Mechanisms of photoisomerization of polyenes in confined media: From organic glasses to protein binding cavities

Article abstract of DOI:10.1562/2006-01-27-RA-786

Photochemical reactivities of model organic systems (stilbene and diphenylbutadiene) in organic glasses were first examined and compared with those in solution and in organized media. These observations were in turn compared with reactivities of polyene chromophores in protein binding cavities (specifically PYP, rhodopsin and bacteriorhodopsin). The obvious conclusion is that the preference for the most volume-conserving Hula-twist mechanism isomerization in organic glasses is because of the close interaction between the guest and the host molecules. In organized media (zeolites, crystals and protein binding cavities), the residual empty space coupled with any specific guest-host interactions that are characteristic of a given system, could lead to involvement of the more volume-demanding one-bond-flip (i.e. torsional relaxation) or bicycle-pedal or an extended HT process in photoisomerization.

Full text of DOI:10.1562/2006-01-27-RA-786

Photochemistry and Photobiology, 2007, 83: 2–10
Mechanisms of Photoisomerization of Polyenes in Confined Media:
From Organic Glasses to Protein Binding Cavities†
Robert S. H. Liu*‡1, Lan-Ying Yang1 and Jin Liu†2
1Department of Chemistry, University of Hawaii, Honolulu, HI, USA
2Department of Chemistry, Murray State University, Murray, KY, USA
Received 27 January 2006; accepted 22 May 2006; published online 23 May 2006 DOI: 10.1562/2006-01-27-RA-786
molecule in-plane (4). The predicted stereochemical conse-
quence of one-photon-two-bond isomerization, however, was
ABSTRACT
Photochemical reactivities of model organic systems (stilbene
and diphenylbutadiene) in organic glasses were first examined
and compared with those in solution and in organized media.
These observations were in turn compared with reactivities of
polyene chromophores in protein binding cavities (specifically
PYP, rhodopsin and bacteriorhodopsin). The obvious conclusion
is that the preference for the most volume-conserving Hula-twist
mechanism isomerization in organic glasses is because of the
close interaction between the guest and the host molecules. In
organized media (zeolites, crystals and protein binding cavities),
the residual empty space coupled with any specific guest–host
interactions that are characteristic of a given system, could lead
to involvement of the more volume-demanding one-bond-flip
(i.e. torsional relaxation) or bicycle-pedal or an extended HT
process in photoisomerization.
not in agreement with nearly all available experimental
evidence obtained with protein bound or solubilized chro-
mophores. Subsequently, the Hula-twist (HT) concept was
introduced by the Hawaii group (3), in which a pair of adjacent
double and single bonds, flanking a single C–H unit, rotates
concertedly. The predicted stereochemical consequence is
simultaneous configurational isomerization of a double bond
and conformational isomerization of an adjacent single bond.
The first unambiguous example demonstrating the predicted
stereochemical consequence of HT was reported by Fuss and
coworkers (5). Many more organic molecules executing HT
photoisomerization in organic glasses have since been des-
cribed in the literature (6–8).
There are, however, less definitive words concerning the
exact mechanism of isomerization of photosensitive protein-
bound polyene chromophores. Since, one by one, the primary
photoproducts of many protein bound polyene chromophores
are being defined by exact structural tools, now appears to be
an appropriate time to examine the mechanisms of the primary
processes in each case. Such questions are in fact part of a
general issue of possible mechanisms of isomerization in
different host systems (9). In this paper, we shall provide an
over-view of mechanisms of photoisomerization conducted in
a variety of media. While some new experimental evidence will
be introduced, this paper is largely a critical review of existing
information in the literature.
INTRODUCTION
Torsional relaxation according to the Mullikan’s energy
diagram for ground and excited states of ethylene has been
the long accepted mechanism for photochemical cis,trans
isomerization (1). However, ever since the first pico-second
time-resolved study on rhodopsin photochemistry (2), the
question how the very volume-demanding torsional relaxation
process which involves turning over of at least one-half of the
chromophore can take place in the rigid binding cavity was
raised. The latter is particularly relevant in view of the short
lifetime of the excited chromophore. Therefore, other mech-
anistic concepts have been introduced in recent decades that
are not as volume-demanding as torsional relaxation. (To
emphasize the rotation around one double bond in a polyene
system, the latter process has also been labeled as one-bond-
flip, OBF (3).) (Fig. 1) Warshel first introduced the bicycle-
pedal process (BP) in which two alternating double bonds
rotate simultaneously while keeping the remaining parts of the
RESULTS
We have carried out photoisomerization of isomers of
1,4-diphenyl-1,3-butadiene (DPB) under three different experi-
mental conditions: in solution, organic glass and in crystal.
The results are described separately below.
Irradiation of DPB in solution. Detailed photoisomerization
studies of isomers of DPB in solution are available in the
literature (10). We have confirmed their conclusion that for
direct irradiation, the isomerization takes place one double
bond at a time starting from any one of the three isomers
producing eventually a mixture of all three isomers. The data
for the c,c isomer is shown in Fig. 2. Because of the efficient
conversion of the c,t isomer to the t,t, the accumulation of
the intermediate c,t isomer was not immediately obvious.
