This Article Abstract Full Text (PDF) Supplemental material Other Versions of this Article: JB.01322-07v1 190/5/1792    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Google Scholar Articles by Bendezú, F. O. Articles by de Boer, P. A. J. PubMed PubMed Citation Articles by Bendezú, F. O. Articles by de Boer, P. A. J.

Conditional Lethality, Division Defects, Membrane Involution, and Endocytosis in mre and mrd Shape Mutants of Escherichia coli ^,

Case Western Reserve University, School of Medicine, Department of Molecular Biology and Microbiology, Cleveland, Ohio 44106

Received 15 August 2007/ Accepted 26 October 2007

	ABSTRACT

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Maintenance of rod shape in Escherichia coli requires the shape proteins MreB, MreC, MreD, MrdA (PBP2), and MrdB (RodA). How loss of the Mre proteins affects E. coli viability has been unclear. We generated Mre and Mrd depletion strains under conditions that minimize selective pressure for undefined suppressors and found their phenotypes to be very similar. Cells lacking one or more of the five proteins were fully viable and propagated as small spheres under conditions of slow mass increase but formed large nondividing spheroids with noncanonical FtsZ assembly patterns at higher mass doubling rates. Extra FtsZ was sufficient to suppress lethality in each case, allowing cells to propagate as small spheres under any condition. The failure of each unsuppressed mutant to divide under nonpermissive conditions correlated with the presence of elaborate intracytoplasmic membrane-bound compartments, including vesicles/vacuoles and more-complex systems. Many, if not all, of these compartments formed by FtsZ-independent involution of the cytoplasmic membrane (CM) rather than de novo. Remarkably, while some of the compartments were still continuous with the CM and the periplasm, many were topologically separate, indicating they had been released into the cytoplasm by an endocytic-like membrane fission event. Notably, cells failed to adjust the rate of phospholipid synthesis to their new surface requirements upon depletion of MreBCD, providing a rationale for the "excess" membrane in the resulting spheroids. Both FtsZ and MinD readily assembled on intracytoplasmic membrane surfaces, and we propose that this contributes significantly to the lethal division block seen in all shape mutants under nonpermissive conditions.

	INTRODUCTION

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
As in many other bacterial species, maintenance of cell shape in Escherichia coli requires an intact peptidoglycan (murein) layer of the envelope and, at least, five well-conserved proteins (PBP2, RodA, MreB and -C, and D) that are each required to prevent a cell shape conversion from rod to prolate spheroid or sphere (for reviews, see references 9, 13, 14, 40, 55, and 90).

PBP2 and RodA are encoded by the mrdA (pbpA) and mrdB (rodA) genes residing in the mrd (murein D) operon (75, 76). Penicillin binding protein 2 (PBP2) is a bitopic integral cytoplasmic membrane (CM) species with a large periplasmic domain that possesses murein DD-transpeptidase activity, binds β-lactams, and has a particularly high affinity for the amidino-penicillin amdinocillin (mecillinam) (FL1060) (42, 74). PBP2 is unique among the PBPs in E. coli in that it is specifically required for cylindrical murein synthesis during cell elongation but dispensable for septal murein synthesis during cell constriction. Its counterpart, PBP3 (FtsI), is similarly unique in that it is specifically required for septal murein synthesis but dispensable for cylindrical murein synthesis (28, 40, 74). RodA belongs to a family of polytopic membrane proteins which also includes the division protein FtsW (39, 56), and evidence suggests that RodA and FtsW are required for proper functioning of PBP2 and PBP3, respectively (31, 40-42, 58).

MreB, -C, and -D are encoded by the mre operon. MreB is the sole known bacterial actin (79) in E. coli. As in Bacillus subtilis (26, 46, 73) and Caulobacter crescentus (32, 35), MreB localizes just underneath the CM in a spiral/banded-like pattern along the length of the cell (50, 71) and is implicated in both shape maintenance and chromosome segregation (48, 85). MreC is a bitopic CM protein and MreD a polytopic one (49, 53, 84). A crystal structure of the large periplasmic domain of Listeria monocytogenes MreC revealed a dimer with some structural similarities to chymotrypsins, though it is unlikely to be a protease (80). Affinity purification and bacterial two-hybrid analyses indicate that MreC interacts with MreD as well as with several of the high-molecular-weight murein synthases (PBPs), including the PBP2 homologues in C. crescentus and B. subtilis (29, 49, 80). An interaction between MreC and PBP2 is further supported by colocalization experiments with C. crescentus and Rhodobacter sphaeroides cells (30, 72). Like MreB, PBP2, MreC, and MreD appear to accumulate in a spotty or helical fashion along the cell envelope in E. coli, B. subtilis, and/or C. crescentus (27, 29, 30, 32, 52), and these localization patterns are reminiscent of those of new murein insertion in the cylindrical portion of B. subtilis cells (22, 78). These and other observations support models in which the helical organization of bacterial actins in the cytoplasm topologically constrain murein synthase and/or hydrolase activities in the periplasm, resulting in growth of the sacculus as a cylinder in between periods of cell constriction (15, 22, 32). How the location of MreB polymers in the cytoplasm is coupled to that of (mostly) periplasmic MreC/PBP complexes is unclear. Coupling could be quite direct in E. coli, as suggested by an MreC-MreB interaction in a bacterial two-hybrid assay (49), but this is probably not the case in C. crescentus and R. sphaeroides (30, 72).

How loss of rod shape affects the ability of E. coli to propagate has been a confusing issue. Loss of the Mre proteins has variously been reported to yield stably propagating spheres (50, 61, 70, 84) or to be lethal (49, 86), unless cells are supplied with extra copies of ftsQAZ (49). Inactivation of PBP2 and/or RodA, by treatment of wild-type (wt) cells with amdinocillin or by mutation of mrdA or mrdB, typically results in the formation of giant nondividing spheroids/spheres, which eventually lyse and die (5, 62, 76, 82). However, PBP2^– cells were found to stably propagate as smaller dividing spheres under several conditions, including (i) simultaneous increases in the essential division proteins FtsQ, -A, and -Z (59, 82); (ii) a low growth rate (4, 47); and/or (iii) an increase in the level of the alarmone ppGpp above a certain threshold (12, 47, 81). RodA^– spheres were similarly reported to survive on poor medium or upon overexpression of ftsQAZ. (5, 28, 82).

Taking care not to select for secondary suppressing alterations, we created sets of mre and mrd mutants in two genetic backgrounds and compared their properties. Our results indicate that unsuppressed cells lacking either of the Mre proteins behave very similarly to those lacking PBP2 and/or RodA. Thus, like mrd cells (28, 47, 82), mre cells were conditionally viable in that they propagated stably as small dividing spheres at low growth rates on poor media but formed giant nondividing spheroids at higher growth rates. The lethality of mre cells at higher growth rates could be partially suppressed by a supply of an overactive form of (p)ppGpp synthase (RelA'). In addition, we found that increased expression of just FtsZ was sufficient to suppress the lethality of both mre and mrd mutants on rich medium.

One striking feature, common to all unsuppressed shape mutants under nonpermissive conditions, was the extensive invasion and elaboration of the CM into the spheroid's cytoplasm. Some of these elaborations were continuous with the exterior CM, while others were topologically separate. Several lines of evidence indicate that the latter derived from the exterior CM by endocytic-like membrane fission events that release periplasm-filled vesicles in the cytoplasm. Interestingly, MreBCD-depleted spheres synthesized phospholipid at about the same rate per unit of cell mass as wt rods, providing a rationale for the "excess" membrane in their interiors. This failure to properly adjust membrane lipid synthesis to actual surface requirements under nonpermissive conditions is likely to contribute to the accompanying lethal division defect in the shape mutants. Both FtsZ and MinD assembled aberrantly on internal membrane systems, suggesting that the latter directly interferes with proper assembly of a division apparatus on external segments of the CM by diverting significant fractions of division proteins into nonproductive assemblies.

