Induction of wavelength-dependent photochemistry in bilirubins by serum albumin

Article abstract of DOI:10.1007/PL00013473

Quantum yields for Z ? E photoisomerization of the natural bi-chromophore bilirubin IXα in ammoniacal methanol solution were found to vary with excitation wavelength, whereas no variation (within experimental error) was observed for xanthobilirubic acid and for symmetrically substituted bilirubins (bilirubin IIIα, bilirubin XIIIα, and mesobilirubin XIIα) in the same solvent. The quantum yield for Z ? E photoisomerization of xanthobilirubic acid bound to human serum albumin was also invariant with excitation wavelength. In contrast, quantum yields for Z ? E photoisomerization of the symmetrically-substituted bilirubins did show marked excitation wavelength dependency when they were bound to human serum albumin. These results show that quantum yields for Z ? E photoisomerization of bilirubins are markedly sensitive to protein binding and to the nature of pyrrole ring β-substituents. They also demonstrate that wavelength-dependent photochemistry is characteristic of bilirubins with non-identical pyrromethenone chromophores, as expected from exciton coupling theory, and that complexation with albumin induces wavelength-dependent photochemistry in symmetrically substituted bilirubins.

Full text of DOI:10.1007/PL00013473

Monatshefte fuÈr Chemie 129, 649±660 (1998)
Induction of Wavelength-Dependent
Photochemistry in Bilirubins
by Serum Albumin
Antony F. McDonagh1;Ã, Giovanni Agati2, and David A. Lightner3
1
Division of Gastroenterology, Box 0538, S-357, University of California, San Francisco,
CA 94143-0538, USA
2
Istituto di Elettronica Quantistica, CNR, Sezione INFM di Firenze, I-50127 Firenze, Italy
3
Department of Chemistry, University of Nevada, Reno, NV 89557-0020, USA
Summary. Quantum yields for Z > E photoisomerization of the natural bi-chromophore bilirubin
IXꢀ in ammoniacal methanol solution were found to vary with excitation wavelength, whereas no
variation (within experimental error) was observed for xanthobilirubic acid and for symmetrically
substituted bilirubins (bilirubin IIIꢀ, bilirubin XIIIꢀ, and mesobilirubin XIIIꢀ) in the same solvent.
The quantum yield for Z > E photoisomerization of xanthobilirubic acid bound to human serum
albumin was also invariant with excitation wavelength. In contrast, quantum yields for Z > E
photoisomerization of the symmetrically-substituted bilirubins did show marked excitation
wavelength dependency when they were bound to human serum albumin. These results show that
quantum yields for Z > E photoisomerization of bilirubins are markedly sensitive to protein binding
and to the nature of pyrrole ring ꢁ-substituents. They also demonstrate that wavelength-dependent
photochemistry is characteristic of bilirubins with non-identical pyrromethenone chromophores, as
expected from exciton coupling theory, and that complexation with albumin induces wavelength-
dependent photochemistry in symmetrically substituted bilirubins.
Keywords. Bile pigments; Exciton coupling; Human serum albumin; Phototherapy; Pyrromethe-
none.
È È
Induktion einer wellenlangenabhangigen Photochemie in Bilirubin durch Serumalbumin
Zusammenfassung. Die Quantenausbeuten der Z > E Photoisomerisierung des natuÈrlichen
È
bichromophoren Bilirubins IXꢀ in ammoniakalisch-methanolischer Losung variieren mit der
Anregungswellenlange, wogegen (innerhalb des experimentellen Fehlers) fur Xanthobilirubinsaure
È
È È
und symmetrisch substituierte Bilirubine (Bilirubin IIIꢀ, und XIIIꢀ, sowie Mesobilirubin XIIIꢀ) im
È
selben Losungsmittel keine solche Variation zu beobachten war. Die Quantenausbeute der Z > E
È
Photoisomerisierung von Xanthobilirubinsaure, die an Serumalbumin gebunden ist Wurde ebenfalls
È
È
als invariant in Bezug auf die Anregungswellenlange gefunden. Demgegenuber zeigten die
Quantenausbeuten der Z > E Photoisomerisierung symmetrisch substituierter Bilirubine eine
È È È
ausgepragte Abhangigkeit von der Wellenlange, wenn sie an Serumalbumin gebunden waren. Diese
Ergebnisse zeigen, daû die Quantenausbeuten der Z > E Photoisomerisierung von Bilirubinen sehr
Ã
Corresponding author

