Binding and killing of bacteria by bismuth subsalicylate.
ABSTRACT Bismuth subsalicylate (BSS) is a compound without significant aqueous solubility that is widely used for the treatment of gastrointestinal disorders. BSS was able to bind bacteria of diverse species, and these bound bacteria were subsequently killed. A 4-log10 reduction of viable bacteria occurred within 4 h after a 10 mM aqueous suspension of BSS was inoculated with 2 x 10(6) Escherichia coli cells per ml. Binding and killing were dependent on the levels of inoculated bacteria, and significant binding but little killing of the exposed bacteria occurred at an inoculum level of 2 x 10(9) E. coli per ml. Intracellular ATP decreased rapidly after exposure of E. coli to 10 mM BSS and, after 30 min, was only 1% of the original level. Extracellular ATP increased after exposure to BSS, but the accumulation of extracellular ATP was not sufficient to account for the loss of intracellular ATP. The killing of bacteria exposed to BSS may have been due to cessation of ATP synthesis or a loss of membrane integrity. Bactericidal activity of BSS was also investigated in a simulated gastric juice at pH 3. Killing of E. coli at this pH was much more rapid than at pH 7 and was apparently due to salicylate released by the conversion of BSS to bismuth oxychloride. It is proposed that the binding and killing observed for BSS contribute to the efficacy of this compound against gastrointestinal infections such as traveler's diarrhea.
- SourceAvailable from: Todd A Houston
Article: 2-Propynyl 2-hydroxy-benzoate.[Show abstract] [Hide abstract]
ABSTRACT: The title compound, C(10)H(8)O(3), has been synthesized as part of our investigations into the generation of new anti-bacterial agents and serves as a building block for the synthesis of compound libraries. The compound crystallizes with two independent mol-ecules in the asymmetric unit. The transoid propynyl ester groups are coplanar with the 2-hydroxy-benzoate group with maximum deviations of -0.3507 (3) and 0.1591 (3) Å for the terminal carbons, with intra-molecular O-H⋯O hydrogen bonding providing rigidity to the structure and ensuring that the reactivity of the alkyne is not compromised by steric factors. The propynyl group forms inter-molecular C-H⋯O inter-actions with the phenolic O atom. Supra-molecular chains along the b axis are found for both mol-ecules with links by weak O-H⋯O inter-molecular inter-actions in the first independent mol-ecule and C-H⋯O inter-actions in the second.Acta Crystallographica Section E Structure Reports Online 01/2009; 66(Pt 1):o226-7. · 0.35 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Adherence ofEscherichia coli isfacilitated byfimbriae andseveral outermembraneproteins (OMPs). Hypertonic conditions, salicylate, andMarmutations areknowntoreduce OmpFexpression. We speculated thatOMPsinvolved inexport orassembly offimbrial subunits might besimilarly affected. Toexplore this hypothesis, E.coli expressing P,type1,S,colonization factor antigen I(CFA/I), orCFA/II fimbriae wasgrown inthepresence ofsalicylate, bismuth salts, NaCl, andnonfermented sugars. Tetracycline-resistant clones were derived fromseveral P-fimbriated strains. Thebacteria weretested fortheability toagglutinate erythrocytes, yeast cells, anda-D-Gal(-4)-,I-D-Gal-bonded latex (Gal-Gal) beadsandwereexamined forfimbriae byelectron microscopy. Hyperosmolar conditions decreased fimbrial expression forallstrains. Expression ofPfimbriae bypyelonephritic strains, allofwhichwereOmpF+,wasreversibly repressed bysalicylate andbismuth salts. CFAstrains weresimilarly affected. Tetracycline-resistant P-fimbriated strains wereOmpFdeficient, were unable toagglutinate erythrocytes andGal-Gal beads, andlacked fimbriae asobserved byelectron microscopy. Strains withplasmid-encoded P-fimbrial genesdidnotdemonstrate OmpFonpolyacrylamide gelelectro- phoresis profiles andwerenotaffected bysalicylate. Thetype1-fimbriated phenotype wasnotaffected by salicylate orbismuth unless thestrains alsoexpressed Pfimbriae. S-fimbriated strains werenotaffected. The mechanism bywhichsalicylates, bismuth salts, andtetracycline resistance inhibit ormodulate theexpression ofPfimbriae maybemediated through OmpFandother OMPs.
