Journal of Antimicrobial Chemotherapy (1996) 37, 911-918
Antimicrobial activities in vitro and in vivo of transition element complexes
containing gold(I) and osmium(VI)
Amanda M. Elsome*, Jeremey M. T. Hamilton-Miller*, William Bmmfitt* and
William C. Noble'
'Johnson Matthey Technology Centre, Biomedical Department, Blonnts Court, Sonning
Common, Reading, RG4 9NH; bRoyal Free Hospital School of Medicine, Department of
Medical Microbiology, Pond Street, London, NW3 2QG; 'Department of Microbial
Diseases, St. Thomas Hospital UMDS, St. John's Institute of Dermatology, Lambeth
Palace Road, London, SE1 7EH, UK
Metal compounds have been used as antibacterial agents for centuries. The in-vitro
activity of two metal containing complexes, one gold, the other osmium, was
investigated using a panel of clinically isolated bacteria and Candida aJbicans. Twenty
strains of each organism were used and MIC and MBC values determined using the
agar plate dilution method. Protein binding effects on the activity of the compounds
were also investigated using media supplemented with 5% human blood. In-vivo
activity of the two compounds was subsequently determined in a hairless-obese
mouse skin-surface activity model. Both compounds were highly active against the
Gram-positive organisms and Candida albicans in vitro. The gold compound had
some Gram-negative activity but the osmium complex was inactive against these
organisms. Both were extensively protein bound. In the in-vivo experiment the gold
compound achieved a 2-3 log reduction for all the test organisms and was at least
as good as or superior to mupirocin in its eradication rate. The osmium compound
was inactive.
Introduction
The antibacterial properties of metals have been recognised for centuries and have
represented some of the most fundamental breakthroughs in medicinal history.
Probably the first laboratory antibacterial experiments involved a metal compound,
namely Koch's investigation into the activity of mercuric chloride on anthrax spores
(Koch, 1890); and the introduction of the metalloid organoarsenical compound
Salvarsan™ in 1912 for the treatment of syphilis is considered to be the first example
of the use of a synthetic therapeutic agent.
The medicinal use of mercury can be traced back to at least the 10th century.
Inorganic mercury compounds were used to treat skin diseases and syphilis (O'Shea,
1990), and more recently organo-mercurial compounds have been used as antiseptics
and biocides. Organic salts of copper and zinc also have mild antiseptic properties
although more commonly they are used in the treatment of fungal infections such as
athlete's foot. Other metals reported to exert a toxic effect on bacteria include the
911
0305-7453/96/050911 + 08 S12.00/0
£, 1996 The British Society for Antimicrobial Chemotherapy
912
A. M. Elsome et al.
precious metals platinum and silver (Ferguson, Murray & Lancy, 1979; Thurman &
Gerba, 1989). The original observations that platinum-ammine complexes had
antibacterial properties (Rosenberg et al., 1967) led to the discovery of their anti-tumour
properties and the development of the highly successful platinum anti-cancer drugs
cisplatin and carboplatin. The antimicrobial properties of silver have been exploited for
centuries; silver lined containers were once used to store water and the introduction of
topical silver nitrate for the treatment of infantile blindness in 1884 was seen as a
significant medical breakthrough.
The concept of gold as an antibacterial therapy can be traced back to the 8th century
(Sadler, 1976). The first scientific examination of gold compounds as antibacterial
agents came in 1890 with the discovery that gold cyanide was inhibitory to 'tubercle
bacilli' in vitro but not in vivo (Koch, 1890). Indeed, the introduction of gold compounds
for the treatment of rheumatoid arthritis was based on their antibacterial activity as it
was assumed that bacteria were responsible for this condition (Lande, 1927). Recently,
complexes of gold have been reported to have a wide range of antimicrobial activities
(Elsome et al., 1991). We report here on the in-vitro and in-vivo activities of two
compounds containing transition elements, namely a gold(I) complex JM 1796, and an
osmium(VI) compound, JM 1397.
Materials and methods
Test compounds
JM 1397 [OsO2(2,6-dimethylphenyl)2] (Longley el al., 1988) and JM 1796
[Au(SCN)(PMe3)] (Melpolder & Burmeister, 1981) were synthesised at the Johnson
Matthey Technology Centre and characterised using appropriate, standard, analytical
methods. When preparing test solutions for the in-vitro studies, acetone was used as
the compound solvent.
Mes
Me-^P—Au—SCN
JM 1397
JM 1796
Media
All media were supplied by Unipath, Basingstoke, Hants, UK.