650
A. F. McDonagh et al.
emp®ndlich in Hinblick auf Proteinbindung und die Natur der Pyrrol-ꢁ-Substituenten sind. Sie
È
È È
demonstrieren weiters daû die wellenlangenabhangige Photochemie charakteristisch fur Bilirubine
mit nicht-identischen Dipyrrin-chromophoren ist, wie dies aus der Excitonkopplungstheorie erwartet
È È
werden kann, und daû die Komplexierung mit Albumin auch eine wellenlangenabhangige
Photochemie in symmetrisch substituierten Bilirubinen induziert.
Introduction
Quantum yields for photochemical reactions do not generally vary with excitation
wavelength [1]. But exceptions to this generalization are not uncommon ± one
being the Z > E photoisomerization of bilirubin (BR) (1, Fig. 1) in solutions
containing equimolar or greater concentrations of human serum albumin (HSA)
[2]. Bilirubin is a bichromophoric molecule, biosynthesized from heme, that
circulates in blood as a reversible association complex with albumin. Its complex
with human serum albumin (BR Á HSA) has a broad absorption band peaking at
about 460 nm and extending to about 530 nm. When people are exposed to light
with wavelengths in this region, bilirubin in blood and other tissues undergoes
reversible photoisomerization from the biosynthetic 4Z,15Z form (1) to the
isomeric 4Z,15E (2) and 4E,15Z (3) forms [3, 4]. Because these three diastereomers
have extensively overlapping absorption bands, a photoequilibrium mixture of all
of them (along with a small amount of the 4E,15E isomer) is eventually formed
when BR Á HSA solutions are irradiated in vitro [5]. Complexation of bilirubin with
proteins has a strong in¯uence on the regioselectivity of its photochemistry. In the
case of the BR Á HSA complex, the proportion of 4E,15Z and 4E,15E isomers that is
present at photoequilibrium is very small compared to the proportion of the 4Z,15E
isomer [3, 6]. Photoisomerization of bilirubin in humans has no known
physiological function, but the reaction is important in phototherapy of newborn
babies with severe jaundice and in the treatment of patients with the rare hereditary
Crigler-Najjar syndrome [7, 8].
The quantum yield (ÈZ;Z!Z;E) for Z!E photoisomerization of BR Á HSA in
vitro varies with excitation wavelength, decreasing from 0.1 to 0.05 with increasing
Fig. 1. Linear representations of 4Z,15Z-bilirubin IXꢀ (1) and its 4Z,15E (2) and
4E,15Z (3) con®gurational isomers

Wavelength-Dependent Photochemistry in Bilirubins
651
Fig. 2. Chemical structures of 4Z,15Z-bilirubin IIIꢀ (4), 4Z,15Z-bilirubin XIIIꢀ (5), 4Z,15Z-
mesobilirubin XIIIꢀ (6), and Z-xanthobilirubic acid (7)
excitation wavelength over the range 457.9±514.5 nm [2]. This phenomenon has
been attributed to exciton coupling [9] and intramolecular partitioning of excitation
energy between the two similar, but non-identical, dipyrrinone chromophores of
the protein-linked pigment [2, 10±12]. The exciton model, which is based on a
more rigorous quantum mechanical treatment, also predicts wavelength-dependent
photochemistry for BR even when it is not bound to protein. However, that pre-
diction has not been demonstrated experimentally, and it is unclear whether the
observed variation of quantum yield with excitation wavelength for photoisome-
rization of BR Á HSA in aqueous solutions is an inherent property of the chromo-
phore or is engendered solely by its association with protein. To distinguish these
contributions, we have measured quantum yields for the photoisomerization of BR
and several closely related model compounds (Fig. 2, 4±7) in protein-free
methanolic solutions and in aqueous solutions containing excess HSA. Our results
are consistent with the exciton coupling explanation for the wavelength-dependent
photochemistry of bilirubin. They show that wavelength-dependent photochemistry
is an inherent property of the pigment which is augmented by binding to albumin.
For symmetrically substituted bilirubins, such as 4, 5 and 6, quantum yields for
photoisomerization are independent of excitation wavelength in the absence of
protein, but wavelength dependent in its presence. Protein binding, therefore, can
effectively ``de-symmetrize'' symmetrical bichromophoric molecules and induce
wavelength-dependent photochemistry.
Results
For all compounds studied, E photoisomers were readily separated from the less
polar parent Z isomers by HPLC. Formation of photooxidation products and of
isomers other than Z,E was negligible under the experimental conditions used.
Xanthobilirubic acid
Quantum yields for Z!E photoisomerization of xanthobilirubic acid (XBR, 7) at
different excitation wavelengths are given in Table 1. In both methanolic and HSA
solutions, values of ÈZ!E were substantial (0.19 and 0.12, respectively) and

