Protein Engineering vol.13 no.8 pp.565–574, 2000
`
`ScFv multimers of the anti-neuraminidase antibody NC10:
`shortening of the linker in single-chain Fv fragment assembled in
`VL to VH orientation drives the formation of dimers, trimers,
`tetramers and higher molecular mass multimers
`
`Olan Dolezal1, Lesley A.Pearce, Lynne J.Lawrence2,
`Airlie J.McCoy3, Peter J.Hudson and Alexander A.Kortt
`
`CSIRO Health Sciences and Nutrition and CRC for Diagnostic
`Technologies, 343 Royal Parade, Parkville, Victoria, 2Biomolecular
`Research Institute, 343 Royal Parade, Parkville, Victoria, Australia 3052 and
`3Department of Haematology, Cambridge University, Hills Road, Cambridge
`CB2 2XY, UK
`1To whom correspondence should be addressed.
`E-mail: olan.dolezal@hsn.csiro.au
`Synthetic genes encoding single-chain variable fragments
`(scFvs) of NC10 anti-neuraminidase antibody were con-
`structed by joining the VL and VH domains with linkers of
`fifteen, five, four, three, two, one and zero residues. These
`VL–VH constructs were expressed in Escherichia coli and
`the resulting proteins were characterized and compared
`with the previously characterized NC10 scFv proteins
`assembled in VH–VL orientation. Size-exclusion chromato-
`graphy and electron microscope images of complexes
`formed between various NC10 scFvs and anti-idiotype Fab⬘
`were used to analyse the oligomeric status of these scFvs.
`The result showed that as the linker length between VL
`and VH was reduced, different patterns of oligomerization
`were observed compared with those with VH–VL isomers.
`As was the case for VH–VL orientation, the scFv-15 VL–
`VH protein existed mainly as a monomer whereas dimer
`(diabody) was a predominant conformation for the scFv-5,
`scFv-4 and scFv-3 VL–VH proteins. In contrast to the VH–
`VL isomer, direct ligation of VL to VH led to the formation
`of predominantly a tetramer (tetrabody) rather than to an
`expected trimer (triabody). Furthermore, the transition
`between dimers and higher order oligomers was not as
`distinct as for VH–VL. Thus reducing the linker length in
`VL–VH from three to two residues did not precisely dictate
`a transition between dimers and tetramers. Instead, two-
`residue as well as one-residue linked scFvs formed a
`mixture of dimers, trimers and tetramers.
`Keywords: antibody/diabody/single-chain Fvs/tetrabody/tria-
`body
`
`Introduction
`The antigen-binding portion of an antibody molecule (Fv
`fragment) is formed by the association of the heavy chain
`variable region (VH) and the light chain variable region (VL).
`Fv fragments are the smallest entities that consistently maintain
`the binding specificity of the whole antibody. Although recom-
`binant DNA techniques have facilitated individual Fv domain
`production, the non-covalently associated VL and VH domains
`in an individual Fv fragment tend to dissociate from one
`another (Glockshuber et al., 1990). To improve stability,
`recombinant single chain Fv fragments (scFvs) have been
`engineered with two variable domains covalently joined via a
`flexible peptide linker (Bird et al., 1988; Huston et al., 1988).
`In general, scFvs with linkers ⬎12 amino acid residues in
`
`© Oxford University Press
`
`length form stable monomers, which usually exhibit similar
`antigen binding affinity to the parent antibody (Skerra et al.,
`1991; Kortt et al., 1994). However, scFv monomers have also
`been observed to form active dimers and higher molecular
`mass multimers upon freezing at high protein concentrations
`(Kortt et al., 1994).
`To provide increased avidity to target antigens, recent
`attention has focused on the design of linkers which generate
`multivalent scFv molecules (reviewed by Hudson, 1999).
`Multivalent scFv molecules are sufficiently large to avoid the
`fast clearance through the kidneys observed for scFv monomers
`and thereby have potential application for tumour imaging and
`radiotherapy (Adams et al., 1998; Colcher et al., 1998; Wu
`et al., 1999). Construction of stable multimeric scFv molecules
`can be achieved by the shortening of the linker length to ⬍12
`residues such that the VH domain is unable to associate with
`its attached VL domain and thus generate a monomeric Fv
`fragment. Instead, VH and VL domains from one scFv molecule
`associate with VH and VL domains from a second scFv
`molecule to form a bivalent dimer, termed a diabody (Hollinger
`et al., 1993). When the linker is shortened to ⬍3 residues or
`when VH and VL domains are directly ligated to each other,
`scFv molecules associate to form a trimer, termed a triabody
`(Iliades et al., 1997; Kortt et al., 1997).
