Membrane Topology of the Streptococcus pneumoniae FtsW Division Protein

Abstract

The topology of FtsW from Streptococcus pneumoniae, an essential membrane protein involved in bacterial cell division, was predicted by computational methods and probed by the alkaline phosphatase fusion and cysteine accessibility techniques. Consistent results were obtained for the seven N-terminal membrane-spanning segments. However, the results from alkaline phosphatase fusions did not confirm the hydropathy analysis of the C-terminal part of FtsW, whereas the accessibility of introduced cysteine residues was in agreement with the theoretical prediction. Based on the combined results, we propose the first topological model of FtsW, featuring 10 membrane-spanning segments, a large extracytoplasmic loop, and both N and C termini located in the cytoplasm.

The ftsW gene was originally identified in Escherichia coli (13) as a member of the division and cell wall (dcw) cluster, which contains genes required for peptidoglycan synthesis and cell division (ftsL, ftsI, murE, murF, mraY, murD, ftsW, murG, murC, ddl, ftsQ, ftsA, and ftsZ) (19), and was later found to be essential in this bacterium (2). The deduced membrane protein FtsW is similar to the RodA protein of E. coli and to the SpoVE protein of Bacillus subtilis (12), and it has been suggested that these proteins have similar roles in cell division, cell elongation, and spore formation, respectively (12, 22). More recently, FtsW was shown to be localized to the septa of dividing E. coli cells (34). These authors proposed that FtsW may play a role in the stabilization of the FtsZ ring during cell division. Because the elongation mode of peptidoglycan synthesis requires penicillin-binding protein 2 (PBP2) and RodA in E. coli (7, 28) and because PBP2 and RodA are thought to act in concert (7, 14), it was suggested that FtsW similarly interacts with PBP3 during septal peptidoglycan synthesis (12). It was proposed that FtsW and RodA might be connected to the flipping of N-acetylglucosamine-N-acetylmuramoyl-(pentapeptide)-pyrophosphoryl undecaprenol (lipid II) through the cytoplasmic membrane (9, 10). The lipid II which is assembled on the cytoplasmic side of the membrane is translocated by an unknown mechanism to the extracytoplasmic side, where its disaccharide-pentapeptide moiety is the substrate of the PBPs for the synthesis of the peptidoglycan. In gram-negative bacteria, ftsW is always physically linked to murG, which is responsible for the final cytoplasmic step in the synthesis of lipid II.

Streptococcus pneumoniae is one of the major human pathogens of the upper respiratory tract, causing pneumonia, ear infections, and meningitis. In S. pneumoniae, alteration of PBPs leads to increasing resistance to β-lactam antibiotics. FtsW, which is essential, is conserved among cell wall-synthesizing bacteria, and has no known human homologue, could be a target for novel nonpenicillin antibiotics. A prerequisite for the identification of regions involved in the binding of cytoplasmic and extracytoplasmic molecules is the determination of the membrane topology of FtsW. The hydropathy profile of the amino acid sequence of FtsW suggests the presence of 8 to 10 α-helical transmembrane (TM) domains connected by hydrophilic loops.

The topology of cytoplasmic membrane proteins has often been studied by the phoA gene fusion method in bacteria. Alkaline phosphatase (PhoA) is enzymatically active after translocation to the periplasm but is inactive when localized cytoplasmically (20). To obtain experimental evidence about the topology of FtsW, we first created and characterized a series of fusions between the N terminus of FtsW and PhoA. To overcome inconsistencies between the fusion protein results and the hydropathy profile, the analysis was extended by determining the accessibility of introduced cysteine to membrane-permeant and membrane-impermeant sulfydryl-specific reagents (27). Based on the results obtained with both strategies and the theoretical prediction, we report here the first topological model of FtsW, which consists of 10 TM segments, a large extracytoplasmic loop, and both N and C termini located in the cytoplasm.

MATERIALS AND METHODS

Materials.

