Glucose-sensitive holographic sensors for monitoring bacterial growth

Article abstract of DOI:10.1021/ac049334n

A glucose sensor comprising a reflection hologram incorporated into a thin, acrylamide hydrogel film bearing the cis-diol binding ligand, 3-acrylamidophenylboronic acid (3-APB), is described. The diffraction wavelength (color) of the hologram changes as t

Full text of DOI:10.1021/ac049334n

Anal. Chem. 2004, 76, 5748-5755
Glu c o s e -S e n s it ive Ho lo g ra p h ic S e n s o rs fo r
Mo n it o rin g Ba c t e ria l Gro w t h
Me i-Ching Le e , Sa tya m oorthy Ka bila n, Abid Hus s a in, Xia oping Ya ng, J e ff Blyth, a nd
Chris tophe r R. Low e *
Institute of Biotechnology, University of Cambridge, Tennis Court Road, Cambridge, CB2 1QT, U.K.
A glucose sensor comprising a reflection hologram incor-
porated into a thin, acrylamide hydrogel film bearing the
cis-diol binding ligand, 3 -acrylamidophenylboronic acid
Glucose levels in fermentation have also been monitored using
high-performance liquid chromatography. However, this would
not be appropriate for use in small-scale bioreactors as it requires
4
(
3 -AP B), is described. The diffraction wavelength (color)
destructive sampling. Similarly, while near-infrared spectroscopy
of the hologram changes as the polymer swells upon
binding cis-diols. The effect of various concentrations of
glucose, a variety of mono- and disaccharides, and the
r-hydroxy acid, lactate, on the holographic response was
investigated. The sensor displayed reversible changes in
diffraction wavelength as a function of cis-diol concentra-
tion, with the sensitivity of the system being dependent
on the cis-diol tested. The effect of varying 3 -AP B con-
centration in the hydrogel on the holographic response
to glucose was investigated, and maximum sensitivity was
observed at a functional monomer concentration of 2 0
mol %. The potential for using this holographic sensor to
detect real-time changes in bacterial cell metabolism was
demonstrated by monitoring the germination and subse-
quent vegetative growth of Ba cillu s su btilis spores.
has also been utilized to monitor glucose levels in bacterial
fermentations, this technique suffers from interference from other
biological constituents and requires costly detection equipment.5
Thus, there is a clear need for stable, inexpensive, and
“nonconsumptive” sensors for use at physiological pH values in
bioprocess monitoring in small-volume bioreactors. Since synthetic
recognition elements are generally more robust than their biologi-
cal counterparts, and do not consume substrate by virtue of the
fact they tend to be equilibrium rather than catalytic systems, they
are particularly suited for use in monitoring small-volume biosys-
tems. The most effective synthetic ligands for binding glucose in
aqueous media are based on the use of boronic acids.6
The ability of boronic acid to reversibly bind cis-diols such as
glucose has long been established.7
-10
At low pH, boronic acid
exists in an uncharged, trigonal planar configuration (Figure 1a,
), which does not readily complex with cis-diols, although it can
1
result in the formation of a strained molecule (Figure 1a, 3 ),11
which is easily hydrolyzed. The strained complex 3 can, however,
react with OH- to form the more stable tetrahedral state (Figure
There is a requirement for sensor technologies that can be
exploited to monitor cell metabolism in microbial fermentation
processes and cell cultures. Such systems could be utilized in high-
throughput screening to monitor the response of cells to potential
therapeutics, optimize protein production, or monitor antibiotic
or end product formation. The sensors may be used in small-
volume microbial systems (nL) that can provide continuous
measurements of metabolites of fermentation, particularly glu-
cose,1 and allow small-scale bioprocesses to be monitored,
optimized, and controlled.
1
a, 4 ), which possesses a negative charge.12 At pH values above
13
the pK
a
, ∼8.8, boronic acid exists in a tetrahedral state (Figure
1
a, 2 ) which binds cis-diols more readily.13 Consequently, in most
applications of these boronate analogues, the pH must be basic,
i.e., >8, for effective binding of sugars. This presents a challenge
for monitoring glucose in biological matrixes.
Thus, there has been a growing interest in the use of boronic
acid ligands for detecting glucose in biological samples. Most of
these detection systems have been based around the use of
fluorescence,14 which can suffer from the large background
Conventionally, glucose sensors utilize the enzyme glucose
oxidase; however, the use of catabolic enzymes is problematic
because they consume their substrate and would significantly
reduce substrate levels in small-volume bioreactors and thereby
alter the metabolism of living cells. Furthermore, enzyme-based
(
4) Larson, T. M.; Gawlitzek, M.; Evans, H.; Albers, U.; Cacia, J. Biotechnol.
Bioeng. 2 0 0 2 , 77, 553-563.
sensors often display poor stability as a result of long-term contact
with process media 2 and problems associated with sterilization.3
(5) Arnolda, S. A. C. J.; Vaidyanathan, S.; Matheson, L.; Mohan, P.; Hall, J. W.;
Harvey, L. M.; McNeil, B. Enzyme Microb. Technol. 2 0 0 0 , 27, 691-697.
(
(
6) Davis, A. P.; Wareham, R. S. Angew. Chem., Int. Ed. 1 9 9 9 , 38, 2978-2996.
