Chimeric Rabbit/Human Fab Antibodies Against the Hepatitis B e-antigen and Their Potential

29 Pages·2017·1.38 MB·English

JBC Papers in Press. Published on August 23, 2017 as Manuscript M117.802272
The latest version is at http://www.jbc.org/cgi/doi/10.1074/jbc.M117.802272
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
Chimeric Rabbit/Human Fab Antibodies Against the Hepatitis B e-antigen and Their Potential
Applications in Assay, Characterization, and Therapy

Xiaolei Zhuang‡1, Norman R. Watts‡1, Ira W. Palmer‡, Joshua D. Kaufman‡, Altaira D.
Dearborn‡, Joni L. Trenbeath§, Elif Eren¶, Alasdair C. Steven¶, Christoph Rader|| and Paul
T. Wingfield‡2

From the ‡Protein Expression Laboratory, NIAMS, §Department of Transfusion Medicine,
Warren Grant Magnuson Clinical Center, ¶Laboratory of Structural Biology Research, NIAMS,
National Institutes of Health, Bethesda, Maryland 20892, and the ||Department of Immunology
and Microbiology, The Scripps Research Institute, Jupiter, Florida 33458

Running Title: Chimeric Rabbit/Human Fab Antibodies Against HBeAg

1 These authors contributed equally to this work.

2 To whom correspondence should be addressed: Protein Expression Laboratory, NIAMS,
National Institutes of Health, Bethesda, Maryland 20892. Telephone: [phone]; E-mail:
D
[email protected] ow
n
lo
a
d
e
d
Keywords: antibody engineering, Hepatitis B virus (HBV, Hep B), phage display, single-domain fro
m
antibody (sdAb, nanobody), viral immunology, e-antigen (HBeAg), core antigen (HBcAg), Fab, h
ttp
ELISA, surface plasmon resonance ://w
w
w
.jbc
.o
Abstract standard, rHBeAg, and patient plasma samples. rg
Hepatitis B virus (HBV) infection afflicts We found that the specificity and sensitivity are by/
g
millions worldwide, causing cirrhosis and liver superior to those of existing commercial assays. ue
s
cancer. HBV e-antigen (HBeAg), a clinical To identify potential fine differences between t o
n
marker for disease severity, is a soluble variant rHBeAg and HBeAg, we used these Fabs in A
p
of the viral capsid protein (HBcAg). HBeAg is microscale immunoaffinity chromatography to ril 3
not required for viral replication, but is purify HBeAg from individual patient plasmas. , 2
0
1
implicated in establishing immune tolerance Western blotting and MS results indicated that 9
and chronic infection. The structure of rHBeAg and HBeAg are essentially
recombinant e-antigen (rHBeAg) was recently structurally identical, although HBeAg from
determined, yet, to date the exact nature and different patients exhibits minor C-terminal
quantitation of HBeAg still remain uncertain. heterogeneity. We discuss several potential
Here, to further characterize HBeAg, we used applications for these humanized Fab/scFv.
phage display to produce a panel of chimeric
rabbit/human monoclonal antibody fragments

(both Fab and scFv) against rHBeAg. Several
Hepatitis B Virus (HBV) is a small
of the Fab/scFv, expressed in Escherichia coli,
partially double-stranded DNA pathogen which
had unprecedentedly high binding affinities (K
d poses a health burden on a global scale. Viral
~ 10-12 M) and high specificity. We used these
hepatitis is the seventh leading cause of death
Fab/scFv in the context of an enzyme-linked
in the world and may be increasing (1). HBV is
immunosorbent assay (ELISA) for HBeAg
non-cytopathic but chronic infection, which
quantification, which we compared with
affects over 350 million people, can ultimately
commercially available kits and verified with
lead to liver cirrhosis and hepatocellular
seroconversion panels, the WHO HBeAg
carcinoma (HCC). Liver cirrhosis and HCC do
1
Copyright 2017 by The American Society for Biochemistry and Molecular Biology, Inc.
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
not stem from acute infection but from repeated vitro can cause a conformational switch leading
cycles of hepatocyte destruction and to the assembly of capsid-like structures
regeneration during the immune clearance (11,12). The different redox conditions in the
phase of chronic infection. The first ER (oxidizing) and the cytoplasm (reducing)
recombinant vaccine was the HBV surface could mediate similar structural changes in
antigen (HBsAg) produced in yeast and has vivo.
