Curvature-Dependent Recognition of Ethanolamine Phospholipids by
Duramycin and Cinnamycin
Kunihiko Iwamoto,* Tomohiro Hayakawa,yMotohide Murate,* Asami Makino,* Kazuki Ito,zTetsuro Fujisawa,z
and Toshihide Kobayashi*y§
*Supra-Biomolecular System Research Group, RIKEN (Institute of Physical and Chemical Research) Frontier Research System, Saitama,
Japan;yLipid Biology Laboratory, RIKEN, Saitama, Japan;zRIKEN SPring-8 Center, Hyogo, Japan; and§INSERM UMR 870, INRA U1235,
INSA-Lyon, University Lyon 1 and Hospices Civils de Lyon, Villeurbanne, France
bind phosphatidylethanolamine (PE). The lipid specificity of duramycin was not established. The present study indicates that
both duramycin and cinnamycin exclusively bind to ethanolamine phospholipids (PE and ethanolamine plasmalogen). Model
membrane study indicates that the binding of duramycin and cinnamycin to PE-containing liposomes is dependent on membrane
curvature, i.e., the lantibiotics bind small vesicles more efficiently than large liposomes. The binding of the lantibiotics to
multilamellar liposomes induces tubulation of membranes, as revealed by electron microscopy and small-angle x-ray scattering.
These results suggest that both duramycin and cinnamycin promote their binding to the PE-containing membrane by deforming
Duramycin is a 19-amino-acid tetracyclic lantibiotic closely related to cinnamycin (Ro09-0198), which is known to
Duramycin is a 19-amino-acid tetracyclic peptide produced
by Streptoverticillium cinnamoneus and is closely related to
cinnamycin (Ro09-0198) (Fig. 1 A) (1–5). Both compounds
characterized by the presence of a high proportion of unusual
phosphatidylethanolamine (PE) (6–8). Because of this char-
acteristic, cinnamycin has been employed to study the
distribution and metabolism of PE (9–14). Duramycin is
specificity of duramycin is not well established. Previously it
was proposed that duramycin recognizes a particular mem-
brane conformation determined by the presence of PE or
monogalactosyl diglyceride (MGDG) (15). Analysis of the
membranes of the duramycin-resistant Bacillus subtilis
mutants revealed that they had little or no PE and cardiolipin
(15,16). In contrast, mutation of alkalophilic Bacillus firmus
to duramycin resistance resulted in a substantial replacement
of PE by its plasmalogen form (17).
In eukaryotic cells, PE is mainly restricted to the inner
leaflet of the plasma membrane (19–21). Recently we showed
that cinnamycin induces transbilayer phospholipid move-
ment of target cells in a PE-dependent manner (8). This
causes exposure of the inner leaflet PE to the peptide and
promotes binding of cinnamycin. When the surface concen-
tration of PE is high, cinnamycin induces membrane reor-
ganization such as membrane fusion and the alteration of the
membrane gross morphology (8). However, the detailed
membrane ultrastructure induced by cinnamycin binding is
not well determined. Although duramycin was known to alter
the membrane permeability of mammalian cells (18,22,23),
the precise mechanism(s) of duramycin-induced membrane
damage is not yet determined.
cinnamycin with model membranes. The results indicate that
both duramycin and cinnamycin selectively bind ethanola-
mine phospholipids, irrespective of whether they are of
diacyl- or plasmalogen type. The binding of the lantibiotics
induces reorganization of the membrane into highly curved
tubular structures as revealed by electron microscopy and
small-angle x-ray scattering (SAXS). In addition, we found
the lantibiotics preferentially bind PE in the highly curved
membranes. Thus, both duramycin and cinnamycin promote
ment and by changing membrane curvature.
MATERIALS AND METHODS
The following were purchased from Avanti Polar Lipids (Alabaster, AL):
L-a-phosphatidylcholine (egg, chicken; egg PC).
L-a-phosphatidylethanolamine (egg, chicken; egg PE).
L-a-phosphatidylethanolamine (liver, bovine; liver PE).
Submitted November 21, 2006, and accepted for publication April 27, 2007.
Address reprint requests to Toshihide Kobayashi, Tel.: 81-48-467-9534;
Editor: Michael Edidin.
