We previously demonstrated that the blood-borne cancer cells become arrested in the lung vasculature via adhesion between the lung endothelial adhesion receptor dipeptidyl peptidase IV D
Trang 1monoclonal antibody 6A3 that blocks pericellular
fibronectin-mediated cancer cell adhesion
Ting-Ting Hung1,*, Jun-Yi Wu1,*, Ju-Fang Liu1and Hung-Chi Cheng1,2
1 Department of Biochemistry and Molecular Biology, College of Medicine, National Cheng Kung University, Tainan, Taiwan
2 National Cheng Kung University Hospital Cancer Center, National Cheng Kung University, Tainan, Taiwan
Introduction
Specific endothelial⁄ cancer cell–cell adhesions dictate
organ-preference cancer metastases [1] We previously
demonstrated that the blood-borne cancer cells become
arrested in the lung vasculature via adhesion between the lung endothelial adhesion receptor dipeptidyl peptidase IV (DPP IV) and pericellular polymeric
Keywords
dipeptidyl peptidase IV; epitope mapping;
monoclonal antibody; polymeric fibronectin;
steric hindrance
Correspondence
H.-C Cheng, Department of Biochemistry
and Molecular Biology, College of Medicine,
National Cheng Kung University, Tainan
70101, Taiwan
Fax: +886 6 274 1694
Tel: +886 6 235353
E-mail: hungchi@mail.ncku.edu.tw
*These authors contributed equally to this
work.
(Received 9 June 2009, revised 27 July
2009, accepted 3 September 2009)
doi:10.1111/j.1742-4658.2009.07352.x
We previously showed that the rat dipeptidyl peptidase IV (rDPP IV) monoclonal antibody (mAb) 6A3 greatly inhibits the pericellular polymeric fibronectin-mediated metastatic cancer cell adhesion to rDPP IV
L311QWLRRI in rDPP IV has been proposed as the putative fibronectin-binding site However, the inhibitory mechanism of 6A3 has been elusive Epitope mapping of 6A3 may help to understand the interaction between fibronectin and rDPP IV In the present study, we showed that 6A3 spe-cies-specifically recognized rDPP IV but inhibited fibronectin⁄ rDPP IV-mediated cell adhesions of various cancer types and species, which was independent of rDPP IV enzymatic activity The 6A3 epitope was stably exposed in both native and denatured rDPP IV On the basis of the resolved structures and the species variations in DPP IV sequences, we finely mapped the 6A3 epitope to a surface-exposed Thr331-dependent motif D329KTTLVWN, only 11 amino acids away from L311QWLRRI on the same plane as the fifth b-propeller blade The functionality of 6A3 epitope in rDPP IV was ultimately demonstrated by the ability of 6A3-rec-ognizable fragments to interfere with the inhibitory effect of 6A3 on full-length rDPP IV binding to pericellular polymeric fibronectin On the basis
of structural analysis, and the fact that the preformed fibronectin frag-ment⁄ rDPP IV complex was co-immunoprecipitated by 6A3 and fixing the rDPP IV structure with paraformaldehyde did not avert the inhibitory effect, the mechanism of 6A3 inhibition may not be the result of complete competition or conformational change
Structured digital abstract
l MINT-7261577 : DppIV (uniprotkb: P14740 ) binds ( MI:0407 ) to FNIII14 (uniprotkb: P04937 )
by anti bait coimmunoprecipitation ( MI:0006 )
Abbreviations
ADA, adenosine deaminase; DPP IV, dipeptidyl peptidase IV; FACS, fluorescence-activated cell sorting; FN, fibronectin; hDPP, human dipeptidyl peptidase; Hm, human mutant; mAb, monoclonal antibody; MBP, maltose-binding protein; mDPP, mouse dipeptidyl peptidase;
Mm, mouse mutant; pAb, polyclonal antibody; PFD, paraformaldehyde; polyFN, polymeric fibronectin; rDPP, rat dipeptidyl peptidase.
