Extragenic Suppressors of Growth Defects in msbB Salmonella

Abstract

Lipid A, a potent endotoxin which can cause septic shock, anchors lipopolysaccharide (LPS) into the outer leaflet of the outer membrane of gram-negative bacteria. MsbB acylates (KDO)₂-(lauroyl)-lipid IV-A with myristate during lipid A biosynthesis. Reports of knockouts of the msbB gene describe effects on virulence but describe no evidence of growth defects in Escherichia coli K-12 or Salmonella. Our data confirm the general lack of growth defects in msbB E. coli K-12. In contrast, msbB Salmonella enterica serovar Typhimurium exhibits marked sensitivity to galactose-MacConkey and 6 mM EGTA media. At 37°C in Luria-Bertani (LB) broth, msbB Salmonella cells elongate, form bulges, and grow slowly. msbB Salmonella grow well on LB-no salt (LB-0) agar; however, under specific shaking conditions in LB-0 broth, many msbB Salmonella cells lyse during exponential growth and a fraction of the cells form filaments. msbB Salmonella grow with a near-wild-type growth rate in MSB (LB-0 containing Mg²⁺ and Ca²⁺) broth (23 to 42°C). Extragenic compensatory mutations, which partially suppress the growth defects, spontaneously occur at high frequency, and mutants can be isolated on media selective for faster growing derivatives. One of the suppressor mutations maps at 19.8 centisomes and is a recessive IS10 insertional mutation in somA, a gene of unknown function which corresponds to ybjX in E. coli. In addition, random Tn10 mutagenesis carried out in an unsuppressed msbB strain produced a set of Tn10 inserts, not in msbB or somA, that correlate with different suppressor phenotypes. Thus, insertional mutations, in somA and other genes, can suppress the msbB phenotype.

Lipopolysaccharide (LPS) forms the outer leaflet of the outer membrane in gram-negative bacteria. Although the outer membrane is more permeable to small hydrophilic molecules (because of the presence of porin channels) than the inner membrane, an intact LPS can protect gram-negative bacteria from bile salts, hydrophobic antibiotics, and complement (12) and is associated with microbial virulence (27). In short, the LPS layer is a complex structure which is crucial for survival, and its properties determine the permeability of the outer membrane to a wide variety of substances.

LPS consists of three major components: lipid A, core polysaccharides, and O-linked polysaccharides. Lipid A is an endotoxin, and its fatty acids (lauric, myristic, and sometimes palmitic acid) anchor LPS into the outer membrane. Under non-cold-shock conditions, the tightly regulated addition of fatty acids to the lipid A precursor is catalyzed by the enzymes HtrB (lauric acid [3]), MsbB (myristic acid [4]), and PagP (palmitic acid [9]). htrB Escherichia coli and Salmonella are nonpermissive for growth on rich agar at or above 37°C (14, 30). However, this growth defect can be suppressed in E. coli with the msbB gene on a high-copy-number plasmid. msbB, also known as mlt (7), waaN (17), and lpxM (1), is one of two multicopy suppressors of htrB E. coli isolated by Karow and Georgopoulos (15).

MsbB can enzymatically add myristic (14:0; fast reaction) or lauric (12:0; slow reaction) acid to different positions on the lipid A precursor (demonstrated in vitro), whereas HtrB has been shown to add lauric acid only to the lipid A precursor (4). MsbB's addition of lauric acid (slow reaction) to the lipid A precursor at the same position normally acylated by HtrB may explain msbB's high-copy-number suppression of the htrB temperature-sensitive growth defect.

Several groups have studied msbB mutants in E. coli K-12 and Salmonella choleraesuis (also known as enterica) serovar Typhimurium. Groups which have studied the growth of msbB knockouts have mentioned that there is no growth defect in msbB E. coli K-12 (15, 28, 32). Khan et al. (17) concluded that msbB Salmonella serovar Typhimurium has a wild-type growth rate in BALB/c mice. In addition, Vaara and Numinen (32) noted that there is no defect in outer membrane permeability barrier function in msbB E. coli K-12, and Somerville et al. (28) found no difference between minimum inhibitory concentrations for a spectrum of antibiotics in wild-type and msbB E. coli K-12. The only reported phenotypes attributed to msbB E. coli K-12 strains were increased deoxycholate resistance (15) and a nonpyrogenic LPS (28). However, when Somerville et al. (29) knocked out the msbB gene in a clinical isolate of E. coli strain H16, they found that msbB H16 formed filaments at 37°C and had a reduction in the level of the K1 capsule, an increase in complement C3 deposition, and increases in opsonic and nonopsonic phagocytosis. Thus, filamentation in msbB E. coli H16 was the first report of a growth defect in an msbB strain.

