Release of Lipopolysaccharide in Escherichia coli Resistant to the Permeability Increase Induced by Ethylenediaminetetraacetate

Mutants of Escherichia coti that fail to show the characteristic increase in permeability induced by brief treatment with ethylenediaminetetraacetate have been isolated. They were examined to see whether they release lipopolysaccharide as expected during such a treatment. Mutants derived from two strains released only 20 to 40% less lipopolysaccharide than their parents, but further analysis of one mutant and its parent showed that this represented reduction of a specific lipopolysaccharide fraction. Both organisms released two lipopolysaccharide components differing in sedimentation rate in sucrose density gradients, but while both released similar amounts of the faster moving component, the mutant released only about one-third as much of the slower moving component. These results suggest that release of the slower component may be required for the permeability increase induced by EDTA.


SUMMARY
Mutants of Escherichia coti that fail to show the characteristic increase in permeability induced by brief treatment with ethylenediaminetetraacetate have been isolated. They were examined to see whether they release lipopolysaccharide as expected during such a treatment. Mutants derived from two strains released only 20 to 40% less lipopolysaccharide than their parents, but further analysis of one mutant and its parent showed that this represented reduction of a specific lipopolysaccharide fraction. Both organisms released two lipopolysaccharide components differing in sedimentation rate in sucrose density gradients, but while both released similar amounts of the faster moving component, the mutant released only about one-third as much of the slower moving component. These results suggest that release of the slower component may be required for the permeability increase induced by EDTA.
Brief treatment of Escherichia coli with ethylenediaminetetraacetate increases cell permeability and releases part of the lipopolysaccharide of the cell wall into the medium (1,2). The relationship of these two phenomena has not been defined. If they are related, then mutants selected for absence of the permeability effect might not be expected to release LPS.1 This paper describes the isolation and characterization of two mutants resistant to the permeability-promoting effect of EDTA. These mutants release LPS, but in reduced amounts.
Further study of one mutant revealed that its LPS, like that of its parent, yielded two fractions on sucrose gradient ultracentrifugation, and that one of these was released in normal, and the other in reduced amounts. EXPERIMENTAL PROCEDURE Materials and IWethods-Actinomycin D was the gift of Dr. A. Patchett of Merck Sharp and Dohme, and colitose and 3-deoxy-* Present address, Department of Biology, Johns Hopkins University, Baltimore, Maryland 21218.
1 The abbreviation used is: LPS, lipopolysaccharide.
Yeast transfer ribonucleic acid was obtained from Miles Laboratory.
Additional colitose was prepared from LPS released by EDTA treatment of E. coli Olll:B4 (3). LPS was hydrolyzed in 0.2 N HzS04 at 95" for 15 min and neutralized with saturated Ba(OH)*.
The sugars thus released were subjected to descending chromatography with the solvent of Colombo et al. (4) and detected with alkaline AgN03.
The area corresponding to authentic colitose was eluted. After periodate oxidation both at 55 and 23", the isolated material and authentic colitose gave identical absorption spectra in the thiobarbituric acid assay. The isolated material showed no detectable contamination by other sugars (less than 5%) on rechromatography.
Cultures were grown at 37" with shaking to provide aeration.
Turbidity was measured at 530 rnp.

1109
EDTA Treatment-Cells were harvested at room temperature in the exponential phase of growth, washed and resuspended in 0.12 M Tris-HCl, pH 8.0, and treated with EDTA for 2 min as previously described (10). EDTA action was stopped by adding MgCls at 2 to 5 times the EDTA concentration or by dilution into growth medium.
Determination of Permeability-The degree of permeability induced by EDTA was generally measured by assaying for uptake of actinomycin.
This was done in three ways. (a) Cells were measured for loss in viability after exposure to the antibiotic.
EDTA-treated cell suspensions were diluted at least lo-fold into growth medium containing 10 pg of actinomycin per ml and incubated in this medium at 37" for half a doubling time. Viable centers were assayed before and after incubation with actinomycin.
(b) Uptake of actinomycin was assayed directly by exposing cells to 14C-actinomycin in growth medium at 37" for times noted in the text. Uptake was terminated by chilling the cells to 0" (1). The cells were centrifuged at 4", washed twice with 0.15 M NaCl at O-4", suspended in 0.15 M NaCl, and counted.