proven highly effective. However, in many
parts of the world the vaccine is either not The HBeAg is not an essential structural
available or it is too expensive. In areas with a component of the virion nor does it appear to
large population of HBV carriers, more than take part in the viral replication cycle (13). Its
90% of perinatal transmissions result in chronic role appears to be linked to long-term viral
infection (CHB). The factors which appear to persistence in the host, as manifest in chronic
establish CHB are the HBV e-antigen (HBeAg) HBV infection. Unlike the related HBcAg
and subviral particles composed of HBsAg which activates type 1 T helper (Th1) cells
(2,3). The relationship between these antigens leading to immune attack, the HBeAg activates
and the progression of CHB is complex (4). Th2 cells which promote immune tolerance
(14). HBeAg may cross the placenta and
HBV is an enveloped virus with a establish immune tolerance in the developing
D
proteolipid surface glycoprotein (HBsAg) and fetus and thereby suppress innate signaling ow
n
a core composed of the core antigen (HBcAg) pathways (15). The molecular details of these lo
a
d
which forms an icosahedral structure complex events are not fully elucidated and ed
containing the viral genome. The structural better tools are required to assess the specific fro
m
biology of HBV has been recently reviewed (5) roles of HBeAg. Identification and structural h
ttp
and the present study focuses on HBcAg and determinations of the key interactions of ://w
the closely related HBeAg (Figs. 1; S1). Both HBeAg would provide a clearer path to targeted w
w
proteins are derived from the C gene but therapeutics. .jb
c
initiated from two different start codons. The .o
rg
HBcAg is expressed as a 183-residue protein In CHB infection, the long-term b/
y
which polymerizes to form the viral capsid. persistence of HBeAg is associated with the g
u
e
HBeAg is expressed as a preprotein and, development of HCC (16). On the other hand, st o
compared to HBcAg, has a 29-residue signal HBeAg seroconversion (from HBeAg carrier to n A
stheeq ueenndceo pulapssmtreica mr etthicaut lutamrg e(tEs Rth).e Fporolltoewini ntog asunctic-eHsBsfeuAl gt hecraarpryie ro) f iCs HaB pmaatrieknetrs f(o1r6 ,1t7h)e. pril 3, 2
0
processing of the signal sequence by the ER Therefore, the quantitative assay of HBeAg is 1
9
signal peptidase, oxidation and additional of clinical importance. All currently available
Carboxyl-terminal processing occurs in the ER HBeAg assays have two shortcomings: (1) they
lumen (6). The secreted HBeAg retains 10 are non-quantitative in that their readout is a
Amino-terminal residues from the preprotein value relative to a defined but arbitrary standard
but the position of the Carboxyl-terminal rather than a mass quantity; and (2) neither the
processing is unclear, occurring between V149 antigen nor the antibodies are structurally
and R154 (7). While this processing scenario is defined, the latter often also cross-reacting with
true for most of HBeAg, about 15% of the the HBcAg. This situation persists despite the
protein in the endoplasmic reticulum returns to advances made in defining in structural detail
the cytoplasm (8). Cytoplasmic forms of the antigenic determinants of rHBcAg capsids
HBeAg appear to be able to form DNA- and rHBeAg dimers (12,18).
deficient capsids, due to the lack of Carboxyl-
terminal arginine residues, and can be Chimeric rabbit/human Fabs consist of
enveloped and released as decoy particles, rabbit variable domains V and V and human
H K
thereby playing a role in maintaining viral constant domains C 1 and C (19,20). It has
H L
persistence (9,10). It has been shown that been shown that libraries of such chimeric
reduction of the soluble oxidized rHBeAg in hybrid antibodies can be generated from the
2
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
spleen and bone marrow of immunized rabbits terminal His-tag on the heavy chain enables
and subsequently selected by phage display protein purification by immobilized metal-
(20,21). These Fabs can be conveniently affinity chromatography on Ni+2-nitrolo-
expressed in E. coli using an expression triacetic acid agarose. Usually this one step of
cassette with two signal sequences, pelB and purification was adequate but for structural
ompA, which direct secretion of the two chains studies an additional gel filtration step was
into the oxidizing periplasmic space. A included. With Western blot analysis, none of
Carboxyl-terminal His-tag on the V -CH the Fab panel detected rHBeAg following SDS-
H 1
chain allows purification of the soluble protein. PAGE, indicating that they were directed
Such chimeric antibodies often have both high against conformational epitopes.
affinity and specificity and can be fully
humanized, making them both powerful Characterization of Chimeric Fab Antibodies
research tools and of therapeutic potential. Here ELISA Screening. The characterization of
we describe the preparation of a panel of the Fab panel had two initial goals: first, to
chimeric Fabs against the rHBeAg. We have assess affinity against the rHBeAg standard;
studied these antibodies in terms of their and second, to check specificity, primarily
binding affinity and stoichiometry and cross- against the closely related rHBcAg (Fig. 1) but
reactivity with the closely related rHBcAg. also against HBsAg which is often present
D
From these characterizations, we developed a together with HBeAg in clinical samples ow
n
sensitive and quantitative assay for the HBeAg (24,25). For initial screening, we used an lo
a
d
and compared its clinical performance with indirect ELISA microtitre plate assay where ed
commercial tests. The protein constituents of rHBeAg was coated by adsorption onto the fro
m
the assay are of known sequence and structure plate and the chimeric rabbit/human Fab h
ttp
and can be produced indefinitely by binding was detected using HRP-labeled ://w
recombinant expression. These features are human IgG. The strongest responses were from w
w
unique and constitute a powerful incentive for Fabs e1, e8, e13, e21 and e38 (Fig. 2). The Fab .jb
c
additional development and application. For panel was then assayed with plates coated with .o
rg
further characterization of HBeAg, we either rHBcAg (capsids) or HBsAg. As b/
y
immunoaffinity purified the protein from expected, the panel only presented background- g
u
e
single-patient plasma samples for mass level binding to HBsAg. In contrast, many Fabs st o
spectrometry. The results indicate that the displayed above background binding to n A
Cpraervbiooxuyslly-t edremteirnmusin eedx,t etnod ast lbeeaysto nRd1 5V1.1 49, as reH29B, cAe3g8, e(sFpige.c i2al)l.y TFyapbisc ael1ly, ,e 5F,a be8 ,b ien1d3i,n ge2 t1o, pril 3, 2
0
rHBeAg was higher than to rHBcAg except 1
9
Results Fabs e5, e16 and e30, which all gave low
Panel of rHBeAg-specific Fab. The Fab responses to either antigen. From this basic
CDR sequences of a panel of 24 clones are survey, Fabs e13, e21 and e38 gave the highest
shown in Table 1. Included in the Table, for response to rHBeAg and with the lowest cross
comparison, are the murine/human Fab me6 reactivity to rHBcAg. To put the specificity of
derived from the murine Mab e6, which binds these selected Fabs in context, we can compare
rHBeAg with high specificity (22), and with the murine/human chimeric Fab me6,
chimeric rabbit/human Fab Rev, which binds to which does not bind to rHBcAg (capsids) (22).
