The SCO2102 Protein Harbouring a DnaA II Protein-Interaction Domain Is Essential for the SCO2103 Methylenetetrahydrofolate Reductase Positioning at Streptomyces Sporulating Hyphae, Enhancing DNA Replication during Sporulation

Streptomyces DNA replication starts with the DnaA binding to the origin of replication. Differently to most bacteria, cytokinesis only occurs during sporulation. Cytokinesis is modulated by the divisome, an orderly succession of proteins initiated by FtsZ. Here, we characterised SCO2102, a protein harbouring a DnaA II protein–protein interaction domain highly conserved in Streptomyces. The ΔSCO2102 knockout shows highly delayed sporulation. SCO2102-mCherry frequently co-localises with FtsZ-eGFP during sporulation and greatly reduces FtsZ-eGFP Z-ladder formation, suggesting a role of SCO2102 in sporulation. SCO2102 localises up-stream of SCO2103, a methylenetetrahydrofolate reductase involved in methionine and dTMP synthesis. SCO2102/SCO2103 expression is highly regulated, involving two promoters and a conditional transcription terminator. The ΔSCO2103 knockout shows reduced DNA synthesis and a non-sporulating phenotype. SCO2102-mCherry co-localises with SCO2103-eGFP during sporulation, and SCO2102 is essential for the SCO2103 positioning at sporulating hyphae, since SCO2103-eGFP fluorescent spots are absent in the ΔSCO2102 knockout. We propose a model in which SCO2102 positions SCO2103 in sporulating hyphae, facilitating nucleotide biosynthesis for chromosomal replication. To the best of our knowledge, SCO2102 is the first protein harbouring a DnaA II domain specifically found during sporulation, whereas SCO2103 is the first methylenetetrahydrofolate reductase found to be essential for Streptomyces sporulation.


Introduction
Streptomycetes are important biotechnological bacteria from which two thirds of the bioactive secondary metabolites used in clinics (mainly antibiotics, but also antitumourals, immunosuppressors, etc.) have been discovered [1,2]. They have complex developmental cycles that include programmed cell death, hyphae differentiation and sporulation [3,4]. One of the most intriguing processes in Streptomyces biology is cell division, since this bacterium grows in the form of multinucleated hyphae with sporadic septae that, at the end of the development, form individual spores (reviewed in [5]).
Bacterial chromosomal replication is led by the binding of the DnaA AAA + ATPase to the origin of replication (oriC). DnaA is a complex protein constituting four structural domains [6,7]: domain I participates in orisome formation, domain II modulates proteinprotein interaction, domain III participates in ATP binding, and domain IV controls DNA

SCO2102/2103 Genes Are Highly Conserved in Streptomyces and Expressed during Sporulation
SCO2102 and SCO2103 show an average amino acid similarity of 72.5 and 96.14%, respectively, among the S. griseus, S. avermitillis, S. lividans, S. clavuligerus and S. venezuelae model strains. SCO2102 harbours a putative transmembrane domain (100% average similarity in Streptomyces) at its amino end, and a DnaA domain (70.7% average similarity in Streptomyces) centred in the protein ( Figure 1a). As introduced above, the canonical S. coelicolor DnaA chromosomal replication-initiator protein (SCO3879) harbours four domains [6]. The SCO2102 dnaA domain shows a 46.6% similarity with the SCO3879 dnaA II domain involved in protein-protein interaction [7]. Interestingly, the S. coelicolor DnaA II domain presents an insertion of 160 amino acids not present in other bacteria [7] (Figure 1b); this additional region is also present in the SCO2102 protein (the SCO3879 region homologous to SCO2102 is highlighted in red in Figure 1b; the SCO2102 and SCO3879 amino acid sequence alignment is shown in Figure S1).  [7]). Amino acid numbering in the scheme correspond to E. coli DnaA [6]. The S. coelicolor DnaA (SCO3879) region homologue to the SCO2102 DnaA domain is highlighted in red. (c) Terminator/antiterminator conformations of the conditional terminator identified in the SCO2103 ORF (predicted using the online algorithm developed by Millman et al. [18]).
The SCO2103 protein harbours a conserved MTHFR domain (96.5% average similarity in Streptomyces) (Figure 1a). Interestingly, SCO2102/2103 synteny was maintained in all the above Streptomyces model strains. The SCO2102/2103 gene transcription is complex and modulated by at least two promoters. The S. lividans SCO2103 orthologue is transcribed from a leaderless mRNA (labelled as P1 in Figure 1a) [16]. Jeong et al. [19] described the existence of at least one other promoter located in the S. coelicolor SCO2103 ORF (P2 in Figure 1a). SCO2102/2103 transcription might be even more complex; our bioinformatic analysis using the online algorithm developed by Millman et al. [18] revealed the existence of two mutually exclusive putative terminator and antiterminator RNA structures just before P2 (Figure 1c).
Both genes, SCO2102 and SCO2103, are more highly expressed at the aerial mycelium sporulating hyphae (48 h in GYM) compared to at the substrate mycelium (16 and 24 [7]). Amino acid numbering in the scheme correspond to E. coli DnaA [6]. The S. coelicolor DnaA (SCO3879) region homologue to the SCO2102 DnaA domain is highlighted in red. (c) Terminator/antiterminator conformations of the conditional terminator identified in the SCO2103 ORF (predicted using the online algorithm developed by Millman et al. [18]).
The SCO2103 protein harbours a conserved MTHFR domain (96.5% average similarity in Streptomyces) (Figure 1a). Interestingly, SCO2102/2103 synteny was maintained in all the above Streptomyces model strains. The SCO2102/2103 gene transcription is complex and modulated by at least two promoters. The S. lividans SCO2103 orthologue is transcribed from a leaderless mRNA (labelled as P1 in Figure 1a) [16]. Jeong et al. [19] described the existence of at least one other promoter located in the S. coelicolor SCO2103 ORF (P2 in Figure 1a). SCO2102/2103 transcription might be even more complex; our bioinformatic analysis using the online algorithm developed by Millman et al. [18] revealed the existence of two mutually exclusive putative terminator and antiterminator RNA structures just before P2 (Figure 1c).
Both genes, SCO2102 and SCO2103, are more highly expressed at the aerial mycelium sporulating hyphae (48 h in GYM) compared to at the substrate mycelium (16 and 24 h in solid GYM cultures) ( Figure 2 [20] was used as control. Three biological replicates were processed.

