Gene-Specific Random Mutagenesis of Escherichia coli In Vivo: Isolation of Temperature-Sensitive Mutations in the Acyl Carrier Protein of Fatty Acid Synthesis

Abstract

Acyl carrier proteins (ACPs) are very small acidic proteins that play a key role in fatty acid and complex lipid synthesis. Moreover, recent data indicate that the acyl carrier protein of Escherichia coli has a large protein interaction network that extends beyond lipid synthesis. Despite extensive efforts over many years, no temperature-sensitive mutants with mutations in the structural gene (acpP) that encodes ACP have been isolated. We report the isolation of three such mutants by a new approach that utilizes error-prone PCR mutagenesis, overlap extension PCR, and phage λ Red-mediated homologous recombination and that should be generally applicable. These mutants plus other experiments demonstrate that ACP function is essential for the growth of E. coli. Each of the mutants was efficiently modified with the phosphopantetheinyl moiety essential for the function of ACP in lipid synthesis, and thus lack of function at the nonpermissive temperature cannot be attributed to a lack of prosthetic group attachment. All of the mutant proteins were largely stable at the nonpermissive temperature except the A68T/N73D mutant protein. Fatty acid synthesis in strains that carried the D38V or A68T/N73D mutations was inhibited upon a shift to the nonpermissive temperature and in the latter case declined to a small percentage of the rate of the wild-type strain.

Two systems for fatty acid biosynthesis, called types I and II, are found in nature. Type I systems occur in the cytosols of mammalian and plant cells, in some plant plastids, and in a few bacteria. In these systems, the steps of fatty acid biosynthesis are catalyzed by one or two very large polypeptides that contain multiple active sites (47). In contrast, the type II systems which occur in bacteria, mitochondria, most plant plastids, and protozoan apicoplasts produce fatty acids using a series of soluble enzymes in which each protein has a discrete enzymatic activity (38). A key component of both systems is acyl carrier protein (ACP) which constitutes a domain of the type I multifunctional proteins but is a very small, soluble, and highly acidic protein in type II systems. ACP is functional only in fatty acid biosynthesis after it has been posttranslationally modified by covalent attachment of a 4′-phosphopantetheinyl (4′-PP) moiety. The 4′-PP prosthetic group is attached to the hydroxyl group of a centrally located serine residue by the AcpS 4′-PP transferase. Acyl intermediates are bound to the 4′-PP thiol in a thioester linkage that allows ACP to shuttle intermediates among the fatty acid synthetic enzymes. The thioester linkage also serves to facilitate chemistry at acyl chain β-carbons.

Escherichia coli contains a single ACP encoded by the acpP gene (35). This was the first ACP discovered (40, 52) and remains the best studied of these proteins. Despite the breadth and depth of studies of E. coli ACP and its acylated derivatives, no mutant strain encoding a mutant ACP has been reported. However, ACP seemed certain to be an essential protein since mutants in several fatty acid synthetic genes that encoded proteins requiring ACP-bound substrates were essential genes (10, 24, 43, 45, 46). We and others assumed that conditionally lethal (temperature-sensitive) acpP mutants would be isolated by the general selections that successfully gave strains containing conditionally lethal defects in other genes acting early in (or throughout) the fatty acid synthetic pathway (14, 24). However, such mutants were not forthcoming. Later some of the other genes defined by the mutants isolated in these selections were found to map in a cluster of fatty acid synthesis genes that includes acpP, thereby suggesting that acpP was in some way refractory to mutant isolation.

The lack of conditionally lethal acpP mutants is a serious shortcoming of the bacterial physiologists toolbox not only because of the importance of lipid synthesis to cell structure and function, but also because ACP has recently been reported to have more interactions with other proteins than any other discrete E. coli protein (1). The only rivals in terms of the numbers of interacting partners are all large complexes containing multiple proteins: DNA polymerase III, RNA polymerase, and the ribosome. The list of proteins that interact with ACP included not only the expected lipid synthetic proteins but also proteins involved in diverse cellular functions (1). Some of these diverse proteins were MukB, a protein involved in the partition of chromosomes between daughter cells; SecA, a key protein translocation component; and SpoT, a protein required for the adaptation of cell physiology to nutrient limitation. Moreover, ACP had previously been reported to be an essential cofactor for the synthesis of a periplasmic oligosaccharide in vitro (49, 50), to stimulate the nicking reaction of transposon Tn3 (26), and to give improved binding of Tn7 transposase to its target sequence (44).

In addition to allowing us to test the more global effects of the loss of ACP, temperature-sensitive acpP mutants would permit more-thorough structure-function studies of the interactions of ACP with the enzymes of lipid synthesis. Although a large number of ACP mutant proteins containing targeted mutations have been made (8, 17, 58, 59), these mutant proteins have been tested only in vitro as substrates for one or two enzymes or with the E. coli in vitro fatty acid synthetic synthesis system, an insensitive test because the system requires only a subset of the enzymes required in vivo and functions at only a few percentages of the cellular rate of fatty acid synthesis (11, 25).