The reported apparent lack of growth defects, in msbB E. coli K-12 or Salmonella, was quite surprising because the other reported lipid A mutations were either lethal or exhibited conditional (temperature-sensitive) phenotypes, such as that seen with htrB (14, 30). The published data seem to suggest that the myristic acid moiety, added to lipid A by the MsbB enzyme, does not play a significant role in outer membrane barrier function, since msbB E. coli K-12 mutants were reported to have no growth phenotype other than increased deoxycholate resistance. As we report below, newly purified msbB Salmonella cultures actually exhibit significant specific growth abnormalities in vitro, whereas msbB E. coli K-12 (and E. coli B) does not.

In the course of tests of Salmonella as an anticancer agent (26), an msbB knockout mutational block in lipid A biosynthesis was introduced in order to decrease the stimulation of the septic shock response and thereby increase the safety for use of Salmonella in humans (21). Spontaneous faster growing derivatives of msbB recombinants were described and used in the construction of some of these strains, which in spite of their good growth characteristics are still as nonpyrogenic as the parental unsuppressed msbB strains (2, 21). We find that the use of Luria-Bertani (LB) broth and no salt (LB-0) or MSB (LB-0 supplemented with Mg²⁺ and Ca²⁺) agar and MSB broth support good growth of unsuppressed msbB strains. Since there is little enrichment for derivatives in these media, the primary de novo phenotype of msbB can be studied. We find that msbB mutations, especially in Salmonella, do in fact confer marked changes in physiology. As we report here, extragenic suppressor mutations occur at a high frequency and confer a more normal growth phenotype in the msbB background.

MATERIALS AND METHODS

Bacterial strains, phage, and media.

The bacterial strains used in this study are listed in Table 1. The Salmonella msbB insertion-deletion for tetracycline resistance was constructed as described by Low et al. (21), and the E. coli msbB insertion-deletion was provided by Costa Georgopoulos and is described by Karow and Georgopoulos (15). P22 mutant HT105/1int201 (obtained from the Salmonella Genetic Stock Center, Calgary, Canada) was used for Salmonella transductions, and P1vir (gift of J. Tomizawa) was used for E. coli transductions. Salmonella serovar Typhimurium and E. coli strains were grown on LB-0 or MSB agar or in MSB broth. MSB medium consists of LB medium (22) with no NaCl and supplemented with 2 mM MgSO₄ and 2 mM CaCl₂. LB-0 is LB medium with no NaCl. MSB broth and agar were used for the growth of strains under nonselective conditions. LB-0 agar was used when using selective antibiotics in transductions and transformations; Mg²⁺ and Ca²⁺ were found to increase phage contamination in transductions (5) and to decrease the effectiveness of certain antibiotics, such as ampicillin and tetracycline. Plates were solidified with 1.5% agar. LB-0 agar or MSB broth were supplemented as needed with ampicillin or carbenicillin (20 or 50 μg/ml), tetracycline (3, 5, or 20 μg/ml), chloramphenicol (15 μg/ml in broth; 25 μg/ml in agar), ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA, free acid) (Sigma, St. Louis, Mo.) (6 mM or 6.5 mM), or deoxycholate (80,000 μg/ml). A 350 mM stock of EGTA, pH 8.0 (adjusted with NaOH), was dissolved and then autoclaved. Antibiotics and sodium deoxycholate were added to LB-0 agar after cooling to 45°C. MacConkey agar base (Difco) was used to prepare galactose-MacConkey agar.

TABLE 1.

Bacterial strains

Strain	Parental strain	Genotype or phenotype	Derivation or source
S. enterica serovar Typhimurium
ATCC 14028	14028	Wild type	ATCC
YS8211	14028	msbB1::Ωtet	Low et al. (21)
YS1	14028	msbB1::Ωtet	P22 · YS8211 × 14028→Tet5r, where YS8211 = donor and 14028 = recipient in P22 transduction
YS1456	14028	ΔmsbB2 ΔpurI3252 somA1	Spontaneous EGTAr derivative of msbB1::Ωtet ΔpurI3252 derivative of 14028; replacement of msbB1::Ωtet by ΔmsbB2 by homologous recombination (6)
YS871	14028	ΔmsbB2 ΔpurI3252 somA+ zbj10:Tn10	P22 · 14028 Tn10 pool × YS1456→Tet20r (EGTAs)
YS872	14028	ΔmsbB2 ΔpurI3252 somA1 zbj10:Tn10	P22 · YS871 × YS1456→Tet20r (EGTAr)
YS873	14028	msbB1::Ωtet somA1 zbj10:Tn10	P22 · YS871 × YS1→Tet20r (EGTAr)
YS1170	14028	msbB1::Ωtet; unknown suppressor(s)	From YS8211, spontaneous selection on EGTA plates
TT16812	LT2	recD541::Tn10dCm	Salmonella Genetic Stock Center, Calgary, Canada
YS2	LT2	msbB1::Ωtet recD541::Tn10dCm	P22 · YS8211 × TT16812→Tet5r
SL1344	SL1344	his Strr; mouse virulent	Sunshine et al. (30)
YS3	SL1344	msbB1::Ωtet his Strr; mouse virulent	P22 · YS8211 × SL1344→Tet5r
E. coli K-12
MLK1067	W3110	F− λ−rph-1 IN(rrnD-rrnE) msbB1::Ωcam	Karow and Georgopoulos (15)
MG1655	MG1655	F− λ−rph-1	Guyer et al. (10)
KL423	MG1655	F− λ−rph-1 msbB1::Ωcam	P1vir · MLK1067 × MG1655→Camr