Trang 2fibronectin (polyFN) [1–5] This cancer cell adhesion to
DPP IV can be greatly blocked by monoclonal antibody
(mAb) 6A3 directed against rat DPP IV (rDPP IV) [2,3]
DPP IV is a homodimeric type II transmembrane
serine protease with multiple functions [6] For
exam-ple, it plays roles in proteolytic cleavage and
inactiva-tion of glucagon-like peptide 1 to maintain glucose
homeostasis [7] and in inactivating other bioactive
pep-tides and various cytokines for sustaining normal
phys-iological conditions [8] Much of the drug development
carried out with respect to type II diabetes has aimed
to effectively block the enzymatic activity of DPP IV
[9] However, the binding between DPP IV and
pericel-lular polyFN appears to be independent of this
enzy-matic activity [3,4] Indeed, the putative FN-binding
site in DPP IV has been proposed as a seven amino
acid sequence, L311QWLRRI, located in the fifth blade
of the b-propeller domain, which is structurally distinct
from the a⁄ b hydrolase domain that harbors the
cata-lytic triad [6,10]
FN, a large, multifunctional glycoprotein, is
secreted as the soluble, dimeric plasma FN or as the
insoluble polyFN PolyFN is either deposited in the
extracellular matrix or assembled on the surfaces of
metastatic cancer cells in the lung to which the
endo-thelial adhesion receptor DPP IV binds [11,12]
A consensus DPPIV-binding motif in FN has been
located in the 13th, 14th and 15th FN type III
repeats [1] Recently, we demonstrated that the
assem-bly of hematogenous cancer pericellular polyFN is
regulated by protein kinase Ce [5]
Several mAbs against DPP IV have been found to
interrupt extracellular matrix attachment and to arrest
the cell cycle of DPP IV-expressing cancer cells [13,14]
One of these mAbs was engineered into humanized
mAb and exerted an inhibitory effect on tumor growth
in a xenograft model [15] Before considering 6A3 as a
base for designing anti-adhesion drugs [16,17], we need
to better understand how 6A3 exerts its inhibitory
effect on the interaction between DPP IV and polyFN
For example, although metastatic cancer cells of
vari-ous species adhere to DPP IV [1], it is unclear whether
the inhibitory effect of 6A3 is also a cross-species
phe-nomenon in preventing cancer cell adhesion to DPP
IV Although 6A3 strongly inhibits DPP IV⁄ FN
bind-ing, the inhibitory mechanism of 6A3 still remains
elu-sive Furthermore, although DPP IV enzymatic
activity plays such important roles in many
physiologi-cal functions [18], it is not known whether 6A3-binding
affects this catalytic activity Moreover, because the
structural stabilities of eptiopes recognized by mAbs
are indispensable to the drug efficacy [19,20], it is
important to examine the stability of the 6A3 epitope
Epitope mapping of 6A3 may be the most direct approach for answering the above questions [21]
In the present study, we first show that the inhibi-tory effect of 6A3 on rDPP IV⁄ FN is a general phe-nomenon 6A3 species specifically recognized a structurally stable epitope in both native and dena-tured rDPP IV, independent of DPP IV enzymatic activity, which is suggestive of a noncompetitive inhibi-tion mechanism The 6A3 epitope in full-length rDPP
IV was finely mapped to the surface-exposed Thr331-dependent eight amino acid sequence D329KTTLVWN near the putative FN-binding site, L311QWLRRI, and only 11 amino acids apart According to structural analysis and co-immunoprecipitation of the preformed
FN fragment⁄ rDPP IV complex with 6A3, we suggest that the competitive mechanism is not responsible for 6A3 inhibition Preventing conformational change of the rDPP IV structure did not avert the inhibitory effect of 6A3 These observations suggest that the inhibitory effect of 6A3 is a result of steric hindrance rather than conformational change
Results
The inhibitory effect of 6A3 on rDPP IV⁄ cancer pericellular polyFN-mediated cell adhesion is a general phenomenon
To determine whether the 6A3 inhibition is a general phenomenon, we selected another two metastatic can-cer cell lines, human breast cancan-cer cells (MDA-MB-231) and mouse melanoma cells (B16F10), for rDPP
IV adhesion assays Similar to MTF7, MDA-MB-231 and B16F10 exhibited high adhesion activities to rDPP
IV, which was specifically abolished by 6A3 in a dose-dependent manner (Fig 1A) The adhesion activities exerted by these cells of various species and cancer types were also greatly blocked by a polyclonal anti-body (pAb) that recognizes FN of multiple species [3] (and data not shown) We next reconfirmed the inhibi-tory effect of 6A3 on binding between rDPP IV and pericellular polyFN After incubation with 6A3, solu-ble rDPP IV lost the ability to bind to immobilized MTF7 cell surfaces to the same degree that polyclonal anti-FN serum inhibited rDPP IV-binding (Fig 1B) These data suggest that the decreased adhesion activity
by 6A3 was indeed a result of the blockage of binding between rDPP IV and pericellular polyFN Biochemi-cally, we also demonstrated that the same FN co-immunoprecipitated with polyclonal anti-FN serum was affinity-precipitated by DPP IV-conjugated beads, which was abrogated by preincubating the beads with 6A3 (Fig 1C) This rDPP IV-precipitated FN was
Trang 3demonstrated to be pericellular polyFN because it was removed from the cell surfaces by pretreating the [35S]methionine metabolically labeled cells with a-chy-motrypsin (Fig 1C) These results suggest that 6A3 directly inhibits binding of rDPP IV to pericellular polyFN of cancer cells