Microbial strains and culture techniques
Strains of methicillin-resistant Staphylococcus aureus (MRSA) were taken from
countries around the world (Maple, Hamilton-Miller & Brumfitt, 1989). Other strains
tested in vitro consisted of Enterococcus faecalis, Proteus mirabilis, Pseudomonas
aeruginosa and Candida albicans that had been isolated recently from clinical material
sent to the Department of Medical Microbiology, Royal Free Hospital. These bacteria
were cultured in Nutrient Broth or on Iso-Sensitest Agar.
Antimicrobial activities of a gold(I) and osmlum(VI) complex
913
The organisms used for the in-vivo studies at the Institute of Dermatology were also
evaluated in vitro.
Microbiological methods
Minimum inhibitory concentration (MIC) values were determined by the agar plate
dilution technique (Woods & Washington, 1995), using an inoculum of lO'cfu applied
with a multi-point inoculating device (Denley, Billinghurst, UK). Plates were read after
18 h incubation at 37°C. Minimum bactericidal concentration (MBC) values were
determined by replication from MIC plates using velvet pads, onto fresh medium; the
end-point was taken as 99.9% killing. Results were summarised by calculating MIC»/9o
and MBC X/K values by cumulation and interpolation (Hamilton-Miller, 1991). Control
plates, incorporating solvent only, were included to ensure that this was not having an
inhibitory effect.
Protein binding was calculated from determinations of MICJO for 20 strains of
S. aureus in both the presence and absence of 5% human blood, thus:
% protein binding = 100 x (MICo with blood—MIC50 without blood)/MIC» with blood.
Microbial strains and culture techniques for in-vivo studies
Antibiotic-sensitive laboratory strains of S. aureus, P. mirabilis, P. aeruginosa,
E. faecalis and C. albicans were used. These were routinely grown on Cystine Lactose
Electrolyte Deficient (CLED) medium (S. aureus, P. mirabilis and P. aeruginosa), Blood
Agar Base supplemented with 7% horse blood (BAB) (5. aureus), MacConkey Agar
No. 2 (E. faecalis) or Sabouraud Dextrose Agar (C. albicans).
Animal model
Pairs of hairless-obese mice, bred in the Institute of Dermatology, were inoculated on
intact skin of the back over an area of 2 x 2 cm with cultures varying in density from
10* to 10* cfu/mL using a cotton tipped swab. Inocula were prepared by scraping
overnight growth (2 day growth for C. albicans) from an agar plate and suspending it
in Nutrient Broth. Cetomacrogol cream (BP) with or without test compound ( 1 % w/w)
was then applied to the same area at a rate of 0.01 g/cm2. In one instance commercial
2% mupirocin cream (Bactroban, SmithKline Beecham) was used. The area was then
covered with a square of sterile plastic held in place by Steridrape (3M). Mice were
allowed food and water without restraint.
After 24 h the dressing was removed and the inoculated area sampled with a sterile
cotton swab moistened in Nutrient Broth. The swab was shaken vigorously in 2 mL
Nutrient Broth for 1 min using a vortex mixer. The suspension was diluted in serial
ten-fold steps, quantitatively inoculated onto agar and incubated at 37°C. After either
24 h (bacteria) or 48 h (C. albicans) the plates were examined and colonies counted at
the most suitable dilution.
To ensure that neither compound exerted an inhibitory effect on the growth of
recovered cells, a comparative count with and without a suitable inactivator was
performed. Using 5. aureus, BAB containing 0.5% dithiothreitol was used in parallel
to plain BAB for JM 1796 and results from each system compared. A comparison of
914
A. M. Eteome et al.
Table I. Minimum inhibitory and minimum bactericidal concentrations (mg/L) for JM
1397 [Os02(2,6-dimethylphenyl)2] against a range of clinical isolates. The number of
strains tested of each organism is given in parentheses. NC indicates not calculable
MBC
MIC
Organism
ICM
IC,
Range
MRSA aureus (20)
E. faecalis (20)
P. mirabilis (20)
P. aeruginosa (20)
C. albicans (20)
0.6
0.5
NC
NC
1.4
0.9
1.4
NC
NC
1.9
0.5-1.0
0.25-2.0
>100
> 100
0.5-2.0
BO.
BC«,
Range
1.0
2.9
1.7
5.4
1.0-2.0
2-8
1.4
2.0
0.5-4.0
cell recovery for CLED and BAB was used for JM 1397, since this compound is
inactivated in the presence of serum (see 'protein binding studies').
