Synthesis and self-aggregation of zinc chlorophylls possessing an ω-hydroxyalkyl group: Effect of distance between interactive hydroxy group and chlorin moiety on aggregation

Article abstract of DOI:10.1039/b107902f

Zinc 13¹-oxochlorins 2,3 and 3D possessing 2-hydroxyethyl, 3-hydroxypropyl and 3-hydroxyprop-1-enyl groups, respectively, at the 3-position are synthesized as models for self-aggregative antenna chlorophylls in green photosynthetic bacteria. Self-aggregation of 2,3 and 3D in nonpolar organic solvents and in the solid state is compared with that of 1 possessing a 3-hydroxymethyl group to determine the effect of the distance between the hydroxy group and the chlorin moiety on the self-aggregation. Visible spectral analyses in hexane containing a small amount of THF reveal that the aggregation abilities decrease in the order of 1 → 2 → 3, with an increase of conformational flexibility of the ω-hydroxyalkyl group in a molecule. Aggregated 2 and 3 give absorption maxima at 701 and 702 nm, respectively, red-shifted from the corresponding monomeric absorption (644 nm). These red-shifts are smaller than that of 1 (647 → 740 nm), which is attributable to the expanded chlorin π-π plane distance in the self-aggregates of 2 and 3. Furthermore, their CD and IR spectra reveal that the aggregates of 3 are relatively disordered and have weak intermolecular noncovalent interactions among the hydroxy group, central zinc and keto group in the supramolecule. Aggregated 3D shows a relatively small red-shift of absorption from monomer to aggregates (654 → 680 nm) due to the decreased overlap between the chlorin π-planes in the aggregated state. However, 3D easily forms precipitates composed of structurally ordered large aggregates, indicating that 3D is favorable for molecular packing in the aggregates.

Full text of DOI:10.1039/b107902f

View Article Online / Journal Homepage / Table of Contents for this issue
Synthesis and self-aggregation of zinc chlorophylls possessing an
ꢀ-hydroxyalkyl group: effect of distance between interactive
hydroxy group and chlorin moiety on aggregation†
1
Shiki Yagaia and Hitoshi Tamiaki*a,b
a Department of Bioscience and Biotechnology, Faculty of Science and Engineering,
Ritsumeikan University, Kusatsu, Shiga 525-8577, Japan
b “Form and Function” PRESTO, Japan Science and Technology Corporation (JST), Kusatsu,
Shiga 525-8577, Japan
Received (in Cambridge, UK) 5th September 2001, Accepted 15th October 2001
First published as an Advance Article on the web 8th November 2001
Zinc 131-oxochlorins 2, 3 and 3D possessing 2-hydroxyethyl, 3-hydroxypropyl and 3-hydroxyprop-1-enyl groups,
respectively, at the 3-position are synthesized as models for self-aggregative antenna chlorophylls in green photo-
synthetic bacteria. Self-aggregation of 2, 3 and 3D in nonpolar organic solvents and in the solid state is compared
with that of 1 possessing a 3-hydroxymethyl group to determine the effect of the distance between the hydroxy group
and the chlorin moiety on the self-aggregation. Visible spectral analyses in hexane containing a small amount of
THF reveal that the aggregation abilities decrease in the order of 1
2
3, with an increase of conformational
flexibility of the ω-hydroxyalkyl group in a molecule. Aggregated 2 and 3 give absorption maxima at 701 and 702 nm,
respectively, red-shifted from the corresponding monomeric absorption (644 nm). These red-shifts are smaller than
that of 1 (647
740 nm), which is attributable to the expanded chlorin π–π plane distance in the self-aggregates of 2
and 3. Furthermore, their CD and IR spectra reveal that the aggregates of 3 are relatively disordered and have weak
intermolecular noncovalent interactions among the hydroxy group, central zinc and keto group in the supramolecule.
Aggregated 3D shows a relatively small red-shift of absorption from monomer to aggregates (654
680 nm) due to
the decreased overlap between the chlorin π-planes in the aggregated state. However, 3D easily forms precipitates
composed of structurally ordered large aggregates, indicating that 3D is favorable for molecular packing in the
aggregates.
Introduction
Spontaneous molecular self-assembly to construct elaborate
supramolecular structures is a scientifically interesting phen-
omenon.1 A minimal modulation of building blocks can be
amplified in a drastic change of the supramolecular structure
through noncovalent interactions. Natural systems offer good
examples of sophisticated molecular self-assembly. One of the
best examples can be found in the organization of chloro-
phyllous pigments in extramembranous light-harvesting
antenna complexes of green photosynthetic bacteria. In the
antenna complex called chlorosome, chlorophyllous pigments
self-assemble to function as light-harvesting and energy-
transferring devices without structural supports by any pro-
teins,2,3 while the core of antenna functions of all other photo-
synthetic organisms such as higher plants and algae is served by
chlorophyll(Chl)–protein complexes.4,5
The main light-harvesting pigments in the chlorosome are
bacteriochlorophyll(BChl)s-c, -d and -e (Fig. 1), which are
found exclusively in chlorosomes and are therefore called
chlorosomal Chls. The molecular structures of chlorosomal
Chls strikingly differ from those of protein-binding Chl-a and
-b; BChl-c, -d and -e possess a 1-hydroxyethyl group at the
3-position instead of the vinyl group that Chls-a and -b have. It
has been established that the hydroxy group of a chlorosomal
Chl molecule simultaneously forms a coordinate bond with the
central magnesium of the second molecule and a hydrogen
Fig.
1
Molecular structures of naturally occurring chlorosomal
chlorophylls.
bond with the 13-keto group of the third molecule as in
C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Mg, and thereby chlorosomal Chls can con-
᎐
struct highly ordered supramolecular self-assemblies2,6 which
are stabilized by π–π interaction between the chlorin macro-
cyclic π-conjugates.7 Investigation of molecular–structural
specificity of chlorosomal Chls would not only facilitate
elucidation of the supramolecular structure of chlorosomal
aggregates but also provide insight into the design of building
blocks in any synthetic self-assemblies.
In the self-assemblies, one of the important factors in the
construction of a well-organized supramolecular structure may
lie in characteristics of the linker unit attaching the interactive
† Alternative synthetic approach for 12 and 12D and IR spectra of
the precipitates of 3D are available as supplementary data. For direct
electronic access see http://www.rsc.org/suppdata/p1/b1/b107902f
DOI: 10.1039/b107902f
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144
This journal is © The Royal Society of Chemistry 2001
3135

View Article Online
elucidate the importance of the distance between the interactive
hydroxy group and the chlorin macrocyclic moiety in chloro-
somal Chls.
Results and discussion
Synthesis of the zinc chlorin 2
Synthesis of the 3-(2-hydroxyethyl)chlorin 6 (Scheme 1) has
been reported by Smith et al.12 The synthetic route was as
follows: treatment of methyl pyropheophorbide-a (31,32-dide-
hydrophytochlorin methyl ester 4D in Scheme 1) with thallium
trinitrate trihydrate
dioxide gas acid treatment of the resulting dimethoxy-
acetal reduction of the resulting 3-(formylmethyl)chlorin
removal of chelated thallium by sulfur
Fig. 2 Synthetic zinc chlorins 1–3 and 3D.
sites in a composition molecule with the building block matrix.
This factor can determine the spatial ‘intra’ supramolecular dis-
tance and orientation of building blocks in a supramolecule.
In chlorosomal Chls, the hydroxy group, the most important
interactive site, is simply attached to the chlorin macrocyclic
moiety through one carbon atom (Fig. 1). It is reasonable
to assume that coordination of such a hydroxy group to the
central magnesium of the adjacent molecule could keep the
chlorin molecules very close to each other, accompanied by
overlap of the chlorin π-systems. In this situation, these
molecules must be stabilized by strong π–π interaction, which
leads to formation of large and stable aggregates as well as to
strong excitonic interaction among pigments. This leads to a
large red-shift of Qy absorption maxima (≈ 750 nm), which
is vital for green bacterial survival in view of efficient energy
transfer to the acceptor BChl-a; this has a near-infrared
absorption band in the baseplate that situates in a chlorosomal
membrane and binds to the membranous surface of the cyto-
plasmic side.3 Despite the importance of such a methylene
linkage, no report has been available on the effect of the linker
between the interactive hydroxy group and the chlorin macro-
cyclic moiety.
