Streptomyces coelicolor Dps-like proteins: differential dual roles in response to stress during vegetative growth and in nucleoid condensation during reproductive cell division


*E-mail; Tel. (+44) 1792 295191; Fax (+44) 1792 602147.


The Dps protein, a member of the ferritin family, contributes to DNA protection during oxidative stress and plays a central role in nucleoid condensation during stationary phase in unicellular eubacteria. Genome searches revealed the presence of three Dps-like orthologues within the genome of the Gram-positive bacterium Streptomyces coelicolor. Disruption of the S. coelicolor dpsA, dpsB and dpsC genes resulted in irregular condensation of spore nucleoids in a gene-specific manner. These irregularities are correlated with changes to the spacing between sporulation septa. This is the first example of these proteins playing a role in bacterial cell division. Translational fusions provided evidence for both developmental control of DpsA and DpsC expression and their localization to sporogenic compartments of aerial hyphae. In addition, various stress conditions induced expression of the Dps proteins in a stimulus-dependent manner in vegetative hyphae, suggesting stress-induced, protein-specific protective functions in addition to their role during reproductive cell division. Unlike in other bacteria, the S. coelicolor Dps proteins are not induced in response to oxidative stress.


The volume of an unconstrained bacterial chromosome significantly exceeds the volume of the cell where it is contained, hence the need for a packaging mechanism to reduce chromosomal volume sufficiently to make it fit inside the cell. Such compaction is achieved by the concerted action of negative DNA supercoiling, molecular crowding and nucleoid-associated proteins (Luijsterburg et al., 2006). The functional role of nucleoid-associated proteins extends beyond their contribution to the physical structure of the nucleoid, also influencing global transcription levels by a variety of mechanisms that involve cooperation as well as antagonism between the various proteins (McLeod and Johnson, 2001; Dorman and Deighan, 2003).

Initially classified as histone-like proteins, nucleoid-associated proteins are high-abundance, low-molecular-weight, basic proteins that bind DNA with low specificity. Studies of the proteins associated with the Escherichia coli nucleoid identified high intracellular abundance of H-NS (histone-like structuring protein), HU (heat-unstable protein), Fis (factor for inversion stimulation) and IHF (integration host factor) (Varshavsky et al., 1977; Murphy and Zimmerman, 1997; Dame, 2005). Determination of the relative abundance of E. coli nucleoid-associated proteins in a growth phase-dependent manner revealed that the most abundant protein during stationary phase is Dps (DNA-binding protein from starved cells), when nucleoid compaction is more advanced (Ali Azam et al., 1999).

Dps-like proteins are encoded by most bacterial genomes and confer DNA protection by a dual mechanism. Non-specific interaction with DNA mediates the formation of a crystalline structure during stationary phase in E. coli that results in nucleoid hyper-condensation, physically shielding the DNA from damage (Almiron et al., 1992; Frenkiel-Krispin et al., 2001; Nair and Finkel, 2004). Additionally Dps protein levels increase when cells are subject to stress. Protection in this case results from the Fe(II) chelating activity provided by the Dps ferroxidase centre, which inhibits Fenton-type reactions and the formation of free radicals (Zhao et al., 2002).

The crystal structure of several Dps proteins shows a four-helix bundle motif subunit that assembles into a dodecamer, resulting in the formation of a hollow core with a net negative charge where mineralization takes place. The surface of Dps dodecamers exhibits a negative charge; hence Dps–DNA co-crystallization can only be mediated by the formation of divalent cation ion bridges (Frenkiel-Krispin and Minsky, 2006). The DNA binding ability of some Dps proteins is influenced by an N-terminal tail, rich in positively charged residues, that protrudes from the four-helix bundle although some family members possessing very short tails are unable to bind DNA in vitro (Ceci et al., 2003; 2005).

Soil, the natural niche for most Streptomyces species, provides a vast array of ever-changing physical and chemical challenges and hence very efficient DNA protection mechanisms must be activated in a dynamic manner to cope with each particular stress. These saprophytic bacteria grow vegetatively as filamentous branching syncytial hyphae in which nucleoids are very loosely compacted. In response to nutrient depletion or other stresses, they undergo morphological differentiation, erecting aerial hyphae. Multiple synchronized cell division then subdivides the apical compartment of an aerial hypha into a chain of pre-spore compartments each of which contains a single highly compacted nucleoid (Flärdh, 2003). Hence the production of spores involves a massive re-structuring of chromosome topology. DNA integrity must be protected during both vegetative and reproductive phases and specific mechanisms must be in place to protect topologically distinct nucleoid organizations.

Here we analyse the functional role of the three Dps proteins encoded by the Streptomyces coelicolor genome, focusing on the differential contribution of these proteins to nucleoid condensation during sporulation. Moreover, we demonstrate a link between various environmental stimuli and Dps expression during vegetative growth.


S. coelicolor Dps orthologues display an unusual sequence diversity

Analysis of the S. coelicolor genome revealed three Dps-like encoding genes, hereby named dpsA (SCO0596), dpsB (SCO5756) and dpsC (SCO1050). clustalw was used to align the putative protein products encoded by these genes with various Dps-like proteins that have been functionally characterized and whose 3D structure has been determined (Fig. 1). Conserved features of the S. coelicolor orthologues include the presence of helical segments structurally homologous to ferritin, identified using the secondary structure prediction tool PSIPRED ( The previously described ferroxidase centre or ‘Dps signature motif’ L(X)17HW(X)3 G(X)6H(X)14 D(X)3ER(X)5961D(X)18[W/H] (Roy et al., 2008) is also conserved, hence reinforcing their classification as members of the Dps protein family.

Figure 1.

Protein sequence alignment of Dps orthologues. Conserved ferroxidation site residues are indicated bellow the aligned sequences. Black bars indicate conserved helices. Amino acid tails described in Results are indicated in bold, with positively charged residues grey-shaded.

The protein alignment also revealed a remarkable diversity in protein sequence and architecture among the S. coelicolor paralogues (ScDps) (Fig. 1). DpsA has a 15-amino-acid N-terminal tail (N-tail) and a 25-residue C-terminal tail (C-tail), both putatively extending beyond the four-helix bundle. Each tail contains only one positively charged amino acid (lysine in the case of the N-tail and arginine for the C-tail). Contrastingly, DpsB possesses only a short eight-residue N-tail containing a single lysine, albeit at a different position. DpsC possesses a relatively long N-tail (around 42 residues) with six positively charged amino acids, four of which are lysine. The diversity in length and amino acid composition of these tails may reflect different functionalities of each ScDps.