Given that prior attempts to obtain acpP(Ts) mutant strains had consistently failed, we developed a new localized mutagenesis technique that takes advantage of mutagenesis by error-prone PCR (54), overlap extension PCR (18), and λ Red-mediated recombination. Using this approach, three acpP conditional mutants were obtained. We report our mutagenesis approach in addition to our characterization of these acpP mutants.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains used in this study were all derivatives of strain MG1655, an E. coli K-12 strain of known genome sequence. The strains and plasmids used in this study are listed in Table 1. The primers used in this study (Table 2) were purchased from Integrated DNA Technologies, Inc. The panD deletion strain NRD1 was created via λ Red recombinase-mediated gene replacement using the PCR product generated from the template, pKD3, using primers panD2K12-For and panD2K12-Rev (55). The ΔpanD mutation was transduced via a phage P1vir lysate into strain MG1655 to yield strain NRD38, and the cat resistance cassette was subsequently removed using the FLP recombinase encoded by pFT-A (32) to yield strain NRD44. Strain NRD23 was generated via λ Red recombinase-mediated gene replacement using the PCR product produced from the template, pKD3, using primers fabFcmIns-For and fabFcmIns-Rev. The strain was subsequently cured of this plasmid by growth at 42°C. The ΔfabF mutation was transduced from strain NRD23 into strain MG1655 via P1vir to generate strain NRD52. The acpP(Ts) mutations were transduced via P1 lysates grown on strains NRD51 and NRD53 into strain MG1655 to yield strains NRD28 and NRD29, respectively. Strain NRD62 was generated by first transforming strain DY330 with plasmid pNRD25, which carries the wild-type acpP gene, and subsequently deleting the chromosomal copy of acpP by λ Red-mediated gene replacement using a PCR product generated from pKD3 using primers acpKO-for and acpKO-rev, whereas the synthetic acpP gene (4, 17, 35) was expressed from the pBAD promoter of pNRD25. Sequencing of both strands of the acpP alleles was performed by the Core Sequencing Unit at the Keck Center for Comparative and Functional Genomics at the University of Illinois using either pNRD26seq-For and -Rev primers or M13For (−21) and M13Rev (−24) primers.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant features	Source or reference
Strains
DY330	W3110 λc1857 Δ(cro-bioA)	55
DH5α	recA1 endA1	Strain collection
CY321	acpP V43I fabF1	51
MC1061	araD139 Δ(ara-leu)7696	Strain collection
MG1655	Wild type	Strain collection
Top10F′	F′ lacIq Tn10 (Tetr)/φ8::lacZ ΔM15 ΔlacX74 recA1 endA1	Invitrogen
NRD1	DY330 ΔpanD::cat	This study
NRD23	MC1061 ΔfabF::cat	This study
NRD28	acpP(Ts) A68T N73D; ΔfabF::cat was transduced via P1vir from NRD51 into MG1655	This study
NRD29	acpP(Ts) D38V; ΔfabF::cat was transduced via P1vir from NRD53 into MG1655	This study
NRD38	ΔpanD::cat transduced via P1vir from NRD1 into MG1655	This study
NRD44	MG1655 ΔpanD	This study
NRD50	MC1061 acpP(Ts) E49G ΔfabF::cat	This study
NRD51	MC1061 acpP(Ts) A68T N73D ΔfabF::cat	This study
NRD52	ΔfabF::cat transduced via P1vir from NRD23 into MG1655	This study
NRD53	MC1061 acpP(Ts) D38V ΔfabF::cat	This study
NRD62	DY330 ΔacpP::cat carrying pNRD25	This study
Plasmids
pBAD322	Ampr, araBAD promotor-based expression vector having a complete pBR322 origin derived from pBAD24 (13)	2
pDK685	Site-directed acpP mutant, encodes AcpP S36A	23
pKD3	oriRγ, Cmr cassette flanked by FRTa sites	5
pKD46	Ampr, RepA1019(Ts), λ γ, β, and exo expressed from an araBAD promotor	5
pMR19	Ampr, synthetic acpP allele under the control of the tac promotor	23
pNRD25	EcoRI-HindIII acpP-containing fragment from pMR19 cloned into the same sites of pBAD322	This study
pNRD77	EcoRI-HindIII acpP-containing fragment from pDK685 cloned into the same sites of pBAD322	This study
pNRD119	Site-directed mutation in acpP in pNRD25 was generated by Quickchange PCR mutagenesis using primers ACPD38V For and Rev; encodes AcpP D38V	This study
pNRD122	Mutant of acpP in pNRD25 was made using primers ACPN73D For and Rev; encodes AcpP A68T	This study
pNRD122	Mutant of acpP in pNRD25 was made using primers ACPN73D For and Rev; encodes AcpP N73D	This study
pNRD128	ClaI fragment from pNRD121 containing G205A mutation cloned into the ClaI fragment from pNRD122, containing A220G mutation; encodes AcpP A68T N73D	This study

^aFRT, FLP recombinase target.

TABLE 2.

