Topological characterization of the essential Escherichia coli cell division protein FtsW


  • Beatriz Lara,

    1. Centro de Biología Molecular ‘Severo Ochoa’, Consejo Superior de Investigaciones Científicas, Universidad Autónoma de Madrid, Cantoblanco, 28049 Madrid, Spain
    Search for more papers by this author
  • Juan A Ayala

    Corresponding author
    1. Centro de Biología Molecular ‘Severo Ochoa’, Consejo Superior de Investigaciones Científicas, Universidad Autónoma de Madrid, Cantoblanco, 28049 Madrid, Spain
    Search for more papers by this author

*Corresponding author. Tel.: +34 (91) 3978083; Fax: +34 (91) 3978087, E-mail address:


The membrane topology of Escherichia coli FtsW, a 46-kDa essential protein, was analyzed using a set of 28 ftsW–alkaline phosphatase (ftsW–phoA) and nine ftsW–β-lactamase (ftsW–bla) gene fusions obtained by in vivo and in vitro methods. The alkaline phosphatase activities or resistance pattern of cells expressing the FtsW–PhoA or FtsW–Bla fusions confirmed only eight out of 10 transmembrane segments predicted by computational methods. After comparison with the recent topology of Streptococcus pneumoniae FtsW, we could identify all the fusions in absolute agreement with the predicted model: N-terminal and C-terminal ends in the cytoplasm, 10 transmembrane segments and one large loop of 67 amino acids (E240–E306) located in the periplasm.


The ftsW gene[1] mapped between murD and murG at the dcw cluster was originally identified and sequenced by Matsuhashi's group[2]. The E. coli FtsW amino acid sequence showed a high homology to the Bacillus subtilis SpoVE and the Escherichia coli RodA proteins that are required for asymmetric division during sporulation and for elongation and maintenance of the rod shape during cell growth, respectively [3,4]. Based on these similarities, it has been suggested that these three membrane proteins have similar roles in cell elongation (RodA), cell division (FtsW), and spore formation (SpoVE), respectively. Moreover, a signature pattern located in the C-terminal region has been described for these cell cycle proteins FtsW/RodA/SpoVE in the PROSITE database (PROSITE: PDOC00352, PS00428) that identified 58 homologous proteins (update of January 2002).

It has been predicted that FtsW might interact with PBP3 in septum formation [1,5]. Membranes prepared from cells overproducing both PBP3 and FtsW showed an increased synthesis of peptidoglycan as well as of undecaprenol-linked intermediates in vitro[6]. It has been shown that the putative enhancer function of FtsW in the transglycosylase activity of PBP3 is not reproducible[7]. However, the increased formation of lipid-linked intermediates suggested that FtsW might function in supply of the lipid-linked precursor to PBP3 for the formation of the septal peptidoglycan.

Genetic analysis of two thermosensitive alleles of the gene (ftsW 201[8] and ftsW 263[9]) revealed that these mutations caused a division block at an earlier stage than that caused in strains carrying the ftsZ 84 allele, and suggested a role of FtsW at the initiation stage of cell division. However, another allele of the gene (ftsW 1640) caused the division block later on[9], which suggests some role of FtsW at a late step of septum formation. It has been proposed that the early function of FtsW in cell division is the establishment of a stable FtsZ structure, and that the late function is a link between the events in the cytoplasm (FtsZ-ring formation) and the initiation of septal peptidoglycan biosynthesis in the periplasm[10]. By immunofluorescence microscopy, FtsW has been shown to be located in the middle of cell[11]. Septal localization of FtsW required FtsZ, FtsA, FtsQ, and FtsL but not FtsI. Thus, FtsW is a late recruit to the division site and is essential for subsequent recruitment of its cognate transpeptidase FtsI (PBP3) but not for stabilization of FtsZ rings. It has been suggested that a primary function of FtsW homologs is to recruit their cognate transpeptidases to the correct subcellular location[12].

An important step to elucidate the presumable double function of this protein is to establish with certainty its topology in the membrane. Based on the combined assays of alkaline phosphatase fusion and cysteine accessibility techniques the topology of Streptococcus pneumoniae FtsW has been recently described[13]. Here we present the results obtained by using ftsW–phoA and ftsW–bla gene fusions in E. coli FtsW.

