APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Apr. 2009, p. 2091–2098
Vol. 75, No. 7
Structural Analysis of Biofilm Formation by Rapidly and Slowly Growing
Margaret M. Williams,1* Mitchell A. Yakrus,2Matthew J. Arduino,1Robert C. Cooksey,2
Christina B. Crane,1Shailen N. Banerjee,1Elizabeth D. Hilborn,3and Rodney M. Donlan1
Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia1; Division of
Tuberculosis Elimination, Centers for Disease Control and Prevention, Atlanta, Georgia2; and National Health and
Environmental Effects Research Laboratory, Office of Research and Development,
United States Environmental Protection Agency, Research Triangle Park,
Received 23 January 2009/Accepted 2 February 2009
Mycobacterium avium complex (MAC) and rapidly growing mycobacteria (RGM) such as M. abscessus, M.
mucogenicum, M. chelonae, and M. fortuitum, implicated in health care-associated infections, are often isolated
from potable water supplies as part of the microbial flora. To understand factors that influence growth in their
environmental source, clinical RGM and slowly growing MAC isolates were grown as biofilm in a laboratory
batch system. High and low nutrient levels were compared, as well as stainless steel and polycarbonate
surfaces. Biofilm growth was measured after 72 h of incubation by enumeration of bacteria from disrupted
biofilms and by direct quantitative image analysis of biofilm microcolony structure. RGM biofilm development
was influenced more by nutrient level than by substrate material, though both affected biofilm growth for most
of the isolates tested. Microcolony structure revealed that RGM develop several different biofilm structures
under high-nutrient growth conditions, including pillars of various shapes (M. abscessus and M. fortuitum) and
extensive cording (M. abscessus and M. chelonae). Although it is a slowly growing species in the laboratory, a
clinical isolate of M. avium developed more culturable biofilm in potable water in 72 h than any of the 10 RGM
examined. This indicates that M. avium is better adapted for growth in potable water systems than in
laboratory incubation conditions and suggests some advantage that MAC has over RGM in low-nutrient
Many species of nontuberculous mycobacteria (NTM) are
commonly isolated from potable water (PW) supplies and have
been implicated in both community-acquired and health care-
associated infections (7, 13, 18, 21, 30, 31). Much attention has
been paid to slowly growing mycobacteria, especially Myco-
bacterium avium (1, 10). However, several health care-re-
lated outbreaks and pseudo-outbreaks caused by rapidly
growing mycobacteria (RGM; e.g., M. abscessus, M. chelo-
nae, M. mucogenicum, and M. fortuitum) demonstrate the
importance of these organisms in causing infections (9).
Examples of diseases caused by RGM in PW supplies in-
clude infections in hemodialysis patients (23), postsurgical
wound infections (6), furunculosis caused by M. fortuitum
(33), and bacteremia caused by M. mucogenicum (20).
Although many studies have linked environmental mycobac-
teria to clinical isolates, NTM can be difficult to culture from
the complex community found in most drinking water dis-
tribution systems (WDS) due to competition on media from
many faster growing fungi and heterotrophic bacteria (7, 18,
25). In addition to linking infections to their source, quan-
tification of environmental NTM will help to determine
their ecological role in WDS biofilms, possibly leading to
more-effective point-of-use treatment to prevent transmis-
sion to susceptible populations. For example, previous work
has demonstrated a positive correlation between lower lev-
els of assimilable organic carbon and the concentration of
NTM in WDS biofilms (25, 30). In other work, clinical isolates
of M. avium formed more biofilm when incubated in water
than when incubated in Middlebrook 7H9 broth (4, 22). The
presence of divalent cations and carbon in the water also in-
creased biofilm production (4). Although laboratory studies
help define parameters for NTM biofilm growth, little is known
regarding environmental settings, such as the numbers of each
species in multispecies biofilms, how often they are sloughed
off into the water supply, and most importantly for human
health, what their virulence is when they reach the user during
bathing/showering, reprocessing of medical devices, or other
exposures. Given that free-living mycobacteria are part of the
water flora, it may ultimately be more relevant to determine
the virulence of NTM reaching exposed individuals than to
merely confirm their presence and numbers.
Some researchers have linked biofilm formation ability, gly-
copeptidolipid (GPL) production, and in some cases, micro-
colony morphology to virulence in NTM (19, 34). Yamazaki et
al. (34) created mutants of M. avium that could not form as
much biofilm as the wild type, and these mutants were also less
infective than the wild type. The opposite was found for M.
abscessus, where more-invasive strains formed less biofilm in a
static laboratory model (19). The rough colony type of M.
abscessus was associated with virulence more than the smooth
colony type was, as tested in human monocyte and mice mod-
* Corresponding author. Mailing address: Centers for Disease Control
and Prevention, 1600 Clifton Rd., MS-C16, Atlanta, GA 30333. Phone:
(404) 639-2697. Fax: (404) 639-3822. E-mail: MWilliams7@cdc.gov.
