INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Since the discovery of phosphatidylglycerol by Benson and Maruo in 1958 (3), the acidic phospholipid has been found in almost every organism and is believed to play essential roles in various cellular processes, such as SecA protein-dependent translocation of proteins across the membrane and rejuvenation of DnaA protein into an active ATP-bound form in the initiation of oriC DNA replication, based on results obtained mainly from in vitro studies (9, 32).

The Escherichia coli pgsA3 allele encodes a defective phosphatidylglycerophosphate synthase with low activity, resulting in cells defective in the major acidic phospholipids, phosphatidylglycerol and cardiolipin (23). The pgsA mutation is lethal in wild-type cells, while the cls mutations, resulting in reduction of cardiolipin content, affect their growth only slightly (13, 24). Therefore, the lethality and growth arrest phenotype of acidic phospholipid-deficiency caused by pgsA3 and reduced expression of pgsA, which is under the control of the lac promoter, are considered indications of the involvement of the acidic phospholipid in essential cellular functions (9, 11, 12, 32). The lethality is suppressed by the lack of or by certain mutational changes in the major outer membrane lipoprotein (Braun's lipoprotein), which consumes equimolar amounts of phosphatidylglycerol for processing of prolipoprotein, the primary gene product of lpp (2, 31). There are two alternatives to explain the suppression: (i) a drain of the limited acidic phospholipid pool required for certain essential functions in pgsA mutants is relieved by a lack of prolipoprotein or (ii) the limitation causes a harmful accumulation of unprocessed prolipoprotein in the inner membrane (9, 32). The null pgsA30::kan allele, which inactivated pgsA by the insertion of a kanamycin resistance gene (11), was reported not to be suppressed by the lack of the major outer membrane lipoprotein (36). Based on this result, Xia and Dowhan (36) suggested that there must be requirements for the acidic phospholipids even in the presence of the suppressor.

However, in this paper, we present evidence that E. coli null pgsA mutants are viable and grow almost normally, with no detectable phosphatidylglycerol and cardiolipin, in the presence of the suppressor. Furthermore, the previous inability to construct a null pgsA mutant, concluding in an incorrect notion that phosphatidylglycerol is essential, is explained by the characteristics of the null mutant. The results indicate that the quantitatively major acidic phospholipids, phosphatidylglycerol and cardiolipin, are not essential for the viability or basic life functions of E. coli.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacterial strains. Table 1 summarizes the E. coli K-12 strains used in this study. Strain S300 is a derivative of W3110 with a ksgB1 marker and is wild type with regard to phospholipid synthesis, and S301 is an lpp-2 derivative of W3110 (25). S303 is a pgsA3 derivative of S301 (25), and S330 is the pgsA30::kan derivative we constructed by P1 phage transduction (22) of an S301 recipient (see below). Another set of K-12 strains, JE5512 (Hfr man-1 pps) and JE5513 (JE5512 lpp-2) (2), were also used. Strain MDL12 [pgsA30::kan (lacOP-pgsA⁺)1 lacZ' lacY::Tn9] (36) was the gift from William Dowhan. We will deposit our construct of S330 strain with the parental strain in the E. coli genetic stock center after publication of this paper.

Media and growth conditions. Luria-Bertani (LB) medium (398 mosM/kg) containing 1% tryptone (Difco, Detroit, Mich.), 0.5% yeast extract (Difco), and 1% NaCl and minimal A medium supplemented with 0.02% MgSO₄, 0.2% glucose, 0.02% amino acid mixture, and 0.0001% thiamine hydrochloride (110 mosM/kg) were prepared as described previously (22). NBY medium of low osmolarity (108 mosM/kg) (34) and NBY medium supplemented with 1% NaCl (370 mosM/kg) were also used. Kanamycin and chloramphenicol were added, when necessary, to concentrations of 30 mg per liter. The motility test plate of LB medium contained 80 g of gelatin per liter, as described previously (18). Cells were grown at 37°C unless otherwise specified, and growth was monitored with a Klett-Summerson photoelectric colorimeter (no. 54 filter).

Phosphatidylglycerophosphate synthase assay. Phosphatidylglycerophosphate synthase in cell extract was assayed by measurement of the CDP-diacylglycerol-dependent incorporation of sn-[U-¹⁴C]glycerol-3-phosphate into the chloroform-soluble lipid fraction, as described previously (26). Specific activity is defined as nanomoles of the labeled substrate incorporated per milligram of protein per minute at 37°C.

