Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
www.elsevier.com/locate/jphotobiol
Component analysis of the fluorescence spectra of a lignin
model compound
Ksenija Radotić
a,*
, Aleksandar Kalauzi a, Daniela Djikanović a, Milorad Jeremić b,
Roger M. Leblanc c, Zoran G. Cerović d
a
d
Centre for Multidisciplinary Studies, University of Belgrade, Despota Stefana 142, 11000 Belgrade, Serbia and Montenegro
b
Faculty of Physical Chemistry, University of Belgrade, Studentski Trg 12-16, Belgrade, Serbia and Montenegro
c
Department of Chemistry, University of Miami, Coral Gables, FL 33124, USA
Ecology, Systematics and Evolution Laboratory, CNRS UMR8079; Bât 362, Centre Universitaire Paris-Sud, 91405 Orsay Cedex, France
Received 26 July 2005; received in revised form 2 December 2005; accepted 4 December 2005
Available online 9 January 2006
Abstract
In order to test whether lignin fluorescence originates from discrete fluorophores, fluorescence emission spectra of the lignin model
dehydrogenative polymer (DHP) were analyzed by the band deconvolution method and time-resolved analysis of both the excitation
and emission spectra. Two series of 22 fluorescence emission spectra of DHP in chloroform/methanol (3:1, v/v) solution, and as a solid
suspension in water, were deconvoluted into three fluorescence and one Raman Gaussian components. Emission spectra were obtained
by stepwise variation of the excitation wavelength from 360 to 465 nm. Deconvolution was performed by nonlinear fitting of all three
Gaussian parameters: area, width and position. Position of all components in a series was treated as a random variable and its approximate probability distribution (APD) calculated from a series of histograms with increasing number of abscissa intervals. A five peak
multimodal APD profile was obtained for both series of DHP emission spectra. The mean fluorescence lifetime varied with wavelength
both in the emission and the excitation decay-associated spectra (DAS), where four kinetic components were resolved. The shapes of the
excitation spectra of the four components were quite different and gradually shifted bathochromically. The multicomponent nature of the
DHP emission spectra along with the changes in the mean fluorescence lifetime and the form of the excitation DAS of the four components give evidence of the heterogeneous origin of fluorescent species emitting in the visible.
2005 Elsevier B.V. All rights reserved.
Keywords: Blue-green fluorescence; Fluorophores; Gaussian components; Lignin model compound; Nonlinear fitting; Time-resolved fluorescence
1. Introduction
Lignin, as a major structural polymer in the plant cell
walls, is the second most abundant polymer on Earth. Lignin is highly branched and random polymer composed of
cross-linked phenyl-propanoid units derived from coniferyl, sinapyl and p-coumaryl alcohols as precursors. In
plants, it is intertwined and cross-linked with other macro*
Corresponding author. Tel.: +381 11 2078451; fax: +381 11 3055289.
E-mail addresses: [email protected] (K. Radotić), kalauzi@ibiss.
bg.ac.yu (A. Kalauzi), [email protected] (D. Djikanović), jeremic
@ffh.bg.ac.yu (M. Jeremić), [email protected] (R.M. Leblanc), zoran.
[email protected] (Z.G. Cerović).
1011-1344/$ - see front matter 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jphotobiol.2005.12.001
molecules in the cell walls [1]. Various types of inter-unit
bonds are possible in lignin, leading to different types of
substructures [2]. Fluorescence is an intrinsic property of
lignin [3]. The structural complexity of lignin makes its
fluorescence spectra difficult to interpret. In order to interpret the results, fluorescence spectra of a variety of lignin
model compounds were examined [3–6]. Fluorescence spectroscopy was used as a sensitive analytical tool in the studies of lignin constituents in waters and soils, as well as in
photochemistry of wood fibres and paper [7–9].
Blue-green fluorescence of leaves (BGF) has been proposed to originate from ferulic acid, often present in the
cell walls [10–12]. Nevertheless, the contribution of lignin
2
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
fluorescence to BGF should not be disregarded in view of
the recent results showing that the most fluorescing structures on the leaf surface are lignified [13]. Indeed, when
excited by UV-A radiation, leaves emit an important
BGF that depends on plant species and the stage of development. The new interest in the origin and possible heterogeneity of this fluorescence has risen with the analysis of
in vivo BGF of plants in the framework of active remote
sensing of vegetation [14,15]. The blue-green fluorophores
are distributed in different compartments of the leaf and
the plant cell [15], complicating further the quest for the
origins and causes of BGF variability. One way of
approaching the problem is to combine spectrofluorimetric
measurements with lifetime measurements in the sub-nanosecond time domain, to differentiate more precisely the
BGF components [12,16,17]. In the cited studies, synchrotron radiation and the time-correlated single-photoncounting technique were used to investigate the spectral
and time-resolved characteristics of the BGF emitted from
spinach and sugar beet leaves. Four kinetic components
were resolved. Lignin exhibits a strong UV-B-excited UVA fluorescence and a much weaker visible fluorescence
either UV-A-excited or as a tail of the UV-B-excited fluorescence. The UV-A-emitting lignin fluorophore has
received much attention and putative candidates have been
periodically proposed: anion of coniferyl alcohol [18], stilbene structures [3,19,20], phenylcoumarone structures [20].