7) Lorand, J. P.; Edwards, J. O. J. Am. Chem. Soc. 1 9 5 9 , 24, 769-774.
*
To whom correspondence should be addressed. Phone: +44 (0)1223
(8) James, T. D.; Harada, T.; Shinkai, S. J. Chem. Soc., Chem. Commun. 1 9 9 3 ,
3
34160. Fax: +44 (0)1223 334162. E-mail: crl1@biotech.cam.ac.uk.
857-860.
(
1) Matanguihan, C.; Michaels, J.; Thrift, J.; Joeris, K.; Heidemann, R.; Zhang,
C.; Reddy, C.; Wu, P.; Konstantinov, K. Biotechnology 2000 Congress, Berlin,
Germany, September 3-8, 2000.
(9) James, T. D.; Linnane, P.; Shinkai, S. Chem. Commun. 1 9 9 6 , 281-288.
(10) Shinkai, S.; Tsukagoshi, K.; Ishikawa, Y.; Kunitake, T. J. Chem. Soc., Chem.
Commun. 1 9 9 1 , 1039-1041.
(
(
2) DiCesare, N.; Lakowicz, J. R. Anal. Biochem. 2 0 0 1 , 294, 154-160.
3) von Woedtke T. J. W.; Hartmann, V.; Stieber, M.; Abel, P. U. Biosens.
Bioelectron. 2 0 0 2 , 17, 373-382.
(11) Springsteen, G.; Wang, B. Tetrahedron 2 0 0 2 , 58, 5291-5300.
(12) Hisamitsu, I. K. K.; Okano, T.; Sakurai, Y. Pharm. Res. 1 9 9 7 , 14, 289-293.
(13) Liu, X. C.; Scouten, W. J. Chromatogr., A 1 9 9 4 , 687, 61-69.
5748 Analytical Chemistry, Vol. 76, No. 19, October 1, 2004
10.1021/ac049334n CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/03/2004

wavelength or color of diffracted light. At least one embodiment
of this sensor functions at physiological pH and ionic strength,
with the diffraction wavelength blue shifting at low glucose
concentrations (up to 10 mM) and then red shifting at higher
glucose concentrations. However, since the majority of these
studies were performed at pH 8.5 and there is a change in
direction of the diffraction wavelength shift at high glucose
concentrations, this system may not be applicable to some
bacterial fermentations, where the glucose concentrations can
range from very high (>25 mM) to near zero. The fabrication of
these sensors is also time-consuming and not easily amenable to
mass manufacture.
An alternative method for detecting glucose using boronic acid
ligands would be through the use of glucose-sensitive holographic
sensors.24 These sensors are based on planar small-volume
polymer hydrogels containing silver halide, which acts as a
holographic recording material. A reflection hologram is recorded
within the polymer hydrogel matrix, producing an optical sensor
that only reflects a narrow band of wavelengths when illuminated
with white light.25 The hydrogels used are termed “smart”
polymers with ligands that interact with specific analytes incor-
porated into the polymer matrix.2
5-32
This allows the hydrogel to
Fig u re 1 . (a) Reversible binding that occurs between boronic acid
and cis-diols in aqueous media. (b) Structure of 3-acrylamidophen-
ylboronic acid.
respond to specific stimuli by changing its swelling state or
hydration upon analyte binding.33 The change in swelling state of
the polymer causes a change in the wavelength of light diffracted
by the holographic grating within the hydrogel. The narrow band
of wavelengths diffracted is determined by the spacing between
silver (Ag ) nanoparticle fringes. Thus, the hologram acts as a
monochromatic mirror, in that light with a wavelength of ap-
proximately double the distance between the fringes will emerge
in phase at a specific angle and constructive interference occurs.
The diffraction wavelength of the hologram is governed by Bragg’s
law:
fluorescence often found in complex biological media and may
also be affected by temperature and oxygen concentration. Many
fluorescence receptors often display low solubility in aqueous
media, have low photochemical stability,15 and may require the
presence of solvents such as methanol to function.16,17 Further-
more, the turbidity of cell growth media will also affect the
performance of fluorescence-based optical sensors.18 Electro-
0
29
chemical boronic acid-based glucose sensors have also been
reported,19,20 but these can suffer from surface fouling and
interference from redox-active biological components.21 A glucose
λ ) 2nD cos θ
sensor based on a boronic acid-containing hydrogel on a quartz
crystal microbalance (QCM)22 has also been reported. QCM
transduction systems, however, are particularly difficult to imple-
ment in real-time and low-volume sensor systems.
where θ is the angle of incidence, D is the distance between
fringes, λ is the diffraction wavelength, and n is the average
refractive index in vacuo.
A glucose sensor has been developed using phenylboronic acid
groups copolymerized within an acrylamide hydrogel containing
a colloidal crystalline array.23 The colloidal crystalline array reports
on the swelling state of the hydrogel by a change in the
(
23) Alexeev, V. L.; Sharma, A. C.; Goponenko, A. V.; Das, S.; Lednev, I. K.;
Wilcox, C. S.; Finegold, D. N.; Asher, S. A. Anal. Chem. 2 0 0 3 , 75, 2316-
2323.
