cellular responses ofbacillus subtilis …cellular response to th gram stain 847 treated with...

Vol. 156, No. 2JOURNAL OF BACTERIOLOGY, Nov. 1983, p. 846-8580021-9193/83/110846-13$02.00/0Copyright © 1983, American Society for Microbiology

Cellular Responses of Bacillus subtilis and Escherichia coli tothe Gram Stain

T. J. BEVERIDGE'* AND J. A. DAVIES2Department of Microbiology, College of Biological Science, University of Guelph, Guelph, Ontario, Canada

NJG 2WI1' and Department of Chemistry, The University of Toledo, Toledo, Ohio 436062

Received 11 April 1983/Accepted 22 June 1983

Exponentially growing cells of Bacillus subtilis and Escherichia coli were Gramstained with potassium trichloro('q2-ethylene)platinum(II) (TPt) in place of theusual KI-12 mordant. This electron-dense probe allowed the staining mechanismto be followed and compared with cellular perturbations throughout the stainingprocess. A crystal violet (CV)-TPt chemical complex was formed within the cellsubstance and at the cell surface of B. subtilis when the dye and Pt mordant wereadded. The ethanol decolorization step dissolved the precipitate from the cellsurface, but the internal complex was retained by the cell wall and remainedwithin the cell. This was not the case for E. coli; the ethanol decolorization stepremoved both surface-bound and cellular CV-TPt. During its removal, the outermembrane was sloughed off the cells until only the murein sacculus and plasmamembrane remained. We suspect that the plasma membrane was also perturbed,but that it was retained within the cell by the murein sacculus. Occasionally, smallholes within the murein and plasma membrane could be distinguished throughwhich leaked CV-TPt and some cellular debris. Biochemical identification ofdistinct envelope markers confirmed the accuracy of these images.

Although the Gram reaction is a technique onwhich much of the criteria for bacterial taxono-my resides, the actual effect of the stain on cellsis imperfectly understood. There is generalagreement that the staining response betweengram-negative and -positive bacteria is due to afundamental permeability difference betweenthe two types of cells and that this differenceresides within the molecular fabric of their en-veloping layers (1, 2, 4, 26, 28, 32). Clearly thereare distinct structural differences between thetwo wall types (4); these unique structural traitsare the result of profoundly different chemistries(4, 11, 25). Presumably, it is these features that,in concert, control the end result of the Gramstain.

Recently, we have defined the chemical mech-anism of the Gram stain and, using this informa-tion, have designed potassium trichloro('q2-eth-ylene)platinum(II) (TPt) as an electron-densechemical mordant to replace the KI-12 of Gram'siodine solution (9). Since crystal violet (CV) is achloride salt, it interacts with iodide in solutionby simple metathetical exchange (i.e., the bulky1- replaces the smaller Cl-, and a CV-I precipi-tate results) (9). The TPt anion (TPt-) is also abulky anion and complexes with CV+ to pro-duce a CV-TPt precipitate which is completelyanalogous to CV-I (9). In this paper, we useTPt- as an electron-dense probe to study the

cellular responses of Bacillus subtilis and Esche-richia coli to the various steps of the Gram stain.

MATERIALS AND METHODS

Growth of cells, the Gram staining method, process-ing for electron microscopy, and conditions of electronmicroscopy and EDS. B. subtilis 168 and E. coli K-12stain AB274 were used for this study. The cells weregrown and processed through the Gram reaction asoutlined according to Davies et al. (9). LB broth (3)was used with both bacteria for the 3H-labeled phos-pholipid experiment. All processing for electron mi-croscopy, actual imaging, and energy dispersive X-rayanalysis (EDS) was done as previously described (9).All images from thin sections reported in this paperhad the contrast increased by uranyl acetate and leadcitrate staining, except for the elemental distributionmaps which used unstained thin sections.

Platinum distribution maps were produced with thePt (Mo, and La) lines. Lines slightly removed fromthe Pt lines (both upstream and downstream) wereused to produce maps of the continuum to check thatthe Pt maps were real. The scanning transmissionelectron microscope (STEM) was operated at 80 kVand 85 ,uA with a 50-,um condenser aperture at a magnifi-cation of x37,500. The total scan time was 500 s.Although some Epon 812 embeddings were attempt-

ed, Durcupan (Fluka AG, Buchs SF, Switzerland),which is a water-miscible resin, proved to be theplastic of choice, since it did not leach the CV-TPtfrom the cells. A time sequence of embeddings wasdone on E. coli during the ethanol decolorization step.In this case, separate batches of CV-TPt cells were

846

on March 1, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

CELLULAR RESPONSE TO TH GRAM STAIN 847

treated with ethanol for 15, 30, 60, 120, and 300 sbefore equilibration into (N-2-hydroxyethylpipera-zine-N'-2-ethanesulfonic acid) HEPES buffer and fixa-tion.Atomic absorption spectroscopy. Both E. coli and B.

subtilis were estimated for the TPt content after theGram reaction by detection of total Pt per cellular dryweight by atomic adsorption. Cells were well washedafter the Gram reaction in 0.05 M HEPES (pH 6.8)buffer until the supernatant was free of material whichabsorbed at 660 nm (a Xmax for CV). They were thenwashed two times in water, hydrolyzed in fumingHNO3, and the residue was dissolved in 0.01 N HNO3containing 250 p.g of CsCl ml-' as a carrier. Flameatomic absorption was performed on a Perkin-Elmermodel 2390 spectrophotometer working in the C2H4-N20 flame mode.