The UV-A-excited visible emitting lignin fluorophores
remain more elusive. According to the published literature,
it is unclear whether lignin fluorescence is produced by a
charge-transfer mechanism [21] or distinct molecular species within lignin polymer [20]. Therefore, the real origin
of lignin fluorescence is still ambiguous.
In this work we analyzed two sets of fluorescence emission spectra of a lignin model polymer, known as dehydrogenative polymer (DHP). One set is obtained by
spectrophotometric measurements of continuously excited
emission, the other set is obtained by pulsed synchrotron
excitation of fluorescence in sub-nanosecond time domain.
Namely, we combined the spectrofluorimetric (e.g., Ref.
[6]) with lifetime (e.g., Ref. [22]) measurements, by adding
a time-resolved analysis of both the excitation and the
emission spectrum. DHP was obtained by peroxidase catalyzed polymerization of coniferyl alcohol. Since natural lignin cannot be isolated in unaltered form, DHP is widely
accepted as the best model compound for lignin studies
[23,24]. The other advantage of DHP is that it can be studied both in solution and in solid state. Once the conditions
for recording fluorescence of solid DHP have been
acquired, they can also be applied for isolated lignins,
which are only partly soluble. Our results presented here
show that even the simple DHP lignin model has several
visible emitting fluorophores, implying that this heterogeneity in emission will be present also in vivo. This may have
significant consequences, not only for the interpretation of
the leaf BGF heterogeneity, but also for the putative structure of natural lignin.
2. Materials and methods
2.1. Synthesis of the dehydrogenative polymer
Lignin model dehydrogenative polymer (DHP) was synthesized according to the procedures of Freudenberg [25]
and Wayman and Obiaga [26], as was reported in the publications of Radotić et al. [27,28]. DHP was synthesized
from coniferyl alcohol, using horseradish peroxidase as
an enzymatic catalyst. The reaction mixture contained
5 · 103 M coniferyl alcohol, 5 · 103 M H2O2 and
2.5 · 108 M horseradish peroxidase (all from Fluka
Chemical Corp., New York) in 50 mM phosphate buffer.
The reaction mixture was prepared by simultaneous addition of H2O2 and coniferyl alcohol solutions to peroxidase
solution. After mixing, the solution was shaken constantly
for 48 h. Polymerization occurs in solution phase, at a temperature of 25 C. Reactions took 48 h to complete. The
precipitate was washed twice in deionized water and evaporated in a vacuum at 5 C.
2.2. Steady-state fluorescence spectroscopy
Fluorescence spectra were collected using a Fluorolog-3
spectrofluorimeter (Jobin Yvon Horiba, Paris, France)
equipped with a 450 W xenon lamp and a photomultiplier
tube. The lignin model compound was either suspended in
deionized water or dissolved in chloroform/methanol (3:1,
v/v), in a 1-cm optical path length quartz cuvette. The suspension was stirred for 1 min before each measurement, in
order to avoid precipitation. Both suspension and solution
contained 0.5 mg mL1 DHP. The slits on the excitation
and emission beams were fixed at 4 and 2 nm, respectively.
The spectra were corrected for the dark counts. In each
measurement seven scans were averaged. The emission
spectrum of the solvent (chloroform/methanol) was subtracted. All measurements were performed at controlled
temperature of 25 C by means of a Peltier element.
A total of Nsol = Nsus = 22 emission spectra were collected for each DHP solution and DHP suspension by excitation at different wavelengths, starting from excitation
maximum at 360 up to 465 nm, with a 5 nm-step. Nonlinear fitting of each of the 44 spectra (Nsol + Nsus) was performed using the Nelder–Mead algorithm implemented in
Matlab, version 6. Spectra were deconvoluted into Gaussian components:
IðkÞ ¼
n
X
i¼1
2
1
ðk k0i Þ
Ai pffiffiffiffiffiffi exp
;
2r2i
2pri
where I(k) is the relative fluorescence intensity at wavelength k, Ai, ith component area, ri, ith component standard deviation (width), k0i, wavelength of ith component
maximum (position) and n, number of components. For
each component, all three parameters Ai, ri and k0i were fitted, so that in the case of DHP solution (nsol = 3), fitting
error was minimized in a nine-dimensional parameter
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
space, while for DHP suspension (nsus = 4, since one component was assigned to water Raman band), there were 12
dimensions. Raman band was not subtracted prior to
deconvolution, because it was used as an internal probe
to check the validity of mathematical procedure. It has to
be noted that its position can be theoretically precisely
calculated.