Plasmids.

A multiple cloning site containing NotI and SfiI sites on each side of a BamHI site was cloned into the EcoRI and HindIII sites of the high-copy-number vector pSP72 (Promega) and the low-copy-number vector pHSG576 (31) (∼8 copies per cell, as reported in reference 19) to facilitate shuttling inserts between the two vectors. These new vectors are named pSM1 and pSM2, respectively. pSM2 containing the cloned wild-type msbB gene is designated pSM21, and pSM2 with the cloned wild-type somA gene is designated pSM22 (Table 2).

TABLE 2.

Plasmids

Plasmid	Relevant characteristic(s)	Source
pNK2883	Ampr, inducible mini-Tn10 with isopropyl-β-d-thiogalactopyranoside promoter	Kleckner et al. (18)
pSP72	High copy number, Ampr	Promega
pHSG576	Low copy number, Cmr, pSC101 derivative	Takeshita et al. (31)
pSM1	pSP72 with a modified multiple cloning site replacing the pSP72 sites between the EcoRI and HindIII sites with the following restriction sites: NotI, SfiI, BamHI, NotI, and SfiI	This study
pSM2	pHSG576 with a modified multiple cloning site replacing the pHSG576 sites between the EcoRI and HindIII sites with the following restriction sites: NotI, SfiI, BamHI, NotI, and SfiI	This study
pSM3	pSP72 with msbB+	This study
pSM21	pSM2 with msbB+	This study
pSM22	pSM2 with somA+	This study

Growth analysis.

Phenotypes of strains were confirmed by replica plating. Master plates were made on either MSB or LB-0 agar. Replica plating was performed using the double velvet technique (20). To test for LB agar sensitivity, triple velvet replica plating, which uses an unincubated double velvet plate to replica plate onto various media, was used. Plates were incubated for 10 h at 28, 30, 37, or 42°C or for 1 to 2 days at 23°C. To generate growth curves, 10-ml MSB broth tubes were inoculated from patches from new clones, with verified phenotypes, and grown on a slant without movement overnight at 37°C. Tubes (2.5-cm diameter) with 10 ml of broth were then inoculated with cells to achieve an optical density at 600 nm (OD₆₀₀) of 0.05. Cells were held on ice until all inoculations were completed. Then the cultures were placed in a 37°C or room temperature (21 to 23°C) water bath on a 30° angle with 100 rpm of translational movement. OD₆₀₀ was measured every 30 min for 420 min.

Restoring msbB⁺ genotype.

In order to confirm that the observed MsbB phenotypes result simply from knocking out MsbB function, a fragment containing wild-type msbB (21) was digested with EcoRV and cloned into pSP72 to test for complementation in YS1 on EGTA and galactose-MacConkey plates. pSP72 carrying wild-type msbB was named pSM3. After observing complementation in YS1 and confirming the insert by sequencing, wild-type msbB was again EcoRV digested and blunt-end ligated into HindIII-digested pSM2, thus producing plasmid pSM21.

Microscopic observation.

Strains ATCC 14028, YS1 (msbB1::Ωtet), and YS1456 (msbB2 purI3252 somA1) were grown, as described above for growth curves, to an OD₆₀₀ of 0.40. Then the cells were stained with nigrosin (11) and observed with an Olympus AX70 microscope.

Mutation frequency determination.

A frozen stock of YS1 was streaked on MSB medium and incubated overnight at 37°C to isolate individual clones. Ten milliliters of MSB broth, in 2.5-cm diameter tubes, were inoculated with independent YS1 colonies. They were grown in tubes at a 30° angle in a 37°C water bath with 100 rpm of translational movement until an OD₆₀₀ of 0.10 was achieved. The tubes were then placed on ice and diluted in ice-cold MSB broth. Dilutions (2 × 10⁻⁶) were plated onto MSB agar to calculate the number of CFU per milliliter. Dilutions (10⁻², 10⁻³, and 10⁻⁴) were plated on 6.5 mM EGTA, galactose-MacConkey, and LB plates and incubated overnight at 37°C. This concentration of EGTA (6.5 mM) was used to minimize the chances of getting growth of nonmutated survivors (based on a series of CFU tests, data not shown). Approximately 20 clones arising on each type of plate were used to make a master plate on MSB medium to determine what percentage of the clones were mutants.

