Role of SufI (FtsP) in Cell Division of Escherichia coli: Evidence for Its Involvement in Stabilizing the Assembly of the Divisome

Bacterial strains and phages.All strains used in this study are derivatives of E. coli K-12 and are listed in Table 1. Strain MR2 (MG1655 ΔlacI) was used as the wild-type strain (38). The construction of the sufI deletion strain is described below. The division-defective ftsZ84, ftsA12, ftsK44, ftsQ1, and ftsI23 mutations were introduced by phage P1-mediated transduction into other strain backgrounds, with the aid of linked antibiotic markers from strains DRC14, EC290, TOE44, JOE86, and LMG64 (obtained from the laboratory of Jon Beckwith), respectively. sodA sodB mutants (10) were obtained from the Coli Genetic Stock Centre (CGSC), and sodC mutants (19) were obtained from the laboratory of James Imlay. Phage P1kc was obtained from a laboratory stock.

Plasmids.pBAD18, an Amp^r, ColEI-based, arabinose-inducible vector (22), was used to clone sufI under the control of the ara promoter. The sufI gene was amplified with primers sufIFP and sufINRP (5′-GGAATTCCATGGAGCAAATATGTCACTC and 5′-GGAATTCGAAGATTACGGTACCGGATTG, respectively, with EcoRI restriction sites underlined), using high-fidelity Pfx polymerase (Invitrogen), digested with EcoRI, and cloned at the same site in pBAD18 to create pMN58. Plasmid pMU2385, a Tp^r, very-low-copy-number IncW derivative carrying a promoterless lacZ reporter gene (45), was used for the construction of a sufI-lac promoter fusion. A 230-bp fragment encompassing the putative promoter region of sufI was amplified with sufFP4 (5′-CGGGATCCGCCAATTGACGTCAGTC) and sufRP4 (5′-GGAATTCTGACGCCGACTGAGTGACA) (restriction sites are underlined), digested with BamHI and EcoRI, and cloned at the corresponding sites in pMU2385 to create pMN59.

Growth media and conditions.Unless otherwise indicated, Luria-Bertani (LB) medium (with 1% NaCl) was used (33), and the growth temperature was 30°C. LBON medium is the same as LB, but without NaCl. The culture media were obtained from BBL-Difco (Madison, WI). Supplementation of LBON with osmolytes (i.e., glucose, sucrose, glycerol, or NaCl) was done at 0.4 M. H₂O₂ (30% [wt/vol]) was used at 0.5 mM. Paraquat (PQ; methyl viologen) and mitomycin C (MMC) were used at the indicated concentrations. l-Arabinose or d-glucose was used at 0.2% for induction or repression of sufI cloned into the pBAD18 vector. Sodium ascorbate was used at 15 mM. The following antibiotics were used at the indicated concentrations: ampicillin (Amp), 50 μg/ml; kanamycin (Kan), 50 μg/ml (in LB) and 10 μg/ml (in LBON); tetracycline (Tet), 15 μg/ml; chloramphenicol (Cam), 30 μg/ml; spectinomycin (Spc), 50 μg/ml; and trimethoprim (Tp), 60 μg/ml.

Construction of a chromosomal sufI deletion-insertion mutant.A complete deletion of the sufI locus was made on the chromosome by recombineering as described previously (50). A pair of primers having homology at the 5′ end to the flanking region of the sufI locus and at the 3′ end to the sequence of the kanamycin resistance gene cassette (sufIFPKan [5′-TGCGGGGAACACTTTCCTGCACGGTATTACTTTAGCCAGTTTTACATGGATTGTGTCTCAAAA TCTCTGAT-3′] and sufIRPKan [5′-GCCCTCCTCGGGCGAGTATGAAGATTACGGTACCGGATTGACCAACAGTTAGAAAAACTCATCGAGCATCA AATG] [underlined sequences are homologous to the Kan^r gene of plasmid pUC4K]) were employed to amplify a 0.9-kb Kan^r cassette of pUC4K. This linear PCR product with flanking homologous sequences of sufI was electroporated into strain HME63, and Kan^r transformants were obtained at 30°C. The putative sufI::Kan deletion-substitution was transferred to a fresh background by P1 transduction, and subsequently, the presence of the deletion was confirmed by sequencing the PCR product amplified with the flanking primers (sufIFP3 [5′-CGGAATTCGGCAATCTGTATTTTTGC] and sufIRP3 [5′-CGGAATTCGCCCTCCTCGGGCGAGT] [EcoRI restriction sites are underlined]). This deletion removed the entire sufI gene.

