7296
Langmuir 2006, 22, 7296-7301
Specific and Nonspecific Interaction Forces Between Escherichia coli
and Silicon Nitride, Determined by Poisson Statistical Analysis
Nehal I. Abu-Lail† and Terri A. Camesano*
Department of Chemical Engineering, Worcester Polytechnic Institute, Worcester, Massachusetts 01609
ReceiVed December 9, 2005. In Final Form: March 31, 2006
The nature of the physical interactions between Escherichia coli JM109 and a model surface (silicon nitride) was
investigated in water via atomic force microscopy (AFM). AFM force measurements on bacteria can represent the
combined effects of van der Waals and electrostatic forces, hydrogen bonding, steric interactions, and perhaps ligandreceptor type bonds. It can be difficult to decouple these forces into their individual components since both specific
(chemical or short-range forces such as hydrogen bonding) and nonspecific (long-range colloidal) forces may be
present in the overall profiles. An analysis is presented based on the application of Poisson statistics to AFM adhesion
data, to decouple the specific and nonspecific interactions. Comparisons with classical DLVO theory and a modified
form of a van der Waals expression for rough surfaces were made in order to help explain the nature of the interactions.
The only specific forces in the system were due to hydrogen bonding, which from the Poisson analysis were found
to be -0.125 nN. The nonspecific forces of 0.155 nN represent an overall repulsive interaction. These nonspecific
forces are comparable to the forces calculated from DLVO theory, in which electrostatic-double layer interactions
are added to van der Waals attractions calculated at the distance of closest approach, as long as the van der Waals
model for “rough” spherical surfaces is used. Calculated electrostatic-double layer and van der Waals interactions
summed to 0.116 nN. In contrast, if the classic (i.e., smooth) sphere-sphere model was used to predict the van der
Waals forces, the sum of electrostatic and van der Waals forces was -7.11 nN, which appears to be a large overprediction.
The Poisson statistical analysis of adhesion forces may be very useful in applications of bacterial adhesion, because
it represents an easy way to determine the magnitude of hydrogen bonding in a given system and it allows the
fundamental forces to be easily broken into their components.
Introduction
The use of atomic force microscopy (AFM) to probe the
physical properties of microbial surfaces has been in continuous
evolution since the first AFM studies on microbes were performed.
Quantitative information is now easily obtained as the AFM can
detect nano- or picoNewton microbial interaction forces under
physiological conditions.1 Such force measurements are used to
gain insight on microbial adhesion forces and on the physicochemical nature of microbial surface macromolecules.2-5
For a sample that is heterogeneous and somewhat poorly
defined, i.e., a microbe, it can be difficult to relate the measured
forces to their fundamental components. In one study on
Enterococcus faecalis, AFM force profiles were integrated, and
the resulting energy profiles were compared with classical DLVO
(Derjaguin-Landau-Verwey-Overbeek) predictions of such
profiles.5 When bacterial sticking efficiencies were calculated
based on AFM data or DLVO predictions, the DLVO-predicted
values were many orders of magnitude smaller than those
predicted from AFM force data. The discrepancy could be due
to the fact that AFM force profiles on most bacterial species will
display forces in addition to van der Waals and electrostatic
forces, which are not accounted for in the DLVO theory. There
also may be discrepancies due to the assumptions needed to
* To whom correspondence should be addressed. Ph: 508.831.5380.
Fax: 508.831.5853. E-mail: [email protected].
† Current address: Center for Biologically-Inspired Materials and Material
Systems, Duke University, Durham, NC.
(1) Dufrêne, Y. F. Micron 2001, 32, 153-165.
(2) Van der Aa, B. C.; Michel, R. M.; Asther, M.; Zamora, M. T.; Rouxhet,
P. G.; Dufrêne, Y. F. Langmuir 2001, 17, 3116-3119.
(3) Abu-Lail, N. I.; Camesano, T. A. Langmuir 2002, 18, 4071-4081.
(4) Lower, B. H.; Yongsunthon, R.; Vellano, F. P.; Lower, S. K. J. Bacteriol.
2005, 187, 2127-2137.