Preparation of electroporation-competent cells.

The technique of O'Callaghan and Charbit (25) was used for preparation of electroporation-competent cells with the following modifications. Overnight cultures in MSB broth were prepared as described above. The next morning, 2 ml of the overnight culture was used to inoculate 100 ml of MSB broth, which was grown in a 37°C water bath with 100 rpm of translational movement until the cells reached an OD₆₀₀ of 0.6. Cells were rinsed with ice-cold 1% glycerol instead of distilled water, because 1% glycerol was found to increase the survival of msbB Salmonella and to help maintain the unsuppressed phenotype (data not shown).

Transduction and transformation.

Salmonella P22 transductions were performed by the method of Davis et al. (5), and E. coli P1 transductions were performed by the method of Miller (22), except that LB-0 plates supplemented with the appropriate antibiotic were used. EGTA was not added to the antibiotic plates for transductions. A Bio-Rad Gene Pulser was used for electroporation with the following settings: 2.5 kV, 1,000 Ω, and 25 μF for transformation of YS1, and 2.5 kV, 400 Ω, and 25 μF for YS1456 and 14028.

Tn10 mutagenesis.

A transposon pool of ATCC 14048 was made using pNK2883 by following the technique of Kleckner et al. (18), except that MSB broth and LB-0-Tet₂₀ agar (contains 20 μg of tetracycline per ml) were used instead of LB broth and LB-Tet₂₀ agar (contains 20 μg of tetracycline per ml). Over 65,000 tetracycline resistant (Tet^r) clones of ATCC 14028 were pooled, and a P22 lysate was made. The pool was screened for auxotrophy for different biosynthetic pathways by replica plating onto minimal media and media containing various pools of amino acids and bases (5).

Linkage of the YS1456 suppressor mutation to a Tn10.

A P22 lysate was made from the ATCC 14028 Tn10 library and transduced into strain YS1456. Tet^r transductants were screened for EGTA sensitivity by replica plating. Upon isolation of a Tn10 which showed linkage to a suppressor gene, the formula of Wu (34) was used to estimate the distance between the Tn10 and the suppressor gene. Additional P22 lysates were made from both suppressed and nonsuppressed recombinants; thus, the Tn10 was linked to the wild-type allele of the YS1456 suppressor gene (strain YS871) and to the YS1456 suppressor mutation (strain YS872).

Cloning of somA.

YS871 (with a Tn10 approximately 3.0 kb away from the wild-type allele of the YS1456 suppressor gene) genomic DNA was cloned into pSM1, and YS872 (with a Tn10 approximately 3.0 kb away from the YS1456 suppressor mutation) genomic DNA was cloned into pSP72 to test for complementation. Genomic DNA was partially digested with Sau3AI, size selected, and cloned into the BamHI site of pSM1 or pSP72. The libraries were transformed into maximum efficiency DH5α (Gibco-BRL) and plated onto LB-0-Amp₂₀-Tet₃ agar in order to select for inserts with Tn10s. Tn10 results in 20 μg/ml Tet^r (resistance to 20 μg of tetracycline per ml) when incorporated into the chromosome but 2 to 5 μg/ml Tet^r when present on a high-copy-number plasmid (18). PCR was used to test transformants arising on LB-0-Amp₂₀-Tet₃ agar to confirm the presence of the Tn10 before sequencing and transforming the inserts into either YS1456 (YS871 library) or YS1 (YS872 library). The YS871 library was tested for complementation in strain YS1456 to screen for an EGTA-sensitive phenotype, and the YS872 library was transformed into YS1 to screen for the suppressor phenotype. The location of the suppressor gene was mapped by sequencing and aligning inserts which did or did not complement the phenotype. Inserts with the suppressor or wild-type allele of the YS1456 suppressor gene were sequenced at Yale's Keck Facility. The putative suppressor gene, somA, was then cloned into a low-copy-number vector, pSM2, via PCR and retested for complementation. Genomic DNA clones were sequenced on both strands and submitted to GenBank under S. enterica serovar Typhimurium somA accession number AF360548.

PCR.