`ScFvs of both VH–linker–VL and VL–linker–VH orientation
`have been produced by various research groups (Huston et al.,
`1991). Our laboratory has to date almost exclusively produced
`scFvs with the VH domain at the amino terminus (Malby et al.,
`1993, Lilley et al., 1994; Coia et al., 1997; Iliades et al.,
`1997). In particular, NC10 anti-neuraminidase scFv antibody,
`in a VH–linker–VL orientation, has been used extensively to
`elucidate the oligomeric nature of scFvs with reduced linker
`lengths (Kortt et al., 1997; Atwell et al., 1999). To investigate
`the oligomerization phenomenon further, the NC10 scFv frag-
`ment was assembled in a reverse, VL–linker–VH, orientation.
`The observed result showed that as the linker length between
`VL and VH was reduced, different patterns of oligomerization
`were observed compared with the original VH–linker–VL
`orientation. Furthermore, the direct ligation of VL to VH led
`to the formation of a tetramer, rather than to a trimer as
`observed for direct ligation of VH to VL.
`
`Materials and methods
`Sequence numbering
`Antibody residues were numbered according to Kabat et al.
`(1991) and for NC10 correspond exactly to Malby et al. (1993).
`Residues in the VL domain of the scFv were superscripted with
`L and the residue number; for example, ArgL107 signifies
`arginine in position 107 of the VL domain. Similarly residues
`in the VH domain of the scFv were superscripted H and the
`residue number.
`General cloning procedures
`Unless stated otherwise, all DNA manipulations were carried
`out according to standard protocols (Sambrook et al., 1989)
`
`565
`
`UPenn Ex. 2062
`Miltenyi v. UPenn
`IPR2022-00855
`
`

`

`O.Dolezal et al.
`
`with reagents purchased from New England Biolabs. All
`polymerase chain reactions (PCRs) were performed with Pfu
`DNA polymerase (Stratagene). The PCR-amplified DNA frag-
`ments were digested with appropriate restriction enzymes, run
`on 1% (w/v) agarose gel and purified from the gel using
`BRESAclean purification kit (Bresatech). The purified DNA
`fragments were ligated into similarly prepared expression
`vectors using reagents and protocols supplied by Gibco-BRL.
`The ligation mixtures were transformed by the electroporation
`method (Dower et al., 1988) into Escherichia coli XL 1-Blue
`MRF⬘ cells (Stratagene) and recombinant clones were identified
`by colony PCR using primers complementary to 5⬘ and 3⬘
`ends of recombinant gene inserts. All DNA sequences of
`various scFv constructs were verified using Dye Terminator
`Cycle Sequencing kits with AmpliTaq (PE Applied Bio-
`systems).
`Construction of NC10 scFv gene fragments with 0 and 15
`residue linkers
`The previously described pPOW–NC10 scFv-15 (VH–VL) gene
`construct (Malby et al., 1993) was used as a source of VL and
`VH gene fragments. To generate the NC10 scFv-0 (VL–VH)
`synthetic gene, the VL and VH gene fragments were PCR-
`amplified using primers N4311 and N4341 for VL, and N4342
`and N4293 for VH (Table I). The resulting VL and VH PCR
`products were gel-purified and joined into an scFv-0 using
`PCR overlap extension (Horton et al., 1989). To create the
`NC10 scFv-15 (VL–VH) synthetic gene, VL and VH gene
`fragments were PCR-amplified using primers N4311 and
`N4535 for VL, and N4534 and N4293 for VH (Table I). The
`resulting VL and VH gene fragments each contained part of
`the linker sequence at the 5⬘ and 3⬘ ends, respectively, as well
`as the BamHI restriction site which allowed for correct in
`frame ligation of these V fragments via a linker sequence
`encoding 15 amino acid (GGGGS)3. The NC10 scFv-0 and
`scFv-15 (VL–VH) synthetic genes were then digested with
`NcoI and NotI restriction enzymes and cloned separately into
`a likewise digested pGC-4C2 E.coli vector (Coia et al., 1997)
`to create pGC-NC10 scFv-0 and scFv-15 plasmids (Figure 1a-
`i). This pGC-4C2 vector backbone incorporates two, rather
`than one, C-terminal FLAG peptide epitopes (Hopp et al.,
`1988) for improved purification efficacy of FLAG-fusion
`proteins. The resulting NC10 scFv-0 and scFv-15 gene con-
`structs encoded N-terminal cleavable pelB periplasmic target-
`ing signal followed by NC10 scFv and two C-terminal FLAG
`peptide epitope tags which were linked to VH domain at the
`C-terminus by three alanine residues and separated from one
`another by two alanine residues (Figure 1a-iii). The NC10
`scFv-15 and scFv-0 (VL–VH) gene fragments were also inserted
`into a heat-inducible pPOW vector (Power et al., 1992). This
`was done by digesting the above described pGC-NC10 scFv-
`15 and scFv-0 (VL–VH) plasmids with NcoI and EcoRI
`restriction enzymes and ligating the scFv synthetic genes into
`a likewise digested pPOW vector. Similarly as for pGC
`constructs,
`the resulting gene fragment (Figure 1a-ii) was
`designed to express pelB signal sequence followed by NC10
`scFv and two C-terminal FLAG tags.
