Báo cáo khoa học: Human lactoferrin activates NF-jB through the Toll-like receptor 4 pathway while it interferes with the lipopolysaccharide-stimulated TLR4 signaling potx

Human lactoferrin activates NF-jB through the Toll-like
receptor 4 pathway while it interferes with the
lipopolysaccharide-stimulated TLR4 signaling
Ken Ando
1
, Keiichi Hasegawa
1
, Ken-ichi Shindo
1
, Tomoyasu Furusawa
1
, Tomofumi Fujino
1
, Kiyomi
Kikugawa
1
, Hiroyasu Nakano
2
, Osamu Takeuchi
3
, Shizuo Akira
3
, Taishin Akiyama
4
, Jin Gohda
4
,
Jun-ichiro Inoue
4
and Makio Hayakawa
1

1 School of Pharmacy, Tokyo University of Pharmacy and Life Sciences, Japan
2 Department of Immunology, Juntendo University School of Medicine, Japan
3 Department of Host Defense, Research Institute for Microbial Diseases, Osaka University, Japan
4 Division of Cellular and Molecular Biology, Department of Cancer Biology, Institute of Medical Science, The University of Tokyo, Japan
Keywords
human lactoferrin; innate immunity;
lipopolysaccharide; nuclear factor-jB
(NF-jB); Toll-like receptor 4 (TLR4)
Correspondence
Makio Hayakawa, School of Pharmacy,
Tokyo University of Pharmacy and Life
Science, 1432-1 Horinouchi, Hachioji, Tokyo
192-0392, Japan
Fax: +81-42-676-4508
Tel: +81-42-676-4513
E-mail:
(Received 26 August 2009, revised 15
February 2010, accepted 18 February 2010)
doi:10.1111/j.1742-4658.2010.07620.x
Lactoferrin (LF) has been implicated in innate immunity. Here we reveal
the signal transduction pathway responsible for human LF (hLF)-triggered
nuclear factor-jB (NF-jB) activation. Endotoxin-depleted hLF induces
NF-jB activation at physiologically relevant concentrations in the human
monocytic leukemia cell line, THP-1, and in mouse embryonic fibroblasts
(MEFs). In MEFs, in which both tumor necrosis factor receptor-associated
factor 2 (TRAF2) and TRAF5 are deficient, hLF causes NF-jB activation
at a level comparable to that seen in wild-type MEFs, whereas TRAF6-
deficient MEFs show significantly impaired NF-jB activation in response
to hLF. TRAF6 is known to be indispensable in leading to NF-jB activa-
tion in myeloid differentiating factor 88 (MyD88)-dependent signaling

pathways, while the role of TRAF6 in the MyD88-independent signaling
pathway has not been clarified extensively. When we examined the
hLF-dependent NF-jB activation in MyD88-deficient MEFs, delayed, but
remarkable, NF-jB activation occurred as a result of the treatment of cells
with hLF, indicating that both MyD88-dependent and MyD88-independent
pathways are involved. Indeed, hLF fails to activate NF-jB in MEFs lack-
ing Toll-like receptor 4 (TLR4), a unique TLR group member that triggers
both MyD88-depependent and MyD88-independent signalings. Impor-
tantly, the carbohydrate chains from hLF are shown to be responsible for
TLR4 activation. Furthermore, we show that lipopolysaccharide-induced
cytokine and chemokine production is attenuated by intact hLF but not by
the carbohydrate chains from hLF. Thus, we present a novel model con-
cerning the biological function of hLF: hLF induces moderate activation of
TLR4-mediated innate immunity through its carbohydrate chains; however,
hLF suppresses endotoxemia by interfering with lipopolysaccharide-depen-
dent TLR4 activation, probably through its polypeptide moiety.
Abbreviations
ActE, actinase E; bLF, bovine lactoferrin; EMSA, electrophoretic mobility shift assay; GAPDH, glyceraldehyde-3-phosphate dehydrogenase;
hLF, human lactoferrin; IKK, IjB kinase; IL, interleukin; IP10, interferon-c-inducible protein-10; IRF, interferon regulatory factor; JNK,
c-Jun N-terminal kinase; LBP, LPS-binding protein; LF, lactoferrin; LPS, lipopolysaccharide; LRP, low-density lipoprotein receptor-related
protein; MD-2, myeloid differentiation-2; MEF, mouse embryonic fibroblast; MyD88, myeloid differentiating factor 88; NF-jB, nuclear
factor-jB; PMB, polymyxin B; TLR, Toll-like receptor; TNF, tumor necrosis factor; TRAF, TNF receptor-associated factor; TRIF,
Toll ⁄ interleukin-1 receptor-domain-containing adaptor inducing interferon-b.
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2051
Introduction
Lactoferrin (LF) is an iron-binding glycoprotein that is
abundant in exocrine secretions, including milk and
the fluids of the digestive tract [1]. Although LF
belongs to a family of transferrins, its biological func-
tion is not limited to the regulation of iron metabo-

lism; it also plays multiple roles in host defense, and in
immune and inflammatory reactions [1–4].
LF shows antimicrobial activities against many
pathogens, including different types of bacteria [2]. In
Gram-negative bacteria, it was observed that LF
specifically binds to porins present on the outer
membrane [5] and induces the rapid release of lipo-
polysaccharide (LPS), resulting in enhanced bacterial
susceptibility to osmotic shock, lysozyme, or other
antibacterial molecules [6]. The LPS-binding activity of
LF may account for the other properties of this pro-
tein in the modulation of the inflammatory process [3].
The stimulation of mammalian cells by LPS occurs
through a series of interactions with several proteins,
including the LPS-binding protein (LBP), CD14, mye-
loid differentiation-2 (MD-2) and Toll-like receptor
(TLR)4 [7]. Na et al. [8] reported that macrophages
pretreated with the LF–LPS complex were rendered
tolerant to LPS challenge. They suggested that the
down-regulation of TLR4 signaling is responsible for
this tolerance. Alternatively, serum LBP may partici-
pate in the LF-dependent modulation of the inflamma-
tory response. Elass-Rochard et al. [9] showed that LF
prevented the LBP-mediated binding of LPS to CD14.
In addition, Baveye et al. [10] demonstrated that LF
interacted with soluble CD14, resulting in the inhibi-
tion of signal transduction mediated by the CD14–LPS
complex.
In contrast to the anti-inflammatory roles noted
above, LF is known to activate immune cells to pro-

duce several cytokines, such as tumor necrosis factor
(TNF), interleukin (IL)-8 and IL-12 [11–13]. However,
the molecular mechanism of how LF activates the
intracellular signaling pathway to induce the produc-
tion of these cytokines remains to be elucidated.
At the surface of cells, molecules should exist that
bind to LF and transduce intracellular signals to evoke
LF-dependent biological responses. One such candi-
date LF receptor is nucleolin [14], a 105 kDa nuclear
protein that has also been described as a cell-surface
receptor for several ligands, such as matrix laminin1
and midkine [15,16]. However, it is unlikely that nucle-
olin directly transduces the intracellular signals in
response to LF, because nucleolin lacks the mem-
brane-spanning region and the cytoplasmic domain
responsible for signal transduction. Another candidate
LF receptor has been described, namely the low-den-
sity lipoprotein receptor-related protein (LRP ⁄ LRP1)
[4]. LRP recognizes more than 30 different ligands,
including LF, and acts as a ‘cargo’ receptor, removing
such ligands from the cell surface [17]. However, the
involvement of LRP in the production of cytokines in
response to LF has not yet been described. A third
candidate molecule is the intestinal LF receptor identi-
fied by Suzuki et al. [18]. They indicate that this recep-
tor is responsible for taking up iron from LF into cells
in infants [19]. However, the intestinal LF receptor is
described as a GPI-anchored protein that lacks the
cytoplasmic domain responsible for signal transduction
[18]. Thus, the molecular nature of the cell-surface

receptor that is involved in the LF-induced cytokine
production is still obscure.
In this study, we demonstrated that endotoxin-free
human LF (hLF) directly activates nuclear factor-jB
(NF-jB), which acts as the master regulator of
immune and inflammatory responses, in the human
monocytic leukemia cell line, THP-1, and in mouse
embryonic fibroblasts (MEFs). By characterizing vari-
ous MEFs which lack adaptor or receptor molecules
that trigger NF-jB activation, we found that TLR4 is
responsible for hLF-induced NF-jB activation.
Furthermore, the carbohydrate chains of hLF were
shown to play a crucial role in hLF-induced NF-j B
activation. Thus, we assume that the carbohydrate
chains of hLF activate TLR4, which mediates the
production of cytokines and chemokines. Notably,
when cells were simultaneously treated with LPS and
endotoxin-depleted hLF, the levels of cytokines and
chemokines produced were significantly lower than
those of cells treated with LPS alone, suggesting that
hLF may have a role as a moderate activator of the
immune system while it can suppress the strong inflam-
matory reactions induced by LPS.
Results
LF has been shown to induce the expression of various
cytokines such as TNF, IL-1 and IL-8 [11]. Extensive
research has established that NF-jB plays a critical
role in the inducible expression of these cytokines
involved in immune function and inflammation [20].
Here we clearly demonstrated that hLF significantly