the HIV-1 Rev protein with pM affinity (23). For example, with Fabs e13 and e38, which are
We selected 17 of the clones from Table 1 for the strongest rHBeAg binders, the ratios of
further analysis based on their sequence binding to rHBeAg versus rHBcAg (capsids)
diversity. The chimeric rabbit/human Fabs were both ~ 2.5, compared to 4.5 obtained with
were expressed in E. coli using the expression the Fab me6. In general, when antigens are
cassette pC3C which comprised: ompA-V -C - adsorbed onto a plate, some protein
k k
pelB-V -C 1-polyHis where the ompA and denaturation can occur which may change the
H H
pelB leader sequences direct secretion of the L presentation of epitopes, potentially decreasing
and H chains into the periplasm. The Carboxyl- or increasing antibody binding.
3
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
analytical ultracentrifugation. The
Binding Kinetics Studied by SPR. The sedimentation equilibrium profile, for example,
binding characteristics of the Fabs which from Fab e13 (Fig. 3) was used to determine a
appeared to be the strongest binders to rHBeAg mass of 127 kDa, which is close to that
in the ELISA assay were next examined in predicted for a complex of 2 Fab and one
more detail using the Biacore system, which rHBeAg dimer (131 kDa). The 2:2 binding
detects binding by surface plasmon resonance stoichiometry was also observed with Fabs e38,
(SPR). Fab was immobilized on the chip e1 and Fab me6. For Fabs e8 and e21, 1:2
(ligand) and titrated with antigen (analyte). The binding was observed indicating that only one
dissociation constants (k ) are given in Table 2. Fab bound to dimeric rHBeAg (Table 2). The
d
Binding affinities to the rHBeAg ranged from structural implications of these binding modes
K ~ 10-7 – 10-12 M and were ranked (high to are discussed later.
d
low affinity): Fab e38 > e13 > e21 > e1 > e8,
which closely matches the ELISA data (Fig. 2). A New and Specific HBeAg Assay
The tightest binders (Fab e13 and Fab e38) Assay Development. One of the aims of this
exhibited very low dissociation rates, which is study was to apply the high-affinity Fabs
a characteristic of high-affinity binding (Fig. ). towards the development of a specific and
However, we note that with such low off-rates, quantitative assay for the HBeAg. We used the
D
and with the technical limitations of the traditional ELISA sandwich format for the ow
n
method, there is some uncertainty about the formulation of this assay. Based on its lo
a
d
eaxcthuiabli tk tdh vea hluigehs bauffti nthiteirees itsy pnioc adlo oufb at nthtiabto tdhieeys smpoencoifcicloitnya lf oarn tthibeo rdHy BMeAabg ,e 6w e( IguGse2da )a (m22u,r2in7e) ed fro
m
selected from rabbit immune repertoires by for antigen capture even though it does not have h
ttp
phage display (26). the highest binding affinity (Table 2). We ://w
found in initial screening trials that microtiter w
w
Binding of Fabs to rHBcAg capsids was plates coated with Mab e6 gave more reliable .jb
c
generally lower, ranging from K ~ 10-5 – 10-8 results than the chimeric Fab me6, suggesting .o
d rg
M and were ranked: Fab e38, e13 > e1, e21 > that more correctly orientated binding sites are b/
y
e8, which again is similar to the ELISA data presented with the larger adsorbed molecule. g
u
e
(Fig. 2). With, for example, Fab e38 and Fab For detection, we screened the highest affinity st o
e13, the off-rates are approximately four orders binders identified in Table 2, namely, Fabs e38, n A
otof mrHagBneiAtugd.e Thihgehreerf ocroem, paaltrheodu tgoh t hbeiirn dbiinngd intgo eh2u1m aann dI gGe1 3fo, r andedt euctsieodn . HORvPe-rc othuep lreHd BaenAtig- pril 3, 2
0
rHBeAg is not absolutely specific it is concentration range 0 – 1 µg/ml, Fab e13 gave 1
9
significantly stronger than to rHBcAg capsid. the strongest response and Fab e21, which
The murine/human chimeric Fab me6 exhibited binds rHBeAg with 1:2 stoichiometry, gave
no binding to rHBcAg capsid (Table 2). This only a weak signal (Fig. 4). Based on these
confirms our previous finding with the murine results, the pairing of Mab e6 and Fab e13 for
Mab e6 (22). To generate chimeric Fab me6 we capture and detection, respectively, of HBeAg
sequenced Mab e6 (12) and used this for gene was further developed.