`Construction and cloning of NC10 scFv gene fragments with
`shorter linkers
`The pGC-NC10 scFv-0 (VL–VH) gene construct (Figure 1a)
`was used for insertion of linkers of increasing length. This
`vector was digested simultaneously with XhoI and PstI restric-
`tion enzymes and the remaining plasmid DNA was purified
`
`566
`
`Fig. 1. (a) Schematic diagram of the NC10 scFv (VL–VH) expression unit
`in pGC (i) and pPOW (ii) vectors. (iii) Important parts of nucleotide and
`amino acid sequences of the NC10 scFv (VL–VH) unit in pGC and pPOW,
`outlining restriction sites for cloning (underlined), C-terminal region of pelB
`sequence, N-terminal sequence of VL gene and C-terminal sequence of VH
`gene terminal (in bold), and two C-terminal octapeptide FLAG epitopes (in
`bold). (b) Amino acid sequences of the C-terminus of the VL domain,
`N-terminus of the VH domain and of the linker peptide (in bold) used in
`each of the NC10 scFv constructs.
`
`using agarose gel. Five sets of synthetic oligonucleotides
`(Table I) were phosphorylated at the 5⬘ termini by incubation
`at 37°C for 30 min with 0.5 units of T4 polynucleotide kinase
`and 1 mM rATP (Biotech International) in 1⫻ PNK buffer.
`Pairs of complementary phosphorylated oligonucleotides were
`pre-mixed in equimolar ratios to form DNA duplexes that
`encoded linkers of increased length. The one-residue linker
`construct was designed to encode a Ser residue in between
`codons for C-terminal VL ArgL107 and N-terminal VH GlnH1,
`whereas the two-, three-, four- and five-residue linker constructs
`were designed to encode additional glycine residues immedi-
`ately preceding the Ser linker residue (Figure 1b). All five of
`these duplexes were designed to contain a ‘sticky end’ overlap
`compatible with XhoI and PstI restriction enzyme sites at the
`5⬘ and 3⬘ ends, respectively. This allowed for direct cloning
`of these duplexes into XhoI/PstI restricted pGC-NC10 scFv-0
`(VL/VH) plasmid.
`
`UPenn Ex. 2062
`Miltenyi v. UPenn
`IPR2022-00855
`
`

`

`Table I. Synthetic oligonucleotides used for construction of NC10 scFv VL–VH gene fragments
`
`scFv multimers NC10scFVmultimers
`
`TCG AGA TAA GAG GTG GAG GTG GAT CCC AGG TGC AGC TGC A
`GCT GCA CCT GGG ATC CAC CTC CAC CTC TTA TC
`
`TCG AGA TAA GAG GAG GTG GAT CCC AGG TGC AGC TGC A
`GCT GCA CCT GGG ATC CAC CTC CTC TTA TC
`
`TCG AGA TAA GAG GTG GAT CCC AGG TGC AGC TGC A
`GCT GCA CCT GGG ATC CAC CTC TTA TC
`
`scFv–15 and scFv–0
`N4311:
`AAA ACC ATG GCC GAT ATC GAG CTC ACA CAG
`N4341:
`AGA CTG CTG CAG CTG CAC CTG TCT TAT CTC GAG CTT GGT
`N4342:
`GGG ACC AAG CTC GAG ATA AGA CAG GTG CAG CTG CAG CAG TCT
`N4293:
`TTT TGC GGC CGC GGA GAC GGT GAC CGT GGT CCC
`N4535:
`TTT GGA TCC ACC TCC ACC GCA ACC GCC TCC ACC TCT TAT CTC GAG CTT GGT CCC
`N4534:
`AAA GGA TCC GGT GGA GGC GGT TGC CAG GTG CAG CTG CAG CAG
`scFv–5
`N5006:
`N5009:
`scFv–4
`N5007:
`N5010:
`scFv–3
`N5008:
`N5004:
`scFv–2
`N5011:
`N5003
`scFv–1
`N5005
`N5002
`
`TCG AGA TAA GAG GAT CCC AGG TGC AGC TGC A
`GCT GCA CCT GGG ATC CTC TTA TC
`
`TCG AGA TAA GAT CCC AGG TGC AGC TGC A
`GCT GCA CCT GGG ATC TTA TC
`
`Oligonucleotide sequences are listed 5⬘ to 3⬘. Linker- and vector-related gene sequences are in italics.
