|
|
|
|||||||
|
|||||||||
(Received for publication, June 27, 1994; and in revised form, January 9, 1995) From the
Granulocyte-macrophage colony-stimulating factor (GM-CSF) is
important in many immune and inflammatory processes. GM-CSF binds to
specific cellular receptors which belong to a recently described
supergene family. These receptors are potential targets for
pharmacologic design, and such design depends on a molecular
understanding of ligand-receptor interactions. One approach to
dissecting out critical intermolecular interactions is to develop
analogs of specific interaction sites of potential importance.
Monoclonal antibodies have been employed for these purposes in prior
studies. Here we present application of recombinant antibody technology
to the development of analogs of a site on GM-CSF bound by a
neutralizing anti-GM-CSF monoclonal antibody. Polyclonal antisera
with high titer neutralizing activity against human GM-CSF were
developed in BALB/c mice. Purified immunoglobulins were prepared and
used to immunize syngeneic mice. Anti-anti-GM-CSF was developed which
demonstrated biological antagonist activity against GM-CSF-dependent
cellular proliferation. RNA was extracted from spleen cells of mice
with biologically active anti-anti-GM-CSF, cDNA synthesized, and
polymerase chain reaction performed with primers specific for murine
Development of small molecular mimics of larger, polypeptide
ligands is one approach to pharmacophore design. Several strategies are
available for the development of such mimics, including the use of
small oligopeptide analogs derived from native
sequence(1, 2, 3, 4, 5) ,
development of peptidic and non-peptidic analogs based on molecular
structure data(6, 7) , and analysis of alternative
ligands(8) . Alternative ligands that bind to the same site as
the native ligand provide the opportunity to investigate structural and
chemical constraints for binding in the setting of diverse backbone
geometries. This has the potential to identify critical contact
residues based on similar structural and chemical characteristics
between the diverse ligands. Prior studies have investigated a
monoclonal antibody (mAb)( Recombinant antibodies have been developed
which are expressed in bacteria (18, 19) or on the
surface of filamentous bacteriophage (20, 21, 22, 23) . The advantages of
recombinant approaches to antibody development include the ability to
rapidly screen thousands of clones simultaneously, the potential to
detect binding moities poorly represented in the initial repertoire,
and the potential to express isolated variable regions. While intact
mAbs contain both light and heavy chain variable regions (V Here we describe the
development of a recombinant light chain library in Escherichia
coli derived from mice immunized with polyclonal anti-GM-CSF. This
library was screened with a previously described neutralizing
anti-GM-CSF mAb 126.213 (24) which inhibits GM-CSF binding to
HL-60 cells, neutralizes GM-CSF induced colony formation, and competes
with the
Figure 1:
Biological
activity of antisera. Proliferation of the human GM-CSF-dependent cell
line MO7E was performed as noted under ``Materials and
Methods'' in the presence of varying dilutions of murine
anti-GM-CSF (following the fifth boost) and murine anti-anti-GM-CSF
(following the ninth boost). Counts/min incorporated ± the
standard deviation of triplicate wells is shown for various dilutions
of antisera. In similar experiments, the inhibition induced by
anti-GM-CSF titered out at 1:20,000 to-:100,000
dilutions.
For reverse transcription, 10-20 µg of RNA in 10 µl
was utilized to synthesize cDNA primed with random hexamers in the
following reaction mixture: 3 µl of Maloney murine leukemia virus
reverse transcriptase with 6 µl of 5 For PCR amplification, the oligonucleotide primers
3315 and 5591 listed in Table 1were employed at 0.5
nM/ml final concentrations. The relative position of these
primers on Ig
Figure 2:
Library screening. A, first round
library screening was carried out on 30 filters lifted from 30 LB/amp
plates representing a total of
For some experiments, lysates were prepared of bacteria expressing
the recombinant antibody fragments. Lysates of E. coli XL1
Blue cells (Stratagene, La Jolla, CA) were prepared either from
unmanipulated bacteria or E. coli transformed with pDAB
The PCR products were ethanol precipitated
(to remove residual primer DNA) and digested with appropriate
restriction endonucleases (XbaI and EcoRI). These
were ligated into similarly restricted, alkaline phosphatase-treated
pDAB
Figure 3:
Characterization of rAb V
The
neutralizing mAb 126.213 specifically immunoprecipitates
Figure 4:
Inhibition of
Figure 5:
Nucleic acid and derived amino acid
sequences of clone 23.2. Sequencing was performed by double-stranded
DNA sequencing with Taq polymerase, as described previously,
using both the PCR primers and primers derived from the pDAB
The intact 23.2 sequence and the
individual CDR sequences were compared with the human GM-CSF sequence
using the Bestfit, Gap, Wordsearch, and Segments programs of the
Wisconsin package(41) . Several regions of weak sequence
similarity were noted which involved CDR regions of 23.2. Prior studies
of 126.213 used murine/human chimeric forms of GM-CSF to map
interaction sites(24) . These studies suggested that residues
77-83 were critical for 126.213 binding to GM-CSF. We noted weak
homology of the CDR I and CDR II with this epitope. An additional
region of weak sequence similarity was also seen between amino acids
54-61 of GM-CSF and the CDR III of 23.2. Interestingly, amino
acids 54-61 (on the B helix of GM-CSF) lie immediately adjacent
to amino acids 77-83 (on the C helix) in the crystal structure of
GM-CSF(42) . However, the weak sequence similarity seen here
indicated that the mimicry of GM-CSF by 23.2 might be better accounted
for on a structural level.