- BMJ Clinical Research 01/1996; 311(7021):1660-1. · 14.09 Impact Factor
Vol. 33, No. 12
ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Dec. 1989, p. 2075-2082
Copyright © 1989, American Society for Microbiology
Binding and Killing of Bacteria by Bismuth Subsalicylate
THOMAS E. SOX* AND CHRISTY A. OLSON
Sharon Woods Technical Center, Procter and Gamble Company, Cincinnati, Ohio 45241
Received 8 May 1989/Accepted 5 September 1989
Bismuth subsalicylate (BSS) is a compound without significant aqueous solubility that is widely used for the
treatment of gastrointestinal disorders. BSS was able to bind bacteria of diverse species, and these bound
bacteria were subsequently killed. A 4-log1o reduction of viable bacteria occurred within 4 h after a 10 mM
aqueous suspension of BSS was inoculated with 2X106 Escherichia coli cells per ml. Binding and killing were
dependent on the levels of inoculated bacteria, and significant binding but little killing of the exposed bacteria
occurred at an inoculum level of 2 x 109 E. coli per ml. Intracellular ATP decreased rapidly after exposure of
E. coli to 10 mM BSS and, after 30 min, was only 1% of the original level. Extracellular ATP increased after
exposure to BSS, but the accumulation of extracellular ATP was not sufficient to account for the loss of
intracellular ATP. The killing of bacteria exposed to BSS may have been due to cessation of ATP synthesis or a
loss of membrane integrity. Bactericidal activity of BSS was also investigated in a simulated gastric juice at pH
3. Killing ofE. coli at this pH was much more rapid than at pH 7 and was apparently due to salicylate released
by the conversion of BSS to bismuth oxychloride. It is proposed that the binding and killing observed for BSS
contribute to the efficacy of this compound against gastrointestinal infections such as traveler's diarrhea.
Insoluble salts of bismuth have been extensively used as
antimicrobial agents. A product containing bismuth subsa-
licylate (BSS) as the active ingredient has been shown to be
clinically effective in the prevention and treatment of a
number of gastrointestinal infections, including traveler's
diarrhea (3, 4, 7) and Campylobacter pylori gastritis (16).
Despite a long history of therapeutic use of BSS and other
bismuth compounds, relatively little is known about their
antimicrobial mode of action. The bacteriostatic (14-16) and
bactericidal (7, 10) activities of these compounds have been
reported previously. Also, electron microscopy of C. pylori
present in gastric biopsies taken after oral administration of
bismuth compounds indicated both intracellular and extra-
cellular association of bismuth with C. pylori (14, 19).
Concomitantly, bacteria lost adherence to the gastric epithe-
lium, and structural degradation of the bacteria was evident.
Although the antimicrobial mode of action of bismuth com-
pounds has received relatively little attention, the actions of
other metals have been examined in more depth. Mercury
exerts antimicrobial activity as a result of covalent attach-
ment to sulfhydryl groups on essential macromolecules (5).
Several other metals such as uranium (20) and nickel (1) are
taken up in large quantities by microorganisms; these metals
do not become covalently bound to the cells, and viable cells
are not needed for this uptake to occur (20). Micrococcus
luteus and an Azotobacter sp. were reported to bind large
amounts oflead (22); this uptake had no effect on growth rate
or viability. In addition to containing a metal with antimicro-
bial properties, BSS also contains salicylate, an effective
antimicrobial agent under acidic pH conditions (9).
Our work has focused on determining the events that
occur during the killing of bacteria by BSS. This work
demonstrates the bactericidal activity ofBSS at both neutral
and acidic pHs.
(This investigation was presented in part at the 28th
Interscience Conference on Antimicrobial Agents and Che-
motherapy, Los Angeles, Calif., 23 to 26 October 1988 [T. E.
Sox and C. A. Olson, Program Abstr. 28th Intersci. Conf.
Antimicrob. Agents Chemother., abstr. no. 391, 1988].)
MATERIALS AND METHODS
BSS. BSS, in the form of a 10% aqueous slurry, was
manufactured by Procter and Gamble Co., Cincinnati, Ohio.
BSS was present as needlelike crystals of about 2 by 7 ,um.
Bacterial strains. Strain H10407, an enterotoxigenic Esch-
erichia coli strain, was obtained from R. Leunk (Procter and
Gamble). All other strains were obtained from the American
Type Culture Collection, and strain designations are pro-
vided in the text.