Results
In-vitro antimicrobial activity
The solvent had no antimicrobial effect at the concentrations employed. JM 1397 was
highly active against MRSA, enterococci and C. albicans (Table I), but did not inhibit
the growth of the Gram-negative species tested. It is of interest that the compound was
cidal not only against staphylococci and C. albicans, but also for enterococci, at 4 x
the respective MIC value.
JM 1796 inhibited MRSA, enterococci and P. mirabilis at low concentrations
(MIC <, 4 mg/L), but was less active against P. aeruginosa and C. albicans. MBC values
were generally close to MIC values, except for the enterococci (Table II).
The MIC and MBC values obtained for the strains used for the in-vivo studies were
consistent with these findings. JM 1397 was highly active against both the Gram-positive
organisms and the C. albicans and inactive against the two Gram-negative bacteria. JM
1796 had the same spectrum of activity for the laboratory strains as measured for the
clinical isolates (Table III).
Table II. Minimum inhibitory and minimum bactericidal concentrations (mg/L) for JM
1796 [AU(SCN)(PMe3)J against a range of clinical isolates. The number of strains tested
of each organism is given in parentheses
MIC
Test
Organism
MRSA (20)
E. faecalis
P. mirabilis (20)
P. aeruginosa (20)
C. albicans (20)
MBC
IC»
IC«
Range
0.33
0.77
1.47
26.4
27.3
0.46
1.58
2.0
51.7
52.5
0.25-0.5
0.25-2.0
2-4
16-64
32-64
BC»
BC*.
Range
1.3
8
2.2
38.3
28.5
1.8
16
3.6
57.7
53.8
0.5-2.0
4-64
1-A
32-64
32-64
915
Antimicrobial activities of a gold(I) and osmlum(VD complex
Table HI. Minimum inhibitory and minimum bactericidal concentrations (mg/L) for JM 1397 and JM 1796
against the antibiotic sensitive laboratory strains of
the Institute of Dermatology, subsequently used for
evaluation in vivo
JM 1397
JM 1796
Test
Organism
MIC
MBC
MIC
MBC
S.
E.
P.
P.
C.
1
0.5
> 128
1
> 128
1
8
> 128
1
> 128
1
2
32
32
4
1
8
64
64
4
aureus
faecalis
aeruginosa
mirabilis
albicans
Protein binding studies
Results of experiments using 5% whole human blood showed that JM 1397 and JM
1796 were extensively bound to protein at 97% and 9 1 % respectively.
Table IV. In vivo activity of JM 1397 and JM 1796 using a murine skin surface activity
model. The results are given as the mean number of colonies recovered from paired mice
24 hours following inoculation and treatment. The study was performed in triplicate and
the result of each experiment is given, (a) indicates parallel counts using inactivator and
revealed statistically undistinguishable results
Recovered organisms (cfu/mL) from 3 replicate
experiments
Test
Organism
Experiment
1
Experiment
2
Experiment
3
Base control
JM 1397
JM 1796
Mupirocin
4.0 x 10* (a)
5.4 x 103 (a)
2.0 x 103 (a)
2.6 x 10s
3.2 x 10*
1.6 x 104
9.0 x 10*
Base control
JM 1397
JM 1796
1.2 x 10*
1.4 x 10*
4.4 x 103
4.2 x 10*
1.6 x 106
3.1 x 103
—
Base control
JM 1397
JM 1796
1.2 x 10'
2.0 x 10'
6.0 x 10*
8.2 x 104
2.0 x 103
1.6 x 10*
—
P. mirabilis
Base control
JM 1397
JM 1796
3.6 x 103
7.4 x 103
1.0 x 102
1.0 x 10*
1.7 x 10*
8.0 x 102
7.5 x 103
1.1 x 10*
1.0 x 10*
C. albicans
Base control
JM 1397
JM 1796
3.8 x 103
1.0 x 103
5.7 x 103
3.8 x 103
1.3 x 103
1.8 x IV
z
S. aureus
E. faecalis
P. aeruginosa
Treatment
8.5 x 102
916
A. M. Elsome et at.
ln-vivo antimicrobial activity
Viable counts of recovered organisms from three experiments are shown in Table IV.
Colonies recovered from paired mice are here averaged. Initial inoculation counts (i.e.
time 0 h) are not given as it is difficult to determine what proportion of the culture used
to inoculate the skin (by rolling a swab dipped in culture over the skin surface) remains
on the skin after inoculation. However, the similarity of replicate counts of recovered
organisms at 24 h gives a good indication that the initial inoculum was standard. There
was no evident difference in counts with or without inactivator. JM 1796 was the most
efficacious compound in all instances, reducing numbers of all the test organisms with
respect to the controls. JM 1397 was inactive under the conditions of this test.
