Báo cáo khoa học: Binding of hemolin to bacterial lipopolysaccharide and lipoteichoic acid An immunoglobulin superfamily member from insects as a pattern-recognition receptor doc

Binding of hemolin to bacterial lipopolysaccharide and lipoteichoic
acid
An immunoglobulin superfamily member from insects as a pattern-recognition
receptor
Xiao-Qiang Yu and Michael R. Kanost
Department of Biochemistry, Kansas State University, Manhattan, KS, USA
Hemolin, a plasma protein from lepidopteran insects, is
composed of four immunoglobulin domains. Its synthesis i s
induced by microbial challenge. We investigated the
biological functions of hemolin in Manduca sexta. It was
found to bind to the surface of bacteria and yeast, and
caused these micro-organisms to aggregate. Hemolin was
demonstrated to bind to lipopo lysaccharide (LPS) from
Gram-negative b acteria and to lipoteichoic acid from G ram-
positive bacteria. Binding of hemolin to smooth-type forms
of LPS was competed for efficiently by lipoteichoic acid and
by rough mutant (Ra and R c) forms of LPS, which differ in
polysaccharide length. Binding of hemolin to LPS w as
partially inhibited by calcium and phosphate. Hemolin
bound to the lipid A component of LPS, and this binding
was completely b locked by free phosphate. O ur results
suggest that hemolin has t wo binding sites for LPS, one that
interacts with t he phosphate groups of lipid A and one that
interacts w ith t he O-specific antigen and the outer-core
carbohydrates of LPS. The binding properties of M. sexta
hemolin suggest that it functions as a pattern-recognition
protein with broad specificity in the defense against micro-
organisms.
Keywords: h emolin; i nsect immunity; lipopolysaccharide;
lipoteichoic a cid; pattern recognition r eceptor.
Upon microbial infection, insects synthesize d efensive

plasma proteins, which include antimicrobial peptides and
proteins, lectins, and cell a dhesion molecules [1–3]. One such
protein is hemolin, a member of the i mmunoglobulin (Ig)
superfamily. H emolin contains four Ig do mains of the I-set
type which are most similar t o t hose in neural cell adhesion
molecules [4–6]. Hemolin has been isolated from hemo-
lymph of two immune-challenged lepidopteran insects,
Hyalophora cecropia and Manduca sexta [7,8]. A hemolin-
like cDNA was also cloned from the fall webworm,
Hyphantria c unea [9]. Hemolin is synthesized mainly in fat
body in response to microbial chal lenge [4,8], but it is also
synthesized in the absence of infection in embryos [10] and
in fat body and midgut during metamorphosis [3,11,12].
Hemolin expressed a t d ifferent developmental s tages o f
M. sexta differs in carbohydrate content. Hemolin isolated
from adult moths and from bacteria-induced larvae con-
tains noncovalently bound carbohydrates, whereas hemolin
from wandering stage (prepupal) larvae lacks carbohydrates
[11].
Available data suggest that hem olin functions in immune
responses by interacting with insect hemocytes and with
bacteria. It binds to hemocytes and bacteria, a nd its binding
to hemocytes inhibits hemocyte aggregation [5,13–15].
Hemolin from H. cecropia interacts with bacterial lipopoly-
saccharide (LPS) and its lipid A component [15,16] and
binds to hemocytes in a calcium-dependent manner [17]. A
membrane-bound form of hemolin has also been reported
[18]. It h as been suggested that hemolin may modulate
hemocytic activities in development and during immune
responses [12], and may function as an opsonin or as a

pathogen-recognition molecule in the defense against infec-
tion [14–16]. The horseshoe-shape arrangement of t he Ig
domains in the structure of H. cecropia hemolin suggested a
mechanism for homophilic binding of hemolin to mole cules
on the surface of hemocytes or micro-organisms [6].
However, the b iological functions of hemolin i n insects
are still not well understood.
Recognition of nonself plays an essential role in
initiating immune responses. The vertebrate innate immune
system and invertebrate immune responses rely on a s et of
proteins known as pathogen-recognition receptors. These
proteins bind to conserved features o f m icrobial surfaces
such as LPS, lipoteichoic acid (LTA) a nd peptidoglycan
from bacterial cell walls, and b-1,3-glucan from f ungal cell
walls [19,20]. Such recognition may initiate a variety of
immune responses in insects, including prophenoloxidase
activation, p hagocytosis, nodule formation, and encapsu-
lation.
In this paper, we focus on t he biological functio ns of
hemolin in de fense against microbial infection. We investi-
gated its binding to Gram-negative and Gram-positive
bacteria, yeast, and bac terial LPS and LTA. Our results
indicate that hemolin functions as a pattern-recognition
receptor with a broad specificity for diverse pathogens in t he
defense against micro-organisms.
Correspondence to M. R. Kanost, Department of Biochemistry,
Kansas State University, Manhattan, KS 66506, USA.
Fax: + 7 8 5 532 7278, Tel.: + 785 532 6964,
E-mail:
Abbreviations: KDO, 2-oxo-3-deoxyoctanoate; LPS, lipopolysac-