stimulates NF-jB DNA binding in THP-1 cells
(Fig. 1A). In mammals, the family of NF-jB proteins
comprises five members: RelA ⁄ p65, RelB, c-Rel,
p50 ⁄ NF-jB1 and p52 ⁄ NF-jB2. Homodimers or hetero-
dimers of these proteins are active forms of NF-jB
Biological action of hLF is mediated through TLR4 K. Ando et al.
2052 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS
with DNA-binding activity. The most intensively stud-
ied NF-jB dimer is RelA:p50 and the activating pro-
cess of this classical NF-jB dimer is described as the
‘canonical pathway’ that is initiated by the activation
of the IjB kinase (IKK) complex that is composed of
two catalytic subunits – IKKa (also known as IKK1)
and IKKb (also known as IKK2) – and a regulatory
subunit, IKKc (also known as NEMO) [21]. As shown
in the bottom panel of Fig. 1A, nuclear extract from
cells treated with hLF contained RelA, suggesting that
hLF activates NF-jB through the ‘canonical pathway’.
Figure 1B shows that significant NF-jB activation was
induced by hLF at physiologically relevant concentra-
tions (12.5–100 lgÆmL
)1
). Significant activation of
NF-jB was also observed in hLF-treated MEFs
(Fig. 1C). Furthermore, IKK, which is responsible for
the phosphorylation-induced degradation of IjB, was
activated in response to hLF (Fig. 1C, top panel). These
results suggest that hLF can stimulate the ‘canonical
pathway’, leading to the activation of the classical
NF-jB dimer, RelA:p50, in various types of cells.

It should be noted that the hLF used in our experi-
ments was prepared by passing it through a polymyxin
B (PMB)–agarose column, which is used to remove
contaminating endotoxin from the solution. As shown
in Fig. 1D, dose-dependent activation of NF-jB was
observed in MEFs treated with various concentrations
of LPS, whereas no activation of NF-jB was observed
in MEFs treated with fractions passed through a
PMB–agarose column, indicating that the PMB–aga-
rose column effectively absorbed LPS. After passing
through this column, the level of endotoxin detected in
the hLF preparation was usually < 0.6 EUÆmg
)1
.To
prove that hLF-induced NF-jB activation is not
caused by traces of residual endotoxin, we compared
the levels of NF-jB activation in cells treated with
fractions of hLF passed through PMB–agarose versus
cells treated with fractions of hLF not passed through
PMB–agarose. As shown in Fig. 1E, NF-jB activation
occurred at a similar magnitude independently of
PMB–agarose purification of hLF, demonstrating that
hLF, but not the contaminating endotoxin, causes NF-
jB activation. Furthermore, the treatment of cells with
actinomycin D or cycloheximide did not affect the
hLF-induced NF-jB activation (Fig. 1F), indicating
that hLF directly triggers NF-jB activation without
requiring newly synthesized proteins such as TNF or
IL-1, well-known activators for NF-jB. Thus, we have
demonstrated that hLF has the activity to stimulate

the canonical NF-jB-activating pathway directly.
In our previous study, we indicated that the carbo-
hydrate chains of hLF play an important role in the
recognition of hLF by THP-1 macrophages [22]. In
order to evaluate the role of carbohydrate chains of
hLF, hLF was treated with actinase E (ActE) (which is
a nonspecific protease derived from Streptomyces
griseus and is also known as Pronase E [23]) in order
to digest the polypeptide region of hLF while the
carbohydrate chains of hLF remain intact (Fig. 2A).
As shown in Fig. 2B, ActE-digested hLF significantly
stimulated NF-jB DNA binding in MEFs. When anti-
RelA was added to the electrophoretic mobility shift
assay (EMSA) reaction mixture, the bands were super-
shifted to the top of the gel, confirming that the classi-
cal NF-jB dimer, RelA:p50, was activated (Fig. 2B). It
should be noted that ActE alone did not stimulate
NF-jB activation (data not shown). Furthermore, the
purified hLF carbohydrate chain fraction, in which
hLF-derived oligopeptides or amino acids were not
detectable, induced marked NF-jB activation in
THP-1 cells (Fig. 2C). By contrast, when hLF was
treated with endo-b-galactosidase, which is known to
cleave the carbohydrate chains at the internal
Galb1-4GlcNAc position, IKK activation and nuclear
translocation of RelA were significantly impaired,
while the same treatment did not affect LPS-induced
activation (Fig. 2D), suggesting that the carbohydrate
chains of hLF are critical for inducing NF-jB activa-
tion. Furthermore, the observation that NF-jB activa-

tion induced by hLF is ‘endo-b-galactosidase sensitive’,
whereas activation induced by LPS is ‘endo-b-galacto-
sidase resistant’, enables us to rule out the possibility
that the trace amount of residual LPS in the
post-PMB agarose fractions of hLF is responsible for
NF-jB activation.
We next focused on the molecular mechanism of
how initial signaling that leads to the activation of
IKK is triggered after hLF is recognized by cells.
Although a remarkable diversity of stimuli lead to the
activation of NF-jB, many of the signaling intermedi-
ates, especially those just upstream of the IKK com-
plex, are thought to be shared [24]. In particular,
TNF receptor-associated factor (TRAF) families of
proteins are key intermediates in nearly all NF-jB
signaling pathways [24]. Among seven TRAF proteins
identified to date, TRAF2, TRAF5 and TRAF6 have
been most extensively characterized as positive regula-
tors of signaling to NF-jB. From the study using
TRAF2 ⁄ TRAF5 double-knockout mice, TRAF2 and
TRAF5 were shown to be involved in TNF-induced
NF-jB activation. [25]. However, TRAF6 has the
most divergent TRAF-C domain, which mediates the
interaction between TRAF proteins and the tails of
cell-surface receptors, and is the only TRAF that
is involved in the signal from the members of the
Toll ⁄ IL-1 receptor [26]. In order to verify the roles of
K. Ando et al. Biological action of hLF is mediated through TLR4
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2053
TRAF proteins in hLF-induced NF-jB activation, we

first examined whether or not overexpression of domi-
nant negative forms of TRAF2 or TRAF6, which lack
RING finger and zinc finger domains, impaired the
NF-jB activation in THP-1 cells in response to hLF.
As shown in Fig. 3A, similar amounts of dominant
negative forms of TRAF2 or TRAF6 were expressed
in THP-1 cells. While dominant negative TRAF2
effectively suppressed the TNF-induced NF-jB activa-
tion, no inhibition was observed in cells treated with
hLF or hLF-derived carbohydrate chains, as in the
case of IL-1-stimulated cells (Fig. 3B). By contrast,
THP-1 cells overexpressing dominant negative TRAF6
did not show NF-jB activation in response to IL-1,
hLF or hLF-derived carbohydrate chains, whereas
their response to TNF was comparable to that of con-
trol cells (Fig. 3B). These results suggest that TRAF6,
but not TRAF2, is involved in hLF-triggered NF- jB
activating signals. Then, we further investigated the
role of TRAFs in hLF-triggered NF-jB activation by
characterizing cells lacking TRAF isoforms. As
reported previously [25], TNF-stimulated NF-jB acti-
vation did not occur in MEFs lacking both TRAF2
and TRAF5 (Fig. 4A). By contrast, hLF and ActE-
treated hLF significantly stimulated NF-jB DNA
binding and nuclear translocation of RelA in
TRAF2 ⁄ TRAF5-deficient MEFs at levels comparable
to those in wild-type MEFs (Fig. 4A). In TRAF6-defi-
cient MEFs, neither hLF nor IL-1 induced NF-jB
activation (Fig. 4B and Fig. S1). As shown in Fig. 4C,
ActE-digested hLF also failed to induce NF-jB

activation in TRAF6-deficient MEFs. However, by
introducing a TRAF6 cDNA into those cells, NF-jB
activation was restored (Fig. 4C), suggesting that
TRAF6 has an important role in hLF-induced NF-jB
activation in MEFs.
TRAF6 is known to be indispensable for NF-jB
activation in the myeloid differentiating factor 88
(MyD88)-dependent signaling pathway; however, the
role of TRAF6 in the MyD88-independent ⁄ Toll ⁄ IL-1
receptor-domain-containing adaptor inducing inter-
feron-b (TRIF)-dependent signaling pathway has not
been clarified extensively [27]. TRIF interacts directly
with TRAF6 via its TRAF6-binding motifs in the
N-terminal region [28,29]. Jiang et al. [29] showed
that TRAF6-deficient MEFs that overexpressed TLR3
failed to activate NF-jB in response to poly(I:C),
indicating that TRAF6 is critical in TRIF-dependent
NF-jB activation downstream of TLR3. However, in
our previous study, using macrophages isolated from
TRAF6-deficient mice, TRAF6 was not required in
the TRIF-dependent signaling including NF-jB acti-
vation [30]. This discrepancy concerning the require-
ment of TRAF6 may reflect the cell-type-specific
regulation of TRIF-signaling. Indeed, in contrast to
the results using TRAF6-deficient MEFs shown in
Fig. 4C and Fig. S1, TRAF6-deficient macrophages
clearly responded to hLF in terms of IKK activation
and nuclear translocation of RelA, although the
earlier responses observed in TRAF6
+ ⁄ )