	MATERIALS AND METHODS

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
E. coli plasmids and phages. The most relevant plasmids, phages, and strains used in this study are listed in Table 1 and depicted in Fig. 1.

View larger version (23K): [in this window] [in a new window]   FIG. 1. Genetic constructs. Shown are the E. coli mre (A) and mrd (B) loci, chromosomal deletion-replacements, and inserts present on plasmids and phages. Portions of the chromosome that were replaced with an aph cassette, or an frt scar sequence remaining after eviction of the cassette, are indicated by brackets. Numbers next to brackets refer to the base pairs replaced, counting from the start of mreB (A) or mrdA (B). Inserts were placed downstream of PR (pFB124, pFB128, and pFB194), PBAD (pFB174), or Plac (all others) control regions of appropriate plasmid/phage vectors (Table 1). LE*, in-frame CTCGAGTAA sequence appended to the end of the gene; M, start codon of mreD changed from GTG to ATG.

Unless indicated otherwise, MG1655 chromosomal DNA was used as a template in amplification reactions. Sites of interest (e.g., relevant restriction sites) are underlined in primer sequences.

To construct pCH221 (P_lac::gfpmut2-T7-mrdB), mrdB (rodA) was amplified using primers 5'-TATAGAATTCATATGACGGATAATCCGAATAAAAAAACATTCTGG-3' and 5'-CATTGTCGACTTACACGCTTTTCGACAACATTTTCC-3'. The product was treated with EcoRI and SalI, and the 1,127-bp fragment was used to replace the 13-bp EcoRI-SalI fragment of pDR107c, yielding pCH218 (P_T7::gfpmut2-T7-mrdB). The 1,970-bp BglII-HindIII fragment of pCH218 was next used to replace the 20-bp BamHI-HindIII fragment of pMLB1113, resulting in pCH221.

For pCH222 [P_lac::mrdB-gfpmut2], mrdB was amplified using primers 5'-TATAGAATTCATATGACGGATAATCCGAATAAAAAAACATTCTGG-3' and 5'-GCACCTCGAGCACGCTTTTCGACAACATTTTCC-3'. The product was treated with NdeI and XhoI, and the 1,119-bp fragment was used to replace the 77-bp NdeI-XhoI fragment of pET21a, yielding pCH219 (P_T7::mrdB-His₆). The 1,152-bp XbaI-XhoI fragment of pCH219 was used to replace the 1,025-bp XbaI-XhoI fragment of pCH151 (P_lac::zipA-gfpmut2), resulting in pCH222.

Construction of plasmid pCH235 (P_lac::mreD-LE) involved several steps. The annealed product of oligonucleotides 5'-TCGAGTAAGTCGACACGGTACCA-3' (sense) and 5'-AGCTTGGTACCGTGTCGACTTAC-3' (antisense) was used to replace the 122-bp XhoI-HindIII fragment of pCH157 (P_lac::gfpmut2-T7-minD minE-His₆). This resulted in pCH181 (P_lac::gfpmut2-T7-minD minE-LE), in which the His₆ tag sequence in pCH157 was replaced with an XhoI site, encoding the dipeptide LE, followed by the TAA stop codon. The mreD gene was amplified using primers 5'-TATAGAATTCATATGGCGAGCTATCGTAGCCAGGGACGCTG-3' and 5'-CGTTCTCGAGTTGCACTGCAAACTGCTGACGGAC-3' and digested with EcoRI and XhoI. The 494-bp fragment was used to replace the 34-bp EcoRI-XhoI fragment of pDR107c, resulting in pCH217 (P_T7::gfpmut2-T7-mreD-His₆). Circularization of the 5,837-bp NdeI fragment of pCH217 yielded pCH223 (P_T7::mreD-His₆). pCH235 was finally obtained by replacing the 1,859-bp XbaI-XhoI fragment of pCH181 with the 512-bp XbaI-XhoI mreD fragment of pCH223.

Plasmid pCH244 (P_lac::mreB mreC mreD yhdE) was obtained after several steps. The 3,359-bp ApoI fragment of pMEL1 was inserted in the EcoRI site of pDR107a, yielding pDB364 [P_T7::gfpmut2-T7-mreB(5-347) mreC mreD yhdE]. The 4,223-bp BglII-HindIII fragment of pDB364 was next used to replace the 20-bp BamHI-HindIII fragment of pMLB1113, resulting in pDB366 [P_lac::gfpmut2-T7-mreB(5-347) mreC mreD yhdE]. The mreB gene was amplified using primers 5'-CGACTCTAGACAGCTTTCAGGATTATCCCTTAGTATG-3' and 5'-GCAAAAGCTTACTCTTCGCTGAACAGGTCGCC-3'. The product was treated with XbaI and HindIII, and the 1,072-bp fragment ligated to the 7,639-bp XbaI-HindIII fragment of pCH151 (P_lac::zipA-gfpmut2), generating pCH214 (P_lac::mreB). Finally, replacement of the 534-bp KpnI-HindIII fragment of pCH214 with the 2,879-bp KpnI-HindIII fragment of pDB366 resulted in pCH244.

To obtain pCH268 (P_lac::gfpmut2-T7-zapA), zapA was amplified using primers 5'-GAAGGATCCATGTCTGCACAACCCGTC-3' and 5'-CGAGTCGACTCATTCAAAGTTTTGGTTAG-3'. The product was treated with BamHI and SalI, and the 336-bp fragment was used to replace the 1,164-bp BamHI-SalI fragment of pDR120 (P_lac::gfpmut2-T7-ftsZ).

For plasmid pFB112 (tet sdiA), the 1,312-bp EcoRI-PstI fragment of pCX19 was ligated to the 3,615-bp EcoRI-PstI fragment of pBR322.

For pFB118 (P_lac::mreB), the 2,696-bp ClaI fragment of pCH244 was deleted.

To obtain pFB120 (P_lac::mreC-LE), mreC was amplified using primers 5'-CTAGTCTAGAATACGAGAATACGCATAACTT-3' and 5'-CGTTCTCGAGTTGCCCTCCCGGCGCACGCGCAGGC-3'. The product was treated with XbaI and XhoI, and the 1,128-bp fragment was used to replace the 512-bp XbaI-XhoI fragment of pCH235.

For pFB121 (P_lac::mreC mreD-LE), an mreCD fragment was amplified using primers 5'-CTAGTCTAGAATACGAGAATACGCATAACTT-3' and 5'-CGTTCTCGAGTTGCACTGCAAACTGCTGACGGAC-3'. The product was treated with XbaI and XhoI, and the 1,616-bp fragment was used to replace the 512-bp XbaI-XhoI fragment of pCH235.

Plasmid pFB124 [cI857(Ts) P_R::mreC, mreD-LE] was obtained by replacing the 1,196-bp XbaI-SalI fragment of pDB346 [cI857(Ts) P_R::ftsZ] with the 1,625-bp XbaI-SalI fragment of pFB121.

In turn, pFB128 [cI857(Ts) P_R::mreD-LE] was created by replacing the 1,625-bp XbaI-SalI fragment of pFB124 with the 521-bp XbaI-SalI fragment of pCH235.

Plasmid pFB142 (P_lac::mreB, mreC-LE) was created in two steps. The 1,271-bp XbaI-XhoI fragment of pCH217 was used to replace the 1,859-bp XbaI-XhoI fragment of pCH181, yielding pCH233 (P_lac::gfpmut2-T7-mreD-LE). An mreBC fragment was amplified using primers 5'-CGACTCTAGACAGCTTTCAGGATTATCCCTTAGTATG-3' and 5'-CGTTCTCGAGTTGCCCTCCCGGCGCACGCGCAGGC-3'. The product was treated with XbaI and XhoI, and the 2,240-bp fragment was used to replace the 1,271-bp XbaI-XhoI fragment of pCH233.

