United States
Environmental Protection
Agency
National Health and Environmental
Effects Research Laboratory
Gulf Breeze, FL 32561
Research and Development
EPA/600/S-98/008 July 1998
ENVIRONMENTAL
RESEARCH BRIEF
In-situ Bioremediation of Trichloroethylene Using Burkholderia cepacia G4
PR1: Analysis of Transport Parameters for Risk Assessment
J.R. Lawrence1 and M.J. Hendry2
Abstract
Transport of bacteria through geologic media may be viewed
as being governed by sorption-desorption reactions. In this
investigation, four facets of the process were examined: (I) the
impact of sorption on bacterial transport under typical ground
water flow velocities and at different transport distances, (II)
the impact of water velocity on bacterial sorption and thus
transport, (III) the impact of other species of bacteria on bacte-
rial transport through porous media, and (IV) the ability of the
bacterium to bind to biofilms and integrate into existing micro-
bial communities. The experiments used three bacteria, Kleb-
siella oxytoca, Burkholderia cepacia G4 PR1, and Pseudomonas
#5, a subsurface isolate. The modeling results suggested that
irreversible sorption (kirr) was a function of mean transit time (t0)
whereby the product (t0-kirr), which is defined as the equivalent
irreversible sorption parameter (A), was constant (mean value
of 3.36) at the scales of this investigation. The migration of K.
oxytoca was predictable under a broad range in ground-water
velocities, and the transport of G4 PR1 was predictable at
velocities >3.5 cm-hr1. Sorption characteristics were bacteria-
specific, and bacterial interactions during transport of bacteria
included: no impact, increased/decreased peak concentration
and tailing, and displacement of resident bacteria by planktonic
populations during attachment. Thus the transport of G4 PR1
may be facilitated by the presence of specific bacteria and
retarded by the presence of others.
Introduction
Early research on the fate and transport of bacteria in geologic
media was stimulated by traditional concerns for disease, dis-
'National Hydrology Research Institute, 11 Innovation Blvd., Saskatoon,
Saskatachewan, S7N 3H5, Canada
department of Geological Sciences, University of Saskatachewan, Saskatoon,
Saskatchewan, S7N OWO, Canada /TV
«3D Printed on Recycled Paper
posal of sewage, and the use of infiltration to purify microbially
and organically contaminated waters. More recently, issues
such as in situ bioremediation, facilitated transport of radionu-
clides and organics, release of genetically engineered microor-
ganisms, the use of bacteria for selective plugging in'
environmental applications, and the use of bacteria for
microbially-enhanced oil recovery have generated interest in
the area. There is a need to accurately predict the rate and
extent of microbial transport through geologic media. In all
cases, our ability to predict the migration of bacteria in geologic
media is limited by our inability to quantify the transport pro-
cesses.
Transport of bacteria through geologic media may be viewed
as being governed by sorption-desorption reactions. Chemical
and physical factors influence the sorption of bacteria to geo-
logic and other media. These factors include: the nature of the
porous medium, such as soil type, grain size, clay and organic
matter content, iron coatings, mineralogy, the chemistry of the
solute, pH, ionic strength, and the presence of surfactants. In
addition, a range of microbiological factors also influence sorp-
tion. These include: cell viability, nutritional status, electrostatic
charge on the cell surface, hydrophobicity, hydrophilicity, cell
size and cell shape, predation, parasitism, motility and chemo-
taxis, and cell cycle events which may influence any of these
factors.
Studies of bacterial transport generally use bacterial break-
through curves to assess the controls on bacterial transport in
geologic media. A few studies have used direct microscopic
observation of bacterial sorption at the solid-liquid or gas-water
interface to examine bacterial sorption at the pore scale.
Currently, bacterial transport models are based on the advec-
tive-dispersion equation. This equation has been modified to
-------
include a variety of terms to reflect biological processes such
as growth, death and predation. Despite the large number of
models and approaches, there are few empirical tests of any
models because of the inherent difficulty in understanding and
quantifying all the relevant parameters. The theories available
are incomplete and the tests of existing models against labora-
tory or field data are few. Recent efforts have concentrated on
advective transport either in the absence of biological pro-
cesses, or through the incorporation of phenomenological coef-
ficients. Thus, in the field of bacterial transport, there is a need
for fundamental empirical studies, theoretical development, and
the creation and testing of numerical models.