Primers used in this study

Primer	Sequence (5′−3′)
panD2K12 For	CTGGAGACGATGTCGCGGCTGGTGAGTAACCAGCCGCAGGGATAAGTGTAGGCTGGAGCTGCTTC
panD2K12 Rev	TCCTCAAAGCAGGCATCGCCTGCTTCGTTAACGACAGGGTAGAAACATATGAATATCCTCCTTAG
fabFcmins For	ATGTTGTCTCCTGTCGGCAATACCGTAGAGTCTACCTGGAAAGCTGTGTAGGCTGGAGCTGCTTC
fabFcmins Rev	TACGCGCAGCATGGCCAATGTTGTGCACGCCGGAAGTACAGGCAGCATATGAATATCCTCCTTAG
acpKO For	ACGAAAACCATCGCGAAAGCGAGTTTTGATAGGAAATTTAAGAGTGTGTAGGCTGGAGCTGCTTC
acpKO Rev	AAAGATAAAACTCAGGCGGTCGAACGACCGCCTGGAGATGTTCACCATATGAATATCCTCCTTAG
acp2 For	CGAGTTTTGATAGGAAATTTAAGAGTATG
acp2 Rev	CTTAGACACGTTTGTCCTCC
acp3 For	CGTAAAATCGTGGTAAGACCTG
acpoverlap3comp Rev	AACTCAGGCGGTCGAACG
acpoverlap3 For	CGTTCGACCGCCTGAGTT
fabFmid Rev	CACGTCAGCATCGCCATA
ACPD38V For	GGTGCTGACTCTCTGGTCACCGTTGAGCTGGTTATGG
ACPD38V Rev	CCATAACCAGCTCAACGGTGACCAGAGAGTCAGCACC
ACPA68T For	CACCACCGTACAGGCTACTATCGATTACATCAACGG
ACPA68T Rev	CCGTTGATGTAATCGATAGTAGCCTGTACGGTGGTG
ACPN73D For	GCTGCTATCGATTACATCGACGGTCACCAGGCTTAAGC
ACPN73D Rev	GCTTAAGCCTGGTGACCGTCGATGTAATCGATAGCAGC
pNRD26seq For	GATTAGCGGATCCTACCTGAC
pNRD26seq Rev	CAGACCGCTTCTGCGTTC

Culture media and growth conditions.

Strains were grown in liquid medium or on agar plates of rich broth (RB) (24), Luria-Bertani broth (LB), or minimal E salts (53). E medium was supplemented with final concentrations of glycerol (0.8%), glucose (0.2%), and arabinose (0.2%), unless specified otherwise. Antibiotics were used at the following final concentrations (in mg/liter): chloramphenicol, 30; kanamycin, 50; and a combination of the β-lactam ticarcillin and the β-lactamase inhibitor clavulanate (Timentin; GlaxoSmithKline), 25. For β-alanine starvation, β-alanine was used at a final concentration of 0.5 nM. For in vivo phosphopantetheinylation experiments, β-alanine (specific activity, 60 Ci/mmol; American Radiolabeled Chemicals) was used at a final concentration of 0.5 μM.

Mutagenesis.

The mutagenesis procedure used combined error-prone PCR mutagenesis (54) and λ Red recombinase-mediated allele replacement (Fig. 1) (5). Briefly, acpP was amplified from NRD23 using primers ACP3-For and ACPoverlap3comp-Rev. fabF::cat was amplified from NRD23 using primers ACPoverlap3 and FabFmid-Rev. Primers ACPoverlap3 comp-Rev and ACPoverlap3 contained 20 bp of overlapping sequence to facilitate overlap extension PCR as described below. The products of both PCRs were fractionated by agarose gel electrophoresis, and the bands of appropriate size were excised from the agarose gels and purified using the QIAquick gel extraction kit. The acpP fragment was then used as the template for either 12 or 35 rounds of random PCR mutagenesis at a final concentration of 200 pg/μl (54). The products of these mutagenic PCRs was fractionated by agarose gel electrophoresis and purified as described above.

FIG. 1.

PCR mutagenesis coupled to λ Red recombinase-mediated allele replacement. (A) The acpP gene and the downstream fabF::cat gene were amplified to produce fragments that overlapped by 20 bp. The acpP genes were then mutagenized using 12 or 35 rounds of mutagenic PCR. The mutagenized acpP genes were then combined with the fabF::cat PCR product, and overlapping extension PCR was performed. (B) The resulting population of mutagenized acpP genes linked to fabF::cat was then transformed into strain MC1061 induced for λ Red recombinase. After recovery, the cells were plated at 30°C with selection for chloramphenicol resistance. The colonies that formed after 48 h were patched onto three plates, each of which was incubated at a different temperature (30, 37, or 42°C) to screen for colonies showing temperature-sensitive growth. TS, temperature-sensitive.

Overlap extension PCR (18) using primers ACP3-For and FabFmid-Rev was performed to generate DNA fragments containing the acpP fragment from either 12 or 35 rounds of mutagenic PCR plus the fabF::cat fragment (Fig. 1). The final concentrations of the acpP fragment and the fabF fragment in the PCR mixture were 30 pg/μl and 60 pg/μl, respectively. Ethanol precipitation of the products of each PCR was performed, and the DNA was resuspended in 8 μl of water. λ Red-induced MC1061 (pKD46) cells were transformed with the PCR product (5). After phenotypic expression, the cells were plated at 30°C on RB-chloramphenicol plates. The colonies that arose were screened for the ability to grow at 37°C or 42°C. Three temperature-sensitive strains, NRD50, NRD51, and NRD53, were isolated using this procedure. With the primers ACP2 For and ACP2 Rev, the acpP genes from these strains were amplified by PCR. The products from at least two independent PCRs generated from each strain were cloned into pCR2.1 using a TOPO TA cloning kit (Invitrogen), and both strands of each cloned PCR product were subsequently sequenced using pNRD26seq For and pNRD26seq Rev primers. Strains NRD50, NRD51, and NRD53 possessed the following base substitutions in their acpP genes: A149G, G205A and A220G, and A116T, respectively. Sequencing as described above was done of clones derived from two independent PCRs amplified from the genomic DNAs of the original strains and, in the case of the D38V and A68T/N73D mutants, of two clones amplified from the genomic DNAs of each of the MG1655 strains derived by transduction from the original strains. Consistent sequences were obtained for each mutation.