†This invited paper is part of the Symposium-in-Print: Photobiology in Asia.
‡Presented at the Photoisomerization Symposium of the PacifiChem Confer-
ence, Honolulu, Hawaii, USA, December 18, 2005.
§Current address: Jin Liu, P.O. Box 3304, Alhambra, CA, USA.
*Corresponding author email: rshl@hawaii.edu (Robert S. H. Liu)
ꢀ 2007 American Society for Photobiology 0031-8655/07
2

Photochemistry and Photobiology, 2007, 83
3
Figure 1. Three possible mechanisms of photoisomerization. Top: torsional relaxation where one-half of the molecule flips around the formal
double bond (or one-bond-flip, OBF). Middle: the bicycle-pedal (BP) process that results in isomerization at two formal double bonds. Bottom: the
Hula-twist (HT) that results in configurational isomerization at one double bond and conformational isomerization at an adjacent single bond. The
cartoon figures on the right emphasize the parts that change in space.
However, it is clear that the initial difference in spectrum was
rather featureless (characteristic of a cis product) which is
different from that of c,t (see Fig. 3 below).
initial product with a featureless absorption band (consistent
with the c,t isomer) (Fig. 4).
Irradiation of DPB crystals. Small crystals of three isomers
of DPB were irradiated with >300 nm light (Pyrex filter).
Progress of reaction was followed by dissolving small amounts
of the powder and analyzed by HPLC or by H NMR (10,11).
Both the t,t and the c,t crystals were found to be un-reactive
under the UV light. However, crystals of the c,c isomer were
found to undergo stereospecific isomerization to the t,t isomer.
We interrupted irradiation at early stages of irradiation with
conversions ranging from 10–15%. It was clear that the c,t
isomer was never present throughout these periods of irradi-
ation (based on hplc and H NMR analyses of dissolved
crystals). This photochemical behavior was also observed in
isomers of two derivatives of DPB: p,p¢-difluoro-DPB and p,p¢-
dimethoxy-DPB.
Irradiation of 1,4-diphenyl-1,3-butadiene in organic
glasses. Preliminary results on irradiation of 1,4-diphenyl-
1,3-butadiene (DPB) isomers in organic glasses at liquid
nitrogen temperature were reported earlier (11). Below we des-
cribe some of the observations on each isomer in detail.
The t,t isomer of DPB was found to be un-reactive in EPA
(ether: isopentane: ethanol = 5 : 5 : 2) glass at 77 K. This
observation is consistent with what is known for diarylethy-
lenes (12). On the other hand, both the c,t or the c,c isomers of
DPB were found to be reactive under the same condition. For
c,t-DPB, the conversion to the t,t isomer was a clean one-to-
one conversion (Fig. 3). On the other hand, for the c,c isomer
the conversion to the t,t is clearly stepwise going through an
1.2
0.6
1.8
1.2
A
D
Figure 2. Changes in UV absorption spectra
0.6
0.0
of c,c-DPB in EPA solution at 200 K with
Corning O-54 filter (>310 nm). Left: absorp-
tion spectra. t = 0, 2, 4, 6, (in red) 8, 12, 17,
23, 43, 53, 73 s (in purple). Right: difference
spectra from left (aqua for the early periods).
Notice the featureless broad spectrum at the
earliest stage of irradiation indicating possible
formation of a cis isomer (i.e. c,t).
0.0
250
300
350
250
300
350
Wavelength (nm)

4
Robert S. H. Liu et al.
1.0
0.5
0.0
1.5
1.0
0.5
0.0
1
0
300
350
A
D
250
300
350
250
300
350
Wavelength (nm)
Figure 3. Changes in UV absorption spectra recorded during irradiation of c,t-DPB in EPA glass at 77 K with O-54 filter. Left: absorption spectra.
t = 0, 30, 80, 140, 220, 310, 550, 790, 1030 and 1250 s. Right: difference spectra after the same time intervals. Notice the appearance of fine
structure at the earliest stage of irradiation (blue) showing formation of the t,t-isomer. Insert: absorption spectra of the photoproduct before (solid
line) and after (dashed) warming to room temperature and re-cooling to 77 K.
1.8
1.2
0.6
0.0
1.2
0.6
0.0
1
0
350
300
A
D
250
300
350
250
300
350
Wavelength (nm)
Figure 4. Changes in UV absorption spectra of c,c-DPB recorded during its irradiation in EPA glass at 77 K with >310 nm light. Left: absorption
spectra. t = 2, 6, 10, 20, 30, 50, 110 (in red), 230, 350, 470, 590, 710, 830, 950, 1010 and 1250 s (purple). Right: difference spectra. Aqua, early stage
of irradiation; blue, late stage of irradiation. Notice that the fine structures only appeared at later stages. Insert: absorption spectra of photoproduct
before (solid) and after (dashed) warming to rt and re-cooling to 77 K.