`Important restriction sites are underlined.
`
`Expression and purification of the scFvs
`NC10 scFv-15 and scFv-0 gene constructs (VL–VH) were
`initially expressed in a heat-inducible pPOW vector (Power
`et al., 1992). Each pPOW-NC10 scFv construct was expressed
`in 1 l of 2YT/amp100 as described by Malby et al. (1993),
`using E.coli strain TOP 10F⬘ (Invitrogen) as host. The scFv-15
`and scFv-0 proteins (VL–VH) were located in the periplasm as
`insoluble aggregates associated with the membrane fraction,
`as found previously for NC10 scFv VH–VL proteins (Malby
`et al., 1993; Kortt et al., 1997). Recombinant scFv proteins
`were purified using a modified protocol of Kortt et al. (1997).
`Cells were disrupted by sonication in 100 ml of phosphate-
`buffered saline, pH 7.4 (PBS), centrifuged and the insoluble
`aggregates of NC10 scFvs were solubilized in 100 ml of 6 M
`guanidine hydrochloride, 0.1 M Tris–HCl, pH 8.0. The resulting
`mixture was then dialysed against PBS and insoluble material
`removed by centrifugation. Recombinant scFvs were purified
`from soluble fractions (supernatant) using an M2 anti-FLAG
`IgG-Sepharose column (5⫻1 cm; Brizzard et al., 1994). The
`affinity column was equilibrated in PBS and bound proteins
`were eluted with 100 mM glycine, pH 3.0. The eluted proteins
`were neutralized with 1/10 volume of 1 M Tris–HCl (pH 8.0)
`and dialysed extensively against PBS–0.02% (w/v) sodium
`azide. Proteins samples were concentrated by ultrafiltration
`(Amicon) over a 10 kDa cut-off membrane (YM10, Diaflo) to
`~1 mg/ml and stored at 4°C.
`Expression and purification of NC10 scFvs (VL–VH) with
`shorter linkers
`Each pGC-NC10 scFv (VL–VH) construct was expressed in
`500 ml of 2YT/amp100 ⫹ 0.1% (w/v) D-glucose as described
`in Dolezal et al. (1995), using E.coli strain TOP 10F⬘ cells as
`host. Expression experiments were terminated 3 h post-induc-
`tion and proteins were isolated from the periplasmic space
`using a modified method of Minsky et al. (1986). Briefly, cells
`were centrifuged at 5000 g for 10 min and supernatant was
`discarded. After washing the cell pellet in ice-cold PBS, cells
`
`were resuspended in 20 ml of ice-cold spheroplast buffer (100
`mM Tris–HCl, pH 8.0, 0.5 M sucrose, 2 mM EDTA, 100 µg/
`ml lysozyme). After incubating the cell mixture on ice for 20
`min, ice-cold half-strength spheroplast buffer (50 mM Tris–
`HCl, pH 8.0, 1 mM EDTA, 0.25 M sucrose) was added to a
`final volume of 100 ml. Cells were mixed gently and left on
`ice for a further 30 min. Periplasmic proteins were harvested
`by centrifugation at 10 000 g. The supernatant (100 ml) was
`sonicated briefly to shear the remaining DNA/RNA and filtered
`through 0.45 µm filter. An Amicon concentrator unit (with
`YM10 membrane) was used to concentrate the periplasmic
`fraction to ~15 ml. Recombinant NC10 scFvs (VL–VH) proteins
`were affinity purified from the periplasmic fraction using an
`M2 anti-FLAG IgG-Sepharose column as described in the
`previous section. Eluted scFvs proteins were dialysed against
`PBS–0.02% (w/v) sodium azide, concentrated by ultrafiltration
`(YM10, Diaflo) to ~1 mg/ml and stored at 4°C.
`Biochemical and biophysical characterization of NC10 scFv
`proteins
`The purity of the NC10 scFvs was monitored by SDS–PAGE
`and Western blot analysis as described previously (Kortt
`et al., 1994). The concentrations of the scFv fragments were
`determined spectrophotometrically, using values for the extinc-
`tion coefficient (ε0.1%) at 280 nm of 1.66 for VL–VH scFvs
`and 1.70 for VH–VL scFvs calculated as described by Gill and
`von Hippel (1989). The relative molecular mass of each
`affinity-purified NC10 scFv was estimated using size-exclusion
`chromatographic columns (Superose 12 HR10/30 and/or Super-
`dex 200 HR 10/30, Pharmacia) on an HPLC system (Bio-Rad
`Model 700) at 21°C in PBS which were previously calibrated
`with Bio-Rad Gel Filtration Standard proteins. The flow-rate
`was 0.5 ml/min, and the absorbance of the effluent stream was
`monitored at both 214 and 280 nm. Elution times of various
`VL–VH scFv oligomers were compared with those already
`established for VH–VL scFv monomers, dimers and trimers
`(Kortt et al., 1994, 1997; Atwell et al., 1999).
