Langmuir 2009, 25, 2081-2089

2081

Competitive Protein AdsorptionsMultilayer Adsorption and Surface

Induced Protein Aggregation
Maria Holmberg*, and Xiaolin Hou
Department of Micro- and Nanotechnology and, Ris National Laboratory for Sustainable Energy,
Technical UniVersity of Denmark, FrederiksborgVej 399, DK-4000 Roskilde, Denmark
ReceiVed February 28, 2008. ReVised Manuscript ReceiVed NoVember 14, 2008
In this study, competitive adsorption of albumin and IgG (immunoglobulin G) from human serum solutions and
protein mixtures onto polymer surfaces is studied by means of radioactive labeling. By using two different radiolabels
(125I and 131I), albumin and IgG adsorption to polymer surfaces is monitored simultaneously and the influence from
the presence of other human serum proteins on albumin and IgG adsorption, as well as their mutual influence during
adsorption processes, is investigated. Exploring protein adsorption by combining analysis of competitive adsorption
from complex solutions of high concentration with investigation of single protein adsorption and interdependent
adsorption between two specific proteins enables us to map protein adsorption sequences during competitive protein
adsorption. Our study shows that proteins can adsorb in a multilayer fashion onto the polymer surfaces and that the
outcome of IgG adsorption is much more sensitive to surface characteristics than the outcome of albumin adsorption.
Using high concentrations of protein solution and hydrophobic polymer surfaces during adsorption can induce IgG
aggregation, which is observed as extremely high IgG adsorptions. Besides using a more hydrophilic substrate,
surface-induced IgG aggregation can be inhibited by changing the adsorption sequence of albumin and IgG.

Introduction
Today it is widely accepted that when an artificial material is
put in contact with a biological fluid, protein adsorption is the
first step in a cascade of events that eventually decide how the
system reacts on the material.1,2 Many studies on protein
adsorption carried out today are pointed toward the development
of biomaterials and artificial organs, where the focus is to fabricate
nonfouling materials and minimize protein adsorption through
surface modifications.3-5 There are also a range of more or less
systematic studies on how specific proteins react when adsorbed
onto surfaces with specific characteristics. Isotherms of single
protein adsorption have been established, and competitive protein
adsorption from protein mixtures, as well as from serum and
plasma, has been studied.6-11 One of the dogmas regarding
competitive adsorption is the so-called Vroman effect, which
states that small proteins present in high concentration will
dominate the adsorption at early stages of the competitive
adsorption processes, where after larger and less abundant protein
will replace some of the already adsorbed proteins.12,13
* Correspondind author. Tel. +45 4677 4735. E-mail. maria.holmberg@
nanotech.dtu.dk.

Department of Micro- and Nanotechnology.

Ris National Laboratory for Sustainable Energy.

(1) Anderson, J. M.; Rodriguez, A.; Chang, D. T. Semin. Immunol. 2008, 20,
86100.
(2) Latour, R. A., Jr. Encyclopedia of Biomaterials and Biomedical Engineering;
2005, 1
(3) Ratner, B. D.; Hoffman, A. S.; Schoen, F. J.; Lemons, J. E., Biomaterials
Science. An Introduction to Materials in Medicine, 2nd ed.; Elsevier Academic
Press: San Diego, CA, 2004.
(4) Valdes, T. I.; Ciridon, W.; Ratner, B.; Bryers, J. D. Biomaterials 2008, 29,
1356.
(5) Castner, D. G.; Ratner, B. D. Surf. Sci. 2002, 500, 28.
(6) Zhang, Z.; Zhang, M.; Chen, S.; Horbett, T. A.; Ratner, B. D.; Jiang, S.
Biomaterials 2008, 29, 4285.
(7) Ariola, F. S.; Krishnan, A.; Vogler, E. A. Biomaterials 2006, 27, 3404
3412.
(8) Malmsten, M. J. Colloid Interface Sci. 1998, 207, 186.
(9) Malmsten, M.; Muller, D.; Lassen, B. J. Colloid Interface Sci. 1997, 193,
88.
(10) Lassen, B.; Malmsten, M. J. Colloid Interface Sci. 1997, 186, 9.
(11) Green, R. J.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B. Biomaterials
1999, 20, 385.

