Mechanism of hemolysis and erythrocyte transformation caused by lipogrammistin-A, a lipophilic and acylated cyclic polyamine from the skin secretion of soapfishes (Grammistidae)

Article abstract of DOI:10.1016/S0968-0896(99)00128-5

The mechanism of hemolysis and erythrocyte transformation caused by lipogrammistin-A (LGA), a lipophilic and acylated cyclic polyamine from the skin secretion of soapfishes (Grammistidae), was investigated. The dependency of hemolysis on the erythrocyte concentration indicated that the amount of membrane-bound LGA required for 50% hemolysis is about 13% of the total phospholipids in erythrocytes on a molar basis. A synthetic analogue which lacked a long alkyl chain exhibited much less activity, suggesting that the alkyl chain is important for membrane-binding. In addition, microscopic observations showed that LGA elicited the invagination of erythrocytes at sublytic concentrations, which makes LGA one of the most potent agents with this transforming activity known to date. Its protonated secondary amino group is responsible for the unequal distribution of LGA in the inner leaflet of the lipid bilayer, which leads to invagination, since acetylation at the amino group markedly reduced the invagination activity. Furthermore, the size of LGA-induced lesions on erythrocyte membrane was estimated to be 7-29 A based on osmotic protection experiments, where the external addition of isotonic molecules in this size range gradually increased the effective dose of LGA. Based on these lines of evidence, the mode of LGA action on erythrocytes is deduced to be as follows. First, LGA molecules bind to erythrocyte membrane by lipophilicity. Second, the molecules accumulate in the inner leaflet of the lipid bilayer by interaction of their cationic ammonium groups with acidic residues of membrane lipid in the inner surface. This uneven distribution of LGA distorts the bilayer structure, and results in a change in cell shape and consequent small lesions. Third, small solutes permeate through the lesions, which induces an osmotic change across the membrane, which leads to colloid-osmotic rupture. This mode of action of LGA on erythrocytes accompanied by cell invagination is the first reported example for natural defense substances.

Full text of DOI:10.1016/S0968-0896(99)00128-5

Bioorganic & Medicinal Chemistry 7 (1999) 2073±2081
Mechanism of Hemolysis and Erythrocyte Transformation Caused
by Lipogrammistin-A, a Lipophilic and Acylated Cyclic Polyamine
from the Skin Secretion of Soap®shes (Grammistidae)
y
Yoshimasa Kobayashi, Hiroyuki Onuki and Kazuo Tachibana*
Department of Chemistry, School of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan
Received 10 March 1999; accepted 7 May 1999
AbstractÐThe mechanism of hemolysis and erythrocyte transformation caused by lipogrammistin-A (LGA), a lipophilic and acy-
lated cyclic polyamine from the skin secretion of soap®shes (Grammistidae), was investigated. The dependency of hemolysis on the
erythrocyte concentration indicated that the amount of membrane-bound LGA required for 50% hemolysis is about 13% of the
total phospholipids in erythrocytes on a molar basis. A synthetic analogue which lacked a long alkyl chain exhibited much less
activity, suggesting that the alkyl chain is important for membrane-binding. In addition, microscopic observations showed that
LGA elicited the invagination of erythrocytes at sublytic concentrations, which makes LGA one of the most potent agents with this
transforming activity known to date. Its protonated secondary amino group is responsible for the unequal distribution of LGA in
the inner lea¯et of the lipid bilayer, which leads to invagination, since acetylation at the amino group markedly reduced the inva-
Ê
gination activity. Furthermore, the size of LGA-induced lesions on erythrocyte membrane was estimated to be 7±29 A based on
osmotic protection experiments, where the external addition of isotonic molecules in this size range gradually increased the e€ective
dose of LGA. Based on these lines of evidence, the mode of LGA action on erythrocytes is deduced to be as follows. First, LGA
molecules bind to erythrocyte membrane by lipophilicity. Second, the molecules accumulate in the inner lea¯et of the lipid bilayer
by interaction of their cationic ammonium groups with acidic residues of membrane lipid in the inner surface. This uneven dis-
tribution of LGA distorts the bilayer structure, and results in a change in cell shape and consequent small lesions. Third, small
solutes permeate through the lesions, which induces an osmotic change across the membrane, which leads to colloid-osmotic rup-
ture. This mode of action of LGA on erythrocytes accompanied by cell invagination is the ®rst reported example for natural defense
substances. # 1999 Elsevier Science Ltd. All rights reserved.
Introduction
the largest group of ichthyocrinotoxins and are the best
studied. These include pardaxins from the Moses sole
(Pardachirus pavoninus and P. marmoratus).6 The third
largest group is steroid glycosides, including pavoninins
±8
Certain species of ®sh (ichthyocrinotoxic species) are
known to produce ichthyotoxins for defense against
1
±3
predation. Although several species have been repor-
ted to secrete toxic mucus from their skin, there have
been very few studies of their detailed chemical struc-
tures and mechanisms of action. The ®rst of these to be
chemically investigated was the cationic surfactant
pahutoxin from the box®sh, Ostracicon lentiginosus,
which is a choline ester of fatty acid. A series of its
analogues were subsequently reported in the secretions
from box®sh and trunk®sh. Polypeptide surfactants are
and mosesins from the Moses sole (Pardachirus pavoni-
nus and P. marmoratus).9
±12
All of these toxins have
amphiphilic structures, and show characteristics com-
mon to detergents.13
Soap®shes (Grammistidae) are another well-known ich-
thyocrinotoxic family. The peptidic ichthyotoxins called
grammistins have been reported in four typical Gram-
mistid ®shes: Pogonoperca punctata, Grammistes sex-
4
,5
lineatus, Diploprion bifasciatum, and Aulacocephalus
Recently, several grammistins from G.
temmincki.1
4±16
sexlineatus and P. punctata have been clari®ed to be
Key words: Lipogrammistin-A; erythrocyte; shape transformation;
colloid-osmotic hemolysis.
