Content uploaded by Antoine van Oijen
Author content
All content in this area was uploaded by Antoine van Oijen
Content may be subject to copyright.
Newcomb, and H. E. Dregne [Int. J. Remote Sens.15,
3547 (1994)]. SOI and SST data are produced and
archived by the National Oceanographic and Atmo-
spheric Administration/Climate Prediction Center
(http://www.cpc.ncep.noaa.gov/). A previous paper
(4) proposed the use of a satellite-derived potential
virus activity factor. We now use monthly NDVI data,
normalized to represent departures from the 1982–
95 mean, to better characterize rainfall anomalies
associated with RVF activity.
14. T. M. Logan, K. J. Linthicum, F. G. Davies, Y. S. Binepal,
C. R. Roberts, J. Med. Entomol.28, 293 (1991); T. M.
Logan, F. G. Davies, K. J. Linthicum, T. G. Ksiazek,
Trans. R. Soc. Trop. Med. Hyg.86, 202 (1992).
15. M. A. Cane, G. Eshel, R. W. Buckland, Nature 370, 204
(1994).
16. S. E. Nicholson, Int. J. Climatol. 17, 345 (1997);
iiii and J. Kim, ibid., p. 117.
17. There is a direct relation between rainfall and green
vegetation growth, between green vegetation growth
and the NDVI, and hence between rainfall and the
NDVI. This relation applies to areas receiving precip-
itation of ,800 mm/year. W. K. Lauenroth, in Per-
spectives in Grassland Ecology, N. French, Ed. (Springer-
Verlag, New York, 1979), pp. 3–24; H. N. Le Houerou
and C. H. Hoste, J. Range Manage.30, 181 (1977);
A. R. Malo and S. E. Nicholson, J. Arid Environ.19,1
(1990); S. E. Nicholson, M. L. Davenport, A. R. Malo,
Clim. Change 17, 209 (1990); C. J. Tucker and S. E.
Nicholson, Ambio, in press.
18. B. N. Holben, Y. J. Kaufman, J. D. Kendall, Int. J.
Remote Sens.11, 1511 (1990); E. F. Vermote and Y. J.
Kaufman, ibid. 16, 2317 (1995); S. O. Los, ibid. 14,
1907 (1994); N. Che and J. C. Price, Remote Sens.
Environ.41, 19 (1992).
19. Correlation coefficients were determined for a data
series calculated by the differences between adja-
cent values with SPSS Trends 6.1 software (SPSS,
Chicago, 1994). Nairobi NDVI anomalies were de-
rived from average monthly composite data within
8 by 8 grid cells, each with a spatial resolution of
8 km centered close to Nairobi, Kenya. Monthly
AVHRR data were derived from global area cover-
age data that are produced by the on-board pro-
cessing of large area coverage data (1.1 km by 1.1
km) and subsequently transmitted to receiving sta-
tions in Virginia or Alaska. Composite data were
formed by selecting the highest NDVI for each grid
cell location from daily data for that month to
minimize cloud and atmospheric contamination.
NDVI data were calculated and mapped to a Ham-
mer-Aitof projection. The highest value during a
monthly period was selected to represent the
monthly composite for each grid cell location.
20. AutoRegressive Integrated Moving Average (ARIMA)
analysis determined by SPSS, Trends 6.1 software.
21. T. M. Logan et al.,J. Am. Mosq. Control Assoc.6, 736
(1990).
5 February 1999; accepted 5 May 1999
Unraveling the Electronic Structure
of Individual Photosynthetic
Pigment-Protein Complexes
Antoine M. van Oijen,
1
* Martijn Ketelaars,
2
Ju¨rgen Ko¨hler,
1
†Thijs J. Aartsma,
2
Jan Schmidt
1
Low-temperature single-molecule spectroscopic techniques were applied to a
light-harvesting pigment-protein complex (LH2) from purple photosynthetic
bacteria. The properties of the electronically excited states of the two circular
assemblies (B800 and B850) of bacteriochlorophyll a (BChl a) pigment mole-
cules in the individual complexes were revealed, without ensemble averaging.
The results show that the excited states of the B800 ring of pigments are mainly
localized on individual BChl a molecules. In contrast, the absorption of a photon
by the B850 ring can be consistently described in terms of an excitation that
is completely delocalized over the ring. This property may contribute to the high
efficiency of energy transfer in these photosynthetic complexes.
