*Corresponding author: Mailing address: Department of
Orthopedic Surgery, Graduate School of Biomedical
Science, Nagasaki University, 1-7-1, Sakamoto, Nagasaki
852-8501, Japan. E-mail: t.tamai＠ind.bbiq.jp
Jpn. J. Infect. Dis., 64, 304-308, 2011
Biofilm Deficiency in Polysaccharide Intercellular Adhesin-Negative
Variants of Staphylococcus epidermidis Selected by
Subminimal Inhibitory Concentrations of Gentamicin
Takashi Tamai*, Toshiyuki Tsurumoto1, Shiro Kajiyama,
Shinji Adachi, Toshiyuki Sakimura, and Hiroyuki Shindo
Department of Orthopedic Surgery and1Department of Gross Anatomy, Graduate School of
Biomedical Science, Nagasaki University, Nagasaki 852-8501, Japan
(Received August 9, 2010. Accepted May 11, 2011)
SUMMARY: Staphylococcus epidermidis is a cause of orthopedic device-related infection, and to treat
such infection, biofilms should be controlled. Polysaccharide intercellular adhesin (PIA) is associated
with the biofilm-forming ability of staphylococcal strains. PIA in biofilm-positive staphylococcal
strains can be detected by the Congo red agar (CRA) method. In this study, we used the CRA method to
examine the effects of subminimal inhibitory concentrations (sub-MICs) of 11 antibacterial agents on
PIA production by S. epidermidis. We found that the PIA-negative variants were selected only by sub-
MICs of gentamicin (GM). This PIA-negative phenotype was maintained over several generations in the
absence of GM. Such selection occurred in six of eight clinical isolates, as well as in the biofilm-positive
control strain. No such selection occurred with aminoglycoside antibiotics except for GM. Most of the
PIA-negative variants that were selected by GM showed a markedly lower biofilm-forming ability on
stainless steel washers than their untreated parent strains. In conclusion, variants with lower biofilm-
forming ability may be selected by a sub-MIC of GM. Investigation of the reason why variants with
reduced biofilm-forming ability can be selected in the presence of sub-MICs of GM may contribute to
strategies against biofilm-related infections.
Orthopedists want to ensure the eradication of
postoperative infections. Recently, biofilms produced
by bacteria have attracted attention as a cause of chron-
ic and intractable device-related infections (1). It has
been suggested that the bacteria in biofilms are 100- to
1,000-fold more resistant to antibacterial agents than
planktonic bacteria (2). It has also been suggested that
the biofilm-forming ability of some staphylococcal
strains is dependent in part on polysaccharide intercellu-
lar adhesin (PIA). Various reports have been published
on the effects of subminimal inhibitory concentrations
(sub-MICs) of antibacterial agents on biofilm formation
and PIA production; however, the results are still con-
Here, we investigated whether sub-MICs of 11 anti-
bacterial agents affect PIA production, and whether al-
tered PIA production reflects biofilm-forming ability
using a biofilm-positive control strain and eight clinical
MATERIALS AND METHODS
Bacterial strains and chemicals: The eight clinical iso-
lates of Staphylococcus epidermidis used in this study
were obtained from infected sites in patients with bone
and joint infections at Nagasaki University Hospital. S.
epidermidis American Type Culture Collection (ATCC)
35984 (RP62A), which is a well-known biofilm producer
(3), was used as a positive control for biofilm forma-
tion. These strains were stored in beads (MicroBankTM;
Pro-Lab Diagnostics, Richmond Hill, Canada) at
－809 C. Congo red (1z Congo red solution) was pur-
chased from Muto Pure Chemicals (Tokyo, Japan).
Mueller-Hinton broth (MHB) and trypticase soy broth
(TSB) were obtained from Becton-Dickinson and Com-
pany (Sparks, Md., USA). Agar was purchased from
Ina Food Industry (Nagano, Japan). D-(＋)-glucose was
obtained from Wako Pure Chemical Industries (Osaka,
Japan). Crystal violet (CV, 1z Pyoktanin Blue Solu-
tion) was purchased from Kanto Chemical (Tokyo,
Japan). Phosphate-buffered saline (PBS) and blood
agar were purchased from Gibco (Gaithberg, Md.,
USA) and Eiken (Tochigi, Japan), respectively.
