The Journal of Immunology
`
`3079
`
`.
`
`r
`
`L
`
`L
`
`i
`
`.
`
`..
`
`L
`
`.-
`
`L
`
`
`
`I?E‘v‘i.-3'Q.‘§E‘1“\;‘SE.{YSLLS‘ FLAT 1
`«
`- vah . ““““““ A
`
`.‘éLI
`'
`
`
`
`(113g;
`.s...
`..,
`‘ WE’QQRPQQPPPLLLifi' 1.5551435
`_KW‘L=.V-»»1‘RE.~
`.DK .......... i-n?-:‘E.Qt
`a..».§ ..........
`.afiLQ
`
` .
`
`F‘CSGQTKLE 1K
`‘R[F 3?}
`.«......v...
`gr"
`t,
`vHSREIJ‘ eeeeee lime
`digital... ...... {,..
`
`\‘1\\,,
`r..\
`EN! N 132‘? "EMZ-i WYKQRP EQGLSEWI '23
`Y'E‘l’f NW ~A§€~§tai3~Avv~Mv
`twang I34»~a
`~»!$-.-‘«-‘3~-»-Mi«-
`
`(Tim
`‘t’t‘i 7‘5” E NU {J'E'F E’RP
`:‘Rfikuw
`"“E~P
`
`-
`-
`-»
`.
`(13m
`Eiiii- S 33%“ 'f Z'RWF E’I‘J‘e' R (LAY 1‘3 VGV’S 'e
`
`-:’\xvl‘-F13'I'GI\\T-Y ":1‘~L\'Y'
`E
`5m-S»~M-TTR\WF-V‘
`-
`...\«-»t.»-t-
`1m- 3 nM- ‘1"1’RWSI F - v
`- Mir - L N? -
`
`.
`
`FIGURE 2. Humanization protocols for mAb COL—1. A, Amino acid sequences of the VL regions of mCOL—1, human Ab VJI’CL, HuCOL—1 derived
`from mCOL—1 and VJI’CL, and the HuCOL—1 variant 24’25’27L. B, Amino acid sequences of the VH regions of mCOL—l, human Ab M030, HuCOL—1
`derived from mCOL—1 and M030, and the variant 61H. Dashes indicate residues that are identical in mCOL—1, M030, HuCOL—1, and variant. Asterisks
`mark framework residues that are deemed essential for maintaining the combining site structure of mCOL—l. Murine framework residues retained in the
`HuCOL—1 are shown in bold.
`
`Abs to the anti—V region Abs. To circumvent this problem, the circulating
`CEA and anti—murine Fc Abs were removed by sequential preadsorption of
`the sera with purified mCOL—6 and mCOL—4 mAbs, two Abs that react with
`epitopes of CEA different from the one recognized by mCOL—1 (28). The
`mAb mCOL—4 has the same isotype as that of mCOL—l. For preadsorption,
`serum samples were added to mCOL—6 coupled to Reacti—gel according to
`the method of Hearn et al. (39). The mixtures were incubated overnight at
`4°C with end—to—end rotation and were centrifuged at 1000 X g for 5 min.
`Preadsorption was repeated until the supernatants displayed no detectable
`anti—murine Fc activity. The procedure was then repeated using mCOL—4
`coupled to Reacti—gel. To detect anti—V region Abs by surface plasmon
`resonance (SPR), the preadsorbed serum was used as a mobile reactant.
`Proteins were immobilized on carboxymethylated dextran CM5 chips
`(BIAcore, Piscataway, NJ) by amine coupling using standard procedure
`(40, 41). HuCOL—1 was immobilized on the surface of flow cell 1, whereas
`the surface of flow cell 2 was coated with an unrelated protein, rabbit y
`globulin (Bio—Rad, Hercules, CA).
`
`Sera reactivity
`
`The reactivity of COL—1 variants to anti—V region Abs was determined
`using a recently developed SPR—based competition assay (58). Competition
`experiments were performed at 25°C using a CM5 sensor chip containing
`either mCOL—1 or HuCOL—1 in flow cell 1 and rabbit y globulin (Bio—Rad),
`as areference, in flow cell 2. Typically, mCOL—1, HuCOL—1, or its variants
`were used at diiferent concentrations, to compete with the Ab immobilized
`on the sensor chip for binding to serum anti—V region Abs. Patient’s serum
`with or without the competitor (mCOL—l, HuCOL—1, or its variants) was
`applied across the sensor surface using a recently developed sample ap—
`plication technique (42) at the unidirectional flow of 1 til/min. After the
`binding was measured for 1000 s, the samples were washed from the sur—
`faces with running buffer using a flow rate of 100 til/min, and the surfaces
`were regenerated with 10 mM glycine (pH 2.0) for the HuCOL—1 sensor
`chip or with HCl (pH 2.3) for the mCOL—1 sensor chip. The percent bind—
`ing at each Ab concentration was calculated as follows: % binding =
`[slope of the signal obtained with competitor (serum + mCOL—1, Hu—
`COL—1, or HuCOL—1 variants)/slope of the signal obtained without com—
`petitor (serum only)] X 100. IC50 for each Ab, the concentration required
`for 50% inhibition of the binding of the serum to either mCOL—1 or Hu—
`COL—1, was calculated.
`
`Results
`Isolation of COL—I H and L chain genes
`
`The genes encoding the L chain and the Fd region of the H chain
`of mAb COL—l were generated by repertoire cloning methodology
`(43), using synthetic oligonucleotides described in Materials and
`Methods. The PCR products of the appropriate size were cloned in
`
`A phage vector, and phagemids carrying the target genes were
`subsequently excised. The cloned genes were sequenced (data not
`shown) before the phagemids were used as templates for the sub—
`sequent PCR amplification.
`
`Generation of genes encoding humanized COL—1 VL and VH
`domains
`
`The mCOL—l was humanized by grafting the CDRs of the L and
`H chains onto the VL and VH frameworks of the appropriate hu—
`man Abs, but retaining those framework residues that were
`deemed essential for preserving the structural integrity of the com—
`bining site (44—46). The lg CDRs have been defined as compris—
`ing residues 31—35b, 50—65, and 95—102 in the H chain and res—
`idues 24—34, 50—56, and 89—97 in the L chain (47). The
`framework residues that were deemed critical were identified on
`the basis of the atomic coordinates of the Abs of known structures
`
`available in the database (for example, see Ref. 48). The human Ab
`sequences that are most similar to mCOL—l are VJI’CL (49) (Gen—
`Bank accession number Z00022) for VL and M030 (50) (GenBank
`accession number A32483) for VH. The alignment of the VL se—
`quences of mCOL—l and VJI’CL, and the VH sequences of
`mCOL—l and M030 are shown in Fig. 2. Also indicated in Fig. 2
`are the locations of the framework residues that are critical for Ag
`binding. The humanization protocols for the VL and VH genes,
`shown in Fig. 2, are based on putting the CDR sequences of mAb
`COL—1 together with the frameworks of the human VL and VH
`templates, while replacing some of the human framework residues
`with those murine framework residues that may be critical for Ag
`binding.
`A nucleotide sequence was then deduced from the amino acid
`sequence of each of the designed humanized VL and VH domains.
`The nucleotide sequences were refined to provide high frequency
`usage of codons and by eliminating, with the help of programs
`FOLD and MAPSORT (51), any self—annealing regions and any
`sites for restriction endonucleases that might complicate cloning of
`the designed genes in the desired vectors. Using the four overlap—
`ping oligonucleotides (shown by long arrows in Fig. 1) that en—
`compassed, on alternating strands, the entire sequence of either the
`VL or VH region and its leader and the respective end primers
`
`