Membrane cyclopropane fatty acid content is a major factor in acid resistance of Escherichia coli

E. coli strains lacking CFA are more sensitive to acid shock

Acid tolerance is very strongly dependent on growth phase, with stationary phase cultures being much more resistant than log phase cultures (Small et al., 1994; Lee et al., 1995; Slonczewski and Foster, 1996; Bearson et al., 1997). We tested the acid tolerance of early stationary phase cultures of two wild-type E. coli strains MG1655 and ZK126. We chose these two strains because strain ZK126 is known to have the full RpoS response (Zambrano et al., 1993) and is derived from W3110 (Bohannon et al., 1991), the strain that provided the original genomic restriction map and much sequence information, whereas the MG1655 genome is that completely sequenced by Blattner et al., (1997). We found that, although both are considered wild-type strains, the sensitivity of the two strains towards extreme acid (pH 3) was very different, and the two strains had differing dependencies on the test protocol used. In the course of this work, we transduced null mutations in either cfa or rpoS (or both) into the two strains. The rpoS null mutation used is the rpoS ::Tn10 insertion of strain UM122 (Mulvey et al., 1988) whereas the cfa ::kan mutation was constructed as described in the Experimental procedures. It should be noted that the cfa insertion mutation cannot have polar effects on downstream genes because cfa is expressed as a monocistronic mRNA (Wang and Cronan, 1994) and the gene immediately downstream (which encodes a riboflavin synthase subunit) is divergently transcribed (Blattner et al., 1997). Strains carrying the cfa ::kan mutation lacked detectable CFA synthase activity and made no CFA detectable by gas chromatography (Cronan et al., 1974) or collision-induced dissociation electrospray mass spectroscopy (Sweetman et al., 1996) (data not shown) as shown for other null mutants (Grogan and Cronan, 1986).

Two general types of protocols have been used for acid shock of bacterial cultures (Small et al., 1994; Lee et al., 1995; Bearson et al., 1997; Brown et al., 1997) and we have tested both types. In protocol A, cultures grown in Luria–Bertani (LB) medium are simply adjusted to pH 3 with HCl, whereas in protocol B the cultures were grown in glucose minimal medium E. These cells were pelleted, washed and then resuspended in medium E salts at either pH 7 or pH 3. In both protocols, the cell suspensions were held at pH 3 for various time periods and samples taken, diluted in neutral medium and spread on RB plates.

We first tested early stationary phase cultures of strain ZK126 and its derivatives because this strain has the full RpoS response, which is known to play a major role in acid shock tolerance (Slonczewski and Foster, 1996). When tested by protocol A, strain ZK126 was very resistant to acid and had a survival of 0.6–0.7 upon exposure to pH 3 for 40 min at 37°C. In contrast, when protocol B was used, the survival was only about 0.01 (Fig. 1B). Because the major difference in the protocols was the medium used, we performed a hybrid experiment in which the cultures were grown as in protocol A, then acid shocked as in protocol B (Fig. 1A). Strain ZK126 again survived this treatment well, suggesting that the growth medium plays a major role in survival of this strain (Fig. 1A). However, when strain YYC1272, the cfa null mutant of strain ZK126, was examined, a different pattern was seen. When tested by protocol A, there was little or no difference in survival of YYC1272 and ZK126, whereas when tested by protocol B the cfa strain survived at a 100-fold lower rate (Fig. 1B). When tested by the hybrid protocol, there was a 20-fold difference in survival (Fig. 1A). Therefore, the presence of LB medium before and during acid shock increased the survival of the cfa strain. Strain ZK126 seems to have at least two mechanisms to survive acid conditions, one requiring the presence of LB medium both before and during acid shock and a second dependent on CFA. We attribute the LB medium effect to the known acid defence mechanisms, one of which, increased expression of basic amino acid decarboxylases (Olson, 1993; Lin et al., 1995; Park et al., 1996; Slonczewski and Foster, 1996), requires the presence of amino acids. When cultures were grown in minimal medium, these defences were inoperative and the effects of CFA content were unmasked. We then turned to early stationary phase cultures of strain MG1655 to test the generality of these observations and found the behaviour of this strain to be quite different, although the cfa derivative was always more sensitive to acid shock than the wild-type strain (Fig. 1C).