Even though the detailed molecular mechanisms of the Vroman

effect are still unresolved, the outcome of competitive adsorption
has been confirmed experimentally through various techniques
and methods.9,14-16 Furthermore, despite the efforts made to
reveal the mechanisms of protein adsorption, some more or less
fundamental aspects regarding protein adsorption, and in particular
competitive protein adsorption, are still not fully understood and
are enthusiastically debated in the literature. Reversibility of
protein adsorption and whether or not proteins adsorbs in a
monolayer or in a multilayer fashion, as well as experimental
protocols for quantitative determination of protein adsorption to
surfaces, are some of the issues that are discussed and remarked
upon.17,18
Some experimental results in the literature correlate well with
a model proposing a single layer (monolayer) of irreversible
adsorbed proteins onto surfaces after typical rinsing procedures,19
and it has been stated that, after the formation of a protein
monolayer on the surface, some increase in the amount of protein
adsorbed can be obtained by rearrangement of the proteins on
the surface.20,21 Proteins with asymmetric conformation, and
thereby with the possibility to significantly change contact area
between protein and surface by changing orientation, will have
enhanced possibilities to increase the amount of proteins adsorbed
after a first monolayer is formed. The saturation value will be
influenced by protein concentration18 and protein characteristics,
where a fast adsorption gives less opportunity for the protein to
relax on the surface, and thereby there will be less spreading,
(12) Vroman, L.; Adams, A. L. Surf. Sci. 1969, 16, 438.
(13) Horbett, T. A. Thromb. Haemostasis 1984, 51(2), 174.
(14) Noh, H.; Vogler, E. A. Biomaterials 2007, 28, 405422.
(15) Krishnan, A.; Liu, Y.-H.; Cha, P.; Allara, D.; Vogler, E. A. J. R. Soc.
Interface 2006, 3, 283.
(16) Sevastianov, V. I.; Tremsina, Y. S.; Eberhart, R. C.; Kim, S. W. ACS
Symp. Ser. 1995, 602, 195.
(17) Noh, H; Vogler, E. A. Biomaterials 2006, 27, 57805793.
(18) Norde, W. Colloids Surf. B 2008, 61, 1.
(19) Balcells, M.; Klee, D.; Fabry, M.; Hocker, J. Colloid Interface Sci. 1999,
220, 198.
(20) Hylton, D. M.; Shalaby, S. W.; Latour, R. A. J. Biomed. Mater. Res. Part
A 2005, 73A(3), 349.
(21) Kleijn, M.; Norde, W. Heterog. Chem. ReV. 1995, 2, 157.

10.1021/la8031978 CCC: $40.75 2009 American Chemical Society

Published on Web 01/21/2009

2082 Langmuir, Vol. 25, No. 4, 2009

where the level of spreading is a function of the relaxation rate

relative to the adsorption rate.22-25 It is known that during protein
adsorption there are contributions from several different forces,
such as for example electrostatic and van der Waals forces, but
the mutual influence of the different forces during the different
processes is not yet established. Several papers argue that protein
adsorption is controlled by content and structure of water on the
surface,4,14,26,27 and it is speculated that a higher and stronger
protein adsorption to hydrophobic surfaces compared to hydrophilic surfaces can been considered to be driven by exclusion
of water molecules from the surface, and therefore all globular
proteins will exclude approximately the same amount of water
per gram protein.27
There are numerous examples of results where the amount of
protein adsorbed is much too high to represent a monolayer and
where dynamic multilayer adsorption is more realistic for
explaining the obtained results.28-30 The amount of proteins
adsorbed onto the polymer surfaces detected in our laboratory
is often in a range that would require a nonrealistic protein density
if adsorbed in a monolayer fashion. We consider the adsorbed
proteins to be part of an interface region between the polymer
surface and protein solution where this interface represents the
adsorbed protein layer. Dependent on characteristics of the system
(surface wettability, ion concentration in solution, protein
flexibility, etc.) as well as the distance from the surface, the
proteins are considered to be more or less strongly connected to
the surface.
Here we present results from albumin and IgG adsorption
from single protein solutions, albumin and IgG mixtures, as well
as human serum solutions onto two different polymer surfaces.
By using two different isotopes in radioactive experiments,
albumin and IgG can be detected simultaneously and competitive
protein adsorption can be investigated. Furthermore, by using
rather high concentration solutions (corresponding to 25% human
serum) we also aimed at estimating the influence of protein
concentration on the adsorption pattern. Often protein solutions
of low concentration (e1 mg/mL) are used for protein adsorption
and for validation of antifouling characteristics of possible
biomaterials, even though the trend and outcome of competitive
protein adsorption are not necessarily the same from solutions
of low and high concentration. The result presented in this study,
as well as earlier performed investigations in our laboratory,
indicates that both albumin and IgG adsorbs in a multilayer fashion
on many polymer surfaces. Furthermore, using high IgG
concentration and adsorbing onto the more hydrophobic PET
surface can cause surface induced IgG aggregation, which is
expressed as very high adsorption onto the surface.

Experimental Section
A disk of polyethylene terephthalate (PET) film (Trafoma A/S,
Denmark) with diameter of 13 mm was used as substrate. To obtain
hydrophilic substrates for adsorption experiments, soft AC plasma
(22) Norde, W.; Giacomelli, C. E. Macromol. Symp. 1999, 145, 125.
(23) Lassen, B.; Malmsten, M. J. Colloid Interface Sci. 1996, 180, 339.
(24) van der Veen, M.; Stuart, M. C.; Norde, W. Colloid Surf. B 2007, 54, 136.
(25) Wertz, C. F.; Santore, M. M. Langmuir 1999, 15, 8884.
(26) Cha, P.; Krishnan, A.; Fiore, V. F.; Vogler, E. A. Langmuir 2008, 24,
25532563.
(27) Krishnan, A.; Cha, P.; Liu, Y.-H.; Allara, D.; Vogler, E. A. Biomaterials
2006, 27, 31873194.
(28) Zhou, C.; Friedt, J.-M.; Angelova, A.; Choi, K.-H.; Laureyn, W.; Frederix,
F.; Francis, L. A.; Campitelli, A.; Engelborghs, Y.; Borghs, G. Langmuir 2004,
20, 58705878.
(29) Lu, C. F.; Nadarajah, A.; Chittur, K. K. J. Colloid Interface Sci. 1994,
168, 152.
(30) Lee, S. H.; Ruckenstein, E. J. Colloid Interface Sci. 1988, 125, 365379.