?Published ahead of print on 6 February 2009.
els. The rough phenotypes formed microscopic cording struc-
tures, while the smooth phenotype did not. GPL was expressed
in smooth types but little in rough types. Rough types formed
little biofilm compared to smooth types. The “hypervirulence”
of a rough colony morphotype was also observed in another
strain of M. abscessus (5), indicating that biofilm formation
ability, biofilm structure, and virulence can be linked in at least
some mycobacteria. The link between biofilm formation and
structure has been examined previously in M. chelonae, M.
fortuitum, and M. marinum (2, 14, 15, 16). M. marinum formed
cords similar to that of the rough M. abscessus strains men-
tioned above (16).
The surfaces to which mycobacteria attach are likely deter-
mined, at least in part, by the hydrophobicity and mycolic acid
composition of the organisms’ cell walls. The environmental
conditions and nutrients available inside pipes and other sur-
faces in health care environments also affect mycobacterial
growth. In the health care setting, some important surface
materials include the pipes comprising the WDS, shower fix-
tures, sink faucets, ice machines, and medical devices. Rele-
vant materials include metals, such as stainless steel (SS) and
copper, and plastics, such as polyvinyl chloride (PVC) and
polycarbonate (PC). In a previous study, M. fortuitum devel-
oped more biofilm on SS, PVC, and PC than on copper or glass
(32). Similar amounts of biomass were measured on PVC and
PC. PC was chosen as a substrate for biofilm growth for this
investigation, along with SS, because of their inclusion in many
medical devices as materials that can be disinfected through
steam autoclaving, the use of ethylene oxide, or irradiation.
Given the association between NTM infections and drinking
water, clinical isolates of NTM would be expected to form
biofilm in WDS. The goal of this study was to determine the
effects of two substrate materials (SS and PC) and nutrient
level (autoclaved municipal tap water and a microbiological
culture medium) on the ability of NTM to form biofilms in a
laboratory model. A second objective of this study was to
evaluate biofilm structure to clarify the role that microcolony
morphologies such as cord and pillar formation may play in the
survival and maintenance of mycobacteria in WDS biofilms.
MATERIALS AND METHODS
Mycobacterial isolates and culture conditions. The 14 mycobacterial isolates
included in this study are listed in Table 1. All non-ATCC (American Type
Culture Collection, Manassas, VA) isolates were obtained from health care-
related outbreaks investigated by the CDC, except for three. Mycobacterium
avium EPA 61151, M. avium EPA 88126, and M. intracellulare EPA 88144 were
obtained during a research study (18). All isolates were cultivated on Middle-
brook 7H10 agar (Becton, Dickinson and Co., Sparks, MD). RGM were incu-
bated at 35°C; M. avium complex (MAC) species isolates (M. avium and M.
intracellulare) were incubated at 37°C.
Method for growth and quantification of biofilms. (i) Preparation of materi-
als. PC and SS grade 316L disks, each measuring 13 mm in diameter and 4 mm
in thickness (BioSurface Technologies, Bozeman, MT), were washed in dilute
laboratory soap (Versa-Clean; Fisher Scientific, Pittsburgh, PA), rinsed at least
five times in reverse osmosis-purified water, rinsed once in 70% ethanol, air
dried, and autoclaved before use.
(ii) Method for growth of biofilms. Biofilms were developed on autoclaved SS
or PC disks incubated in a 24-well tissue culture plate (Corning Incorporated,
Corning, NY), with one disk per well. Each disk was covered with 1.5 ml of either
sterile R2A broth (R2A medium without the agar) (26) or autoclaved PW.
Suspensions of each isolate, collected from the surface of a Middlebrook 7H10
agar plate, were made in Middlebrook 7H9 broth; concentration was determined
by measuring absorbance in a MicroScan turbidity meter (Dade Behring, Deer-
field, IL). Suspended cells were diluted in 0.00425% monopotassium phosphate
(Butterfield’s buffer; Becton, Dickinson and Co.), and approximately 104CFU
were inoculated per well. Inoculation concentration was confirmed by enumer-
ating CFU from a subsample on Middlebrook 7H10 agar. The well plates were
incubated for 72 h at 35°C with gentle shaking on a rocker platform (setting 9, 15
rotations/min; Cole-Parmer, Vernon Hills, IL). At the end of the incubation
time, disks were removed with sterile forceps, dipped three times with gentle up
and down motions in a beaker containing phosphate-buffered saline (PBS) to
remove loosely attached bacteria, and processed for either plate count enumer-
ation or microscopy analysis. Disks destined for microscopic structural analysis
were fixed in 4% formaldehyde and stored at 4°C until examined.
(iii) Biofilm quantification by viable plate counting. After disks were rinsed in
PBS, mycobacteria were enumerated by culturing on Middlebrook 7H10 agar.