24) Kabilan, S.; Blyth, J.; Lee, M. C.; Marshall, A.; Hussain, A.; Yang, X. P.;
Lowe, C. R. J. Mol. Recognit. 2 0 0 4 , 17, 162-166.
25) Mayes, A. G.; Blyth, J.; Millington, R. B.; Lowe, C. R. J. Mol. Recognit. 1 9 9 8 ,
11, 168-174.
(
(
(
(
(
(
(
(
14) Karnati, V. V.; Gao, X.; Gao, S.; Yang, W.; Ni, W.; Sankar, S.; Wang, B. Bioorg.
Med. Chem. Lett. 2 0 0 2 , 12, 3373-3377.
15) Yang, W. Q.; Yan, J.; Springsteen, G.; Deeter, S.; Wang, B. H. Bioorg. Med.
Chem. Lett. 2 0 0 3 , 13, 1019-1022.
16) James, T. D.; Sandanayake, K. R. A. S.; Shinkai, S. Angew. Chem., Int. Ed.
Engl. 1 9 9 4 , 33, 2207-2209.
(26) Marshall, A,; Blyth, J.; Davidson C. A. B.; Lowe, C. R. Anal. Chem. 2 0 0 3 ,
75, 4423-4431.
(27) Marshall, A. J.; Young, D. S.; Blyth, J.; Kabilan, S.; Lowe, C. R. Anal. Chem.
2 0 0 4 , 76, 1518-1523.
(28) Millington, R. B.; Mayes, A. G.; Blyth, J.; Lowe, C. R. Anal. Chem. 1 9 9 5 ,
17) Badugu, R.; Lakowicz, J. R.; Geddes, C. D. Anal. Chem. 2 0 0 4 , 76, 610-
67, 4229-4233.
6
18.
18) Junker, B. H.; Wang, D. I. C.; Hatton, T. A. Biotechnol. Bioeng. 1 9 8 8 , 32,
5-63.
(29) Millington, R. B.; Mayes, A. G.; Blyth, J.; Lowe, C. R. Sens. Actuators, B
1 9 9 6 , 33, 55-59.
(30) Blyth, J.; Millington, R. B.; Mayes, A. G.; Frears, E.; Lowe, C. R. Anal. Chem.
5
19) Kikuchi, A.; Suzuki, K.; Okabayashi, O.; Hoshino, H.; Kataoka, K.; Sakurai,
1 9 9 6 , 68, 1089-1094.
Y.; Okano, T. Anal. Chem. 1 9 9 6 , 68, 823-828.
(31) Mayes, A. G.; Blyth, J.; Kyrolainen-Reay, M.; Millington, R. B.; Lowe, C. R.
Anal. Chem. 1 9 9 9 , 71, 3390-3396.
(32) Mayes, A. G.; Blyth, J.; Millington, R. B.; Lowe, C. R. Anal. Chem. 2 0 0 2 ,
(
(
20) Norrild, J. C.; Sotofte, I. J. Chem. Soc., Perkin Trans. 2. 2 0 0 2 , 303-311.
21) Kress-Rogers, Chemosensors, Biosensors and Immunosensors. In Instru-
mentation and Sensors for the Food Industry; Butterworth-Heinemann Ltd.:
Oxford, U.K., 1993; Chapter 17.
74, 3649-3657.
(33) Lowe, C.; Blyth, J.; Marshall, A.; James, A.; Kabilan, S.; Sartain, F.; Lee M.
C.; Madrigal-Gonzalez, B.; Yang, X.; Davidson, C. Int. Soc. Opt. Eng. 2 0 0 3 ,
3, 20-23.
(
22) Malitesta, C.; Losito, I.; Zambonin, P. G. Anal. Chem. 1 9 9 9 , 71, 1366-
1
370.
Analytical Chemistry, Vol. 76, No. 19, October 1, 2004 5749

As the hologram swells, the diffraction wavelength increases
and the color red shifts. Contraction of the polymer causes a
decrease in diffracted wavelength and a blue shift. The changes
in diffraction wavelength can be quantified using a suitable
reflection spectrophotometer.
Instrumentation. Holograms were analyzed using an Avantes
AVS-MC2000-2 reflectance spectrometer and AvaSoft 5 processing
software (Knight Scientific).
Synthesis of 3 -Acrylamidophenylboronic Acid. A well-
stirred, cooled solution of 3-aminophenylboronic acid monohydrate
(5.0 g, 32.2 mmol) in aqueous sodium hydroxide (5.16 g, 129.04
mmol, 50 mL) was treated with dropwise addition of acryloyl
chloride (5.2 mL, 64.4 mmol) over a period of ∼10 min. After 30
min of stirring on ice, the reaction mixture was allowed to reach
room temperature and stirred for a further 2 h. The mixture was
adjusted to pH 8 using dilute HCl (0.1 M). The resulting beige
precipitate was filtered off and washed 5 times with 25 mL of cold
water. Drying in air resulted in a fine powder, which was then
dissolved in hot 20% (v/ v) aqueous ethanol (∼50 mL) and filtered
through fluted filter paper. The filtrate was left to stand overnight
at room temperature, and the resulting crystals were filtered,
A holographic glucose sensor was produced initially by
copolymerizing 4-vinylphenylboronic acid with acrylamide and
N,N′-methylenebisacrylamide.24 This sensor only responded well
to glucose at pH 9 as the boronic acid ligand used was thought
to have a high pK
a
value.20 It has been shown, however, that
hydrogels made from 3-acrylamidophenylboronic acid (3-APB;
Figure 1b) copolymers bind glucose at physiological pH.12,34 Thus,
copolymers of 3-APB and acrylamide with N,N′-methylenebis-
acryalmide as a cross-linker were synthesized and holograms
recorded within them for use as glucose sensors. The response
of the 3-APB copolymers to glucose, as well as a variety of
saccharides, lactate, and a glycosylated protein, all of which
contain cis-diols and can be found in cell growth media, was
investigated. The effect of altering the mole percent of 3-APB in
the hydrogel on the sensitivity to glucose was also studied. Finally,
glucose depletion was monitored continuously as a function of
bacterial growth using the holographic glucose sensor, to dem-
onstrate utility of the new optical sensor under physiological
conditions.
