82
C1b3f) DNA exposure in full particles.
For full particles the exposure or release of DNA occurs during
infection. To test for changes in the DNA exposure, wild type or mutant capsids would be examined, and
capsids bound to wild type or mutant TfRs, or treated with proteases or with Ca
2+
removed. The 3’ end of the
viral DNA can be exposed without particle disintegration, and we would detect that release by using it as a
template for synthesis with the T7 DNA polymerase. Synthesis initiates on the 3’ DNA hairpin, so that DNA
synthesis is proportional to the amount of the 3’ hairpin exposed (13, 74). The capsids would be incubated
with T7 polymerase and dNTPs, and the amount of product determined by incorporation of
32
P- or Cy5.5-
labeled nucleotides which would indicate the degree of 3’-end exposure.
The 5’-end of the DNA is exposed on the outside of the full capsid, with the NS1 protein attached to that DNA
in newly-made particles. We would detect the NS1-associated with the capsid using the gold labeled Fab of
the CE10 antibody against NS1 protein, and examine in EM. We can also detect the 5’ end of the DNA directly
using an hnRNP A/B protein that recognizes a sequence close to the 5’-end of the viral DNA that is exposed
on the outside of the capsid (68). That protein would be purified and gold labeled, then used to detect the DNA
on the outside of the capsid. The protein binds close to the surface of the capsid, and should be tightly
localized adjacent to the capsid. After initially screening the binding by negative staining, cryoEM analysis
should allow the position of DNA exposure to be seen, most likely at or adjacent to a fivefold axis of the capsid.
C1b4) Dynamic properties of digested or mutant capsids.
For the cleavage site variants and Ca
2+
binding
mutants, protease treatments and probing with a large panel of specific Mab will be used to examine for
changes in the structure. Capsid stability can be measured by resistance to heating, either testing for effects
on infectivity, or by changes in Trp fluorescence using spectrofluorometry, as we have described previously
(45). The level of cleavage would be determined by SDS-PAGE analysis after protease treatments. To
confirm the sites of cleavage seen, we would use micro-N-terminal sequencing or size analysis by mass
spectrometry. We would particularly focus on cleavages generated during limited digestions, which would
show the structures within the capsid that are changing in conformation.
C1b5) Receptor binding properties of the mutant or digested capsids
.
We would determine the binding affinity
of wildtype or variant capsids to the cell receptors (19, 49). Plasmid expression efficiently displays the receptor
on the surface of TfR-negative TRVb cells, and we can also readily purify TfR ectodomains after expression in
insect cells from baculoviruses (49). The purified TfRs bind the capsids in vitro with similar relative affinities to
those seen for the same receptors expressed on cells (49). To measure the affinity of virus binding to the
wildtype TfR or mutants we would also use flow cytometry to measure binding and release of capsids to the
receptors expressed on cells. In an alternative solid-phase binding assay, the His-tagged TfR ectodomain
would be immobilized on lipid bilayers with Ni
2+
-NTA, and fluorescently labeled capsids flowed over the
receptor layer. Labeled viruses would be followed by fluorescent microscopy using TIRF, and the on- and off-
rates determined directly, allowing the affinity of binding to be determined directly (9). Additional studies are
described in Section C2, below.
C1b6) Changes in capsid permeability.
Changes in capsid permeability are associated with several structural
changes in the virus, including cleavage of VP2 to VP3 in full capsids, the release of the VP1 N-termini from
inside the capsid, and the release of the 3’-end of the viral DNA (see above). Some or all of these releases
occur through pores at the 5-fold axes of symmetry. Capsid
permeability can therefore be increased by mutations in the
loops surrounding the pores at the fivefold axes of
symmetry. We already have mutants of residues involved in
the stabilizing the pore structure (“gatekeepers”) including
Asn167Ala, Asp168Ala and Thr170Ala, and those capsids
would be examined for their permeability and the degree of
exposure of the VP2 and VP1 N-termini or DNA under mild
heating or other conditions. The permeability of empty
capsids can be semi-quantitatively monitored using negative
staining EM at neutral pH. Stains with differing properties
include methylamine tungstate (NanoW), ammonium
molybdate, uranyl acetate, NiSO
4
, and sodium tungstate. NanoW does not enter intact empty particles at
neutral pH, but can enter after different treatments, while NiSO
4
readily enters empty particles under most
conditions (45)
(Fig. 3).