Construction of ftsA* derivative of strain MR22.The ftsA* mutation was transferred into the sufI mutant MR22 from strain WM1659 (obtained from William Margolin) with the aid of the leuB82::Tn10 marker, which is approximately 50% linked to the ftsA locus. Initially, strain WM1659 was transduced to Tet^r with a P1 lysate prepared on a strain carrying the leuB::Tn10 marker. The presence of the ftsA* mutation in the Tet^r transductants was examined by preparing P1 lysates on six of these and checking for the ability to suppress the temperature sensitivity of an ftsK44 mutant. Of the six Tet^r transductants examined, two were shown to have the ftsA* allele, and subsequently, the P1 lysates made with these two strains were used to transduce strain MR22. The presence of the ftsA* allele in these sufI Tet^r transductants was tested by preparing P1 lysates on six of them and transducing them again into the ftsK44 mutant strain. Of the six sufI Tet^r colonies, three were shown to carry the ftsA* allele, and one of these was designated MR29. The isogenic Tet^r transductant that retained the wild-type ftsA locus was designated MR30.

LBON-Ts phenotype of sufI mutant.For most of the experiments, the growth of the sufI mutant was examined on LBON plates prepared from BBL-Difco Laboratories medium, on which growth was significantly inhibited at high temperatures (LBON-Ts phenotype). However, on LBON plates prepared with medium components sourced indigenously (Hi-Media, Mumbai, India), the growth inhibition was very severe, and therefore we used this medium for the isolation of multicopy suppressor plasmids as described below.

Screening for multicopy plasmids that suppress the phenotype of the sufI mutant.A multicopy plasmid library carrying approximately 3- to 5-kb E. coli genomic DNA fragments generated by partial Sau3A digestion and cloned at the BamHI site in a p15A-based plasmid, pACYC184, was a gift from M. Radman's laboratory. This plasmid pool was introduced into the sufI mutant MR22, and transformants were plated on LBON-Cam plates at 42°C. Plasmids were isolated from colonies that grew to different extents, and their ability to suppress was reconfirmed after an additional round of transformation. Subsequent restriction analysis and sequencing with the vector primers (184TetA [5′-CGCCGAAACAAGCGCTCATGAGCC] and 184TetB [5′-CTATGCGCACCCGTTCTCGGAGCAC]) allowed the identification of various classes of plasmids that restored the viability of the sufI mutant on LBON medium. One such suppressor plasmid, pMN60, obtained in this screen carried the complete dapE gene and the C-terminal part of the adjacent acrD gene. Deletion of dapE from pMN60 eliminated the ability to suppress the phenotype of MR22, indicating that dapE was responsible for the multicopy suppression.

Other techniques.Standard protocols were followed for experiments involving recombinant DNA and plasmid manipulations (39). Transpositions, transductions, P1 phage preparations, and β-galactosidase assays were performed using standard methods, as described previously (33).

Growth and viability measurements.The viability of each strain was measured by applying 8-μl aliquots of various dilutions (10⁻², 10⁻⁴, 10⁻⁵, 10⁻⁶, and 10⁻⁷) of overnight cultures to appropriate plates and incubating them for 20 to 36 h. The relative plating efficiency and growth were determined for each strain based on comparison with the control strain on the same plate.

