| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
*
Department of Microbiology and Molecular Genetics and the Molecular Biology Institute, University of California, Los Angeles, CA 90095;
Department of Microbiology, Columbia University College of Physicians and Surgeons, New York, NY 10032; and
Department of Genetics, Stanford University School of Medicine, Stanford, CA 94305
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and the variable region of the light chain
on different human 
constant regions (creating inside-out
molecules). Although fully assembled molecules were secreted, this
variable region exchange slowed the kinetics of Ab assembly. Although
the variable region exchange does not lead to a detectable change in
the microenvironment of the combining site, it did alter the kinetic
parameters of binding to immobilized Ag, slowing both the on and off
rates. When effector functions were evaluated, inside-out IgG1 and IgG3
were more effective in complement-mediated cytolysis than their
wild-type counterparts. Variable region domain exchange may be one
approach to obtaining Abs of identical specificity with altered binding
characteristics. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
class consist of two heavy
(H)3 chains and two
light (L) chains. The H and L chains fold into 12,000-Da globular ß
barrel domains with homologous sequences but different functions. There
are four domains for the H chain and two for the L chain. H chain and L
chain are held together both by disulfide bonds and by extremely
strong, noncovalent interactions (1). Ig domains other than
CH2 interact strongly in a lateral fashion to form modules:
VH-VL, CL-CH1, and
CH3-CH3. This lateral pairing buries
hydrophobic residues that would be exposed in isolated domains.
Longitudinal interactions along the H chain or L chain are weak or
nonexistent. Within the IgG1 myeloma protein Kol, there are no
nonbonded longitudinal contacts (2).
Although Ab molecules and their domains have been described as pearls
on a string with weak longitudinal noncovalent interactions between the
domains, but strong lateral interactions (except for CH2),
the questions still remain as to whether longitudinal domain
interaction exerts any influence and whether the context of the
variable regions influences their functional properties. An earlier
study (3) in which variable regions from
anti-(4-hydroxy-3-nitrophenyl)acetyl (anti-NP) Abs were
exchanged between H chains and L chains of the murine or 
1
isotype had concluded that the binding sites were independent from the
constant region context. In the current study we have extended these
studies by characterizing a family of anti-dansyl Abs with the
murine VH on human C and the murine
VL on different human constant regions (creating
inside-out (io) molecules). We found that this variable region exchange
alters the kinetics of assembly of the io protein compared with that of
their isotype-matched wild-type (wt) protein. Although the variable
region exchange does not lead to a detectable change in the
microenvironment of the combining site, we found that it does change
the kinetic parameters of binding to immobilized Ag, slowing both the
on and the off rate. Moreover, io IgG1 and IgG3 are more effective in
complement-mediated cytolysis than their comparable wild types. These
results indicate that for some variable regions, associated functions
can be influenced by their constant region context and suggest that
variable region exchange may be one strategy for changing the Ag
binding characteristics of Abs while maintaining their specificity.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The expressed V and VH genes
from the mouse anti-dansyl (anti-DNS) hybridoma 27–44 had
previously been joined to human C in the
pSV184
Hneo expression vector and to human IgG heavy chain
in the pSV2Hgpt vector, respectively (4). To produce the
vectors encoding the io molecules, the EcoRI in the
intervening sequence immediately 5' of C in
pSV184Hneo was converted to a SalI site using
oligonucleotide linkers (Fig. 1
). As a
result of this manipulation, C and the four human IgG
constant regions existed as SalI-BamHI cloning
cassettes (5). Using these cassettes, the IgG constant regions were
then joined to VL, and C was joined to
VH.