synthesis and E. coli expression. The SPR data
clearly shows that specificity was retained and Epitope Mapping. In a sandwich ELISA,
this was also true for the scFv e6 (data not the capture and detection antibodies should
shown). ideally have different, non-overlapping
epitopes. Although this requirement is not
Binding Stoichiometry. As the rHBeAg is a essential for antibodies which can bind to both
homodimer it, in principle, has two copies of epitopes of a dimeric protein, we sought to
most epitopes. This was studied by mixing combine specific capture with high affinity
rHBeAg with a two-fold molar excess of Fab detection and, as rationalized above, the
and resolving complexes from excess reagents combination of Mab e6 and Fab e13 was the
by gel filtration and then analyzing by combination of choice. The cross-reactivity of
4
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
these two antibodies was studied by rHBeAg (S/CO) (Fig. S6A) indicated a linear response
neutralization as monitored by SPR. First, up to 1 µg/ml (56 nM, monomer) with a lower
immobilized Fab e13 was titrated with a fixed detection limit of ~ 4 ng/ml (0.2 nM). A
amount of rHBeAg together with increasing calibration curve was also prepared using the
concentrations of Mab e6. There was a positive WHO HBeAg international standard from the
non-competitive response proportional to Mab Paul-Ehrich-Institut which has an activity of
e6 concentration (Fig. 5A). Namely, rHBeAg 100 PE IU/ml. The linear plots for 0 – 1.0 PE
with bound Mab e6 could still bind to the Fab IU/ml and 0 – 10 PE IU/ml are shown (Fig.
e13 ligand. Second, immobilized Mab e6 was S6B, C), respectively. The lower detection
titrated with a fixed amount of rHBeAg level is ~ 0.02 PE IU /ml. From these plots, 1
together with varying concentrations of either PE IU/ml corresponds to ~ 0.2 µg/ml (~ 10 nM)
Fab e13 or Fab me6: non-competitive and of the rHBeAg (Fig. 7).
competitive bindings, respectively, were
observed (Fig. 5B). These results indicate that Assay Clinical Performance. To test the
Mab e6 and Mab e13 bind to rHBeAg, and utility of the assay on patient samples, we first
presumably HBeAg, at non-overlapping checked the matrix effect by titrating rHBeAg
epitopes. to 1 µg/ml in the presence of undiluted and 1:1
diluted non-immune human plasma. No
D
Calibration Curves. A sandwich ELISA differences were observed in the linear plots ow
n
which incorporated Mab e6 for antigen capture, (Fig. S5C). Our next check of the assay was to lo
a
d
Fab e13 for detection and HRP-anti-human IgG detect HBeAg content in patient plasma using ed
for signal generation, was used to titrate a negative-to-positive seroconversion panel. fro
m
rHBeAg over the range of 0 – 100 µg/ml (Fig. Two commercial panels from ZeptoMetrix h
ttp
6). The absorbance response was linear from were used which also included assay data for ://w
0.1 – 1 µg/ml and at > 50 µg/ml the “hook HBsAg and HBeAg determined using the w
w
effect” (28) was observed indicating saturation. Abbott Laboratory assay systems. There was a .jb
c
To simplify the assay, rather than using high correlation between our assay and the .o
rg
secondary antibody detection, HRP was Abbott assay in detecting (or not) patient b/
y
conjugated to Fab e13. Over the linear response plasma HBeAg (Fig. S7). g
u
e
range (0.1 – 1 µg/ml), the conjugate gave only st o
a slightly lower response than the non- The 67 HBV patient plasma samples n A
cfoorn jaullg asutebds eaqnuteibnot dsytu, dthieesr e(fFoirge. Sit3 w). aFso alldoowpitnedg pCrleinviicoauls lCy enatsesra yweder ef oars sHayBeedA agg abiny wthieth NtwIHo pril 3, 2
0
further optimization, including, for example, commercial kits and with the system described 1
9
antibody concentrations, selection of blocking above. A comparison of a subset of the data (25
reagents and incubation times, the ELISA samples) is shown in Table S2. The data shows
protocol was established as described in that the detection of positive samples is similar
Experimental Procedures and as shown in the with all three systems. The Vitros system had
schematic (Fig. S4). the strongest signal as it employs
chemiluminescent detection rather than the
For standardization of the calibration chromogenic detection used in the other two.
curves we determined a cut-off value (CO) for
the assay (see Experimental Procedures) and as The assay data described above, albeit
illustrated in Fig. S5A. We also checked intra- limited in number, indicate that our assay
and inter-plate variation using samples with matches the commercial systems in terms of the
negative, low, medium and high signals (Fig. simple detection of HBeAg in plasma;
S5B). The intra- and inter-assay coefficient of however, our system appears superior in the
variance (CV) values were < 5% and < 7.2%, following ways. First, our system is more
respectively, which indicated high sensitive based on detection limits using both
reproducibility (Fig. S5C). Using rHBeAg, the WHO PE standard and the rHBeAg (Fig.
calibration curves of signal to cut-off ratio S8A). Second, our assay, based on capture with
5
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
Mab e6, is more selective for rHBeAg than monoclonal antibody fragments directed
rHBcAg (capsids), and therefore presumably against the rHBeAg through phage display.