Preparation of simulated gastric juice. A simulated gastric
juice was prepared to mimic the composition of human
gastricjuice (11). This simulated gastricjuice contained 0.1%
pepsin (Sigma Chemical Co., St. Louis, Mo.), 0,1% porcine
gastric mucin (Sigma), 20.5 mM NaCl, and 2.7 mM KCl.
Concentrated hydrochloric acid was then added to a final
level of0.74%, and the pH was adjusted to the desired value
with aqueous sodium hydroxide. Gastric juice was made
fresh daily because preliminary experiments indicated that it
developed bactericidal activity when allowed to age for
Neutralization of antimicrobial activity. Preliminary exper-
iments indicated that dilution of inoculated BSS suspensions
into D/E neutralizing broth (Difco Laboratories, Detroit,
Mich.) neutralized the bactericidal activity of BSS. Conse-
quently, samples were diluted into D/E broth to halt bacte-
Determination of binding and killing. Bacterial cultures
were grown overnight at 37°C with shaking in tryptic soy
broth (Difco). Broth cultures were diluted in water, and A420
values were measured. Previously developed calibration
data were used to adjust absorbances of these suspensions
so that they contained the desired number of bacteria for the
binding and killing assay. BSS was added to water or
ANTIMICROB. AGENTS CHEMOTHER.
simulated gastric juice and mixed thoroughly. The pH of the
resulting suspension was adjusted as necessary, and the
suspension was warmed to 37°C prior to addition ofbacteria.
At indicated times, the solutions were mixed thoroughly
before removal of samples that were then diluted 1:10 in D/E
broth. These saniples were used to determine the total
number of viable organisms present in the suspension. An
additional sample was removed and centrifuged at 190 x g
for 5 min (1,000 rpm in a TH-4 swinging-bucket rotor, TJ-6
centrifuge; Beckman Instruments, Inc., Palo Alto, Calif.).
Preliminary experiments (data not shown) indicated that this
centrifugation did not pellet bacteria free in suspension, but
these conditions were sufficient to pellet BSS. Bacteria
remaining in the supernatant were considered unbound, and
those present in the pellet were considered bound to BSS.
Immediately after centrifugation, the supernatant was di-
luted 1:10 in D/E broth. The pellet was also suspended with
D/E broth. Samples in D/E broth were inoculated onto
150-mm-diameter plates of tryptic soy agar (Difco) with a
model C spiral plater (Spiral Systems Instruments Inc.,
Bethesda, Md.) set to deliver 40 pAl of sample. After over-
night. incubation at 37°C, plates were scanned with a Spiral
Systems laser colony counter, and Spiral Systems CASBA
bacterial enumeration software, version 1.2, was used to
determine the number of viable bacteria in the original
C. pylori ATCC 43504 was tested in the above procedure
with the following exceptions. C. pylori was inoculated on
Campy chocolatized agar (Remel Laboratories, Lenexa,
Kan.) and incubated in a microaerophiic atmosphere (Cam-
pylobacter microaerophilic system; BBL Microbiology Sys-
tems, Cockeysville, Md.) at 37°C for 5 days. Cells were
suspended in water, and absorbances were adjusted as
described above. Polysorbate 80 was found to be inhibitory
to C. pylori, and C. pylori samples were diluted in D/E broth
lacking Polysorbate 80 and lecithin. Samples were spiral
plated onto brucella agar containing 3% laked sheep blood
and 4% sheep serum and incubated in a microaerophilic
atmosphere'at 37°C for 7 days. The procedures described
above were used for quantitation of C. pylori colonies on
Effects of albumin on binding. BSS (10 mM) was sus-
pended in a 10-mg/ml solution of bovine serum albumin
(Sigma). The suspension was mixed and then heated to 37°C
in a water bath. Bacteria were added, and samples were
assayed for binding and killing as described above.
Solubilization of BSS. In some experiments, bacteria
bound to BSS were released by solubilizing the BSS with
sodium tartrate (biotech grade; Fisher Scientific Co., Pitts-
burgh, Pa.). A volume of filter-sterilized 1 M sodium tartrate,
pH 7, was added to an equal volume of 10 mM BSS
suspension. This resulted in the solubilization of the BSS in
approximately 1 min at 37°C.