(5) Cail, T. L.; Hochella, M. F. Geochim. Cosmochim. Acta 2005, 69, 29592969.
apply these models to AFM data. In other studies where it has
not been possible to decouple AFM force curves into fundamental
components of the interaction, correlations have been developed
to help explain the origin of the various forces. For example, the
adhesion force from AFM retraction profiles on nine different
strains of Streptococcus mitis, each interacting with silicon nitride,
could be correlated with the water contact angle on bacterial
lawns and with the nitrogen-to-carbon ratio on the bacterial
surface, the latter parameter quantified by X-ray photoelectron
spectroscopy (XPS).6 Despite some success with such correlations, it is generally not possible to quantitatively describe AFM
force data on bacteria in terms of the underlying fundamental
components.
One approach to quantifying information from AFM force
data is to concentrate on statistical analysis from many adhesion
force measurements. Several approaches that rely either on a
quantized distribution of discrete single-bond contact forces or
on the histogram of the distribution of rupture forces have been
used to determine individual bond forces between ligand-receptor
interactions and self-assembled monolayers (SAMs) of carboxylic
groups 7-9 or amino-terminated SAMs.10 This analysis has the
most obvious physical meaning for well-defined samplesubstrate interactions, such as ligand-receptor pairs,8 but some
success has been found in applying these approaches in bacterial
adhesion studies.3,11 Limitations of this approach are that sev(6) Vadillo-Rodriguez, V.; Busscher, H. J.; Norde, W.; de Vries, J.; van der
Mei, H. C. Langmuir 2003, 19, 2372-2377.
(7) Hoh, J. H.; Cleveland, J. P.; Prater, C. B.; Revel, J.-P.; Hansma, P. K. J.
Am. Chem. Soc. 1992, 114, 4917-4918.
(8) Lee, G. U.; Kidwell, D. A.; Colton, R. J. Langmuir 1994, 10, 354-357.
(9) Moy, V., T.; Florin, E.-L.; Gaub, H. E. Science 1994, 266, 257-259.
(10) Wei, Z. Q.; Wang, C.; Zhu, C. F.; Zhou, C. Q.; Xu, B.; Bai, C. L. Surf.
Sci. 2000, 459, 401-412.
(11) Abu-Lail, N. I.; Camesano, T. A. EnViron. Sci. Technol. 2003, 37, 21732183.
10.1021/la0533415 CCC: $33.50 © 2006 American Chemical Society
Published on Web 07/18/2006
Interaction Forces between E. coli and Silicon Nitride
eral hundreds to thousands of force measurements must be
performed, requiring a long measurement time that could damage
the sample,12 and for bacterial adhesion studies, it still can be
difficult to relate the force distributions to their fundamental
components.
Another approach, which does not require as many measurements and can be used to quantify the individual strength of
bond forces, is derived based on Poisson statistics.12,13 Applying
a Poisson distribution to AFM pull-off data can allow an accurate
estimation of the individual bond strengths and provide detailed
information on the magnitude of the forces involved in the
adhesion between the sample and tip.13 This method represents
one of the only available ways to quantify hydrogen bonding
between bacteria and surfaces from AFM data.
The current study provides a detailed investigation of the
interaction forces that affect the adhesion of E. coli JM109 to
a model surface, a silicon nitride AFM tip. Poisson statistical
analysis was used to decouple the specific and nonspecific forces
that make up bacterial force profiles.
Materials and Methods
Cultures. Escherichia coli JM109 (K-12) was provided by
Professor Kristin N. Wobbe of the Department of Chemistry and
Biochemistry at Worcester Polytechnic Institute (Worcester, MA).
Cells were grown in Luria broth [5 g of NaCl, 5 g of tryptone,
2.5 g of yeast extract in 1 L of milli-Q water (Millipore)] at 37 °C
and 200 rpm and harvested when the OD600 reached 0.9. After
being harvested, cells were centrifuged and washed with ultrapure
water (Milli-Q water, Millipore Corp.) to remove excessive salts.
The cells were then resuspended in ultrapure water until they were
ready to be used.
Atomic Force Microscopy Experiments. All AFM experiments
were performed with a Dimension 3100 Nanoscope IIIa (Digital
Instruments/Veeco) and silicon nitride tips (Digital Instruments/
Veeco). As a pretreatment, silicon nitride cantilevers were irradiated
with ultraviolet light in air for 15 min to remove any organic
contaminates prior to use.14 The spring constants of the tips were
measured to be 0.13 ( 0.02 N/m.15 Bacterial cells from suspension
were attached to cleaned, silanized glass slides by covalent bonding
between bacterial carboxylic groups and amino groups of the
aminosilane compound.3,16 Cells remained hydrated with water prior
to AFM measurements, and AFM imaging was performed in tapping
mode using ultrapure water. Force measurements were made on a
bacterium-free area of the glass slide before and after making the
measurement on the bacterium, to ensure that the tip’s properties
had not been altered by contact with the sample.11 Although we have
performed our force measurements in ultrapure water, we know that
cells retained their structure and did not lyse over the time scale of
these experiments, because this was verified with AFM imaging.