2Materials and methods

2.1Bacterial strains and growth conditions

The bacterial strains used were E. coli CC118 (araD139, Δ(ara, leu)7697, ΔlacX74, ΔphoA20, galE, galK, thi, rpsE, rpoB, argE (am), recA1), CC202 (F42, lacI3, zzf2::TnphoA/CC 118) and TG1 (F′traD36 lacIqΔ(lacZ)M15 pro A+B+/supE Δ(hsdM-mcrB)5 (rkmkmcrB) thi Δ(lac-proAB) that were grown in the Luria broth (LB) medium at 37°C. Antibiotics were used as follows: ampicillin, 100 μg ml−1, and kanamycin, 30 μg ml−1 in liquid media and 300 μg ml−1 in solid media.

5-Bromo-4-chloro-3-indolylphosphate disodium salt (XP) (Sigma) was added at a concentration of 40 μg ml−1. p-Nitrophenyl phosphate (Sigma 104), a substrate for alkaline phosphatase assay, also was from Sigma.

The ECF substrate (Amersham®) from the ECF (enhanced chemifluorescence) signal amplification module was purchased from Amersham and used at a concentration of 12 μg ml−1 for detection of the phosphatase activity of the fusions in intact cells.

2.2Isolation of in vivo ftsW–phoA fusions

The strain CC202 was transformed with pBL1, a pUC19 derivative plasmid containing a fragment of the murD gene and the entire ftsW gene. As described by Manoil and Beckwith[15], transformation was plated on a medium containing ampicillin (100 μg ml−1), XP (40 μg ml−1), and a high concentration of kanamycin (300 μg ml−1) to isolate the colonies, in which transposition of TnphoA into the multicopy plasmid occurred. After incubation at 37°C, blue colonies were selected, and plasmid DNA was purified and used to transform the strain CC118. Selection was done on the same plates, and plasmid DNA was prepared and characterized by restriction analysis and sequencing by standard methods[14].

2.3Construction of gene fusions by ExoIII-nested deletions

Plasmid pJS521 (a gift of G.R. Jacobson,[16]) was first digested with Pst I and Bam HI to clone the phoA gene in pUC19 (we called this plasmid pBL8). pBL8 was treated with Hin dIII, the 3′-recessed ends were filled in by using Klenow, and the subsequent digestion with Sac I removed the phoA-containing fragment. This fragment was then ligated with pBL1 that was cut at the 3′ end of the ftsW gene (codon 402) with Mlu I, filled in, and then digested with Sac I. The new plasmid was called pBL10 and was used to construct a series of fusion plasmids by nested deletions from the 3′ end of the ftsW gene. Exonuclease III (ExoIII) digests linear DNA with 5′ overhanging ends or blunt ends, whereas 3′ overhanging ends are resistant to the nuclease action. Then 7 μg of pBL10 was cut with Sal I and placed within codon 388 to create a 5′ overhanging end that could be used as a substrate for the ExoIII; digestion with Pst I that creates a 3′ overhang at the codon 27 (proline) of phoA was used to protect bases from ExoIII. The digested DNA resuspended in 40 μl of ExoIII buffer was then incubated with 1 μl of ExoIII at 37°C. Aliquots of 2.5 μl were taken every 15 s, and placed on ice in tubes containing 2.25 units of S1 nuclease. When all aliquots had been taken, tubes were incubated for 30 min at 30°C. Then 1 μl of S1 stop mixture (0.3 M Tris–HCl, 50 mM EDTA, pH 8) was added, and the tubes were heated for 10 min at 70°C. Klenow (0.125 units per tube) was used to fill in ExoIII-digested DNA in the presence of deoxynucleotides (0.05 mM), and blunt ends were ligated overnight at room temperature. DNA was transformed into TG1 cells, and transformants were selected on LB plates containing ampicillin (100 μg ml−1) and XP (40 μg ml−1). A restriction analysis was first made, and in frame fusions were confirmed by sequencing.

2.4In vitro construction of ftsW–phoA and ftsW–blaM fusions

To complete FtsW topology, we made four in vitro fusions at different amino acid positions of interest, which define the topology more precisely. Fusion A82 was made by digesting pBL10 with Nru I (codon 82) and Pst I, each with a unique restriction site in the plasmid. Protruding ends were made blunt by the Klenow enzyme, were ligated together, and DNA was transformed first into TG1 cells and then into CC118.