(c) Uptake was also assayed by determining the sensitivity of RNA synthesis to the antibiotic.
Incorporation of uracil-l-14C into acid-precipitable material in the presence and absence of actinompcin was determined, and actinomycin sensitivity was calculated as previously described (10).
Permeability to o-nitrophenylgalactoside was assayed by comparing P-galactosidase activit)y of intact cells with that of toluenized cells as previously described (10).
Release of LPS-EDTA-treated and control cell suspensions were separated into cell pellet and supernatant fluid by centrifugation.
The amount of LPS in each was determined by assaying for 3-deoxyoctulosonate (AB1105 and NLlOl) or for colitose (J-5 and NL102) after hydrolyzing samples for 15 min at 95'. Colitose was determined by the method of Cynkin and Ashwell (11) directly on resuspended pellets and supernatant fluids as previously described (3). 3-Deoxyoctulosonate was determined by the method of Weissbach and Hurwitz (12) on the supernatant fluid and on LPS extracted from the pellet. These fractions were prepared as follows. The supernatant fluid was dialyzed at 4" against 0.10 M sodium phosphate buffer, pH 7.0 (100 volumes), and then against three portions of distilled water.
The pellet (containing cells from 100 ml of culture) was suspended in 1.0 ml of water and extracted twice with hot phenol (13). The combined aqueous phases were dialyzed against distilled water (250 volumes) yielding a preparation containing approximately equal amounts of LPS and RNA (13). When 3-deoxyoctulosonate was added to the crude LPS, and assayed, the values obtained were additive, indicating that the extract did not interfere with the assay.
Sucrose Density Gradient Centrifugation-Cells were grown in medium containing 14C-galactose, treated with EDTA, and centrifuged.
The supernatant fluids were dialyzed for 24 hours against 0.05 M sodium phosphate buffer, pH 7.0 (two 250-volume portions).
Approximately 0.20 ml was applied to 4.6 ml of a 5 to 20% sucrose gradient in 0.05 M sodium phosphate buffer, pH 7.0. Gradients were centrifuged in the SW39 head of a Beckman model L-2 ultracentrifuge.
Fractions of 0.2 ml of volume were collected through t.he bottom of the centrifuge tubes. Water (1.8 ml) was added to each fraction and 1.0 ml of the diluted fractions was counted.
No appreciable quenching by sucrose was observed.
An average of 85% of the counts applied to the gradients were recovered.
Lipid Extraction and Analysis-Samples were extracted at room temperature two to three times with an equal volume of 1-butanol.
The combined butanol extracts were washed two to three times with water, concentrated by evaporation, applied to Silica Gel G thin layer plates (Mann Research Laboratories), and the chromatographs developed in a chloroform-methanolwater (65:25:4, v/v) solvent as described by Ames (14). After development, the plates were exposed to iodine vapors to stain lipid components, sprayed with ninhydrin to detect amino groups, and sprayed with the reagent of Dittmer and Lester (15) to detect organic phosphate.

Selection of Nutants
Although normal E. coli cells are impermeable to actinomycin, EDTA-treated cells admit enough actinomycin to cause a 95% or greater loss in viability.
Mutants which do not show such increased permeability (per mutants) were selected by their ability to survive exposure to actinomycin after EDTA treatment. Since a small percentage of cells survive but are not mutants (lo), the survivors were repeatedly exposed to the selection procedure until mutants became numerous.
Two classes of mutants were obtained: permeability mutants of the type being sought, and mutants permeable to actinomycin after EDTA treatment but resistant to its action (actinomycin-resistant mutants). These latter will be described in a separate communication.
Per mutants were distinguished from resistant mutants by screening for uptake of actinomycin after EDTA treatment.
Mutants of l?. coli '4B1105 were selected by the following procedure.
A fully grown culture was treated with nitrous acid by the method of Kaudewitz (16). Approximately 2% of the cells survived.
The cells were suspended in enriched minimal medium, to a density of about 3.0 X lo8 cells per ml, incubated for 2 hours, and treated with EDTA.