For pFB149 (P_lac::mreB mreC mreD-LE), the 1,033-bp BamHI-SalI fragment of pCH244 was replaced with the 359-bp BamHI-SalI fragment of pFB124.

To create pFB174 (P_BAD::mreB mreC mreD-LE), the 1,451-bp XbaI-HindIII fragment of pLL116 (a pBAD33 derivative that will be described elsewhere) was replaced with the 2,743-bp XbaI-HindIII fragment of pFB149.

For pFB185 (P_lac::mrdB), the 508-bp NsiI-HindIII fragment of pCH221 was used to replace the 1,252-bp NsiI-HindIII fragment of pCH222.

To construct pFB190 (P_lac::mrdA), pTB59 (P_lac::mrdAB) was used as a template to amplify mrdA (pbpA) with primers 5'-CTCTGAATTCCCGTGAGTGATAAGGGAGCTTTGAGTAG-3' and 5'-GCCAAGCTTGGTCGACTTAATGGTCCTCCGCTGCGGC-3'. The product was treated with EcoRI and HindIII, and the 1,954-bp fragment was used to replace the 3,084-bp EcoRI-HindIII fragment of pTB59.

For pFB194 [cI857(Ts) P_R::mrdB], the 1,155-bp XbaI-SalI fragment of pFB185 was used to replace the 1,625-bp XbaI-SalI fragment of pFB124.

Plasmid pTB182 (ftsQAZ) was obtained in several steps. The HindIII site within ftsA on pZAQ was removed by the QuikChange procedure (Stratagene), using the mutagenic primers 5'-CAGTTGCAGGAAAAGCTCCGCCAACAAGGGG-3' and its reverse complement, resulting in a silent change (underlined) of FtsA codon 319 (Leu). The resulting plasmid (pTB178) was next mutagenized using primers 5'-TTATGAGGCCGACGATCTAGACGGCCTCAGGCGACAG-3' and its reverse complement, creating an XbaI site in between ftsA and ftsZ. The 4,377-bp PstI-HindIII fragment of the resulting plasmid (pTB179) was then used to replace the 12-bp PstI-HindIII fragment of pGB2, yielding pTB182. The direction of ftsQAZ transcription from this plasmid is opposite that of the aadA gene.

For pTB188 (P_R::ftsZ), pDB346 [cI857(Ts) P_R::ftsZ] was used as a template in a PCR with 5'-CGTAGGATCCGCATGCGGGATAAATATCTAACACCGTGCGTG-3' and 5'-GCTCAAGCTTGTCGACTTAATCAGCTTGCTTACGCAGGAATG-3'. The product was treated with BamHI and HindIII, and the 1,359-bp fragment was used to replace the 20-bp BamHI-HindIII fragment of pGB2, yielding pTB188. Note that this plasmid lacks a lambda repressor and that ftsZ is constitutively transcribed in the direction opposite that of aadA.

For pYT11 (P_tac::relA'), a portion of relA was amplified with primers 5'-CTTTTCTAGATTTCGGCAGGTCTGGTCCCTAAAGG-3' and 5'-GGTCCTCGAGCTGGTAGGTGAACGGCACAATGCGCCC-3'. The product was treated with XbaI and XhoI, and the 1,401-bp fragment was used to replace an XbaI-XhoI fragment of pCH276, a plasmid whose construction will be detailed elsewhere. The 1,500-bp EcoRI-HindIII fragment of the resulting plasmid (pYT5) was next used to replace the 30-bp EcoRI-HindIII fragment of pJF118EH, yielding pYT11. The plasmid encodes the first 455 residues of RelA, followed by a glutamic acid residue and a stop codon.

Phages CH221, CH235, CH268, FB120, FB185, FB190, and TB59 were obtained by crossing NT5 with pCH221, pCH235, pCH268, pFB120, pFB185, pFB190, and pTB59, respectively, as described previously (24).

E. coli strains. mre knockout strains were constructed by red recombineering, using pKD13 as a template for amplification of an aph cassette consisting of aph flanked by FLP recombinase substrate sites (frt) and appropriate mre sequences (23, 91). Knockout alleles on linear fragments were recombined with the chromosome of strain DY329 carrying plasmid pCX16 [sdiA]. With plating under standard conditions (LB-kanamycin [Kan] at 30°C), the number of recombinants recovered in the presence of pCX16 was, at least, 2 to 3 logs higher than in its absence.

We used the following primer sets (chromosomal sequences are underlined): for mreB<>aph, 5'-GACCTGGGTACTGCGAATACCCTCATTTATGTAAAAGGACAAGGCATCGTGTGTAGGCTGGAGCTGCTTC-3' [primer mreB(KO)5'] and 5'-AGCCATCGGTTCTTCAATCAGGAAGACTTCACGGGCACCAGCGCCCTGCGATTCCGGGGATCCGTCGACC-3' [mreB(KO)3']; for mreC<>aph, 5'-ATCGGATGCAGGCAGGGGAAGTGTCTGTTTACCCTGCCTGGTCTGATACGATAAGTGTAGGCTGGAGCTGCTTC-3' [mreC(KO)5'] and 5'-AGCGATCCCCGTTGCCGGTTCAGGTAACTTTGGCCCCATCGCGTCTGGCGAATTCCGGGGATCCGTCGACC-3' [mreC(KO)3']; for mreD<>aph, 5'-GTGGCGAGCTATCGTAGCCAGGGACGCTGGGTAATCTGGCTCTCTTTCCTCTAAGTGTAGGCTGGAGCTGCTTC-3' [mreD(KO)5'] and 5'-TCAGCAAGAAAATCCACGGCCAGAGCACCCCATTGACTACACTACTCCAGAATTCCGGGGATCCGTCGACC-3' [mreD(KO)3']; for mreBCD<>aph, primers mreB(KO)5' and mreD(KO)3'; for mreBC<>aph, primers mreB(KO)5' and mreC(KO)3'; and for mreCD<>aph, primers mreC(KO)5' and mreD(KO)3'.

Recombination yielded a set of six mre<>aph derivatives of DY329/pCX16, which all showed a spherical phenotype. The six strains were transformed with pCH244 (P_lac::mreB mreC mreD yhdE), and transformants of each reverted to a rod shape in an IPTG (isopropyl-β-D-thiogalactopyranoside)-dependent manner. Phage P1 was grown on a transformant (containing both pCX16 and pCH244) of each strain in the presence of 250 µM IPTG, resulting in a high-titer transducing lysate for each mre<>aph allele. These lysates were then used to transduce Mre⁺ strains, PB103, or TB28, using various strategies to avoid selective pressure for the accumulation of undesired suppressor mutations. Generally, this was accomplished by the introduction of appropriate correcting or suppressing mre, sdiA, or ftsZ plasmids or phages into the Mre⁺ recipient before introduction of a chromosomal mre<>aph allele by transduction. For example, to obtain the MreBCD depletion strain FB30/pFB174 (mreBCD<>aph/cat araC P_BAD::mreBCD), TB28 was transformed with pFB174 prior to transduction of mreBCD<>aph and transductants were recovered at 30°C on LB-Kan supplemented with chloramphenicol and 0.5% arabinose.

