Correspondence: Manas K. Chattopadhyay, NIDDK, National Institutes of Health, 8 Center Drive, Bldg. 8, Room 223, Bethesda, MD 20892, USA. Tel.: 301 496 1058; fax: 301 402 0240; e-mail: email@example.com
Glutathionylspermidine synthetase/amidase (Gss) and the encoding gene (gss) have only been studied in Escherichia coli and several members of the Kinetoplastida phyla. In the present article, we have studied the phylogenetic distribution of Gss and have found that Gss sequences are largely limited to certain bacteria and Kinetoplastids and are absent in a variety of invertebrate and vertebrate species, Archea, plants, and some Eubacteria. It is striking that almost all of the 75 Enterobacteria species that have been sequenced contain sequences with very high degree of homology to the E. coli Gss protein. To find out the physiological significance of glutathionylspermidine in E. coli, we have performed global transcriptome analyses. The microarray studies comparing gss+ and Δgss strains of E. coli show that a large number of genes are either up-regulated (76 genes more than threefold) or down-regulated (35 genes more than threefold) by the loss of the gss gene. Most significant categories of up-regulated genes include sulfur utilization, glutamine and succinate metabolism, polyamine and arginine metabolism, and purine and pyrimidine metabolism.
Earlier work from this laboratory showed that 95% of the intracellular spermidine and a large fraction of the intracellular glutathione are converted to monoglutathionylspermidine in Escherichia coli at the end of logarithmic growth (Dubin, 1959; Tabor & Tabor, 1970). Bollinger et al. (1995) and Kwon et al. (1997) reported the purification of glutathionylspermidine synthetase/amidase of E. coli and showed that the bifunctional enzyme had a separate amidase and synthetase domains. Later, Pai et al. (2006), reported crystal structures of E. coli Gss in complex with substrate, product, and inhibitor.
In 1985, Fairlamb et al. (1985) reported that glutathionylspermidine and diglutathionylspermidine (trypanothione) are present in trypanosomes and that diglutathionylspermidine disulfide, rather than glutathione disulfide, is the substrate for a glutathionyl-like reductase in trypanosomes. These findings probably account for the therapeutic efficacy of difluoromethylornithine, an inhibitor of polyamine biosynthesis, in African trypanosomiasis (Fairlamb, 1988; Wyllie et al., 2009).
Trypanothione is not present in E. coli. In contrast to the large amount of glutathionylspermidine found in stationary and near-stationary E. coli cultures, the earlier studies indicated that logarithmically growing cultures of E. coli contain very little (Smith et al., 1995) or no detectable (Tabor & Tabor, 1976) glutathionylspermidine. As the formation of glutathionylspermidine affects the intracellular levels of both spermidine and glutathione, we felt that it is important to test whether the Gss is only present in certain bacteria and Kinetoplastids. Therefore, we have carried out blast searches of the NCBI databases and have found that the distribution of the Gss is indeed very limited. The small amount of glutathionylspermidine present in logarithmically growing cultures poses the question of whether glutathionylspermidine synthetase has any physiological function in logarithmically growing E. coli. Therefore, we have carried out microarray studies of E. coli, comparing a strain with a deletion in the gene coding for glutathionylspermidine synthetase (Δgss) with a gss+ strain and have found that a large number of genes are up-regulated or down-regulated in the Δgss strain compared to the gss+ strain.
Materials and methods
Strains and culture medium
Strains used in this study are listed in Table 1. Cultures were grown in M9 medium (Miller, 1992) containing 0.4% glucose; incubation was at 37 °C with shaking.
For a comparison of the different phyla, blast searches were carried out comparing the E. coli Gss amino acid sequences (accession number AAC76024.1) with the nonredundant protein databases of the National Center for Biotechnology Information (NCBI). The cutoff level for significant homology, as defined by Hall (Hall, 2011), is e < 10−3 and query coverage > 55%.
