Streptomyces coelicolor Genes ftsL and divIC Play a Role in Cell Division but Are Dispensable for Colony Formation

ABSTRACT We have characterized homologues of the bacterial cell division genes ftsL and divIC in the gram-positive mycelial bacterium Streptomyces coelicolor A3(2). We show by deletion-insertion mutations that ftsL and divIC are dispensable for growth and viability in S. coelicolor. When mutant strains were grown on a conventional rich medium (R2YE, containing high sucrose), inactivation of either ftsL or divIC resulted in the formation of aerial hyphae with partially constricted division sites but no clear separation of prespore compartments. Surprisingly, this phenotype was largely suppressed when strains were grown on minimal medium or sucrose-free R2YE, where division sites in many aerial hyphae had finished constricting and chains of spores were evident. Thus, osmolarity appears to affect the severity of the division defect. Furthermore, double mutant strains deleted for both ftsL and divIC are viable and have medium-dependent phenotypes similar to that of the single mutant strains, suggesting that functions performed by FtsL and DivIC are not absolutely required for septation during growth and sporulation. Alternatively, another division protein may partially compensate for the loss of both FtsL and DivIC on minimal medium or sucrose-free R2YE. Finally, based on transmission electron microscopy observations, we propose that FtsL and DivIC are involved in coordinating symmetrical annular ingrowth of the invaginating septum.

Streptomyces coelicolor A3(2) is a gram-positive, mycelial soil bacterium that has a complex life cycle culminating in sporulation (15,68). On solid media, the vegetative mycelium is a dense network of branched multigenomic hyphae that are divided by occasional irregularly spaced cross-walls. As the colony matures, multigenomic aerial hyphae grow away from the substrate mycelium surface. Finally, each aerial hypha undergoes synchronous division at evenly spaced sites to form uninucleoid prespore compartments, which then develop into mature spores.
Cell division is essential for colony formation in unicellular rod-shaped bacteria, but not in S. coelicolor (for reviews, see references 11, 23, 49, and 50). Deletions of S. coelicolor ftsZ and ftsQ result in viable strains which are completely or largely blocked for septation, respectively (51,52). In this study, we have characterized ftsL and divIC (ftsB/ygbQ in Escherichia coli) homologues from the complete S. coelicolor genome sequence (6). Both S. coelicolor gene products are predicted to be bitopic membrane proteins, as they are in E. coli and Bacillus subtilis (13,19,32,40), possessing a short cytoplasmic N-terminal domain, a single membrane-spanning segment, and a coiled-coil region in the larger C-terminal domain that is predicted to be located outside the membrane. All identified homologues contain predicted coiled-coil regions in their Cterminal domains (13,25,46). The precise functions of FtsL and DivIC during cell division are unknown, although it has been suggested that FtsL stabilizes DivIC in the division complex (62). Only the external C-terminal domain is required for DivIC in B. subtilis (41), and a C-terminal truncation of 10 amino acids from FtsL results in loss of function (61). Both proteins localize to nascent division sites in E. coli and B. subtilis (13,26,40,62), and their localization is codependent. B. subtilis DivIC and E. coli FtsL are unstable in the absence of their partner (13,19). In B. subtilis, FtsL is highly unstable (18) and perhaps rate limiting for division because it is a substrate for a membrane metalloprotease (10). Proper mid-cell localization requires the membrane-spanning segment and the Cterminal domain for FtsL of E. coli (25) but only the C-terminal domain for DivIC of B. subtilis (41). Initial studies suggested that B. subtilis FtsL and DivIC interact with each other through their coiled-coil domains, forming heterodimers or higher-order oligomers (62). More recent evidence indicated that in both E. coli and Streptococcus pneumoniae, FtsL and FtsB (DivIC) are present in a complex with the division protein FtsQ (DivIB) (12,54) and that this complex can be formed before localization to the division site (12). Most recently, an FtsQ-independent interaction between FtsL and FtsB was detected in E. coli, using a novel method that prematurely targets proteins to the division site (27,28). Finally, interactions between these and other divisome proteins have been observed by using bacterial two-hybrid analyses (20,38). The demonstration of direct interactions between FtsL and FtsB, and also the formation of a subcomplex comprised of these two proteins with FtsQ (DivIB) in both gram-negative and gram-positive bacteria, indicates that protein assembly dynamics in this portion of the divisome assembly pathway are widely conserved among prokaryotes.
In this study we demonstrate that the cell division genes ftsL and divIC are dispensable for growth and viability but are required for proficient septation during growth and sporulation in S. coelicolor. Null mutations in these genes resulted in medium-dependent phenotypes with a more severe division defect when strains were grown on a rich osmotically enhanced medium used for routine growth of S. coelicolor. Furthermore, a double mutant strain displayed a division phenotype nearly indistinguishable from that of either single mutant. Finally, because of the mutant phenotypes observed by transmission electron microscopy, we propose that FtsL and DivIC are involved in coordinating symmetrical annular ingrowth of the invaginating septum. This is the first concrete suggestion of a function for these components of the divisome.