As a result of the nonlinear fitting, the total number of
components (ntot) for the two series of emission spectra
differed: (ntot)sol = Nsol * nsol = 66, (ntot)sus = Nsus * nsus =
88. For each of the two cases, positions of all components
were treated as a random variable, so that histograms of
component positions Nc(k0,k0 + Dk0) = f(k0) were calculated (Nc denoting number of components with positions
falling within the k0 and (k0 + Dk0) interval). For both,
solution and suspension, a series of histograms was constructed with an increasing number of abscissa intervals
(decreasing Dk0). An approximation of the probability density for a component to assume a position on the k0-axis
could then be reconstructed from the histogram data.
Time-correlated single-photon-counting measurements
were performed in darkness on the same apparatus, as
described previously [12,17] with some modifications.
Decay-histograms were acquired until 1,200,000 counts
were accumulated. Thirty-nine nanoseconds were covered
by 2048 channels (0.019 ns per channel). Iterative convolution of BGF decays was performed as previously described
[17] using proprietary program based on Marquardt search
algorithm for non-linear parameters [31]. The correctness
of the convolution was judged from the examination of
the v2 value, distribution of the weighted residuals between
the calculated and the experimental function, and the autocorrelation function of the residue [32]. The accumulation
of over 50,000 counts at the fluorescence timing histogram
maximum allowed the fitting of a four exponential decay
model at all wavelengths. The v2 value increased each time
an additional component was added, from two to three to
four decay components, justifying the four-component
model. Further addition of components neither ameliorated the statistics nor was always converging. We have
2.3. Excitation and emission decay-associated spectra
(DAS)
Fluorescence intensity (a.u.)
3.5
x 10
5
3
2.5
2
1.5
1
0.5
0
350
400
450
500
550
600
550
600
Wavelength (nm)
Fluorescence intensity (a.u.)
Spectra were recorded in darkness on the SA4 line of the
SUPERACO synchrotron in Orsay (France), as described
previously [12,17,29] with some modifications. For the
excitation part of the set-up, a double-grating monochromator (6 nm band pass, H10 D UV, Jobin-Yvon, Longjumeau, France) was used, instead of a single-grating
monochromator. For measurements of excitation spectra,
a KV418 (Schott) anti-UV filter was placed before the excitation monochromator. For measurements of emission
spectra, a KV389 (Schott) anti-UV filter was placed before
the emission monochromator. Excitation spectra from 300
to 400 nm were recorded by ‘scrolling’ the excitation monochromator, and integrating the number of counts during
1.1 s at each new wavelength spaced by 2 nm. Emission
spectra from 400 to 600 nm were recorded in the same
way by ‘scrolling’ the emission monochromator. For most
cases, two forward and two backward scans were averaged.
Excitation spectra were corrected on-line for instrumental
response using rhodamine B as a photon counter [30]; a
part of the excitation beam was deviated toward a cell filled
with a solution of rhodamine B (3 g L1), whose fluorescence was continuously recorded. Emission spectra were
corrected for instrumental response using quinine sulphate
as a standard [30]. All measurements were made at controlled room temperature (20 C).
Fluorescence was expressed in quinine sulphate equivalent units (QSEU); 1000 QSEU correspond to the fluorescence of 1 lM quinine sulphate dehydrate in 1 cm layer
0.105 M perchloric acid, excited at 347.5 nm and emitted
at 450 nm, under identical measuring conditions. Quinine
sulphate is a fluorescence standard of known emission
spectrum that is used and distributed by the United States
National Institute of Standard and Technology (NIST).
3
18
x 10
4
16
14
12
10
8
6
4
2
0
350
400
450
500
Wavelength (nm)
Fig. 1. Two series of emission spectra: DHP dissolved in chloroform/
methanol (3:1, v/v) (upper panel) and suspended in deionized water (lower
panel), for excitation wavelengths in the range 360–465 nm with 5 nm step.
In the solution, solvent emission spectra had been subtracted; in the
suspension, traveling water Raman peak is visible.
4
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
shown previously that both the maximization of entropy
[33] and the least square minimization approach yield the
same results, in the present case four components [17].
Decays were analyzed individually and then a global fit
was performed to all decays in a spectrum simultaneously
[17,29]. The absolute contribution to the emission of the
individual component was calculated by multiplying the
fractional intensity by the emission spectra at the corresponding wavelength recorded separately.
Pn
Fractional intensities were defined P
as: fi ¼ ai siP
= i¼1 ai si
n
n
2
and
Pn the mean lifetime as: sm ¼ i¼1 fi si ¼ i¼1 ai si =
i¼1 ai si where ai represent the
Pnrelative pre-exponential factors of the decay function ( i¼1 ai ¼ 1).