Fusions T38, P375, and L403 were constructed by replacing the ftsW gene in pBL10 with a polymerase chain reaction (PCR) fragment coding the desired region to be involved in the fusion. Three fragments, 600, 1590, and 1680 bp long, were first amplified, using oligonucleotides A51 (5′-ACGCCAAGCTTGCATGCCGGG-3′) and A52 (5′-CATGATCAGGCTGCAGGTATCTTTTTCCCG-3′), A54 (5′-ACCGTAACTGATCTGCAGCAATGTCAGACC-3′), A53 (5′-GCCTGCGCTTTCTGCAGACGCGTTTCA-3′), respectively, that contain a Hin dIII and a Pst I site, (underlined). This PCR fragments was cloned after restriction in pBL10 also cut with Pst I and Hin dIII, creating an in frame fusion with phoA at codons 40 (fusion T38), 375 (fusion P375), and 403 (fusion L403) of the ftsW gene.

To generate β-lactamase fusions we followed the method described by[17]. Various truncated forms of the ftsW gene (at the 3′ end) were amplified by PCR from the E. coli chromosomal DNA using as primers oligonucleotides generating a Kpn I site. The oligonucleotide A60 (5′-CGTACCAGGTACCGCCTGTGCCAGCAGTAACACTGC-3′) used as the primer for the 5′ end of the gene in all constructions introduced a Bam HI site upstream from the initiation codon of ftsW, within the sequence of the preceding murD gene[18]. The generated DNA fragments were purified, cut with both Bam HI and Kpn I enzymes and then inserted between the same sites of the pNF150 vector. In each case, the in frame insertion resulted in the expression of a truncated form of the ftsW gene product fused to the mature form of β-lactamase (HPETLVKVK_) via a short intermediate peptide linker AVPHAISSSPLR originating from the sequence of the pNF150 plasmid vector. In all cases, the site of fusion junction was confirmed by DNA sequencing.

2.5DNA sequence analysis of fusions

Double-stranded ftsW–phoA or ftsW–blaM fusion plasmid DNA was sequenced according to the Sequenase version 2.0 sequencing protocol (United States Biochemicals) by the chain-terminator method[19]. The sequencing primer hybridizing at the 5′ end of ′phoA (5′-CCCCCATCCCATCGCCAATCA-3′) was obtained from Isogen. Deduced amino acid sequences were analyzed by using the WISCONSIN computer programs and molecular biology services accessible on the World Wide Web: ExPASy (

2.6Alkaline-phosphatase activity of fusions

Activity of in frame fusions was measured as described by Beckwith[20]. Overnight cultures were diluted 100-fold in the LB medium and the appropriate antibiotic and allowed to grow at 37°C to OD550= 0.5. The cells were harvested and resuspended in the same volume of 1 M Tris–HCl, pH 8. Phosphatase activity was assayed by measuring the rate of p-nitrophenyl phosphate hydrolysis.

The specific activity of the fusion proteins was determined as the ratio of the measured phosphatase activity to the amount of each fusion protein detected, at the exponential state of growth, as described below.

We developed a new assay to visualize the phosphatase activity in intact cell, using a fluorescent substrate of the Amersham ECF signal amplification module. This substrate is a phospho-derivative that, when cleaved by phosphatase, allows detection of fluorescence in the cell envelope by fluorescence microscopy as well as the quantification of the activity in a spectrofluorometer (AMINCO Bowman series 2 luminescence spectrometer, excitation at 420 nm, emission at 560 nm). One ml of the appropriate culture (109 cells ml−1) was centrifuged for 10 min at 4°C in an Eppendorf microfuge, 3000 rpm. Cell pellet was resuspended in sodium bicarbonate 0.1 M buffer, pH 7.5, washed once by centrifugation under the same conditions as above, and finally resuspended in 1 ml of the same buffer. The ECL substrate (Amersham®) was added to a final concentration of 12 μg ml−1, and incubated for 10 min at 4°C. Then 10 μl of the cell suspension were placed on a slide of agarose for observation and photography in a fluorescence microscope (Olympus System Microscope BX50-FLA and Olympus Photomicrography System PM10AD). The rest was centrifuged for 10 min at 4°C at 15 000 rpm to separate bacteria. Supernatant was used directly to measure the fluorescence emission at 560 nm, with an excitation wavelength of 420 nm and a detector high voltage of 535 V.

