Structure, function and regulation of the DNA-binding protein Dps and its role in acid and oxidative stress resistance in Escherichia coli: a review


Young Min Kwon, 1260 W. Maple Ave, O-219, Department of Poultry Science, University of Arkansas, Fayetteville, AR 72701, USA. E-mail:


Dps, the DNA-binding protein from starved cells, is capable of providing protection to cells during exposure to severe environmental assaults; including oxidative stress and nutritional deprivation. The structure and function of Dps have been the subject of numerous studies and have been examined in several bacteria that possess Dps or a structural/functional homologue of the protein. Additionally, the involvement of Dps in stress resistance has been researched extensively as well. The ability of Dps to provide multifaceted protection is based on three intrinsic properties of the protein: DNA binding, iron sequestration, and its ferroxidase activity. These properties also make Dps extremely important in iron and hydrogen peroxide detoxification and acid resistance as well. Regulation of Dps expression in E. coli is complex and partially dependent on the physiological state of the cell. Furthermore, it is proposed that Dps itself plays a role in gene regulation during starvation, ultimately making the cell more resistant to cytotoxic assaults by controlling the expression of genes necessary for (or deleterious to) stress resistance. The current review focuses on the aforementioned properties of Dps in E. coli, its prototypic organism. The consequences of elucidating the protective mechanisms of this protein are far-reaching, as Dps homologues have been identified in over 1000 distantly related bacteria and Archaea. Moreover, the prevalence of Dps and Dps-like proteins in bacteria suggests that protection involving DNA and iron sequestration is crucial and widespread in prokaryotes.


The DNA-binding protein from starved cells (Dps) was previously described by Almiron et al. (1992) as a novel DNA-binding protein with regulatory and protective roles in starved E. coli. There is a wealth of knowledge that has been obtained about Dps since its discovery, particularly pertaining to the protein’s structure and its interaction with DNA. A great deal of attention has also been devoted to the role of Dps in oxidative stress resistance, as Dps was first described as a protein vital for stationary phase-induced hydrogen peroxide resistance (Almiron et al. 1992). For Dps proteins capable of binding DNA, as all Dps proteins are not (reviewed by Haikarainen and Papageorgiou 2010), DNA shielding is a major mode of protection employed by the protein and ameliorates chemical or reactive damage that can occur upon exposure to oxidative stress. Recent studies have painted a clearer picture of how Dps-mediated oxidative stress resistance is accomplished beyond physical shielding and established the ability of Dps to provide bimodal protection during times of stress (Zhao et al. 2002). However, the biochemical and molecular mechanisms (beyond shielding) employed by Dps to provide resistance against other stresses, such as acid resistance, continue to be somewhat enigmatic and are currently under investigation. Regulation of Dps has also been intensely scrutinized, and as a result, we have gained insight into how the protein is regulated during various biological situations. Upon discovery of Dps, Almiron et al. (1992) reported evidence for a possible regulatory role for the protein itself. Yet the loss or overexpression of Dps has not been directly associated with the downregulation or upregulation of any specific gene target to date. Thus, the ability of Dps to perform in a regulatory capacity remains, to some extent, speculative. The current review discusses the structure, functions, and properties of Dps as elucidated in the model organism E. coli that enable it to provide protection during certain stressful conditions. Also, the complex regulation of Dps expression and its possible role in gene regulation in starved E. coli are discussed as well.

Dps structure and its capacity as a DNA-binding protein

The structural information for several Dps proteins (isolated from different organisms) has been elucidated and deposited into the Protein Data Bank ( Additionally, intensive structural research has revealed a highly conserved structural fold of proteins within the Dps protein family (Zanotti et al. 2002; Ren et al. 2003; Kauko et al. 2004; Zeth et al. 2004; Franceschini et al. 2006; Gauss et al. 2006; Romao et al. 2006; Thumiger et al. 2006). Escherichia coli Dps shares many structural features common to other ferritins characterized in bacteria, including the paradigmatic ferritin (FtnA) and the heme-containing bacterioferritin (Bfr) (Grant et al. 1998). Dps, FtnA and Bfr are all composed of at least 12 subunits folded into four-helix bundles and assembled into spherical protein shells (reviewed by Andrews et al. 2003). Such similarity in the structure among these ferritins is suggestive of common ancestry (Peña and Bullerjahn 1995). Escherichia coli Dps in particular has a shell-like structure (approx. 80–90 Å in diameter) assembled from 12 identical subunits whose assembly creates a spherical hollow cavity (approx. 40–50 Å in diameter) that serves as an iron storage compartment (Zhao et al. 2002; Haikarainen and Papageorgiou 2010). Once assembled, E. coli Dps is very compact and stable with the highly flexible and lysine-rich N-terminus of each of its monomers protruding out from the dodecamer (Haikarainen and Papageorgiou 2010). The DNA-binding ability of Dps was initially discovered when purified Dps was added separately to supercoiled plasmid DNA and linear DNA (Almiron et al. 1992). From this simple yet elegant binding assay, several critical DNA-binding properties of Dps were revealed; one of the most important properties being the intense stability of the DNA–Dps complex. Dps bound to DNA prior to heating is able to withstand intense acute heat shock in excess of 100°C and continues to be a highly stable complex, even after prolonged heating at 65°C. An additional critical binding property of Dps is that Dps-bound DNA reveals no clear footprint after digestion with DNAse I (Almiron et al.1992); a binding characteristic very similar to that observed with other histone-like proteins (Schmid 1990). This property of E. coli Dps demonstrates its ability to bind DNA without any apparent sequence specificity.