∆SCO2103 Knockout Shows a Bald Phenotype
The ∆SCO2103 knockout mutant shows a bald phenotype, i.e., the absence of aerial mycelium (white colour in culture plates) and sporulation (grey colour in culture plates) in all culture media analysed (GYM, R5A, SFM, TBO) (compare the wild-type strain, Figure 3a-d, with the ∆SCO2103 knockout, Figure 3e-h). Spore chains were never observed in the ∆SCO2103 hyphae (see laser-scanning confocal fluorescence and phase-contrast images in Figure 3). Actinorhodin production (blue colour in Petri plates) seems to be higher in GYM, R5A, and SFM solid cultures of the ΔSCO2103 mutant than in the wild-type strain Figure 2. SCO2102/2103 gene expression in the S. coelicolor wild-type strain. SCO2102 and SCO2103 transcript abundance in GYM solid cultures. Asterisks indicate significant differences compared to the 16 h sample. Transcript abundance of the SCO4758 housekeeping gene [20] was used as control. Three biological replicates were processed.

∆SCO2103 Knockout Shows a Bald Phenotype
The ∆SCO2103 knockout mutant shows a bald phenotype, i.e., the absence of aerial mycelium (white colour in culture plates) and sporulation (grey colour in culture plates) in all culture media analysed (GYM, R5A, SFM, TBO) (compare the wild-type strain, Figure 3a-d, with the ∆SCO2103 knockout, Figure 3e-h). Spore chains were never observed in the ∆SCO2103 hyphae (see laser-scanning confocal fluorescence and phase-contrast images in Figure 3). Actinorhodin production (blue colour in Petri plates) seems to be higher in GYM, R5A, and SFM solid cultures of the ∆SCO2103 mutant than in the wild-type strain cultures, which might be a consequence of the absence of sporulation: wild-type sporulating hyphae would stop metabolism and antibiotic production, while ∆SCO2103 non-sporulating hyphae would produce antibiotics for a longer time. Since the S. lividans SCO2103 orthologue has a methylenetetrahydrofolate reductase involved in methionine synthesis [16], we investigated whether methionine might restore the S. coelicolor ∆SCO2103 knockout mutant phenotype. However, sporulation was not restored in all methionineamended media (Figure 3i-l). As expected, the ∆SCO2103 knockout mutant was unable to grow in minimal medium without methionine (Figure 3m), corroborating the finding that SCO2103 methylenetetrahydrofolate reductase activity participates in methionine biosynthesis [16].

SCO2102 and SCO2103 Gene Expression Is Highly Regulated, Involving at Least Two Promoters and a Conditional Transcriptional Terminator
In contrast to ∆SCO2103, the ∆SCO2102 knockout mutant can sporulate ( Figure 3n). As mentioned above, the expression of the SCO2102 and SCO2103 genes is controlled by at least two promoters (P1 and P2 in Figure 1a) [16,19]. Bioinformatics revealed the existence of a putative conditional transcriptional terminator between P1 and P2 (Figure 1c).

SCO2102 and SCO2103 Gene Expression Is Highly Regulated, Involving at Least Two Promoters and a Conditional Transcriptional Terminator
In contrast to ∆SCO2103, the ∆SCO2102 knockout mutant can sporulate ( Figure 3n). As mentioned above, the expression of the SCO2102 and SCO2103 genes is controlled by at least two promoters (P1 and P2 in Figure 1a) [16,19]. Bioinformatics revealed the existence of a putative conditional transcriptional terminator between P1 and P2 (Figure 1c). To further understand the interaction between SCO2102 and SCO2103 gene transcription, we quantified their mRNA abundances in the wild-type strain and the ∆SCO2102 and ∆SCO2103 knockout mutants growing on solid GYM 48 h cultures (Figure 4a), the timepoint showing the maximum SCO2102 and SCO2103 transcriptions ( Figure 2). We also quantified SCO2102 mRNA abundance in the ∆SCO2102 mutant complemented with SCO2102 and controlled by promoters 1 and 2 (∆SCO2102 (pNG3-p1-p2-SCO2102) strain) and with SCO2102 and P1, P2, as well as the putative conditional terminator (∆SCO2102 (pNG3-P1-Ter-P2-SCO2102)) ( Figure 4b). The identified conditional terminator is functional, as the expression of the SCO2102 gene in the ∆SCO2102 mutant complemented with SCO2102 under the control of the two promoters and the conditional terminator (∆SCO2102 (pNG3-P1-Ter-P2-SCO2102) strain) was 33% lower than in the construction without the putative conditional terminator (∆SCO2102 (pNG3-p1-p2-SCO2102) strain) (Figure 4b). Interestingly, SCO2102 expression in the ∆SCO2102 mutant complemented with P1 and P2-SCO2102 was around 10 times lower than that in the wild-type strain (3.8 × 10 -5 vs. 5.9 × 10 -4 , respectively, Figure 4). This result again indicates the existence of further unknown regulation that could not be reproduced in the ∆SCO2102 mutant complemented with P1 and P2-SCO2102. transcript abundances in the ∆SCO2102 knockout complemented with the SCO2102 gene controlled by P1 and P2 with and without the conditional transcriptional terminator. Asterisks indicate significant differences compared to the wild-type strain. Three biological replicates were processed.