The primary process in bacterial photosynthesis
is the absorption of a photon by the light-
harvesting antenna system, followed by the rap-
id and efficient transfer to the reaction center
where the charge separation takes place. Typi-
cally, photosynthetic purple bacteria contain
two types of antenna complexes, light-harvest-
ing complexes 1 and 2 (LH1 and LH2, respec-
tively), both of which are integral membrane
proteins. The reaction center is presumed to be
surrounded by the LH1 complex, whereas the
LH2 complexes are arranged around the perim-
eter of the LH1 ring in a two-dimensional struc-
ture (1). The structure of the LH2 complex of
the purple bacterium Rhodopseudomonas aci-
dophila is known in great detail from x-ray
crystallography (2), which has shown that the
LH2 complex comprises 27 BChl a and (pre-
sumably) 18 carotenoid molecules nonco-
valently bound to the protein matrix. The BChl
a molecules are organized in two concentric
rings (Fig. 1). One ring, referred to as B800,
features a group of nine well-separated BChl a
molecules with an absorption band at ;800
nm. The other ring, referred to as B850, consists
of 18 closely interacting BChl a molecules with
an absorption band at ;860 nm. The entire
LH2 complex is cylindrically symmetric with a
ninefold symmetry axis. Upon excitation, ener-
gy transfer occurs from B800 to B850 mole-
cules on a picosecond time scale (3–5), whereas
among the B850 molecules, it is an order of
magnitude faster (6–8). The transfer of energy
from LH2 to LH1 and subsequently to the
reaction center occurs in vivo on a time scale of
5to25ps(9), very fast in comparison to the
decay of B850 in isolated LH2, which corre-
sponds to a lifetime of 1.1 ns.
Despite the fact that the LH2 complex has
been intensively investigated in recent years
with a wide variety of spectroscopic tools, in-
cluding the observation of the fluorescence dy-
namics of single LH2 complexes (10), no clear
picture of the electronic structure of its excited
states exists. Here, we present the results of a
study of isolated single LH2 complexes by
single-molecule fluorescence-excitation spec-
troscopy, a method successfully applied in re-
cent years to the detection of single guest mol-
ecules in crystalline and amorphous matrices
(11). This technique allows the observation of
optical spectra of individual complexes devoid
of the ensemble averaging over static intercom-
plex disorder, thus directly revealing the salient
properties of the electronic structure of the ex-
cited states.
The LH2 complexes of R. acidophila were
prepared as described elsewhere (3). Hydro-
lyzed poly(vinyl alcohol) (PVA) with a weight-
average molecular weight of 125,000 (obtained
from British Drug House) was purified over a
1
Centre for the Study of Excited States of Molecules,
2
Department of Biophysics, Huygens Laboratory, Lei-
den University, Post Office Box 9504, 2300 RA Leiden,
Netherlands.
*To whom correspondence should be addressed. E-
mail: antoine@molphys.leidenuniv.nl
†Present address: Ludwig-Maximilians-Universita¨t
Mu¨nchen, Sektion Physik und Center for NanoScience,
Lehrstuhl fu¨r Photonik und Optoelektronik, Amalien-
strasse 54, 80799 Mu¨nchen, Germany.
zy
xx
Fig. 1. Geometrical ar-
rangement of the 27
BChl a molecules of
the LH2 complex of R.
acidophila obtained
by x-ray crystallogra-
phy. The B800 BChl a
molecules are depict-
ed in blue, and the
B850 pigments are
red. The phytol chains
of the BChl a mole-
cules are omitted for
clarity. The data have
been taken from the Protein Data Bank (identification code: 1kzu).
REPORTS
16 JULY 1999 VOL 285 SCIENCE www.sciencemag.org400
mixed resin in order to remove ionic impurities.
Thin polymer films, with a thickness of ,1
mm, were prepared by adding 1% w/w purified
PVA to a solution of 5 310
211
M LH2 in
buffer (0.1% lauryldimethylamine N-oxide, 10
mM tris, and 1 mM EDTA, with pH 8.0), which
was then spin coated on a LiF substrate (12).
The samples were mounted in a cryostat, cooled
to 1.2 K, and illuminated with a tunable con-
tinuous wave Ti-sapphire laser (spectral band-
width of 1 cm
21
). Microscopic images could be
obtained by wide-field illumination of the sam-
ple and imaging of the fluorescence at 890 nm
on a charge-coupled device camera. Fluores-
cence-excitation spectra were then acquired by
confocally exciting a spatially well-isolated
complex and detecting its fluorescence (also at
890 nm) with an avalanche photodiode. In both
cases, the detection bandwidth was 20 nm.
More details can be found in (13).