Prolonged incubation of purified E. coli Dps molecules results in self-aggregation in solution culminating in the formation of two-dimensional crystals with hexagonal packing and spacing of 78 ± 1 Å (Wolf et al. 1999). Self-aggregation of Dps is the driving force in the formation of large DNA–Dps complexes (Ceci et al. 2004). Immediately upon the addition of DNA, Dps dodecamers undergo extensive aggregation and quickly form multilayered plate-like crystals thereafter (Wolf et al. 1999). This highly organized co-crystallization process is produced with closed supercoiled plasmids, linear double-stranded DNA, as well as single-stranded RNA molecules with little to no difference in the resultant crystalline structure (Wolf et al. 1999). While the crystallization of vital macromolecules is normally considered to be incompatible with prokaryotic life, the co-crystallization of DNA and Dps is representative of a generic defence strategy whereby the sequestration of DNA provides a highly efficient means of protection against an array of environmental assaults. Experimental evidence suggests that DNA and Dps form stacked alternating layers within which DNA is effectively and rapidly sequestered (Wolf et al. 1999). However, DNA is not thought to bind directly to the surface of the protein; as the surface of Dps dodecamers does not display DNA-binding motifs and is dominated by negative charges that would likely repel negatively charged DNA molecules (Grant et al. 1998). Frenkiel-Krispin et al. (2001) proposed that Dps is unable to directly bind DNA and that DNA-Dps complex formation relies on ion bridges formed by Mg2+. In actuality, it is the protruding lysine-containing N-terminus of Dps dodecamers (specifically, the last three N-terminus lysine residues) that is crucial to Dps–DNA co-crystallization and Dps self-aggregation in solution (Ceci et al. 2004). N-terminal deletion mutant studies revealed the importance of N-terminus lysine residue multiplicity and physiological pH in the formation of large Dps–DNA complexes that measure upwards of hundred of nanometres (as indicated by atomic force microscopy imaging). At physiological pH, wild-type Dps carries three positively charged lysine residues at its N-terminus and forms large Dps–DNA complexes in vitro. As pH increases beyond the physiological range, these lysine residues become deprotonated, and the formation of large DNA-Dps complexes is no longer possible. However, simple DNA binding (DNA void of Dps self-aggregation or co-crystallization) is still able to take place. Dps mutants possessing only one of the three terminal lysine residues are also able to bind DNA in a simplified manner, but are unable to sequester and condense it. Mutants lacking all three terminal lysine residues are unable to bind or condense DNA whatsoever. Similar packing of Dps dodecamers in protein crystals and within Dps–DNA complexes led to the belief that the two phenomena (Dps self-aggregation and DNA condensation) are coupled events and are promoted by the same type of protein–protein interactions between individual Dps monomers (Ceci et al. 2004). Furthermore, the common structural basis of Dps–DNA complexes and Dps protein crystals generates a similar dependence upon salts or small charged molecules that are capable of binding the macromolecular structures involved in complex formation (Ceci et al. 2004), a discovery based upon prior research by Frenkiel-Krispin et al. (2001).

Dps self-aggregation results in a crystal lattice whereby three adjacent dodecamers create holes lined by lysine-rich N-termini (Grant et al. 1998). DNA is assumed to be threaded through these holes and to interact with lysine-rich N-termini upon doing so. This hypothesis is supported by the results of differential interference contrast and fluorescent microscopy studies performed by Wolf et al. (1999) that suggest DNA is threaded through the Dps crystalline structure and by the fact that members of the Dps family that do not possess positive N-termini (i.e., Listeria innocua ferritin, Bacillus anthracis Dlp-1 and Dlp-2, and Helicobacter pylori HP-NAP) do not appear to bind DNA (Bozzi et al., 1997; Papinutto et al. 2002; Zanotti et al. 2002). The result of these elegant molecular interactions is a remarkably efficient Dps and DNA co-crystallization process in which DNA is efficiently protected against deleterious environmental assaults. Furthermore, the protein–protein interactions that occur in formation of Dps-DNA complexes are additional proof that protein–protein interactions play a major role in nucleoid structuring (Ceci et al. 2004).