6A3 species-specifically recognizes both native and denatured rDPP IV without interfering with its enzymatic activity
We next examined the species-specificity of 6A3-recog-nition of DPP IV in immunoprecipitation and in immunoblotting assays We found that 6A3 strongly recognized rDPP IV but only negligibly bound to native human (h)DPP IV and mouse (m)DPP IV in immunoprecipitation assays (Fig 2A–C) In immuno-blotting assays where DPP IVs were subjected to the denatured condition in SDS–PAGE, 6A3 exclusively recognized rDPP IV (Fig 2D) The ability of 6A3 to recognize both native and denatured rDPP IV suggests that the 6A3 epitope is stably exposed in full-length rDPP IV By contrast to the species-conserved FN-binding sequence, the rat-specific recognition of DPP
IV by 6A3 implies that the 6A3 epitope does not totally overlap with the putative FN-binding site This possibility was further supported by the fact that, although the FNIII14 (a DPP IV-binding competent
FN fragment) [1] did not to bind to the 6A3-preincu-bated rDPP IV (data not shown), the pre-bound FNIII14⁄ rDPP IV complex was
co-immunoprecipitat-ed by 6A3, even in the presence of high salt (200 mm NaCl) solution (Fig 2E) To determine whether the inhibitory effect of 6A3 interferes with the peptidase function, we set out to measure DPP IV enzymatic activity in the presence of 6A3 In line with the previ-ous results showing that FN⁄ DPP IV binding is inde-pendent of DPP IV catalytic activity [3], 6A3-binding
of rDPP IV also had no effect on its enzymatic activ-ity, indicating that this process may occur outside the enzymatic active site (Fig 2F)
The eight-bladed b-propeller domain harbors the 6A3-binding site
To identify the 6A3-binding site, we first generated recombinant maltose-binding protein (MBP)-fusion fragments of extracellular rDPP IV based on the reported structural and functional domains of the pre-dicted version 1 [22] and the resolved version 2 [23] (Fig 3A) These two versions differ mainly in the numbers of blades in the b-propeller domain Although version 1 was predicted to be seven-bladed
No DPP IV
Bio-DPP IV alone
Bio-DPP IV + Contr
ol 50 300 50 300
Bio-DPP IV + 6A3 (µg·mL
–1 )
Bio-DPP IV +
αFN pAb (µg·mL
–1 )
DPP IV binding (O
IP: αFN +Contr
ol +6A3
DPP IV-conjugated Affi-Gel
FN monomer
180 kDa
Cells alone Control
MD A-MB-231
MTF7 B16F10
A
B
C
Fig 1 6A3 inhibits pericellular FN-mediated lung-metastatic
cancer cell adhesion (A) rDPP IV adhesion activities of
MDA-MB-231, MTF7 and B16F10 (5 · 10 4 cells per well) in the absence or
presence of various concentrations of 6A3 or 0.3 mgÆmL)1
non-immune isotype IgG1 (control) at 37 C for 30 min were
mea-sured as described in the Experimental procedures (B) Binding of
soluble biotinylated rDPP IV (2 lg) to MTF7 (5 · 10 4 cells per
well), grown at 37 C overnight on 96-well plates (after PFD
fixa-tion) in the absence or presence of 6A3 (50 or 300 lgÆmL)1), the
same isotype mouse IgG1 (control; 300 lgÆmL)1) or polyclonal
FN antibody (50 or 300 lgÆmL)1), was detected with horseradish
peroxidase-conjugated streptavidin (C) Whole cell lysates from
1 · 10 6 2-h recovered suspended S 35 -labeled MTF7 cells,
pretreat-ed without ( )) or with (+) 10 lgÆmL )1 a-chymotrypsin (1 h at
37 C), were immunoprecipitated with 2 lg polyclonal FN antibody
or pulled-down with DPP IV-conjugated Affi-Gel 10 beads in the
presence of 0.3 mgÆmL)1 6A3 or control before being subjected
to SDS–PAGE and radiography Note that Affi-Gel conjugated with
control protein phosphorylase b did not pull down any pericellular
polyFN [1] (and data not shown) and the FN monomeric bands
shown in the radiography were reduced from high molecular
weight pericellular polyFN by b-mecaptoethanol [3,37].
Trang 4(amino acids 131–502) [22], version 2 has been shown
to be eight-bladed (amino acids 49–502) [24] (Fig 3A)
The reason that we took the predicted version into
consideration was that other members of the propyl
oligopeptidase family to which DPP IV belongs
con-tain a seven-bladed b-propeller domain and that
delet-ing portions of the predicted N-terminal a⁄ b-hydrolase
domain (amino acids 29–130; the first propeller blade
of version 2) resulted in a loss of enzymatic activity
and protein integrity, implying that this region,
although structurally belonging to the eight-bladed
b-propeller domain, may be functionally involved in
a⁄ b-hydrolase catalytic activity [22] Because the
a⁄ b-hydrolase domains of the two versions both
include certain portions of the N- and C-terminal
regions, we constructed two chimeric MBP-fusion frag-ments, A+D and G+D, to cover these domains (Fig 3A) After purification, we found that all but fragment A were endogenously degraded (Fig 3B), which is not uncommon in bacterial systems expressing mammalian proteins [25] The high structural stability
of fragment A explains why it is required for stabiliz-ing the a⁄ b-hydrolase structure and enzymatic activity even within the b-propeller domain [22,24] From the results of both immunoblotting and immunoprecipita-tion, we found that fragments B and C, and extracellu-lar full-length, were recognized by 6A3 (Fig 3B), suggesting that the 6A3 epitope resides within the last seven propeller blades in the eight-bladed b-propeller domain The fragments containing the a⁄ b-hydrolase