Biochemical analyses of the Gram reaction superna-tant. (i) Protein. Protein was estimated by the methodof Markwell et al. (21), employing lysozyme (SigmaChemical Co.) as a standard. CV and its variouscomplexes have 'max at ca. 580 and 660 nm (9) and willinterfere with the 600-nm line of the Folin-phosphomo-lybdate-phosphotungstate complex for protein deter-mination. Therefore, to compensate for this spectralinterference, CV was added to the protein standards inappropriate concentrations; i.e., the CV concentrationof each sample was estimated before determinationswere begun.

(ii) KDO. The thiobarbituric acid procedure ofWeissback and Hurwitz (33) was used, employingpurified 2-keto-3-deoxyoctonate (KDO) (Sigma Chem-ical Co.) as a standard. Since this method forms achromogen with a Xmax at 548 nm, we also had spectralinterference with this assay in the presence of CV.Addition of CV to the standard solutions provedunreliable. We therefore were able to estimate KDOonly under low CV concentrations in the supernatant(before ethanol decolorization) or in the control sam-ples (no CV) before and after ethanol decolorization.The ethanol control sample is a good estimate of KDOextraction during the decolorization of CV cells.

(iii) Incorporation of [3H]glycerol into the phospholip-ids of both bacteria and estimation of [3H]phospholipidrelease during the Gram reaction. [2-3H]glycerol (200.0mCi mmol-') was obtained from New England Nucle-ar Corp. For the labeling of B. subtilis, the methods ofboth Sargent (27) and Smit et al. (29) were used. Bothmethods incorporated 3H into the plasma membranefraction of the cells, and since the LB broth (3) of themethod of Smit et al. (29) gave a higher cell yield, itwas used to label cells for the Gram reaction. Thesame method was used to label E. coli membranes.The bacteria were grown overnight in 250 ml of LB

broth containing 0.01% glycerol at 22°C for B. subtilisand 37°C for E. coli. Each culture was pelleted bycentrifugation at 10,000 x g for 15 min, suspended in120 ml of LB broth containing 5 ,uCi of [2-3H]glycerolml-' and grown at the appropriate temperature for 4 h.Each culture was then washed two times with 100 mlof 0.05 M HEPES (pH 7.0) buffer, and the cells werechecked for the incorporation of the label (which waspositive in both cases).The cells were then put through the Gram regimen

by the method of Davies et al. (9), and the superna-tants of the various steps of the reaction were checkedfor radioactivity by liquid scintillation by using a

Beckman model LC250 counter. Aquasol (New En-gland Nuclear) was used as the scintillation fluid.

Preparation of potassium TPt. TPt was prepared bythe method of Davies et al. (9).SDS-PAGE. The discontinuous sodium dodecyl sul-

fate-polyacrylamide gel electrophoresis (SDS-PAGE)system of Laemmli (20) was used to monitor theprotein composition to the Gram reaction supernatantof E. coli. Much of the extracted material was mem-branous and contained high concentrations of CV thatcould not be removed by either extensive dialysis (10 x5 liters of buffer) or treatment with mixed bead resins(e.g., Amberlite MB-3 or ABX; British Drug Houses).This CV contamination effected the electrophoreticproperties of the proteins (smearing was common) andthe clarity of Coomassie brilliant blue staining. Theresulting gels were of poor quality (Fig. 8 is repre-sentative), but the major outer membrane proteinscould be differentiated above background.

RESULTSCellular CV-TPt content of B. subtilis and E.

coli. B. subtilis and E. coli were monitored forthe CV-TPt complex by atomic absorption de-tection of Pt during the Gram reaction beforeand after ethanol decolorization. In B. subtilis,1.98 and 1.71 mmol of CV-TPt per mg (dryweight) of cells were found before and afterethanol treatment, respectively. In E. coli, 2.44and 1.14 mmol of CV-TPt per mg (dry weight) ofcells were found before and after ethanol treat-ment, respectively. These values were calculat-ed from atomic absorption analysis of total Pt incells. Atomic absorption analysis also confirmedthat the CV-to-TPt ratio was 1:1. After ethanoltreatment, B. subtilis lost ca. 14% of its totalcellul-associated CV-TPt. The amount of precip-itate that remained associated with these cellsproduced the typical purple color of gram-posi-tive cells (9). E. coli, on the other hand, lost ca.53% of the CV-TPt precipitate. This must be alow estimate since large numbers of these cells(a total of 0.154 g [dry weight] of cells) tended toaggregate during ethanol treatment to trap CV-TPt within the central mass of bacteria. Thebrief ethanol treatment (9) was not long enoughto completely decolorize all of these cell masses,but light microscopy confirmed that a majorityhad been washed free of the dye.

Electron microscopy of B. subtilis during theGram reaction. Previous work had demonstratedthat CV-TPt could be detected within cells byelectron microscopy (9). The staining of thesesame sections of cells with uranyl acetate andlead citrate added enough contrast to the cellularstructure, without displacing the Pt probe (asmonitored by EDS), that ultrastructural detailcould be distinguished.