Similarly, derivatives of PB103 carrying chromosomal mre<>frt alleles (Table 1; also see Table S1 in the supplemental material) were obtained by introduction of pFB112 (tet sdiA) prior to transduction with the corresponding mre<>aph lysates. The resulting mre<>aph/pFB112 strains were then transformed with pCP20 [bla cat repA(Ts) cI857(Ts) P_R::flp] (16, 23) and plated at 30°C on LB containing ampicillin (Amp) and tetracycline. Transformants were streaked on LB lacking Amp and incubated at 42°C to simultaneously induce production of Flp recombinase and block replication of pCP20. Kan- and Amp-sensitive clones were purified, resulting in the desired mre<>frt/pFB112 strains. These strains were transformed with appropriate mre plasmids and used for complementation analyses (see Table S1 in the supplemental material). The growth of some of these transformants at 37°C and in the presence of IPTG led to simultaneous correction of the rod phenotype and competitive loss of pFB112 (see Table S1 in the supplemental material), giving rise to depletion strains that lacked extra copies of sdiA, such as the MreB depletion strain FB17/pFB118/pFB124 (mreBCD<>frt/P_lac::mreB/P_R::mreCD).

For construction of mrd mutants, we used the following primer sets (chromosomal sequences are underlined): for mrdAB<>aph, 5'-CATCCTTATCACCGTGAGTGATAAGGGAGCTTTGAGTAGAAAACGCAGCGGGTGTAGGCTGGAGCTGCTTC-3' [pbp2(KO)5'] and 5'-CGCCAGCCATGACGCGACCAAAGGTGGTTTGCGCTCTGGCGGCTATCCATTCCGGGGATCCGTCGACC-3' [rodA(KO)3']; and for mrdB<>aph, 5'-CGATCTGCCTGCGGAAAATCCAGCGGTTGCCGCAGCGGAGGACCATTAAGTGTAGGCTGGAGCTGCTTC-3' [rodA(KO)5' and rodA(KO)3'].

Recombination with the chromosome of DY329/pCX16 resulted in FB29/pCX16 (mrdAB<>aph/sdiA) and FB20/pCX16 (mrdB<>aph/sdiA), which propagated as spheres. These strains were transformed with pTB59 (P_lac::mrdAB), which caused transformants to revert back to a rod shape in the presence of IPTG. P1 lysates were prepared on FB29/pCX16/pTB59 and FB20/pCX16/pTB59 transformants, and these were used to transduce mrdAB<>aph and mrdB<>aph into PB103 or TB28 derivatives carrying appropriate complementing plasmids and/or phages.

For the P1 transduction experiments whose results are shown in Tables 3 and 5, we used the mre<>aph and mrd<>aph lysates described above, except for the mrdB<>aph and lacIZYA<>aph transducing lysates (Table 3), which were prepared on strains FB22(CH221) and TB12, respectively.

Immunofluorescence and confocal microscopy. On-slide immunofluorescence staining methods that work well with rods and filaments (1, 37) did not result in consistent labeling of FtsZ structures in large spheroids. We therefore developed a protocol for immunostaining of cellular structures in which incubations with lysozyme and antibodies are done in solution. Briefly, cells were fixed by adding 1 ml of culture directly to a mixture of formaldehyde and glutaraldehyde in NaPO₄ buffer (pH 7.5), giving final concentrations of 2.4%, 0.04%, and 30 mM, respectively. The suspension was incubated for 10 min at room temperature (RT), followed by incubation for 50 min on ice. Cells were washed twice in 1 ml phosphate-buffered saline (PBS) (10.1 mM Na₂HPO₄, 1.8 mM KH₂PO₄, 137 mM NaCl, 2.7 mM KCl, pH 7.4), once in 1 ml GTE (20 mM Tris-Cl, 50 mM glucose,10 mM EDTA, pH 7.5), and resuspended in 0.4 ml of GTE. Aliquots (0.1 ml) of cells were treated with egg white lysozyme from a freshly prepared stock of 0.4 mg/ml in GTE to give a final concentration of 16 µg/ml and incubated for 2 to 4 min at RT. Cells were washed once in 1 ml PBS and resuspended in 1 ml PBS containing 2% bovine serum albumin (BSA). During the subsequent blocking and antibody incubation steps, samples were gently mixed in an end-over-end tube mixer. After 1 h at RT, affinity-purified rabbit polyclonal anti-FtsZ antibodies (37) were added, and incubation was continued overnight at 4°C. Cells were washed once with 1 ml PBS, resuspended in 1 ml PBS containing 2% BSA and a 1:2,500 dilution of Alexa-488 conjugated anti-rabbit immunoglobulin G (Molecular Probes), and incubated for 2 h at RT. Cells were washed twice with 1 ml PBS and resuspended in 50 µl PBS, and aliquots were spotted onto poly-L-lysine-coated coverslips. Confocal microscopy was performed with a Zeiss LSM 510 inverted laser-scanning microscope using a 100x (1.45-numerical-aperture) oil immersion objective. Images were collected using 488-nm excitation light from an argon-krypton laser, a 560-nm dichroic mirror, and a 500- to 550-nm band pass barrier filter. For all images, a z series was collected at 0.2-µm increments. Image processing, including projections and three-dimensional rotations, were performed using LSM 510 software (version 2.5).

Wide-field microscopy. The cells in Fig. 5 and 7G were imaged on a Leica DM IRE2 microscope outfitted with a CoolSnap HQ camera (Photometrics) and a piezo-driven 100x (1.4-numerical-aperture) oil objective. Optical sections were collected at a 0.2-µm step size and with Cy3- and/or green fluorescent protein (GFP)-specific filter sets. Images were deconvolved through 40 iterations of a blind deconvolution algorithm provided in the Leica AS MDW package. As indicated, either deconvolved individual slices or maximum projections of the deconvolved image stack are shown. All other wide-field images, including time lapse series, were obtained with a Zeiss Axioplan-2 microscope setup as previously described (44). Live cells were imaged using clean but otherwise untreated microscope slides.

View larger version (45K): [in this window] [in a new window]   FIG. 5. Aberrant FtsZ assemblies in live MreBCD-depleted spheres. The MreBCD depletion strain FB30(CH268)/pFB174 [mreBCD(Plac::gfp-zapA)/PBAD::mreBCD] was grown at 37°C in M9-maltose medium lacking arabinose and containing 50 µM IPTG. Live cells were imaged early (A and B) (OD600 = 0.2) and later (C and D) (OD600 = 0.4) during depletion of MreBCD. Cells were mixed with FM4-64 (0.5 µg/ml) immediately prior to imaging. Maximum projection GFP (2), FM4-64 (3), and merged (4) as well as corresponding DIC (1) fluorescence images are shown. Panels D5 to D8 show merged fluorescence images of individual optical slices from the top to the bottom of the cell. The arrows in panel C4 highlight some odd-looking GFP-ZapA accumulations. The arrows in panels D1 and D5 point at small vesicle-like bodies that are both visible by DIC and outlined with FM4-64 fluorescence, while the arrowheads point at much larger bodies that failed to be outlined by the dye. The arrow in panel D8 points at a membrane involution near the bottom of the cell. Bar = 2 µm.

View larger version (136K): [in this window] [in a new window]   FIG. 7. Topological separation of internal membrane systems from the external cell membrane. Shown are live MreBCD-depleted spheroids of strain FB30/pFB174 (mreBCD/PBAD::mreBCD) that were grown at 37°C in LB to an OD600 of 0.2. (A to C) Cells were mixed with FM4-64 and imaged immediately. Panels show DIC (1) and FM4-64 (2) fluorescence images. Arrowheads in panels A and B point to large vesicles that were not outlined by FM4-64, while the arrow in panel C points out a vesicle that was. (D to G) Cells were pulse labeled with FM4-64 for 5 min, 30 min prior to imaging. Shown are both DIC (1) and FM4-64 (2 to 6) fluorescence images. Note that all vesicles visible by DIC are now also outlined by FM4-64 fluorescence. In addition, FM4-64 stains structures not readily resolvable by DIC (e.g., arrows in panels D and F). For the cell in panel G, individual z slices from top to bottom (G2 to G5) as well as a maximum projection image (G6) are shown. Note that FM4-64-stained material is present throughout the interior of the cell. (H and I) CellTrace BTME was added 15 min, and FM1-43 immediately, before imaging. DIC (1), FM1-43 (2), and BTME (3) images are shown. Arrowheads point to material labeled with BTME but not with FM1-43. (J to L) The growth medium was supplemented with LY. Cells were gently washed in prewarmed medium lacking LY and imaged immediately. Many vesicles visible by DIC contained trapped LY (e.g., arrowhead in panel K). Some bodies that appeared as a vesicle by DIC did not retain the dye, suggesting that they were still continuous with the external CM and periplasm (arrow in panel L). Panel J illustrates that cells of the wt parent control (TB28) completely failed to retain the dye. Bar = 2 µm.