IP: mα hDPP IV 6A3 Contr mα hDPP IV 6A3
ol
IB: IF7 hDPP IV-Jurkat Mock-Jurkat
IB: 6A3 Contr
ol 6A3 CU31 IP:
Rat Kidney Extract
IP:
IB: r αmDPP IV
Contr
ol Rat IgG
Mouse Kidney Extract
IB: 6A3
Rat Mouse Human DPP IV
F
DPP IV + 6A3 DPP IV alone
IP: 6A3 IB: MBP
MT
NaCl 150 m M 200 m M E
110 kDa
110 kDa
110 kDa
110 kDa
55 kDa
IB: 6A3
1.00
0.25 0.50 0.75
0.00
Fig 2 6A3 specifically recognizes both native and denatured rat DPP IV independently of its enzymatic activity (A) Three hundred micro-liters of Sprague-Dawley (SD) rat kidney extracts (one kidney per milliliter being homogenized, as previously described [3]) was subjected
to immunoprecipitation at 4 C overnight with 2 lg of purified control mouse IgG1, 6A3 mAb or CU31 pAb and then to immunoblotting with 6A3 (B) Three hundred microliters of C57BL6 mouse kidney extracts (two kidneys per milliliter [3]) was subjected to immunoprecipi-tation at 4 C for overnight with 2 lg of purified 6A3, control, mouse DPPIV (mDPP IV) mAb from rat or rat IgG, and then to immunoblot-ting with mDPP IV mAb from rat (C) Three hundred microliters of human Mock-Jurkat or DPP IV-Jurkat cell lysates (1 mgÆmL)1from
1 · 10 7
cells) was subjected to immunoprecipitation at 4 C for overnight with 2 lg of 6A3, control, or human DPP IV (hDPP IV) mAb from mouse and then to immunoblotting with hDPP IV mAb IF7 from mouse (D) Mouse, rat kidney extracts or DPP IV-Jurkat cell lysates (made as described in A, B, and C) were subjected to immunoprecipitation at 4 C for overnight with 2 lg of 6A3, mDPP IV mAb from rat or hDPP IV mAb from mouse, respectively, and then to immunoblotting with 6A3 (E) rDPP IV preincubated with DPP IV-binding com-petent FN fragment MBP-FNIII14 (WT) or with DPP IV-binding sequence-mutated MBP-FNIII14 (MT) [1] were co-immunoprecipitated by 6A3 in the presence of 150 or 200 m M NaCl and then subjected to immunoblotting with a-MBP pAb (upper panel) or with 6A3 (lower panel) (F) DPP IV enzymatic activities were measured in the absence or presence of 6A3 as described in the Experimental procedures.
Trang 5domain were not recognized by 6A3, which is
consis-tent with the results showing that 6A3-binding did not
interfere with DPP IV enzymatic activity (Fig 2E)
6A3-binding epitope is located within a
Thr331-dependent linear eight-amino acid
region near the proposed FN-binding site
Because 6A3 specifically recognized rDPP IV
(Fig 2A–D), we next compared the aligned secondary
structures and sequences of the three species and
selected three nonconserved rDPP IV regions to test
6A3-binding (Fig S1A) Unfortunately, amino acid
swapping in these three regions in rDPP IV with those
of hDPP IV did not affect 6A3 recognition (Fig S1B) Therefore, we began with a serial deletion scheme and generated two truncated B fragments: B-144 and B-272 6A3 only bound to B-144 but not to B-272 (Fig S2A), suggesting that the 128 amino acid region between Val231 and Ala358 contains the 6A3 epitope
To further narrow down the 6A3 epitope, we then identified two nonconserved clusters within this 128 amino acid region (Fig S2B) Because the first cluster harboring B⁄ c was invalid for 6A3-binding (Fig S1),
we focused on the second cluster, according to which
we generated B-159 and B-166, and a 28 amino acid
IB:6A3 IP:6A3 IB:MBP
IB:MBP
1
1
Ver 1
Ver 2
7
7
D C
G
A B C D A + D G + D ex FL
50
100 75 kDa
50 75 100
50 75 100
*
*
* *
*
A
B
50
75 kDa
IB:6A3 B-159 B-166 B-272 B (329~358)
T331I V334T B-166
75 kDa
IB:6A3 IB:6A3
75
WT Hm Mm
B-166 kDa
C
Fig 3 6A3 recognizes the linear Thr331-dependent eight-amino acid epitope in the eight-bladed b-propeller domain (A) Scheme of rDPP IV molecular dissecting for constructing MBP-fusion proteins based on the predicted version 1 [22] and resolved version 2 [24] of the reported structural and functional domains In both versions, black rectangles represent intracellular domains and white rectangles transmembrane domains Although fragments A and G belong to N-terminal portions of a ⁄ b-hydrolase domains in both versions of DPP IV, fragment D is the invariant C-terminal a ⁄ b-hydrolase domain Fragment B represents the seven-bladed b-propeller domain of version 1 and fragment C is the eight-bladed b-propeller domain of version 2 Chimera fragments A+D and G+D represent the complete a ⁄ b-hydrolase domains exFL repre-sents the extracellular full-length DPP IV fragment (B) 6A3 immunoblotting (upper panel), 6A3 immunoprecipitation followed by anti-MBP immunoblotting (middle panel) and anti-MBP immunoblotting (lower panel) for 50 ng of amylose agarose bead-purified MBP-fusion fragments
as described in the Experimental procedures As a result of general protein degradations, all fragments were loaded so that the amounts of full-length proteins were approximately equal Note that full-length fragments are indicated by an asterisk to the right of the individual protein bands (C) 6A3 immunoblotting of B-159, B-166, B-272 and B(329–358) (for detailed positioning, see Fig S2B) Note that binding of B-166 and B(329–358) to 6A3 indicates that the eight-amino acid sequence (Asp329 to Asn336) within the 106-amino acid region between Val231 and Asn336 of the eight-bladed b-propeller domain harbors the 6A3 epitope (D) 6A3 immunoblotting of T331I and V334T mutants (for detailed positioning, see Fig S2D) (E) 6A3 immunoblotting of wild-type (WT), Hm and Mm (for detailed positioning, see Fig S2E).