? 2007 by the Biophysical Society
1608 Biophysical JournalVolume 93 September 20071608–1619
1-oleoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine (18:1 lyso-PE).
L-a-phosphatidylserine (brain, porcine; brain PS).
L-a-phosphatidylinositol (liver, bovine; liver PI).
L-phosphatidyl-DL-glycerol (egg, chicken; egg PG).
L-a-phosphatidic acid (egg, chicken; egg PA).
glycerol (18:1 BMP).
cardiolipin (heart, bovine; heart CL).
chicken; egg SM).
porcine; brain SM).
Total cerebrosides (brain, porcine; GalCer).
From Matreya (Pleasant Gap, PA):
Glucosylceramide (human; GlcCer), lactosylceramide (LacCer), and
monogalactosyl diglyceride (plant, hydrogenated; 18:0 MGDG).
From Larodan Fine Chemicals (Malmo ¨, Sweden):
Monogalactosyl diglyceride (plant MGDG).
From Wako Pure Chemical Industries (Osaka, Japan):
Ganglioside GM1(bovine brain; GM1), ganglioside GM2(NeuAc) (bovine
brain; GM2), and ganglioside GM3(NeuAc) (bovine; GM3).
From Sigma (St. Louis, MO):
1,2-didodecanoyl-sn-glycero-3-phosphoethanolamine (DLPE), 1,2-dimyr-
istoyl-sn-glycero-3-phosphoethanolamine (DMPE), and duramycin and
Abu, a-aminobutyric acid; Ala, alanine; Asn, asparagine; Asp, aspartic acid; Gln, glutamine; Gly, glycine; Lys, lysine; Phe, phenylalanine; Pro, proline; Val,
valine. Ala6is linked to Lys19as lysinoalanine. Ala-S-Ala: lanthionine, Ala-S-Abu: b-methyllanthionine, X2: lysine (duramycin) or arginine (cinnamycin)
(4,5). (B) Rabbit erythrocytes (final 3 3 107cells/ml) were incubated with various concentrations of duramycin for 30 min at 4?C or 37?C. Hemolysis was
measured as described in Materials and Methods.(C) Duramycinwas preincubated with various concentrations of MLVs composed of POPC,90 mol % POPC
and 10 mol % SOPE, or 90 mol % POPC and 10 mol % C18(plasm)-18:1 PE for 1 h at 37?C. The mixtures (final concentration of duramycin was 5 mM) were
then added to rabbit erythrocytes (final 3 3 107cells/ml) and further incubated for 30 min at 37?C, followed by the measurement of hemolysis. Horizontal axis
indicates the final concentration of the total lipids in MLVs. (D) Duramycin was preincubated with MLVs containing 90 mol % POPC and 10 mol % of
indicated lipids, followed by the measurement of hemolysis, as described in panel C. Final concentrations of duramycin and total lipids were 5 mM and
500 mM, respectively. Data are means 6 SD of at least three independent experiments.
Ethanolamine phospholipids inhibit hemolytic activity of duramycin. (A) Structure of duramycin and cinnamycin (Ro09-0198). Abbreviations:
Curvature-Dependent PE-Binding Peptides 1609
Biophysical Journal 93(5) 1608–1619
From Nacalai Tesque (Kyoto, Japan):
Sodium hydrosulfite (sodium dithionite), o-phenylenediamine, and
Dulbecco’s phosphate-buffered saline (?) (PBS). (The PBS, pH
7.4, contained 200 mg/l potassium chloride, 200 mg/l potassium
dihydrogenphosphate, 8000 mg/l sodium chloride, and 1150 mg/l di-
From CovalAb (Lyon, France):
Rabbit polyclonal antisera against duramycin.
Preparation of lipid vesicles
Multilamellar vesicles (MLVs) were prepared by hydrating a lipid film with
20 mM HEPES-NaOH (pH 7.4) and 100 mM NaCl, unless otherwise
indicated, and vortex mixing. To prepare large vesicles, MLVs were sub-
jected to extrusion through polycarbonate filters (Nucleopore, Maidstone,
UK) for 25 times using a two-syringe extruder (Avanti Polar Lipids,
Alabaster, AL). To prepare small vesicles, MLVs were subjected to
sonication using a XL-2020 sonicator (Misonix, Farmingdale, NY) until
peak diameters (nm) of the size distribution by number became ,50 nm.