Bacteria typically encode for one or two Dps-like proteins. Indeed, a search of available genome sequences revealed that only one other actinobacterial genome encodes three of these proteins (Streptomyces ghanaensis). Analysis of the phylogenetic relationship between Dps orthologues identified a distinct distribution pattern, with most actinobacterial sequences clustering around DpsB and DpsA, while DpsC forms part of a very distinctive clade of only three sequences (Fig. S1). DpsC mapped at a distal position among sequences from Proteobacteria; however, low bootstrap values do not support the hypothesis of DpsC having a proteobacterial ancestry.

It is evident from the phylogenetic distribution that Dps sequences from other Streptomyces species do not follow the distribution pattern exemplified by the partitioning of S. coelicolor DpsA, DpsB and DpsC into separate clades. For example, the two Dps-like sequences of Streptomyces scabies, Nocardia farcinica and Streptomyces avermitilis are all contained within the DpsB clade while two of S. ghanaensis paralogues localize with DpsB and the third with DpsA.

S. coelicolor Dps proteins have antagonistic roles during nucleoid compaction

Both single mutants in which each dps gene was disrupted and double mutants encompassing all combinations were constructed (Table 1, Fig. S2). Macroscopic phenotypic analysis of the mutant strains grown as patches on several different solid media revealed no major differences between parental and mutant strains, although the dpsA mutant aerial mycelium exhibited a delay (∼1 day) in developing the grey pigment-associated with spores when grown on MS sporulation medium.

Table 1.  Bacterial strains, plasmids and cosmids.
Strain or plasmidDescriptionTransposon insertiona (genome position), GenBank accessionSource
 S. coelicolor A3(2) M145Prototrophic SCP1– SCP2– Pgl+ Kieser et al. (2000)
 DSC0596, dpsA-M145 dpsA-::Tn50625G5.2.E08 (637508)This study
 DSC5756, dpsB-M145 dpsB-::Tn5062SC7C7.1.A06 (6293980)This study
 DSC1050, dpsC-M145 dpsC-::Tn5062G20A.F03 (1106709)This study
 DSC0596/5756 dpsA/B-M145 dpsA-::Tn5062 /dpsB-::Tn5062 Hyg5G5.2.E08 (637508) SC7C7.1.A06 (6293980)This study
 DSC0596/1050 dpsA/C-M145 dpsA-::Tn5062 /dpsC-::Tn5062 Hyg5G5.2.E08 (637508) G20A.F03 (1106709)This study
 DSC5756/1050 dpsB/C-M145 dpsB-::Tn5062 /dpsC-::Tn5062 HygSCG20A.F03 (1106709) 7C7.1.A06 (6293980)This study
 DSCO2085-1M145 ftsW::Tn5062(1)SC4A10.2.H05 (2239002)Mistry et al. (2008)
 DSCO2083M145 ftsQ::Tn5062SC4A10.2.B05 (2236542)Mistry et al. (2008)
 J2400 (whiG)M145 whiG::hyg Flärdh et al. (1999)
 DSCO5749 osaB-M145 osaB::Tn5062SC7C7.1.G11 (6285443)Fernandez-Martinez et al. (2009)
 DSCO5747-1 osaC-M145 osaC::Tn5062SC7C7.1.D06 (6278735)Fernandez-Martinez et al. (2009)
 M145ΔsigB sigB-M145 ΔsigB::apra Fernandez-Martinez et al. (2009)
E. coli
F′traD36 proA+B+lacIqΔ(lacZ)M15/Δ
(lac-proAB) glnV44 e14- gyrA96 recA1
relA1endA1 thi hsdR17
 Yanisch-Perron et al. (1985)
 E. coli ET12567  (pUZ8002)
Dam13::Tn9 dcm6 hsdM hsdR recF143 16 zjj201::Tn10 galK2 galT22 ara14 lacY1
xyl5 leuB6 thi1 tonA31 rpsL136 hisG4
tsx78 mtli glnV44, containing the non-transmissible oriT mobilizing plasmid, pUZ8002
 Flett et al. (1997)
 BW25113 (pIJ790)
K12 derivative: ΔaraBAD, ΔrhaBAD containing λRED recombination plasmid pIJ790 Gust et al. (2003)
 pQM5062pMOD+Tn5062, AmpicillinR and ApramycinRAJ566337.1Bishop et al. (2004)
pMOD+Tn5066 (Tn5062 Hyg), HygromycinR and AmpicillinR This study
 SCF55.2.F07ApramaycinRSCF55.2.F07::Tn5062 (638384)Bishop et al. (2004)
 7C7.1.A06HHygromycinR7C7.1.A06, Tn5066 replaces Tn5062 (6293980)This study
 G20.1.F03HHygromycinRG20.1.F03, Tn5066 replaces Tn5062 (1106709)This study
 pGEM-T EasyAmpicillinR Promega Corp.
 pSH152HygromycinR Mistry et al. (2008)
 pIJ8600tipA promoter, ApramycinR Sun et al. (1999)
 pALTER1TetracyclineR Promega Corp.
 pNA585mCherry Ausmees et al. (2007)
 pNA303eGFP Ausmees et al. (2007)
 pRWHis1ApramycinR Del Sol et al. (2006)
 pRWHis2ApramycinR Del Sol et al. (2006)
 pDpsA1pALTER containing dpsA This study
 pDpsA2pSH152 containing dpsA This study
 pDpsA4dpsA in pGEM-T Easy This study
 pDpsA6pSH152, dpsA::mCherry This study
 pDpsA7dpsA::His6, ApramycinR This study
 pDpsA7HdpsA::His6, HygromycinR This study
 pDpsA8dpsA coding sequence in pGEM-T Easy This study
 pDpsA9pIJ8600, tipA-dpsA::His6 This study
 pDpsB1dpsB in pGEM-T Easy This study
 pDpsB2pNA303, dpsB::eGFP This study
 pDpsB3pSH152, dpsB::eGFP This study
 pDpsB4dpsB::His6, ApramycinR This study
 pDpsB4HdpsB::His6, HygromycinR This study
 pDpsB8dpsB coding sequence in pGEM-T Easy This study
 pDpsB9pIJ8600, tipA-dpsB::His6 This study
 pDpsC1dpsC in pGEM-T Easy This study
 pDpsC2dpsC::His6, ApramycinR This study
 pDpsC2HdpsC::His6, HygromycinR This study
 pDpsC3pDpsC1, dpsC::mCherry This study
 pDpsC4pSH152, dpsC::mCherry This study
 pDpsC8dpsC coding sequence in pGEM-T Easy This study
 pDpsC9pIJ8600, tipA-dpsC::His6 This study