Since the unusual two-bond isomerization of the c,c
isomer is different from that of the crystals, we proceeded
to determine the crystal structures of two c,c-DPB’s (Figs. 5
DISCUSSION
Formation of unstable products during irradiation of isomers
of DPB only in a low temperature glass (Figs. 3 and 4) showed
and 6).
that these unique photochemical properties are associated with
the restricting media. Similar unstable products were also
obtained for styrenes (7) and dienes (8) in organic glasses. It
became logical to ask whether such photochemical observa-
tions are to be expected in other confining media such as
zeolite, crystals and even the binding cavities of proteins.
Earlier, one of us (9) reasoned that the reaction sites within the
amorphous media of organic glasses are very different from
those of the organized media. The collapsing solvent cage
surrounding a substrate during the course of freezing must be
constantly changing in shape accompanied by a decrease in the
size of the enclosed cavity. On the other hand, for an organized
medium, the shape of the cavity is defined by the unique
architecture of the host molecule. It is usually rigid in shape
and the confined cavity is not likely to change much with
temperature or after occupancy of a substrate molecule or
Figure 5. The X-ray crystal structure of c,c-p,p¢-difluoro-DPB. The
distance between the diene units in two molecules is 4.908 A.
˚

Photochemistry and Photobiology, 2007, 83
5
Figure 6. The X-ray crystal structure of c,c-p,p¢-dimethoxy-DPB. The distances between the double bonds of two sets of non-equivalent
molecules average: 4.46 and 5.08 A.
˚
other (e.g. solvent) molecules. In most cases, the guest
molecules are usually introduced into the cavity through
diffusion. There is no reason to suspect that the binding cavity
can be fully occupied (unless there are unique specific
attractive interactions), leaving the possibility of substantial
1.5
A
unfilled space. The residual space can be filled by solvent
molecules if they were present at the time when the substrate
was introduced. This would suggest that for an organized host,
the inhibiting effect of the wall of the host cavity is likely to be
less than that of an amorphous organic glass. However, it
should also be recognized that specific interactions between
part of the ‘‘walls’’ of the organized hosts with the substrate
molecules could dictate the manner of photochemical trans-
formation of the substrate (specifically the isomerization
pathway) in that particular guest-host pair. We would like to
examine available experimental evidence that might shed light
to these generalizations.
1.0
0.5
0
500
1000
1500
T (s)
Figure 7. Changes of absorbance at 305 nm during irradiation
(>310 nm) of isomers of stilbene at 77 K in EPA glass: round dots,
trans isomer; triangles, cis isomer. The slight rise in absorption during
early stage of irradiation of the trans isomer was probably due to local
thawing and re-freezing (15).
Photoisomerization of simple organic polyenes
Stilbenes and 1,4-diphenylbutadienes. 1,2-Diphenylethylene,
or stilbene and 1,4-diphenylbutadiene (DPB) are model organic
compounds frequently used for studies of the photoisomeriza-
tion reaction. It will be of interest to compare their photo-
chemical behavior in solution and in various confined media.
In the case of the stilbene, ample data are available in the
literature (13). In solution, its two isomers, upon direct
irradiation, undergo ready inter-conversion giving a photosta-
tionary state containing the two isomers. Additionally, there is
a slower 6e-cyclization reaction from the cis isomer giving the
colored dihydrophenanthrene (not formed in low temperature
glass). Upon increase of solvent viscosity, it is known that the
trans to cis photoisomerization becomes progressively less
efficient (14). Eventually, in frozen media, the reaction stopped
completely giving way to a high fluorescence yield of the trans
isomer. The cis isomer, however, retained partly (ꢀ30% at 77 K
in EPA) (6) its photoreactivity. Hence, in a low temperature
glass, the eventual product mixture, or the photostationary
state mixture, contained only the trans isomer (Fig. 7).
Ramamurthy et al. described results of photoisomerization of
stilbene imbedded in the Na Y zeolite (16), prepared by the
interaction of the vacuum heated zeolite with a hexane
solution of t-stilbene. Ready photoisomerization was
observed, yielding an eventual photostationary state mixture
identical to that of a hexane solution of stilbene. In other
words, confinement did not lead to reduced efficiency of trans
to cis isomerization. It is a clear indication that the wall of the
zeolite must be too remote to provide any inhibiting effects to
the stilbene substrate. And, any void space in the zeolite cavity
must have been filled by hexane molecules.
hν
Because of its symmetrically substituted rings, for stilbene,
it is not possible to ascertain whether the low temperature
isomerization of the cis isomer proceeds by way of the OBF
process (torsional relaxation) or the volume-conserving HT
process. However, if one infers from the results of o-substi-
tuted stilbenes (6,15), it seems safe to conclude that in a low
temperature glass cis-stilbene also isomerizes by way of HT.
For DPB isomers, the three isomers are known to undergo
ready photochemical inter-conversion in solution giving a
photostationary state mixture containing all three isomers
(10). For direct irradiation, the isomerization takes place one
double bondat a time. However, under the confined condition of
organic glass, the t,t isomer is photostable while the c,t and c,c

6
Robert S. H. Liu et al.
isomers isomerize readily. Thus, prolonged irradiation resulted
in exclusive formation of the t,t isomer. That a stable t,t isomer
was obtained from c,t-DPB (insert of Fig. 3) indicates that the
reaction did not proceed by the way of HT-2 isomerization. The
HT-1 process however gives the same stable t,t isomer as the
OBF process. Only after the use of o,o¢-disubstituted DPBs (11),
the latter was successfully eliminated. Therefore, same as cis-
stilbene, the isomerization reaction has taken place in the least
volume-demanding HT manner for the more restricting media
of organic glasses. And, for c,c-DPB in EPA glass, the reaction
proceeded sequentially first yielding the stable c,t isomer thenthe
t,t isomer culminated with the structured absorption band that is
characteristic of the t,t isomer (Fig. 4).