∆SCO2103 Knockout Mutant Sporulation Is Restored by the SCO2103 Gene
As stated above, SCO2103 ORF removal also affects SCO2102 due to the elimination of one of the known promoters of SCO2102 (P2 in Figure 1a), considerably reducing transcript abundances in the ∆SCO2102 knockout complemented with the SCO2102 gene controlled by P1 and P2 with and without the conditional transcriptional terminator. Asterisks indicate significant differences compared to the wild-type strain. Three biological replicates were processed.
As expected, the ∆SCO2102 mutant lacked SCO2102 expression, whereas the ∆SCO2103 mutant lacked SCO2103 transcription ( Figure 4a). The ∆SCO2103 mutant had highly reduced SCO2102 expression (98% less than the wild-type strain), indicating the importance of P2 in the expression of this gene ( Figure 4a). Interestingly, SCO2103 expression was highly reduced in the ∆SCO2102 mutant (87% less than the wild-type strain) (Figure 4a), indicating that further unknown regulation, beyond the P1 and P2 promoters, exists.
The identified conditional terminator is functional, as the expression of the SCO2102 gene in the ∆SCO2102 mutant complemented with SCO2102 under the control of the two promoters and the conditional terminator (∆SCO2102 (pNG3-P1-Ter-P2-SCO2102) strain) was 33% lower than in the construction without the putative conditional terminator (∆SCO2102 (pNG3-p1-p2-SCO2102) strain) ( Figure 4b). Interestingly, SCO2102 expression in the ∆SCO2102 mutant complemented with P1 and P2-SCO2102 was around 10 times lower than that in the wild-type strain (3.8 × 10 -5 vs. 5.9 × 10 -4 , respectively, Figure 4). This result again indicates the existence of further unknown regulation that could not be reproduced in the ∆SCO2102 mutant complemented with P1 and P2-SCO2102.

∆SCO2103 Knockout Mutant Sporulation Is Restored by the SCO2103 Gene
As stated above, SCO2103 ORF removal also affects SCO2102 due to the elimination of one of the known promoters of SCO2102 (P2 in Figure 1a), considerably reducing SCO2102 expression ( Figure 4a). Four combinations of the SCO2102/2103 genes with the P1 and P2 promoters and the putative conditional transcriptional terminator were used in this study to complement the ∆SCO2102 and ∆SCO2103 knockout mutants (outlined in Figure 5). The SCO2103 + (pNG3-p1-SCO2103) strain (Construction 2 in Figure 5) restored wild-type sporulation at 72 h (compare Figure 5d,e with Figure 5l). However, when both SCO2102 and SCO2103 genes were used together (Construction 1 in Figure 5), sporulation was accelerated to 48 h (Figure 5i,j). The SCO2102 gene alone was not able to restore ∆SCO2103 sporulation (Figure 5m,n). These results indicate that SCO2103 is the gene responsible for the bald phenotype and that an increase in the dose of the SCO2102 gene in the ∆SCO2103 mutant complemented with SCO2102 and SCO2103 accelerates sporulation timing.

∆SCO2102 Knockout Shows a Delay in Sporulation, Which Is Complemented by the SCO2102 Gene
The ∆SCO2102 knockout mutant was complemented with SCO2102 under the control of one or two promoters (constructions 3 and 4 in Figure 5). ∆SCO2102 sporulation was delayed to 96 h in GYM ( Figure 5p) compared to the 72 h in the wild-type strain (Figure 5d,e). This delay in sporulation was complemented with SCO2102 controlled by P2 ( Figure 5r). Interestingly, when the two promoters were used in the complementation, the delay in sporulation was not complemented and it was even more delayed than in the ∆SCO2102 mutant, since after 96 h, culture spores were hardly observable ( Figure 5t). This result again indicates an intriguing dose-dependent effect of SCO2102 on sporulation.