charide; LTA, lipoteichoic acid.
(Received 2 2 October 200 1, revised 6 February 2002, a ccepted 8
February 2 002)
Eur. J. Biochem. 269, 1827–1834 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.02830.x
EXPERIMENTAL PROCEDURES
Hemolin and microbial components
Hemolin from hemolymph of naive wandering stage
larvae (hemolin form W1 which lacks bound carbohy-
drate) was purified as described previously [11], and
used for all experiments. Smooth LPS (S-LPS) from
Escherichia coli strains 026:B6 and 0111:B4, LPS r ough
mutan ts Ra (E. coli EH100), Rc (E. coli J5) and Rd2
(E. coli F583), diphosphoryl lipid A (E. coli F583), LTA
from Staphylococcus aureus, 2-oxo-3-deoxyoctanoate
(KDO), laminarin, curdlan, zymosan, chitosan, mannan,
glucose, galactose, mannose, xylose, N-acetylglucosamine
(GlcNAc), and N-acetylgalactosamine (GalNAc) were
purchased from Sigma. The Re m utant of LPS (E. coli
D31m4) was from List Biological Laboratory Inc.
(Campbell, CA, USA). Peptidoglycan (from S. aureus)
was purchased from Fluka.
Agglutination of bacteria and yeast by hemolin
Fluorescein isothiocyanate-labeled S. aureus, E. coli,or
Saccharomyces cerevisiae (Molecular Probes) were sus-
pended in Tris-buffered saline ( Tris/NaCl; 25 m
M
Tris/
HCl, 137 m
M
NaCl and 3 m

M
KCl, pH 7.0) and used
for the agglutination assay. Hemolin at 0.5 m gÆmL
)1
or
BSA at 1 . 0 mgÆmL
)1
(asacontrol)wasusedin
agglutination of micro-organisms as described previously
[21].
Binding of
125
I-labeled hemolin to bacteria and yeast
Hemolin was labeled with
125
I using Iodobead (Piece) as
an iodinatio n reagent. One Iodobead was washed with
500 lL iodin ation buffer (100 m
M
sodium phosphate
buffer, pH 7.0). The washed bead was added to 1 mCi
Na
125
I (Dupont NEN) in 434 lL iodination buffer, and
incubated for 5 min at room temperature. Then 250 lg
purified hemolin in 66 lL deionize d water was mixed
with the s olution containing the Iodobead, f ollowed by
incubation for 3 min at room temperature. The Iodo-
bead was r emoved, a nd
125

I-labeled he molin was sepa-
rated from free
125
I by applying the solution to an
equilibrated Sephadex G -25 d esalting c olumn (PD10,
Pharmacia). The column was eluted with NaCl/P
i
(25 m
M
sodium phosphate buffer, 137 m
M
NaCl and
3m
M
KCl, pH 7.0), and 0.6 mL fractions were collected.
Samples of 5 lL were removed from each fraction
and c ounted in a c counter. F ractions containing
125
I-labeled hemolin were pooled and stored at )20 °C.
The specific activity of the labeled hemolin was
3.9 · 10
5
c.p.m.Ælg
)1
.
125
I-Labeled hemolin at a concentration of 1.0 l
M
was
incubated with formalin-killed E. coli strain XL1-blue,

Micrococcus lysodeikticus,orS. cerevisiae (yeast) (each at
3 · 10
5
cells) in 50 lL Tris/NaCl containing 1 m gÆmL
)1
BSA, in the absence or presenc e of 50 l
M
unlabeled
hemolin. The mixture was incubated for 30 min at room
temperature, then centrifuged for 5 min at 10 000 g.The
supernata nt was removed by aspira tion, and the cells were
washed four times with Tris/NaCl, then counted in a
c coun ter for bound hemolin.
Binding of hemolin to immobilized LPS, LTA or lipid A
Wells of a flat-bottom 96-well assay plate (Costar, Fisher)
were coate d with 2 lgLPS(E. coli 026:B6), LTA or
diphosphoryl lipid A as described p reviously [22]. H emolin
at different concentrations prepared in binding buffer
(50 m
M
Tris/HCl, 50 m
M
NaCl, pH 8.0, 0.1 mg ÆmL
)1
BSA) co ntaining 0 or 10 m
M
CaCl
2
, o r in phosphate buffer
(50 m

M
sodium phosphate, 50 m
M
NaCl, pH 8.0,
0.1 m gÆmL
)1
BSA) was added at 50 lL per well, and
binding was allowed to occur for 6 h at room temperature,
before washing as described by Yu & Kanost [22]. B ound
hemolin was m easured by fi rst incubating wells with rabbit
anti-hemolin serum (diluted 1000-fold with binding buffer),
then with alkaline phosphatase-conjugated goat anti-(rabbit
IgG) Ig (Bio-Rad) (diluted 3 000-fold with binding buffer),
and bound alkaline phosphatase activity was determined by
hydrolysis of p-nitrophenyl phosphate, all as described
previously [22]. The A
405
value o f each well was determined
using a microtiter plate reader (Bio-Tek Instrument, Inc.).
Binding of hemolin to immobilized LPS in the presence
of competitors
The w ells of a 96-well plate w ere coated with LPS from
E. coli 026:B6 (2 lg per well). Hemolin at a concentration of
30 lgÆmL
)1
was p reincubated w ith S-LPS (E. coli strains
026:B6 and 0111:B4), LPS from rough mutants (Ra, Rc,
Rd2 and Re), diphosphoryl lipid A, LTA, peptidoglycan,
zymosan, laminarin, KDO (each at 0 .8 mg ÆmL
)1