The detergent 3-(laurylamido)-N,N′-dimethylaminopropylamine oxide (LAPAO) was a gift from G. Brandolin and was synthesized as described previously (3). The membrane-impermeant sulfydryl-specific reagent [2-(trimethylammonium)ethyl]methanethiosulfonate bromide (MTSET) was purchased from Toronto Research Chemicals. The membrane-permeant sulfydryl-biotinylating reagent N-(6-(biotinamido)hexyl)-3′-(2′-pyridylthio)propionate (biotin-HPDP) was from Uptima.

Bacterial strains and culture conditions.

S. pneumoniae strain R6 was used as a source of chromosomal DNA for PCR amplification of the ftsW gene. E. coli MC1061 [hsdr araD139 Δ(araABC-leu) ΔlacX74 galU galK rpsL thi] was used as a host for plasmid construction, for PhoA assays, and for expression of cysteine mutants. E. coli was cultured aerobically at 37°C in Luria-Bertani (LB) medium supplemented with ampicillin (100 μg/ml). PhoA activity was detected as blue colonies on LB agar plates containing 5-bromo-4-chloro-3-indolyl-phosphate (toluidine salt) (XP), ampicillin, and arabinose at concentrations of 40 μg/ml, 100 μg/ml, and 0.01%, respectively.

Construction of Φ(ftsw′-phoA)(Hyb) fusion vectors.

Standard procedures were used to prepare and handle recombinant DNA and to transform the E. coli cells. A sequence containing the E. coli phoA gene, lacking the first 28 codons, was PCR amplified as a BglII -HindIII fragment and cloned into the arabinose-inducible plasmid pARA14 (4) to create pARAPHOA. Hybrid genes coding for different fusions of the FtsW amino-terminal region to the 29th residue of PhoA were constructed by designing PCR primers (Table 1) that placed N-terminal sections of FtsW in frame with PhoA. A series of 28 3′ truncates of ftsW were then PCR amplified from S. pneumoniae R6 chromosomal DNA as PstI-BglII fragments and ligated into vector pARAPHOA. The reporter fusion sites were chosen such that each putative TM segment, based on the hydropathy profile, was flanked by PhoA fusions. All fusion sites and inserted PCR fragments were sequenced to ensure that no undesired mutation was introduced.

TABLE 1.

Oligonucleotides used for the construction of ftsW-phoA fusions

Site of fusion in FtsW	Sequencea
Ala34	TTAGATCTGCACTGGTGGTCGAA
Leu46	TTAGATCTAACTGCAAGGCGCTC
Leu72	TTAGATCTAAAAAATCTAGTCTCAATT
Gly101	TTAGATCTCCGTTTACTGAAATACC
Lys133	TTAGATCTTTGGAGAATCGGTGAG
Lys152	TTAGATCTAGCCATTGATTTTGAGTC
Pro174	TTAGATCTGGGAAAATTCCCAAACT
Arg198	TTAGATCTCGATAAGCGATTCCAC
Thr225	TTAGATCTGTCTCAACACCGATTAG
Arg238	TTAGATCTCGTTTGGCTACATAGCC
Ala247	TTAGATCTGCAAAAGGATTAAAA
Ala250	TTAGATCTGCACGATCGGCAAA
Gly253	TTAGATCTCCTGCATCAGCACGA
Gly267	TTAGATCTCCACCATTGACCATGG
Lys279	TTAGATCTTTTTCAATCGAGTTTCC
Ser293	TTAGATCTGAAAAGACAAAGTCTGTA
Glu298	TTAGATCTTCTTCAATCACGATAG
Gly303	TTAGATCTCCAACAAAGCCAAATTC
Ile315	TTAGATCTATCATGAAAAACAAGAGAG
Ala325	TTAGATCTGCTCGAATACCGACC
Pro328	TTAGATCTGGATTCTCCGCTCGAA
Ile349	TTAGATCTATATTGACAAATACCTGAA
Pro357	TTAGATCTGGAATCAAGCCCGAAA
Leu366	TTAGATCTAAGAAGGGGAAGGTTAC
Val375	TTAGATCTACTAGAAGACTATTTCCA
Ala388	TTAGATCTGCATCAATATTTAAGACAA
Arg397	TTAGATCTCGGTACAACTTAGC
Lys409	TTAGATCTTTCAACAGAAGGTTCATT

^aBoldface nucleotides indicate BglII sites.

