Approval for the research protocol was received from the animal
care and use committee. Adult male Sprague-Dawley rats weighing
approximately 300 g (Sasco, Williamstown, Massachusetts) were housed
individually and allowed free access to food and water. The rats
were divided into two groups. Ninety-seven were inoculated with
one of the test organisms (a monomicrobial inoculation), either Staphylococcus
aureus (number 29213; American Type Culture Collection,
Manassas, Virginia) or Pseudomonas aeruginosa (number
27853; American Type Culture Collection); ninety-four animals received
a polymicrobial inoculation with both test organisms in varying concentrations.
Six animals were inoculated with each ratio of Staphylococcus
aureus to Pseudomonas aeruginosa. Each
ratio was administered on two different days—that is, three
animals were inoculated on one day, and three were inoculated on
the other. After all inoculations were completed, ten additional
animals were inoculated with the combination of 103 colony-forming
units (CFU) of Staphylococcus aureus and 103 CFU
of Pseudomonas aeruginosa. All procedures were
performed in a designated animal operating room.
Preparation of Bacterial Inoculum
Staphylococcus aureus and Pseudomonas
aeruginosa were obtained from the Harry S. Truman Veterans Administration
Hospital inpatient laboratory and were maintained on trypticase-soy-agar
plates impregnated with 5% sheep’s blood (BBL
Microbiology Systems, Cockeysville, Maryland). Twenty-four hours
prior to use, strains were streaked and were transferred to fresh
trypticase-soy-agar plates. On the day of inoculation, a portion
of the freshly transferred organisms was suspended in 1.0 mL of sterile
normal saline solution in a 2-mL microcentrifuge tube (Fisher Scientific,
Pittsburgh, Pennsylvania). This suspension was centrifuged at 6000 rpm
for two minutes (Eppendorf 5415; Brinkman Instruments, Westbury,
New York). The supernatant was removed with a micropipette (Pipetman;
Rainin Instrument, Woburn, Massachusetts), and the pellet was resuspended
in 1 mL of fresh normal saline solution. This process was repeated
twice for a total of three washings. The final pellet was resuspended
in normal saline solution. The washed bacterial suspension was then
diluted in normal saline solution, and the absorbance at 530 nm
was measured with use of a spectrophotometer (Spectronic 20+;
Spectronic Instruments, Rochester, New York). A standard curve was
determined for each organism, and the absorbance required to produce
a concentration of 1.0 ¥ 108 CFU/mL
was determined. Serial dilutions were performed to obtain the desired
concentration of colony-forming units. This suspension of known bacterial
concentration was transferred to 2-mL microcentrifuge tubes (Fisher
Scientific), placed on ice, and taken to the animal operating room.
Each concentration was verified by serially diluting the experimental
suspension onto trypticase-soy-agar plates. The number of colonies
on these dilution plates was manually counted after incubation at 37°C
for twenty-four hours.
The inoculum for each animal was suspended in 100 L of normal
saline solution. For the polymicrobial inocula, each organism was
suspended individually in 50 L of normal saline solution and remained
separate until each was placed individually into the operative wound.
Operative Procedure
Once the animal was anesthetized with 3% isoflurane
(Baxter Pharmaceutical Products, Liberty Corner, New Jersey), the
lower back was shaved and was prepared with povidone-iodine solution
and 70% ethanol. The animal was transferred to a sterile
operating table. The operative field was draped with a small, sterile
towel, and the operative site was covered with a 6.0 ¥ 7.0-cm
Tegaderm (3M Healthcare, St. Paul, Minnesota). An incision was made
over the lumbar spine with use of a number-15 blade. The largest
lumbar spinous process was palpated and exposed. A 20-gauge needle
(Becton Dickinson, Rutherford, New Jersey) was placed through the
exposed spinous process, and a 22-gauge stainless-steel wire was
placed through the needle. The needle was then removed, leaving
the wire in the spinous process. The wire was tied, cut to a length
of 2 cm, folded back on itself, and laid flat in the wound. The
bacterial inoculum was then placed into the wound with use of a
micropipette (Pipetman; Rainin Instrument).
