Structural and biochemical analysis of a phosin from Streptomyces chartreusis reveals a combined polyphosphate- and metal-binding fold.

X-ray crystallographic analysis of a phosin (PptA) from Steptomyces chartreusis reveals a metal-associated, lozenge-shaped fold featuring a 5-10 Å wide, positively charged tunnel that traverses the protein core. Two distinct metal-binding sites were identified in which the predominant metal ion was Cu2+ . In solution, PptA forms stable homodimers that bind with nanomolar affinity to polyphosphate, a stress-related biopolymer acting as a phosphate and energy reserve in conditions of nutrient depletion. A single protein dimer interacts with 14-15 consecutive phosphate moieties within the polymer. Our observations suggest that PptA plays a role in polyphosphate metabolism, mobilisation or sensing, possibly by acting in concert with polyphosphate kinase (Ppk). Like Ppk, phosins may influence antibiotic synthesis by streptomycetes.

Abbreviations

CHAD, conserved histidine α‐helical domain

polyP, polyphosphate

Streptomyces are Gram‐positive, filamentous soil bacteria, capable of producing a large variety of secondary metabolites. Many of these bioactive molecules, which enable the genus to thrive in highly competitive natural environments, are used as therapeutics or phytosanitary agents in medicine and agriculture 1, 2. Although Streptomyces have been amongst the most prolific sources of clinically relevant antibiotics for decades 3, a recent estimate suggests that only 10% of the metabolic diversity of the genus has been explored, with the vast majority of biosynthetic pathways remaining silenced under typical laboratory conditions 4, 5, 6. To unlock the full chemical repertoire of Streptomyces and reveal new molecules of biomedical interest, a better understanding is needed of the intricate regulatory network that governs secondary metabolism.

Interestingly, the most efficient trigger of antibiotic biosynthesis in Streptomyces appears to be a nutritional limitation in inorganic phosphate (P), which correlates with reduced ATP levels and, consequently, energetic stress 7, 8, 9. As in other bacteria, growth under P‐limiting conditions leads to the induction of the Pho regulon, governed by the two‐component regulatory system PhoR/PhoP 10, 11. As soon as falling P concentrations render uptake via low‐affinity transport systems inadequate, the sensory kinase PhoR autophosphorylates and transfers its phosphoryl group to the response regulator PhoP 12, 13. This transcription factor represses a range of metabolic pathways 10, 14, 15 while activating the expression of proteins involved in re‐establishing P homeostasis 16 and the scavenging of P from external sources. The latter include secreted phosphatases 10, 17, 18 as well as low‐ and high‐affinity phosphate transport systems, PitH2 and PstSCAB 19, 20.

In the model strain Streptomyces lividans, one of the genes strongly upregulated in conditions of P limitation 21 and found to exert a major negative influence on antibiotic production 22, 23, 24 is ppk. This gene encodes a polyphosphate (polyP) kinase (Ppk), which uses ATP to synthesise polyP 22. Under conditions of low ATP content, Ppk mainly acts as an adenosyl diphosphate kinase to catalyse the reverse reaction and regenerate ATP from ADP and polyP 22, 23. As a consequence, the enzyme plays an important role in maintaining energetic homeostasis during phosphate limitation 23.

Interestingly, the gene located immediately downstream of ppk encodes a protein comprised of a so‐called CHAD (conserved histidine α‐helical domain). Although the exact function of CHADs is unclear, these domains frequently occur in conjunction with CYTH (CyaB and thiamine triphosphatase homology) domains and are thought to participate in phosphate and nucleotide metabolism 25. Based on their unusually high histidine content, CHADs have also been predicted to bind divalent metal ions 25. More recently, CHAD‐containing proteins were found to co‐localise with polyP granules in bacteria 26. They were therefore redesignated ‘phosins’ or polyP‐targeting proteins (PptA or PptB, depending on the presence of additional domains). In the present study, we have structurally and biochemically characterised PptA from a streptomycete, as we hypothesised that this protein, like Ppk, might play a key role in the response to P‐limiting conditions and the regulation of secondary metabolism.