macrophages
Fig. 1. Human LF induces canonical NF-jB activation. (A) THP-1 cells were treated with TNF (3 ngÆmL
)1
) or endotoxin-depleted hLF
(500 lgÆmL
)1
) for the indicated periods of time and then nuclear extracts were prepared. The NF-jB DNA-binding activities in the nuclear
extracts were determined using EMSA and the RelA levels in the nuclear extracts were determined using immunoblotting. Using the same
nuclear extracts, EMSA was used to determine the activity of the constitutively produced DNA-binding protein, Oct-1, as a loading control.
(B) THP-1 cells were treated with endotoxin-depleted hLF, at the indicated concentrations, for 90 min. Separately, cells were stimulated with
TNF (3 ngÆmL
)1
) for 20 min. Nuclear extracts were prepared and analyzed by immunoblotting for the presence of RelA. Using the same
nuclear extracts, immunoblotting was carried out to detect histone H-1 as a loading control. (C) MEFs were treated with TNF (3 ngÆmL
)1
)or
endotoxin-depleted hLF (500 lgÆmL
)1
) for the indicated periods of time, and nuclear and cytoplasmic extracts were prepared as described in
the Materials and Methods. The IKK activities in the cytoplasmic extracts were determined. The NF-jB and Oct-1 DNA-binding activities of
the nuclear extracts were measured using EMSA and the RelA levels of the nuclear extracts were determined by immunoblotting. (D) An
LPS solution containing 0.1 mgÆmL
)1
of BSA was loaded or not loaded onto a PMB–agarose column. After determining the protein concen-
tration, the eluate was used to treat MEFs for 60 min. Separately, MEFs were stimulated with IL-1b (3 ngÆmL
)1
) for 20 min. Nuclear extracts
were prepared and analyzed by immunoblotting to detect the levels of RelA. Using the same nuclear extracts, immunoblotting was carried
out to detect histone H-1 levels as a loading control. Quantification of the bands was performed using densitometric analysis (Image Gauge
4.0). Similar results were obtained in three separate experiments. (E) NaCl ⁄ P

i
containing hLF was loaded or not loaded onto a PMB–agarose
column. After determining the protein concentration, the eluate was used to treat MEFs for 60 min. Separately, MEFs were stimulated with
IL-1b (3 ngÆmL
)1
) for 20 min. Nuclear extracts were prepared and analyzed by immunoblotting to detect the levels of RelA. Using the same
nuclear extracts, immunoblotting was carried out to detect histone H-1 levels as a loading control. Quantification of the bands was per-
formed using densitometric analysis (Image Gauge 4.0). Similar results were obtained in three separate experiments. (F) MEFs were pre-
treated with actinomycin D (5 lgÆmL
)1
) or cycloheximide (5 lgÆmL
)1
) for 30 min and then treated with IL-1b (3 ngÆmL
)1
) or endotoxin-
depleted hLF (500 lgÆmL
)1
) for the indicated periods of time. Nuclear extracts were prepared and the RelA levels were determined using
immunoblotting. Using the same nuclear extracts, immunoblotting was carried out to detect the levels of histone H-1 as a loading control.
IB, immunoblotting.
Biological action of hLF is mediated through TLR4 K. Ando et al.
2054 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS
were weakened (Fig. 4D). These results suggest that
hLF-induced NF-jB activation may involve the
TRIF-dependent pathway.
TRIF-dependent NF-jB activation proceeds down-
stream of TLR3 or TLR4. However, TLR3-triggered
signaling is independent of MyD88, whereas TLR4
activates both MyD88-dependent and MyD88-indepen-
dent ⁄ TRIF-dependent signaling pathways [27]. There-

fore, we next examined the role of MyD88 in the
hLF-stimulated signaling pathway leading to NF-jB
A
C
B
D
E
F
K. Ando et al. Biological action of hLF is mediated through TLR4
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2055
activation. When MyD88-deficient MEFs were stimu-
lated with hLF, the earlier activation observed at
40 min was not obvious; however, significant IKK
activation occurred 80 min after stimulation with hLF
(Fig. 5A). Delayed, but significant, NF-jB DNA bind-
ing and nuclear translocation of RelA was also
observed in MyD88-deficient MEFs (Fig. 5A). By con-
trast, when TRIF-deficient MEFs were stimulated with
AB C
D
Fig. 2. Carbohydrate chains of hLF are responsible for NF- jB activation. (A) hLF was digested with actinase E as described in the Materials
and methods. The resultant sample was subjected to SDS ⁄ PAGE followed by Coomassie Brilliant Blue (CBB) staining. (B) MEFs were trea-
ted for 60 min with endotoxin-depleted hLF (500 lgÆmL
)1
) containing 12.5 lgÆmL
)1
of oligosaccharides or with the endotoxin-depleted frac-
tion of ActE-digested hLF (ActE–hLF) containing 12.5 lgÆmL
)1
of oligosaccharides. Separately, cells were stimulated with TNF (3 ngÆmL

)1
)
for 20 min. Nuclear extracts were prepared and EMSA was performed in the presence or absence of anti-RelA, as described in the Materials
and methods. SS, supershifted band. (C) THP-1 cells were treated for 90 min with endotoxin-depleted hLF (500 lgÆmL
)1
), ActE–hLF contain-
ing 12.5 lgÆmL
)1
of oligosaccharides, or purified carbohydrate chains derived from endotoxin-depleted hLF (hLF–CC) containing 50 lgÆmL
)1
of oligosaccharides. Separately, cells were stimulated with TNF (3 ngÆmL
)1
) for 20 min. Nuclear extracts were prepared and then subjected
to EMSA to analyze the NF-jB DNA-binding activities or to immunoblotting to detect the RelA levels. (D) LPS or hLF were treated or left
untreated with endo-b-galactosidase and then hLF was subjected to endotoxin depletion as described in the Materials and methods. MEFs
were stimulated with various concentrations of LPS or hLF for 60 min. Nuclear and cytoplasmic extracts were prepared as described in the
Materials and methods. The IKK activities in the cytoplasmic extracts were determined using the IKK immunoprecipitation assay. Using the
same cytoplasmic extracts, immunoblotting was carried out to detect glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels as a load-
ing control. Nuclear extracts were analyzed by immunoblotting to detect the RelA levels. Using the same nuclear extracts, immunoblotting
was carried out to detect histone H-1 levels as a loading control. Quantification of the bands was carried out using densitometric analysis
(Image Gauge 4.0). Similar results were obtained in three separate experiments. IB, immunoblotting.
Biological action of hLF is mediated through TLR4 K. Ando et al.
2056 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS
hLF, IKK activation was observed at 40 min,
but declined rapidly (Fig. 5B). From these results,
hLF-induced NF-jB activation may occur through the
MyD88-dependent earlier step and through the
MyD88-independent ⁄ TRIF-dependent later step. Thus,
TLR4 could be a possible candidate as the receptor
responsible for hLF-stimulated signal transduction.

Figure 6A clearly shows that hLF did not induce
IKK activation, NF-jB DNA binding or nuclear
translocation of RelA in TLR4-deficient MEFs at any
time-point studied. In addition, hLF failed to activate
c-Jun N-terminal kinase (JNK) in TLR4-deficient
MEFs, whereas it significantly induced JNK activation
in wild-type MEFs (Fig. S2). By contrast, hLF-stimu-
lated NF-jB activation was not impaired in MEFs
lacking TLR2, which triggers only the MyD88-depen-
dent signaling pathway (Fig. 6B). These results demon-
strated that TLR4 is responsible for hLF-evoked
signal transduction.
TLR4 can activate two separate transcription factors:
NF-jB and interferon regulatory factor 3 (IRF3); the
former is activated by the MyD88-dependent pathway
and the TRIF-dependent pathway and the latter is acti-
vated by the TRIF-dependent pathway [27]. NF-jB
induces several pro-inflammatory cytokines, such as
TNF or IL-1b, whereas IRF3 induces interferon-b,
thereby leading to the induction of interferon-inducible
genes such as interferon-c-inducible protein-10 (IP10).
Therefore, we next examined whether or not hLF indeed
induced TNF and IP10 production. As shown in
Fig. 7A, 500 lgÆmL
)1
of hLF induced the production of
a large amount of IP10 in THP-1 cells, although the
amount produced was lower than that in cells treated
with 75 EUÆmL
)1