To examine the morphology of nucleoids during vegetative and reproductive developmental stages, a live-cell staining approach was used to avoid potential artefacts caused by fixative agents. The dpsA mutant displayed large irregularities in pre-spore compartment size when cell walls where stained with FITC-WGA (Fig. 2A). The DNA distribution in vegetative hyphae and pre-sporogenic aerial hyphae remained unaffected irrespective of the dps mutation combination. However, the DNA content or nucleoid morphology of pre-spore compartments was altered in a mutant-dependent manner. For the dpsA mutant, individual pre-spore compartments contained apparent larger than normal nucleoids flanked by adjacent compartments with smaller nucleoids. A correlation between nucleoid and spore compartment size was evident using a dual cell wall-nucleoid live stain (Fig. 2A). Atomic force microscopy, examining the surface topography of sporogenic hyphae, confirmed the irregularities in the dimensions of individual pre-spore compartments of the dpsA mutant (Fig. 2C). To assess if the viability of the small spore compartments was compromised, serial dilutions of a dpsA- spore suspension were quantified both by Neubauer haemocytometer spore counting and by plating on MS to determine colony-forming units (cfu). No significant differences were observed when determinations from both cell-counting methods were compared (not shown).

Figure 2.

Cytological analysis of S. coelicolor dps mutants.
A. FITC-WGA live staining of cross-walls in M145 and dpsA mutant aerial hyphae (an arrow points at typical large compartment flanked by smaller ones).
B. Dual cell wall (WGA-Texas Red) and nucleoid (Syto9, green) live staining of M145 and dpsA mutant. A white arrow points at a larger than normal spore compartment (TxR-S9: Texas Red-Syto9).
C. Atomic force microscopy topographic image of M145 and dpsA mutant aerial hyphae showing shape and size of spore compartments. Regular indentations indicate positions of sporulation septa.
D. Syto9 live staining of aerial hyphae of M145 and mutant strains under study. Insets show fourfold digital magnification of M145 and dpsA- nucleoids respectively. White arrow on inset indicates partially condensed nucleoids in dpsA-. Bar: 10 μm.

Laser scanning confocal microscopy of the dpsA mutant confirmed differences in DNA content between large and small compartments, providing an indication that the former contain two or more adjacent nucleoids (Fig. 2D, indicated by arrow on inset). The fluorescence signal profiles of nucleoids in a dpsA mutant spore chain also provide a representation of different DNA contents in individual compartments (Fig. 3). The parental strain M145 displays an expected regular contour in width and height for each peak, suggesting similar DNA content per spore compartment. In contrast, the dpsA mutant fluorescence profile shows big irregularities in both signal intensity and peak shape. The broad and double-peaked shape of the profile is consistent with the presence of more than one nucleoid per compartment. Adjacent peaks where the fluorescence signal is reduced by three- to fourfold compared with normal indicate the presence of small, possibly partial nucleoids. We did not detect an increase in the frequency of empty spore compartments in the mutant relative to the parental strain (data not shown). Genetic complementation of the dpsA mutant, integrating the gene under control of its native promoter at the phiC31 att site, resulted in the formation of pre-spore compartments resembling those of the wild type in terms of their regularity in size and DNA content (data not shown).

Figure 3.

Fluorescence profiles of Syto9-stained nucleoids. Images used to determine the profiles are shown next to the relative fluorescence plotted along a 12 μm line bisecting the stained nucleoids. Bar: 2 μm.

Similar analysis of dpsB and dpsC mutants revealed regularity in DNA content, nucleoid morphology and pre-spore compartment size. However, in contrast to the wild type, an apparent reduction in nucleoid width along the axis of the hypha with transversally wider nucleoids, resulting in an overall rounder shape, was evident in the dpsB mutant (Fig. 2D). A more significant and regular increase in nucleoid compaction was detected in the dpsC mutant, as estimated from the rounder shape of the nucleoids when compared with M145 (Fig. 2D). The genetic organization of both the dpsB and dpsC loci (both genes define the start of putative operons) indicates the possibility that the mutant phenotypes could result from polar effects (Fig. S1). To discount this, each mutant was genetically complemented by the introduction of an integrative plasmid carrying a copy of the corresponding dps gene with its native promoter sequence. In these strains, the appearance of nucleoids reverted back to the wild-type phenotype (data not shown), proving the link between each gene and the corresponding mutant phenotype.

Analysis of the dpsAB and dpsAC double mutant strains revealed regularly shaped, more compact nucleoids as in dpsB or dpsC single mutants (Fig. 2D). Apparently, disruption of dpsB or dpsC in a dpsA mutant background abolishes the irregular DNA content observed in individual pre-spore compartments of the latter, indicating that DpsB and DpsC together cannot drive nucleoid condensation in the dpsA mutant. Only while on their own (dpsAB or dpsAC double mutants) DpsB or DpsC can mediate nucleoid condensation, suggesting antagonizing roles for both proteins during nucleoid compaction in S. coelicolor, although other nucleoid-associated proteins may also intervene.