B. subtilis showed a surprising retention ofcellular detail when treated with CV (no TPt)and decolorized (Fig. la). Ribosomes, plasmamembrane, and the wall fabric could be detect-

VOL. 156, 1983


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

848 BEVERIDGE AND DAVIES

la

FIG. 1. (a) Thin section of Bacillus subtilis which has been treated with CV alone and then decolorized withethanol. This and all other sections have been contrasted with uranyl acetate and lead citrate. The dark areaswithin the cytoplasm are presumed to be ribosomes. In this and all other micrographs, the bar represents 100 nm.(b) This micrograph, from the same preparation as (a), reveals a high magnification of the wall to show that itremains intact after ethanol treatment. (c) This is a CV cell that has been treated with Gram's iodine to form theCV-I complex and then decolorized. The wall remains intact, as does the PM, but the cytoplasm contains voids(arrows) that once contained CV-I complexes.

ed. The wall remained intact (Fig. ib) and re-tained its typical 22- to 25-nm thickness (5).Cells that had been stained by the CV-iodidemethod resembled the CV cells in every wayexcept that the cytoplasm contained voids (Fig.lc). These were attributed to areas which hadpreviously contained the CV-I precipitate untilthe energy of the electron beam sublimed theiodine from the section to leave empty spaces(9).CV-TPt sections revealed electron-dense de-

posits scattered throughout the cytoplasm. Theplasma membrane was retracted from the wall(Fig. 2a). EDS of the electron-dense depositsconfirmed that they contained Pt (9). STEM-

EDS platinum maps of bacteria before and afterethanol decolorization failed to reveal differ-ences between the two treatments (Fig. 2b andc). Unlike atomic absorption spectropho-tometry, this technique is not sensitive enoughto differentiate between untreated and decolor-ized B. subtilis cells.

Plasmolysis was not unique to CV-TPt cells,but was seen in CV and CV-I cells as well (Fig.2d is representative). This phenomenon wasonly seen in decolorized cells and was, there-fore, a common response to the ethanol treat-ment. The ethanol had denatured and condensedthe cytoplasmic substance away from the plas-ma membrane (PM). Cells that were not decolor-

J. BACTERIOL.


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

F..

_I,I,~-

*, -

FIG. 2. (a) A B. subtilis cell that has been treated with CV and TPt and then decolorized. The CV-TPtcomplexes (arrows) stain dark in this preparation and were not removed by the ethanol treatment. In this and inother CV-TPt images, the dark areas have been identified as containing Pt by EDS. (b) STEM-EDS elementaldistribution map of a cell for Pt (i.e., CV-TPt) before ethanol treatment. Each dot represents a high Ptconcentration within the section and the highly concentrated area conforms to the shape of a bacterial cell. Mapsof the X-ray continuum (i.e., background radiation) for this and other samples demonstrated that the pointdistribution was due to Pt and was not an artifact of the mass thickness of the specimen. All maps are of

849


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/


TABLE 1. Protein extracted from B. subtilis byGram reaction

Total % TotalTreatment protein cellular

(mg)a protein

Control 0.334 2.0CV 0.356 2.1CV-l 0.350 2.1CV-TPt 0.382 2.3Control decolorized by alcohol 0.549 3.2CV decolorized by alcohol 0.736 4.3CV-I decolorized by alcohol 1.319 7.8CV-TPt decolorized by alcohol 1.104 6.5

a Data were derived from 10 ml of a culture with anoptical density at 600 nm of 0.600 with a total dryweight of 32.4 mg. Figures represent the average ofthree protein estimations for each sample by theMarkwell et al. method (21).

ized often contained CV-TPt deposits on thesurface of the cell wall (presumably this wasexternal precipitate that had condensed on to thecell surface during chemical metathesis of exog-enous CV+ and TPt-) (Fig. 2f). These were notseen after ethanol treatment (the ethanol haddissolved and washed them away) (Fig. 2a), butoccasionally, the wall was buckled and con-tained electron-opaque material within its lumen(Fig. 2e). EDS proved this material to be CV-TPt. This suggested that it was the fabric of thewall, and not the PM, which retained CV-TPtduring decolorization.

Biochemical identification of cellular productsextracted from B. subtilis during the Gram reac-tion. Table 1 shows the amount of protein whichwas extracted from B. subtilis during each stepof the Gram reaction. Both the CV-I and CV-TPtmethods were employed. In all cases, whetheriodide or TPt were used or whether the cellswere decolorized or not, the percent proteinextracted did not exceed 7.8% of the total cellu-lar protein. Presumably, it was the wall fabricwhich prevented more substantial loss of proteinfrom cells during the Gram reaction.

Table 2 shows the amount of phospholipidwhich was removed during each step of theGram reaction. Since ethanol is a membraneperturbant (12, 18) it is expected that lipid wouldbe removed. Of the cellular lipid 30 to 43% wasremoved by CV, CV-I and CV-TPt without

J. BACTERIOL.

ethanol treatment. This amount, no doubt, wasdue to the 20% ethanol within the CV stainingsolution (9). Ethanol treatment did not increasethe amount of lipid extracted. Presumably, thistreatment further contracted the wall fabric andmade it impermeable to the passage of additionallipid. No attempt was made to identify theextracted lipids.