2.7Quantification of the expression of FtsW–PhoA fusions

The level of expression of each individual fusion protein was quantified by immunoblotting of total exponential cell culture. The CC118 strains containing plasmids harboring in frame fusions were grown until late mid-exponential phase. Total proteins were separated in the 10% acrylamide SDS–PAGE gel and transferred by blotting to Immobilon paper. The PhoA fragment in the fusion protein was detected by hybridization of the membrane with a polyclonal anti-alkaline phosphatase antiserum and revealed by the chemiluminescence method. Bands corresponding to the fusion proteins were quantified (arbitrary units) by densitometric analysis of the film, using a Mustek 1200SP scanner system and the software imaging-analysis TINA version 2.09e.

2.8Ampicillin resistance of cells expressing β-lactamase fusion proteins

The ampicillin resistance of individual cells of DH5α carrying the various ftsW–blaM plasmids listed in Table 1 was determined by plating appropriate dilutions of exponential phase cultures onto 2YT plates containing 0, 30, 60, 100 or 200 μg ml−1 ampicillin. Growth was observed after 24 h of incubation at 37°C, and the resistance was estimated as the ratio of colony numbers on ampicillin plates to colony numbers on control plates.

Table 1.  Specific activity of the FtsW–PhoA fusions obtained by two different methods
  1. aCodon of ftsW into which ′phoA is fused and the amino acid change produced, PVL indicates three first amino acids of PhoA, and [*] (DSYTQVASWTEPFPFC) is the amino acid sequence of the stretch introduced by the Tnpho A derived from the IS50L insertion sequence.

  2. bThe AP activity is the mean of two to five values from independent determinations.

  3. cThe arbitrary unit obtained by densitometric analysis of the film, as described in Section 2.

  4. dIntensity of the yellow color of the cells after 10 min at 4°C in the presence of the ECL substrate, as described in Section 2; − colorless, +/− very pale, + pale, ++ intense, +++ very intense.

  5. eArbitrary units of the fluorescence intensity of the supernatants measured at 560 nm.

FusionMethod of constructionJunctionaAP activity units/ODbAmount detected with anti-PhoAcSpecific AP activityFluorescence
L100Tnpho AL100P-[*]-PVL52.71.8513.5
L121Tnpho AL121P-[*]-PVL3<0.115.0
V165Tnpho AV165A-[*]-PVL33.220.932.50
V233Tnpho AV233A-[*]-PVL252.88.93++38.05

2.9Transmembrane segment prediction

The following available programs on the World Wide Web were used to predict the membrane topology of FtsW: Tmpred, TMAP, SOSUI, DAS and pHDtopology accessible on the World Wide Web: ExPASy ( We used 58 homologous proteins to FtsW identified in COG0772 database at the National Center for Biological Information (

3Results and discussion

3.1Prediction of the membrane topology of E. coli FtsW

The hydrophatic profile of E. coli FtsW, previously identified as a membrane protein, revealed three large hydrophobic regions (M1-V132, P181-A239, and L307-R414) separated by two hydrophilic regions of 48 (K133-K180) and 67 (E240-E306) residues, respectively. We could predict that the first N-terminal hydrophobic region contained three or four membrane-spanning segments, the central hydrophobic region might contain from one to three; and the third C-terminal hydrophobic region contained three such segments. The two remaining parts of the protein show a high degree of hydrophilicity and should represent water-soluble domains (Fig. 1A).

Figure 1.

Predicted transmembrane segments of FtsW by different methods and the predicted and proposed models. A: Open bars represent transmembrane or uncharged segments. The prediction methods DAS, SOSUI, TMpred, TMAP, and pHDtopology are described in the text. The HYDROPHOBIC prediction was obtained by visual inspection of the sequence and represents segments of at least 14 uncharged residues, flanked by charged amino acids. Ratios of positive to negative charged residues on the hydrophilic segments of the HYDROPHOBIC prediction (upper line) and pHDtopology prediction (lower line) are shown. Marks + and − indicate the type of residue bordering the hydrophobic segments (+, R or K; −, D or E). B: Predicted topological model of E. coli FtsW. C: Proposed model of E. coli FtsW. Numbers on the transmembrane segments correspond to those displayed in panel A.