Trang 6fragment B (amino acids 329–358) (Fig S2B) We
found that 6A3 recognized all of them (Figs 4C and
S2C), indicating that the 6A3 epitope is located within
the linear eight-amino acid region between Asp329 and
Asn 336 This conclusion was well supported by the
fact that swapping the eight-amino acid sequence in
B-166 (wild-type) with that of human (Hm) or mouse
(Mm) (Fig S2D) lead to a loss of their 6A3-binding
abilities (Fig 3D) We then constructed B-166
mutants, T331I and V334T, where the only amino
acids distinct from those in Mm were individually swapped with Ile and Thr (Fig S2E) The results demonstrate that 6A3 recognizes V334T but T331I (Fig 3E), strongly suggesting that the 6A3-binding epitope lies within a Thr331-dependent eight-amino acid region
In the compiled ribbon and surface representations
of the resolved DPP IV structure, this region is located
at the fifth propeller blade of the eight-bladed propel-ler domain (Fig 4A, B) Consistently, it resides at the
A
R 315
R 316
I 317
T 331
2 1
3
4
5 6
7
8
T 331
B
T 331
Cells alone Control 6A3 50 µg·mL –1 6A3 100 µg·mL –1 6A3 300 µg·mL –1
MTF7 cell adhesion to PFD-fixed DPP IV
E
Fig 4 6A3 epitope is in close proximity
with the proposed FN-binding site
L311QWLRRI but is far from the enzymatic
active site (A) Ribbon representation for the
eight-bladed b-propeller domain (dark yellow)
of the resolved rDPP IV (Protein Data Bank
code: 2GBC) The view is from the top of
the propeller domain with individual
propel-ler blades being alphabetically numbered.
The Thr331-dependent eight-amino acid 6A3
epitope located at upper portion of the fifth
propeller blade is labeled in green with
Thr331 highlighted in red The proposed
FN-binding sequence L 311 QWLRRI in cyan
is located at lower portion of the fifth
pro-peller blade The rest of the fifth propro-peller
blade is labeled in blue (B) Side view of the
eight-bladed b-propeller domain as
presented in ribbon mode (upper panel) or
in surface contour mode (lower panel) (C)
Top view and (D) side view of the surface
contour image for the entire rDPP IV
struc-ture The catalytic triad (S631, D709 and
H741) labeled in marine blue can be
visual-ized through the narrower channel that is
formed by the eight b-propeller blades (the
white arrow) where the catalytic dipeptide
products leave (C), or through the open
active site cleft into which DPP IV
sub-strates may enter (the yellow arrow) (D) All
the views were rendered with PYMOL 0.99
and color representations are the same as
those shown in (A) (E) rDPP IV-coated wells
were first pre-treated with 2% PFD at room
temperature for 30 min Percent specific
adhesion activities of MTF7 were then
measured similar to Fig 1A.
Trang 7outer surface of the eight-bladed propeller domain,
which is totally opposite to the inner surface of the
a⁄ b-hydrolase domain (Fig 4C, D) Interestingly, this
6A3 epitope and the proposed FN-binding site,
L311QWLRRI, both belong to exactly the same
propel-ler blade, and are only 11 residues apart (Fig 4A, B)
This structural proximity suggests that the inhibitory
effect of 6A3 is a result of steric hindrance To rule
out the remote possibility that the inhibitory effect is a
result of conformational change, we used
paraformal-dehyde (PFD), which forms methylene bridges between
any two residues with an amino group in their side
chains [26], to prevent conformational change of the
DPP IV structure 6A3 was still able to inhibit the
can-cer cell adhering to the PFD-fixed DPP IV (Fig 4E),
further supporting the idea that 6A3 most likely
exerts steric hindrance in the inhibition of FN⁄ DPP IV
binding
The Thr331-dependent linear eight-amino acid
region mediates the 6A3-binding in full-length
DPP IV
Although we identified the 6A3 epitope in a loop area
of the rDPP IV molecule according to the in silico
analysis (Fig 4A–D), we did not know whether this
epitope is available for 6A3-binding in the full-length
rDPP IV structure Therefore, we first ectopically
expressed full-length rDPP IV on HEK293 cell
sur-faces After preincubation with 6A3, B-166 wild-type
but not Hm, Mm or T331I, greatly inhibited 6A3
immunofluorescent staining of rDPP IV-expressing
HEK293 cells (Fig 5A) Next, purified full-length
rDPP IV together with reference mouse IgG were
immunoblotted with 6A3 that was preincubated with
B-166 wild-type, Hm or Mm Normalized with the
55 kDa IgG heavy chain, the wild-type demonstrated a
greater blocking effect compared to the other two
mutants (Fig 5B) Consistently, 6A3⁄ rDPP IV
immu-noprecipitation was inhibited in the presence of the
wild-type fragment (Fig 5C) The essential role of
Thr331 in 6A3-recognition was reconfirmed by the
results showing that T331I failed to inhibit 6A3⁄ rDPP
IV binding in immunofluorescent staining,
immunocy-tochemistry, immunoblotting and immunoprecipitation
assays (Fig 5A–C) To further corroborate the specific
binding between 6A3 and its epitope in rDPP IV, we
performed competition assays between DPP IV protein
and the T331I orV334T B-166 mutant peptides for
6A3 binding T331I, but not V334T, blocked the
inhi-bition of cancer cell adhesion to DPP IV and soluble
DPP IV binding to polyFN-expressing cancer cells
(Table 1) Taken together, these data indicate that the