Complementation assays.

Derivatives of strain NRD29, each harboring a plasmid containing the acpP allele to be tested under the control of the araBAD promoter, was streaked on two RB-ticarcillin-clavulanate-glucose plates and an RB-ticarcillin-clavulanate-arabinose plate. One glucose-supplemented plate was incubated at 30°C for two nights, and the other two plates were incubated at 42°C overnight.

Analysis of in vivo phosphopantetheinylation.

The in vivo analysis of phosphopantetheinylation was performed using strains of NRD44 harboring an acpP allele that encoded the wild-type, S36A, D38V, E49G, A68T, or A68T N73D AcpP from a vector paraBAD promoter. The strains were first starved for β-alanine overnight at 30°C on minimal E-glycerol-ticarcillin-clavulanate medium plates containing trace amounts of β-alanine. The cells of each strain were resuspended in two tubes to an optical density at 600 nm (OD₆₀₀) of 0.6 to 0.7 in the same medium lacking β-alanine. Arabinose and [³H]β-alanine were then added to all tubes, and one tube of each strain was incubated at 30°C and the other tube was incubated at 42°C for 3 h. The cells were harvested by centrifugation, resuspended in 100 mM sodium 2-(N-morpholino)ethanesulfonic acid (MES) buffer (pH 6.1), and lysed by sonication. The insoluble cell debris was removed by centrifugation. The amount of protein present in the supernatant was quantified using the Bio-Rad protein assay kit. Equal amounts of protein from each sample were then fractionated on a 20% native polyacrylamide gel. The gels were fixed, soaked in NRAMP100V (Ambion), dried, and then exposed to X-ray film (Kodak BIOMax XAR film).

Measurements of fatty acid synthetic rates.

Strains NRD28, NRD29, and NRD52 were grown with shaking overnight at 30°C in RB medium. The overnight cultures were then diluted 100-fold into fresh RB liquid medium and grown with shaking at 30°C to an OD₆₀₀ between 0.4 and 0.6. These cultures were subsequently diluted 10-fold into fresh RB liquid medium. The strains were grown while shaking at 30°C to an OD₆₀₀ of ∼0.2. A 1.0-ml sample was then removed and added to a prewarmed tube containing 5.0 μCi of [1-¹⁴C]acetate. The tube was incubated at 30°C for 10 min while being shaken at 215 rpm and then treated as described below. The rest of the culture was divided into two flasks. One flask was incubated at 30°C, whereas the other was incubated at 44°C. The OD₆₀₀ of each culture was measured at various time points after incubation. Additional 1.0-ml samples were removed from the flask incubated at 44°C after 30, 60, and 120 min of incubation. During the labeling experiment, the culture of the wild-type strain, NRD52, was periodically diluted with prewarmed medium in order to maintain exponential growth. Those samples were added to prewarmed flasks containing 5 μCi of [1-¹⁴C]acetate and incubated at 44°C for 10 min while being shaken at 215 rpm. After incubation with [1-¹⁴C]acetate, 2.0 ml of methanol and 1.0 ml of chloroform were added to each tube. The tubes were then incubated at 30°C for at least 1 h. The insoluble debris was removed by centrifugation at 5,000 × g, and 1.0 ml of H₂O and 1.0 ml of chloroform were added to each tube. The tube was vortexed and then centrifuged at 5,000 × g for 5 min. The organic phase was placed into a new tube and evaporated under a stream of N₂. The samples were then suspended in 40 μl of a 2:1 mixture of chloroform-methanol, and the entire sample was loaded onto an Analtech Silica Gel G plate that had previously been activated by heating at 80°C for 1 h. The thin-layer chromatography plate was subsequently developed with a mobile phase of chloroform, methanol, and acetic acid (65:25:8). The plate was dried and then exposed to KODAK BioMax XAR film. The sections of the thin-layer chromatography plate containing phospholipids for each strain were then scraped off the plate into a scintillation vial. Bio-Safe II scintillation fluid (Research Products International Corp.) was then added to each tube, and the amount of ¹⁴C incorporated into the phospholipids of each strain was quantified using a Beckman LS 6500 multipurpose scintillation counter.

Measurement of ACP turnover.