*
8
y
Corresponding author. Tel.: +81-3-5841-4366; fax: +81-3-5841-
380; e-mail: ktachi@chem.s.u-tokyo.ac.jp
Present address: Marine Biotechnology Institute, Sodeshi, Shimizu,
polypeptides, which respectively contain 13±25 amino
acid residues.
1
7,18
They are reported to form amphiphilic
20
19
Shizuoka, 424-0037, Japan.
a-helices and disrupt the cell membrane.
0968-0896/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(99)00128-5

2
074
Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
synthesis of which has been previously reported.22 The
resulting amide 7 was converted to aldehyde 8 by
reduction with LiBH and then SO -pyridine oxidation.
4
3
After deprotection, reductive cyclization was carried out
via a Schi€ base to furnish 2 in modest yield. An N-acetyl
analogue (3; Scheme 2) was prepared from LGA by
acetylation with Ac O. The structures of analogues 2 and
2
1
3
logue (4; Scheme 2) was derived from LGA by Hofmann
were con®rmed by MS and H NMR. An acyclic ana-
In addition to grammistins, the secretions from D. bifas-
ciatum and A. temmincki have been shown to contain a
degradation,23 which was carried out by quaternization
of its secondary amine with CH I. The structure was
characterized by MS and by its UV spectrum, which
3
1
6
large amount of lipophilic hemolysin. We recently
succeeded in determining the chemical structure of the
lipophilic hemolysin, named lipogrammistin-A (LGA,
showed the presence of an a,b-unsaturated amide. Its
trimethylammonium group was con®rmed by C NMR
13
), by spectroscopic methods and total synthesis.2
1,22
of a C-labeled derivative of LGA prepared by the
same procedure as for analogue 4, except for using
13
1
LGA does not belong to any of the classes of ichthyo-
crinotoxins described above, and is the ®rst polyamine
lactam, as an acylated form, to be isolated from animal
sources. Interestingly, LGA was more hydrophobic
than other amphiphilic ichthyotoxins, which implies
that its mechanism for cytolysis is di€erent than those
of detergents.
13
CH3I.
Membrane anity and intrinsic hemolytic activity of LGA
Figure 1 shows the dose-dependency of LGA-induced
hemolysis for three di€erent densities of erythrocytes.
Based on the shift of these curves along the increased
density of erythrocytes, the relationship between the
bound LGA per cell and percent hemolysis was
obtained as shown in Figure 2. The binding isotherm of
LGA was also obtained as shown in Figure 3.
In this paper, we ®rst report the potent erythrocyte-
transformation activity of LGA, and then describe the
respective functions of structural components of LGA in
its hemolytic and erythrocyte-transformation activities.
Figure 2 shows the intrinsic hemolytic activity of LGA
after binding to erythrocytes. Here, the concentration of
LGA in erythrocytes required for 50% hemolysis is
Results
7
Preparation of LGA analogues
5Â10 molecules/cell, which is estimated to be about
13% of the lipid molecules comprising the membrane of
a single cell.
24
An analogue bearing all of the functional groups of
LGA except for a long alkyl chain (2; Scheme 1) was
synthesized by the condensation of N-protected b-ala-
nine (6) and the acylated polyamine unit of LGA (5),
Linearity of the binding isotherm (Fig. 3) implies that
2
5
LGA is bound as a monomer in the membrane, where
Scheme 1.

Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
2075
Scheme 2.
Figure 1. Dose±response curves for LGA-induced hemolysis at an ery-
7
Figure 2. Relationships between the percentage of hemolysis and the
amount of membrane-bound substances; LGA (1; squares), N-acety-
lated analogue (3; circles), and ring-cleaved analogue (4; triangles).
7
throcyte density of 1Â10 cells/mL (squares), 2Â10 cells/mL (closed
7
7
circles), 4Â10 cells/mL (triangles), and 8Â10 cells/mL (open circles),
ꢀ
mean‹SD, n=4. Cells were incubated with LGA at 37 C for 12 h. Other
conditions were as in Experimental.
Membrane-perturbing activity of LGA in erythrocytes and
SUVs
most of the molecules incorporated in erythrocytes are
presumed to be partitioned. If LGA formed aggregates
in the membrane, the binding isotherm would follow an
exponential function, considering that LGA should be
monomeric in the outer medium at this concentration
range. The partition coecient of LGA from the aqu-
eous medium to cell membrane of erythrocytes is esti-
To estimate the contribution of the membrane-perturb-
ing activity of LGA to hemolysis, LGA-induced ¯uo-
rescence leakage from calcein-encapsulating SUVs
(small unilamellar vesicles) consisting of eggPC (egg-
yolk phosphatidylcholine) was evaluated and compared
with LGA-induced hemolysis. The dependence of ¯uo-
rescence leakage on the concentration of LGA is shown
in Figure 4 together with hemolysis at the same lipid
density for comparison. As shown, these two dose±
response curves are very close. As with hemolytic activity,
4
mated from Figure 3 as 7Â10 when the avarage volume
of the lipid bilayer in an erythrocyte cell is assumed to
be 5Â10
�
16
L. This can be converted to the association
8
constant (K ) between LGA and lipid molecule (4Â10
a
4
� 1
.
molecules/cell) as 5Â10 M

2
076
Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
thermodynamic analysis revealed that two factors, i.e.
intrinsic membrane-disrupting activity and membrane-
binding anity, made independent contributions (data
not shown). In the case of eggPC-SUV, the association
erythrocytes (i.e. ghosts) were observed. Figure 6 shows
the dose-dependency of the LGA-induced shape change.
This activity of LGA was compared with that of chlor-
promazine, which was previously reported to be a
potent erythrocyte-invagination inducer. LGA was
one order of magnitude more potent than chlorproma-
zine on a molar basis.