View larger version (121K): [in this window] [in a new window]   FIG. 8. Vesicle formation does not require FtsZ polymerization. Shown are spheroids of strain FB30/pTB63/pDR144 (mreBCD/ftsQAZ/Plac::sfiA). (A to D) Cells were inoculated to an OD600 of 0.05 in LP maltose containing no (A) or 0.5 mM (B) IPTG and then grown at 37°C to an OD600 of 0.3, fixed, and labeled with BTME. Shown are DIC (1) and BTME (2) fluorescence. One hundred cells of each culture were further analyzed to determine the average lengths of their long and short axes (C) and the percentages of cells containing internal membrane and/or showing signs of constriction (D). Arrowheads in panel B point to examples of vesicles visible by both DIC and BTME fluorescence. (E to F) Cells were diluted to an OD600 of 0.025 in LB supplemented with 0.5 mM IPTG and either no (E) or 50 µg/ml (F) LY and grown at 37°C to an OD600 of 0.2 to 0.3. For panel E, cells were incubated with BTME for 15 min at 37°C and then treated with FM1-43 immediately before imaging live. Shown are DIC (1), FM1-43 (2), and BTME (3) fluorescence. Note the intracytoplasmic membrane stained by BTME that was inaccessible to FM1-43. For panel F, cells were gently washed in prewarmed growth media prior to imaging. Shown are DIC (1) and LY (2) fluorescence images. Note the trapped LY in intracytoplasmic vesicles. Bar = 2 µm.

Cell dimensions and geometrical considerations. Cell dimensions were measured from differential interference contrast (DIC) images by using Object-Image 2.15 (83). The position of the long axis in spheroids was judged by eye, and the short axis was measured perpendicular to the long one. The volume (V), surface (S), and circumference (C) of a sphere were calculated using V_s = 4/3r³, S_s = 4r², and C_s = 2r and those of a rod (capsule) with V_c = 4/3r³ + r²h, S_c = 4r² + 2rh, and C_c = 2r, with r representing radius and h cylinder length. The volume of a prolate spheroid (c > a) was obtained using V_sph = 4/3a²c, with c representing polar radius and a equatorial radius.

Other methods. Whole-cell extracts were prepared as described previously (37). Protein concentrations were measured using the noninterfering protein assay (NI; G-Biosciences), with BSA as a standard. Quantitative Western analyses were done essentially as before (45).

	RESULTS

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Construction of unsuppressed mre- and mrd-null and depletion strains of E. coli. To study the physiological relevance of the Mre proteins in E. coli, we used red-mediated recombineering to create sets of strains carrying chromosomal mre mutations. Careful construction and maintenance of these strains was prompted by our initial observations, consistent with those by others (49, 70), that propagation of mre mutants under common growth conditions appears to select for progeny that produce an elevated level of the division protein FtsZ. Thus, our first attempts to create mre<>aph lesions (Fig. 1A) in the recombineering strain DY329 [cI857 (cro-bioA)] yielded no or very few recombinants (data not shown), suggesting the lesions caused death. This result was not in agreement with the viability of strains carrying the classical mreB129 or mre-678 [(mreB-rng)] alleles (85).

A clue to what hindered the construction of mre knockouts came from observing strain PA340-678/pCH244 (mreBCD/P_lac::mreBCD), which carries the classical chromosomal mre-678 allele (84) and a complementing plasmid containing mreBCD downstream of the lac promoter (Fig. 1A). Cells grew as spheres in the absence of IPTG and as rods in its presence. In the latter case, however, a small but significant fraction of rods contained polar septa, leading to the production of minicells. Minicells were not observed in strain PA340/pCH244, indicating that their formation by the shape-corrected mutant was a property neither shared by its Mre⁺ parent nor induced by expression of the mre genes from the plasmid (data not shown). Because cooverexpression of the ftsQ, -A, and -Z genes both suppresses the lethality of mrd lesions (59, 82) and induces minicell formation (88), we hypothesized that similar to what occurs with mrd, (i) the mre genes in E. coli might be essential, explaining our difficulties in obtaining null alleles; (ii) existing mre mutants may have accumulated second-site mutations that lead to an elevated level of FtsQ, FtsA, and/or FtsZ; and (iii) elevated levels of the division proteins restore viability and allow mre mutants to propagate as spheres.

Accordingly, quantitative Western analyses showed that the classical mreB129 and mre-678 derivatives of strain PA340 (85) contained about two times more FtsZ than the parent (Table 2). Moreover, introduction of plasmid pCX16 (sdiA) in the recombineering strain DY329 now allowed the recovery of viable mre knockout derivatives at frequencies that were at least 2 logs higher than without the plasmid (data not shown). SdiA positively regulates a promoter (ftsQ2p) upstream of ftsQAZ, and cells carrying pCX16 contained three- to fourfold the normal level of FtsZ (87) (Table 2). The use of DY329/pCX16 for recombineering allowed for viable mrd knockout derivatives (Fig. 1B) to be readily obtained as well.

The presence of the sdiA plasmid was not absolutely required for allelic transfer, as transduction of mre<>aph alleles to PB103/pGB2 did yield rare spherical transductants (Table 3). Further analyses of one of these (FB2sup [mreB<>aph]) showed that its FtsZ level was over twofold higher than normal (Table 2), indicating it had undergone a second alteration, leading to increased production of the division protein. To avoid a selective advantage for such undefined (and undesired) suppressors of Mre^– or Mrd^– lethality, care was taken to provide all knockout strains with complementing (Mre⁺ or Mrd⁺) and/or lethality-suppressing (SdiA⁺ or Fts⁺) plasmids or phages during subsequent strain manipulations.

Polarity of mreB and mreC lesions. Initial complementation experiments indicated that the mre<>aph alleles were polar on expression of downstream genes, precluding a firm conclusion as to the role of each gene (not shown). We proceeded with complementation experiments using strains in which the aph gene had been evicted by FLP recombinase, leaving only the 82-bp frt scar sequence on the chromosome (23). To maintain viability, these strains also carried the sdiA plasmid pFB112 (tetA sdiA), a ColE1 derivative conferring tetracycline resistance. As expected, each of the mre<>frt/pFB112 strains grew as spheres. Complementation was studied with a set of six ColE1 derivatives which confer resistance to Amp and carry one or more of the mre genes downstream of the lac promoter (Table 1 and Fig. 1A). The mre<>frt/pFB112 strains were transformed with each one of the P_lac::mre plasmids. Cells were plated on LB agar containing Amp and IPTG, and transformants were examined for both cell morphology and loss of the incompatible pFB112 plasmid.

As summarized in Table S1 in the supplemental material, plasmid pCH244 (P_lac::mreBCD) was capable of restoring a rod shape to each of the mre<>frt strains. In addition, pCH235 (P_lac::mreD) restored the rod shape in the mreD<>frt strain. However, the mreB<>frt lesion failed to be restored by pFB118 (P_lac::mreB) unless cells also harbored pFB124 (P_R::mreCD), a compatible plasmid carrying mreC and mreD downstream of a temperature-inducible P_R promoter. Similarly, the mreC<>frt allele could be corrected only by P_lac::mreC plasmids that carry mreD in cis (pCH244 or pFB121) or when mreD was coexpressed in trans from pFB128 (P_R::mreD). Others previously noted that frameshift or frt deletion-replacement lesions in mreB are polar on the expression of mreC and mreD (61, 86). Our complementation results are consistent with this and further show that the chromosomal mreC<>frt lesion (Fig. 1A) is similarly polar on the expression of mreD.