The cultures were incubated with shaking in air until the OD600 nm was 0.7–0.8 (log-phase culture) or 2.8–3.0 (stationary-phase culture). The cells were collected by centrifugation, extracted with perchloric acid, and 5 μL of the 10% perchloric acid extract, representing 1 mg of cells (wet weight), was then analyzed by ion exchange chromatography essentially as described earlier (Murakami et al., 1989; Chattopadhyay et al., 2009b) using a Shim-pack column (Shimadzu, ISC-05/S0504); the eluting buffer was 1.6 M NaCl, 0.2 M sodium citrate.
Radioactive glutathionylspermidine analysis
Four milliliter of the M9 medium was inoculated with a single colony of each strain and grown at 37 °C overnight. The overnight cultures were diluted into 50 mL of medium to an OD600 nm of 0.05 and grown for several hours to an OD600 nm of 0.2. To determine the relative amounts of glutathionylspermidine and of spermidine in each strain, cells were prelabeled with 1.25 μCi of [14C]-spermidine trihydrochloride (12.5 nmoles), and the incubation was continued for either 2 h (‘log-phase culture’ OD600 nm = 0.7) or 20 h (‘overgrown culture’). The cultures were rapidly centrifuged at room temperature. The pellets were washed twice with medium and re-suspended in 10% perchloric acid (1 : 5 wt/vol); the supernatants were subjected to HPLC chromatography on a Shim-pack cation exchange column with the elution system described in the previous section but with 1.0 M NaCl-0.2 M sodium citrate as the elution buffer. The elutes were collected at 2-min intervals (0.7 mL min−1), and a 100-μL aliquot from each fraction was counted in a Beckman scintillation counter (LS6500).
RNA isolation, microarray analysis
Three independent cultures (109–1010 cells) from the E. coli gss+ and Δgss cells (OD600 nm of 0.7–0.8) were harvested and re-suspended in Tris-EDTA buffer (100 mM Tris, 10 mM EDTA, pH 8.0) containing 2 mg mL−1 lysozyme (Sigma). The cell suspensions were incubated for 5 min at room temperature to digest the cell wall. Total RNA was isolated according to the protocol described in the RNeasy mini kit (Qiagen, Germantown, MD). The mRNAs were enriched from total RNA by removing the 16S and 23S ribosomal RNAs using the MICROBExpress method and kit (part no. AM1905; Ambion). The quantity and quality of RNA were evaluated by OD260 nm/OD280 nm assays and by RNA capillary electrophoresis (Agilent Biotechnologies). Enriched mRNAs were reverse-transcribed by Superscript II and random hexanucleotide primer (Invitrogen) and used for microarrays as described earlier (Chattopadhyay et al., 2009a) using Affymetrix (Santa Clara, CA) E. coli GeneChip arrays (Genome 2.0 array; n = 3 each for gss+ and Δgss). anova (analysis of variance) was performed, and P-values were calculated using Partek Pro-software (Partek, St. Louis, MO) and plotted in negative log scale on y-axis against the Affimetrix signal ratios for each probe set on x-axis. Up- and down-regulated genes were selected based on P-values of <0.05 and fold change > +2 or −2. The complete microarray data can be obtained from GEO (accession number GSE30679).
Glutathionylspermidine synthetase/amidase is only found in two distinct groups: Eubacteria and Kinetoplastida
Most striking is that sequences homologous to E. coli Gss are only found in Eubacteria and the very distantly related Kinetoplastids (plus two fungal species with relatively low homology; Table 2). No homologous sequences (as defined by the blast-p program) were found when the E. coli Gss sequence was compared with the human, rat, mouse, Arabidopsis, rice, worm, and Drosophila sequence databases (Table 2).
Table 2. Distribution of Gss in different taxonomic groupsa
The taxonomic classifications are based on the taxonomic groupings in NCBI database and blast-p using NCBI database. The distribution of sequences homologous to the Gss amino acid sequences is based on blastp analyses of the various microbial sequences in the NCBI database, omitting any redundant sequence for specific species.
Calculated for those species showing significant homology to the Escherichia coli Gss amino acid sequence.
No significant homology was found when the E. coli Gss amino acid sequence was used to blast search in different vertebrate (mammals, zebra fish, chicken, western clawed frog, etc.) and invertebrate (arthropode, nematode, echinoderm, chidaria, etc.) protein sequence databases.
