points in different part of the revised manuscript (Lines: 235-241, 431-438).

6. Figure 4 and associated text. This feels like a list of genes that went up or down, with little explanation of the expected result based on the prior observations in the paper. Hence, the significance of these observations is lost on the reader.

Added lines 215-218 in the revised main text.

Figure 1C and 1E. The graphs would be easier to view if the lines were color-coded, especially for Figure 1E.

Now revised panels were color-coded in Figure 1.

Line 143: "Therefore, spermidine stress would likely evoke O2- radicals in the S. aureus RN4220 but not in USA300 strain". This statement is too strong, since it is based on data from E. coli.

S. aureus data has been removed as recommended by Editor and reviewer I.

Line 163. "Confirms" is too strong a word to use here without measuring intracellular levels of the relevant molecules with/without each of the treatments. I recommend softening the wording to "is consistent with".

Replaced accordingly.

Figure 3 and associated description in the Results. The authors should make it clear at the start of this section and in the figure title that these experiments were done with E. coli.

Since S. aureus data has been removed, this will not be a problem now.

Figure 4A. The authors should add a description of each of the numbered gene categories.

The numbered gene categories are listed in the supplementary information section. Now, we mentioned the same thing in the Figure 4A legend.

Figure 4C. The quantification graph is too small to easily interpret.

We redistributed the panels of Figure 4 into two different figures (Figure 4 and a new Figure 5) to show the quantification graph prominently.

Figure 4G. The authors should plot data for Mn on a separate y-axis scale.

Done. Instead of putting a separate Y axis, two different plots were made; one for Fe (Figure 4B) and another for Mn (Figure 4D).

Figure 4G. This panel includes data for E. coli and data for S. aureus, which is very confusing. The authors should use a separate panel for data from each species.

S. aureus data is removed as recommended by the reviewer and Editor.

Line 335. Should read "spermidine is highly toxic to cells".

Removed the article “The” from the sentence.

Reviewer #2:

The authors have investigated why the polyamine spermidine is toxic to the bacterium Escherichia coli when intracellular spermidine is present in excess. To obtain cells where spermidine might be present in excess, they sought to create a gene deletion strain for the speG gene that encodes spermidine N-acetyltransferase, an enzyme that N-acetylates spermidine when spermidine levels are in excess, thereby neutralizing whatever function of spermidine is deleterious to cell growth. In principle, a speG gene deletion strain of E. coli is likely to have higher intracellular levels of spermidine. The authors also grew the speG strain in media contain high levels of added spermidine, which would be expected to further increase intracellular spermidine levels. A large number of physiological experiments were performed by the authors to demonstrate whether excess spermidine affects oxidative stress, and they concluded that it caused the generation of superoxide. Other analyses were performed to assess the affect of spermidine on iron oxidation. The global impact of spermidine toxicity was assessed by performing a transcriptional microarray experiment where the speG gene deletion strain was grown with and without 3.8 mM added spermidine in the growth medium, and the authors concluded that spermidine affects iron-sulfur cluster biogenesis.

Unfortunately (and inexplicably), the authors at no point measured spermidine or N-acetylspermidine levels in the cells that they were working with. The gene deletion regions for spermidine N-acetyltransferase (speG) and for spermidine synthase (speE) were transferred from KEIO collection strains into the authors' model E. coli wildtype strain by phage P1 transduction but were not confirmed by PCR after transduction, or by measuring spermidine or N-acetylspermidine levels. Thus, there is no proof that spermidine levels are altered in the speG or spermidine synthase (speE) gene deletion strains. The gene deletion strains were not complemented by the wildtype genes to show that any growth effects were not due to off-target damage cause by the P1 transduction. Furthermore, the supposed gene deletion strains were grown in LB medium, which contains spermidine, instead of in chemically defined M9 medium that does not contain spermidine. Because the spermidine content of the gene deletion strains is unknown, one can have very little confidence in any result that is supposedly based on whether cells contain excess spermidine or not. Although the microarray data looked at the effect of {plus minus} 3.8 mM spermidine on gene expression in the speG gene deletion strain, without knowledge of the intracellular spermidine content, any transcriptional effect could be due to indirect effects of spermidine on the cell wall, the outer membrane, on pH, and osmotic effects. In conclusion, the authors have performed a large number experiments where any conclusions on outcomes are in doubt due to the complete lack of knowledge of the content of intracellular spermidine, the metabolite that is supposedly responsible for the observed effects. Since, presumably, the authors do not have the data for intracellular spermidine content in their experiments, it is difficult to see how the data obtained can be used to make physiological conclusions, other than relating the observed effects to the presence or absence of added spermidine in the growth medium.