Approach
/. The impact ofsorptfon on bacterial transport
under typical ground water flow velocities and
at different transport distances.
This objective was accomplished by: (1) obtaining visual infor-
mation on the sorption of the bacteria to the granular, porous
medium; (2) conducting bacterial breakthrough experiments at
three scales of investigation (3.8,10, and 40 cm long columns)
to further characterize the bacterial sorption processes; and (3)
mathematically modeling the bacterial concentration curves
measured in the outflow from the columns as a function of
time. All experiments were conducted under conditions that
eliminated the influence of biological processes such as preda-
tion, growth, and death of the bacteria, thereby allowing the
study of the effects of sorption on microbial transport. A me-
dium-to-coarse grained silica sand was used as the geologic
medium. To approximate typical saturated ground water flow
velocities in the saturated sand, velocities ranging from 0.1 to
0.2 nvd"1 were used in the visualization and column experi-
ments.
II. The impact of water velocity on bacterial
sorption and thus transport.
The goal of this study was accomplished by conducting a
series of bacterial breakthrough experiments (in quadruplicate)
for each bacterium at four linear ground-water velocities rang-
ing from 0.5 to 14 cm-hr'. The resulting bacterial breakthrough
data, measured in the outflow from the columns as a function
of time, were modeled in conjunction with an associated con-
servative tracer to determine the sorption parameters for each
bacteria as a function of velocity.
///. The impact of other species of bacteria on
bacterial transport through porous media.
To investigate these interactions we carried out column break-
through experiments using a well-characterized silica sand and
3 bacterial species with demonstrated differences in transport
characteristics. In the present study, we sought to demonstrate
whether there was a systematic effect of the presence of one
bacteria on the transport parameters of the second during
simultaneous transport and sequential transport experiments.
The only variable in the experimental design was the presence
or sequence of addition of the selected bacteria to the column.
The experiments reported here used three bacteria, Klebsiella
oxytoca, Burkholderia cepacia G4 PR1, and Pseudomonas #5,
a subsurface isolate, to assess the effect of cell type on
sorption. K. oxytoca is a non-motile gram-negative rod approxi-
mately 1.4 ±0.8 urn x 0.8 ±0.1 fim. G4 PR1 is a Tn5 insertion
mutant (2 ±0.5 urn x 0.9 ±0.1 urn) which constitutively ex-
presses the toluene-ortho-monooxygenase responsible for TCE
degradation. As such, this organism is a candidate for use in
remediation of contaminated aquifers, and determination of its
transport characteristics is a prerequisite for its application in
the subsurface. Pseudomonas #5 is a deep subsurface isolate
(National Hydrology Research Institute, Saskatoon, SK, Canada)
3.1 + 1.4 urn x 1.1 ± 0.1 urn that produces a'surfactant and
may have applications in bioremediation protocols, and thus its
transport parameters and impact on transport of other bacteria
is of concern. ;
IV. The ability of the bacterium to bind to
biofilms and integrate into existing microbial
communities.
The objective of this study was to utilize confocal scanning
laser microscopy (CSLM) and Mab techniques to assess the
survival and integration of G4 PR1 into this community, when
TCE was provided as an exogenous carbon source with and
without provision of a co-metabolite. G4 PR1 was introduced
into a continuously-fed culture with a defined 9 member
degradative community previously isolated from soils. This
simulated conditions that could be encountered within an in
situ subsurface reactor or a biocassette.
Materials and Methods
Porous Medium and Aqueous Solutions
A medium-coarse grained commercially available silica sand
was used to create the porous medium. Sieve analysis showed
that 77% of the sand was in the size range of 500 to 1000 urn
diameter. The remainder of the sand was in the size range of
250 to 500 urn diameter. The total inorganic carbon content
(TIC) of the sand was 0.11% (ww1) and total organic carbon
(TOC) was below detection (<0.1 ppm).