Dps as a ferritin

Iron is an essential element for life; playing an important role in processes such as respiration, oxygen transport and DNA biosynthesis. However, the availability of free iron is extremely limited. This vital element can become particularly toxic when present in an excess amount because of its ability to drive the production of highly deleterious hydroxyl radicals via the Fenton reaction (Fe2+ + H2O2  Fe3+ + OH + OH). Structural conservation of Dps family members is again exemplified within the core structure of the protein, as several core amino acids located near its ferroxidase centre (H51, H63, D78, E82 and W52 by E. coli numbering) are highly conserved (Chiancone and Ceci 2010). Iron is tightly bound within this ferrihydrite core until mobilization becomes necessary for relevant biological processes. Hydrophilic and hydrophobic channels connect the protein core to the external medium and form at the junction between the threefold symmetry of related subunits. Ferrous ions sequestered into the core of the protein enter via hydrophilic channels and utilize two histidine residues and one glutamic acid as iron ligands for entry.

Dps has been identified as a ferritin-like protein partly because of its ferroxidase activity, or more specifically, its ability to oxidize bound ferrous ions to the ferric state (Nair and Finkel 2004). The twelve putative catalytic ferroxidase sites of Dps are each located at each of the six interfaces of adjacent subunits of the protein (two identical ferroxidase centres at each twofold subunit (Ilari et al. 2002; Zhao et al. 2002). Escherichia coli Dps ferroxidase sites are believed to be bimetallic, with a high-affinity iron-binding site A and a low-affinity site B (Ilari et al. 2000; Nordlund and Eklund 2005). Site A binds one iron ion coordinated by His, Asp and Glu residues, while site B (also capable of binding iron) often contains one water molecule. The low affinity for iron observed in E. coli Dps site B is attributed to the presence of a lysine residue which forms a salt bridge with an iron coordinating Asp residue (Ilari et al. 2002); however, full metal saturation of the both sites is inducible via anaerobic titration with Fe2+ (Zhao et al. 2002). Ferroxidase sites catalyse the initial reaction in the pathway that converts soluble ferrous ions to a Fe3+-saturated ferrihydrite mineral core on the inner surface of the protein (Velayudhan et al. 2007). Within each ferroxidase site, two ferrous ions are oxidized for every one molecule of H2O2 reduced, thereby avoiding hydroxyl radical formation via Fenton chemistry (Fig. 1) (Zhao et al. 2002). Ferrous ions bound at ferroxidase sites are also capable of being oxidized by O2; however, this process occurs at a much slower rate and is quite inefficient when compared to oxidation utilizing H2O2 (Zhao et al. 2002). H2O2-mediated oxidation is contradictory to that of other ferritins, where the oxidation of ferrous ions is most effectively accomplished by O2. Perhaps it is for this reason that Dps is often referred to as a ‘ferritin-like’ protein and rarely as a true ferritin. Spin trapping experiments have demonstrated that Dps greatly attenuates hydroxyl radical production by oxidizing ferrous ions with H2O2 at a stoichiometric ratio of one H2O2 per two Fe2+ ions oxidized (Zhao et al. 2002, 2006). Consequently, this oxidation method utilized by Dps protects DNA because it greatly ameliorates the potentially lethal combination of Fe2+ and H2O2. The ferritin-like properties of Dps also allow it to play a major role in hydrogen peroxide detoxification. In fact, the primary role of Dps may not be iron storage, but to protect macromolecules from the combined lethality of ferrous ions and H2O2. This notion is supported by the additional presence the iron storage proteins Ferritin A and bacterioferritin in E. coli that account for the bulk of iron storage in the bacterium [up to 50% of cellular iron during sufficient growth is stored by Ferritin A alone (Abdul-Tehrani et al. 1999)] and by the fact that Dps strongly prefers H2O2 over O2 as an oxidant when preparing iron for storage. With iron storage most likely being a secondary role of Dps in E. coli, it is unlikely that deletion of dps in E. coli will have a substantial effect on the intracellular level of iron. Some fluctuation in iron levels following deletion of dps is almost a certainty; yet, there is no evidence within the current literature to support this proposition. If deletion of dps does indeed effect the intracellular concentration of iron in E. coli, this is certain to have a correlated effect on the expression of genes dependent upon the cellular concentration of iron; i.e., those genes whose regulation is mediated by the ferric uptake regulator protein that controls the iron-dependent regulation of over 90 genes in E. coli (Hantke and Braun 2000; Hantke 2001). In order to determine any effects that a dps mutation may have on the iron storage capacity of E. coli, whole cell iron contents should be carefully monitored over varying growth conditions.

Figure 1.