Materials and methods

Recombinant protein expression and purification

A codon‐optimised synthetic gene corresponding to the full‐length PptA protein (RefSeq WP_010036273.1) was purchased from Eurofins Genomics (Ebersberg, Germany). This gene was subcloned into pET‐28wTEV, a derivative of pET‐28b (Merck Chemicals and Life Science GmbH, Darmstadt, Germany) encoding a TEV‐protease site in place of the original thrombin recognition sequence. The His6‐tagged protein was expressed in Escherichia coli (E. coli) strain BL21 (DE3). An overnight preculture, grown from a single colony in LB medium supplemented with 50 μg/ml kanamycin, was diluted 1:100 in antibiotic‐containing medium and incubated shaking at 37 °C until the OD600 nm reached 0.4. Protein expression was induced by adding IPTG to a final concentration of 1 mm, followed by incubation for a further 4 h. Bacteria were collected by centrifugation for 15 min at 4 °C, 5000 , resuspended in buffer A (20 mm Tris/HCl pH 7.4, 400 mm NaCl) and lysed by sonication (Bandelin Sonopuls HD 2070, 10 min, 50% bursts, 50% power) on ice. Cell debris was removed by centrifugation for 30 min at 4 °C, 40 000 . The cleared lysate was then applied to a cobalt column (HisPur Cobalt Resin, Thermo Fisher Scientific GmbH, Dreieich, Germany) equilibrated in buffer A. The protein was eluted in buffer A containing 300 mm imidazol. Fractions containing PptA were pooled and incubated for 16 h at room temperature with TEV protease (0.7 mg per 100 mg of PptA), followed by further purification using gel filtration (Superdex 200, GE Healthcare GmbH, Munich, Germany). The resulting protein appeared as a single band on Coomassie‐stained SDS/PAGE gels.

X‐ray crystallography

Prior to crystallisation, purified PptA was concentrated to 25 mg/ml by means of ultrafiltration (Vivaspin Turbo 15, 10 kDa molecular weight cut‐off, Sartorius Stedim Biotech GmbH, Goettingen, Germany). The protein was crystallised using the sitting drop method at 22 °C, using a reservoir solution containing 100 mm MES/HCl pH 6.5 and 1.6 m MgSO4. Before flash‐freezing in liquid nitrogen, crystals were incubated for several seconds in five successive buffers containing 100 mm MES/HCl pH 6.5, 1.6 m MgSO4 and increasing amounts (6, 12, 18, 24 or 30%) of glycerol. A bromide derivative was prepared by complementing these buffers with 150, 300, 450, 600 and 750 mm NaBr, respectively.

All crystallographic measurements took place at 100 K. Native and bromide derivative data were recorded on beamline BL14.1 operated by the Joint Berlin MX‐Laboratory at the BESSY II electron storage ring (Berlin‐Adlershof, Germany) 27. All data were processed using the program XDS 28. An overview of data collection statistics can be found in Table 1.

Table 1

Data collection and refinement statistics for the structure determination of PptA. Values in parentheses pertain to outer resolution shell reflections.

	Native	NaBr
	PDB code 6RN5
Data collection
Source	BESSY BL14.1	BESSY BL14.1
Wavelength	0.905002 Å	0.920113 Å
Detector	PILATUS 6M	PILATUS 6M
Space group	C2221	C2221
Cell dimensions
a	49.3 Å	48.7 Å
b	132.2 Å	132.6 Å
c	116.4 Å	116.0 Å
Resolution range	46.2–2.04 Å (2.16–2.04 Å)	45.8–2.83 Å (3.00–2.83 Å)
R sym	10.6% (91.6%)	10.4% (43.9%)
<I/σ(I)>	19.2 (2.3)	15.8 (4.3)
Data completeness	98.9% (94.5%)	99.7% (98.4%)
Average redundancy	12.3 (8.9)	6.8 (6.6)
Wilson B	38.1 Å2	41.9 Å2
Refinement
Resolution	46.2–2.04 Å
Total no. reflections	24,546
Test set	5% (1228)
R work	19.1%
R free (5% test set)	23.7%
Protein atoms	2155
Other non‐solvent atoms	28
Solvent atoms	134
RMSD bond lengths	0.0138 Å
RMSD bond angles	1.53°
Average B	43.2 Å2

The structure was solved by means of single‐wavelength anomalous diffraction (SAD), using Phaser 29 in combination with Phenix 30. Multiple rounds of manual model building and refinement were carried out using Coot 31, 32, Refmac 33, 34 and the CCP4 package 35. Structure refinement statistics have been summarised in Table 1. Figures showing molecular models were prepared using pymol 0.99rc6 (DeLano Scientific LLC, Palo Alto, CA, USA). The surface charge distribution of PptA was calculated using ABPS 36 with an ionic strength value of 150 mm.