Whereas each transformant in which the rod shape was restored had lost pFB112, all transformants that remained spherical had retained this sdiA plasmid, even though the antibiotic in the medium (Amp) favored maintenance of the incompatible P_lac::mre competitors (see Table S1 in the supplemental material). Apparently, pFB112 provided all spherical Mre^– cells with a selective advantage, supporting the conclusion that extra copies of sdiA allowed them to propagate.

Each mre gene is required for both maintenance of rod shape and normal viability. To study unsuppressed Mre^– phenotypes, we used strains that lack an sdiA plasmid and in which transcription of one or all of the mre genes can be shut off by omitting IPTG or arabinose from the growth medium. Specific depletion of MreB or MreC was accomplished by supplying cells with an appropriate source of MreC and/or MreD to compensate for the polarities associated with the chromosomal mreB and mreC lesions described above.

As shown in Fig. 2, cells of strain FB17/pCH244 (mreBCD<>frt/P_lac::mreBCD) (row 8) grew about as well as the control strain PB103/pCH244 (wt/P_lac::mreBCD) (row 7) on LB at 37°C in the presence of IPTG (columns A to C). Growth of FB17/pCH244 was negligible in the absence of the inducer (columns D to F), however, confirming that depletion of all three Mre proteins severely limits the ability of cells to propagate. Identical results were obtained upon the specific depletion of MreB, MreC, or MreD separately, using strains FB17/pFB118/pFB124 (mreB<>frt/P_lac::mreB/P_R::mreCD) (row 2), FB10(FB120)/pFB128 [mreC<>aph(P_lac::mreC)/P_R::mreD] (row 4), or FB11(CH235) [mreD<>aph(P_lac::mreD)] (row 6), respectively. As detailed further below, depletion of any of the Mre proteins caused cells to grow into large spheres that failed to divide properly (see Table S2 in the supplemental material).

View larger version (96K): [in this window] [in a new window]   FIG. 2. Mre– lethality and suppression of MreBCD– lethality by extra SdiA. Even-numbered rows show that depletion of MreB (row 2), MreC (row 4), MreD (row 6), or MreBCD (row 8) causes a severe growth defect and that the latter can be suppressed by extra SdiA (row 10). Odd-numbered rows show appropriate wt controls. Cultures were grown to density at 37°C in LB supplemented with appropriate antibiotics and IPTG and were diluted 104 (columns A and D), 105 (B and E), or 106 (C and F)-fold. Aliquots (10 µl) were spotted on LB plates containing either 0.1% glucose (D to F) or IPTG (A to C) at 100 µM (rows 3 to 6) or 250 µM (other rows). Plates were incubated overnight at 37°C and photographed. The strains used were PB103/pFB118/pFB124 and FB17/pFB118/pFB124 (rows 1 and 2), PB103(FB120)/pFB128 and FB10(FB120)/pFB128 (rows 3 and 4), PB103(CH235) and FB11(CH235) (rows 5 and 6), PB103/pCH244 and FB17/pCH244 (rows 7 and 8), and PB103/pCH244/pCX16 and FB17/pCH244/pCX16 (rows 9 and 10).

Overexpression of FtsZ is sufficient to restore viability to Mre^– cells. In addition to stimulating transcription of the ftsQAZ division genes (87), SdiA affects the expression of many other genes as well (89).

To test whether an increased level of just FtsZ is sufficient to suppress Mre^– lethality, we used strain FB30/pFB174 (mreBCD<>aph/P_BAD::mreBCD) carrying either pDR3 (P_lac::ftsZ) or the vector control pMLB1113. Aliquots of serially diluted cultures were spotted on LB agar supplemented with either 0.5% arabinose, 0.1% glucose, or 0.1% glucose plus 100 µM IPTG. Both strains grew well in the presence of arabinose (MreBCD⁺) but failed to form colonies in the presence of glucose (MreBCD^–) when IPTG was absent. The presence of IPTG, however, specifically allowed pDR3-carrying cells to grow in the presence of glucose (MreBCD^– FtsZ⁺⁺), showing that overproduction of FtsZ is indeed sufficient to overcome the growth defect of Mre^– cells (Fig. 3A).

View larger version (42K): [in this window] [in a new window]   FIG. 3. Conditional viability of Mre– cells. (A) The MreBCD depletion strain FB30/pFB174 (mreBCD/PBAD::mreBCD) carrying either plasmid pDR3 (Plac::ftsZ) (row 2) or a vector control (row 1) was grown to density at 37°C in LB with 0.5% arabinose. Cultures were diluted and spotted on LB plates containing 0.5% arabinose (columns A to C), 0.1% glucose (columns D to F), or 0.1% glucose plus 100 µM IPTG (columns G to I). Plates were incubated at 37°C. (B) Overnight cultures of strain TB28 (wt) carrying either pDR3 (Plac::ftsZ) (row 2) or a vector control (row 1) were diluted and spotted on LB plates containing the A22-stock solvent methanol (columns A to C), 10 µg/ml A22 (columns D to F), or 10 µg/ml A22 plus 50 µM IPTG (columns G to I). Plates were incubated at 37°C. (C) The MreBCD depletion strain FB21/pFB149 (mreB/Plac::mreBCD) (rows 2 and 4) and its wt parent TB28/pFB149 (rows 1 and 3) were grown overnight at 37°C in LB with 250 µM IPTG. Cultures were diluted and spotted on M9-maltose plates containing 250 µM IPTG (columns A to C) or 0.1% glucose (columns D to F). Plates were incubated at 30°C (rows 1 and 2) or 20°C (RT; rows 3 and 4). (D) An overnight culture of strain TB28 (wt) was diluted and spotted on M9-maltose plates containing methanol (row 1) or 5 µg/ml A22 (row 2). Plates were incubated at RT or 37°C, as indicated. (E) The MreBCD depletion strain FB30/pFB174 (mreBCD/PBAD::mreBCD) (rows 1 and 2) and its wt parent TB28/pFB174 (rows 3 and 4) carrying either plasmid pYT11 (Ptac::relA') (rows 2 and 4) or pJF118EH (vector) (rows 1 and 3) were grown to density at 30°C in M9-maltose with 0.5% arabinose. Cultures were diluted in LB and spotted on an LB plate with 50 µM IPTG, which was incubated at 30°C. Overnight cultures were serially diluted 104 (columns A, D, and G)-, 105 (columns B, E, and H)-, and 106 (columns C, F, and I)-fold (A to D) or 103 (column A)-, 104 (column B)-, 105 (column C)-, and 106 (column D)-fold (E) in LB, and 10-µl aliquots were spotted in each case. Plates were incubated for 2 days (M9 at RT) or overnight (all others).