It should be noted that bond order reversal upon light
excitation facilitates this two-bond isomerization process.
Furthermore, the observed reaction is an energy down-hill
c,c to t,t conversion that conveniently overcomes any activa-
tion energy suspected to be present in a BP process of an
excited polyene as shown by the MNDO calculations (17).
The bulky phenyl end groups in the crystal prevents the
molecule from undergoing the OBF process of which one of
the two phenyl rings must turn over. The HT process is also
not favored, most likely due to the substantial lateral dis-
placement of the phenyl ring involved in such a process
(Fig. 1). Consequently, the BP process presents itself as a least
motion compromise, uniquely fitting within the intermolecular
empty space present in c,c-DPB crystals.
Other organic examples of bicycle-pedal process motion in the
literature. There are, in fact, other examples of two-bond
isomerization in the literature. Several years ago, the direct
conversion of c,c muconate salt (1) to its t,t isomer was
reported (18) (Fig. 8c), the first example of two-bond BP
isomerization. It is closely related to the current case of
c,c-DPB in that both involve stereospecific conversion from
the higher energy c,c isomer to the t,t isomer. In 2003, the
surprising result of conformation flexibility in crystals of trans-
stilbene and trans-azobenzene was described (19). The rotation
of the ethylenic unit is in fact a BP motion (Fig. 8a) turning
simultaneously, in this case, two alternating single bonds (for a
ground state process). These observations underscore the large
intermolecular space present around the ethylenic unit, made
possible by the bulky end groups (the phenyl groups), a
situation similar to those in crystals of c,c-DPB and muconate
salt.
In organic crystals, the isomers of DPB exhibited different
photoreactivities: the c,t isomer being photostable while the c,c
isomer isomerizing stereospecifically to the t,t isomer. The
cause for the latter new photochemical reactivity becomes
evident when one compares the crystal structure of c,c-DPB
with different conformers of t,t-DPB. Only the s-trans con-
former (red) is close in shape with that of the stable conformer
of the c,c isomer (black). In crystals, the molecule is likely to be
‘‘anchored’’ at the two bulky ends (no room for rotation of the
phenyl groups), leaving considerable intermolecular space
between the connecting butadiene chains (Figs. 5 and 6). It
allows rotational motions of these atoms in the form of
volume-conserving bicycle-pedal (BP) process for the conver-
sion of c,c-isomer directly to the t,t-isomer. As no flipping over
of the phenyl rings is involved, it is a least motion process in
the chemical transformation.
It should also be pointed out that the photoisomerization
observed in crystals of the photochromic salicylideneaniline (3)
(20) and trans-dibenzoylethylene (4), crystals (21) are likely the
result of BP motion of the ethylenic bond coupled with a
nearby single bond (see red arrows). (HT was proposed for the
Figure 8. The common bicycle-pedal-like motion in four different systems. Ground state conformational equilibration in crystals of (a) t-stilbene
(19) and (b) the trans thiocinnamate (2) chromophore in crystals of PYP (26); (c) photoisomerization of c,c-muconate salt (1), black, to the t,t-
isomer, red (18) and (d) photoisomerization of the thiocinnamate chromophore of PYP (2), black, to the cis isomer, red (24,25).

Photochemistry and Photobiology, 2007, 83
7
latter (21).) Both examples are in reality a double bond twist
coupled with a nearby single bond twist, however, rotating the
middle two-carbon fragment in a manner similar to a BP
process. We might add that in a recent conference report,
Harada reached the similar conclusion in their X-ray crystal
study of (4) (22).
re-investigation of the crystal structure of PYP at low tempera-
ture revealed the presence of two conformers of the cinnamate
chromophore which inter-convert at slightly elevated tempera-
ture (26). It involves the motion of connecting ethylenic unit only
(around the dotted single bonds), equivalent to the BP-like
motion in t-stilbene and t-azobenzene (see Fig. 8b). We
emphasize that this observation again revealed a large empty
space surrounding the ethylenic unit of the PYP chromophore.