Actinorhodin and Undecylprodigiosin Productions Were Altered in the ∆SCO2102 and ∆SCO2103 Knockout Mutants
Actinorhodin production was slightly reduced in the ∆SCO2103 knockout in liquid sucrose-free R5A non-sporulating cultures ( Figure 6a). This phenotype is complemented by the SCO2102 gene controlled by P1 and P2 (Construction 4 in Figure 5), but not by the other complementation constructions (Figure 6a). Undecylprodigiosin production was not significantly altered in the ∆SCO2103 knockout compared to the wild-type strain ( Figure 6c). Interestingly, some of the ∆SCO2103-complemented strains (constructions 2 and 4 in Figure 5) had increased undecylprodigiosin production (Figure 6c). wild-type sporulation at 72 h (compare Figure 5d,e with Figure 5l). However, when both SCO2102 and SCO2103 genes were used together (Construction 1 in Figure 5), sporulation was accelerated to 48 h (Figure 5i,j). The SCO2102 gene alone was not able to restore ∆SCO2103 sporulation (Figure 5m,n). These results indicate that SCO2103 is the gene responsible for the bald phenotype and that an increase in the dose of the SCO2102 gene in the ∆SCO2103 mutant complemented with SCO2102 and SCO2103 accelerates sporulation timing.  Actinorhodin and undecylprodigiosin production were considerably reduced in the ∆SCO2102 mutant (Figure 6b,d). Antibiotic production could not be restored to wild-type levels with the SCO2102 gene being controlled by one or two promoters (Figure 6b,d).
not significantly altered in the ∆SCO2103 knockout compared to the wild-type strain (Fig  ure 6c). Interestingly, some of the ∆SCO2103-complemented strains (constructions 2 and 4 in Figure 5) had increased undecylprodigiosin production (Figure 6c).
Actinorhodin and undecylprodigiosin production were considerably reduced in th ∆SCO2102 mutant (Figure 6b,d). Antibiotic production could not be restored to wild-typ levels with the SCO2102 gene being controlled by one or two promoters (Figure 6b,d). Figure 6. Antibiotic production in ∆SCO2102, ∆SCO2103 and their complemented strains. Antibioti production was measured in three biological replicates from liquid sucrose-free R5A medium at 16 h once the maximum production was reached. (a,b) Actinorhodin. (c,d) Undecylprodigiosin. Th constructions used to complement the ∆SCO2102/SCO2103 mutants are numbered as in Figure 5 Significant differences compared to the wild-type strain are labelled by asterisks. Figure 6. Antibiotic production in ∆SCO2102, ∆SCO2103 and their complemented strains. Antibiotic production was measured in three biological replicates from liquid sucrose-free R5A medium at 168 h once the maximum production was reached. (a,b) Actinorhodin. (c,d) Undecylprodigiosin. The constructions used to complement the ∆SCO2102/SCO2103 mutants are numbered as in Figure 5. Significant differences compared to the wild-type strain are labelled by asterisks.

SCO2102 Co-Localises with FtsZ during Sporulation
To determine whether SCO2102 co-localises with FtsZ, we created a SCO2102-mCherry gene that was introduced into the S. coelicolor FM145 strain harbouring ftsZ-eGFP created by Willemse and Wezel [21]. We followed SCO2102-mCherry (red) and FtsZ-eGFP (green) during development under a fluorescence microscope (Figure 7). At early time points (16 h), FtsZ-eGFP spots were observed inside some hyphae (arrowheads in Figure 7a), but there was no trace of mCherry fluorescence. At 48 h, when DNA replication associated with sporulation starts in S. coelicolor GYM-confluent cultures [22], some hyphae showed eGFP-FtsZ-or SCO2102-mCherry-fluorescent spots (arrowheads in Figure 7c,f). There were also hyphae in which green and red spots co-localised (arrows in Figure 7i,k), indicating that FtsZ and SCO2102 can co-localise during part of the complex succession of proteins modulating sporulation. Under the culture conditions, developmental time points and the fluorescence microscope settings detailed in the methods, there was no detectable autofluorescence in the S. coelicolor wild-type control cultures ( Figure S2). Although SCO2102-mCherry and eGFP-FtsZ fluorescence was not abundant, it was unequivocally above the fluorescence of the controls observed under the same conditions.
The FtsZ ladders (the succession of 1 µm-separated Z-rings) were easily observed in the S. coelicolor FM145 strain expressing FtsZ-eGFP (Figure 7p-r), but we never observed them in sporulating hyphae of the S. coelicolor strain expressing FtsZ-eGFP and SCO2102-mCherry (Figure 7j). This result indicates some interference between SCO2102-mCherry and FtsZ-eGFP which precludes, or at least diminishes, FtsZ-eGFP polymerisation.
The FtsZ ladders (the succession of 1 µm-separated Z-rings) were easi the S. coelicolor FM145 strain expressing FtsZ-eGFP (Figure 7p-r), but we n them in sporulating hyphae of the S. coelicolor strain expressing FtsZ-eGFP a mCherry (Figure 7j). This result indicates some interference between SCO2 and FtsZ-eGFP which precludes, or at least diminishes, FtsZ-eGFP polymer

SCO2102 and SCO2103 Co-Localise during Sporulation
Subsequently, we analysed whether SCO2102 and SCO2103 co-localise, investigating the dynamics of SCO2102-mCherry and SCO2103-eGFP in S. coelicolor (Figure 8a-f). Interestingly, all SCO2102-mCherry spots observed at the sporulating hyphae (48 h) co-localised with SCO2103-eGFP (arrows in Figure 8d,e). In contrast, there were several SCO2103-eGFP spots that did not co-localise with SCO2102-mCherry (Figure 8e). The SCO2103-eGFP spots were observed in the wild-type strain (Figure 8g,h), but they were never observed in the ∆SCO2102 knockout mutant expressing SCO2103-eGFP (Figure 8i). These results reveal that SCO2102 is essential for the SCO2103 positioning at sporulating hyphae. The wild-type control strain observed under the same fluorescence microscope settings did not show observable fluorescence ( Figure S2).