Construction of unique cysteine FtsW mutants.

The S. pneumoniae ftsw gene was PCR amplified using the following primers 5′-GCTGCAGGACACCATCACCATCACCATAAGATTAGTAAGAGGCAC-3′ and 5′-GAGATCTACTTCAACAGAAGGTTCATTGGTTG-3′. The PCR product was cloned as a PstI-BglII fragment into pARA14. The resulting construct, pARAW, contained codons for AlaAlaGly(His)₆ introduced between the first Met codon and the second Lys codon of ftsW. This construct, encoding a His-tagged and Cys-free protein, was then used as a template for mutagenesis, which was performed with the Quick Change site-directed mutagenesis kit (Stratagene), using overlapping primer pairs (Table 2) .

TABLE 2.

Cysteine mutagenesis

Site of cysteine mutation	Sequencea
Tyr141	GCAAGAAGAAATAGCTACATGTGATTTTCAAGTTTTGAC
Tyr197	GTATACAGTTAGTGGAATCGCCTGCAGATGGTTTTCAACCATTCTG
Ser227	CAGCCTAATCGGTGTTGAAACGTTTTGCAAAATTCCAGTATTCGG
Ser240	GGCTATGTAGCCAAGCGCTTTTGCGCATTTTTTAATCCTTTTGCCG
Ala252	CCGATCGTGCTGATTGTGGCCACCAGTTAGCTAATTCTTATTTTGC
Asn265	GCTAATTCTTATTTTGCTATGGTCTGTGGCGGTTGGTTTGGTC
Ala286	CGATTGAAAAACGAGGTTATTTTCCGGAATGTCATACAGACTTTGTC
Ser305	GGCTTTGTTGGTGCATGCCTTATTTTAGCTCTCTTGTTTTTC
Arg324	GATTATCTTGGTCGGTATCTGCGCAGAGAATCCTTTCAATGCC
Asn330	CGAGCGGAGAATCCTTTCTGCGCAATGGTTGCACTCGGTGTC
Ser358	GATTTCGGGCTTGATTCCGTGCACAGGAGTGACTTTCCCC
Ser389	GCCTTTGTCTTAAATATTGATGCATGCGAAAAACGCGCTAAATTG
C-Ter	CAATGAACCTTCTGTTGAAGTGCTAGCATCTGGTACCAAGCTTG

^aOnly one of the complementary oligonucleotides is shown. The cysteine codons are shown in boldface. C-Ter, C terminus.

Assay of alkaline phosphatase activity.

E. coli MC1061 harboring the fusion vectors was grown for 2 h in LB medium in the presence of arabinose (0.01%) and ampicillin (100 μg/ml). Cells were collected by centrifugation and resuspended in 1.0 M Tris-HCl (pH 8.0) to give an optical density of 1 at 600 nm. The PhoA activity of the cells was then assayed by measuring the rate of p-nitrophenyl phosphate hydrolysis essentially as described previously (23). Activities are expressed in Miller units. Each value given is the mean value of triplicate determinations corrected for the background activity measured with cells harboring plasmid pARA14. The PhoA activity determinations presented in this study were performed on aliquots of the same samples that were used for Western blot analysis of protein expression.

Chemical labeling and blocking of cysteine residues.