We would therefore examine the various mutant capsids under controlled conditions
including heating, protease digestion, and receptor or antibody binding. For empty particles we would
Fig. 3.
Assessing the penetration of capsids
using NanoW or NiSO
4
, distinguishing the
different particle forms; NanoW will penetrate
empty particles after various treatments.
Research Strategy Page 36
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
73
particularly examine the penetration of NanoW, carefully examining the capsids by electron microscope and
the proportion of penetrated particles determined.
C1c)
Outcomes, potential problems and difficulties, and alternative approaches
.
The goals of this first section are the characterization of any important variation, flexibility, and asymmetry in
the parvovirus capsid structures, and effects on capsid permeability and the exposure of internal components
important for infection. Structural variation clearly occurs in the capsid, and many of the variants or mutants
we will be examining are already known to show differences in flexibility and/or protease cleavage
susceptibility. The effects of structural changes on the capsid functions are harder to predict, but functions
known to be altered include the affinity of receptor binding (in particular the host-specific binding to the
glycosylated canine TfR), as well as sialic acid binding, antibody binding, and exposure of various peptides and
of the viral DNA. Receptor binding will also be examined in detail in the experiments in Section C2, and the
relationships between antibody and TfR binding will be examined in Section C3.
There are some potential challenges in these analyses of the capsids. First, we cannot predict the functional
effects of some of the structural changes, and some effects may be relatively subtle. It is clear that the capsids
act as finely regulated machines and perform their functions through controlled transitions, and do not (for
example) simply fall apart, even to release the DNA, but likely expose internal structure or the DNA 3’-end only
under controlled circumstances during infection. Asymmetry can be difficult to detect and quantify, but in this
system we should be able to detect that using a variety of biochemical or other approaches. We know some of
the capsid changes that control protease cleavages and can use that information to predict alternative sites,
and have enough preliminary information to be able to prepare informative mutants for our functional analyses.
The divalent ion binding sites are well defined from the X-ray structures, as are natural mutations are known
that prevent one ion binding and also increase the flexibility of the adjacent loop that controls sialic acid binding
(60). Altering the binding sites of the other two ions is quite straightforward, although we do not know if
capsids with mutations affecting the binding of those ions will be viable. However, we can also control ion
binding by chelation with EDTA or EGTA, or by incubation at low pH, so there are several alternative
approaches available. If mutant viruses are non-viable, we can test many functions by expression of capsids
by plasmid transfection or by expression from baculoviruses (58).
C2)
Aim 2.
To define the structural interactions between various parvovirus capsids and variants of
the transferrin receptor or artificial receptors.
Hypothesis: That specific binding of capsids to the feline or
canine TfRs is required for successful cell infection, and those interactions are controlled by viral structures
varying in structure and flexibility.
C2a) Background on receptor and capsid structures involved in binding.
The FPV and CPV capsids bind the TfR on the surface of their host cells, and it appears that specific TfR-
capsid interactions are required for infection, as other ligands that bind the virus at the cell surface and mediate
uptake do not result in infection (26). The TfR is large butterfly shaped homodimer (11nm span), where each
monomer is comprised of protease-like, apical and helical domains
(Fig. 4)
. While FPV and CPV both bind the
feline TfR, only CPV binds the canine TfR, an interaction that was critical for the host range shift of CPV (53).
We have prepared a low resolution structure of the feline TfR:capsid structure determined by cryoEM which
shows the footprint of
the feline TfR on the
capsids (22), and have
also defined sites on
the receptor that control
interactions with the
virus (19, 49, 50). The
canine TfR has a
distinct interaction with
the CPV capsid
compared to the feline
TfR, as many viral
changes that alter
canine TfR binding do
not change feline
receptor binding (25,
51). Mutational studies
Fig. 4.
.
The TfR structure, based on a model of the human receptor (domains are
colored on one monomer – protease-like = red, apical = green, helical = yellow).
Residues tested for effects on the binding of CPV are colored: no clear effect on
virus binding dark blue, alteration of virus binding red or purple. Known Asn-
linked glycosylation sites labeled in orange. A portion of the stalk domain is
colored in cyan (left), and transferrin binding sites are brown (right). B) The
position of residue Leu 221 in the apical domain.