These results indicated that other physiological processes can mask the effects of CFA content on acid shock survival. Because rpoS mutants are known to be defective in acid shock survival (Small et al., 1994; Slonczewski and Foster, 1996), it seemed likely that these other processes might be regulated, at least in part (as is cfa expression), by RpoS. If so, we would expect that strains lacking RpoS as well as CFAs should be more sensitive than either single mutant strain. The results of analysis of strain ZK126 and its derivatives confirmed this prediction, the double mutant was more sensitive to acid shock than either single mutant strain when tested by any of our protocols. We attribute the lesser sensitivity of the rpoS strain relative to the double mutant to the P1 cfa promoter, which functions in the absence of RpoS (Wang and Cronan, 1994). The double mutant strain therefore completely lacked cfa expression, whereas residual cfa expression remained in the rpoS strain. Cultures grown on either LB or minimal medium both gave striking differences in survival when the mutant cultures were compared with the wild-type strain, although overall survival values were generally increased if the cultures were grown in LB (Fig. 1A and B). Strain MG1655 and its derivatives gave very similar results except that this strain and its derivatives survived acid shock much better when protocol B was used rather than protocol A (the opposite of the results obtained with strain ZK126 and its derivatives). The rpoS derivative of MG1655 (YYC1276) was almost 100-fold more acid sensitive than its wild-type parent (Fig. 1C), whereas the rpoS cfa derivative had an acid survival rate 700-fold lower than the wild parental strain when tested by protocol A (Fig. 1C). Protocol B gave the same order of acid sensitivity: wild type > cfa > rpoS > cfa rpoS (Fig. 1C).

We then tested whether the defect in survival of the cfa strains was directly due to the loss of this gene by introducing the cfa plasmid pAYW19 into these strains. Both of the transformed derivatives of strains ZK126 and MG1655 became as acid resistant as their wild-type parental strains (data not shown), indicating that the effect of the cfa null mutation was specific to the loss of the cfa gene product. An acid shock experiment was also performed on a mixed culture of strain YYC1273 (cfa::kan ) and its parental strain MG1655 as a further test. The number of surviving cells of each genotype was determined by using media with or without kanamycin. Only the cfa mutant strain could form colonies on kanamycin-supplemented medium, and thus plating on these media differentiated the wild type and cfa::kan null mutant survivors. The survival rate of both strains were the same as when they were grown and acid shocked separately (data not shown), excluding the possibility that molecules produced by one strain diffuse through the medium and influence acid shock survival of the other strain.