The treated cells were diluted with an equal volume of enriched minimal medium containing actinomycin (final concentration, 10 kg per ml), incubated for 45 min, further diluted with 4 volumes of enriched minimal medium and incubated overnight. The selection procedure was repeated four times; the survivors were diluted into enriched minimal medium, grown to the midexponential phase, treated with EDTA, diluted lo-fold into enriched minimal medium, and incubated overnight in the presence of 10 pg of actinomycin per ml. About 3% of the cells survived the first, second, and third treatments.
Both the fourth and fifth treatments gave about 40% survival.
Survivors from the final selection were screened for uptake of actinomycin, taking advantage of t,he observation that, actinomytin colors cells that take it up. Cultures grown from single colonies and from the parent strain were treated with EDTA, incubated with actinomycin (5 pg per ml) for 15 to 20 min, and centrifuged.
A mutant (NLlOl), carrying a mutation designated per-l on the basis of data presented below, gave a paler pellet than AB1105 cells. A single colony isolate was used to prepare the stock strain.
Mutants of E. coli J-5 were selected as described for AB1105 except cells were not exposed to mutagen prior to the selection procedure, and cells were grown in minimal, as opposed to enriched minimal, medium prior to EDTA treatment (following EDTA treatment, cells were incubated in enriched minimal medium).
It was found with J-5 that the number of survivors Release in Permeability Mutants 1. Uptake of 14C-actinomycin by EDTA-treated cells of NLlOl and AB1105. NLlOl and AB1105 were grown to approximately equal turbidity in minimal medium and treated with EDTA as described under "Experimental Procedure." Exposure to EDTA was terminated by dilution into 4f volumes of previously warmed minimal medium.
The diluted cells were incubated at 37", 2 to 3 min later 4 rg of l*C-actinomycin per ml were added (0 time), and samples were taken at various times thereafter.
increased successively with each treatment: 0.0001, 0.0005, 0.75, and 36%. Survivors from the final passage were further screened. In this case, to distinguish per mutants from actinomycin-resistant mutants, survivors were treated with EDTA and tested for the absence of a noticeable lag in growth in medium containing actinomycin, since actinomycin-resistant mutants show such a lag.2 One survivor (NL102) which did not show a lag was designated as carrying a per-2 mutation on the basis of data presented below.
A single colony isolate was used to prepare the stock strain.
The organisms thus isolated are derivatives of the starting strains, and not contaminants, since NLlOl has the same auxotrophic requirements for proline, histidine, and thiamine as AB1105; and NL102, like J-5, contains colitose as a constituent of its LPS.
In the growth media employed in this study, the per mutants grow at 37" at virtually the same rate as the parental strains.
In minimal medium NLlOl and AB1105 have a doubling time of 60 min and, in enriched minimal medium, both have a doubling time of 35 min. In trypticase soy minimal galactose medium, NLl02 and J-5 have a doubling time of 53 min. In minimal medium, J-5 has a doubling time of about 75 min and NL102 has a generation time about 10% greater than this.

Response of NLIOI to EDTA Treatment
Permeability-In contrast to the parent strain AB1105, NLlOl (per-l) grown in minimal medium is virtually impermeable to actinomycin after EDTA treatment. Following EDTA treatment, uptake of 14C-actinomycin by NLlOl was less than 5% that of the parent strain (Fig. 1). In the parent strain uptake of actinomycin by EDTA-treated cells halts RNA synthesis (1) EDTA treated mutant cells (Fig. 2), a result consistent with little uptake of the antibiotic. EDTA treatment of E. coli increases permeability to a variety of compounds (1). To see if impermeability of NLlOl to actinomycin truly reflects resistance of a general permeability barrier to EDTA, entry of another normally excluded compound was studied.
Since AB1105 is lactose permease negative, o-nitrophenylgalactoside normally cannot be hydrolyzed by internal /3-galactosidase, but can be hydrolyzed after the cells are treated with EDTA (1). In contrast, NLlOl remained impermeable to this compound after EDTA treatment (Table I).
The response of NLlOl to EDTA was influenced by the composition of the growth medium.
When grown in minimal medium, EDTA-treated cells of NLlOl were practically impermeable to actinomycin (Figs. 1 and 2)) but when grown in enriched medium they became partially permeable, incorporating about 30% as much actinomycin as the parent strain (Table II).