5
� 1
26
constant K was shown to be 1.5Â10 M , which is
a
somewhat higher but at the same magnitude as that
obtained for hemolysis. In addition, the concentration
of LGA molecules in membrane required for the 50%
leakage of calcein was about 20% of the lipid molecules.
This value is also close to that obtained for hemolysis.
These results indicate that LGA-induced hemolysis is
caused not by interaction with membrane-bound
proteins, but essentially by membrane disruption.
Size of LGA-induced lesions on erythrocyte membrane
To evaluate the size of the lesions induced by LGA on
erythrocyte membrane, osmotic protection experiments
were performed. When the lesions are smaller than
hemoglobin, a colloid-osmotic process is involved in
hemolysis, which can be prevented by adding solutes
larger than the lesions to the medium.27
Transformation of the shape of erythrocytes by LGA
To investigate the mode of action of LGA on ery-
throcyte membranes in further detail, we examined the
shape-transforming activity of LGA on erythrocytes at
sublytic doses. Microscopic observation of LGA-treated
cells showed that LGA changed the shapes of ery-
throcytes from normal discoid to an invaginated form
Figure 7 shows LGA-induced hemolysis in the presence of
osmotic protectants of various sizes. Hemolysis was not
prevented by mannitol, but was prevented by substances
larger in diameter than PEG 400 (polyethylene glycol
400); PEG 2000 (polyethylene glycol 2000) almost com-
pletely prevented lysis at the dose for 100% hemolysis
(stomatocytes) at 2±3 mM (Fig. 5). At higher doses of
LGA, swollen spherical cells (spherocytes) and lysed
Figure 4. Comparison of the dose±response curve for calcein leakage
from liposomes and that for hemolysis; hemolysis induced by LGA
(
squares), n=4, and calcein leakage induced by LGA (circles), n=2.
Figure 3. Binding isotherms of LGA (squares), N-acetylated analogue
(
Erythrocytes were suspended in bu€er at a density of 12 mM EggPC
equivalent. Liposomes were diluted to a concentration of 12 mM
EggPC equivalent. Other conditions were as in Experimental. Leakage
was measured 2 min after a sample was added.
3) (circles), and ring-cleaved analogue (4) (triangles), where partitions
between membrane-bound substances and free substances are shown
at di€erent percentages of hemolysis.
Figure 5. LGA-induced shape changes in erythrocytes. Erythrocytes were incubated for 1 h without (a) and with (b) 3.1 mM. (c) 5.5 mM. (d) 13.1 mM
of LGA. Pictures were photographed under a phase-contrast microscope after being ®xed with glutaraldehyde.

Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
2077
Structural e€ects on hemolytic and shape-transforming
activity
I: Long alkyl chain of LGA. The chemical structure of
LGA is characterized by its (i) long alkyl chain, (ii) sec-
ondary amine, and (iii) N-acyl cyclic polyamine struc-
ture. The contribution of each of these structural
features to the hemolytic activity and erythrocyte-
transforming activity of LGA was then investigated.
The synthetic analogue 2, which has all of the functional
groups of LGA except for its long alkyl chain, and is
thus less lipophilic, was subjected to hemolysis and
liposome assays. These revealed that 2 is devoid of
hemolytic activity, and produced much less dye leakage
than LGA (7% leakage at 1 mM of 2, data not shown).
II: Secondary amine. To examine the e€ect of the basic
secondary amino group of LGA on the above biological
activities, N-acetyl LGA (3), a neutral analogue under
physiological conditions, was tested in the same manner
as LGA. This non-cationic analogue showed less ery-
throcyte-transforming activity, as shown in Figure 6. It
also showed somewhat less hemolytic activity. The
thermodynamic data in Figures 2 and 3 revealed that
the membrane anity of 3 is comparable to that of
LGA. Therefore, the reduction in both of the activities
is believed to be due to the intrinsic potency of 3 in the
membrane. The concentration of 3 in the membrane
Figure 6. Dose-dependency of changes in erythrocyte shape induced
by LGA (squares), N-acetylated analogue 3 (circles), and chlorpro-
mazine (triangles). Erythrocytes were suspended in Tris±HCl bu€er at
7
a density of 1Â10 cells/mL. After incubation for 1 h, cells were
observed as in Experimental. The extent of transformation was quan-
ti®ed using the morphological index de®ned by Fujii and co-workers:
P
(
negative morphological score) Â (number of transformed cells) /
26
(total cell number). Morphological scores of � 1 to � 4 were assigned
to the shapes shown along the vertical axis (mean‹SD, n=3).
8
required for 50% hemolysis is 1Â10 molecules/cell,
which is nearly double that of LGA (Fig. 2)
III: Ring structure. To evaluate the contribution of the
cyclic structure of LGA to its activities, an acyclic ana-
logue (4) was prepared from LGA by quaternization
and consequent Hofmann degradation.23 The analogue
4
showed lower hemolytic and transformation-inducing
activities (data not shown). However, a thermodynamic
analysis revealed that this reduction in the activities was
not due to intrinsic activities, but to a decrease in
membrane-binding anity (Figs 2 and 3).
Discussion
The present experimental results enable us to deduce
how LGA induces hemolysis. The mechanism can be
divided into three steps; (i) LGA binds to erythrocyte
membrane, where (ii) it induces membrane-distortion
and erythrocyte-transformation, which (iii) leads to
colloid-osmotic hemolysis. Each step is discussed below.