The exact nature of the photochemical transformation was
made clear by the crystal structure of its primary photoproduct
(24,25). The process takes advantage of the s-cis ene-ester
conformation present in the pigment, that makes the single
bond thioester linkage parallel to the alkene double bond (bold
lines), a pre-requisite for BP-like motion. The structures of the
chromophore before and after irradiation (according to
crystallographic data) are shown in the form of overlaid
bond-line structures (Fig. 8d): the trans to cis isomerization
accompanied by flipping over of the carbonyl group of the
thioester, an expected consequence of the BP-like motion. The
motion is clearly shown in the overlaid structures (Fig. 8d). It
involves motion only of the middle two-carbon fragment (an
ethylenic carbon and the carbonyl carbon). It is a motion
closely similar to that in t-stilbene except in this case a
concerted rotation of only one formal double bond along with
a single bond is involved (Fig. 8d). The clear advantage of such
a transformation is that it does not lead to disruption of the
cysteine (covalent) and phenoxy (ionic) anchors at the two
ends. (Since only one double bond changes its configuration
one might also label this process a modified OBF.) This
isomerization mechanism is apparently uniquely fitting for the
short, anchored chromophore of PYP. It should also be clear
that the short Cys tether and the short chromophore do not
permit a HT process that would lead to significant lateral
displacement of either one of the anchors.–
Rhodopsin photoisomerization. The HT concept was origin-
ally introduced (3) to account for the facile isomerization of
the confined 11-cis-retinyl chromophore (black) in rhodopsin
to the all trans form (red). (See structures below for the
confined and anchored chromophores.) It requires the forma-
tion of an s-cis linkage (whether 10-s-cis or 12-s-cis) in the all-
trans photoproduct. However, such a conformation was not
detected in vibrational structural analyses of the stable
photoproduct bathorhodopsin (28). Whether such an unstable
conformation can be present in the unstable earlier interme-
diate photorhodopsin (29), as suggested by Fuss (5), remained
Photoisomerization of polyene chromophore within protein
binding cavities
Protein binding cavities are also host systems with rigid, well-
defined structural framework. Hence, they are similar to
cavities in organic crystals that are usually larger than the
substrates that they accept. Thus even when a protein binding
site is ‘‘filled,’’ there should still be empty space surrounding
the trapped substrate for reaction. Hence, the surrounding
environments cannot be as restrictive as the collapsing solvent
molecules in formation of organic glasses. We would like to
show below that the recently reported photochemical struc-
tural studies of three photoactive pigments (photoactive yellow
protein, rhodopsin and bacteriorhodopsin) revealed larger
empty space available for reaction for each case. In fact, their
primary photochemical reactions are often dictated by the
need to execute a least motion of the chromophore so as not to
disturb the anchors that position the chromophore at the
unique, defined position. (See below on implication of specific
model of protein–substrate interactions.)
The photoactive yellow protein. The facile trans to cis
isomerization of the small and anchored thiocinnamate chro-
mophore (2) of photoactive yellow protein (PYP) within its
binding cavity remains an exciting and unexpected observa-
tion. The specific nature of this conversion has become
increasingly clear through several elegant steady state (23,24)
and time-resolved X-ray crystallographic studies (25) on this
protein.
The thiocinnamate chromophore was shown to exist
exclusively in the s-cis ene-ester conformation. It is remarkably
close in shape to those of trans-stilbene and azobenzene
(Fig. 8a). Therefore, it is perhaps not surprising that a recent
–During the early stage of development of the concept of medium directed
isomerization, one of us (Liu, 27) incorrectly suggested the involvement of HT
process in the primary process of PYP pigment. The mistake was partly due to
an incorrect interpretation, on his part, of the crystal structure of the
photoproduct.

8
Robert S. H. Liu et al.
unanswerable. Subsequent theoretical studies (30,31) also
suggested direct formation of the all-trans photoproduct with
an s-cis linkage present only at the 6,7-single bond.
with the motion of the C–H12 fragment (36), a process made
clear in the most recent ultrafast time-resolved RR study (37).
The preferred direction of rotation of the C12–H fragment to
initiate isomerization must be into the empty space encircled
by TM-4,5 (yellow arrow in Fig. 9). Again, we emphasize the
presence of a large empty space surrounding the retinyl
chromophore.
A crystal structure of bathorhodopsin, the primary product,
has recently become available (38). A comparison of the
structures of the 11-cis-retinyl chromophore in rhodopsin and
the all-trans-retinyl chromophore in bathorhodopsin from the
crystal structures revealed several interesting features. First,
the 11,12 cis linkage, twisted even in rhodopsin ()41ꢁ) (27,28),
rotates by 115ꢁ to another twisted structure in bathorhodopsin
()152ꢁ), much less than the usual 180ꢁ rotation in a cis to trans
isomerization. This twist is negated by many small twists in the
opposite direction of the nearby single and double bonds. As a
result, it is possible for this lengthy chromophore to execute
the OBF motion without disrupting either end of the anchors
(the hydrophobic pocket for the cyclohexenyl ring or the
lysine-iminium terminal). Given the empty space surrounding
the chromophore mentioned above, the possibility of carrying
out the usually volume-demanding OBF process is perhaps not
surprising.
The empty space revealed by the crystal studies also
required modification of the view on the underlying causes
for the highly stereoselective recognition of the 11-cis retinal by
the binding pocket of opsin. It is clearly not because of the
snug interactions between the binding cavity and the substrate
(the lock-&-key model). Rather, the recognition is achieved
through the fulfillment of several key contact points (the lysine
anchor, the hydrophobic pocket recognizing the ring, the 9-
methyl recognition site, the protein wall near C-12 and the
chiral twists of the polyene chain) as pointed out in a recent
paper (39).