Phospholipid analysis. Cells were labeled in NBY medium supplemented with 1% NaCl and with 7.5 µCi of ³²P_i/ml for six generations to late stationary phase at 37°C. Lipids were extracted by the method of Ames (1) and separated by two-dimensional thin-layer chromatography on silica gel plates (Silica gel 60; Merck) as described previously (33). The plates were developed first (x dimension) with chloroform-methanol-water (65:25:4 [vol/vol/vol]) and then (y dimension) with chloroform-methanol-acetic acid (65:25:10 [vol/vol/vol]), dried, and exposed to X-ray film for 30 h. After the labeled spots were scraped off the plate, the radioactivities were counted and calculated (see Table 3).

PCR analysis. DNA from fresh colonies of the strains to be examined was used as a template for PCR amplification with Ex Taq DNA polymerase (Takara, Tokyo, Japan). Primer FPPG5 (5'-CCGTCACCATGGAATTTAATATCCCTAC-3') starts from 8 bases upstream of the initiation codon of pgsA, and antisense primers ASFPPG1 (5'-CCCGAATTCATCAAGCAATCAG-3') and ASFPPG3 (5'-GTCAGTACTGCAGCCACAAAG-3') start from 156 bases downstream and 55 bases upstream, respectively, of the stop codon TGA (35). With the primer pair FPPG5 and ASFPPG1, the expected size of the product from the wild-type pgsA gene is 712 bp and that from the pgsA30::kan allele is 1.9 kbp by the insertion of a kan cassette (ca. 1.2 kbp) (11). With primer pair FPPG5 and ASFPPG3, the expected sizes of the products from the wild-type pgsA and the pgsA30::kan alleles are 499 bp and 1.7 kbp, respectively. MDL12 has the pgsA30::kan allele and lacOP-controllable pgsA, but the latter allele has no ASFPPG1 site.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Construction of a null pgsA mutant. During the course of the construction of a strain whose acidic phospholipid content is controlled by an exogenously added inducer, we transferred by P1 phage transduction the null pgsA30::kan allele of strain MDL12 (36) into lpp-2 mutants with or without a plasmid-borne and araBAD promoter-controllable pgsA gene (unpublished data). The null pgsA allele was constructed by Heacock and Dowhan by inserting a kanamycin resistance gene into a functionally important region of the pgsA gene (11). Unexpectedly, we found that a recipient strain, S301, that did not contain plasmid-borne pgsA, which was included as a negative control, yielded a large number (comparable to those of recipient strains harboring the plasmid-borne pgsA gene) of Kan^r transductants, indicating an efficient transfer of the pgsA30::kan allele into the recipient strain, irrespective of the presence of the intact pgsA gene. Since acidic phospholipids were considered essential for cell viability, the emergence of Kan^r transductants was rather surprising. Table 2 shows a typical result of P1 transduction, which demonstrated the viable nature of the null (pgsA30::kan) mutant when lpp, the structural gene for the major outer membrane lipoprotein (14), is defective (lpp-2). The table also shows that an unidentified mutation in strain JE5512 that partially suppresses a leaky mutation, pgsA3 (2), did not suppress the lethal nature of the null pgsA. The suppressive effect of the lpp-2 mutation was similarly observed in another lpp-2 mutant (JE5513) with a different genetic background. PCR analysis of three independent null transductants (strain S330) indicated that they all have pgsA30::kan alleles of the same size as that of strain MDL12 (1.9 kbp) and have no wild-type pgsA (0.71 kbp), as assessed with the primer pair FPPG5 (the sense primer at the initiation codon) and ASFPPG1 (the antisense primer at 156 bases downstream from the stop codon) (Fig. 1). With the primer pair FPPG5 and ASFPG3 (the antisense primer within pgsA located 214 bases upstream from ASFPPG1), a 1.7-kbp fragment of the pgsA30::kan allele alone (no DNA fragment of wild-type pgsA) was produced from the null transductants. Strain MDL12 gave DNA fragments (1.7 and 0.5 kbp) corresponding to pgsA30::kan and lacOP-controllable wild-type pgsA contained in MDL12 with the latter primer pair (12, 36).