Alternative models were also generated by searching the
crystallographic data base for loops of the same size as the CDR I
region. The spatially conserved Cartesian positions at the
NH
Figure 6:
Structural basis for mimicry. The
structure of GM-CSF was determined from coordinates derived from the
crystal structure (J. M. LaLonde, K. Swaminathan, and D. Voet,
manuscript in preparation), displayed on the MacImdad program
(Molecular Applications Group, Palo Alto, CA) on a Macintosh Quadra 950
computer. The 23.2 V
Prior studies investigating
the epitope on GM-CSF recognized by 126.213 by mutagenesis (24) implicated residues 77-83, located on the C helix of
GM-CSF. Peptide mapping studies of this antibody suggest recognition by
the B and C helices as well as an epitope representing the first
Figure 7:
Binding of 126.213 to synthetic peptides
derived from the 23.2 sequence. Binding was performed by ELISA assay as
described under ``Materials and Methods.'' The values shown
are A
The ability of these peptides to
compete with GM-CSF for binding to HL-60 cells was examined using a
radioreceptor assay. HL-60 cells were preincubated with peptides prior
to the addition of
Figure 8:
Inhibition of
The bioactivity of
these peptides was assessed by their effect on GM-CSF-dependent
cellular proliferation. This was compared with their effect on
interleukin-2-dependent proliferation by the CTLL cell line, to control
for nonspecific toxic effects. The results are shown in Fig. 9.
At the concentrations used, none of the peptides were toxic to CTLL
cells with the exception of the CDR III peptide at 2 mg/ml. The CDR II
peptide had no inhibitory effect on either cell line. In contrast, the
CDR III and CDR I peptides inhibited GM-CSF-dependent cellular
proliferation. For the CDR III peptide, the IC
Figure 9:
Inhibition of GM-CSF-dependent cell
proliferation by peptides. The proliferation assay was performed as
noted under ``Materials and Methods'' on AML193 cells
(GM-CSF-dependent) and CTLL cells (IL-2-dependent), in the presence or
absence of increasing amounts of peptides as noted. Results from two
experiments are combined, with the mean ± standard error percent
inhibition of proliferation shown versus increasing peptide
concentration. A, CDR I peptide; B, CDR II peptide; C, CDR III peptide.
GM-CSF
binding and bioactivity have been analyzed at a molecular level.
Mutagenesis studies implicate the first (A) helix in binding of GM-CSF
to the high affinity GM-CSFR Recent studies from our group
have used synthetic peptides, anti-peptide antisera, and neutralizing
mAbs to map epitopes on GM-CSF critical for bioactivity.
In
prior studies, we described the molecular basis for antibody mimicry of
a viral
hemagglutinin(9, 10, 11, 12) . Other
groups have applied this technology to platelet fibrinogen
receptor(14) , the thyroid-stimulating hormone
receptor(15) , and epitopes on the hepatitis B surface
antigen(17) . Monoclonal antibodies were utilized in these
studies as mimics and to derive sequence information. The studies
presented here are the first to suggest that recombinant antibodies can
be similarly employed to develop alternative ligands. The prior studies
of antibody mimicry in general described mimicry of structures either
known or predicted to represent reverse turns. As antibody CDRs are
generally reverse turns, the ability of antibody CDRs to mimic other
reverse turn regions does not necessarily imply that CDRs can mimic
amino acid residues presented by other diverse backbone geometries. The
epitopes involved in this study are largely
Volume 270,
Number 12,
Issue of March 12, 1995 pp. 6628-6638
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
light chain V regions. Polymerase chain reaction products were
cloned into the pDAB
vector and an expression library
developed. This was screened with anti-GM-CSF neutralizing mAb 126.213,
and several binding clones isolated. One clone (23.2) which inhibited
126.213 binding to GM-CSF was sequenced revealing a murine light
chain of subgroup III. Comparison of the 23.2 sequence with the human
GM-CSF sequence revealed only weak sequence similarity of specific
complementarity determining regions (CDRs) with human GM-CSF.
Structural analysis revealed potential mimicry of specific amino acids
in the CDR I, CDR II and FR3 regions of 23.2 with residues on the B and
C helices of GM-CSF. A synthetic peptide analog of the CDR I was bound
by 126.213, specifically antagonized GM-CSF binding to cells and
blocked GM-CSF bioactivity. These studies indicate the feasibility of
using recombinant antibody libraries as sources of interaction site
analogs.
), 87.92.6, which mimicked a
neutralizing epitope on the reovirus type 3
hemagglutinin(9, 10, 11, 12) .
87.92.6 was bound both by a reovirus type 3 neutralizing mAb and the
reovirus type 3 receptor. Sequence similarity between 87.92.6 light
chain second complementarity determining region (CDR II) and the
reovirus type 3 hemagglutinin (13) allowed the development of
synthetic peptides and peptidomimetics which bound both the
neutralizing mAb and the reovirus type 3 receptor. These peptides and
peptidomimetics also demonstrated biological activity on reovirus type
3 receptor bearing cells. The use of anti-receptor mAbs as a source of
sequence-structural information to aid in peptide design has allowed
the development of similar biologically active peptides in several
systems, including the platelet fibrinogen receptor(14) , the
thyroid-stimulating hormone receptor(15) , and epitopes on the
human immunodeficiency virus (16) and hepatitis B surface
antigen(17) . and V
, respectively), recombinant antibodies can be
developed which express both V and V, or
V or V alone. This limits the potential
interaction sites of the recombinant antibody, allowing more precise
delineation of critical interaction regions.
chain of the GM-CSF receptor for GM-CSF
binding(25) . Screening with radioiodinated 126.213 yielded
several binding clones, including one that inhibited
immunoprecipitation of GM-CSF by 126.213. Comparison of the recombinant
V sequence with the human GM-CSF sequence revealed only
weak similarity with GM-CSF, but structural analysis suggested mimicry
of residues on the B and C helices of GM-CSF by a site chiefly made up
of the CDR I region of 23.2. A synthetic peptide corresponding to the
CDR I was bound by the neutralizing anti-GM-CSF mAb and specifically
inhibited GM-CSF binding and the growth of GM-CSF-dependent cells.