Fluorescence signal profiles of nucleoids in pre-spore compartments of single dpsB or dpsC mutants and in double dpsAB or dpsAC mutants confirmed that these mutants contain small compact nucleoids (Fig. 3). This is reflected in the number of nucleoids and hence pre-spore compartments in a given interval of a sporogenic hypha. For example, M145 and the dpsBC double mutant have nine nucleoids or pre-spore compartments in a 12 μm interval, dpsB or dpsC single mutants and the dpsAB double mutant have 11 nucleoids or pre-spore compartments per 12 μm interval, and the dpsAC double mutant has 11 nucleoids or pre-spore compartments per 12 μm interval. These differences were further quantified by determining for each strain the lengths of approximately 400 Syto9-stained nucleoids. Only nucleoids that formed part of a pre-spore chain were analysed. Measurements were used to calculate the percentage of nucleoids per 0.1 μm length interval and plotted in the 0–2 μm range. The histogram generated confirmed the differences in nucleoid size distribution between the mutants and parental strains (Fig. 4A). Measurements of nucleoid length for all strains displayed significant deviations from normality (P < 0.05; Kolmogorov–Smirnov test) thus statistical comparisons were confined to non-parametric tests. Significant differences (P < 0.05; Kruskall–Wallis test) were uncovered among all strains thus pair-wise Mann–Whitney tests with corrections for multiple comparisons were used post hoc. All strains differed significantly from each other in terms of nucleoid length. FITC-WGA staining of the cross-walls for each strain, followed by pre-spore compartment size measurement, revealed a similar distribution to that observed for the nucleoids, confirming the correspondence between nucleoid size and pre-spore compartment size (Fig. 4B). Interestingly, the measurements obtained from dpsA mutant compartments spread across a broad size range, as observed for nucleoid size measurements, but rather than a single population distribution as observed for the nucleoids, three peaks were clearly discernible at 0.8, 1.2 and 1.5 μm sizes, indicating three subpopulations of pre-spore compartment sizes in this mutant (Fig. 4B). The approximately regular distance between peaks (around 0.3–0.4 μm) indicates that pre-spore compartment size (as defined by cross-walls) can only increase or decrease as a function of discrete positional changes in septum location, rather than a continuous variation.

Figure 4.

Histograms showing distribution of nucleoid (A) and pre-spore compartment (B) sizes of S. coelicolor M145 and dps mutant strains under study. Both histograms were generated plotting the per cent of nucleoids or compartments per 0.1 μm size interval, across the 0.1–2 μm size interval.

Developmental regulation and localization of DpsA and DpsC to sporogenic hyphae

To localize their expression, C-terminal translational fusions between each Dps protein and autofluorescent proteins were engineered. The gene fusions, under control of the respective native promoters, were expressed in both the parental strain M145 and the corresponding dps mutant. This approach permitted evaluation of the functionality of the tagged protein, as measured by its ability to restore parental nucleoid morphology. Fluorescence microscopy of the M145/pDpsA6 strain (expressing DpsA::mCherry, DpsAmCh) revealed no detectable fluorescence due to mCherry in vegetative or pre-sporogenic aerial hyphae (Fig. 5A and B). However, spore chains exhibited strong red fluorescence due to expression of the fusion protein. Expression of DpsAmCh was limited to the apical pre-spore compartments, as no fluorescence was detected in compartments below the basal septum (Fig. 5C). The functionality of DpsAmCh was confirmed by its ability to restore regular nucleoid morphology to the dpsA mutant (data not shown). To examine the relationship between this developmental regulation of dpsA and sporulation septation, the translational fusion was expressed in an ftsW mutant that is unable to initiate septation and an ftsQ mutant that forms incomplete septa (Mistry et al., 2008). In both ftsW and ftsQ mutants DpsAmCh expression was localized to the apical aerial compartment, but was not detectable in the subapical compartment (Fig. 5D and E). The distribution of DpsAmCh in the apical aerial compartments of the ftsW mutant was discontinuous (Fig. 5F), reflecting the distribution of partially compacted nucleoids in the ftsW mutant aerial hyphae, as revealed by Syto9 DNA staining (Fig. 5G). In contrast, in a whiG mutant, no DpsAmCh or DpsCmCh expression was detected in aerial apical filaments, or in any tissue (results not shown).

Figure 5.

Expression of DpsA and DpsC is developmentally controlled. Bright-field and corresponding fluorescence images from S. coelicolor M145 (A, B and C), ftsQ- (D) and ftsW- (E) strains carrying a dpsA::mCherry fusion and M145 carrying a dpsC::mCherry fusion (H) are shown. Strain names are indicated on each panel. Insets show fourfold digital magnification. (F) shows discontinuous localization of DpsAmCh associated to partially condensed nucleoids. Syto9-stained DNA in an ftsW aerial hypha confirms partial nucleoid condensation in this mutant (G).

Analysis of expression of a translational fusion of DpsC::mCherry (DpsCmCh) revealed a similar pattern of protein localization in the sporogenic hyphal tip (Fig. 5H). The fluorescence signal intensity observed with DpsCmCh was noticeably lower than that observed with DpsAmCh, indicating either a lower level of expression or higher turnover of the fusion protein. Attempts to visualize a DpsBeGFP fusion were unsuccessful as it was not possible to differentiate between the background fluorescence control (M145 carrying the empty vector) and the recombinant strain carrying the dpsBeGFP fusion. This could be attributed to very low expression levels or high turnover of the DpsB fusion protein.

Gene-specific stress induction of dps expression during vegetative growth

The protective role of Dps proteins during oxidative and osmotic challenges has been documented in several unicellular eubacteria. Consequently, we investigated if stresses could also induce expression of the S. coelicolor dps genes in a developmentally independent manner. For this purpose the three proteins were translationally coupled with a hexa-histidine tag and expressed in the corresponding dps single mutant. Expression of each His-tagged protein, in each case controlled by its relevant native promoter, was tested by performing a time-course experiment to quantify basal expression levels throughout development.

For DpsA, a relatively constant protein abundance level was observed throughout vegetative and early aerial growth, followed by a sharp increase in the fraction prepared from spores (Fig. 6A). A very clear protein degradation product detected in the 24 h sample but fading or absent at later developmental stages hints at a post-translational control mechanism that could keep DpsA levels in check during growth phases characterized by non-condensed, diffuse, nucleoid conformation (Fig. 6A, black arrows). The abundance of DpsB remained constant throughout development, albeit at barely detectable levels when compared with DpsA (data not shown). Similarly, DpsC displayed a constitutive expression pattern at equally low abundance levels when compared with DpsA and, with this approach, there was no apparent increase in expression during sporulation (data not shown). Determination of the relative transcript abundance corresponding to each dps gene using quantitative real-time PCR confirmed the developmental regulation observed for dpsA and revealed a similar pattern for both dpsB and dpsC (Fig. S3A).