Imaging of each cell was always performed prior to the force
measurement, and revealed that E. coli JM109 cells retained their
ellipsoid shapes and had dimensions within those reported in the
literature.17 All force measurements were performed using a 2.39
µm/s pulling rate and with applied loads that varied between 3.7 and
8 nN. A typical AFM experiment lasted 2-3 h. A total of 11 bacterial
cells from cultures prepared on 3 days were examined. For each
bacterium, 3 force profiles were captured, always over the center
of the cell.
Bacterial Surface Roughness. From AFM images in tapping
mode, we analyzed the roughness of the bacterial surfaces, quantified
(12) Williams, J. M.; Han, T.; Beebe, J. T. P. Langmuir 1996, 12, 1291-1295.
(13) Han, T.; Williams, J. M.; Beebe, J. T. P. Anal. Chim. Acta 1995, 307,
365-376.
(14) Tomitori, M.; Arai, T. Appl. Surf. Sci. 1999, 140, 432-438.
(15) Cleveland, J. P.; Manne, S.; Bocek, D.; Hansma, P. K. ReV. Sci. Instrum.
1993, 64, 403-405.
(16) Camesano, T. A.; Logan, B. E. Langmuir 2000, 16, 4563-4572.
(17) Amro, N. A.; Kotra, L. P.; Wadu-Mesthridge, K.; Bulychev, A.; Mobashery,
S.; Liu, G. Langmuir 2000, 16, 2789-2796.
Langmuir, Vol. 22, No. 17, 2006 7297
in terms of the root-mean-square roughness (rms) parameter. Images
of five bacterial cells were captured (scan size ) 15 µm, scan speed
) 1.01 Hz, 512 samples/line). The rms was measured on five replicate
200 nm × 200 nm areas.
Contact Angle Measurements. Contact angles of ultrapure
water (polar), formamide (polar), and diiodomethane (apolar) were
measured on E. coli JM109 lawns, using a Ramé-Hart NRL Contact
Angle Goniometer (model #100, Mountain Lakes, NJ). A 2 mL
sample of ∼108 cells/ml was filtered onto a 0.45-µm silver membrane filter (Osmonics Inc.). At least 20 separate measurements
with each liquid (2 µL droplets) were taken on both sides of each
droplet.
Poisson Analysis of the Adhesion Force Data from AFM
Retraction Curves. Retraction curves were analyzed to quantify
and characterize the adhesion events. In each cycle, at least one, but
often multiple, adhesion events were observed, which represent an
attachment between the AFM silicon nitride tip and a bacterial surface
biopolymer (or biopolymers; Figure 1). Each adhesion event is
characterized by a pull-off distance and a pull-off force as represented
by the circles in panels A and B in Figure 1. The pull-off force is
equivalent to the adhesion force and represents the sum of all
interaction forces (specific and nonspecific) between the bacterial
surface biopolymers and the silicon nitride cantilever. Beebe et al.
developed a method for characterizing AFM pull-off force data using
a Poisson statistical distribution function,12,13,18 which allows for the
detection of the strength of individual chemical bond forces. The
main assumptions underlying the application of the Poisson analysis
to force data are (1) the total adhesive force (F) develops as the sum
of discrete bonds, both chemical and noncovalent bond interactions,
such as hydrogen bonding and van der Waals forces10 and (2) these
bonds form randomly with a small probability and all have similar
force values (Fi). This methodology was shown to be valid for
characterizing single bond forces between carboxylic-functionalized
surfaces,19 between silicon nitride and either gold or mica,20 for
surfaces functionalized with organosilanes,21 and between biotin
and avidin or streptavidin.22
The probability function that is used to describe the Poisson
distribution of the forces formed at the pull-off point can be expressed
as23
exp(-Fav)
P(F) ) (Fav)n
n!
(1)
where P(F) represents the possibility that an event with an adhesive
force (F) will form, Fav is the average pull-off force of all individual
pull-off forces, and n is the number of pull-off forces that occur in
a certain range. In our case, Fav was taken as the average value for
all of the 100 points shown in Figure 1C.