Discussion
Gram-negative organisms have an outer membrane, which provides an effective
control over the uptake of toxic substances, and as a consequence are generally more
resistant to antibacterial agents (Vaara, 1992). This was certainly the case for the
two metal complexes investigated here. The Gram-positive organisms used were
more sensitive to both compounds tested whilst the Gram-negative bacteria were more
resistant, particularly to the osmium(VI) complex, JM 1397.
This specificity of activity suggests that the sensitivity of the test organisms to the test
compounds is associated with the different cell wall structures characterised by the
Gram stain. Previous studies looking at the interactions of gold(I) compounds with
biological systems suggest that the major mode of action involves the highly specific
coordination of gold(I) to thiol groups on proteins, and particularly with L-cysteine,
although other groups are thought to be involved (Smith Brown & Cappell, 1985).
Studies looking at microbially-deposited gold have shown that positively charged
residues such as RNH3+, R^N+ and coordinating ligands such as RSH and R2S are
important for gold binding (Watkins et al., 1987). The reduced activity of the gold(I)
compound against the Gram-negative organisms may be due to a lack of availability
of such groups to the complex.
The potent activity that both of the compounds demonstrated against Gram-positive
bacteria was of particular interest. The incidence of multiply antibiotic-resistant
Gram-positive cocci is a major and growing clinical concern, demanding new strategies
in antimicrobial chemotherapy.
The therapeutic efficacy of the drugs currently available for the treatment of
Gram-positive organisms is limited in the treatment of the 'problem Gram-positive
cocci'. Many are inactive against MRSA, MARSA and/or enterococci; there is the
emergence of resistant strains; intense monitoring is required during treatment; many
are unproven in severe infections; and they are expensive. Vancomycin is the drug of
choice in MRSA and MARSA infections. However, administration must be by iv
infusion and resistance to vancomycin and teicoplanin has been widely reported,
particularly amongst enterococci (Leclerq et al., 1988; Greenwood, 1989; Johnson et al.,
1990). The activity of both compounds against MRSA and enterococci in vitro indicate
the potential for metal coordination complexes to overcome the limitations experienced
with the antibiotics currently available to treat these organisms.
In general, the in-vivo results of all the test organisms showed an approximately 2-3
log decrease in count for JM 1796 but a failure to decrease with JM 1397. The log cell
Antimicrobial activities of a gold(I) and osmlum(VI) complex
917
kill for JM 1796 was similar to that of mupirocin against S. aureus and superior
to mupirocin for the other organisms (W. C. Noble, personal communication). JM
1397 showed no activity against any of the test organisms in vivo despite obvious
activity in vitro. The extent to which this compound is protein bound and hence
inactivated, in the presence of serum, is the most probable explanation for this
observation.
The hairless-obese mouse has been shown previously to be a valuable model for the
study of gene transfer between staphylococci (Naidoo & Noble, 1978, 1987; Noble,
Virani & Cree, 1992) and in this to parallel the results obtained with human skin.
Antibiotic or bacteriocin production and the resultant interaction between organisms
has also been studied using this model (Noble & Willie, 1980; Noble, 1988).
In conclusion, the data presented here both in vitro and in vivo has identified JM 1796
[Au(SCN)(PMe3)] as a metal containing complex of potential therapeutic benefit,
particularly for the topical treatment of multiply antibiotic resistant Gram-positive
bacteria.
Acknowledgements
We would like to thank Dr P. D. Savage for his skill in synthesising the test compounds,
Mrs S. Shah for her invaluable help and Dr B. R. C. Theobald for his assistance in
preparing the manuscript.
References
Elsome, A. M., Brumfitt, W., Hamilton-Miller, J. M. T., Savage, P. D., King, R. O. & Frickcr,
S. P. (1991). Antimicrobial activity and potential therapeutic use of new gold coordination
complexes. In Program and Abstracts of the Thirty-First Interscience Conference on
Anti-Microbial Agents and Chemotherapy, Chicago, 1991. Abstract 387, p. 163. American
Society for Microbiology, Washington, DC.
Ferguson, C. A., Murray, R. G. E. & Lancy, P. (1979). Effects of some platinum (IV) complexes
on cell division of Escherichia coli. Canadian Journal of Microbiology 25, 545-59.
Greenwood, D. (1989). Antibiotic resistance in enterococci. Journal of Antimicrobial
Chemotherapy 24, 631-5.