Holmberg and Hou

polymerization,31 using the monomer diethylene glycol vinyl ether
(DEGVE), was performed. The polymerization was performed using
a power of 1.4 W for 30 min, resulting in a surface with a contact
angle around 20-60 32 compared to the 80-90 normally
observed for the unmodified PET surface. Both unmodified and
modified surfaces was characterized using an OCA 15 plus contact
angle microscope (Dataphysics Instruments GmbH), a Perker-Elmer
Spectrum One fourier-transform infrared (FTIR) spectrometer
(Perkin-Elmer Instruments), and a monochromatic X-ray photoelectron spectrometer (XPS) (K-Alpha from Thermo Fisher Scientific)
before radioactive labeling experiments.
The roughness of PET surfaces before and after protein adsorption
was analyzed using atomic force microscopy (AFM) (MultiMode
Nanoscope III AFM, Vecco Instruments) in tapping mode in air.
Adsorption experiments and rinsing procedures were performed as
described for radioactive labeling experiments (but without label),
and as reference we used a PET disk that only had been in contact
with phosphate-buffered saline (PBS) buffer and otherwise treated
as the PET disks used for adsorption experiments. Surfaces were
scanned over areas from 0.5 0.5 m to 5 5 m, and roughness
data are obtained from a surface area of 0.5 1 m. The software
package SPIP (Image Metrology A/S) was used for analysis of the
obtained AFM data. First-order linewise correction (LMS fit) before
roughness (root-mean-square roughness, Sq) of the surface was
obtained, and the uncertainty of approximately 10% on Sq values
originates from instrumentation.
Albumin, immunoglobulin G (IgG), human serum, and PBS were
obtained from Sigma-Aldrich. Human serum diluted to 0.25% and
25% in PBS buffer, as well as albumin and IgG solutions with
corresponding concentrations (0.10 mg/mL and 10 mg/mL for
albumin and 0.03 mg/mL and 3 mg/mL for IgG), with added 125Ialbumin and 131I-IgG, were used for adsorption experiments to
polymer surfaces. The Iodo-Gen method was used to radiolabel the
proteins with radioactive iodine; 0.15 mL of albumin or IgG solution
and 0.05 mL of iodine solution (450 kBq/L 125I or 450 kBq/L 131I,
Perkin-Elmer Life and Analytical Science) are added to an IodoGen tube (Pierce) for iodine labeling at room temperature for 15
min, after which the solution is removed from the reaction tube. The
labeled proteins are separated from free iodine and other chemical
reagents using gel chromatography (a PD-10 desalting column from
Amersham Bioscience, UK). Typical concentrations of labeled protein
were around 500 g/mL for 125I-albumin and 200 g/mL for 131IIgG. Labeled proteins where checked for radiolysis using TCA
(trichloroacetic acid) precipitation33 and showed less than 7% free
iodine in the solution over a period of 5 days.
Adsorption was performed at room temperature and with
adsorption times ranging from approximately 5 s to 24 h. The polymer
substrates were placed in ELISA wells, to which 0.5 mL of protein
or human serum solution used for adsorption was added. For twostep adsorption experiments including more than one protein solution,
the second protein solution was added to the same ELISA well with
a pipet after a selected time interval (see Table 1). When adsorption
has been carried out for the time intended, the protein solution is
removed from the ELISA well with a pipet. Immediately after the
removal of protein solution, 2 mL of PBS buffer is added to the well,
after which the surfaces are rinsed (exchange of the PBS buffer in
the well two times and rinsing of each polymer disk in three beakers
of PBS buffer). The radioactivity of 125I and 131I on each surface is
measured by a Canberra 20 -counter (Canberra) and used to
determine the amount of albumin and IgG adsorbed to the surface.
Each measuring point presented in adsorption graphs from
radioactive labeling experiments represents the average of measurements performed on three surfaces treated in the same way during
the same adsorption experiments. The error bars represents the
standard deviation of the mean. For transforming the -count into
(31) Ademovic, Z.; Wei, J.; Winther-Jensen, B.; Hou, X. L.; Kingshott, P.
Plasma Process. Polym. 2005, 2, 53.
(32) Holmberg, M.; Stibius, K. B.; Larsen, N. B.; Hou, X. L. J. Mater. Sci.:
Mater. Med. 2008, 19, 2179.
(33) Sivaraman, T.; Kumar, T. K. S.; Jayaraman, G.; Yu, C. J. Protein Chem.
1997, 16(4), 291.

CompetitiVe Protein Adsorption

Table 1. Adsorption Experiments on PET and DEGVEa
expt
no.
1
2
3
4
5
6
7
8
9
10

experimental procedure
1
1
1
1
1
1
1

min adsorption in albumin solution

min adsorption in IgG solution
min adsorption in albumin and IgG solution (mixture)
h adsorption in albumin solution
h adsorption in IgG solution
h adsorption in albumin and IgG solution (mixture)
min adsorption in albumin solution followed by 1 h
adsorption in albumin and IgG solution (mixture)
1 min adsorption in IgG solution followed by 1 h
adsorption in albumin and IgG solution (mixture)
1 min adsorption in albumin solution followed by 1 h
adsorption in 25% human serum
1 min adsorption in IgG solution followed by 1 h
adsorption in 25% human serum

a
Adsorption experiments performed on PET and DEGVE surfaces for
analysis of mutual influence on albumin and IgG adsorption in the presence
of the other protein (IgG or albumin) or 25% human serum. In all performed
experiments, the albumin concentration was 10 mg/mL and the IgG
concentration was 3 mg/mL.

protein mass per area, a surface area of 2.694 cm2 is used for total
disk area. For calculations and estimations of protein density, surface
coverage, etc. performed in this paper, albumin is considered to be
a sphere having a diameter of 7 nm and molecular weight of 67 kDa,
while IgG is considered to have a diameter of 20 nm and a molecular
weight of 150 kDa.
Even though most results on protein adsorption in this paper are
presented as ng/cm2, thus as mass per area, the results should not
be considered to be interpreted as protein monolayers. The methods
and materials used during experiments underlying the presented
results do not allow us to measure the thickness of adsorbed protein
layers, and therefore, the adsorption cannot be presented as mass per
volume or amount per volume. However, the analyzed surfaces are
considered to be a three-dimensional systems consisting of a polymer
surface, a protein solution, and an interface between them where
adsorbed proteins can be more or less strongly associated with the
surface, dependent on the distance between polymer surface and
protein, as well as protein and surface characteristics.