Incorporation of exogenous CFAs renders cells more acid resistant

The above experiments indicated that cells lacking CFA synthase were more sensitive to acid shock. It remained possible that these results were due to lack of the enzyme protein per se or of an enzyme product other than CFA (e.g. S-adenosylhomocysteine) rather than the lack of CFA. To test these possibilities, we bypassed the cfa null mutation by providing chemically synthesized CFA in the growth medium. To demand efficient incorporation of the exogenous acids, we used a strain unable to make unsaturated fatty acids (UFAs). Such strains are UFA auxotrophs and require an appropriate UFA or CFA for growth (in the absence of supplementation cultures of these, strains lyse) (Magnuson et al., 1993). The strains used carried the cfa::kan null mutation (to prevent conversion of the UFA supplement to CFA) and the temperature-sensitive fabA2 mutation (fabA encodes the enzyme that introduces the double bond of the UFAs). We grew fabA cfa derivatives of strains ZK126 and MG1655 (strains YYC1283 and YYC1284 respectively) in LB medium supplemented with either a CFA or the UFA analogue for many generations of growth at 42°C (a restrictive temperature for the fabA2 allele) to ensure that the added fatty acid was the sole CFA (or UFA) component of the membrane lipids. Early stationary phase cultures of these cells then were subjected to acid shock at pH 3 for 20 min at 42°C (Fig. 2). Strain YYC1283, the fabA cfa ZK126 derivative, had showed a 330-fold greater survival when grown with C17 CFA supplement than when the strain was grown with the corresponding UFA, which was palmitoleic acid (C16:1) (Fig. 2A). The longer chain length homologues gave a lesser effect; cells grown in LB media supplemented with the C19 CFA survived only fivefold better than those grown with the corresponding UFA (C18:1). Cultures of strain YYC1284, a fabA cfa derivative of strain MG1655, survived fivefold better when supplemented with either C17 CFA or C19 CFA than when supplemented with the corresponding UFA (Fig. 2B). These data therefore directly demonstrate that cells with membranes containing CFA resist acid shock better than the corresponding UFA-containing cells. We analysed the fatty acid compositions of the supplemented cultures by gas chromatography and found that the anticipated UFAs were present when the cultures were supplemented with either palmitoleic acid or oleic acid, whereas only the anticipated CFA was found in the CFA-supplemented cultures (data not shown).

Acid adaptation and CFA synthesis

The tolerance of early log phase cultures of enterobacteria to pH 3 treatment is greatly increased if cultures are adjusted to a moderately acid pH and allowed to grow for a period before shift to pH 3 (Foster and Hall, 1990; Slonczewski and Foster, 1996). This phenomenon, called acid adaptation or acid habituation, requires active metabolism during the incubation (Foster, 1991; 1993; Lee et al., 1994). Previous workers (Brown et al., 1997) showed that cellular CFA levels increase during acid adaptation, and they attributed this increase to the known accumulation of RpoS protein that would increase cfa expression by activating the RpoS-dependent P2 promoter (Wang and Cronan, 1994). We tested this premise by use of the cfa–lacZ transcriptional fusions constructed previously (Eichel et al., 1999). Early log phase cultures of strain ZK126 showed only low-level cfa gene expression in either a wild-type or rpoS::kan background. However, when the wild-type cultures were subjected to mild acid (pH 5) exposure for one generation, expression from the RpoS-dependent P2 promoter was increased 10-to 25-fold depending on whether the lacZ fusion construct included the interpromoter region (IPR) or not, whereas expression from the RpoS-independent P1 promoter was only slightly increased (Fig. 3A). In the absence of RpoS, no cfa expression from the P2 promoter was seen, whereas LacZ expression from the P1–lacZ and P1–P2–lacZ fusions was increased two- to threefold under adapted conditions (Fig. 3A). Surprisingly, the level of cfa expression of the P1–P2–lacZ fusion in the rpoS::kan background reached almost the same level as that of the wild-type strain (Fig. 3A). The reason for this is unclear, but suggests that a factor that stimulates expression from the P1 promoter accumulates during acid adaptation. Nevertheless, it seems that during acid adaptation of rpoS::kan strains, expression of cfa from the P1 promoter can produce enough CFA to account for protection from the extreme acid shock.