Disks were placed in 50-ml polypropylene centrifuge tubes containing 10 ml of
PBS plus 0.1% Tween 80. Bacteria were removed from the disk surface with
three cycles of sonication in a water bath sonicator (frequency of 42 kHz [?6%];
Branson Ultrasonics Corp., Danbury, CT) for 1 min, followed by vortexing
TABLE 1. Clinical, environmental, and reference strains of NTM examined for biofilm formation ability
M. abscessus 23007
M. abscessus BF6
NY type 2 strain; one of two types isolated from patients who received
illegal cosmetic procedures in 2002
Contaminated vials caused multistate outbreak, 1995–1996
Strain associated with contaminated contact lenses used during illegal
Strain associated with contaminated contact lenses used during illegal
Infection associated with indwelling catheter
M. abscessus 4AU
M. chelonae 35752
M. chelonae 34
Vial of adrenal cortex extract
Cornea of LASIK patient
M. chelonae 56Cornea of LASIK patient
M. chelonae 99 Blood of bone marrow transplant
Blood of bone marrow transplant
Leg of nail parlor patient
M. fortuitum 32Infection associated with indwelling catheter
M. fortuitum 89 Caused outbreak of furunculosis of the legs in patients using
Serotype 4 strain; most common type isolated from patients with AIDS
Strain persistent at point of use for at least 26 mo
M. smegmatis 19420
M. avium 91
M. avium EPA 61151
M. avium EPA 88126
M. intracellulare EPA 88144
aLASIK, laser-assisted in situ keratomileusis.
2092 WILLIAMS ET AL.APPL. ENVIRON. MICROBIOL.
(maximum setting) for 30 s. This bacterial suspension was diluted in 0.00425%
monopotassium phosphate and spread on 7H10 agar. Plates were incubated at
35°C or 37°C for rapid growers or slow growers, respectively, until colonies were
observed (3 to 5 days for rapid growers and 10 to 14 days for slow growers,
typically). Removal efficiency was estimated by microscopic examination of 12
isolates grown in triplicate on each substrate under each nutrient condition.
Percent coverage at the substrate level was calculated as described in “Micros-
copy and image analysis” below.
(iv) Biofilm measurement by microscopic methods. Bacterial DNA within
biofilms was stained with Sybr green I (Molecular Probes, Inc., Carlsbad, CA) for
visualization of all bacteria attached to the SS or PC surface. Disks were removed
from formaldehyde, dipped in PBS, and placed in a 5? solution of Sybr green I
in room temperature Tris-EDTA buffer at pH 8 (10 mM Tris-1 mM EDTA;
Mediatech, Inc., Manassas, VA). Staining was carried out for 10 min in the dark,
followed by a 1-min rinse in sterile distilled water. Disks were air dried in the
dark and mounted onto glass microscope slides with double-sided tape. Cover-
slips (22 mm2, with 1-mm thickness) were mounted onto the disks with ProLong
Gold antifade mounting fluid (Molecular Probes, Inc.); the coverslips were taped
down on either side of the disk to maintain stability.
(v) Microscopy and image analysis. Biofilms were examined with a Zeiss
Axioplan epifluorescence microscope equipped with a fluorescein filter set
(480/40 nm excitation, 505-nm long-pass dichroic mirror, 535/50 nm emission)
and an ApoTome, an attachment that reduces light scatter in fluorescing samples
(Carl Zeiss, Inc., Thornwood, NY). Images were obtained through a 40? oil
objective, a Zeiss AxioCam HRm digital camera, and AxioVision imaging soft-
ware. For each microscope field (x-y plane) imaged, optical sections were ob-
tained through the thickness of the biofilm (z), from the disk substratum to the
top of the biofilm, in a Z-stack. At least five Z-stacks were sampled per disk, and
three disks per isolate per condition were examined. To view a representative
section of each disk, at least one image was obtained from the center, and the
others were obtained from different quadrants of the disk. For measuring percent
biofilm removal, 10 substratum images were obtained from each of three disks
for 30 images total per nutrient level and surface material. All images were
exported in gray-scale tagged image file format for analysis using COMSTAT
Image analysis was performed on each image stack with manual thresholding.
Variables such as maximum thickness, biomass estimate, and percent coverage
were calculated using COMSTAT. For percent coverage, the maximum percent
coverage of each image stack was included, rather than that of the layer closest
to the substratum. The exception to this was in calculating the efficiency of
biofilm removal from disks. Percent coverage was measured at the substratum on
disks after sonication/vortexing and compared to those of disks containing bio-
film by subtracting percent coverage after biofilm removal from percent coverage
of intact biofilm, dividing by percent coverage of intact biofilm, and multiplying
(vi) Statistical analysis. Multiple comparisons of nutrient level and substrate
material within each isolate were performed on image analysis data using the
general linear model with the Bonferroni adjustment in SigmaStat (version 3.5;
Systat Software Inc). Statistical comparisons of mycobacterial isolate biofilm
growth detected by culturing were performed with SAS (version 9.1; SAS Insti-
tute Inc.). Significance was indicated if P values were ?0.05.
Enumeration of mycobacterial biofilm by plate counting.