Results
The variations of albumin and IgG adsorption from both single
protein solutions and human serum solutions onto PET surfaces
with adsorption times ranging from 5 s to 24 h are shown in
Figure 1.
Figure 1a shows the variation of albumin and IgG adsorption
from 0.25% human serum solution and single protein solutions
with corresponding albumin and IgG concentrations (0.1 mg/
mL albumin and 0.03 mg/mL IgG). In Figure 1b, the variation
of albumin adsorption from 25% human serum and from albumin
solution of corresponding concentration (10 mg/mL albumin) is
shown. It is observed in Figure 1 that the adsorption of albumin
and IgG to the PET surface is a fast process and already after
5 s of adsorption a layer of at least 100-200 ng/cm2 protein is
detected on the surface for all solutions. For the high concentration
of albumin solution an even higher adsorption (around 650 ng/
cm2) is detected after the first 5 s of adsorption.
For adsorption from single protein solutions, the albumin and
IgG adsorption increases with increased adsorption time, as
expected. After 24 h of adsorption the albumin adsorption from
0.1 mg/mL albumin ends at around 600 ng/cm2, and the IgG
adsorption from the 0.03 mg/mL IgG solution has reach a value
of approximately 1500 ng/cm2. The albumin adsorption from 10
mg/mL albumin solution is rather stable around 650-700 ng/
cm2 during the first hour of adsorption, after which there is an

Langmuir, Vol. 25, No. 4, 2009 2083

increase in the adsorption and after 24 h there is more than 900

ng/cm2 albumin detected on the surface.
Lower amounts of albumin and IgG are observed on the PET
surface when adsorption is performed from human serum
compared to that from single protein solutions. As already
mentioned, after 5 s of adsorption the amount of albumin and
IgG adsorbed from human serum is similar to that adsorbed from
single protein solution (around 100 ng/cm2). The IgG adsorption
from 0.25% human serum increased with increased adsorption
time and ends around a value of 400 ng/cm2 after 24 h of adsorption
and thus at a lower value than what is observed for adsorption
from IgG solution with similar IgG concentration. Additionally,
the adsorption rate of IgG adsorption from 0.25% human serum
is lower at adsorption times above 1 h (0.3 ng/cm2/min)
compared to adsorption times below 1 h (17 ng/cm2/min).
Adsorption of albumin from human serum increases for the
first 5-10 min of adsorption, after which there is a decrease in
albumin adsorption with increased adsorption time for adsorption
from 25% human serum and saturation in albumin adsorption
for adsorption from 0.25% human serum. After 24 h of adsorption,
the albumin adsorption is below 200 and 100 ng/cm2 from 25%
and 0.25% human serum, respectively.
From results presented in Figure 1, it can be concluded that
the presence of other proteins during adsorption influences the
outcome of albumin and IgG adsorption onto PET. However, the
mutual influence on albumin and IgG adsorption cannot be
withdrawn from these results. A series of experiments on protein
adsorptions to both unmodified PET surfaces and DEGVEmodified PET surfaces (from now on called DEGVE surfaces)
were performed at room temperature as the scheme presented
in Table 1. In all performed experiments, 10 mg/mL albumin
and 3 mg/mL IgG are used (corresponding to the concentration
in 25% human serum). Figure 2 shows the results obtained from
adsorption experiments listed in Table 1, performed on PET and
DEGVE surfaces.
Generally, the albumin and IgG adsorption is lower on the
more hydrophilic DEGVE surface than on PET, as expected.
The amount of protein detected on the DEGVE surface is in the
range of what could be considered to represent protein monolayers,
while most adsorptions onto PET show higher amounts of protein
adsorbed and cannot be considered to represent monolayer
adsorption. For example, a monolayer of 500 ng/cm2 albumin
would require a PET surface area 1.7 times larger than the one
available. IgG adsorption onto PET shows higher adsorption
than albumin in most experiments presented in Figure 2. The
adsorption of 5550 ng/cm2 IgG onto PET corresponds to
approximately 6 1013 protein molecules, which would cover
an area of 70 times larger than that available in the experiment
if adsorbed in a monolayer fashion. The results are interpreted
as albumin and IgG adsorption in a multilayer fashion, where
the proteins do not necessarily have physical contact with the
surface to be strongly connected to it, as discussed in detail later
in the text (see Discussion).
Results from most experiments presented in Table 1 follow
the same trend on both PET and DEGVE surfaces. Thus, the
adsorption from single protein solutions, as well as from mixtures
of albumin and IgG, show an increase in protein adsorption with
increased adsorption time on both surfaces. Additionally, the
adsorption of both albumin and IgG onto both surfaces is lower
when adsorption is performed from a mixture of the two proteins
compared to that from a single protein solution (see experiments
1-6). However, experiments with a head start of one protein are
somewhat different on the DEGVE surface compared to the PET
surface. When adsorption is performed onto the PET surface and