of LPS. Similarly to IP10, hLF stimu-
lated the production of a significantly higher amount of
TNF than present in the control but the levels were
lower than in LPS-stimulated cells (Fig. 7B).
LF has been described as the molecule that inter-
feres with the biological actions of LPS [3,8–10].
Indeed, the levels of IP10 and TNF produced by
THP-1 cells simultaneously treated with LPS and
hLF were significantly lower than those produced by
cells treated with LPS alone (Fig. 7A,B). When the
NF-jB activities were examined using the 3x jB-Luc
luciferase reporter vector, the action of LPS was also
impaired in the presence of hLF, which alone induced
a lower, but significant, level of NF-jB activation
(Fig. 7C). By contrast, hLF failed to inhibit TNF-
induced NF-jB activation (Fig. S3), suggesting that
hLF may specifically attenuate the LPS–TLR4 signal-
ing pathway. Interestingly, ActE-digested hLF, in
which the amount of oligosaccharide was equivalent
to that of intact hLF, failed to inhibit LPS-dependent
NF-jB activation, while it induced NF-jB activation
at a level comparable to that induced by intact hLF
(Fig. 7C). Similarly, purified hLF carbohydrate chains
that can stimulate NF-jB activation also failed to
inhibit the action of LPS (Fig. 7D). By contrast,
endo-b-galactosidase treatment did not affect the
inhibitory action of hLF on LPS-stimulated NF-jB
activation, whereas it impaired hLF-dependent NF-jB
activation (Fig. 7E). These results suggest that the
polypeptide moiety of hLF is required for inhibiting

LPS action, whereas the carbohydrate chains of hLF
act to stimulate TLR4.
Spik et al. [31] reported the primary structures of
LF glycans from humans, mice, cows and goats. They
A
B
Fig. 3. TRAF6, but not TRAF2, is involved in hLF-triggered NF-jB
activation. (A) THP-1 cells were co-transfected with the 3x jB-Luc
luciferase reporter vector and the b-galactosidase expression vec-
tor, together with the expression vector encoding the FLAG-tagged
dominant negative form of TRAF2 (TRAF2DN) or TRAF6
(TRAF6DN), as described in the Materials and methods. After 24 h,
cells were lysed in SDS ⁄ PAGE sample buffer, and the resultant cell
lysates were subjected to immunoblotting to detect the FLAG epi-
tope or to detect b-actin levels as a loading control. (B) THP-1 cells
were co-transfected with the 3x jB-Luc luciferase reporter vector
and the b-galactosidase expression vector, together with the
expression vector encoding TRAF2DN or TRAF6DN, as described
for Fig. 3A. After 24 h, TNF (10 ngÆmL
)1
), IL-1b (6 ngÆmL
)1
), endo-
toxin-depleted hLF (500 lgÆmL
)1
) or endotoxin-depleted hLF-CC,
containing 50 lgÆmL
)1
of oligosaccharides, was added to the
culture and incubated for a further 5 h. Cells were harvested and

NF-jB-dependent luciferase production was measured as described
in the Materials and methods. Data are expressed as the mean ±
SD of triplicate determinations. Bars represent fold induction com-
pared with the control (*P < 0.01). IB, immunoblotting.
K. Ando et al. Biological action of hLF is mediated through TLR4
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2057
described the species-specific differences in the struc-
tures of LF glycans (i.e. all LFs contain biantennary
glycans of the N-acetyllactosamine type; however, only
hLF contains a-1,3-fucosylated N-acetyllactosamine
residues within them). In addition, only hLF possesses
poly-N-acetyllactosaminic glycans. By contrast, bovine
AB
CD
Fig. 4. TRAF6 is indispensable in hLF-induced NF-jB activation in MEFs, whereas TRAF6-independent NF-jB activation occurs in mouse
macrophages stimulated with hLF. (A) Wild-type and TRAF2 ⁄ 5
) ⁄ )
MEFs were treated for 60 min with endotoxin-depleted hLF (500 lgÆmL
)1
)
or with endotoxin-depleted ActE–hLF containing 12.5 lgÆmL
)1
of oligosaccharides. Separately, cells were stimulated with TNF (3 ngÆmL
)1
)or
IL-1b (3 ngÆmL
)1
) for 20 min. Nuclear extracts were prepared and then subjected to EMSA to analyze NF-jB DNA-binding activities or to
immunoblotting to detect the RelA levels. Using the same nuclear extracts, immunoblotting was carried out to detect histone H-1 levels as
a loading control. (B) Wild-type and TRAF6

) ⁄ )
MEFs were treated with endotoxin-depleted hLF (500 lgÆmL
)1
) for 60 min. Separately, cells
were stimulated for 20 min with TNF (3 ngÆmL
)1
) or IL-1b (3 ngÆmL
)1
). Nuclear extracts were prepared and then subjected to EMSA to ana-
lyze NF-jB DNA-binding activities or to immunoblotting to detect RelA levels. Using the same nuclear extracts, immunoblotting was carried
out to detect histone H-1 levels as a loading control. (C) Wild-type MEFs, TRAF6
) ⁄ )
MEFs and TRAF6
) ⁄ )
MEFs ectopically overexpressing
TRAF6 were treated for 60 min with endotoxin-depleted ActE–hLF containing 12.5 lgÆmL
)1
of oligosaccharides. Separately, cells were stimu-
lated with IL-1b (3 ngÆmL
)1
) for 20 min. Nuclear extracts were prepared and then subjected to EMSA to analyze NF-jB DNA-binding activi-
ties or to immunoblotting to detect RelA levels. Using the same nuclear extracts, immunoblotting was carried out to detect histone H-1
levels as a loading control. (D) Spleen macrophages, differentiated from the splenocytes of TRAF6
+ ⁄ )
and TRAF6
) ⁄ )
mice, were treated
with TNF (3 ngÆmL
)1
), IL-1b (3 ngÆmL

)1
), or endotoxin-depleted hLF (500 lgÆmL
)1
) for the indicated periods of time, and then nuclear and
cytoplasmic extracts were prepared as described in the Materials and methods. The IKK activities were determined in the cytoplasmic
extracts using the IKK immunoprecipitation assay and nuclear extracts were used for immunoblotting to detect RelA levels. Using the same
nuclear extracts, immunoblotting was carried out to detect histone H-1 levels as a loading control. IB, immunoblotting.
Biological action of hLF is mediated through TLR4 K. Ando et al.
2058 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS
A
B
Fig. 5. NF-jB activation induced by hLF
occurs through MyD88-dependent and
MyD88-independent ⁄ TRIF-dependent path-
ways. (A) Wild-type and MyD88
) ⁄ )
MEFs
were treated with endotoxin-depleted hLF
(500 lgÆmL
)1
), TNF (3 ngÆmL
)1
) or IL-1b
(3 ngÆmL
)1
) for the indicated periods of time
and then cytoplasmic and nuclear extracts
were prepared as described in the Materials
and methods. The IKK activities were deter-
mined in the cytoplasmic extracts using the

IKK immunoprecipitation assay. The NF-jB
DNA-binding activities in the nuclear
extracts were determined using EMSA and
the RelA levels in the nuclear extracts were
determined using immunoblotting. Using
the same nuclear extracts, immunoblotting
was carried out to detect histone H-1 levels
as a loading control. (B) TRIF
) ⁄ )
MEFs were
treated with LPS (75 EUÆmL
)1
), TNF
(3 ngÆmL
)1
), poly(I:C) (50 lgÆmL
)1
)or
endotoxin-depleted hLF (500 lgÆmL
)1
) for
the indicated periods of time. Cytoplasmic
extracts were prepared and the IKK
activities were determined. Using the same
cytoplasmic extracts, immunoblotting was
carried out to detect GAPDH levels as a
loading control. IB, immunoblotting.
A
B
Fig. 6. TLR4 is responsible for hLF-induced

NF-jB activation. (A) Wild-type and
TLR4
) ⁄ )
MEFs were treated with LPS
(1500 EUÆmL
)1
), IL-1b (3 ngÆmL
)1
) or endo-
toxin-depleted hLF (500 lgÆmL
)1
) for the
indicated periods of time. Nuclear extracts
were prepared and then subjected to EMSA
to analyze NF-jB DNA-binding activities or
to immunoblotting to detect RelA levels.
Using the same nuclear extracts, immuno-
blotting was carried out to detect histone
H-1 levels as a loading control. Cytoplasmic
extracts from TLR4
) ⁄ )
MEFs were prepared
and the IKK activities were determined.
(B) Wild-type and TLR2
) ⁄ )
MEFs were trea-
ted with LPS (1500 EUÆmL
)1
), peptidoglycan
(PGN) (10 lgÆmL