Culturable biofilm was formed by all isolates of Mycobacterium
spp. incubated at each nutrient level and in substratum mate-
rial, except for M. smegmatis grown in low-nutrient conditions
(PW), as measured by plate counting (Fig. 1). The lower de-
tection limit for the plate count method was 100 CFU per disk
or approximately 24 CFU/cm2. In general, nutrient level had
more of an effect on biofilm plate counts than did substrate
material. For all RGM, biofilm plate counts on PC were sig-
nificantly higher under high-nutrient conditions (R2A) than
those in PW (P ? 0.05). Biofilm plate counts were also signif-
icantly higher (P ? 0.05) on steel (SS) in R2A than those in
PW for several of the strains (M. abscessus strains ATCC
23007, BF6, and 4AU; M. chelonae 99; M. fortuitum strains 32
and 89; and M. smegmatis ATCC 19420). PC also yielded
significantly higher (P ? 0.01) biofilm plate counts than SS for
several of the isolates (M. abscessus ATCC 23007, M. chelonae
strains 56 and 99, and M. smegmatis ATCC 19420) in R2A. In
PW, the type of material did not affect RGM biofilm formation
Neither nutrient nor substratum had a significant effect on
viable biofilm counts for the three M. avium strains (M. avium
strains 91, EPA 61151, and EPA 88126), though these organ-
isms tended to develop higher biofilm plate counts than the
RGM in PW. However, M. intracellulare EPA 88144, a MAC
organism, formed significantly more viable biofilm in R2A than
in PW when grown on PC (P ? 0.01).
Description of microcolonies observed by epifluorescence
microscopy. The largest biofilm structures were observed in
samples grown in R2A on PC, displaying several microcolony
morphologies (Fig. 2). For instance, M. abscessus ATCC 23007
formed large, diffuse microcolonies (Fig. 2A). In contrast, M.
abscessus strains BF6 and 4AU and M. chelonae ATCC 35752
formed cord structures in R2A on PC and, to a lesser extent,
on SS. M. abscessus BF6 also formed cords occasionally in PW
on SS. Examples of cording structure are shown in Fig. 2B and
C. M. fortuitum strains 89 and 32 formed tall, narrower micro-
colonies, but M. fortuitum 32 also had curved, fingerlike pro-
jections that extended along the surface at the base of many
structures (Fig. 2E). Other isolates formed moderate to very
sparse biofilms, as represented by M. smegmatis ATCC 19420
and M. avium 91 in Fig. 2D and F, respectively. Although
culturable M. smegmatis ATCC 19420 was not recovered from
disks incubated in PW, small amounts of attached bacteria
were observed directly (data not shown). For EPA strains M.
avium EPA 88126 and M. intracellulare EPA 88144, it was
necessary to scan much of each disk to find any evidence of
cells or microcolonies. Image stacks of these two mycobacteria
were not analyzed quantitatively.
Image analysis. With the exception of M. chelonae 56, bio-
mass for the RGM was highest on PC in R2A and lowest on PC
in PW. The high-nutrient biomass measurements are in agree-
ment with the biofilm plate counts for PC and SS, whereas in
PW, the substratum had less of an effect on plate count than on
biomass (Table 2). Maximum thickness (Fig. 3) and maximum
percent coverage (Table 3) for all RGM, with the exception of
M. chelonae 56, were also highest for biofilms grown on PC in
R2A (P ? 0.05). Biomass and percent coverage for M. avium
EPA 61151 were also significantly higher on PC in R2A. These
results suggest that quantitative structural analysis methods
(i.e., biomass, maximum thickness, and maximum percent cov-
erage) can predict viable count results for these organisms, that
nutrient level may predict attachment and biofilm formation,
and that PC is generally most conducive to biofilm formation.
Image analysis parameters and plate counts were compared
among non-cord-forming M. abscessus ATCC 23007 and cord-
forming M. abscessus BF6 and 4AU biofilms incubated in R2A
on PC. Plate counts for M. abscessus ATCC 23007 were not
significantly higher than those for M. abscessus BF6, despite
vast differences in microcolony structure. However, M. absces-
sus ATCC 23007 produced significantly more biomass and
maximum percent coverage than M. abscessus BF6 (P ? 0.05).
Maximum thickness measurements for M. abscessus ATCC
23007, however, were not significantly different than those for
M. abscessus BF6. M. abscessus 4AU, another cord former,
VOL. 75, 2009BIOFILM FORMATION BY NONTUBERCULOUS MYCOBACTERIA2093
FIG. 1. Culturable biofilm of Mycobacterium clinical isolates and reference strains after growth for 3 days at 35°C in R2A medium (a) or PW
(b) on PC or SS disks. Data were transformed by addition of 1 and converting values to log10values. The detection limit was 100 CFU/disk or
approximately 24 CFU/cm2(n ? 3). Strain identifiers are as follows: Mab23, M. abscessus ATCC 23007; MabBF6, M. abscessus BF6; Mab4AU,
M. abscessus 4AU; Mch35, M. chelonae ATCC 35752; Mch34, M. chelonae 34; Mch56, M. chelonae 56; Mfo32, M. fortuitum 32; Mfo89, M. fortuitum
89; Msm19, M. smegmatis ATCC 19420; Mav91, M. avium 91; Mav61, M. avium EPA 61151; Mav26, M. avium EPA 88126; and Min44, M.
intracellulare EPA 88144.
2094 WILLIAMS ET AL.APPL. ENVIRON. MICROBIOL.
FIG. 2. Compiled biofilm images of six mycobacteria isolates grown in R2A medium on PC disks. Biofilm was stained with Sybr green I before
image stacks were obtained. (A) M. abscessus ATCC 23007; (B) M. abscessus BF6; (C) M. chelonae ATCC 35752; (D) M. smegmatis ATCC 19420;
(E) M. fortuitum 32; (F) M. avium 91.