Among the bacterial species, the homology was highest in the Proteobacteria and was particularly high in almost all of the 75 Enterobacteriales species that have been sequenced (27–100% homology in all of the Enterobacteria species and > 65% identity in 50% of the species). A comparison of different species (Table 2) shows that outside of the Proteobacteria, homologous sequences are only found in a few other bacterial species, and these have much less homology.
Glutathionylspermidine and spermidine levels in gss+ and Δgss strains
To measure the conversion of intracellular spermidine to glutathionylspermidine in stationary- vs. log-phase cultures of gss+ and Δgss strains of E. coli, [14C]-spermidine labeled cells were analyzed on a cation exchange column as described in 'Materials and methods'. As shown in Fig. 1a, confirming previous results from this and other laboratories, most of the spermidine in stationary cultures of a gss+ (wild type) strain was converted to glutathionylspermidine, and a much smaller amount of conversion was found in log-phase cultures (Dubin, 1959; Tabor & Tabor, 1970, 1971, 1976; Bollinger et al., 1995; Smith et al., 1995). No conversion of spermidine to glutathionylspermidine was found in cells containing a deletion in the gss gene (gss−; Fig. 1a).
As shown in Fig. 1b, there was a very large decrease (85–90%) in the spermidine content of gss+ cells observed in a stationary-phase culture (compared to a gss− control), but only a small decrease (10–15%) in the spermidine content of a log-phase culture.
Comparison of global gene expression in gss+ and Δgss Escherichia coli
We have applied microarray analysis to study the global gene expression profile of logarithmically growing E. coli cultures (OD600 nm of 0.7–0.8). We used logarithmically growing cultures because, as shown in Fig. 1, in stationary-phase wild-type E. coli converts most of the spermidine into glutathionylspermidine, and global gene expression might be affected by a decrease in the glutathione and spermidine levels; in contrast, only 10–15% of spermidine is converted to glutathionylspermidine in logarithmically growing cells.
The effects of the gss deletion on gene expression are shown in Fig. 2 and Tables 3, 4 and 5. There was no expression of the gss gene in Δgss cells, as compared to a high level of expression of gss in gss+ cells (Fig. 2, Table 4). It is evident from the volcanic graph (Fig. 2) that the gss deletion has a pronounced effect on gene expression. To facilitate data analysis, the genes were grouped into functional categories based on Ecogene database, Affymetrix gene ID, and gene ontology (GO). Top GO categories of up- and down-regulated genes are presented in Tables 3, 4 and 5 and Supporting Information, Fig. S1.
Table 3. Genes up-regulated (fourfold or more) in gss mutant (Δgss) over wild-type (gss+)
In these tables, data are the average of three independent experiments. ‘Annoted functions’ are based on Ecogene and Affymetrix databases.
Effect of gss gene deletion on up-regulation of gene expression
When compared with the levels in the gss+ cells, in the Δgss cells, transcripts of 183 genes were up-regulated more than twofold. A total of 76 genes were up-regulated greater than threefold, and 24 genes were up-regulated greater than fivefold. Most significant categories of up-regulated genes include sulfur utilization, glutamine and succinate metabolism, polyamine and arginine metabolism, and purine and pyrimidine metabolism. As shown in Tables 3 and 4 and Fig. S1a–d, many genes involved in polyamine and arginine metabolism, purine and pyrimidine biosynthesis, sulfate and succinate metabolism show 2- to 10-fold higher induction of gene expression in Δgss cells as compared to gss+ cells.
Effect of gss gene deletion on repression of gene expression
Deletion of gss gene resulted in down-regulation of 134 genes twofold as compared to wild-type cells. A total of 35 genes were down-regulated more than threefold, and 12 genes were down-regulated more than fourfold. Several genes related to molybdate transporters (Table 4, heat-map in Fig. S1e), nitrate transporters, copper transport/efflux (Table 4 and heat-map in Fig. S1f), and C4-dicarboxylate transporters were repressed in Δgss cells. Several oxidoreductases such as fumarate HCP oxidoreductase and glutaredoxin were also repressed.