The Mre proteins are dispensable for viability at low growth rates. In the experiments described thus far, cells were cultured with rich (LB) medium at 30°C or 37°C. The growth phenotypes of Mre^– cells were characterized more rigorously by depleting MreBCD from derivatives of two parent strains, PB103 and TB28, on both rich (LB) and poor (M9) media and at three different temperatures (37°C, 30°C, and 20°C). The parent strains have distinct backgrounds, and TB28 grows significantly faster than PB103, especially on minimal medium (Table 4). Depletion strains FB17/pFB149 and FB21/pFB149 (mreBCD/P_lac::mreBCD) and their respective parent controls, PB103/pFB149 and TB28/pFB149 (wt/P_lac::mreBCD), were subjected to spot titer analyses on medium containing either 0.1% glucose or 250 µM IPTG. The results (shown in Fig. S1 in the supplemental material) are summarized in Fig. 3C and Table 4. As anticipated, FB21/pFB149 failed to grow in the presence of glucose (MreBCD^–) under most conditions (row 2 in Fig. 3C and even rows in Fig. S1B in the supplemental material). Strikingly, however, it grew almost as well as the parent control on minimal medium at RT (20°C) (row 4 in Fig. 3C and row 12 in Fig. S1B in the supplemental material). Strain FB17/pFB149 (see Fig. S1A in the supplemental material) similarly failed to grow in the presence of glucose on LB at 37°C and 30°C (rows 2 and 4) but grew about as well as its parent control on M9 at both 30°C and 20°C (rows 10 and 12) and even formed some tiny colonies on LB at 20°C and on M9 at 37°C (rows 6 and 8). When growing in the presence of glucose under permissive conditions, cells of each depletion strain propagated as spheres (not shown), indicating that expression of mreBCD from the plasmid was sufficiently repressed. Compared to the doubling times of the parent strains in liquid medium under comparable conditions (Table 4), these results indicated that while Mre functions are essential for viability at moderate-to-high growth rates (approximate mass doubling time [T_d] <150 min), they become dispensable during slow growth (T_d > 300 min).

The ability of unsuppressed Mre^– cells to propagate under slow-growth conditions suggested that slow growth might also reduce the toxicity of A22. Figure 3D shows that while A22 prevented growth of TB28 on LB agar at 37°C, the drug was indeed markedly less effective at inhibiting colony formation at 20°C.

Suppression of Mrd^– lethality by ftsZ or slow growth. Cooverexpression of ftsQ and ftsA with ftsZ was previously found to be required to restore viability to amdinocillin-treated cells on rich medium (59). This predicted that, in contrast to what was found for Mre-depleted spheres (see above), overexpression of ftsZ by itself might not be sufficient to rescue Mrd-depleted ones. We addressed this issue by using strain FB40(TB59) [mrdAB<>frt (P_lac::mrdAB)] carrying either pTB182 (P_QAZ::ftsQAZ), pTB188 (P_R::ftsZ), or a vector control (pGB2). Aliquots of serially diluted cultures, including the Mrd⁺ parent controls, were spotted on LB agar containing either 100 µM IPTG or 0.1% glucose. All strains grew well in the presence of IPTG (MrdAB⁺). MrdAB-depleted cells carrying the vector control failed to grow, but those that carried either plasmid pTB182 or pTB188 (MrdAB^– FtsQAZ⁺⁺ or MrdAB^– FtsZ⁺⁺, respectively) plated efficiently (see Fig. S2 in the supplemental material). Moreover, pTB188 (P_R::ftsZ) rescued MrdAB-depleted cells about as well as cells that were depleted for MreBCD in a parallel control experiment (see Fig. S2 in the supplemental material). We conclude that elevated expression of ftsZ is sufficient to alleviate the growth defects of both Mre^– and Mrd^– cells.

wt cells resist killing by amdinocillin on poor medium (4, 47), and a mrdB (rodA)-null mutant was reported to survive on poor medium as well (28). Hence, it was likely that slow-growth conditions would also allow unsuppressed MrdAB^– spheres to propagate. To verify this, spot titer analyses were performed with the MrdAB depletion strains FB39(TB59) and FB40(TB59) [mrdAB(P_lac::mrdAB)] and their parent controls, PB103(TB59) and TB28(TB59) [wt(P_lac::mrdAB)], respectively. As summarized in Table 4, the results (provided in Fig. S3 in the supplemental material) were similar to those obtained with the MreBCD depletion strains. Although the depletion strains failed to grow in the presence of glucose (MrdAB^–) on LB agar at 37°C and 30°C, they grew about as well as the parent controls under conditions favoring slower mass increase. Transduction experiments again supported these observations in that stable mrdAB<>aph derivatives of both PB103 and TB28 could be obtained at normal frequencies as long as they were selected for and maintained under conditions avoiding too-rapid growth (Table 5, and data not shown).

Suppression of MreBCD^– lethality by RelA'. Cells become resistant to killing by amdinocillin when concentrations of the stringent alarmone ppGpp rise above a threshold that is still well below that needed to stop growth altogether (47, 81). Given the similar growth requirements of Mrd^– and Mre^– spheres noted above, we suspected that the lethality associated with depletion of the MreBCD proteins on rich medium might be suppressed by increased ppGpp as well. To test this, we stimulated alarmone synthesis in wt and Mre-depleted cells by IPTG-induced expression of an overactive form of (p)ppGpp synthase (RelA', lacking residues 456 to 744) (47, 69) from plasmid pYT11 (P_tac::relA'). As expected, the inducer caused a reduction in the growth rate of pYT11-carrying cells, and growth ceased completely at 250 µM IPTG. In LB medium without arabinose and containing only 50 µM IPTG, T_ds increased from 59 to103 min in wt rods (TB28) and from 76 to 108 min in MreBCD-depleted spheres (FB30/pFB174) (Table 6). Spot-titer analyses showed that expression of RelA' under these conditions was sufficient to suppress the lethality of MreBCD-depleted spheres (Fig. 3E). Although the modestly reduced growth rate might have contributed to the ability of Mre^– spheres to survive in this experiment, this is unlikely to be the sole explanation, as spheres in which relA' is not artificially induced fail to survive unless the T_d value surpasses 150 min, at least (Table 4).

We conclude that the growth requirements of genetically unsuppressed Mre^– and Mrd^– spheres are quite similar. Both are viable at low growth rates, and their deaths at high growth rates can be prevented by increases in ppGpp levels and/or an extra supply of just the FtsZ division protein.

Conditional lethality of Mre^– cells is associated with a division defect and aberrant assembly of FtsZ. The finding that ftsZ overexpression allows propagation of Mre^– cells under nonpermissive growth conditions suggested that, as what was inferred for spherical mrd mutants (82), the lethality associated with loss of mre might be primarily caused by a division defect. The phenotype of Mre^– cells supported this possibility. Strains completely lacking one or more of the Mre proteins, but carrying an sdiA plasmid, grew as spheres of various sizes. Many of these appeared to be in the process of constriction, and immunostaining with anti-FtsZ antibodies showed the protein associated with these sites. In most of the smaller spheres, FtsZ had accumulated in well-defined rings, although some rings showed atypical branches (see Fig. S4A in the supplemental material). About 10 to 20% of these populations consisted of distinctly larger cells, likely due to unequal inheritance of the suppressing sdiA plasmid. In these cells, FtsZ invariably appeared assembled in more-complex patterns that often included isolated patches and foci as well as more-extended structures that failed to span the girth of the cell but appeared branched and/or folded back on themselves (see Fig. S4B in the supplemental material).

Depletion of each (or all) of the three Mre proteins under nonsuppressing conditions resulted in a uniform giant-sphere phenotype (see Table S2 in the supplemental material). For example, when cells of the MreB depletion strain FB17/pFB118/pFB124 (mreB<>frt/P_lac::mreB/P_R::mreCD) were shifted from LB medium containing IPTG to medium lacking the inducer, cells initially grew and divided as rods but then lost the rod shape and ultimately formed very large spherical cells. FtsZ assembled in typical rings early during depletion (Fig. 4A and B), but the nondividing large cells that formed later on again contained the protein in more-complex patterns as described above (Fig. 4C to E).