Antibacterial agents used in this study and determina-
tion of minimal inhibitory concentrations (MICs) of
gentamicin (GM): The antibacterial agent-containing
discs (BBL Sensi-Disc) used in this study were purchased
from Becton-Dickinson Company, and the amount of
each agent per disc was as follows: GM, 120 mg; amika-
cin (AMK), 30 mg; arbekacin (ABK), 30 mg; vancomycin
(VCM), 30 mg; tetracycline (TC), 30 mg; chloram-
phenicol (CP), 30 mg; cefazolin (CEZ) 30 mg; fosfomy-
cin (FOM) 50 mg; minocycline (MINO) 30 mg; rifampin
(RFP), 5 mg; and clarithromycin (CAM), 15 mg.
MICs for suspended bacteria were measured by the
microdilution method (4).
Fig. 1. The effects of subminimal inhibitory concentrations (sub-MICs) of gentamicin (GM) on polysaccharide in-
tercellular adhesin (PIA) production were assessed using Congo red agar (CRA) method. (A) ATCC 35984, (B)
strain no. 1, (C) strain no. 2, (D) strain no. 3, (E) strain no. 4, (F) strain no. 5, (G) strain no. 6, (H) strain no. 7, (I)
strain no. 8. PIA-negative variants grow as red colony on CRA. PIA-negative variants appeared around inhibitory
zones for all strains, with two exceptions (B and C).
Assessment of PIA production using the Congo red
agar (CRA) method: PIA production was assessed using
the CRA method as previously described (5). Briefly,
colonies on blood agar were inoculated into 10 mL of
TSB and incubated at 379 C for approximately 3 h
(OD600＝ 0.2). This culture was vortexed and then dilut-
ed 10-fold with PBS. Approximately 0.1 mL of the
diluted suspension was dropped onto CRA plates
(0.08z Congo red, 2.1z MHB, 0.5z glucose, and
1.7z agar), the plates were briefly air-dried, and then
the antibacterial discs were put in place. The plates were
incubated at 379 C for 48 h and then incubated at room
temperature for 24 h. PIA production was determined
by macroscopic observation since PIA-negative variants
form red colonies, whereas PIA-positive variants form
black colonies (5). PIA-negative variants were used in
the subsequent experiments.
Assessment of biofilm formation: The biofilm-form-
ing ability of the PIA-negative variants was compared
with that of their cognate parent strains using a previ-
ously described method (6). Briefly, colonies were in-
oculated in antibacterial agent-free TSB and incubated
at 379 C for approximately 3 h (OD600＝ 0.2). Sterilized
stainless steel washers (UW–0303–05; 6.0 mm diameter
and 0.5 mm thickness, SUS304 quality containing 18z
chrome and 8z nickel; Wilco, Tokyo, Japan) were im-
mersed in these cultures for 10 min to allow bacteria to
adhere. Subsequently, the bacteria adhering to the
washers were incubated in fresh TSB at 379 C for 24 h
with the adhered surface down in order to facilitate
The washer to which the biofilms adhered was washed
gently with PBS to remove the planktonic bacteria, and
then dried with a drier for approximately 5 min to fix
the bacteria to the washer. The biofilm adhering to the
washer was stained with CV for 2 min, and then washed
twice with PBS to remove the excess stain.
Eight areas (660 mm × 480 mm) of each washer were
arbitrarily selected and full-color pictures were obtained
using a digital optical microscope (VHX-100; Keyence,
Osaka, Japan). The biofilm coverage ratio (BCR) was
calculated as the ratio of the CV-stained area to total
area of the washer. The color photograph was then con-
verted to a grayscale TIFF image using Photoshop Ele-
ments 6 (Adobe Systems, San Jose, Calif., USA), and
the BCR was measured with Scion Imaging software
(Scion, Frederick, Md., USA). The data are presented
as percentages and the mean ± standard deviation (SD)
of four replicates. The data were analyzed using
Student's t test.
Assessment of revertants from serial passage of PIA-
negative variants: PIA-negative variants underwent seri-
Fig. 2. Assessment of biofilm-forming ability of PIA-negative variants compared with that of the cognate parent
strains using crystal violet (CV) staining of biofilms formed on stainless steel washers. The macroscopic images of
the CV-stained biofilms formed on stainless steel washers showed that the PIA-negative variant (B) impaired the
biofilm-forming ability compared with the parent strain of ATCC35984 (A). The biofilm coverage ratios (BCRs)
were compared between the PIA-negative variants and the cognate parent strains derived from the PIA-positive
control strain (ATCC35984) and clinical strains in which sub-MICs of GM selected PIA-negative variants (C).
*P º 0.05 compared with parent strains.**P º 0.01 compared with parent strains.
al passage every 24 h in antibacterial-free TSB until
PIA-positive prototypes (revertants) appeared (for a
maximum of 2 weeks). The CRA method described
above was also used to determine reversion. Revertants
were identified when more than approximately 10z of
the colonies that appeared were PIA-positive.