These studies suggest a structural basis for recombinant antibody
mimicry of a predominately helical molecule (human GM-CSF), demonstrate
a bioactive peptide analog of a GM-CSF site implicated in receptor
binding, and indicate the feasibility of using recombinant antibody
libraries as sources of interaction site analogs.
Bacterial Strains
E. coli DH5
competent cells (Life Technologies, Inc.) were used for transformation.
Bacteria were grown in Luria broth containing 100 mg/ml ampicillin
(LB/amp)(26, 27) . Solid media contained 1.5% agar
(Difco Inc.).Enzymes and Oligonucleotides
Restriction
endonucleases and T4 DNA ligase were purchased from Life Technologies,
Inc. Enzyme reaction conditions were according to those of the
supplier. Oligonucleotides for PCR primers and for Southern blotting
were synthesized by the DNA Synthesis Facility of the Wistar Institute.
The primers were selected by analysis of immunoglobulin sequences as
published by Kabat et al.(28) . The specific primers
are listed in Table 1. PCR amplification employed primers 3315
(relatively specific for the murine V III family) and 5591 (near
the 3` end of the C coding region). Note that primer 5591
introduces a stop codon at codon 207, resulting in a truncated light
chain lacking the carboxyl-terminal 8 amino acids including the
cysteine at position 214. This should result in production of light
chains which are predicted to remain monomers. Primers were tested on
various hybridoma cell lines in the laboratory prior to their use in
library construction. In these studies amplified sequences were
isolated from the gels cloned and sequenced to verify the utility of
the primers and their specificity for amplification of Ig variable
regions (data not shown).
Cell Lines and Proliferation Assay
CTLL cells and
the proliferation assay in response to IL-2 or rat spleen concanavalin
A was described previously(29) . AML 193 cells were obtained
from the American Type Culture Collection (ATCC), and MO7E cells were
from R. Zollner, Genetics Institute (Cambridge MA). AML 193 was grown
serum free in Iscove's modified Dulbecco's medium with
insulin (10 µg/ml), transferrin (5-10 units/ml), 1% OPI media
additive (oxalate, pyruvate, and insulin), and GM-CSF (0.5 ng/ml). MO7E
was grown in RPMI 1640 with 10% heat-inactivated fetal calf serum,
Pen/strep, L-glutamine, and GM-CSF 0.5 ng/ml. For
proliferation assays, 2 
10
AML 193 or MO7E cells
were cultured per well in 96-well round bottomed plates in the above
medium along with test antisera in a final volume of 200 µl.
Following a 3-5 day incubation, tritiated thymidine (1
µCi/well) was added for an additional 18 h, the cells harvested
onto glass fiber filters utilizing a PhD cell harvester, and
counts/minute incorporated determined in a standard liquid
scintillation system.Development of Anti-GM-CSF and
Anti-anti-GM-CSF
Recombinant human GM-CSF (obtained from Bachem
Biosciences, Philadelphia, PA) was used to immunize BALB/c mice as
described previously(30) . Serum was obtained 1 week following
each boost with antigen. Antisera from three to five animals were
pooled for the assays performed. Following the third boost, significant
neutralizing titers against human GM-CSF-dependent cellular
proliferation were demonstrated (Fig. 1). The mice were bled
after five boosts and IgG purified from serum by affinity
chromatography with Sepharose-protein A. This was used to immunize
syngeneic BALB/c mice (50 µg of purified IgG/immunization) and
serum obtained following each boost. The sera were assayed for
inhibition of GM-CSF-dependent proliferation (see below), and
significant (>50%) inhibition was seen following the eighth boost
against both MO7E and AML193 cells ( Fig. 1and data not shown).
Mice that exhibited neutralizing activity on this assay served as
spleen cell donors.
Amplification of Anti-anti-GM-CSF Immunoglobulin Light
Chain Variable Regions (V
Spleenocytes were
isolated from four anti-GM-CSF immunized mice who displayed
neutralizing activity against GM-CSF-dependent proliferation. A cell
suspension was prepared and lymphocytes isolated by Ficoll-Hypaque
density gradient centrifugation. RNA was extracted with the RNAzol kit
(Biotecx Laboratories Inc., Houston, TX), according to the
manufacturer's instructions. Following isolation, the RNA was
precipitated with isopropyl alcohol, pellets washed in 70% ethanol, and
rotary evaporated. The dried pellets were resuspended in 50 µl of
diethylpyrocarbonate-water and RNA quantified spectrophotometrically. ) reverse transcriptase
buffer, 1.5 µl of RNase inhibitor, and 3 µl of 0.1 M dithiothreitol (all from Life Technologies, Inc.), 3 µl (100
pmol) of random hexamers (from Pharmacia LKB Biotechnol), and 1 µl
of 40 mM dNTPs (10 mM in each dNTP,
from Boehringer Mannheim, GmbH W., Germany). Following a 10-min
preincubation at 25 °C, the reaction was carried out for 1 h at 42
°C, then 95 °C for 5 min followed by storage at -20
°C until use. cDNA is shown in Fig. 2. The PCR mixture (100
µl) consisted of 10 µl of PCR primers, 16 µl of dNTPs
(final concentration 200 µM in each dNTP), 10 µl of
PCR buffer (10 ; Perkin-Elmer Cetus), 61.5 µl of
dH
O, 2 µl cDNA, and 1.2 units of Taq polymerase (Perkin-Elmer Cetus). Amplification was carried out in
a Programmable Thermal Cycler (MJ Research, Watertown, MA). The
amplification program was 94 °C for 3 min followed by five cycles
of 94 °C for 60 s, 52 °C for 60 s, 72 °C for 60 s; followed
by 25 cycles of 94 °C for 60 s, 52 °C for 90 s, and 72 °C
for 120 s. Following 30 cycles, the temperature was held at 72 °C
for 7 min. Positive amplification was determined by agarose gel
electrophoresis. The PCR products were cloned into the pDAB plasmid, which is of utility for protein expression as has been
published previously (31, 32) . PCR products and
plasmid DNA were cut with the appropriate endonucleases and plasmid DNA
was treated with calf intestinal phosphatase (Boehringer Mannheim),
followed by ligation using 1 unit of T4 DNA ligase overnight at 16
°C. Ligation mixtures were transformed into E. coli DH5 competent cells as described by the manufacturer.