`
`567
`
`UPenn Ex. 2062
`Miltenyi v. UPenn
`IPR2022-00855
`
`

`

`O.Dolezal et al.
`
`Formation of complexes with 3–2G12 anti idiotype Fab⬘
`Fab⬘ fragments of the NC10 anti-idiotype antibody, 3–2G12,
`were prepared as described in Kortt et al. (1997). Purified
`NC10 scFv-0, scFv-1, scFv-2 and scFv-5 (VL–VH) proteins
`were mixed with a small molar excess of 3–2G12 anti-idiotype
`Fab⬘, as described previously by Kortt et al. (1997). The
`complexes were separated from excess Fab⬘ by size-exclusion
`chromatography on Superdex 200 (HR 10/30) in PBS, pH 7.4,
`with a flow-rate of 0.5 ml/min. The column had previously
`been calibrated with uncomplexed scFvs and 3–2G12 Fab⬘.
`Electron microscopy
`Complexes of scFv-0 and scFv-1, scFv-2 and scFv-5 with 3–
`2G12 Fab⬘ and the complex of influenza virus neuraminidase
`(soluble tetrameric extracellular domain) with NC10 Fab⬘
`(Malby et al., 1994) were examined by electron microscopy
`(EM). EM imaging and data analysis were performed as
`described previously (Lawrence et al., 1998; Atwell et al.,
`1999).
`Molecular modeling
`Computer-generated models of NC10 scFv triabodies and
`tetrabodies were constructed using Fv modules that corre-
`sponded to the coordinates of the NC10 Fv domain in PDB
`entry 1NMB (Malby et al., 1994, 1998). Fv modules were
`manipulated as rigid bodies with the O molecular graphics
`package (Jones et al., 1991). Triabody structure corresponded
`to the model described by Kortt et al. (1997) comprising three
`Fv modules with threefold symmetry. Tetrabodies comprised
`four Fv modules with fourfold symmetry. No attempt was
`made to model conformational changes in the Fv domains.
`
`Results and discussion
`Expression of NC10 scFv-15 and scFv-0 constructs
`For direct comparison with the original NC10 scFv-15 and
`scFv-0 (VH–VL) constructs (Malby et al., 1993; Kortt et al.,
`1997), the NC10 scFv-15 and scFv-0 (VL–VH) gene fragments
`were initially constructed and expressed in a heat-inducible
`expression vector pPOW (Figure 1a-iii). As for VH–VL orienta-
`tion, in the VL–VH construct of the NC10 scFv, the VL domain
`was linked to the VH domain using the classical linker design
`of Huston et al. (1988, 1991), whereby the codon for C-
`terminal VL ArgL107 was linked to the codon for N-terminal
`VH GlnH1 via a 15 amino acid residue linker ([G4S]3; Figure
`1b). The ArgL107 was defined from the NC10 scFv crystal
`structure to be the last residue in the VL that made intra-
`domain contacts within VL and additional residues therefore
`formed a true linker (Malby et al., 1998). In case of scFv-0,
`the C-terminal VL ArgL107 was ligated directly to VH GlnH1
`(Figure 1b).
`The pPOW/NC10 scFv-15 and scFv-0 (VL–VH) constructs
`were expressed under the same conditions as those used
`previously for the VH–VL constructs (Malby et al., 1993; Kortt
`et al., 1997). As was the case for the VH–VL proteins, the
`majority of expressed scFv-15 and scFv-0 (VL–VH) proteins
`were located in the periplasm as insoluble protein aggregates
`which were solubilized by extraction with 6 M GuHCl and
`on dialysis refolded into soluble, functional scFv entities.
`Furthermore, there was no apparent difference in expression
`levels between the VL–VH and VH–VL scFvs proteins. The
`resulting soluble scFvs were purified by affinity chromato-
`graphy on an M2 anti-FLAG IgG-Sepharose column. Yields
`of 2–5 and 1–2 mg of soluble affinity-purified protein per litre
`
`568
`
`Fig. 2. Size-exclusion chromatography on a calibrated Superose 12 HR 10/
`30 column of affinity-purified NC10 scFv-15 proteins. Superimposed on the
`same scale are runs for scFv-15 VL–VH (solid line) and scFv-15 VH–VL
`(dashed line). The VL–VH protein eluted as monomer at 28.0 min and dimer
`at 25.5 min. These elution times are consistent with calculated molecular
`masses of 28.5 and 57 kDa, respectively. The VH–VL protein eluted as
`monomer at 28.6 min and dimer at 26.2 min. These elution times are
`consistent with calculated molecular masses of 27 and 54 kDa, respectively.