The structure and X‐ray diffraction data for PptA have been deposited in the Protein Data Bank (PDB) with accession number 6RN5.

Metal analysis

X‐ray fluorescence analysis 37 of PptA crystals was performed at beamline BL14.1 operated by the Joint Berlin MX‐Laboratory at the BESSY II electron storage ring (Berlin‐Adlershof) 27. Metal concentrations were determined under strongly denaturing conditions (in the presence of 8 m urea) according to the Zincon method described by Säbel et al. 38.

Small‐angle X‐ray scattering

Small‐angle X‐ray scattering (SAXS) data were recorded at beamline P12 of the EMBL outstation at PETRA III, DESY, Hamburg 39, using a PILATUS 2M pixel detector, a sample‐to‐detector distance of 3.1 m and a wavelength of 1.24 Å. Measurements covered the momentum transfer range 0.008–0.47 Å−1 (defined as 4π sin(θ)/λ, where 2θ is the scattering angle and λ is the X‐ray wavelength). Solutions contained 20 mm Tris/HCl pH 7.4, 400 mm NaCl and 1.3–5.5 mg/ml PptA. The sample temperature was 283 K in all experiments. A capillary flow cell was used to collect 20 successive 50 ms exposures for each sample, revealing no significant change in the course of any one experiment. The data were normalised to the intensity of the transmitted beam and radially averaged. Scattering of the buffer was subtracted and the difference curves were scaled to unity protein concentration (1 mg/ml). Merging and further data analysis were carried out with version 2.8.4 of the ATSAS package 40. The program CORAL 40 was used to compute theoretical SAXS curves based on the crystallographic homodimer, as well as to model missing loops and extremities. The SAXS data and corresponding CORAL model of the PptA dimer have been deposited in the SASBDB data base with accession number SASDF76.

Polyphosphate interaction

PolyP was obtained from (Sigma‐Aldrich Chimie SARL, Saint‐Quentin‐Fallavier, France). To detect polyphosphatase activity, polyP was incubated with increasing amounts of PptA, in a buffer containing 20 mm Tris/HCl pH 7.4 and 200 mm NaCl. The effect of protein activity on polyP length was subsequently analysed using denaturing PAGE. An equal volume of denaturing sample buffer (89 mm Tris, 89 mm boric acid, 2 mm EDTA, 7M urea, 12% Ficoll) was added to the reactions, which were then applied to a gel containing 10% polyacrylamide (1 : 29), 89 mm Tris, 89 mm boric acid, 2 mm EDTA and 7M urea. The running buffer contained 89 mm Tris, 89 mm boric acid, 2 mm EDTA (1 × TBE). Electrophoresis at 15 V/cm was continued until the bromophenol blue that was loaded separately in a neighbouring lane had reached the bottom of the gel. PolyP was visualised using the negative staining method of Smith and Morrissey 41. Similarly, physical interaction between polyP and PptA was analysed using an electrophoretic mobility shift assay (polyP‐EMSA) based on native PAGE. The experimental set‐up, buffers and gels were identical to those used for the polyphosphatase assays, except that urea was omitted from both gel and sample buffer.

Calorimetric experiments were carried out using a MicroCal PEAQ‐ITC calorimeter (Malvern Pananalytical), according to the instructions provided by the manufacturer. All analyte samples contained 20 mm Tris/HCl pH 7.4 with 200 mm NaCl and were degassed immediately before use. The cell was loaded with 400 μL of a 100 μm PptA sample, whereas the syringe contained either 400 μm or 300 μm polyP with a chain length of 25 or 45 phosphates, respectively. Experiments were performed at 25 °C and involved successive 4 μL injections with 150 s intervals and a mixing speed of 500 Hz. Data were analysed using the MicroCal PEAQ‐ITC software, assuming a single set of equivalent binding sites.

Results

Crystallographic analysis of PptA reveals a novel fold with unique structural features

For our structural studies, we chose a phosin from Streptomyces chartreusis. Compared to orthologs from other streptomycetes, this protein contains fewer regions of low complexity that are likely to interfere with crystallisation. His6‐tagged PptA was expressed in E. coli and purified to homogeneity. Crystals were obtained that diffracted to 2.04 Å, allowing us to solve the structure by means of bromide SAD as described in Materials and methods.