MATERIALS AND METHODS
Bacterial strains, plasmids, and media. The prototrophic, SCP1 Ϫ SCP2 Ϫ S. coelicolor A3(2) strain M145 was used as the parent strain for mutant isolation (6,43). A marker replacement (double) recombination strategy for division mutant isolation was similar to that previously used for the isolation of ftsZ and ftsQ null mutants (51,52). DU151 and DU152 (both ⌬divIC::aphI) (this study) were isolated by transforming M145 to neomycin resistance with pJA19 (described below) and screening for those sensitive to spectinomycin (6 spectinomycin-sensitive colonies were identified among the 40 transformants tested). DU190 and DU191 (both ⌬ftsL::hyg) (this study) were isolated by transforming M145 to hygromycin resistance with pJA25 (described below), screening for those sensitive to apramycin (2 apramycin-sensitive colonies were identified among the 32 transformants tested). The new division mutant strains were readily isolated on the first attempt. Double mutant strains DU219 and DU220 (both ⌬divIC::aphI ⌬ftsL::hyg) (this study) were readily isolated on the first attempt by transforming DU152 to hygromycin resistance with chromosomal DNA prepared from DU191 (50 to 60 transformants obtained). Plasmid-containing derivatives of the multiply auxotrophic plasmid-free S. coelicolor strain 2709 (43) were used as the donor strains for conjugal transfer of self-transmissible plasmids for genetic complementation experiments.
Standard procedures for conjugal transfer of plasmids, protoplast preparation, and transformation were used (43). Double-stranded plasmid or chromosomal DNA was alkaline denatured before transformation of S. coelicolor protoplasts to stimulate homologous recombination (55). Streptomyces strains were grown at 30°C. YEME (liquid), R2YE (agar), MS (agar), and minimal medium (MM) (agar) for growth of S. coelicolor were as described previously (43,52). Glucose or mannitol was added to 0.5% (wt/vol) in MM. Final concentrations of antibiotics used for Streptomyces were apramycin at 25 g ml Ϫ1 , thiostrepton at 50 g ml Ϫ1 , spectinomycin at 100 g ml Ϫ1 in MM and 200 g ml Ϫ1 in R2YE, hygromycin at 50 g ml Ϫ1 in MM and 200 g ml Ϫ1 in R2YE, and neomycin at 10 g ml Ϫ1 in R2YE. For experiments performed to identify the source of mediumdependent division phenotypes, certain individual components were omitted when preparing R2YE. The compounds individually tested by omission were sucrose, proline, CaCl 2 , yeast extract, and N-[Tris(hydroxymethyl)methyl]-2 aminoethane-sulfonic acid.
E. coli strains LL308 (70) and TG1 (58) were used for standard plasmid manipulation. To bypass the methyl-specific restriction system of S. coelicolor, the mutant dam dcm E. coli strain ER 2 -1 was used for preparation of unmodified plasmid DNA (5). E. coli strains were propagated in Luria-Bertani medium. Antibiotic concentrations used for E. coli were ampicillin at 100 g ml Ϫ1 and apramycin at 100 g ml Ϫ1 .
pBluescript II SK(ϩ) (Stratagene) and pNEB193 (New England Biolabs) were used as standard cloning vectors. PCR products were cloned into pCR2.1 (Invitrogen). pIJ2925 is a pUC-like plasmid with BglII sites flanking the multiple cloning site (37). pOJ260 is a pUC-like plasmid which contains the aac(3)IV gene in place of bla (8). pHP45⍀ was the source of the aadA omega fragment (56). pAH137 (A. Hausler) was the source of aphI (52). pOJ427 (B. Schoner) was the source of aac(3)IV. pKC1053 was the source of hyg (44). pJRM10 (51) is a bifunctional replicon consisting of the SCP2*-derived low-copy-number plasmid pIJ922 (43) fused to pBluescript II SK(ϩ). Cosmids E25 and C69 were used as sources of DNA containing divIC (SCO3095) and ftsL (SCO2091), respectively (57). pRK20 contains a PCR-amplified 520-bp insert from 451 nucleotides upstream until 67 nucleotides downstream of the translation start codon of yllC (mraW; SCO2092), the first gene in the division and cell wall cluster of S. coelicolor (R. Kuennen and J. R. McCormick, unpublished data). This fragment appears to contain a promoter for dcw (P dcw ) because it has been shown to be active in an enhanced green fluorescent protein gene (egfp) promoter probe analysis (J. A. Bennett and J. R. McCormick, unpublished data) using integrating vectors pIJ8660 and pIJ8668 (63). The position of a transcription start site(s) has not been mapped.
General DNA techniques. DNA restriction and modifying enzymes were used according to the manufacturer's recommendations. Chromosomal DNA was prepared from YEME-grown cultures according to the methods of Kieser et al. (43). Plasmid DNA was prepared from E. coli using the QIAprep spin kit (Qiagen). The QIAquick gel extraction kit was used to purify DNA fragments fractionated by agarose gel electrophoresis. Platinum Pfx DNA polymerase (Invitrogen) was used to amplify DNA by PCR. When required for cloning, nontemplated adenine overhangs were added to PCR-amplified products by treatment with Taq DNA polymerase (Promega). PCR-amplified products were cloned into pCR2.1 using the TOPO TA cloning kit (Invitrogen).