Sodium chloride (NaCl) (maximum, 3z) has been
reported to promote reversion of some variants (7). To
assess the effect of NaCl, serial passages were per-
formed using antibacterial-free TSB with 0.5z NaCl
until reversion occurred (for a maximum of 2 days).
PIA-negative variants appear at sub-MICs of GM:
The MICs of GM against clinical strain No. 1, 2, 3, 4, 5,
6, 7, and 8 and strain ATCC35984 were Ã0.5, Ã0.5,
Æ32, Æ32, Æ32, Æ32, 16, Æ32, and 8 mg/mL, respec-
tively. PIA-negative variants appeared around the GM
inhibitory zone of neally all strains, with two exceptions
(Fig. 1). In contrast, PIA-negative variants did not ap-
pear around the inhibitory zone of the other antibacteri-
al agents used in this study (data not shown).
This PIA-negative phenotype was maintained in the
absence of GM over several generations.
PIA-negative variants have impaired biofilm-forming
ability compared with their cognate parent strains: The
macroscopic images of CV-stained biofilms formed on
stainless steel washers showed that the PIA-negative
variant of ATCC35984 (Fig. 2B) had impaired biofilm-
forming ability compared with the parent strain (Fig.
2A). The PIA-negative variants of the clinical isolates
also had macroscopically impaired biofilm-forming
ability compared with the parent strains. The BCRs of
the PIA-negative variants, except for No. 5, were sig-
nificantly lower than those of their cognate parent
strains, and the BCR of strain No. 5 also tended to be
lower than that of the parent strain (P ＝ 0.056) (Fig.
PIA-negative variants reverted to the PIA-positive
phenotype after serial passages; however, reversion was
not promoted by NaCl: PIA-negative variants reverted
to the PIA-positive phenotype after four or more pas-
sages. ATCC35984 reverted after 10 passages, No. 4, 5,
6, 7, and 8 reverted after 12 or 13 passages, and No. 3
reverted after four passages. NaCl did not promote the
reversion of any strain tested in our study.
Although Rachid et al. previously reported that sub-
MICs of GM had no effect on the expression of the ica
genes that encode enzymes involved in PIA production
(8), we found that PIA-negative variants were selected
by sub-MICs of GM. It is unlikely that the PIA-negative
variants are produced due to the action of Congo red at
the concentration employed in the CRA method (9).
Although the method used in our study is different from
that used in their study, we confirmed that PIA-negative
variants were selected by GM in the absence Congo red
(data not shown). Therefore, we could not clarify the
discrepancy between these studies; however, it was not
dependent on the combination of GM and Congo red
that we used.
The mechanism by which PIA-negative variants are
selected by sub-MICs of GM is not fully understood;
however, in a previous report by Ziebuhr et al., three
possibilities were suggested. They classified naturally
occurring PIA-negative variants into three groups. In
the first group, the PIA-negative variants revert to the
PIA-positive phenotype after only two serial passages or
with the addition of NaCl, and expression of the ica
operon is inhibited (7). In the second group, 9 to 12 seri-
al passages are required for PIA-negative to PIA-posi-
tive reversion (7); this group includes phase variants
proven to be caused by rearrangement of the ica operon
(10,11). In the third group, reversion never occurs, and
the ica operon is either completely or partially deleted
(7). According to their classification, our variants might
belong to the second group; however, gene arrangement
was not confirmed, and further investigation is re-
The reason why PIA-negative variants were not ob-
tained from two of the clinical isolates is unclear.
However, the sensitivities of these strains to GM were
higher (MICs Ã0.5 mg/mL) than those of other strains.
Therefore, for the two clinical isolates from which no
PIA variant was selected, the antibacterial effect of GM
may have been dominant to the PIA-negative variant-
We also showed that most PIA-negative variants had
significantly impaired biofilm-forming ability. Accord-
ingly, our results suggest that the impaired biofilm-
forming ability of these variants may be due in part to
the inhibition of PIA production.
Because inhibition of protein synthesis is the antibiot-
ic mechanism of action for GM, we initially considered
that inhibition of protein synthesis resulted in the selec-
tion of PIA-negative variants of the first group. It has
also been reported that a protein synthesis inhibitor,
CP, inhibited biofilm synthesis (12). However, our
study showed that PIA-negative variants were not
selected by CP or other aminoglycosides except GM.
Therefore, the impaired biofilm-forming ability of our
PIA-negative variants might be independent of protein
synthesis inhibition. In addition, biofilm production is
dependent on not only PIA production due to ica gene
expression but also agr genes and other factors (13).