The in vivo analysis of ACP turnover was performed using strain CY321 carrying a plasmid containing the acpP allele to be tested under the control of the araBAD promoter. The strains were grown to early log phase in RB-ticarcillin-clavulanate liquid medium, and then arabinose was added to induce ACP expression. After 1 h, the cells from each 40-ml culture were pelleted at 6,000 × g for 5 min at 6°C, washed once with 40 ml of RB-ticarcillin-clavulanate-glucose medium, and resuspended in 40 ml of RB-ticarcillin-clavulanate-glucose medium. A 4-ml sample was removed from the cell suspension, which was subsequently divided into two cultures. One culture was incubated at 30°C, and the other was incubated at 42°C. Samples of 4 ml were then removed from each culture after 30, 60, 120, and 240 min. The cells from each of the samples removed from these cultures were pelleted at 4°C. The cells were suspended in 0.3 μl of 100 mM sodium MES buffer (pH 6.1). The cell suspensions were sonicated, followed by removal of the cell debris by centrifugation at 18,000 × g for 30 min and addition of dithiothreitol to a final concentration of 100 mM. An equal volume of cell extract from each sample was fractionated on a 20% native polyacrylamide gel at 120 V and then transferred to an Immobilon-P membrane (Millipore). The membrane was blocked with TTBS (25 mM Tris-HCl, pH 7.4, 17.1 mM NaCl, 2.7 mM KCl, 0.05% Tween 20) containing 5% nonfat milk powder and then washed with TTBS. The ACP proteins were detected by Western blotting using the primary antibody anti-AcpP immunoglobulin Y (prepared by Aves Laboratories), the secondary antibody anti-immunoglobulin Y conjugated to horseradish peroxidase (Pierce), the ECL Plus chemiluminescent substrate (Amersham Biosciences) and ECL Hyperfilm (Amersham Biosciences).

RESULTS

Isolation of acpP(Ts) mutants via λ Red-mediated recombination of mutagenized acpP genes.

We believed that the prior lack of success in obtaining acpP(Ts) mutants was due to the inadequacy of existing techniques to generate temperature-sensitive mutations in small essential genes. To address this possibility, we developed a novel combination of existing techniques (error-prone PCR mutagenesis, overlap extension PCR, and λ Red-mediated recombination) to generate temperature-sensitive acpP mutants (Fig. 1). From the 100 colonies produced using this procedure with 12 rounds of mutagenic PCR, two temperature-sensitive mutants were obtained, strains NRD53 and NRD50. A third temperature-sensitive mutant, strain NRD51, was obtained by screening the 44 colonies that arose from the MC1061(pKD46) cells transformed with the PCR product obtained after 35 rounds of mutagenic PCR. Strain NRD50 showed leaky growth at 42°C, whereas mutant strain NRD53 failed to grow at 42°C (Fig. 2) and strain NRD51 failed to grow at either 37°C or 42°C. Introduction of a plasmid-borne wild-type acpP gene restored the ability of all three temperature-sensitive mutants to grow at nonpermissive temperatures (Fig. 2A).

FIG. 2.

Complementation analysis of the temperature-sensitive mutants and demonstration of the essentiality of acpP. (A) The three temperature-sensitive mutants isolated by mutagenesis were tested for the ability to grow at 42°C. Each of these mutant strains was then transformed with a plasmid expressing wild-type E. coli AcpP under the control of the arabinose-inducible paraBAD promoter. The resulting strains were streaked onto RB medium containing either arabinose (induction conditions) or glucose (repression conditions), and the plates were incubated at 42°C. (B) The plasmids encoding wild-type, A68T, N73D, or A68T N73D AcpP species were tested for the ability to allow the growth of the acpP(Ts) mutant strain NRD29 at 42°C. (C) Strains DY330 and NRD62, a derivative of DY330 harboring an acpP deletion, carrying a wild-type acpP allele under the control of an araBAD promoter (pNRD25) were tested for the ability to grow in the presence of arabinose or glucose. Note that the growth of the wild-type strain MC1061 containing the wild-type AcpP-encoding plasmid was inhibited by arabinose but that a similar derivative of the other wild-type strain, DY330, was not. Inhibition of growth by overproduction of ACP is due to the accumulation of apo-ACP (the unmodified form), which inhibits phospholipid synthesis (22). We attribute the differential effects seen to the inability of strain MC1061 to metabolize arabinose, which increases the extent of induction compared to that of arabinose-catabolizing strains such as DY330 (13, 27). WT, wild type.

Upon sequencing, each of the three temperature-sensitive mutant strains was found to carry a different acpP mutation(s). The acpP genes of strains NRD50 and NRD53 contained the base substitutions A149G and A116T, respectively, and thus encoded proteins having a single mutation, E49G and D38V, respectively. Strain NRD51 contained two base substitutions, G205A and A220G, that resulted in the expression of a protein having two mutations, A68T and N73D. Reconstruction experiments done by introduction of each of two mutations of strain NRD51 into the wild-type gene showed that the A68T mutation was mainly responsible for the temperature-sensitive phenotype (Fig. 2B), as measured by complementation of the D38V strain. Although the N73D mutation had little or no effect on growth at 42°C, this mutation seemed to potentiate the effects of the A68T mutation since the acpP(Ts) strain carrying the doubly mutant gene grew more poorly at the nonpermissive temperature than did the same strain with the plasmid encoding the A68T protein. Each of the acpP(Ts) mutations was backcrossed into the sequenced wild-type strain MG1655 using phage P1 transduction and selection of the linked chloramphenicol resistance determinant. The transductants, all of which showed temperature-sensitive growth, were used to perform the following experiments. Since strain NRD50 (the E49G mutant) showed leaky growth at the nonpermissive temperature, this strain was not characterized as fully as the other two mutant strains.