Research Strategy Page 37
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
66
of the feline and canine TfRs have defined receptor residues that control virus binding (19, 49, 50), and those
show that one site is particularly critical for the interaction. When that site was changed to the other amino
acids, a series of receptors with single changes were obtained that varied in affinity of binding, from no binding,
low binding, to close to wild type (19). Those mutant TfRs are all available for the proposed studies.
These studies will provide fundamental information about the processes of receptor-virus interactions. In the
case of TfR, the rodent TfR-1 is used for cell binding and infection by New World arenaviruses and by mouse
mammary tumor virus, and those viruses also bind to the apical domain of the receptor, in some cases
interacting with the same residues in the structure as CPV (1, 69). While the normal function of the apical
domain is unknown, we have examined TfR sequences from many different large and small cats which likely
have been infected by FPV for very long periods, and have identified a region in the apical domain with
substitutions of residues 378 (Gln-Arg), 379 (Asn-Ile), and 380 (Trp-Ser) within the virus binding region, that is
clearly under selection, perhaps by the virus. Those changes will therefore also be tested in our future studies
to see whether they affect virus binding and infection.
C2b) How does capsid bind the TfR, and is that specific interaction important for infection of cells?
We will further define the TfR-capsid interactions by analysis of the receptor and viral structures using a variety
of approaches to alter the viral and TfR structures involved in the interaction, to explain how the functional
contacts are controlled, and the effects on binding or infection of the structural variants being examined. In
addition to studying specific mutants we will also prepare higher resolution cryoEM structures of various TfR-
capsid complexes. The overall goals are to create a complete dynamic structural model of the interaction that
explains the functions in binding and infection.
C2b1) Defining the functional capsid-TfR interactions.
We have
identified several residues in the receptor and capsid that control binding,
and have determined the 27Å resolution asymmetric structure of the
dimeric feline TfR ectodomain interacting with the CPV capsid (22)
(Fig. 5)
.
Here we will define the interactions between specific residues on the
receptor and capsid in more detail, and will complement the mutational
studies with higher resolution cryoEM structures of the complexes, so as to
achieve a complete understanding of the interactions and their control of
various viral functions.
C2b1a) Viral mutants to examine.
Some cleavage site and flexibility
mutants will be derived from Aim 1 by mutation of sites in the capsid
structure. Other viruses will have altered affinities of TfR binding or TfR
contacts. The CPV-2a natural variant is altered at VP2 residues 87, 101,
300 and 305 and shows reduced affinity for the feline TfR, but retains high
infectivity (49). We have prepared intermediate mutants with single or
double changes in the capsid (e.g. of VP2 residues Ile101Thr and
Ala300Gly) that alter the affinity of receptor binding. VP2 residues 93 and
323 together control binding to the canine TfR but not the feline receptor
(20).
C2b1b) Selection for viruses with altered binding properties by passaging on mutant receptors.
To identify the
capsid residues that interact with specific sites on the TfR, we will isolate mutant viruses by growth in CHO
cells that are expressing TfR variants which show reduced virus binding and infection due to point mutations in
the apical domain. We have prepared cell clones that express the altered affinity variants of the feline TfR
(19). Those will be inoculated with CPV or FPV and then the virus passaged repeatedly to isolate variants that
are better adapted. Once isolated those viruses will be sequenced to identify the candidate contacts between
the receptor and virus that are selected for increased infection.
C2b1c) TfR variants with increased binding prepared by in vitro evolution and selection.
To alter TfR binding
we would use yeast surface display and selection to prepare receptor variants with altered capsid-binding
properties, including those with higher affinities. Clones of the feline TfR ectodomain and of the apical domain
alone have been prepared and those are displayed on the surface of yeast as fusions with the Aga2p mating
agglutinin protein (4, 12). To select for altered binding variants of the protein, TfR sequences around the apical
domain would be subjected to error-prone PCR, and then the mutant products transformed into yeast along
with the expression plasmid to give incorporation by the homologous recombination (4). The mutant TfR yeast
library would be selected by capsid binding, and fluorescence-based sorting used to enrich for yeast
Fig. 5.
The cryoEM structure
of the feline TfR in complex
with the CPV capsid.