Electron microscopy of E. coli during the Gramreaction. The response of E. coli to the Gramstain was much different than that of B. subtilis.These bacteria tended to agglutinate upon theaddition of CV and TPt to form large aggregatesof cells and CV-TPt precipitate. The precipitatewas found on the cell surface, in the cytoplasm(Fig. 3a) and within the periplasm (Fig. 3b).Alcohol decolorization reduced the amount ofcytoplasmic substance, congealed the remainingcellular contents, and produced large regionsdevoid of electron-scattering power (Fig. 3c).These cells did not produce a strong Pt signalwith EDS, and presumably the cytoplasmicvoids had initially contained CV-TPt (9). Inmany instances only the ghosts of cells remained(Fig. 3d).

Close examination of these bacteria duringdecolorization revealed a disruptive membraneeffect (Fig. 4). When a time sequence of embed-dings was made during the decolorization proc-ess, the membrane perturbation could be moreaccurately followed (Fig. 5a to f). Initially, theouter membrane (OM) began to buckle and folduntil it eventually sloughed off the cell surface.This continued until only small OM fragmentswere left, or until the murein layer (peptidogly-gan [PG]) was laid bare. In some cells, smallholes were eventually produced in the remaininglayers (PG and PM) through which CV-TPt and

TABLE 2. Amount of lipid ([3H]glycerol labeled)extracted from B. subtilis by Gram reaction"

Treatment No decolorizing With decolorizingalcohol alcohol

Control 31CV 30 32CV-I 43 39CV-TPt 42 39

a All data are expressed as the percent of totalcounts (lipid) extracted from the cells.

unstained thin sections which were ca. 60 nm thick. (c) Same as (b) but after ethanol treatment. Highconcentrations of Pt remain in the cell. (d) A cell treated with CV-I and decolorized which shows severeplasmolysis. (e) A cell that has been treated with CV-TPt and decolorized. The wall has buckled and has trappedCV-TPt within its lumen (arrow). The dark areas within the cytoplasm are also CV-TPt. These point EDSanalyses were done with a 5-nm beam size so that the spatial resolution of the technique was ca. 10 nm. (f). Thiscell has been treated with CV-TPt but has not been subjected to the ethanol decolorization step. Surface CV-TPtis evident (arrow). Beam size was also 5 nm.


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

FIG. 3. (a) An E. coli cell which has been treated with CV-TPt but not subjected to ethanol decolorization.The darkly stained areas are CV-TPt complexes (arrows). (b) A high magnification of the envelope from one ofthe cells of the preparation used in (a). CV-TPt is located within the periplasmic space (arrow). (c) The samepreparation after ethanol treatment. Only congealed material remains within the cytoplasmic space and the CV-TPt precipitate can no longer be distinguished. (d) A cell ghost after cellular lysis during ethanol treatment.

851


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/


FIG. 4. A CV-TPt E. coli cell that has been subjected to ethanol treatment for 120 s. The OM has separatedfrom the PG layer.

cellular debris extruded (Fig. 6). The end resultwas a lysed ghost that retained the original shapeand form of the bacterium (Fig. 3d).

Semi-quantitative STEM-EDS analyses ofCV-TPt cells before and after ethanol treatmentconfirmed that there was a drastic reduction ofPt (i.e., CV-TPt) in bacteria that had undergonedecolorization. In fact, elemental maps of thecells revealed that the CV-TPt distribution wasmore widely dispersed in the treated cells (Fig.7a and b). Point EDS analyses corroborated theelemental maps (Fig. 7c and d) and this wasespecially apparent when the spectrum of thetreated cells was subtracted from that of theuntreated cells (Fig. 7e); a substantial Pt peakremained.

Biochemical identification of cellular productsextracted from E. coli during the Gram reaction.Table 3 shows the amount of protein which wasremoved from E. coli during each step of theGram reaction. The decolorization step had aprofound effect on the amount of cellular proteinwhich was leached since up to 36% of the totalprotein was removed. Likewise, there was atremendous increase in the [3H]glycerol-labeledphospholipid that was liberated during ethanoltreatment (Table 4). The amount of radioactivelabel which was released from the CV, CV-I,and CV-TPt cells without ethanol treatment,once again, is a reflection of the ethanol contentof the CV staining solution (9).Although electron microscopy clearly re-

vealed that ethanol treatment perturbed the OM,

chemical evidence was desirable. KDO is aconstituent of the lipopolysaccharide (LPS)which is unique to the OM. Analyses for thiscomponent in the Gram supernatant revealedthat KDO was extracted by ethanol and CVtreatment (CV contains 20% ethanol) (Table 5).In addition, SDS-PAGE of the released productsdemonstrated that the OmpA, OmpC and possi-bly OmpF proteins were abundant in the Gramsupernatant (Fig. 8). Taken in concert, thesebiochemical data support the electron micro-scopic results and suggested that the OM hadbeen disrupted during the Gram stain.