)1
), or endotoxin-depleted
hLF (500 lgÆmL
)1
) for the indicated periods
of time. Nuclear extracts were prepared and
analyzed by immunoblotting to detect RelA
levels. Using the same nuclear extracts,
immunoblotting was carried out to detect
histone H-1 levels as a loading control. IB,
immunoblotting.
K. Ando et al. Biological action of hLF is mediated through TLR4
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2059
A
DE F
BC
Fig. 7. Human LF moderately activates TLR4 via its carbohydrate chains whereas it attenuates LPS-triggered TLR4 activation independently
of the carbohydrate chains. (A) THP-1 cells were treated for 24 h with endotoxin-depleted hLF (500 lgÆmL
)1
), LPS (75 EUÆmL
)1
), or endo-
toxin-depleted hLF (500 lgÆmL
)1
) plus LPS (75 EUÆmL
)1
). The levels of IP10 released in the media were determined by ELISA. Data are
expressed as the mean ± SD of triplicate determinations. (B) THP-1 cells were treated with endotoxin-depleted hLF (500 lgÆmL
)1
), LPS

(75 EUÆmL
)1
), or endotoxin-depleted hLF (500 lgÆmL
)1
) plus LPS (75 EUÆmL
)1
) for 24 h. The levels of TNF released in the media were deter-
mined by ELISA. Data are expressed as the mean ± SD of triplicate determinations. (C) THP-1 cells were co-transfected with 3x jB-Luc
luciferase reporter vector and b-galactosidase expression vector. After 24 h, endotoxin-depleted hLF (500 lgÆmL
)1
), endotoxin-depleted
ActE–hLF containing 12.5 lgÆmL
)1
of oligosaccharides, LPS (150 EUÆmL
)1
), endotoxin-depleted hLF (500 lgÆmL
)1
) plus LPS (150 EUÆmL
)1
),
or endotoxin-depleted ActE–hLF containing 12.5 lgÆmL
)1
of oligosaccharides plus LPS (150 EUÆmL
)1
) was added to the culture, which was
incubated for a further 5 h. NF-jB-dependent luciferase production was measured as described in Fig. 3B. Data are expressed as the
mean ± SD of triplicate determinations. Bars represent fold induction compared with the unstimulated control. (D) THP-1 cells were co-trans-
fected with the 3x jB-Luc luciferase reporter vector and the b-galactosidase expression vector. After 24 h, endotoxin-depleted hLF-CC con-
taining 50 lgÆmL
)1

of oligosaccharides, LPS (150 EUÆmL
)1
), or hLF-CC containing 50 lgÆmL
)1
of oligosaccharides plus LPS (150 EUÆmL
)1
)
was added to the culture, which was incubated for a further 5 h. NF-jB-dependent luciferase production was measured as described in the
legend to Fig. 7C. Data are expressed as the mean ± SD of triplicate determinations. Bars represent fold induction compared with the
unstimulated control. (E) Human LF was treated or left untreated with endo-b-galactosidase and subjected to endotoxin depletion as
described in the Materials and methods. THP-1 cells were then stimulated for 90 min with endo-b-galactosidase-untreated ⁄ -treated hLF
(300 lgÆmL
)1
), LPS (75 EUÆmL
)1
), or endo-b-galactosidase-untreated ⁄ -treated hLF (300 lgÆmL
)1
) plus LPS (75 EUÆmL
)1
). Nuclear extracts
were prepared, and the levels of RelA were analyzed using immunoblotting. Using the same nuclear extracts, immunoblotting was carried
out to detect histone H-1 levels as a loading control. Quantification of the bands was carried out using densitometric analysis (Image Gauge
4.0). Similar results were obtained in three separate experiments. (F) THP-1 cells were co-transfected with 3x jB-Luc luciferase reporter
vector and b-galactosidase expression vector. After 24 h, endotoxin-depleted hLF (500 lgÆmL
)1
), endotoxin-depleted bLF (500 lgÆmL
)1
), LPS
(150 EUÆmL
)1

), endotoxin-depleted hLF plus LPS, or endotoxin-depleted bLF plus LPS was added to the culture, which was incubated for a
further 5 h. NF-jB-dependent luciferase production was measured as described in Fig. 3B. Data are expressed as the mean ± SD of tripli-
cate determinations. Bars represent fold induction compared with the unstimulated control. Endo-b, endo-b-galactosidase; IB, immunoblotting.
Biological action of hLF is mediated through TLR4 K. Ando et al.
2060 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS
LF (bLF) contains glycans of the oligomannosidic
type, representing the unique member of the transfer-
rin family containing both N-acetyllactosamine and
oligomannosidic-type glycans [31]. As shown in
Fig. 7F, bLF did not induce NF-jB activation but
strongly attenuated LPS-induced NF-jB activation,
similarly to hLF. These results clearly demonstrate
that both hLF and bLF conserve the polypeptide
structure that contributes to the impairment of LPS
action, while only hLF contains ‘active carbohydrate
chains’ to stimulate TLR4-mediated signaling path-
ways. As the candidate of ‘active carbohydrate chains’,
we focused on poly-N-acetyllactosaminic glycans.
Besides hLF, poly-N-acetyllactosaminic glycans are
known to exist in human erythrocytes [32,33]. There-
fore, we examined whether or not carbohydrate chains
isolated from human erythrocytes induce NF-jB acti-
vation. When wild-type MEFs were treated with
human erythrocyte-derived carbohydrate chains,
nuclear translocation of RelA was clearly demon-
strated, whereas the same treatment did not induce the
nuclear translocation of RelA in TLR4-deficient MEFs
at any time-point indicated (Fig. S4A). Human eryth-
rocyte-derived carbohydrate chains also induced
nuclear translocation of RelA in THP-1 cells (Fig. S4B,

lane 1 versus lane 2). However, simultaneous addition
of the carbohydrate chains did not affect LPS-induced
RelA nuclear translocation (Fig. S4B, lane 3 versus
lane 4). From these results, we postulate that the poly-
N-acetyllactosaminic carbohydrate moiety may act as a
moderate activator of TLR4.
Discussion
LF has been described as the molecule that modulates
our immune system in vivo. However, two contrasting
descriptions for this have been put forward: one con-
cerns immuno-activating (or pro-inflammatory) prop-
erties and the other concerns immunosuppressive (or
anti-inflammatory) properties that are based on attenu-
ation of the LPS action [1,4]. It is necessary to deter-
mine the precise molecular mechanism of how these
contrasting functions are exerted. Most importantly,
the cell-surface receptor that recognizes LF and then
triggers intracellular signals to induce immunological
reactions must be identified. However, this has not yet
been fully elucidated.
The immuno-activating (or pro-inflammatory) func-
tion of LF represents the pro-inflammatory cytokine
production induced by LF. In this study, we demon-
strated that hLF induced significant activation of
NF-jB, a master regulator of immune reactions that
plays a critical role in pro-inflammatory cytokine pro-
duction. By characterizing MEFs in which the adaptor
or receptor molecules involved in NF-jB activation
are genetically deficient, we were able to narrow down
the signaling process responsible for hLF-induced

NF-jB activation, and reached the conclusion that
TLR4 is indispensable.
TLR4 is an essential receptor for LPS recognition
[34,35]. In addition, TLR4 is implicated in the recogni-
tion of taxol, a diterpene purified from the bark of the
western yew (Taxus brevifolia) [36,37]. Furthermore,
TLR4 has been shown to be involved in the recogni-
tion of endogenous ligands, such as heat shock
proteins (Hsp60 and Hsp70), the extra domain A of
fibronectins, oligosaccharides of hyaluronic acid, hepa-
ran sulfate and fibrinogen. However, very high concen-
tration of these endogenous ligands are required to
activate TLR4 [38]. In addition, it has been shown that
contamination of the Hsp70 preparation with LPS
confers the ability to activate TLR4 [39].
In this study, we used hLF passed through a
PMB–agarose column to minimize contamination with
endotoxin. It is of note that the magnitude of NF-jB
activation induced by a pre-PMB–agarose fraction of
hLF (500 lgÆmL
)1
), which contained 13 EUÆmL
)1
of
endotoxin, was comparable to that induced by a post-
PMB–agarose fraction of hLF (500 lgÆmL
)1
), which
contained < 0.3 EUÆmL
)1

of endotoxin (Fig. 1E). This
result indicates that the effect of endotoxin
(13 EUÆmL
)1
) on NF-jB activation is negligible in the
presence of hLF (500 lgÆmL
)1
). By contrast, the action
of hLF (500 lgÆmL
)1
) is not influenced by endotoxin
(13 EUÆmL
)1
). Indeed, endotoxin-depleted hLF reduced
the levels of TNF and IP10 production induced by LPS
(Fig. 7, A and B). LF is secreted in the apo-form from
epithelial cells in most exocrine fluids, such as saliva,
bile, pancreatic and gastric fluids, tears and, particu-
larly, in milk [1]. In human milk, the hLF concentration
may vary from 1 mgÆmL
)1
(mature milk) to 7 mgÆmL
)1
(colostrum) [40]. We have shown, in the present study,
that hLF can induce NF-jB activation at much lower
concentrations (i.e. 25–500 lgÆmL
)1
) than found in
human milk, and therefore it is likely that hLF acts as
the endogenous activator for TLR4 in the intestines of