The prediction of the secondary structure of the polypeptide chain by the Chou and Fasman[21] algorithm confirmed that the most hydrophobic regions had a highly α-helical nature. We also used five computer programs available on the Word Wide Web (see Section 2) to predict transmembrane segments. All five programs gave similar, but not identical, predictions, ranging from nine to 11 transmembrane segments (TM) for E. coli FtsW, which are summarized in Fig. 1A. A preliminary topological model (called 10-TMS model) suggested by most of the prediction methods contains 10 transmembrane segments, 0 (L13-S33), 1 (L48-M68), 2 (V87-W107), 3 (A112-V132), 5 (P181-A194), 6 (L198-G216), 7 (W220-A239), 8 (L307-F324), 9 (G342-A364), 10 (T373-L395), two large hydrophilic loops, 3/5 (K133-K180), 7/8 (E240-E306) at the cytoplasm and periplasm, respectively, and both the N-terminus and C-terminus facing the cytoplasm (Fig. 1B).

To check validity of this model, we used alkaline phosphatase and β-lactamase as topological reporters for analysis of topology of FtsW in the membrane.

3.2Comparison with other putative FtsW proteins

We first performed the search on the SWISS-PROT, release 40.12, TrEMBL, release 19.10 databases with the BLAST program, using the sequence from E. coli ftsW gene. One hundred and six homologous genes corresponding to putative FtsW or RodA proteins were identified by this method, and also several genes by searching on the Web page several other unfinished sequence genome projects ( Also, by HMM search at the TIGR databases (, 83 homologs to E. coli FtsW corresponding to 42 species and 32 major phylogenetic lineages were identified, and the COG0772 database at the National Center for Biological Information ( contains 58 homologous proteins corresponding to 30 different species and 18 major phylogenetic lineages. Comparison analysis of the amino acid sequence with the PILE-UP and DIVERGE programs clearly separated those with the highest homology to FtsW protein from those with the strongest homology with RodA and divided the phylogenetic tree into two sections (data not shown). This separation on the phylogenetic tree of the two families of homologous proteins clearly indicates that, although the two proteins (FtsW and RodA) must have quite similar functions, they have remained as individual entities in the course of evolution. This analysis also indicated that the gene products with the strongest homology with E. coli FtsW are currently located in a cluster of conserved cell division genes (dcw cluster), in which there was also ftsZ or mraW. In those species where a single FtsW/RodA homolog is found (Buchnera sp. APS TOKIO 1998, Cyanophora paradoxa (Cyanella), Deinococcus radiodurans R, Neisseria gonorrhoeae FA1090, Neisseria meningitidis Z2491, N. meningitidis MC58 and Synechocystis sp. PCC6803) or where there was a third homolog (Bacillus halodurans C-125, Bacillus subtilis 168, Enterococcus faecalis V583, Lactococcus lactis IL1403, and Streptomyces coelicolor A3(2)) all those proteins had a higher homology to FtsW than to RodA.

Then we performed a comparison analysis among the proteins, using the CLUSTAL-W program[22]. Although the primary structures of the transmembrane segments greatly diverge among orthologues (with the exception of Ec-TM10), most of the predicted transmembrane segments and loops are well conserved, in terms of hydrophobicity, charge, number, and size, among the 29 orthologues FtsW-like proteins (data not shown). This suggests that general topology of all these proteins should be identical.