SCO2102 and SCO2103 Co-Localise during Sporulation
Subsequently, we analysed whether SCO2102 and SCO2103 co-localise, investigating the dynamics of SCO2102-mCherry and SCO2103-eGFP in S. coelicolor (Figure 8a-f). Interestingly, all SCO2102-mCherry spots observed at the sporulating hyphae (48 h) co-localised with SCO2103-eGFP (arrows in Figure 8d,e). In contrast, there were several SCO2103-eGFP spots that did not co-localise with SCO2102-mCherry (Figure 8e). The SCO2103-eGFP spots were observed in the wild-type strain (Figure 8g,h), but they were never observed in the ∆SCO2102 knockout mutant expressing SCO2103-eGFP (Figure 8i). These results reveal that SCO2102 is essential for the SCO2103 positioning at sporulating hyphae. The wild-type control strain observed under the same fluorescence microscope settings did not show observable fluorescence ( Figure S2). (a-f) SCO2103-eGFP (green) and SCO2102-mCherry (red) cellular localisation in the S. coelicolor wildtype strain. All SCO2102-mCherry spots co-localise with SCO2103-eGFP spots (arrows), but there are SCO2103-eGFP spots that do not co-localise with SCO2102-mCherry spots. (g-l) SCO2103-eGFP expressed in the wild-type and the ∆SCO2102 knockout strains. Fluorescence and phase-contrast microscope images are shown. Scale bars 5 µm.
2.9. Chromosomal DNA Is Reduced in the Hyphae of the ∆SCO2103 Knockout Compared to the ∆SCO2102 Knockout and the Wild-Type Strains SCO2103 methylenetetrahydrofolate reductase participates in methionine biosynthesis [16], which was corroborated in this study (Figure 3m). Methylenetetrahydrofolate reductases also participate in other biosynthetic pathways, such as dTTP biosynthesis [17], which is necessary for DNA replication. We tested whether SCO2103 participates in DNA biosynthesis during sporulation, comparing the amounts of chromosomal DNA in the SCO2102 and SCO2103 knockout mutants and the wild-type strain (Figure 9). The amount of chromosomal DNA was highly reduced in the ∆SCO2103 mutant hyphae compared to the wild-type strain (19.4% reduction), indicating a role of SCO2103 in DNA synthesis, necessary for sporulation. Chromosomal DNA was also reduced in the ∆SCO2102 mutant; however, the difference with the wild-type strain was not significant (Figure 9). A reduction in chromosomal DNA in the ∆SCO2102 mutant would correlate with the reduction observed in the SCO2103 expression in this mutant (see above and Figure 4a). ductases also participate in other biosynthetic pathways, such as dTTP biosynthe which is necessary for DNA replication. We tested whether SCO2103 participates biosynthesis during sporulation, comparing the amounts of chromosomal DNA SCO2102 and SCO2103 knockout mutants and the wild-type strain (Figure 9). The of chromosomal DNA was highly reduced in the ΔSCO2103 mutant hyphae comp the wild-type strain (19.4% reduction), indicating a role of SCO2103 in DNA sy necessary for sporulation. Chromosomal DNA was also reduced in the ΔSCO2102 however, the difference with the wild-type strain was not significant (Figure 9). A tion in chromosomal DNA in the ΔSCO2102 mutant would correlate with the re observed in the SCO2103 expression in this mutant (see above and Figure 4a). Figure 9. Chromosomal DNA at the wild-type sporulation time points in GYM cultures of type strain and the ∆SCO2102/2103 knockout mutants. Significant difference compared to t type strain is labelled by an asterisk.

Discussion
The expression of the SCO2103-SCO2102 genes is modulated by at least two p ers identified by Blanco et al. [16] and Jeong et al. [19] (P1 and P2 in Figure 1a). H identified a conditional terminator upstream of P2 and inside the SCO2103 ORF 1c), which reduces SCO2102 expression when present (Figure 4b). The antitermina formation would be necessary for SCO2103 ORF expression. Conditional trans termination is a strategy of cis-acting RNA-based regulation consisting of two m exclusive terminator and antiterminator RNA structures, that depending on a spec nal (small ligands, depletion of amino acids, antibiotics), can activate premature tr tion termination [18]. Further work is necessary to characterise the signal/s mod the conditional terminator/antiterminator identified in the SCO2103 ORF. A SCO2102 is downstream to SCO2103, SCO2103 expression was highly reduced ∆SCO2102 mutant (87% less compared to the wild-type strain) (Figure 4a), indicat additional unknown regulation, beyond the P1 and P2 promoters and the conditio minator, exists. This complex regulation modulates the SCO2102 and SCO2103 tr dose producing a delay (Figure 5s,t) or acceleration (Figure 5i) of the sporulation Figure 9. Chromosomal DNA at the wild-type sporulation time points in GYM cultures of the wild-type strain and the ∆SCO2102/2103 knockout mutants. Significant difference compared to the wild-type strain is labelled by an asterisk.