the breast-fed infants while concomitantly acting as a
competitor of LPS. Interestingly, ActE-digested hLF
does not inhibit LPS and it potently activates NF- jB,
similarly to intact hLF (Fig. 7C). By contrast, endo-
b-galactosidase-treated hLF retained the inhibitory
property towards LPS, whereas its activity, in terms of
NF-jB activation, was impaired compared with that of
intact hLF (Fig. 7E). These results suggest that hLF
may have divergent intramolecular modules in its carbo-
hydrate chains and its polypeptide moiety: the former
K. Ando et al. Biological action of hLF is mediated through TLR4
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2061
with an immuno-activating property and the latter with
an anti-inflammatory property.
The current view on the TLR4 activation induced by
LPS is as follows: LBP (a serum protein or a cell-associ-
ated protein) binds to the LPS aggregates. An LPS
monomer is then transferred to CD14. CD14 is required
for the efficient transfer of the LPS monomer to MD-2,
which is either soluble form or associated with the ecto-
domain of TLR4. The formation of a trimeric
LPS–MD-2–TLR4 complex is thought to be the final
event of the recognition of extracellular LPS [41]. In the
present study, we were unable to reveal the molecular
mechanism of how hLF triggers TLR4 activation. Inter-
estingly, Baveye et al. [10] reported that hLF interacts
with soluble CD14 and inhibits the expression of
E-selectin and intercellular adhesion molecule (ICAM)
induced by the CD14–LPS complex. Therefore, we can
postulate that the CD14–hLF complex, instead of the

CD14–LPS complex, may be presented to TLR4,
thereby triggering TLR4 activation through the carbo-
hydrate chains of hLF. However, hLF may interfere
with the formation of the CD14–LPS complex, resulting
in the attenuation of LPS signaling. As LF is known to
interact with LPS, the binding of LPS with LBP may
also be interfered with in the presence of hLF [9].
Although the involvement of TLR4 in LF signaling
has already been reported by Curran et al. [42], their
conclusion was different from ours. They examined the
bLF-activated signal transduction, including NF-jB
signaling, using macrophages from congenic TLR4
) ⁄ )
C.C3-Tlr4
lps-d
mice. In contrast to our results, they
showed that bLF-stimulated IjB degradation was
rather enhanced in TLR4
) ⁄ )
macrophages, suggesting
that TLR4 is not required for bLF-induced NF-jB
activation [42]. At present, we cannot explain this dis-
crepancy. However, we should emphasize the role of
the carbohydrate chains of LF in TLR4 activation.
Species-specific differences exist in the structure of the
carbohydrate chains of LF [31]. As shown in Fig. 7F,
bLF, lacking poly-N-acetyllactosaminic glycans, fails
to activate NF-jB and attenuates the action of LPS.
By contrast, the poly-N-acetyllactosaminic glycan-
enriched fraction derived from human erythrocytes

activated NF-jB, but did not affect the LPS-induced
activation of NF-jB (Fig. S4A, B).
There is intense interest in the effects of breast-feeding
on infants and the mechanisms behind these effects.
Compared with formula-fed infants, breast-fed infants
are known to have a high level of resistance to infectious
diseases [43]. In particular, several lines of evidence sug-
gest that breast-feeding provides strong protection
against diarrheal disease [44]. Human milk contains
numerous components that support the infant’s host
defense [44]. Among them, hLF may play an important
role in stimulating the infant’s own immune system by
activating TLR4-mediated signaling while preventing
the high levels of inflammation induced by LPS. Thus,
we should highlight TLR4 as the key molecule required
for the biological activity of hLF.
Materials and methods
Reagents
LF from human milk, LF from bovine milk, LPS from
Escherichia coli serotype O111:B4 and anti-FLAG M2
were purchased from Sigma Chemical Co. (St Louis, MO,
USA). Peptidoglycan from Staphylococcus aureus was
obtained from Fluka. The human IP10 ELISA kit was
obtained from Invitrogen Corporation (Carlsbad, CA,
USA). Recombinant human IL-1b and recombinant human
TNF-a were purchased from R&D Systems, Inc. (Minneap-
olis, MN, USA). Antibodies specific for NF-jB p65 (C20),
histone H-1 (AE-4) and glyceraldehyde-3-phosphate dehy-
drogenase (GAPDH) (clone 6C5) were from Santa Cruz Bio-
technology (Santa Cruz, CA, USA). Anti-IKKc (C73-764)

was from BD Biosciences Pharmingen (San Diego, CA,
USA). Antibodies against JNK and phospho-JNK
(Thr183 ⁄ Tyr185) were obtained from Cell Signaling Tech-
nology (Danvers, MA, USA). The TNF-a Human Biotrak
Easy ELISA kit was obtained from GE Healthcare Bio-
Sciences KK, Japan. Actinase E was obtained from Kaken-
Seiyaku Co., Japan. Endo-b-galactosidase (EC 3.2.1.103)
from Escherichia freundii and the Endospecy
Ò
ES-50M kit
were purchased from Seikagaku Co., Japan. The PMB–aga-
rose column (Detoxi-GelÔ Endotoxin Removing Gel) was
from Pierce Biotechnology Inc. (Rockford, IL, USA).
Removal of contaminating endotoxins from hLF
First of all, we examined the endotoxin content in commer-
cially obtained hLF using the Endospecy
Ò
ES-50M kit.
While the endotoxin activity of E. coli LPS was approxi-
mately 1 510 000 EUÆmg
)1
, hLF usually contained endo-
toxin activity of 15–26 EUÆmg
)1
of protein. The NaCl ⁄ P
i
(PBS) solution containing 20 mg of hLF was loaded onto
the PMB–agarose column (1 mL of gel). The protein con-
centration of the eluate was then determined according to
the method of Bradford [45]. After passing through the col-

umn, the endotoxin activity in hLF was < 0.6 EUÆmg
)1
(i.e. more than 96% of LPS was removed from hLF).
ActE digestion of hLF
NaCl ⁄ P
i
containing hLF (15 mgÆmL
)1
) was subjected to
digestion with ActE (5 mgÆmL
)1
) for 48 h at 37 °C. After
inactivating the enzyme by boiling for 10 min, the resultant
Biological action of hLF is mediated through TLR4 K. Ando et al.
2062 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS
reaction mixture was centrifuged for 20 min at 3000 g. The
supernatant was filtered through a 0.22-lm filer membrane
(MILLEX-GV; Millipore, Billerica, MA, USA), the filtrate
was loaded onto the PMB–agarose column and the eluate
was collected. Then, the concentration of the saccharides
was determined using a phenol ⁄ sulfuric acid method [46]
and expressed as the equivalent concentration of glucose.
Isolation of the carbohydrate chains from hLF
In order to obtain peptide ⁄ amino acid-free N-linked
oligosaccharides from hLF, hLF was subjected to hydrazin-
olysis and purified as described previously [22]. After con-
firming that amino acids were not detectable in the purified
fraction, the concentration of the oligosaccharides was
determined using a phenol ⁄ sulfuric acid method [46] and
expressed as the equivalent concentration of glucose.

Endo-b-galactosidase treatment of hLF and LPS
hLF (4 mgÆmL
)1
) or LPS (15–450 EUÆmL
)1
) was treated
with endo-b-galactosidase (0.1 UÆmL
)1
) in 0.1 m sodium
acetate buffer (pH 5.8), containing 0.07 m NaCl, for 48 h at
37 °C. The resultant reaction mixtures were then dialyzed
against NaCl ⁄ P
i
. The dialyzed endo-b-galactosidase-treated
hLF was passed through the PMB–agarose column, as
described above, to remove contaminating endotoxin.
Cell culture, transient transfection and reporter
gene assay
The human monocytic leukemia cell line, THP-1, was main-
tained in RPMI-1640 supplemented with 10% fetal bovine
serum, penicillin (50 UÆmL
)1
) and streptomycin
(50 lgÆmL
)1
) in a humidified atmosphere of 5% CO
2
at
37 °C. MEFs were maintained in DMEM supplemented
with 10% fetal bovine serum, penicillin (50 UÆmL