As Fig. 1 shows, the PptA fold consists essentially of α‐helices, pseudosymmetrically arranged to form a flat, lozenge‐like structure. Arguably the most striking feature of the structure is a central tunnel, 5–10 Å in diameter that traverses the protein core. Its inner surface is highly positively charged as it is almost entirely made up of Arg side chains (Fig. 2). Three ordered sulfate ions, originating from the crystallisation buffer, are present within the confines of the tunnel and partially compensate its charge. The ions are contacted by a total of nine Arg residues, as well as a His and a Tyr side chain (Fig. 2B).

Figure 1

The crystal structure of PptA, shown in two 90°‐rotated views. Cu2+ ions are depicted in green, sulfates in yellow (sulfur) and red (oxygen). Dotted lines represent disordered loops for which no electron density was observed. Disordered regions at the N and C terminus are not shown.

Figure 2

PptA contains a positively charged tunnel traversing the protein core. (A) Surface charge distribution of PptA, represented by a colour spectrum from red (‐5 kT/e) via white (0 kT/e) to blue (+5 kT/e). Panels correspond to 180°‐rotated views. (B) Stereo view of the central tunnel and the Arg side chains lining it. Sulfate ions sequestered from the crystallisation buffer are shown. Also depicted are His and Tyr side chains that participate in sulfate binding.

Intriguingly, two metal‐binding sites could be identified in the structure. One of these is located in the interior of the protein, whereas the other lies on a crystallographic symmetry axis between adjacent molecules in the crystal lattice (Fig. 3). The presence of metal ions was initially deduced from unusually high‐local electron densities (16 σ and 6.8 σ respectively) as well as the characteristic coordination of His side chains. Since it was not possible to establish the identity of the bound ions on the basis of our crystallographic data and the observed coordination chemistry alone 42, we analysed the metal content of PptA crystals by X‐ray fluorescence (XRF) 37. This experiment unambiguously revealed the presence of Cu2+. In addition, a smaller signal corresponding to Zn2+ could be detected. Specific protein‐metal ratios were determined by chemical means 38, revealing averages of 1.2 atoms of Cu2+ and 0.18 atoms of Zn2+ per protein molecule. Based on these results, we included the predominant metal ion, Cu2+, in our final crystallographic model of PptA.

Figure 3

The two metal‐binding sites identified in PptA, with Cu2+ shown in green. (A) The intramolecular metal‐binding site, with the final 2mFo‐DF c electron density map contoured at 2.0 σ. (B) The intermolecular metal‐binding site, located on a crystallographic 2‐fold symmetry axis (arrow). The 2mFo‐DF c electron density map shown here is contoured at 1.5 σ.

PptA forms stable homodimers in solution

The PISA analysis 43 of protein–protein contacts in the crystal suggested stable association of PptA into homodimers (Fig. 4A), with a buried interaction surface of 965 Å2. The program predicted a ΔG o of dissociation amounting to 6.5 kcal/mol, or 28.3 kcal/mol if metal ions were taken into account. In addition, gel filtration results, obtained using a Superdex 200 matrix (data not shown) pointed at a molecular mass of approximately 80 kDa, are consistent with dimer formation.

Figure 4

PptA forms a stable dimer in solution. (A) The PptA dimer as observed in the crystal lattice, with subunits in blue and pink. The two images show 90°‐rotated views. (B) Concentration‐normalised scattering from a PptA solution (black dots). The momentum transfer is defined as s = 4π sin(θ)/λ, where 2θ is the scattering angle and λ = 1.24 Å is the X‐ray wavelength. Theoretical scattering data for the dimer (red curve, χ2 = 1.1) were generated using the program CORAL 40, which accounted for loops that were missing in the crystallographic model. (C) Surface representation of the final model produced by CORAL, oriented as in the upper image of panel (A). Subunits of the dimer are shown in blue and pink, with regions added by CORAL in dark blue.