In Fig. 2, the fluorescence-excitation spectra
of several single LH2 complexes and an ensem-
ble of LH2 complexes are compared. The en-
semble spectrum features two broad structure-
less bands at ;800 and 860 nm, corresponding
to the absorptions of the B800 and B850 pig-
ments, respectively. When observing the single
complexes, the ensemble averaging in these
bands is removed, and remarkable spectral fea-
tures become visible. The striking differences
between the two absorption bands can be ratio-
nalized by considering the intermolecular inter-
action strength Jbetween neighboring BChl a
molecules in a ring and the spread in transition
energies D.Jis mainly determined by the inter-
molecular distance and the relative orientation
of the molecular dipole moments. Variations in
site energies Dcan often be attributed to struc-
tural variations in the environment of the BChl a
molecules, resulting in changes in the electro-
static interaction with the surrounding protein. If
the ratio J/Dis small, it is expected that the
excitations are mainly localized on individual
BChl a molecules. If the coupling strength J
between the BChl a molecules is much larger
than D, the description should be in terms of
delocalized excited-state wave functions with
relatively short energy relaxation times.
As can be seen in Fig. 2, the B800 band of
an individual LH2 complex consists of several
relatively narrow spectral lines. From the width
of the B800 ensemble line, a value of ;125
cm
21
for the diagonal disorder Dcan be ex-
tracted, and from the x-ray structure, it can be
calculated, with a point-dipole approximation,
that the interaction energy Jbetween neighbor-
ing pigments amounts to 224 cm
21
(14). The
ratio ?J/D?'0.2 is characteristic for electroni-
cally excited states, which are largely localized
on individual pigments. Therefore, the narrow
lines around 800 nm can be attributed to the
absorptions of individual BChl a molecules in
the B800 ring. This interpretation is corroborat-
ed by the strong dependence of the relative
intensities of these lines on the polarization of
the incident radiation, consistent with the dif-
ferent directions of the dipole moments of lo-
calized transitions of BChl a molecules in the
ring (15).
In the B850 band, the interaction strength
between the BChl a molecules is determined to
be ;300 cm
21
(14), that is, considerably larger
than the disorder (estimated to be ;125 cm
21
).
Therefore, we have to consider excitonic inter-
actions in order to understand the optical spec-
tra. As a starting point, we calculated the excit-
ed-state manifold of a cylindrically symmetric
B850 assembly with zero disorder. Of the two
nondegenerate (denoted as k50 and k59)
and eight pairwise degenerate (k561, k5
62, ..., k568) exciton states, only the
low-energy degenerate pair k561 will carry
appreciable oscillator strength (Fig. 3A, left).
Upon introducing diagonal disorder in the ring,
the pairwise degeneracies will be lifted, and the
oscillator strength is redistributed over adjacent
exciton states (16) (Fig. 3A, right). The transi-
tion dipole moments associated with the k5
61 transitions will have orthogonal polariza-
tions. This orthogonality is maintained when
disorder is introduced, assuming that the diag-
onal disorder is dominated by variations in
electrostatic interactions and possibly intermo-
lecular distances, rather than by changes in the
orientations of the BChl a molecules.
Fig. 2. Comparison of fluorescence-excitation
spectra for an ensemble of LH2 complexes (top
trace) and several individual LH2 complexes at 1.2
K. The vertical scale applies to the bottom spec-
trum; all other spectra are offset for clarity. cps,
counts per second.
k
kk
kk
AC
B
Fig. 3. (A) Schematic representation of the energy-level scheme of the lowest states in the excited-state
manifold of the B850 ring in LH2 of R. acidophila. Compared are the relative positions of the lowest
levels in the presence (left) and in the absence (right) of ninefold rotational symmetry. The gray circles
indicate the initial population of a given excited state, and the arrows indicate the relative orientation
of the transition dipole moments in the plane of the ring. (B) Fluorescence-excitation spectrum of the
long-wavelength region of an individual LH2 complex for mutually orthogonal polarized excitation as
schematically indicated in (A) by the colored arrows. (C) Fluorescence-excitation spectrum of the red
wing of the long-wavelength absorption in the B850 band. In the bottom panel, a stack of 200
consecutively recorded spectra (3 s per scan) is shown where the fluorescence intensity is given by the
color code (yellow corresponds to high intensities). The spectrum in the top panel corresponds to an
average of only those scans that are covered by the box. For this particular complex, the whole set of
lines in the B850 band is shifted toward higher energies in comparison to the complex shown in (B).