2084 Langmuir, Vol. 25, No. 4, 2009

Holmberg and Hou

Figure 1. Variation of (a) albumin and IgG adsorption onto PET surfaces from 0.25% human serum solution and single protein solutions with
concentrations of albumin and IgG corresponding to the human serum solution, and (b) albumin adsorption onto PET surfaces from 25% human
serum and single protein solutions with albumin concentration corresponding to the human serum solution (10 mg/mL albumin).

albumin or IgG have a head start before the other protein, the
adsorbed amount of the first introduced protein is higher than
when both proteins are introduced to the surface simultaneously.
At the same time, the adsorption of the second added protein is
lower than when both proteins are added simultaneously (see
experiments 6-8). For adsorption onto PET, a head start of IgG
over albumin results in approximately 48% higher adsorption of
IgG and around 19% lower adsorption of albumin compared to
when both proteins are added to the surface simultaneously. A
head start of albumin, on the other hand, does not show an
appreciable increase in the amount albumin adsorbed (around
5% increase) but a rather large decrease in the amount of IgG
adsorbed (around 79% decrease) compared to when both albumin
and IgG are introduced to the surface at the same time. Looking
at adsorption onto DEGVE surface, the results from experiments
7 and 8 look similar, and no increase or decrease in albumin
and/or IgG adsorption is observed when one protein is allowed
to adsorb for 1 min before the second protein is added compared
to when both proteins are added to the surface simultaneously.
The somewhat large error bars observed in Figure 2b are believed

to be caused by nonhomogenous DEGVE surfaces, where the

surface is not completely covered by the plasma polymerized
layer. Replacing the second added protein with 25% human serum
results in a decrease in adsorption of the first added protein (the
one that has a head start) compared to when the second added
solution is a single protein solution (see experiments 7-10). For
adsorption onto PET, the decrease is around 25% for albumin
adsorption and around 65% for IgG adsorption. For DEGVE
surfaces, the decrease is around 62% for both proteins.

Discussion
Results from single protein adsorption onto a typical polymer
surface normally show an increase in protein adsorption with
increased protein concentration in solution, as well as with
increased adsorption time. However, protein adsorption performed
from protein mixtures implies competitive protein adsorption,
where different proteins with different characteristics will compete
with each other for the available space on the surface. Aside
from the molecular weight and conformation of the proteins,

CompetitiVe Protein Adsorption

Langmuir, Vol. 25, No. 4, 2009 2085

Figure 2. Adsorption of albumin and IgG on (a) PET and (b) DEGVE surfaces from different combinations of protein adsorption processes, as
presented in Table 1.

surface and protein characteristics such as charge and wettability

are expected to have an influence on the different adsorption
processes. Thus, when our polymer surfaces are put in contact
with human serum, a whole range of different blood proteins
will interact with the surface and compete with each other for
adsorption sites. Being present in high concentrations and not
being the largest protein in the solution, albumin and IgG are
expected to dominate the surface at early stages of the competitive
protein adsorption. Later in the process other proteins will be
able to compete successfully for space on the surface and replace
some of the adsorbed albumin and IgG molecules, which is also
what is observed in the presented results, as well as former
experiments performed in our laboratory.32
Thus, adsorption from single protein solution increases with
increased adsorption time, while adsorption from human serum
shows competitive processes where adsorption of a specific
protein saturates, or even decreases, with increased adsorption
time. This is also illustrated in Figure 3, where the ratios of the
amounts of albumin and IgG adsorbed on the PET surface from
human serum (0.25% and 25%) to that from single protein

solutions (with corresponding albumin and IgG concentrations

as the human serum solutions) at adsorption times ranging from
5 s to 1 h are shown. The ratios decrease with increased adsorption
time, reflecting a situation where an increase in adsorption time
is followed by an increase in protein adsorption from single
protein solution simultaneously with a slower increase, saturation,
or a decrease in protein adsorption from human serum.
The slightly negative slopes of the curves (around 0.001) for
the adsorption from 25% human serum and corresponding single
protein solutions are rather gentle, while the slopes for 0.25%
human serum and corresponding single protein solution are more
steep (around 0.01 for IgG and around 0.005 for albumin). The
steeper slope of the 0.25% curves is explained by the fact that,
for low protein concentrations, the adsorption at short adsorption
times from single protein solutions and human serum is similar,
and at the same time the adsorption of albumin and IgG from
single protein solution increases more rapidly with adsorption
time than that from human serum solution (see Figure 1a). The
more gentle slope of the 25% curves is a consequence of a situation
where adsorption from the high concentration single protein

2086 Langmuir, Vol. 25, No. 4, 2009

Holmberg and Hou

Figure 3. Ratio of the amount of protein (albumin or IgG) adsorbed on PET surfaces from diluted human serum solutions to that from single protein
solution with the corresponding albumin and IgG concentration as the diluted human serum, as a function of adsorption time.