 Iron oxidation in E. coli. (a) Oxidation of ferrous ions in E. coli can occur via the Fenton Reaction. In this instance, the stoichiometeric ratio of iron to hydrogen peroxide is 1 : 1 and thus results in the production of hydroxyl radicals that can lead to damage of vital macromolecules and cell death. (b) Oxidation of ferrous ions occurs in the ferroxidase centre of Dps and proceeds via a reaction whereby the stoichiometric ratio of iron to hydrogen peroxide is 2 : 1. This reaction does not produce hydroxyl radicals; thus, the potential damage to cellular components is alleviated.

Reduction in the mineralized core of Dps consequentially permits H2O2 consumption, thereby affording the protein low-level catalase activity as well (Zhao et al. 2002). The catalase-like activity of Dps (that is actually stronger than that observed in human H-chain ferritin) is demonstrated by the addition of H2O2 to a solution of Dps-containing 24 Fe3+/Dps molecule. In this instance, hydrogen peroxide is utilized as an oxidizing agent, and O2 is generated by its decomposition according to the following equation: H2O2 → H2O + ½O2 (Zhao et al. 2002). The catalase activity of Dps seems to be significant in Dps-mediated protection against hydrogen peroxide stress, as Zhao et al. (2002) has identified its ability to decompose hydrogen peroxide as a significant method of detoxification in E. coli.

Regulation of Dps expression

Regulation of dps expression in E. coli is complex and dynamic (Fig. 2). Although little research has been conducted on the expression of Dps in response to fluctuations in cellular iron levels, it has been shown that Dps expression is highly dependent upon the growth phase in which it is expressed (Table 1). Expression is induced by OxyR in exponentially growing cells and σs and the histone-like integration host factor (IHF) during stationary phase (Altuvia et al. 1994). Although OxyR and σS both induce dps expression, their regulatory actions are independent of each other during these distinct growth phases. The redox-sensitive OxyR regulator is a member of the LysR family of transcriptional activators and regulates genes involved in hydrogen peroxide stress, including catalase HPI and alkyl hyperoxide reductase (Tartaglia et al. 1989; Storz et al. 1990). Following exposure to hydrogen peroxide during exponential growth, dps transcription is induced by OxyR, which activates σ70-RNA polymerase. dps mRNA levels during stationary phase are controlled by the rpoS-encoded σS, the stationary phase-specific sigma factor (Altuvia et al. 1994). Additionally, consensus sequences for IHF are found within the dps promoter, and additional examination of the heterodimeric IHF protein revealed that it is required for σS-mediated induction of dps in starved cells (Freundlich et al. 1992; Altuvia et al. 1994).

Figure 2.

 Dps regulation in E. coli. (a) (i) Following exposure to hydrogen peroxide during exponential growth OxyR is activated, which induces expression of the sigma 70 RNA polymerase and ultimately expression of dps. (ii) Dps is expressed upon entry into stationary phase following expression of sigma S. Heterodimeric integration host factor protein consensus sequences located within the dps promoter are required for sigma S-mediated induction of dps in stationary phase cells. (b) Downregulation of dps expression occurs when hydrogen peroxide levels are low during exponential phase and involves the nucleoid-associated proteins Fis and H-NS. (i) H-NS binds to the promoter region of Dps and prevents binding by sigma 70. Transcription of dps is thereby blocked, and Dps protein is not produced. (ii) Fis blocks transcription of dps by binding to sigma 70 at the dps promoter. An inoperable complex is formed between sigma 70 and Fis that is unable to initiate transcription of dps. (c) Post-translational regulation of Dps during exponential growth involves the ClpAP and ClpXP proteases and occurs in the absence of oxidation stress. Both the ClpAP and ClpXP proteases degrade Dps during exponential growth, thereby drastically reducing the level of Dps within the cell. It should be noted that the degradation of Dps by ClpAP depicted here plays a minor role in comparison with degradation of Dps by ClpXP in the overall proteolytic regulation of Dps.

Table 1.   Regulatory elements controlling Dps expression in Ecoli
Dps regulatorRegulatory roleLevel of regulationInducing conditionReference
OxyRUpregulationTranscriptionalOxidative stress in exponentially growing cellsAltuvia et al. (1994)
σsUpregulationTranscriptionalStarvationAltuvia et al. (1994)
Integration host factorUpregulationTranscriptionalStarvationAltuvia et al. (1994)
FisDownregulationTranscriptionalExponential phaseGrainger et al. (2008)
H-NSDownregulationTranscriptionalExponential phaseGrainger et al. (2008)
ClpAP ProteaseDownregulationTranslationalExponential phaseStephani et al. (2003)
ClpXP proteaseDownregulationPost-translationalExponential phaseStephani et al. (2003)