`The column was equilibrated in PBS, pH 7.4, and the flow-rate was 0.5 ml/
`min.
`
`of shake-flask culture were typically obtained for scFv-15 and
`scFv-0, respectively.
`Biophysical analysis of NC10 scFv-15 and scFv-0 proteins
`Samples of affinity-purified scFv-15 proteins (VL–VH and
`VH–VL) were shown by SDS–PAGE to comprise essentially
`homogeneous scFv preparations with a main protein band at
`~27.0 kDa for VH–VL and ~28.5 kDa for VL–VH (data not
`shown). This apparent difference in protein mobility on SDS–
`PAGE was attributed to the additional sequence associated
`with the second FLAG epitope that has been added to the C-
`terminus of scFv VL–VH proteins (see Materials and methods
`and Figure 1a). Size-exclusion chromatography on a calibrated
`Superose 12 column showed that the oligomeric status of the
`NC10 VH–VL and VL–VH scFv-15 proteins was similar in
`solution (Figure 2). Gel filtration of affinity-purified scFv-15
`VL–VH protein showed the presence of two main peaks with
`apparent molecular masses of ~28.5 kDa (major peak) and
`~57.0 kDa (minor peak), corresponding to scFv-15 monomer
`and dimer, respectively. The scFv-15 VH–VL protein when
`analysed on the same column eluted mainly as monomer at
`~27.0 kDa with a small amount of dimer at ~54.0 kDa. The
`relative differences in elution times for VL–VH and VH–VL
`monomers and dimers were again attributed to the additional
`FLAG sequence at the C-terminus of scFv-15 VL–VH protein.
`Both scFv-15 protein samples also contained traces of higher
`order oligomers, but because of small quantities of these
`species it was not possible to assign their oligomeric status
`unequivocally. Similarly as shown for NC10 scFv-15 VH–VL
`protein (Kortt et al., 1994), the formation of NC10 scFv-15
`VL–VH dimers and higher molecular mass multimers was
`induced by storing the sample at higher concentrations (⬎1
`mg/ml) or by repeated freezing and thawing (data not shown).
`In contrast to scFv-15 proteins, size-exclusion chromato-
`graphy of scFv-0 (VL–VH and VH–VL) proteins demonstrated
`a significant difference in elution profiles (Figure 3a). The
`chromatography of the NC10 scFv-0 (VL–VH) protein yielded
`a major protein peak eluting at ~21.9 min with a distinct
`
`UPenn Ex. 2062
`Miltenyi v. UPenn
`IPR2022-00855
`
`

`

`scFv multimers NC10scFVmultimers
`
`Fig. 4. Reducing SDS–PAGE of affinity-purified NC10 scFvs (VL–VH)
`stained with Coomassie Brilliant Blue G-250. Lanes: 1, scFv-0; 2, scFv-1;
`3, scFv-2; 4, scFv-3; 5, scFv-4; 6, scFv-5; 7, scFv-15.
`
`vector because of its capacity to produce soluble and active
`scFv proteins in the bacterial periplasm. A schematic diagram
`depicting the general outline of these short-linkered NC10
`scFv expression units in pGC is shown in Figure 1b. The pGC
`NC10 scFv VL–VH constructs were expressed in E.coli TOP
`10F⬘ cells using the expression protocol of Dolezal et al.
`(1995). The expressed scFv products were isolated from the
`periplasmic fraction by affinity chromatography as described
`in Materials and methods. The yield of soluble affinity-purified
`protein decreased progressively as the linker length was
`shortened,
`from 2 mg/l bacterial culture for scFv-15 to
`0.5 mg/l bacterial culture for scFv-0.