Inhibition of growth and fatty acid synthesis of the acpP(Ts) mutants at the nonpermissive temperature.

The growth of the MG1655 derivatives carrying the D38V mutation or the A68T/N73D mutations was followed before and after the shift to the nonpermissive temperature of 44°C (Fig. 3A and B). Both mutant strains grew more slowly than the wild-type strain at 30°C, the permissive temperature (Fig. 3A). Upon the shift to 44°C (Fig. 3B), strain NRD28, which carried the A68T/N73D lesions, ceased normal growth after 30 min and ceased growth altogether within about 150 min. In contrast, although strain NRD29 carrying the D38V mutation also ceased normal growth after 30 min, a lower growth rate then ensued, suggesting that the mutant ACP retained some residual function at the nonpermissive temperature (although growth ceased after about 9 h at 44°C). The rates of fatty acid synthesis in these strains were assayed by incorporation of [1-¹⁴C]acetate (Fig. 3C). At each of the time points given, a sample of each culture was removed to a tube containing [¹⁴C]acetate in the same shaking water bath. After we allowed incorporation for 10 min, a mixture of chloroform and methanol was added to each tube to terminate the incorporation and extract the lipids. At 30°C, both mutant strains made fatty acids at about one-third the rate of the wild-type strain. Upon the shift to 44°C, the fatty acid synthetic rate of the D38V mutant strain declined to about 10% of that of the wild-type strain and stayed at that rate for the 2-h duration of the experiment. In contrast, the synthetic rate of the A68T/N73D mutant strain progressively declined to a rate that was only a small percentage of that of the wild-type strain. Therefore, these mutations inhibited lipid synthesis as expected. It should be noted that E. coli can grow and synthesize fatty acids almost normally with only about 30% of its normal content of holo-ACP (9, 19).

FIG. 3.

Analysis of growth and fatty acid synthesis in two acpP(Ts) mutants and the parental strain. Strains NRD28 (encoding AcpP A68T/N73D), NRD29 (encoding AcpP D38V), and NRD52 (encoding wild-type AcpP) were grown at 30°C to an OD₆₀₀ of ∼0.2, and the culture was then split into two flasks. One flask was incubated at 30°C, and the second flask was incubated at 44°C. The turbidities of the cultures were then measured at various time points. (A) Growth curves of strains NRD52 (squares), NRD28 (triangles), and NRD29 (circles) at 30°C. (B) Growth curves of the same strains after a shift of the exponentially growing cultures from 30°C to 44°C. The symbols are the same as those used for panel A. (C) Rates of fatty acid synthesis (labeled acetate incorporation [incorp.]) at various time points after shifting of the cultures of the three strains to 44°C. Fatty acid synthesis rates were determined as described in Materials and Methods. The symbols are the same as those used for panel A. Note that the y axes are log scales.

acpP is an essential gene.

The isolation of three conditional acpP mutants indicted that acpP is an essential gene. To confirm the essentiality of acpP, we deleted the chromosomal copy of acpP in the presence of a plasmid-borne synthetic acpP gene that was expressed from an arabinose-inducible promoter and then tested the ability of the resulting strain, NRD62, to grow in the presence or absence of inducer (Fig. 2C). The ΔacpP strain NRD62 grew only in the presence of arabinose, whereas an isogenic strain having an intact chromosomal acpP gene in addition to that carried by pNRD25 grew in the presence and absence of arabinose.

Phosphopantetheinylation of the temperature-sensitive AcpP proteins in vivo.

To be functional in lipid synthesis, ACPs must be posttranslationally modified via the transfer of a 4′-PP group from coenzyme A (9, 31, 48). The temperature sensitivity of our AcpP mutants may have resulted from the inability of these altered ACPs to be posttranslationally modified at 42°C. Therefore, we tested the ability of these three mutant ACPs to be phosphopantetheinylated at 30°C and 42°C. To do this, we expressed the plasmid-encoded E49G, A68T, and D38V ACPs in NRD44, a panD deletion strain, in minimal medium containing β-alanine, and the cell extracts were then fractionated on native polyacrylamide gels and exposed to film. All three mutant ACP proteins demonstrated greater levels of 4′-PP attachment at 42°C than at 30°C (Fig. 4). Therefore, the inability of these acpP(Ts) strains to grow at nonpermissive temperatures could not be attributed to a lack of prosthetic group attachment.

FIG. 4.

Analysis of 4′-PP attachment to the temperature-sensitive ACPs (with the mutations D38V, E49G, and A68T) at 30°C and 42°C. The protein encoded by the plasmid is shown above the lane (in the lane marked “None,” the strain carried the empty vector). The ΔpanD strain NRD44 harboring plasmids that encoded the various mutant AcpPs expressed from the araBAD promoter was first starved for β-alanine at 30°C on minimal E medium-glucose plates. The cells were then resuspended in minimal E liquid medium. Arabinose and β-alanine were then added to final concentrations of 0.2% and 0.5 μM, respectively. One culture of each strain was incubated at 30°C and the other at 42°C for 3 h. The cells were then harvested by centrifugation, resuspended in 100 mM MES (pH 6.1), and sonicated. The cell debris was pelleted, and the soluble cell extract was fractionated on a 20% native polyacrylamide gel. The gel was then fixed, treated with fluorography solution, dried, and exposed to film. The autofluorogram is shown. WT, wild type.