The vesicle size was examined by dynamic light scattering measurements at
37?C using a Zetasizer Nano ZS (Malvern, Worcestershire, UK). In Fig. 5,
the vesicles were also examined by freeze-fracture electron microscopy.
Measurement of the amount of phospholipids residing in the
outer leaflet of the most external layer of liposomes
Two milliliters of the liposome suspensions (total 50 mM phospholipids)
containing 1 mol % N-NBD-PE were mixed with 20 ml of 1 mM sodium
hydrosulfite at 25?C and the fluorescence was measured using an FP-6500
spectrofluorometer (Jasco, Tokyo, Japan) with excitation and emission
quenches N-NBD-PE, which is localized in the outer leaflet of the most
external layer of the liposomes (24). The fluorescence was monitored until it
Measurement of hemolysis
Rabbit erythrocytes were prepared by washing rabbit whole blood (Nippon
Bio-Supply Center, Tokyo, Japan) with PBS. Measurement of hemolysis of
hemolysis was determined by incubating rabbit erythrocytes (3 3 107cells/
ml) on ice for 30 min, whereas 100% hemolysis was measured after three
on hemolytic activity of duramycin were examined as described previously
(25) with some modifications. In brief, 40 ml of duramycin solution in PBS
without 100 mM NaCl were mixed and incubated for 1 h at 37?C or at 4?C.
After the addition of 160 ml of rabbit erythrocyte suspensions in PBS, the
was measured. Final concentrations of erythrocytes, duramycin and lipo-
somes in the mixture are indicated in the figure legends. In the case of using
determined by measuring phosphorus content (26).
High-sensitivity titration calorimetry
Isothermal titration calorimetry (ITC) was performed using a Microcal
VP-ITC titration calorimeter (MicroCal, Northampton, MA) as described
previously (27) with some modifications. Duramycin was titrated with
vesicles at 37?C. Injection volumes were 8 ml. The calorimeter cell had a
reaction volume of 1.4034 ml. As duramycin was known to give a positive
Biuret test (2), the concentrations of duramycin were measured by using
BCA ProteinAssayReagent(Pierce Biotechnology,Rockford,IL)and/or by
determining dry weight of duramycin. The lipid concentrations of liposomes
were determined by measuring phosphorus content by phosphorus assay
(26). The heats of dilution were determined in the control experiments using
the buffer (20 mM HEPES-NaOH (pH 7.4) and 100 mM NaCl) in place of
duramycin solutions, and were subtracted from the heats determined in the
corresponding duramycin-lipid binding experiments. Binding constants (Ka)
of duramycin and ethanolamine phospholipids were calculated as
assuming duramycin and ethanolamine phospholipid form 1:1 complex (see
Results). Binding constants were estimated from ITC results by curve-fitting
analysis with ‘‘One Set of Sites’’ model from Origin, Ver. 5.0 (MicroCal,
Liposome binding assay using gel filtration
Various liposomes were incubated with duramycin for 30 min at 37?C in 20
mM HEPES-NaOH (pH 7.4), 100 mM NaCl buffer. After incubation,
liposome-bound duramycin was separated from free duramycin by gel fil-
tration as described previously (29,30) with some modifications. Five mil-
liliters polypropylene column (Pierce Biotechnology) was filled with 3 ml of
Bio-Gel A-15m Gel (Bio-Rad, Hercules, CA) equilibrated with the buffer.
After applying 100 ml of the reaction mixture, 200 ml of the buffer was
added and the eluent was collected. This step was repeated until majority of
duramycin was eluted from the column. Each fraction was analyzed for the
amounts of liposomes and duramycin. For quantification of the liposomes,
50 ml of each fraction was diluted with 50 ml of the buffer, followed by the
addition of 10 ml 10% Triton X-100. The fluorescence was measured using
an ARVO SX Multilabel Counter (Wallac, Turku, Finland) with excitation
and emission wavelengths at 485 and 535 nm, respectively. Duramycin was
quantified by enzyme-linked immunosorbent assay.