Microscopy.The strains to be examined by microscopy were diluted into appropriate medium from fresh overnight cultures and processed after 6 to 8 h of growth. They were fixed in 2% formaldehyde for 1 h at 37°C, washed twice with phosphate-buffered saline, and suspended in the same buffer containing 50% glycerol. For DAPI (4′,6′-diamidino-2-phenylindole dihydrochloride) staining of nucleoids, fixed cells were treated with 0.25 μg/ml DAPI for 15 min at room temperature and mounted on slides. Differential interference contrast (DIC) and fluorescence images were taken using a Zeiss Axioplan fluorescence microscope and were processed in Adobe Photoshop.

A sufI deletion mutant is sensitive to oxidative stress and DNA damage at high temperature.In order to study the phenotypes of a sufI mutant, a complete deletion of the sufI locus was made in the chromosome of strain MR2 as described in Materials and Methods. This mutant, designated MR22 (MR2 ΔsufI222::Kan), grew quite well on LB plates at all temperatures and on LBON plates at 30°C; however, on LBON at 37°C or 42°C, growth was significantly inhibited (Fig. 1A). This growth inhibition (LBON-Ts) was osmoremedial, as supplementation with any of a variety of osmolytes, such as glucose (Fig. 1A), sucrose, glycerol, or NaCl (data not shown), completely rescued the growth of MR22. The LBON-Ts phenotype of the sufI mutant was seen only on medium exposed to good aerobic conditions; the generation of partially anaerobic conditions by tightly sealing the plate with Parafilm alleviated the LBON-Ts phenotype (Fig. 1A). Furthermore, the addition of a chemical reductant, ascorbate, to the growth medium eliminated the LBON-Ts phenotype (Fig. 1A). Since these observations suggested that molecular oxygen or oxidative stress could be toxic to the sufI mutant, the effect of reactive oxygen radicals on the growth of this mutant was examined by adding either PQ, a redox cycling agent that generates superoxide radicals (O₂⁻), or H₂O₂, which generates hydroxyl radicals (OH·). The results showed that the sufI mutant was extremely sensitive to PQ (PQ^s) at 37°C or 42°C (Fig. 1A), indicating that superoxide-generated stress is lethal at high temperatures. It was only slightly sensitive to hydrogen peroxide up to 0.5 mM (data not shown).

• Open in new tab
• Download powerpoint

FIG. 1.

Growth and filamentation of sufI mutant. (A) Growth of MR2 (wt) and MR22 (sufI) on LBON plates at 30, 37, and 42°C and on LBON plates (at 42°C) with 0.4 M glucose, tightly sealed with Parafilm (to create partial anaerobic conditions), or with 15 mM ascorbate. (Bottom) Growth of wild-type, sufI, and sufI sulA (MR32) strains on LB plates supplemented with either PQ (20 μM) or MMC (0.3 μg/ml) at 42°C. (B) Growth curves for wild-type and sufI strains at 30 or 42°C in LBON broth or LB broth plus PQ (40 μM; LBPQ) at 42°C. (C) DIC micrographs of sufI (a, b, and c) and wild-type (d, e, and f) cells taken out after 90 min, 3 h, or 6 h of growth in LBON at 42°C. The inset in panel c shows a very rare and exceptionally long sufI mutant filament.

We also examined the effect of DNA damage on the growth of the sufI mutant strain MR22 by exposing it to the DNA-damaging agent MMC. As shown in Fig. 1A, this mutant was extremely sensitive to MMC at a concentration of 0.3 μg/ml at 42°C. It is known that treatment of E. coli with DNA-damaging agents such as MMC elicits a strong SOS response, thereby inducing SulA (44), an inhibitor of FtsZ polymerization (34). In order to examine whether the MMC sensitivity (MMC^s) was mediated by SulA, a mutation in sulA was introduced into MR22, and it was shown that the sufI sulA double mutant (strain MR32) was insensitive to MMC treatment, clearly demonstrating that this process occurs by SOS-mediated induction of SulA (Fig. 1A). Likewise, the introduction of a mutation in lon that causes elevated levels of SulA (because lon encodes a protease whose normal substrate is SulA) (44) into strain MR22 resulted in extreme sickness at high temperature, which could be alleviated by a mutation in sulA (data not shown).