A major difference between the two spectrofluorimeters
was the power of the light source. The 450-W xenon lamp
of the Fluorolog-3 allows the use of a double monochromator both at the excitation and the emission side. This
yields a signal-to-noise ratio (SNR) based on water Raman
scatter of 5000, compared to the SNR of the FLU3 set-up
of 500. The output of the synchrotron Super-ACO SA4 line
λ
5
x 10
λ = 428.0
3.5
°
3. Results and discussion
A set of fluorescence spectra, recorded by excitation of
DHP at different wavelengths (kex), starting from excitation
maximum at 360 nm, with 5 nm-step, is shown in Fig. 1.
The starting excitation wavelength 360 nm was chosen on
the basis of our previous measurements of excitation spectra. The excitation at shorter wavelengths revealed no addi-
= 365 (nm)
λ
ex
460.6
12000
507.1 (nm)
Fluorescence intensity (a. u.)
Fluorescence intensity (a. u.)
ex
filtered through a H10 D UV Jobin-Yvon double monochromator yielded only 5 lW of luminous power at the
level of the sample in the FLU3 set-up, which precluded
the use of a double monochromator at the emission side.
The FLU3 set-up was designed for the recording of timeresolved decay-associated excitation and emission spectra
(DAS) of photo-labile samples by time-correlated singlephoton-counting. The irradiation is so low that it can
hardly lead to photochemically-induced isomerization of
chromophores even during long exposures needed to
acquire the fluorescence spectra or decays.
3
2.5
2
1.5
1
0.5
0
400
450
500
550
λ = 490.6
513.0
°
533.5 (nm)
10000
8000
6000
4000
2000
0
600
= 455 (nm)
480
500
Wavelength (nm)
520
540
560
580
600
580
600
Wavelength (nm)
5
x 10
1.8
λ° = 414.6
15000
423.8
462.6
511.7 (nm)
λ = 486.0
°
Fluorescence intensity (a. u.)
Fluorescence intensity (a. u.)
2
1.6
1.4
1.2
1
0.8
0.6
0.4
528.9
536.9
563.5 (nm)
10000
5000
0.2
0
400
450
500
Wavelength (nm)
550
600
0
480
500
520
540
560
Wavelength (nm)
Fig. 2. Examples of DHP emission spectra deconvoluted into three (solution, upper panels) and four (suspension, lower panels) Gaussian components.
Excitation wavelength was 365 nm for the left and 455 nm for the right panel. Circles, measured spectra; solid lines, separate components and their sum
fitted to the experimental data. Although all three Gaussian parameters were fitted per component (area, width and position), only the position results (k0)
are given.
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
580
Component position (nm)
560
540
520
500
480
460
440
420
400
360
380
400
420
440
460
480
460
480
Excitation wavelength (nm)
580
560
Component position (nm)
tional emission bands above 400 nm. In solution spectra,
the Raman band originating from the C–H vibration of
the solvent is not visible, since solvent spectra had been
subtracted (Fig. 1, upper panel). In all recorded emission
spectra of DHP in suspension, a Raman scattering signal
was observed originating from the O–H vibration of water
(Fig. 1, lower panel). A Raman band of the solvent has
been previously noticed in fluorescence of lignin and
related compounds [9]. We have not subtracted the water
Raman signal from the spectra, in order to use it as a control in the proposed model. The Raman band was mathematically treated equally to all other components of the
emission spectra. The shape of emission spectra changed
from one-peak asymmetric to more or less multi-peak profile, moving from low to high excitation wavelengths, suggesting a complex structure of lignin energy levels that are
involved in fluorescence emission. This is in accordance
with the fact that lignin is built from about 20 multifunctional phenyl-propanoid units, which can interact in a variety of ways, and can be attached to different side groups. It
is reasonable to assume that they absorb at different wavelengths producing a complex lignin spectrum. Absorbed
light at a given wavelength is proportional to the excitation
coefficients of all phenyl-propanoid groups that absorb at
that wavelength, as well as to their concentration, while
the fluorescence intensity is proportional to the absorbed
light. Different chromophores are excited to a different
extent when excitation wavelength is varied. Consequently,
a series of emission spectra, obtained by varying kex, should
consist of components with relatively stable maxima,
implying unequal probability distribution of their positions. A typical result of component Gaussian deconvolution of DHP emission spectra obtained by excitation at
365 and 455 nm, of both suspension and solution, is shown
in Fig. 2. Deconvolution of DHP solution was performed
with three components (Fig. 2, upper panel), because sufficient accuracy could not be achieved with one- and twocomponent analysis. This fact is also in accordance with
the results of Machado et al. [22]. In the case of DHP suspension, the spectra were deconvoluted into four components (Fig. 2, lower panel), because one of the
components has been assigned to the Raman band.