3.3Activities of FtsW–PhoA and FtsW–BlaM fusion proteins

In this work, we have applied a gene-fusion method[23] to extensively investigate the membrane topology of the essential E. coli cell division protein FtsW. The fusion proteins and their activities are listed in Tables 1 and 2. When activities of these fusion proteins were superimposed on the predicted 10-TMS model of FtsW, the results obtained turned out to be incompatible with the predicted topology (Fig. 1B). The reversed orientation of the Ec-TM1, Ec-TM2 and Ec-TM3 segments, the predicted orientation for Ec-TM9 and Ec-TM10, and two large periplasmic and cytoplasmic loops were more strongly confirmed by the phoA fusion results. To further characterize expression and localization of the PhoA moiety of the fusion proteins, we developed a method to visualize the alkaline phosphatase (AP) activity by fluorescence microscopy and to measure the activity by a fluorescence emission of a cleaved substrate (ECF fluorescence substrate of the Amersham's ECF signal amplification module, see Section 2). After cleavage, the substrate can diffuse to the periplasm and immediately outside the cell, and only after 90–100 min it could also be found in the cytoplasm. All those fusions with a fluorescence emission of at least 10.00 (arbitrary units) in the supernatant developed a fluorescence pattern on the cell envelope. Fig. 3 shows the current pattern of fluorescence observed with FtsW–PhoA fusion protein I146.

Table 2.  Construction of the ftsW–blaM gene fusions and response of fusion-containing cells to ampicillin
  1. aRestriction sites for Bam HI (GGATCC) and Kpn I (GGTACC) that were introduced in oligonucleotide sequences are indicated in bold.

  2. bThe junction site corresponds to the C-terminal amino acid of the truncated FtsW sequence, which has been fused to the mature form of βlactamase, via an intermediate peptide linker consisting of AVPHAISSSPLR (originating from the pNF150 vector sequence).

  3. cThe DH5αE. coli strains harboring plasmids bearing fusions between complete or truncated ftsW from E. coli and blaM were grown as 100–200 separate single colonies on plates containing ampicillin at 100 μg ml−1 and kanamycin at 50 μg ml−1. After 24 h of incubation at 37°C, strains growing in the presence of 100 μg ml−1 ampicillin were considered as resistant (R) and otherwise sensitive (S). Strains harboring plasmids pFTSWB2 and pFTSWB3 also grew at 200 μg ml−1. Strains harboring plasmids pFTSWB1, pFTSWB4, pFTSWB5, pFTSWB6, pFTSWB7, pFTSWB8 and pFTSWB9 did not grow at 30 μg ml−1.

PlasmidOligonucleotideaJunction sitebResponse to ampicillinc
 3′ end PCR primer  
 5′ end PCR primer  
Figure 3.

Detection of the AP activity in the cell envelope by fluorescence microscopy. One ml of an LB culture (109 cells ml−1) of the strain CC118/pBL10(I146) was centrifuged in an Eppendorf microfuge. Cell pellet was washed and resuspended in the 0.1 M sodium bicarbonate buffer, pH 7.5, and the ECL substrate (Amersham®) was added and incubated for 10 min at 4°C. Then 10 μl of the cell suspension were placed on a slide of agarose for observation by phase contract (A) and fluorescence (B) microscopy. The bar in panel A represents 2 μm.

A high AP activity of fusion R73 (279 units/OD) and A82 (119 units/OD) clearly indicates that loop 1/2 (13 residues) is located in the periplasm, while the low activity of fusion T38 (15 units/OD) suggests that loop 0/1 (13 residues) must be in the cytoplasm. According to the positive-inside von Heijne's rule[24], topology of these loops is in a good agreement with the ratio of positive to negative charge of 3:2 and 8:4, respectively. These data confirm the idea that Ec-TM1, and not Ec-TM0, could be the first topological signal for insertion of FtsW and the first anchoring segment to the membrane. The homology comparison analysis of the N-terminal region of E. coli FtsW (M1-R46) with 29 FtsW-homologous proteins reveals no homologous hydrophobic sequence to the first predicted transmembrane segment (Ec-TM0, L13-S33), so we postulate that this region should remain in the cytoplasm. Moreover, a FtsW form of a lower molecular mass (FtsWS) has a start of translation (M42) after this segment, and this form seems to be completely functional in vivo[11], which indicates no essential role of this region, nor topogenic determinant for Ec-TM0. However, we cannot exclude that this region, in the case of E. coli FtsW, could be associated with the membrane after insertion of the protein.