Figure 6.

Expression of Dps proteins is induced by different stimuli. Plasmids encoding C-terminal His-tagged DpsA, DpsB and DpsC were introduced into the corresponding S. coelicolor dps mutant strains and detected by immunoblot. When relevant, conditions used and treatment duration are indicated.
A. Time-course expression of DpsAHis throughout development. Arrows indicate putative degradation products during vegetative growth.
B. Stress-dependent induction of DpsAHis.
C. Osmotic shock-dependent induction of DpsAmCherry expression in S. coelicolor dpsA-/pDpsA6 germinating spores. Bright-field and fluorescence images are shown, as well as a fourfold magnified view of the fluorescent field showing uncondensed nucleoids.
D. Oligomeric state of DpsAHis in vivo is not modified by expression-inducing stimuli. Immunoblot of a Blue Native PAGE gel detecting DpsAHis under the conditions indicated. Arrow indicates putative DpsAHis monomer. Native molecular weight reference proteins used were: carbonic anhydrase (29 kDa), bovine serum albumin (66 kDa), β-amylase (200 kDa), apoferritin (443 kDa) and thyroglobulin (669 kDa).
E and F. Induction of expression of DpsBHis(E) and DpsCHis (F) after treatment under conditions indicated.

DpsA abundance increased dramatically when vegetative hyphae were subjected to osmotic stress and, to a lesser extent, to high temperature, but remained unaffected in response to oxidative stress (Fig. 6B, summarized in Table S1) or nutritional downshift (data not shown). The incremental increase in DpsA abundance caused by osmotic stress was not medium dependent and was also observed on a medium that supports only vegetative growth (data not shown). Osmotic stress induction of dpsA was confirmed using quantitative real-time PCR. A sharp increase in dpsA transcript abundance was induced by KCl treatment, while the abundance of dpsB and dpsC transcripts remained similar to the non-induced state (Fig. S3B and C), indicating that only dpsA expression is induced as part of the osmotic stress response. We subsequently analysed nucleoids in vegetative hyphae expressing DpsAmCh after osmotic stress. Red fluorescence due to DpsAmCh was clearly visible in hyphae subjected to osmotic stress, while no fluorescence was detected in the untreated control (Fig. 6C). Despite induction of DpsA no nucleoid condensation was detectable, implying that any protective role attributable to DpsA after osmotic stress is achieved in the absence of noticeable nucleoid compaction. The native state of DpsA was also determined before and after stress. The abundance of DpsAHis in a Blue Native gel corresponded to what was observed after SDS-PAGE. Migration of the protein in the gel was the same in all the conditions tested, at a size corresponding to a dodecameric state (240 kDa). A small amount of monomeric protein (20 kDa) could be detected in the salt-stressed sample (Fig. 6D, black arrow), indicative of some dissociation from the higher-order protein assembly. These observations, monitoring for the first time the in vivo oligomeric state of a Dps protein, suggest that dodecameric assembly per se is not enough to drive nucleoid condensation in vivo and that the contribution of other proteins/factors is needed.

DpsB expression was not significantly influenced by osmotic shock and only high temperature and nutritional downshift caused a small increase in protein levels (Fig. 6E). Except for a marked increase in abundance caused by temperature upshift, DpsC expression was unaffected by the other stimuli (Fig. 6F). It is noticeable that none of the Dps proteins was induced by H2O2, a typical inducer of dps gene expression in other bacteria. The viability of all dps mutant strains was similar to that of the parental M145 strain when exposed to increasing concentrations of H2O2. This indicates that in S. coelicolor DNA protection during oxidative stress is achieved without the contribution of Dps proteins, in contrast to their role in other bacteria.

The viability and phenotypes of the spores of dps mutant strains were tested after exposure to inducing stress conditions. Only heat stress affected spore viability and specifically for the dpsB mutant (Fig. S4). Growth on MS supplemented with 250 mM KCl did not affect significantly the growth or phenotypes of the mutant strains, except for a decrease in actinorhodin production by the dpsA mutant (data not shown).

To investigate induction of DpsA after osmotic stress, the plasmid encoding for DpsAHis was introduced into known osmoregulatory sigB, osaB and osaC mutants (Fernandez-Martinez et al., 2009). Expression of DpsA was induced after osmotic stress by a similar increment in these mutants as it was in the wild-type strain M145 (Fig. S5). SigB-independent induction of DpsA expression after osmotic stress is in contrast to how dps genes are induced after osmotic stress in other bacteria such as Bacillus subtilis (Antelmann et al., 1997). Disruption of the paralogous genes also had no influence on DpsA expression, as both basal and stress-induced expression of DpsA in dpsB and dpsC mutant backgrounds mirrored the expression profiles in the wild-type strain (data not shown). This confirms that regulation of dpsA expression is not modified in the absence or presence of osmotic stress to compensate for the missing paralogues.


Induction of expression of S. coelicolor Dps proteins is independent of both oxidative stress and SigB

Given the free-living non-motile lifestyles of Streptomyces, necessitating an ability to adapt to various environmental stresses in situ, it is perhaps not surprising that members of the genus possess multiple dps genes that can potentially contribute in combination to maintaining genome integrity. Comparative genome and phylogenetic analyses indicate that most members of the genus have two dps genes, orthologues of dpsA and dpsB, whereas a minority of species possess a third dpsC orthologue. The latter is related to a dps gene found in various species of Proteobacteria, including other inhabitants of soil ecosystems such as Bradyrhizobia, and its position near one end of the S. coelicolor genome is consistent with previous acquisition by lateral gene transfer. The ferroxidase centre is well conserved in all three S. coelicolor Dps proteins, hence an expectation that a primary role for these proteins would be in DNA protection, particularly during oxidative stress, in parallel to the established function of Dps proteins in bacteria such as E. coli and B. subtilis (Chen and Helmann, 1995; Nair and Finkel, 2004). However, we observed no induction of expression of any of the three S. coelicolor Dps proteins as a result of oxidative stress and also that the respective mutants do not exhibit any sensitivity to H2O2. Hence we conclude that DNA protection during oxidative stress is mediated by alternative functions that are likely regulated by the transcription regulators CatR and OxyR (Hahn et al., 2000; 2002). In contrast, induction of DpsA during vegetative growth as a result of osmotic stress, and of all three proteins by heat stress, suggests that these proteins have differential roles in the adaptive responses to these stresses.