), glucose,
galactose, mannose, GlcNAc, GalNAc, or xylose (each at
0.4 m
M
)in50lL binding buffer for 3 h at room temper-
ature. The mixture was then added to S-LPS (E. coli
026:B6)-coated w ells, and the binding was allowed t o occur
at room temperature f or 6 h before washing and detection
of bound hemolin as described above.
RESULTS
Agglutination of bacteria and yeast by hemolin
To inve stigate whether hemolin can bind t o bacteria or yeast
and cause aggregation of t hese micro-organisms, we
performed an agglutination assay. When E. coli, S. aureus,
or S. cerevisiae cells we re incubated with hemolin at a
concentration of 0.5 mg ÆmL
)1
, aggregates of bacteria and
yeast were observed (Fig. 1) . The size of the aggregates
correlated with hemolin concentration, with higher hemolin
concentration resulting in larger aggregates (data not
shown). When these micro-organisms were incubated with
a control protein, B SA, no obvious aggregates were
observed ( Fig. 1). A gglutination of bacteria by hemolin
was not affected by addition of EDTA (data not shown),
indicating that hemolin does not require bivalent cations for
its a gglutination activity. The observed agglutination of
bacteria and yeast by hemolin may be due to binding and
crossing-linking of these micro-organisms by hemolin.
Binding of hemolin to bacteria and yeast cells

M. sexta hemolin was reported to b ind to E. coli [23] and to
enhance the association o f E. coli with hemocytes [ 14].
To assess the binding of hemolin to different types of
1828 X Q. Yu and M. R. Kanost (Eur. J. Biochem. 269) Ó FEBS 2002
micro-organisms, we performed a binding assay using
125
I-labeled hemolin. Radioactively labeled h emolin bound
to a Gram-negative b acterium, E. coli, a Gram-positive
bacterium, M. lysodeikticus,andtoayeast(S. cerevisiae)
(Fig. 2 ). Specific binding (demonstrated by competition
with unlabeled hemolin) accounted f or 70% of the h emolin
binding to M. lysodeikticus, 46% of the binding to E. coli,
and 25% of the binding to S. cerevisiae . Thus, b inding of
hemolin to Gram-positive and Gram-negative bacteria as
well as yeast appears to involve specifi c binding sites, with a
lower degree of specific binding to yeast cells than to
bacteria.
Binding of hemolin to immobilized LPS
Daffre & Faye [16] reported that H. cecropia hemolin
interacts with bacterial LPS. We performed an enzyme-
linked immunosorbent assay to measure bind ing of
M. sexta hemolin to immobilized LPS. Hemolin at different
concentrations was a dded to w ells of a m icrotiter plate
coated with S-LP S from E. coli (strain 026:B6). After an
incubation period and washing, the bound hemolin was
detected using a ntiserum to hemolin. Binding of hemolin to
LPS was concentration-dependent and saturable, reaching a
maximum at 80 lgÆmL
)1
hemolin (Fig. 3). Nonlinear

regression analysis of the binding data showed that binding
of hemolin to LPS fits a two-site binding model (R
2
¼
0.92), with a high-affinity site (K
d1
¼ 0.041 ±
0.065 lgÆmL
)1
) and a low-affinity site (K
d2
¼ 53.2 ±
20.1 lgÆmL
)1
). As these binding studies were performed
under nonequilibrium conditions, the calculate d binding
constants s hould be c onsidered rough estimates. Binding of
hemolin to LPS in buffer containing 10 m
M
CaCl
2
was
approximately half of that observed in the absence of
calcium ( Fig. 4). When the binding assay was performed i n
phosphate buffer instead of Tris buffer, we also observed an
 50% decrease in binding of hemolin to LPS (Fig. 4).
Hemolin binds to the O-specific antigen, outer-core,
and lipid-A moieties of LPS
Bacterial LPS consists of three moieties: lipid A, the core
carbohydrate, and the O-specific antigen (Fig. 5) [24]. Lip-

id A i s composed of a b-glucosaminyl-(1,6)-a-
D
-glucosamine
Fig. 1. Agglutination of bacteria and yeast by
hemolin. BSA ( 1 mgÆmL
)1
) or h emolin
(0.5 mg ÆmL
)1
) was incubated w ith fluorescein
isothiocyanate-labeled E. coli (1.0 · 10
9
cellsÆmL
)1
), S. aureus (1.0 · 10
9
cellsÆmL
)1
)or
S. cerevisiae (yeast) (1.0 · 10
8
cellsÆmL
)1
)in
Tris/NaCl. After i ncubation for 45 min at
room temperature, cells we re observed by
fluorescence microscopy.
Fig. 2. Binding of
125
I-labeled hemolin to bacteria and y east.