652
A. F. McDonagh et al.
Table 1. Quantum yields (ÈZ!E) for Z!E photoisomerization of
xanthobilirubic acid (7) in 0.1% NH4OH/MeOH and HSA solutions
ÈZ!E
Excitation
wavelength
(nm)
NH4OH/MeOHa
HSAa
390
410
430
440
460
470
0.197 Æ 0.010
0.190 Æ 0.006
0.192 Æ 0.010
0.184 Æ 0.006
0.182 Æ 0.002
0.127 Æ 0.003
0.122 Æ 0.005
0.122 Æ 0.005
a
Mean Æ standard deviation
constant within the experimental error over the range of excitation wavelengths
investigated. The value for XBR Á HSA is much smaller than that reported
previously (0.4; [13]). However, as noted elsewhere [2], the earlier value was based
on an estimated ÈZ;Z!Z;E value of 0.2 for BR Á HSA which has subsequently been
shown to be incorrect.
Symmetric bilirubins
Quantum yields for Z!E photoisomerization of bilirubin IIIꢀ (BR-III, 4),
bilirubin XIIIꢀ (BR-XIII, 5), and mesobilirubin XIIIꢀ (MBR-XIII, 6) are listed in
Table 2. For all three compounds, ÈZ!E was signi®cantly (P > 0.1) independent of
excitation wavelength in NH4OH/MeOH, with average values of 0.058, 0.013 and
0.041, respectively. When the pigments were complexed with HSA, however,
ÈZ!E decreased signi®cantly with increasing excitation wavelength (P < 0.0005),
the greatest effect being observed with BR-III (Fig. 3). The maximum ÈZ!E values
for BR-III and MBR-XIII in HSA solutions were only slightly different from those
in NH4OH/MeOH. In contrast, ÈZ!E at 460 nm for BR-XIII was four times greater
in HSA solution than in NH4OH/MeOH.
Bilirubin-IXꢀ
Irradiation of BR (4Z,15Z) in NH4OH/MeOH generated the corresponding 4E,15Z
and 4Z,15E stereoisomers 2 and 3. Quantum yields for their formation are given in
Table 3 and plotted in Fig. 4 as a function of excitation wavelength. The quantum
yield ÈZ;Z!E;Z for formation of the 4E,15Z isomer 3 is about one third of the
quantum yield ÈZ;Z!Z;E for formation of the 4Z,15E stereoisomer 2 for excitation
in the 430±470 nm range, but only about one half at 500 nm because of a decrease
in ÈZ;Z!Z;E and an increase in ÈZ;Z!E;Z. Our measured quantum yields are similar
to those reported for the photoisomerization of bilirubin in chloroform/
triethylamine solution at a single excitation wavelength (436 nm) [14]. ÈZ;Z!Z;E
showed a weak but signi®cant wavelength dependence, decreasing with increasing
excitation wavelength. The effect was not so marked as for BR Á HSA (reported