|
P3X63Ag8.653, an Ig-nonproducing mouse myeloma cell line, was transfected with the H and L chain expression vectors (4), transfectants were selected, and surviving clones were screened for Ab production by ELISA using DNS/BSA-coated plates. DNS/BSA was prepared using dansyl chloride and BSA of RIA grade from Sigma Chemical Co. (St. Louis, MO). The amount of bound chimeric Ab was determined using alkaline phosphatase-conjugated polyclonal goat Ab (Sigma Chemical Co.) against human IgG constant regions. Clones producing large quantities of anti-DNS Ab were expanded and maintained in Iscove’s modified Dulbecco’s medium containing 5% calf serum. Chimeric Abs were purified by DNS-coupled affinity chromatography as described previously (6). The concentrations of purified Abs were determined by a bicinchoninic acid-based protein assay (BCA Protein Assay, Pierce, Rockford, IL). Concentration and purity were confirmed by SDS-PAGE analysis or by chromatography on a Superose 12 column (Pharmacia, Piscataway, NJ).
Biosynthetic labeling
Transfectomas were washed twice in methionine-free DMEM (Irvine Scientific (Irvine, CA)) supplemented with nonessential amino acids (Life Technologies, Grand Island, NY) and glutamine (29.2 µg/ml). Cells were labeled in 1 ml of DMEM with 15 µCi of [35S]methionine (Amersham, Arlington Heights, IL). For labeling of secretions, [35S]methionine was added to 15 µCi/ml, and cells were labeled for 3 h at 37°C. For pulse-chase experiments, 107 cells were incubated in 50 µCi/ml [35S]methionine for 5 min, at which time they were diluted to 10 ml in Iscove’s modified Dulbecco’s medium containing a 100-fold excess of unlabeled methionine and 10% horse serum. Cells were harvested onto ice at the indicated times and pelleted by centrifugation. For cytoplasmic Ig, the cell pellet was lysed in 0.5 ml of 1% Nonidet P-40, 0.4% deoxycholate, 66 mM EDTA, and 10 mM Tris, pH 7.4; nuclei were pelleted by centrifugation; and the cytoplasmic lysate was transferred to a fresh tube. To immunoprecipitate the cytoplasmic IgG, a mixture of rabbit anti-human Fab and Fc antisera was added followed by precipitation with Staphylococcus aureus protein A (IgGsorb, The Enzyme Center, Boston, MA) and washing. The precipitates were resuspended in sample buffer (25 mM Tris, pH 6.7; 2% SDS; 10% glycerol; and 0.008% bromophenol blue), the Abs were eluted from the IgGsorb by boiling, and the samples were analyzed by SDS-PAGE and autoradiography.
Fluorescence emission spectrum
To determine the fluorescence emission spectra of
-dansyl-L-lysine bound to the proteins, protein at
a concentration of 54 µg/ml was mixed with hapten at a concentration
sufficient to saturate the binding sites. The excitation wavelength was
340 nm. The emission spectrum was determined using a
spectrofluorometer.
Biosensor analysis of Ag binding
DNS-BSA80 was immobilized to a commercially available IAsys cuvette precoated with a high m.w. carboxymethylated dextran (Fisons Applied Sensor Technology, Cambridge, U.K.), using the conditions recommended by the supplier. The covalently bound DNS-BSA80 increased the sensorgram response by 3500 Arc s. The resonance response increases 163 Arc s/ng/mm2 of bound protein; therefore, the cuvette’s 18-mm2 surface was coated with 386 ng or approximately 2.6 x 10 12 molecules of DNS-BSA80.
For binding studies, the cuvette was washed with PBST (0.02 M NaPO4, 0.85% NaCl, 0.02% Tween, pH 6.8), and a stable baseline was obtained before each addition of ligand. Ligand was added to 200 µl of PBST in volumes between 1 and 5 µl at time zero, and association was observed. The cuvette was then washed with PBST to begin dissociation. For dissociation in the presence of soluble competitor, after the wash with PBST the signal was allowed to stabilize for 30 to 60 s and then 1 µl of a 38 µg/ml solution of DNS-lysine was added. Dissociation was measured for both 23 and 103 nM added Ab. Any remaining bound ligand was removed with Gentle Elution Buffer (Pierce). Association constants were determined using four to six different Ab concentrations in the 12 to 500 nM range. All calculations were performed using the FASTfit software program (Fisons Applied Biosensor Technology, Cambridge, U.K.). Before performing binding experiments, equal amounts of Ab were run on SDS-PAGE to assure that the relative concentrations were accurate and that the Abs were intact. The gel was dried using Gel Drying Film (Promega, Madison, WI) and scanned using Desk Scan II (Hewlett-Packard, Palo Alto, CA). Band intensities were determined using National Institutes of Health IMAGE software (National Institutes of Health, Bethesda, MD), and the results were used to make any necessary corrections in concentrations.