HBeAg and HBcAg. It might be assumed that Rabbit monoclonal antibodies often have both
commercial systems would be highly specific high affinity and specificity and can recognize
for the target antigen but two of the systems that epitopes conserved between human, rat and
we tested showed high cross reactivity with mouse antigens (26,29). In a previous study, we
rHBcAg (capsids) (Fig. S8B). For the testing of produced a chimeric rabbit/human Fab with an
clinical plasma this is not critical as HBcAg exceptionally high affinity for the HIV-1
(i.e. free nucleocapsids) is usually not present protein Rev, which served as a crystallization
(discussed further below) but for basic research chaperone mediating the first structural
purposes this may be of importance. Third, our determination of this key protein (30). Here we
system is quantitative and we have shown have generated a repertoire of chimeric Fab
calibration curves for the rHBeAg and the molecules, consisting of rabbit variable
WHO PE standard (Figs. 7, S6). The domains and human constant domains, against
quantitative aspect of our assay should be the rHBeAg. The 50 kDa chimeric Fab
useful for monitoring the treatment of CHB molecules can be fully humanized (19,31) and
(24) and for basic research, as discussed below. converted to an IgG by fusion with Fc coding
sequences (31). This can be advantageous in
D
Identification of HBeAg in Patient Plasma. applications where greater size and/or ow
n
Another aim of this study was to isolate HBeAg bivalency are necessary. Alternatively, the 50 lo
a
d
from patient plasma to confirm its sequence, kDa Fab can be downsized to a 25 kDa scFv, ed
especially at the Carboxyl-terminal end. The which is sometimes more suitable as a fro
m
antigen was isolated from HBeAg-positive crystallization chaperone as it can form a more h
ttp
plasma samples (S/CO ~ 1000 – 3000; ~ 1 – 3 closely packed crystal (32), as discussed below. ://w
µg/ml HBeAg (Fig. 8)) by affinity w
w
chromatography using Mab e6 immobilized on The specificities of commercially available .jb
c
resin. SDS PAGE of the eluted protein under antibodies and assay systems against HBcAg .o
rg
reducing and non-reducing conditions followed and HBeAg are usually either not clearly b/
y
by Western blot analysis showed a ~ 19 kDa defined or not known. For example, detailed g
u
e
species that migrated slightly faster under the binding kinetics using two well-known st o
non-reducing condition (Fig. 8A), consistent commercial anti-HBeAg antibodies (904 and n A
wdiistuhl fidthee b opnrde,s eans coec cuorfs iann r HiBnterAa-gm boeletwcueleanr 9b0et5w) eiennd ictahtee d rtHhaBt etAhegy adnidd norHt BdicsAcrgi mi(n2a2t)e. pril 3, 2
0
C(-7) and C61 (11,12). The protein bands were Although commercial kits have longstanding 1
9
digested with trypsin and analyzed by mass utility for assaying these antigens in the clinic
spectrometry. The peptides were compared to a they may be unsuited for research purposes due
database of peptides derived from rHBeAg and to such cross-reactivity. In part, the problem
rHBcAg where full coverage had been lies in the close sequence similarity between the
established. Three peptides were identified, two proteins although at a structural level they
corresponding to residues 28-40, 82-98 and are vastly different: a 4 MDa 240-subunit
127-151 (Fig. S9B). The 127-151 peptide capsid (HBcAg) versus a soluble 35 kDa
corresponds to the Carboxyl-terminal sequence dimeric protein (Fig. 1). Also, until recently
of the rHBeAg (147TTVV149) plus an additional (12), the structure of the rHBeAg was unknown
two arginine residues (147TTVVRR151). This and selection of the protein preparation for any
result was obtained with HBV genotypes C immunization protocol and screening may not
(two independent determinations) and F. We have been correct.
failed to detect this peptide in the E genotype.
In characterizing the Fab panel we have
Discussion paid particular attention to the binding
Antibodies and the HBeAg Protein. We stoichiometry. Using sedimentation
have generated chimeric rabbit/human equilibrium ultracentrifugation we determined
6
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
the mass of Fab:rHBeAg complexes. In is formed. The 10-residue propeptide hinders
previous work, we used the murine Fab e6 formation of rHBcAg-like dimers that can
(obtained by digestion of Mab e6) as a assemble to form capsid-like structures.