Fig. 3 shows the positions of components of DHP emission
spectra, for all excitation wavelengths. Component positions exhibited relative stability, although statistical variations of their positions were present. In solution spectra
(Fig. 3, upper panel), positions of the components seem
to be more stable while varying kex than in suspension
(Fig. 3, lower panel). There are several reasons to explain
this observation: better quality of solution spectra; more
intermolecular contacts in solid DHP that broaden and
shift bands; presence of very weak and very broad band
in DHP that escapes Gaussian analysis. Excitation wavelength range is clearly divided in two domains, one below
400 nm, and the other one above 420 nm, with the 400–
420 nm transition interval. The two domains are not well
pronounced in suspension spectra due to reasons men-
5
540
520
500
480
460
440
420
400
360
380
400
420
440
Excitation wavelength (nm)
Fig. 3. Fitted Gaussian component positions (k0) as a function of the
excitation wavelength (kex), after decomposing two series of DHP emission
spectra (solution, upper panel; suspension, lower panel) from Fig. 1. For
each particular spectrum, serial positions of the Gaussian components,
with increasing k0, are marked with *, +, s and · (only in suspension).
Components exhibited higher (solution) or lower (suspension) stability of
positions when kex was varied. However, position of the Raman water
band, present only in suspension, was markedly different from all the
others, increasing linearly with kex. The Raman band is marked with
different symbols, depending on its position relative to the other three
components, as its relative position within a spectrum may change.
tioned above. The presence of traveling Raman component
is clearly visible in the suspension, as a set of points linearly
dependent on kex. We tested the origin of the Raman band
by averaging differences between reciprocal positions of
these 22 Raman components and the corresponding 1/kex
and found agreement with water O–H stretch.
In order to acquire a distribution profile for all the positions of the components from Fig. 3, upper and lower panel
separately, we constructed corresponding histograms, presented in Fig. 4. Histogram profiles indicate that there
are intervals of grouping for the position of the components, both solution and suspension. However, since positions and relative amplitudes of histogram maxima
6
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
0.016
0.014
Probability distribution
Number of components
15
10
5
0
400
450
500
550
0.012
0.01
0.008
0.006
0.004
0.002
Component position (nm)
0
Number of components
15
450
500
550
Component position (nm)
Fig. 5. Approximate distribution of the probability (APD) that a fitted
Gaussian component of all DHP emission spectra occupies a position on
the k0-axis (reconstructed from the histograms presented in Fig. 4). Solid
and dashed lines refer to DHP dissolved in chloroform/methanol (3:1, v/v)
and suspended in deionized water, respectively. In both cases, APD is
multimodal, indicating that component positions are not equally probable. Separation between APD peaks is better in the case of the solution,
probably because position of the traveling water Raman component has a
uniform probability in the suspension.
10
5
0
400
400
450
500
550
Component position (nm)
Fig. 4. Histograms of component positions presented in Fig. 3, for all
excitation wavelengths. Solution, upper panel; suspension, lower panel.
depended on the number of histogram abscissa intervals,
we calculated approximate probability distribution (APD)
for position components by weighed averaging histogram
values for a set of histograms, where interval number varied from 2 to 25. Results of this procedure are presented in
Fig. 5. Both profiles showed a multimodal APD, the solution profile being characterized by better separation of
peaks. This phenomenon is probably due to the traveling
water Raman component in suspension, which has a uniform probability distribution of positions. One APD maximum in solution was positioned very close to the
corresponding maximum in suspension at 460 nm. However, a shift of 5–8 nm could be observed between four
other APD maxima in solution and suspension (around
425–430, 483–491, 501–507 and 535–540 nm). This could
be attributed to the difference in intermolecular forces
between molecules in solution and in solid state.
Parallel time-resolved analysis showed that the mean
lifetime varied with wavelength both in the emission
(Fig. 6) and the excitation (Fig. 7) DAS, indicating that
DHP fluorescence is heterogeneous. This is an important
observation because the lifetime is an inherent characteristic of the decays that does not depend on the convolution
model and the number of fitted components. In addition
to an unresolved short-lived component in unreduced
Abies wood, and two medium-lived components (1.03
and 3.30 ns) resolved by Castellan and Davidson [34] in
reduced Abies wood, Machado et al. [22] added a longlived component (8.48 ns), found in Eucalyptus grandis lignin, to the list of lignin fluorescence lifetime components.