View larger version (39K): [in this window] [in a new window]   FIG. 4. MreB-depleted cells contain normal levels of FtsZ but form aberrant FtsZ structures. (A to E) Immunofluorescence confocal microscopy of MreB-depleted cells with anti-FtsZ antibody, Strain FB17/pFB124/pFB118 (mreBCD/PR::mreCD/Plac::mreB) was grown in the presence of either 250 µM IPTG (A) or 0.1% glucose (B to E). Samples for staining were taken both early (B) (OD600 = 0.2) and late (A and C to E) (OD600 = 0.6) during growth/depletion. Maximum projection (A1, B1, C1, D1, and E2), DIC (A3, B3, and C4), and merged (A2, B2, C3, D2, and E1) fluorescence images are shown. In panel C2, the image in panel C1 is rotated 90° about the y axis. Panels E3, E4, and E5 show y axis rotations of the left-hand (168°), middle (60°), and right-hand (60°) cell in panel E2, respectively. Bar = 2 µm. (F) Anti-FtsZ immunoblot of whole-cell extracts of MreB-depleted spheres (lanes 3 and 6) and rod-shaped controls (other lanes). Extracts were prepared on strain PB103 (wt) (lanes 1 and 4), and its MreB depletion derivative FB17/pFB124/pFB118 was grown in the presence of either 250 µM IPTG (lanes 2 and 5) or 0.1% glucose (lanes 3 and 6). Cells were harvested both early (OD600 = 0.2) (lanes 1 to 3) and late (OD600 = 0.4 to 0.5) (lanes 4 to 6) during growth/depletion, and each lane received 10 µg total protein. Measured intensity values of FtsZ bands in lanes 2 and 3 were normalized to that in lane 1 and those in lanes 5 and 6 to that in 4. Resulting relative values are shown below the lanes. Cells were grown in LB at 37°C in each case (A to F).

Figure 5D highlights features of a large Mre^– spheroid that appears in the process of division. It contains a shallow constriction perpendicular to the middle of its long axis, and the bulk of FtsZ appears to have assembled in a zone around the constriction. The constriction is asymmetric, however, in that the bottom part of the cell shows a clear invagination that is associated with a fairly well-defined arc of FtsZ (D7 and D8), whereas invagination is less obvious in the top part, where FtsZ seems present in ill-defined clusters scattered about midcell (D5 and D6). As Mre^– cells grow into very large spheres under these nonpermissive conditions, we imagine that many such constriction attempts eventually abort.

One possible explanation for the failure of Mre^– spheres to divide properly under nonpermissive growth conditions, and for the fact that extra FtsZ can restore division, is that the absence of the Mre proteins somehow caused a drop in the level of FtsZ. This was not supported by Western analyses, however, as we detected no significant change in the level of FtsZ in nondividing MreB-depleted spheres (Fig. 4F, lane 6) compared to that in dividing rod-shaped control cells (lanes 4 and 5).

Vesicle-like bodies in E. coli spheres. The cell in Fig. 5D also shows another striking feature of Mre^– spheres under nonpermissive conditions, which is the presence of vesicle-like bodies in their interiors. Imaging by DIC indicated the presence of a large vesicle-like body in the left-hand half of the cell and that of smaller ones elsewhere (D1). Most of these were not stained by the membrane-impermeable FM4-64 dye, suggesting that if these compartments were surrounded by membrane, it was discontinuous with the externally accessible CM. One of the optical slices shows a clear small circle of FM4-64 staining near the cell center, however, suggesting the presence of a finger-like involution of the CM at this site that reaches well into the body of the cell. Some FtsZ clusters surrounding this FM4-64-stained material can be seen as well (D5). Another projection of FM4-64 stain that appears continuous with the CM is visible in a plane near the cell bottom (D8).

Vacuole-like inclusions were previously noted upon inactivation of PBP2 (MrdA) by amdinocillin in E. coli (54) and in a Salmonella enterica serovar Typhimurium rodA (mrdB) mutant grown on soft agar (20). In addition, they were observed in a number of E. coli shape mutants with ill-characterized lesions (2, 3, 38, 54, 63). Two of these older studies included thin section transmission electron microscopy analyses of the mutant cells. Allison (3) performed these studies on a mon (envB) mutant (2) that may have been allelic with one of the mre genes (54, 85), while Henning et al. (38) studied a temperature-sensitive shape mutant (lss12) that may have been allelic to mrdA as it produced a thermolabile PBP2 protein (75). Both studies showed the presence of CM involutions, stacked cisternae, and vesicle-like compartments traversing the cytoplasmic space of large spherical cells. These compartments appeared lined by a unit membrane, and their lumens lacked ribosomes, suggesting that they formed by involutions of the CM. Whether all intracytoplasmic membrane was continuous with the CM was not assessed (3, 38).

To better define the genetic requirements for vacuolization in shape mutants, we depleted each of the Mre and Mrd proteins separately under nonsuppressing conditions and observed cells by both membrane staining and DIC. In each case, cells formed large spheres with readily apparent vesicle-like inclusions (see Table S2 in the supplemental material). Therefore, the phenomenon is not provoked by the absence of any of the shape proteins specifically but is more likely a general consequence of growth as a nondividing sphere per se. As this phenotype is inherently interesting and correlates with the failure of spheres to divide properly, we studied the formation of vesicle-like inclusions in Mre^– spheres in more detail.

Involution of the CM and endocytosis in E. coli spheres. To ensure that the vesicle-like bodies that we observed in spheres were bounded by CM, we visualized the latter in live MreBCD-depleted cells with a fusion of GFP to the N-terminal transmembrane domain of the bitopic CM protein ZipA (GFP) (45). The fluorescent fusion accumulated around each vesicle that was visible by DIC, suggesting that they were indeed surrounded by CM (Fig. 6B and C). Topologically, the lumens of these bodies are expected to correspond to extracytoplasmic space. If they are bounded by CM only, this space should correspond to the periplasm. Though unlikely, it is also conceivable that they are lined with both CM and outer membrane (OM), in which case the lumen is expected to be compartmentalized further. To probe these possibilities, we used a GFP fusion that is targeted to the periplasm via the twin arginine transport system (^TTGFP) (6). Panels G to I of Fig. 6 illustrate that ^TTGFP indeed accumulated in the lumen of each vesicle. In addition, the fusion distributed evenly within vesicular space, indicating that it was not compartmentalized further.

View larger version (93K): [in this window] [in a new window]   FIG. 6. Intracytoplasmic membrane compartments in MreBCD-depleted spheroids. Shown are live cells of the MreBCD depletion strain FB30/pFB174 (mreBCD/PBAD::mreBCD) producing either transmembrane GFP from lysogenic phage CH178 [Plac::zipA(1-183)-gfp] (A to C) or periplasmic TTGFP from plasmid pTB6 (Plac::sstorA-gfp) (D to J). Cells were grown to an OD600 of 0.2 at 37°C in LB with 50 µM IPTG and either with (A and D) or without (other panels) 0.5% arabinose. Both DIC (1) and GFP (2) fluorescence images are shown in panels A to I. Panels J show a time lapse series of a spheroid exposed to lysozyme. For this experiment, 2 µl of culture was applied to a slide and covered with a coverslip. A 1-µl aliquot of egg white lysozyme (100 µg/ml in GTE) was then pipetted against the edge of the coverslip, where it was drawn under by capillary action. GFP fluorescence was recorded immediately at 1-s intervals (J1 to J6). The arrowhead in panel J1 points to a compartment that quickly released TTGFP into the medium, in contrast to other compartments that retained the fusion throughout the procedure (arrows). Panel J7 shows a DIC image of the lysozyme-treated sphere a few seconds after the image in panel J6 was taken. Bar = 2 µm.

As mentioned above, only a subset of vesicle-like bodies showed accumulation of fluorescence at their peripheries when the membrane-impermeable CM dye FM4-64 was added to Mre^– spheres immediately before microscopy (Fig. 7A to C). In contrast, when spheres were pulse labeled with the dye 30 min prior to observation, virtually all bodies that were visible by DIC were now also clearly outlined by a fluorescent border (Fig. 7D to G). In addition, as was observed with the GFP and ^TTGFP probes, the dye accumulated in various other patterns that traversed the sphere's interior and that were often quite extensive. At the time of observation, the spheres in this experiment (Fig. 7D to G) had an average