View larger version (45K): [in this window] [in a new window]   FIG. 1.   PCR analysis of the pgsA alleles of E. coli pgsA mutants. PCR products amplified with Ex Taq DNA polymerase were subjected to 1.2% agarose gel electrophoresis. Lanes 1, W3110; lanes 2, S301; lanes 3, MDL12; lanes 4 to 6, independent clones of the transductants (S330). Primer pairs FPPG5-ASFPPG1 (a) and FPPG5-ASFPPG3 (b) were used. The design and sequences of the primers are described in Materials and Methods. With the former primer pair, the wild-type pgsA allele and the pgsA::kan allele gave products of 0.71 and 1.9 kbp, respectively. With the latter primer pair, the wild-type pgsA and the pgsA::kan alleles gave products of 0.5 and 1.7 kbp, respectively. A DNA fragment of ca. 1.5 kbp which appeared in MDL12 (panel b, lane 3) may be the product of a false annealing of the antisense primer with a site in lacZ' fused to pgsA (12). The molecular size markers included (two left lanes of each gel) were -HindIII digest (23.1, 9.4, 6.6, 4.4, 2.3, 2.0, and 0.56 kbp) and -EcoT14 I digest (19.3, 7.7, 6.2, 4.3, 3.5, 2.7, 1.9, 1.5, 0.93, and 0.42 kbp).

Phospholipid biosynthesis and composition of the null pgsA mutant. Figure 2 shows autoradiograms of lipid fractions heavily and uniformly labeled with ³²P_i. Strain S303 (pgsA3 lpp-2) contained very low but definitely detectable levels of phosphatidylglycerol and cardiolipin, as previously described (23), but a transductant strain, S330 (pgsA30::kan lpp-2), did not contain detectable levels of these acidic phospholipids. We calculated phospholipid compositions from the radioactivities of spots of these autoradiograms (Table 3); strain S303 contained phosphatidylglycerol and cardiolipin at 0.3 and 0.2 molar percent, respectively, of the total phospholipids, whereas in strain S330 we did not detect these acidic phospholipids at all. In strains S303 and S330, biosynthetic precursors for the major phospholipids, phosphatidic acid and (d)CDP-diacylglycerol, accumulated. Hence, the phospholipid polar head group composition of the mutant was 4.0% phosphatidic acid, 3.2% CDP-diacylglycerol, 91% phosphatidylethanolamine, and others. Despite these drastic abnormalities in polar head group composition, the total phospholipid contents in pgsA null mutants were not much different from that of the wild-type strain (150 nmol of lipid phosphorus per mg of cellular protein in both S303 and S330 compared to 180 nmol in S301). The activity of phosphatidylglycerophosphate synthase was undetectable in strain S330 (below the detection limit of 0.01 nmol of sn-[U-¹⁴C]glycerol 3-phosphate incorporated into chloroform-soluble products per min per mg of protein at 37°C, or less than 0.3% of that of the wild-type strain, W3110). The activity of strain S303 (pgsA3) was 0.22 nmol per min per mg of protein. These findings strongly suggested that the null pgsA mutants do not contain phosphatidylglycerol and cardiolipin at all, although we cannot exclude the possibility that very low levels of these lipids, at the most 0.01% of the total phospholipids or a few thousand molecules per cell (32), are present in the null mutants, satisfying the absolute requirements, if any, for the lipids.

View larger version (39K): [in this window] [in a new window]   FIG. 2.   Autoradiograms of 32P-labeled phospholipids of pgsA mutants. Cells of E. coli S301 (wild type) (a), S303 (pgsA3) (b), and S330 (pgsA30::kan) (c) were grown at 37°C in NBY medium supplemented with 1% NaCl in the presence of 7.5 µCi of 32Pi/ml for six generations (to the late-exponential growth phase). The lipids were extracted and separated by two-dimensional thin-layer chromatography (Silica gel 60; Merck) as described previously (34). After the preparation of autoradiograms, the spots were scraped off the plates for measurement of their radioactivities (Table 3). CL, cardiolipin; PE, phosphatidylethanolamine; PA, phosphatidic acid; PG, phosphatidylglycerol.

Phenotypes of the null pgsA mutant. Despite the apparent absence of the major acidic phospholipids, cells of the null pgsA mutant grew almost normally at both 30 and 37°C in a rich LB medium (Fig. 3). However, they did not grow at higher temperatures: the turbidity increase ceased in 2 h when the cells were shifted from 37 to 40°C and more rapidly when they were shifted to 42°C (Fig. 3c). Two hours after the temperature shift to 42°C, the culture turbidity began to decrease. The plating efficiency of the null mutants at 42°C on LB plates was 10⁵ relative to that at 37°C. This temperature sensitivity of the null mutants explains the previous inability to cure a temperature-sensitive covering plasmid (bearing the wild-type pgsA gene) in a pgsA30::kan lpp-2 double mutant at a high temperature, thus leading to the incorrect conclusion that phosphatidylglycerol was essential (36). In order to cure the covering pgsA⁺ plasmid, Xia and Dowhan (36) raised the temperature to 42°C. This treatment should have hampered the growth and colony formation of the pgsA30::kan lpp-2 double mutant, thus yielding no viable colonies. Hence, they were led to the incorrect notion that the pgsA gene was not cured, i.e., that phosphatidylglycerol was indispensable.