Here, we address the extension of C3–C31 covalent linkage
between the hydroxy group and the chlorin moiety in chloro-
somal Chls. Zinc chlorin 1 possessing a hydroxymethyl group at
the 3-position is known to be a good model compound for
naturally occurring chlorosomal Chls (Fig. 2).8 Visible absorp-
tion, CD and IR spectroscopic properties of self-aggregates
of 1 formed in nonpolar organic solvents such as 1% tetra-
hydrofuran (THF)–hexane are very similar to those of isolated
chlorosomal Chls aggregates and the in vivo chlorosomal aggre-
gates. Though such spectroscopic similarities generally provide
no direct evidence of structural resemblance between their
supramolecules, it can be clear that the aggregates of 1 are large
and highly ordered as chlorosomal aggregates. Moreover, the
aggregates of 1 proved to transfer the excitation energy to an
acceptor bacteriochlorin molecule as a functional model of
natural chlorosomal antennas.9–11 We chose zinc chlorin 1 as a
natural-type model compound possessing the shortest linker
(3-CH2) between the hydroxy group and the chlorin macro-
cyclic moiety. We prepared zinc chlorins 2 and 3 possessing a
flexible linker (3-CH2CH2 and 3-CH2CH2CH2, respectively),
and 3D‡ possessing a fairly rigid linker (3-CH᎐CH–CH ) by
᎐
2
novel synthetic approaches for functionalization of the 3-vinyl
and 3-formyl groups of chlorophyll derivatives (Fig. 2). Self-
aggregation behavior of zinc chlorins 2, 3 and 3D in nonpolar
organic solvents was examined and compared with that of 1 to
‡ The compound numbering in the present zinc chlorins also indicates
the carbon number in the covalent linker and “D” denotes that the C31–
C32 covalent linkage is a double bond.
Scheme 1 Reagents and conditions: (i) (CH2OH)2–(CH3)3SiCl, dry
CH2Cl2; (ii) BH3–THF, then H2O2–NaOH; (iii) aq. HCl–acetone; (iv)
Zn(OAc)2ؒ2H2O, CH3OH–CH2Cl2.
3136
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144

View Article Online
with NaBH4 to give 6. To convert 4D to 6 more easily, we
attempted direct introduction of a hydroxy group into the 32-
position by anti-Markovnikov hydration of the 3-vinyl group.
The reactive 13-keto group of 4D was protected as its ethylene
ketal according to the reported procedure to give 5D in 90%
yield (step i in Scheme 1).13,14 Chlorin 5D was carefully treated
with BH3–THF and successively treated with H2O2–NaOH
(step ii). After the reaction mixture had been treated with dil.
aq. HCl to eliminate the ketal-protection (step iii), a regio-
isomeric mixture of 3-(2-hydroxyethyl)chlorin 6 and 3-(1-
hydroxyethyl)chlorin 7 (methyl bacteriopheophorbide d) was
obtained in 51% yield from 5D. The ratio of 6 : 7 was 33 : 1 by
1H-NMR, indicating that the hydroboration regioselectively
proceeded at the 3-vinyl group in anti-Markovnikov manner
as expected. Chlorin 4D (18%) was also obtained by ketal-
deprotection of unchanged 5D in the hydroboration (step ii).
Slight reductive cleavage of either C131–O bond of the
dioxolane ring in 5D occurred during step ii to give 131-(2-
hydroxyethoxy)chlorins (bottom drawing in Scheme 1).15
Treatment of 5D with excess of BH3–THF gave a large amount
of the undesired cleavage compounds. Regioisomers 6 and 7
were zinc-metallated and pure zinc chlorin 2 possessing a
2-hydroxyethyl group at the 3-position was isolated by high-
performance liquid chromatography (HPLC). Overall yield
from 4D to 6 in the present synthetic procedures was 45%
(= 0.9 × 0.51 × 33/34 × 100) and 4D was recovered to the extent
of 26% [= (0.1 ϩ 0.9 × 0.18) × 100]. Therefore, yield of 4D
6
based on consumed 4D is estimated to be 61% [= 0.45/(1 Ϫ 0.26)
× 100], which is higher than that reported by Smith et al. (51%
yield). Moreover, our method is advantageous over the previous
procedures from the viewpoint of laboratory safety because
highly toxic thallium compound and sulfur dioxide gas are not
required.
Synthesis of zinc chlorins 3 and 3D
To synthesize 3-(3-hydroxypropyl)chlorin 3 (or 12), we chose
methyl pyropheophorbide-d (3-deethyl-3-formylphytochlorin
methyl ester, 8)8,16 as a starting material (Scheme 2). We first
examined a Wittig reaction of 8 with commercially available
stable formylmethyl ylide W-a (Ph P᎐CHCHO, Chart 1). The
᎐
3
Scheme 2 Reagents and conditions: (i) CH2᎐᎐CHCH2Br, In, aq. THF;
(ii) OsO4, NaIO4 aq. AcOH–THF; (iii) PTSA, benzene, 50 ЊC; (iv)
t-BuNH2ؒBH3, CH2Cl2; (v) H2/PtO2, THF; (vi) Zn(OAc)2ؒ2H2O,
CH3OH–CH2Cl2.
2-hydroxyethylidenetriphenylphosphorane (W-b, Chart 1)17
and 1,3-dioxolan-2-ylmethylenetriphenylphosphorane (W-c,
Chart 1).18 Ylides W-b and W-c were prepared from the corre-
sponding phosphonium bromides by treatment with n-BuLi
and lithium hexamethyldisilazane (LiHMDS) in THF, respec-
tively, according to literature methods. When 8 was treated with
the above two unstable ylides in dry THF, few desired coupling
products were obtained, and 8 was predominantly recovered
with a small amount of uncharacterizable degradation prod-
ucts. Unstable ylides W-b and W-c must have been decomposed
before reacting with the less reactive 3-formyl group of 8, which
is consistent with our previous observation.16 Treatment of 8
with in situ preparation of the above two ylides resulted in a
drastic increase of the degradation products, indicating that 8 is
sensitive to strong bases.
Chart 1
reaction in dichloromethane at room temperature did not
progress and most of the starting material 8 was recovered.
Increasing the reaction temperature (≈50 ЊC) did not promote
the Wittig reaction and a further increase in temperature
(≈70 ЊC) gave rise to degradation of 8. Although other stable
Wittig reagents such as Ph P᎐CHCOOCH and Ph P᎐CHCN
᎐
᎐
3
3
3
smoothly reacted with the 3-formyl group of 8 to provide
the corresponding 3-(2-substituted ethenyl)chlorins,16 highly
stable Ph P᎐CHCHO (W-a) did not react because of the
᎐
3
lower reactivity of the 3-formyl group of 8 than that in
benzaldehyde.
We next examined Wittig reaction of 8 with unstable ylides,
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144
3137

View Article Online
As an alternative method for the preparation of 3, we
examined Grignard reaction of 8 with commercially available
1,3-dioxolan-2-ylmethylmagnesium bromide (G-a, Chart 1).
However, the reaction of 8 with G-a in dry THF provided no
desired Grignard adduct, and undesired C132–C173 cyclized
derivative of 8 was obtained in low yield together with
recovered 8.19
Our attempts at Wittig reaction of 8§ described above led to
the observation that the chlorin 8 is sensitive to strong bases
and high reaction temperature (>70 ЊC) in addition to the low
reactivity of the 3-formyl group in comparison with other
formyl groups conjugated with a benzene-like aromatic ring.
Mild conditions are necessary for the reaction of 8. Indium-
mediated Barbier-type reaction of allyl or propargyl halides
with carbonyl compounds is one of the useful synthetic
methods for functionalization of carbonyl groups under mild
conditions. In the reaction, carbonyl compounds smoothly
react with allyl or propargyl halides in an aqueous organic
solution in the presence of indium at an ambient temperature
to give Barbier/Grignard-type adducts in a considerably high
yield.20 We examined the indium-mediated Barbier type
reaction of 8. When an aq. THF solution of 8 and excess of
allyl bromide (8 eq.) were stirred in the presence of indium
powder (8 eq.) for 2 hours at room temperature, allylated
chlorin 9 was obtained in a surprisingly high yield of over
80% (step i in Scheme 2). Prolonged reaction time (overnight)
gave rise to additional allylation at the 13-keto group. Dia-
stereomeric ratio of the 31-stereochemistry was almost 1 : 1 by
1H-NMR. Subsequently, the vinyl group of 9 at the 32-position
was oxidatively cleaved by OsO4–NaIO4 to give the correspond-
ing aldehyde.10 Although successful oxidation of 9 was
ascertained by the typical CHO signal (δ 10 in CDCl3) in the
1H-NMR spectrum of the crude reaction mixture, isolation of
10 was difficult by flash column chromatography (FCC) due to
its broadened elution with uncharacterizable side-products.