Stabilities of the mutant ACPs.

Although the β-alanine labeling studies suggested that the mutant proteins were stable at 42°C, the large coenzyme A pools of E. coli (20) preclude a more definitive pulse-chase analysis. We therefore expressed the wild-type or mutant ACP proteins from the tightly regulated araBAD promoter at 30°C in strain CY321 and then blocked further expression by repressing the promoter (13). Half of each culture remained at 30°C, whereas the other portion was shifted to 44°C. The ACP levels were then followed over 4 h by Western blotting (Fig. 5). The chromosomal acpP allele of strain CY321 encodes the fully functional V43I mutant ACP (23), which migrates faster than the wild-type ACP and the newly isolated mutant ACPs. This difference in mobility allowed us to unambiguously determine the stability of these ACPs. As expected (19, 33), the wild-type ACP was stable for the duration of the experiment. All of the mutant proteins are stable at the permissive temperature and for at least two h at 42°C, except the A68T/N73D AcpP, which showed a fourfold decline in 120 min and was completely degraded within 4 h (Fig. 5). Since inhibition of fatty acid biosynthesis began before significant turnover of the A68T/N73D AcpP was observed, this mutant protein may have been unable to function in lipid synthesis prior to its degradation (Fig. 3C).

FIG. 5.

Stabilities of the mutant ACPs. Western blot analysis of the in vivo turnover of wild-type (WT) E. coli AcpP and the three temperature-sensitive AcpP species. Derivatives of strain CY321 cells harboring plasmids expressing wild-type, D38V, E49G, A68T, or A68T N73D AcpP from the E. coli araBAD promoter were grown in RB-ticarcillin-clavulanate medium to an OD₆₀₀ of ∼0.4. Expression of the ACPs was induced for 1 h with arabinose (0.2%). The cells were washed once with RB-ticarcillin-clavulanate-glucose (0.2%) medium and then suspended in this medium. A sample of culture was then removed. The cell culture was then divided, and half the culture was incubated at 30°C and the other half was incubated at 42°C. Additional samples were removed from both cultures after 30 min, 60 min, 120 min, and 240 min of incubation. The soluble fraction of the cell extracts was then fractionated on a 20% native polyacrylamide gel, and AcpP species were detected by Western blotting as described in Materials and Methods. Note that the apparently decreased intensities of the E49G protein bands are due to a shorter exposure of the film in that experiment. The initial band intensities were comparable to those of the wild-type strain in that experiment. Strain CY321 encodes a functional mutant ACP that has an unusually fast migration on native polyacrylamide gels (23) and thus allows resolution of the chromosomally and plasmid-encoded ACPs.

DISCUSSION

We obtained three conditional acpP mutants by utilizing a combination of error-prone PCR, overlap extension PCR, and λ Red-mediated recombination. Three different amino acid substitutions (D38V, E49G, and A68T) each resulted in a temperature-sensitive E. coli strain. All three mutations are located in the helical regions of ACP, which lends credence to the hypothesis that a few amino acid substitutions within the loop regions would disrupt ACP function. Therefore, the prior failures to isolate acpP mutants previously seem likely to be due to the extremely small size of the target for mutagenesis (perhaps as little as 100 bp) and perhaps also to a limited range of substitutions that result in proteins that retain sufficient function under permissive conditions. It may seem surprising that the mutant proteins accept the prosthetic group at the nonpermissive temperature. However, neither of the helix II mutations that we isolated is involved in contacts between ACP and AcpS, the enzyme responsible for modification. As shown by the crystal structure of the ACP-AcpS complex (29), the two proteins interact strongly by two salt bridges plus several hydrogen bonds plus two “knob and socket” hydrophobic interactions, almost all of which involve helix II. This complex seems quite strong as judged from the low crystallographic temperature factors (B-factor) of the ACP-AcpS interface (29). It therefore seems possible that the strength of the interactions with AcpS might overcome any distortion of the mutant ACPs and thereby allow their modification. Indeed, unlike an enzyme which has a discrete intrinsic activity, demonstration of increased thermolability of a mutant ACP protein is problematical. ACP function can be assayed only indirectly by its role as an enzyme cofactor. Therefore, a test enzyme as well as experimental conditions that do not stabilize the mutant protein must be chosen (e.g., the wild-type ACP is greatly stabilized by mono- and divalent cations as well as by acylation [3, 34, 37, 41, 42]). An increase in thermolability may not be observed because the interacting enzyme could stabilize the mutant ACP and thereby permit function. Indeed, by site-directed mutagenesis we have isolated mutant ACPs that result in the accumulation of acyl-ACP intermediates in vivo that are not seen in wild-type cells (N. R. De Lay and J. E. Cronan, Jr., unpublished data). From their electrophoretic migration rates, the acyl chains of these intermediates are the products of several rounds of the fatty acid biosynthesis cycle, indicating that at a discrete stage during the synthesis of the acyl chain, an enzyme failed to interact properly with the mutant ACP, although other enzymes must have tolerated the mutation in order to account for the chain length of the product. For these reasons, we believe that the biochemical-phenotype ACP mutants are best studied in vivo.