PCR was performed using whole bacteria. Clones were tested for the presence of Tn10 using primers specific for the regulatory region. The Tn10 primers 5′-GGATCCTTAAGACCCACTTTCACATTTAAGT-3′ and 5′-GGTTCCATGGTTCACTTTTCTCTATCAC-3′ yield a 721-bp product. somA was amplified with primers containing NotI restriction sites: 5′-GGGGGCGGCCGCCGGATTTGGCGATTGAAGTC-3′ and 5′-GGGGGCGGCCGCGATAAGTTGGCAGCGGGG-3′. These primers generate a 1,329-bp product when amplifying the wild-type allele and a 2,298-bp product from the YS1456 suppressor allele. The Tn10 primers were kindly provided by Caroline Clairmont, Vion Pharmaceuticals. All primers were made by the Yale University Keck Facility. PCRs were performed using Ready-To-Go PCR beads (Amersham Pharmacia Biotech Inc.).

DNA sequencing.

DNA sequencing was performed at the Yale University Keck Facility using fluorescent dye terminated thermocycle sequencing. To sequence DNA flanking Tn10s on cloned inserts, the following primers were used: TnL1, 5′-CCCACCTAAATGGAACGGCGTT-3′, and TnR2, 5′-GGCACCTTTGGTCACCAACGCTT-3′. These primers were provided by Stanley Lin of Vion Pharmaceuticals. SP6 and T7 primers were used to sequence from the ends of inserts in pSM1, which is a derivative of pSP72. For sequencing inserts in the low-copy-number vector, pSM2 (derivative of pHSG576), the following primers, obtained from Joann Sweasy, were used: M13, 5′-GCGGATAACAATTTCATATAGG-3′, and U17, 5′-GTAAAACGACGGCCAGT-3′.

RESULTS AND DISCUSSION

Growth phenotypes of msbB strains.

During the engineering of msbB Salmonella a variety of colony sizes were observed on LB plates. As shown in Fig. 1, when plating dilutions onto LB agar from an msbB broth culture with an OD₆₀₀ of 0.1 (using growth conditions as described in the “Mutation frequency determination” section of Materials and Methods), a few colonies arise after 15 h, and by 27 h a variety of colony sizes are apparent. Subsequent experiments revealed that the smaller colonies (unsuppressed msbB clones) were sensitive to 6 mM EGTA and galactose-MacConkey media, whereas the larger colonies (suppressed msbB clones) grew well on these media.

FIG. 1.

Dilution (10⁻⁴) of a YS1 culture (OD₆₀₀ of 0.10) plated onto LB agar after a 15-h (A) or 27-h (B) incubation at 37°C. Arrows point to colonies with spontaneous suppressor mutations arising after 15 h.

Sensitivity of YS1 (msbB1) to 6 mM EGTA (Fig. 2B), galactose-MacConkey (Fig. 2C and D), and LB medium (Fig. 2F) is shown in a replica plate series. Growth of YS1 is inhibited or retarded on these media after 10 h of incubation at 28, 30, 37, or 42°C. Similarly, strains YS2 (LT2 msbB1) and YS3 (SL1344 msbB1) are also sensitive to 6 mM EGTA, galactose-MacConkey, and LB medium (data not shown). Thus, these growth defects, which are observed in three different strain backgrounds, are not strain-specific phenomena. No growth inhibition is observed on 6 mM EGTA or LB plates at 23°C, but sensitivity to galactose-MacConkey is maintained at 23°C (data not shown). As shown in Fig. 2A, YS1 grows well on LB-0 agar at 30°C (and also at 23, 28, 37, and 42°C; data not shown). The phenotype of msbB Salmonella strains is in sharp contrast to the reported phenotype of mutants of htrB, which encodes another LPS biosynthetic enzyme, since htrB Salmonella has a temperature-sensitive growth defect on rich agar at or above 37°C (30).

FIG. 2.

Replica plate series. (A) LB-0 medium, 30°C; (B) 6 mM EGTA medium, 30°C; (C) galactose-MacConkey (Gal. Mac.) medium, 30°C; (D) galactose-MacConkey medium, 28°C; (E) LB-0 medium, triple velvet (t.v.), 30°C; (F) LB medium, triple velvet, 30°C. The master plate was made on MSB agar. w.t., wild type.

In contrast to the above phenotypes of simple msbB mutants, spontaneously derived suppressor strains YS1456 and YS1170 (both msbB with suppressor mutations) have intermediate resistance phenotypes, showing that suppressor mutations can partially compensate for the msbB growth defect. Strain YS1456 grows stronger on 6 mM EGTA than YS1170 (Fig. 2B), and YS1170 grows stronger on galactose-MacConkey than YS1456 (Fig. 2C and D).

In order to test if the phenotypes observed result solely from the disruption of the msbB gene, we cloned msbB⁺ into pSM2 (generating the vector pSM21) and tested for complementation. Wild-type msbB on a low-copy-number vector complements the 6 mM EGTA and galactose-MacConkey growth defects (data not shown), demonstrating that the phenotypes observed are due to disruption of the msbB gene and not a polarity effect.