However, both the LBON-Ts (data not shown) and PQ^s phenotypes of MR22 (Fig. 1A) were not suppressed by sulA mutation, indicating that sensitivity to oxidative stress is not due to the SOS response caused by DNA damage. On the other hand, strain MR22 was not sensitive to alkylating agents, such as MNNG (N-methyl-N′-nitro-N-nitrosoguanidine) and EMS (ethyl methane sulfonate) (data not shown). An insertion mutation in tatB that affects the formation of a functional TatABC translocase, thereby abolishing the export of SufI into the periplasm, also conferred sensitivity to PQ or MMC at high temperatures; however, the tatB mutant required slightly higher concentrations of these agents for its sensitivity than did the sufI mutant (data not shown).

The sufI mutant is filamentous at high temperatures.Although the growth of strain MR22 was appreciably inhibited on LBON plates at 42°C, the cultures grown in LBON broth showed only a minor decrease in absorbance (A ₆₀₀) values (and also a corresponding reduction in the number of CFU) in the exponential phase and almost no difference in values in the stationary phase compared to those for the wild-type strain, MR2 (Fig. 1B). Likewise, the growth rate of MR22 was also not altered by the addition of PQ (up to 40 μM) to the broth cultures at 42°C (Fig. 1B).

We examined the cell morphology of the sufI mutant by taking aliquots from the cultures growing in LBON at 42°C at various time points. As shown in Fig. 1C, the cells from early exponential phase appeared to be significantly longer and were heterogeneous in size. Nevertheless, the filamentation of these mutants slowly disappeared as the culture entered into stationary phase. At an A ₆₀₀ value of around 1, the cells started regaining their normal shape, and by an A ₆₀₀ value of 1.5, the wild-type and mutant cells were almost indistinguishable, though the latter were slightly elongated (Fig. 1C). However, a few exceptionally long filaments could be seen at a very low frequency (approximately 1 in 10⁴ cells) in the stationary-phase cultures of the sufI mutant (inset in Fig. 1C). A similar pattern of filamentation was seen in cultures grown in either LB or minimal medium (with 0.2% glucose as the C source) at 42°C, demonstrating that cell elongation is a function of temperature and does not depend on the growth rate in the medium (data not shown). Likewise, the presence of PQ in the growth medium had no effect on the filamentation of the sufI mutant; it neither enhanced nor allowed filamentation to persist in the stationary phase (data not shown). Conversely, MMC treatment greatly enhanced the filamentation of the sufI mutant (because of SulA-mediated inhibition of FtsZ), and as expected, this increase was abolished in the sufI sulA double mutant, MR32 (data not shown). Nevertheless, strain MR32 was as filamentous as MR22, clearly indicating that the filamentation seen in the sufI mutant (Fig. 1C) was not a consequence of SOS induction and that SufI as such is required for some step in the process of cell septation.

Visualization of the nucleoids by DAPI staining followed by fluorescence microscopy revealed that they were normal and regularly spaced, without any significant chromosomal aberrations or partition defects (data not shown).

Identification of multicopy suppressors of sufI.To understand the basis of sufI phenotypes, we identified multicopy suppressor plasmids that restored viability to the mutant on LBON at 42°C as described in Materials and Methods. One major class of suppressor plasmids was found to carry the complete ftsN gene. Since multicopy ftsN is known to suppress the defects of many division mutants (11, 14, 38), we tested the effect of plasmid pMN14, a medium-copy-number vector with cloned ftsN (38), on the growth of the sufI strain MR22. The results showed that multicopy ftsN was able to suppress the LBON-Ts phenotype of MR22 very efficiently (Fig. 2). Another set of plasmids conferring multicopy suppression of sufI carried the sdiA locus, which encodes a positive regulator of the ftsQAZ operon (46). Thereafter, the effect of increased ftsQAZ expression was directly examined by introducing plasmid pMN8 (38), and it was also found to confer viability to MR22 on LBON medium. However, plasmids carrying ftsQ, ftsA, ftsQA, and ftsAZ (38) did not restore growth to MR22 on LBON.