Furthermore, ica-independent biofilm-forming strains
have recently been reported (14). These complicated
mechanisms might also contribute to the variation in
biofilm-forming ability among our PIA-negative vari-
There was no significant difference between the BCRs
of the parent strain and the PIA-negative variant of No.
5. We employed the BCR method because our laborato-
ry is familiar with this method, and it is relatively relia-
ble and reproducible (15). However, this method does
not reflect the three-dimensional mass of biofilms, and
thus it might have underestimated significant differ-
ences in biofilm-forming ability between the parent
strain and the PIA-negative variant of isolate No. 5.
In conclusion, we reported that PIA-negative variants
selected by sub-MICs of GM markedly impaired
biofilm-forming ability. Our finding is clinically intri-
guing because biofilm formation frequently causes in-
tractable infections. The mechanism underlying biofilm
formation should be determined in order to provide ef-
fective anti-biofilm strategies against intractable infec-
(Division of Laboratory Medicine, Nagasaki University Hospital,
Nagasaki, Japan) for storing and supplying strains.
We are grateful to Katsunori Yanagihara
Conflict of interest
None to declare.
1. Stewart, P.S. and Costerton, J.W. (2001): Antibiotic resistance of
bacteria in biofilms. Lancet, 358, 135–138.
2. Gristina, A.G. and Costerton, J.W. (1985): Bacterial adherence
to biomaterials and tissue: the significance of its role in clinical
sepsis. J. Bone Joint Surg. Am., 67, 264–273.
3. Gäotz, F. (2002): Microreview Staphylococcus and bioflims. Mol.
Microbiol., 43, 1367–1378.
4. Clinical and Laboratory Standards Institute (2009): Methods for
dilution antimicrobial susceptibility tests for bacteria that grow
aerobically; approved standard-8 ed. Document M7-A8. Clinical
and Laboratory Standards Institute, Wayne, Pa.
5. Freeman, D.J., Falkiner, F.R. and Keane, C.T. (1989): New
method for detecting slime production by coagulase negative
staphylococci. J. Clin. Pathol., 42, 872–874.
6. Adachi, K., Tsurumoto, T., Yonekura, A., et al. (2007): New
quantitative image analysis of staphylococcal biofilms on the sur-
faces of nontranslucent metallic biomaterials. J. Orthop. Sci., 12,
7. Ziebuhr, W., Loessner, I., Krimmer, V., et al. (2001): Methods to
detect and analyze phenotypic variation in biofilm-forming
staphylococci. Methods Enzymol., 336, 195–205.
8. Rachid, S., Ohlsen, K., Witte, W., et al. (2000): Effects of subin-
hibitory antibiotic concentrations on polysaccharide intercellular
adhesin expression in biofilm-forming Staphylococcus epidermi-
dis. Antimicrob. Agents Chemother., 44, 3357–3363.
9. Arciola, C.R., Campoccia, D., Gamberini, S., et al. (2002): De-
tection of slime production by means of an optimised Congo red
agar plate test based on a colourimetric scale in Staphylococcus
epidermidis clinical isolates genotyped for ica locus. Biomaterials,
10. Ziebuhr, W., Krimmer, V., Rachid, S., et al. (1999): A novel
mechanism of phase variation of virulence in Staphylococcus
epidermidis: evidence for control of the polysaccharide intercellu-
lar adhesin synthesis by alternating insertion and excision of the
insertion sequence element IS256. Mol. Microbiol., 32, 345–356.
11. Ziebuhr, W., Dietrich, K., Trautmann, M., et al. (2000): Chro-
mosomal rearrangements affecting biofilm production and an-
tibiotic resistance in a Staphylococcus epidermidis strain causing
shunt-associated ventriculitis. Int. J. Med. Microbiol., 290,
12. Dobinsky, S., Kiel, K., Rohde, H., et al. (2003): Glucose-related
dissociation between icaADBC transcription and biofilm expres-
sion by Staphylococcus epidermidis: evidence for an additional
factor required for polysaccharide intercellular adhesin synthesis.
J. Bacteriol., 185, 2879–2886.
13. Boles, R.B. and Horswill, A.R. (2008): agr-mediated dispersal of
Staphylococcus aureus biofilms. PLoS Pathog., 4, e1000052.
14. O'Gara, J.P. (2007): ica and beyond: bioflm mechanisms and
regulation in Staphylococcus epidermidis and Staphylococcus
aureus. FEMS Microbiol. Lett., 270, 179–188.
15. Kajiyama, S., Tsurumoto, T., Osaki, M., et al. (2009): Quantita-
tive analysis of Staphylococcus epidermidis biofilm on the surface
of biomaterial. J. Orthop. Sci., 14, 769–775.