The oligomerisation state of PptA in solution was further investigated by small‐angle X‐ray scattering (SAXS), results of which are shown in Fig. 4B. Comparison of the Guinier‐derived forward scattering (I 0) to that of a bovine serum albumin (BSA) reference indicated a particle mass of 58.8 kDa, whereas a recently developed concentration‐independent Bayesian method 44 yielded a value of 78.5 kDa. These results are in good agreement with the expected mass of a PptA dimer (79.2 kDa). A more detailed analysis was carried out with the program CORAL, which was used to add strings of dummy residues to the high‐resolution crystallographic structure in order to account for missing loop regions as well as the undefined N‐ and C‐terminal ends of the protein. The optimised model generated by CORAL (Fig. 4C) explains the experimental data to within experimental error (Fig. 4B, fitted curve in red), corroborating the dimerisation mode observed in the crystal lattice.

PptA binds with nanomolar affinity to polyphosphates

In view of the proximity of the pptA and ppk genes and their potential co‐regulation, we hypothesised that the PptA protein might interact with polyP. The presence of a string of sulfate ions inside the tunnel also supports this idea, as occupation of phosphate‐binding sites by sulfate is frequently observed in protein crystals obtained with sulfate as the main precipitant 45.

We first tested the hypothesis that PptA acts as a polyphosphatase, by analysing the stability of polyP in the presence and absence of the protein. However, the length of polyP molecules, reflected in their electrophoretic mobility under strongly denaturing conditions, was not affected by a 1 h preincubation with PptA (data not shown). We next asked whether PptA is able to stably interact with polyP. In contrast to results obtained under denaturing conditions, native PAGE of PptA‐polyP mixtures (Fig. 5A) revealed a dramatic change in polyP electrophoretic mobility with increasing protein concentration, indicative of a direct and highly stable interaction. At low protein‐polyP ratios, a complex characterised by a limited size and net negative charge migrated several millimetres below the top of the gel, as a single, discrete band. At higher protein‐polyP ratios, as the available polyP started to become saturated, a low‐mobility complex appeared that barely migrated into the gel at all. Presumably, the latter complex consisted of multiple PptA molecules bound to a single polyP chain, resulting in an increased overall size of the complex as well as a further reduction in its overall negative charge.

Figure 5

Interaction of PptA with polyP. (A) PolyP electrophoretic mobility shift assay (polyP‐EMSA). (B) Raw ITC data for 25‐mer polyP. (C) Fitted binding curves for 25‐mer (black) and 45‐mer (red) polyP. The molar ratio is defined as the polyP concentration divided by the PptA dimer concentration. (D) Summary of the thermodynamic characteristics of the interaction, determined using 25‐mer and 45‐mer polyP (colours as in C). According to these independent analyses, a single PptA dimer binds 15.0 and 14.4 phosphate units respectively.

The interaction of PptA with polyP was analysed in more detail by means of isothermal calorimetry (ITC). Independent titrations with two polyP preparations of different chain length (Fig. 5B–D) produced similar results, indicating a K d value around 100 nM and a binding site size of 14–15 phosphate moieties per PptA homodimer. Binding is entirely entropy‐driven, with a positive ΔH o and a strongly negative ‐TΔSo contribution to ΔG o, the standard free energy change (Fig. 5D).

Discussion

In the study presented here, we analysed the structural and biochemical properties of a phosin (PptA) from S. chartreusis. The highly unusual polyphosphate‐ and metal‐binding fold that we identify has not been described in the literature, but bears similarity to a previously deposited crystal structure (PDB entry 3E0S). The latter, annotated as an uncharacterised protein from Chlorobaculum tepidum, was solved in the context of a structural genomics initiative. Superposition of the two structures via secondary structure matching in the program Coot 31, 32 yields a Cα‐RMSD of 2.9 Å for a total of 244 aligned residues (Fig. 6). Like PptA of S. chartreusis, 3E0S consists entirely of α‐helices that together form a thin, pseudosymmetrical lozenge with a central tunnel running through its core. As the superposition shows, the relative positioning of most α‐helices varies considerably between the two structures, as does their length. Moreover, several α‐helices are missing in PptA. In spite of these marked differences and the very limited sequence identity between the two proteins (14.5% in the structure‐based alignment shown in Fig. 6B, excluding gaps), the net positive charge of the tunnel surface is essentially conserved. Similar to PptA, 3E0S features multiple sulfate ions apparently sequestered from the crystallisation buffer. These observations strongly suggest that 3E0S and other divergent members of the CHAD family are also capable of polyP binding. Dimerisation appears to be conserved, as a very similar 2‐fold symmetrical arrangement is present in the 3E0S crystal lattice. The two homodimeric assemblies can be superimposed with a Cα‐RMSD of 4.2 Å for a total of 490 aligned residues. In contrast, the two metal‐binding sites that we identified in PptA have no direct counterparts in 3E0S, perhaps reflecting the existence of multiple CHAD families that differ in their metal‐binding ability or specificity. Like other phosins, 3E0S is rich in His residues that may constitute alternative binding sites, while the absence of metal ions in the experimentally determined structure could be a consequence of the protein purification or crystallisation procedure.