Downregulation at the dps promoter occurs during exponential phase when cells are not actively being exposed to oxidative stress and is accomplished by the nucleoid-associated proteins Fis and H-NS (Ali Azam et al. 1999; Grainger et al. 2008). The level of Dps present in cells during log phase growth is known to be minute (about 6000 Dps molecules per cell), when Fis and H-NS levels are at their highest. Conversely, Fis levels are low in stationary phase cells when Dps levels are known to soar (approx. 180 000 Dps molecules per cell) (Ali Azam et al. 1999; Grainger et al. 2008). At this point, Dps becomes the most abundant nucleoid-associated protein in the cell. H-NS levels are most abundant in exponentially growing cells, but levels remain significant during stationary phase as well. Fis and H-NS bind at adjacent sites within the core dps promoter, and each is able to repress dps expression by preventing transcription initiation by σ70 RNA polymerase. Although Fis and H-NS are both responsible for dps repression during log phase, they utilize two distinct mechanisms. Binding of H-NS to the promoter region of dps prevents binding by σ70 (Grainger et al. 2008). However, this repression can be overcome by binding of σS, a stationary phase RNA polymerase that also recognizes the dps promoter. Fis represses dps transcription by binding to its promoter in the spacer region between its extended -10 region (Jeong et al. 2006) and the -35 hexamer, ultimately trapping σ70 at the promoter. Downregulation of dps occurs as Fis forms an inoperable closed complex with σ70, thereby blocking transcription by σS. In this instance, σ70 actually functions as a co-repressor to block transcription of dps by operating in conjunction with Fis to block transcription initiation by σS (Grainger et al. 2008).

Dps levels are not only regulated at the transcriptional level, but are also controlled post-translationally via proteolysis (Stephani et al. 2003). Growth phase-dependent proteolysis of Dps is dependent on the physiological state of the cell and involves the ClpXP and ClpAP proteases. ClpP14 is a 14 subunit serine peptidase (Maurizi et al. 1990) that separately binds the ATPase orthologs ClpX and ClpA to form the ClpXP and ClpAP proteases, respectively (Katayama et al. 1988; Grimaud et al. 1998). Dps is directly degraded by ClpXP and ClpAP during exponential growth (Schmidt et al. 2009). Furthermore, ClpXP-mediated degradation of Dps is also shown to occur upon re-entry into the logarithmic phase and also involves direct degradation of σs (Flynn et al. 2003). Dps is also strongly accumulated during stationary phase with the help of ClpAP, which, in this instance, controls Dps regulation post-transcriptionally. The ClpAP protease facilitates the accumulation of Dps by indirectly maintaining the ongoing translation of dps mRNA in long-term stationary phase cells, yet, never directly degrades the Dps protein itself during starvation. Although Stephani et al. (2003) has shown that the ClpAP protease is important for stationary phase stability of Dps because of its ability to facilitate translation of the protein, it has also been suggested that a translational repressor of Dps and/or an RNase targeting dps mRNA for degradation may also be targets of ClpAP proteolysis as well during starvation.

In E. coli, the RssB protein acts as a recognition factor for regulated degradation of RpoS by ClpXP during exponential growth (Muffler et al. 1996; Becker et al. 1999; Hengge 2008). Yet Dps still remains highly unstable in an rssB deletion mutant during exponential phase and continues to be rapidly degraded in cells possessing this genetic background (Stephani et al. 2003). Studies involving RssB and its role in the ClpXP proteolysis of Dps revealed that RssB does not play a role in proteolysis of Dps and is not involved in its turnover whatsoever (Stüdemann et al. 2003). And so, with regard to Dps proteolysis during exponential growth, at least one crucial question remains unanswered: is there a specific recognition factor for Dps that functions similarly as RssB and responds to growth phase-specific signals to control ClpXP-mediated proteolysis of Dps? To date, this crucial question remains unanswered. More knowledge has been accrued about ClpAP proteolysis of Dps during exponential phase. Ninnis et al. (2009) and Schmidt et al. (2009) recently identified Dps as a target of N-rule degradation in E. coli and demonstrated that ClpS acts as an adapter protein for ClpAP-mediated degradation of Dps during exponential growth. The N-rule degradation pathway, one of the most profound processes identified for the degradation of proteins in E. coli, is based upon the primary structure of targeted proteins and states that the half-life of a protein is determined by the nature of its N-terminal residue (reviewed by Varshavsky 1996). With regard to removal of Dps via the N-rule pathway in E. coli, ClpS targets a truncated version of Dps (Dps6-167) and interacts directly with the N-terminal residue, Leu6, to target the protein for ClpAP degradation (Schmidt et al. 2009). Removal of Dps N-terminal residues 1–5 occurs via an undiscovered mechanism, and the purpose(s) for truncating the protein during growth are unknown as well. It suggested that removal of residues 1–5 is the result of an endoproteolytic event resulting in a Dps truncation variant that possesses highly unstable residues at the N-terminus of the protein (Ninnis et al. 2009; Schmidt et al. 2009). It should also be mentioned that ClpXP utilizes the N-terminal tail of Dps as a recognition signal for degradation; however, this recognition motif differs from the one recognized in ClpS-mediated degradation in that N-terminal residues 1–5 of Dps are required for ClpXP proteolysis (Flynn et al. 2003; Schmidt et al. 2009). Essentially, N-terminal cleavage of Dps prevents ClpXP-mediated turnover, which may indicate that the removal of Dps residues 1–5 is carried out partially for the purpose of ClpAP-mediated degradation of Dps during growth.