Complement-mediated hemolysis
SRBC (Pocono Rabbit Farm, Canadensis, PA) were coated with DNS-BSA (0.25 mg/ml DNS-BSA and 5% SRBC in 150 mM NaCl and 0.25 mM CrCl3, pH 7.0) for 1 h at 30°C and loaded with [51Cr]sodium chromate (Amersham Corp., Arlington Heights, IL). The free [51Cr]sodium chromate was removed by washing the cells three times in 10 ml of fresh Gel-HBS buffer (0.01 M HEPES, 0.15 M NaCl, 0.5 mM MgCl2, 0.15 mM CaCl2, and 0.1% gelatin, pH 7.4). Chimeric Abs in Gel-HBS at various concentrations were added to round-bottom, 96-well plates (Corning Glass Works, Corning, NY) in a volume of 50 µl. Then, 50 µl of 2% 51Cr-loaded SRBC and 10 CH50 units of guinea pig complement (Colorado Serum Co., Denver, CO), preabsorbed against unlabeled SRBC, in a volume of 25 µl were added to each well sequentially. The plates were incubated at 37°C for 45 min, unlysed SRBC were pelleted by centrifugation of the plate, and 50 µl of supernatant were counted in a gamma counter. Each point was assayed in triplicate, and the percent lysis was calculated.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Expression vectors were constructed with the four IgG constant
regions joined to VLDNS in pSV184Hneo
and the human C joined to VHDNS in
pSV2HgptVHDNS (Fig. 1
). When only
VLC was expressed, it was secreted primarily
as nondisulfide-linked light chain (Fig. 2, lanes 1 and2). When VHC alone was
expressed, it remained essentially intracellular (Fig. 2, lanes
3 and 4). However, when
VHC and VLC were
expressed together, both chains were secreted in approximately
equimolar amounts, but most lacked intermolecular disulfide bonds. Both
cytoplasmic lysates and secretions from the
VHC+VLC
transfectants bound dansyl when assayed by ELISA; cytoplasmic lysates
and secretions from cells expressing only
VHC or VLC did
not bind Ag (data not shown). This suggests that
VHC+VLC
heterodimers assembled functional DNS binding sites, although it is
unclear whether an interchain disulfide bond is required.
|
Clones expressing VH joined to C
along with VL joined to the four human isotypes were
also isolated and characterized. The proteins produced by these
transfectomas were designated io to distinguish them from their wt
isotype-matched controls consisting of classical
VHCH and VLCL heavy and
light chains. To characterize the proteins produced, io and wt cell
lines were grown overnight in medium containing
[35S]methionine, and the secreted Igs were
immunoprecipitated with DNS-BSA-Sepharose and analyzed by SDS-PAGE
(Fig. 3). IgG1, IgG2, and IgG3 io Abs
assemble and secrete H2L2 heterodimers that
bind Ag. Both HL half-molecules and fully assembled
H2L2 that bind Ag are produced by io and wt
IgG4. About 10% of wt IgG4 anti-DNS and 30% of io IgG4
anti-DNS consist of HL molecules (Fig. 3A). The
presence of the HL half-molecule is a consequence of the 4 hinge (7, 8). The nature of the noncovalent interactions between HL
half-molecules remains unclear. Following reduction, the io heavy
chains were similar in apparent m.w. to their wt counterparts (Fig. 3B); as shown in Figure 2, VHC migrated with a larger apparent m.w.
than VLC.
|
A pulse-chase experiment was performed to determine whether
switching the variable regions to different constant regions had any
affect on either the kinetics or the pathway of assembly of Ab. In each
case the io Ab-producing cell lines were compared with an
isotype-matched wt Ab-producing cell line. Two to four independent
clones of each io isotype were examined, and within an isotype similar
assembly patterns were observed. Results from a representative clone of
each isotype are shown in Figure 4.