The adhesive force, measured as a pull-off force event in forcedistance profile, is related to the number of bonds ruptured during
the pull-off event by
F ) niFi
(2)
where Fi represents the average individual-bond rupture force due
to specific forces and ni represents the number of individual bonds.
An additional parameter, Fo, can be added to account for nonspecific
interactions. Based on the relationship between the measured force
and the number of bonds ruptured (eq 2), one can derive relationships
(18) Wenzler, L. A.; Moyes, G. L.; Olson, L. G.; Harris, J. M.; Beebe, T. B.
Anal. Chem. 1997, 69, 2855-2861.
(19) Han, T.; Williams, J.; Beebe, T. P., Jr. Anal. Chim. Acta 1995, 307,
365-376.
(20) Williams, J. M.; Han, T.; Beebe, T. P., Jr. Langmuir 1996, 12, 12911295.
(21) Wenzler, L. A.; Moyes, G. L.; Raikar, G. N.; Hansen, R. L.; Harris, J.
M.; Beebe, T. P., Jr. Langmuir 1997, 13, 3761-3768.
(22) Lo, Y.-S.; Huefner, N. D.; Chan, W. C.; Stevens, F.; Harris, J. M.; Beebe,
T. P., Jr. Langmuir 1999, 15, 1373-1382.
(23) Barlow, R. J. Statistics. A guide to the use of statistical methods in physical
sciences; Wiley: New York, 1989.
7298 Langmuir, Vol. 22, No. 17, 2006
Abu-Lail and Camesano
Table 1. Summary of Adhesion Forces between E. coli JM109
and Silicon Nitride, in Water, Measured by AFM
cell
avg. F, nN
variance F, nN2
set sizea
1
2
3
4
5
6
7
8
9
10
11
All
-1.01
-6.36
-0.94
-1.13
-1.22
-4.26
-0.75
-0.64
-0.44
-0.64
-0.66
-1.30
0.11
0.75
0.08
0.15
0.14
0.54
0.07
0.09
0.01
0.03
0.07
18
2
7
17
22
7
5
4
7
4
7
100
a The total number of adhesion peaks observed in three force curves
measured on each bacterial cell. Bacterial cells were from different
cultures and measurements were performed on different days. Individual
data points are shown in Figure 1C for all cells.
Figure 2. Distribution of absolute values of pull-off forces measured
during retraction portions of force cycles between E. coli JM109
surface biopolymers and silicon nitride cantilevers in water. The
solid line represents the theoretical Poisson distribution of the
adhesion forces.
Figure 1. Examples of AFM retraction curves measured between
a silicon nitride AFM tip and E. coli JM109 bacterial surface
biopolymers, in water. (A) Example of data set that shows a single
adhesion event. (B) Example of data set that shows multiple adhesion
events. Each adhesion event is highlighted with circles. (C) A
summary of all the adhesion events measured on 11 bacterial cells.
This plot shows all of the data that was used for the analysis represented in Figures 2 and 3 and is also described in Table 1. Each different type of symbol corresponds to measurements on a single cell.
for the mean (µF) and variance of the distribution (σF2), which are
equivalent for a Poisson distribution
µ ) µ FFi + F o
(3)
σ2 ) σF2Fi2 ) µFFi - FiFo
(4)
Linear regression is used to calculate Fi and Fo for a given data set.
For our measurements, µ and σ2 were taken as the average and
the variance of all of the pull-off forces measured on one cell (Table
1). A plot of σ2 versus µ is shown in Figure 3. The slope of the linear
regression line that passes through the data points shown in Figure
3 was taken as the specific force, Fi, and the intercept was used to
calculate the nonspecific forces, Fo.
Interfacial Free Energy Calculations. For two surfaces at
equilibrium and in contact, the net free energy of adhesion between
the two surfaces can be described as a function of the apolar (van
der Waals) and polar (Lewis acid-base) free energies. Free energies
are related to the surface tension components of the two surfaces.
The various surface tension components of the bacterial cell can be
calculated using the Young-Dupré equation24
LW
+ - +
(1 + cosθ)γL ) 2(xγLW
s γL + xγs γL + xγs γL )
(5)
where θ is the contact angle, γL is the total surface tension of the
liquid, γLW
is the Lifshitz-van der Waals, or a polar surface tension
i
component of condensed material (i), γ+
i and γi are the electronacceptor and electron-donor parameters of the Lewis-acid base
components of the surface tension of condensed material (i),25 and
the subscripts “S” and “L” refer to the bacterial and liquid phases.