Crude aldehyde 10 was easily dehydrated by mild heating (50
ЊC) in benzene containing a catalytic amount of toluene-p-
sulfonic acid (PTSA) to afford pure chlorin 11D in 59% isolated
yield from 9 after simple FCC purification. As expected,
dehydration of 10 smoothly progressed in comparison with that
of the 3-(1-hydroxyethyl)chlorin 7 because of formation of a
stable α,β-unsaturated aldehyde as the dehydration product.
The resulting 11D was a single stereoisomer with respect to
the eth-3-enyl moiety and showed a coupling constant 3J(31H–
32H) = 16 Hz in its 1H-NMR spectrum, indicating the E-
configuration as drawn in Scheme 2. Chlorin 11D shows a Qy
absorption maximum at 691 nm in CH2Cl2, which is red-shifted
from that of unsubstituted vinylchlorin 4D (667 nm) by the
direct conjugation of the 32-formyl group with the chlorin
chromophore through the eth-3-enyl moiety.16 The red-shift is
larger than those of the corresponding E-32-methoxycarbonyl
(684 nm in CH2Cl2) and E-32-cyano derivatives (689 nm
in CH2Cl2), which are obtained by Wittig reaction of 8 with
Ph P᎐CHCOOCH or Ph P᎐CHCN,16 probably due to the
Fig. 3 Visible absorption (a) and CD spectra (b) of monomeric zinc
chlorin 2 (solid curve), 3 (broken curve) and 3D (dotted curve) in neat
THF. Concentration of each solution is 15 µM.
of allylation is very high. This reaction is useful for the intro-
duction of various allyl-type substituents into a formyl group
directly bound to a chlorin moiety. The indium-mediated
Barbier-type reaction of 8 with various allyl and propargyl
compounds will be reported elsewhere. Moreover, chlorin 11D
is a versatile precursor for preparation of intramolecular
electron/energy-transfer systems because the reactive formyl
group is directly linked to a rigid eth-3-enyl group that could
spatially fix donor and acceptor chromophores.
Optical properties of monomeric zinc chlorins
Visible absorption spectra of monomeric 2 and 3 in neat THF
are shown in Fig. 3a by solid and broken curves, respectively.
Both 2 and 3 gave the same Soret and Qy absorption maxima
at 424 and 644 nm, respectively, which are almost identical
to those of monomeric 1 in neat THF (424 and 646 nm). Both
the full widths at half maxima (FWHMs5) of Qy absorption
bands of 2 and 3, which result from molecular transition along
with Qy direction (N21---N23 axis in left drawing of Fig. 2), are
323 cmϪ1 and also identical to that of 1 (320 cmϪ1). Extension
of covalent linkage between the hydroxy group and the chlorin
moiety does not electronically affect the absorption properties
of monomeric zinc chlorins because there is no direct conjug-
ation of the hydroxy group with the chlorin π-system and no
strong inductive effect of the hydroxy group. Monomeric 3D in
THF has Soret and Qy absorption maxima at 427 and 654 nm
(dotted curve in Fig. 3a), which are almost identical to those
of zinc complex of 4D (427 and 655 nm). Red-shifts of these
bands from those of monomeric 2 and 3 are attributed to direct
conjugation of the C31–C32 double bond with the chlorin π-
system.21 The FWHM of Qy absorption band of 3D is broader
(406 cmϪ1) than those of 2 and 3 and also identical to that of
zinc complex of 4D (419 cmϪ1). The fact that there is no absorp-
tion spectral difference between 3D and the zinc complex of
4D indicates that the 32-hydroxymethyl group does not elec-
tronically affect the absorption properties of eth-3-enylchlorins.
CD spectra of monomeric 2, 3 and 3D in neat THF gave
small positive/negative and negative CD signals in the regions
of the corresponding monomeric Soret and Qy absorption
bands, respectively (Fig. 3b); these are attributable to the small
rotational strength of monomeric chromophores. Similarly to
the visible absorption spectra, CD spectra of monomeric zinc
chlorins 2/3 and 3D were also identical to those of monomeric 1
and the zinc complex of 4D, respectively.
᎐
᎐
3
3
3
stronger electron-withdrawing effect of the formyl group
compared with that of methoxycarbonyl and cyano groups.
Chlorin 11D was easily converted into 12D by reduction with
t-BuNH2ؒBH3 in 92% yield. Hydrogenation of 12D with H2
over PtO2 in THF gave the 3-(3-hydroxypropyl)chlorin 12 in
moderate yield (81%). Finally, zinc-metallation of 12 gave the
desired zinc chlorin 3. In addition, zinc chlorin 3D was pre-
pared by zinc-metallation of 12D.
To our knowledge, this is the first attempt of indium-
mediated allylation of methyl pyropheophorbide-d 8. The yield
Self-aggregation of zinc chlorin 2
§ Alternative synthetic approach via reduction of 3-[2-(methoxy-
carbonyl)ethenyl]chlorin by LiA1H4 obtained by Wittig reaction of
8 with Ph3᎐᎐CHCOOCH3 (ref. 18) gave desired chlorin 12 and 12D
(see supplementary data). However, overall yield from 8 to 12 was only
1.1%.
Visible absorption spectra of zinc chlorins 2 and 1 in 1% THF–
hexane under the same concentration (15 µM) are shown in
Fig. 4. In 1% THF–hexane, zinc chlorin 2 gave new absorption
bands in the longer-wavelength region in addition to the
3138
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144

View Article Online
Fig. 4 Visible absorption spectra of 2 (a) and 1 (b) in 1% THF–
hexane. Concentration of both the solutions is 15 µM.
Fig. 6 CD spectra of 2 in 0.1% THF–hexane (a) and 1 in 1% THF–
hexane (b). Concentration of both the solutions is 15 µM.
Fig. 5 Visible absorption (solid curve) and fluorescence spectra
(excited at 440 nm, dotted curve) of 2 (15 µM) in 0.1% THF–hexane.
residual monomeric bands at 421 and 643 nm (Fig. 4a). The
new absorption bands growing in the nonpolar organic solvents
indicate that 2 self-aggregated to form oligomers in which
composition molecules electronically interacted to absorb
longer-wavelength light in comparison with the monomer
in neat THF. Comparison with the intensity of the residual
monomeric band at around 643 nm in 1% THF–hexane
demonstrates that roughly 20% (calculated from the intensity
of residual monomeric absorption) of 2 remained as the
monomer despite only a slight proportion (< 5%) of mono-
meric 1 being present in 1% THF–hexane (Figs. 4a and 4b). The
Fig. 7 IR spectra of monomeric 2 in THF (upper) and the solid thin
film of 2 (lower) in the hydroxy (left), and carbonyl and chlorin skeletal
C–C/C–N vibrational regions (right).
hexane containing a larger amount of aggregates, such CD
signals were enhanced and the signal shape remained nearly
unchanged (Fig. 6a). This demonstrates that the observed CD
signals were derived from the 701-nm-absorbing aggregates of
2: the negative small peak at around 650 nm (arrow in Fig. 6a)
is ascribable to residual monomeric 2. The CD signals are
composed of a sharp and intense reverse S-shaped signal in the
longer wavelength region and a relatively broad and small
S-shaped signal in the shorter-wavelength region. Such a signal
pattern resembles that of the aggregates of 1 (Fig. 6b). The
observation is ascribed to the strong exciton coupling among Qy
transition moments of composition molecules, which are close
and parallel to each other and aligned to the long axis of aggre-
gates in a highly ordered supramolecular structure.8,19 Both
the observed spectra for the aggregates of 2 and 1 could be
simulated by the sum of two sign-reversed exciton-type com-
ponents. These CD components would originate from the non-
degenerate excitonic transitions of a major oligomeric species.22
Thus, this CD study strongly suggests that the orientation of
the pigments in the aggregates of 2 is not greatly different from
that of the aggregates of 1.
monomer
aggregate equilibrium of 2 in solution lies toward
a more monomeric state than that of 1, indicating a lower
aggregation ability for 2 than for 1.
Visible absorption and fluorescence emission spectra of 2
in 0.1% THF–hexane are shown by solid and dotted curves in
Fig. 5, respectively. Decrease of THF content in hexane (1
0.1%) enhanced the absorption bands of the aggregates of 2
(15 µM in monomer base) in compensation for monomeric
bands (solid curve in Fig. 5). In comparison with the monomer
in THF, the absorption bands of the aggregates were broadened
(FWHM of Qy band is 765 cmϪ1), moved to 442 (Soret) and 701
nm (Qy), and red-shifted by 960 (Soret) and 1260 cmϪ1 (Qy).