I appreciate the concern of the Reviewer. In fact, many years ago, we have noticed that some knockouts in Keio collection (e.g ∆hns) are not perfect. Therefore, our default lab practice is to make fresh strains by P1 transduction in an isogenic WT E. coli background (Either MG1655 or BW25113) followed by confirming the allelic replacement by PCR using flanking primer-pairs and Sanger sequencing. Although we had already mentioned the point of PCR verification in the Methods section, we forgot to mention the Sanger sequencing part. Now we clearly mentioned the sanger sequencing part (Line number: 416-420). in Methods section.

Now we have performed HPLC coupled MS analyses of the tri-dansylated-spermidine to determine its intracellular level to support our observations (in collaboration with my colleague Dr. Vinod D. Chaudhari; See the new author list for the detail). Di-dansylated-hexane diamine was used as an internal standard. Polyamines were extracted from about 50mg wet cell pellets adding internal standard and then fully dansylated. The tri-dansylated spermidine peak areas were first normalized with the area of the internal standards, and then with the wet mass of the cell pellets. A standard curve was also generated using different amounts of tri-dansylated spermidine using the HPLC. The cellular levels of spermidine were represented as µmol/100mg of wet cell pellets (Figure 1—figure supplement 1C and Source data Figure 1—figure supplement 1C).

We could not resolve N-acetyl spermidine peak using the protocol used by Dr. Chaudhari lab. We see an old literature where authors have shown two different, N1 and N8-acetyl spermidine peaks. In our case these two acetylated peaks but were merged with many other components, as detected in MS data.

We agree with the reviewers that the spermidine can exerts effect on cell wall, outer membrane, pH and osmotic balance but how it will cause different effects to WT and ∆speG cells when present in extracellular environment is not conceivable. They can cause differential effect on those components only when accumulate intracellularly which was possible in ∆speG cells, but not in a wild type background. For example, spermidine also has an effect on translation. However, we have shown here that the superoxide generation is one of the major cause of spermidine toxicity. Thus, effects of spermidine in two isogenic strains (WT and DspeG) are different.

Our data now shows the intracellular level of spermidines in the different strains (Figure 1—figure supplement 1C and Source data Figure 1—figure supplement 1C).

It is difficult to see how the data from this study can be salvaged. The observed effects cannot be attributed to excess spermidine or the complete absence of spermidine, since the level of spermidine in cells was never measured. Furthermore, the purported spermidine synthase (speE) deletion strain was grown in LB medium, which is a rich medium containing spermidine. Growth of the supposedly spermidine deficient speE deletion strain in LB medium will result in spermidine uptake from the LB medium, such that the supposedly depleted spermidine will be replaced by spermidine from the LB medium. The authors should have used chemically defined M9 medium for any spermidine-related experiment. Not only should spermidine levels and N-acetylspermidine levels have been measured in the speG deletion strain, but the levels of these metabolites should have been measured over the growth curve because N-acetylspermidine is accumulated more in stationary phase.

It may be possible to express the physiological changes observed in this study to the presence or absence of added spermidine to the growth medium, but the changes observed cannot be attributed to the levels of intracellular spermidine, since they were never measured.

Now we have performed MS-coupled HPLC analyses of the spermidine peaks to determine its intracellular level to support our observations.