(24) de Gennes, P. G. ReV. Mod. Phys. 1985, 57, 827-863.
(25) van Oss, C. J. Interfacial Forces in Aqueous Media; Marcel Dekker:
New York, 1994.
Interaction Forces between E. coli and Silicon Nitride
Langmuir, Vol. 22, No. 17, 2006 7299
Aii ) 24πHo2γLW
i
(9)
Electrostatic-Double-Layer Interactions. The electrostaticdouble-layer interactions between the bacterium and tip are calculated
from the linearized version of the Poisson-Boltzmann expression,
for sphere-sphere interactions.29
Ee )
2πapamNA
(am + ap)κ
(φm2 + φp2)
2
{
2φmφp
φm + φ p
2
2
(
ln
)
1 + exp(-κh)
1 - exp(-κh)
ln[1 + exp(-2κh)]
Figure 3. Linear relationship between the mean and the variance
of the pull-off forces, a requirement of the Poisson distribution. The
error bars represent the standard errors of mean. The solid line (linear
regression) was used to calculate the specific and nonspecific
components of the adhesion, with resulting parameters being, the
specific force, Fi ) -0.125 nN, the intercept (-FiFo) ) -0.019
nN2, and the nonspecific forces (Fo) ) 0.155 nN. R2 for the application
of the Poisson distribution to these data ) 0.99.
}
(10)
where φm and φp are the normalized bacterial and tip surface
potentials, respectively, κ is the inverse Debye screening length, and
NA is Avogadro’s number. Surface potentials and radii for the
bacterium and tip were measured by us previously.11 The values of
the parameters used in our analysis were -20.2 mV, -16 mV, 500
nm, and 250 nm for the surface potentials, followed by the radii of
the bacterium and the silicon nitride tip, respectively. An ionic strength
of 0.0027 M was used for the ionic strength of ultrapure water, as
previously estimated.30
Results
where A132 is the Hamaker constant for media 1 (bacterium) and 2
(silicon nitride) interacting across media 3 (solvent). The individual
Hamaker constant for each of the interacting components can be
calculated as26
Specific and Nonspecific Forces. The rupture forces between
the silicon nitride tip and biopolymers were characterized by
applying a Poisson distribution function to the distribution of
adhesion forces (Figure 2). A summary of the mean and standard
deviation of the adhesion forces per set is given in Table 1. A
linear plot of mean versus variance of the force (Figure 3) was
used to determine the specific and nonspecific components of
the overall interaction force, which were -0.125 and +0.155
nN, respectively.
Through the use of the Poisson-based analysis, the individual
components of the overall interaction could be broken into their
fundamental parts. Since there are no chemical bonds to be
expected between silicon nitride and the bacterial surface, the
specific forces can be attributed to hydrogen bonding. The
measured nonspecific forces include van der Waals and
electrostatic interactions, which can be compared with theoretical
predictions.
In calculating the theoretical van der Waals attractive force,
two different models were used, both the classic model for smooth
spheres, and a model accounting for the roughness of the
interacting surface. To use the latter model, an estimation of the
rms and the distance between asperities on the bacterial surface
is necessary to calculate the van der Waals force component.
AFM images on E. coli JM109 revealed that the rms value in
water was 3.7 ( 0.5 nm, based on examining several 200 nm
× 200 nm areas (images not shown). The asperity height was
obtained from tapping-mode AFM images of a similar E. coli
K-12 strain, HB101 (16.3 ( 2.8 nm;31). The Hamaker constant
used in the van der Waals calculations was estimated from the
liquid contact angle measurements on the bacterial lawns (Table
2) and was found to be 3.80 × 10-20 J for this bacterium in water,
interacting with silicon nitride.
With these parameters, van der Waals forces based on the
“rough” model were -0.046 nN (Table 3). This value is much
smaller than that predicted by the classical or smooth, spheresphere van der Waals model, which gives a value of -7.21 nN.
Electrostatic-double layer interactions were found to be repulsive,
(26) Israelachvili, J. N. Intermolecular & Surface Forces, 2nd ed.; Academic
Press: New York, 1992.