molecular chaperone to solve the structure of Although the intra-molecular C(-7) – C61
the rHBeAg (12). One Fab binds to the disulfide bond stabilizes the position of the
Carboxyl-terminal helix-loop region of each of propeptide this is not essential as the reduced
the two subunits; 2 Fabs are accommodated protein can maintain its fold and tertiary
without any steric hindrance. Although most of structure, however upon warming, the rHBeAg
the chimeric Fabs also bind with 2:2 dimer can rearrange into the assembly-
stoichiometry (Table 2) they do not necessarily competent rHBcAg dimer-like form (12). Thus,
share the same epitope and Fab e13, for rHBeAg purification requires careful
example, shows no cross reactivity with Fab characterization to ensure that preparations are
me6 (Fig. 5). Two of the Fabs (e8 and e22) not contaminated with structural isomers,
exhibited 1:2 binding; these are of interest different oxidation states, and aggregated
because they mimic the monovalent binding of protein, work that is rarely performed, or at
the traditional anti-HBeAg antibodies least not reported. We have used physically
described in the literature (33). This mode of homogenous, dimeric, fully-oxidized rHBeAg
binding suggested that HBeAg was a for immunization of rabbits used for
D
monomeric protein and this belief persisted construction of the phage libraries, for ow
n
until recently. As the rHBeAg is a dimeric screening of antibody panels, for generating lo
a
d
molecule, the binding of one antibody either calibration curves and for all other studies ed
sterically blocks binding of a second antibody, described. It should be noted that the WHO fro
m
or binding of one antibody is accompanied by HBeAg standard used in clinical assays is not a h
ttp
an allosteric change sensed in the second purified protein but rather patient(s) plasma ://w
binding site thereby reducing its affinity. With with high HBeAg titre. w
w
the rHBcAg capsid, we have previously .jb
c
described examples of these binding An assay for HBeAg. By applying the anti- .o
rg
characteristics (18,34,35). rHBeAg Fab panel we have devised a sensitive b/
y
and specific assay for HBeAg. In principle, g
u
e
The rHBcAg expressed in E. coli appears during both acute and chronic HBV disease, the st o
structurally and antigenically similar to HBcAg clinical assay of HBeAg does not require high n A
aanssda yis k uitsse (d3 6a)s. aO np otshieti voeth estra hnadnadrd, aitnte HmBptcsA tgo sapltehcoiufigchit y pbreecsaeunste tihne cvliorsael lyp raerltaitceldes H B(DcAange, pril 3, 2
0
isolate the HBeAg from HBV-positive plasma particles), is normally not exposed. However, 1
9
(37-39) have not to date yielded protein for disruption of the fragile viral particles during
detailed characterization to guide recombinant sample acquisition and handling could expose
expression. The HBeAg is derived from an viral nucleocapsids, thereby leading to an
alternatively translated C-gene with a 29- artificially elevated measurement. With
residue signal sequence that is subsequently commercial kits, potential non-specific binding
processed at the endoplasmic reticulum to a 10- is unlikely to change the basic assessment of
residue propeptide (6,40). The processing at the whether a plasma sample is HBeAg-positive or
Carboxyl-terminus is not completely negative (see for example, Table S2), whereas
understood (as discussed below) but the protein in research this selectivity may be more
extends to at least residue 149 (37). Hence, we important. The basis of selectivity in our
used the consensus (-10) – 149 sequence (Fig. ELISA is derived from the Mab e6 used for
S1) which also corresponds to the published HBeAg capture. We have previously shown
structure (12). Based on our understanding of that this antibody does not bind to assembled
the rHBeAg structure, we have carefully rHBcAg (22). In that study, we also observed
monitored the oxidation and physical state of that Mab 3120 (obtained from the Institute of
the protein to ensure that it is dimeric and that Immunolgy, Tokyo) has the opposite
the intra-molecular C(-7) – C61 disulfide bond specificity, only binding to assembled rHBcAg
7
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
and not to the dimeric protein. In combination Assay of Anti-HBeAg. The seroconversion
with specific antigen capture we used for of HBeAg status is of clinical importance and
detection the very high affinity chimeric Fab commercial kits are available for detecting this
e13 which does not cross-react with Mab e6 (46). These kits employ a variation of the
(Fig. 5). Although the sensitivity of our ELISA ELISA format: plasma samples are mixed with
assay appears to match, and in some instances (usually) a fixed amount of rHBeAg. Excess or
even exceed, the commercial systems that we non-neutralized HBeAg is then captured with
have compared it with, the assay is by no means solid phase anti-rHBeAg antibody and the
optimized yet. First, the affinity of the capture second conjugated anti-rHBeAg antibody is
Mab e6 (K ~ 10-8 M) is not as high as many of used for detection. This is termed a
d
the other chimeric Fabs, including Fab e13 neutralization anti-e-antigen immunoassay and
(Table 2) but unlike the others it has the highest a positive sample gives no color reaction. A
specificity (as defined above). In this respect, previous study compared commercial systems
the rabbit phage display library is a valuable and discussed their limitations; the major one
resource for further screening for unique being that anti-HBeAg can only be detected in
HBeAg epitopes. As the conversion of murine the absence of excess HBeAg (46).