125
I-hemolin (1.0 l
M
) w as incubated with formalin-killed E. coli,
M. lysodeikticus or S. cerevisiae (each at 3 · 10
5
cells) in T ris/NaCl in
the presence or absence of 50 l
M
unlabeled hemolin. The c ells were
washed four times and counted in a c c ounter to detec t bound hemolin.
Total binding re presents the amount of hemolin bo und in the absence
of unlabeled hemolin. Specific binding was calculated by subtracting
the amount of h emolin bo und in t he pre sence of a 5 0-fo ld excess of
unlabeled hemolin (nonspecific binding) from t otal binding.
Fig. 3. Binding of h emolin to immobilized LPS. He mo lin at different
concentrations prep ared in binding buffer was a dded to LPS-coated
microtiter plates and incubated for 6 h at room temperature. The
binding assa y was performed as describ ed in Experime ntal Proce dure s.
Each point represents the mean ± SD from four i ndividual mea-
surements. The solid line represents a nonlinear regression calculation
of a two-site b inding curve (R
2
¼ 0.92 ), and the dotted line represents
the curve calculated for one-site binding ( R
2
¼ 0.85).
Ó FEBS 2002 M. sexta hemolin as a pattern-recognition protein (Eur. J. Biochem. 269) 1829
disaccharide backbone which carries up to seven fatty
acids. The c ore carbohydrate i s further divided into an

inner-core and an outer-core s ubdomain. The inner core i s
composed of KDO a nd heptoses, while the outer core
contains hexoses, primarily glucose, galactose, and
GlcNAc. The O-specific antigen consists of a distinct
repeating o ligosaccharide of up to 40 units [24]. O-specific
antigen structures are highly variab le compared with other
moieties of LPS. LPS from s mooth c olony forms of Gram-
negative bacteria (S-LPS) c ontains all of t hese components,
whereas LPS from rough m utants (R-LPS) lack the
O-antigen and may a lso lack parts of the outer and inner
core polysac charide.
We per formed competitive binding assays to test binding
of different moieties of LPS to M. sexta hemolin. Binding of
hemolin to immobilized LPS from a smooth strain of E. coli
(026:B6) was measured in the presence of a 20-fold excess of
different f orms of LPS o r lipid A as competitors. S-LPS
from E. coli st rain 0111:B4 competed more efficiently (82%)
for hemolin binding than did S-LPS from strain 026:B6
(64%) (Fig. 6). Because these two types of LPS differ in
O-specific antigen structure [25,26], t his result suggests t hat
hemolin can bind to the O-specific antigen of 0111:B4.
Ra-LPS, w hich lacks an O -specific antigen, and Rc-LPS,
which also lacks part of the outer core, competed for
hemolin binding (59%) about as well as did S-LPS from
strain 026:B6 (Fig. 6), indicating that the O-specific antigen
of 026:B6 may not contribute significantly t o hemolin
binding. However, Rd
2
-LPS, which lacks the entire outer
core and two heptose residues from the inner core, was

significantly less efficient as a competitor (30%), suggesting
that hemolin may bind to g alactose, glucose, or GlcNAc
residues in the outer cor e or to heptose residues i n the inner
core. The finding that glucose and galactose inhibited
binding of hemolin to LPS (Fig. 7) is consistent with this
idea. H owever, KDO, a component of the inner core of
LPS, did not inhibit binding of hemolin to LPS (Fig. 7).
Re-LPS and lipid A a lone were approximately e quivalent to
Fig. 6. Binding o f hemolin t o LPS in the presence o f different forms of
LPS as competitors. Hemolin (30 lgÆmL
)1
) was preincubated with
S-LPS (from E. coli strains 026:B6 and 0111:B4), Ra-LPS, Rc-LPS,
Rd2-LPS, Re-LPS, or diphosphoryl lipid A (each at a final concen-
tration of 0.8 mgÆmL
)1
)in50lL binding buffer for 3 h at room tem-
perature. Th e mixture was then ad de d to wells of LPS-coated microtiter
plate and incubated for 6 h at room temperatu re. The binding assay
was performed as described in Experimental procedures. The bars
represent the mean ± SD from four individual measurements.
Fig. 7. Binding of hemolin to LPS in the presence of saccharides as
competitors. Hemolin (30 lgÆmL
)1
) was preincubated with glucose,
galactose, mannose, GlcNAc, GalNAc, xylose (each at final 400 m
M
),
KDO, or LPS (E. c oli 026:B6) (each at 800 lgÆmL
)1

)in50lLbinding
buffer for 3 h at room temperature. The mi xture was th en ad ded to
wells of L PS-coated microtite r plate and incubated for 6 h at room
temperature. The binding assay was performed as i n Fig. 6 . The bars
represent the mean ± S D from four individual m easuremen ts.
Fig. 5. Schematic d iagram of ba cterial LPS (mod ified from [24 ]).
Fig. 4. Binding of h emolin to LPS in the presenc e o f calcium or phos-
phate. Hemolin at differen t c on centrations was prepared in binding
buffer without c alcium (solid line) o r with 10 m
M
calcium (dotted line),
or in phosphate buffer (dashed line). The binding assay was performed
the same as i n Fig. 3 . Each point represents th e mean ± SD from fo ur
individual measurements.
1830 X Q. Yu and M. R. Kanost (Eur. J. Biochem. 269) Ó FEBS 2002
Rd
2
-LPS as competitors for hemolin binding (Fig. 6). These
results are consistent with a hypothesis t hat the binding of
hemolin to Rd
2
-LPS and Re-LPS is a result of interaction
with their lipid A moiety. T o further i nvestigate the
interaction of M. sexta hemolin with lipid A, we assayed
direct binding of hemolin to immobilized lipid A (Fig. 8).
Hemolin binding to lipid A in Tris buffer was concentra-
tion-dependent, but was not satu rated at 80 lgÆmL
)1
hemolin. When t he assay w as carried out in phosphate
buffer, binding of hemolin to lipid A was nearly eliminated