These spectral changes are characteristic of the self-aggregates
of chlorosomal chlorophylls. However, the red-shifted values
of the absorption bands of the aggregates of 2 from the
monomeric bands are smaller than those of 1 [1360 (Soret)
and 1940 cmϪ1 (Qy)] in 1% THF–hexane (Fig. 4b), indicating
weaker excitonic interaction in the aggregates of 2 and smaller
aggregation number. The fluorescence spectrum of 2 in 0.1%
THF–hexane showed the emission at 709 nm attributed to the
701-nm-absorbing aggregates of 2 (dotted curve in Fig. 5).
The emission is also less red-shifted from the monomeric Qy
emission (644 nm in THF) than those of the aggregates of 1
Fig. 7 shows the IR spectra of monomeric 2 in THF and the
solid film of 2 in the C᎐O, chlorin skeletal C–C and C–N (right)
᎐
and O–H (left) vibrational regions. Zinc chlorin 2 in the solid
film showed three bands in the range of 1500–1630 cmϪ1, of
which peak positions are identical to those of monomeric 2 in
(646 nm in THF
745 nm in 1% THF–hexane).
In 1% THF–hexane, 2 exhibited specific CD signals in the
THF (Fig. 7). These three bands at 1540, 1553 and 1617 cmϪ1
,
red-shifted Qy band region (data not shown). In 0.1% THF–
arising from the C–C and C–N vibrations of the chlorin
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144 3139

View Article Online
skeleton, show the presence of a 5-coordinated central Zn atom
in comparison with previous assignments for the solid film
aggregates of 1.8,23 Therefore, 2 in the solid film as well as in
THF possesses a central zinc with a single axial O-ligand
(O ؒ ؒ ؒ ZnN4). The band peaking at 1660 cmϪ1 in the solid
film of 2 is attributed to the 13-C᎐O stretching mode, which
᎐
is drastically downshifted (30 cmϪ1) from that in monomer
solution at 1690 cmϪ1, free from any interaction. The large
downshifted value indicates the formation of the C᎐O ؒ ؒ ؒ H–
᎐
O ؒ ؒ ؒ Zn linkage in the film of 2.8 The bands peaking at 1738
cmϪ1 in both IR spectra are attributed to a carbonyl stretching
mode of the 172-COOCH3 group. The lack of shift of these
bands demonstrates that the 172-COOCH3 is free from any
interaction in the film and monomeric state. The O–H stretch-
ing band in the solid film of 2 peaked at 3200 cmϪ1, which is
greatly downshifted by 300 cmϪ1 from the O–H stretching band
in monomer solution at 3500 cmϪ1, normally hydrogen-bonding
with a THF molecule. The large downshift is ascribable to
strong hydrogen bonding of the 32-OH group in the film. The
hydrogen bond is enhanced by the coordination of the hydroxy
oxygen to the central zinc. From the visible absorption and CD
spectral similarity between the solid film and the solution
aggregates, it is plausible that IR spectra of the solid film reflect
the structural characteristics of solution aggregates. Thus, the
Fig. 8 Visible absorption and fluorescence spectra of 3. (a) Time-
dependent visible absorption spectral change of 3 (15 µM) in 0.5%
THF–hexane; solid curve, immediately after preparation of the
solution; dotted curves, 3–9 min later; broken curve, steady state 10 min
later. (b) Visible absorption (solid curve) and fluorescence spectra
(excited at 440 nm, dotted curve) of 3 (15 µM) in 0.1% THF–hexane.
IR study revealed that 2 in aggregates formed the C᎐O ؒ ؒ ؒ H–
O ؒ ؒ ؒ Zn linkage as well as 1 did.
monomeric absorption bands (dotted curves in Fig. 8a), and
the spectral change finished 10 minutes later (broken curve in
Fig. 8a). During the spectral change, the Qy maxima of the
red-shifted bands remained in the range 702–704 nm, indicating
an increase of the same aggregated species. Such a time-
dependent aggregation was not observed in the aggregation of
1 and 2 in THF–hexane admixture in any ratio. Therefore,
formation of the aggregates in 3 is slower than in 1 and 2.
Visible absorption and fluorescence emission spectra of 3
in 0.1% THF–hexane are shown by solid and dotted curves in
Fig. 8b, respectively. In 0.1% THF–hexane, 3 (15 µM) gave
dominant absorption bands of the aggregates having absorp-
tion maxima at 440 (Soret) and 702 nm (Qy), immediately after
preparation of the solution (solid curves Fig. 8b). The Qy
absorption maximum is almost identical to that in 0.5% THF–
hexane. In comparison with the monomer in THF, the absorp-
tion bands of the aggregates were more broadened (FWHM of
Qy band is 730 cmϪ1) and red-shifted by 909 (Soret) and 1280
cmϪ1 (Qy), respectively. The red-shifted values of the absorption
bands of the aggregates of 3 from the monomeric bands are
considerably smaller than those of 1 [1360 (Soret) and 1940
cmϪ1 (Qy)] in 1% THF–hexane (Fig. 4b). This is attributed to a
weaker excitonic interaction and a smaller aggregation number.
However, these red-shifted values of 3 are nearly identical to
those of 2 [960 (Soret) and 1260 cmϪ1 (Qy)], indicating the same
degree of excitonic interaction and aggregation number in each
aggregate. The aggregates of 3 emitted at 707 nm (dotted curve
in Fig. 8b), for which the red-shifted value from the monomeric
emission (644 nm in THF) is smaller than that of the aggregates
᎐
One of the reasons for the lower aggregation ability of 2 than
of 1 is an enhancement of conformational freedom of the
interactive hydroxy group in a molecule, allowed by the more
flexible ethylene linker of 2 than the methylene linker of 1. This
is in good agreement with our previous study; conformationally
flexible interactive sites in a molecule suppress the aggregation
ability of metallochlorins because they can take various
molecular conformations.21 Another reason may lie in the
ethylene linker of 2 being bulkier than the methylene linker of
1. The bulkier linker is disadvantageous for molecular packing
in the self-aggregates, suppressing the formation of tight self-
aggregates. CD and IR studies showed that 2 can form aggre-
gates that possess the same molecular orientation (ordered
orientation) and the same interaction motif among the three
interactive sites (the intermolecular C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Zn
᎐
linkage) as the aggregates of 1, while 2 aggregated with a
weaker excitonic interaction state and in smaller aggregation
number than 1 did, from the visible spectral analyses. The
longer ethylene linker must expand the face-to-face distance
between the neighboring molecules linked by an HO ؒ ؒ ؒ Zn
bond in the aggregates of 2. Obviously, this situation weakens
the π–π interaction between the chlorin π-planes, decreasing
the excitonic interaction and the size of aggregates. Both the
weaker excitonic interaction and the smaller aggregation
number diminish the red-shift of electronic absorption of the
aggregates of 2.
of 1 (646 nm in THF
almost the same as that of the aggregates of 2 (644 nm in THF
745 nm in 1% THF–hexane) but
Self-aggregation of zinc chlorin 3
The visible absorption spectrum of zinc chlorins 3 (15 µM) in
1% THF–hexane gave no new red-shifted absorption bands
(data not shown), and was almost identical to that of the
monomer in THF except for slight blue-shifts probably due to a
decrease in the solvent polarity. In 1% THF–hexane, almost all
3 remained as the monomer, while 20% of 2 and a slight pro-
portion (<5%) of 1 remained as the monomer in 1% THF–
709 nm in 0.1% THF–hexane).
The 0.5% THF–hexane solution of 3 (15 µM) immediately
after preparation gave small CD signals in the red-shifted Soret
and Qy band-regions (solid curve in Fig. 9a). The CD peaks
gradually increased in intensity while retaining the signal shape
(dotted curves in Fig. 9a) to give finally the CD spectrum shown
by the broken curve in Fig. 9a (10 min after preparation). This
result corresponds to the slow aggregation process of 3 shown
by time-dependent visible spectral changes in the same solution
(Fig. 8a). The 0.1% THF–hexane solution of 3 gave CD signals
whose shape was almost identical to those observed in
0.5% THF–hexane (solid curve Fig. 9b). This indicates that the
aggregate species formed in 0.5% and 0.1% THF–hexane are
identical. The CD signal in Qy region is composed of a single
reverse S-shaped signal. The signal shape is quite asymmetric,
hexane (Figs. 4a and 4b). Therefore, the monomer
aggregate
equilibrium of 3 in solution lies in the greatest extent of mono-
meric state among the three zinc chlorins, indicating the lowest
aggregation ability of 3. With a decrease of the THF concentra-
tion in the hexane solution, 3 (15 µM) began to associate and
gave broadened absorption bands in 0.5% THF–hexane (solid
curve in Fig. 8a). After preparation of the solution, these new
bands gradually grew concomitantly with the decrease of the
3140
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144

View Article Online
solid film of 2, a clear shoulder was exhibited in the higher
frequency region of the major 13-C᎐O band (solid arrow in
᎐
Fig. 10). The observation indicates that a small proportion of
the 13-C᎐O groups in the solid aggregates of 3 is in a different
᎐
situation from the C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Zn where a large pro-
᎐
portion of the 13-C᎐O group is involved. Such a chemical
᎐
environment is ascribable to a relatively weak C᎐O ؒ ؒ ؒ H–
᎐
O( ؒ ؒ ؒ Zn) bond as well as normal hydrogen bond (C᎐O ؒ ؒ ؒ
᎐
H–O) and direct coordination to central zinc (C᎐O ؒ ؒ ؒ Zn).