An artificial ground water (AGW) was used as the solute in all
experiments. Chemical concentrations (mg-L1) of the AGW
were Ca=22.2, Mg=3.43, K=1.41, Cl=0.1, PO4=0.10, alkalinity
(as CaCO3)=66.65, and pH=7.3. NO and SO4 concentrations
were both <1.0 mg-L1. Cl (100 mg-U) was added to the AGW
to act as a conservative tracer (AGWt). The ionic strengths of
the AGW and AGWt were determined to be 0.002 and 0.005,
respectively.
Column Experiments
Column experiments were conducted at three scales of investi-
gation (4.7 cm dia x 3.8 cm long, 4.7 cm dia x 10 cm long and
5.0 cm dia x 40 cm long) to investigate the sorption controls on
the transport of K. oxytoca. All experiments were conducted in
quadruplicate. Each replicate experiment was conducted on
freshly repacked columns. Silica sand was emplaced under
standing water to minimize air entrapment and tamped down
with a glass rod during filling to attain a bulk density of about
1.5 g-crrr3 and a porosity of about 0.4.
Constant bottom-to-top water fluxes (to minimize the effects of
gravity) were maintained through the columns. Approximately
1-4 pore volumes (PV) of AGW were flushed through the
columns prior to introducing 0.8 PV of the bacteria suspended
in AGWt. Bacterial suspensions (C0) were pulsed through the
columns. After introducing the bacterial suspensions, AGW
was flushed through the columns for approximately 10 PV.
Column effluent samples were collected in sterile vials using a
fraction collector. The number of bacteria in the effluent frac-
-------
lions were determined by plate counts on 20% brain-heart
infusion (BHI) agar with 50 mg-L-1 kanamycin added. Chloride
concentrations were measured using a Ag/AgCI chloride elec-
trode calibrated against Cl analyses determined by ion chro-
matography.
Bacterial Culture, Media, and Growth Rates
Three bacteria were used to assess the interactions between
species during bacterial transport: Burkholderia cepacia G4
PR1, Pseudomona sp#5 and Klebsiella oxytoca. To facilitate
recovery and enumeration of the bacteria, a K. oxytoca with an
antibiotic resistance marker (kanamycin) was prepared by
transposon 5 mutagenesis. K. oxytoca was routinely cultured in
half strength brain heart infusion (BHI) media at pH 7.4. Flasks
containing 500 ml_ of 50% BHI were inoculated from an iso-
lated colony of the kanamycin resistant K. oxytoca growing on
BHI agar plates. The flasks were incubated for 48 hr on an
orbital shaker at 23+2°C. Cultures were harvested by centrifu-
gation (t=10 min at 25,OOOxG), resuspended, and washed with
AGWt three times to remove all available carbon.
G4 PR1 and Pseudomonas #5 were prepared in the same
manner as K. oxytoca, with the exception that the growth
medium used was 10% strength tryptic soy broth or agar as
appropriate. The cells were added to the AGWt to produce
input solutions with similar bacterial concentrations. The con-
centrations of viable cells in the input solutions were monitored
for 200 hr by plating dilutions on selective media. All solutions
were used immediately after preparation. All stock cultures
were stored at -80°C in half strength BHI in 15% glycerol.
Results and Discussion
/. Effect of Scale.
Experiments were performed to better understand bacterial
sorption in saturated porous media. The experimental approach
consisted of visualization of the sorption-desorption process
using a CSLM, column experiments at three scales (4.7 cm
dia. x 3.8 cm, 4.7 cm dia x 10 cm and 5.0 cm dia x 40 cm), and
numerical modeling of bacterial transport and sorption. Kleb-
siella oxytoca, a non-motile gram-negative rod was used in the
experiments. The porous medium consisted of a medium-to-
coarse grained commercially-available silica sand. To approxi-
mate natural ground water flow conditions through sands, all
experiments were conducted at typical ground water velocities
(0.4 to 0.8 crn-hr1). Because the focus of the present study
was sorption, the impacts of other biochemical and biological
processes on bacterial transport were eliminated.