We also studied the acid survival of strain ZK126 and its three mutant derivatives using both of the acid adaptation protocols described in the Experimental procedures. When these strains were adapted at pH 5 in LB medium and subjected to pH 3 acid shock, all four strains had similar survival values (about 0.5; data not shown). It should be noted that, without acid adaptation, early log phase cultures of these strains survive acid shock very poorly (survival of about 10⁻⁶). The high-acid survival values obtained after adaptation were in part due to the use of acid shock protocol A. When these four strains were grown on glucose in minimal medium E, acid adapted in the same medium at pH 5, and then subjected to acid shock protocol B, the survival values differed greatly (Fig. 3B). The survival rates of cfa::kan or rpoS::Tn10 derivatives of strain ZK126 were one or two orders of magnitude lower, respectively, than the survival of the wild-type strain, whereas the rpoS::Tn10 cfa::kan derivative of ZK126 survived even more poorly (Fig. 3B). The survival rates had the same dependence on rpoS and cfa alleles seen in the acid shock experiments; wild type > cfa > rpoS > cfa rpoS. For example, in two separate experiments with adapted strains, the wild type, cfa, rpoS and cfa rpoS strains had survival values (± standard deviation) of 0.70 (± 0.06), 0.09 (± 0.016), 0.0061 (± 0.0014) and 0.000012 (± 0.000005), respectively, after 2 h of exposure to pH 3.0. We also found that when cultures at low cell densities (A₆₀₀ = 0.2) were acid shocked without adaptation the cfa or rpoS derivatives of strain ZK126 showed lower acid survival rates than the wild-type strain, whereas the double mutant had the lowest acid survival rate (Fig. 3C). After adaptation, survival of all strains except the double mutant strain increased greatly. We conclude that during acid adaptation the induced expression of cfa and the subsequent synthesis of CFA in cell membrane phospholipids is an important factor in protection of cells from death by extreme acid shock.

Other strains lacking CFA show different degrees of acid sensitivity: the role of RpoS

The acid resistance of strain ZK126 was three or four orders of magnitude greater than strain MG1655 when tested by acid shock protocol A. An obvious explanation for this difference would be the presence of different rpoS alleles in the two strains. Indeed, there are differences in the rpoS DNA sequences for the two strains reported in GenBank that result in changes in two amino acid residues in the N-terminal half of the protein plus a frameshift such that the MG1655 protein is 66 amino acids shorter than that of strain ZK126. In order to test whether the differences in acid sensitivity between these two strains was owing to the activities of their RpoS proteins, we transformed strain MG1655 with the rpoS plasmid pMMKatF2 (Mulvey et al., 1988) (Fig. 4). The rpoS gene of this plasmid is reported to have the same sequence as that of strain ZK126 (Zambrano et al., 1993). The strain carrying the rpoS plasmid did indeed become 100-fold more acid resistant, however the introduced rpoS gene did not increase the acid resistance of strain MG1655 to the level seen for strain ZK126 under the same conditions (Fig. 1), indicating that strains MG1655 and ZK126 differed in a factor(s) other than RpoS that plays an important role in acid resistance.

We also tested the acid shock response of a number of other E. coli strains carrying a wild-type cfa gene or the cfa::kan allele. Previous workers in this laboratory (Taylor and Cronan, 1976; Grogan and Cronan, 1986) used strain FT1 and a derived cfa mutant strain FT17 for study of CFA function. We found that FT1, like MG1655, was much more acid sensitive (100- to 1000-fold) than strain ZK126 (Fig. 5). Strain YYC1106, a cfa::kan derivative of strain FT1, had a survival only two- to threefold lower than that of strain FT1 (Fig. 5) when tested by acid shock protocol A. We also constructed rpoS::Tn10 and cfa::kan rpoS::Tn10 derivatives of strain FT1 (strains YYC1256 and YYC1257 respectively), which were found to be 100-fold and 1000-fold more sensitive to acid than the parental strain FT1 (Fig. 5). We attribute the behaviour of FT1 and its derivatives to a deficiency of RpoS function because strain FT1 acquired an acid resistance similar to that of strain ZK126 upon introduction of the rpoS plasmid pMMKatF2 (data not shown). Therefore, the different acid sensitivities of strains FT1 and ZK126 can be attributed solely to RpoS function.

Finally, to examine the generality of the role of CFA in protection from acid shock, we examined one additional strain, K19, which is a direct descendent of Y-Mel, one of the earliest strains in the E. coli K-12 lineage. Strain K19 was found to be 100-fold more sensitive than strain ZK126 to acid shock protocol A, and the cfa::kan derivative of K19, strain JWC255, was five- to 10-fold lower in acid survival than strain K19 when tested by acid shock protocol A and 100-fold lower when tested by acid shock protocol B (data not shown). Therefore, CFA also plays an important role in this strain.