REPORTS
www.sciencemag.org SCIENCE VOL 285 16 JULY 1999 401
In all spectra of the single LH2 complexes
we observed, the B850 band consisted of two
broad absorption lines at ;860 nm, some-
times accompanied by a weaker third transi-
tion at the higher energy side. These obser-
vations can be explained in terms of the
exciton model. The two absorptions corre-
spond to the k561 transitions, with their
degeneracy lifted. By performing polariza-
tion-dependent experiments on these two
bands (17), the orthogonality of the associat-
ed transition dipole moments, predicted by
the exciton model, could be ascertained (Fig.
3B). This orthogonality was observed in all
individual LH2 complexes that we studied
and is a strong indication for a high degree of
delocalization of the excitation. The observed
homogeneous linewidth of the k561 tran-
sitions of ;50 cm
21
is consistent with an-
isotropy decay times of ;100 fs found in
pump-probe experiments (8). The extent of
delocalization will decrease, and the dynam-
ical properties will change at a higher tem-
perature, where mixing of the exciton states
by vibronic coupling will occur (18).
Another observation supporting the ex-
citonic level scheme is the detection of the
lowest exciton state k50. By repeatedly
scanning the excitation wavelength quickly
through the low-energy side of the k561
pair and following the spectral features
through time, the presence of the spectrally
rapidly diffusing lowest exciton state could
be made visible in a fraction (;25%) of the
studied complexes (Fig. 3C). The low in-
tensity of the k50 transition, which in
principle is dipole forbidden, and spectral
diffusion on a time scale faster than that of
the experiment explain the absence of this
lowest exciton transition in most of the
complexes. The linewidth of the k50 state
should be ;0.005 cm
21
, as determined by
the 1.1-ns fluorescence lifetime of the sys-
tem, but the observed value of ;5cm
21
is
mainly determined by residual spectral dif-
fusion and the bandwidth of the excitation
source.
The energy splitting dE
61
between the
k511 and k521 states was measured for
all complexes investigated (Fig. 4). As men-
tioned previously, the presence of the disor-
der Dforms a plausible cause for the lifting of
the degeneracy of these exciton states. How-
ever, simulations show that the observed av-
erage dE
61
of ;110 cm
21
cannot be ex-
plained by taking into account only random
disorder (as depicted in Fig. 4) for a simulat-
ed Dwith a full width at half-maximum
(FWHM) of 125 cm
21
(19). Even an unrea-
sonably large value of D'500 cm
21
for the
width of the distribution of site energies did
not result in an energy separation between the
k561 exciton states, as observed experi-
mentally. To exclude the possibility that
these abnormally high splittings are caused
by an anisotropic environment in the poly-
mer matrix, we repeated the experiment on
single LH2 complexes in a glycerol matrix,
which resulted in similar values of dE
61
.A
Jahn-Teller–like deformation in the excited
state can probably be ruled out in view of
the unrealistically high values needed for
the electronic-nuclear coupling strength.
Although random disorder can give rise to
large variations in the absorption wave-
lengths of the B850 bands (Fig. 2), the ob-
served energy separation of the k561 states
can only be explained in terms of largely
correlated disorder, such as a static symmet-
ric distortion of the protein complex in the
ground state. In the case of an elliptical de-
formation of the ring, it can be shown, on the
basis of symmetry arguments, that only the
k561 exciton states will be split. This is
consistent with the absence in the spectra of a
splitting and a polarization effect of the k5
62 states. The eigenfunctions of the k561
states belong to the long and short axes of the
ellipse and hence exhibit orthogonal polariza-
tion of their transition moments, as observed
experimentally. Our simulations show that
the observed splittings can be explained by
assuming an eccentricity «of the ring of 0.52,
corresponding to a ratio of the long and short
radius of 0.85, and a random disorder of 125
cm
21
(Fig. 4); «5(1–a
2
/b
2
)
1/2
, where aand
bare the length of the short and long axes,
respectively. An explanation for the sym-
metry lowering in the LH2 from ninefold in
the crystals used for resolving the x-ray
structure to the twofold symmetry observed
in our experiments may be found in the
extremely dense packing of LH2 in the x-ray
crystals, causing a stabilization of the struc-
ture. In our case of completely isolated com-
plexes, these stabilizing forces are absent,
and the complex deforms. What the symme-
try properties of the LH2 are in a natural
environment, surrounded by a limited num-
ber of LH2 complexes in the photosynthetic
membrane, is therefore an intriguing question
and deserves further study.
This work demonstrates that single-mole-
cule spectroscopy is a powerful tool to reveal in
detail the factors determining the electronic
structure of pigment-protein complexes and,
more generally, of molecular aggregates. Vari-
ous manifestations of disorder can be probed
directly, providing valuable information for the
theoretical modeling of energy-transfer pro-
cesses in these systems, a better understanding
of the structure of these biologically important
systems, and an understanding of how these
systems function.