1
crushed, and left to dry in a desiccator. The H NMR spectrum
3
4
of the product was consistent with previously published data,
1
and the purity of the product was >99% as adjudged by H NMR
integration.
Synthesis of P olymer Films. The appropriate quantity of each
monomer was dissolved to give a prepolymer solution with the
required molar ratios. In each case, the amount of cross-linker,
MBA, was kept constant at 1.5 mol %. The monomers were
dissolved in DMSO containing 2% (w/ v) of the photoinitiator
DMPA. The solid-to-solvent ratio of monomer to DMSO was kept
constant throughout at 0.452:1 (w/ v). A 100 µL droplet of
monomer solution was pipetted onto the polyester surface of an
aluminised polyester sheet sitting on a glass plate. A glass
microscope slide, which had been treated with 3-(trimethoxysilyl)-
propyl methacrylate,31 was then gently lowered, silane-treated side
down, onto the monomer mixture. Films were polymerized by
UV-initiated free radical reaction at 20 °C for 30 min. Polymerized
films were peeled off the polyester sheet while submerged in
deionized water. Prior to hologram construction, the edges of each
film were cleaned with a scalpel blade to remove any excess
material.
EXPERIMENTAL SECTION
Materials. All chemicals were of analytical grade unless
otherwise stated. 1,1′-Diethyl-2,2′-cyanine iodide (QBS, photosen-
sitizing dye), 2,2-dimethoxy-2-phenylacetophenone (DMPA), R -
1
acid glycoprotein (AGP), acrylamide (electrophoresis grade),
ammonium sulfate, â-deoxyribose, dimethyl sulfoxide (DMSO),
dipotassium phosphate, â-
chloric acid, hydroquinone,
maltose, (+)-mannose, monopotassium phosphate, N,N′-meth-
ylenebisacrylamide (MBA), potassium bromide, R- (-)-ribose,
D
(-)-fructose,
D
(+)-galactose, hydro-
L-lactate, lactose, magnesium sulfate,
D
D
silver nitrate (1 M, volumetric standard), sodium citrate, sodium
hydroxide, sodium thiosulfate, sucrose, acryloyl chloride and
tryptophan were purchased from Sigma Chemical Co.. Ascorbic
acid and 3-aminophenylboronic acid monohydrate was obtained
Hologram Construction. Holograms were constructed using
the diffusion method as described by Blyth et al.35 The polymer
films were placed face-down onto 400 µL of 0.2 M silver nitrate
for 2 min. Excess solution was wiped off, and the film was dried
in a stream of warm air. Under red safe lighting, the slides were
placed film side up in 40 mL of 4% (w/ v) KBr and 0.05% (w/ v)
from Avocado Research Chemicals. (+)-Glucose was purchased
D
from ICN Biomedicals. 4-Methylaminophenol sulfate (Metol) was
purchased from Acros Chemicals. Methanol and acetic acid
(
glacial) were obtained from Fisher Scientific Ltd. Phosphate-
buffered saline (PBS) tablets providing 10 mM phosphate, 137
mM NaCl, 2.7 mM KCl (pH 7.4 at 25°C) when dissolved in 200
mL of deionized water were purchased from Fluka. Tryptone soya
broth was purchased from Oxoid Ltd. Bacto peptone digest was
obtained from Difco. Thermo Electron Infinity Glucose Hexo-
kinase kit purchased was from Thermo Trace (Melbourne,
Australia).
Equipment. Microscope slides (Super Premium, 1-1.2 mm
thick, low iron) were purchased from BDH (Merck) Ltd. Alumi-
nized, 100-µm polyester film (grade MET401) was purchased from
HiFi Industrial Film Ltd. A UV Exposure Unit (∼350 nm, model
no. 555-279) was purchased from RS Components. A frequency-
doubled Nd:YAG laser (350 mJ, 532 nm, Brilliant B, Quantel) was
used in hologram construction.
2
ascorbic acid in 1:1 methanol/ H O (v/ v) with 1 mL of 0.1% (w/ v)
QBS in methanol and agitated for 1 min. The films were rinsed
under distilled water and immersed polymer side down into the
hologram exposure bath31,26 containing PBS, pH 7.4, for 10 min.