)1
), strep-
tomycin (50 lgÆmL
)1
) and 50 lm b-mercaptoethanol in a
humidified atmosphere of 10% CO
2
at 37 °C. The 3x
jB-Luc luciferase reporter vector (0.5 lg) and the b-galac-
tosidase expression vector pAct ⁄ b-gal (0.1 lg) were
co-transfected into THP-1 cells using the FuGENE6 trans-
fection reagent (Roche Diagnostics, IN, USA). After 24 h
of incubation, cells were treated with TNF, IL-1b, hLF or
purified carbohydrate chains from hLF (hLF-CC) and then
incubated for a further 5 h. The cells were then harvested
and the cellular luciferase activities were measured using a
chemiluminescence photometer and normalized to the
b-galactosidase expression. Data were analyzed using the
Student’s t-test. To evaluate the effect of dominant negative
forms of TRAF2 or TRAF6 on NF-jB activation, cells
were transfected with pME-FLAG-TRAF2DN or
pME-FLAG-TRAF6DN during transfection with the
reporter plasmids, as described previously [47].
Differentiation of splenocytes from
TRAF6-deficient mice into macrophages
Spleen cells from 2-week-old TRAF6
+ ⁄ )
and TRAF6
) ⁄ )
mice [30] were incubated for 6 h in MEMa (Gibco) con-

taining 10% fetal bovine serum, and nonadherent cells were
further cultured with 10 ngÆmL
)1
of macrophage colony-
stimulating factor. Adherent cells obtained after 6 days of
culture were used as macrophages. The protocol of animal
preparations for the experiment was approved by the Ethics
Committee of our institute.
Preparation of the nuclear and cytoplasmic
extracts
Nuclear extracts were prepared according to the method of
Kida et al. [48], with slight modifications. Cells treated with
various agents were washed with NaCl ⁄ P
i
on ice and then
suspended in 800 lL of buffer A (10 mm Hepes ⁄ KOH, pH
7.9, 10 mm KCl, 1.5 mm MgCl
2
,1mm dithiothreitol, 0.2 mm
phenylmethanesulfonyl fluoride, 1 lgÆmL
)1
of aprotinin,
1 lgÆmL
)1
of leupeptin and 1 lgÆmL
)1
of pepstatin). After
incubation for 35 min on ice, Nonidet P-40 was added to the
samples to a concentration of 0.1%, then the samples were
left on ice for 5 min. After centrifugation (5 min, 2000 g,

4 °C), the cytoplasmic extracts were immediately adjusted to
the condition for immunoprecipitation by adding the immu-
noprecipitation (2·) buffer described below. The precipitates
containing nuclear proteins were further washed once with
buffer A containing 0.1% Nonidet P-40 and then extracted
with buffer C (20 mm Hepes ⁄ KOH, pH 7.9, 420 mm NaCl,
1.5 mm MgCl
2
, 0.2 mm EDTA, 1 mm dithiothreitol, 0.2 mm
phenylmethanesulfonyl fluoride and 25% glycerol).
EMSA
EMSA was performed as described previously [49]. Binding
reaction mixtures containing 4 lg of nuclear extract protein,
2 lg of poly(dI-dC) and
32
P-labeled probe, were incubated
for 20 min at room temperature. For the antibody-mediated
supershift assay, nuclear extracts were pre-incubated with
1 lg of anti-RelA for 20 min at 4 °C before adding the
32
P-labeled probe. The reaction mixtures were analyzed
electrophoresis through native 4% polyacrylamide gels. The
sequence of the NF-jB probe was 5¢-AATTCTCAGAG
GGGACTTTCCGAGAGG-3¢. The sequence of the Oct-1
probe was 5¢-CTAGATATGCAAATCATTG-3¢.
Immunoblotting
Cells were washed with NaCl ⁄ P
i
and cell extracts were pre-
pared using the SDS ⁄ PAGE sample buffer described below.

After normalization of protein content according to the
protein assay, samples were resolved by SDS ⁄ PAGE and
K. Ando et al. Biological action of hLF is mediated through TLR4
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2063
subjected to immunoblotting analyses. The immunocom-
plexes on the poly(vinylidene difluoride) membranes were
visualized using enhanced chemiluminescence detection.
Quantification of the bands was performed using densito-
metric analysis (Image Gauge 4.0).
IKK assay
Cells extracts were prepared using immunoprecipitation
buffer [50] with a slight modification (the Nonidet P-40
concentration was increased to 1.0%). After normalization
of the protein content according to the protein assay, endog-
enous IKK was immunoprecipitated with anti-IKKc, and
the in vitro kinase assay was performed as described
previously, using glutathione S-transferase (GST)–IjBa,
a member of IjB family of proteins, as the substrate [50].
Isolation of the carbohydrate chains from human
erythrocyte membrane proteins
The erythrocyte-rich fraction of healthy human blood
(blood group O) was obtained from the Japanese Red
Cross Tokyo Metropolitan Blood Center. Erythrocyte
membranes were then isolated and delipidated. In order to
obtain peptide ⁄ amino acid-free oligosaccharides, the delipi-
dated membranes were subjected to hydrazinolysis after
protease digestion. After confirming that amino acids were
not detectable in the purified fraction, the concentration of
the oligosaccharides was determined.
Acknowledgements

We thank T. Hiraga, K. Ihashi, K. Sato, N. Komagat-
a, H. Majima, Y. Namekawa, A. Suzuki, A. Kashim-
ura, M. Yamamuro, M. Watanabe, K. Ito, Y. Murase
and T. Setoguchi for technical assistance. We also
thank Dr S. Miyamoto for providing us with the 3 x
jB-Luc luciferase reporter vector. This work was sup-
ported in part by a grant from the Japan Private
School Promotion Foundation.
References
1 Legrand D, Elass E, Carpentier M & Mazurier J (2005)
Lactoferrin: a modulator of immune and inflammatory
responses. Cell Mol Life Sci 62, 2549–2559.
2 Ward PP, Uribe-Luna S & Conneely OM (2002)
Lactoferrin and host defense. Biochem Cell Biol 80,
95–102.
3 Vorland LH (1999) Lactoferrin: a multifunctional glyco-
protein. APMIS 107, 971–981.
4 Legrand D, Elass E, Carpentier M & Mazurier J (2006)
Interactions of lactoferrin with cells involved in immune
function. Biochem Cell Biol 84, 282–290.
5 Gado I, Erdei J, Laszlo VG, Paszti J, Czirok E,
Kontrohr T, Toth I, Forsgren A & Naidu AS (1991)
Correlation between human lactoferrin binding and
colicin susceptibility in Escherichia coli. Antimicrob
Agents Chemother 35, 2538–2543.
6 Leitch EC & Willcox MD (1998) Synergic antistaphylo-
coccal properties of lactoferrin and lysozyme. J Med
Microbiol 47, 837–842.
7 Lu Y-C, Yeh W-C & Ohashi PS (2008) LPS ⁄ TLR4 sig-
nal transduction pathway. Cytokine 42, 145–151.

8 Na YJ, Han SB, Kang JS, Yoon YD, Park S-K, Kim
HM, Yang K-H & Joe CO (2004) Lactoferrin works
as a new LPS-binding protein in inflammatory activa-
tion of macrophages. Int Immunopharmacol 4, 1187–
1199.
9 Elass-Rochard E, Legrand D, Salmon V, Roseanu A,
Trif M, Tobias PS, Mazurier J & Spik G (1998) Lacto-
ferrin inhibits the endotoxin interaction with CD14 by
competition with the lipopolysaccharide-binding
protein. Infect Immun 66, 486–491.
10 Baveye S, Elass E, Fernig DG, Blanquart C, Mazurier J
& Legrand D (2000) Human lactoferrin interacts with
soluble CD14 and inhibits expression of endothelial
adhesion molecule, E-selectin and ICAM-1, induced by
the CD14-lipopolysaccharide complex. Infect Immun 68,
6519–6525.
11 Sorimachi K, Akimoto K, Hattori Y, Ieiri T & Niwa A
(1997) Activation of macrophages by lactoferrin: secre-
tion of TNF-a, IL-8 and NO. Biochem Mol Biol Int 43,
79–87.
12 Shinoda I, Takase M, Fukuwatari Y, Shimamura S,
Ko
¨
ller M & Ko
¨
nig W (1996) Effects of lactoferrin and
lactoferricin
Ò
on the release of interleukin 8 from
human polymorphonuclear leukocytes. Biosci Biotechnol