In contrast to Salmonella, KL423 (E. coli msbB) does not show sensitivity to 6 mM EGTA (Fig. 2B) but does have a particular temperature-dependent phenotype for galactose-MacConkey sensitivity: its growth is inhibited on galactose-MacConkey medium at 28 (Fig. 2D) but not at 30 (Fig. 2C), 23, 37, or 42°C (data not shown). As reported by Karow and Georgopoulos (15), we find that KL423 has increased resistance to deoxycholate at 30°C; this is also true for Salmonella suppressor strain YS1456. No increased resistance to deoxycholate was observed in unsuppressed msbB Salmonella strain YS1 (data not shown). Furthermore, KL423 (E. coli msbB) was found to have no apparent growth defect in LB, LB-0, or MSB broth at 37°C (data not shown), confirming the results reported by Karow and Georgopoulos (15). One possible interpretation of the species difference is that wild-type E. coli (at least strains K-12 and B [strain WA837; data not shown]) carry alleles that compensate for msbB in a way similar to that of the Salmonella suppressor mutations, thereby preventing the specific growth defects which are seen in simple msbB Salmonella mutants.

A number of specific growth defects of msbB Salmonella strains were also observed in liquid growth media. LB-0 medium compensates for the LB slow growth defect in agar (Fig. 2A and E) but not always in broth. At 37°C, the growth of YS1 is inhibited in LB broth (Fig. 3A) and the strain lyses in LB-0 broth under our specific conditions (see Materials and Methods and Fig. 3B). If the starter culture is from a plate, the lysis phenomenon is not observed (data not shown). We also tested other unrelated Salmonella serovar Typhimurium strains YS2 (LT2 msbB) and YS3 (SL1344 msbB) and found that they also lyse in LB-0 broth and have poor growth in LB broth (data not shown). The somA1 suppressor mutation in strain YS1456 (see below) compensates for the msbB lysis phenomenon in LB-0 broth (Fig. 3B) and substantially compensates for the LB sensitivity (Fig. 3A). In order to test for the possibility that the purI mutation in YS1456 played a role in these growth properties, a purI⁺ YS1456 strain (YS873) was constructed and was found to have growth curves identical to those observed with purI3252 YS1456 (data not shown). YS1 log phase lysis does not occur at 23°C, but sensitivity to LB broth is maintained at 23°C (data not shown).

FIG. 3.

Growth curves of 14028, YS1, and YS1456 at 37°C in LB broth (A), LB-0 broth (B), or MSB broth (C).

In an attempt to compensate for the specific in vitro msbB growth defects in LB and LB-0 broth (Fig. 3A and B), LB-0 medium was supplemented with 2 mM MgSO₄ and 2 mM CaCl₂ (called MSB medium). In MSB broth (Fig. 3C), YS1 (msbB) has a near-wild-type growth rate when grown in a shaker with 100 rpm of translational movement. However, even in MSB broth YS1 cultures accumulate some debris when grown with rapid rotational movement at 37°C (data not shown). Thus, the extent of shear forces and oxygenation appears to be important for maintaining the de novo msbB phenotype.

The phenotype of msbB cells is presumably a result of the absence of the myristoyl group on lipid A. Lateral interactions between neighboring lipid A molecules can presumably be weakened by having fewer acyl groups anchoring lipid A into the outer membrane or by a shortage in divalent cations. An excess of Mg²⁺ or Ca²⁺ might stabilize the LPS moieties in msbB strains, which have one less fatty acid than the wild type, since it is possible that these divalent cations could increase the number divalent-cation cross-linked phosphate groups that decorate lipid A and the LPS core. This inverse relationship between levels of cations and acyl groups is consistent with the reported increase of lipid A palmitate levels (by PagP) in wild-type Salmonella under low Mg²⁺ conditions (9).

Previous work has shown that Mg²⁺ and Ca²⁺ have different effects on the stability of the outer membrane in Salmonella and E. coli (24). We have found that either CaCl₂ or MgCl₂ can compensate for msbB growth defects. However, much higher concentrations of MgCl₂ are needed to yield compensation similar to that with CaCl₂-containing media (data not shown). Since either Mg²⁺ or Ca²⁺ can cross-link the phosphate groups of lipid A, it is possible that Ca²⁺ may be able to compensate by an additional, Ca²⁺-specific, pathway. For example, Kanipes et al. (13) recently reported that Ca²⁺-, but not Mg²⁺-, containing media resulted in increased phosphoethanolamine substitution on LPS's inner core.

Morphological abnormalities in msbB Salmonella.