᎐
The carbonyl stretching band of the 172-COOCH3 group in the
solid film of 3 emerged at 1738 cmϪ1, unshifted from the
monomeric band in THF. However, a clear shoulder was
observed in the lower-frequency region of the band (broken
arrow in Fig. 10). This indicates that a small proportion of
172-COOCH3 in the solid film of 3 formed a certain kind of
interaction such as 172-C᎐O ؒ ؒ ؒ H–O( ؒ ؒ ؒ Zn) and/or 172-C᎐
᎐
᎐
O ؒ ؒ ؒ Zn, while most of 172-C᎐O was free from any inter-
᎐
Fig. 9 CD spectra of 3. (a) Time dependent CD spectral change of
3 (15 µM) in 0.5% THF–hexane; solid curve, immediately after
preparation of the solution; dotted curves, at 3–9 min later; broken
curve, steady state at 10 min later. (b) CD spectrum of 3 (15 µM) in
0.1% THF–hexane.
action, as in the case of 2. The O–H stretching band of 3 in the
solid aggregates was broadened in the range 3100–3500 cmϪ1
.
This clearly contrasts with the relatively sharp and largely
downshifted O–H stretching band of 2 (3200 cmϪ1) in the solid
film. The broadening is very likely attributable to the overlap of
several O–H stretching bands originating from heterogeneously
interacting O–H groups in strength and variety. This is
consistent with the presence of the less downshifted 13-C᎐O
᎐
stretching band and the slightly downshifted carbonyl
stretching band of the 172-COOCH3 group. Thus, the IR study
demonstrated that the aggregates of 3 contain not only the
major strong C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Zn linkage but also other
᎐
minor interaction motifs such as relatively weak C᎐O ؒ ؒ ؒ H–
᎐
O ؒ ؒ ؒ Zn linkage, C᎐O ؒ ؒ ؒ Zn, 172-C᎐O ؒ ؒ ؒ H–O( ؒ ؒ ؒ Zn)
᎐
᎐
and 172-C᎐O ؒ ؒ ؒ Zn as intermolecular noncovalent linkages.
᎐
The low aggregation ability of 3 is ascribable to much
enhancement of conformational freedom of the interactive
hydroxy group in a molecule and the bulkiness of the linker
trimethylene moiety in 3. Reflecting the long linker, the face-to-
face distance between the composition molecules in the aggre-
gates of 3 was expanded more than that in the aggregates of 1,
as shown by the red-shifted values of the aggregates. Neverthe-
less with the longer trimethylene linker of 3 than the ethylene
linker of 2, both the aggregates of 3 and 2 showed almost
identical red-shifted values, indicating that the face-to-face
distance between the neighboring molecules linked by an
HO ؒ ؒ ؒ Zn bond in the aggregates of 3 is comparable to those
in the aggregates of 2. This finding demonstrates that the tri-
methylene moiety of 3 and the ethylene moiety of 2 in the
aggregates maintain the same distance between the 3-position
and the ω-hydroxy group. Obviously, conformational strain
of the trimethylene moiety in aggregated 3 is larger than that of
the ethylene moiety in aggregated 2. Therefore, conformational
restriction of the linker moiety through coordination of a
ω-hydroxy group to the central zinc of a neighboring molecule
by aggregation of 3 is larger than that by aggregation of 2. This
leads to substantial dissociation of the HO ؒ ؒ ؒ Zn bond in the
aggregation of 3, which may further suppress the aggregation
of 3 and also may make the aggregation of 3 slower than that of
1 and 2 as shown by the time-dependent visible and CD spectral
change in 0.5% THF–hexane. Moreover, the ω-hydroxy group
of 3, attached to the conformationally strained trimethylene
linker in the aggregated state, must more or less diverge from
the Qy molecular axis. This situation may disturb the orienta-
tion of composition molecules and prevent the formation of the
Fig. 10 IR spectra of solid thin film of aggregated 3 in the hydroxy
(left), and carbonyl and chlorin skeletal C–C/C–N vibrational band
regions (right).
where the negative signal peaking at 710 nm is approximately
three-fold larger than the positive signal peaking at 688 nm.
Such a CD signal in the Qy region is clearly distinguishable from
those of 1 and 2 (Fig. 6). Furthermore, it is worth pointing out
that the aggregates of 3 gave intense reverse S-shaped CD signal
in the Soret region in comparison with those of 1 and 2; the
ellipticities of CD peaks in aggregated solutions are 14/Ϫ8 in 1
(Fig. 6b), Ϫ4/12 in 2 (Fig. 6a) and 27/Ϫ33 mdeg in 3 (Fig. 9b).
Thus, CD spectra of the aggregates of 3 are remarkably dif-
ferent from those of 1 and 2. This demonstrates that the type
of excitonic coupling among the pigments in the aggregates of 3
is different from those in the aggregates of 1 and 2, indicating
that orientation of the pigments in the aggregates of 3 is also
different; the aggregates of 3 may be less ordered in the
supramolecular structure than are those of 1 and 2. The dif-
ference between the CD spectra of the aggregates of 2 and 3
whose visible spectra are almost identical to each other does not
conflict with the higher sensitivity of CD spectra for small
structural changes of chlorosomal Chl supramolecules in com-
parison with absorption spectra.24,25
Fig. 10 shows IR spectra of the solid thin film of aggregated
3 whose supramolecular structure had been confirmed to be
identical to that in 0.1% THF–hexane by visible absorption and
CD spectral similarity between the two aggregated states (data
not shown). The solid film of 3 showed three bands at 1540,
1553 and 1617 cmϪ1, arising from the C–C and C–N vibrations
of the chlorin skeleton. These peak positions are identical to
those of monomeric 3 in THF and solid aggregates of 1 and 2,
indicating the presence of a 5-coordinated central Zn atom.
Therefore, the central zinc in the solid film of 3 was also
axially ligated by a single O-ligand (O ؒ ؒ ؒ ZnN4). Although the
homogeneously strong intermolecular C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Zn
᎐
linkage as shown by CD and IR spectral analyses.
Self-aggregation of zinc chlorin 3D
In 1% THF–hexane, 3D (15 µM) gave a Qy absorption maxi-
mum at 680 nm in addition to a residual monomeric band at
652 nm, where the two bands overlapped each other con-
13-C᎐O stretching band in the solid film of 3 dominantly
᎐
emerged at 1660 cmϪ1, which is identical to that observed in the
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144
3141

View Article Online
The smaller red-shift of the Qy absorption band in the aggre-
gates of 3D than those of the aggregates of 2 and 3 as well as 1
despite their large size would be due to the significantly weaker
excitonic interaction between the composition molecules in the
aggregates. As expected, the planar and rigid 3-CH᎐CH- linker
᎐
does not separate the two neighboring molecules linked by an
HO ؒ ؒ ؒ Zn bond toward the vertical direction with respect to
the chlorin π-plane but does separate them toward the hori-
zontal direction compared with those of 1. This situation
decreases the overlapping region between the chlorin π-systems
in the aggregates of 3D, weakening the excitonic interaction in
the supramolecule.
Fig. 11 Visible absorption spectrum of 3D (15 µM) in 1% THF–
hexane.
siderably (Fig. 11). Spectral deconvolution revealed that the
spectrum in the Qy region was the sum of a single Gaussian
component peaking at 680 nm and the monomeric Qy absorp-
tion band peaking at 652 nm observed in 5% THF–hexane
(data not shown). The FWHM of the 680 nm absorption band
was 650 cmϪ1, 1.6-fold broader than that of the monomer
(400 cmϪ1). The band broadening is characteristic of self-
aggregation, showing the 680-nm-absorbing species to be the
self-aggregates of 3D. The 680 nm absorption band is red-
shifted by 580 cmϪ1 from the monomeric Qy band in THF
(654 nm). The significantly smaller red-shift of the Qy absorp-
The importance of a methylene linker between the interactive
hydroxy group and the chlorin moiety for the formation of
antenna aggregates
The flexible, long linker expanded the face-to-face distance
between the composition molecules in the aggregates of 2 and
3 compared with those in the aggregates of 1. Moreover,
composition molecules in the aggregates of 3 were prevented
from associating regularly and from forming a homogeneously
strong intermolecular C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Zn linkage. The rigid,
᎐
long linker of 3D decreases the overlapping region between the
chlorin π-planes in the aggregated state. These aggregations of
2, 3 and 3D described here are disadvantageous for the light-
harvesting antenna pigments in chlorosomes for the following
three reasons. (i) 2, 3 and 3D self-aggregated to form relatively
unstable oligomers. (ii) Unfavorable intermolecular interactions
in the self-aggregates may result in structural defects in the
aggregates like a lattice defect in crystals. Such defects are dis-
advantageous for efficient energy migration in the aggregates.