The wound was allowed to incubate for fifteen minutes. Pulsatile
irrigation was then performed with 3 L of normal saline solution
with use of a Pulsavac and an irrigation gun (Zimmer Patient Care,
Dover, Ohio). Finally, the wound was closed in one layer with use
of 4-0 Monosof nylon horizontal mattress sutures (United States
Surgical, Norwalk, Connecticut). Each animal received 0.05 mg of
Buprenex (buprenorphine) subcutaneously for analgesia. All animals
were individually housed in fresh cages, given free access to food
and water, and allowed to convalesce for two weeks.
A total of 191 animals were studied. We used seventy-seven animals
to determine the doses of Pseudomonas aeruginosa (forty-seven
animals) and Staphylococcus aureus (thirty animals)
that would be required to cause an infection in 50% of
the animals (ID50). Next, two groups of forty-two animals (a total
of eighty-four animals) were inoculated with Staphylococcus
aureus (at a concentration of 103 CFU
in one group and 106 CFU in the other)
and ascending concentrations of Pseudomonas aeruginosa. Finally,
ten animals were inoculated with 103 CFU
of Staphylococcus aureus, 103 CFU
of Pseudomonas aeruginosa, or 103 CFU
of both Staphylococcus aureus and Pseudomonas
aeruginosa, for a total of thirty animals.
Culture Collection and Strain Identification
All animals were killed in a CO2 chamber on postoperative day
14. The animals were shaved and prepared, and a number-15 blade
was used to make a superficial incision just lateral to the original
incision. A sterile cotton applicator was swiped on the subdermal
tissue and then placed into a sterile 15-mL tube (Fisher Scientific)
containing 5 mL of sterile trypticase-soy broth (BBL Microbiology
Systems). With use of a fresh scalpel blade, an incision was made
through the paraspinous fascia and the underlying muscle on each
side of the spinous process that contained the implanted wire. The paraspinal
musculature was swabbed with a cotton applicator, and the swab was
placed into 5 mL of trypticase-soy broth. Finally, the wire was
grasped with a sterile hemostat, pulled out of the spinous process,
and placed in 5 mL of trypticase-soy broth. The trypticase-soy-broth
culture tubes were then placed in an incubator at 37°C.
The presence or absence of turbidity in the trypticase-soy-broth
culture tubes was recorded at twenty-four and seventy-two hours.
Medium from turbid culture tubes was applied to 5% sheep-blood trypticase-soy-agar
plates and incubated at 37°C for twenty-four hours. Colony morphology,
color, and type of hemolysis were recorded. Individual colonies
were removed from the plates with a wire loop. Coagulase (Difco
Laboratories, Detroit, Michigan), catalase, and oxidase (Sigma Chemical,
St. Louis, Missouri) activity was determined. Antibiotic susceptibility
patterns were performed on Mueller-Hinton agar plates (BBL Microbiology
Systems) for strain identification.
Data Analysis
Infection was defined as growth of the test strain from any of
the three sites, in each animal, from which specimens had been removed
for culture (superficial tissue, deep tissue, or wire implant). The
Reed-Muench method was used to determine the number of colony-forming
units required for infection in 50% of the animals (ID50)
for each of the test strains. The Fisher exact test was used to compare
the numbers of infections at the individual culture sites within
each group of animals. Statistical comparisons of infection rates
between groups were made with use of the determination of the likelihood
of independence, which considers -2 times the difference of the
log likelihood for each data set. This test closely approximates
the chi-square test but corrects for the possibility of interaction
among data sets. SAS software (SAS, Cary, North Carolina) was used
for statistical calculations.