Under these conditions, the mutant showed 50% survival after exposure to actinomycin; the parent strain, as expected, showed much less (2%) survival (Table II).
In subsequent studies of LPS release, the more definitive minimal medium was used.
Release of LPS-NLlOl and BB1105 cells were washed and resuspended in Tris, one portion of each treated with EDTA, and the suspensions separated into pellet and supernatant fluid by centrifugation.
To determine the release of LPS, the pellets and the supernatant fluids were assayed for 3-deoxyoctulosonate, a compound which has been found only in LPS (17). As measured by percentage of release of 3-deoxyoctulosonate, AB1105 released about 50% of its LPS upon EDTA treatment, while NLlOl released about 30% (Table III).
The total amount of 3-deoxyoctulosonate in both strains was the same, about 0.0095 rmole per mg of cells, dry weight.
Less than 2y0 of the protein of the   To facilitate further studies of release of LPS by per organisms, NL102 (per-a) and its parent, J-5, were used. Their LPS can readily be assayed by measuring colitose, a sugar found only in the antigenic side chains of the LPS of these organisms; or by incorporation of radioactively labeled galactose, since lack of UDP-galactose 4-epimerase (9)  POh) containing 5.5 pg per ml (A) or 5.9 rg per ml (B) '4C-actinomycin and the amount of label taken up by the cells in 4 min at 37" was determined.
To measure LPS release, the remainder of the samples was separated into pellet and supernatant fluid and the colitose content of each was det,ermined.
Permeability and Release of LPX-Increased permeability, as measured by actinomycin sensitivity, and percentage of LPS release, as measured by colitose content, were determined for EDTA-treated NL102 and J-5 after growth in three different media (Table IV).
In all three media the mutant released 30 to 40% less LPS than the parent strain.
As measured by colitose, no difference in the LPS content of mutant and parent strains was found.
NL102 is much less permeable to actinomycin after EDTA treatment than is its parent.
It is least permeable after growth in minimal medium.
As with NLlOl (per-l), growth on rich medium (Trypticase soy galactose) increases the permeability induced by EDTA.
J-5 and NL102 grow poorly on minimal medium, but it was found that supplementing minimal medium with 10% trypticase soy broth (trypticase soy minimal galactose) both improved growth and preserved the relative impermeability of NLlOQ after EDTA treatment, so trypticase soy minimal galactose medium was used in all further studies.
Kinetics of Increase in Permeability and Release of LPS-The release of LPS and the initial rate of actinomycin uptake as a function of time of exposure to EDTA were measured (Fig. 3). As in E. coli 0111 : 134 (2)

Sucrose Density Gradient
Studies on Released LPS-The LPS of J-5 and NL102 was uniformly labeled by growing cells at least four generations in Trypticase soy minimal galactose medium containing '4Cgalactose. The cells were then treated with EDTA for 30 set or 2 min and the supernatant fluid was dialyzed and layered on a 5 to 20% sucrose gradient.
The LPS released both by J-5 and NL102 sedimented as two components (Fig. 4): a narrow peak (P2) sedimenting slightly faster than a tRNA marker (S N 4) and a faster sedimenting peak of broader distribution (Pl). Both J-5 and NLl02 released roughly equal amounts of the PI component.
In contrast, NLl02 released significantly reduced amounts of 1'2 when compared to the parent strain.
In the experiments presented in Fig. 4, the mut.ant strain released only about one-third as much P2 as the parent (Table V). Similar results have been obtained in other experiments.
The sedimentation constants of Pl and 1'2 were calculated with use of transfer RNA as a reference and assuming simple proportionality between s and distance traveled from the meniscus (19). Sedimentation values of approximately 6 S and 12 S for P2 and Pl, respectively, were obtained.
Like AB1105 and 0111 :B4, the supernatant of EDTA-treated J-5 cells contains noncovalently bound phospholipid.3 When supernatants from J-5 and NL102 were extracted with butanol prior to centrifugation in a sucrose density gradient, a single peak of radioactivity was found at approximately the same position as P2, indicating that removal of this lipid reduces the sedimentation rate of Pl to a value approximating that of P2.