In the column experiments, the peak concentration of bacteria
decreased with increased scale of migration. The bacterial
pulse was sharp after migrating 3.8 cm, broadened after mi-
grating 10 cm, and was poorly-developed at a distance of 40
cm. Well defined tailing was also observed in all three column
experiments. The outflow over inflow concentrations (C/Co) of
the tails were in the same order of magnitude (1Q-3) at all three
scales after ten times the mean transit time(10to). The attenua-
tion of the peaks and the presence of well-defined tailing
supported the presence of both an irreversible (kj and kineti-
cally controlled reversible sorption (kr) of K. oxytoca, as ob-
served in the visualization experiment. In the case of reversible
sorption, the rate of forward sorption (k() was not the same as
the rate for kr.
Best-fit determinations at all three scales indicated that the
reversible sorption parameters were independent of scale (k,=0.1
hr1 and kr=0.02 hr1), while kirr decreased with increasing scale
(0.6, 0.13 and 0.062 hr1 for the 3.8, 10, and 40 cm long
columns, respectively, Table 1). The modeling results sug-
gested that kirr was a function of mean transit time (t0), whereby
the product (t0-kirr), which is defined as the equivalent irrevers-
ible sorption parameter (A), was constant (mean value of 3.36)
at the three scales of investigation. The use of A may prove a
useful approximation in predicting the values of kiir at different
scales and different water flow velocities (from a known or
assumed t0).
Table 1. Sorption parameters determined from best-fit simulations
of the bacterial breakthrough data from the columns packed
with silica sand. A, the parameter calculated from the best-
fit simulations, is also tabulated.
Column length,
L(cm)
3.8
10.0
40.0
Fitted Ic
(hr')
0.600
0.130
0.062
Fitted k,
(hr')
0.1
0.1
0.1
Fitted k
(hr')
0.02
0.02
0.02
Calculated
A value*
3.27
3.60
3.20
As stated in the introduction, our ability to predict microbial
attachment for transport studies is poorly understood. Addi-
tional experimentation is required to address the feasibility of
using sorption parameters to approximate the transport of sev-
eral strains of bacteria in the subsurface. The apparent scale
dependency of kte and our inability to simulate the peak con-
centrations indicated that the sorption process was more com-
plex than can be explained by assuming that the attachment
and detachment of bacteria were solely a function of attached
concentration. Even in the well-controlled environment used in
these experiments, it is evident that other controls on sorption
need to be considered. For example, the detachment of bacte-
ria may also be a function of bacterial residence time on the
grain surfaces.
//. Effect of Velocity.
Column experiments were conducted in 3.3 cm ID x 11.4 cm
long columns to investigate the effect of water velocity on
bacterial sorption (and thus transport) through saturated po-
rous media. The experiments reported here used two bacteria,
Klebsiella oxytoca and Burkholderia cepacia G4 PR1, to as-
sess the effect of cell type on sorption. The porous medium
consisted of a medium-to-coarse grained commercially-avail-
able silica sand. To approximate ground water flow conditions
through sands, experiments were conducted at natural ground-
water velocities (0.5 to 0.6 cm-hr1) and at 3, 6-7, and 13-14
cm-hr1. The focus of the present study was sorption, thus the
impacts of other biochemical and biological processes on bac-
terial transport were minimized.
Minor differences were measured between the peak C/C0 data
for the K. oxytoca breakthroughs at velocities between 3 and
13 cm-hr1. Peak values were about 1x10r1. In the 0.6 cm-hr'
velocity experiment, however, peak C/Co values for K. oxytoca
were about 1x10'2. The measured peak C/C0 values for G4
PR1 decreased as velocity decreased from 1x10'1 to 8x10'5 for
velocities of 14 to 0.5 cm-hr1. Well-defined tailing was ob-
served in all bacterial experiments. In all cases, the C/C
values of the tailing ranged between IxlQ-3 and 5x10"3 and
between IxlO"5 and 5x10'5for K. oxytoca and G4 PR1, respec-
tively.
-------
A 1-D mathematical model for advective-dispersive transport
was used to determine the velocities and dispersivities in the
columns by matching the modeled results to the measured
breakthroughs of a conservative tracer, chloride ion. The
calculated values of velocity and dispersivity were then used in
the 1D mathematical model to determine the sorption param-
eters for each bacteria under each velocity tested by matching
the calculated breakthroughs to the measured breakthroughs.