VOL. 75, 2009BIOFILM FORMATION BY NONTUBERCULOUS MYCOBACTERIA 2095
developed significantly more culturable bacteria than M.
abscessus ATCC 23007 (P ? 0.01). However, no significant
differences between the biomass, maximum percent coverage,
or maximum thickness measurements of M. abscessus ATCC
23007 and those of M. abscessus 4AU were observed.
Biofilm removal efficiency was estimated for 12 isolates
grown in triplicate on each substrate and under each nutrient
condition by comparing the percent coverage at the substrate
level after sonication/vortexing to the percent coverage of in-
tact biofilm. The majority of isolates under each nutrient level
and substratum demonstrated ?95% biofilm removal effi-
ciency, and all but four calculations were above 90%. Catego-
rized by growth condition, isolates incubated in R2A on PC
produced a removal efficiency ranging from 100% (99.95%) for
M. abscessus ATCC 23007 to 90.0% for M. chelonae ATCC
35752; in R2A on SS, the efficiency ranged from 99.6% for M.
abscessus ATCC 23007 to 83.4% for M. smegmatis ATCC
19420; in PW on PC, the efficiency ranged from 99.9% for M.
chelonae 99 to 90.1% for M. abscessus 4AU; and in PW on SS,
the recovery efficiency ranged from 99.6% for M. fortuitum 32
to 74.6% for M. chelonae 56.
All clinical isolates of NTM formed detectable biofilm in
PW. The highest amount of culturable biofilm was formed in
PW in 3 days by the clinical isolate M. avium 91, a slow grower
that takes more than 7 days to form colonies on laboratory
medium. Ironically, the two clinical isolates M. avium 91 and
EPA 88126 formed more culturable biofilm under each con-
dition than the environmental isolate M. avium EPA 61151. A
similar result was obtained by Carter et al. during a comparison
of biofilm formation abilities of M. avium isolates from AIDS
patients (4). This suggests that M. avium may be better adapted
for growth in PW systems than some species of RGM. How-
ever, with the challenges in detecting and accurately enumer-
ating NTM in environmental samples (7, 8, 18), more infor-
mation is required to confirm this in PW supplies. Although
many studies have linked MAC in PW supplies to human
infection (1, 10, 12), outbreak investigations have found health
care-related infections caused increasingly by waterborne
RGM (9, 11, 20). In this study of single-species biofilms in a
simple batch system, RGM biofilm growth was highly influ-
enced by nutrient level, with PW restricting biofilm growth. It
may be that in a multispecies PW biofilm, nutrient exchange
with other organisms may enhance RGM growth. This should
be determined in concert with the study of possible control
measures for these organisms. Since mycobacteria demon-
strate high tolerance for chlorine disinfectants typically present
in PW (3, 4), especially when in biofilms (28), the best infection
control intervention may be provided by measures taken at the
point of use by the individual or health care professionals.
Examples of infection prevention include preventing exposure
of wound or catheter entry sites to PW during bathing, ade-
quately maintaining ice machines that make ice intended for
patient consumption, and performing point-of-use treatment
on PW that will reach patients with compromised immune
TABLE 2. Biomass estimation of 10 RGM and two M. avium isolatesa
Mean biomass vol (?m3/?m2) (SD) grown withb:
M. abscessus ATCC 23007
M. abscessus BF6
M. abscessus 4AU
M. chelonae ATCC 35752
M. chelonae 34
M. chelonae 56
M. chelonae 99
M. fortuitum 32
M. fortuitum 89
M. smegmatis ATCC 19420
M. avium 91
M. avium EPA 61151
aBiomass estimation was calculated for 10 RGM and two M. avium isolates incubated in two nutrient levels and on two substratum materials.
bMean value was obtained from five image stacks from each of three disks for most isolates (n ? 15). R2A and PW were the media (nutrient levels) used; PC and
SS were the surface materials. SD, standard deviation.
FIG. 3. Maximum biofilm thickness of 10 RGM and two M. avium
isolates incubated at two nutrient levels and on two substratum mate-
rials. Maximum thickness was significantly higher for all RGM grown
on PC substratum (P ? 0.05) in R2A (high-nutrient condition) than for
those grown on PC in PW (low-nutrient condition). Substratum ma-
terial was significant for all RGM grown in R2A, except for M. chelo-
nae 56 and for M. avium isolates 91 and EPA 61151, at either nutrient
level. Nutrient level did not significantly affect M. avium biofilm thick-
ness (n ? 15 for most isolates). Legend abbreviations for biofilm
grown: R2A-PC, R2A broth on PC; R2A-SS, R2A broth on SS; PW-
PC, PW on PC; PW-SS, PW on SS. Strain identifiers on the x axis are
defined in the legend to Fig. 1.
2096 WILLIAMS ET AL.APPL. ENVIRON. MICROBIOL.
Temperature may be an important factor in determining
mycobacterial growth in PW biofilms (29). Although the rec-
ommended optimal incubation temperature for cultivation of
most RGM is 28 to 30°C (13), all of the isolates included in this
study are capable of growth in most of the temperature ranges
found in interior plumbing, including the incubation tempera-
ture of 35°C used during this study.