solution very quickly reaches a relatively high value of adsorption,

while adsorption from 25% human serum at the same time lies
around an adsorption value close to what is observed for 0.25%
human serum, around 100 ng/cm2.
Previous experiments on PET surfaces where albumin and
IgG adsorption from human serum solutions was performed for
1 and 24 h showed a decrease in albumin and IgG adsorption
with both increased adsorption time and increased protein
concentration. Hence, competitive adsorption processes occurring
at the interface between polymer surface and protein solution
have a time range of at least 24 h. However, these processes can
also be detected on a shorter time scale,11 and in Figure 1 it is
observed that the influence from competitive adsorption processes
on the content of the protein layer on the surface is in the minute
range. The decrease in albumin adsorption after approximately
5 min adsorption from human serum shows that, at this point of
adsorption, the rate of replacement of albumin by other proteins
present in the solution exceeds the adsorption rate of albumin
onto the PET surface. In a similar way, the decrease in the
adsorption rate of IgG at longer adsorption times observed in
Figure 1a can be interpreted as a situation where other proteins
start to get more successful in interacting strongly with the polymer
surface, and therefore IgG cannot adsorb to the PET surface as
readily as before. However, during the first 24 h of adsorption,
the IgG adsorption from human serum never reaches a situation
where the replacement of IgG by other proteins present in the
solution overcomes the adsorption of IgG onto the PET surface.
Therefore, we conclude that in the presence of human serum at
adsorption times above 5 min, the interaction between PET and
albumin is weaker than the interaction between PET and IgG.
By combining these results of competitive adsorption from human
serum with results from experiments as those described in Table
1, where only albumin and IgG are present in the solution, the
mutual influence of albumin and IgG adsorption can be explored.
It has already been stated that the presentation of albumin and
IgG adsorption as ng/cm2 should not be considered as an
interpretation of the results where the detected protein layer
represents a monolayer of proteins. On the contrary, the adsorbed
protein layer is not considered to be a well-defined two-

dimensional structure with a specified height corresponding to

the thickness of a protein layer but instead to be made up by a
more indistinct interface between polymer surface and protein
solution where the binding strength between protein and surface
decreases with increased distance between the two. In Figure 4,
the ng/cm2 values presented in Figure 1a on albumin and IgG
adsorption have been converted into total picomole values of
albumin and IgG present on the surface.
The adsorption trends of the different adsorption curves in
Figure 4 are the same as those presented in Figure 1a. Albumin
and IgG adsorption from single protein solution increase with
increased adsorption time, albumin adsorption from human serum
shows saturation at adsorption times above 5 min, and the IgG
adsorption rate from human serum decreases with increased
adsorption time. However, when presented as the total amount
(pmol) of protein adsorbed onto the surface, the difference
between albumin and IgG adsorption from single protein solution
at longer adsorption times is not as distinguished as in Figure
1a). Thus, the amount of molecules adsorbed onto the surface
is similar for albumin and IgG (25 pmol of albumin or IgG
represents approximately 1.5 1013 protein molecules), even
though the adsorbed IgG represents a higher mass. Considering
the results in Figure 4, one can speculate that the most important
parameter determining how much protein is adsorbed onto the
surface is the water content and/or water structure on the surface,
as also suggested by Vogler et al.27 However, the adsorptions
presented in Figure 4 are performed at low protein concentrations,
and our experience has shown that the difference in albumin and
IgG adsorption onto PET (both in ng/cm2 and pmol) accelerates
tremendously with increased protein concentration during single
protein adsorptions. Upon performing the calculation from ng/
cm2 to pmol for 24 h of adsorption from 10 mg/mL albumin and
3 mg/mL IgG (not shown), a difference in picomoles of 435%
(!) between IgG and albumin is found, with IgG showing the
larger adsorption.
The adsorption isotherms from single protein solutions
observed in Figures 1a and 4 show a typical development of
increased protein adsorption with increased adsorption time.
Furthermore, there is an increase in albumin adsorption with

CompetitiVe Protein Adsorption

Langmuir, Vol. 25, No. 4, 2009 2087

Figure 4. Adsorption results from Figure 1a presented as picomole, showing the albumin and IgG adsorption onto PET as a function of adsorption
time. Adsorption from both single protein solutions and human serum solutions is plotted in the graph.

increased albumin concentration at each adsorption time detected

(see Figure 1a,b), and thus there is an increase in protein adsorption
with increased protein concentration. Both observed relationships
are expected, and it is also expected that the rinsing procedures
used during the experiments remove all proteins from the interface
except the ones that interacts strongly with the surface.28,30 Thus,
our results indicate that the longer the protein solution is in
contact with the polymer surface and the higher the protein
concentration, the larger the amount of proteins that becomes
strongly associated with the surface. Upon considering the 25
pmol albumin adsorbed onto PET after 24 h and approximating
the surface area that each albumin molecule occupies to be around
3.8 10-13 cm2, it is clear that the albumin cannot be arranged
in a monolayer on the surface (that would require a diameter of
the PET disks around 19 mm). Furthermore, the albumin and
IgG adsorption from single protein solutions does not show
saturation after 24 h of adsorption, strengthening the idea that
the proteins are not adsorbing in a monolayer fashion. Instead,
we would describe the adsorbed proteins to be part of an interface
between polymer surface and protein solution. The proteins that
are in direct physical contact with the polymer surface (or the
water film present at the polymer surface) will be more or less
strongly affected by the surface characteristics. Dependent on
these characteristics, the interaction between protein and surface
can induce conformational changes in the protein molecules,
which in turn result in a new surface with specific characteristics
facing the protein solution, which in turn are influenced by this
new surface and so on. Thus, even though proteins are not in
direct contact with the polymer surface, they can still be affected
by the characteristics of the surface. Eventually, the authority of
the surface will be too weak to be sensed by the proteins and
the proteins can be considered to be in solution instead of
being part of the adsorbed protein layer, the interface. Considering
the results presented in Figures 1 and 4, the increase in protein
adsorption from single protein solutions with increased adsorption
time indicates that this interface is a rather dynamic structure
where proteins over time can become more and more entangled
into the protein interface and thereby be more strongly connected
to the polymer surface. As the protein concentration is increased,
these processes occur more rapidly and a higher adsorption is
observed faster.