Wavelength-Dependent Photochemistry in Bilirubins
653

654
A. F. McDonagh et al.
Fig. 3. Variation in quantum yield with excitation wavelength for Z!E photoisomerization of HSA
complexes of 4Z,15Z-mesobilirubin XIIIꢀ (6), 4Z,15Z-bilirubin XIIIꢀ (5), and 4Z,15Z-bilirubin IIIꢀ
(4). The dashed curve shows previously published [2] quantum yields for the photoisomerization of
4Z,15Z-bilirubin IXꢀ to 4Z,15E-bilirubin IXꢀ under similar conditions
Table 3. Quantum yields for photoisomerization of 4Z,15Z-BR-IX (1) to 4E,15Z-BR-IX (3) (ÈZ;Z!E;Z
and to 4Z,15E-BR-IX (2) (ÈZ;Z!Z;E) in 0.1% NH4OH/MeOH
)
a
a
Excitation
wavelength
(nm)
ÈZ;Z!E;Z
ÈZ;Z!Z;E
430
450
470
480
500
0.0080 Æ 0.0002
0.0085 Æ 0.0005
0.0082 Æ 0.0006
0.0086 Æ 0.0007
0.0103 Æ 0.0004
0.0253 Æ 0.0012
0.0266 Æ 0.0010
0.0238 Æ 0.0016
0.0220 Æ 0.0010
0.0197 Æ 0.0011
a
Mean Æ standard deviation
Fig. 4. Variation in quantum yields with excitation wavelength for the photoisomerization of
4Z,15Z-bilirubin IXꢀ (1) to 4Z,15E-bilirubin IXꢀ (2) and 4E,15Z-bilirubin IXꢀ (3) in 0.1%
NH4OH/MeOH

Wavelength-Dependent Photochemistry in Bilirubins
655
previously [2] and shown in Fig. 3 (dashed curve) for comparison). ÈZ;Z!E;Z was
constant within the experimental error between 430 and 480 nm, but increased by
24% at 500 nm.
Discussion
There is extensive spectroscopic and computational evidence that the methylene-
linked dipyrrinone chromophores in excited-state bilirubins interact by electric
dipole-dipole coupling, forming a molecular exciton [15±18]. Because of this, the
®rst excited singlet state splits into two energy levels [9, 15, 19] and may be
represented simplistically as a double minimum potential well (Fig. 5) from which
decay, via twisted intermediates, can lead to ground-state photoisomers or starting
material [10±12]. Decay to ground-state photoisomer from one well would
correspond to isomerization in one of the rubin chromophores, decay from the
other well to isomerization in the other chromophore. Competing with these decay
pathways, but not depicted in the diagram, are radiationless internal conversions
back to the ground state [20±23]. Luminescent pathways can be ignored because
the quantum yield for ¯uorescence for bilirubin is very small [24].
For symmetrically substituted bilirubins, such as BR-III (4), BR-XIII (5) and
MBR-XIII (6), the same E-isomer would be obtained, via identical twisted-state
intermediates, from either of the two exciton states. Therefore, wavelength-
dependent photochemistry would only be expected if the rate of radiationless decay
from each exciton state to the common ground state were signi®cantly different,
which is unlikely. Consistent with this expectation, we were unable to detect
signi®cant variations in photoisomerization quantum yields with excitation
wavelength for the three symmetrically substituted bilirubins in NH4OH/MeOH.
Fig. 5. Double-minimum potential energy model for the ®rst excited singlet state of 4Z,15Z-bilirubin
IX (1); horizontal dashed lines indicate excited vibronic levels