The whole area of the slide was then exposed to three single
pulses from a frequency-doubled (532 nm) Nd:YAG laser. The
slide was removed from the exposure bath and agitated for ∼10
s in freshly made developer solution (40 mL of 20 g/ L ascorbic
acid, 5 g/ L 4-methylaminophenol sulfate, 20 g/ L sodium carbon-
ate, 6.5 g/ L sodium hydroxide). The developed hologram was
rinsed under distilled water and immersed in stop solution (5%
(
v/ v) acetic acid). After rinsing under distilled water, the film was
immersed for 5 min in agitated 20% (w/ v) sodium thiosulfate to
(
34) Kanekiyo, Y.; Masahito, S.; Ritsuko I.; Shinkai, S. J. Polym. Sci., Part A:
Polym. Chem. 2 0 0 0 , 38, 1302-1310
(35) Blyth, J.; Millington, R. B.; Mayes, A. G.; Lowe, C. R. Imaging Sci. J. 1 9 9 9 ,
47, 87-91.
5750 Analytical Chemistry, Vol. 76, No. 19, October 1, 2004

remove any undeveloped silver and rinsed in methanol and then
distilled water.
HP 8452A diode array spectrophotometer. The configuration of
the hologram in the cuvette precluded any affect of the cell density
on the diffraction wavelength since the optical path for the
reflection hologram did not traverse the cell suspension. The pH
of the medium was also monitored using a Hanna microelectrode.
Monitoring Holographic Responses. Single strips of holo-
gram, ∼8 mm wide were cut from a slide and inserted into a 4-mL
plastic cuvette with the film side facing inward. PBS (1 mL) was
added and the cuvette left overnight to equilibrate at 30 °C in a
temperature-controlled cuvette holder. An Avantes AVS-MC2000-2
reflection spectrophotometer equipped with AvaSoft 5 processing
software was used to measure the wavelength of light reflected
from a white light source by the hologram as determined by
spacing of the fringes within the polymer.32
RESULTS AND DISCUSSION
The response of a 12 mol % 3-APB hologram to glucose in PBS
pH 7.4 at 30°C was investigated. Figure 2a shows that the
diffraction wavelength of the hologram red shifts on addition of
glucose, indicating that the polymer expands upon binding
glucose. A similar result was observed by Asher et al.38 with an
acrylamide copolymer containing pendant phenylboronic acid
groups. They attributed this to the formation of charged, tetra-
hedral phenylboronate groups (Figure 1a, 4 ) on binding of
glucose. The presence of the charged boronate anions creates a
Donnan potential, which results in water partitioning into the
hydrogel and thereby causing it to swell. The wavelength response
of the sensor was directly related to the glucose concentration
(Figure 2b) over the normal physiological range of blood glucose
concentrations in humans39 and could be extended up to 150 mM.
The half time (t1/ 2) required to reach the maximum diffraction
wavelength change at 2 mM glucose in PBS buffer, pH 7.4, with
a 12 mol % 3-APB hologram was 10.5 ( 0.99 min for triplicate
measurements. The calibration of the sensor deviated by 10% or
less for variations in ionic strength of up to 50 mM NaCl in PBS
buffer, pH 7.4, indicating a low susceptibility to changes in ionic
strength within the range normally anticipated in a fermentation
broth.
When the hologram was thoroughly washed in an excess of
buffer, the peak returned to within (3 nm of the initial baseline
wavelength, demonstrating the fully reversible response of the
sensor. The sensor was stored dry at room temperature and could
be used repeatedly over the course of several months without
significant degradation of the response to glucose, attesting to
the robust nature of holographic sensors.26
A control copolymer comprising 12 mol % N-phenylacrylamide
instead of 3-APB in acrylamide produced a sensor that displayed
no response to glucose, indicating that the observed shifts in
diffraction wavelength with 3-APB copolymers were due to the
interaction of glucose with the phenylboronic acid ligand, rather
than nonspecific interaction with the polymer backbone.
The sensitivity of holographic sensors can be controlled by
altering the amount of the active monomer or ligand present in
the hydrogel, with the sensitivity of the system generally increas-
ing with active monomer concentration.26,32 Thus, the mole percent
of 3-APB incorporated into the acrylamide copolymer was varied
and the response to glucose tested in PBS (Figure 3).
The response of the holographic sensors to (+)-glucose in
D
PBS was tested. An overnight solution was used to ensure the
relative proportion of glucose isomers present would mimic the
equilibrium mixture found in media or biological fluids. The
response of the sensor to the addition of six 20-µL aliquots of 0.1
M glucose in PBS to the cuvette was recorded. Between each
addition, the sensor was allowed to equilibrate to a stable
diffraction wavelength. A 2 × 5 mm magnetic follower (Fisher)
and stirrer arrangement was used to ensure constant agitation.
Deoxyribose, fructose, galactose, lactate, lactose, maltose,
mannose, ribose, and sucrose (0.1 M solutions) were also tested
using a 12 mol % 3-APB hologram in a similar manner.
The heavily glycosylated protein (45 wt %), AGP, was selected
as a model glycoprotein. It has a molecular mass of 41-43 kDa,
and 45% of its weight is due to its glycosylated chains.36 The
response of the hologram to e3 mg/ mL AGP was investigated.