DISCUSSION
Mutants resistant to the permeability promoting effect of EDTA (per mutants) were isolated to help determine whether release of LPS is both necessary and sufficient for the EDTAinduced permeability effect. Some earlier observations suggested a functional relationship.
Both the increased permeability and LPS release occur equally well at 0 and 37" (3), and both occur rapidly, being complete within a few minutes of EDTA exposure (2). Both reactions are prevented or terminated by divalent cations but divalent cations do not reverse either effect. Restoration to normal permeability does not require RNA, protein, or peptidoglycan synthesis but does require energy metabolism, an observation which would be compatible with a requirement for resynthesis of cell wall LPS (10).
On the other hand, it was found with E. coli 0111 :B4, that the amount of LPS released by EDTA over a pH range of 7.0 to 8.0 did not vary, whereas over this same range the induced permeability increased with increasing pH, indicating that these two effects could be partly dissociated (3).
The per strains isolated in this study showed normal growth rates and no significant differences in LPS content, but EDTAt,reated cells of the mutants released less LPS than the parent strains.
Further study of one strain revealed that this reflected reduced release of one specific LPS fraction.
The possibility that the phenotype of the per mutants may result from multiple mutations cannot be excluded.
The same phenotype was obtained in two independent selections and in two different strains, making this possibilit#y less likely.
However, since the cells of one st,rain were treated with mutagen, and the mutants selected after several cycles of an enrichment procedure, multiple mutations could accumulate.
Preliminary studies suggest that some revertants of per-2 that become permeable after treatment with 3 S. B. Levy, personal communication.
EDTA have not regained the original pattern of LPS release but instead may have further changes in their LPS. Genetic analysis of per mutants and their revertants, currently under way, will obviously be of great interest and help to clarify these phenomena.
Although the per mutation leads to lesser release of LPS, mere reduction in amount of LPS released in the mutant is insufficient to explain its failure to be permeable.
As shown in Fig. 3, after 13 min of EDTA treatment, NL102 released as much LPS as J-5 did at 30 set but this release was associated with much less permeability in the mutant than in the parent. This result suggested that perhaps the release of a particular fraction of LPS is required for the increased permeability effect.
The LPS released by EDTA treatment of E. coli 0111 :B4, the parent strain of E. coli J-5, was previously found to contain two major components (3). In the analytical ultracentrifuge, one component, Fl, sedimented as a broad heterodisperse peak of 10 to 80 S and the second component, F2, as an ultrasharp peak of 5.4 S at infinite dilution.
If, prior to ultracentrifugation, the released LPS was treated with chloroform-methanol to remove noncovalently bound lipid, Fl sedimented as a single symmetrical peak of 11 S, and F2 was unchanged.
The two components were qualitatively identical in their chemical composition, but differed in the proportion of their carbohydrate components.
When the LPS released by EDTA from J-5 and its per-2 mutant was studied by sucrose density centrifugation, the released LPS showed two major components (Pl and P2) suggestive of the Fl and F2 components released by 0111 :B4. Like Fl, the sedimentation rate of Pl was reduced by organic solvent extraction; however, both before and after such extraction, the sedimentation rate of Pl was lower than found for Fl. This could be the result of an effect of the sucrose gradient on the molecular structure of this component or might reflect disaggregation since the sucrose gradients utilized less than 1% the concentration of LPS as did the analytical ultracentrifuge studies. It is of course possible that Pl and P2 do not represent the two chemically different LPS species, Fl and F2, but only reflect differing degrees of association of LPS with phospholipid.
Although the LPS released by both parent and mutant showed the same two components, one of these components (P2) was appreciably reduced in the mutant.
Thus the rather small reduction in total LPS release shown by the mutant represents a large decrease in a specific fraction.
Also, in J-5, the increase in release of P2 between 30 set and 2 min of EDTA treatment roughly paralleled the percentage increase in permeability during this time ( Fig. 3 and Table V). Both these results suggest a relationship between release of P2 and increase in permeability. It cannot yet be concluded that this LPS release is necessary for the permeability effect to occur. The two effects could be only operationally associated; for instance, they might be independent secondary effects of some primary action of EDTA.
The present results do, however, support a working hypothesis that release of this specific fraction of LPS is in some manner required for the EDTA-caused permeability change.