Measurement of ATP. Intracellular and extracellular ATP
levels were measured to determine the effects of BSS on cell
physiology and integrity (17, 21). Samples of bacteria ex-
posed to BSS were treated with sodium tartrate as described
above to solubilize the BSS. A 50-,J sample was then mixed
with 50 ,u of 20 mM Tris hydrochloride (pH 7.7) and 100ILI
of Microlyse (LKB-Wallac Instruments, Helsinki, Finland).
A 100-,ul sample of lysate was added to 100 plA of ATP assay
mix (Sigma), and light production was quantitated with a
photometer (SAI Instruments, San Diego, Calif.). ATP stan-
dard addition experiments were performed to determine the
degree of quenching caused by sodium tartrate and tartrate-
solubilized BSS. Results were corrected for quenching to
determine the actual amounts ofATP present in the samples.
Extracellular ATP was determined by filtering the tartrate-
solubilized cell suspension through a 0.2-,um Dynagard filter
(Microgon, Inc., Laguna Hills, Calif.) before the assay.
Intracellular ATP was calculated by subtracting the extra-
cellular ATP from the total ATP found in the sample.
Salicylate determination. Soluble salicylate levels were
determined by passing samples through 0.2-pum Dynagard
filters and assaying the filtrates with a salicylate diagnostic
E. coli binding to BSS. E. coli ATCC 10536 attached
rapidly to BSS suspended in water (pH 7). At inoculum
levels of 106 and 5 x 107 CFU/ml, binding was very rapid and
was essentially complete before the first set of data was
collected (Fig. 1A and B). At a greater inoculum level, most
binding occurred within 30 min, but the number ofunbound
bacteria continued to decrease for an additional 90 min (Fig.
1C). At some ofthe time points, the sum ofviable bound and
unbound bacteria was slightly less than the total viable
population observed. Some killing ofbound bacteria by BSS
continued during the centrifugation procedure prior to the
addition ofD/E broth, and this cell death during the centrif-
ugation procedure accounted for the observed lack of addi-
The rapid binding of bacteria was followed by a slower
loss of viability of the bound cells. The extent of killingwas
highly dependent on the inoculum level, as demonstrated by
the different amounts of killing observed for the three
inoculum levels shown in Fig. 1. At high inoculum levels (2
x 109 CFU/ml), there was no detectable killing during the
first 5 h ofexposure despite most ofthe bacteriabeingbound
to BSS. However, some killing of the exposed cells did
occur with longer exposures. Exposure of 2.2 x 109 CFU/ml
to BSS for 24 h resulted in a 90% loss in viability, compared
with a 13% loss of viability in control bacteria exposed only
to water (data not shown). In contrast, at an inoculum level
of approximately 106 CFU/ml, a>4-log1oreduction in viable
bacteria occurred within 4 h (Fig. 1A).
Since bacteria bound to BSS particles, it was possiblethat
the bacterial colonies obtained in this work originated from
BSS particles with one or more bacteria attached, rather
than having developed from individual bacterial cells. This
would have resulted in erroneously low values for the
number of viable bacteria present in samples exposed to
BSS. An alternative experimental approach was to add
samples of suspensions of bacteria in BSS to 1 M sodium
tartrate. Sodium tartrate solubilized the BSS, with concom-
itant release of the bound bacteria. Preliminary studies
indicated that the resulting soluble bismuth solution did not
have bactericidal activity for E. coli. Plating of these tar-
trate-solubilized samples yielded recoveries of viable bacte-
ria very similar to those obtained in D/E broth (Table 1).
Therefore, dilution of BSS-containing samples into D/E
broth yielded accurate values for viable bacteria.
We examined the ability of D/E broth and its antimicrobial
neutralizing components to release bacteria bound to BSS
(Table 1). Addition of E. coli to 10 mM BSS in water resulted
in only 0.1 to 1.0% of the bacteria remaining in suspension
after low-speed centrifugation (second column). D/E broth,
D/E broth components, or water (third column) was used to
suspend the pellets of BSS and bacteria (fourth column).