Monitoring Glucose Depletion in a Culture Medium. A
1
2 mol % 3-APB hologram was equilibrated in 2 mL of fresh
minimal medium. The medium consisted of 7 g/ L K HPO , 2 g/ L
KH PO , 0.5 g/ L C NaO , 0.1 g/ L MgSO4, 1 g/ L (NH4) SO
g/ L glucose, and 5 g/ L Bacto peptone digest at pH 7.1. Bacillus
2
4
2
4
6
H
7
7
2
4
,
5
subtilis NCTC3610 spores were prepared using the CCY medium
method as described by Stewart et al.37 and stored in water at
4
°C. A 4-µL aliquot of spores was added to the medium containing
7
the hologram to give a final concentration of 4.2 × 10 cells/ mL
(
as determined by plate counts).
The cuvette was incubated at 30 °C, agitated using a magnetic
stirrer, and the holographic diffraction wavelength recorded for
0 h. The cuvette was capped with a sterile sponge bung to allow
1
free diffusion of oxygen. The medium was aerated using a
peristaltic pump with a 0.2-µm Minisart (Sartorius) filter attached
to a sterile needle. To verify the glucose concentration of the
medium, 50-µL samples were removed from the cuvette at specific
time points using a sterile syringe. The samples were centrifuged
for 1 min at 16000g, and the supernatant was frozen using liquid
nitrogen. Glucose determination of the supernatant was performed
spectrophometrically using a Thermo Electron Infinity Glucose
Hexokinase kit (Catalog No. TR15421). Glucose concentrations
in the growth medium were interpolated from a calibration curve
comprising culture medium supplemented with 0-25 mM glucose
by reference to a best-fit linear equation. Triplicate measurements
were typically <(5% for glucose measurements with the hexo-
kinase kit. The optical density (A 600 nm) of the growing bacteria
was also monitored in identical experimental conditions using a
Sensitivity to glucose increased as a function of mole percent
of 3-APB in the polymer up to 20 mol % (Figure 3a). Using a 20
mol % 3-APB hologram, a change of ∼0.5 mM glucose could be
resolved, based on the resolution of the detection system used.
The small error bars in Figure 3a demonstrate the reproducibility
of the sensor. Furthermore, as seen in Figure 3b, a maximum
(
38) Asher, S. A.; Alexeev, V. L.; Goponenko, A.. V.; Sharma, A.. C.; Lednev, I.
K.; Wilcox, C. S.; Finegold, D. N. J. Am. Chem. Soc. 2 0 0 3 , 125, 3322-
3329.
(
(
36) Hochepied, T.; Berger, F. G.; Baumann, H.; Libert, C. Cytokine Growth Factor
Rev. 2 0 0 3 , 14, 25-34.
37) Stewart, G. S.; Johnstone, K.; Hagelberg, E.; Ellar, D. J. Biochem. J. 1 9 8 1 ,
(39) Eggert, H. J. F.; Morin, C.; Norrild, J. C. J. Org. Chem. 1 9 9 9 , 64, 3846-
1
98, 101-106.
3856.
Analytical Chemistry, Vol. 76, No. 19, October 1, 2004 5751

Fig u re 2 . (a) Change in diffraction wavelength of a hologram containing 12 mol % 3-APB as a function of glucose concentration in PBS
pH 7.4 at 30 °C. The glucose concentration values are displayed above the diffraction peaks. The hologram red shifts with an increase in
glucose concentration and returns to within 3 nm of the original diffraction wavelength when washed with PBS. (b) Calibration curve for diffraction
wavelength versus glucose concentration.
response toward glucose was observed at 20 mol % 3-APB. Above
this concentration, the sensor became less responsive, presumably
due to alterations in the elastic properties of the polymer at high
ligand concentrations, rendering it less sensitive to modulation
by glucose binding.32 Since 3-APB is a planar, hydrophobic
molecule, it is possible that the polymer simply becomes more
rigid as 3-APB molecules stack together within the polymer at
high mol percentages.
Boronic acids are known to bind a wide variety of cis-diols.7,16,40
As such, any sensor system utilizing boronic acid ligands would
be susceptible to interference from competing cis-diol-containing
compounds in the sensing environment. The sensor response to
various mono- and disaccharides was tested in order to investigate
the cross-reactivity of the holographic glucose sensor with other
analytes commonly found in cell culture media (Figure 4). The
response to lactate, an R-hydroxy acid, which is an important
product of anaerobic metabolism, was also investigated as it has
been shown that lactate can bind to boronic acid.40
The position of the hydroxyl groups on different sugars, such
as mannose and galactose, affects the affinity and thus the
response of the sensor when compared to glucose, as binding is
known to occur only with suitably configured cis-diols.41 The
hologram responded fastest to fructose, which also caused the
greatest swelling and wavelength change of the sensor. Similarly,
ribose gave a large response, suggesting that five-membered
furanose rings offer the most suitable positioning of diols for
binding to boronic acid. It has been suggested by van den Berg
and van Bekkum42 that the cis-1,2 diol configuration on the
furanose ring produces a very stable borate ester with a low ring
strain and a decrease in entropy of the ligand. Deoxyribose
displayed little diffraction wavelength shift compared to ribose,
indicating the importance of the hydroxyl group on carbon 2,
which is present in ribose but absent in deoxyribose (Figure 4).