Construction of plasmids used to create ftsL-null and divIC-null mutants. We constructed deletion-insertion mutations in the S. coelicolor ftsL and divIC genes in E. coli. Plasmid pJA25 contains hyg inserted in place of a 344-bp AgeI-Ecl136II restriction fragment coding for the 5Ј end of the ftsL predicted reading frame, including the start codon, the coding region for the predicted membranespanning segment, and part of the coiled-coil domain (Fig. 1A). The details describing the construction of pJA25 follow. First, an 8.5-kb EcoRI fragment from cosmid C69 was cloned into the EcoRI site of pBluescript II SK(ϩ), resulting in pJR132. A 3-kb KpnI fragment (yllC-ftsL-ftsIЈ) from pJR132 was then cloned into pBluescript II SK(ϩ) to create pABC3 (A. B. Cadwallader and J. R. McCormick, unpublished data). A 1.56-kb Acc65I-Ecl136II fragment (ЈftsL-ftsIЈ) from pABC3 was inserted into pOJ260 digested with HindIII (filled in) and Acc65I, creating pJA21. A 1.2-kb EcoRI-AgeI fragment (yllC-ftsLЈ) from pABC3 was cloned into pBluescript II SK(ϩ) digested with EcoRI and XmaI to create In the shown segments of the chromosome, all the indicated genes are in the same orientation with respect to one another, reading by convention from left to right. The numbered boxes refer to the gene nomenclature used by the Sanger Centre S. coelicolor genome project (e.g., SCO3094). The locations of restriction sites used to construct the mutations or complementation plasmids are shown. A, AgeI; B, BclI, E, EcoRI; H, BamHI; K, KpnI; P, PstI; S, SacI. Horizontal bars positioned below the chromosomal regions correspond to DNA inserts in plasmids used for mutagenesis or genetic complementation experiments. The ftsL SC and divIC SC deletion-insertion mutations are indicated by dashed lines in the plasmids used for mutagenesis (pJA25 and pJA19, respectively). Plasmid pJA105 is similar to pJA89 but contains a 204-bp in-frame deletion in ftsL SC (indicated by an asterisk set off by parentheses) that removes the region coding for the predicted membrane-spanning segment and the coiled-coil region. The results of genetic complementation experiments (see Fig. 2, below) are summarized in the labeled column on the right. ϩ, complementation; ϩ/Ϫ, partial complementation; Ϫ, no complementation.

VOL. 189, 2007
ftsL AND divIC MUTANTS OF STREPTOMYCES 8983 pRA2. The DNA insert of pRA2 was excised with EcoRI and BamHI and ligated to pJA21 digested with the same enzymes to form pJA22. Finally, hyg was inserted in the opposite orientation relative to the cloned division and cell wall genes as a 2.08-kb EcoRV-BamHI fragment from pKC1053 into pJA22 digested with XbaI (filled in) and BamHI, creating pJA25. Plasmid pJA19 contains aphI in place of a 384-bp BamHI-KpnI fragment of the divIC coding region, an internal deletion that encompasses nearly the entire divIC gene (Fig. 1B). The details describing the construction of pJA19 follow. A 4.6-kb PstI-HindIII fragment from cosmid E25 was cloned into pOJ260 to create pJA15. A 1.67-kb KpnI fragment from pJA15 was ligated to pNEB193 to obtain pJA16. pJA17 was created by cloning a 2.2-kb PstI-BamHI fragment from pJA15 into pJA16. The omega fragment of pHP45⍀ containing the selectable marker aadA was cloned as a 2-kb HindIII fragment into pJA17 to create pJA18. The 1.05-kb aphI gene contained in pAH137 was excised with EcoRI and HindIII and cloned into pIJ2925, creating pRA4 with the aphI gene flanked by BglII sites. Finally, aphI was excised with BglII and inserted into BamHI-digested pJA18 to create pJA19. pJA19 has aphI inserted in the same orientation as divIC and flanking genes.
Southern blot hybridization analysis. The ftsL deletion-insertion mutation was verified by Southern blot hybridization analysis using chromosomal DNA double digested with KpnI and EcoRI. Probes used were a 1.2-kb EcoRI-AgeI restriction fragment corresponding to the region upstream of ftsL and a 1.6-kb Ecl136II-KpnI restriction fragment corresponding to the region downstream of ftsL. The probes hybridized to a 3.1-kb fragment for wild-type strain M145 and to 2.35-kb and 2.48-kb fragments for the S. coelicolor ftsL (ftsL SC )-null mutant strains (data not shown). The divIC deletion-insertion mutation was verified by Southern blot hybridization analysis using DNA double digested with NcoI and HindIII. Probes used were a 2.2-kb PstI-BamHI restriction fragment corresponding to the region upstream of divIC and a 1.7-kb KpnI restriction fragment corresponding to the region downstream of divIC. The probes hybridized to a 6.9-kb fragment for wild-type strain M145 and to 4.65-kb and 2.08-kb fragments for the divIC SC -null strains (data not shown). Hybridization reactions were performed at 65°C using buffer that contained 5% sodium dodecyl sulfate (67). Nylon membranes (Hybond-N; Amersham) were used as solid support, and probes were nonisotopically labeled before immunological detection (DIG DNA labeling and detection kit; Boehringer Mannheim).
Construction of plasmids used for genetic complementation experiments. We constructed bifunctional replicons derived from pJRM10 with various DNA inserts in E. coli and then introduced them into S. coelicolor (Fig. 1). The divIC gene was PCR amplified using pJA15 as template with primers divIC1 (5Ј-TCT AGACATCGAGGAGATCCTCGAC-3Ј) and divIC2 (5Ј-GAGCTCGTCTTCC TGTTGAGTCACTG-3Ј), which add an XbaI site 171 bp upstream of divIC and a SacI site 70 bp downstream, respectively (underlined), and cloned the reaction product into pCR2.1 to create pJA26. The SpeI site of pJA26 was then destroyed by treatment of SpeI-digested DNA with Klenow fragment of DNA polymerase I and ligase, creating pJA32. A 720-bp SacI fragment containing divIC was excised from pJA32 and cloned into pBluescript II SK(ϩ), resulting in pJA46. Finally, the pIJ922 portion of pJRM10 was inserted into plasmid pJA46 digested with HindIII and SpeI to form pJA57 (Fig. 1B).