Previous research has demonstrated a link between myco-
bacterial virulence and cord-forming ability (19). In that study,
virulent M. abscessus strains formed cord structures, produced
less biofilm, and produced less GPL than a nonvirulent strain.
The three M. abscessus isolates examined in the current study
all formed biofilm in each nutrient level on both surface
materials. The numbers of culturable biofilm bacteria were
equivalent or higher in the rough, cording, clinical isolates
than in the smooth, noncording isolate. By contrast, quantita-
tive analysis of biofilm structure by biomass estimation and
maximum percent coverage demonstrated that the smooth M.
abscessus 23007 produced more biofilm than the cord-forming
M. abscessus BF6. This suggests that the cording formations
are more compact, containing more cells per unit volume than
the more-diffuse biofilm formed by smooth M. abscessus. These
results suggest that in at least some cord-forming M. abscessus
isolates, biofilm formation ability cannot be predicted by col-
ony morphology. This indicates that the relationship among
cording, biofilm formation, and virulence is not completely
characterized. To understand these relationships more fully, a
larger sampling of M. abscessus strains and other cord-forming
species should be examined. Additionally, more-complete
analysis of components necessary for biofilm formation in
NTM is required.
Biomass and other image analysis measurement parameters
may differ from culturable bacteria numbers for several rea-
sons. Biofilm was imaged after staining with Sybr green I, a
“total” stain of double-stranded DNA that does not indicate
cell viability. It is possible that some attached mycobacteria did
not maintain viability on PC or SS during the 3-day incubation.
Another possibility is that during removal of mycobacteria
from disks for culturing, some biofilm remained clumped, even
after vortexing and sonication, so that the number of culturable
bacteria could have been underestimated. However, the low
variance in CFU per isolate demonstrates that separation of
mycobacteria during sample processing was consistent, if not
100% complete. Also, cell size may vary among species, strains,
or even within a strain grown under different conditions. When
measuring small microcolonies, as is the case for many of the
mycobacteria grown in PW, a small difference in individual cell
size and shape will greatly influence the microcolony dimen-
sions. Although extracellular material was not visualized or
quantified in this investigation, its presence could also affect
Under low-nutrient conditions in PW, the low growth num-
bers and microcolony formation were at the lower limit of what
may be considered biofilm versus attached cells. While cultur-
able numbers of bacteria recovered from disk surfaces were
high enough to indicate biofilm growth in PW for some iso-
lates, particularly for MAC, direct observation of biofilms of-
ten demonstrated the presence of single cells or very small
clusters of bacteria. This limited the use of image analysis
under these conditions, since complex biofilm structure was
absent. This may be due to the short incubation time for
biofilm development in this experiment, so that young biofilm
was observed rather than mature, complex biofilm. This was
demonstrated in a comparison of percent surface coverage and
maximum percent coverage to determine if structure in the x-y
plane differed between the substratum level and elsewhere in
the biofilm. The high correlation between the two measure-
ments indicated either that the maximum percent coverage
occurred at the substratum level, as seen in some samples (data
not shown), or that similar percent coverage existed through-
out the height of the biofilm. This corresponds to the simple
microcolony structure observed for many of the mycobacteria,
particularly under low-nutrient incubation.
Generally, more variability was observed in direct measure-
ments than in plate counts, suggesting that more observations
per sample are required to make direct observation consistent.
This would be possible if a completely automated imaging
system could be employed.
M. smegmatis, frequently chosen as a model organism for
biofilm research (24, 27), may not be the ideal Mycobacterium
species model for the study of biofilm formation in PW, since
no culturable M. smegmatis was recovered in PW biofilms in-
cubated under the conditions in this study.
All mycobacterial clinical isolates formed biofilm under
high- and low-nutrient conditions. Nutrient level was a more
important factor than the two substrate materials tested for
microcolony formation by RGM in this study. Although other
studies have demonstrated a difference in biofilm formation
ability between cord- and non-cord-forming M. abscessus
strains, the three M. abscessus isolates included in this study
formed roughly equivalent amounts of culturable biofilm, de-
spite the measurable differences in microcolony morphology.
Additional study of M. abscessus cording ability and pathoge-
nicity, as well as its ecology in PW supplies, may lead to a
better understanding of the role played by M. abscessus in
health care-related infections. When most measurements
are considered, nutrient level did not significantly affect
TABLE 3. Maximum percent coverage of biofilm in the x-y plane of
each image stack
Mean % coverage (SD) grown witha:
R2A-PC R2A-SSPW-PC PW-SS
M. chelonae 34
M. chelonae 56
M. chelonae 99
M. fortuitum 32
M. fortuitum 89
M. avium 91
M. avium EPA
9.7 (6.6)5.9 (5.2) 0.096 (0.14) 0.54 (0.39)
4.8 (4.1) 0.69 (0.76)0.074 (0.10)0.30 (0.30)
5.8 (3.1)3.3 (3.7) 0.058 (0.083) 0.50 (0.75)
0.85 (0.37)0.72 (0.59)0.031 (0.028) 0.072 (0.074)
aMean values were obtained from five image stacks from each of three disks
for most isolates (n ? 15). R2A and PW were the media (nutrient levels) used;
PC and SS were the surface materials. SD, standard deviation.