15,000-20,000 colonies. A
representative filter is shown here. B, second round screening
of one positive clone (clone 23.2) replated and probed with 
I-126.213. Compare with panel C, second round
screening of a control clone with an irrelevant V region. D, third round screening of clone 23.2. Compare with panel
E, E. coli transformed with pDAB alone.
Library Characterization
Competent Epicurian
coli cells (the Cell Center, University of Pennsylvania)
transformed with the amplified V/pDAB library was
plated on LB/amp plates. Control ligated pDAB vector
produced four to five colonies/plate, while the appropriately ligated
V/pDAB transformants produced 175-465
colonies/plate. Inserts were confirmed by plasmid miniprep
analysis(26, 27) , which revealed appropriately sized
inserts in 75% of colonies. Approximately 2,000 colonies on 10 plates
were screened in this study.Protein Expression
Bacterial clones possessing the
V genes inserted into pDAB were plated onto LB/amp
plates. Control plates contained E. coli transformed with
either pDAB alone, pUC19,or pUC18. Following overnight
growth, replica plating, and additional overnight growth, 0.45-m
nitrocellulose filters were placed on the bacterial plate. Filters were
lifted to other LB/amp plates on which 50 µl of
isopropyl-
-thio-galactopyranoside (IPTG) (25 mg/ml; Stratagene, La
Jolla, CA) had been spread and were then incubated for 4 h at 37
°C. ilters were then exposed to chloroform vapor for 15 min and
incubated overnight (with shaking) in lysis buffer (100 mM Tris-Cl, pH 7.8,150 mM NaCl,5 mM MgCl,1.5% bovine serum albumin (BSA),1 µg/ml
pancreatic DNase I,and 40 µg/ml lysozyme). Filters were then
blocked for 4 h with blocking buffer (5% non-fat dry milk and 0.05% BSA
in phosphate-buffered saline (PBS: 140 mM NaCl, 2.7 mM KCl, 10 mM NaHPO, 1.8 mM KHPO; pH .4)). Following blocking, filters
were screened for specific variable region expression as noted below. alone, or the various V regions ligated into
pDAB. Colonies were grown overnight in LB/Amp, and 500
µl used to seed 5-l cultures grown to 0.6 A
units in Superbroth (Cell Center, University of Pennsylvania),
then induced with 1 mM IPTG for for 4-12 h. The cells
were centrifuged (10,000 revolutions/min for 30 min) and the pellets
dissolved in 2 ml of lysis buffer (10 mM Tris-HCl pH 8.0, 100
mM NaCl, 0.5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and aprotinin diluted 1:100 from a
concentration of 2.1 mg/ml, all from Sigma). These cells were sonicated
for 45 s on ice and clarified by centrifugation (11,000 g for 15 min at 4 °C) and the supernatant (lysate) used
as sources of V fragments.Library Screening
For binding of I-labeled 126.213 or control mAb ID6 (33) blocked
filters were incubated for 2 h at 37 °C with I-labeled mAb (purified by staphylococcal protein A
affinity chromatography), 500,000-1,000,000 cpm/ml, labeled by
the chloramine T method (9) in PBS containing 1% BSA and 0.05%
Tween-20 (PBS-BSA). Filters were washed extensively with PBS-BSA and
autoradiographed (Kodak XRP film) for 2-24 h.DNA Sequencing
Double-stranded DNA sequencing
employed the primers listed in Table 1and followed previously
published protocols (34) . Sequencing proceeded in both the 5`
and 3` orientations to confirm all sequence information.Western Blotting
Bacterial lystaes (prepared as
above) or recombinant human GM-CSF were run on 15% SDS-polyacrylamide
gel in reducing sample buffer (2% SDS, 50 mM Tris-HCl, pH 6.8,
10% glycerol, 0.001% bromphenol blue) with 500 µg of bacterial
protein (as determined by the Bio-Rad protein assay) loaded in each
well. Following electrophoresis, the gel was transferred to Immobilon P
transfer membranes (Millipore) as described(35) , and the blot
blocked with 1% bovine serum albumin, 5% non-fat dry milk, 0.05% Tween
20 in PBS (blocking buffer) for >1 h at 37 °C or overnight at 4
°C. The blots were then incubated with 5 ml of purified mAb 126.213
diluted to 2 µg/ml in blocking buffer for 2 h at 37 °C, and
washed four times in PBS, 0.05% Tween 20 (wash buffer). Polyclonal goat
anti-mouse IgG (Sigma) (radioiodinated as described(9) ) was
then added (1,000 cpm/µl in blocking buffer) and the blot incubated
for 2 h at 37 °C, washed extensively in wash buffer, dried, and
exposed for 12-72 h to Kodak XAR film.Immunoprecipitation
5 µg of purified 126.213
was reacted with protein G beads (Sigma) in Eppendorf tubes and rotated
overnight at 4 °C. The tubes were centrifuged and the liquid
aspirated. The beads were then washed three time with lysis buffer (1%
Triton, 0.05% SDS, 10 mM NaHPO-NaHPO, 150
mM NaCl, 5 mM EDTA, 100 µM Na
VO, and 5 µg/ml aprotinin all from
Sigma) to remove unbound antibody. The beads were resuspended in 100
µl of lysis buffer, and I-GM-CSF was added to the
tubes in the presence or absence of inhibitors (100 µl of total
volume) and rotated at 4 °C for 1 h. The tubes were then
centrifuged, the liquid discarded, and the beads washed three times.