Effect of deletion of gss gene on transcriptional regulators
Increased or decreased transcription of the large number of genes presented above may not be due to a direct effect of gss gene deletion; rather, expression of 12 transcriptional regulators are increased in Δgss cells, and four transcriptional activators are repressed in Δgss cells as compared to gss+ cells during log phase (Table 5).
It is striking that the Gss sequences have been conserved with a high degree of homology throughout the Enterobacteria (including E. coli, Salmonella enteric, and Klebsiella pneumoniae), where both the glutathionylspermidine synthetase and amidase domains have been conserved in most of the species. It seems possible that within the Enterobacteriales, Gss have extensive inheritance, and thus they, show more than 60% identity in many species. In addition, based on blast-p analysis among the closely related bacterial groups in the gamma-proteobacteria, Gss sequences are present in some species of the Pasteurellalel, Pseudomonadale, Vibrionale, and Xanthomonadale groups, but absent in others. Many other bacteria either do not have Gss homologs (Table 2) or only possess lower homology with the synthetase domain (i.e. the C-terminal part).
As opposed to these results in various bacterial species, there are no homologs in nearly all other organisms (including Saccharomyces cerevisiae, mammals, and plants) (Table 2). In contrast however, there is a high degree of homology between the E. coli Gss sequences and both the synthetase and amidase domains of both glutathionylspermidine synthetase (Gss) and trypanothione synthetase (Trs) of Kinetoplastids (Tetaud et al., 1998). The close relationship between Kinetoplastids and bacterial Gss sequences and the absence of such sequences in almost all other organisms suggest that either these organisms lost their respective ancestral sequences early in their lineage or Kinetoplastids have acquired the ability to synthesize both glutathionylspermidine from bacteria followed by gene duplication and modification to synthesize trypanothione. Large-scale phylogenetic analyses on genomic data have demonstrated that several distantly related microbial eukaryotes have acquired mostly metabolic genes from prokaryotic organisms (Opperdoes & Michels, 2007; Andersson, 2009). For instance, many protists encode genes not found universally among other eukaryotes, and these are patchily distributed (such as tyrosyl t-RNA, genes for de novo pyrimidine biosynthesis and others; for references, see Opperdoes & Michels, 2007; Andersson, 2009).
Most organisms contain high concentrations of at least one low-molecular weight thiol for maintenance of an intracellular-reducing environment, such as glutathione (most organisms including E. coli), homoglutathione (mung bean), glutathionylspermidine (E. coli, Crithidia fasciculata), trypanothione (trypanosomatids) and L-γ-glutamyl-cystine (halobacteria) (Fairlamb & Cerami, 1992). Two important functions of these thiols are well-documented-thiol modification of proteins and protection of DNA from ionizing radiation or oxidative damages. The most important function of these compounds is the modification of protein thiols either by the formation of mixed disulfides or by the formation of intramolecular disulfides. These post-translational modifications protect proteins from oxidative stress and can regulate their functions (Fairlamb & Cerami, 1992), at least in part due to presence of trypanothione (Krieger et al., 2000). Thus, when the genes for trypanothione synthetase and reductase from Trypanosoma cruzi were introduced into E. coli, the cells were protected from radiation-induced DNA damage (Fitzgerald et al., 2010).
Although the high homology for the Gss sequences in the Enterobacteria suggests an important physiologic function for glutathionylspermidine in these organisms, no specific function has been described for this system in bacteria. One possible function of the enzyme glutathionylspermidine synthetase in E. coli could be a regulation of metabolites (both spermidine and glutathione) because of the presence of bifunctional activity of the enzyme Gss. It is also clear from our studies and from others that glutathionylspermidine and glutathione are not essential, as mutants of GSH or spermidine grow normally on minimal medium during normal aerobic growth (Greenberg & Demple, 1986; Chattopadhyay et al., 2009b). However, both glutathione and polyamines are absolutely required for protection against oxidative stress (Chattopadhyay et al., 2003; Masip et al., 2006), and polyamines are involved in other cellular functions (such as swarming, (Kurihara et al., 2009). Thus, it could be possible that glutathionylspermidine is essential during environmental stresses. Despite these changes in gene expression, we have not found any difference in the two strains (gss+ vs. gss−) in their growth rate, their sensitivity to oxygen, the toxicity of copper sulfate or cadmium sulfate, or survival after long-time storage (data not shown). As one of the older speculations suggested a function in protecting DNA (Krieger et al., 2000; Fitzgerald et al., 2010), we also tested their sensitivity to UV radiation, but found no significant difference in either survival or development of fluorouracil-resistant mutations (data not shown). Although we could not find any difference in oxygen sensitivity, Chiang et al. (2010) have reported slight increase in the sensitivity of a combined grxΔ and gssΔ double mutant to hydrogen peroxide, but no difference between gss+ and gssΔcells. They have also reported that glutathionyspermidine could form mixed disulfides with proteins, but their results do not exclude the possibility that comparable binding occurs with intracellular glutathione. In our C14-spermidine incorporation assays, we found more than 98% of the counts are in the TCA supernatant, and only < 2% counts in TCA precipitate with the macromolecules. In this experiment, the gss+ cells showed twice more counts than gss− cells in the TCA precipitate (data not shown).