`Molecular mass analysis of scFv proteins with shorter
`linkers
`SDS–PAGE analysis of affinity-purified NC10 scFv-0, scFv-1,
`scFv-2, scFv-3, scFv-4, scFv-5 and scFv-15 (VL–VH) protein
`samples showed that the scFvs comprised predominantly a
`single component of apparent molecular mass ~27.5–28.5 kDa,
`as expected for this series of proteins (Figure 4). When
`these scFvs were subjected to analysis by size-exclusion
`chromatography on a calibrated Superdex 200 column, a
`number of major protein peaks were observed with a significant
`variation in elution times consistent with the presence of scFv
`oligomers (Figure 5). Elution profiles for scFv-5, scFv-4 and
`scFv-3 (Figure 5a) showed the presence of a single major peak
`with an elution time consistent with a molecular mass of ~57
`kDa expected for a dimer. This peak eluted at the same time
`as the dimer peak in scFv-15 (VL–VH) preparations (Figure
`5b). The minor higher molecular mass species observed in
`these three profiles (Figure 5a) eluted at the same elution times
`as the zero-linked NC10 VL–VH trimer and tetramer peaks
`(Figure 5b). These findings are consistent with those observed
`for the NC10 scFv-5, scFv-4 and scFv-3 (VH–VL) proteins
`which mainly formed dimer and a small amount of trimer
`(Atwell et al., 1999). However, no tetramer species was
`observed for scFv-5, -4 and -3 (VH–VL) proteins. Interestingly,
`in the case of scFv-5 (VL–VH) protein, consistently larger
`yields of tetramer species were observed and this allowed for
`the purification and subsequent characterization of this tetramer
`by gel filtration on a Superdex 200 column and by electron
`microscopy (see below). This scFv-5 tetramer was, however,
`relatively unstable as it partially reverted back to dimer (40%
`dimer after 24 h at 4°C). Similarly, a small amount of purified
`scFv-5 dimer species converted to tetramer (~5–10% after
`24 h at 4°C).
`
`569
`
`Fig. 3. Size-exclusion chromatography on a calibrated Superdex 200 HR 10/
`30 column of affinity-purified NC10 scFv-0 proteins. Columns were
`equilibrated in PBS, pH 7.4, and the flow-rate was 0.5 ml/min. (a) Shows
`the scFv-0 VL–VH tetramer (solid line) eluting at 21.9 min with a trimer
`shoulder on the trailing edge. Superimposed on the same scale is a run for
`scFv-0 VH–VL trimer (dashed line) eluting at 24.1 min. (b) Shows
`separation of scFv-0 tetramer from scFv-0 trimer using two Superdex 200
`HR 10/30 columns linked in tandem.
`
`shoulder on the trailing edge of the peak. The NC10 scFv-0
`(VH–VL) protein, previously demonstrated to be a trimer (Kortt
`et al., 1997), eluted as a single peak at 24.1 min on this
`column. The components of NC10 scFv-0 (VL–VH) protein
`were resolved by gel filtration on two Superose 12 HR10/30
`columns linked in tandem, yielding two protein peaks with
`apparent molecular masses of ~108 and ~78 kDa (Figure 3b).
`In contrast to NC10 scFv-0 (VH–VL), the tandem gel filtration
`indicated that NC10 scFv-0 (VL–VH) forms not only the
`expected trimer (~78 kDa) but also a tetramer (~108 kDa).
`Attempts to isolate homogeneous NC10 scFv-0 tetramer were
`only partially successful, as isolated tetramer rapidly inter-
`converted into a mixture of tetramer and trimer (data not
`shown). This observation indicated a relatively rapid equilib-
`rium between NC10 scFv-0 VL–VH tetramer and trimer with
`the tetramer being the predominant species as shown in
`Figure 3.
`Expression of NC10 VL–VH scFvs with variable linker length
`To investigate further the effect of linker length upon multimer-
`ization of NC10 scFvs in reverse (VL–VH) orientation, seven
`different NC10 scFv constructs with linkers from 15 to 0
`amino acid residues were assembled in pGC secretion vector
`(Coia et al., 1997) as described in Materials and methods.
`This vector was chosen for construction and expression of
`these scFvs with shortened linkers in preference to pPOW
`
`UPenn Ex. 2062
`Miltenyi v. UPenn
`IPR2022-00855
`
`

`

`O.Dolezal et al.
`
`Fig. 5. Size-exclusion chromatography on Superdex 200 HR 10/30 column
`of affinity-purified NC10 scFv VL–VH proteins. The column was
`equilibrated in PBS, pH 7.4, and the flow-rate was 0.5 ml/min. (a) Shows
`the scFv-5 (dotted line), scFv-4 (dashed line) and scFv-3 (solid line), all
`eluting predominantly as dimers; (b) shows scFv-15 (solid line) containing
`monomer and dimer, and scFv-0 (dashed line) eluting as a mixture of
`mainly tetramer and trimer; (c) shows the scFv-1 (dashed line) and scFv-2
`(solid line), each containing a mixture of dimers, trimers, tetramers and
`higher molecular mass multimers.
`
`Size-exclusion chromatographic profiles of scFv-1 and
`scFv-2 yielded a series of peaks consistent with an equilibrium
`mixture of scFv dimers, trimers, tetramers and higher molecular
`mass multimers (Figure 5c). In the case of scFv-2 this
`equilibrium mixture was biased towards the smaller scFv
`entities whereas in case of scFv-1 the equilibrium mixture
`shifted towards higher molecular mass multimers. Analysis of
`entire scFv-1 and scFv-2 protein mixtures in 60% ethylene
`glycol and at lower concentration (⬍100 µg/ml) showed that
`the higher molecular mass multimers dissociated primarily
`into dimers and trimers (data not shown). This suggests that
`the dimers and trimers of scFv-2 and particularly scFv-1 are
`
`570
`
`relatively unstable and as a result prefer to assemble into
`higher molecular mass multimers as the linker length is reduced
`from two to one amino acid residue.