To get an idea, we compared the relative spermidine peak area in the LB medium to WT cell. This suggests that dry LB medium contains approx. 25 times lesser spermidine concentration than equal wet mass of WT cell pellets. If we consider the bio-number, The 100 mg wet cell pellet will correspond to about 20 mg dry mass. Therefore, comparing dry masses, LB powder will have 125 time lesser concentration of spermidine than in E. coli Dry cell pellets. Thus, this spermidine contamination in LB medium transported uphill to raise the polyamine level in the ∆speE strain to some extent, but it remains well below WT levels.

We could not resolve N-acetyl spermidine peaks but could detect their presence or absence in the MS data.

Reviewer #3:

In the manuscript by Kumar et al., the authors proposed that sperimidine (SPD) triggers the production of superoxide radicals and claimed that this production is the unique molecular mechanism of toxicity in bacteria (line 81). These conclusions are based on the use of ROS probes. The results are interesting, provocative and conceptually new. However, clarifications, controls and new experiences are needed to support their conclusions.

Now we have come up with new experiments and clarifications as stated point-by point below.

My main concern is that the authors consider the ROS probes as specific for each compound (HO{degree sign}, O2{degree sign}-). The interconnection between ROS in the cells and the absence of great specificity of this type of tools should not make it possible to affirm the dependence of one type of ROS. Other approach would be necessary. Detecting very transient ROS, such as superoxide radicals, is challenging in the study of oxidative stress in vivo and an electron paramagnetic resonance (EPR) approach is considered ideal since it is unique for detecting free radicals.

Thanks to the reviewer for educating us with his helpful comments. We are convinced by the reviewer that the ROS probes, specifically H2DCFDA or DHE are not very specific for OH or superoxide radicals! Although DHE would be somewhat specific for superoxide in our assay condition. However, we cannot take any chance to misinterpret the observations. Therefore, as per the reviewer’s suggestion, we performed EPR spectroscopy experiments for superoxide detection.

We have struggled a lot to execute the suggested EPR experiments for superoxide species detection. This is because the EPR machines are available in a few distantly places in India and either heavily occupied by the physicists and chemists, or usage were restricted due to COVID situation. We just could just manage to get few slots in two different places, to perform the bare minimum experiments to standardize and get the results (with ∆speG strain plus minus spermidine) to prove our points.

We used CMH spin-probe for superoxide detection as it is cell permeable, required to add in low concentration, reacts better with superoxide and half life is greater than the spin traps and other spin probes. Quickly, just raw panels that are shown in Author response image 1, were used to make the new Figure 2 (replacing old figure 2 of S. aureus data) in the manuscript. Also, we added a section in the result and another in the Materials and methods (Line numbers 148-175; 473-485).

Author response image 1
Download asset Open asset

The anaerobic experiments which support the superoxide conclusion need to be mitigated as polyamine uptake has been reported to be PMF dependant (Kashiwagi et al. J. Bact 1986). Therefore, a possible other explanation will be that in anaerobic condition PMF is slow down, polyamine uptake decrease and bacteria are more resistant. This point needs to be integrating in the text.

Now, we have integrated this point in the text as recommended by the reviewer. Also added this reference:

Kashiwagi K, Kobayashi H, Igarashi K. Apparently unidirectional polyamine transport by proton motive force in polyamine-deficient Escherichia coli. J Bacteriol. 1986 Mar;165(3):972-7. doi: 10.1128/jb.165.3.972-977.1986. PMID: 3005244; PMCID: PMC214524.

How the ROS probes response to SPD in anaerobic condition, what happens in the absence of oxygen if nitrate is present?

They did not differ in signal between the conditions.

The absence of SoxR activation in presence of spermidine is difficult to understand. The authors could used the Imlay study (Gu et al. 2011 , DOI: 10.1111/j.1365-2958.2010.07520.x) in which it is show that SoxR is directly activated by redox cycling drugs rather than by superoxide.

Now, we have included this possibility in the results, and discussion that superoxide may not directly activate the SoxR (Lines: 235-241, 431-438).

Figure 1 A and B : add control (wt+SPD ; wt+SPD +/- tiron).

Controls added.

Figure 1C : speE mutant is expected to have less SPD and therefore more H2O2, how the authors explain the opposite result?