Enzyme-linked immunosorbent assay of duramycin
A fifty-microliter sample was added to each well of an Immulon 2HB
(Thermo Fisher Scientific, Waltham, MA) microtiter plate. After overnight
(TBS; 10 mM Tris-HCl, pH7.4, 150 mM NaCl). Two-hundred microliters of
30 mg/ml bovine serum albumin (Fraction V; Sigma, St. Louis, MO) in TBS
was then added to each well. After 2-h incubation at room temperature, the
wells were washed with TBS. The bound duramycin was detected by adding
anti-duramycin antiserum followed by incubation with ECL anti-rabbit IgG,
horseradish peroxidase-linked species-specific whole antibody (from don-
key; GE Healthcare UK, Buckinghamshire, UK). The intensity of the color
developed with o-phenylenediamine as a substrate was measured with a
Microplate Reader model 680 (Bio-Rad), reading the absorption at 490 nm
with reference at 630 nm.
Negative staining was performed as reported previously (8) with some
modifications. MLVs containing 2 mM total lipids were incubated with
various concentrations ofduramycinfor 30minat 37?C.After centrifugation
at 19,000 3 g for 30 min at 4?C, the pellet was suspended in the same
volume of the buffer containing 20 mM HEPES-NaOH (pH 7.4) and 100
mM NaCl buffer, fixed with 2.5% glutaraldehyde for 30 min at room
temperature and washed three times with the same buffer by centrifugation
at 19,000 3 g for 30 min at 4?C. For negative staining images, the
1610 Iwamoto et al.
Biophysical Journal 93(5) 1608–1619
suspensions were adsorbed onto poly-D-lysine-treated formvar-coated grids
and negatively stained with 2% sodium phosphotungstic acid. For freeze
fracture images, the samples were frozen in liquid propane cooled by liquid
nitrogen, fractured in a freeze-etching machine (Balzers BAF400T, Balzers,
Liechtenstein) at ?110?C, and replicated by platinum/carbon. Replicated
sampleswereimmersedinhouseholdbleach todissolvethelipids, washedin
water, and then mounted on formvar-coated copper grids. Both specimens
for negative staining and freeze fracture images were examined under trans-
mission electron microscope (Tecnai 12, Philips, Eindhoven, The Netherlands,
or 1200EX-II, JEOL, Tokyo, Japan). Electron micrographs recorded on im-
aging plates were scanned and digitized by an FDL 5000 imaging system
(Fuji Photo Film, Tokyo, Japan).
SAXS measurements were carried out at RIKEN Structural Biology
Beamline I (BL45XU) at SPring-8, 8 GeV synchrotron radiation source
(Hyogo, Japan) (31). The x-ray wavelength used was 0.9 A˚and the beam
size at the sample position was ;0.4 3 0.7 mm2. The distance of sample-to-
detector was 968 mm. Samples were measured in a sample cell with a path
length of 1.5 mm and a pair of thin quartz windows (30 mm thickness). The
sample temperature was controlled to 37 6 0.01?C with a high precision
thermoelectric device. The samples were allowed to equilibrate for at least
5 min and the diffraction collected with 2 min exposure. Buffer profiles were
also taken for background subtraction purposes. The SAXS patterns were
recorded with an imaging plate (30 3 30 cm2) system of RIGAKU R-Axis
IV21(32). The two-dimensional scattering patterns were circularly aver-
aged and reduced to one-dimensional profiles using FIT2D, Ver. 12.012
(http://www.esrf.fr/computing/scientific/FIT2D/), a two-dimensional data
reduction and analysis program. The reciprocal spacing (s) and scattering
s ¼ 1=d ¼ ð2=lÞsinu;
q ¼ 2ps ¼ ð4p=lÞsinu;
where d is the lattice spacing, 2u is the scattering angle, and l is the
wavelength of x ray, were calibrated with silver behenate by the long-period
spacing of 5.838 nm (34).
Modeling analysis of SAXS data
Small-angle scattering intensity, I(q), can be described as
IðqÞ ¼ kNPðqÞ;
where k is an instrument constant, N is the number density of particles, and
P(q) is the particle scatteringfunction.Of possible structuraltypestested, the
SAXS data of duramycin-membrane complexes was well fitted by a concen-
tric cylindrical shell model. The scattering function of a randomly oriented
cylindrical particle composed of ithshells is expressed as
where F(q) is the form factor of the cylinder with ithshells and 2H is the
length of the cylinder. The values Dri, Vi, and Ricorrespond to the average
excess electron density (contrast), volume, and radius of ithshell, respec-
tively. The value x is the angle between the longest particle axis and the
scattering vector q, and J1corresponds to the first-order Bessel function.