Massive induction of DpsA expression in vegetative hyphae following osmotic stress is particularly intriguing and our data implicate different regulatory mechanisms controlling dpsA expression in response to different environmental stresses. Previous analyses of the osmotic stress response in S. coelicolor have demonstrated a critical role for alternative sigma factors, and in particular SigB (Lee et al., 2005). Expression of this sigma factor is upregulated after either osmotic or oxidative stress, positively regulating both sigB itself and expression of the SigB regulon that includes genes encoding other alternative sigma factors and the response regulator osaB (Fernandez-Martinez et al., 2009). Our results showed that osmotic stress induction of dpsA is, however, independent of this regulatory cascade.

Overexpression of a Dps protein can cause nucleoid condensation during exponential growth of a unicellular bacterium such as Staphylococcus aureus. It is believed that nucleoid condensation in those conditions may be due to the absence of a Fis orthologue in this bacterium (Morikawa et al., 2007). In E. coli this is prevented by Fis; as hyper-condensation of nucleoids is observed only in a fis mutant when the Dps protein is overexpressed (Ohniwa et al., 2006). The absence of detectable nucleoid condensation in vegetative hyphae of S. coelicolor after overexpression of dps genes induced by either osmotic or heat stress or driven by the tipA thiostrepton-inducible promoter suggests that similar mechanisms to prevent Dps-mediated nucleoid compaction may function. As there is no obvious Fis orthologue encoded by S. coelicolor, it remains to be determined what other nucleoid-associated proteins fulfil this role in maintaining diffuse nucleoids during vegetative growth.

S. coelicolor Dps proteins are required for correct spore nucleoid condensation and their expression is developmentally controlled

Developmental upregulation of dpsA and dpsC in S. coelicolor has a quite distinct function that has not been described in other bacteria to date, namely the compaction of nucleoids to permit their efficient segregation during sporulation cell division. In this process, a single apical syncytial aerial compartment is subdivided into 50 or more pre-spore compartments, each containing a single nucleoid. Our data indicate that the three Dps proteins collectively contribute to achieve a critical balance of nucleoid condensation required for this process.

Several other functions have been identified that operate to ensure correct chromosome segregation during sporulation in S. coelicolor. FtsK translocates incompletely segregated nucleoids into daughter cells prior to closure of the newly synthesized septum. In an ftsK mutant, an incompletely segregated chromosome is believed to be trapped by septum closure so that one compartment receives a terminally truncated genome (Wang et al., 2007). A developmental locus encoding SffA, a second FtsK-family protein, and the small membrane protein SmeA, which functions to localize SffA to septa, influences chromosome segregation and condensation (Ausmees et al., 2007). The SMC (structural maintenance of chromosome) protein is implicated in chromosome condensation in many bacteria including S. coelicolor in which it is upregulated during sporulation. A Δsmc mutant has an approximately sevenfold increased frequency of anucleate spores and viable spores also contain less compact nucleoids (Kois et al., 2009). Chromosome segregation is also partially dependent on a partition system involving ParA, an ATPase, and ParB that binds to numerous parS sites close to oriC (Jakimowicz et al., 2002). These proteins are both upregulated during sporulation with ParA likely promoting formation of the nucleoprotein ‘segregosome’ complex of ParB bound to the oriC region. parA or parB mutants display anucleate spores at frequencies of between 20% and 25%.

Collectively, these observations imply that the orchestration of nucleoid condensation and synchronous multiple septation in an aerial filament is dependent on several overlapping and partially redundant functions. Even in a triple ftsK smc parB mutant, the majority of spores were judged to contain complete nucleoids (Dedrick et al., 2009). Evidently DpsA also has a fundamental role in the process, although the frequency of anucleate compartments produced by the dpsA mutant does not differ from the wild-type frequency. Disruption of dpsA affects nucleoid condensation and results in a high frequency of individual pre-spore compartments apparently containing more than one nucleoid, whereas adjacent compartments possess a reduced DNA content, possibly indicative of the presence of an incomplete genome. We believe that in the dpsA mutant SMC, ParAB-, FtsK- and SffA-dependent segregation functions are unable to operate correctly on incompletely condensed nucleoids, thereby producing compartments with either multiple or partial nucleoids. Indeed, the distribution of these nucleoids at the onset of cell division may have a major influence on the positions of where septa are completed and, consequently, the localization of the segregation functions. In all dps mutants there is good size correlation between nucleoids and pre-spore compartments. Both are highly irregular in a dpsA mutant, suggesting that septum formation is blocked in parts of a sporogenic aerial filament occupied by multiple bulky incompletely condensed nucleoids, reminiscent of the nucleoid occlusion mechanism described in other bacteria.

Nucleoid condensation during sporulation is evidently mediated by the Dps proteins acting in combination. The irregularities in DNA content and pre-spore compartment size of a dpsA mutant are absent in either dpsAC or dpsAB double mutants. Indeed, the former has the most distinctly compact nucleoids of any mutant. This hyper-condensation correlates with the smallest compartment sizes of any mutant, again consistent with nucleoid occlusion guiding septum placement. The dramatic effect of mutation of an additional dps gene in the dpsA mutant implies an antagonistic role of either DpsB or DpsC in terms of DpsA function. Hence, in the absence of the condensing function of DpsA, both DpsB and DpsC in combination maintain the nucleoid in an incompletely condensed state. The function of either one of the DpsB or DpsC proteins on its own is to promote much tighter nucleoid condensation, as reflected by the degree of nucleoid compaction and compartment size in dpsB or dpsC single mutants, although the contribution from other nucleoid-associated proteins, still to be characterized, cannot be ignored. Hence we propose a model in which a DpsB and DpsC, either separately or as a putative DpsBC complex, modulate DpsA condensing function in the wild type to ensure an appropriate degree of nucleoid compaction at cell division. Establishing exactly how these proteins act together and in concert both with other nucleoid-associated and the divisome during sporulation cell division is an intriguing area for further functional and structural studies.