656
A. F. McDonagh et al.
Similarly, unsymmetrically substituted bilirubins also have intrinsically two
exciton states with potential energy wells of unequal depths. But, in this case, the
two excitons can decay to non-identical diastereomeric ground states (and starting
material), and their rates of decay might be different. For bilirubin IXꢀ (1), the
excitation energy may become localized either in the dipyrrinone with an exo-vinyl
group or in the dipyrrinone with an endo-vinyl group. Decay from these states
would then generate the respective diastereomeric 4Z,15E and 4E,15Z ground
states or starting material. Thus, for both symmetrically and unsymmetrically
substituted bilirubins the speci®c singlet (exciton) well into which the excitation
energy eventually becomes localized, following decay from upper vibrational
levels, may depend on the wavelength (energy) of excitation, but wavelength-
dependent quantum yields for photoisomerization would be expected only for
unsymmetrically substituted pigments in an achiral isotropic medium. Excitation of
these pigments using energies that are large enough to populate singlet vibrational
levels lying above the barrier between the wells would allow excitation to hop
between the wells before eventually becoming trapped in one of them [25, 26].
Therefore, with short-wavelength excitation, quantum yields for isomerization
would be expected to be independent of wavelength and dependent only on the
relative depths of the wells and the relative rates of radiationless deactivation
processes. But towards longer excitation wavelengths, the probability of excitation
to vibronic levels higher than the barrier between the wells would diminish, and
some variation of the quantum yields for isomerization with wavelength would
be expected. This expectation was borne out for photoisomerization of bilirubin
in NH4OH/MeOH. The quantum yield, ÈZ;Z!Z;E for formation of the major
4Z,15E isomer declined signi®cantly with excitation wavelength over the range
of 450±500 nm. A corresponding increase in the ÈZ;Z!E;Z quantum yield was
not detected for excitation from 430 to 480 nm, but the value for excitation
at 500 nm was signi®cantly greater than that for excitation at 480 nm. The
possibility that the wavelength-dependent photochemistry exhibited by bilirubin in
NH4OH/MeOH stems from the presence of conformational isomers [27] or
aggregates [19] or from other ground-state phenomena is ruled out by the
wavelength-independent photochemistry of BR-III (4), BR-XIII (5) and MBR-XIII
(6) in the same solvent.
When bilirubins bind to albumin, strong intramolecular exciton coupling is
maintained, as shown by CD studies of symmetric (BR-III, MBR-XIII) and
unsymmetric (BR-IX) rubins [15, 16, 18, 28]. Since the two chromophores of each
bound pigment molecule are no longer in identical environments, the relative rates
of radiationless deactivation and the partitioning of de-excitation pathways
between return to the original (Z,Z) ground state and to the photoisomer ground
state would be expected to be different for each of the two exciton states.
Consequently, wavelength-dependent quantum yields for Z!E photoisomeriza-
tion would be expected even for symmetrically substituted rubins bound to
albumin. This was observed for all three of the symmetrical compounds studied,
most markedly for BR-III (4). Thus, complexation with albumin desymmetrizes
symmetrical rubins and induces wavelength-dependent photochemistry.
Bichromophores other than rubins are likely to exhibit the same phenomenon
when they bind to proteins.

Wavelength-Dependent Photochemistry in Bilirubins
657
In contrast to the bichromophoric rubins, the monochromophoric compound 7
(XBR), for which exciton coupling is impossible, showed no detectable variation in
ÈZ!E over a range of wavelengths-spanning its visible absorption band in NH4OH/
MeOH or when bound to HSA.
The quantum yields for Z !E photoisomerization of the model dipyrrinone
XBR (7) in NH4OH/MeOH are considerably larger than those for Z !E
photoisomerization of bilirubins in the same solvent. This is not unexpected,
because intramolecular hydrogen bonding between propionic carboxylates and
pyrrinone NH groups, which would raise the energy barrier for Z !E
isomerization, is possible only in the rubins. Hydrogen bonding could also lower
quantum yields for isomerization of the rubins by facilitating abortive radiationless
relaxation processes [29, 30].
The data in this paper, along with previous work, show that quantum yields for
Z !E photoisomerizations of bilirubins are remarkably sensitive to both protein
binding and to the speci®c nature and position of the substituents on the lactam
rings. Although association of the pigments with albumin might be expected
intuitively to inhibit photoisomerization compared to the reaction in protein-free
solvents, such as methanolic ammonia, the converse is true for BR-IX and BR-XIII.
In contrast, binding to albumin has relatively little effect on photoisomerization of
MBR-XIII, ÈZ!E values for this compound being similar in methanol and aqueous
albumin solutions. However, at present, the individual in¯uences of lactam
substitution, hydrogen bonding, and protein binding on the quantum yields cannot
be quantitated.
Experimental
Chemicals
Bilirubin (Porphyrin Products, Logan, UT, USA) was puri®ed and crystallized as described
previously [31]. Bilirubin IIIꢀ (BR-III, 4) and bilirubin XIIIꢀ (BR-XIII, 5) were isolated by
preparative high pressure liquid chromatography (HPLC) [32] of a puri®ed mixture of bilirubins
IIIꢀ, IXꢀ and XIIIꢀ produced by acid catalyzed scrambling of bilirubin [33]. Xanthobilirubic acid
(XBR, 7) and mesobilirubin XIIIꢀ (MBR-XIII, 6) were prepared by total synthesis [34, 35]. All
pigments were >94% homogeneous by HPLC. Human serum albumin (HSA) was ``Essentially Fatty
Acid Free, Fraction V'' (Sigma Chemical Co., St. Louis, MO, USA).
Solutions
Solutions in NH4OH/MeOH were prepared by dissolving weighed amounts of solute in argon purged
solvent (conc. NH4OH:MeOH ˆ 0.10:99.9, v/v). Solutions of rubins in aqueous HSA were prepared
by dissolving weighed quantities of pigment (0.06±0.3 mg) in 50±300 ml of argon purged 0.05 M
NaOH and diluting this solution immediately to the required concentration with HSA in argon
purged 0.1 M phosphate buffer (pH 7.4). Final pigment:HSA molar concentration ratios were 1:2.
For studies on XBR (7) a mole ratio of pigment to protein of 1:10 was used. With lower protein
ratios, thermal reversion of E-7 to Z-7 was signi®cant during the time required to complete the
measurements. Solutions were freshly prepared each working day and allowed to stand for about an
hour in the dark before starting measurements. Pigment solutions were manipulated under red