The apramycin resistance gene was removed by FLP recombinase. The in-frame deletion was confirmed by DNA sequence analysis.
All plasmids used in complementation experiments were introduced into the S. coelicolor multiple auxotroph 2709 by transformation to thiostrepton resistance and then mated from the 2709 derivative into the desired prototrophic strains, selecting for thiostrepton resistance and prototrophy. Isolated transconjugants were tested for the presence of the antibiotic resistance gene marking the division gene mutation.
Light and electron microscopy. Methods for phase-contrast microscopy using coverslip lift slides were performed as described by McCormick and Losick (51). For transmission electron microscopy (TEM), two different whole colonies for each mutant were processed after growth on R2YE for 6 days as described by Kieser et al. (43), except that phosphate-buffered saline was substituted for sodium cacodylate buffer. Thin sections were viewed using a Hitachi 7100 microscope fitted with an Advanced Microscopy Techniques Corporation Advantage 10 charge-coupled-device camera with digital capture performed on a Macintosh computer using the public domain NIH Image program (developed at the U.S. National Institutes of Health and available on the internet at http://rsb .info.nih.gov/nih-image/). Final digital images were adjusted for brightness and contrast using Adobe Photoshop.
Computer analysis. Gene and predicted protein product sequences were obtained from the Streptomyces coelicolor genome database (ScoDB; http: //streptomyces.org.uk/S.coelicolor/index.html). The program PSI-BLAST (2) was used to identify the putative divIC homologue in S. coelicolor. The program DAS (17) was used to analyze the predicted gene products for potential membranespanning segments, and the programs COILS 2.2 (47), MULTICOIL (69), and PAIRCOIL (7) were used to analyze the predicted gene products for the potential to form coiled-coils.

Identification of ftsL and divIC homologues in S. coelicolor.
The S. coelicolor homologue of the ftsL gene of B. subtilis and E. coli, ftsL SC (assigned ScoDB gene number SCO2091; annotated cosmid clone gene designation SC4A10.24c), is located in the major division and cell wall (dcw) cluster between yllC (yabC, mraW) and ftsI as in other bacteria (Fig. 1A). The ftsL SC gene is predicted to encode a 206-amino-acid, 20.9-kDa protein. FtsL SC has 19% identical residues (13/68) to those representing the membrane-spanning segment and coiled-coil region of the COG4839 consensus sequence (representing 26 FtsL sequences of bacilli) (64). As noted for the B. subtilis homologue (21), FtsL SC has negligible primary sequence similarity to FtsL of E. coli, and primary amino acid sequence alignment to other FtsL proteins shows that few residues are completely conserved (data not shown). This observation is consistent with the fact that few specific residues are important for function in B. subtilis (61). Our assignment is based on a similar argument made for the B. subtilis homologue (21), that of gene position (synteny) in the major division and cell wall cluster, the predicted structure and topology of the product, and the mutant phenotype (see below).
Both FtsL SC and DivIC SC are predicted to be bitopic membrane proteins, each containing a single membrane-spanning segment and a larger C-terminal domain. The single transmembrane segments were predicted using DAS (circa residues 51 to 76 for FtsL SC and residues 57 to 76 for DivIC SC ). Potential coiled-coil regions were identified by the COILS program (circa residues 78 to 109 for FtsL SC and residues 76 to 115 for DivIC SC ); the MULTICOIL program predicts that both proteins are more likely to interact as a trimer than a dimer coiled-coil, and analysis with the PAIRCOIL program suggests that DivIC SC may interact with itself. As was previously pointed out for the homologues of B. subtilis (46), we noted that DivIC SC shows a small degree of sequence similarity to FtsL SC (data not shown).
ftsL SC and divIC SC mutants have medium-dependent phenotypes. To determine whether ftsL SC was functionally equivalent to ftsL of E. coli (32,65) and B. subtilis (19,21), we constructed deletion-insertion mutants in which a 344-bp fragment containing the 5Ј end of the ftsL SC open reading frame was replaced by a hygromycin resistance cassette (Fig. 1A). Two independently isolated strains were named DU190 and DU191. This result verified that ftsL SC was not essential for growth and viability in S. coelicolor. This result is not so surprising, given that similar results were previously obtained for ftsZ and ftsQ (51,52). DU190 and DU191 grew well, and colonies produced an abundant aerial mycelium on MM, R2YE, and MS media similar to that of the wild-type strain. The aerial mycelium was a paler gray relative to the wild-type M145 on MS medium, suggesting that there was a division defect within individual aerial hyphae or a defect in the developmental maturation of prespore compartments. For unknown reasons, but presumably related to their decreased or abolished vegetative crosswall formation, previously isolated ftsQ and ftsZ mutants produce copious amounts of the blue-pigmented antibiotic actinorhodin when grown on minimal media, resulting in colonies that are surrounded by blue halos (51,52). In contrast, ftsL SC -null mutant colonies were not surrounded by blue halos when grown on minimal medium. Presumably, this result is due to the fact that vegetative cross-wall formation is only modestly affected (see below).