Figure 7. Prevention of hemolysis induced by LGA in the presence of
3
mannitol (7 A in diameter, closed circles), ranose (11 A, triangles),
0 mM various colloid-osmotic protectants; none (closed squares),
27
Ê
Ê
Ê
PEG 400 (12 A, open circles), and PEG 2000 (29 A, open squares),
Ê
According to our thermodynamic investigations, com-
mon events are involved in hemolysis and ¯uorescence
leakage from liposomes consisting of only phosphati-
dylcholine. This indicates that hemolysis does not
involve speci®c receptors of LGA or other membrane
proteins, but rather nonspeci®c membrane-permeabili-
zation. The potent activity of LGA also involves its high
membrane anity due to its hydrophobicity. Based on
the experiments using the truncated analogue 2, it
appears that the long alkyl chain of LGA mainly con-
tributes to its high membrane anity. The cyclic structure
also contributes to its membrane anity in some degree.
n=4. Erythrocytes were suspended in Tris±HCl bu€er at a density of
7
2
Â10 cells/mL. Total osmotic pressure was adjusted by reduction of
ꢀ
NaCl. Hemolysis was determined after incubation with LGA at 37 C
for 30 min. Other conditions were as in Experimental.
(
formed by LGA is larger than 7 A (the diameter of man-
Fig. 7). These results indicate that the membrane lesion
Ê
Ê
nitol) and smaller than 29 A (the diameter of PEG 2000).
27
In addition, the required diameter of the protectant
increased with the dose of LGA. This indicates that the
size of the lesion increases as the dose of LGA increases.

2
078
Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
The present study demonstrated that LGA induces cell
transformation with invagination at sublytic doses (Fig.
Next, osmotic protection experiments revealed that
small solutes pass through LGA-induced lesions on the
membranes of distorted erythrocytes without producing
hemolysis. According to these experiments, the lesions
6). Erythrocyte transformation induced by drugs has
often been explained by the `bilayer couple hypothesis'
2
8
Ê
proposed by Sheetz and Singer; i.e. such shape chan-
ges may be due to asymmetric expansion of the lipid
bilayer membrane of erythrocytes caused by the selec-
tive accumulation of a drug in one of the two lea¯ets of
the bilayer. Cationic drugs such as chlorpromazine tend
to reside in the inner half of the bilayer due to Coulombic
interaction with the acidic phospholipids on this side.
This causes expansion of the inner half of the bilayer and
results in the invagination of erythrocytes. Although
some membrane components such as cholesterol play
roles in protecting the membrane against disruption,
LGA induced hemolysis in nearly the same dose as in
liposome experiments (Fig. 4). This might be because the
membranes of LGA-treated erythrocytes are highly dis-
torted due to the uneven accumulation of LGA molecules.
are estimated to be smaller than 30 A, which is too small
to allow hemoglobin to pass through the membrane.
This indicates that `colloid-osmotic hemolysis' occurs
on erythrocytes; i.e. since small solutes enter through
the small lesions by chemical potential, the osmotic
balance between the inside and outside of the cell is lost
due to entrapped hemoglobin, resulting in the permea-
tion of water into the cell and consequent explosion of
the cell. The cells just before explosion can be observed
under a microscope as swollen spherocytes at higher
sublytic doses.
The experiments also showed that the size of the mem-
brane lesions increases as the dose of LGA increases.
This excludes the possibility that LGA forms pores of a
speci®c diameter on the lipid bilayer, such as in ion
channel models. In addition, other experimental results
have indicated that, unlike ionophores, LGA molecules
do not chelate any charged solutes of physiological
abundance (unpublished results); in ion extraction
Since the non-cationic derivative of LGA (3) exhibits
weaker transformation activity (Fig. 6), LGA-induced
transformation can be partly accounted for by the
bilayer couple hypothesis. However, analogue 3 still
induced an invaginated transformation despite its neu-
trality. Factors other than the mere bilayer couple
hypothesis are needed to explain these results. Con-
cerning erythrocyte transformation by a neutral drug,
the involvement of aminophospholipid translocase has
assays,33 ionic solutes such as K , Na , Ca or Mg
+
+
2+
2+
were not extracted by LGA from water to the organic
layer. In experiments with an electrode34 loading LGA,
no potential response was observed against ions such as
�
Cl , various ammonium ions, or the metal ions men-
2
9,30
been proposed,
and this may also be the case with
tioned above. These results suggest that solutes enter
not through channels (or by carriers), but through
`defects' on the membrane, which are due to bilayer
perturbation caused by LGA.
analogue 3; i.e. perturbation of the lipid bilayer induced
by 3 causes a much faster ¯ip-¯op of membrane lipids,
resulting in the loss of their asymmetric distribution in
the bilayer. To restore the original lipid distribution,
aminophospholipid translocase draws phosphatidyl-
ethanolamine or phosphatidylserine from the outer
monolayer into the inner layer. Consequently, it causes
expansion of the inner half of the bilayer and invaginates
the erythrocyte.
Recent investigations35,36 have shown that surfactants
show two types of hemolytic activity; i.e. a `fast' activity
in which surfactant molecules that penetrate into the
lipid bilayer uproot membrane components by forming
mixed micelles with them, resulting in the leakage of
hemoglobins out of erythrocytes, and a `slow' activity,
which refers to colloid-osmotic lysis. The lesions pro-
Analogue 3 induced the same extent of hemolysis at
twice the dose as LGA. This reduction in hemolytic
activity may be due to the di€erence in intrinsic activ-
ities caused by membrane-bound molecules, since the
membrane-perturbing activity of bound LGA is twice
that of the N-acetyl analogue 3. On the other hand, the
cyclic structure of LGA is not essential for hemolytic
activity, since the intrinsic activity of the acyclic analo-
gue 4 is nearly the same as that of LGA once it binds to
the membrane (Figs 2 and 3).
Ê
duced by LGA are smaller than 30 A even at the dose
for 100% hemolysis, since PEG 2000 almost completely
protected against lysis (Fig. 7), which shows that lysis is
mostly due to the latter activity. The fact that the `fast'
activity does not play a role here may be ascribed to the
structure of the LGA molecule, which lacks a distinct
amphiphilic nature to solubilize membrane components
into micelles. In addition, it might also be ascribed to the
shape of the LGA molecule, which is not considered to
be suitable for micelle packing.