The beads were then resuspended in 2 sample buffer (0.5 M Tris-HCl, pH 6.8, 16% glycerol, 3.2% SDS, 8% 2-mercaptoethanol,
and 0.04% bromphenol blue (all from Sigma) in distilled H0)
and heated at 95 °C for 5 min to dissociate bonds. Samples were
then loaded onto 10% SDS-polyacrylamide gel electrophoresis gels and
analyzed by autoradiography as described(35, 36) .Radioreceptor Binding Assay
This was modified from
previously published protocols(37, 38) . Briefly,
HL-60 cells (from ATCC) were grown in RPMI 1640 with 10% fetal calf
serum and added L-glutamine. 1-2 10
HL-60 cells were washed twice in RPMI 1640 with 1% BSA and 25
mM HEPES, pH 7.4 (binding buffer), centrifuged, and incubated
with inhibitors as noted in figure legends in a 25-µl volume for 1
h at room temperature. 0.5 nM of I-GM-CSF (118
µCi/µg, from DuPont NEN) was then added for 30 min at room
temperature, the cells layered over 500 µl of chilled fetal calf
serum, centrifuged, and the pellets counted.Peptide Synthesis
All peptides were synthesized by
solid-phase methods as described previously (9, 10, 11, 12) by the Wistar
Institute Peptide Synthesis Facility or Macromolecular Resources at
Colorado State University, deprotected, and released from the resin
using anhydrous HF.Enzyme-linked Immunosorbent Assay (ELISA)
ELISA
was performed with polystyrene plates (Dynatech Laboratories Inc.)
coated with peptide by evaporation of peptides (at the concentrations
noted) in distilled water overnight at 37 °C. The wells were washed
with PBS, blocked with 0.05% Tween, 2% BSA in PBS, and washed with PBS.
Primary antibodies were added at varying dilutions for >1 h at 37
°C. After washing, secondary antibody, goat anti-mouse conjugated
to horseradish peroxidase (Sigma) was added per well in 1% BSA in PBS
for 1-2 h at 37 °C. The substrate used for color development
was 3,3`,5,5`-tetramethyl-benzidine dihydrochloride (Sigma). The wells
were decanted, washed extensively, and absorbance of samples was
measured in a plate reader (MR 5000; Dynatech Laboratories) and
expressed as A nm. Specific values were
determined by subtracting the absorbance measured from uncoated wells
from the absorbance to peptide coated wells(39) .
Construction and Screening of Anti-anti-GM-CSFV
Polyclonal neutralizing
antibodies against human GM-CSF raised in BALB/c mice were used to
develop syngeneic anti-anti-GM-CSF with neutralizing activity (Fig. 1). The PCR was used to amplify immunoglobulin VC
Library
genes from these mice. Oligonucleotide primers for amplification of
immunoglobulin genes were chosen based on conserved DNA sequences found
in V variable framework regions and from the constant
region domain. The 5` primer used in these experiments was relatively
specific for the murine VIII family. The 3` primer introduced a
stop codon at position 207, eliminating 8 amino acids including the
carboxyl-terminal cysteine residue, thereby eliminating the tendency
for the produced light chains to dimerize. Spleen cells were isolated,
RNA extracted, and cDNA synthesized. This served as a template for PCR
amplification of the VC regions. Bands of
the expected size (680 base pairs) were observed following agarose
gel electrophoresis. This amplification was specific as control
cellular DNA from human T lymphocyte cell lines did not yield a PCR
product (data not shown).. Following ligation, the reaction products were
transformed into E. coli DH5 cells and plated onto 30
LB/amp plates. This VC library was then
screened after induction with IPTG with radioiodinated neutralizing mAb
126.213, which specifically neutralizes GM-CSF activity. Thirty filters
containing 500-1,000 colonies each were screened in this manner.
A representative filter is shown in Fig. 2A. Based on
the observed binding of I-126.213 to colonies we picked
30 reactive colonies. These were expanded and replated and rescreened
using fresh I-126.213 and a control mAb (ID6) specific
for HIV-1 gp120(33) . Approximately 50% of the filters were
bound by I-126.213 but not by I-ID6
following the second round of screening (see Fig. 2, B and C, for representative filters screened with I-126.213). Most of these were bound by I-126.213 in subsequent rounds of screening (Fig. 2, D and E). Ten colonies which were
consistently bound by I-126.213 but not I-ID6 in subsequent assays were selected for further
characterization.Characterization of V
Western blot analysis was performed of bacterial
lysates prepared from the bacterial colonies consistently bound by
126.213. For Western analysis, lysates were prepared from E. coli transformed with the pDABC Regions plasmid alone, or containing
the specific light chain inserts. The cultures were then either left
uninduced or induced with IPTG, bacterial lysates prepared, separated
electrophoretically, and transferred to Immobilon filters. These were
probed with 126.213 followed by I-goat anti-mouse IgG and
analyzed by autoradiography. A typical Western blot is shown in Fig. 3A. This compares cultures of bacteria both
uninduced and following induction with IPTG. As can be seen, IPTG
induces the appearance of a 21-23 kDa band for the clones
containing specific V regions, while only nonspecific bands
are present in the cultures transformed with pDAB alone.