View larger version (27K): [in this window] [in a new window]   FIG. 3.   Growth characteristics of E. coli pgsA mutants. Cells of E. coli S301 (wild type) (a), S303 (pgsA3) (b), and S330 (pgsA30::kan) (c) were grown in LB medium at 37°C to Klett 5 to 7. They were cultured further at 37°C () or transferred to 30 () or 42°C (), and Klett units were measured every hour.

Nonessential nature of the major acidic phospholipids. From the present observation that the major acidic phospholipids are absent in null pgsA mutants, we reach a tentative conclusion that these lipids are not essential for the viability of E. coli cells. This notion, together with the previous observations, now allows us to formulate a working model for the physiological roles of acidic phospholipids in E. coli. The roles are grouped into two categories: (i) headgroup-specific phospholipids as the substrates of enzymatic reactions, which are dispensable if the reaction products are dispensable, and (ii) acidic phospholipids, which are replaceable by other phospholipids with net negative charges. These two groups of putative functions will be discussed in some detail below.

Suppression by lipoprotein deficiency. The lpp gene product is dispensable (14), but it is absolutely related to the viability of pgsA mutants (2) (Table 2 and Fig. 3). Among the headgroup-specific functions, the maturation of the lpp gene product, prolipoprotein, which consumes equimolar amounts of phosphatidylglycerol (31), is the only exception. During the maturation of prolipoprotein, it first receives the diacylglycerol moiety of phosphatidylglycerol on the Cys-21 residue, rendering it susceptible to signal peptidase II (31). In the cells lacking phosphatidylglycerol, the modification should be impaired, and unmodified prolipoprotein seems to remain in the inner membrane, probably because of defective translocation of prolipoprotein due to the lack of phosphatidylglycerol (8, 38). Recently, Yakushi et al. (37) indicated that the inner membrane accumulation of prolipoprotein per se is not lethal but that a covalent linking between the inner membrane and peptidoglycan through the COOH-terminal lysine of the prolipoprotein is lethal, presumably because cell surface integrity is disrupted. Accordingly, we assume that the reason for the lethality due to the lack of phosphatidylglycerol in lpp⁺ cells is a disruption of cell envelope integrity by means of a covalent linking of peptidoglycan and the inner membrane through the unmodified prolipoprotein accumulated in the membrane.

Functional replacement with other acidic phospholipids. As to the functions of phospholipids that require only net negative charges, the accumulation of phosphatidic acid in pgsA mutants (Table 2) should be important for cell viability: functional replacement of phosphatidylglycerol with other acidic phospholipids, including phosphatidic acid, has been described for several peripheral membrane protein functions in vitro (5, 9, 10, 19, 29, 32). One example is the in vitro promotion by acidic phospholipids of the release of ADP from DnaA protein, by which an inactive ADP form of the initiator protein in the oriCDNA replication rejuvenates into an active ATP-bound form (5, 10). The physiologically important states of phospholipids have been shown to be those in the fluid phase (i.e., above the transition temperature) and a negatively charged headgroup, such as phosphatidic acid, phosphatidylserine, and ganglioside G_M1. DnaA protein appears to interact with the negatively charged surface of the membrane by a hydrophobic side of a certain helical-wheel region of this protein (10). As described for SecA (4), DnaA protein may partly insert into the inner membrane after recognizing a domain of several negatively charged headgroups on the surface of the membrane (9). Even if phosphatidylglycerol and cardiolipin are present at levels below the detectable limit (less than 0.01% of total phospholipids or a few thousand molecules per cell) in pgsA30::kan lpp-2 mutants, these amounts seem to be too small to provide DnaA protein with enough headgroups of phosphatidylglycerol and cardiolipin at the membrane surface (9). Most probably with the phosphatidic acid accumulated in the null mutant (4.0% of the total phospholipids [Table 3]), the initiation of DNA replication should take place, since the null mutants, with no phosphatidylglycerol and cardiolipin, grew almost normally. Hence, the lethality and growth arrest phenotype of the lpp⁺ cells with reduced phosphatidylglycerol and cardiolipin (2, 12) should not be linked to the suggested inability to initiate DNA replication (36).