References and Notes
1. M. Z. Papiz et al.,Trends Plant Sci. 1, 198 (1996).
2. G. McDermott et al.,Nature 374, 517 (1995).
3. J. T. M. Kennis, A. M. Streltsov, H. Permentier, T. J.
Aartsma, J. Amesz, J. Phys. Chem. B 101, 8369 (1997).
4. R. Monshouwer, I. Ortiz de Zarate, F. van Mourik, R.
van Grondelle, Chem. Phys. Lett. 246, 341 (1995).
5. H.-M. Wu et al.,J. Phys. Chem. 100, 12022 (1996).
6. M. Chachisvilis, O. Ku¨hn, T. Pullerits, V. Sundstro¨m, J.
Phys. Chem. B 101, 7275 (1997).
7. R. Jimenez, S. R. Dikshit, S. E. Bradforth, G. R. Fleming,
J. Phys. Chem. 100, 6825 (1996).
8. S. I. E. Vulto, J. T. M. Kennis, A. M. Streltsov, J. Amesz,
T. J. Aartsma, J. Phys. Chem. B 103, 878 (1999).
9. V. Nagarajan and W. W. Parson, Biochemistry 36,
2300 (1997).
10. M. A. Bopp, Y. Jia, L. Li, R. J. Cogdell, R. M. Hochstrasser,
Proc. Natl. Acad. Sci. U.S.A. 94, 10630 (1997).
11. Th. Basche´, W. E. Moerner, M. Orrit, U. Wild, Single
Molecule Optical Detection, Imaging and Spectrosco-
py (Verlag-Chemie, Munich, 1997).
12. LiF is a favorable substrate because the inversion
symmetry of the alkali halide crystal prevents first-
order Raman scattering.
13. A. M. van Oijen, M. Ketelaars, J. Ko¨hler, T. J. Aartsma,
J. Schmidt, J. Phys. Chem. B 102, 9363 (1999).
14. K. Sauer et al.,Photochem. Photobiol. 64, 564 (1996).
15. A. M. van Oijen, M. Ketelaars, J. Ko¨hler, T. J. Aartsma,
J. Schmidt, Chem. Phys., in press.
16. R. G. Alden et al.,J. Phys. Chem. B 101, 4667 (1997).
17. The polarization-dependent experiments were per-
formed by rotating the polarization of the incident laser
light in the plane of the sample. As a consequence of
the high-velocity spin coating, the LH2 complexes are
oriented predominantly flat on the substrate surface,
perpendicular to the propagation vector of the excita-
tion light. This was confirmed by similar experiments on
LH2 complexes in non–spin-coated films.
18. J. A. Leegwater, J. Phys. Chem. 100, 14403 (1996).
19. Monte Carlo simulations were performed on the basis
of the crystal structure of LH2 of R. acidophila. Only
nearest neighbor interactions were taken into ac-
count with the dipole-dipole approximation. A ran-
dom diagonal disorder (FWHM 5125 cm
21
) was
introduced for both the undistorted and distorted
rings, assuming that it is centered at the same tran-
sition energy for every B850 pigment. In the case of
the distorted ring, the individual pigments were po-
sitioned on an ellipse, whereas the long-wavelength
(Q
y
) transition dipoles retained the alignment as seen
in the crystal structure.
20. The authors thank J. P. Abrahams (Leiden University,
Leiden, Netherlands), J. Knoester (Groningen Univer-
sity, Groningen, Netherlands), and J. H. van der Waals
(Leiden University, Leiden, Netherlands) for many
helpful discussions. We also thank D. de Wit for the
preparation of the LH2 complexes and M. Hessel-
berth for assistance with the spin coating. This work
is supported by the Stichting voor Fundamenteel
Onderzoek der Materie (FOM) with financial aid from
the Nederlandse Organisatie voor Wetenschappelijk
Onderzoek (NWO). J.K. is a Heisenberg fellow of the
Deutsche Forschungsgemeinschaft.
8 April 1999; accepted 2 June 1999
Fig. 4. The distribution of the energy separations
dE
61
of the k561 transitions. The histogram
represents the experimental data (exp) for all
complexes studied. The values obtained by nu-
merical calculations (19), assuming only a disor-
der of 125 cm
–1
, are depicted by solid black
circles. After an additional elliptic deformation,
with an eccentricity of «50.52, is introduced in
the simulations (sim), one obtains the data rep-
resented by solid black squares.
REPORTS
16 JULY 1999 VOL 285 SCIENCE www.sciencemag.org402