(Fig. 8).
Hemolin binds to LTA on Gram-positive bacteria
To investigate to w hich components o n the sur face of
Gram-positive bacteria and yeast h emolin binds, w e
performed a competitive binding assay, using microbial
components a s c ompetitors for binding of hemolin to S-LPS
(Fig. 9 ). LTA and peptidoglycan, both cell-wall compo-
nents of Gram-positive bacteria, decreased binding of
hemolin to LPS by 86% and 26%, r espectively, suggesting
that they bind to hemolin at the same site as LPS. The
observation that LTA was a more efficient competitor than
was LPS itself suggests that hemolin has a higher affinity for
LTA than for LPS. This i s consistent with the finding that
more hemolin bound to Gram-positive bacteria than to
Gram-negative bacteria (Fig. 2). H emolin bound directly to
immobilized LTA (Fig. 10). The binding was concentra-
tion-dependent and was not saturated at 50 lgÆmL
)1
hemolin. Nonlinear regression analysis of the binding data
showed that bin ding o f h emolin to LTA also fits a two-site
binding model ( R
2
¼ 0.89), with a K
d1
¼ 0.12 ±
0.11 lgÆmL
)1
and K
d2
¼ 110.1 ± 125.3 lgÆmL

)1
.
Yeast cell walls are composed primarily of b-1,3-glucans
and mannans [27]. Zymosan, a y east cell-wall preparation
that contains glucan, mannan and chitin, decreased binding
of hemolin to LPS by 61% (Fig. 9 ). But laminarin, a soluble
form of b-1,3-glucan, did not inhibit hemolin binding to
LPS (Fig. 9), and hemolin did not bind to curdlan, an
insoluble f orm o f b-1,3-glucan ( data not shown), indicating
that hemolin does not bind to b-1,3-glucans. Mannan a nd
chitosan (deacetylated chitin) also did not inhibit hemolin
binding to LPS (data not shown). However, mannose
inhibited binding of hemolin to LPS by 28% (Fig. 7), which
suggests that h emolin may b ind to the mannan on the
surface of yeast.
DISCUSSION
Hemolin synthesis is i nduced by Gram-negative a nd Gram-
positive bacteria, and it is the major protein produced in
response to m icrobial infection i n l epidopteran insects such
as H. cecropia and M. sexta [7,8,28], s uggesting that i t
Fig. 9. Binding o f hemolin to LPS in the presence of microbial c ompo-
nents as competitors. Hemolin (30 lgÆmL
)1
) was preincubated with
LPS (E. coli 026:B6), LTA, peptidoglycan, zymosan, or lamin arin
(each at 800 lgÆmL
)1
)in50lL binding buffer for 3 h at room
temperature. The mixture was then added to we lls of a LPS -co ated
microtiter plate and in cubated for 6 h at ro om temperature. T he

binding assay was performed as in Fig. 6. The bars r epresent the
mean ± SD f orm four individual measurements.
Fig. 8. Binding of hemolin to lipid A. Hemolin at different concentra-
tions prepared in binding buffer or phosphate buffer w as added to
diphosphoryl lipid A-coated microtiter plates and incubated for 6 h at
room te mperature. The binding was performed as described in
Experimental procedures. E ach point represents t he mean ± SD from
four individual measurements.
Fig. 10. Binding of hemol in to immobilized LTA. Hemolin at different
concentrations prepared in binding buffer was added t o L TA-coat ed
microtiter plates and incubated for 6 h at room temperature. The
binding a ssay was p erformed as described in Experimental procedures.
Each point represents the mean ± SD from four i ndividual mea-
surements. The solid line represents a nonlinear regression calculation
of a two-site b inding curve (R
2
¼ 0.89 ), and the dotted line represents
the curve calculated for one-site binding ( R
2
¼ 0.78).
Ó FEBS 2002 M. sexta hemolin as a pattern-recognition protein (Eur. J. Biochem. 269) 1831
functions in the i mmune response o f these insects. However,
hemolin does not display direct antibacterial activity.
Instead, its role may be related to its ability t o bind to the
surface of hemocytes [5,13–15] a nd to bacteria. Hemolin has
beenshowntobindtoE. coli [4,23] and t o increase t he
association of E. coli with hemocytes [ 14]. W e h ave found
that hemolin als o binds to Gram-positive bacteria and to a
lesser degree to yea st. In these studies we ha ve invest igated
the binding of hemolin to LPS and LTA, molecules that are