Research Strategy Page 38
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
59
expressing receptors binding higher levels of virus compared to an HA epitope included in the expression
vector - yeast showing higher virus binding relative to the levels of protein expression would be candidates for
those with increased affinity. A BD-Biosciences FACS Aria high speed flow cytometer/cell sorter at Cornell is
available to the Parrish laboratory, which can sort up to 50,000 events per second, allowing us to enrich for
even rare mutants with increased binding. Repeated selection and additional rounds of mutagenesis would be
used if necessary to isolate the higher binding versions of the receptor. TfR variants with altered binding would
be re-introduced into the TfR gene in an expression plasmid, transfected into TRVb cells, and then tested for
binding affinity and for the ability to mediate infection. The relative binding and infectivity of each of the three
major strains of virus would be compared to those seen for the wildtype TfR, or to the lower affinity versions of
the TfR already produced in other studies (19, 50).
C2c) Use cryoEM and mutant mapping to obtain better structural models of the CPV-feline TfR and
canine TfR interactions.
To directly visualize the TfR-capsid interaction we would prepare higher resolution cryoEM reconstructions of
TfR:capsid complexes, by including more complexes and by using approaches that allow orientation of the
receptor in the complex. We would examine complexes of the feline and canine TfR ectodomains, as well as a
canine TfR mutant that has a single point mutation that removes a glycosylation site, which binds with higher
affinity and which will likely still show the binding of the canine TfR. Traditional cryoEM approaches can be
difficult due to the dimeric and the aggregation-prone properties of the TfR ectodomain, and we would avoid
these problems using a number of approaches. Data would be collected at Purdue University, where Dr.
Hafenstein worked for several years until 2009, so that she is familiar with the microscopes and systems
required for these studies. A letter of support is included from Dr. Paul Chipman, who manages the EM facility
at Purdue, indicating that Dr. Hafenstein would have access to the appropriate equipment for these studies.
In one approach we would purify the capsid-TfR complex in a Separose CL2B column to isolate monomeric
complexes (22), and after concentrating those would be frozen and examined by cryoEM. An EM defocus
level of <4 microns is necessary for high resolution reconstructions, but at those defocus levels there can be
insufficient contrast to easily determine the position of the TfR bound to the capsid. We would therefore seek
to increase the TfR density by saturating the TfR with iron-loaded transferrin, which would both double the
mass of the TfR and add Fe
3+
ions to the complex, allowing better visualization.
In a second approach we would prepare liposomes containing up to 10% (w/w) DOGS-NTA which would be
nickel loaded, incubated with soluble 6-His TfR ectodomain, and then with purified virus, and frozen for cryoEM
(5, 8). A similar approach would be to use “monolayer purification” (32), where a lipid monolayer containing 2
to 20% Ni-NTA lipids would be used to bind the TfR ectodomain, followed by incubation with capsids. That
monolayer could be picked up on an EM grid and examined by negative staining, or frozen for cryoEM. The
latter method has been used to visualize TfR-transferrin complexes and ribosomes (32), giving sufficient
resolution to define their structures. An additional advantage of these methods is that they would result in the
virus-binding apical domain of the TfR orientated away from the lipid surface, and the density of the TfR:capsid
complex can be adjusted by varying the amount of the Ni-NTA lipid in the membrane. Using proper controls
for non-specific capsid binding, this would allow us to ensure that each particle had one or more receptors
located between the capsid and the membrane layer.
C2c1) CryoEM reconstruction methods, asymmetric structures.
The methods to be used here for cryoEM
reconstructions are now well established, including those that do not rely on the standard icosahedral
averaging methods during reconstruction. To visualize the feline TfR interacting asymmetrically with CPV, we
used a reconstruction procedure that was developed at Purdue (22, 35). The program involved a combination
of the programs SPIDER (17) and a modified version of XMIPP (62), and utilized the particle orientations found
in an icosahedral reconstruction of the CPV-feline TfR complex. A spherical mask is defined that
corresponded to one icosahedral asymmetric unit, then the area of each of the 60 possible projections of the
mask is examined to determine the best correlation between the density of the image and the projected density
of the mask, given the angles that define the virus orientation for each specific difference image. These
orientations are used to compute an asymmetric reconstruction. The procedure is iterated using the resultant
map to re-select, from the 60 orientational possibilities, the preferred angles that define the TfR position
relative to the orientation of the virus. In our previous studies, convergence of the particle orientations was
reached after about six cycles. To obtain three-dimensional reconstructions of individual virus symmetry
features, Briggs et al. describe a method of cryoEM reconstruction allowing for the classification and
reconstruction of individual features of icosahedral virus particle (6). This allows examination of features that
Research Strategy Page 39
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
56
deviate from perfect icosahedral symmetry, such as a unique vertex or incorporation of proteins with less than
full occupancy.