The effects of the three ACP mutations can readily be rationalized by reference to the known crystal structure of butyryl-ACP (39). The D38V mutation replaced a residue having a negatively charged side chain with a hydrophobic residue. This new side chain seems likely to undergo hydrophobic interactions with the L15 isobutyl moiety of helix I (Fig. 6A and B) and thereby facilitate the interaction of helices I and II of the mutant protein, thereby giving a relatively more compact structure. Indeed, although expected to migrate more slowly than the wild-type protein due to loss of a negative charge, this mutant AcpP migrated slightly more rapidly than the wild-type protein on native polyacrylamide gels, consistent with it having a more compact structure. The D38V AcpP was normally modified with 4′-PP at the nonpermissive temperature and is stable at both 30°C and 42°C. Therefore, this mutant ACP seems defective in its ability to effectively transfer fatty acid intermediates to one or more lipid synthetic enzymes at the nonpermissive temperature. The lack of complementation ability is not simply due to the loss of the negative charge at this position, since substitution of D38 with polar uncharged residues fails to result in a growth defect (De Lay and Cronan, unpublished data). The E49G substitution eliminates a salt bridge that normally occurs between E49 of helix II and R6 of helix I (Fig. 6C and 6D). The introduction of glycine at residue 49 might also increase the flexibility of the loop between helix II and helix III, which normally maintains the spatial orientations of helixes II, III, and IV via the side chain of I54 (28). The decreased mobility of the E49G protein on native polyacrylamide gels (Fig. 4) may reflect these effects in addition to the elimination of a negative charge. A previous study demonstrated that an I54A substitution in the loop between helix I and helix II caused an increase in the hydrodynamic radius of the recombinant Vibrio harveyi AcpP (8). The A68T amino acid change that contributed to the phenotype of the G205A A220G double mutant introduced a larger polar residue into the hydrophobic core of AcpP (Fig. 6E and F). The results of our modeling suggest that the hydroxyl group of the introduced threonine side chain will sterically clash with the backbone oxygen of V65 and the threonine methyl group will clash with the sec-butyl group of I62. Accommodation of these steric clashes may cause the increased hydrodynamic radius of the A68T AcpP indicated by the decreased electrophoretic mobility of the mutant protein on native polyacrylamide gels (Fig. 4). Interestingly, the introduction of the N73D amino acid substitution into A68T AcpP decreased the stability of the protein (Fig. 5) as well as the ability of this AcpP to complement the D38V mutant, although expression of the AcpP harboring only the N73D mutation resulted in complementation of the D38V mutant (Fig. 2B).

FIG. 6.

Modeling of the amino acid substitutions present in the AcpP(Ts) proteins onto the E. coli butyryl-ACP crystal structure (39). Segments of the crystal structures of wild-type AcpP (A, C, and E) are compared to the corresponding segments containing the following amino acid substitutions present in the AcpP(Ts) proteins: D38V (B), E49G (D), and A68T (F). (A-D) The hydrophobic residues are shown in brown. (E and F) The measurements are in angstroms.

These interpretations aside, the isolation of three conditional acpP mutants and the dependence of strain NRD62 on acpP expression demonstrate that acpP is an essential gene. Therefore, AcpP must be the only carrier protein in E. coli capable of shuttling intermediates among lipid biosynthetic enzymes. It should be noted that prior workers have concluded that ACP is an essential gene in E. coli (6, 7, 12) based on inhibition of growth by peptide nucleic acids targeted to the translation initiation region of acpP mRNA. However, these results led to only indirect inferences because only very small quantities of the inhibitory agents were available, which precluded any measurements other than growth.

The approach that we used to isolate acpP(Ts) mutants should be generally applicable to the isolation of temperature-sensitive alleles of other essential E. coli genes. In our hands the method was efficient and specific. Three temperature-sensitive mutants were isolated by screening only 144 colonies, and all carried mutations within acpP. We believe that conditional lethal mutations are required for valid studies of essential genes. Others have used titration or depletion approaches (e.g., a wild-type gene under the control of a tunable promoter or resident on a plasmid with conditional replication) to study essential genes (see reference 16 for a recent example). This approach, although valid for demonstrating the essential nature (or its lack) of a given gene, is inappropriate for physiological studies because the titrated and control cultures are not physiologically comparable. This is because the titrated cultures undergo a long period of progressively slowing growth before they reach the threshold where growth stops. Since the expression of many genes is regulated by growth rate (15, 30), the progressive slowing of growth will result in many secondary effects that are not reflected in (or adequately controlled by) the normally growing control culture.

The chief impediment to implementation of our approach is the lack of a neighboring nonessential gene within which to insert the selected marker (fabF::cat in the present case). However, insertion into a neighboring essential gene could be used if a second copy of the gene is provided elsewhere, on a low-copy-number plasmid or integrated into a phage attachment site. Insertion of the PCR product into this second copy should readily be avoided if the second copy is the homologous gene from Salmonella enterica (56, 57) or that from another organism. The lack of homology at the DNA level effectively blocks recombination between E. coli and S. enterica by the host recA-mediated recombination pathway (36), and in vitro studies (21) suggest that the λ recombination system should be similarly blocked, although the foreign gene must be carefully chosen. The selectable marker can later be removed by λ Red-mediated recombination, which restores gene function and allows the second copy to be eliminated