Both DU190 and DU191 exhibited a similar cell division defect during sporulation as observed by phase-contrast microscopy. Strain DU191 was chosen for further analysis. After 6 days on MM containing glucose as a carbon source, twothirds of the DU191 aerial hyphae contained shallow constrictions at regularly spaced intervals, suggesting that these hyphae had not completed septation to form individual prespore compartments (similar to those shown in Fig. 2F). Approximately half of these constricted aerial hyphae were straight, while the others were coiled. The remaining one-third of the aerial hyphae contained mature spore chains with occasional lysed compartments (Fig. 2B), indicating that FtsL SC was not absolutely required for division. Similar results were observed for growth on MM containing mannitol as a carbon source (data not shown). Surprisingly, when DU191 was grown on the rich ). The mutants produce apparently normal spore chains on MM agar (in ϳ1/3 of the hyphae) and aerial hyphae containing evenly spaced constrictions which do not appear to have completed division on R2YE agar. Examples of evenly spaced constrictions in aerial hyphae of the mutants are indicated by arrows. Straight and coiled aerial hyphae were seen for the division mutants. Only examples of straight filaments are shown, because they better display the observed division phenotypes. For genetic complementation experiments, shown are phasecontrast micrographs demonstrating the division phenotypes of representative aerial hyphae of plasmid-containing strains. The ftsL-null mutant (DU191) containing pJRM10 (low-copy-number vector lacking an insert) has aerial hyphae with evenly spaced constrictions (I). pJA88 (yllC-ftsL SC ) partially complements the ftsL SC -null mutation (aerial hyphae with evenly spaced constrictions) (J; an infrequent spore chain is shown in the inset), while pJA89 (yllC-ftsL SC -ftsI) appears to complement the ftsL SC -null mutation fully (K). As anticipated, pJA105, containing an in-frame deletion of ftsL, does not complement the ftsL SC -null mutation (L). Also shown are representative aerial hyphae produced by the div-IC SC -null mutant (DU152) containing pJRM10 (vector control), which fails to complement the mutation (M), and pJA57 (divIC SC ), which appears to fully complement the divIC SC -null mutation (N).
To determine whether divIC SC was functionally equivalent to divIC of B. subtilis (46) and ftsB (ygbQ) of E. coli (13), we independently isolated two deletion-insertion mutants (DU151 and DU152) in which a 384-bp fragment internal to the div-IC SC coding region was replaced by aphI (Fig. 1B). Therefore, divIC SC was also dispensable for the growth and viability of S. coelicolor. DU151 and DU152 grew well and colonies produced an abundant aerial mycelium on MM, R2YE, and MS media. The aerial mycelium was a paler gray relative to the wild-type M145 on MS medium, suggesting that there may be a defect in spore formation or maturation. Similar to the ftsL SC -null mutants (see above), divIC SC -null mutant colonies were not surrounded by blue halos when grown on minimal medium.
Both DU151 and DU152 exhibited a similar cell division defect during sporulation, as assayed by phase-contrast microscopy. DU152 was chosen for further analysis. It displayed a medium-dependent division defect indistinguishable from that of the ftsL SC -null mutant strain when grown on minimal as well as rich media ( Fig. 2C and G, respectively). Approximately one-third of the aerial hyphae produced when grown on minimal medium contained mature spore chains with occasional lysed compartments (Fig. 2C), indicating that DivIC SC was not absolutely required for division.
The genetic neighborhood of divIC SC is different from that of ftsB (ygbQ) in E. coli (13) and divIC of B. subtilis (3,46). Upstream of divIC SC is gene SCO3096, coding for enolase (eno), an enzyme in glycolysis. (SCO7638 is a second gene encoding enolase [eno2].) Downstream of divIC SC is an open reading frame (SCO3094, E41.03c) that encodes a hypothetical membrane protein of unknown function (COG1507). The downstream open reading frame is in the same orientation as divIC SC , and the two genes are potentially part of an operon (Fig. 1B). We isolated strains containing an insertion mutation in SCO3094 (approximately two-thirds of the way from the initiation codon) to determine whether it might also be involved in cell division. The mutant strains produced slightly smaller colonies than that of the wild type, but they sporulated at a wild-type level (data not shown). Thus, the gene immediately downstream of divIC SC is most likely not involved in cell division.
Complementation experiments were performed to ensure that the cell division defects observed for the ftsL SC -null mutant (DU191) and divIC SC -null mutant (DU152) were imparted by the introduced deletion-insertion mutations. DNA inserts contained in the low-copy-number complementation plasmids are shown in Fig. 1. Plasmid pJA88 (P dcw -yllC-ftsL SC ) and plasmid pJA107 (P dcw -yllC-ftsL SC -ftsIЈ) partially restored sporulation to DU191 ( Fig. 2I and J, respectively). Thus, the cell division defect of DU191 was not the result of an unlinked mutation. In contrast, pJA89 (P dcw -yllC-ftsL SC -ftsI) substantially restored sporulation to DU191 (Fig. 2K). Finally, pJA105, which is identical to pJA89 except that it contains a 204-nucleotide in-frame deletion in ftsL (P dcw -yllC-⌬ftsL SC -ftsI), did not restore sporulation to DU191 (Fig. 2L). The results obtained from complementation experiments prove that the cell division defect was the result of the deletion of ftsL in DU191 with little or no contribution from a polar effect on the downstream cell division gene ftsI. Plasmid pJA57 containing additional copies of divIC SC failed to complement the ftsL SC mutation (data not shown). Therefore, modestly elevating the copy number of divIC SC did not compensate for the loss of ftsL SC .
When pJA57, which contains divIC SC as the only complete reading frame, was introduced into DU152, spore formation was dramatically increased to a level indistinguishable from that of the wild type (Fig. 2N). Therefore, the cell division defect of DU152 was the result of the deletion of divIC SC . Plasmid pJA89 containing additional copies of ftsL SC failed to complement the divIC SC mutation (data not shown). Therefore, modestly elevating the copy number of ftsL SC did not compensate for the loss of divIC SC .