`As described previously (Figure 1), when compared with
`the original NC10 scFv VH–VL proteins, a second FLAG
`sequence was included in our constructs to improve the
`purification efficiency of various NC10 VL–VH proteins. To
`confirm that this additional FLAG sequence does not affect
`the state of oligomerization, several pGC-NC10 VL–VH clones
`containing a single FLAG sequence were also constructed and
`expressed in E.coli. The ‘single-FLAG’ NC10 VL–VH proteins
`with no, one, two and 15 residue linkers were purified on an
`anti-FLAG affinity column and analysed by gel filtration
`chromatography. The resulting gel filtration profiles indicated
`no apparent difference in oligomerization properties between
`‘single-FLAG’ and ‘double-FLAG’ NC10 VL–VH proteins
`(data not shown). The presence of minor bands at ~26 kDa in
`affinity-purified NC10 VL–VH samples (Figure 4) provided
`further evidence that the additional FLAG peptides do not
`affect the oligomeric status of these proteins. These bands
`occur owing to a gradual cleavage of FLAG peptides from the
`C-terminus of NC10 scFv proteins during extended storage at
`4°C (demonstrated by Western blot analysis with the anti-
`FLAG M2 antibody, data not shown). The presence of these
`bands, however, had no significant effect on the gel filtration
`profiles, and hence multimeric forms, of these NC10 scFvs
`(data not shown).
`Formation of complexes with 3–2G12 anti-idiotype Fab⬘
`fragments
`To analyse further the multimeric status of NC10 scFvs
`(VL–VH), the Fab⬘ fragment of 3–2G12 anti-idiotype mono-
`clonal antibody, which competes with influenza virus neura-
`minidase for binding to NC10, was used to form complexes
`with various scFv VL–VH proteins. Affinity-purified scFv-5
`dimer and scFv-5 tetramer, scFv-0 tetramer as well as the
`scFv-2 and scFv-1 mixtures were complexed with 3–2G12
`Fab⬘ fragment and the resulting complexes analysed by gel
`filtration (data not shown) and by electron microscopy (see
`below). In the case of scFv-5,
`the tetramer species was
`separated from dimer by gel filtration just prior to binding to
`3–2G12 Fab⬘. Complexes were formed by mixing dimers (for
`scFv-5) and tetramers (for scFv-5 and scFv-0) in a 1:2 and
`1:4 molar ratio, respectively, with the 3–2G12 Fab⬘ fragment.
`The Fab⬘ was kept in slight excess to these ratios to ensure
`complete decoration of all antigen binding sites present. The
`resulting scFv dimer–Fab⬘ and scFv tetramer–Fab⬘ complexes
`were then purified by size-exclusion chromatography on a
`Superdex 200 column (data not shown). There was no evidence
`of unbound scFv tetramer (for scFv-0 and scFv-5) or scFv
`dimer (for scFv-5) in any of these preparations. The scFv-5
`dimer–Fab⬘ complex eluted at an elution time corresponding
`to a molecular mass of ~160 kDa, which is consistent with
`the mass of a complex of two Fab⬘ molecules (~52 kDa each)
`and one scFv-5 dimer (~55 kDa). The scFv tetramer–Fab⬘
`complex (both scFv-0 and scFv-5) eluted at an elution time
`corresponding to a molecular mass of ~320 kDa, which is
`consistent with the mass of a complex of four Fab⬘ molecules
`and one scFv tetramer (~108 kDa). The estimated molecular
`mass of scFv–anti-idiotype Fab⬘ complexes is consistent with
`the prediction that scFv-5 dimers are bivalent and bind two
`Fab⬘ fragments and that scFv-0 (and scFv-5) tetramers are
`tetravalent and bind four Fab⬘ fragments. Minor peaks were
`
`UPenn Ex. 2062
`Miltenyi v. UPenn
`IPR2022-00855
`
`

`

`scFv multimers NC10scFVmultimers
`
`Fig. 6. Electron micrographs of scFv–3–2G12 Fab⬘ complexes stained with 2% potassium phosphotungstate. Magnification bar is 50 nm. (a) Field of scFv-5
`dimer complexes. (b) Field of scFv-5 tetramer complexes. (c) Field of scFv-0 tetramer complexes. At least one scFv-0 triabody (circled) is also present.
`(d) Field showing projections of compl