Based on the observation by the electron microscopy, the length of the
straight part of the duramycin-membrane rod was estimated to be ;150 nm,
which is much larger than the rod radius (;10 nm) and is out of limit of our
experimental resolution. Therefore, we set the 2H value with 150 nm for
this fitting analysis. In case of such large H, the axial factor in Eq. 4 drops
to zero very rapidly, unless those orientations where x is very small. This
means that the rods make a contribution to the scattering only when they
are nearly perpendicular to the scattering vector q. That is, the I(q) remains
nearly unaltered with a slight change of H (35,36) in q that we observed.
Duramycin specifically binds
Whereas the specific binding of cinnamycin to PE is estab-
lished, the lipid specificity of duramycin is obscure. Previ-
ously it was proposed that duramycin recognizes a particular
membrane conformation determined by the presence of PE
or MGDG (15). Studies on duramycin-resistant bacteria sug-
gest that duramycin binds to PE (15–17), but not to ethanol-
amine plasmalogen (17). In the present study, we examined
the lipid specificity of duramycin by 1), examination of the
inhibitory effects of liposomes from various lipids on the
toxicity of duramycin; and 2), heat measurement during in-
teraction of duramycin and the liposomes by ITC. Similar to
cinnamycin (6), duramycin exhibited temperature-dependent
hemolytic activity against rabbit erythrocytes (Fig. 1 B).
When duramycin was preincubated with POPC MLVs, the
hemolytic activity was not affected (Fig. 1, C and D). In
contrast, preincubation with PE-containing MLVs inhibited
the hemolytic activity, suggesting that duramycin binds PE
(Fig. 1, C and D). Similar inhibitory effects were observed
with ethanolamine plasmalogen-containing MLVs (Fig. 1, C
and D). Similar to duramycin, cinnamycin-induced hemo-
lysis was inhibited by PE- and ethanolamine plasmalogen-
containing liposomes (Supplementary Material, Fig. S1),
suggesting that cinnamycin also binds the plasmalogen form
of ethanolamine phospholipids. In Fig. 1 D, we examined the
effects of MLVs containing 90 mol % POPC and 10 mol %
of various lipids on duramycin-induced hemolysis. Hemo-
lysis was inhibited by the presence of PEs with different fatty
acids, ethanolamine plasmalogens, and lyso-PE. However,
other lipids, including MGDG, did not affect the hemolysis,
suggesting that duramycin specifically binds ethanolamine
The interaction of duramycin with ethanolamine phos-
pholipids was then examined by ITC. Whereas injections of
POPC (Fig. 2 A) or POPC/MGDG (9:1) (Fig. 2 D) lipo-
somes to duramycin solution exhibited only slight exother-
mic reactions, each injection of POPC/POPE (9:1) (Fig. 2 B)
or POPC/C16(plasm)-18:1 PE (9:1) (Fig. 2 C) liposome
suspension caused a distinct exothermic reaction. The re-
action enthalpy DH? was calculated from the ITC profile. In
Fig. 2 B, the total amount of duramycin in the sample cell
Curvature-Dependent PE-Binding Peptides 1611
Biophysical Journal 93(5) 1608–1619
was nd0¼ 28:2 nmol and the total heat measured until the
equivalent point was +12
thalpy was thus calculated as DH? ¼ +hi=nd0¼ ?5:0 kcal/
mol duramycin. Similarly, the reaction enthalpy DH? was
calculated as DH? ¼ +hi=nd0¼ ?3:8 kcal/mol duramycin
for POPC/C16(plasm)-18:1 PE liposome suspension (Fig.
2 C). In Fig. 2 B, the amount of PE injected in the first 12
steps was 53.3 nmol. Thus, duramycin/PE ratio was 1:1.89
when the peptide was consumed. Cinnamycin is reported to
form 1:1 complex with PE (37,38). Assuming duramycin also
forms 1:1 complex, our results suggest that only the PE in
the outer leaflet is accessible to duramycin when the PE con-
tent was 10 mol %. When the PE content of POPC/POPE
liposomes exceeded 20 mol %, duramycin bound more than
50 mol % of the total PE (data not shown), as reported on
cinnamycin (38). The binding constant of duramycin and
POPE and that of duramycin and C16(plasm)-18:1 PE were
estimated as Ka¼ (2.1 6 0.4) 3 108M?1and Ka¼ (1.1 6
0.2) 3 108M?1, respectively.