ID50 with Either Organism Alone
The number of colony-forming units of Staphylococcus
aureus needed to cause infection in 50% of the
animals was 2.8 ¥ 104 (Fig. 1-A). A bacterial
load of 103 CFU produced a 10% infection
rate, and 106 CFU produced a 100% infection
rate. The distribution among the three culture sites showed a definite predilection
for the wire site. There was at least one positive culture of the
specimens from four of the nineteen superficial sites, five of the
nineteen deep-tissue sites, and all nineteen wire sites (Table I). These differences
were significant (p < 0.001).
The number of colony-forming units of Pseudomonas aeruginosa required
to cause a 50% infection rate (Fig. 1-B) was 4.8 ¥ 105;
103 CFU of bacteria produced a 0% infection
rate, and 106 CFU produced a 68% infection
rate. The culture-site distribution was not similar to that seen
in the evaluations of the Staphylococcus aureus ID50.
Twenty-three animals had infection of at least one culture site.
Among these twenty-three animals, eight had an infection at a superficial
site; fourteen, at a deep-tissue site; and seven, at a wire site
(Table I).
Polymicrobial Inoculation
103 CFU of Staphylococcus
aureus Combined with Pseudomonas aeruginosa
The combination of 103 CFU of Staphylococcus
aureus (one-tenth of the Staphylococcus aureus ID50)
with low concentrations of Pseudomonas aeruginosa (102,
103, or 104 CFU)
yielded infection rates that were higher than those found with either
organism alone at the same concentrations. This trend culminated
with the combination of 103 CFU of Staphylococcus
aureus and 103 CFU of Pseudomonas
aeruginosa. At this concentration, a 75% infection
rate was observed. With 10 CFU of Pseudomonas aeruginosa and
as the concentration of Pseudomonas aeruginosa was
increased (to 105, 106,
and 107 CFU), this trend reversed and
the infection rate fell to 33% at a concentration of 107 CFU.
This rate was well below that with 107 CFU
of Pseudomonas aeruginosa alone (83%)
(Fig. 2).
The differences between the infection rates of animals inoculated
with both species and those inoculated with one species were significant
(p = 0.004).
The only organism isolated from any of the wounds inoculated
with both Staphylococcus aureus and Pseudomonas
aeruginosa was Staphylococcus aureus; Pseudomonas
aeruginosa was not isolated from any of these wounds. The distribution
of infections among the culture sites was similar to that seen in
the animals inoculated with Staphylococcus aureus alone.
Of twenty-four animals for which at least one culture was positive,
five had an infection at a superficial site; eleven, at a deep-tissue
site; and twenty-two, at a wire site (Table I). These differences were significant
(p < 0.001).
At low concentrations of Pseudomonas aeruginosa, the
increased infection rate was greater than the additive value of
the infection rate of each organism individually. An inoculation
with 103 CFU of Staphylococcus
aureus would be expected to result in a 10% infection
rate, and an inoculation with 103 CFU
of Pseudomonas aeruginosa should yield no infections,
yet animals inoculated with 103 CFU of
both Staphylococcus aureus and Pseudomonas
aeruginosa had an infection rate of 75% (p = 0.004)
(Fig. 3).
106 CFU of Staphylococcus
aureus Combined with Pseudomonas aeruginosa
When we repeated the polymicrobial series with an inoculum of
106 CFU of Staphylococcus aureus,
we anticipated an infection rate approaching 100%, on the
basis of the results of the Staphylococcus aureus ID50.
With low concentrations of Pseudomonas aeruginosa (0
to 105 CFU), we observed infection rates
ranging from 83% to 100%. At higher concentrations
of Pseudomonas aeruginosa (106 and
107 CFU), the infection rate paradoxically
decreased (Fig. 4)
compared with the infection rates with either Staphylococcus
aureus or Pseudomonas aeruginosa alone.
The combination of 106 CFU of Staphylococcus
aureus and 107 CFU of Pseudomonas
aeruginosa yielded only a 33% infection rate.