(iii) The excitonic interaction among pigments in the aggregates
of 2, 3 and 3D was weakened and thereby these Qy fluorescence
bands appeared in a shorter-wavelength region. This is
energetically disadvantageous for efficient energy transfer to
BChl-a in a baseplate.26 It is concluded that the methylene
linkage between the interactive hydroxy group and the chlorin
moiety in the naturally occurring chlorosomal Chls as well as
the model pigments such as 1 is necessary for the formation
of the antenna aggregates appropriate for the efficient light-
harvesting, energy-migration and energy-transfer.
tion band of the aggregates of 3D than those of 1 (1940 cmϪ1
,
Fig. 4b), 2 (1260 cmϪ1, Fig. 4a) and 3 (1280 cmϪ1, Fig. 8b)
indicates the considerably smaller aggregation number and/or
weaker excitonic interaction of the aggregates of 3D. The
absorption intensities of the 680-nm-absorbing aggregates and
the residual monomer are 0.14 and 0.12, respectively, and the
quotient of absorption intensity of the aggregates over the con-
sumed monomer is 0.27. This value is much smaller than those
of the aggregation of 1 (0.7), 2 (0.68) and 3 (0.63). This is
probably due to inhomogeneity of the solution by the for-
mation of microscopic precipitates, because the aggregated
solution of 3D gave macroscopic precipitates from the pre-
paration after storage for only a few minutes. This prevented
us from performing any spectral analyses other than rapid
measurements of the visible spectrum in solution. In 0.1%
THF–hexane, 3D immediately formed precipitates just after
preparation. Such a precipitation could not be observed in 0.1%
THF–hexane solution of 2 and 3, even after storage for several
hours. 3D possesses the planar 3-CH᎐CH-linker, which is more
᎐
favorable for the molecular packing in the supramolecule than
are 2 and 3 involving a fully saturated alkyl linker.
Experimental
Instrumentation
During formation of the macroscopic precipitates in the 1%
THF–hexane solution of 3D, no red-shift was observed in the
680 nm absorption band. Moreover, supernatant hexane solu-
tion after several minutes was completely depleted of the 680-
nm-absorbing aggregates. Shaking the precipitated solution
disrupted the macroscopic precipitates and the absorption
spectrum again showed a 680 nm absorption band. These
findings indicate that the 680 nm-absorbing aggregates were
soluble, large aggregates and a building block of macroscopic
precipitates.
Visible absorption and CD spectra were measured in air-
saturated solvents at room temperature on a Hitachi U-3500
spectrophotometer and a JASCO J-720W spectropolarimeter,
respectively. 1H-NMR measurements were performed on a
Bruker AC-300 MHz NMR spectrometer. Fast-atomic
bombardment (FAB)–mass spectra (MS) were recorded on
a JEOL HX-100 spectrophotometer; FAB-MS samples were
dissolved in dichloromethane, and m-nitrobenzyl alcohol was
used as the matrix. HPLC was carried out with a Shimadzu
LC-10AS pump, and an SPD-M10AV visible detector. Fourier
transform (FT)-IR spectra were recorded at room temperature
on a Shimadzu FTIR-8600 spectrophotometer; a solution was
measured in a KBr cell and an aggregate film on a thin-
aluminium-film-coated glass by means of reflection absorption
spectroscopy through a Shimadzu AIM-8000R microscope.
The IR spectrum of the precipitates of 3D was measured (see
supplementary data†). The O–H, 172-C᎐O, 13-C᎐O and the
᎐
᎐
three chlorin-skeletal C–C and C–N vibrational bands emerged
at 3180, 1738, 1653, 1614, 1552 and 1539 cmϪ1, respectively.
These band positions are characteristic of the formation of the
C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Zn linkage as explained in the IR study of 2
᎐
(Fig. 7). Moreover, no shoulder or band broadening in the
13-C᎐O, 172-C᎐O and O–H stretching bands was observed,
᎐
᎐
Materials
while they were observed in the IR spectrum of aggregated 3
(Fig. 10). These results demonstrate that 3D forms a homo-
THF and hexane for visible, CD and IR spectra were freshly
distilled over CaH2 before use. BH3 (THF solution, 1.00 M)
and 1,3-dioxolan-2-ylmethylmagnesium bromide (G-a, hexane
solution, 0.95 M) were purchased from Aldrich. Formyl-
methylenetriphenylphosphorane (W-a) and (2-hydroxyethyl-
(triphenyl)phosphonium bromide for the preparation of ylide
geneous, strong C᎐O ؒ ؒ ؒ H–O ؒ ؒ ؒ Zn linkage in the precipi-
᎐
tates. Therefore, the precipitates and their building blocks, the
680-nm-absorbing aggregates of 3D, are structurally ordered,
large aggregates, similar to the aggregate structure of 1 reported
in our previous study.8
3142
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144

View Article Online
W-b were purchased from Wako Pure Chemical Industries. 1,3-
Dioxolan-2-ylmethyl(triphenyl)phosphonium bromide for the
preparation of ylide W-c was purchased from Tokyo Chemical
Industry. FCC was carried out on silica gel (Merck Kiesel gel
60, 9358). HPLC was performed with packed ODS columns
(Cosmosil 5C18AR, Nacalai Tesque, 6.0 × 250 mm). All
synthetic zinc chlorins and their intermediates were charac-
terized by their visible, 1H-NMR and FAB-MS spectra. Ketal-
protection and -deprotection of 13-keto group were done
according to the procedures reported by Tamiaki et al.14 which
are slight modifications of the original by Pandey et al.13 Oxid-
ative cleavage of the vinyl group,21 reduction of the formyl
group21 and zinc-metallation27 were done according to reported
procedures. Methyl pyropheophorbide-a8 4D, ketal-protected
methyl pyropheophorbide-a 5D,13,14 and methyl pyro-
pheophorbide-d8 8 were prepared from modification of Chl-a
according to reported procedures. All synthetic procedures
were done in the dark.
bromide (300 mg) and indium powder (90 mg). After stirring
of the mixture for 1.5 h at room temperature under nitrogen
atmosphere, indium powder was filtered off and washed with
CH2Cl2. The organic layer was washed with water three times,
and dried over Na2SO4. The solvents were evaporated off
and the residue was purified by FCC (1% CH3OH–CH2Cl2)
and then recrystallized from CH2Cl2–hexane to give 3-(1-
hydroxybut-3-enyl)chlorin 9 (57 mg) in 82% yield. The 31-
stereoconfiguration is a 1 : 1 mixture of R and S. For com-
pound 9: λmax/nm (CH2Cl2) 661 (rel. 45%), 605 (7.4), 537 (8.7),
504 (9.5), 410 (100), 398 sh (87); 1H-NMR (CDCl3) δ 9.54/9.58,
9.24/9.25, 8.44/8.46 (each 1H, s, 5-, 10-, 20-H), 6.09/6.12 (1H, t,
J = 8 Hz, 3-CH), 5.90–6.05 (1H, m, 32-CH), 5.25 (1H, d, J = 17
Hz, E-33-CH), 5.15 (1H, d, J = 10 Hz, Z-33-CH), 5.07/5.13,
4.95/4.99 (each 1H, d, J = 20 Hz, 131-CH2), 4.38 (1H, q, J = 8
Hz, 18-H), 4.08–4.20 (1H, m, 17-H), 3.64/3.65, 3.51, 3.34/3.35,
3.19/3.20 (each 3H, s, 2-, 7-, 12-CH3, COOCH3), 3.60 (2H, q,
J = 8 Hz, 8-CH2), 3.05–3.20 (2H, m, 31-CH2), 2.88/2.96 (1H, s,
OH), 2.45–2.67, 2.03–2.37 (each 2H, m, 17-CH2CH2), 1.72/1.74
(3H, d, J = 8 Hz, 18-CH3), 1.64 (3H, t, J = 8 Hz, 81-CH3), Ϫ0.01/
0.06, Ϫ2.04/Ϫ1.98 (each 1H, s, NH); MS (FAB) (Found: Mϩ,
592. Calc. for C36H40N4O4: M, 592).
Synthesis of zinc methyl 3-devinyl-3-(2-hydroxyethyl)pyropheo-
phorbide-a [32-hydroxyphytochlorinatozinc(II) methyl ester 2]
To a dry THF (15 ml) solution of the chlorin 5D (40 mg)
was added dropwise BH3–THF (0.03 M; 30 ml) under argon
atmosphere at room temperature with vigorous stirring. After
stirring of the mixture for 3 h, ethanol (2 ml) and aq. NaOH
(3.0 M; 2 ml) were added at Ϫ10 ЊC and then ice-chilled 30% aq,
H2O2 (2 ml) was added at 0 ЊC. After stirring for 1.5 h, the
solution was poured into brine and extracted with several
portions of CHCl3 and the combined organic phases were
washed with brine and dried over Na2SO4. After evaporation of
the solvents, the residue was dissolved in acetone (27 ml) and
then conc. HCl (0.2 ml) was added with stirring. The reaction
mixture was stirred for 10 min, poured into water, and extracted
by CH2Cl2, and the extract was washed successively with 4% aq.