Results presented in Figure 2 show that influence from surface

characteristics on albumin and IgG adsorption is quite different
for PET and DEGVE surfaces and that the adsorption behavior
of IgG molecules is much more sensitive to a polymer surface
with hydrophobic character than the behavior of albumin
molecules is. It is known that proteins generally bind more readily
and strongly to hydrophobic surfaces than to hydrophilic surfaces,
as mainly explained by the existence of dehydration forces during
interaction between a hydrophobic surface and proteins.18,25,38,32,35 The influence from dehydration on relaxation
and spreading of proteins on hydrophobic surfaces can even
cause unfolding of the proteins, exposing hydrophobic parts of
the proteins that normally are encapsulated by a hydrophilic
shell in the native protein structure. A globular protein like albumin
is less sensitive to dehydration during adsorption to a hydrophobic
surface, while the more flexible IgG molecule undergoes what
we call surface-induced aggregation, which is expressed as very
high amounts of adsorbed proteins (sometimes 10 times what
normally is observed on similar surfaces).23,36,37 The expression
aggregation should not be considered to represent the formation
of aggregates on the surface but to cover a condition where the
proteins are strongly affected by its environment and thereby
changes conformation, which in turn results in a matrix structure
where proteins interact strongly with each other.
To evaluate changes in surface topography and structure caused
by adsorption of a protein layer onto the PET surface and to
verify that there is no aggregation in the form of protein clusters
on the surface after protein adsorption, tapping-mode AFM
measurements in air were performed. The roughness (Sq) of a
0.5 1 m large area on each surface was obtained and is listed
in Table 2. Figure 5 shows AFM images of a bare PET surface
that only has been in contact with PBS buffer, as well as a PET
surface onto which 3 mg/mL IgG has been adsorbed for 1 h at
room temperature. Figure 5b is representative for all images
(34) Sit, P. S.; Marchant, R. E. Thromb. Haemostasis 1999, 82, 1053.
(35) Wertz, C. F.; Santore, M. M. Langmuir 2001, 17, 3006.
(36) Giacomelli, C. E.; Bremer, M. G. E. G.; Norde, W. J. Colloid Interface
Sci. 1999, 220, 13.
(37) Holmberg, M.; Stibius, K. B.; Larsen, N. B.; Kingshott, P.; Hou, X. L.
Anal. Biochem. 2007, 361, 120.
lvarez, R.; Galisteo-Gonzalez, F. Heterog. Chem. ReV. 1995,
(38) Hidalgo-A
2, 249.

2088 Langmuir, Vol. 25, No. 4, 2009

Holmberg and Hou

Figure 5. AFM images (500 1000 nm) of PET after being in contact with (a) PBS buffer for 1 h and (b) 3 mg/mL IgG solution for 1 h. Both
surfaces have been rinsed in PBS buffer after the same scheme as described for surfaces used in radioactive labeling experiments. Note that the z
axis is showed on a enlarged scale (20) compared to the x and y axes.
Table 2. Roughness of Scanned PET Samplesa
surface
PET
PET
PET
PET

+
+
+
+

PBS buffer for 1 h + rinse

10 mg/mL albumin for 1 h + rinse
3 mg/mL IgG for 1 h + rinse
0.03 mg/mL IgG for 1 h + rinse

roughness (Sq), nm
0.949 ( 0.1
0.874 ( 0.1
0.980 ( 0.1
1.251 ( 0.1

a
Roughness of PET films measured with AFM before and after protein
adsorption experiments. All measurements were performed in tapping mode
in air with a standard probe.

obtained on PET surfaces with adsorbed protein, independent of

protein type and protein concentration used during adsorption.
All surfaces analyzed with AFM showed similar roughness
(Table 2) and both AFM images in Figure 5 resemble each other
in surface structure. However, a surface structure on a smaller
scale than the roughness can be observed on the PET surface
with adsorbed proteins (Figure 5b), but is not observed on the
bare PET surface (Figure 5a). This surface structure can be
explained by a stronger interaction between surface and probe
during analysis of PET with an adsorbed protein layer. When
scanning the surface with adsorbed proteins, the tip will stick
more to the surface than when scanning a bare PET surface, and
thereby more variation in the surface structure will be observed
on the resulting AFM image. No larger objects (bumps) were
observed on any of the analyzed surfaces, indicating that adsorbed
proteins are homogeneously distributed over the surface. To
confirm if material were deposited on the PET surface after protein
adsorption, scratching experiments were performed. Here an
area of 100 100 nm was scanned with a higher scanning force,
after which an area of 500 500 nm was imaged with a lower
scanning force on both bare PET and PET with adsorbed IgG
(3 mg/mL). No difference in roughness or surface structure was
observed on the bare PET surface after scratching, while a 12
nm high structure was observed in connection to the scratched