This rate was significantly different (p = 0.005) from
the rates with Staphylococcus aureus or Pseudomonas
aeruginosa alone. As with the previous series, all of the
organisms grown on culture were Staphylococcus aureus; thirty-six
of thirty-eight wire sites were infected (p < 0.001) (Table I).
The results of our study demonstrate synergy between Staphylococcus
aureus and Pseudomonas aeruginosa when
a low level of each organism is present in the wound. At low bacterial
levels, wound conditions appeared to enhance the pathogenicity of Staphylococcus
aureus while at the same time interfering with the pathogenicity
of Pseudomonas aeruginosa. This suggests that low
concentrations of Pseudomonas aeruginosa somehow
enhance the ability of Staphylococcus aureus to
cause infection in this orthopaedic wound model, while the presence
of Staphylococcus aureus protects against infection
from Pseudomonas aeruginosa.
Bacterial synergy is not a new concept. Reports on bacterial
synergy can be found in the general surgery literature as early
as 193111. Synergy between Escherichia
coli and Bacteroides fragilis has been
documented. Dunn et al.12 showed
this synergy in a rat model, and Kelly13 demonstrated
it in a guinea-pig wound model. Synergy between anaerobes and facultative
aerobes has also been well documented7,12-15.
Hall et al.6 noted that patients
who had mixed anaerobic-aerobic osteomyelitis did not respond to
treatment 61.5% of the time, compared with a 20% rate
of non-response for patients with single-organism osteomyelitis.
Kelly and Warren7 studied cultures
of specimens taken from wounds during abdominal operations, before
they were irrigated with normal saline solution, and found that wounds
from which both aerobes and anaerobes had been isolated had a 71% infection
rate whereas those with aerobes alone were infected only 23% of the
time. These examples demonstrate the clinical relevance of microbial
interactions resulting in enhanced pathogenicity.
Staphylococcus species account for the majority of infections
in orthopaedic wounds. Staphylococcus aureus is
among the most common organisms causing pediatric osteomyelitis16 as well as clinical infections after
total hip arthroplasty3, open
traumatic wounds9, and elective
orthopaedic procedures17. It has
been hypothesized that the ubiquity and virulence of this organism
accounts for its preeminence.
Associations between Staphylococcus aureus and
anaerobic bacteria have been described, and synergy between Staphylococcus
species has been demonstrated in animal osteomyelitis models8,14,18. Rissing et al.19, using a rat model of osteomyelitis,
found that levels of Bacteroides fragilis as low
as 100 organisms per wound greatly enhanced the infectivity of Staphylococcus
aureus. Merritt and Dowd5 discovered,
in a hamster model of open fractures, that the addition of 104 CFU
of Proteus mirabilis to wounds inoculated with Staphylococcus
aureus greatly increased the infection rate. The ability
of multiple organisms to enhance the pathogenicity of Staphylococcus
aureus in these animal models suggests that a similar phenomenon
is possible in humans. Synergy may play a role in the predominance
of Staphylococcus aureus infections in orthopaedic
wounds, especially in the small number of bacteria believed to cause
infection associated with total joint arthroplasty4.
In our study, only Staphylococcus aureus was
isolated from the positive cultures of specimens from the polymicrobially
inoculated animals, even when the concentration of Pseudomonas
aeruginosa had been increased to 107 CFU
in the presence of just 103 CFU of Staphylococcus
aureus. Among the animals infected polymicrobially, the pattern
of positive cultures (that is, the relative preponderance of positive
cultures of specimens from superficial, deep-tissue, or wire sites)
was consistent with that of animals infected with Staphylococcus
aureus alone. In both the polymicrobially inoculated animals
and the animals inoculated with Staphylococcus aureus alone,
there was a marked predilection for wire sites. This was not the
case among the animals inoculated with Pseudomonas aeruginosa alone.