present on the surface of Gram-negative a nd Gram-positive
bacteria, respectively.
LPS on the surface of Gram-negative bacteria is a
potential target for binding of pattern-recognition recep-
tors. The availability of E. coli mutants expressing differ-
ently t runcated forms of LPS makes it possible to identify
the part of the LPS molecule to which a protein binds.
H. cecropia hemolin binds to wild-type E. coli and also to
mutants lacking the c ore carbohydrate [29]. We f ound that
M. sexta he molin bound to immobilized LPS and to its
isolated lipid A component in a concentration-dependent,
saturable manner. Competitive b inding experiments indi-
cated that hemolin binds smooth forms of LPS most
efficiently, but rough forms of LPS, lacking the O-antigen
and p arts of the i nner-core and outer-core p olysaccharide,
and lipid A alone could also p artially compete f or hemolin
binding to smooth LPS. Rough mutants Rd and Re,
containing only the lipid A moiety and part of the inner
core, competed no better than lipid A alone, suggesting
that hemolin does not bin d to KDO in the inner core. This
is consistent with the observation that KDO did not
compete for hemolin binding to LPS and with the results
of Daffre & Faye [16], who showed by photoaffinity
labeling that hemolin from H. cecropia bound to S-LPS
and that the binding could be competed f or by lipid A but
not KDO. Approximately 30 n g hemolin specifically
bound to the surface of 3 · 10
5
E. coli cells (Fig. 2 ),
indicating that  10

6
molecules of hemolin bound to each
E. coli cell. Because Gram-negative b acteria c ontain  10
6
molecules of LPS per cell [30], this result suggests that, on
average, each LPS mole cule w as occupied by one molecule
of hemolin.
Results of binding curves and competition experiments
suggest that hemolin contains two binding sites for LPS.
One site a ppears to bind to the carbohydrate components
in the O-antigen and outer core, a nd the other site binds to
lipid A. Even though isolated lipid A binds to hemolin, it
could only partially compete for LPS binding to S-LPS.
Similarly, Daffre & F aye [16] f ound that a large excess of
lipid A decreased hemolin binding to S-LPS b y only 42%.
The b inding of hemolin to lipid A may involve an
interaction w ith the phosphate groups on lipid A. Free
phosphate decreased hemolin binding to S-LPS by
approximately half and nearly eliminated hemolin binding
to lipid A. An interpretation of these results is that
phosphate disrupts binding of lipid A by competing for a
site that interacts with phosphate groups, and that a
separate binding site that interacts with carbohydrate
components of LPS is not affected by phosphate. In the
crystal structure of H. cecropia hemolin, a phosphate ion
was f ound in the interface of Ig d omains 2 and 3 [6].
Perhaps this r egion o f t he molecule is part of a binding site
for lipid A.
Because the homophilic binding of hemolin to o ther
hemolin molecules was shown to require Ca

2+
[17], we
tested the effect of Ca
2+
on LPS binding. Rather than
enhancing h emolin binding, 10 m
M
Ca
2+
decreased
hemolin binding to LPS by a bout half, very similar to the
effect of phosphate. Electrostatic interactions of Ca
2+
with
phosphate groups of lipid A may mask these groups and
interfere with the lipid A-binding site but not the carbohy-
drate-binding site. The opposing effects o f C a
2+
on hemolin
binding to LPS and other hemolin molecules suggest that
homophilic binding occurs at a site distinct f rom LPS
binding.
More hemolin bound to the Gram-positive bacterium
M. lysodeikticus , which does not contain LPS, than to
E. coli (Fig. 2). When we tested whether cell surface
components of Gram-positive bacteria can compete with
LPS for hemolin binding, we found that peptidoglycan,
the major cell-wall component of Gram-positive bacteria,
inhibited b inding of hemolin t o L PS by 26%, w hereas
LTA, another surface component of Gram-positive

bacteria, inhibited hemolin binding to LPS by 86%.
LTA was more effective than LPS itself as an inhibitor
of hemolin binding to LPS, suggesting that LPS and
LTAbindtothesamesitesonhemolinandthathemolin
may have a higher affinity for LTA. Hemolin was also
observed to bind d irectly to i mmobilized LTA (Fig. 10).
LPS and LTA are s imilar in containing both polysac-
charide components a nd lipid components associated
with phosphate groups [31], and these may occupy the
same binding sites in hemolin. Another insect plasma
protein that has been shown to interact with LTA is
apolipophorin-III of Galleria mellonella, which presum-
ably binds to the hydrophobic components of LTA [32].
To function as a pattern-recognition receptor, a protein
must bind to the s urface of i nvading micro-organis ms.
We showed that hemolin binds to the surface of Gram-
negative and Gram-positive bacteria and yeast, and
caused aggregation of these micro-organisms. Binding
of hemolin to the surface of b acteria appears t o be due
to specific interactions with LPS on Gram-negative
bacteria and to LTA and perhaps also peptidoglycan
from Gram-positive bacteria. Binding of hemolin to yeast
was less efficient, and it is not clear from our
experiments w hat part o f the yeast cell wall is t he
hemolin-binding site. Aggregation of m icro-organisms by
hemolin and the ability of hemolin to bind to hemocytes
may promote phagocytosis and the formation of hemo-
cyte nodules to clear micro-organisms from the insect
hemolymph.
Recognition of micro-organisms by pattern-recognition