We confirm here the presence of these components even
in the model polymer DHP, with the addition of a very
short lived one. In DHP we could resolve four kinetic components in both the emission and the excitation fluorescence. Global fluorescence lifetimes were 0.07, 0.50, 1.9
and 7.7 ns (Fig. 6) and 0.07, 0.57, 2.5 and 12.7 ns (Fig. 7)
in the emission and excitation spectra, respectively. The
major contribution to the fluorescence spectrum (Fig. 6)
comes from the very short-lived component, which has a
lifetime (70 ps) close to the limit of the resolving power
of the FLU3 set-up (40 ps). This component has a blue
maximum (450 nm) as opposed to the short-lived (0.5 ns)
and the medium component (1.9 ns) that have green maxima (530 nm). The difference in both lifetime and emission
maxima of fluorescence components can be explained only
by the presence of two types of fluorophores in DHP, one
blue and the other green emitting. The similarity of the
spectra of the short and medium component can not
exclude the possibility that they originate from the same
fluorophore present in two different physical environments,
e.g., less and more rigid environment, respectively [35]. The
contribution of the fourth, long-lived component, although
real, was very small, so it is difficult to interpret the
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
7
Fig. 6. Decay-associated DHP emission spectra from 400 to 600 nm, excited at 355 nm.
spectrum that seems rather broad. The excitation DAS
(Fig. 7) corroborate the finding of emission DAS. The
shapes of the excitation spectra of the four components,
which in theory should correspond to their absorption
spectra, were quite different. In the measured 300–400-nm
window, the very short-lived component had a single maximum at 385 nm. In accordance with the emission DAS,
this component had the largest contribution to total fluorescence. The main maxima of the three longer-lived components were gradually shifted bathochromically. For the
medium and long component the maximum is probably
at a longer wavelength than 400 nm. The medium component had a pronounced secondary maximum at around
330 nm. The presence of decays recorded at longer wavelengths in the time-resolved excitation spectrum increased
the global lifetime of the medium and long components
to 2.5 and 12.7 ns, respectively. Machado et al. [22] found
little influence of the excitation wavelength on emission
decays. That was hardly compatible with their identification of a triple origin of lignin fluorescence deduced from
synchronous spectra and the quenching of long-lived fluo-
rescence probes, unless the three components had almost
identical absorption spectra.
The data about the origin of lignin fluorescence are
rather scarce. Some of the authors state that lignin fluorescence is produced by charge-transfer mechanism, without
any defined fluorophores present in the polymer [21]. Some
others propose that, in the case of UV-B-excited lignin UVA fluorescence, distinct molecular species within the lignin
polymer, such as phenylcoumarone and stilbene structures,
may be the source of lignin fluorescence [20]. On the other
hand, it was found that the lignin spectra and the spectra of
the phenylcoumarone models do not coincide completely.
This was proposed to be the result of the contribution of
lignin chromophores other than phenylcoumarones to the
lignin fluorescence or the fact that the phenylcoumarone
models differ structurally from the phenylcoumarone structures in the lignin [20]. It was also shown that many simple
phenol compounds involved in the reactions and structure
of the cell walls emit fluorescence in the blue part of the
spectrum [3,5], while in the green part only the compounds
having merged rings (such as flavonoid quercetin and
8
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
Fig. 7. Decay-associated DHP excitation spectra from 300 to 400 nm, emitted at 445 nm.
alkaloid berberin) have found to be emitters [5]. So, the
nature of the lignin fluorophores is presently still unrevealed. Machado et al. [22] used lifetime quenching of
long-lived fluorescent probes to map the energy distribution of the structural units present in lignin fragments from
E. grandis wood. Their results indicated that the emission
spectrum of this lignin might be regarded as a superposition of at least three broad spectral envelopes with slightly
different emission maxima and widths. They concluded that
the majority of the fluorescence complexity of this lignin
seems to be associated with ground state heterogeneity,
due to the complex mixture of the different fluorophores,
on which different fluorophore environments are superimposed [20].
The spectral decomposition performed in the present
study indicates that fluorescence of lignin model compound
originates from several fluorophores that can be divided
into distinct groups emitting at different wavelengths, and
having different absorption spectra. Difference between
fluorophores may be in molecular structure, or it can be
similar structures in different molecular environment, or
both. In different DHP states, such as chloroform/methanol solution and solid suspension in water, APD showed
the same number of peaks (five) with close positions. Small
shifts of peak positions can be attributed to the influence of
different environments. Stepwise increase of wavelength
excitation brings one by one fluorophores in excited state.
It was previously noticed by some authors that the red shift
of the emission band of lignin or related complex compounds, when the intensity of the excitation band was
increased [6,9], was caused by the excitation of different
fluorophores in lignin [9]. The changes in the mean lifetime
and the form of the individual time-resolved excitation
spectra of the four components reported here for DHP,
confirm the heterogeneous origin of fluorescent species
emitting in the visible, both in DHP and lignin. The excitation spectra of different DHP preparations, although similar, were not identical, and in addition changed with time
after initiation of synthesis (data not shown).