658
A. F. McDonagh et al.
Table 4. Absorption maxima (ꢂmax) and millimolar extinction coef®cients ("max) for pigments
studied
0.1% NH4OH/MeOH
max (nm)
HSA/buffer, pH 7.4
max (nm)
Compound
ꢂ
"
ꢂ
"
max
max
BR-IX (1)
BR-III (4)
BR-XIII (5)
MBR-XIII (6)
XBR (7)
453
455
448
430
416
68.4 Æ 1.0
61.1 Æ 2.7
57.5 Æ 2.0
65.3 Æ 0.5
37.5 Æ 0.3
458
456
458
436
426
47.0 Æ 1.3
53.0 Æ 1.0
42.0 Æ 1.2
50.2 Æ 0.9
31.2 Æ 0.4
safelight. Extinction coef®cients and absorption maxima for the compounds used in the study are
listed in Table 4.
Quantum yields
Solutions (1.5±2 ml) were irradiated at room temperature under 1 atmosphere of argon in a 1 cm path
length quartz cuvette by a light beam with a cross section of about 0.2 cm2 inside the solution.
During irradiations, samples were stirred magnetically by a stirring bar which did not obstruct the
light beam. In some experiments, irradiation was performed with 5 nm bandpass light from a Bausch
and Lomb grating monochromator. The light beam was ®ltered to remove harmonic overtones below
360 nm. In other experiments, we used the expanded beam of an argon ion laser (Spectra Physics,
Model 165±008). Energy doses in the quantum yield measurements were kept low enough to
produce a fraction of E isomer that was less than 20% of the photoequilibrium value for each
excitation wavelength. Photoequilibria for Z > E reactions were generated using photon ¯uence
rates about 10 times greater than those used in the quantum yield measurements, and the reactions
were monitored by absorbance difference spectroscopy. Photon dose rates at the sample surface were
measured by potassium ferrioxalate actinometry [36] or by a photodiode power meter (EG & G,
Model 460). The two methods gave similar results. A 0.02 M solution of K3Fe(C2O4)3 in 0.05 M
H2SO4 was used as actinometer for measurements in the 390±460 nm range. The actinometer
concentration was increased to 0.15 M for measurements between 470 and 515 nm. The light
intensity was measured before and after each photoisomerization experiment, and the average value
was used to calculate the incident photon dose rate. In some experiments, the excitation light was
split into two orthogonal beams by a cube beam splitter in order to monitor the light intensity during
the photoisomerization. Photon doses in the quantum yield measurements ranged from 4.4Â1015 to
2.6Â1017 depending on excitation wavelength, pigment concentration, and volume of the sample.
Initial pigment concentrations were 9.3±21.9 mM, 6.5±66 mM, 9.9±22.6 mM, 15.4±23 mM and 4.2±
10.5 mM for BR-III, BR-IX, BR-XIII, MBR-XIII and XBR, respectively, in 0.1% NH4OH/MeOH, and
23.9±51.4 mM, 30.0±44.4 mM, 32.7±36.8 mM and 15.7±41.7 mM for BR-III, BR-XIII, MBR-XIII and
XBR, respectively, in aqueous HSA solutions.
Quantum yields (È) for the Z!E reaction were calculated by
cE Â V Â N
È ˆ
F  t  ꢀ1 ꢁ 10ꢁA
†
where cE is the concentration of the E isomer after irradiation time t, V represents the solution
volume, and N is Avogadro's number. F is the photon ¯ux and A the initial absorbance of the
solution. The concentration of the E isomer in the irradiated solutions was determined by HPLC and