To further investigate the phenotype of msbB cells in LB and LB-0 broth, we observed the morphology of cells grown in these media and also MSB broth at 37°C with 100 rpm of translational movement at an OD₆₀₀ of 0.40. Unusual morphology was observed in YS1 cultures in all three broths tested and suggests that a loss-of-function mutation in msbB can lead to problems in cell division.

At 37°C, ∼3 to 30% of various clones of YS1 cells in MSB broth form filaments (Fig. 4B). Filaments were defined as cells which were at least three times the length of neighboring cells. In addition, many filaments have cross-sections that are larger than normal. The YS1456 suppressor mutation suppresses filamentation, since only ∼1 to 3% of YS1456 cells form filaments in MSB broth compared to ∼3 to 30% in YS1 and ∼0% in wild-type ATCC 14028 (Fig. 4A). The length of wild-type cells ranged from ∼2 to 4 μM, and the average width was ∼1 μM. In contrast, the length of YS1 cells ranged from ∼1 to 36 μM, and the width ranged from ∼1 to 2 μM in MSB broth.

FIG. 4.

Morphology of 14028 and YS1 in MSB or LB broth at 37°C. Magnification, ×4,400. (A) 14028, MSB broth; (B) YS1, MSB broth; (C) 14028, LB broth; (D) YS1, LB broth, Bar, 10 μM.

Filamentation of YS1 may contribute to small (approximately twofold) decreases in CFU per OD unit observed in MSB broth at 37°C. No filaments were observed when YS1 was grown under identical conditions at 23°C (data not shown). A similar finding was observed in the H16 clinical isolate of the E. coli msbB mutant, where filamentation occurs at 37°C but not at 30°C (29). In contrast, Karow and Georgopoulos (15) did not report any filamentation in their K-12 W3110 msbB mutants, and we observed a low percentage (1 to 2%) of filament formation in our K-12 MG1655 msbB mutant strain in MSB broth (data not shown).

In LB broth at 37°C, all YS1 cells appear to be elongated and curved and formed bulges (Fig. 4D). In contrast to YS1, ∼22% of YS1456 cells form filaments, with some swelling in LB broth (data not shown). 14028 cells have normal morphology in LB broth (Fig. 4C). In LB broth, the length of wild-type cells ranged from 2 to 4 μM, and the average width was ∼1 μM. In contrast, the length of YS1 cells ranged from ∼6 to 10 μM, and the average width was ∼2 μM. In LB broth at 23°C, YS1 forms filaments similar to those observed in MSB broth at 37°C but not elongated bulging cells (data not shown). Before log phase lysis in LB-0 broth at 37°C, YS1 forms both filaments and curvy, elongated, bulging cells (data not shown).

Mutation frequency determination.

Having noticed a seemingly high frequency of faster-growing derivatives of msbB strains upon streaking on various media, we then measured the frequency of mutants exhibiting faster growth. Our results (Table 3) show that using freshly isolated clones, the average number of 6.5 mM EGTA-resistant mutants from stabilized clones grown in MSB broth is ∼4 × 10⁻⁴, and the average number of galactose-MacConkey-resistant mutants is ∼1 × 10⁻⁴. However, the observed frequency of LB-resistant mutants grown similarly in MSB broth but plated on LB agar, which allows slow growth, is ∼1.0 × 10⁻², which is approximately 25- to 100-fold higher than the frequency of the directly selected EGTA- or galactose-MacConkey-resistant mutants. After streaking the EGTA-, galactose-MacConkey-, or LB-resistant colonies on MSB agar, we found that ∼88% of the EGTA-resistant colonies, 100% of the galactose-MacConkey-resistant colonies, and ∼73% of the LB-resistant colonies maintained their phenotype after streaking, showing that the majority of resistant colonies are relatively stable mutants. Thus, the data suggest that our mutation frequency estimates reflect the number of mutants in the cultures, except for the selection on LB, which allows slow growth of unmutated clones (as shown in Fig. 1). Since YS1 cells can undergo a significant number of generations of growth on LB plates, there are many more cell divisions in which mutations could occur in the early hours of incubation, thus giving an artificially high calculated frequency of mutation from simply counting the number of colonies after 15 h. This interpretation is strengthened by finding that the vast majority of the suppressed derivatives obtained on LB are resistant to 6 mM EGTA and/or galactose-MacConkey medium, and the frequency of these mutations, when selected directly, was shown to be much lower (10⁻⁴ versus 10⁻²).

TABLE 3.