• Open in new tab
• Download powerpoint

FIG. 2.

Suppressors of sufI. Cultures of MR22 carrying the plasmid pCL1920, its derivatives with cloned sufI, ftsQAZ, or ftsN, or pMN60 (pACYC184-dapE) and strains MR29 (sufI ftsA* leuB::Tn10) and MR30 (sufI leuB::Tn10) were grown at 30°C in LB, and various dilutions (10⁻², 10⁻⁴, 10⁻⁵, 10⁻⁶, and 10⁻⁷) were applied to LBON, LB-plus-PQ (20 μM), or LB-plus-MMC (0.4 μg/ml) plates and grown for 24 h at 42°C.

Another multicopy suppressor plasmid, pMN60, carrying the complete dapE gene, also suppressed the defects of MR22 (Fig. 2). dapE encodes N-succinyl-l-diaminopimelic acid desuccinylase, an enzyme that catalyzes the last step in the synthesis of l-diaminopimelic acid, which is an essential structural component of the peptidoglycan layer (7). It was earlier identified as a multicopy suppressor of a temperature-sensitive allele of the heat shock gene grpE (grpE80), which codes for a cochaperone of the DnaK-DnaJ system (48). DapE was also shown to function as a Mn²⁺-dependent aspartyl peptidase in Salmonella enterica serovar Typhimurium (8); however, it is not very clear which activity of dapE is responsible for the suppression of SufI phenotypes. Multicopy dapE was also identified as a suppressor of ftsEX (M. Reddy, unpublished results).

In addition to conferring growth on LBON at 42°C, all of the above-described plasmids relieved both the PQ^s and MMC^s phenotypes of strain MR22 (Fig. 2). The filamentation of MR22 carrying these plasmids was also completely reduced (data not shown).

The ftsA* allele alleviates the defects of the sufI mutant. ftsA* is a gain-of-function allele that encodes a mutant FtsA protein (R286W) which is capable of suppressing the defects of zipA (15), ftsK (14), and ftsN (4) deletion mutants. The ftsA* allele was introduced into strain MR22 by phage P1-mediated transduction with the linked leuB::Tn10 marker (as described in Materials and Methods), and it was observed that the transductants carrying the ftsA* allele were not sensitive to either PQ or MMC and showed no growth defects on LBON at 42°C (Fig. 2). The double mutants also regained their normal cell shape at high temperatures (data not shown).

Multicopy SufI suppresses the division defects of ftsA12, ftsQ1, and ftsK44 mutants but not that of an ftsZ84 mutant.It was shown earlier that multicopy sufI could partially suppress the thermosensitivity of an ftsI23 mutant (25). In addition, it restored the viability of ΔftsEX mutants under low-osmolarity conditions (38). Here the ability of multicopy sufI to suppress the defects of other division mutants was examined by introducing the plasmid pCL1920 (control vector) or pMN16 (pCL1920 with cloned sufI) into strains MR24, MR25, MR26, and MR27, which carry ftsZ84 (Ts), ftsA12 (Ts), ftsK44 (Ts), and ftsQ1 (Ts) mutations, respectively. As shown in Fig. 3A, strain MR26/pMN16 grew well at 42°C on both LB (Fig. 3A) and LBON (data not shown) plates, and correspondingly, its filamentation was notably decreased (Fig. 3B). Strains MR25/pMN16 and MR27/pMN16 also grew well at 42°C on LB plates (Fig. 3A) but not on LBON plates (data not shown). The filamentation of these strains was not significantly decreased at 42°C, although the filaments appeared to be slightly shorter and healthier (data not shown). However, the filamentation of both of these strains was considerably reduced when they were grown at 37°C (Fig. 3B). The plasmid pMN16 did not suppress the temperature sensitivity of strain MR24 on either LBON (Fig. 3A) or LBON supplemented with 0.5% NaCl (data not shown), and likewise, the filamentation remained unchanged (data not shown).

• Open in new tab
• Download powerpoint

FIG. 3.