E. coli MC1061 harboring plasmids encoding single-cysteine FtsW mutants were grown in LB medium containing ampicillin (100 μg/ml), and expression was initiated by the addition of 0.002% arabinose at mid-log phase. After further incubation for 3 h, cells in 2 ml of culture (optical density of 1 at 600 nm) were collected at 3,500 × g and converted to spheroplasts as follows: the pellets were resuspended in 450 μl of 100 mM Tris-HCl (pH 7.5) and 0.5 M sucrose before the addition of 450 μl of 0.1-mg/ml lysozyme dissolved in 8 mM EDTA (pH 8.0) and were incubated on ice for 20 min. They were pelleted at 6,000 × g for 2 min and washed with 1 ml of phosphate-buffered saline (PBS). Biotinylation and blocking of cysteine residues were performed essentially as described previously (27) with some modifications. For blocking of periplasmic thiol groups, cells were resuspended in 1 ml of 1 mM MTSET in PBS at room temperature. After 3 min, the spheroplasts were washed three times with PBS. Blocked and untreated cells were then incubated with 200 μl of PBS containing 1 mM biotin-HPDP at room temperature for 20 min. Subsequently, the cells were washed three times with PBS and lysed with 1 ml of lysis buffer (150 mM NaCl, 5 mM EDTA, 50 mM Tris [pH 7.5], 1% LAPAO). After centrifugation for 15 min at 15,000 × g, the supernatants were mixed with 75 μl of a 50% slurry of NeutrAvidin beads (Pierce) and incubated overnight at 4°C with mild agitation. The beads were collected by centrifugation and washed three times with lysis buffer, twice with high-salt wash buffer (500 mM NaCl, 5 mM EDTA, 50 mM Tris [pH 7.5], 0.1% LAPAO), and once with salt-free wash buffer (50 mM Tris [pH 7.5], 0.1% LAPAO). Proteins were eluted by adding 50 μl of 2× sodium dodecyl sulfate (SDS) sample loading buffer containing 200 mM dithiothreitol and vortexing the mixture.

Western immunoblotting.

Whole-cell extracts of samples used for the PhoA activity assays and the cysteine experiments, as well as proteins eluted from the NeutrAvidin beads, were subjected to SDS-polyacrylamide gel electrophoresis (PAGE) without prior heating in SDS buffer and transferred to nitrocellulose filters. FtsW-PhoA fusions were then immunostained with a rabbit anti-PhoA immunoglobulin G preparation (Rockland). Following incubation with horseradish peroxidase-coupled anti-rabbit antibodies, Ftsw-PhoA polypeptides were visualized by using the ECL chemiluminescence detection kit (Amersham Pharmacia) as recommended by the manufacturer. Single-cysteine FtsW mutants were immunostained with an anti-polyhistidine peroxidase conjugate mouse immunoglobulin (Sigma) and visualized as described above.

Computer analyses.

Multiple alignments of the FtsW and RodA sequences were performed with ClustalW. The putative membrane-spanning domains of the FtsW protein were predicted using the TopPredII (6), PHD (26), MEMSAT2 (15), and TMHMM (18) programs.

RESULTS

Identification of the putative ftsw gene of S. pneumoniae.

S. pneumoniae is one of the major human pathogens of the upper respiratory tract, causing pneumonia, ear infections, and meningitis. The genetic organization of the dcw cluster has been studied in S. pneumoniae (21). As observed in other gram-positive bacteria (25), some of the dcw genes are located in distinct chromosomal regions. In particular, ftsW is missing from the S. pneumoniae dcw cluster, as previously noted in B. subtilis (12), Staphylococcus aureus (25), and Enterococcus hirae (8). Two genes of the ftsW-rodA-spoVE family are present in the genome of S. pneumoniae. As neither of the two candidates is associated with other genes of the dcw cluster, they were tentatively identified on the basis of their degree of sequence similarity to well-defined FtsW and RodA proteins. Of the 53 proteins of the family present in the Swissprot and Trembl databases, 13 and 8 could be unambiguously ascribed to the FtsW or RodA groups, respectively. After multiple alignments of the sequences with ClustalW, the pairwise identity and similarity were scored and indicated that the protein selected for this study was more FtsW-like than its homologue, which was more RodA-like. For example, our selected FtsW sequence has 27 and 21% identity with E. coli FtsW and RodA, respectively, whereas the corresponding values for the selected RodA sequence are 21 and 26%. During the course of our work, two independent S. pneumoniae genome-sequencing and annotation projects that similarly identified the two proteins were completed (11, 29). No clear consensus sequence could be extracted that would easily discriminate FtsW from RodA. A single position appears to always be occupied by a lysine in FtsW (Lys²⁷⁹ in S. pneumoniae), whereas it is never a lysine in RodA.

Prediction of the membrane topology of FtsW.