The overall topology and midpoint of the actual transmembrane region Ec-TM3 in Fig. 4 was inferred from the activity results, such as an increasing AP activity from the N- to the C-terminal of the segment in fusions A112, (0 unit/OD); S118, (3 units/OD); L121, (3 units/OD); L126, (73 units/OD); and S130, (133 units/OD). This is also confirmed by the published evidence that as few as about 10 residues of an ‘outgoing’ transmembrane segment are sufficient to promote export when fused to AP, whereas 10 residues of an ‘in-coming’ transmembrane region usually are insufficient to translocate AP into the cytoplasm[25]. This effect of the insertions on the potential location of the catalytic activity can be observed in the case of fusions A112, S118, and L121 that clearly disrupt the transmembrane segment, while fusions L126 and S130 allow export of the PhoA moiety. If so, the loop 2/3 (five residues) and loop 3/4 (11 residues) should be located in the cytoplasm and in the periplasm, respectively.

Figure 4.

Membrane topology model of FtsW from E. coli. Predicted topological model derived from AP activity and β-lactam resistance of the fusion proteins and by comparison with the predicted model of FtsW of S. pneumonia. The primary amino acid sequence is given in the single-letter code and in black characters surrounded by a circle, a square or a diamond. Residues in squares indicate a conserved position in at least 60% of the FtsW homologs, while those in diamonds are conserved in more than 90% of the FtsW homologs proteins. Plus and minus signs indicate positively and negatively charged amino acids, respectively. Amino acids corresponding to predicted transmembrane segments are in gray. The positions (in white characters) of the PhoA and BlaM fusions are superimposed to the model. Fusions are indicated according to the AP activity or Bla resistance as black (active or resistant) or gray (inactive or sensitive) background. Positions of the known amino acid change mutations are also indicated: E170K, P181L, and P253L. Conserved residues among the sequence of FtsW from E. coli and S. pneumonia are highlighted by a bold line and white background.

The central region of the protein contains three potential transmembrane segments. However, if we consider that the loops should have at least five residues, the predicted loop 5/6 is too small, and also the length of Ec-TM5 is rather short. The high AP activities of fusion proteins from S130 (specific AP activity 9.70) to T207 (specific AP activity 4.39) allowed us to think that this entire segment should be in the periplasm. However, the resistance pattern of strains harboring plasmids pFTSWB1 (sensitive), pFTSWB2 (resistant) and pFTSWB3 (resistant), permits us to consider the Ec-TM5 as an actual transmembrane segment with the predicted orientation (see Fig. 1). Comparison with the corresponding transmembrane segment (Sp-TM5) in S. pneumoniae (Fig. 2) confirms this assumption and allows to extend the length of the segment up to G177. In this context, a new transmembrane segment Ec-TM4, only predicted by pHDtopology, has been identified by AP fusions I146 (AP activity 226) and V165 (AP activity 3). Considering those AP activities, this segment should have the ‘out-in’ orientation and it must correspond to Sp-TM4 displayed in Fig. 2. High AP activity of fusion R172 (AP activity 60) can be explained by the high charged segment preceding the fusion point that should prevent Ec-TM4, already a charged segment, to function as in-going transmembrane segment.

Figure 2.

Sequence alignment of FtsW from S. pneumoniae and E. coli. The alignment was obtained by the CLUSTAL-W program. Identical or homolog amino acids are marked with a vertical bar and two dots, respectively. Predicted transmembrane segments for S. pneumoniae (Sp-TM) and E. coli (Ec-TM) are marked as black bars under each sequence. Open (inactive) or black (active) triangle shows the positions of PhoA fusion. The positions of Bla fusion are shown as open (sensitive) or black (resistant) diamond. For comparison, the positions of the PhoA fusions recently described in S. pneumoniae FtsW are also shown.

The predicted topology for Ec-TM6 and Ec-TM7 could be confirmed by the activity and resistant pattern of AP fusions V202 (AP activity 118), T207 (AP activity 97), V233 (AP activity 25) and BlaM fusions on plasmid pFTSWB4 (A217) (sensitive) and pFTSWB5 (L219) (sensitive). These results corroborate the actual orientation ‘out-in’ of Ec-TM6 and ‘in-out’ of Ec-TM7 and are in agreement with the corresponding transmembrane segment of S. pneumoniae, Sp-TM6 and Sp-TM7, respectively.