Thanks for pointing to this this crucial point.

We repeated the experiments and again find that the H2DCFDA and DHE signals were higher (see new Figure 1A and B), but H2O2 levels was again lower in ∆speE strain than WT cells under spermidine stress. We also checked KatG and AhpC levels. They do not change in ∆speE strain under spermidine stress! Therefore, indeed these paradoxical observations are difficult to explain. One possibility could be that in the absence of spermidine, the level of free irons are more (as we have shown spermidine interacts with free iron) that may participate in Fenton reaction consuming more H2O2 and liberating Hydroxyl and hydroperoxyl radicals (see the reactions in Author response image 2). However, this is just a hypothesis.

Author response image 2
Download asset Open asset

Figure 1E : add control (WT+pSodA ; speG+pSodA) and all the strains with empty vector.

Now we have added controls as suggested.

Sup1 : C : does the sod(A or B) simple mutant are more sensitive to SPD at higher concentration than 4.5mM?

We did not find any difference (see Author response image 3). This result could be due to non-induction of sodA/B, and/or declined level of manganese. Further, we do not know at this moment if functional gain of SpeG in the sod mutant strains, if any. The part of the Sup1C figure (SPD 4.5mM) has now been moved to main Figure 1. Rest of the Sup1C has been moved to Sup 4.

Author response image 3
Download asset Open asset

Figure 3 B and C : add control (empty vector).

Added and repeated the experiments

Figure 4A : microarray result, the authors could add and comment the data on the suf operon (Fe/S cluster biosynthesis during stress condition, ROS and iron limitation).

Thanks for this suggestion. We have seen suf operon was indifferent except sufA was about 2.5-fold downregulated. The sufA expression data is now added in heat-map (Figure 4a). We explained these results at related texts (270-277). Also, added the sufA expression in microarray heat-map panel (Figure 4A)

[Editors’ note: what follows is the authors’ response to the second round of review.]

Reviewer #1:

The authors have responded positively to the reviewers' comments. I have only a few comments:

1. Figure 1A. MFI levels for H2DCFDA are reduced when spermidine is added to wild-type cells, but the intracellular concentration of spermidine does not detectably increase in these cells (Figure 1 - Figure Supplement 1C). Thus, the data don't support the conclusion that ROS detected by H2DCFDA is reduced as a result of increased intracellular spermidine levels.

Indeed, in the light of the above data, it will be erroneous to say that increased spermidine level in WT cells by exogenous spermidine stress results in decreased H2DCFDA fluorescence. However, one cannot deny that exogenous spermidine supplementation caused the decreased H2DCFDA fluorescence. In fact, SpeG (spermidine N-acetyl transferase) enzyme can quickly acetylate either N1 and N8-amine groups of excess spermidines forming two different species of N-acetyl-spermidines in WT strain keeping almost constant level of spermidine. In eukaryotic system, there is known mechanisms of further degradation/recycling of N-acetyl-spermidines, but nothing is known about the fate of N-acetyl spermidines in prokaryotes. Furthermore, whether N-acetyl spermidines, the products of spermidine overaccumulation in WT cells, have also roles in decreasing ROS is not known so far.

On the other hand, ΔspeG strain is devoid of SpeG function, and therefore, there would be very little (if spontaneous acetylation exists), or no turnover of spermidine to N-acetyl-spermidines. Therefore, one can easily detect spermidine overaccumulation.

Therefore, we never claimed (in line number 123-125) that increased intracellular spermidine level in WT cells caused the decreased H2DCFDA fluorescence.

To further clarify the issue, we have added a few extra texts (Lines: 95-97; 112; 114-118) in the manuscript.

2. Line 124-5, "...although spermidine accumulation in the ΔspeG strain reduces overall ROS levels and oxidative stress...". This statement is misleading. Intracellular levels of spermidine are higher in the ΔspeG strain when spermidine is added exogenously, and these cells do exhibit reduced H2DCFDA fluorescence, but so do wild-type cells, where the intracellular spermidine levels do not increase when spermidine is added exogenously.