Hydropathy analysis of FtsW and RodA proteins of various organisms by the TopPredII (6), PHD (26), MEMSAT2 (15), and TMHMM (18) programs suggested the presence of 8 to 10 TM helices, depending on the program used and the species considered. As the N terminus consists of 12 amino acid residues containing three positively and no negatively charged residues, this segment is unlikely to cross the membrane, according to the positive-inside rule (33). We therefore localized the N terminus of FtsW in the cytoplasm. A consensus predicted topology is presented in Fig. 1. It must be noted that MEMSAT2 failed to predict TM helix 5 (TM5) in all proteins considered.

FIG. 1.

Membrane topology of FtsW from S. pneumoniae. The model is a consensus derived from predictions by the TopPredII, PHD, MEMSAT2, and TMHMM programs after multiple alignments of 13 sequences. Fully conserved residues are shown as solid circles. The experimental results are depicted as follows: PhoA fusion sites are indicated by shaded and open arrowheads for active and inactive constructs, respectively; residues mutated to cysteine are shown as shaded and open ellipses or lightly shaded lozenges if the thiols are extracellular, cytoplasmic, or inaccessible, respectively. EXT, extracytoplasmic side; CYT, cytoplasmic side.

Construction of FtsW-PhoA fusion proteins.

The topology of bacterial membrane proteins has often been studied by the phoA gene fusion method (20). The use of alkaline phosphatase, which is enzymatically active in the periplasm but is inactive when localized cytoplasmically, is advantageous because a positive result requires the active export of the mature reporter enzyme moiety. In contrast, the use of β-galactosidase as an additional cytoplasmic reporter may give a positive phenotype as a result of many artifacts (35). 3′-truncated ftsW fragments were cloned in frame with phoA to construct the fusions. The choice of fusion site was determined from hydrophobicity plots of FtsW and selected so as to bond the carboxyl end of each putative TM segment to the reporter protein. A series of 28 Φ(ftsw′-phoA)(Hyb) fusion plasmids were thus constructed and analyzed after expression in E. coli cells.

Alkaline phosphatase activities and immunodetection of fusion proteins.

The alkaline phosphatase activity of E. coli cells harboring the Φ(ftsw′-phoA)(Hyb) plasmids was determined by blue-white screening on XP plates and by measuring the alkaline phosphatase activity of cells harvested in the early exponential growth phase. Since alkaline phosphatase functions in the periplasmic space, but not in the cytoplasm, the fusion sites of FtsW-PhoA giving high activity are presumably located in the periplasm, and in contrast, those with no activity are expected to be in the cytoplasm.

Fusions at Ala³⁴, Leu⁴⁶, Gly¹⁰¹, Pro¹⁷⁴, Thr²²⁵, Arg²³⁸, Ile³⁴⁹, Pro³⁵⁷, Leu³⁶⁶, Val³⁷⁵, Ala³⁸⁸, Arg³⁹⁷, and Lys⁴⁰⁹ grew as blue colonies on XP plates and showed high PhoA activities (>46 Miller units). The PhoA activities of the fusions located on the C-terminal side of the predicted TM10 were significantly weaker than those of all other periplasmic fusions (Table 3). The remaining fusions, including the Leu⁷², Lys¹³³, Leu¹⁵², Arg¹⁹⁸, Ala²⁴⁷, Ala²⁵⁰, Gly²⁵³, Gly²⁶⁷, Lys²⁷⁹, Ser²⁹³, Glu²⁹⁸, Gly³⁰³, Ile³¹⁵, Ala³²⁴, and Pro³²⁸ sites, presented the white phenotype and showed virtually no activity above background (<10 Miller units). Western blotting of the 28 FtsW-PhoA fusion proteins confirmed the synthesis of all the chimeras (Fig. 2), indicating that the absence of PhoA activity was not due to a very low level of protein expression. As previously observed with TM proteins (1, 24), they had a higher-than-expected electrophoretic mobility due to increased SDS binding to hydrophobic regions. The level of expression was particularly high for the chimeras with fusions to amino acid residues 34 and 46, which explained the higher PhoA activities observed with these constructs (Table 3).

TABLE 3.