(27) Rabinovich, Y. I.; Adler, J. J.; Ata, A.; Singh, R. K.; Moudgil, B. M. J.
Colloid Interface Sci. 2000, 232, 10-16.
(28) Rabinovich, Y. I.; Adler, J. J.; Ata, A.; Singh, R. K.; Moudgil, B. M. J.
Colloid Interface Sci. 2000, 232, 17-24.
(29) Elimelech, M.; Gregory, J.; Jia, X.; Williams, R. A. Particle deposition & aggregation: Measurement, modelling and simulation; ButterworthHeinemann: Woburn, MA, 1995.
(30) Abu-Lail, N. I.; Camesano, T. A. Biomacromolecules 2003, 4, 10001012.
(31) Liu, Y. Personal communication.
van der Waals Forces. The nonretarded Lifshitz-van der Waals
interactions between two spherical surfaces, FV, can be calculated
as26
Fv ) -
A132amap
(6)
6h2(am + ap)
where the separation distance, h, is evaluated as Ho, the theoretical
distance of closest approach (1.57 Å),25 A132 is the Hamaker constant
of the bacterium with the tip in that media, am is the bacterial radius,
and ap is the tip radius.
For consideration of rough surfaces, a special form of the van der
Waals expression can be used, in which the roughness of the adhering
particle (E. coli JM109, in this case) is accounted for. If the roughness
in the interaction is assumed to be primarily due to bacterial surface
macromolecules, then the appropriate model for calculating the
adhesion force has been derived as27,28
Fv,rough ) -
[
A132ap
6Ho2
1+
1
+
32apkvrms
(
λ2
)]
1
kvrms
1+
Ho
) (
2
(7)
where kv is a coefficient equal to 1.817, and the terms rms and λ
refer to the root-mean-square roughness and height of asperities,
respectively, on the bacterial surface.
The Hamaker constant for the system is calculated as
A132 ≈ (xA11 - xA33)(xA22 - xA33)
(8)
7300 Langmuir, Vol. 22, No. 17, 2006
Abu-Lail and Camesano
Table 2. Contact Angle Measurements on E. coli JM109
solvent
contact angle (θ)
Na
formamide
water
diiodomethane
43.0 ( 1.9
17.0 ( 3.8
50.1 ( 1.0
41
21
41
a
N is the number of contact angle measurements averaged in the
reported value.
Table 3. Comparison of Measured Forces, Those Predicted by
Poisson Analysis, and Theoretical Calculationsa
Fadh
Poisson analysis, specific forces
Poisson analysis, nonspecific forces
theoretical van der Waals forces
(rough model)
theoretical van der Waals forces
(classical, smooth model)
theoretical electrostatic forces
theoretical nonspecific:
van der Waals (rough) + electrostatic
theoretical nonspecific:
van der Waals (smooth) + electrostatic
a
-1.30
-0.125
0.155
-0.046
-7.21
0.162
0.116
-7.11
All values are in nN.
with a value of 0.162 nN. Since the nonspecific interactions are
likely to be due primarily to van der Waals and electrostatic
interactions, we summed those contributions. The summed
contributions (when choosing the rough van der Waals model)
agree fairly closely with the nonspecific interactions predicted
through the use of the Poisson statistical analysis (0.116 nN
compared with 0.155 nN), whereas the summed contributions
based on the classical van der Waals model are very far from
(and of opposite sign to) the values suggested by the Poisson
analysis (total force of -7.11 nN for the smooth van der Waals
model).
Discussion
Breaking AFM Force Data into Components. In certain
biological systems, for example, the binding forces between
ligand-receptor pairs streptavidin and biotin, or between proteins
and surfaces, discrete and well-defined forces govern the binding.
These forces can be described using histograms to pick out discrete
intervals of the binding force32 or through a thermodynamic
treatment of bond energies.33 To describe the interactions between
bacteria and a nonbiological surface, nonspecific interactions
are always present, and in most cases, there are several types of
interaction forces operating simultaneously.
Repulsive forces can be observed during the approach portions
of AFM force cycles, especially for bacteria which express large
amounts or long polysaccharide polymers, such as Pseudomonas
strains. In these cases, success has been observed with using a
steric model to describe the repulsive bacterial polymer interactions with a smooth substrate.34,35 For strains that do not exhibit
large steric forces due to surface polymers, such as Enterococcus
faecalis, it has been possible to model the interaction forces from
AFM approach curves using DLVO theory. The adhesion forces
to surfaces could also be described based on the interfacial energy
using the interaction force boundary layer (IFBL) theory. When
both were compared, the latter model gave much more realistic
(32) Sagvolden, G. Biophys. J. 1999, 77, 526-532.