C2d) Can alternative receptor binding ligands in the CPV capsids mediate infection?
An open question is still whether the TfR binding is required for cell infection because of some structural
specificity, or whether other receptors would also work. This would be examined directly by using alternative
receptors to bind the capsid to cells, and examining the effects on cell entry and infection. In studies of other
parvoviruses and AAVs the insertion or replacement of peptides or protein domains into the capsid structures
has allowed the selection of variants with altered receptor binding and tropisms (11, 44). We will examine CPV
variants with a variety of insertions into surface loops that have been shown to allow peptide addition (57). An
initial candidate would be the insertion of the integrin-binding peptide Arg-Gly-Asp into the 4 surface loops,
some of which would be expected to affect the binding to the TfR, while others would be outside the TfR
binding site, so that they may be able to be grown on normal cells, and tested for infectivity on cells expressing
other receptors (such as the αvβ5 integrin). This has been shown to allow some transduction of CPV vectors
in previous studies (43), and also to work for AAV vectors (18). Peptides that bind the human TfR have been
reported (37, 72), and we would insert those sequences into the CPV capsids in the surface loops. Capsids
would be prepared after transfection and tested for their ability to bind the feline, canine or human TfRs when
those were expressed on TRVb cells, and to mediate infection.
We would also use our experience with antibody engineering to prepare bi-specific antibodies that contain the
binding domains of non-neutralizing anti-capsid Mab as well as those of Mab that we have prepared against
the feline TfR (3, 50). Alternatively the anti-TfR scFv itself would be inserted into the capsid structure as an
additional domain, either into one of the surface loops as has been shown to work for various sized peptides,
or added to the C- or N-terminus of VP2 as has been reported for AAV vectors (73).
Each mutant or sequence-inserted capsid would be tested for assembly after transfection, and then for viability
in feline or canine cells. Assembling viruses that do not infect and replicate independently would be prepared
by transfection and the virions purified and concentrated before being used for testing. Binding of the capsids
to cells expressing various receptors (or no receptors as controls) would be tested using our standard methods
of flow cytometry or fluorescence microscopy (19, 24, 26). The TfR binding peptide- or scFv-expressing
capsids, or those tested with the bi-specific antibody linker would be inoculated onto cells expressing mutant
versions of the feline TfR that do not bind capsids [e.g. Ile221Ser (50)]. Infection would be measured by
staining for virus capsids in cells, or the NS1 protein.
C2e) Expected outcomes, possible problems and solutions and alternative approaches.
The studies to be conducted here extend from our previous work which has largely been published, and we
expect that most will be straightforward and lead to clear results. The preparation of mutant and chimeric
capsids containing peptides would be straightforward, as would be the bi-specific antibody linker. The results
are difficult to predict, as it does appear from previous results that the CPV capsid is not as permissive for use
of alternative ligands as is AAV. However, by trying several different approaches we believe that we can
answer the basic question about the structural dependence on TfR binding for infection.
We do not currently have a way to directly connect capsid mutations with specific structures of the TfR, and so
we will select for viral variants using TfRs with known changes in their structure, and also to prepare higher
resolution TfR:capsid complex structures. While it is not certain that the viral and TfR changes selected will
necessarily be in direct contact, that is a reasonable possibility, and the residues identified can be used in
docking and orientation studies of the TfR-capsid complexes. The higher resolution TfR-capsid structures will
involve refinements of methods that we have already developed, as well as preparation of purified complexes,
or of immobilized receptors on lipid layers. These would increase the quality of the results by reducing the
background of the images, and where the TfR was immobilized on a monolayer or liposome, its orientation
relative to the capsid would be constrained to one or a small number of positions in the images collected. We
can also use tomography with a tilting stage to collect more images of each complex, to obtain additional
structural information.