Mab e6 to chimeric murine/human Fab me6 Circumventing this by increasing rHBeAg used
(and to scFv e6) had no effect on either affinity for neutralization led the authors to conclude
D
or specificity, the phage system can also be that the HBeAg in blood is antigenically ow
n
used for affinity maturation (41-43). Further, as different from the rHBeAg used in the lo
a
d
direct adsorption of Fab me6 to microtitre commercial kits. Although we have not applied ed
plates reduced efficacy compared to the Mab our assay for antibody detection, the well- fro
m
e6, a generic Fc domain, for example, could be defined antibodies and highly characterized h
ttp
fused to the chimeric molecule (44,45) Another rHBeAg that we have described could be ://w
assay limitation, namely, the detection signal, applied to develop an antibody detection w
w
could be enhanced by switching from the system which would give less ambiguous .jb
c
current chromogenic to either a fluorescent or performance than those currently in use. .o
rg
chemiluminescent system. b/
y
Immunoaffinity Purification of HBeAg. g
u
e
Assay Performance. The assay was One of our initial aims was to purify HBeAg st o
validated by assessing the matrix effect, from patient plasma for structural comparison n A
rSe5p)r.o Tdhuec iabsislaityy d, eatnedc taionna llyimticita ol fs e0n.0s2it iPvEit yIU (/Fmigl,. wstuitdhi ethse. IrtH mBaeyA sge eumse sdu hreprreis ainngd tihna atl lt hoius rh eaasr lnieort pril 3, 2
0
which corresponds to ~ 0.2 nM of rHBeAg, is been accomplished before and may indeed have 1
9
lower than that of any of the commercial been but, if so, this achievement is buried deep
systems that we tested. Also, the clinical in the vast HBV literature. Our first task was to
performance, albeit limited in absolute see if we could directly identify HBeAg in
numbers, closely matches that of the Ortho- positive plasma using Western blotting (Fig.
Clinical Vitros Eci immunodiagnostic system. S9A). The best commercial antibody we found
Our assay can be improved; it was not our for this task was a polyclonal rabbit anti-
intention to make a commercial grade system rHBcAg. Parenthetically, this antibody, from
but rather a useful research tool with high Dako, was generated using rHBcAg from E.
sensitivity and specificity. We consider the coli and must detect linear epitopes of rHBeAg
most important attribute of our system to be and rHBcAg exposed during SDS-PAGE. We
that it is well defined and accessible. The used a small-scale immunoaffinity adsorption
expression, purification and structure of the method to purify and concentrate HBeAg from
assay constituents have been detailed for Fab e6 individual plasma samples. Although the
+ rHBeAg (12) and will be made available amounts of protein obtained in this way were
elsewhere for scFv e13 + rHBeAg (Eren et al., limited, we identified the HBeAg according to
manuscript in preparation). the following criteria. (1) The antibody we used
for purification (Mab e6 or Fab me6) is specific
8
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
for rHBeAg (and likely, for HBeAg). Even if uncoating of the capsid (the Carboxyl-terminus
HBcAg were present due to, for example, is sequestered inside the capsid) (48).
disruption of any Dane particles, this protein
would not bind. (2) SDS-PAGE and Western Therapeutic Applications. The therapeutic
blot of reduced versus oxidized protein (Fig. targeting of HBeAg is attractive given its key
8A) showed a shift to higher mobility in the role in regulating immune tolerance,
latter. This indicates the presence of the intra- establishing viral persistence and establishing
molecular (C(-7) – C61) disulfide bond; in CHB (49). Further, given the global magnitude
comparison, oxidized rHBcAg migrates as a of the problem (1) any new approach should be
dimer due to an inter-molecular (C61 – C61) considered, including high-affinity antibody
disulfide bond at the dimer interface (Fig. 8B). fragments or the smaller mimetics,
(3) Mass spectrometric analysis of the SDS- administered especially to HBeAg-positive
PAGE band detected peptides corresponding to women to block perinatal transmission. In cell
the HBeAg (Fig. S9B). Although the sequence culture, delivery of scFv against HBcAg
coverage was incomplete, the Carboxyl- inhibited HBV replication, illustrating the
terminal region was identified as extending to potential of direct antibody treatment (50).
R151. There is a consensus that the HBeAg
Amino-terminal region contains the residual 10 Included in the treatments of CHB which
D
residues from processing of the 29-residue are under development and recently ow
n
signal sequence but there is some debate summarized (51), are drugs that target capsid lo
a
d
regarding the Carboxyl-terminal sequence. The assembly and multi-antigen vaccines (52). ed
earliest determination identified cleavage at Related to this, it is worth noting that Fab me6, fro
m
V149 (37). This site also corresponds to the which does not bind to rHBcAg (capsids) and h
ttp
boundary of the HBcAg assembly domain and was used in the ELISA for specific capture of ://w
structural studies have consequently focused on HBeAg, binds to residues at the dimer-dimer w
w
residues 1 – 149 (5). Subsequent work with cell (assembly) interface of rHBcAg and which are .jb
c
culture models showed that cleavage via furin also present on rHBeAg (see Fig. S4 in (12)). .o
rg
processing occurs at residues 154 and 159, and The Mab e6 was first derived from the b/
y
perhaps 164 and 169 (7,47). The difficulty with immunization of mice with HBcAg treated with g
u
e
these studies is that the sequence SDS (27). Historically, Murray and colleagues st o
(149RRRGRSPRRRTPSPRRRRSQ169) is first demonstrated that HBeAg could be n A
swuisthce psteirbilnee top rvoatreiaasbel-el ipkreo caecstsiivnigty b. yW enitzhy mceelsl dexetpeocstiendg byc rtyrpetaitci ngH BHeBAcAg-gli kwei the pSitDoSpe tsh e(r5e3b)y. pril 3, 2
0
culture expression, even if the initial processing Based on our previous structural studies, it 1
9
is specific, subsequent processing may occur by would be predicted that Fab me6 can block
endogenous protease activity and this is also rHBcAg assembly and this has been verified
true of the plasma-derived protein. Despite experimentally (N. Watts, unpublished data).