The crystal structure of rhodopsin became available in 2000
˚
(32). In spite of the fact that its resolutions (2.8 and 2.5 A) (33)
are far from being sufficient in providing finer structural
information such as exact conformational properties of the 11-
cis-retinyl chromophore, there is much interesting new infor-
mation derived from the structure.
A most unusual feature is the location of the retinyl
chromophore. Instead of the anticipated location in the middle
of the heptahelical bundle of the opsin binding site, it sits close
to one end of the helical bundle (34,35). One side of it faces
helix-3 and the other side faces the large trans-membrane loop
that connects helices 4 and 5 (TM-4, 5). The loop, however, is
rather rigid because of the presence of two disulfide linkages.
Encircled by this large loop is a sizable empty space that was
revealed in the crystal structure, as shown in Fig. 9. Only
˚
protein residues within 4.0 A of either side of the 11-cis retinyl
chromophore are shown. On the left side, the 11,12-cis linkage
leans against helix-3 while on the right side, it faces virtually an
empty space. It is known that the isomerization process starts
Bacteriorhodopsin (bR) photoisomerization. The primary
photochemical process for bR is the all-trans (black) to 13-cis
(red) isomerization of the imbedded retinyl chromophore (see
structures below). An s-cis conformation was also suspected to
be present in the primary photoproduct of bR (40). However,
the more recent crystal structure of the stable early interme-
diate K (41) showed an absence of such a conformation.
Glu113
Ser186
Cys187
Lys296
C-12
C-11
Ala117
Tyr268
Tyr191
Gly121
Thr118
Ala269
Ile189
Phe261
Phe212
Met207
Glu122
His211
We have now superimposed the crystal structure of the all-
trans-retinyl chromophore structure of bR (42) with that of K
(13-cis) (41) in Fig. 10. It is clear that the butyl tether is not
fully extended in bR (especially twists at the 16,17 and 17,18
single bonds) which allows the conversion of the longer
all-trans geometry (43) to the shorter 13-cis of the early K
Figure 9. Partial crystal structure of rhodopsin (32) showing the
amino acid residues surrounding the 11-cis-retinyl chromophore
˚
(green). Only atoms within 4.0 A of the chromophore are shown. It
is clear that the chromophore leans against helix-3 (amino acids 113–
122) on the left while facing an open space on the right. The yellow
arrow indicates the preferred direction of isomerization.

Photochemistry and Photobiology, 2007, 83
9
extended HT for bR. Thus, it is not surprising that the new
photochemical results obtained in crystals of diphenylbutadie-
nes and other reported studies of organic crystals are similar to
that of a protein (specifically that of PYP).
Acknowledgements—The work was supported by grants from the
National Science Foundation (CHE-05-14737) and Kentucky NSF
EPSCoR (4-65752-03-397). Daniel Chang provided valuable assistance
in producing computer graphics shown in Figs. 9 and 10. J. Liu thanks
Dr. Victor G. Young, Jr. at X-Ray Crystallographic Laboratory,
University of Minnesota for the determination of the crystal struc-
tures.
14
15
16
14
15
16
REFERENCES
1. Mulliken, R. S. and C. C. I. Roothaan (1947) The twisting fre-
quency and the barrier height for free rotation in ethylene. Chem.
Rev. 41, 219–231.
2. Busch, G. D., M. L. Applebury, A. A. Lamola and P. Rentzepis
(1972) Formation and decay of prelumirhodopsin at room tem-
perature. Proc. Natl Acad. Sci. USA 69, 2802–2806.
3. Liu, R. S. H. and A. E. Asato (1985) The primary process of vision
and the structure of bathorhodopsin: A mechanism for photo-
isomerization. Proc. Natl Acad. Sci. USA 82, 259–263.
4. Warshel, A. (1976) Bicycle-pedal model for the first step in the
vision process. Nature (London) 260, 679–683.
Figure 10. The all-trans-retinyl chromophore from crystal structure of
bR (blue) (42) superimposed with the 13-cis chromophore from the
crystal structure of the K intermediate (maroon) (41). Most of the
atoms remain unperturbed with the exception of the marked 14, 15 and
16 segment of the conjugated chromophore that turned over during the
transition from bR to K, through rotation at the two bonds shown
(green arrows). Three of the closest amino acid residues are also
shown.
5. Muller, A. M., S. Lochbrunner, W. E. Schmid and W. Fuss (1998)
¨
intermediate, compensated by the stretching of butyl tether.
The process involves significant movement of primarily C-14
and the 15,16-imino bond by turning over this segment of the
conjugated system. In other words, it must have involved the
rotation of the 13,14 formal double bond and the 16,17 single
bond. The process can be viewed as an extended HT process
(turning over of a three-carbon fragment rather than a one-
carbon fragment). Apparently, the curved and anchored
nature of the butyl side chain and the empty space surrounding
the imino end of the conjugated system have made this unique
process of photoisomerization the preferred one in bR.
Low-temperature photochemistry of previtamin D: Hula-twist
isomerization of a triene. Angew. Chem. Int. Ed. 37, 505–507.
6. Imamoto, Y., T. Kuroda, M. Kataoka, S. Shevyakov, G.
Krishnamoorthy and R. S. H. Liu (2003) Photoisomerization by
hula-twist. 2,2¢-dimethylstilbene and a related ring-fused analog.