Dps as a regulatory protein

Although it has been suggested, little experimental evidence has been offered which demonstrates the regulatory role of Dps in E. coli during nutritional deprivation or oxidative stress. This lack of evidence has hindered the elucidation of the regulatory abilities of Dps. However, the notion that Dps may play a role in the regulation of gene and/or protein expression during times of stress is not completely unwarranted. In the initial work describing Dps (Almiron et al. 1992), the patterns of proteins synthesized after 3 days of starvation in both wild-type E. coli and a dps::kan null mutant were analysed using two-dimensional gel electrophoresis. Scrutiny of protein expression patterns revealed a dramatic difference in the proteomes of the parental and mutant strains, as well as the pleiotropic phenotype of mutants lacking a functional dps gene. Interestingly, Dps shares several properties of histone-like proteins, such as the heterodimeric protein HU associated with the E. coli nucleoid and the heat-stable nucleoid-structuring protein H-NS (Drlica and Rouviere-Yaniv 1987; Schmid 1990). In addition to binding DNA in a highly stable sequence independent manner, histone-like proteins can also act both as positive and as negative effectors in different systems (Berger 1999). The concentration of the histone-like protein H-NS has been shown to increase slightly during stationary phase – as does that of Dps – and a highly similar pleiotropic phenotype (with regard to gene expression) is observed in cells lacking H-NS (Higgins et al. 1988; Grainger et al. 2008).

Although results of the aforementioned studies have lead to the inference of a regulatory role for Dps in the global regulation of protein expression following prolonged starvation in E. coli, experimental evidence of the direct or indirect role of Dps in gene expression during stationary phase has yet to materialize. It may be more fruitful to examine the regulatory role of Dps via gene expression analysis and/or in vitro DNA-binding assays utilizing specific promoters as binding targets. For instance, DNA microarray analysis, often utilized to measure the global transcriptional response in an organism following a defined treatment, may be employed to monitor changes at the mRNA level that occur in a dps deletion mutant during stationary phase. If performed, microarray would most likely be utilized as a screening tool and could very likely reveal possible regulatory target(s) of Dps. Genes with a reduced level of transcript in the dps mutant when compared to the wild type may be positively regulated by Dps, while those with an increased level of mRNA when compared to the parental strain may be repressed by Dps. Once possible targets of regulation have been identified, further studies would need to be performed to determine the level of regulation, i.e. if the regulation is at the transcriptional, post-transcriptional, translational, or post-translational level, and if the regulation is direct or indirect. Because of the highly pleiotropic phenotype of the dps mutant during starvation, it can be presumed that this is an instance when Dps is most active in its regulatory capacity (Altuvia et al., 1994). Although the proteomic profile of the dps mutant has not been examined during oxidative stress, this may also be a time when Dps functions in a regulatory capacity as the presence of Dps seems to be crucial for survival during oxidative stress in Ecoli (Martinez and Kolter 1997; Choi et al. 2000; Zhao et al. 2002; Nair and Finkel 2004). Also, there is prior evidence of ferritins similar to Dps functioning as genetic activators in cell culture under oxidative stress (Lee et al. 2009).

If the regulatory role of Dps were to be scrutinized utilizing the current knowledge, a persuasive argument could be made that Dps has no true regulatory role, and the observed pattern of protein expression occurs because Dps-bound DNA is simply inaccessible to transcription factors (that positively and negatively regulate gene expression) and/or transcriptional machinery. Even in this instance, a regulatory-like role for Dps cannot be completely ruled out. As previously mentioned, Dps is considered a histone-like protein for its many similarities with the eukaryotic histone. It is therefore possible that Dps again functions in the same manner as the eukaryotic histone which undergoes reversible covalent modifications to make DNA accessible or inaccessible to transcriptional machinery, thereby regulating gene expression (Berger 1999; Nair and Finkel 2004). Dps may also regulate gene expression by interacting directly or indirectly with transcription factors and may even work by recruiting transcription factors to the promoter of genes targeted for regulation (Nair and Finkel 2004). The true mechanism of Dps’ hypothesized regulation may be a combination of all previously mentioned mechanisms. However, at this time, the regulatory capacity of Dps remains greatly uncharacterized and is therefore unsubstantiated and speculative.