Several things are noteworthy. The wt and io IgG1 and IgG2 assemble
through an H2 intermediate. For wt IgG1 and IgG2, the
H2 and the H2L intermediates are rapidly
converted to H2L2, with 80% of the Ig fully
assembled by 30 min. In contrast, assembly of io IgG1 and IgG2 into
H2L2 takes longer and <20% of the Ig is
assembled into H2L2 by 30 min (Fig. 4, A–D, and data not shown). The wt and io IgG3 also assemble
using H2 and H2L as intermediates (Fig. 4, E and F, and data not shown); however, the
difference in assembly kinetics between wt and io IgG3 is less marked,
and both have similar amounts of H2L2 by 60
min. The wt and io IgG4 assemble H2 and H2L but
also accumulate HL half-molecules that are found in the secretions
(Figs. 3 and 4, G and H). HL was an even
more prominent assembly intermediate in an io IgG4 clone in which there
was the synthesis of a large excess of L chain (data not shown).
|
All the proteins containing both VHDNS and
VLDNS could be isolated using Ag. Ag binding was
characterized further by measuring the fluorescence emission spectra of
bound DNS and by measuring Ab binding kinetics. Similar fluorescence
emission spectra were obtained for DNS bound by both io and wt proteins
(Fig. 5), indicating that the
microenvironment of the hapten in the combining site was similar in all
the proteins.
|
). The Abs were also directly compared
by adding equal quantities of the anti-DNS Abs (5 µg) to the
DNS-BSA80-coated cuvette (Fig. 6). The wt and io Abs exhibited different
binding kinetics. The wt Abs showed a more rapid ka than the
io Abs. The wt IgG1, IgG2, and IgG3 approached equilibrium by 300
s; the wt IgG4 required longer, reflecting its smaller initial
ka (Table I). In contrast, the io Abs did not reach
equilibrium during the 1000 s of the experiments. However, when
the cuvette was washed with PBST, the io Abs dissociated more slowly
than the wt (Fig. 6), as reflected in a smaller kd (Table I). Under the conditions of the experiment, rebinding occurred, and the
off-rate represented the competition between dissociation and
rebinding. Addition of DNS-lysine as a soluble competitor to prevent
rebinding increased the kd (Table I), but again, the io Abs
showed a smaller kd than their wt counterparts.
|
|
To evaluate the ability of the io Abs to activate the complement
cascade, we performed a direct lysis assay using
DNS80-BSA-coated SRBCs. Neither io nor wt IgG2 and
IgG4 were able to effect lysis (data not shown), and as we had observed
previously, wt IgG3 was more effective than wt IgG1 in causing lysis
(9, 10). In addition, we found that io IgG1 and IgG3 were significantly
more effective than wt IgG1 and IgG3 in activating complement-mediated
cytolysis, with io IgG1 as effective as wt IgG3, and io IgG3 causing
maximum lysis at a concentration less that one-tenth of that required
for wt IgG3 (Fig. 7).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and with VL joined to C regions of different human
isotypes were secreted, with covalent assembly occurring between the H
and L chains (Fig. 3). The L chain heterodimers of
VHC-VLC, although
mostly lacking covalent bonds, also were secreted and bound Ag (Fig. 2). Presumably the
VHC-VLC form
noncovalent heterodimers capable of binding Ag and being secreted.