Wavelength-Dependent Photochemistry in Bilirubins
659
corrected for the E!Z back photoreaction according to
ꢀ
ꢁ
cPEE
cE ˆ cPEE  ln
ꢀcPEE ꢁ cmE †
where cmE and cPEE are the measured concentrations of the E isomer at irradiation time t and at
photoequilibrium.
HPLC analyses [5, 37]
Photoisomers were analyzed on a 5m, 25Â0.46 cm Ultrasphere C18 IP-ODS column (Beckman
Instruments, Inc., Fullerton, CA, USA) coupled to a 3Â0.46 cm pre-column containing the same
packing material. The solvent ¯ow rate was 1 ml/min. Eluting compounds were detected with a diode
array absorbance detector (Hewlett-Packard, Model 1040A). The solvent was a mixture of water and
0.1 M methanolic di-n-octylamine acetate (Me-Docta), prepared from di-n-octylamine (Aldrich
Chemical Co., St. Louis, MO, USA) and acetic acid.
For analysis of the isomers of XBR the H2O:Me-Docta ratio was 1:4 (v/v) and peaks were
monitored at 416 nm, the absorption maximum for Z-XBR in the eluent. Preliminary studies on the
Z > E photoisomerization of XBR in the HPLC mobile phase showed that the absorbance spectra of
the individual isomers have an isosbestic point at 443 nm and relative extinction coef®cients of
1:1.63 (E:Z) at 416 nm. This correction factor was used in calculating relative isomer concentrations
from the integrated areas of HPLC peaks detected at 416 nm.
For HPLC analysis of BR-III and BR-IX isomer solutions, the H2O:Me-Docta ratio was 8:92
(v/v); for analysis of MBR-XIII solutions it was 5:95 (v/v), and for analysis of BR-XIII solutions it
was either 8:92 (v/v) (photolyses in NH4OH/MeOH) or 5:95 (v/v) (photolyses in HSA solutions). For
BR-III, BR-XIII and MBR-XIII, peaks were monitored at the corresponding isosbestic points (481,
486, and 455 nm, respectively) for the Z > E isomerization in the HPLC solvent, determined by
absorbance difference spectroscopy. Photoisomers of BR-IX were monitored at 450 nm, and relative
HPLC peak areas were corrected for extinction coef®cient differences using the factors of 1.00, 1.02
and 1.25 for 4Z,15Z-, 4E,15Z- and 4Z,15E-BR-IX, respectively [12].
Injection volumes for HPLC were 20 ml. Samples in NH4OH/MeOH were mixed before injection
with enough water to make the proportion of water in injectate and mobile phase the same. This
precaution eliminated ghost peaks that interfered with the quantitation of E isomers. Samples in HSA
solutions were diluted with excess ice-cold mobile phase, vortexed for a few seconds, microfuged for
1 min, and the clear ¯uid phase was injected. Dilution ranged from 1:8 for BR-III/HSA solutions to
1:2 for XBR/HSA solutions.
In photoisomerization experiments, samples were analyzed in duplicate or triplicate either
immediately or after brief storage at 0 ꢂC. Unirradiated solutions were analyzed as controls. Several
experiments (3±5) were run for each excitation wavelength, and average values were calculated. E
photoisomers were identi®ed on the basis of their absorption spectra, HPLC retention times, and
their quantitative reversion to the corresponding Z isomers upon treatment with tri¯uoroacetic acid
[5]. The signi®cance of the wavelength dependence was tested by one-way analysis of variance.
Curves in the displayed ®gures were drawn using the linear ®t command in Kaleidograph (Macintosh
Version 3.0, Abelbeck Software, Reading, PA, USA).
Acknowledgements
This work was supported by grants DK 26307, DK 26743, GM 36633, and HD 17779 from the
National Institutes of Health and by NATO grant 0149/85. G. A. was supported by a fellowship from
the Comitato Nazionale per le Scienze Biologiche e Mediche ± CNR, Italy. Also acknowledged are
the assistance of Dr. Franco Fusi (CNR) with some of the measurements and support and
encouragement to G. A. by Prof. R. Pratesi (CNR).