differences between these reports, they do We are currently screening members of our
indicate some in vivo heterogeneity in the chimeric Fab panel, especially those binding
HBeAg. The microscale purification from with the same stoichiometry as Fab me6, for
individual plasma samples coupled with mass their inhibitory effects on rHBcAg assembly.
spectrometry, as we have described, indicates These antibodies could be useful as they may
that the Carboxyl-terminus of HBeAg extends target the HBeAg as well as HBcAg assembly.
to at least R151 (in at least two genotypes) and
perhaps even further, as suggested by the cell Apart from the direct targeting of HBeAg
culture studies mentioned above (7,47). It was with antibodies and related mimetics there is of
noted that HBcAg expressed in E. coli can be course classical vaccination. This was in fact
partially processed at R151 by bacterial attempted over 30 years ago by Murray and his
proteases and it was speculated that analogous colleagues who immunized chimpanzees with
activity in vivo could proceed following HBcAg (54) and SDS-treated HBcAg (55).
Despite the limited number of animals tested,
9
Chimeric Rabbit/Human Fab Antibodies Against HBeAg
the positive results supported further reducing SDS-PAGE (Fig. 8B) it was incubated
vaccination trails based upon or containing with 1 µM CuCl for 30 min, 5 mM EDTA
2
HBcAg and its HBeAg-like derivative, and added and re-chromatographed on Superdex
perhaps a general role for internal antigens in 200 to remove any aggregated protein.
generating immunity against viral infection.
This combinational approach is currently being rHBeAg with Carboxyl-terminal Avi-tag. A
evaluated against HBV (52) and we propose rHBeAg construct with a 17-residue peptide
that new vaccine formulations should include biotin ligase substrate domain (Avi-tag:
rHBeAg. GGGLNDIFEAQKIEWHE) appended to the
Carboxyl-terminus was expressed in E. coli.
Conclusions. We have produced a panel of Protein purification was as described above.
chimeric rabbit/human Fabs specific for Biotinylation with biotin ligase (Avidity, LLC)
rHBeAg, some with unprecedentedly high was done according to the manufacturer’s
affinities (K ~ 10-12 M). These antibodies have protocol. Following the reaction, the protein
d
both diagnostic and therapeutic potential. We was gel filtrated on Superdex S200. The protein
have described a quantitative assay for HBeAg was characterized by mass spectrometry to
in which both the epitopes and paratopes are confirm protein labeling.
known, and which is superior to existing
D
commercial assays, both in sensitivity and rHBcAg. The construct Cp1 – ow
n
specificity. On a microscale, we have 149.C48C.C107A (Fig. S1), which corresponds lo
a
d
immunoaffinity purified HBeAg from to the capsid assembly domain, was produced ed
individual patient plasmas and subjected it to as previously described and assembled into fro
m
proteomic analysis. Our results suggest that the capsids (22). For many of the assays we used h
ttp
Carboxyl-terminus is heterogeneous and can capsids assembled from Cp1 – ://w
extend to at least R151, consistent with recent 149.C48C.C107A dimers as surrogate HBcAg. w
w
cell culture studies and contrary to the earlier .jb
c
determination from plasma that the terminus is rHBsAg. The Small form (genotype D) was .o
rg
V149. Future studies will determine whether purified from recombinant yeast by b/
y
the additional residues have any structural and immunoaffinity chromatography (56). g
u
e
functional consequences. We have discussed st o
the untested strategy of targeting HBeAg with Selection of Anti-rHBeAg Chimeric n A
a ntibodies and related mimetics. mRaebthboitd/Hs uumseadn wFearbe bays Pprheavgioe usDlyis pdleasyc. riTbehde pril 3, 2
0
Experimental Procedures (23,57). In brief, rabbits homozygous for 1
9
HBV Patient Plasmas. A total of 67 HBV immunoglobulin allotypes V al and C b9 were
H k
patient plasma samples, 35 HBeAg-positive immunized with purified rHBeAg and spleen
and 32 HBeAg-negative, were obtained from and bone marrow were collected and processed
the National Institutes of Health Clinical for total RNA preparation. RT-PCR
Center, in accordance with institutional policies amplification of rabbit V , C , and V encoding
L k H
for protection of human subjects. All plasma sequences was performed using established
samples had been screened by the Clinical primer combinations and protocols (20). V -
L
Center on the Ortho Clinical Vitros Eci C -V cassette assembly and asymmetric Sfil
k H
immunodiagnostic system, and the HBeAg data ligation into the phage display vector pC3C was
corresponding to each of the plasma samples performed as described previously (20,31). The
were provided. library consisting of transformed rabbit/human
Fab clones was selected by phage display on
rHBeAg. The E. coli expression and rHBeAg which had been selectively
purification were performed as previously biotinylated on the Carboxyl-terminal Avi-tag
described (22). We used the construct Cp(-10) (see above) and immobilized on stepavidin-
– 149.C48C.C107A (Fig. S1). If the protein coated plates. After several rounds of panning,
was not fully oxidized as judged by non- selected clones were screened for rHBeAg
10