Discussion
The expression of the SCO2103-SCO2102 genes is modulated by at least two promoters identified by Blanco et al. [16] and Jeong et al. [19] (P1 and P2 in Figure 1a). Here, we identified a conditional terminator upstream of P2 and inside the SCO2103 ORF (Figure 1c), which reduces SCO2102 expression when present (Figure 4b). The antiterminator conformation would be necessary for SCO2103 ORF expression. Conditional transcription termination is a strategy of cis-acting RNA-based regulation consisting of two mutually exclusive terminator and antiterminator RNA structures, that depending on a specific signal (small ligands, depletion of amino acids, antibiotics), can activate premature transcription termination [18]. Further work is necessary to characterise the signal/s modulating the conditional terminator/antiterminator identified in the SCO2103 ORF. Although SCO2102 is downstream to SCO2103, SCO2103 expression was highly reduced in the ∆SCO2102 mutant (87% less compared to the wild-type strain) (Figure 4a), indicating that additional unknown regulation, beyond the P1 and P2 promoters and the conditional terminator, exists. This complex regulation modulates the SCO2102 and SCO2103 transcript dose producing a delay (Figure 5s,t) or acceleration (Figure 5i) of the sporulation in our ∆SCO2102and ∆SCO2103complemented strains. Interestingly, we were unable to create the SCO2102/2103 double mutant, indicating that the combined activity of the SCO2102 and SCO2103 proteins might be essential for Streptomyces viability. We were also unable to overexpress SCO2102, again suggesting that the SCO2102 transcript dose might determine development and viability.
In our study, SCO2102 frequently co-localised with FtsZ (SCO2082) during sporulation ( Figure 6); FtsZ is the first component of the vegetative and sporulation divisomes, i.e., the proteins that control cell envelope invagination during cytokinesis (reviewed in [9]). We were unable to determine the precise FtsZ and SCO2102 kinetics and we do not know if FtsZ starts first or continues after SCO2102 accumulation in sporulating hyphae. However, our results indicate that SCO2102 participates, together with FtsZ, in the complex succession of proteins regulating Streptomyces sporulation [9]. The SCO2102 always co-localised with SCO2103 during sporulation (Figure 8d,e), and positions the SCO2103 methylenete-trahydrofolate reductase at sporulating hyphae, since SCO2103-eGFP fluorescent spots are absent in the ∆SCO2102 knockout (Figure 8i). Methylenetetrahydrofolate reductases participate in the biosynthesis of dTTP [17], and SCO2103 might be crucial to provide the dTTP necessary for DNA biosynthesis accompanying sporulating hyphae. The amount of chromosomal DNA was, in fact, significantly lower in the ∆SCO2103 non-sporulating mutant (Figure 9).
Actinorhodin and undecylprodigiosin productions were affected in the ∆SCO2102/03 knockout mutants (Figure 6). This effect on antibiotic production might be consequence of the reduction in SCO2103 MTHFR-dependant methionine synthesis, since methionine is important for the synthesis of many antibiotics (e.g., via SAM), including acinorhodin [23]. Moreover, methionine visually impacts undecylprodigiosin production (red colour) in the ∆SCO2103 knockout mutant growing on TBO medium (Figure 3l). The effect on antibiotic production might also be indirect, due to the effect on sporulation timing, since hypha differentiation, sporulation and antibiotic production are highly interconnected [24,25].
Overall, the main finding of this study is that SCO2102 (harbouring a DnaA II proteininteraction domain) is essential for the positioning of SCO2103 methylenetetrahydro-folate reductase at sporulating hyphae, and that SCO2103 enhances DNA replication during sporulation. Based on these results, we propose a model in which SCO2102 positions SCO2103 at sporulating hyphae, facilitating the nucleotide biosynthesis necessary for chromosomal DNA replication ( Figure 10). The absence of SCO2102 in the ∆SCO2102 mutant precludes SCO2103 localisation at sporulating hyphae, probably limiting dTMP synthesis [17] and delaying sporulation. The ∆SCO2103 mutant lacks SCO2103 MTHFR activity and displays a strong reduction in DNA synthesis, which would block sporulation. Further work will be necessary to fully understand the interaction between these proteins during sporulation. We can conclude that SCO2102 and SCO2103 cooperate in the modulation of differentiation and sporulation. Normal sporulation requires correct levels of SCO2102 and SCO2103 transcription. To the best of our knowledge, SCO2102 is the first protein harbouring a DnaA II domain specifically found during sporulation, whereas SCO2103 is the first methylenetetrahydrofolate reductase found to be essential for Streptomyces sporulation.  Figure 10. Model outlining the dynamics of FtsZ (green), SCO2102 (red) and SCO2103 (orange) in sporulating hyphae. SCO2102 and FtsZ co-localise in the wild-type sporulating hyphae. SCO2102 is essential for positioning SCO2103 at sporulating hyphae, increasing MTHFR activity and nucleotide synthesis (dTMP), enhancing chromosomal DNA replication accompanying sporulation. SCO2103 is not positioned at sporulating hyphae in the ∆SCO2102 mutant, reducing dTMP accessibility and delaying sporulation. The absence of SCO2103 MTHFR activity in the ∆SCO2103 mutant would block the DNA synthesis enhanced by SCO2103, generating a non-sporulating phenotype.

Bacterial Strains and Culture Conditions
All Streptomyces and Escherichia coli strains used in this work are listed in Table 1. Spores were harvested from SFM solid plates [26] after growth at 30 °C for 7 days. In the bald strains, mycelium was collected from SFM cultures at 7 days. Differentiation analyses Figure 10. Model outlining the dynamics of FtsZ (green), SCO2102 (red) and SCO2103 (orange) in sporulating hyphae. SCO2102 and FtsZ co-localise in the wild-type sporulating hyphae. SCO2102 is essential for positioning SCO2103 at sporulating hyphae, increasing MTHFR activity and nucleotide synthesis (dTMP), enhancing chromosomal DNA replication accompanying sporulation. SCO2103 is not positioned at sporulating hyphae in the ∆SCO2102 mutant, reducing dTMP accessibility and delaying sporulation. The absence of SCO2103 MTHFR activity in the ∆SCO2103 mutant would block the DNA synthesis enhanced by SCO2103, generating a non-sporulating phenotype.