These suspensions were centrifuged again to determine the
SOX AND OLSON
BINDING AND KILLING BY BISMUTH SUBSALICYLATE
. 20- Total
Minutes of Exposure to lOmM BSS (pH7) at 370C
Minutes of Exposure to lOmM BSS (pH7) at 370C
FIG. 1. Effects of inoculum level on binding and killing ofE. coli
ATCC 10536by a 10 mM aqueous suspension of BSS. Limit of
detection ofprocedureis indicated bythe dashed line. Inoculum
levels were 1.73 x 106 (A), 8.35 x 107 (B), and 2.17 x 109 (C)
ability of the various components to release bound bacteria.
The percentages of bacteria that remained in suspension
after centrifugation, indicating release from BSS, are shown
in parentheses in the fifth column. Exposure to D/E broth
released about 44% of the bound bacteria. The combination
of lecithin and Polysorbate 80 was largely responsible for
release of the bound bacteria; these two components re-
\\leasedabout 38% ofthe bound bacteria. Lecithin alone could
not be tested because of its immiscibility in water in the
absence ofan emulsifying agent. The nutrient components of
D/E broth (tryptone and yeast extract) also contributed to
\t'the release. A solution of these materials at the concentra-
tions found in D/E broth resulted in 32% release of the bound
The effects of pretreatment of BSS with 10 mg of bovine
serum albumin per ml on binding and killing of E. coliwere
\examined (Fig. 2). Albumin pretreatment greatly decreased
time needed for killing of the exposed bacteria. However,
exposure to albumin-pretreated BSS eventually resulted in
-killing of all of the exposed bacteria.
A- Not Bound
the extent of binding of bacteria and increased the length of
Binding ofother bacteria to BSS. An enterotoxigenicstUin
ofE. coli (H10407) showed binding and killing similar to that
observed for E. coli ATCC 10536 (data not shown). Binding
of Salmonella typhimurium and Staphylococcus aureus to
BSS and their subsequent killing were also examined (Fig.
3). These strains showed binding to BSS similar to that
Minutes of Exposure to ±OmM BSS (pH7) at 370C
VOL. 33, 1989
ANTIMICROB. AGENTS CHEMOTHER.
TABLE 1. Effects of D/E broth and components on release of bacteria bound to BSS'
1.0 M sodium tartrate
0.5% Polysorbate 80
37.9 mM sodium thiosulfate
8.8 mM sodium thioglycolate
0.7% Lecithin, 0.5% Polysorbate 80
D/E broth without neutralizers
aBacteria (E. coliATCC 10536) were added to 10mM BSS suspended in water, pH 7. Samples were removed immediately and diluted in D/E broth to determine
initial levels of viable bacteria. The remaining suspensions were centrifuged. The supernatants were removed and assayed for viable bacteria. The pellets were
suspended to the original sample volume with the indicated media, and these were assayed for viable bacteria. The resuspended pellets were centrifuged again
to determine the extent ofrelease ofbound bacteria. Supernatants and pellets (after resuspension in D/E broth) were assayed to determine the number of bacteria
released during treatment.
bPercentages of total recovered bacteria found in the supernatant.
c-, 1.0 M sodium tartrate solubilized BSS; no material was present in pellet after centrifugation.
described above for E. coli. Salmonella typhimurium was
killed more rapidly than E. coli; the increased sensitivity of
Salmonella typhimurium was consistently observed in mul-
tiple experiments. Relatively little killing of Staphylococcus
aureus occurred despite substantial binding to BSS. C.
pylori ATCC 43504 was extremely sensitive to BSS, and the
proportions of bound and unbound organisms could not be
determined due to the rapid killing. The survival of equal
quantities of C. pylori added to a BSS suspension or a water
control is presented in Fig. 4. No viable organisms were
detected 15 min after exposure of 2.3 x 106 CFU/ml to BSS.
BSS effects on ATP levels. ATP levels were determined in
E. coli ATCC 10536 exposed to 10 mM BSS in water (pH 7)
(Fig. 5). Total ATP levels decreased to about 10% of the
initial levels after 30 min. Examination of extracellular ATP
levels over time indicated a rapid increase in extracellular
ATP after exposure to BSS, followed by a gradual decrease
in extracellular ATP. After 30 min of exposure, most of the
remaining ATP was extracellular. Other experiments (data
not shown) indicated that exogenous ATP was fairly stable
under these conditions, with 49% of the added ATP remain-
ing after 3 h. The rapid decrease in total ATP after exposure
to BSS was not due to instability of extracellular ATP.
Antimicrobial activity of BSS in simulated gastric juice.