Hamilton-Miller, J. M. T. (1991). Calculating MIC». Journal of Antimicrobial Chemotherapy 27,
863-4.
Johnson, A. P., Uttley, A. H. C , Woodford, N. & George, R. C. (1990). Resistance to
vancomycin and teicoplanin: an emerging clinical problem. Clinical Microbiology Reviews 3 ,
280-91.
Koch, R. (1890). Ueber bacteriologische Forschung. Deutsche Medicinische Wochenschrift 16,
756-7.
Lande, K. (1927). Die Gunstige Beeinflussung schleichender Dauerinfekte durch Solganol.
Munchener Medizinische Wochenschrift 74, 1132-4.
Leclerq, R., Derlot, E., Duval, J. & Courvalin, P. (1988). Plasmid-mediated resistance to
vancomycin and teicoplanin in Enterococcus faecium. New England Journal of Medicine 319,
157—61.
Longley, C. J., Savage, P. D., Wilkinson, G., Hussain, B. & Hursthouse, M. B. (1988). Alkylimido
and oxo aryls of rhenium. X-ray structures of bis(tert-butylimido)dich]oro(o-tolyl)rheniuni
and bis(2,6-dimethylphenyl)dioxometal, metal = rhenium and osmium. Polyhedron 7,
1079-88.
Maple, P. A. C , Hamilton-Miller, J. M. T. & Brumfitt, W. (1989). World-wide antibiotic
resistance in methicillin-resistant Staphylococcus aureus. Lancet i, 537-40.
Melpolder, J. B. & Burmeister, J. L. (1981). Antisymbiosis and the trans-influence in gold(I)
thiocyanate complexes. Inorganic Chimica Acta-Articles 49, 115-20.
918
A. M. Elsome et al.
Naidoo, J. & Noble, W. C. (1978). Transfer of gentamicin-resistance between strains of
Staphylococcus aureus on skin. Journal of General Microbiology 107, 391-3.
Naidoo, J. & Noble, W. C. (1987). Skin as a source of transferable antibiotic resistance in
coagulase-negative staphylococci. Zentralblatt fur Bakteriologie, Suppl. 16, 225-34.
Noble, W. C. (1988). Activity of Corynebacterium jeikeium bacteriocin in vivo. Microbial Ecology
in Health and Disease 1, 201-3.
Noble W. C , Virani, Z. & Crec, R. G. A. (1992). Co-transfer of vancomycin and other resistance
genes from Enlerococcus faecalis NCTC 12201 to Staphylococcus aureus. FEMS
Microbiology Letters 93, 195-8.
Noble, W. C. & Willie, J. A. (1980). Interactions between antibiotic-producing and non-producing
staphylococci in skin surface and sub-surface models. British Journal of Experimental
Pathology 61, 339-43.
O'Shea, J. G. (1990). "Two minutes with venus, two years with mercury"—mercury as an
antisyphilitic chemotherapeutic agent. Journal of the Royal Society of Medicine 83, 392-5.
Rosenberg, B., Rcnshaw, E., Vancamp. L., Hartwick, J. & Drobnik, J. (1967). Platinum-induced
filamentous growth in Escherichia coli. Journal of Bacteriology 93, 716-21.
Sadler, P. J. (1976). The biological chemistry of gold: a metallo-drug and heavy-atom label with
variable valency. Structure and Bonding 29, 171-214.
Smith, W. E., Brown, D. H. & Cappell, H. A. (1985). Gold-binding sites in the plasma of patients
with rheumatoid arthritis undergoing treatment with gold sodium thiomalate. Inorganic
Chimica Ada—Bioinorganic Chemistry Articles and Letters 106, L23—4.
Thurman, R. B. & Gerba, C. P. (1989). The molecular mechanisms of copper and silver ion
disinfection of bacteria and viruses. CRC Critical Reviews of Environment Control 18,
205-315.
Vaara, M. (1992). Agents that increase the permeability of the outer membrane. Microbiology
Reviews 56, 395-411.
Watkins, J. W., Elder, R. C , Greene, B. & Darnall, D. W. (1987). Extermination of gold binding
in an algal biomass using EXAFS and XANES spectroscopies. Inorganic Chemistry 26,
1147-51.
Woods, G. L. & Washington, J. A. (1995). Dilution test procedures. In Manual of Clinical
Microbiology, 6th edn (Murray, P. R., Baron, E. J., Pfaller, M. A., Tenover, F. C , & Yolken,
R. H., Eds), pp. 1327-41.
(Received 31 October 1995; accepted 15 January 1996)