While this invaginating erythrocyte-transformation
activity has been widely found in local anesthetics and
tranquilizers, most of these show weaker hemolytic and
transformation activities against erythrocytes than
Thus, the mode of action of LGA on erythrocytes can
be summarized as follows. First, LGA monomers are
eciently partitioned to the erythrocyte cell membrane
due to their hydrophobic structure. Second, LGA
molecules accumulate in the inner lea¯et of the lipid
bilayer due to electrostatic interaction between their
protonated amino groups and the anionic polar heads
of membrane lipids, and then distort the bilayer struc-
ture to transform the cell shape and create small lesions.
Third, small solutes enter through these lesions, which
leads to an osmotic imbalance across the membrane.
LGA. Furthermore, only a few examples from natural
sources are known,2
6,31
and none of them are known to
show as potent an invagination-inducing activity as
LGA. This potent transforming activity can be mostly
accounted for by its high membrane anity. The num-
ber of LGA molecules required to cause this shape
7
change, which is estimated to be 10 molecules/cell, is
roughly the same as that of chlorpromazine.32

Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
2079
Consequently, the erythrocytes undergo colloid-osmotic
lysis. This mode of action of LGA on erythrocytes
accompanied by cell invagination is the ®rst reported
example for natural defense substances.
14-[N-(tert-Butoxycarbonyl)-ꢀ-alanyl]-4,10-bis[(2S)-2-
methylbutyryl]-4,10,14-triazatetradecanal (8). To a solu-
tion of 7 (320 mg, 0.535 mmol) in THF (15 mL) and
MeOH (0.1 mL) was added LiBH (60 mg, 2.8 mmol).
4
After stirring at room temperature overnight, 1 M aqu-
eous potassium sodium tartrate (30 mL) was added.
Extraction with CHCl and concentration a€orded a
Experimental
Chemistry
3
crude residue, which was then dissolved in dry CH Cl
2
2
(
added Et N (0.59 mL, 4.2 mmol) and SO -pyridine
10 mL) and DMSO (5 mL). To this solution were
General methods. 13CH I was purchased from Shoko
3
3
3
ꢀ
0 C for 2 h and then at room temperature for 1 h, the
Tsusho, Tokyo, Japan. All other chemicals were pur-
chased from Kanto Chemicals, Tokyo, Japan. NMR
spectra were recorded on a Bruker AM-500 spectro-
complex (307 mg, 1.93 mmol) at 0 C. After stirring at
ꢀ
reaction mixture was diluted with EtOAc (30 mL), and
washed successively with H O, 5% aqueous KHSO ,
saturated aqueous NaHCO , and brine. Concentration
1
13
meter at 500 MHz for H, 125 MHz for C with CHCl3
3
2
4
1
at 7.24 ppm, CHD OD at 3.30 ppm, CDCl at 77.0
2
3
3
1
3
ppm, or CD OD at 49.0 ppm. NMR data are those of
3
and ¯ash chromatography on silica gel (10 g, CHCl3:
MeOH, 95:5) a€orded 225.4 mg (76%) of 8 as a color-
equilibrated mixtures of conformers due to amide/ure-
thane bond rotation. Reactions requiring anhydrous con-
ditions were carried out in dry, freshly distilled solvents
under a nitrogen or argon atmosphere.
2
4
ꢀ
less oil: [a] +23.4 (c 1.39, MeOH); IR (®lm) 3315,
d
�
1 1
1710, 1625 cm ; H NMR (CDCl , 500 MHz) d 9.7±9.6
3
(m, 1H), 7.2±7.0 (br, 1H), 5.4±5.3 (br, 1H), 3.62±3.30
(m, 12H), 2.77±2.60 (m, 2H), 2.55±2.40 (m, 2H), 2.35
(m, 2H), 1.80±1.45 (m, 8H), 1.36 (m, 2H), 1.35 (brs,
Isolation and puri®cation of LGA (1). LGA was isolated
from the skin mucus of Aulacocephalus temmincki, a
species of soap®sh (Grammistidae), as reported pre-
9H), 1.02 (m, 6H), 0.80 (m, 6H); 13C NMR (CDCl , 125
3
MHz) 200.8, 199.4, 177.6, 176.8, 176.5, 176.5, 171.7,
156.0, 79.0, 48.4, 47.3, 45.5, 43.9, 43.0, 42.4, 40.5, 37.4,
37.3, 37.3, 37.2, 36.8, 36.2, 35.5, 29.6, 29.4, 29.2, 28.4,
27.3, 27.3, 24.1, 18.0, 17.9, 17.8, 17.7, 12.1, 12.1, 12.0.
2
1
viously. Brie¯y, the acetone extract of mucus scraped
from the ®sh skin was partitioned between water and
ethyl acetate, and the ethyl acetate layer was fractionated
through an alumina column (chloroform:methanol, 98:2
to 92:8) followed by repeated reversed-phase HPLC on
ODS to yield LGA (0.5% yield based on wet mucus).
8,14-Bis[(2S)-2-methylbutyryl]-4,8,14-triaza-17-hepta-
decanolactam (2). A solution of aldehyde 8 (28.9 mg,
0
.052 mmol) in dry CH Cl (1 mL) and TFA (1 mL)
2 2
N-(tert-Butoxycarbonyl)-ꢀ-alanine (6). b-Alanine (1.78
was stirred at room temperature for 20 min. After
removal of the solvents, the residue was dissolved in dry
MeOH (10 mL). MS3A (1 g) and Et N (14 mL) were
g, 20 mmol) was dissolved in dioxane:H O (2:1) (60
mL). To this solution were added 1 M aqueous NaOH
2
3
(
20 mL) and Boc O (4.7 mL, 20 mmol). The reaction
2
added to give pH 7. After stirring at room temperature
for 45 min, NaBH CN (12 mg, 0.19 mmol) was added.
mixture was stirred at room temperature for 30 min.