Figure 6

Comparison of the PptA structure solved in the present study (6RN5, blue) and PDB entry 3E0S, PptB from Clorobium tepidum (yellow). (A) Stereo image of the superposed structures, shown as Cα traces. (B) Structure‐based alignment of the protein sequences. Metal‐binding His residues are shown in red, whereas sulfate‐binding and positively charged residues lining the tunnel surface are shown in blue.

The biological consequences of polyP binding by phosins remain unknown. PolyP is an ancient and ubiquitous biological polymer, thought to act as a reserve of both phosphate and energy during nutritional stress 46, 47. In addition, a role in signalling has been proposed 48. Earlier work has revealed that Ppk, the bacterial kinase responsible for polyP synthesis and the regeneration of ATP from ADP and polyP, plays a negative role in the regulation of antibiotic production by Streptomyces lividans in conditions of phosphate limitation 22, 23, 24. The fact that in streptomycetes PptA is encoded by a gene immediately downstream of ppk may point at related functions of the proteins in phosphate metabolism and, conceivably, in the regulation of antibiotic synthesis. Interestingly, recent microscopic studies revealed a striking co‐localisation of PptA and polyP granules in several prokaryotes 26. The direct, high‐affinity interaction between PptA and polyP that we describe in the present study readily explains these findings. In earlier studies, Ppk co‐localised in similar fashion with polyP granules 49, 50, consistent with the idea that the two proteins act in concert. One attractive possibility is that PptA assists Ppk in the mobilisation of polyP reserves at the onset of P limitation. Intriguingly, the gene immediately downstream of ppk in E. coli and most other bacteria does not encode PptA but a polyphosphatase, Ppx. However, we failed to detect any polyphosphatase activity in PptA. This could indicate that PptA acts as a cofactor or processivity factor, either to Ppk or to a true polyphosphatase. PptA forms stable dimers in solution, which implies that the protein could, in principle, bind two strands of polyP simultaneously and thereby physically link the polymer chains in a non‐covalent manner. A critical role of such a mechanism in the nucleation or growth of aggregates seems unlikely, given that polyP granules were also found in PptA knock‐out mutants 26. However, this observation does not rule out a qualitative impact of PptA binding on granule structure. PptA might, for instance, render polyP molecules at the surface of granules more accessible to Ppk or polyphosphatases.

Our observation that a phosin from S. chartreusis binds Cu2+ is intriguing, as this particular metal ion is seldom found in bacterial cytoplasmic proteins. In most cases, the metal either acts as a cofactor in enzymes or is sequestered by specialised proteins responsible for its storage and detoxification. In Streptomyces, considerable variations in the cytoplasmic Cu2+ concentration accompany development 51 and strongly influence germination, differentiation and antibiotic synthesis 51, 52, 53, 54. Consistent with these key regulatory roles, Cu2+ levels are precisely controlled 51 by means of a dedicated import and export machinery that includes a recently discovered chalkophore 55. It is conceivable that PptA acts as a sensor or modifier of the intracellular Cu2+ concentration, while simultaneously measuring or affecting polyphosphate availability. Thus, the protein might mediate cross‐talk between two critical nutrient‐sensing pathways, both of which have been linked to morphological development and antibiotic production. It is also possible that PptA is involved in stripping divalent metal ions from polyP, which is itself a strong chelator with known roles in metal storage 46. Such an activity might not only serve to increase free metal ion concentrations in the cytosol, but also to bring polyP into an extended form that may be more readily metabolised. Further studies will be required to investigate the effects of the interaction of PptA with metals and polyP in vivo, as well as to establish if the protein influences development and antibiotic production in Streptomyces.

Author contributions

SW conceived and supervised this study; SW, NHR and MG performed experiments and analysed data; MJV and WH provided reagents, facilities and helpful advice; SW wrote the manuscript and made revisions.