Already clarified this issue in the above-mentioned response against reviewer’s recommendation point no 1 (Lines: 95-97; 112; 114-118).

3. The authors should show their HPLC traces used for quantifying intracellular spermidine levels. These data should include the control data using purified spermidine, and should indicate the quantified area of the HPLC trace.

We have added this HPLC profile pictures in a separate source data (Figure 1 – figure supplement 1C - Source data 2)

4. Lines 140-1, "The double and triple mutants containing empty vector exhibited higher growth defects than ΔspeG strain on LB-agar plate supplemented with spermidine (Figure 1E)." I am not convinced by this conclusion; the sodA deletion has a very small effect when combined with the speG deletion, and the sodB deletion has no effect. However, what is clear is that overexpression of sodA rescues the growth defect of a speG mutant grown with exogenous spermidine. The authors don't mention this important result.

We have modified the text mentioning the strain names (Lines 142-144). Also, we have mentioned that the sodA overexpression rescued the growth defect of speG mutant (Lines 146-149)

5. Figure 4 - Figure Supplement 1. The authors should repeat this restreak so that individual colonies for each strain can be more easily compared.

Repeated and the old figure has been replaced.

6. It seems likely that Fis-regulated and IHF-regulated genes are enriched in the set of genes identified as being differentially expressed in the microarray analysis. However, the authors have not done a statistical test to address this. A Fisher's exact test would suffice for this.

First, we distributed 49 Fis-enriched genes in 32 transcriptional units/operons as per EcoCyc database. Similarly, 17 IHF-enriched genes were distributed in 13 transcriptional units/operons. We also checked the total number of transcriptional units/operons that are upregulated or downregulated by the above two transcription regulators. Then Fisher’s exact test for Fis-enriched and IHF-enriched transcriptional units were performed. The calculated P-values for the Fis-enriched and IHF-enriched transcriptional units were 0.0023 (i.e., significant) and 0.7714 (i.e., not significant), respectively. We have mentioned this Fisher’s exact test in the result section (Line number 224-225) to draw its significance.

Reviewer #2:

I'm very happy to see that some of my suggestions from the first review were implemented by the authors.

Attached are a few more comments.

1) Claiming that the production of superoxyde is the unique molecular mechanism of toxicity is always too strong.

lines 35-36: "Therefore, we propose that the spermidine-induced superoxide radicals cause spermidine toxicity in E. coli"

line 84: "we decipher a unique molecular mechanism of spermidine toxicity in bacteria"

whereas in the rebuttal letter, the authors are more cautious, which is more appropriate: "superoxide generation is one of the major causes of spermidine toxicity" (just before reviewer#2 recommendations for the authors).

We have made the necessary changes in those lines, as recommended.

2) The comments on S. aureus in the introduction (l.73-77) are not any more necessary

We have now deleted this text from the revised manuscript.

3) Fig.1E: all the strains carrying plasmids (pDAK1 or pSodA), the plasmids are missing for ∆speG ∆sodB and ∆speG ∆sodA ∆sodB: is it a mistake? I hope so because otherwise, it lacks this control. Moreover, it is not indicated the presence of antibiotic to maintain plasmids in the agar plate +/- SPD (methods line 448 and legend).

We are extremely sorry for this inadvertent mistake! Now we rectified it.

4) lines 140, 146 and 297: the use of single, double or triple mutants is not clear, please rephrase

We modified the lines by mentioning the exact names of those mutants (new Lines: 140-143; 298-300).

5) lines 168 to 176: errors on the panels of EPR experiments in the text:

(2B and 2C) instead of (2B and 2E)

(2D and 2E) instead of (2C and 2F)

(2F and 2G) instead of (2D and 2G)

Thank you for pointing out this error! We changed them accordingly (new lines 171-180).

6) Line 362 : "This is why spermidine is a double-edged sword where in excess, it provokes O2- anion production, and in scarcity, it leads to general ROS production."

Lower SPD level does not lead to ROS production but to a higher ROS level, please rephrase.