The ftsL SC divIC SC double mutant has a phenotype indistinguishable from that of the single mutants. Previously, it was suggested that FtsL and DivIC may interact through their coiled-coil domains in B. subtilis to form heterodimers (62). Similarly, it has also been proposed that in E. coli the DivIC homologue FtsB (YgbQ) may interact with FtsL (13). Additional research has shown this to be the case. Perhaps the equivalent proteins are functionally redundant under certain conditions in S. coelicolor, explaining the ability of a single mutant to divide on minimal medium. We noted above that FtsL SC and DivIC SC have slight sequence similarities. To investigate the possibility of functional redundancy, double mutant strains were created by transformation using chromosomal DNA prepared from the ftsL SC mutant and the divIC SC mutant as recipient. Independently isolated double mutants were named DU219 and DU220. Retention of the divIC SC -null mutation was verified by the antibiotic resistance phenotype. The introduction of the ftsL SC -null allele for both strains was verified by Southern blot hybridization (see Materials and Methods), and DU220 was chosen for further analysis. Interestingly, the absence of both proteins did not result in a more severe division defect. The ftsL SC divIC SC double mutant strain had a phenotype indistinguishable from that of either single mutant on minimal and rich media, suggesting that these genes do not have redundant functions. The phenotype of the double mutant is more consistent with a requirement for an FtsL-DivIC interaction at a specific stage of septation. The double mutant was capable of producing approximately one-third of the aerial hyphae with distinctly separated prespore compartments on glucose MM (Fig. 2D). The remainder of the aerial hyphae contained evenly spaced constrictions. However, when grown on R2YE, the double null strain was largely blocked for sporulation, producing only aerial hyphae with evenly positioned constrictions about one normal spore length apart (Fig. 2H).
Septal morphology of ftsL SC and divIC SC mutants. We anticipated that the evenly spaced constrictions in aerial hyphae of division mutants resulted from a failure to complete septum invagination. However, we could not rule out the possibility that the sporulation septa were complete and the resulting prespore compartments had failed to metamorphose into separated spores. Therefore, TEM was employed to examine the morphologies of sporulation septa and vegetative cross-walls of the mutants grown on R2YE, the medium in which a more severe division phenotype was seen (Fig. 3).
Wild-type strain M145 produced aerial hyphae that were divided into distinct spore compartments by normal sporulation septa (Fig. 3D and G). Approximately 97% of the observed sporulation septa fell into this category ( Table 1). The remaining 3% of observed division sites appeared to be at an early stage with slight constrictions but no obvious invagination. These examples occurred adjacent to one another within the same hypha (i.e., the observed aerial hypha was undergoing synchronous septation). Consistent with observations made by phase-contrast microscopy, the aerial hyphae of the ftsL SC -null and divIC SC -null mutants were mainly blocked at an intermediate stage of sep-tation, as follows. Within most aerial hyphae, partial sporulation septa had begun to form but had not completely traversed the width of the filament (circa 92% of observed septa) ( Table  1). These partial septa fell into three main categories. First, circa 80% of the total observed division sites were only partially constricted, with little or no obvious ingrowth (Fig. 3E). Second, other partially formed sporulation septa had invaginated but from only one side of the division site (approximately 9% of the cases) and not at the other (Fig. 3F and I for the ftsL SC mutant and Fig. 3E and H for the divIC SC mutant). Third, partially formed septa with invagination from both sides were observed in 4% of the cases. We noted that when invagination from both sides was observed, one side frequently appeared to be longer than the other. Complete sporulation septa were also observed and fell into two categories. First, apparently normal sporulation septa that had completely traversed the filament were occasionally observed (approximately 6% of the observed septa, similar to those shown in Fig. 3G). Second, in the remaining 3% of the cases, aberrant sporulation septa that completely traversed the filament were evident (Fig.  3E). It is possible that some of these septa that appeared to be complete were actually tangential sections through incomplete septa. In contrast, cross-wall formation within vegetative hyphae was not as severely affected as sporulation septation in aerial hyphae (Table 1). Cross-walls observed by TEM fell into three classes. Approximately 51% of vegetative cross-walls in the mutants appeared morphologically normal (not shown but similar to the wild type in Fig. 3A). An additional 20% of the cross-walls completely traversed the hypha but were aberrant in appearance (i.e., thick, wavy, or not perpendicular to the long axis). (Again, it is possible that some of these cross-walls that appeared to be complete were actually tangential sections through incomplete septa.) In the remaining 29%, potentially arrested or slow ingrowth of cross-walls was observed in vegetative hyphae for both the divIC SC and the ftsL SC mutants ( Fig. 3B and C, respectively). For slightly more than one-third of this last class (and 11% of the total), ingrowth from only one side of the filament was observed (Fig. 3C). For the other two-thirds of this last class (18% of the total), ingrowth from both sides of the filament was observed (Fig. 3B). We noted that when invagination from both sides was observed, one side frequently appeared to be longer than the other (Fig. 3C). In contrast, 97% of the wild-type vegetative septa were complete (Table 1). Invagination from only one or both sides was observed in only 2% of the cases for the wild type (one example for each type).
Vegetative cross-wall formation in the double mutant strain was essentially identical to either single mutant (Table 1). Interestingly, sporulation septation was noticeably less impaired when both genes were inactivated. Two to three times as many septa appeared to be complete in a double mutant compared to either single mutant. An additional 14% of the septa traversed the hyphae and were aberrantly thick, resulting in spore-like compartments that failed to separate. The frequency observed for classes of normal and aberrantly com-pleted septa increased at the expense of the class blocked early in constriction.