660
A. F. McDonagh et al.: Wavelength-Dependent Photochemistry in Bilirubins
References
[1] Morrison H, Bernasconi C, Pandey G (1984) Photochem Photobiol 40: 549
[2] Agati G, Fusi F, Pratesi R, McDonagh AF (1992) Photochem Photobiol 55: 185
[3] Lightner DA, McDonagh AF (1984) Acc Chem Res 17: 417
[4] McDonagh AF (1986) N Engl J Med 314: 121
[5] McDonagh AF, Palma LA, Trull FR, Lightner DA (1982) J Am Chem Soc 104: 6865
[6] Kanna Y, Arai T, Tokumaru K (1993) Bull Chem Soc Jpn 66: 1586
[7] McDonagh AF, Lightner DA (1985) Pediatrics 75: 443
[8] Ennever JF (1992) Phototherapy for Neonatal Jaundice. In: Polin R, Fox W (eds) Fetal and
Neonatal Physiology, vol 2. Saunders, New York, p 1165
[9] Harada N, Nakanishi K (1983) Circular Dichroic Spectroscopy ± Exciton Coupling in Organic
Stereochemistry. University Science Books, Mill Valley, CA
[10] McDonagh AF, Lightner DA (1985) Mechanism of Phototherapy of Neonatal Jaundice.
Regiospeci®c Photoisomerization of Bilirubins. In: Blauer G, Sund H (eds) Optical Properties
and Structure of Tetrapyrroles. Walter de Gruyter, Berlin, p 297
[11] McDonagh AF, Lightner DA (1985) Intramolecular Energy Transfer in Bilirubins. In: Bensasson
R, Jori G, Land E, Truscott T (eds) Primary Photoprocesses in Biology and Medicine. Plenum
Press, New York, p 321
[12] McDonagh AF, Agati G, Fusi F, Pratesi R (1989) Photochem Photobiol 50: 305
[13] Lamola AA, Braslavsky SE, Schaffner K, Lightner DA (1983) Photochem Photobiol 37: 263
[14] Kanna Y, Arai T, Tokumaru K (1993) Bull Chem Soc Jpn 66: 1482
[15] Blauer G (1983) Israel J Chem 23: 201
[16] Person RV, Peterson BR, Lightner DA (1994) J Am Chem Soc 116: 42
[17] Boiadjiev SE, Lightner DA (1997) Chirality 9: 604
[18] Lightner DA, Reisinger M, Landen GL (1986) J Biol Chem 261: 6034
[19] Kasha M (1963) Radiation Res 20: 55
[20] Falk H, Neu®ngerl F (1979) Monatsh Chem 110: 987
[21] Falk H, Neu®ngerl F (1979) Monatsh Chem 110: 1243
[22] Falk H, Grubmayr K, Neu®ngerl F (1979) Monatsh Chem 110: 1127
[23] Kaschke M, Kleinschmidt J, Riegler M (1986) Chem Phys 107: 89
[24] Lamola AA (1985) Effects of Environment on Photophysical Processes of Bilirubin. In: Blauer G,
Sund H (eds) Optical Properties and Structure of Tetrapyrroles. Walter de Gruyter, Berlin, p 311
[25] Ikeda T, Lee B, Kurihara S, Tazuke S, Takenaka A, Ito S, Yamamoto M (1988) J Am Chem Soc
110: 8299
[26] Ikeda T, Lee B, Tazuke S, Takenaka A (1990) J Am Chem Soc 112: 4650
[27] Holzwarth AR, Schaffner K (1981) Photochem Photobiol 33: 635
[28] Agati G, McDonagh AF (1995) J Am Chem Soc 117: 4425
[29] Falk H (1989) The Chemistry of Linear Oligopyrroles and Bile Pigments. Springer, Wien
[30] Lewis FD, Yoon BA, Arai T, Iwasaki T, Tokumaru K (1995) J Am Chem Soc 117: 3029
[31] McDonagh AF, Assisi F (1972) Biochem J 129: 797
[32] DiCesare JL, Vandemark FL (1987) Perkin Elmer Chromatog Newslet 9: 7
[33] McDonagh AF, Assisi F (1972) J Chem Soc Chem Comm 117
[34] Grunewald J, Cullen R, Bredfeldt J, Strope ER (1975) Org Prep Proc Int 7: 103
[35] Trull FR, W FR, Lightner DA (1987) J Heterocyclic Chem 24: 1573
[36] Murov SL (1973) Handbook of Photochemistry. Dekker, New York
[37] McDonagh AF, Palma LA, Lightner DA (1982) J Am Chem Soc 104: 6867
Received January 13, 1998. Accepted January 27, 1998