The mathematical model accounted for both irreversible (kj
and reversible (k, and kt) sorption. The applications of both
Irreversible and reversible sorption were required to obtain
good fits to the measured K. oxytoca data (Table 2 and 3).
Results of this modeling suggested that kltr and kr were inde-
pendent of velocity (0.062 and 0.02 hr1) and an empirical
relationship was developed relating kf to velocity. The best fits
between the modeled and observed breakthroughs for G4 PR1
were obtained using only reversible sorption (Table 4 and 5).
The kr and kf values were found to be independent of velocity
at velocities between 3 and 13 crn-hr1 (0.0008 hr1 and 2.45 hr
1, respectively). At the lowest velocity investigated (0.5 cm-hr
1), the kf value was six times smaller. These results suggest
that we can predict the migration of K. oxytoca under a broad
range of ground-water velocities and the transport of G4 PR1
at velocities of >3.5 cm-hr1. Additional experimentation will be
required, however, to predict the transport of G4 PR1 at lower
velocities.
Table 2. Best-fit sorption parameters determined from the modeling of the bacteria] breakthrough of K. oxytoca from 11.4 cm-long columns
packed with silica sand.
Concentration
of K. oxytoca,
CJCFUmL-*]
8.55 10'
5.70 10'
8.5510'
9.40 10'
8.55 10'
9.40 10'
9.40 10T
5.70 107
Flow
Velocity,
v [cm-hr1]
0.6
0.6
3.1
3.1
6.0
6.0
13.4
12.7
k,[hr']
0.23
0.14
0.69
0.55
1.16
0.92
2.00
1.50
Fitted
K,[hr<]
0.026
0.016
0.010
0.014
0.018
0.016
0.024
0.033
KJhr']
0.062
0.062
0.062
0.062
0.062
0.062
0.062
0.062
Calculated
Recovery*
2.0
5.0
7.0
13.0
10.0
19.0
19.0
32.0
Model
Efficiency,
E
0.886
0.754
0.889
0.954
0.961
0.972
0.982
0.941
•Recovery calculated only to the end of the experiment.
E « a best (it parameter describing how well the calculated values fit the observed.
Table 3. Mean best-fit values of sorption parameters obtained by
modeling of the bacterial breakthrough of K. oxytoca.
Flow Velocity
v [cnvhr1] k,[hr']
0.6
3.1
6.0
13.0
0.19
0.62
1.04
1.75
Calculated
kr [hr1]
0.021
0.012
0.017
0.028
Mnr']
0.062
0.062
0.062
0.062
Empirical
Mhr1]*
0.18
0.63
1.04
1.85
*k,values calculated using the empirical relationship presented in the
text.
Table 4. Best-fit sorption parameters determined from the modeling of the bacterial breakthrough of G4 PR1 from 11.4 cm-long columns packed with
silica sand.
Concentration of PR1
Bacteria CJCFU mL-1]
1.37 10s
0.77 10"
1.3710"
0.77 108
1.0510*
1.37108
0.6610s
Flow Velocity v
[cm-hr1]
0.5
3.3
3.8
7.1
7.1
7.1
13.4
Fitted
k,[hr1]
0.40
2.13
2.32
2.83
2.47
2.64
2.41
kr[hr1]
0.0010
0.0013
0.0008
0.0005
0.0010
0.0005
0.0005
Calculated
Recovery*
0.01
0.12
0.17
1.4
1.4
2.0
12.6
Model
Efficiency E
0.871
0.983
0.842
0.994
0.976
0.983
0.928
•Recovery calculated only to the end of the experiment.
E - a best fit parameter describing how well the calculated values fit the observed.
-------
Table 5. Mean best-fit values of sorption parameters obtained by
modeling of the bacterial breakthrough of G4 PR1.
Flow Velocity
v [cm-hr1]
u.s
3.5
7.1
13.4
Calculated
k([hr']
U.4U
2.23
2.65
2.41
kr[hr1]
U.UU1U
0.0010
0.0007
0.0005
This study showed that sorption characteristics are bacteria-
specific, likely related to the surface chemistry because G4
PR1 is more hydrophobic than K. oxytoca. The study also
showed that for bacterial transport experiments to be directly
applicable to the subsurface environment, they should be con-
ducted at typical ground-water velocities, or that a relationship
between the sorption parameter(s) and velocity be known.