NaHCO3 and water, and dried over Na2SO4. The solvents were
evaporated off and the residue was purified by FCC to give 4D
(6.7 mg, 18% yield, 5% EtO2–CH2Cl2) and a 33 : 1 mixture of
3-(2-hydroxyethyl)chlorin 6 and 3-(1-hydroxyethyl)chlorin 7
(19.5 mg, 51% yield, 1% CH3OH–CH2Cl2); λmax/nm (CH2Cl2) =
658 (relative intensity, 39%), 602 (7.0), 536 (7.8), 503 (9.9), 408
The above vinylchlorin 9 (60 mg) was oxidized21 by OsO4
(ca. 5 mg), NaIO4 (140 mg) and AcOH (180 µl) in water
(1.1 ml)–THF (10 ml) and roughly purified by FCC (0.5–2.5%
CH3OH–CH2Cl2) to give a crude diastereomeric mixture of
the 3-(1-hydroxy-2-formylethyl)chlorin 10 [1H-NMR (CDCl3)
δ = 10.03 (1H, s, CHO)].
To a benzene solution (25 ml) of the entire crude sample was
added PTSA (20 mg) with stirring. After stirring for 30 min at
50 ЊC under nitrogen atmosphere, the solution was poured into
ice–water, the aqueous layer was extracted with several portions
of CH2Cl2 and combined organic fractions were washed
successively with 4% aq. NaHCO3 and water, then dried over
Na2SO4. The solvents were evaporated off and the residue
was purified by FCC (5–10% Et2O–CH2Cl2) and recrystallized
from CH2Cl2–hexane to give the 3-(2-formylvinyl)chlorin 11D
(34 mg) in 59% yield from 9 in the above two steps: λmax/nm
(CH2Cl2) 691 (rel. 65%), 629 (8.9), 552 (13), 520 (14), 420 (100);
1H-NMR (CDCl3) δ 10.14 (1H, d, J = 7 Hz, CHO), 9.59, 9.46,
8.75 (each 1H, s, 5-, 10-, 20-H), 8.89 (1H, d, J = 16 Hz, 3-CH),
7.36 (1H, dd, J = 7, 16 Hz, 31-CH), 5.32, 5.17 (each 1H, d,
J = 20 Hz, 131-CH2), 4.55 (1H, dq, J = 2, 7 Hz, 18-H), 4.36 (1H,
d, J = 8 Hz, 17-H), 3.72 (2H, q, J = 8 Hz, 8-CH2), 3.70, 3.62,
3.54, 3.28 (each 3H, s, 2-, 7-, 12-CH3, COOCH3), 2.50–2.80,
2.20–2.40 (each 2H, m, 17-CH2CH2), 1.84 (3H, d, J = 7 Hz,
18-CH3), 1.72 (3H, t, J = 8 Hz, 81-CH3), 0.02, Ϫ1.94 (each 1H,
s, NH); MS (FAB) (Found: Mϩ, 576. Calc. for C35H36N4O4: M,
576).
1
(100), 396 sh (80); H-NMR (CDCl3) δ (major 6) 9.48, 9.26,
8.50 (each 1H, s, 5-, 10-, 20-H), 5.23, 5.07 (each 1H, d, J = 20
Hz, 131-CH2), 4.46 (1H, q, J = 7 Hz, 18-H), 4.37 (2H, t, J = 7
Hz, 31-CH2), 4.27 (1H, d, J = 8 Hz, 17-H), 4.12 (2H, t, J = 7 Hz,
3-CH2), 3.70 (2H, q, J = 8 Hz, 8-CH2), 3.64, 3.60, 3.26, 3.25
(each 3H, s, 2-, 7-, 12-CH3, COOCH3), 2.45–2.73, 2.14–2.36
(each 2H, m, 17-CH2CH2), 1.80 (3H, d, J = 7 Hz, 18-CH3), 1.69
(3H, t, J = 8 Hz, 81-CH3), 0.45, Ϫ1.68 (each 1H, s, NH); MS
(FAB) (Found: Mϩ 566, Calc. for C34H38N4O4: M, 566).
The above mixture of metal-free chlorins 6 and 7 was zinc-
metallated27 and purified by HPLC (Cosmosil, CH3OH–water
8 : 1; 1.0 ml minϪ1) to give the title compound 2 (tR 6.6 min) and
the zinc complex of 7 (tR7.7 min). The separation ratio was 1.6.
2: λmax/nm (THF) 644 (rel. 73%), 599 (10), 564 (5.5), 520 (3.6),
424 (100), 403 (57); 1H-NMR (10% v/v CD3OD–CDCl3) δ 9.35,
8.95, 8.16 (each 1H, s, 5-, 10-, 20-H), 5.10, 4.92 (each 1H, d,
J = 20 Hz, 131-CH2), 4.29 (1H, q, J = 7 Hz, 18-H), 4.12 (2H, t,
31-CH2), 4.05–4.20 (3H, m, 3-CH2 ϩ 17-H), 3.61 (2H, q, J =
8 Hz, 8-CH2), 3.50, 3.41, 3.12, 3.10 (each 3H, s, 2-, 7-, 12-CH3,
COOCH3), 2.36–2.60, 2.07–2.34 (each 2H, m, 17-CH2CH2),
1.66 (3H, d, J = 7 Hz, 18-CH3), 1.58 (3H, t, J = 8 Hz, 81-CH3);
MS (FAB) (Found: Mϩ, 628. Calc. for C34H36N4O464Zn: M,
628).
The above formylchlorin 11D (10 mg) was reduced21 by t-
BuNH2ؒBH3 (20 mg) in dry CH2Cl2 (10 ml). Purification by
FCC (0.75–1% CH3OH–CH2Cl2) and recrystallization from
CH2Cl2–hexane gave the corresponding 3-(3-hydroxyprop-1-
enyl)chlorin 12D (9.2 mg) in 92% yield; λmax/nm (CH2Cl2) 667
(rel. 41%), 612 (7.4), 539 (8.7), 509 (10), 414 (100), 402 sh (84);
1H-NMR (CDCl3) δ 9.45, 9.29, 8.53 (each 1H, s, 5-, 10-, 20-H),
7.86 (1H, d, J = 16 Hz, 3-CH), 6.88 (1H, dt, J = 5, 16 Hz,
31-CH), 5.26, 5.10 (each 1H, d, J = 20 Hz, 131-CH2), 4.75 (2H,
d, J = 5 Hz, 32-CH2), 4.47 (1H, dq, J = 2, 7 Hz, 18-H), 4.28 (1H,
d, J = 8 Hz, 17-H), 3.64 (2H, q, J = 8 Hz, 8-CH2), 3.65, 3.62,
3.38, 3.20 (each 3H, s, 2-, 7-, 12-CH3, COOCH3), 2.50–2.77,
2.20–2.40 (each 2H, m, 17-CH2CH2), 1.80 (3H, d, J = 7 Hz,
18-CH3), 1.67 (3H, t, J = 8 Hz, 81-CH3), 0.39, Ϫ1.75 (each 1H, s,
NH); MS (FAB) (Found: Mϩ, 578. Calc. for C35H38N4O4: M,
578).