The disaccharides, lactose, maltose, and sucrose, showed poor
response, probably due to steric hindrance caused by the glyco-
(
41) Soh, N.; Kitano, K.; Imato, T. Anal. Sci. 2 0 0 2 , 18, 1159-1161.
(
40) Lavigne, J. J.; Anslyn, E. V. Angew. Chem., Int. Ed. 1 9 9 9 , 38, 3666-3669.
752 Analytical Chemistry, Vol. 76, No. 19, October 1, 2004
(42) van den Berg, R, P. J.; van Bekkum H. Carbohydr. Res. 1 9 9 4 , 253, 1-12.
5

Fig u re 3 . (a) Response of holograms containing 5, 8, 10, 12, 15, and 20 mol % 3-APB to glucose in PBS pH 7.4 at 30 °C. Error bars represent
1 standard deviation (n ) 3). (b) Change in diffraction wavelength shift as a function of mol % 3-APB with 10.7 mM glucose in PBS pH 7.4
(
at 30 °C.
many microbial cell cultures, this would not be a significant issue
as the main carbon source in growth media is glucose.
The sensor was found to respond to the R-hydroxy acid, lactate,
as it also possesses a suitable cis-diol structure.43 The system
responded faster to lactate (t1/ 2 0.7 min; 2 mM lactate) than
glucose (t1/ 2 10.5 min; 2 mM glucose), possibly due to flexible
orientation of cis-diols in lactate and its small size, which would
allow it to diffuse more rapidly into the polymer than the larger
glucose molecule. To investigate whether lactate binds to 3-APB
using its carboxylic acid group only or using its two cis-diols, a
â-hydroxy acid, 3-hydroxybutyric acid, which possesses a car-
boxylic acid group and a â-hydroxyl group but no cis-diols, was
tested. The sensor showed no response to 3-hydroxybutyric acid,
suggesting that binding of lactate occurs through the cis-diols
present and not just the carboxylic acid group. A similar result
was observed with the monocarboxylic acid acetic acid.
Fig u re 4 . Response of a 12 mol % 3-APB hologram to various
saccharides and lactate in PBS pH 7.4 at 30 °C. Key: s, fructose;
9, ribose; +, galactose; ], lactate; 2, mannose; - - -, glucose; 0,
lactose; b, maltose; O, sucrose; ×, deoxyribose.
Although the holographic glucose sensor does respond to a
variety of monosaccharides, in most cases, these will not be found
in abundance in growth media unless they are used as the primary
carbon source. However, saccharides are present on the surface
of cells and on proteins, some of which may be produced and
sidic bond and the fact that certain diol groups were no longer
available for binding. The order of binding affinities deduced from
the magnitudes of the holographic response to various analytes
agrees with the data obtained by Springsteen and Wang,11 using
a competitive fluorescence assay. These results highlight the
relatively low selectivity of the sensor, but with mammalian and
(43) Gray, C. W., Jr.; Houston, T. A. J. Org. Chem. 2 0 0 2 , 67, 5426-5428.
Analytical Chemistry, Vol. 76, No. 19, October 1, 2004 5753

Fig u re 5 . (a) Change in diffraction wavelength of a 12 mol % 3-APB hologram in the presence of growing B. subtilis and decreasing glucose
levels determined by an enzymatic hexokinase method; 9, hexokinase; s, hologram. The inset shows the correlation between real glucose
concentrations and diffraction wavelength at specific time points during growth. (b) Optical density and pH of the growing spores under the
same conditions; 2, OD; s, pH.
secreted by the cells, as glycosylated products.44 These structures
could potentially interfere with the operation of the sensor as they
would be able to bind to 3-APB. Thus, the response of the sensor
this sensor to monitor fermentation processes. Previously, holo-
graphic sensors were successfully used to monitor the homolactic
fermentation of milk by Lactobacillus casei through the change in
pH generated during the fermentation process.26 In the present
report, the growth of B. subtilis was chosen as a model system
for cell growth, since this particular strain is simple to grow in a
reproducible fashion under well-defined conditions. B. subtilis is
an obligate aerobe; therefore, products of anaerobic fermentation
such as lactate, which may interfere with the sensor, are unlikely
to be produced in significant quantities.
1
to a model glycoprotein, R -acid glycoprotein, was assessed to see
if it could interfere with glucose detection.
AGP is an acute phase protein present at a concentration of
0
.2-1 mg/ mL in normal human blood, which increases by 2-3
4
5
times during inflammation. The sensor displayed no response
to AGP in PBS. This was probably due to the large size of the
protein, which was unable to penetrate significantly into the cross-
linked polymer matrix. Thus, glycosylated proteins, whether
present as part of a complex growth medium, or excreted during
cell growth, should not affect the response of the sensor. In
addition, glycosylated structures, which are normally present on
the surfaces of cells, would not be expected to interfere with
glucose detection.