Thr331-dependent linear eight-amino acid region indeed mediates 6A3-binding in full-length DPP IV
Discussion Specific adhesion between cancer cells and endothelia contributes to organ-preference cancer metastasis [1,3]
We previously generated rDPP IV mAb 6A3 that blocks the DPP IV⁄ FN-mediated adhesion of lung-metastatic cancer cells in the lungs [3] In the present study, we finely mapped the 6A3 epitope and analyzed the inhibitory mechanism of 6A3
One of the three mechanisms, namely competition, conformational change and steric hindrance, may explain the inhibitory effect of 6A3 The distinct spe-cies-specificities of the 6A3 epitope and the FN-binding and the co-immunoprecipitation of preformed
FN⁄ DPP IV complex with 6A3 (Fig 2E) make it less likely that 6A3 inhibits the FN⁄ rDPP IV binding via competition Nevertheless, before the putative FN-binding sequence is firmly validated, we cannot totally rule out that both binding sites partially overlap
On the other hand, most residues in the putative FN-binding sequence L311QWLRRI [10] are apparently buried in the propeller core except the R315RI (Fig 4B, lower panel) FN remains to bind to the PFD-fixed DPP IV, where no conformation can be changed to expose the buried residues (Fig 4E), indicating that additional residues, other than R315RI, may contrib-ute to the FN-binding However, the DPP IV frag-ment B, which contains L311QWLRRI, did not exhibit significant FN-binding activity (data not shown), suggesting either that the true FN-binding site is located outside the fragment B and affected by 6A3 via conformational change or that L311QWLRRI
is part of the true-binding site, the presentation of which might only be supported by the full-length DPP IV structure and affected by 6A3 via steric hindrance
Although the exact inhibitory mechanism of 6A3 cannot be proclaimed for certain, the preservation of the inhibitory effect on FN binding to PFD-fixed DPP
IV by 6A3 (Fig 4E) appears to disfavor the effect of conformational change By contrast, the R315RI together with several nearby sequences appear to belong to a discrete surface-exposed motif, which is located near the 6A3 epitope at a distance of approxi-mately 30 A˚ (Fig S3A, C), likely representing a dis-continuous FN-binding domain and favoring the mechanism of steric hindrance Interestingly, based on structural comparison and superimposition of rDPP
IV (2GBC) and hDPP IV (1NU6), this putative FN-binding motif appears to be relatively conserved
Trang 8(Fig S3A–D), implying that this motif may be a more
appropriate FN-binding domain
General application of antigen-specific mAbs in
can-cer therapy is highly anticipated [27] For example,
bevacizumab is generally used to inhibit angiogenesis
in treating patients with various cancer types [28] In
line with this concept, 6A3 inhibits the rDPP
IV-bind-ing of rat and human breast cancer cells, mouse
mela-noma cells (Fig 1A) and several other types of cancer
cells (data not shown) However, the failure of 6A3 in
recognizing hDPP IV makes it impossible to use for
direct application in cancer therapies Nevertheless, the
superimposition of rDPP IV and hDPP IV reveals that
the overall conformations of the two molecules are
rather similar, with subtle differences as a result of
amino acid side chain variations (Fig S3D) We
specu-late that mAbs generated against peptides containing
the 6A3 epitope-corresponding sequence D331
ESS-GRWN in hDPP IV may be evaluated in the future
for therapeutic purposes One important consideration
is that, before clinical application, pretests of these mAbs in animal models are required Accordingly, the antigenic peptide sequences should be carefully deter-mined so that the generated mAbs will also recognize rDPP IV and⁄ or mDPP IV and block lung metastases
of cancer cells in these animals
To serve as a safe drug, a potential mAb is expected
to exert its effect without causing adverse consequences
of physiological functions [29] DPP IV is involved in adenosine deaminase (ADA)-mediated T cell prolifera-tion [6] Antibodies recognizing epitopes including resi-dues Leu294 and Val341 were found to have inhibitory effects on ADA-binding of hDPP IV [30] These epi-topes do not overlap with the 6A3 epitope-correspond-ing sequence in hDPP IV, implyepitope-correspond-ing that the mAb generated against this hDPP IV sequence should not interrupt ADA binding to hDPP IV Although mono-clonal DPP IV antibody 4D10 arrests the cell cycle
A
C
Binding of 6A3 preincubated with B-166 mutants
IB: 6A3 preincubated with B-166 mutants
WT Hm T331I V334T IP: 6A3 preincubated with B-166 mutants
Ig heavy chain rDPP IV
rDPP IV
B-166 mutants
WT (A.F.I = 8.75) T331I (A.F.I = 22.53)
Mm (A.F.I = 20.46) Control (A.F.I = 1) MBP (A.F.I = 20.26) Hm (A.F.I = 20.26)
B
Fig 5 6A3-recognizable DPP IV fragments
block the bindings of 6A3 to both native and
denatured full-length DPP IV (A) FACS
analyses of ectopically full-length rDPP
IV-expressing HEK293 cells by staining the
cells with 6A3 preincubated with MBP,
B-166 Hm, Mm, wild-type (WT), T331I
mutants or the same isotype mouse IgG
(control) as described in the Experimental
procedures The dot plots were generated
with PE-Cy5 fluorescent intensities
(repre-senting 6A3-binding abilities) against FSC-H.