We have several versions of the feline and canine TfRs that bind with lower affinities than the wildtype due to
single changes in their sequence (19). The yeast expression technology has been used for many studies
expressing cell surface glycoproteins [e.g. (28, 33)], and the method is being used in the Parrish laboratory for
antibody engineering (see also Section C3). The TfR is a complex protein structure, and it may not fold
correctly or bind virus efficiently in yeast, and there may be differences in glycosylation. Despite these
difficulties if these studies worked they would give us valuable data, and so we see this as a “high risk but high
Research Strategy Page 40
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
77
payoff” experiment, but not essential for the studies - even if the yeast expression does not result in an
improved receptor, we have many other TfR mutants and capsids with variable properties that will give us most
of the information we are seeking about the roles of the receptor:capsid interactions in virus infection.
C3) Aim 3. Use antibodies to probe the capsid structure, and also to determine how binding to
overlapping sites leads to neutralization in some cases, but not others.
Hypothesis: That antibodies can
be used to detect variant structures in the viral capsid, and that the specific position and orientation of binding
controls the likelihood of competition with the TfR, and neutralization of infection.
C3a) Background.
The properties of viral antigens and the effects of antibody binding vary, with some viruses being efficiently
neutralized, while others show varying degrees of escape from antibody binding or inactivation (23, 54). The
parvovirus capsid is a potent multivalent antigen which rapidly stimulates strong antibody responses, and each
virion can bind up to 60 antibody molecules (21). We have prepared many different antibodies that recognize
the CPV capsid, and have recently mapped the binding footprints of 8 separate antibody Fabs on the virion
surface structure using cryoEM (21). Those antibodies covered a total of ~70% of the capsid surface, and
could be divided into two major clusters of overlapping binding sites (termed “A” and “B”) (21, 63). Several of
the antibody footprints clearly overlapped the binding footprint of the feline TfR
(Fig. 6)
. Those antibodies are
very specific for particular structures in the capsids, and therefore can be used to probe for changes (21, 70).
While 7 of 8 antibodies bound with similar affinities and overlapped the TfR binding site on the capsids, 4
showed little neutralization of the virus when tested as Fabs,
while two showed efficient neutralization (46). Some
neutralizing and non-neutralizing Fabs represent pairs with
essentially the same footprint, indicating that there can be
distinct interactions between TfR and antibody binding that give
different effects on viral infection. One model derived from the
cryoEM structures is that the angle of attachment is important in
neutralization, and that Fabs binding at an oblique angle may
block TfR binding to both the capsid subunit where the Fab
bound, as well as to an adjacent two-fold related asymmetric
unit (21)
(Fig. 7)
. Competition between the Fabs and the feline
TfR was much more efficient for the highly neutralizing Fabs
suggesting that those had a different mechanism of blocking
receptor attachment (46). While other possibilities such as
differences in the affinity of binding of the neutralizing and non-
neutralizing Fabs seem unlikely due to the similar affinities of
most of the antibodies examined, we will also be able to test that
possibility directly in the studies proposed by selecting for
increased affinity variants (see below).
C3b) Antibody and TfR binding – uneven Fab binding and examining for unoccupied sites.
The poor neutralization by several antibody
Fabs indicates that even in the presence of
excess Fab (when all 60 capsid sites appear
occupied by cryoEM), there are still sites on the
capsids available for receptor binding. As
described above, cleaved VP2 surface loops
would likely form such open sites, as they are
unlikely to allow antibody attachment, but those
may still allow (or perhaps even favor) TfR
binding. Such a mechanism would explain why
Fabs that bind at oblique angles could sterically
block TfR access to an adjoining cleaved site on
the capsid which did not directly bind that
antibody. Here we would use the capsids with
either varying numbers of cleavages or different
levels of cleavage susceptibility to examine for
effects on antibody and TfR binding.
Fig. 6.
Binding footprints on one
e
asymmetric unit of the CPV capsid,
showing a B-site antibody (blue), an
A-site antibody (red), and of the feline
TfR (black). Residues 297 and 300
control protease cleavages.
Fig. 7.
The cryoEM structures of the complexes
s
between the Fab of Mab F (neutralizing) and Mab8
(largely non-neutralizing as Fab), showing the different
orientations of those Fabs, which otherwise bind with
similar footprints on the capsid.