The role of Dps in stress resistance in Escherichia coli

Careful examination of Dps has revealed its ability to protect in a variety of stressful conditions in several organisms. During stress, Dps is one of the principally overexpressed proteins and plays a crucial role in protecting E. coli from numerous stresses; including oxidative stress, high pressure, UV and gamma irradiation, thermal stress, and copper and iron toxicity (Martinez and Kolter 1997; Wolf et al. 1999; Choi et al. 2000; Ishikawa et al. 2003; Nair and Finkel 2004; Hong et al. 2006; Malone et al. 2006; Jeong et al. 2008; Yu et al. 2009). The protective role of Dps in extreme acidic (pH 2.0) and alkaline conditions (pH 12.0) has been demonstrated as well in E. coli (Nair and Finkel 2004). dps mutants are hypersensitive when subjected to each of the aforementioned stresses and experienced a sharp reduction in survival as a result of being challenged (Martinez and Kolter 1997; Choi et al. 2000; Nair and Finkel 2004; Jeong et al. 2008). However, Dps is most notably associated with protection against oxidative stress. All aerobic organisms are subjected to oxidative stress, a stress characterized by the production of hydroxyl radicals that can inflict lethal damage by binding and altering vital macromolecules. The toxicity of hydroxyl radicals is broadly dissipated because they react with virtually every biomolecule that they encounter. Another commonly encountered oxidative stress characterized by potent antimicrobial activity is exposure to hypochlorous acid (HOCl). During host infection, E. coli are subjected to this acid-based oxidative stress following uptake by phagocytic cells, after which neutrophils and macrophages release high concentrations of HOCl during their characteristic oxidative burst. HOCl (formed by the peroxidation of chloride ions) has a severe and damaging effect on cell membranes, proteins and nucleotides, making it extremely lethal (Klebanoff 1968; Foote et al. 1983). Hydrogen peroxide and HOCl generate similar reactive oxygen species in vivo according to the following formulas:


and consequently, resistance against both oxidative stresses is largely mediated by the same genes, including dps (Dukan and Touati 1996). Aerobic organisms have developed methods to protect vital macromolecules from oxidative agents generated during normal aerobic metabolism and those encountered from exogenous sources. One such protective mechanism is to induce the expression of genes encoding proteins capable of detoxifying oxidative agents and those involved in the repair of damaged macromolecules. Dps is a protein normally associated with stress resistance during stationary phase; however, it is not surprising that it is also crucial for hydrogen peroxide resistance during the log phase, as the protein is part of the OxyR regulon and is activated accordingly during log phase oxidative stress (Altuvia et al. 1994). Furthermore, it should be noted that stationary phase cultures are intrinsically more resistant to hydrogen peroxide than actively growing cultures because the stationary phase-specific sigma factor (σS) directs the expression of vital hydrogen peroxide detoxifying agents including the HPII catalase (katE) and the DNA repair enzyme endonuclease III (Farr and Kogoma 1991; Goodrich-Blair et al. 1996). Although these mediators of hydrogen peroxide resistance are in play and may alleviate the effects of hydrogen peroxide exposure, the importance of Dps for starvation-induced resistance to hydrogen peroxide has been clearly demonstrated via deletion mutant analyses (Almiron et al. 1992; Nair and Finkel 2004).

Martinez and Kolter (1997) suggested that the protective effect of Dps during hydrogen peroxide exposure is attributable to Dps physically binding and protecting groups within the major and minor grooves of DNA from the deleterious effects of hydroxyl radicals. However, because Dps does not protect DNA from other agents with the potential to inflict physical damage (i.e., alkylating agents), it is speculated that its protective effect is not solely because of shielding of DNA from oxidizing agents. Recently, it has been established that Dps prevents Fenton-mediated oxidative damage by trapping hydroxyl radicals within the protein shell (Bellapadrona et al. 2010). Hydroxyl radicals are generated via Fenton chemistry when hydrogen peroxide is used in the oxidation of ferrous ions within the core of Dps. Elegant stopped-flow analyses show that a conserved tryptophan residue (Trp52) located near the ferroxidase centre of Dps is in radical form after iron oxidation by hydrogen peroxide in E. coli. Essentially, Trp52 prevents the diffusion of newly generated hydroxyl radicals from the ferroxidase centre into solution, where they can significantly damage macromolecules. In E. coli, the end result of Trp52 binding of hydroxyl radicals produced via Fenton chemistry is that only a minute amount of these deleterious agents are released into solution following hydrogen peroxide exposure (Zhao et al. 2002; Bellapadrona et al. 2010). Not only is the presence of Dps Trp52 highly conserved throughout the prokaryotic world (present in over 99% of reported Dps sequences), the residue represents a means to trap and detoxify dangerous hydroxyl radicals generated during iron oxidation, and establishes a profound role for the protein shell of Dps.