The arbitrary fluorescent intensities (A.F.I.)
of the DPP IV-positive HEK293 cells are
calculated as average fluorescent
inten-sity · total DPP IV-positive HEK293 cell
numbers for each staining Representative
images of each immunofluorescent staining
results are inserted inside each
correspond-ing dot plot (B) Immunoblottcorrespond-ing of the
purified full-length rDPP IV and the mouse
IgG heavy chain as a quantitative reference
protein with 6A3 preincubated with the
same MBP-fusion DPP IV fragments as in
(A) (C) 6A3 immunoblotting of the
immuno-precipitates where 0.2 lg of purified
full-length DPP IV was immunoprecipitated with
0.5 lg of 6A3 preincubated with 2 lg of
WT, Hm, T331I or V334T mutants Note
that, although the 6A3 binding-competent
WT and V334T blocked rDPP IV
immuno-precipitation by 6A3, Hm and T311I, which
were incapable of binding to 6A3, did not
exert this inhibitory effect.
Trang 9progression of cancer cells and reduces in vivo tumor
growth, it does not cause any apparent adverse effect
in nude mice [15] It is likely that it causes distinct
effects of cytotoxicity against DPP IV-expressing
can-cer cells and normal cells Altogether, we propose a
hypothesis that a mAb raised against the 6A3
epitope-corresponding sequence in hDPP IV may functionally
be unique and safe in cancer patients
The surface-exposed mAb epitope must be stable
enough in the circulation to resist those
cancer-associ-ated protein modification factors, such as the reactive
oxygen species resulting from pro-inflammatory
oxida-tive stress [31] and mechanical stress toward endothelia
[32,33] Together with shear stress from the capillary
circulation and the respiratory pressure of the lungs,
all the above inflammatory factors are potent with
respect to causing various degrees of conformation
alterations of endothelial proteins [34] Because 6A3 is
able to recognize either the native form of DPP IV in
immunoprecipitation assays or the denatured form in
immunoblotting assays (Fig 2A–D), the 6A3 epitope
in DPP IV appears to be relatively stable This
stabil-ity is further confirmed by the abilstabil-ity of 6A3 to
recog-nize cell-surface expressed DPP IV either in live cells
(Fig 5A) or formaldehyde-fixed tissues [2,4,35]
In conclusion, we have successfully identified the
epitope of the potent anti-metastatic mAb 6A3 that
blocks metastatic cancer cell adhesions to rat lung
endothelial DPP IV, most likely via steric hindrance
The 6A3-epitope corresponding sequence in hDPP IV
may potentially be used to generate functionally
unique and safe monoclonal anti-metastatic sera
Experimental procedures
Cell lines, antibodies and reagents
Rat lung-metastatic MTF7 cells, mouse B16F10 cells
and human MDA-MB-231 cells were obtained from
Dr B U Pauli (Cornell University, Ithaca, NY, USA) [1] They were grown in DMEM (Invitrogen, Carlsbad, CA, USA) containing 5% fetal bovine serum (FBS) Mock (Mock-Jurkat) or hDPP IV-expressing (hDPP IV-Jurkat) Jurkat cells were generous gifts from Dr C Morimoto (University of Tokyo, Tokyo, Japan) They were grown in RPMI 1640 medium (Invitrogen) containing 5% FBS mAb 6A3 and pAb CU31 were generated against rDPPIV [1] mDPP IV mAb from rat was from R&D Systems (Minne-apolis, MN, USA); hDPPIV mAb from mouse was from Santa Cruz Biotechnology (Santa Cruz, CA, USA); human
FN pAb from rabbit was from Sigma (St Louis, MO, USA); rabbit polyclonal anti-MBP serum and amylase aga-rose beads were from New England Biolabs Inc (Ipswich,
MA, USA); and IF7 mAb was a generous gift from Dr
C Morimoto (University of Tokyo, Tokyo, Japan) All other reagents were purchased from Sigma [35S]methionine was from ICN Biochemicals (Irvine, CA, USA) and Affi-Gel 10 beads were obtained from Bio-Rad (Hercules,
CA, USA)
Plasmid construction
pMAL-c2 vector and full-length rDPP IV cloned in pRC-CMV vector were obtained from Dr B U Pauli [1] All MBP-fusion fragments of DPP IV were PCR-amplified and inserted into the EcoRI and HindIII sites of pMAL-c2 vec-tor as described previously [1] Using wild-type B-166 as template, overlap extension PCR amplification was per-formed to generate B-166 Hm mutant, B-166 Mn mutant, B-166 T331I (T331) and B-166 V334T (V334T) as MBP-fusion proteins, as described above [1] The amino acids included in the PCR-amplified fragments are described
in Doc S1
Purification of MBP-fusion fragments and rDPP IV
The fusion protein purification procedures have been described previously [1] Briefly, the cell lysates of Escheri-chia coli cells expressing various MBP-fusion proteins were passed through amylose columns and eluted with 10 mm maltose in column buffer [1] Immunoblotting with poly-clonal anti-MBP serum was used to verify the purified pro-teins Rat lung DPP IV was purified from rat lung extracts by 6A3 immunoaffinity chromatography [3]
[35S]methionine metabolic labeling of MTF7 for rDPP IV-conjugated Affi-Gel 10 affinity precipitation
MTF7 Cells were first starved for methionine before addi-tion of 50–100 lCi of [35S]methionine at 37C overnight Cells were recovered in DMEM containing 20% FBS as previously described [3] Before making cell lysates [36],
Table 1 6A3-recognizable DPP IV fragments competitively
neutral-ize the 6A3 inhibitory effects on cancer cell adhesion to DPP IV.