Strains, media and chemicals

All bacterial strains were derivatives of E. coli K-12 and are listed in Table 1. Genetic markers were transferred among strains by phage P1vir transduction. Strains carrying the cfa::kan mutation were made by P1 transduction from strain JC7623 containing the kanamycin insertion within the cfa gene. Strains carrying the rpoS::Tn10 mutation were made by P1 transduction from strain UM122, which was kindly provided by P. Loewen (Mulvey et al., 1988). Strains YYC1283 and YYC1284 carry the fabA2 temperature-sensitive mutation, which was transduced from strain DC308 (Clark et al., 1983) into strains YYC1272 and YYC1273, respectively, with selection for tetracycline resistance followed by screening for the temperature-sensitive fabA2 phenotype.

Table 1. . List of strains and plasmids (Clark et al. (1983); Taylor and Cronan (1976); Winans et al. (1985); Mulvey et al. (1988); Bohannon et al. (1991); Eichel et al. (1999); Wang et al. (1992)).

The liquid media used were LB or medium E (Vogel and Bonner, 1956). When minimal medium E was used, glucose (0.4%) was added as carbon source. The solid medium was RB agar, a version of LB containing less yeast extract (Chang and Cronan, 1982). The concentration of antibiotics used in media were 50 mg l⁻¹ kanamycin sulphate, 12 mg l⁻¹ tetracycline hydrochloride, and 100 mg l⁻¹ sodium ampicillin.

Tergitol NP-40 detergent, palmitoleic acid (C16:1) and oleic acid (C18:1) were from the Sigma Chemical. The 17-carbon CFA, cis-9,10-methylenehexadecanoic acid (C17-CFA), was chemically synthesized from palmitoleic acid in this laboratory by the method of Goff (1964) followed by purification by argentation chromatography (DeVries, 1963). The sample of cis-9,10-methyleneoctadecanoic acid (dihydrosterculic acid) (C19-CFA) was obtained from Supelco.

Fatty acids were first neutralized with 2 N KOH in 80% ethanol to form the more soluble potassium salt that was mixed with a volume of 20% Tergitol NP-40 (sufficient to give a final NP-40 concentration of 0.1%) and this solution was then added to LB medium.

Acid shock, acid adaptation and enzyme assay

Acid shock of cell cultures was carried out by two different protocols. In protocol A, HCl (6 N) was gradually added to cultures grown in LB medium until a pH of 3.1–3.2 was reached. The pH values were monitored by pH measurements on an identical, but separate, culture. After acid addition, the cultures were shaken at 37°C for 40 min. The untreated control (neutral pH) cultures were immediately diluted and plated for colony formation. The cultures were 25-fold dilutions of a overnight culture that were grown into early stationary phase (approximate A₆₀₀ of 2.0 or 1–2 × 10⁹ ml⁻¹). Early stationary phase was defined as 1 h after the turbidity of the culture ceased to increase. Cultures grown longer into stationary phase gave erratic results, presumably because of heterogeneity of the cells of the culture.

In protocol B, 1 ml samples of cultures grown in minimal medium E supplemented with glucose were harvested by centrifugation at room temperature in an Eppendorf minicentrifuge. The cells were then washed with medium E at pH 3 (adjusted with HCl) and resuspended in 1 ml of pH 3 medium E. The cell suspensions were shaken at 37°C for 1, 2 or 3 h, and the cells were diluted into medium E at pH 7 and plated on RB agar medium. The early stationary phase cultures were 100-fold dilutions of an overnight culture grown to an approximate A₆₀₀ of 1.5, when the growth of the cultures were beginning to slow from exponential growth. Control samples received the same treatment except that medium E at pH 7 was used throughout the procedure. Survival is defined as the ratio of colonies formed on RB agar medium after acid treatment to the number of colonies formed on the same medium by cultures that were not treated with acid. The survival values given were the average of at least two experiments and the range of the values were ± 50% of the values. We also performed protocol B with the resuspended cells at 1 × 10⁸ cells ml⁻¹ as carried out by others (Lee et al., 1994; Small et al., 1994; Lin et al., 1995) and obtained results consistent with those obtained using the more concentrated cell suspensions.