DISCUSSIONThe results reported in this paper provide an

understanding toward the events that occur tobacteria during the Gram stain. Two bacteriawere chosen as representatives of the two Gramtypes, B. subtilis and E. coli, and they weresubjected to the Gram reaction using a newelectron-opaque probe to replace the iodide ofGram's iodine (9). Under the conditions em-ployed, these bacteria reacted unequivocally tothe stain; B. subtilis stained gram-positive, andE. coli stained gram-negative (9).With B. subtilis, a metathetical ion exchange

product (CV-TPt or CV-I) forms both at the cellsurface and within the cell substance during theGram stain. Ethanol treatment dissolves thesurface-bound product and releases small quan-tities of cellular protein and phospholipid intothe external milieu. It does not remove the

J. BACTERIOL.


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

PG PM1I

OM PGPM

FIG. 5. A series of micrographs of CV-TPt E. coli cells depicting the sequence of OM perturbation during theethanol decolorization step. Exposure times to ethanol were as follows: (a), 15 s; (b), 30 s; (c and d), 60 s; (e), 120s; and (f), 300 s. After 300 s, little, if any, OM remains on the cells. All images are the same magnification.

853

Om,

AL.


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/


PGtOM

*'

rrtC 'A;7:. t . 7R .tFIG. 6. A hole in the polar end of a CV-TPt cell after decolorization with ethanol. The debris that is leaking

out of the cell (large, left arrow) gave a strong Pt signal by EDS.

product from within the bacteria which is re-stricted from migration by the molecular fabricof the wall; presumably, the CV-TPt complex istoo large to penetrate through the interstices ofthe PG-teichoic acid matrix. (Under our growthconditions, the walls should be in the teichoic

C

acid form, and teichuronic acid should not bepresent [6]). The CV+ (1.88 by 0.65 nm; seereference 9) and TPt- (0.82 by 0.62 nm; seereference 9) ions permeate through the wallfabric and interact and combine with one anoth-er inside the cell. Here, a large precipitate of

d e1L1z /'C,Ct>ta Cu

PPtPL

FIG. 7. (a) STEM-EDS platinum map of an E. coli cell before ethanol decolorization. (b) Same as (a) but afterdecolorization. The dots are not as clustered and are more widely disbursed, making them appear smaller. (c)EDS spectrum of the cell in (a). (d) EDS spectrum of the cell in (b). The copper peaks in this and in (c) are derivedfrom the copper EM grids. The copper peak in (c) is larger since the specimen was closer to a grid bar. Thechlorine is from the plastic of the section. (e) EDS spectrum after (d) has been subtracted from (c). There is astrong Pt signal left which confirms that there was much less platinum in cells after ethanol decolorization.

J. BACTERIOL.


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/


TABLE 3. Protein extracted from E. coli by Gramreaction

Total % TotalTreatment protein cellular

(mg)a protein

Control 0.178 3.0CV 0.184 3.0CV-I 0.210 3.5CV-TPt 0.232 3.9Control decolorized with alcohol 2.040 34.0CV decolorized with alcohol 2.021 34.0CV-I decolorized with alcohol 2.130 36.0CV-TPt decolorized with alcohol 2.000 30.0

a Data derived from 10 ml of a culture of an opticaldensity of 0.630 with a total dry weight of 11.4 mg.Figures represent the average of three protein estima-tions for each sample by the Markwell et al. method(21).

CV-TPt is formed which can be detected byelectron microscopy. This precipitate is spreadthroughout the cell substance and occupies asubstantial volume of the cell; very little isremoved by ethanol treatment (the 14% differ-ence [see Results] is most probably due tosurface-bound CV-TPt). Our conception ofthese cells after ethanol treatment is dia-grammed in Fig. 9.At least three explanations could explain the

retention of the complex within B. slubtilis dur-ing ethanol treatment. First, it is possible thatthe CV-TPt precipitate is not dissolved by theethanol within the interior of the cell. If thiswere true, the large dye aggregate would be toolarge to exit from the cell. Second, it is possiblethat the CV-TPt aggregate is dissolved but thatthe dissociated complex (ca. 2.50 by 1.00 nm) istoo bulky to penetrate through the wall matrix.As the ethanol is replaced by an aqueous coun-terstain (e.g., carbol fuchsin) during the nextstep of the Gram reaction, the CV-TPt wouldagain precipitate out of solution within the con-tents of the cell. The last possibility is a combi-nation of the first two; in this case, only a smallproportion of the CV-TPt complex would besolubilized by the ethanol, and this proportionwould be free to migrate from the bacteria.

TABLE 4. Amount of lipid ([3Hlglycerol labeled)extracted from E. coli by Gram reaction"

Treatment No decolorizing With decolorizingalcohol alcohol

Control 95CV 34 96CV-I 42 96CV-TPt 44 92

a All data are expressed as the percent of totalcounts (lipid) extracted from the cells.