Alkaline phosphatase activities of FtsW-PhoA fusion proteins

Fusion site	Colony color on XP platea	PhoA activity (Miller units)
Ala34	Blue	1,183
Leu46	Blue	858
Leu72	White	10
Gly101	Blue	210
Lys133	White	0
Leu152	White	1
Pro174	Blue	225
Arg198	White	2
Thr225	Blue	186
Arg238	Blue	222
Ala247	White	4
Ala250	White	1
Gly253	White	2
Gly267	White	4
Lys279	White	7
Ser293	White	7
Glu298	White	1
Gly303	White	0
Ile315	White	3
Ala325	White	1
Pro328	White	3
Ile349	Blue	212
Pro357	Blue	203
Leu366	Blue	217
Val375	Blue	228
Ala388	Blue	46
Arg397	Blue	57
Lys409	Blue	79

^aIsolates were plated on LB agar plates containing the chromogenic substrate XP for alkaline phosphatase (40 μg/ml).

FIG. 2.

Western blot analysis of FtsW-PhoA fusion proteins expressed in E. coli. Cell extracts were separated by SDS-12.5% PAGE and visualized by immunoblotting with anti-PhoA antibody. The position of the fusion junction is indicated above each lane.

The PhoA activities obtained with the first 10 N-terminal fusions are consistent with the predicted model and confirmed the existence and positions of TM1 to TM7. In contrast, the large extracellular loop which was predicted from Gly²²² to Glu²⁹⁸ was not confirmed by the PhoA results, as all the fusions between residues Ala²⁴⁷ and Glu²⁹⁸ showed no activity despite the correct expression of the chimeras. It is unlikely that FtsW crosses the membrane between Arg²³⁸ and Ala²⁴⁷, as this segment is too short to form a TM helix and no conserved hydrophobic cluster is located between these positions. The important discrepancy between the hydropathy-based prediction and the PhoA activities in the C-terminal half of the protein prompted us to probe the topology of this part of the molecule further by the accessibility of introduced cysteine residues.

Construction of cysteine variants and labeling with sulfydryl-specific reagents.

Unique cysteine residues were introduced into FtsW by replacement of the residues Tyr¹⁴¹, Tyr¹⁹⁷, Ser²²⁷, Ser²⁴⁰, Ala²⁵², Asn²⁶⁵, Ala²⁸⁶, Ser³⁰⁵, Arg³²⁴, Asn³³⁰, Ser³⁵⁸, and Ser³⁸⁹ and by adding a cysteine at the C terminus (Fig. 1). The positions were chosen in order to probe the computational model as well as a tentative topology derived from the PhoA fusion results. All of the mutants were well expressed in E. coli MC1061 cells (data not shown). Spheroplasts of E. coli MC1061 producing the desired FtsW variants were preincubated with or without the membrane-impermeant reagent MTSET to either block periplasmic thiols or leave all cysteine residues unblocked. Treated and untreated spheroplasts were then incubated with membrane-permeant biotin-HPDP, and labeled FtsW was isolated and detected as described in Materials and Methods. The results presented in Fig. 3 show that thiols at the following positions reacted with both biotin-HPDP and MTSET, indicating a periplasmic location: Ser²²⁷, Ala²⁵², Ala²⁸⁶, and Ser³⁵⁸. In contrast, cysteine residues in place of Tyr¹⁴¹, Tyr¹⁹⁷, Asn³³⁰, and Ser³⁸⁹ could be labeled by biotin-HPDP but were not blocked completely by pretreatment with MTSET, indicating the cytoplasmic location of these residues. The other cysteine mutants that were investigated could not be labeled (data not shown), indicating that residues Ser²⁴⁰, Asn²⁶⁵, Ser³⁰⁵, and Arg³²⁴ and the C terminus are inaccessible to the reagent biotin-HPDP and therefore are probably located within the membrane or buried within the protein structure.

FIG. 3.

Accessibility of introduced cysteine residues. Spheroplasts producing FtsW mutants with single cysteines at the given positions were treated with (+) or without (−) 1 mM MTSET for 3 min at room temperature. They were then incubated with 1 mM biotin-HPDP for 20 min at room temperature and lysed with the lysis buffer. Biotinylated proteins recovered with NeutrAvidin resin were separated by SDS-PAGE, and FtsW proteins were visualized by immunoblotting with an anti-polyhistidine immunoglobulin.