///. Bacterial interactions during transport in
porous media.
A series of breakthrough experiments were conducted to as-
sess the influence of planktonic or attached cells of one spe-
cies on the transport of a second species. The experimental
design examined combinations of Klebsiella oxytoca,
Burkholderia cepacia G4 PR1 and Pseudomonas #5. The
organisms were either mixed and input simultaneously or one
was input, allowed to breakthrough, and the second organism
added. In the latter case, the impact of the presence of the
first organism on attachment and breakthrough of the second
and the influence of the second on the tailing portion of the
breakthrough of the first organism could be assessed. The
basic hydrogeologic and physical parameters used in the ex-
periments were: groundwater velocity =10 cm/hr; dispersivity =
0.3 cm; bulk density = 1.8 gm/cm3; porosity = 0.4; distance =
11 cm. Simultaneous transport experiments indicated that
peak concentration during breakthrough of G4 PR1 was not
significantly influenced by the presence of the other bacteria in
the planktonic phase; however, tailing concentrations were
increased an order of magnitude in the presence of Pseudomo-
nas #5. When K. oxytoca was input with G4 PR1 and
Pseudomonas #5, its breakthrough pattern was unaffected. In
contrast, Pseudomonas #5, exhibited an increase in peak con-
centration during breakthrough in the presence of K. oxytoca
but not G4 PR1. When G4 PR1 was resident in the column, it
increased both peak breakthrough and tailing of Pseudomonas
#5 but not that of K. oxytoca. If K. oxytoca was resident in the
column, then breakthrough of G4 PR1 was unaffected, but
Pseudomonas #5 exhibited a one-to-two order of magnitude
increase in peak concentration. The presence of Pseudomo-
nas #5 increased breakthrough of G4 PR1 and, in this case,
the breakthrough of G4 PR1 resulted in an increase in tailing
concentration of Pseudomonas #5. During parallel experi-
ments, the presence of Pseudomonas #5 resulted in both
decreased peak concentration and tailing concentration of K.
oxytoca. These results indicate that bacterial interactions dur-
ing transport of bacteria are species specific and the patterns
of these interactions included: no impact, increased/decreased
peak concentration and tailing, and displacement of resident
bacteria by planktonic populations during attachment. Thus
the transport of G4 PR1 may be facilitated by the presence of
specific bacteria and retarded by the presence of others.
IV. Integration ofG4 PR1 into a degradative
microbial community.
The fate of genetically engineered microorganisms is con-
trolled by their ability to integrate into and survive in existing
microbial communities. The following studies examined the
fate of Burkholderia cepacia G4 PR1, a bacterium constitutive
for the toluene-ortho-monooxygenase responsible for trichloro-
ethylene degradation. To examine its survival potential, the
organism was introduced into a defined microbial community
grown with TCE and dibutyl phthalate as the major source of
carbon and energy, or with TCE alone. Studies utilizing Mabs
and CSLM showed that after one and two years the G4 PR1
could be found in biofilms grown from continuously-fed cultures
of the defined microbial community. Further CSLM investiga-
tions showed that G4 PR1 was a common member of the
community and formed microcolonies located throughout the
biofilm. In addition, with time G4 PR1 became part of mound
structures characteristic of the biofilm community. This posi-
tion within the biofilm community may be consistent with its
role as a primary member when TCE was provided as a
carbon source. Comparisons of Biolog profiles of the defined
community with and without G4 PR1 indicated that its pres-
ence altered community carbon utilization, confirming changes
in community composition. Thus, in the presence of TCE/
phthalate or TCE alone, G4 PR1 integrated and became a
significant member of an established microbial community.
Acknowledgments
The U.S. Environmental Protection Agency through its Office of
Research and Development, National Health and Ecological
Effects Laboratory, Gulf Ecology Division, partially funded this
research under assistance agreement CR822568 to the Uni-
versity of West Florida. Mention of trade names or commercial
products does not constitute endorsement or recommendation
for use.
-------
-------
-------
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use
$300
EPA/600/S-98/008
------- |