Biochem 60, 521–523.
13 Actor JK, Hwang S-A, Olsen M, Zimecki M, Hunter
RL Jr & Kruzel ML (2002) Lactoferrin immunomodu-
lation of DTH response in mice. Int Immunopharmacol
2, 475–486.
14 Legrand D, Vigie K, Said EA, Elass E, Masson M,
Slomianny M-C, Carpentier M, Briand J-P, Mazurier J
& Hovanessian AG (2004) Surface nucleolin partici-
pates in both the binding and endocytosis of lactoferrin
in target cells. Eur J Biochem 271, 303–317.
15 Kleinman HK, Weeks BS, Cannon FB, Sweeney TM,
Sephel GC, Clement B, Zain M, Olson MO, Jucker M
& Burrous BA (1991) Identification of a 110-kDa non-
integrin cell surface laminin-binding protein which rec-
ognizes an A chain neurite-promoting peptide. Arch
Biochem Biophys 290, 320–325.
16 Take M, Tsutsui J, Obama H, Ozawa M, Nakayama T,
Maruyama I, Arima T & Muramatsu T (1994) Identifi-
cation of nucleolin as a binding protein for midkine
(MK) and heparin-binding growth associated molecule
(HB-GAM). J Biochem 116, 1063–1068.
Biological action of hLF is mediated through TLR4 K. Ando et al.
2064 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS
17 Schneider WJ & Nimpf J (2003) LDL receptor relatives
at the crossroad of endocytosis and signaling. Cell Mol
Life Sci 60 , 892–903.
18 Suzuki YA, Shin K & Lo
¨
nnerdal B (2001) Molecular
cloning and functional expression of a human

intestinal lactoferrin receptor. Biochemistry 40, 15771–
15779.
19 Suzuki YA, Lopez V & Lo
¨
nnerdal B (2005) Mamma-
lian lactoferrin receptors: structure and function. Cell
Mol Life Sci 62, 2560–2575.
20 Baldwin AS Jr (1996) The NF-jB and IjB proteins:
new discoveries and insights. Annu Rev Immunol 14,
649–681.
21 Hayden MS & Ghosh S (2004) Signaling to NF-jB.
Genes Dev 18, 2195–2224.
22 Eda S, Kikugawa K & Beppu M (1996) Binding charac-
teristics of human lactoferrin to the human monocytic
leukemia cell line THP-1 differentiated into macrophag-
es. Biol Pharm Bull 19, 167–175.
23 Narahashi Y, Shibuya K & Yanagita M (1968) Studies
on proteolytic enzymes (pronase) of Streptomyces
griseus K-1. II. Separation of exo- and endopeptidases
of pronase. J Biochem 64, 427–437.
24 Hayden MS & Ghosh S (2008) Shared principles in
NF-jB signaling. Cell 132, 344–362.
25 Tada K, Okazaki T, Sakon S, Kobarai T, Kurosawa K,
Yamaoka S, Hashimoto H, Mak TW, Yagita H,
Okumura K et al. (2001) Critical roles of TRAF2 and
TRAF5 in tumor necrosis factor-induced NF-jB
activation and protection from cell death. J Biol Chem
276, 36530–36534.
26 Arch RH, Gedrich RW & Thompson CB (1998) Tumor
necrosis factor receptor-associated factors (TRAFs) – a

family of adapter proteins that regulates life and death.
Genes Dev 12, 2821–2830.
27 Kawai T & Akira S (2007) Signaling to NF-jB by Toll-
like receptors. TRENDS Mol Med 13, 460–469.
28 Sato S, Sugiyama M, Yamamoto M, Watanabe Y,
Kawai T, Takeda K & Akira S (2003) Toll ⁄ IL-1 recep-
tor domain-containing adaptor inducing IFN-b (TRIF)
associates with TNF receptor-associated factor 6 and
TANK-binding kinase 1, and activates two distinct
transcription factors, NF-jB and IFN-regulatory fac-
tor-3, in the Toll-like receptor signaling. J Immunol 171,
4304–4310.
29 Jiang Z, Mak TW, Sen G & Li X (2004) Toll-like
receptor 3-mediated activation of NF-jB and IRF3
diverges at Toll-IL-1 receptor domain-containing adap-
ter inducing IFN-b.
Proc Natl Acad Sci USA 101 ,
3533–3538.
30 Gohda J, Matsumura T & Inoue J (2004) TNFR-associ-
ated factor (TRAF) 6 is essential for MyD88-dependent
pathway but not Toll ⁄ IL-1 receptor domain-containing
adaptor-inducing IFN-b (TRIF)-dependent pathway in
TLR signaling. J Immunol 173, 2913–2917.
31 Spik G, Coddeville B & Montreuil J (1988) Compara-
tive study of the primary structures of sero-, lacto-, and
ovotransferrin glycans from different species. Biochimie
70, 1459–1469.
32 Fukuda M, Dell A, Oates JE & Fukuda MN (1984)
Structure of branched lactosaminoglycan, the carbohy-
drate moiety of band 3 isolated from adult human ery-

throcytes. J Biol Chem 259, 8260–8273.
33 Beppu M, Mizukami A, Ando K & Kikugawa K (1992)
Antigenic determinants of senescent antigen of human
erythrocytes are located in sialylated carbohydrate
chains of band 3 glycoprotein. J Biol Chem 267, 14691–
14696.
34 Poltorak A, He X, Smirnova I, Liu M-Y, van Huffel C,
Du X, Birdwell D, Alejos E, Silva M, Galanos C et al.
(1998) Defective LPS signaling in C3H ⁄ HeJ and
C57BL ⁄ 10ScCr mice: mutation in Tlr4 gene. Science
282, 2085–2088.
35 Hoshino K, Takeuchi O, Kawai T, Sanjo H, Ogawa T,
Takeda Y, Takeda K & Akira S (1999) Cutting Edge:
Toll-like receptor 4 (TLR4)-deficient mice are hypore-
sponsive to lipopolysaccaride: evidence for TLR4 as the
Lps gene product. J Immunol 162, 3749–3752.
36 Byrd-Leifer CA, Block EF, Takeda K, Akira S & Ding
A (2001) The role of MyD88 and TLR4 in the
LPS-mimetic activities of Taxol. Eur J Immunol 31,
2448–2457.
37 Kawasaki K, Akashi S, Shimazu R, Yoshida T, Miyake
K & Nishijima M (2000) Mouse Toll-like receptor
4 ⁄ MD-2 complex mediates lipopolysaccharide-mimetic
signal transduction by Taxol. J Biol Chem 275, 2251–
2254.
38 Takeda K & Akira S (2005) Toll-like receptors in innate
immunity. Int Immunol 17, 1–14.
39 Gao B & Tsan M-F (2003) Endotoxin contamination in
recombinant human heat shock protein 70 (Hsp70)
preparation is responsible for the induction of tumor

necrosis factor a release by murine macrophages. J Biol
Chem 278, 174–179.
40 Houghton MR, Gracey M, Burke V, Bottrell C &
Spargo RM (1985) Breast milk lactoferrin levels in
relation to maternal nutritional status. J Pediatr
Gastroenterol Nutr 4, 230–233.
41 Jerala R (2007) Structural biology of the LPS recogni-
tion. Int J Med Microbiol 297, 353–363.
42 Curran CS, Demick KP & Mansfield JM (2006)
Lactoferrin activates macrophages via TLR4-dependent
and -independent signaling pathways. Cell Immunol 242,
23–30.
43 Howie PW, Forsyth JS, Ogston SA, Clark A & Florey
CD (1990) Protective effect of breast feeding against
infection.
BMJ 300, 11–16.
44 Schack-Nielsen L & Michaelsen KF (2007) Advances in
our understanding of the biology of human milk and its
effects on the offspring. J Nutr 137, 503S–510S.
K. Ando et al. Biological action of hLF is mediated through TLR4
FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS 2065
45 Bradford MM (1976) A rapid and sensitive method for
the quantitation of microgram quantities of protein
utilizing the principle of protein-dye binding. Anal
Biochem 72, 248–254.
46 Dubois M, Gilles KA, Hamilton JK, Rebers PA &
Smith F (1956) Colorimetric method of determination
of sugars and related substances. Anal Chem 28, 350–
356.
47 Ishida T, Mizushima S, Azuma S, Kobayashi N, Tojo T,

Suzuki K, Aizawa S, Watanabe T, Mosialos G, Kieff E
et al. (1996) Identification of TRAF6, a novel tumor
necrosis factor receptor-associated factor protein that
mediates signaling from an amino-terminal domain of the
CD40 cytoplasmic region. J Biol Chem 271, 28745–28748.
48 Kida Y, Kuwano K, Zhang Y & Arai S (2001) Achole-
plasma laidlawii up-regulates granulysin gene expression
via transcription factor protein-1 in a human monocytic
cell line, THP-1. Immunology 104, 324–332.
49 Schreck R, Rieber P & Baeuerle PA (1991) Reactive
oxygen intermediates as apparently widely used messen-
gers in the activation of the NF-jB transcription factor
and HIV-1. EMBO J 10, 2247–2258.
50 DiDonato JA, Hayakawa M, Rothwarf DM, Zandi E
& Karin M (1997) A cytokine-responsive IjB kinase
that activates the transcription factor NF-jB. Nature
388, 548–554.
Supporting information
The following supplementary material is available:
Fig. S1. Kinetics of hLF-induced NF-jB activation in
wild type and TRAF6
) ⁄ )
MEFs.
Fig. S2. Kinetics of hLF-induced JNK activation in
wild type and TLR4
) ⁄ )
MEFs.
Fig. S3. Human LF does not attenuate TNF-induced
NF-jB activation.
Fig. S4. Carbohydrate chains from human erythrocytes

induce NF-jB activation.
This supplementary material can be found in the
online version of this article.
Please note: As a service to our authors and readers,
this journal provides supporting information supplied
by the authors. Such materials are peer-reviewed and
may be re-organized for online delivery, but are not
copy-edited or typeset. Technical support issues arising
from supporting information (other than missing files)
should be addressed to the authors.
Biological action of hLF is mediated through TLR4 K. Ando et al.
2066 FEBS Journal 277 (2010) 2051–2066 ª 2010 The Authors Journal compilation ª 2010 FEBS