However, L chains consisting of VHC were
unable to bind Ag or be secreted. The precise structure of the
VHC appears to determine whether it is
secreted from cells. Another VHC with the
anti-NP variable region was secreted in the absence of an
associated L chain (3), while a VHC with a
human variable region was not secreted even with the endogenous
J558L L chains (11). In several cases failure to secrete L chains
can be traced to amino acid sequences present in the variable region
(12, 13, 14). Changes in the kinetics of assembly
The change in domain position results in a slowed kinetics
of covalent intracellular assembly. In contrast with an earlier study
of domain exchanged anti-NP Abs, where L, H, HL,
H2L, and H2L2 were present in
the secretions (3), only fully assembled H2L2
and free L chain are seen in the secretions of io IgG1, io IgG2, and io
IgG3. Like wt IgG4, io IgG4 also secretes HL half-molecules; the
formation of half-molecules has been shown to be a consequence of the
structure of the 4 hinge (7, 8).
During in vivo assembly of the molecule, noncovalent association of the subunits must precede interchain disulfide bond formation, since sulfhydryl groups must be brought close together before oxidation. Covalent assembly of H chains can begin while they are still nascent chains, and disulfide-linked nascent H chains are present in the murine cell line MOPC-21 that assembles its Ig through an H2 intermediate (15). Similarly, H-H disulfide bond formation occurs early in the assembly of the io molecules. Strong noncovalent interactions in the Fc also serve to bring H chains into close proximity, and they appear to be dimers at concentrations as low as 10-10 M, which would facilitate their covalent assembly (16). The L chain is released from the ribosome before the terminal Cys involved in disulfide bond formation has transited the membrane of the endoplasmic reticulum and is available. Therefore, covalent attachment of L chain must take place after its translation has been completed. Excess L chain has been observed to facilitate Ig assembly and secretion (17), and a variant of the murine cell line MPC-11 which synthesizes less L chain has a delay in the secretion of its Ig, with the predominant intermediate in assembly now H2 instead of the normally occurring HL. However, for the io Abs, the L chain appears to be present in excess, so the quantity of L chain present should not delay their assembly. In fact, additional transfectants making a large excess of L chain showed similar H chain assembly kinetics (data not shown).
During in vitro reassembly, H-L interaction is accompanied by
conformational changes in both subunits (18, 19), and previous studies
(20) have suggested that the binding of one -chain domain, either
VL or C to Fd induces a conformational
change in the adjacent domain in Fd, modifying its reactivity toward
the complementary -chain domain. In the io molecules, Fd now
consists of VL and CH1 and the L chain consists
of VH and C. In this altered orientation,
the interaction of CH1 and C and that of
VL and VH may fail to induce the appropriate
conformational changes, thereby slowing the covalent assembly
process.
Earlier pulse-chase analysis of cells from a patient with a 1,
paraprotein had indicated HL half-molecules as an intermediate in the
assembly of human IgG1 (21). In contrast, with all the chimeric
proteins, both wt and io, H2 is the predominant assembly
intermediate; only wt and io IgG4 show significant quantities of HL
molecules, which probably do not represent assembly intermediates but,
instead, a final product. This difference from the previous studies may
reflect the use of a instead of a constant region on the L
chain. Alternatively, there may be allotypic differences between the
1 H chain used in these experiments and the 1 H chain present in
the patient from which the cells were obtained for the earlier
pulse-chase experiments.
Ag reactivity of io anti-DNS molecules
Previous studies using anti-NP Abs in which the variable
regions had been exchanged between the heavy and light chains had shown
no alteration in the functional properties of these Abs and had
suggested autonomy of the Ag binding domains (3). Similarly, the io
anti-DNS Abs retain the local environment of the hapten binding
site as measured by fluorescence enhancement, but, unlike the
anti-NP Abs, they have been shown to exhibit altered binding
kinetics. The wt 27.44 DNS combining site is believed to bind the
dansyl moiety in a lateral position near the rim of the binding pocket
(22). In the io molecules, the combining sites may be positioned
differently, and this difference in position could alter the structure
of the array of Ab molecules on the Ag. This altered positioning could
then affect the on and off rates by influencing the ease both of
association with immobilized Ags and of rebinding. The constant region
of murine Abs has been shown to influence Ag binding (23). At high
epitope density, murine IgG3 was seen to have a significantly greater
apparent functional affinity than murine IgG1, IgG2b, or
F(ab')2; these results are consistent with a model for
cooperative binding by IgG3 that depends on noncovalent interactions
between 3 Fc regions of Ab bound to epitopes in close proximity.