These results confirmed the locations of residues Tyr¹⁴¹, Tyr¹⁹⁷, and Ser²²⁷ determined by the PhoA fusion method and the theoretical prediction. In contrast, the location of Ala²⁵², Ala²⁸⁶, Ser³⁰⁵, Asn³³⁰, Ser³⁵⁸, and Ser³⁸⁹ did not match the topology derived from the PhoA activities, whereas they were in agreement with the hydropathy-based prediction. The extracellular location of Ala²⁵² and Ala²⁸⁶ confirmed the exposure to the extracellular space of the loop between TM7 and TM8 predicted by the computational model, while the existence of TM8, TM9, and TM10 is confirmed by the inaccessibility of Ser³⁰⁵, the cytoplasmic location of Asn³³⁰ and Ser³⁸⁹, and the extracytoplasmic location of Ser³⁵⁸.

DISCUSSION

FtsW is an essential membrane protein involved in bacterial cell division which is homologous to RodA and SpoVE. It has been proposed that FtsW, RodA, and SpoVE could play the same molecular role in cell division, cell elongation, and spore formation, respectively (12). Because RodA is thought to interact with PBP2 in E. coli during the elongation mode of peptidoglycan synthesis (7, 14), it was suggested that FtsW similarly interacts with PBP3 in septum formation.

Topological information on FtsW should be helpful in identifying protein domains or segments that might interact with other proteins involved in cell division. After multiple alignments of FtsW and RodA sequences and analysis with four topology prediction programs, a consensus model was obtained for the topology of FtsW from S. pneumoniae (Fig. 1). Several techniques exist to probe the topology of multispan TM proteins, including the use of epitope-specific antibodies, gene fusions with reporter enzymes, protease protection assays, and cysteine scanning (32). All of these methods have their specific limitations, and sometimes different techniques have suggested different topologies. Here, we used PhoA fusions and cysteine accessibility to test the theoretical prediction. Based on sequence alignment, hydropathy profiles, and both experimental procedures, we propose the first topological model of a member of the FtsW/RodA/SpoVE family, featuring 10 membrane-spanning segments, a large extracellular loop, and both N and C termini located in the cytoplasm.

The proposed TM helices contain between 19 and 23 amino acids, the average length being that required to span the hydrophobic region of the membrane. The distribution of positive charges in the model follows the positive-inside rule (33), with basic residues predominantly located on the cytoplasmic side (Δ9) and acidic residues more evenly spread (Δ1).

The model includes a large extracytoplasmic loop of 77 residues between TM7 and TM8, which may be involved in interactions with FtsW partners. The critical role of this region is supported by the fact that half of the fully conserved residues are located in this extracellular loop (Fig. 1). Most of the other fully conserved residues are located in other short extracytoplasmic loops, whereas none are found on the cytoplasmic side. Contrary to the general high accessibility of thiol groups in hydrophilic regions, the variable efficiency of biotinylation of introduced cysteines suggests that the large extracellular loop may adopt a structure in which some amino acid residues are not readily accessible to sulfydryl-specific reagents. Interestingly, one of the mutations in the ftsw gene of E. coli that results in a block to division (16) is located in this extracytoplasmic loop.

The two other mutations in the ftsw gene that cause a temperature-dependent block in E. coli cell division (16, 17) are located in the cytoplasmic loop linking TM4 and TM5 and in TM5, respectively. It has been proposed previously that FtsZ, a cytoplasmic protein which forms a ring along the membrane at the septum site (19), may interact with FtsW (16, 17). It was also proposed that FtsW and RodA might be connected to the flipping of lipid II through the membrane (9, 10). MurG, which is responsible for the final cytoplasmic step in the synthesis of lipid II and whose gene is always physically linked to ftsw in gram-negative bacteria, might be another cytoplasmic partner of FtsW. In addition, we found two charged and one polar fully conserved residues in TM4. The localization of these residues in a TM segment suggests that they may play an important role in substrate binding and/or transport.