VOL. 75, 2009 BIOFILM FORMATION BY NONTUBERCULOUS MYCOBACTERIA2097
MAC biofilm development. This study indicated that M. Download full-text
avium is better equipped to grow in warm PW supplies than
RGM is, offering an explanation for the greater occurrence
of disease caused by MAC.
The findings and conclusions in this report are those of the authors
and do not necessarily represent the views of the U.S. Centers for
Disease Control and Prevention or the Environmental Protection
Agency. Use of trade names and commercial sources are for identifi-
cation only and do not constitute endorsement by the Public Health
Service, the Centers for Disease Control and Prevention, or the En-
vironmental Protection Agency.
1. Aronson, T., A. Holtzman, N. Glover, M. Boian, S. Froman, O. G. W. Berlin,
H. Hill, and G. N. Stelma, Jr. 1999. Comparison of large restriction frag-
ments of Mycobacterium avium isolates recovered from AIDS and non-AIDS
patients with those of isolates from potable water. J. Clin. Microbiol. 37:
2. Bardouniotis, E., H. Ceri, and M. E. Olson. 2003. Biofilm formation and
biocide susceptibility testing of Mycobacterium fortuitum and Mycobacterium
marinum. Curr. Microbiol. 46:28–32.
3. Carson, L. A., N. J. Petersen, M. S. Favero, and S. M. Aguero. 1978. Growth
characteristics of atypical mycobacteria in water and their comparative re-
sistance to disinfectants. Appl. Environ. Microbiol. 36:839–846.
4. Carter, G., M. Wu, D. C. Drummond, and L. E. Bermudez. 2003. Charac-
terization of biofilm formation by clinical isolates of Mycobacterium avium.
J. Med. Microbiol. 52:747–752.
5. Catherinot, E., J. Clarissou, G. Etienne, F. Ripoll, J.-F. Emile, M. Daffe, C.
Perronne, C. Soudais, J.-L. Gaillard, and M. Rottman. 2007. Hypervirulence
of a rough variant of the Mycobacterium abscessus type strain. Infect. Immun.
6. Chadha, R., M. Grover, A. Sharma, A. Lakshmy, M. Deb, A. Kumar, and G.
Mehta. 1998. An outbreak of post-surgical wound infections due to Myco-
bacterium abscessus. Pediatr. Surg. Int. 13:406–410.
7. Covert, T. C., M. R. Rodgers, A. L. Reyes, and G. N. Stelma, Jr. 1999.
Occurrence of nontuberculous mycobacteria in environmental samples.
Appl. Environ. Microbiol. 65:2492–2496.
8. Dailloux, M., C. Laurain, M. Weber, and P. Hartemann. 1999. Water and
nontuberculous mycobacteria. Water Res. 33:2219–2228.
9. De Groote, M. A., and G. Huitt. 2006. Infections due to rapidly growing
mycobacteria. Clin. Infect. Dis. 42:1756–1763.
10. du Moulin, G. C., K. D. Stottmeier, P. A. Pelletier, A. Y. Tsang, and J. Hed-
ley-Whyte. 1988. Concentration of Mycobacterium avium by hospital hot
water systems. JAMA 260:1599–1601.
11. Falkinham, J. O., III. 1996. Epidemiology of infection by nontuberculous
mycobacteria. Clin. Microbiol. Rev. 9:177–215.
12. Glover, N., A. Holtzman, T. Aronson, S. Froman, O. G. W. Berlin, P.
Dominguez, K. A. Kunkel, G. Overturf, G. Stelma, Jr., C. Smith, and M. A.
Yakrus. 1994. The isolation and identification of Mycobacterium avium com-
plex (MAC) recovered from Los Angeles potable water, a possible source of
infection in AIDS patients. Int. J. Environ. Health Res. 4:63–72.
13. Griffith, D. E., T. Aksamit, B. A. Brown-Elliott, A. Catanzaro, C. Daley, F.
Gordin, S. M. Holland, R. Horsburgh, G. Huitt, M. F. Iademarco, M.
Iseman, K. Olivier, S. Ruoss, C. F. von Reyn, R. J. Wallace, Jr., and K.
Winthrop on behalf of the ATS Mycobacterial Diseases Subcommittee. 2007.
An official ATS/IDSA statement: diagnosis, treatment, and prevention of
nontuberculous mycobacterial diseases. Am. J. Respir. Crit. Care Med. 175:
14. Hall-Stoodley, L., and H. Lappin-Scott. 1998. Biofilm formation by the
rapidly growing mycobacterial species Mycobacterium fortuitum. FEMS Mi-
crobiol. Lett. 168:77–84.
15. Hall-Stoodley, L., C. W. Keevil, and H. M. Lappin-Scott. 1999. Mycobacte-
rium fortuitum and Mycobacterium chelonae biofilm formation under high
and low nutrient conditions. J. Appl. Microbiol. 85:60S–69S.
16. Hall-Stoodley, L., O. S. Brun, G. Polshyna, and L. P. Barker. 2006. Myco-
bacterium marinum biofilm formation reveals cording morphology. FEMS
Microbiol. Lett. 257:43–49.