(33) Moy, V. T.; Florin, E.-L.; Gaub, H. E. Science 1994, 266, 257-259.
(34) Camesano, T. A.; Logan, B. E. EnViron. Sci. Technol. 2000, 34, 33543362.
(35) Emerson, R. J.; Camesano, T. A. Appl. EnViron. Microbiol. 2004, 70,
6012-6022.
results than the DLVO model.5 Another study showed fairly
good agreement between extended-DLVO calculations (including
acid/base interactions) and AFM approach curves of three different
E. coli strains with a 1 mm colloidal probe (Pyrax), although the
theoretical primary energy minimum, which would extend to
negative infinity, could not be demonstrated from the experiments.36
AFM retraction force portions of the cycle between bacteria
and surfaces always show some degree of attraction, due to surface
molecules forming weak physical attachments with the AFM
tip. Histograms and statistical tests can be used to distinguish the
adhesion forces among different bacterial strains or among
different experimental conditions.11 It has been difficult to model
the adhesion forces more explicitly, since the number and type
of polymers present on a bacterial surface is generally not known,
and several types of interactions are occurring simultaneously.
Therefore, a key advantage of the application of the Poisson
statistical analysis techniques to AFM adhesion force distributions
is that one does not need to know such detailed information on
the nature of the polymers present. Regardless, one is still able
to decouple the overall forces into their specific and nonspecific
components.
Importance of Hydrogen Bonding in Bacterial Adhesion.
Applying the Poisson statistical analysis to the adhesion force
distribution from AFM data allowed us to decouple the
components of the specific and nonspecific forces involved in
the adhesion of E. coli to silicon nitride. In addition, this analysis
technique represents a very simple method to estimate hydrogen
bond forces from AFM data for bacteria, since, when considering
bacterial adhesion to inert surfaces, we usually do not have other
types of specific forces present (i.e., chemical bonds). The
technique was also successfully applied to biological systems
which do exhibit chemical bonding, for example, to the study
of binding between biotin-avidin or biotin-streptavidin.22,37
However, such specific chemical bond forces did not exist in the
system we studied.
The role of hydrogen bonding in controlling bacterial adhesion
to inert surfaces has been less studied than some of the other
components of the interaction, such as electrostatic-double layer
forces or steric repulsion, in part because there has been no
simple method to measure and/or quantify these forces. One
study addressed very systematically the ability to form hydrogen
bonds with metal oxide surfaces (TiO2, Al2O3, and SiO2) for the
isolated O-antigens of several bacterial strains, including E. coli,
Citrobacter freundii, and Stenotrophomonas maltophilia.38
Infrared spectroscopy was used to demonstrate that surface
hydroxyl groups from the isolated saccharides interacted with
water bound to the surface of the metal oxides. Although a very
interesting result, the work was not extended to whole bacteria,
so it is not known how other surface molecules may form hydrogen
bonds with their contacting surfaces.
From our analysis, we learned that the specific forces and
nonspecific forces governing the adhesion of E. coli JM109 to
silicon nitride are of a very similar magnitude. Therefore, it will
be important to consider forces such as hydrogen bonding in
predicting the overall adhesion or attachment behavior of bacteria
to inert surfaces. The hydrogen bond value we estimate for a
given experiment (0.125 nN) is similar to literature values, which
reportedly range from 0.094 to 0.377 nN.26 AFM analysis has
not been used to quantify hydrogen bonding between bacteria
(36) Morrow, J. B.; Stratton, R.; Yang, H.-H.; Smets, B. F.; Grasso, D. EnViron.
Sci. Technol. 2005, 39, 6395-6404.
(37) Lo, Y.-S.; Zhu, Y.-J.; Beebe, T. P., Jr. Langmuir 2001, 17, 3741-3748.
(38) Jucker, B. A.; Harms, H.; Hug, S. J.; Zehnder, A. J. B. Colloids Surf.
B-Biointerfaces 1997, 9, 331-343.