Further work is needed to examine the potential interaction of the FtsW/RodA/SpoVE family members with penicillin-binding proteins via the large extracellular loop and with FtsZ or MurG via the small cytoplasmic loops.

To propose the model with 10 TM segments depicted in Fig. 1, preference was given to the results of the cysteine experiments rather than to the PhoA fusions. This technique has been used in a large number of topology studies, and several of the models thus obtained have been confirmed, even by X-ray crystallography (5, 30), indicating that the method is valid for a number of proteins. However, different types of anomalous behavior of fusion proteins have been reported (32). The major assumption in the fusion approach is that truncation of a membrane protein does not affect its native topology. In our study, the PhoA fusions gave expected results for the N-terminal half of FtsW but failed to confirm the predicted topology of the C-terminal part. This observation is consistent with the possibility that the correct insertion of the C-terminal TM segments requires interaction of protein segments located downstream of the fusion sites. Deletion of these sequences may prevent or decrease the probability of the correct insertion (31). Indeed, PhoA fusions in the most C-terminal part of FtsW, from Ile³¹⁵ to the C terminus, may be adopting the correct topology, in contrast to fusions from Ala²⁴⁷ to Gly³⁰³. The low PhoA activity of the three C-terminal fusions, compared to their normal expression level, might be interpreted as resulting from the partial insertion of TM10.

In the long periplasmic loop, the PhoA fusions at Thr²²⁵ and Arg²³⁸ are periplasmic, whereas fusions following Ala²⁴⁷ indicated instead a cytoplasmic location of the reporter enzyme. The sequence between Arg²³⁸ and Ala²⁴⁷ thus appears to prevent the translocation of PhoA, which is conceivable considering the chargeless and hydrophobic nature of this segment. This explanation may also account for the anomalous location of the PhoA fusions after Gly²⁶⁷, as the programs TopPredII and TMHMM calculate ambiguous TM probabilities for the segments spanning residues 250 to 261 and 256 to 277, respectively. The membrane insertion of TM8 may tip the scale towards the complete translocation of the long periplasmic loop. The inaccessibility of the thiol introduced at position 265 may also be explained by the hydrophobicity of the adjacent sequences. Indeed, a chargeless hydrophobic segment is likely to be buried within the protein fold.

Because of the ambiguous results with the gene fusion approach, the topological arrangement of the C-terminal part of FtsW was further investigated by a cysteine accessibility analysis. For membrane topological studies, thiol modification has become an alternative to fusion of marker proteins. The advantage of using introduced cysteines is that the native fold is less perturbed by point mutations than by large truncations. Moreover, thiol-specific reagents are available in a variety of coupling chemistries, sizes, hydrophilicities, and labels that can suit diverse experimental situations, such as the level of expression or the availability of a short purification procedure or specific antibodies. Given the difficulty of isolating FtsW, a method was favored whereby thiols are biotinylated, all tagged proteins are isolated by affinity chromatography, and FtsW is revealed by immunoblotting.

Despite the smaller number of cysteine mutations than PhoA fusions, precedence was given to the cysteine results, as the method is less disruptive of the protein integrity and it supports the hydropathy-based prediction. Another instance where the same choice has been made is a study of the Na⁺-citrate transporter CitS (31). The various cysteine accessibility methods are more difficult to implement than the gene fusion techniques, as they require fine tuning of several reaction conditions. Therefore, it can be recommended that an investigation start with gene fusions. However, in case of disagreement with a theoretical prediction, and if reasonable confidence can be given to the prediction (such as if it was performed using several algorithms on multiple homologous sequences), it is not worth multiplying the gene fusions but is wiser to switch to an appropriate cysteine accessibility method.

In conclusion, we have determined the membrane topology of S. pneumoniae FtsW by applying a gene fusion approach in combination with a cysteine accessibility analysis. The model presented is composed of 10 membrane-spanning segments, a large extracytoplasmic loop, and both N and C termini located in the cytoplasm. Based on the sequence alignments of FtsW/RodA/SpoVE proteins, it is likely that the model presented here can be applied to other family members.