When the effects of fatty acid-supplemented medium were tested, strains YYC1283 and YYC1284 were first grown in LB medium at 30°C overnight and the cells were diluted 100-fold into LB medium containing the proper fatty acid supplements and grown for 16 h at 42°C. The cells were subsequently subcultured by a 10-fold dilution into the fatty acid-supplemented LB medium and grown to early stationary phase (approximate cell density at A₆₀₀ of 1.7–2.0) at 42°C. Acid shock to pH 3 was carried out on fatty acid-supplemented cultures by protocol A at 42°C for 20 min.

Acid adaptation requires early log phase cultures, and when performed in LB medium the cells were subcultured three times at 37°C while in early log phase (a total dilution of 10⁵) and the final culture was grown to early log phase (A₆₀₀ of about 0.15). The cultures were adjusted to pH 5 by gradual addition of 6 N HCl and allowed to continue growth to an approximate A₆₀₀ of 0.3 and then subjected to pH 3 acid shock using protocol A. The control and unadapted cultures were treated similarly to the adapted cultures, except that the cultures were maintained at pH 7 and, at an approximate A₆₀₀ of 0.3 (about 3 × 10⁸ cells ml⁻¹), the unadapted cultures were acid shocked to pH 3. When acid adaptation was carried out in minimal glucose medium, the cells were subcultured twice (a total dilution of 10³) and the cultures grown to an A₆₀₀ of 0.10. The adapted cultures were adjusted to pH 5 and, after one doubling, these cultures were subjected to pH 3 acid shock using protocol B. The control and unadapted cultures were treated similarly to the adapted cultures, except that the cultures were maintained at pH 7 throughout and, at an approximate A₆₀₀ of 0.2, the unadapted cultures were acid shocked to pH 3.

β-Galactosidase activity was used to measure the hydrolysis of o-nitrophenyl-β-d-galactopyranoside as described by Miller (1972) and is expressed as Miller units.

CFA synthase activity was assayed as described previously by Taylor and Cronan (1976).

Construction of the cfa insertion null allele

Plasmid pAYW60 carrying a kanamycin-resistant gene in the middle of the cfa gene was constructed by Dr A.-Y. Wang of this laboratory. A 2 kbp BamHI DNA fragment encoding the kanamycin resistance determinant of plasmid pHP45Ω-Km (Fellay et al., 1987) was converted to blunt ends by T4 DNA polymerase and ligated to linearized plasmid pAYW19 (Wang et al., 1992) that had been cut with KpnI (a site within the cfa gene) and converted to blunt ends by T4 DNA polymerase. This disruption would block transcription (Fellay et al., 1987) of the cfa gene after codon 17 of the 382-codon cfa gene and thus none of the regions conserved in this protein family (Grogan and Cronan, 1997) would be expressed. This plasmid, pAYW60, was linearized by digestion with SacI (a site located outside the cfa gene), transformed into strain JC7623 and kanamycin-resistant ampicillin-sensitive colonies were isolated (Winans et al., 1985).

The construction of the cfa promoter–lacZ fusion lambda phages in which lacZ expression is driven by P1–I–P2 (the intact cfa promoter), P2 alone, or I–P2 and P1–I, respectively, was described previously (Eichel et al., 1999). These phages were then used to lysogenize the wild-type strain ZK126 and its isogenic rpoS strain ZK1000 (see Table 1).