There are two possible explanations for our inability to isolate Pseudomonas
aeruginosa after polymicrobial inoculations: (1) Pseudomonas
aeruginosa was in the wounds and we were unable to isolate
it in culture, and (2) no Pseudomonas aeruginosa was
in the wounds. We do not believe that the first possibility is true
because our ability to isolate Pseudomonas from the animals inoculated
with Pseudomonas aeruginosa alone suggests that
an error in culture methods does not account for the failure to
isolate Pseudomonas aeruginosa in the polymicrobially
inoculated animals. However, if the number of Pseudomonas organisms
in the wounds was below the sensitivity of the cotton-swab technique,
these very small numbers of organisms would have been undetectable.
Finally, if the second possibility were true, wound ecology may have
adversely affected Pseudomonas aeruginosa in such
a way that no organisms were viable at the time that the specimens
were taken for culture.
We think that very little or no Pseudomonas aeruginosa was
present in the wounds at fourteen days. The phenomenon of only one
organism growing after the use of a mixed inoculum is not unique
to our study. Merritt and Dowd5 noted
microbial specificity in their previously mentioned hamster model.
This specificity depended on the presence or absence of internal fixation.
Of fourteen animals that underwent internal fixation and were inoculated
with 104 CFU of Staphylococcus
aureus and Proteus mirabilis, ten had
growth of Proteus mirabilis alone on culture at
two weeks, one had growth of Staphylococcus aureus alone,
and three had growth of both organisms. In sixteen animals that
did not receive internal fixation, a different pattern was seen:
eight had growth of only Staphylococcus aureus, three
had growth of only Proteus mirabilis, and five
had growth of both organisms. The authors concluded that the presence
of Staphylococcus aureus potentiated infection
with Proteus mirabilis, but they did not comment
on the differences seen between the animals that had internal fixation
and those that received no fixation. In their animal study, Rissing
et al.19 noted smaller numbers
of Bacteroides fragilis than anticipated when compared
with the numbers of Staphylococcus aureus in final
cultures. They attributed this to the liberation of prostaglandin
E2 by Bacteroides fragilis, which then facilitated
an infection with Staphylococcus aureus.
Interplay between the host and pathogens as well as interaction
among pathogens seems to substantially influence the local ecology
of the wound, according to the data from our study and that of Merritt
and Dowd5. The presence of orthopaedic
implants has been shown to decrease the number of bacteria required to
cause infection19, further complicating
the already manifold interactions among the pathogens and of the
host to these pathogens. We hypothesize that interactions taking place
not only among the pathogens but also with the host immune response
play a substantial role in the wound environment and alter the wound
ecology. Seemingly small changes in the wound environment can greatly
affect the microbiology of the wound.
We think that host interactions combined with the interplay of
pathogens may explain why synergy was demonstrated at low levels
of Pseudomonas aeruginosa and apparent inhibition
was shown at higher levels. The exact mechanism is unknown. A possible explanation
for this phenomenon lies in the specificity of the immune response.
The immune system may target one pathogen preferentially at certain inoculum
concentrations, with the other "slipping by" undetected.
As the level of inoculation increases, the inflammatory response
is greater and both pathogens are affected. Changes in the wound, which
mediate the inflammatory response, can then affect which organism
predominates in the wound.
The results of our study demonstrated (1) synergy at low concentrations
of Staphylococcus aureus and Pseudomonas
aeruginosa, (2) inhibition of Pseudomonas aeruginosa by Staphylococcus
aureus, and (3) predominance of Staphylococcus
aureus in wounds infected with both Staphylococcus
aureus and Pseudomonas aeruginosa in a
rat model of orthopaedic wounds.
Elucidation of the mechanisms underlying our findings is beyond
the scope of this project. However, we hypothesize that the host
immune response may play an important role in the presence of polymicrobial
infection and may influence the microbiology of the wound. More
study is required to unravel the mechanisms underlying our findings,
but doing so may shed light on the reason for the preponderance
of Staphylococcus species found in orthopaedic infections.
Note: The authors thank J. Glenn Phaup, David Lane, Justin Ogden,
and Minitia Chahal for their help with the operative procedures,
and Brian Conroy, MD, for assistance with the manuscript.