We rephrased the text accordingly (new Lines 361-362)

https://doi.org/10.7554/eLife.77704.sa2

Article and author information

Author details

  1. Vineet Kumar

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Present address
    Regional Centre for Biotechnology, Faridabad, India
    Contribution
    Investigation, Methodology, Validation, Writing – original draft
    Contributed equally with
    Rajesh Kumar Mishra
    Competing interests
    No competing interests declared

  2. Rajesh Kumar Mishra

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Formal analysis, Investigation, Methodology, Validation
    Contributed equally with
    Vineet Kumar
    Competing interests
    No competing interests declared

  3. Debarghya Ghose

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Investigation, Methodology, Validation
    Competing interests
    No competing interests declared

  4. Arunima Kalita

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Investigation, Methodology, Validation
    Competing interests
    No competing interests declared

  5. Pulkit Dhiman

    1. Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    2. Division of Medicinal Chemistry, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared

  6. Anand Prakash

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared

  7. Nirja Thakur

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared

  8. Gopa Mitra

    Clinical Proteomics Unit, Division of Molecular Medicine, St. John's Research Institute, St John's Medical College, Bangalore, India
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared

  9. Vinod D Chaudhari

    1. Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    2. Division of Medicinal Chemistry, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Investigation, Methodology, Supervision
    Competing interests
    No competing interests declared

  10. Amit Arora

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Present address
    Department of Medical Microbiology, Post Graduate Institute of Medical Education and Research, Chandigarh, India
    Contribution
    Formal analysis, Investigation, Methodology
    For correspondence
    aarora.pgi@gmail.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3503-4695

  11. Dipak Dutta

    Division of Molecular Biochemistry and Microbiology, CSIR Institute of Microbial Technology, Chandigarh, India
    Contribution
    Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Writing – review and editing
    For correspondence
    dutta@imtech.res.in
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0458-4109

Funding

Council of Scientific and Industrial Research (CSIR), India (MLP-042)

  • Dipak Dutta

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors are grateful to Dr Debashish Adhikari, Division of Chemical Sciences, IISER Mohali, for their critical inputs on the plausible binding mechanism of spermidine and iron; to Prof. Kaushik Ghosh and Dr JS Meena, IIT Roorkee, India, and Mr LM Jha, IISER Bhopal, India, for their sincere support in EPR analyses. The work has been funded by CSIR IMTECH, India to DD. VK was an ICMR fellow, RKM and PD is a UGC fellow, DG is a CSIR fellow, AK is a DST-Inspire fellow, and AP is a DBT fellow.

Ethics

Polyclonal antibodies were raised using NZW rabbits in an in-house animal facility. This animal handling was approved by the Institutional Animal Ethics Committee (IAEC) and performed according to the National regulatory guidelines issued by Committee for the Purpose of Supervision of Experiments on Animals (CPSEA), Govt. of India.

Senior Editor

  1. Gisela Storz, National Institute of Child Health and Human Development, United States

Reviewing Editor

  1. Joseph T Wade, New York State Department of Health, United States

Reviewer

  1. Joseph T Wade, New York State Department of Health, United States

Version history

  1. Preprint posted: September 5, 2021 (view preprint)
  2. Received: February 8, 2022
  3. Accepted: April 11, 2022
  4. Accepted Manuscript published: April 13, 2022 (version 1)
  5. Accepted Manuscript updated: April 14, 2022 (version 2)
  6. Version of Record published: April 25, 2022 (version 3)

Copyright

© 2022, Kumar et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,993
    Page views
  • 261
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Vineet Kumar
  2. Rajesh Kumar Mishra
  3. Debarghya Ghose
  4. Arunima Kalita
  5. Pulkit Dhiman
  6. Anand Prakash
  7. Nirja Thakur
  8. Gopa Mitra
  9. Vinod D Chaudhari
  10. Amit Arora
  11. Dipak Dutta
(2022)
Free spermidine evokes superoxide radicals that manifest toxicity
eLife 11:e77704.
https://doi.org/10.7554/eLife.77704

