structures of the Lhcas are highly conserved
but their spectroscopic properties differ subs-
tantially ( 52 ), especially because both Lhca3
and Lhca4 contain Chls that absorb >700 nm
(Fig. 2); these are called red Chls. Their excited-
state energies are lower than those of most
of the Chls of PSI (absorbing in the 670 to
680 nm range; these are called bulk Chls) and
thus strongly influence EET by introducing
energetically uphill transfer steps ( 10 ). The
intrasubunit equilibration time in all Lhcas is
very fast (<10 ps), whereas the intersubunit
time of EET within the Lhca dimers is some-
what slower ( 53 ). When they are part of the
PSI supercomplex, these Lhcs mainly transfer
directly to the core. The transfer is very fast for
Lhca1 and Lhca2, whereas from Lhca3 and
Lhca4, it appears to be slower, mainly because
of their red Chls. Despite several slow EET
transfer steps, the overall trapping time in
the PSI-LHCI supercomplex is only ~50 ps,
with quantum efficiency still close to 1 ( 54 – 56 ).
Considering the experimental results men-
tioned above, the organization of the Chls in
the x-ray structures of PSI not showing con-
nections between Lhca2 and the core was
puzzling ( 25 , 26 ). This apparent inconsistency
was solved by the most recent cryo-EM
structure, which revealed the position of one
additional subunit (PsaN, in pink in Fig. 4),
which is located between Lhca2 and the core
and binds two Chls ( 27 ). By applying the cal-
culation method provided in Box 2, we obtain
effective times of transfer within the dimers
and from individual complexes to the core
that are all close to what was determined ex-
perimentally ( 54 , 55 ), demonstrating that the
simplified way of estimating transfer times
can help us to better understand the origin of
the experimental findings. An overview of the
various transfer times and pathways in PSI is
given in Fig. 6.
For a long time, the structure described
above was considered to represent the defin-
itivePSIcomplexofplants.However,thepicture
that is now emerging is that in most condi-
tions, at least one LHCII trimer is also present,
forming PSI-LHCI-LHCII complexes ( 57 – 59 )
(Fig. 4). This addition results in an increase of
the absorption cross section by 25% without
substantially influencing the charge separa-
tion efficiency. It is important to note here
that LHCII does not contain any red Chls
( 45 ) (Fig. 3). LHCII is energetically very well
connected to PSI, transferring its energy
directlytothecorewhileincreasingtheoverall
trapping time of the complex by only ~10 to
20 ps ( 57 , 60 ). In a more recent study, the
transfer from LHCII to PSI appeared to be
heterogeneous, suggesting the presence of two
particles with slightly different LHCII-core con-
nections ( 61 ). The structure of the PSI-LHCI-
LHCII complex of maize ( 27 )showsthatLHCII
is directly connected tothecorethroughtwo
Chls on a newly identified subunit (PsaO, in
brown in Fig. 4), and the calculated transfer
time is ~75 ps, which is relatively close to the
slowest time obtained by spectroscopy ( 61 ).
Newstructural data from organisms that
can experience low-light conditions show
that the antenna size of PSI can be further
increased by the association of additional Lhcas
( 37 , 38 , 62 – 64 ). In the green algaeC. reinhardtii
( 37 , 38 )andBryopsis corticulans( 63 ), four more
Lhcas form a second belt associated with the
PSI-LHCI basic structure as present in plants,
whereas two more Lhcas are directly con-
nected to the core, which is also the case in PSI
from the red algaCyanidioschyzon merolae(Fig.
4). The absorption cross section is increased by
>50% compared with that of plant PSI, but
the trapping time of the two complexes is very
similar ( 65 ). This may not seem intuitive because
the trapping time usually increases with an-
tenna size [see, e.g., ( 66 )], but the reason is
that the Lhcas inC. reinhardtiido not contain
the same red Chls that are present in plants.