Bacterial Strains and Culture Conditions
All Streptomyces and Escherichia coli strains used in this work are listed in Table 1. Spores were harvested from SFM solid plates [26] after growth at 30 • C for 7 days. In the bald strains, mycelium was collected from SFM cultures at 7 days. Differentiation analyses were carried out on GYM [27] (5 g/L glucose, 4 g/L yeast extract, 5 g/L malt extract, 0.5 g/L MgSO 4 .7H 2 O, 20 g/L agar, after autoclaving supplemented with sterile 0.5 g/L K 2 HPO 4 ) plates covered with cellophane, inoculated with 10 7 spores from a freshwater suspension and cultured at 30 • C. In the bald strains, an aliquot of the mycelium collected from 7-dayold SFM plates was used to quantify the protein per mL, and plates were inoculated with mycelium equivalent to 26.3 mg protein. Samples for quantification of actinorhodin and undecylprodigiosin production were obtained from 20 mL sucrose-free R5A [28] cultures grown at 30 • C and 200 rpm in 100 mL flasks inoculated with 10 7 spores/mL, or with mycelium equivalent to 26.3 mg protein/mL in the bald strains. The bald phenotype was analysed in TBO [29], SFM and MM supplemented with 20% mannitol [26]. When indicated, plates were amended with 50 µg/mL methionine. with the primers 2103 leftF2103/2103RighR. The PCR product was cloned and sequenced in pCR TM -Blunt II-TOPO, followed by sequencing with primers M13F and M13R to confirm the absence of mutations. The insert was released with StuI and cloned into pCRISPR-Cas9 harbouring the 20 nt target sequence digested with StuI. The final vector pCRISPR-2103 was introduced into the S. coelicolor wild-type strain by conjugation. The conjugants harbouring the plasmid were selected using apramycin resistance, plated in SFM and grown at 37ºC for 3 days for plasmid clearance. Mutations were confirmed by PCR using the primers CAS91F and CAS91R (Table 1).

SCO2102 Mutagenesis
The SCO2102 knockout (∆SCO2102 mutant) was created following the workflow described above for the ∆SCO2102 mutant: the 20 nt target sequence was selected inside the SCO2102 gene and amplified by PCR with the primers sg2102F and sgRNA-R. The regions surrounding SCO2102 were amplified by PCR with 2102 leftF/2102 LeftR and 2102 RightF/ 2102 RightR primers, and combined by overlap-extension PCR [36] with the primers 2102 leftF/ 2102 RightR. Fragments were cloned to create the pCRISPR-2102 plasmid, which was introduced into the S. coelicolor wild-type strain by conjugation; conjugants were selected, the plasmid was eliminated by growing the strain at 37 • C, and the mutations were confirmed by PCR using the primers CAS91F and CAS91R (Table 1).
Synthesis of the p 1 p 2 SCO2102 gene was ordered to BGI Genomics (Hong Kong). This synthetic gene had the SpeI recognition sequence, 294 bps including p1 (from S. coelicolor chromosomal position 2,262,210 to 2,261,917), 236 bps including p2 (from S. coelicolor chromosomal position 2,261,172 to 2,260,937), the SCO2102 ORF plus 63 bps (from S. coelicolor chromosomal position 2,260,936 to 2,259,881), and the EcoRV recognition sequence. The synthesised gene was cloned into pNG3 [32] between EcoRV and SpeI.
The p 1 -conditional terminator p 2 SCO2102 was constructed as follows. A DNA fragment containing the conditional terminator and p2-SCO2102 was amplified using primers TerF/R and cloned into pCR TM -Blunt II-TOPO. The conditional terminator was released from pCR TM -Blunt II-TOPO using EcoRV and XbaI and cloned into pKMV harbouring p1 digested with the same enzymes (pKMV-p1 was synthetised by BGI Genomics, Hong Kong) ( Figure S3). The p 1 -conditional terminator p 2 SCO2102 was cloned into pNG3 at the SpeI and EcoRV cloning sites.
The pNG3 plasmids harbouring the genes used for complementation were introduced into the ∆SCO2103 bald mutant by protoplast transformation, as described in Kieser et al. [26]. Regarding the ∆SCO2102 mutant, conjugation of the spores using E. coli ET12567/pUZ8002 as a donor strain was used to introduce the pNG3 plasmids following the protocol described by Kieser et al. [26]. The integration of these plasmids into the ΦBT1 integration site (gene SCO4848) [37] was verified by PCR using primers SCO4848F and pMS82R.

DNA and RNA Extraction
Genomic DNA isolation was performed following standard methods [26]. RNA was isolated using Direct-zol™ RNA columns (Zymo-Spin™, Irvine, CA, USA) and treated with DNase I (Qiagen, Germany). The quantity and integrity of the RNA samples were measured with Nanodrop 2000 (Thermo Scientific, Waltham, MA, USA) and 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA).