Effects of multicopy sufI on growth and filamentation of division mutants. (A) Viability of division mutants carrying the vector, pCL1920 (−), or pMN16 (+) on LB plates at 30 or 42°C. ftsZ84 mutants were plated on LBON. (B) DIC micrographs of ftsA12, ftsK44, and ftsQ1 strains carrying either pCL1920 (a, b, and c) or pMN16 (d, e, and f). Cells were grown to mid-exponential phase in LB at 37°C (ftsA12 and ftsQ1 cells) or 42°C (ftsK44 cells) and then taken for microscopy.

Multicopy SufI relieves the MMC and PQ sensitivity of division mutants.We examined the sensitivity of other division mutants (carrying the ftsZ84, ftsA12, ftsK44, ftsQ1, ftsI23, ΔftsEX, or ΔsufI allele) to DNA damage or oxidative stress by exposing them to either MMC or PQ at 30°C. Compared to the wild-type strain MR2, all of the division mutants, excepting the ftsZ84 mutant, were sensitive to both agents at a range of concentrations. Of all the mutants, strains MR28 (ftsI23) and MR10 (ΔftsEX) were extremely sensitive to MMC (at 0.4 μg/ml) and PQ (at 15 μM), and this sensitivity was abolished by overexpression of SufI. Plasmid pMN58, a derivative of pBAD18 with sufI placed under the control of the arabinose-regulated promoter, conferred a growth ability on MR28 and MR10 upon the addition of 0.2% arabinose, whereas the addition of glucose had no effect (Fig. 4).

• Open in new tab
• Download powerpoint

FIG. 4.

Effects of multicopy sufI on MMC and PQ sensitivities of division mutants. Cultures of ΔftsEX::Kan, ftsI23, and ΔsufI mutants carrying plasmid pMN58 (pBAD18-sufI) were grown, and dilutions were applied to LB-MMC (0.4 μg/ml) or LB-PQ (15 μM) plates supplemented with either 0.2% l-arabinose or d-glucose and incubated for 24 h at 30°C.

SufI is indispensable for viability of the ftsI23 mutant.To check the effect of sufI deletion on the growth and viability of other division-defective mutants, we attempted to introduce the sufI::Kan deletion directly by phage P1-mediated transduction. It could be introduced into strains MR24 (ftsZ84) and MR25 (ftsA12) but not into MR26 (ftsK44), MR27 (ftsQ1), or MR28 (ftsI23). The reciprocal transductions were also done by using the linked leuB::Tn10 marker for selection in the case of mutations in MR24, MR25, MR27, and MR28 and the zbj::Tn10 marker in the case of mutation in MR26, and similar results were obtained (data not shown). However, the sufI deletion could be introduced into strains MR26 and MR27 in the presence of increased concentrations of osmolytes (i.e., on LB plates supplemented with 0.2 M glucose or NaCl), but not into MR28 (although very tiny transductants appeared after 48 h of incubation, they failed to grow upon further purification [data not shown]). Even in the presence of multicopy plasmids carrying ftsN, ftsQAZ, or dapE, the sufI deletion could not be introduced into strain MR28 (data not shown), suggesting that sufI is absolutely required for the viability of the ftsI23 mutant. It was shown earlier that a sufI ftsEX double mutant is also inviable (38). All of these observations indicate that the absence of SufI causes sickness to the division mutants to different extents.

Transcriptional regulation of sufI.To examine whether the expression of sufI is regulated by conditions of stress, the promoter of sufI was cloned into the single-copy promoter-probe vector pMU2385 (45), and its expression was measured by β-galactosidase reporter gene assays. The basal level of β- galactosidase activity of the vector was in the range of 3 to 5 Miller units, whereas plasmid pMN59 (with the sufI promoter) showed activity of 25 to 30 Miller units, indicating that the promoter is weak. None of the perturbations tested, including altered temperature, osmolarity, stationary phase, oxidative stress, DNA damage, or mutations in either the stationary-phase sigma factor gene rpoS or soxR, encoding an activator of the superoxide regulon, affected the promoter strength of the sufI promoter.