1hi¼ ?142 mcal. The reaction en-
The binding of duramycin and cinnamycin to PE is
dependent on the physical properties and the
curvature of the membrane
We examined the effect of physical properties of PE-containing
liposomes on the inhibition of duramycin-induced hemoly-
sis. As shown in Fig. 1 D, POPC/DSPE MLVs inhibited the
hemolytic activity of duramycin. In contrast, DSPC/DSPE
(9:1) MLVs did not inhibit duramycin-induced hemolysis
(Fig. 3 A). DSPC/POPE (Tm¼ 20?C (39)) and DSPC/DOPE
(Tm¼ ?16?C (39)) inhibited hemolysis (Fig. 3 A). It is
speculated that DSPC (Tm¼ 54.5?C)/DSPE (Tm¼ 74?C)
(39) provides tightly packed surfaces at 37?C. These results
suggest that the binding of duramycin to PE is dependent on
the physical properties of the membrane. In Fig. 3 B, the
effect of membrane curvature of DSPC/DSPE liposomes on
the inhibition ofduramycin-induced hemolysis wasexamined.
The parentheses in the figure legend indicate the diameters
of the liposomes examined by dynamic light scattering. In
contrast to MLVs, both small vesicles and large vesicles
inhibited hemolysis, small vesicles being more effective than
large vesicles. Similar results were obtained with cinnamycin
(Supplementary Material, Fig. S2).
The curvature-dependent binding of duramycin to DSPC/
DSPE membranes was then measured by ITC. Injections of
DSPC/DSPE (9:1) large vesicles of ;700 nm diameters into
duramycin solution resulted in slight exothermic reactions
(Fig. 4 B). In contrast, injections of DSPC/DSPE small
vesicles of ;40 nm diameters revealed distinct exothermic
reactions until the sixth injection (Fig. 4 D). We also found
that injections of DSPC small vesicles revealed, to some
extent, exothermic reactions, whereas those of DSPC large
vesicles did not (Fig. 4, A and C). We next investigated
whether duramycin preferentially binds highly curved mem-
branes when lipids contain unsaturated fatty acid. Fig. 3 C
shows the inhibition of duramycin-induced hemolysis by
liposomes with different sizes. POPC/POPE small vesicles
were slightly more effective than large vesicles and MLVs
when the liposomes were preincubated with duramycin at
37?C. When POPC/POPE liposomes were preincubated with
duramycin at 4?C, a clear difference was observed in the
inhibitory effects of small vesicles, large vesicles, and MLVs
with the small vesicles showing the most prominent effect
(Fig. 3 D). When duramycin was titrated with POPC/POPE
(9:1) large vesicles, the peak diameter of which was evaluated
as ;700 nm by dynamic light scattering, the binding constant
of duramycin to POPE was estimated to be Ka¼ (4.0 6 1.9) 3
107M?1from the ITC result (data not shown). This value is
five times smaller than that of the titration with POPC/POPE
vesicles in Fig. 2 B. The peak diameter of the POPC/POPE
vesicles used in Fig. 2 B was evaluated as ;100 nm by
dynamic light scattering. These results suggest that duramycin
preferentially binds PE in fluid and highly curved membranes.
In Fig. 5, A–D, large vesicles and small vesicles were
further characterized by freeze-fracture electron microscopy.
lipids. Lipid vesicles were prepared by extrusion through polycarbonate
filters with 100-nm pore size. ITC was performed as described in Materials
and Methods. The values 20.1 mM (A, B, and D) or 13.7 mM (C) duramycin
in the reaction cell (1.4034 ml) was titrated with 5.30 mM POPC (A), 5.55
mM POPC/POPE (9:1) (B), 5.05 mM POPC/C16(plasm)-18:1 PE (9:1) (C)
or ;6 mM POPC/plant MGDG (9:1) (D) at 37?C. Each peak corresponds to
the injection of 8 ml of liposomes. Data are representatives of three inde-
pendent experiments (A and B).
Duramycin specifically interacts with ethanolamine phospho-
1612 Iwamoto et al.
Biophysical Journal 93(5) 1608–1619