Research Strategy Page 41
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
75
C3b1a) TfR binding to Fab-occupied capsids.
To test this model, we would examine for such open sites on the
capsids using viruses with alterations in capsid cleavages (derived in Section C1 studies). Mutant or wildtype
capsids would be digested with proteinases to cleave varying proportions of the surface loop around VP2
residue 300. After purification, those capsids would be incubated with excess Fab, and then with purified TfR
labeled with 5 nm colloidal gold. After washing, the complexes would be examined in the EM either unstained
or with light negative staining, looking for capsid-associated gold particles, compared to control capsids, or
those incubated with labeled TfR in the presence of non-specific Fabs. A prediction of our hypothesis is that
most Fabs would allow some TfR binding, but that binding would be blocked by the highly neutralizing Fabs of
antibodies E and F.
C3b1b) Are there open sites on Fab-saturated capsids that can be detected with antibodies of different
specificities?
Open sites on capsids incubated with excess Fab would also be assayed by leaving the Fab with
the capsids in large excess, and incubating with gold labeled Fabs of antibodies binding to different regions of
the viral asymmetric unit. For example, cleavages around VP2 residue 300 would likely prevent the B-site
antibodies from binding, while most A-site specific Fabs would still bind to those digested capsid subunits. As
some A-site antibodies (e.g. shown in red in
Fig. 6
) clearly overlap the binding footprints of some B-site Fabs
(as shown in blue in
Fig. 6
) (21), they could be used as specific probes for the open sites not occupied by the
B-site Fabs. For these assays controls are critical to distinguish possible artifacts. One would be provided by
the comparison of the antibodies with different effects on receptor binding. The possibility that the second
antibody binds due to the natural exchange due to low affinity of the different antibodies would be examined by
swapping the two antibodies in the study (gold labeling the first antibody and adding excess second antibody),
or by using antibody variants with increased affinities selected as described below.
C3b1c) Detecting low levels of open sites by precipitating capsids with biotinylated antibodies.
An additional
assay for open sites would be to use biotinylated Fabs or TfR, and to incubate those with the Fab:capsid
complexes in the presence of excess unlabeled Fab. We would then precipitate the biotinylated protein in the
complex with streptavidin beads, or would precipitate the TfR with an anti-TfR monoclonal antibody we have
prepared (50). The amount of capsid precipitated would be determined by Western blotting. This would show
the proportion of capsids with sites that were accessible to the probe in question, even if present at low levels
or transiently.
C3c) Is the affinity of the antibodies important for the TfR competition and neutralization?
An alternative mechanism would be that neutralization is related to the specific affinity of the antibody, with
increased affinity allowing a more efficient blocking of viral functions - such as receptor binding or capsid
trafficking and specific structural changes in the endosome. We could test this by preparing antibody domains
with higher and lower affinities and testing their abilities to neutralize the virus as scFv or Fabs. We have
cloned the light and heavy variable chains of several antibodies and expressed those domains in yeast, as
described for TfR (Section C2b1c). We would initially concentrate on neutralizing:non-neutralizing pairs of
antibodies that bind to essentially the same sites, but that differ in neutralization - such as antibodies F and 8
(21). Initial studies show that the initial set of 4 antibodies that we have expressed are produced in good
amounts on the surface of yeast cells, and that they show functional capsid binding when expressed (e.g.
Fig
8
). Those antibody V
H
and V
L
domains would therefore be readily mutated by error prone PCR and then
introduced back into the yeast
to allow recombination,
producing a library that can be
selected for altered (generally
higher) affinity forms (9). We
have already produced such a
library for the clone of antibody
8 in yeast, and started to select
for higher affinity versions.
Antibodies selected would be
expressed as either scFvs, or
linked to the heavy and light
chain constant regions to
produce Fabs (71), and tested
for binding affinity and for
Fig. 8.
The functional antibody scFv expression on the surface of yeast,
t,
where expression was detected with an anti-Fab antibody staining, and
the binding activity detected using Alexa-fluor labeled capsids. Three
examples are shown for of the antibodies to be examined here.
Research Strategy Page 42
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
59
neutralization using standard methods (46). The affinities of binding can be directly measured for the antibody
expressed on the yeast surface using flow cytometry to estimate on- and off-rates (9, 67).