Although we have not been able to define a specific function for the gss gene, we feel that the microarray results clearly show that this gene has a considerable effect on the physiology and gene expression of the bacteria. Comparison of the gss+ and gss− strains in the microarray studies showed marked differences in the regulation of different mRNAs. These differences have been listed in Tables 3, 4 and 5. Some of the gene expression changes in gss+ vs. gss− cells are in the polyamine metabolisms and arginine metabolisms pathways, as expected. We felt that it was important to show that glutathionylspermidine is not just an inactive end-product, but is metabolically active. Our isotope exchange experiments show that glutathionylspermidine is metabolically active in both logarithmically growing (data not shown) and stationary cultures (Tabor & Tabor, 1975). Thus, it could be possible that even in the log-phase cells, where glutathionylspermidine content is < 10%, there is always some change in spermidine and glutathione pools due to activities of both the synthetase and amidase domains of gss+ as compared to gss− cells. For further understanding of regulatory pathways involved in the gene expression pattern of up- or down-regulated genes in gss+ vs. gss− cells, we performed bioinformatics analyses. The microarray results show an up-regulation of succinate metabolism (sdhD, sdhC, sdhA), which increases fumarate synthesis in the cells and on the other hand down-regulation of fumarate metabolism (frdC, frdD, and frdB), which could increase fumarate level in the gss− cells. The transcription of sdhCABD is enhanced during a switch from aerobic to anarerobic growth by ArcA transcriptional regulators (Maklashina et al., 1998). The carAB regulon is regulated by arginine, pyrimidine, and purine levels (Devroede et al., 2004). The genes for purine metabolisms (e.g. purM, purD, and purH) are regulated by PurRP (Meng et al., 1990). The precise mechanism of how these genes are regulated by gss gene deletion is not known. However, as shown in Table 5, fourteen transcriptional regulators are either up- or down-regulated in gss− culture. These transcriptional regulators may be directly or indirectly affected by the presence or absence of glutathionylspermidine in these two strains. Some of these transcriptional factors are related to growth in low oxygen or low pH. For example, pdhR, a repressor involved in respiratory control of pyruvate dehydrogenase complex (Ogasawara et al., 2007), is induced in the gss− cells; ttdR, a transcriptional activator required for tartrate utilization (Kim et al., 2009), and, cadC, a transcriptional activator for cadBA induced during low oxygen and low pH (Haneburger et al., 2011) are repressed in gss− cells. Apart from these, the puuR transcriptional repressor of the putrescine utilization pathway (Kurihara et al., 2008) is also induced in gss cells.
The phylogenetic data showing that the full Gss sequences are mainly found in two phyla, Enterobacteria and Kinetoplastida, and not in most other species, indicate that glutathionylspermidine and diglutathionylspermidine are not necessary for most species, but have specialized functions in Enterobacteria and Kinetoplastids. We do not know the function of glutathionylspermidine in Enterobacteria, but it seems possible that it is important for survival of these organisms (such as E. coli) in the crowded, anaerobic environment in the intestinal lumen.
This research was supported by the Intramural Research Program of the National Institutes of Health (National Institute of Diabetes, Digestive and Kidney Diseases).
Conflict of interest
The authors declare no conflict of interest in this study.