Notably, this gel was run under non-reducing conditions, suggesting
that the light chains do not dimerize, instead remaining as monomers.
The molecular mass of the band detected is somewhat lower than the 23
kDa predicted for the isolated light chain. This may be due to
inaccuracy of the molecular weight markers used or could reflect
compact folding of the VC fragments.
regions. A, Western blot analysis of rAb fragments. E.
coli transformed with various plasmids were induced or left
uninduced, lysates prepared, and Western blotting performed with
126.213 as the primary antibody as noted under ``Materials and
Methods.'' Lanes were as follows: 1, clone 23.2
uninduced; 2, clone 23.2 induced; 3, clone 5.1
uninduced; 4, clone 5.1 induced; 5, pDAB alone uninduced; 6, pDAB alone induced; 7, 300 ng of GM-CSF (positive control). Molecular weight
markers are indicated. The arrow indicates the band
specifically induced. B, inhibition of immunoprecipitation by
23.2. Immunoprecipitation of I-GM-CSF was performed as
noted under ``Materials and Methods.'' Lysates of E. coli expressing 23.2 or control (irrelevant clone) were prepared,
protein quantified, and 400 µg used to inhibit immunoprecipitation.
Inhibitors were added as follows: 1, pDAB alone
induced; 2, 300 ng of GM-CSF; 3, clone 25.1
uninduced; 4, clone 25.1 induced; 5, clone 23.2
uninduced; 6, clone 23.2 induced; 7, 
C
molecular weight markers.
I-GM-CSF. This assay allowed investigation of the ability
of various rAb VC regions to compete with I-GM-CSF binding to 126.213. Of the 10 rAb
VC regions screened, only one (clone 23.2)
reproducibly inhibited immunoprecipitation by 126.213 (Fig. 3B). Inhibition with the lysates from bacteria
transformed with 23.2 reproducibly inhibited immunoprecipitation on
multiple experiments (Fig. 3B and data not shown).
Inhibition was much greater for IPTG-induced cell lysates compared with
uninduced lysates. Clone 23.2 was selected for further
characterization.Inhibition of GM-CSF Binding to HL-60 Cells by
23.2
GM-CSF specifically binds GM-CSF receptors present on HL60
(human myelomonocytic leukemia) cells, and this binding is inhibited by
126.213(24) . We examined the ability of rAb 23.2 to inhibit
binding of I-GM-CSF to HL60 cells on a standard cellular
binding assay. In this assay, HL-60 cells were preincubated with
lysates from E. coli induced with IPTG following
transformation with the 23.2 plasmid or an irrelevant plasmid (pUC18).
The counts/minute bound are shown versus increasing amounts of
lysate added in Fig. 4. 23.2 transformed bacterial lysates
inhibited binding of I-GM-CSF to HL-60 cells, while
control lysates had no effect. This result indicates that 23.2 competes
with GM-CSF for binding to a site on HL-60 cells and may bind to the
GM-CSF receptor present on these cells.
I-GM-CSF
binding to HL-60 cells by 23.2. The binding assay was performed as
noted under ``Materials and Methods'' using 2
10 HL-60 cells, in the presence or absence of increasing
amounts of 23.2 or control (pUC18) lysates. The counts/min (cpm) bound ± standard error of replicate
determinations for two lysate preparations are
shown.
Sequence of Clone 23.2
As clone 23.2 was
specifically bound by mAb 126.213, and competed with GM-CSF for binding
to 126.213 and to HL-60 cells, the 23.2 insert was sequenced. The
nucleic acid sequence and derived amino acid sequence of 23.2 is shown
in Fig. 5. The 23.2 V region is a member of the
murine V III family as defined by Kabat et
al.(28) , or the V 21 group as defined by Weigert et al. (40), with the J region derived from the J1
family(40) . Data base searching reveals that the 23.2 V/J
amino acid sequence is very similar to the previously described V
21 hybridoma light chains 6684 and 7940 derived from NZB mice (40) , differing by only 6 amino acid substitutions from 6684
and 8 substitutions from 7940.
plasmid. FR, framework; codon numbering (above the
sequence) is according to Kabat et al.(28) with codon
one corresponding to the first amino acid residue of the FR1 region.
Leader peptide sequence is not shown.
Structural Analysis of GM-CSF Mimicry by
23.2
Structural analysis of 23.2 was carried out following
development of molecular models of the V domain. A
molecular model of the light chain was developed by examining sequence
homologies of the 23.2 sequence with sequences of crystallographically
known light chain structures. Identification of crystallographic
templates for the light chain model included examining the length of
the respective CDRs to match those of the template. As many light chain
structures display equivalent CDR II and CDR III lengths, several
alternative models were developed. Model 1 was developed using as a
template the antibody 50.1 (Iggb), an antibody directed against the V3
loop of HIV gp120. This template displayed equivalent CDR lengths with
the 23.2 sequence. The CDRs and framework (FR) regions of the 50.1
template were mutated to those of 23.2 using the program Insight
(Biosym Technologies). The side chain angles of the substituted
residues were set according to angles identified in a data base of side
chains. Each CDR and FR region were changed individually, followed by
1000 cycles of energy minimization to eliminate close contacts between
atoms. As in our prior studies (43, 44, 45) ,
the program Discover (version 2.8, Biosym Technologies) was used for
energy minimization with supplied constant value force field. - and COOH-terminal regions of CDR I were held fixed in
the search procedure. A Cartesian distance matrix was constructed for
combinations of the residues on the NH- and COOH-terminal
regions of the CDR I and compared to a precalculated Cartesian distance
matrix data base of high resolution protein structures(46) .