receptors is a crucial function of the innate immune
system of vertebrates a nd invertebrates [19,20]. Pattern-
recognition receptors identified in M. sexta and other
insect species include C-type lectins [9,21,22,33–35], b-1,3-
glucan-binding proteins [36,37], and peptidoglycan-binding
proteins [38–41]. The rapid induction of hemolin to high
concentration in hemolymph (1.5 mgÆmL
)1
in M. sexta
larvae) [5,8] and its broad specificity f or binding to
different types of micro-organisms suggests that it func-
tions as a pattern-recognition receptor that p articipates in
detection and elimination o f a variety of pathogens in
lepidopteran insects.
1832 X Q. Yu and M. R. Kanost (Eur. J. Biochem. 269) Ó FEBS 2002
ACKNOWLEDGEMENTS
We than k Maureen Gorman and Neal Dittm er for helpful comments
on the manuscript. This work was supported b y Nation al Institutes of
Health G rants AI 31084 and GM41247. This is contribution 00-320-J
from the K ansas Agricultural E xperiment Station.
REFERENCES
1. Gillespie, J.P., Kanost, M.R. & Trenczek, T. (1997) Biological
mediators of insect immunity. Annu. Rev. Entomol. 42 , 611–643.
2. Bulet, P., Hetru, C., Dimarcq, J.L. & Hoffmann, D. (1999)
Antimicrobial peptides in insects: structure and function. Dev.
Comp. Immunol. 23 , 329–344.
3. Trenczek, T. (1998) Endogenous defense mechanisms of insects.
Zoology 101, 298 –315.
4. Sun, S C., Lind strom, I., Boman, H.G., Faye, I. & Schmidt, O.
(1990) Hemolin: an insect-immune p rotein belonging t o the

immunoglobulin superfamily. Science 250, 1729–1732.
5. Ladendorff, N.E. & Kanost, M.R. (1991) Bacteria-induced
protein P4 (Hemolin) f rom Manduca sexta: a member of the
immunoglobulin superfamily which can inhibit hemocyte
aggregation. Arch. Insect B iochem. Physiol. 18 , 285–300.
6. Su, X D., Gastinel, L.N., Vaughn, D.E., Faye, I., Poon, P. &
Bjorkman, P.J. (1998) Crystal structure of hemolin: a horseshoe
shape with implications for homophilic adhesion. Science 281,
991–995.
7. Rasmuson, T. & Boman, H.G. (1979) Insect immunity. V. P ur-
ification and some properties of immune protein P4 from
hemolymph of Hyalophora cecropia pup ae. Insect Biochem. 9,
259–264.
8. Ladendorff, N.E. & Kanost, M.R. (1990) Isolation and char-
acterization of b acteria-induce d protein P4 f rom hemolymph of
Manduca s exta. Arch. Insect Biochem. Physiol. 15, 33–41.
9. Shin, S.W., Park, S S., Park, D S., Kim, M.G., Kim, S.C., Brey,
P.T. & Park, H Y. (1998) Isolation and characterization of
immune-related genes f rom the fall webworm, Hyphantria cunea,
using PCR-based differential display and subtractive cloning.
Insect Biochem. Mol. Biol. 28, 827–837.
10. Bettencourt, R., Assefaw-Redda, Y. & Faye, I. (2000) The i n sect
immune prote in h emolin is expressed during oogenesis and
embryogenesis. Mech. Dev. 95, 301–304.
11.Yu,X Q.&Kanost,M.R.(1999)Developmentalexpression
of Manduca s exta hemolin. Arch. Insect Biochem. Physiol. 42,
198–212.
12. Lanz-Mendoza, H.L. & Faye, I. ( 1999) Physiological a spects of
the immunoglobulin superfamily in invertebrates. Dev. Comp .
Immunol. 23 , 359–374.

13. Faye, I. & Kan ost, M .R. (1 998 ) Funct ion a nd r egul ation of
hemolin. In Molecular Mechanisms of Immune Responses i n Insects
(Brey, P.T. & H ultmark, D ., e ds), pp. 173–188. Chapman & Hall,
New York.
14. Kanost, M.R. & Zhao, L. (1996) Insect h emolymph proteins from
the immunoglobulin superfamily. Adv. Comp. Environ. Physiol. 23,
185–197.
15. Lanz-Mendoza, H., Bettencourt, R., Fabbri, M. & Faye, I. (1996)
Regulation of the insect immune response: the effect of hemolin on
cellular immune m echanisms. Cell Immunol. 169, 47–54.
16. Daffre, S. & Faye , I . (1997) Lipopolysa ccharide interaction with
hemolin, a n insect mem ber of the Ig-superfamily. FEBS Lett. 408,
127–130.
17. Bettencourt, R., Gunne, H., Gastinel, L., Steiner, H. & Faye, I.
(1999) Implications of hemolin glycosylation and Ca
2+
-binding
on homophilic and cellular interactions. Eur. J. Biochem. 266 , 964–
976.
18. Bettencourt, R., Lan z-Mendoz a, H., Lindquist, K.R. & Faye,
I. (1997) Cell adhesion properties of hemolin, an insect
immune protein in the Ig superfamily. Eur. J. Biochem. 250,
630–637.
19. Janeway, C.A. (1989) Approaching the asymptote? Evolution and
revolution in immunology. Cold Spring H arbor Symp Quant. B iol.
54, 1–13.
20. Janeway, C.A. (1992) The i mmune system evolved to discriminate
infectious nonself from noninfectious self. Immunol. Today 13,
11–16.
21. Yu, X Q., Gan, H. & Kanost, M.R. (1999) Immulectin, an