DHP seems as a good working model that could help
identify the different fluorophores by changing the conditions and time of its synthesis, coupled to other analytical
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
techniques, in addition to fluorescence. However, the putative lignin or DHP fluorophore might never have the same
spectral and lifetime characteristics once isolated or synthesized as a monomer. Influences of the neighbours in the
matrix can be multiple, from copigmentation, hydrogen
bonding, dipole induction to steric hindrance [35]. It can
also be a very small quantity but with a high quantum yield.
Using two experimental approaches, deconvolution of a
series of emission spectra obtained by excitation at different
wavelengths, and time-resolved decay-associated excitation
and emission fluorescence spectra, here we show the existence of distinct fluorophores in the lignin model polymer,
emitting in both blue and green part of the spectrum (Figs.
5–7). Similarity of results obtained with solution and solid
samples indicate that the latter procedure for recording
fluorescence spectra can be used when natural lignins are
studied, since they are only partly soluble in the organic
solvents.
We have shown that even a ‘‘simple’’ DHP lignin model
polymer has a complex fluorescence in the visible that can
contribute to the detected heterogeneity BGF of the leaf.
Still, a quantitative analysis of both leaf BGF and isolated
lignin is needed.
Acknowledgements
Grant 1911 (Cellular response to pollution stress in
trees. Possibility of application in biomonitoring of the
environment) from the Ministry of Science and Technology
of the Republic of Serbia supported this study. We gratefully acknowledge partial funding to Z.G.C. through the
CNRS GDR 1536 FLUOVEG. This work was supported
by the Access to Research Infrastructure action of the
Improving Human Potential Program of the European
Community to Ksenija Radotić (LURE project BF 02699 & 2365). Authors wish to thank Dr. Aurélie Cartelat
for her help during DAS acquisition and Dr. Ismaël Moya
for TCSPC expertise and for the use of his proprietary convolution program ‘‘fluomarqdt’’.
References
[1] N. Lewis, E. Yamamoto, Lignin: occurrence, biogenesis and biodegradation, Annu. Rev. Plant Physiol. Plant Mol. Biol. 41 (1990) 455–
496.
[2] W. Boerjan, J. Ralph, M. Baucher, Lignin biosynthesis, Annu. Rev.
Plant Biol. 54 (2003) 519–546.
[3] K. Lundquist, B. Josefsson, G. Nyquist, Analysis of lignin products
by fluorescence spectroscopy, Holzforschung 32 (1978) 27–32.
[4] H. Konschin, F. Sundholm, G. Sundholm, Fluorescence characteristics of lignin model compounds, Acta Chem. Scand. B 30 (1976)
262–266.
[5] M. Lang, F. Stober, H.K. Lichtenthaler, Fluorescence emission
spectra of plant leaves and plant constituents, Radiat. Environ.
Biophys. 30 (1991) 333–347.
[6] H. Tylli, I. Forsskåhl, C. Olkkonen, The effect of photoirradiation on
high-yield pulps: spectroscopy and kinetics, J. Photochem. Photobiol.
A: Chem. 87 (1995) 181–191.
[7] S.I. Falkehag, Lignin in materials, Appl. Polym. Symp. 28 (1975)
247–257.
9
[8] R.S. Davidson, The photodegradation of some naturally occurring
polymers, J. Photochem. Photobiol. B: Biol. 33 (1996) 3–25.
[9] S. Patsayeva, V. Yuzhakov, M. Lamotte, R. Fantoni, A. Lai, A. Palucci,
Variation of the UV-to-blue fluorescence ratio for organic matter in
water under conditions of fluorescence saturation, in: Proceedings
of EARSeL-SIG-Workshop LIDAR, Dresden/FRG, 2000.
[10] P.J. Harris, R.D. Hartley, Detection of bound ferulic acid in the cell
walls of the Gramineae by ultraviolet fluorescence microscopy,
Nature 259 (1976) 508–510.
[11] H.K. Lichtentaler, J. Schweiger, Cell wall bound ferulic acid, the
major substance of the blue-green fluorescence emission of plants, J.
Plant Physiol. 152 (1998) 272–282.
[12] F. Morales, Z.G. Cerović, I. Moya, Time-resolved blue-green
fluorescence of sugar beet (Beta vulgaris L.) leaves: spectroscopic
evidence for the presence of ferulic acid as the main fluorophore in the
epidermis, Biochim. Biophys. Acta 1273 (1996) 251–262.
[13] S. Meyer, A. Cartelat, I. Moya, Z.G. Cerović, UV-induced blue-green
and far-red fluorescence along wheat leaves: a potential signature for
leaf ageing, J. Exp. Bot. 54 (2003) 757–769.
[14] C. Buschmann, H.K. Lichtenthaler, Principles and characteristics of
multi-color fluorescence imaging of plants, J. Plant Physiol. 152
(1998) 297–314.