After evaporation of the solvent, the residue was adjus-
ted to pH 3 with 5% aqueous KHSO , and extracted
with EtOAc. Concentration and recrystallization from
hexane±EtOAc a€orded 6 (3.25 g, 87%) as white nee-
3
The mixture was stirred at room temperature overnight,
and ®ltered through a Celite pad, followed by the addi-
tion of saturated aqueous NaHCO . After insoluble
4
3
materials were ®ltered o€, the reaction mixture was
dried over Na SO , and the concentrated crude product
was chromatographed through an alumina column (neu-
1
dles: H NMR (CDCl , 500 MHz) d 11±8.5 (br, 1H),
3
3
2
4
13
.36 (br, 2H), 2.54 (br, 2H), 1.40 (brs, 9H); C NMR
(
CDCl , 125 MHz) 177.4, 156.0, 79.7, 35.9, 34.4, 28.3.
3
tral, activity II, 10 g, CHCl :MeOH, 9:1) and further
3
puri®ed by HPLC (SiO , YMC A-024, CHCl :MeOH:
3
2
Ethyl 14-[N-(tert-butoxycarbonyl)-ꢀ-alanyl]-4,10-bis[(2S)-
2-methylbutyryl]-4,10,14-triazatetradecanoate (7). To a
ꢁ
solution of crude 5 HCl (41.9 mg, 0.090 mmol) in dry
i-PrNH , 9:1:0.2) to a€ord 3.0 mg (14%) of 2 as a color-
2
1
less oil: H NMR (CDCl , 500 MHz) d 4.09 and 3.99
3
(q each, 2H, J=7 Hz), 3.45 (m, 2H), 3.40 (m, 2H), 3.35±
3.13 (m, 4H), 2.58 (m, 2H), 2.54±2.36 (m, 2H), 1.63±1.40
(m, 6H), 1.30 (m, 2H), 1.20 (m, 2H), 1.15 and 1.12 (t
each, 3H, J=7 Hz), 0.97 (m, 6H), 0.74 (m, 6H); MS (EI)
m/z (rel. intensity) 438 (32), 423 (7), 381 (13), 353 (100),
CH Cl (2 mL) were added Et N (15 mL, 0.11 mmol),
2
2
3
HOBt (21 mg, 0.14 mmol), WSCI (30 mg, 0.16 mmol),
and N-Boc-b-alanine (6) (23 mg, 0.12 mmol). After
stirring at room temperature overnight, the solvent was
evaporated. The crude residue was dissolved in EtOAc,
+
57 (87); HRMS (EI) m/z calcd for C H N O (M )
3
24
46
4
and washed successively with 5% aqueous KHSO , satu-
4
438.3570, found 438.3610.
rated aqueous NaHCO , and brine. Concentration and
3
column chromatography of the residue on silica gel (2 g,
CHCl :MeOH, 98:2) a€orded 7 (35.3 mg, 65%) as a col-
Preparation of acetylated analogue (3) from LGA. To a
solution of LGA (295 mg, 0.5 mmol) in dry pyridine (250
3
2
7
ꢀ
732, 1713, 1641, 1633 cm ; H NMR (CDCl , 270
ꢀ
stirring at room temperature for 15 h, complete removal
orless oil: [a] +15.0 (c 0.753, CHCl ); IR (®lm) 3315,
mL) was added Ac O (10 mL, 100 mmol) at 0 C. After
d
3
2
�
1
1
1
3
MHz) d 7.03 (br, 1H), 4.13 and 4.09 (each q, 2H, J=7.0
Hz), 3.57 (m, 2H), 3.42±3.07 (m, 10H), 2.57 and 2.53 (each
t, 2H, J=7 Hz), 2.60±2.42 (m, 2H), 2.38 (t, 2H, J=7 Hz),
of the solvent and ¯ash chromatography on silica gel
(100 mg, CHCl :MeOH, 95:5) a€orded 308 mg (97%) of
3
3 as a yellow oil. HRMS (FAB, 3-nitrobenzylalcohol
+
1
1
.70±1.50 (m, 8H), 1.39 (s, 9H), 1.25 (m, 2H), 1.25 and
.21 (each t, 3H, J=7.0 Hz), 1.07 (m, 6H), 0.85 (m, 6H).
matrix) m/z calcd for C H N O ([M+H] ) 633.5319,
4
3
7
69
4
1
found 633.5292. Acetyl signals in H NMR of the acetyl

2
080
Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
group (CD OD, 500 MHz) d 2.12 (s), 2.15 (s), 2.20 (s),
of a solution of 10% (v/v) polyethylene glycol mono-p-
octaphenyl ether was taken as 100% leakage.
3
2.22 (s), 2.48 (s), 2.50 (s), 2.54 (s), 2.55 (s). This signal
multiplication is due to restricted rotation around three
tertiary amide groups, since mutual saturation transfers
were observed among these signals.
Hemolysis tests. After erythrocytes were diluted with the
same bu€er as in Erythrocytes to a known concentra-
tion, 1.25 mL of a solution of the tested substance in
MeOH was added to a 200-mL aliquot of the suspen-
Preparation of acyclic analogue (4) from LGA. To a
solution of LGA (301 mg, 0.5 mmol) in dry MeOH was
added KHCO (2 mg, 20 mmol), and then CH I (5 mL,
ꢀ
sion. Cells were incubated at 37 C and sedimented by
centrifugation, and then 100 mL of the supernatant was
transferred for measurement of the concentration of
hemoglobin at 420 or 540 nm. One hundred percent
hemolysis was obtained by incubating the cells with a
greater than 20-fold volume of distilled water for the
same period. The sediments were subjected to the mor-
phological observations described below.