Another stress often encountered by E. coli that represents a formidable barrier against infection is extreme acidity. Infecting cells must face environments with extremely low pH values within the stomach of hosts (around pH 1.0–2.0) and within the phagolysosome (pH 4.5) following uptake by phagocytic cells. The extreme toxicity of highly acidic environments results in damage to essential macromolecules and the cytoplasmic acidification that is associated with acid exposure (reviewed by Smith 2003). Acid stress is particularly damaging to DNA and results in depurination and depyrimidination that increases linearly with decreasing pH (Lindahl and Nyberg 1972; Friedberg et al. 2006). Such genetic alterations lead to unpaired nitrogenous bases and mismatches in repaired DNA sequences (Raja et al. 1991). Also, cytoplasmic acidification resulting from acid stress is detrimental to the activity of vital enzymes that perform optimally at neutral or near neutral pHs. In E. coli and other pathogenic bacteria alike, three main defence strategies are employed to combat severe acidity. Exposed cells have the ability to change membrane composition to protect cells from acid exposure, initiate enzymatic homoeostasis of cytoplasmic pH, and repair and/or prevent damage to vital macromolecules (Raja et al. 1991; Brown et al. 1997; Castanie-Cornet et al. 1999; Jordan et al. 1999). Our current knowledge suggests that Dps-mediated protection against acid falls into the latter category of protection, as Dps has shown a great ability to protect exposed cells against DNA strand breakage under conditions of intense acidic stress (Jeong et al. 2008). dps mutants are much more susceptible to strand breaks than the wild type in acidic conditions that ultimately results in a sharp reduction in dps mutant survivability (Choi et al. 2000; Nair and Finkel 2004; Jeong et al. 2008). DNA in acid stressed cells is protected by direct interaction with Dps via co-crystallization, which helps maintain genetic integrity during this time (Jeong et al. 2008). It is possible that Dps alleviates the elevated rate of depurination/depyrimidination via shielding as well. Dps may also function in its suggested regulatory role and induce genes that mitigate acid damage or those needed to keep cytoplasmic pH within a physiological range.

Within the last 4 years, an interesting discovery has been made with regard to Dps and stress resistance. Although Dps is primarily a cytoplasmic protein, its presence in the outer membrane has been repeatedly reported (Lacqua et al. 2006; Li et al. 2008). Changes in Dps levels in the outer membrane protein compartment have a pronounced effect on antibiotic sensitivity and resistance to bacteriophages. These observations are consistent with the protective role of Dps during stress responses and may result in the identification and characterization of novel protective mechanisms employed by Dps during stressful conditions.

Concluding remarks and future perspectives

The protective ability of Dps stems, in part, from its ability to effectively sequester iron atoms that would alternatively be used in the production of highly toxic free radicals via the Fenton reaction. The capacity to sequester and bind DNA in an extremely organized and rapid manner during times of nutritional deprivation and/or environmental stress is an additional approach utilized by Dps to provide protection. These protective strategies have proven to be essential for stress resistance in E. coli. During nutritional stress, when Dps is one of the predominate proteins produced by starved cells, the protective functions of Dps may reveal a general Dps-dependent defence mechanism utilized by bacteria and Achaea during times of starvation and oxidative stress. Although a great amount of knowledge regarding the DNA-binding ability and the ferritin-like properties of Dps has been acquired, there are significant questions that remain concerning the properties of Dps discussed in this review. For instance, what are the mechanisms behind the disassembly of DNA-Dps complexes that occurs upon nutrient up-shift and what other proteins are involved in this process? Also, what is the role of the Dps protein (beyond physical shielding) during acid stress and does the protein assuage or alleviate depurination/depyrimidanation or induce genes vital for acid resistance? Questions concerning Dps regulation by ClpAP and ClpXP remain as well. Specifically, in the process of maintaining Dps stability during starvation, does ClpAP act upon translational repressors of Dps or an RNAase targeting dps mRNA? Also, what are the mechanisms involved in ClpXP-mediated degradation of Dps during growth and is there an adapter protein that aids in the delivery of Dps to the ClpXP protease? Finally, what is/are the purpose or purposes for Dps truncation in ClpAP-mediated proteolysis? Answers to these questions are needed for full clarification of the protein’s functions and regulation. Knowledge is especially lacking with regard to the regulatory role of Dps. Proteomic studies performed by Almiron et al. 1992 are an excellent beginning for establishing a regulatory role for Dps in starved E. coli, and it is extremely probable that Dps does indeed display inductive and repressive capabilities that aid in protection and/or repair during environmental stress. Identification of Dps as a global regulator in E. coli would substantiate the ferritin-like protein as a major player in stress-related microbial physiology and further enhance its already diverse functional repertoire.


This research was supported by The Arkansas Bioscience Institute.