Experimental procedures

Bacterial strains and media

Streptomyces coelicolor A3(2) and E. coli strains are listed and described in Table 1. Cloning procedures were performed in E. coli JM109, while E. coli ET12567/pUZ8002 was used for intergeneric conjugative transfer of plasmid DNA into Streptomyces strains (Kieser et al., 2000). Gene replacement experiments were performed in BW25113 (pIJ790) strain as described (Gust et al., 2003). Culturing of E. coli strains was as recommended (Sambrook and Russel, 1989). S. coelicolor strains were grown on the surface of MS (mannitol soya flour) agar (Kieser et al., 2000). Streptomyces mutant strains were obtained using Tn5062-mutagenized cosmids (Table 1). Double mutants were generated by replacing the existing apramycin resistance gene in Tn5062 insertions with a hygromycin resistance gene, followed by conjugation into previously created and confirmed apramycin-resistant mutants (Bishop et al., 2004). The identity of all mutants was confirmed by Southern blot (Sambrook and Russel, 1989).

DNA manipulation and plasmid construction

Standard DNA procedures were performed as described (Sambrook and Russel, 1989). Plasmid pQM5066 was constructed by excising the apramycin coding sequence from pQM5062 using EcoICRI and replacing it with a blunt-ended BglII/BamHI fragment from pIJ963, containing the Streptomyces hygroscopicus hygromycin resistance gene (Kieser et al., 2000). The resulting plasmid contains a hyg gene flanked by the up- and downstream apramycin sequences present in Tn5062. In order to construct double mutants, the apramycin resistance gene on cosmids mutagenized with Tn5062 was replaced by a hygromycin resistance gene, adapting the PCR targeting system (Gust et al., 2003). Briefly, the hygR gene was excised from pQM5066 as a HindIII fragment, which provides enough flanking homologous sequences for λRed-mediated replacement of the apramycin gene in Tn5062.

The DNA fragments used to construct all plasmids are illustrated in Fig. 1. A 5.5 kb KpnI fragment from transposon-mutagenized cosmid SCF55.2.F07 ( was cloned into pALTER 1 resulting in pDpsA1, from which a 2.9 kb HindIII fragment containing dpsA and flanking sequences was excised, blunt-ended and ligated to pSH152 restricted with EcoRV, generating pDpsA2.

In order to create C-terminus translational fusions, each dps-like gene and corresponding upstream sequence containing the respective putative native promoter region were amplified by PCR. The forward primer in all cases incorporated an XbaI site while the reverse primers used were designed to replace the native stop codons with a BglII site (Table 1, Fig. S1). PCR amplifications were performed using Finnzymes' Phusion® DNA Polymerase (Finnzymes) following the manufacturer's recommendations. PCR products were subcloned into pGEM-TEasy, generating, respectively, pDpsA4, pDpsB1 and pDpsC1. The inserts contained on the above plasmids were extracted by XbaI/BglII restriction and ligated to pRWHis2, previously digested with XbaI/BamHI to remove an existing insert. In the resulting plasmids (pDpsA7, pDpsB4 and pDpsC2) the C-end of each dps gene is translationally fused to 6xHistidines. The apramycin-resistant marker on these plasmid was replaced by a hygromycin resistance gene using the PCR targeted system (Gust et al., 2003), producing the plasmids pDpsA7H, pDpsB4H and pDpsC2H.

C-end fusions to autofluorescent proteins were constructed as follows. Plasmids pDpsA7 and pDpsC1 were linearized with BglII and ligated to the mCherry coding sequence, recovered from pNA585 digested with BamHI/BglII. The plasmids obtained (pDpsA5, pDpsC3) were then digested with EcoRI to excise the insert, blunt ended and ligated to the integrative vector pSH152 digested with EcoRV, resulting in pDpsA6 and pDpsC4 respectively. A different approach was followed for dpsB, where pDpsB1 was digested with PstI/BglII and the dpsB containing insert was ligated to pNA303 digested with PstI/BamHI in order to generate pDpsB2, containing a C-end fusion to eGFP. The dpsB::eGFP fusion was subsequently excised with XbaI restriction, blunt ended and ligated to pSH152 EcoRV giving pDpsB3.

Plasmids used in overexpression studies were prepared by amplifying each dps coding sequence by PCR, using forward primers that introduced an NdeI site overlapping the corresponding start codon (DpsAF2, DpsBF2, DpsCF2) and the same reverse primers that replaced the stop codon with a BglII site. Each PCR product was cloned into pGEM-Teasy to obtain pDpsA8, pDpsB8 and pDpsC8, and then followed by NdeI/BglII restriction to release the dps coding sequence that was then ligated to pRWHis1, digested NdeI/BamHI. The plasmids obtained (pDpsA9, pDpsB9 and pDpsC9) allow the tipA promoter-mediated overexpression of each dps::His6.

Phylogenetic reconstruction

Amino acid sequences were retrieved from NCBI (, MBGD (, BROAD Institute ( and organism-specific genome sequence databases.

Phylogenetic reconstruction using dps protein sequences was performed in MEGA (Molecular Evolutionary Genetic Analysis, ver. 3.1) using the Neighbour-Joining method. The complete deletion option was selected to remove all missing data or alignment gaps, therefore eliminating the influence of different length N- or C-end tails present in only some of the sequences. Bootstrap values (500 replicates) were used to assess the robustness of the inferred phylogeny.

RNA isolation and quantitative RT-PCR

Total RNA isolation and quantitative RT-PCR (Q RT-PCR) procedures were performed as previously described (Del Sol et al., 2006; Fernandez-Martinez et al., 2009). Briefly, sterile cellophane discs placed on top of MS plates were inoculated with Streptomyces spores, grown overnight and then transferred to MS or MS-250 mM KCL. Discs remained in place for a specified time, after which, mycelia were collected and total RNA isolated with a Qiagen RNeasy mini kit. cDNAs were obtained using a RETROscript reverse transcription kit (Ambion) with random decamers. Real-time PCR was carried out on an iCycler iQ real-time PCR detection system (Bio-Rad) using SYBR-Green Supermix 2× containing Thermo-Start DNA Polymerase (ABgene). Gene-specific primers for Q PCR (Table 2) were designed using Beacon Design 2.0 (Premier Biosoft, USA). Known concentration dilutions of S. coelicolor genomic DNA were used as standard and total RNA samples were included as negative controls. S. coelicolor hrdB gene was used as internal control.