Measurements of mutation frequency in strain YS1 (msbB; LB, EGTA, and galactose-MacConkey sensitive) to LB, EGTA, or galactose-MacConkey resistance

Selective medium	Frequency of mutants for YS1 clone:			Avg mutation frequency	Avg no. of mutants per 10,000 cells
	A	B	C
EGTA (6.5 mM)	3.8 × 10−4	2.8 × 10−4	4.3 × 10−4	3.6 × 10−4	4
Galactose-MacConkey	2.0 × 10−4	1.2 × 10−4	4.5 × 10−5	1.2 × 10−4	1
LB	1.0 × 10−2	1.7 × 10−2	2.9 × 10−3	9.9 × 10−3	99

High mutation frequencies in YS1 cultures for resistance to 6.5 mM EGTA (4 × 10⁻⁴) and galactose-MacConkey (1 × 10⁻⁴) agar suggest that there may be many genetic targets for suppression of the msbB Salmonella phenotype. The EGTA and galactose-MacConkey mutation frequencies are similar to that observed for temperature-sensitive compensatory mutation in htrB E. coli, which was ∼1.0 × 10⁻⁴ (16). It is possible that many of the YS1 suppressor mutations are loss-of-function mutations, since there are many more targets for loss-of-function than gain-of-function mutations. Furthermore, the high frequency of spontaneous suppressor mutations may explain why other groups (8, 17, 33) did not mention the phenotypes discussed in this report. It is possible that these authors used cultures that had been inadvertently overgrown with faster growing derivatives, which rapidly arise on LB plates and do not exhibit strong specific growth defects in vitro.

Mapping of one of the mutations which suppresses msbB.

We used a Tn10 pool to obtain linkage to the suppressor mutation, denoted somA1 for suppressor of msbB, in YS1456. The presence of 0.7% auxotrophy involving a variety of biosynthetic pathways in our 14028 transposon pool (data not shown) suggested that Tn10s had integrated into the 14028 chromosome rather randomly. In order to link a Tn10 marker to the YS1456 suppressor mutation, we transduced DNA fragments, from this transposon library in 14028, into YS1456. Two out of 300 transductants were EGTA sensitive from a transduction bringing the wild-type P22 Tn10 library into strain YS1456. The EGTA-sensitive phenotype results from replacing the YS1456 suppressor gene with wild-type DNA linked to particular Tn10s from the library. Cotransduction frequencies of 79 and 82% were obtained, suggesting that the two Tn10s lie ∼3.2 and ∼3.0 kb, respectively, away from the YS1456 suppressor gene. The Tn10 lying ∼3.0 kb away was selected for use in gene linkage experiments. Transductants with the Tn10 linked to wild-type or mutant alleles of somA were used to make genomic libraries.

Cloning and sequencing of somA.

When plasmids with Tn10 linked to the YS1456 allele of somA were electroporated into YS1, no suppressor phenotypes were observed (data not shown). (The electroporation conditions used select for the transformation of unsuppressed msbB cells, therefore making spontaneously suppressed mutants nearly undetectable in our screens.) However, when plasmid clones carrying the Tn10 linked to wild-type alleles of the YS1456 suppressor gene were electroporated into strain YS1456, EGTA sensitivity was observed, thus showing that YS1456 has a recessive suppressor mutation. Both mutant and wild-type versions of the 19.8-centisome (Cs) region were sequenced, and the sequence data revealed that YS1456 has an IS10 insertion sequence at nucleotide 922 in an open reading frame homologous to ybjX of E. coli, which we call somA (suppressor of msbB). The IS10 insertion may have occurred at some point in the ancestry of YS1456 in which a purI::Tn10 mutation was present. The Salmonella somA gene, which maps close to 19.8 Cs, has ∼59% nucleotide homology to E. coli ybjX and at the level of protein has 56% identity. somA was then amplified by PCR and cloned into a low-copy-number vector to demonstrate complementation (restoration of EGTA and galactose-MacConkey sensitivity [data not shown]) in strain YS1456.

somA, a gene encoding a protein of unknown function, is the first example of an identified spontaneous suppressor of msbB. No growth defect was found for the somA1 mutation in an msbB⁺ background (data not shown), and no obvious transmembrane domains or signal sequences were apparent from basic sequence analyses. In a search of GenBank sequences, Salmonella somA was found to have ∼46% nucleotide homology to serovar Typhi virK, and at the level of protein there is 34% identity. Serovar Typhi virK maps at approximately 61 Cs, in the iroN upstream region, on the Salmonella chromosome (GenBank accession no. AF029845). As part of the Shigella virulence plasmid, virK was found to be a virulence factor which mediates intracellular spreading by posttranscriptionally regulating the virG gene product (23). A possible relationship between somA and virK is, at this point, unclear.

We have begun further analysis along these lines, and Tn10 mutagenesis in YS1 has produced Tn10-induced suppressor mutations which yield phenotypes and genetic targets distinct from that of the somA suppressor mutation, indicating that somA is only one example of an extragenic insertional suppressor mutation (data not shown). Experiments to infer the function of somA are in progress.