Addition ofE. coli ATCC 10536 to 10 mM BSS suspended in
simulated gastric juice (pH 3) resulted in much more rapid
killing than was observed for BSS in water at pH 7 (Fig. 6).
Since essentially all of the added bacteria were killed within
a few minutes after addition to gastric juice containing BSS,
it was not possible to assess the proportions of bound and
unbound bacteria with the binding assay described above.
Control experiments indicated that freshly prepared gastric
juice at pH 3 had slow bactericidal activity against E. coli,
and after 2 h at 37°C 57% of the added bacteria remained
We examined the release of salicylate from a 10 mM BSS
suspension in simulated gastric juice (pH 3) that was inocu-
lated with 106 CFU of E. coli ATCC 10536 per ml (Fig. 7).
Salicylate release was rapid and continued for at least 1 h. A
concentration of 1.09 mM salicylate was present after 5 min
of exposure; approximately 90% of the exposed bacteria
were dead at this time. A subsequent experiment was
performed in which bacteria were exposed to 1.09 mM
salicylate in gastric juice to determine whether the levels of
salicylate released in gastricjuice could account for the rapid
Total, 10mM BSS
\ ^A - Not Bound, 10mM BSS
El- Bound, 10mM BSS
Minutes of Exposure at 37°C
FIG. 2. Effects of pretreatment of 10 mM BSS with 10 mg of
bovine serum albumin (BSA) per ml on binding and killing of E. coli
ATCC 10536 by BSS.
SOX AND OLSON
BINDING AND KILLING BY BISMUTH SUBSALICYLATE
\ A- Not Bound
Minutes of Exposure to lOmM BSS (pH7) at 370C
Minutes of Exposure to lOmM BSS (pH7) at 370C
FIG. 3. Binding and killing of Salmonella typhimurium and Staphylococcus aureus by a 10 mM aqueous suspension of BSS.
killing observed (Fig. 8). The killing by salicylate alone at pH
3 was similar to that observed for bacteria exposed to 10 mM
BSS in gastric juice at pH 3. Controls of 1.09 mM salicylate
in water (pH 7) or gastric juice (pH 7) did not show any
The antimicrobial mode of action of BSS has received
little attention. The effects of BSS we have described,
including its ability to bind several species of bacteria and
the subsequent killing of the bound organisms, may contrib-
ute to the observed in vivo efficacy of this compound. Our
work utilized bacteria in stationary growth phase, and we did
not examine the responses of bacteria in other growth
phases. Extensive literature exists on the binding of micro-
organisms to other surfaces. Binding may occur as a result of
binding to specific sites (lectinlike interactions) or nonspe-
cific processes such as electrostatic and hydrophobic inter-
actions, hydrogen bonds, and van der Waals interactions
(12). The range of organisms observed to bind to BSS in this
work suggests that nonspecific binding is occurring. Pre-
treatment ofBSS with a protein (albumin) decreased the rate
of binding and killing, but complete killing of the exposed E.
coli was still observed. This suggests that albumin and
bacteria compete to some extent for the same binding sites,
and albumin may sterically inhibit binding of bacteria (13).
Polysorbate 80 is reported (18) to be effective in desorbing
bacteria attached to glass surfaces by hydrophobic interac-
tions. Polysorbate 80 was ineffective in releasing E. coli
bound to BSS, suggesting that hydrophobic interactions
were not responsible for binding. It is unclear whether
binding is essential for bactericidal activity of BSS. In the
presence of0.1% albumin (Fig. 2), all ofthe exposed bacteria
were killed despite the incomplete binding observed. In the
absence of albumin, more extensive binding occurred and
the rate ofkilling was much greater. Hence, it appears that
bound bacteria are killed more rapidly than unbound bacte-
Although binding of microorganisms to surfaces has been
extensively investigated, the rapid killing of bound bacteria
we observed has not been reported previously. This may be
a unique instance of bacteria binding to the surface of an
antimicrobial but insoluble material. The killing of bacteria
we observed at neutral pH is not due to the salicylate
moiety, since binding and killing have also been observed
with other insoluble bismuth compounds that do not have
antimicrobial counterions (unpublished results). Also, salic-
ylate lacked detectable bactericidal activity at pH 7. Bismuth
therefore appears to be responsible for the bactericidal
activity against the bound organisms.
VOL. 33, 1989