[15] Z.G. Cerović, G. Samson, F. Morales, N. Tremblay, I. Moya,
Ultraviolet-induced fluorescence for plant monitoring: present state
and prospects, Agron. Agr. Environ. 19 (1999) 543–578.
[16] Y. Goulas, I. Moya, G. Schmuck, Time-resolved spectroscopy of the
blue fluorescence of spinach leaves, Photosynth. Res. 25 (1990) 299–
307.
[17] Z.G. Cerović, F. Morales, I. Moya, Time-resolved spectral studies of
blue-green fluorescence of leaves, mesophyll and chloroplasts of sugar
beet (Beta vulgaris L.), Biochim. Biophys. Acta 1188 (1994) 58–68.
[18] H. Konschin, F. Sundholm, Protolytic dissociation of electronically
excited phenols related to lignin. II. Isoeugenol and coniferyl alcohol,
Finn. Chem. Lett. 1 (1974) 46–49.
[19] S.I. Falkehag, J. Marton, E. Adler, Chromophores in Kraft lignin.
Lignin structure and reactions, Adv. Chem. Ser. 59 (1966) 75–89.
[20] B. Albinsson, S. Li, K. Lundquist, R. Stomberg, The origin of lignin
fluorescence, J. Mol. Struct. 508 (1999) 19–27.
[21] S. Barsberg, T. Elder, C. Felby, Lignin–quinone interactions:
implications for optical properties of lignin, Chem. Mater. 15
(2003) 649–655.
[22] A.E. Machado, D.E. Nicodem, R. Ruggiero, D.S. Perez, A. Castellan, The use of fluorescent probes in the characterization of lignin: the
distribution, by energy, of fluorophores in Eucaliptus grandis lignin,
Photochem. Photobiol. A: Chem. 138 (2001) 253–259.
[23] R.H. Hwang, An approach to lignification in plants, Biochem.
Biophys. Res. Commun. 105 (1982) 509–514.
[24] N.G. Lewis, J. Newman, G. Just, J. Ripmeister, Determination of
bonding patterns of carbon-13 specifically enriched dehydrogenatively polymerized lignin in solution and solid state, Macromolecules
20 (1988) 1752.
[25] K. Freudenberg, Lignin im rahmen der polymeren neturstoffe,
Angew. Chem. 68 (1956) 84–92.
[26] M. Wayman, T.I. Obiaga, The modular structure of lignin, Can. J.
Chem. (1974) 2102–2110.
[27] K. Radotić, J. Simić-Krstić, M. Jeremić, M. Trifunović, A study of
lignin formation at the molecular level by scanning tunneling
microscopy, Biophys. J. 66 (1994) 1763–1767.
[28] K. Radotić, S. Todorović, J. Zakrzewska, M. Jeremić, Study of
photochemical reactions of coniferyl alcohol. II. Comparative structural study of a photochemical and enzymatic polymer of coniferyl
alcohol, Photochem. Photobiol. 68 (1998) 703–709.
[29] G. Latouche, Z.G. Cerović, F. Montagnini, I. Moya, Light-induced
changes of NADPH fluorescence in isolated chloroplasts: a spectral
and fluorescence lifetime study, Biochim. Biophys. Acta 1460 (2000)
311–329.
[30] G.C.K. Roberts, Correction of excitation and emission spectra, in:
J.N. Miller (Ed.), Standards in fluorescence spectrometry. Techniques
10
K. Radotić et al. / Journal of Photochemistry and Photobiology B: Biology 83 (2006) 1–10
in visible and ultraviolet spectrometry, Chapman & Hall, London,
1981, pp. 49–67.
[31] I. Moya, P. Sebban, W. Haehnel, Lifetime of excited states and
quantum yield of chlorophyll a fluorescence in vivo, in: A.J.
Govindjee, J. Amesz, D.C. Fork (Eds.), Light Emission by Plants
and Bacteria, Academic Press, Orlando, 1986, pp. 161–190.
[32] A. Grinwald, I. Steinberg, On the analysis of the fluorescence decay
kinetics by the method of least squares, Anal. Biochem. 59 (1974)
583–598.
[33] J.C. Brochon, Maximum entropy method of data analysis in timeresolved spectroscopy, Meth. Enzymol. 240 (1994) 262–311.
[34] A. Castellan, R.S. Davidson, Steady-state and dynamic fluorescence
emission from Abies wood, J. Photochem. Photobiol. A. Chem. 78
(1994) 275–279.
[35] B. Valeur, Fluorescent probes for evaluation of local physical and
structural parameters, in: S.G. Schulman (Ed.), Molecular Luminescence Spectroscopy Methods and Applications: Part 3, Wiley, New
York, 1993, pp. 25–84.