3
3
ꢀ
8
0 mmol) slowly at 0 C. After stirring under darkness at
room temperature overnight, the reaction mixture was
concentrated, and CHCl extract was subjected to ¯ash
3
chromatography on silica gel (100 mg, CHCl :MeOH:
3
i-PrOH, 15:5:1) and further puri®ed by HPLC (ODS,
YMC AM-323, CH CN:H O:TFA, 60:40:0.05) to give
3
2
170 mg (40%) of 4 as a TFA salt. HRMS (FAB, 3-
nitrobenzylalcohol matrix) m/z calcd for C H N O
+
Thermodynamic analysis of hemolysis. Dose-dependence
curves for substance-induced hemolysis of erythrocytes
at di€erent cell densities may re¯ect not only the intrin-
sic hemolytic activity of the added sample but also the
degree of its distribution in the cells. To evaluate the
intrinsic hemolytic ecacy of a substance that is
3
8
73
4
3
(
M ) 633.5683, found 633.5710. UV (MeOH) l
208
max
3
nm, (e=9.0Â10 ), a,b-unsaturated amide. For the ana-
1
3
13
logue obtained using CH I, a tri- C-methyl ammoni-
3
13
um group was characterized in the C NMR spectrum
CD OD, 125 MHz) d 53.5 (brs).
(
3
assumed to be incorporated equally into cells, the fol-
lowing calculations were applied.1
balance equation is:
2,25,37
The material
Biological assay
Materials and instruments. EggPC was purchased from
Sigma, MO, USA. Calcein and EDTA were purchased
form Dojindo, Kumamoto, Japan. Polyethylene glycol
mono-p-octaphenyl ether was purchased from Tokyo
Chemical Industry, Tokyo, Japan. All other chemicals
were purchased from Kanto Chemicals, Tokyo, Japan.
Fluorescence was recorded on a Hitachi 650-60 spec-
trophotometer. Cell morphology was observed under a
phase-contrast microscope, OLYMPUS BX-40.
‰
SŠ ˆ ‰SŠ ‡ ‰SŠ ˆ ‰SŠ ‡ r‰erythrocyteŠ
t
f
b
f
where the total concentration of added substance, [S]t,
consists of the fraction in the test media, [S] , and the
cell-bound fraction, [S] , at the speci®c erythrocyte
b
density, [erythrocyte]. On the assumption that the per-
centage of hemolysis is determined solely by the average
bound amount of substance molecules per cell,
f
‰SŠ
b
r ˆ ‰
, linearity between [S] and [erythrocyte] is
t
erythrocyteŠ
Erythrocytes. Freshly drawn, heparinized human blood
was washed three times with Tris±HCl bu€er (150 mM
NaCl, 10 mM Tris/HCl, 1 mM EDTA, 22 mM glucose,
pH 7.4) to remove plasma and bu€y coat. In each
experiment, the erythrocytes were washed more than two
times with the same bu€er and resuspended in it. The cell
density was determined by visually counting the cells in
expected at any given percentage of hemolysis. Since [S]
t
and [erythrocyte] are measurable, the values r and [S]f
can be obtained as the slope and intercept, respectively,
by plotting the [S] required for each percentage of
t
hemolysis against [erythrocyte]. Thus, the relationship
between r and the percentage hemolysis can be obtained
as the intrinsic hemolytic activity of a substance against
0.01 mL of the suspension under a light microscope.
[erythrocyte]. The relationship between r and [S] , which
f
is known as the binding isotherm, can be obtained in the
same manner. The partition coecient is obtained from
Preparation of liposomes. SUVs were prepared from a
lipid ®lm of EggPC that had been thoroughly dried in
vacuo and then hydrated with the Tris±HCl bu€er con-
taining 60 mM calcein. After vortex mixing and ultra-
sonication at 0 C under a nitrogen stream for 60 min, the
resultant SUVs were separated from untrapped calcein
and multilamellar vesicles by gel ®ltration (Sepharose
r
the slope of the binding isotherm ( ˆ ‰ ) by converting
SŠf
the unit of r to molarity in erythrocyte membrane.
ꢀ
Morphological observation of erythrocytes undergoing
lysis. To a small amount (ca. 5 mL) of the erythrocyte
suspension was added 5 mL of 0.9% glutaraldehyde
solution in Tris±HCl bu€er (150 mM NaCl, 10 mM
Tris/HCl, 1 mM EDTA, 22 mM glucose, pH 7.4). The
4
B, 2Â40 cm, 10 mM Tris±HCl bu€er solution with 150
mM NaCl and 1 mM EDTA as an eluent).
®xed erythrocytes were observed microscopically. The
Measurements of ¯uorescence leakage. An aliquot of the
SUV suspension was diluted with 2 mL of the Tris±HCl
bu€er to a certain lipid concentration in a 3-mL cuvette
for ¯uorescence measurements. After the addition of a
sample solution in 10 mL of MeOH, calcein leakage
from SUVs was monitored ¯uorometrically (excitation
at 490 nm and emission at 520 nm). The ¯uorescence
increase reached a plateau within 2 min in all cases. The
changes in the shapes of erythrocytes were quanti®ed
using the morphological index de®ned by Fujii and co-
workers.26
Acknowledgements
The authors gratefully acknowledge Professor Y. Ume-
zawa and Dr K. Tohda in their department and Pro-
fessor M. Sugawara, Nihon University, Tokyo, Japan,
¯
uorescence intensity obtained by the addition of 20 mL

Y. Kobayashi et al. / Bioorg. Med. Chem. 7 (1999) 2073±2081
2081
who allowed us to use their equipment and provided
helpful suggestions. The authors are also grateful to
Professor M. Murata, Osaka University, Osaka, Japan,
for his valuable discussion and critical reading of this
manuscript.