However, in this case IgG3 was found to have advantages in both
association and dissociation rate constants.
Changes in complement activation
Human IgG isotypes differ in their ability to fix complement. IgG1 and IgG3 are most effective; IgG2 is much less effective, functioning under only some assay conditions (6); and IgG4 is unable to bind C1q and activate the complement cascade. The CH2 domain is the binding site for Clq with the exposed ß strand of Glu318, Lys320, and Lys322 important (24). Amino acid changes within CH2 result in the inability of IgG4 to activate complement (9. 10) and contribute to the differential abilities of the other isotypes to activate complement (6, 7, 25, 26, 27). The io IgG1 and io IgG3 molecules were significantly more effective than their wt counterparts in their ability to effect complement-mediated cytolysis. Both io and wt IgG2 and IgG4 remained nonfunctional in this assay. Several aspects of the changed structure in the io Abs could contribute to these differences. The geometric presentation of hapten or antigenic epitopes recognized by Abs can determine their effectiveness in activating complement (28), probably due to the nature of the lattice formed by the Ag-antibody complexes. Switching the domains between the H and L chains in the anti-DNS Abs could alter the position of the binding site relative to the remainder of the molecule. This is possible given the reported position of the combining site at the edge of the variable region (22) (discussed above). A different array of io IgG1 and IgG3 more effective in complement activation could form on the surface of Ag coated-SRBC. It is also possible that the slower off-rate of the io molecules leads to a more stable Ab-C1 complex.
These studies demonstrate that it is possible to alter some of the functional properties of Ab molecules by changing the disposition of the different domains. While maintaining identical amino sequences in the variable region, it was possible to change the kinetics of Ag binding by exchanging the variable regions of the H and L chains. The resulting Abs had altered functional properties, including increased efficacy in complement mediated cytolysis. Domain exchange may provide an additional approach to producing Abs with the appropriate constellation of functional properties.
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Sherie L. Morrison, Department of Microbiology and Molecular Genetics, University of California, 405 Hilgard Ave., Los Angeles, CA 90095-1489.
3 Abbreviations used in this paper: H, Ig heavy chain; L, Ig light chain; NP, (4-hydroxy-3-nitrophenyl)acetyl; io, inside-out; wt, wild-type; DNS, 1-dimethylamino- naphthalene-5-sulfonyl.
Received for publication November 19, 1996. Accepted for publication November 19, 1997.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2 immunoglobulin chain is prevented by a single amino acid substitution in its variable region. Cell 33:77.[Medline]
-chains and light chains. J. Exp. Med. 132:1482.
This article has been cited by other articles:
|
|
 |
|
  M. Torres, R. May, M. D. Scharff, and A. Casadevall Variable-Region-Identical Antibodies Differing in Isotype Demonstrate Differences in Fine Specificity and Idiotype J. Immunol., February 15, 2005; 174(4): 2132 - 2142. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. F. Montano and S. L. Morrison Influence of the Isotype of the Light Chain on the Properties of IgG J. Immunol., January 1, 2002; 168(1): 224 - 231. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Vidarte, C. Pastor, S. Mas, A. B. Blazquez, V. de los Rios, R. Guerrero, and F. Vivanco Serine 132 Is the C3 Covalent Attachment Point on the CH1 Domain of Human IgG1 J. Biol. Chem., October 5, 2001; 276(41): 38217 - 38223. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