Angew. Chem. Int. Ed. 42, 3630–3633.
7. Krishnamoorthy, G., A. E. Asato and R. S. H. Liu (2003) Con-
formational isomerizations of symmetrically substituted styrenes.
No-reaction photoreactions by hula-twist. Chem. Commun., 2170–
2171.
8. Krishnamoorthy, G., S. Schieffer, S. Shevyakov, A. E. Asato, K.
Wong, J. Head and R. S. H. Liu (2004) S-cis conformers in low
temperature irradiation of dienes. Hula-twist mechanism of
isomerization. Res. Chem. Interm. 30, 397–405.
9. Liu, R. S. H. and G. S. Hammond (2005) Reflection on medium
effects on photochemical reactivity. Acc. Chem. Res. 38, 396–403.
10. Yee, W. A., S. L. Hug and D. S. Kliger (1988) Direct and sensi-
tized photoisomerization of 1,4-diphenylbutadienes. J. Am. Chem.
Soc. 110, 2164–2169.
11. Yang, L.-Y., R. S. H. Liu, K. J. Bowman, N. L. Wendt and J. Liu
(2005) New aspects of diphenylbutadiene photochemistry. Reg-
ioselective hula-twist photoisomerization. J. Am. Chem. Soc. 127,
2404–2405.
12. Mazzucato, U. and F. Momicchioli (1991) Rotational isomerism
in trans-1,2-diarylethylenes. Chem. Rev. 91, 1679–1719.
13. Waldeck, D. H. (1991) Photoisomerization dynamics of stilbenes.
Chem. Rev. 91, 415–436.
14. Malkin, S. and E. Fischer (1964) Temperature dependence of
photoisomerization. III. Direct and sensitized photoisomerization
of stilbenes. J. Phys. Chem. 68, 1153–1163.
15. Yang, L.-Y., M. Harigav, Y. Imamoto, M. Kataoka, T.-I. Ho, E.
Andrioukhina, O. Federova, S. Shevyakov and R. S. H. Liu (2006)
1,2-Diarylethylenes in hula-twist photoisomerization. Photochem.
Photobiol. Sci. 5, 874–882.
16. Syamala, M. S., S. Devanathan and V. Ramamurthy (1986)
Modification of the photochemical behavior of organic molecules
by cyclodextrin: Geometric isomerization of stilbenes and alkyl
cinnamates. J. Photochem. 34, 219–229.
17. Seltzer, S. (1987) MNDO barrier height for catalyzed bicycle-
pedal, hula-twist and ordinary cis–trans isomerizations of proto-
nated retinal Schiff base. J. Am. Chem. Soc. 109, 1627–1631.
18. Odani, T., A. Matsumoto, K. Sada and M. Miyata (2001) One
way E,Z-isomerization of bis(n-butylammonium) (Z,Z)-muconate
under photoirradiation in the crystalline state. Chem. Commun.,
2004–2005.
CONCLUSION
Photoisomerization from the excited singlet state is a diabatic
process proceeding directly from the excited potential surface
to the ground state product. Different mechanisms of reactions
are possible under different confined conditions for the
reacting polyene. We argued that the close interactions
between host molecules and the substrate in amorphous
organic glasses have made the most volume-conserving HT
process a preferred process of isomerization in such media. It
differs from the common OBF process for photoisomerization
under the non-restrictive condition of common solutions. In
organized media (zeolites, crystal, protein binding cavities,
etc.), the rigid host structures are likely to retain some empty
space even when the cavity is occupied by a substrate. This
residual empty space plus other specific interactions between
the host and the substrate (e.g. specific points of recognition
and the rigid anchors as well as the length of the chromophore
and the appended tether) could lead to uniquely different
photoisomerization mechanism(s) suitable for that particular
pair of guest and host molecules. When coupled with a second
single bond rotation, the usual volume-demanding OBF
process can become part of a least motion process. Interest-
ingly, for each of the three well-known photosensitive biopig-
ments, a different mechanism of photoisomerization is
involved: modified BP for PYP, OBF for rhodopsin and

10 Robert S. H. Liu et al.
19. Harada, J. and K. Ogawa (2004) X-ray diffraction analysis of non-
equilibrium states in crystals: Observation of an unstable con-
former in flash-cooled crystals. J. Am. Chem. Soc. 126, 3539–3544.
20. Harada, J., K. Uekusa and Y. Ohashi (1999) X-ray analysis of
structural changes in photochromic salicylideneaniline crystals.
J. Am. Chem. Soc. 121, 5809–5810.
21. Kaupp, G. and J. Schmeyers (2000) The solid-state E ⁄ Z photo-
isomerization of 1,2-dibenzoylethene. J. Photochem. Photobiol. 59,
15–19.
32. Palczewski, K., T. Kumasaha, T. Hori, C. A. Behnke, H. Moto-
shima, B. A. Fox, I. Le Trong, D. C. Teller, T. Okada, R. E.
Stenkamp, M. Yamamoto and M. Miyano (2000) Crystal struc-
ture of rhodopsin: A G protein-coupled receptor. Science 289,
739–745.