Categories and tags

Research organism

Further reading

    1. Biochemistry and Chemical Biology
    2. Cell Biology

    Mitochondrial temperature homeostasis resists external metabolic stresses

    Mügen Terzioglu, Kristo Veeroja ... Howard T Jacobs
    Research Article

    Based on studies with a fluorescent reporter dye, Mito Thermo Yellow (MTY), and the genetically encoded gTEMP ratiometric fluorescent temperature indicator targeted to mitochondria, the temperature of active mitochondria in four mammalian and one insect cell line was estimated to be up to 15°C above that of the external environment to which the cells were exposed. High mitochondrial temperature was maintained in the face of a variety of metabolic stresses, including substrate starvation or modification, decreased ATP demand due to inhibition of cytosolic protein synthesis, inhibition of the mitochondrial adenine nucleotide transporter and, if an auxiliary pathway for electron transfer was available via the alternative oxidase, even respiratory poisons acting downstream of oxidative phosphorylation (OXPHOS) complex I. We propose that the high temperature of active mitochondria is an inescapable consequence of the biochemistry of OXPHOS and is homeostatically maintained as a primary feature of mitochondrial metabolism.

    1. Biochemistry and Chemical Biology
    2. Evolutionary Biology

    Evolution towards simplicity in bacterial small heat shock protein system

    Piotr Karaś, Klaudia Kochanowicz ... Krzysztof Liberek
    Research Article

    Evolution can tinker with multi-protein machines and replace them with simpler single-protein systems performing equivalent functions in an equally efficient manner. It is unclear how, on a molecular level, such simplification can arise. With ancestral reconstruction and biochemical analysis, we have traced the evolution of bacterial small heat shock proteins (sHsp), which help to refold proteins from aggregates using either two proteins with different functions (IbpA and IbpB) or a secondarily single sHsp that performs both functions in an equally efficient way. Secondarily single sHsp evolved from IbpA, an ancestor specialized in strong substrate binding. Evolution of an intermolecular binding site drove the alteration of substrate binding properties, as well as the formation of higher-order oligomers. Upon two mutations in the α-crystallin domain, secondarily single sHsp interacts with aggregated substrates less tightly. Paradoxically, less efficient binding positively influences the ability of sHsp to stimulate substrate refolding, since the dissociation of sHps from aggregates is required to initiate Hsp70-Hsp100-dependent substrate refolding. After the loss of a partner, IbpA took over its role in facilitating the sHsp dissociation from an aggregate by weakening the interaction with the substrate, which became beneficial for the refolding process. We show that the same two amino acids introduced in modern-day systems define whether the IbpA acts as a single sHsp or obligatorily cooperates with an IbpB partner. Our discoveries illuminate how one sequence has evolved to encode functions previously performed by two distinct proteins.

    1. Biochemistry and Chemical Biology
    2. Computational and Systems Biology

    Death by a thousand cuts through kinase inhibitor combinations that maximize selectivity and enable rational multitargeting

    Ian R Outhwaite, Sukrit Singh ... Markus A Seeliger
    Research Article

    Kinase inhibitors are successful therapeutics in the treatment of cancers and autoimmune diseases and are useful tools in biomedical research. However, the high sequence and structural conservation of the catalytic kinase domain complicates the development of selective kinase inhibitors. Inhibition of off-target kinases makes it difficult to study the mechanism of inhibitors in biological systems. Current efforts focus on the development of inhibitors with improved selectivity. Here, we present an alternative solution to this problem by combining inhibitors with divergent off-target effects. We develop a multicompound-multitarget scoring (MMS) method that combines inhibitors to maximize target inhibition and to minimize off-target inhibition. Additionally, this framework enables optimization of inhibitor combinations for multiple on-targets. Using MMS with published kinase inhibitor datasets we determine potent inhibitor combinations for target kinases with better selectivity than the most selective single inhibitor and validate the predicted effect and selectivity of inhibitor combinations using in vitro and in cellulo techniques. MMS greatly enhances selectivity in rational multitargeting applications. The MMS framework is generalizable to other non-kinase biological targets where compound selectivity is a challenge and diverse compound libraries are available.