17. Heydorn, A., A. T. Nielsen, M. Hentzer, C. Sternberg, M. Givskov, B. K.
Ersboll, and S. Molin. 2000. Quantification of biofilm structures by the novel
computer program COMSTAT. Microbiology 146:2395–2407.
18. Hilborn, E. D., M. A. Yakrus, T. C. Covert, S. I. Harris, S. F. Donnelly, M. T.
Schmitt, S. Toney, S. A. Bailey, and G. N. Stelma, Jr. 2008. Molecular
comparison of Mycobacterium avium isolates from clinical and environmen-
tal sources. Appl. Environ. Microbiol. 74:4966–4968.
19. Howard, S. T., E. Rhoades, J. Recht, X. Pang, A. Alsup, R. Kolter, C. R.
Lyons, and T. F. Byrd. 2006. Spontaneous reversion of Mycobacterium
abscessus from a smooth to a rough morphotype is associated with reduced
expression of glycopeptidolipid and reacquisition of an invasive phenotype.
20. Kline, S., S. Cameron, A. Streifel, M. A. Yakrus, F. Kairis, K. Peacock, J.
Besser, and R. C. Cooksey. 2004. An outbreak of bacteremias associated with
Mycobacterium mucogenicum in a hospital water supply. Infect. Control
Hosp. Epidemiol. 25:1042–1049.
21. Le Dantec, C., J.-P. Duguet, A. Montiel, N. Dumoutier, S. Dubrou, and V.
Vincent. 2002. Occurrence of mycobacteria in water treatment lines and in
water distribution systems. Appl. Environ. Microbiol. 68:5318–5325.
22. Limia, A., F. J. Sangari, D. Wagner, and L. E. Bermudez. 2001. Character-
ization and expression of secA in Mycobacterium avium. FEMS Microbiol.
23. Lowry, P. W., C. M. Beck-Sague, L. A. Bland, S. M. Aguero, M. J. Arduino,
A. N. Minuth, R. A. Murray, J. M. Swenson, and W. R. Jarvis. 1990. Myco-
bacterium chelonae infection among patients receiving high-flux dialysis in a
hemodialysis clinic in California. J. Infect. Dis. 161:85–90.
24. Martinez, A., S. Torello, and R. Kolter. 1999. Sliding motility in mycobac-
teria. J. Bacteriol. 181:7331–7338.
25. Norton, C. D., M. LeChevallier, and J. O. Falkinham III. 2004. Survival of
Mycobacterium avium in a model distribution system. Water Res. 38:1457–
26. Reasoner, D. J., and E. E. Geldreich. 1985. A new medium for the enumer-
ation and subculture of bacteria from potable water. Appl. Environ. Micro-
27. Recht, J., A. Martinez, S. Torello, and R. Kolter. 2000. Genetic analysis of
sliding motility in Mycobacterium smegmatis. J. Bacteriol. 182:4348–4351.
28. Steed, K. A., and J. O. Falkinham III. 2006. Effect of growth in biofilms on
chlorine susceptibility of Mycobacterium avium and Mycobacterium intracel-
lulare. Appl. Environ. Microbiol. 72:4007–4011.
29. Torvinen, E., M. J. Lehtola, P. J. Martikainen, and I. T. Miettinen. 2007.
Survival of Mycobacterium avium in drinking water biofilms as affected by
water flow velocity, availability of phosphorus, and temperature. Appl. En-
viron. Microbiol. 73:6201–6207.
30. Torvinen, E., S. Suomalainen, M. J. Lehtola, I. T. Miettinen, O. M. Zacheus,
L. Paulin, M.-L. Katila, and P. J. Martikainen. 2004. Mycobacteria in water
and loose deposits of drinking water distribution systems in Finland. Appl.
Environ. Microbiol. 70:1973–1981.
31. Tsintzou, A., A. Vantarakis, O. Pagonopoulou, A. Athanassiadou, and M.
Papapetropoulou. 2000. Environmental mycobacteria in drinking water be-
fore and after replacement of the water distribution network. Water Air Soil
32. Williams, M. M., and R. M. Donlan. 2005. Mycobacterium fortuitum biofilm
formation on water distribution pipe materials. In Distribution system bio-
film control. Proceedings of the American Water Works Association Water
Quality Technology Conference and Exposition, Que ´bec City, Canada, 6 to
10 November 2005.
33. Winthrop, K. L., M. Abrams, M. A. Yakrus, I. Schwartz, J. Ely, D. Gillies,
and D. J. Vugia. 2002. An outbreak of mycobacterial furunculosis associated
with footbaths at a nail salon. N. Engl. J. Med. 346:1366–1371.
34. Yamazaki, Y., L. Danelishvili, M. Wu, E. Hidaka, T. Katsuyama, B. Stang,
M. Petrofsky, R. Bildfell, and L. E. Bermudez. 2006. The ability to form
biofilm influences Mycobacterium avium invasion and translocation of bron-
chial epithelial cells. Cell. Microbiol. 8:806–814.
2098 WILLIAMS ET AL.APPL. ENVIRON. MICROBIOL.