At the C-terminal of the protein there is a region from E240 up to F340, where all the fusions, both PhoA and BlaM, give low AP activity or sensitive colonies. These results could be interpreted as that the whole region is located in the cytoplasm, but this topology is at odds with the rest of model. The major assumption in the fusion approach is that truncation of the membrane protein does not affect its native topology. However in our analysis, as it has been already shown in the case of S. pnemoniae FtsW[13], the PhoA and BlaM fusions failed to confirm the predicted topology of this C-terminal region. This remark is consistent with the possibility that the correct insertion of the loop 7/8 and transmembrane Ec-TM8 is dependent on or requires interactions of protein segments located downstream of the fusion sites. In this sense there are two cysteine residues, one located at the cytoplasmic inlet within Ec-TM4 (C158) and the other in Ec-TM9 (C346), that could establish a covalent link between these two transmembrane segments and be required for stabilization. Deletion of these sequences may prevent or decrease the probability of the correct insertion. Indeed, PhoA fusions in the most C-terminal part of FtsW, from Q355 to the C-terminus, may be adopting the correct topology. Thus, the periplasmic location of loop 7/8 and the right prediction for Ec-TM8 have been inferred from the high homology of this region (50% identity, 71% homology) with the corresponding segment (loop 7/8 and Sp-TM8) of the published topology of S. pneumoniae FtsW[13].

In the last segment of the protein (F340-R414) the high AP specific activity of fusion P375 (9.16) allows us to consider the right prediction of the loop 9/10 in the periplasm, whereas the low activity of fusion L403 (0.75) predicts cytoplasmic location of the C-terminal end of the protein. Location of the C-terminal end of the protein in the cytoplasm also conforms to the positive-inside von Heijne's rule (ratio 5:3)[24]. Considering the cytoplasmic location of loop 8/9 and the C-terminal end, and the periplasmic location of loop 9/10, the predicted ‘in-out’ orientation of transmembrane Ec-TM9 and the ‘out-in’ orientation of transmembrane Ec-TM10 are expected. This topology for Ec-TM9 and Ec-TM10 is also confirmed by a low fluorescence of fusion Q355 (0.21) and a relatively high fluorescence of fusion M387 (30.84). A high activity of the last fusion M387 indicates disruption of the ‘in-going’α-helix of Ec-TM10, which allows the PhoA moiety to remain in the periplasm.

When all the fusions were analyzed by the in vivo fluorescence method, the data correlated precisely (see Table 1) with those obtained by the colorimetric method. However, a lower activity for fusion Q355 (0.21) was found, which may result from a disruption of the out-going α-helix Ec-TM9 retaining PhoA in the cytoplasm. A higher AP specific activity detected for this fusion by the colorimetric method was actually due to the low level of the fusion detected by immunoblotting of the total cell extract, as the AP activity also was low.

From these data, we propose a topological model with 10 transmembrane segments and a large periplasmic loop (Fig. 4). The topological model would have the following structure: transmembrane segments (Ec-TM1: T47-A66, Ec-TM2: G86-M104, Ec-TM3: Y110-G129, Ec-TM4: L141-N162, Ec-TM5: G177-Q195, Ec-TM6: T200-G216, Ec-TM7: W220-A239, Ec-TM8: L307-F324, Ec-TM9: F343-A364, and Ec-TM10: L374-L394), cytoplasmic segments (N-terminal: M1-R46, loop 2/3: E105-R109, loop 4/5: Y163-R176, loop 6/7: A217-L219, loop 8/9: R325-G342, C-terminal: L395-R414) and periplasmic loops (loop 1/2: S67-D85, loop 3/4: S130-D140, loop 5/6: P196-G199, loop 7/8: E240-E306, loop 9/10: G365-T373]. Although this proposed model (Figs. 1C and 4) is at odds with the results of hydrophobicity analysis, which predicts some transmembrane segments not present in our final model (Ec-TM0) and the reversed orientation of some others (Ec-TM1, Ec-TM2 and Ec-TM3), the consistency of the properties of the fusions (steady-state levels, PhoA activities and BlaM resistance, and fluorescence on the cell membrane), as well as the high correlation of our results with those previously published for S. pnemoniae FtsW (see Fig. 2), allow us to be confident about the model. These data suggest a general similar topology for all member of the FtsW/RodA/SpoVE family.


This work was supported by grants BIO97-0665 and PB97-1193 from the Dirección General de Enseñanza Superior e Investigación Científica (Ministerio de Educación y Cultura, Spain). The financial support of Fundación Ramón Areces to the Centro de Biología Molecular ‘Severo Ochoa’ is greatly acknowledged.