Role of SufI in cell division.The sulA-independent filamentation phenotype of the sufI deletion mutant at 42°C indicates that some step of cell septation is blocked at high temperatures (Fig. 1C). The filamentation pattern of the sufI mutant is interesting in that cells in the early exponential phase exhibit cell elongation but gradually regain their normal cell shape at later stages of growth (Fig. 1C), suggesting the existence of an adaptation mechanism that operates in stationary phase. Since the stationary-phase sigma factor RpoS regulates a morphogene, bolA, that controls cell shape in stationary phase (27), an rpoS deletion was introduced but was shown not to alter the viability or filamentation pattern of MR22 (data not shown). One other possible reason for this growth phase-dependent filamentation of the sufI mutant could be the generation of partially anaerobic conditions in the medium with increasing culture density; however, cells grown with high or low aeration did not show significant variation in filamentation (data not shown).

The suppression of sufI mutant filamentation by factors that stabilize the FtsZ ring assembly strongly argues for the functional involvement of sufI in the division process. Most of the phenotypes of the sufI mutant (Fig. 1) were suppressed by multiple copies of FtsQAZ, showing that a coordinated increase of all these divisomal proteins removes the defect of sufI deletion (Fig. 2). Likewise, multicopy FtsN also abolished all the defects of the sufI mutant. Furthermore, the presence of the gain-of-function ftsA* allele, which codes for an altered FtsA (R286W) protein, alleviated the phenotypes of the sufI mutant very efficiently (Fig. 2). This allele emerged as a nonspecific suppressor of most of the division mutants and, interestingly, was able to eliminate the requirement for several essential division proteins, such as ZipA (15), FtsK (14), FtsN (4), and FtsEX (Reddy, unpublished observations), most likely by stabilizing the FtsZ ring assembly (16). In accordance with all of the above observations, a SufI-green fluorescent protein fusion protein has been shown to localize to the division septum (David Weiss, personal communication).

SufI is required for growth under conditions of stress.Most of the phenotypes of the sufI mutant were more striking at 37 or 42°C, reflecting the need for SufI at high temperatures (Fig. 1A) and/or implicating high temperature as a sensitizing factor for the assembly of the divisome (38). It appears that SufI is rendered essential for the growth of E. coli whenever the division process is compromised by various conditions, as described below.

(i) Decreased FtsZ function through high levels of SulA, generated either by treatment with the DNA-damaging agent MMC or by a mutation in Lon protease, is detrimental to the sufI mutant (Fig. 1A). This premise is also strengthened by the extreme sickness of an ftsZ84 sufI double mutant at 42°C (on LBON plates that contained 0.7% NaCl, on which both the single mutants grew reasonably well) and also by its filamentation phenotype in LB at 30°C, where the single mutants showed normal cell morphology (data not shown). Hence, it seems that the MMC^s phenotype of the sufI mutant is due to two overlapping mechanisms, i.e., compromised division in the absence of SufI coupled to the lowered activity of FtsZ.

(ii) SufI is also required for the growth and viability of the various division-defective mutants carrying the ftsQ1 (Ts), ftsK44 (Ts), ftsI23 (Ts), or ΔftsEX allele. The extent of dependence on SufI may vary with the severity of the division-defective mutation, and this could be the most likely basis for the absolute requirement of SufI for the viability of either the ΔftsEX or ftsI23 mutant (38; this study).

(iii) Unlike the MMC^s phenotype, both the PQ^s and LBON-Ts phenotypes of the sufI mutant are not suppressed by a mutation in sulA, showing that these are not the result of a DNA damage response (Fig. 1A). This observation permitted us to speculate that oxidative damage may cause stress to the divisomal assembly, and the fact that most of the division mutants are sensitive to superoxide radicals (Fig. 4) supports the idea that the division process could be intrinsically sensitive to oxidative stress.