Of the three possibilities, the second choice isthe most attractive. In vitro, CV-TPt is solublein ethanol, and it is difficult to imagine that thecellular precipitate could be different. The meth-od of decolorization used in our experimentsrelied on the fact that excess ethanol was used(see reference 9), and this should be sufficient todissolve all cellular precipitate. Therefore, pos-sibilities 1 and 3 seem untenable. The latticeconstant of dried PG is smaller than that of thehydrated form (8, 14). Ethanol should replace orreduce the content of water within the wallfabric, condense the polymeric network (23),and shrink the dimensions of the lattice constant(4, 8, 14). The end result is that the fabric of thewall should become even more impermeable tothe dye complex. The lattice constant in driedPG foils suggests that the glycan strands areseparated by 0.44 nm and that the dimer repeatalong the strand is 0.90 nm. Of course, not allstrands of PG are cross-linked by tetrapeptidestems (B. subtilis has a cross-linking efficiencyof 30%), but the limiting dimension should be0.44 nm. Indeed, disregarding the secondarypolymers, up to 24 separate sheets of PG piledon top of one another could fit into the thicknessof the B. subtilis wall (4). The planar alignmentof these sheets within the wall would also affect

TABLE 5. KDO extracted from E. coli by Gramreaction

Treatment nmol" Cellular

Control 0 0CV 240 14CV-1 250 15CV-TPt 200 12Control decolorized by alcohol 660' 41CV decolorized by alcohol ND"CV-I decolorized by alcohol NDCV-TPt decolorized by alcohol ND

" Data derived from 10 ml of a culture with anoptical density of 0.630 with a total dry weight of 11.4mg. Figures represent the average of three KDOestimations for each sample.

b Based on a total of 1,624.5 nmol of KDO whichwas found in 11.4 mg of dry weight culture.

' Although alcohol extracted a high proportion ofKDO, this amount is only 40% of the total availableKDO in the culture. Electron microscopy showed thatethanol produced large aggregates of cells which con-tained trapped OM. This membrane was not cell-bound but was difficult to remove from the aggregates.Consequently, it did not contribute to the KDO esti-mations of the extract. We believe the observed dis-crepancy between the extract and available KDO isdue to this nonspecific binding to the cellular aggre-gates.

d ND, Not determinable because of spectral inter-ference from chemically bound crystal violet.

VOL. 156, 1983


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/


I 2

66 0K-m

.OmpF--- C

----A

.-%"w1

450 mm A

34 7 - 3..i.

24-0 -

18.4-

14-3-

FIG. 8. SDS-PAGE of the ethanol decolorizationextract of E. coli. The major polypeptide bands arelabeled in lane 2. Lane 1 contains molecular weightmarkers. All bands are smeared and lightly stainedbecause of the ionic effect of the crystal violet.

the permeability characteristics and further in-hibit the dye complex from migrating out of thecells.The molecular architecture of the E. coli wall

is different than that of B. subtilis. Instead of anamorphous matrix of interwoven peptidoglycanand teichoic (or teichuronic) acid strands, the E.coli wall is partitioned into two separate entities,the PG (which is most probably a monolayer [7])and the OM (an asymmetric LPS-phospholipidbilayer which is interspersed with protein, themajority of which consists of three or four majorpolypeptide species [4, 11, 15]). The OM ischemically linked to the PG (murein) layer bylipoprotein (through covalent interaction withthe diaminopimelic acid of the PG peptide stem)and matrix (or porin) protein (through magne-sium salt bridging) (4, 11).

Clearly, the Gram reaction affects these wallsin a profoundly different manner than the gram-positive variety. It is the OM which is immedi-ately affected by the ethanol. This is one of theprime differences between the two systems. Thismembrane eventually disappears from the cellsurface to reside permanently in the externalmilieu as membrane sheets or vesicles afterrehydration. The resulting cell becomes leaky atthis stage, and exudes the CV-TPt complex andsome cellular constituents (protein and nucleicacid can be detected by their UV spectra).Eventually, in some cells, small holes occur in

the PG and PM through which particulate cellu-lar debris and CV-TPt are seen to escape.Two general mechanisms of membrane inter-

action with ethanol can be assumed. First, itshould intercalate into membrane with concomi-tant disturbance of native membrane structure.Second, the polar hydroxyl group shouldinteract with the hydrophilic head groups ofmembrane lipids resulting in a disruption of theirnormal ionic affinity for small ions, such asMg2+ or Ca2 , or charged molecules such asproteins or water. The latter possibility wouldperturb the boundary water of the membrane bythe dehydration effect of ethanol. Althoughmembrane fluidity does not seem to be affected,a variety of complex alterations within and with-out the membrane are possible (12, 16, 17, 18).Our perceptual view of the OM of E. coli is

that it is a more rigid membrane than the PM,that it is physically supported by chemical link-age to the PG, and that, consequently, it is moreable to withstand environmental stress than thePM (4, 11). This presents a conundrum; why isthe PM not disturbed to a greater degree than theOM? It is our contention that the PM is per-turbed; after all, it is not very different from theB. subtilis PM, and our B. subtilis results indi-cated that intracellular CV-TPt breached the PMbut was trapped by the wall. We believe that theE. coli PM is disrupted in a similar manner butthat the sheets and vesicles thus formed are toolarge to penetrate through the PG layer. In fact,PM disruption during the decolorization step

FIG. 9. Artist's rendering of a B. subtilis cell afterthe ethanol decolorization step of the Gram reaction.The cell wall (W) remains intact. Although in thisdrawing and in the electron microscopy (Fig. lc and2a) the PM appears to be intact, it is more probablethat the ethanol has disrupted this membrane withconcomitant release of some of the CV-TPt complexesinto the periplasm where it is trapped by W. Theunlabeled arrows point towards intracellular and wall-bound CV-TPt. The intracellular CV-TPt is embeddedin the cell substance (CS) matrix. No attempt has beenmade to accurately define the dimensional aspects ofthis cell, and the W, PM, and periplasmic space havebeen exaggerated to clarify the staining results.