The 20 best matches were examined using the program Insight II and
appropriate choices were made based upon similarities in chiralities of
side chains at the junctures of the CDR I loop. The choice was spliced
into the template using the program Insight II. Two alternative models
were constructed using this approach. The one involved splicing a loop
identified in the immunoglobulin Fc fragment 1Fc2 (Model 2).
The other involved the heavy chain CDR I of 50.1 (Model 3). It
is well known that light chains can adopt heavy chain conformations in
the absence of heavy chain(47) . The CDR I of the heavy chain
50.1 was spliced into the template. The alternative structures were
mutated to the 23.2 sequence and the structures energy minimized. These
models are presented in Fig. 6.
models were derived as described under
``Results.'' The GM-CSF view is directed at the B and C
helices, while the 23.2 models' view is directed at the CDR I
region. Specific residues implicated in mimicry are indicated. The
models are further discussed in the text.
strand, which are all structurally adjacent. ()Analysis of
the 23.2 models suggests a structural basis for mimicry of this site.
This is shown in Fig. 6. All three models center our attention
on residues Thr
, Glu
, and Lys63 of
GM-CSF. The proximity of Lys
and Glu suggest
a charge-charge interaction. In all three models, these 2 residues are
mimicked by Arg
and Asp
of 23.2. For Models 1
and 3, the mimicry suggests similar orientations of 23.2, while for
Model 2 the structure is rotated 90°. The other GM-CSF
residues mimicked include: Thr mimicked by Ser 
in Models 1 and 3, and by Thr
in Model 2; Lys mimicked by Arg in Models 1 and 3, and Lys
in Model 2; Thr
and Ser
mimicked by
Ser
in Models 1 and 3, and by Ser
and
Tyr
in Model 2; Lys
mimicked by Lys in Models 1 and 3, and by Ser
in Model 2; and
Glu
mimicked by Ser
and Ser in
Model 1, and by Ser
in Model 3. Thus, while sequence
similarity between GM-CSF and 23.2 is quite low, structural similarity
is suggested centered on the B and C helices of GM-CSF and the 23.2 CDR
I.Binding and Bioactivity of CDR Peptides
To further
investigate the basis for mimicry by 23.2, synthetic peptides were
developed based on the 23.2 CDR sequences. These are shown in Table 2. The CDR peptides were used in an ELISA assay to
determine binding by 126.213 (Fig. 7). Binding to the CDR II and
CDR III peptides was not higher than binding to the control peptide
used, although it was higher than the isotype matched control mAb used (Fig. 7). However, the CDR I peptide was bound at higher levels
than the other CDR peptides and the control peptide, and was bound by
126.213 but not the control mAb. Additional studies using a competitive
ELISA indicate that this peptide blocks GM-CSF binding by 126.213 (data
not shown). This suggests that the CDR I region of 23.2 is the major
recognition site for 126.213.
nm binding to the peptides at the
concentration noted minus A nm binding to
BSA-coated control plates. Results are compared for 126.213 versus an isotype matched control mAb (D1.H3) specific for a peptide
derived from the hamster -adrenergic receptor. The mean ±
S.D. of triplicate wells is shown for increasing amounts of purified
126.213 added. A and B, binding to CDR I peptide. C and D, binding to the CDR II peptide. E and F, binding to the CDR III peptide. G and H, binding to the control peptide. The mAbs used were: 126.213
in A, C, E, and G and D1.H3 in B, D, F, and H.
I-GM-CSF and specific binding
determined in the presence of excess unlabeled GM-CSF. A representative
experiment is shown in Fig. 8. Increasing amounts of CDR I
peptide were able to specifically inhibit GM-CSF binding in a
dose-dependent manner, while CDR II and CDR III peptides did not
demonstrate any specific binding inhibition. Thus, the CDR I peptide
antagonizes I-GM-CSF binding to HL-60 cells, suggesting
interaction of this peptide with the GM-CSFR.
I-GM-CSF
binding to HL-60 cells by CDR peptides. The radioreceptor assay was
performed as noted under ``Materials and Methods,'' using
10 HL-60 cells. The cells were preincubated with peptides
at varying dilutions for 60 min at room temperature prior to the
addition of I-GM-CSF. The specific proportion of cpm
bound was determined by subtracting the proportion of cpm bound under
identical conditions in the presence of saturating amounts (50
nM) of unlabeled GM-CSF. The standard deviation of this assay
was 10% on multiple determinations. The percent inhibition of
binding is shown versus increasing amounts of
peptides.
was 1
mg/ml (approximately 400 µM), while for the CDR I peptide,
it was 50 µg/ml (approximately 21 µM). These data
indicate that the CDR I peptide is a specific antagonist of
GM-CSF-dependent cellular proliferation in a micromolar concentration
range.
GM-CSF-Receptor Interactions
GM-CSF activity is
mediated by binding to specific cellular receptors (GM-CSFR) which
belong to a recently described supergene
family(38, 49, 50, 51, 52, 53, 54) .