area on the PET surface with adsorbed proteins, indicating that

adsorbed proteins have been scratched off in the area scanned
with higher scanning force (not shown). Thus, no globular
aggregates are observed on the PET surfaces after protein
adsorption, and we consider the high amount of proteins detected
during radioactive labeling experiments to be evenly distributed
over the PET surfaces in a multilayer fashion. It should also be
noted that there can be multilayer protein adsorption onto a
polymer surface without induced protein aggregation.
No surface-induced aggregation is observed on the DEGVE
surface (see Figure 2), supporting the hypothesis that it is the
hydrophobic character of the PET surface that results in the very
high IgG adsorption. Furthermore, the result from experiment
7 in Figure 2a points toward the large influence of the environment
the adsorption is taking place within on IgG adsorption. In this
experiment, albumin is given a 1 min head start over IgG, resulting
in a more hydrophilic environment for the IgG to adsorb within,
and thereby the surface induced IgG aggregation is inhibited.
Giving one protein a head start over the other will give the
first introduced protein possibility to relax and spread on the
surface. Blocking of surfaces with proteins (normally bovine
serum albumin, BSA) to reduce unspecific protein adsorption is
routinely used in laboratories all over the world. Thus, the decrease
in adsorption of the second added protein in experiments 7 and
8 in Figure 2 is expected, both in relation to single protein
adsorption and when both proteins are added to the surface
simultaneously. However, the characteristics of the surface seem
to have an influence on these processes. Comparing adsorption
experiments 6 and 8 on PET and DEGVE in Figure 2 shows that
a head start of one protein does not have that large an influence
on the following adsorption of the second added protein on the
DEGVE surface. Our interpretation of the results is that on a
more hydrophilic surface the proteins are not that strongly

CompetitiVe Protein Adsorption

connected to the surface and are thereby easier to replace by

other proteinsseven if they have a head startswhich also is
expressed as a lower protein adsorption on the more hydrophilic
DEGVE surface compared to PET. Looking at competitive
adsorption and head start experiments involving human serum,
one can see that spreading of proteins on the surface seems to
make the proteins more difficult to replace and that the
replacement of adsorbed albumin onto PET is a rather slow
process, even in the presence of high concentration human serum
solutions with many high-affinity proteins. Results on albumin
adsorption from human serum where all proteins are added to
the surface simultaneously (Figure 1) and results on albumin
adsorption when albumin has a head start over human serum
(Figure 2, experiment 9) show that more albumin is detected on
the surface in the experiment using a head start. Here the amount
of adsorbed albumin is 422 ng/cm2, while it is under 100 ng/cm2
when all proteins are added to the surface simultaneously. Thus,
giving albumin the opportunity to spread on the surface can
enhance the blocking capacity of albumin appreciable.
However, it should be noted that even though the head start
of one protein results in a reduction in adsorption of the second
added protein, the second introduced protein is still detected on
the surface. Besides indicating that there is competitive adsorption,
the results also strengthen the idea that proteins can adsorb in
a multilayer fashion and thus that proteins can adsorb onto
proteins. In our experience, protein adsorption only shows values
low enough for representing monolayer adsorption on rather
hydrophilic polymer surfaces, as also observed in Figure 2, where
the amount of adsorbed albumin and IgG is reduced when
adsorption is performed on the DEGVE surface compared to
being performed on the PET surface. Furthermore, when high
concentration protein solutions, such as 50%-100% human serum
are used, even quite hydrophilic and normally nonfouling surfaces
show tendencies to adsorb more protein than what is compatible
with monolayer protein adsorption.
Besides the obvious value in obtaining knowledge regarding
competitive adsorption, the information obtained from these
studies is relevant for materials and protocols used within the
biotechnology and biomedical area today, as well as for evaluating
possible biomaterials in the future. The relatively large influence
on IgG structure and IgG adsorption from the characteristics of
a polymer sample should be taken into consideration when

Langmuir, Vol. 25, No. 4, 2009 2089

working with systems, including polymer surfaces and unspecific

protein adsorption. Even though our results show that blocking
the surface using albumin molecules results in a lower amount
of IgG adsorbed to the polymer surface, it is also observed from
the analysis that the surface is not completely blocked and that
it is still possible for IgG to adsorb to the surface after blocking
is performed. Furthermore, when working with systems involving
unspecific protein adsorption to polymer surfaces, such as, for
example, ELISA setups, the influence on protein conformation
from surface characteristics is of great importance. Errors in
detected signals from these assays can be caused by both
inefficient blocking of the surface, resulting in signals from
unspecific adsorbed molecules, as well as from unfolding of
adsorbed proteins, resulting in, for example, inactivation of their
binding sites and thereby less of a possibility for recognition by
other molecules.

Conclusions
Competitive adsorption among albumin, IgG, and other proteins
present in human serum is monitored by using radioactive labeling,
where different proteins are labeled with different isotopes.
Through this technique it is shown that the time frame within
which adsorption of albumin onto PET is overtaken by
replacement of albumin from the surface by other proteins present
in human serum is in the minute range. Both albumin and IgG
adsorption can be reduced by making the PET surface more
hydrophilic through plasma polymerization, and results show
that the rather flexible IgG molecule is much more sensitive to
surface wettabiliy than the more rigid and globular albumin
molecule. Our results indicate that the adsorbed proteins left on
the surface after rinsing represent an interface between polymer
surface and protein solution with proteins that are strongly
connected to the surface and that albumin and IgG can adsorb
in a multilayer fashion onto the polymer surfaces.
Acknowledgment. We acknowledge the Danish Agency for
Science, Technology, and Innovation for funding the project
Competitive Protein Adsorption to Polymer Surfaces, grant
no. 274-07-0350, as well as Jrgen Garns at DFM A/S
(www.dfm.dtu.dk) for valuable support and discussion regarding
AFM measurements and analysis of obtained AFM data.
LA8031978