inducible C-type lectin from an insect, Manduca sexta, stimulates
activation of plasma prophenol oxidase. Insect Biochem. Mol.
Biol. 29, 5 84–597.
22. Yu, X Q. & Kanost, M.R. (2000) Immulectin-2, a lipopoly-
saccharide-specific lectin from an insect, Manduca sexta, is
induced in response to Gram-negative bacteria. J. Biol. Chem. 275,
37373–37381.
23. Zhao, L. & Kanost, M.R. (1996) In search of a function for
hemolin, a hemolymph protein from the immunoglobulin super-
family. J. Insect Physiol. 42, 7 3–79.
24. Raetz, C .R.H. (1990) Biochemistry of endotoxins. Annu. Rev.
Biochem. 59, 129–170.
25. Manca, M.C., Weintraub, A. & Widmalm, G. (1996) Structural
studies of the Escherichia coli 026 O-antigen p olysaccharide.
Carbohydr. Res. 281, 155–160.
26.Gupta,R.K.,Egan,W.,Bryla,D.A.,Robbins,J.B.&Szu,
S.C. (1995) Escherichia coli 0111 O-specific polysaccharide,
prepared by treatment with acetic acid or hydrazine, bound
to tetanus t oxoid b y t wo syn thetic s chem es. Inf ect . Im mun. 63 ,
2805–2810.
27. Stratford, M. (1994) Another brick in the wall? Recent
developments concerning the yeast cell envelope. Yeast 10 , 1741–
1752.
28. Andersson, K . & Steiner, H. ( 1987) Structure and properties of
protein P4, the major bacteria-inducible protein in pupae of
Hyalophora c ecropia. Insect Bi ochem. 17, 133–140.
29. Schmidt, O., Faye, I., Lindstrom-Dinnetz, I. & Sun, S.C. (1993)
Specific immune recognition of insect hemolin. De v. Comp.
Immunol. 17 , 195–200.
30. Inouye, M. (1979) What is the outer membrane? In Bacterial Outer

Membranes (Inouye, M., ed.), pp . 1–12. J ohn Wile y & S ons, N ew
York.
31. Fischer, W., Mannsfeld, T. & Hagen, G. (1990) On the basic
structure of p o ly ( glyceropho sphate) l ipote ichoic a cid s. Biochem.
Cell Biol. 68, 3 3–43.
32. Halwani, A.E., Niven, D.F. & Dunphy, G.B. (2000) Apolipo-
phorin-III and the interactions of lipoteichoic acids with the
immediate i mmun e r esponse s of Galleria mellonella. J. Invertebr.
Pathol. 76 , 233–241.
33. Jomori, T. & N atori, S. (1992) Function of the lipopolysaccharide-
binding protein of Periplaneta americana as an opsonin. F EBS
Lett. 296, 283–286.
34. Koizumi,N.,Imamura,M.,Kadotani,T.,Yaoi,K.,Iwahana,H.
& Sato, R. (1999) The lipopolysaccharide-binding protein parti-
cipating in hemocyte nodule formation in the silkworm Bombyx
mori is a novel member of the C-type lectin superfamily with two
different tandem carbohydrate-recognition domains. FEBS Lett.
443, 139–143.
35. Shin, S.W., P ark, D.S., Kim, S.C. & Park, H .Y. (2000) Two car-
bohydrate recognition d omains o f Hyphantria cunea lectin bind to
bacterial lipopolysaccharides through O -specific ch ain. FEBS
Lett. 467, 70–74.
36. Ochiai, M. & Ashida, M. (200 0) A pattern recognition p rotein for
b-1,3-glucan. J. Biol. Chem. 275, 4 995–5002.
37. Ma, C. & K anost, M.R. (2000) A b-1,3-glucan-recognition protein
from an in sect , Manduca s exta, agglutinates microorganisms and
activates the phenoloxidase cascade. J. Biol. Chem. 27 5, 7505–
7514.
Ó FEBS 2002 M. sexta hemolin as a pattern-recognition protein (Eur. J. Biochem. 269) 1833
38. Yu,X Q.,Zhu,Y F.,Ma,C.,Fabrick,J.A.&Kanost,M.R.

(2002) Pattern recognition proteins in Manduca sexta plasma.
Insect Biochem. Mol. Biol. in press.
39. Werner, T., Liu, G., Kang, D., Ekengren, S., Steiner, H. &
Hultmark, D. (2000) A family of peptidoglycan recognition pro-
teins i n the fruit fly Drosophila melanogaster. Proc. N atl A cad. Sci.
USA 97, 13772–13777.
40. Ochiai, M. & Ashida, M. (1999) A pattern re cognition p rotein for
peptidoglycan. Cloning the cDNA and the gene of the silkworm,
Bombyx mor i. J. Biol. Chem. 274, 118 54–11858.
41. Kang, D., Li u, G ., L undstro
¨
m,A.,Gelius,E.&Steiner,H.(1998)
A peptidoglycan recognition protein in innate immunity conserved
from insects to humans. Proc. Natl Acad. Sci. USA 95 , 10078–
10082.
1834 X Q. Yu and M. R. Kanost (Eur. J. Biochem. 269) Ó FEBS 2002