Transmission electron microscopy also indicated that the frequency of division initiation in vegetative hyphae of the division mutants was not dramatically decreased compared with the wild-type strain (data not shown).
Osmolarity of the growth medium appears to affect the division phenotypes of ftsL SC and divIC SC mutants. Of the media tested, division phenotypes were most severe on R2YE, a conventional rich medium with high osmolarity. In addition to being used for standard strain propagation, this osmotically enhanced medium is used for protoplast regeneration following polyethylene glycol-mediated transformation. To determine the parameters that affect the division phenotypes of the ftsL SC and divIC SC mutants on different media, we dissected the R2YE medium effect by individually omitting components. We found that omitting the sucrose from R2YE (normally present at 0.3 M) resulted in a division phenotype for both mutants that closely resembled that of the same strains grown on MM (Fig. 4). Omitting several other individual components of R2YE (see the list in Materials and Methods) had no noticeable effect on the division phenotypes of the mutants (data not shown). Because sucrose is not utilized as a carbon source by S. coelicolor (33), we infer that osmolarity was the parameter affecting the severity of the division defect on different types of media.

DISCUSSION
In this study, we report the characterization of cell division genes ftsL SC and divIC SC (ftsB/ygbQ) in S. coelicolor A3(2), a sporulating mycelial bacterium. These genes are essential for colony formation of the model unicellular rod-shaped bacteria E. coli (13,32) and B. subtilis (21,46). Here, we have demonstrated that ftsL SC and divIC SC are dispensable for growth and viability in S. coelicolor, as has been previously shown for ftsZ and ftsQ (51,52). Furthermore, strains mutant for both "normally essential" genes were no more impaired for division and colony formation than either single mutant, suggesting that the genes may code for proteins involved in the same aspect of cell division. Nonetheless, ftsL SC and divIC SC are required for the a Division pattern scoring: complete, normal division; constricted, aerial filament indented without obvious ingrowth; two sides, ingrowth from both sides; one side, ingrowth from only one side; other, division appeared to be complete but septa were obviously aberrant in some way (i.e., thick, wavy, or not perpendicular to long axis of cell). NA, not applicable (because vegetative filaments do not constrict).
b Values in parentheses are the number of vegetative cross-walls or sporulation septa counted. efficient conversion of multigenomic aerial hyphae into chains of unigenomic spores ( Fig. 2 and 3). It was surprising to discover that the severity of the division phenotypes for the new S. coelicolor mutants were dependent on the medium, because the previously analyzed ftsZ-null and ftsQ-null mutants were the same on all media tested (51,52). We found that strains mutant for ftsL SC , divIC SC , or both genes can divide reasonably well, forming some chains of spores from aerial hyphae when growth occurs on a medium lacking sucrose ( Fig. 2B to D). Therefore, under certain physiological circumstances, reasonably proficient division can be accomplished in the complete absence of the functions usually supplied by the deleted genes.
Osmolarity of the medium appears to affect the severity of the division defect (Fig. 4). However, we anticipate that further dissecting the parameters affecting physiology and gene expression on osmotically enhanced media may prove to be complicated. This difficulty may be related to the fact that S. coelicolor has a large genome containing multiple sigma factors involved in the regulation of osmotic stress responses (66) and a plethora of uncharacterized transcriptional regulators (34). Several candidates (sigB, sigH, and osaB) for important genetic regulators of an osmotic response have been identified and studied (9,45,60). To further complicate matters, each of these regulators has additional roles in morphological differentiation and/or oxidative responses. Microarray analysis has helped to identify additional candidate genes likely to be involved in transient osmoregulation (39,45). Physiological regulators are likely to play a role as well. For example, a natural intracellular osmolyte has been shown to have an effect on the stability of FtsZ polymers in vitro (53). Interestingly, the original characterization of ftsL (mraR) mutants of E. coli also showed that they had medium-dependent phenotypes (32,36). Lysis, branching, and deformation of the cell wall were observed. Most likely, those phenotypes were in response to differences in the salt (NaCl) concentration. We do not know how those results relate to our present observations. What roles do FtsL and DivIC perform during division? At present, their functions are not known (11,20,23,54), and few clues are discernible in the smooth filaments that result following thermal inactivation or blocking synthesis of either protein in rod-shaped bacteria. These products appear to be structural components of the multisubunit division apparatus, important for recruiting other division proteins (see below) and perhaps indirectly acting through the division apparatus to stabilize the cytokinetic Z ring during constriction. For example, the action of DivIC of B. subtilis would have to be indirect, because only the extracellular domain is required for division (41). Alternatively, because these proteins have coiled-coil regions, it was suggested that they might be involved in membrane fusion at the culmination of septation (13,25,46). The two proteins are not absolutely required for either process during division for S. coelicolor, especially when growth occurs on a low-osmolarity medium, conditions where approximately one-third of the aerial hyphae of the mutants can divide nearly as well as the wild type (Fig. 2). Apparently, when growth occurs on such a medium, the ladder-like array of Z rings that form in aerial filaments of S. coelicolor (59) is relatively stable during constriction without the two proteins, and membrane fusion can occur.