Interaction Forces between E. coli and Silicon Nitride
Langmuir, Vol. 22, No. 17, 2006 7301
and silicon nitride previously. However, the technique has been
applied to measure intramolecular hydrogen bonding in biological
molecules. For example, by using an AFM force cycle to stretch
the peptide cysteine3-lyseine30-cysteine from an alpha helix to
a linear chain, intramolecular hydrogen bonds were broken, and
such forces were quantified.39
With knowledge of the magnitude of hydrogen bonding in
bacterial adhesion systems, which can easily be obtained from
Poisson analysis of AFM force spectra, numerous applications
can be affected. For example, one strategy to prevent bacterial
adhesion to biomaterial surfaces relied on the creation of thin
polymer films that lacked functional groups capable of forming
hydrogen bonds, thus reducing the adhesion of bacteria (Staphylococcus aureus and Staphylococcus epidermidis) and proteins
to the film surfaces.40 As another example, the resistance of
bacteria to the antibiotic vancomycin has been shown to be related
to the ability of certain bacteria to avoid forming a hydrogen
bond with the antibiotic.41,42 These examples emphasize how
important it is to consider hydrogen bonding, along with the
typically considered van der Waals and electrostatic interactions,
in applications that are controlled by bacterial adhesion to a
surface.
Symbols Used
Acknowledgment. This publication was made possible in
part by a CAREER Award to T.A.C. from the National Science
Foundation (BES-0238627). We also acknowledge the donors
of the Petroleum Research Fund of the American Chemical
Society, for partial support of this work (PRF 38988-G2). The
authors also thank Mr. Ray Emerson (WPI) for helpful discussions
throughout the course of this work, and Dr. Jayne Morrow and
Prof. Domenico Grasso, both formerly of Smith College, for
their assistance with the contact angle measurements.
(39) Lantz, M. A.; Jarvis, S. P.; Tokumoto, H.; Martynski, T.; Kusumi, T.;
Nakamura, C.; Miyake, J. Chem. Phys. Lett. 1999, 315, 61-68.
(40) Chapman, R. G.; Ostuni, E.; Liang, M. N.; Meluleni, G.; Kim, E.; Yan,
L.; Pier, G.; Warren, H. S.; Whitesides, G. M. Langmuir 2001, 17, 1225-1233.
(41) Loll, P. J.; Axelsen, P. H. Ann. ReV. Biophys. Biomol. Struct. 2000, 29,
265-289.
(42) Walsh, C. T.; Fisher, S. L.; Park, I. S.; Prahalad, M.; Wu, Z. Chem. Biol.
1996, 3, 21-28.
am ) bacterial radius (nm)
ap ) tip radius (nm)
A132 ) Hamaker constant for interacting media (J)
Aii ) individual Hamaker constant of each component (J)
F ) adhesion force (nN)
Fav ) average adhesion force, from AFM measurements (nN)
Fe ) calculated electrostatic-double layer force (nN)
Fi ) specific force component from Poisson distribution (nN)
Fo ) nonspecific force component from Poisson distribution (nN)
Fv ) van der Waals force for sphere-sphere interactions (nN)
Fv,Rough ) van der Waals force for two spheres, accounting for
surface roughness (nN)
h ) separation distance between two interacting bodies (nm)
Ho ) distance of closest separation (0.157 nm)
kv ) dimensionless coefficient used in rough van der Waals force
calculation (1.817)
n ) number of occurrences of certain pull-off event
ni ) number of individual bonds
NA ) Avogadro’s number (#/mol)
rms ) root-mean-square roughness of the bacterium (nm)
φm ) Normalized surface potential of the bacterium
φp ) normalized surface potential of the tip
γL ) total surface tension of liquid (mJ/m2)
γL+ ) electron-acceptor parameter of the Lewis acid-base surface
tension component of the liquid (mJ/m2)
γL- ) electron-donor parameter of the Lewis acid-base surface
tension component of the liquid (mJ/m2)
γLW
L ) Lifshitz-van der Waals surface tension component of the
liquid (mJ/m2)
γs+ ) electron-acceptor parameter of the Lewis acid-base surface
tension component of the bacterium (mJ/m2)
γs- ) electron-donor parameter of the Lewis acid-base surface
tension component of the bacterium (mJ/m2)
γLW
s ) Lifshitz-van der Waals surface tension component of the
bacterium (mJ/m2)
λ ) peak distance between asperities (nm)
λc ) retardation correction (nm)
θ ) contact angle
µF ) mean of the adhesion forces (nN)
σ2F ) variance of the adhesion force (nN2)
LA0533415