18. Yokota, H.; Shimakura, K.; Nagashima, Y.; Shiomi, K.
Spring meeting of the Japanese Society of Fisheries Science,
Tokyo, 1998, Abstract 1054.
19. Shiomi, K.; Igarashi, T.; Shimakura, K.; Nagashima, Y.
Autumn meeting of the Japanese Society of Fisheries Science,
Hiroshima, 1997, Abstract 832.
2
0. Yokota, H.; Shimakura, K.; Nagashima, Y.; Shiomi, K.
Autumn meeting of the Japanese Society of Fisheries Science,
Hakodate, 1998, Abstract 823.
References
1
. Tachibana, K. In Bioorganic Marine Chemistry; Scheuer, P.
J., Ed.; Springer-Verlag: Berlin, 1988; Vol. 2, pp. 117±138.
. Nair, M. S. R. In Handbook of Natural Toxins; Tu, A. T.,
Ed.; Marcel Dekker: New York, 1988; Vol. 3, pp. 211±226.
. Hashimoto, Y. Marine Toxins and Other Bioactive Marine
Metabolites; Japan Scienti®c Societies Press: Tokyo, 1979; pp.
21. Onuki, H.; Tachibana, K.; Fusetani, N. Tetrahedron Lett.
1993, 34, 5609±5612.
22. Onuki, H.; Ito, K.; Kobayashi, Y.; Matsumori, N.;
Tachibana, K.; Fusetani, N. J. Org. Chem. 1998, 63, 3925±
3932.
23. Chen, F. C. M.; Benoiton, N. L. Can. J. Chem. 1976, 54,
3310±3311.
24. Beutler, E. In Hematology, 4th ed. Williams, W. J.; Beutler,
E.; Erslev, A. J.; Lichtman, M. A., Eds.; McGraw-Hill: New
York, 1990; pp. 317±322.
25. Matsuzaki, K.; Nakai, S.; Handa, T.; Takaishi, Y.; Fujita,
T.; Miyajima, K. Biochemistry 1989, 28, 9392±9398.
26. Fujii, T.; Sato, T.; Tamura, A.; Wakatsuki, M.; Kanaho,
Y. Biochem. Pharmacol. 1979, 28, 613±620.
27. Katsu, T.; Ninomiya, C.; Kuroko, M.; Kobayashi, H.;
Hirota, T.; Fujita, Y. Biochim. Biophys. Acta 1988, 939, 57±63.
28. Sheetz, M. P.; Singer, S. J. Proc. Natl. Acad. Sci. USA
1974, 71, 4457±4461.
29. Schrier, S. L.; Zachowski, A.; Devaux, P. F. Blood 1992,
79, 782±786.
30. Loh, R. K.; Huestis, W. H. Biochemistry 1993, 32, 11722±
11726.
2
3
3
4
5
1
6
12±332.
. Boylan, D. B.; Scheuer, P. J. Science 1967, 155, 52±56.
. Yoshikawa, M.; Sugimura, T.; Tai, A. Agric. Biol. Chem.
989, 53, 37±40.
. Thompson, S. A.; Tachibana, K.; Nakanishi, K.; Kubota, I.
Science 1986, 233, 341±343.
. Thompson, S. A.; Tachibana, K.; Kubota, I.; Zlotkin, E. In
Peptide Chemisty 1987; Shiba, T.; Sakakibara, S., Eds.; Protein
7
Research Foundation: Osaka, 1988; pp. 127±132.
8
9
2
1
1
1
1
2
1
1
1
1
. Shai, Y. Toxicology 1994, 87, 109±129.
. Tachibana, K.; Sakaitani, M.; Nakanishi, K. Science 1984,
26, 703±705.
0. Tachibana, K.; Sakaitani, M.; Nakanishi, K. Tetrahedron
985, 41, 1027±1038.
1. Tachibana, K.; Gruber, S. H. Toxicon 1988, 26, 839±853.
2. Ohnishi, Y.; Tachibana, K. Bioorg. Med. Chem. 1997, 5,
251±2265.
3. Abdul-Haqq, A. J.; Shier, W. T. J. Toxicol., Toxin Rev.
991, 10, 289±320.
31. Thimon, L.; Peypoux, F.; Exbrayat, J. M.; Michel, G.
Cytobios 1994, 79, 69±83.
32. Seeman, P. Pharmacol. Rev. 1972, 24, 583±655.
33. Moore, S. S.; Tarnowski, T. L.; Newcomb, M.; Cram, D.
J. J. Am. Chem. Soc. 1977, 99, 6398±6405.
4. Hashimoto, Y.; Kamiya, H. Toxicon 1969, 7, 65±70.
5. Randall, J. E.; Aida, K.; Hibiya, T.; Mitsuura, N.;
Kamiya, H.; Hashimoto, Y. Publ. Seto Mar. Bio. Lab. 1971,
1
1
34. Ito, T.; Radecka, H.; Tohda, K.; Odashima, K.; Umezawa,
Y. J. Am. Chem. Soc. 1998, 120, 3049±3059.
35. Bielawski, J. Biochim. Biophys. Acta 1990, 1035, 214±217.
36. Chernitsky, E. A.; Senkovich, O. A. Membr. Cell Biol.
1997, 11, 475±485.
9, 157±190.
6. Oshima, Y.; Shiomi, K.; Hashimoto, Y. Bull. Jpn. Soc. Sci.
Fish. 1974, 40, 223±230.
7. Shiomi, K.; Igarashi, T.; Shimakura, K.; Nagashima, Y.
Spring meeting of the Japanese Society of Fisheries Science,
1
37. Hu, M.; Konoki, K.; Tachibana, K. Biochim. Biophys.
Acta 1996, 1299, 252±258.
Tokyo, 1997, Abstract 1005.