An additional variant on this protein engineering approach would involve determining the virus-specific contacts
and specificity determinants of the antibodies. We have already solved the X-ray crystal structure of the Mab
14, which specifically binds CPV but not FPV (21). Other specific antibodies do not recognize certain natural
or antibody-selected antigenic variants (escape mutants) of the capsid (63). We would therefore use the same
yeast expression library and mutagenesis to select antibody mutants that recognize those variant capsids, and
thereby determine the antibody structural determinants of site-specific recognition. This would allow better
models of the antibody-capsid contacts to be obtained.
C3c1) Functional testing of the mutated antibodies.
To determine the effects of the different changes in the
capsids and antibodies on the process of infection and the relationship to neutralization, we would examine the
effects of the purified antibodies or antibody domains (as scFv or Fab) on the viral functions. We would purify
the wild type and mutant forms of the capsids and antibodies. Antibodies would be tested for their ability to
compete with TfR in binding assays, and for their ability to neutralize standard virus preparations. Wildtype or
mutant capsids would also be used, and treated with proteases to cleave varying proportions of the VP2 in
some studies. Those would then be examined for their ability to bind the TfR in solid phase or on cells, for
uptake into the normal pathways of cell entry, or to be neutralized, using our well established methods (24, 46).
C3d) Expected outcomes, potential problems and their solutions or alternative approaches.
These studies will use existing well-characterized antibodies and engineered variable domains to answer
important questions about the mechanisms of antibody attachment, recognition of specific capsid structures,
and their interactions with receptor binding leading to neutralization. We already have the antibody-capsid
complex structures of a representative set of 8 antibodies (21), and have several of the antibodies expressed
in either bacteria or yeast and have confirmed that those are expressed and bind the viral capsids (e.g. Fig. 8).
We are therefore well positioned to carry out the studies proposed.
While the lack of binding by some of the antibodies to sites with cleaved loops has not been formally shown,
that is highly likely given the properties of the known escape mutations and binding sites. The preparation of
the gold-labeled Fabs should be quite straightforward, while the TfR labeling may be more challenging.
However, the TfR ectodomain dimer, as expressed, displays two 6-His tags on the underside of the receptor,
and that should be readily labeled by Ni
2+
-gold conjugates. The selection for antibody domains with altered
affinities should be straightforward given that we already know that the most interesting antibody pairs are
expressed in a functional form on the surface of yeast, and our close collaborator, Dr. Moonsoo Jin, has used
all of the methods proposed for preparing and screening mutant libraries for ligands with altered affinities (see
letter of collaboration).
C4) Overall Summary and Conclusions.
These studies will integrate our understanding of the structures of viral capsids with a more detailed knowledge
of the binding properties and functions of host cell receptors and antibodies. The results would be correlated
with biochemical and structural analyses of the capsid flexibility, variation, and/or asymmetry. These are
central questions that apply to any non-enveloped animal virus, and have parallels to the structural changes
and interactions seen for many enveloped virus glycoproteins, and so the results will clarify some of the
underlying rules about how viruses interact with their host ligands and infect cells.
The work builds on a solid intellectual and methodological foundation resulting from our previous studies, and
we have most of the materials and background information required. For each of the projects we combine well
established methods with new approaches, and have alternative approaches for each of the experiments
where the technology is novel or untested. Studies already underway would be continued in the first phase of
the funding period, while studies requiring the development of reagents or information from previous studies
will be done later in the project.
TIMELINE
This project would take 5 years to complete. The sequence of studies will initiate in years 1 and 2 with the
preparation of the capsid mutants and their testing, along with development of the new methods for sample
preparation for cryoEM, and collection of the cryoEM data for analysis. Analysis of the role of cleaved and
stabilized capsids would initiate with the currently available mutants, and continue through years 3 and 4.
Mutant forms of the TfR and antibodies would be prepared in the first years, and tested in later years up to
year 5. The preparation of capsids with altered receptor binding sites (peptides or domains), and selected on
mutant receptors, would occur during years 3 to 5.
Research Strategy Page 43
Principal Investigator/Program Director (Last, first, middle): Parrish, Colin, R.
Documents you may be interested
Documents you may be interested