Therefore, the energetic uphill step is far less
steep, leading to faster EET, which compensates
for the larger antenna size ( 65 ). Unfortunately,
the EET rates cannot be calculated from the
availableC. reinhardtiistructures because the
resolution does not allow discrimination be-
tween Chlsaandb, information that is crucial
to permit the identification of the transfer
pathways, and only ~50% of the expected Chls
b( 67 )wereassigned( 38 ). Recently, the struc-
ture of an even larger complex from the moss
Physcomitrella patens,containing12LHCs,
was determined. Although the current resolu-
tion does not allow identification of the pig-
ments, this structure shows new docking sites
for the antennas ( 64 , 68 ), suggesting that the
antenna size of PSI can be even larger than that
observed in the complexes purified up to now.Light harvesting of PSII supercomplexes
ThecorecomplexofPSII( 69 ) (Fig. 5) is highly
conserved in all organisms performing oxy-
genic photosynthesis [for a recent review, see
( 70 )]. Its average excited-state lifetime is typically
~60 to 100 ps ( 66 , 71 – 73 ), and the observed
variation may be due to different stability of
the isolated complexes ( 74 ). This lifetime is
dominated by the relatively slow transfer from
the internal antennas CP43 and CP47 to the
RC, whereas the process of CS itself is much
faster ( 70 ). Although the core complex is mainly
present as a dimer, calculations based on the
available structures indicate that EET between
both adjacent monomers is relatively slow.
Compared with PSI, the association of the
outer antenna to PSII is weaker, easily leading
to disassembly during purification. As a result,
the first homogeneous preparation of the PSII
supercomplex was not reported until 2009 ( 75 ),
and the number of spectroscopic studies has
been limited. In a study comparing differentsupercomplexes of plants, an increase in average
lifetimes was reported going from the core to
C 2 S 2 and C 2 S 2 M 2 ( 66 ), consistent with the
expectation that a larger antenna size cor-
responds to a longer excited-state lifetime
(trapping time) ( 76 ).
Applying the calculation method presented
in Box 2 to the PSII C 2 S 2 structure ( 29 )(Fig.5)
shows that the minor complexes CP26 and CP29
arewellconnectedtothecore.However,in
contrast to what has been reported previously
( 28 , 29 ), they do not function as bridges be-
tween S-LHCII and the core: All LHCII pig-
ments that are close enough for efficient transfer
to CP26/CP29 were reported to be Chlbmole-
cules ( 28 , 29 ) and therefore S-LHCII transfers
directly to the core.
In the C 2 S 2 M 2 complex ( 28 , 30 )(Fig.5),the
M-LHCII trimer is physically connected to
CP24, CP29, and S-LHCII, but it transfers ex-
citations mainly (80%) to CP29 and the connec-
tion is excellent. According to the structures,
the close connections between the Chls of
CP24 and those of CP29 all involve Chlb
molecules, so EET to the core should proceed
through M-LHCII. It cannot be ruled out that
the assignment of the Chlband Chlapig-
ments in the structure of CP24 is not entirely
correct. It was experimentally determined that
in CP24 after exciting Chlb, >~85% of the
excitation energy arrives on Chlawith a time
constant of 600 fs ( 77 ); however, based on the
published structure, this percentage is calcu-
lated to be <50%, whereas most EET is sub-
stantially slower than 1 ps.
In 2019, several groups published the
structure of the PSII supercomplex C 2 S 2 M 2 N 2
of the green algaC. reinhardtii( 34 – 36 ) (Fig. 5).
This complex is not only larger, containing
an additional N-LHCII trimer (also called an
L-trimer) compared with plants, but it also
lacks CP24 and shows a different orientation
of the M-trimer. This different orientation leads
to different excitation energy transfer pathways
in plants and algae, as will be discussed below.
Its C 2 S 2 part is very similar to that of plants and
the calculated effective rates of EET from S-
LHCII, CP29, and CP26 to the core are nearly
identical. Although it was originally believed
that excitations from M-LHCII and N-LHCII
flow to the core through CP29 ( 35 ), closer in-
spection of the structures shows that all close
contacts between CP29 and the S-, M-, and N-
trimers involve Chlb. Conversely, the N-LHCII
trimer has many good connections with M-
LHCII, and M-LHCII is also better connected
to one of the S-LHCIIs than to CP29; there-
fore,S-LHCIIfunctionsasthebridgebetween
the N- and M-trimers and the core.
Fig. 6 provides an overview of the EET
pathways in the various supercomplexes. It
was unexpected that there are mainly parallel
pathways to the core. It had always been ac-
cepted that the minor complexes functionallyCroceet al.,Science 369 , eaay2058 (2020) 21 August 2020 5of9
RESEARCH | REVIEW