Real-Time Quantitative Reverse-Transcription PCR (qRT-PCR)
A High-Capacity cDNA Reverse-Transcription Kit (Applied Biosystems, Waltham, MA, USA) was used to synthetise cDNA from 0.5 µg of RNA from three biological replicates. Real-time PCRs were carried out on a fluorescent quantitative-detection system FQD-96A (BIOER). The reactions were performed in triplicate, using 2 µL of two-fold diluted cDNA, 10 µL of PowerTrack™ SYBR Green Master Mix (ThermoFisher, Bremen, Germany), and 300 nM of specific primers (listed in Table 1) in a final volume of 20 µL. SCO2102 and SCO2103 were amplified with q2102F/R and q2103F/R primers, respectively; SCO4758 (amplified using primers q4758F/R) was used as a control since its expression is constitutive during development [20]. DNA contamination and primer dimer amplification were tested in negative controls by replacing cDNA with RNA or water. The amplification conditions were as follows: 2 min at 50 • C, 10 min at 95 • C, 40 repetitions of 15 s at 95 • C and 1 min at 60 • C.
Primer efficiencies were measured using serial dilutions of genomic DNA as a template and differed between the two primer pairs used. Consequently, we performed absolute quantification generating standard quantification curves using PCR-amplified SCO2102/03 and SCO4758 DNA fragments as templates for each primer pair [38]. All the standard curves used had a linear correlation coefficient greater than 0.99. Three biological replicates were processed. Statistical significance was determined by comparing transcript abundances using a t-test.

Antibiotic Quantification
Undecylprodigiosin and actinorhodin were quantified spectrophotometrically, according to Tsao et al. [39] and Bystrykh et al. [40]. For actinorhodin quantification, KOH was added to the culture samples at a final concentration of 1N. Cellular pellets were discarded by centrifugation, and the actinorhodin concentration was spectrophotometrically determined at 640 nm, applying the linear Beer-Lambert relationship (ε 640 = 25,320). The culture samples for undecylprodigiosin quantification were lyophilised, resuspended in methanol, acidified with 0.5N HCl and spectrophotometrically assayed at 530 nm, using the Beer-Lambert relationship to estimate the concentrations (ε 530 = 100,500). Three biological replicates were processed. Antibiotic concentrations were normalised by total protein. Three biological replicates were analysed. The reliability of the differences in antibiotic production compared to the wild-type strain was analysed using Student's t-tests. Differences were considered significant if the p value was equal to or less than 0.05 (asterisks in Figures 1d and 4m-r).

Protein Quantification
Protein concentration was quantified by the Bradford [41] method using bovine serum albumin standard (Sigma Aldrich, Burlington, MA, USA). Total protein extracts were obtained by mixing a volume of culture with a volume of 1 M NaOH, boiling for 5 min and removing cell debris by centrifugation at 7740× g.

S. coelicolor SCO2102 mCherry/eGFP and SCO2103 eGFP Constructions
We designed the SCO2102 ORF including the P2 promoter ( Figure 1a) together with the SCO2102 and the mCherry ORFs ( Figure S3). In addition, we designed an eGFP ORF ( Figure S3). The eGFP and mCherry codon usages were manually optimised to Streptomyces, using the Streptomyces codon usage table from 100 Streptomyces genes reported by Kieser et al. [26]. These genes were synthesised by BGI Genomics (Hong Kong), who cloned them into pMV. P2-SCO2102-mCherry was released from pMV using SpeI and EcoRV and cloned into pNG3 [32] digested with the same enzymes. The pNG3 harbouring P2-SCO2102-eGFP was created by releasing mCherry using XhoI and EcoRV and replacing it with eGFP released from pMV with the same enzymes.
The above pRAS-and pNG3-integrative plasmids were introduced to the S. coelicolor wild-type or ∆SCO2102 strains via conjugation [26].

Viability Staining, mCherry and eGFP Visualisation
Culture samples were obtained and processed to analyse cell viability and sporulation at various incubation time points, as previously described [43]. Cell viability and sporulation were analysed by staining the cells with propidium iodide and SYTO 9 (LIVE/DEAD Bac-Light Bacterial Viability Kit, Invitrogen, L-13152, Waltham, MA, USA). The samples were observed under a Leica TCS-SP8 confocal laser-scanning microscope with wavelengths of 488 and 568 nm excitations and 530 nm (green) or 640 nm (red) emissions [43] for SYTO 9/PI staining.
SCO2102-mCherry and SCO2103-eGFP were analysed in SFM plates without cellophane. After spore inoculation, cover glasses were positioned at an angle of 45 degrees. At the indicated time-points, the cover glasses were removed and the samples were mounted with ultrapure mQ water. Then, they were observed using a Leica DMRXA fluorescence microscope with FITC and Texas Red filters. Pictures were taken with an ORCA-Flash4.0 V3 Digital CMOS camera.
Culture controls without SYTO 9/PI staining and without eGFP or mCherry were used to fix the levels at which autofluorescence was detected.
Microscopy images were processed (histogram intensity levels were adjusted, and the scale was added) using Fiji software [44]. Figure composites were made using AdobePhotoshop CS5.1.

Quantification of Chromosomal DNA
Hyphae were resuspended in Tris-HCl 20 mM, pH 8, EDTA 1mM, complete mini EDTA-free (Roche) protease inhibitor cocktail, and broken via sonication (MSE soniprep 150) in an ice-bath for 6 cycles comprising 10 s of power and a 60-second break. Cellular debris was removed by centrifugation (10,000g for 10 min). Chromosomal DNA was quantified using the Burton assay, hydrolysing DNA with perchloric acid and quantifying deoxyribose [45,46]. The DNA standard curves were created using a 1 mg/mL calf thymus DNA (Sigma Aldrich) stock solution. The DNA concentration was normalised relative to the total protein measured by Bradford [41], and samples were analysed in triplicate; statistical significance was estimated using a t-test.