Table 2.  Oligonucleotides.
NameSequence (5′ to 3′ direction)

Protein methods

Total protein was used for immunodetection of proteins. In all cases, cellophane cultures were set up on MS agar as described above. Following incubation overnight (approximately 16 h) discs were transferred to plates providing the stimulus assessed or to stimulus-free plates acting as controls for basal expression levels. In the case of overexpression studies, discs were transferred to plates containing the antibiotic thiostrepton at 5 and 25 μg ml−1. In studies looking at negative supercoiling, discs were transferred to plates containing (i) 250 mM KCL, (ii) MS-containing Novobiocin (10, 50 and 100 μg ml−1) and (iii) MS-containing KCL Novobiocin at the previously stated concentrations. The search for conditions that induce expression of Dps proteins followed an experimental format focused on short-term physiological responses. The conditions tested were osmotic up-shock (MS/250 mM KCl), high temperature (MS/42°C), starvation (transfer to Minimal Medium without a Carbon Source), nutritional upshift (transfer to Nutrient Agar) and oxidative stress (MS/1 mM H2O2). After incubation mycelia were scraped from the cellophane discs and suspended in Sonication Buffer [50 mM Tris-HCl, pH 8, 200 mM NaCl, 15 mM EDTA, Complete protease inhibitor cocktail (Roche Diagnostics)]. In time-course expression experiments, mycelia were collected at 16, 24, 36 and 48 h. One final sample was taken at a later stage during sporulation. When assessing spore-associated protein levels the plates were incubated for 3–4 days and the spores collected in water. Mycelium-free, spore suspensions were created as described in Kieser et al. (2000). Briefly, after filtering through cotton-wool, spores were pelleted by centrifugation and re-suspended in Sonication Buffer. Cells or spores were disrupted by sonication (20 s burst on ice) until a clear lysate was obtained. Cell-free extracts were obtained by centrifugation (13 000 r.p.m. for 3 min) and recovery of the supernatant. Total protein concentration was determined using the Bradford method (Bio-Rad). SDS-PAGE was performed as previously described (Sambrook and Russel, 1989), loading 10 μg of total protein per lane in 15% SDS-PAGE gels. Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Hybond-P, Amersham) using a semi-dry electrophoretic transfer cell (Trans-Blot SD, Bio-Rad). Immunological detection was performed using an ECL Advance Western blotting detection kit (Amersham Pharmacia Biotech). His-tagged proteins were detected with a Penta-His peroxidase conjugate (QIAGEN). The relative abundance of bands was estimated using Quantity One software (Bio-Rad). Due to low abundance levels, detection of DpsBHis and DpsCHis required the use of lower of Penta-His peroxidase conjugate dilution (1/3000) than when detecting DpsA (1/5000). Additionally, blocking of membranes for DpsAHis detection was achieved using non-fat dry milk, while for DpsBHis and DpsC 5% BSA was employed to increase sensitivity.

Blue Native PAGE was performed using Invitrogen's NativePAGE™ Novex® Bis-Tris Gel System following the manufacturer's recommendations.

Assessment of mutant phenotypes

Differences in colony and nucleoid phenotypes of dps mutants were compared with the wild-type M145 strain. To assess colony morphology, strains were plated on MS and R5 agar (Kieser et al., 2000). Strains used in fluorescent microscopy were taken as coverslip impressions on the surface of MS growing colonies after 3 days or grown in the acute angle of a coverslip inserted in the agar medium. Live-cell nucleoids were stained with Syto9 (Invitrogen) (5 μM diluted in 20% Glycerol). Dual staining of live-cell walls and nucleoids was performed using Syto9 (nucleoid) and WGA-Texas Red (cell wall), both from Invitrogen. Briefly, coverslip impressions were immersed in 2% BSA/Texas Red/PBS solution and incubated for 1 h. Excess stain was removed with gentle washes in PBS and coverslips were mounted on slides in a drop of Syto9/Glycerol solution. Single staining of live-cell walls was performed as above but using FITC-WGA (Invitrogen). Images were obtained using a Nikon Eclipse E600 epifluorescence microscope fitted with a Coolsnap microscope camera (RS Photometrics., Tucson, AZ).

Laser scanning confocal microscopic (LSCM) images of Syto9-stained coverslip impressions were obtained using a Zeiss LSM510 META (Carl Zeiss) through a 63×/1.4 NA plan apochromat DIC objective lens. Samples were visualized using an argon laser (488 nm) and a long-pass filter of 505 nm. Z-stack images though spore chains were generated using a constant z-stack interval (0.28 μm), with 15 z-steps and a pixel resolution of 1024 × 1024. Projections of z-stacks were compiled using Zeiss LSM image browser.

Fluorescence intensity profiles along spore chains stained with Syto9 were created in ImageJ (Abramoff et al., 2004). Briefly, freehand lines, approximately 12 μm in length, were drawn along the spore chains, passing through the centre of adjacent nucleoids. Fluorescent profiles were created using the ‘plot profile’ function. Measurements of nucleoid length were performed using Scion Image software (Scion Corp., Frederick, MD) obtained from

Atomic force microscopy samples were prepared and imaged as described previously (Del Sol et al., 2007).

Data analysis

Data for all nucleoid measurements were initially analysed using the Kolmogorov–Smirnov test for normality. Data that displayed extreme skewness and differed significantly from normality (i.e. P > 0.05) were analysed using non-parametric tests. Where significant differences were observed under Kruskall–Wallis tests, pair-wise Mann–Whitney tests were performed post hoc to identify strains that differed most significantly. To test the probability of making a type I error under multiple comparison procedures, family-wise type 1 errors were calculated using the Dunn-Sidak method (Sokal and Rohlf, 2001). All statistic procedures were undertaken in Statistical Package for Social Sciences (SPSS) for windows version 13.0.

Spore viability assay

Spore suspensions were incubated at 60°C for different time intervals, serially diluted and plated on MS agar plates. Colonies were counted and the cfu percentage per treatment, referred to an untreated sample (t = 0), was determined. The cfu percentage data were plotted against time, using a log scale.


This research has been supported by grants from the BBSRC (BB/E019242/1) and European Commission (LSH-IP 005224).