Status of 6A3 DPP IV binding (A 492 )a DPP IV adhesion (%)b
6A3 + B-166 V334T 0.67 ± 0.06 c 11.2 ± 6.49 c
6A3 + B-166 T331I 1.60 ± 0.20 69.23 ± 14.75
a
Binding of biotinylated rDPP IV to MTF7 in the absence or
pres-ence of 6A3 preincubated with B-166 mutants b rDPP IV% specific
adhesion activities of MTF7 in the absence or presence of 6A3
preincubated with B-166 mutants. cCompared to the values of
assays without 6A3, these values are significantly decreased
(P < 0.05).
Trang 10some cells were first treated with 10 lgÆmL)1
a-chymotryp-sin for 30 min at 37C Cell lysates were then subjected to
rDPP IV-conjugated Affi-Gel 10 affinity precipitation in the
absence or presence of 0.3 mgÆmL)16A3 and then to SDS–
PAGE and autoradiography The aliquots of cell lysates
were also subjected to immunoprecipitation with polyclonal
anti-FN serum [3]
DPP IV enzymatic activity assays
Purified rDPP IV (0.5 lg per assay) in the presence or
absence of a two-fold molar ratio excess of 6A3 was
incu-bated in 250 lL of assay buffer for 30 min at 37C,
fol-lowed by stopping the reaction with 750 lL of 1 m acetate
buffer and then measuring the absorption at 405 nm [4,35]
Cancer cell adhesion assays
rDPP IV adhesion activities of MDA-MB-231, MTF7 and
B16F10 (5· 104
cells per well) in the absence or presence
of 0.3, 0.1 or 0.05 mgÆmL)1 6A3 or the same isotype IgG1
described previously [1,3,35]
Immunoprecipitation and immunoblotting
[35S]methionine metabolically labeled MTF7 cell lysates
(0.1 lgÆmL)1) were incubated with various antibodies (1–
2 lg) The immunoprecipitates were subjected to
immuno-blotting, as described previously [36] Full-length rDPP IV
(0.2 lg) and 0.05 lg of mouse IgG were subjected to 6A3
immunoblotting (dilution 1 : 1000) in the presence of
vari-ous MBP-fusion fragments (0.1 lg)
ELISA
A modified ELISA [1,3] was used to measure the binding
(5· 104cells per well) grown at 37C overnight on 96-well
plates (after PFD fixation) in the absence or presence of 50
same isotype mouse IgG1 (control) with horseradish
peroxi-dase-conjugated streptavidin
Immunofluorescent staining and
fluorescence-activated cell sorting (FACS) analysis
rDPP IV-expressing HEK293 cells were stained with
V334T mutants, followed by PE-Cy5-conjugated goat
anti-mouse IgG and FACS analysis (detected using a FL3
Bio-sciences, San Jose, CA, USA) [1,36]
Software for DPP IV structural analysis
The ribbon- and surface-representations of rDPP IV(Pro-tein Data Bank code: 2GBC) X-ray structure was visualized with pymol, version 0.99 (http://www.pymol.org/) The superimposition of rDPP IV and hDPP IV was performed
php)
Acknowledgements These studies were supported by National Science Council Grant NSC94-2314-B-006-122, National Sci-ence Council Grant NSC 95-2320-B-006-075-MY3, and The Program for Promoting Academic Excellence
& Developing World Class Research Centers and Cen-ter of Excellence for Clinical Trial and Research in Oncology Specialty DOH-TD-B-111-004 The authors wish to thank Dr Iain C Bruce for his professional editing on our manuscript and Dr Ming-Derg Lai for his thoughtful discussion regarding the manuscript The authors also thank Ms Megan Cheng for her language assistance in manuscript writing and Mr Chien-Fang Kuo and Mr Chu-kuei Lin for their help
in preparation and assembly of figures
References
1 Cheng HC, Abdel-Ghany M & Pauli BU (2003) A novel consensus motif in fibronectin mediates dipeptidyl peptidase IV adhesion and metastasis J Biol Chem 278, 24600–24607
2 Johnson RC, Zhu D, Augustin-Voss HG & Pauli BU (1993) Lung endothelial dipeptidyl peptidase IV is an adhesion molecule for lung-metastatic rat breast and prostate carcinoma cells J Cell Biol 121, 1423–1432
3 Cheng HC, Abdel-Ghany M, Elble RC & Pauli BU (1998) Lung endothelial dipeptidyl peptidase IV pro-motes adhesion and metastasis of rat breast cancer cells via tumor cell surface-associated fibronectin J Biol Chem 273, 24207–24215
4 Cheng HC, Abdel-Ghany M, Zhang S & Pauli BU (1999)
Is the Fischer 344⁄ CRJ rat a protein-knock-out model for dipeptidyl peptidase IV-mediated lung metastasis of breast cancer? Clin Exp Metastasis 17, 609–615
5 Huang L, Cheng HC, Isom R, Chen CS, Levine RA & Pauli BU (2008) Protein kinase Cepsilon mediates poly-meric fibronectin assembly on the surface of blood-borne rat breast cancer cells to promote pulmonary metastasis J Biol Chem 283, 7616–7627