Figure 5a shows a gradual decrease in the diffraction wave-
length of the hologram after addition of bacterial spores, suggest-
ing a depletion of glucose in the medium. The determination of
glucose with the independent hexokinase method suggested that
the decrease in diffraction wavelength observed was due to
decreasing glucose concentrations. Bacterial growth was con-
firmed by monitoring the increase in optical density at 600 nm of
a separate experiment performed under identical conditions, in
the presence of a hologram; the population was observed to enter
stationary phase after 10 h (Figure 5b). With an OD ≈ 1, the cell
The depletion of glucose during the growth of a model bacterial
system was monitored in order to prove the feasibility of using
(
(
44) Voet, D.; Voet, J. G. Biochemistry, 2nd ed.; John Wiley and Sons: New York,
995.
45) Shiyan, S. D.; Bovin, N. V. Glycoconjugate J. 1 9 9 7 , 14, 631-638.
1
8
density was found to be 1.8 × 10 cfu/ mL as determined by plate
5754 Analytical Chemistry, Vol. 76, No. 19, October 1, 2004

counts. Figure 5a also shows that the trend observed with the
diffraction wavelength shift mirrored that of the external glucose
determination method. The inset in Figure 5a suggested a linear
correlation between the measured glucose concentrations and
diffraction wavelength values during the growth period.
It should be possible to correlate the diffraction wavelength
to actual glucose concentrations by use of a calibration curve,
which would provide the diffraction wavelength for known
concentrations of glucose in the growth media. However, the pH
of the culture medium was found to decrease during spore
sensor does not display unique specificity for glucose, glucose is
normally the predominant free sugar used in the majority of cell
growth media as the primary carbon source. The sensor does not
respond to macromolecular glycosylated structures, such as
glycoproteins, which cannot diffuse into the hydrogel matrix.
Characterization of the sensor revealed a high affinity for furanose
ring structures, such as those of fructose and ribose, and an affinity
for the R-hydroxy acid, lactate, which maybe problematic as it is
a product of anaerobic cell metabolism. Preliminary studies using
a live aerobic culture of B. subtilis spores demonstrated that the
sensor could be used to monitor glucose depletion in the culture
media with a good correlation between the sensor response and
glucose concentration.
Unlike enzyme-based glucose sensors, the holographic sensor
is nonconsumptive, amenable to mass manufacture, and easily
miniaturizable, which makes it suitable for use in small-volume
microfermentor systems. The absence of a biological recognition
system also makes it more robust, stable, and less expensive to
manufacture. The sensor can be read using a simple reflection
spectrometer and does not require large, dedicated pieces of
equipment. Although the current sensor displays low selectivity
and is affected by pH, these variables could be measured through
the use of other holograms in an array and compensated for by
use of suitable algorithms.
Forthcoming studies will focus on improving both the selectiv-
ity and response kinetics of the sensor. Modification of the ligand,
the polymer hydrogel, or both could be used to reduce or
eliminate interference and improve the kinetics of the sensor. The
sensor will also be tested on a wider range of cell fermentations,
including other bacterial and mammalian cells, and in small-
volume systems. Furthermore, the sensor should be applicable
to blood glucose monitoring since it displays a progressive
response to glucose concentrations up to at least 25 mM at
physiological pH (7.4) and is only minimally affected by ionic
strength at this pH. It is anticipated that the sensor will function
in whole blood.
germination and growth (Figure 5b). The hydrogel contracts as
the pH decreases26 since 3-APB is an acid with a pK
of ∼8.6.
46
a
Furthermore, as seen in Figure 1a, the sensitivity of boronic acids
toward cis-diols decreases with pH. Both of these factors would
8
alter the overall response to glucose of the holographic glucose
sensor as a function of pH. These two factors may be the cause
of the change in the diffraction wavelength seen at ∼4 h in Figure
5a. Thus, it would be necessary to calibrate the sensor for changes
in pH over the glucose concentration range to convert the
diffraction wavelength observed directly into a true glucose
concentration. However, despite this response of the sensor to
changes in pH, there is a clear correlation between the holo-
graphic sensor response and glucose depletion (Figure 5a).
One of the key advantages of the holographic sensor system
is that it is possible to produce multiplexed sensor systems with
an array of different holograms for different analytes. Thus, a
suitable pH hologram, such as that described by Marshall et al.,26
could be used in tandem with the glucose sensor to record the
pH of the medium continuously. The pH data could then be used
to correct the response of the sensor for the changes in pH during
cell growth. A suitable lactate sensor could also be used in tandem
to quantitate lactate in anaerobic fermentations and correct for
responses to lactate. Ultimately, the system would operate through
an array of holographic sensors, to detect a variety of parameters
associated with cell metabolism, for use in monitoring cell growth
and fermentation processes. The array could be customized to
accommodate selected analytes associated with specific cell
growth processes.
ACKNOWLEDGMENT
We thank Alex J. Marshall, Graham Christie, Felicity Sartain,
and Colin A.B Davidson for useful discussions and input. This
work was supported by the BBSRC, Smart Holograms Ltd., and
GlaxoSmithKline plc.
CONCLUSIONS
A reversible, real-time holographic sensor for glucose was
fabricated, using 3-APB as a glucose binding ligand. Although the
Received for review May 6, 2004. Accepted July 29, 2004.
AC049334N
(
46) Otsuka, H.; Uchimura, E.; Koshino, H.; Okano, T.; Kataoka, K. J. Am. Chem.
Soc. 2 0 0 3 , 125, 3493-3502.
Analytical Chemistry, Vol. 76, No. 19, October 1, 2004 5755