Our TEM observations of the phenotypes of S. coelicolor mutants suggest a function for FtsL and DivIC. In the complete absence of the proteins, we frequently observed ingrowth of septal material from only one side for both sporulation septum formation (for approximately 8%) as well as vegetative cross-wall formation (for approximately 9%). In the cases where we observed ingrowth from two sides, one side was frequently longer than the other. Therefore, it appears that a coordination process has been disturbed and symmetrical annular ingrowth has been impaired. When division proceeds beyond the initial constriction phase in aerial filaments, which is FtsL and DivIC independent in aerial hyphae, it appears that only part but not all of the constricting septum-synthesizing machinery is halted (or drastically slowed) from proceeding. In the absence of FtsL and DivIC, further ingrowth becomes FIG. 4. Phase-contrast microscopy of ftsL SC -null and divIC SC -null mutants, showing the effects of omitting sucrose from R2YE medium on aerial hyphal morphology. Shown are representative aerial hyphae produced by the wild-type and mutant strains when grown for 6 days on standard R2YE (A to C) or R2YE prepared without sucrose (D to F). The wild-type strain (M145) produces aerial hyphae that metamorphose into chains of spores when grown on both types of media (A and D). The ftsL SC -null (DU191) and divIC SC -null (DU152) mutants produce constricted aerial hyphae (straight and coiled) that are blocked for division on standard R2YE (B and C, respectively). These same mutants are able to complete division in a fraction of the aerial hyphae when grown on R2YE prepared without sucrose (E and F, respectively).

VOL. 189, 2007
ftsL AND divIC MUTANTS OF STREPTOMYCES 8989 skewed. Based on these observations, we propose that FtsL and DivIC function to coordinate the symmetrical ingrowth of the constricting division apparatus (Fig. 5). Perhaps FtsL and DivIC (FtsB) function as the switch for the proposed second step for the maturation of the divisome (1). Consistent with this idea, FtsL is rate limiting for division in B. subtilis (20), overexpression of ftsL BS in ezrA-disrupted cells stimulates Z ring constriction (42), and FtsL SP only appears transiently at the division site in S. pneumoniae (54). At present our data are unable to distinguish if the entire divisome fails to invaginate coordinately and symmetrically or if the Z ring constricts symmetrically and the remainder of the division apparatus invaginates asymmetrically. We were unable to detect localized signals using green fluorescent protein fusions to FtsL and DivIC to help provide evidence (data not shown). The TEM analysis illustrates two additional points about S. coelicolor development. First, there is a genetic or physiological distinction between the two temporally and spatially separated types of division in this mycelial organism (see also reference 29). Sporulation septation was more sensitive to the loss of ftsL SC or divIC SC than vegetative cross-wall formation. For each single mutant, more than 50% of vegetative cross-walls were normal in appearance, while an additional 20% were complete but aberrant in some way. In contrast, only 9% of sporulation septa were complete and normal in appearance, while more than 75% of division sites in aerial filaments were partially constricted. Second, TEM analysis indicates that division may not be absolutely required to regulate developmental gene expression necessary for spore maturation. Despite the inability to complete division in most cases, differentiation of the cell wall of aerial filaments continues (for examples, see Fig. 3E and H). The cell wall of the division mutants has a thickened appearance similar to that of mature spores (Fig.  3G) and is not thin, as in early prespore compartments (Fig.   3D). Furthermore, many of these compartments were rounded, more like mature spores than the rod-shaped prespore compartments that normally result after sporulation septation divides an aerial hypha (compare Fig. 3D with 3E). These observations indicate that there may not be an absolute requirement (i.e., a checkpoint) to complete the subdivision of aerial filaments into unicellular compartments before activating "late" developmental gene expression in aerial filaments.
In recent years, numerous studies using cell biology tools have shown that the known division proteins are recruited to the middle of the cell in paradigm rod-shaped bacteria E. coli and B. subtilis (references 11 and 23 and references therein). Additional methods demonstrate that divisome assembly in E. coli and B. subtilis results from many unexpected interactions (12,20,24,27,38). We do not know yet the order of division protein recruitment to division sites in S. coelicolor or if localization of FtsL and DivIC are codependent as they are in B. subtilis and FtsL and FtsB (YgbQ) are in E. coli. In E. coli and B. subtilis, FtsL and DivIC (FtsB) are required so that FtsI and FtsW (and FtsN in E. coli) are recruited to the division site. If division proteins are recruited in a similar order with similar dependencies in S. coelicolor, then our current observations lead one to consider an additional puzzle. When both FtsL SC and DivIC SC proteins are missing from the division complex for an S. coelicolor double mutant, how are the anticipated later-recruited proteins attracted to the division complex? If FtsI and FtsW are not recruited to division sites in the ftsL SC divIC SC double mutant, then S. coelicolor is able to divide without at least four "essential" division proteins (i.e., FtsL, -I, and -W and DivIC).
FIG. 5. Model proposing a function for FtsL and DivIC during prokaryotic cytokinesis. During division in a wild-type aerial filament, the nascent invaginating septum constricts symmetrically (top). In aerial filaments mutant for ftsL SC or divIC SC , symmetrical invagination is impaired and coordination is lost, resulting in division sites where the septum is frequently synthesized from one side of the cell cylinder (bottom). Sagittal (A and B) and longitudinal (C) cross-sections of division sites within aerial filaments are depicted. The newly synthesized cell wall material during constriction is hatched in the sagittal sections through the septum (A and B). For a wild-type hypha, regardless of the orientation of the longitudinal section, invagination appears to occur evenly from both sides. In a longitudinal section through the vertical plane of a division mutant (C), invagination predominantly appears asymmetrical. Note that if a longitudinal section were in the horizontal plane for a division mutant, it would appear that invagination was occurring from both sides. These patterns would be expected and do occur with approximately equal frequencies for the division mutants (Table 1).