J. BACTERIOL.


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/


' ) (c)

FIG. 10. Artist's rendering of the sequence ofevents that occur to E. coli during the ethanol decolor-ization step of the Gram stain. (a) The OM is sloughingoff from the cell, whereas the PG and PM remainintact. As explained in the discussion of the text, theintact PM may be an artifact of electron microscopy.The CV-TPt complex is the electron dense precipitatethat is scattered throughout the internal cell substance(CS). (b) Most of the OM has been removed (only afew fragments remain) and much of the CV-TPt hasbeen leached from the CS. (c) A small localized holehas developed (usually at one pole) in the PG and PMthrough which the remaining CV-TPt and some of theCS is passing. The dimensional aspects of these draw-ings are by no means accurate, and certain structures(e.g., the periplasmic space) have been exaggeratedfor the clarity of the image. Like all artistic representa-tions and model systems, we have no doubt simpli-fied the series of events. As our tools for decipheringmolecular events at the subcellular level become moreacute, we expect more definitive explanations to beobtained.

would leave the PG as the only remaining obsta-cle to the removal of intracellular CV-TPt. Sincethis layer is most probably a monolayer (7),since the PG is only 20 to 30% cross-linked (10,22, 30), and since discontinuities within themonolayer must exist to accommodate lyso-zyme and autolysin digestion (4, 13), this layer

could not present the same sort of barrier to thedye complex as the B. subtilis wall. Once theethanol-treated bacteria were reequilibratedback to aqueous conditions, the fragments ofPM would reanneal to form an encompassingmembrane. By electron microscopy, this mem-brane would resemble the original PM in shapeand form.The observation that some cells of E. coli

develop small localized lesions within the PGand PM during ethanol treatment is of interest.There is the possibility that these are distinctsites of autolytic activity within the PG fabric.Our subjective analyses (no statistics were at-tempted) by electron microscopy suggested thatthese occurred most frequently at one polar endof the cell (the newly synthesized end?). Theydid not develop until the OM had been removed(a stringent control?). Ethanol decreases theextent of cross-linkage in the PG causing eventu-al lysis (16, 17). The reduction of cross linkage isdue to a displacement and concomitant foldingof the glycopeptide carboxypeptidase andtranspeptidase (16) which are membrane bound(19, 24, 31). Since ethanol has disrupted the E.coli membranes during the decolorization step,it is possible that the localized lesions seen in thePG are the end result of carboxypeptidases andtranspeptidases displaced from the membrane.If a hole did occur, then the underlying PMcould bleb and escape, exposing the cell interior.A model system for the decolorization processin E. coli is presented in Fig. 10.

In summary, using electron microscopy andTPt (which is apparently chemically analogousto the iodide of Gram's iodine [9]), we havefollowed the stain complex of the Gram reactionand related it to the ultrastructural alterationsmanifest within one gram-positive and one gram-negative bacterium. These observations havebeen related to biochemical changes during thestaining response. The difference between thetwo cell types appears to reside solely in thefundamental molecular dissimilarities withintheir wall fabrics. Milton Salton knew the an-swer all along (26)!

ACKNOWLEDGMENTS

This work is dedicated to M. Salton, Department of Micro-biology, New York University School of Medicine, for hiscontribution to our fundamental knowledge of bacterial wallsand because he knew the answer all along. This work wassupported by a Natural Science and Engineering ResearchCouncil of Canada (NSERC) operating grant to T.J.B. ThePhilips 400T-STEM-EDS system used in this study is aNSERC regional facility which was funded by NSERC andwhich is maintained by the University of Guelph.We are indebted to R. Humphrey for the EDS analyses, the

STEM dot maps, and the atomic absorption spectroscopy, toR. Harris for the thin sections, to G. Ferris for his perserver-ance with the SDS-PAGE analyses, and to C. Forsberg forsuggesting the [3H]glycerol experiment (all are with the De-partment of Microbiology, University of Guelph).

VOL. 156, 1983


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/


LITERATURE CITED

1. Bartholomew, J. W., and T. Mittwer. 1952. The Gramstain. Bacteriol. Rev. 16:1-29.

2. Benians, T. H. C. 1920. A further investigation into theprinciples underlying Gram's stain with special referenceto the bacterial cell membrane. J. Pathol. Bacteriol.23:401-412.

3. Bertani, G. 1951. Studies on lysogenesis. I. The mode ofphage liberation by lysogenic Escherichia coli. J. Bacteri-ol. 62:293-300.

4. Beveridge, T. J. 1981. Ultrastructure, chemistry and func-tion of the bacterial wall. Int. Rev. Cytol. 72:229-317.

5. Beveridge, T. J., and R. G. E. Murray. 1979. How thick isthe Bacillus subtilis cell wall? Curr. Microbiol. 2:1-4.

6. Beveridge, T. J., and R. G. E. Murray. 1980. Sites ofmetal deposition in the cell wall of Bacillus subtilis. J.Bacteriol. 141:876-887.

7. Braun, V., H. Gnirke, U. Henning, and K. Rehn. 1973.Model for shape-maintaining layer of the Escherichia colicell envelope. J. Bacteriol. 114:1264-1270.