33. Okada, T., Y. Fujiyoshi, M. Silow, J. Navarro, E. M. Landau and
Y. Shichida (2002) Functional role of internal water molecules in
rhodopsin revealed by x-ray crystallography. Proc. Natl. Acad.
Sci., USA 99, 5982–5987.
22. Harada, J. (2005) A conference report presented at the PacifiChem
Conference, Honolulu, December 18, 2005.
23. Borgstahl, G. E. O., D. R. Williams and E. D. Getzoff
34. Teller, D. C., T. Okada, C. A. Behnke, K. Palczewski and R. E.
Stenkamp (2001) Advances in determination of a high-resolution
three-dimensional structure of rhodopsin, a model of G-protein-
coupled receptors (GPCRs). Biochemistry 40, 7761–7772.
35. Liu, R. S. H. and L. U. Colmenares (2003) The molecular basis for
the high photosensitivity of rhodopsin. Proc. Natl. Acad. Sci.,
USA 100, 14639–14644.
36. Mathies, R. A. and J. Lugtenburg (2000) The primary photore-
action of rhodopsin. In Handbook of Biological Physics (Edited by
D. G. Stavenga, W. J. deGrip and E. N. Pugh, Jr. ), vol. 3, pp. 55–
90. Elsevier Science, Amsterdam.
37. Kukura, P., D. W. McCamant, S. Yoon, D. B. Wandschneider
and R. A. Mathies (2005) Structural observation of the primary
isomerization in vision with femtosecond-stimulated Raman.
Science 310, 1006–1009.
38. Nakamichi, H. and T. Okada (2006) Crystallographic analysis of
primary visual photochemistry. Angew. Chem. Intern. Ed., 45,
4270–4273, and personal communication by Dr. Okada to RL.
39. Liu, R. S. H., G. S. Hammond and L. U. Colmenares (2005)
Possible role of the 11-cis-retinyl conformation in controlling the
dual decay processes of excited rhodopsin. Proc. Natl. Acad. Sci.,
USA 102, 10783–10787.
˚
(1995) 1.4 A Structure of photoactive yellow protein, a cytosolic
photoreceptor: Unusual fold, active site and chromophore. Bio-
chemistry 34, 6278–6287.
24. Genick, U. K., S. M. Soltis, P. Kuhn, I. L. Canestrelli and E. D.
˚
Getzoff (1998) Structure at 0.85 A resolution of an early protein
photocycle intermediate. Nature 392, 206–209.
25. Genick, U. K., G. E. O. Borgstahl, K. Ng, Z. Ren, C. Pradervand,
P. M. Burke, V. Srajer, T. Y. Teng, W. Schildkamp, D. E. McRee,
K. Moffat and E. D. Getzoff (1997) Structure of a protein phot-
ocycle intermediate by millisecond time-resolved crystallography.
Science 275, 1471–1475.
26. Anderson, S., V. Srajer and K. Moffat (2004) Structural hetero-
geneity of cryotrapped intermediates in the bacterial blue light
photoreceptor, photoactive yellow protein. Photochem. Photobiol.
80, 7–14.
27. Liu, R. S. H. (2001) Photoisomerization by hula-twist: A funda-
mental supramolecular photochemical reaction. Acc. Chem. Res.
34, 555–562.
28. Paling, I., J. A. Pardoen, E. van den Berg, C. Winkel, J. Lugten-
burg and R. A. Mathies (1987) Assignment of fingerprint vibra-
tions in the resonance Raman spectra of rhodopsin, isorhodopsin
and bathorhodopsin: Implications for chromophore structure and
environment. Biochemistry 26, 2544–2556.
40. Schulten, K. and P. A. Tavan (1978) Mechanism for light-driven
proton pumping of Halobacterium halobium. Nature (London)
272, 85–86.
41. Edman, K., P. Nollert, A. Royant, H. Belrahli, E. Peboy-Peyoula,
J. Hadju, R. Neutze and E. M. Landau (1999) High resolution X-
ray structure of an early intermediate in the bacteriorhodopsin
photocycle. Nature 401, 822–826.
29. Shichida, Y., S. Matuoka and T. Yoshizawa (1984) Formation of
photorhodopsin, a precursor of bathorhodopsin. Photobiochem.
Photobiophys. 7, 221–228.
30. Warshel, A. and N. Barboy (1982) Energy storage and reaction
pathways in the first step of the vision process. J. Am. Chem. Soc.
104, 1469–1476.
42. Lueke, H., B. Schobert, H. T. Richter, J. P. Cartailler and J. K.
˚
Lanyi (1999) Structure of bacteriorhodopsin at 1.55 A resolution.
J. Mol. Biol. 291, 899–911.
31. Yamada, A., T. Yamato, T. Kakitani and S. Yamamoto (2004)
Torsional potential works in rhodopsin. Photochem. Photobiol. 79,
476–486.
43. Liu, R. S. H. and D. Browne (1986) A bioorganic view of the
chemistry of vision: HT-n and BP-m,n mechanisms for reactions
of confined, anchored polyenes. Acc. Chem. Res. 19, 42–48.