The high affinity GM-CSFR is comprised of an chain (GM-CSFR)
specific for GM-CSF (38) , and a chain
(
), which can also associate with the IL-3 and IL-5
receptor chains(52) . The GM-CSFR imparts
specificity to the interaction with GM-CSF, and when expressed without
is able to bind GM-CSF, albeit with lower affinity
than the heterodimeric receptor(55) . The high affinity
receptor (GM-CSFR and ) appears to be the signal
transducing unit(56, 57) , with a sequential binding
of GM-CSF to GM-CSFR followed by binding to postulated. The formation of a ternary complex of GM-CSF with
GM-CSFR and implies that more than one site on
GM-CSF is needed for receptor binding and bioactivity./ complex, but not to
the low affinity receptor (GM-CSFR
alone)(55, 58, 59) . This is illustrated most
strikingly by studies using mutants of residue Glu
of
GM-CSF, which inhibit binding of GM-CSF to the low affinity receptor,
but display little activity in inhibiting binding to the high affinity
receptor(58) . Based on these experiments, it has been proposed
that the first helix of GM-CSF is responsible for binding to
(59) . Murine and human GM-CSF display species
specificity and are not cross-reactive. As the substitutions are
scattered throughout the molecule, it was possible to swap regions of
murine and human GM-CSF to locate sites critical for receptor
interaction(37) . These studies indicated a critical role for
amino acids 21-31 and 77-94 in mediating the activity of
human GM-CSF, suggesting that the second site may be involved in
binding to the GM-CSFR. However, other potential GM-CSFR
interaction sites have also been suggested in mutagenesis
studies(60, 61, 62) , mapping of neutralizing
mAbs(24, 63, 64, 65) , and synthetic
peptide studies(48, 63, 66) . Thus, in spite
of considerable study, the GM-CSFR interaction site(s) on GM-CSF
remain incompletely characterized. The major findings were: a peptide derived from the sequence of
the A helix (residues 17-31) and antibodies to this peptide
inhibited GM-CSFdependent cellular proliferation; a peptide comprising
portions of the B and C helices (residues 54-78) was recognized
by two neutralizing monoclonal antibodies (including 126.213) and
exhibited biological antagonist activity. Other peptides were also
bound by 126.213 corresponding to residues 78-99 and 31-54,
but were not specific antagonists of GM-CSF bioactivity. These three
peptides together constitute a ``face'' on GM-CSF centered on
the B and C helices and opposite the A helix. Together with the prior
studies noted above, these studies suggest two binding sites on GM-CSF
important in receptor binding: the A helix which likely interacts with
, and the opposite face centered on the B and C
helices which we propose interacts with the GM-CSFR. The ability
of synthetic peptides corresponding to these epitopes to specifically
inhibit GM-CSF bioactivity strongly supports their role in receptor
interaction.GM-CSF Mimicry by Recombinant Antibody Light
Chain
The studies described here suggest that the 23.2 rAb
fragment mimics a binding site on GM-CSF involved in interaction with
the GM-CSFR. 23.2 was selected to bind to the antigen binding
idiotopes of the neutralizing mAb 126.213, which competes with a
soluble form of the GM-CSFR for binding to GM-CSF. 23.2 displays
several features characteristic of an ``internal image'' of
the antigen including competition with GM-CSF for binding to 126.213
and to HL-60 cells ( Fig. 3and Fig. 4). The sequence
similarity of 23.2 with GM-CSF noted, while weak, is spread out over
residues 53-98, which comprises the B and C helices as well as
the BC interhelical loop, and represent one ``face'' of the
GM-CSF molecule(42) . The weak sequence similarity seen led to
the development of molecular models of 23.2 to investigate a potential
structural basis for the mimicry observed (Fig. 6). This
suggests mimicry of specific residues on the GM-CSF B and C helices by
specific residues in the 23.2 CDR I, CDR II, and FR3 regions. Synthetic
peptides corresponding to the 23.2 CDR regions were developed and
evaluated. This analysis led to the observation that the CDR I peptide
is recognized by 126.213 (Fig. 7) and is a biological and
receptor antagonist of GM-CSF ( Fig. 8and Fig. 9). The
CDR I region of 23.2 contributes most of the residues implicated in the
structural analysis. The activity of the CDR I peptide confirms the
importance of these residues and suggests that this peptide interacts
with the GM-CSFR, functioning as a receptor antagonist. helical in nature. In
spite of this, molecular modeling of this epitope suggests a structural
basis for mimicry as noted above. This indicates that antibody mimicry
of amino acid arrays on helical regions can be understood on a
molecular-structural level. The application of recombinant antibody
technology to development of such mimics should broaden the
applicability of alternative ligand development in the analysis of
active site structures.
)-thio-galactopyranoside; BSA, bovine serum albumin; PBS,
phosphate-buffered saline; ELISA, enzyme-linked immunosorbent assay;
rAb, recombinant antibody.)
We wish to thank L. Marie for her helpful comments and
S. Fish for her most kind assistance.
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  Q. Fang, Y.-Y. Sun, W. Cai, G. R. Dodge, P. A. Lotke, and W. V. Williams Cartilage-reactive T cells in rheumatoid synovium Int. Immunol., May 1, 2000; 12(5): 659 - 669. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Monfardini, M. Ramamoorthy, H. Rosenbaum, Q. Fang, P. A. Godillot, G. Canziani, I. M. Chaiken, and W. V. Williams Construction and Binding Kinetics of a Soluble Granulocyte-Macrophage Colony-stimulating Factor Receptor alpha -Chain-Fc Fusion Protein J. Biol. Chem., March 27, 1998; 273(13): 7657 - 7667. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Panda, G. C. Kundu, B. I. Lee, A. Peri, D. Fohl, I. Chackalaparampil, B. B. Mukherjee, X. D. Li, D. C. Mukherjee, S. Seides, et al. Potential roles of osteopontin and alpha Vbeta 3 integrin in the development of coronary artery restenosis after angioplasty PNAS, August 19, 1997; 94(17): 9308 - 9313. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Monfardini, T. Kieber-Emmons, D. Voet, A. P. Godillot, D. B. Weiner, and W. V. Williams Rational Design of Granulocyte-Macrophage Colony-stimulating Factor Antagonist Peptides J. Biol. Chem., February 9, 1996; 271(6): 2966 - 2971. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASBMB Journals | Molecular and Cellular Proteomics |
| Journal of Lipid Research | ASBMB Today |