However, the effect of oxidative stress on the division assembly could be indirect, as growth at high temperatures coupled with increased oxidative stress or low osmotic strength may require a higher activity of chaperones and therefore may affect the proper folding of division proteins. It was recently shown that the chaperonin GroE is involved in folding of the division protein FtsE, and the filamentation phenotype of a GroE depletion mutant is due to the improper folding of FtsE (13). It has been shown that the rate of superoxide formation in the periplasm is fairly high (∼3 μM/s, when normalized to the estimated periplasmic volume, whereas the rate in the cytosol is ∼5 μM/s), and it also contains a Cu, Zn superoxide dismutase (SodC) for scavenging the periplasmic superoxides (19, 26). Yet sodC mutants are not sensitive to superoxide-generating agents, indicating that there probably should be some hitherto unidentified factor(s) that protects the periplasm from superoxide stress (19). However, the preliminary observations show that a sufI sodC double mutant is viable and just as PQ sensitive as the sufI single mutant. Likewise, mutations in sodAB which abolish the cytoplasmic superoxide dismutase activity (10) also did not alter the viability or filamentation pattern of the sufI mutant (data not shown).

SufI stabilizes the divisomal assembly.In an earlier study (38), we speculated that both SufI and FtsEX may have redundant functions, based on the findings that (i) the presence of sufI is essential for the viability of an ftsEX deletion mutant and (ii) the multicopy SufI protein is able to substitute for the functions of FtsEX. However, in this study, it was observed that the absence of SufI is detrimental to the growth of several division-defective mutants. Hence, it is possible that the function of SufI may not be particularly redundant or overlapping with that of FtsEX alone but may have a wide-ranging role in protecting the divisomal assembly or its components. In support of this, multicopy sufI suppressed the growth defects of division mutants carrying the ftsA12, ftsEX, ftsK44, ftsQ1, or ftsI23 allele (25, 38; this study). In addition to suppressing the thermosensitivity of the division mutants (Fig. 3A), multicopy sufI also abolished the PQ and MMC sensitivity of ftsI23 and ftsEX mutants (Fig. 4), validating the idea that SufI protects the divisomal components against damage caused by various stress conditions. Preliminary observations from this laboratory showed that increased FtsQAZ or FtsN also suppresses the PQ and MMC sensitivity of ftsI23 and ftsEX mutants.

The restoration of the growth defects of the division mutants by multicopy sufI is reminiscent of the suppression shown by multicopy ftsN (11). Both sufI and ftsN in multiple copies are able to confer growth to all of the above mutants, except the ftsZ84 mutant. The role of FtsN in division is not clear; in vitro, it binds murein sacculus and is presumed to be involved in peptidoglycan assembly or stabilization of the divisome components (42). It is possible that SufI may also function in an analogous way. As proposed earlier (14), division proteins may perform two main functions: one is to stabilize the interactions between divisome components, and the second is to generate the division septum. SufI may fall into the former class of divisome proteins.

However, the precise biochemical function of SufI does not emerge from these studies. A homology search with SufI has shown that it belongs to the multicopper oxidase family. It closely resembles the blue copper oxidase (CueO) of E. coli and, to a lesser extent, the spore coat protein (CotA) of Bacillus subtilis. Both CueO and CotA are copper-dependent oxidoreductases. CueO is a TatABC substrate that confers copper tolerance to E. coli by oxidizing the toxic cuprous ions into cupric ions in the periplasm (20), whereas CotA is a thermostable laccase that protects the spores of B. subtilis against UV irradiation and oxidative stress (23). Unlike its homologs, SufI is not known to bind any cofactor (3). SufI does not appear to be a ubiquitous protein. In this context, it is interesting that both FtsN and SufI are present only in organisms belonging to the orders Enterobacteriales and Pasteurellales of the Gammaproteobacteria.

Our results, along with the localization data of David Weiss, strongly argue for a role of sufI in the cell division of E. coli. Since sufI fulfils the requirements of an fts gene (as defined in reference 9), we propose that it be redesignated ftsP. In addition, this will avoid confusion with other known suf genes (the sufABCDSE operon located at 37.9 min) that are involved in the biogenesis of iron-sulfur (Fe-S) clusters in E. coli (37).