8. Burge, R. E., A. G. Fowler, and D. A. Reaveley. 1977.Structure of the peptidoglycan of bacterial cell walls. I. J.Mol. Biol. 117:927-953.

9. Davies, J. A., G. K. Anderson, T. J. Beveridge, and H. C.Clark. 1983. Chemical mechanism of the Gram stain andsynthesis of a new electron-opaque marker for electronmicroscopy which replaces the iodine mordant of thestain. J. Bacteriol. 156:837-845.

10. dePedro, M. A., and U. Schwarz. 1981. Heterogeneity ofnewly inserted and pre-existing murein in the sacculus ofEscherichia coli. Proc. Natl. Acad. Sci. U.S.A. 78:5856-5860.

11. DiRienzo, J. M., K. Nakamura, and M. Inouye. 1978. Theouter membrane proteins of Gram-negative bacteria: bio-synthesis, assembly, and functions. Annu. Rev. Biochem.47:481-532.

12. Elworthy, P. H., and D. S. McIntosh. 1961. Micelle forma-tion by lecithin in some aliphatic alcohols. J. Pharm.Pharmacol. 13:663-669.

13. Formanek, H. 1978. A three-dimensional model of thedigestion of peptidoglycan by lysozyme. Biophys. Struct.Mech. 4:1-14.

14. Formanek, H., S. Formanek, and H. Wawra. 1974. Athree-dimensional atomic model of the murein layer ofbacteria. Eur. J. Biochem. 46:279-294.

15. Funahara, Y., and H. Nikaido. 1980. Asymmetric localiza-tion of lipopolysaccharides on the outer membrane ofSalmonella typhimurium. J. Bacteriol. 141:1463-1465.

16. Ingram, L. 0. 1981. Mechanism of lysis of Escherichiacoli and other chaotropic agents. J. Bacteriol. 146:331-336.

17. Ingram, L. O., and N. S. Vreeland. 1980. Differential

J. BACTERIOL.

effects of ethanol and hexanol on the Escherichia coli cellenvelope. J. Bacteriol. 144:481-488.

18. Jain, M. K., J. Glaser, A. Upreti, and G. C. Upreti. 1978.Intrinsic perturbing ability of alkanols in lipid bilayers.Biochim. Biophys. Acta 509:1-8.

19. Kozarich, J. W., and J. L. Strominger. 1978. A membraneenzyme from Staphylococcus aureus which catalyzestranspeptidase, carboxypeptidase, and penicillinase activ-ities. J. Biol. Chem. 253:1272-1278.

20. Laemmli, U. K. 1970. Cleavage of structural proteinsduring the assembly of the bead of bacteriophage T4.Nature (London) 227:680-685.

21. Markwell, M. A. K., S. M. Haas, L. L. Birber, and H. E.Tolbert. 1978. A modification of the Lowry procedure tosimplify protein determination in membrane and lipopro-tein samples. Anal. Biochem. 87:206-210.

22. Olijhock, A. J. M., S. Klencke, E. Pas, N. Nanninga, andU. Schwarz. 1982. Volume growth, murein synthesis, andmurein cross-linkage during the division cycle of Esche-richia coli PA3092. J. Bacteriol. 152:1248-1254.

23. Ou, L.-T., and R. E. Marquis. 1972. Coccal cell-wallcompactness and the swelling action of denaturants. Can.J. Microbiol. 18:623-629.

24. Rogers, H. J. 1979. Biogenesis of the cell wall in bacterialmorphogenesis. Adv. Microb. Physiol. 19:1-62.

25. Rogers, H. J., J. B. Ward, and I. D. J. Burdett. 1978.Structure and growth of the walls of gram-positive bacte-ria. Symp. Soc. Gen. Microbiol. 28:139-176.

26. Salton, M. R. J. 1963. The relationship between the na-ture of the cell wall and the Gram stain. J. Gen. Microbiol.30:223-235.

27. Sargent, M. G. 1973. Membrane synthesis in synchronouscultures of Bacillus subtilis 168. J. Bacteriol. 116:397-409.

28. Scherrer, R. 1963. Cell structure and quantitative Gramstain of Bac illus megaterium. J. Gen. Microbiol. 31:135-145.

29. Smit, J., Y. Kamio, and H. Nikaido. 1975. Outer mem-brane of Salmonella typhimurium: chemical analysis andfreeze-fracture studies with lipopolysaccharide mutants.J. Bacteriol. 124:942-958.

30. Takabe, I. 1965. Extent of cross-linkage in the mureinsacculus of Escherichia coli B cell wall. Biochim.Biophys. Acta 101:124-126.

31. Tamura, T., Y. Imae, and J. L. Strominger. 1976. Purifi-cation to homogeneity and properties of two D-alaninecarboxypeptidase I activities from E. coli. J. Biol. Chem.251:414-423.

32. Webb, M. 1948. The action of lysozyme on heat killedgram positive microrganisms. J. Gen. Microbiol. 2:260-274.

33. Weissback, A., and J. Hurwitz. 1959. The formation of 2-keto-3-deoxyheptonic acid in extracts of Escherichia coliB. J. Gen. Microbiol. 254:705-709.


http://jb.asm.org/

Dow

nloaded from

http://jb.asm.org/

cellular responses ofbacillus subtilis …cellular response to th gram stain 847 treated with...

Documents