Synthesis of zinc 3-devinyl-3-(3-hydroxypropyl)pyropheo-
phorbide-a [32-(hydroxymethyl)phytochlorinatozinc(II)
methyl ester 3]
To a dry THF solution (10 ml) of the above chlorin 12D
(7 mg) was added PtO2 (10 mg). After stirring of the mixture
for 1.5 h under H2 atmosphere, the PtO2 was filtered off and
the filtrate was washed with water three times and dried over
Na2SO4. The solvents were evaporated off and the residue was
To a stirred aq. THF solution (15 ml; THF–water 2 : 1) of
methyl pyropheophorbide-d 8, (65 mg) were added allyl
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144
3143

View Article Online
purified by FCC (1% CH3OH–CH2Cl2) and recrystallized from
CH2Cl2–hexane to give the 3-(3-hydroxypropyl)chlorin 12
(5.7 mg) in 81% yield; λmax/nm (CH2Cl2) 657 (rel. 43%), 602
sions. S. Y. thanks the Japan Science Society for a Sasakawa
Scientific Research Grant.
1
(7.7), 535 (8.6), 504 (9.0), 471 (3.6), 410 (100), 396 (77); H-
NMR (CDCl3) δ 9.46, 9.27, 8.47 (each 1H, s, 5-, 10-, 20-H),
5.24, 5.09 (each 1H, d, J = 20 Hz, 131-CH2), 4.45 (1H, dq, J = 2,
7 Hz, 18-H), 4.27 (1H, d, J = 8 Hz, 17-H), 3.85–3.98 (4H, m,
3-CH2 ϩ 32-CH2), 3.68 (2H, q, J = 8 Hz, 8-CH2), 3.65, 3.61,
3.30, 3.24 (each 3H, s, 2-, 7-, 12-CH3, COOCH3), 2.20–2.75
(6H, m, 31-CH2 ϩ 17-CH2CH2), 1.80 (3H, d, J = 7 Hz, 18-CH2),
1.69 (3H, t, J = 8 Hz, 81-CH3), 0.57, Ϫ1.63 (each 1H, s,
NH); MS (FAB) (Found: m/z 581. Calc. for C35H41N4O4: MHϩ,
581).
References
1 J.-M. Lehn, Supramolecular Chemistry. Concepts and Perspectives,
VCH, Weinheim, 1995.
2 H. Tamiaki, Coord. Chem. Rev., 1996, 148, 183.
3 J. M. Olson, Photochem. Photobiol., 1998, 67, 61.
4 G. McDermott, S. M. Prince, A. A. Freer, A. M. Hawthornthwaite-
Lawless, M. Z. Papiz, R. J. Cogdell and N. W. Isaacs, Nature
(London), 1995, 374, 517.
5 P. Jordan, P. Fromme, H. T. Witt, O. Klukas, W. Saenger and
N. Krauβ, Nature (London), 2001, 411, 909.
6 P. Hildebrandt, H. Tamiaki, A. R. Holzwarth and K. Schaffner,
J. Phys. Chem., 1994, 98, 2192.
7 A. R. Holzwarth and K. Schaffner, Photosynth. Res., 1994, 41, 225.
8 H. Tamiaki, M. Amakawa, Y. Shimono, R. Tanikaga, A. R.
Holzwarth and K. Schaffner, Photochem. Photobiol., 1996, 63, 92.
9 H. Tamiaki, T. Miyatake, R. Tanikaga, A. R. Holzwarth and
K. Schaffner, Angew. Chem., Int. Ed. Engl., 1996, 35, 772.
10 T. Miyatake, H. Tamiaki, A. R. Holzwarth and K. Schaffner,
Photochem. Photobiol., 1999, 69, 448.
11 T. Miyatake, H. Tamiaki, A. R. Holzwarth and K. Schaffner, Helv.
Chim. Acta, 1999, 82, 797.
12 K. M. Smith, D. A. Goff and D. J. Simpson, J. Am. Chem. Soc.,
1985, 107, 4946.
The above metal-free chlorin 12 was zinc-metallated27, puri-
fied by FCC (1% CH3OH–CH2Cl2) and recrystallized from
CH2Cl2–hexane to give the title compound 3; λmax/nm (THF)
644 (rel. 77%), 599 (10), 564 (5.5), 519 (3.2), 423 (100), 403 (59);
1H-NMR (10 v/v % CD3OD–CDCl3) δ 9.44, 9.06, 8.23 (each
1H, s, 5-, 10-, 20-H), 5.11, 4.97 (each 1H, d, J = 20 Hz, 131-
CH2), 4.37 (1H, dq, J = 2, 7 Hz, 18-H), 4.19 (1H, d, J = 8 Hz,
17-H), 3.60–3.90 (6H, m, 3-CH2 ϩ 32-CH2 ϩ 8-CH2), 3.56, 3.42,
3.21, 3.16 (each 3H, s, 2-, 7-, 12-CH3, COOCH3), 2.20–2.77
(6H, m, 31-CH2 ϩ 17-CH2CH2), 1.75 (3H, d, J = 7 Hz, 18-CH3),
1.67 (3H, t, J = 8 Hz, 81-CH3); MS (FAB) (Found: Mϩ, 642.
Calc. for C35H38N4O464Zn: M, 642).
13 R. K. Pandey, N. Jagerovic, J. M. Ryan, T. J. Dougherty and
K. M. Smith, Tetrahedron, 1996, 52, 5349.
14 H. Tamiaki, T. Watanabe and T. Miyatake, J. Porphyrins
Phthalocyanines, 1999, 3, 45.
15 R. J. Abraham, A. E. Rowan, N. W. Smith and K. M. Smith,
J. Chem. Soc., Perkin Trans. 2, 1993, 1047.
Synthesis of zinc methyl 32-(hydroxymethyl)pyropheophorbide-a
[31,32-didehydro-32-(hydroxymethyl)phytochlorinatozinc(II)
methyl ester 3D]
Metal-free chlorin 12D was zinc-metallated,27 purified by FCC
(1% CH3OH–CH2Cl2) and recrystallized from CH2Cl2–hexane
to give 3; λmax/nm (THF) 654 (rel. 66%), 608 (11), 568 (5.6), 522
(3.8), 427 (100), 407 (78); 1H-NMR (10 v/v % CD3OD–CDCl3)
δ 9.49, 9.22, 8.34 (each 1H, s, 5-, 10-, 20-H), 7.84 (1H, d, J =
16 Hz, 3-CH), 6.76 (1H, dt, J = 5, 16 Hz, 31-CH), 5.15, 5.02
(each 1H, d, J = 20 Hz, 131-CH2), 4.67 (2H, d, J = 5 Hz,
32-CH2), 4.41 (1H, dq, J = 2, 7 Hz, 18-H), 4.22 (1H, d, J = 7
Hz, 17-H), 3.72 (2H, q, J = 8 Hz, 8-CH2), 3.60, 3.40, 3.30, 3.23
(each 3H, s, 2-, 7-, 12-CH3, COOCH3), 2.20–2.65 (4H, m,
17-CH2CH2), 1.76 (3H, d, J = 7 Hz, 18-CH3), 1.69 (3H, t,
J = 8 Hz, 81-CH3); MS (FAB) (Found: Mϩ, 640. Calc. for
C35H36N4O464Zn: M, 640).
16 H. Tamiaki and M. Kouraba, Tetrahedron, 1997, 53, 10677.
17 B. E. Maryanoff, A. B. Reitz and B. A. Duhl-Emswiler, J. Am. Chem.
Soc., 1985, 107, 217.
18 B. Fraser-Reid, B. F. Molino, L. Magdzinski and D. R. Mootoo,
J. Org. Chem., 1987, 52, 4505.
19 S. Yagai, T. Miyatake, Y. Shimono and H. Tamiaki, Photochem.
Photobiol., 2001, 73, 153.
20 C.-J. Li, Tetrahedron, 1996, 52, 5643.
21 S. Yagai, T. Miyatake and H. Tamiaki, J. Photochem. Photobiol. B:
Biol., 1999, 52, 74.
22 D. R. Buck and W. S. Struve, Photosynth. Res., 1996, 48, 367.
23 H. Tamiaki, M. Amakawa, A. R. Holzwarth and K. Schaffner,
Photosynth. Res., in press.
24 R. G. Alden, S. H. Lin and R. E. Blankenship, J. Lumin, 1992, 51,
51.
25 H. Tamiaki, S. Miyata, Y. Kureishi and R. Tanikaga, Tetrahedron,
1996, 52, 12421.
26 Y. Saga, K. Matsuura and H. Tamiaki, Photochem. Photobiol., 2001,
74, 72.
27 H. Tamiaki, S. Yagai and T. Miyatake, Bioorg. Med. Chem., 1998, 6,
2171.
Acknowledgements
We thank Drs Tadashi Mizoguchi, Yoshitaka Saga and Luke
Ueda-Sarson of Ritsumeikan University for helpful discus-
3144
J. Chem. Soc., Perkin Trans. 1, 2001, 3135–3144