S=2~2~~~~~~~~~~~~~~~~~~LETTERSTONATURE~~~~~~~~~~NA_T_U_R_E_V_O_L_.3_2_1_29~M_AY~19~86
`
`differed between NMDA and KA (Fig. 3c, d), and in individual
`spinal cord neurones KA-evoked increases in [Ca2+]; were
`always much smaller than those evoked by NMDA. These
`experiments suggest that Na+ is a poor trigger for inducing an
`increase in [Caz+];, since in several neurones the inward (Na+)
`current activated by KA produced no detectable arsenazo III
`signal. However, our results do not exclude the possibility that
`Caz+ influx through ion channels activated by NMDA triggers
`release of Ca2+ from intracellular stores27, contributing further
`to the NMDA-evoked arsenazo III signals reported here.
`Although the present results suggest a high Ca2+ permeability
`of NMDA-receptor-activated channels (Fig. 3), the net flux of
`monovalent cations (that is, conductance) decreases in the pres(cid:173)
`ence of Ca2+. This reflects interactions between permeant ions
`within the channel with Ca2+ acting as both a permeant ion and
`as a blocker of monovalent cation flux25·26·28
`•
`The experiments reported here provide evidence for an
`agonist-triggered increase in [Caz+]; in mammalian spinal cord
`neurones. Previously, ion-sensitive microelectrodes were used
`to measure changes in intracellular ionic activity triggered by
`excitatory amino acids in frog motoneurones9
`. The latter experi(cid:173)
`ments suggested an increase in both [Na+]; and [Ca2+]; during
`perfusion with L-glutamate but the results were difficult to
`interpret clearly as (1) neurones were not voltage-clamped and
`thus it is difficult to separate the relative contributions of Ca2+
`influx via voltage-dependent calcium channels and agonist-acti(cid:173)
`vated channels, and (2) L-glutamate is a mixed agonist that acts
`at multiple subtypes of excitatory amino-acid receptor2·6·7.
`The response to NMDA-receptor activation thus provides a
`second source of calcium flux, distinct from that resulting from
`conventional voltage-dependent calcium channels, which may
`have important long-term effects on excitability. Our finding
`that the ion channels linked to the NMDA receptor subtype are
`more permeable to Caz+ than those linked to KA receptors, has
`implications for the role of excitatory amino-acid receptors in
`CNS function. It is possible that Caz+ influx activated by NMDA
`receptors underlies the synaptic plasticity generating long-term
`potentiation, as the latter is prevented by intracellular injection
`of EGTA to chelate Ca2+ (ref. 29), or by blocking NMDA
`receptors with selective antagonists30. For example, Ca2+ influx
`localized at transmitter-operated ion channels could have a role
`in organizing and regulating postsynaptic structures in an
`appropriate spatial relation to transmitter-releasing presynaptic
`terminal boutons, and it is important to consider that Ca2+ influx
`occurring at NMDA receptors located on dendritic spines might
`produce an especially large but localized elevation in intracel(cid:173)
`lular Caz+ concentration, due to restriction of Ca2+ diffusion
`along the narrow shaft of the spine. In addition, our results
`have some bearing on the mechanisms of desensitization of
`NMDA receptors, as the link that has been demonstrated
`between [Caz+]; and desensitization ofnicotinic receptors at the
`neuromuscular junction31·32 may occur also for other receptor(cid:173)
`ionophore complexes. Thus our results may help to explain the
`similar desensitization evoked by either large doses of NMDA
`or depolarizing voltage jumps 7, which trigger Ca2+ entry through
`NMDA channels and voltage-dependent calcium channels,
`respectively.
`
`Received 3 January; accepted 1 April 1986.
`
`1. Krogsgaard-Larsen, P., Honore, T., Hansen, J. J., Curtis, D. R. & Lodge, D. Nature 284,
`64-66 (1980).
`2. Watkins, J. C. & Evans, R. H. A Rev. Pharmac. Tox. 21, 165-205 (1981).
`3. McLennan, H. Prog. Neurobiol. 20, 251-271 (1983).
`4. Nowak, L., Bregestovski, P., Ascher, P., Herbet, A. & Prochiantz, A. Nature 307, 462-465
`(1984).
`5. Mayer, M. L., Westbrook, G. L. & Guthrie, P. B. Nature 309, 261-263 (1984).
`6. Mayer, M. L. & Westbrook, G. L. J. Physiol., Lond. 354, 29-53 (1984).
`7. Mayer, M. L. & Westbrook, G. L. J. Physiol., Lond. 361, 65-90 (1985).
`8. Dingledine, R. J. PhysioL. Lond. 343, 385-405 (1983).
`9. Biihrle, C. P. & Sonnhof, U. Pfliigers Arch. ges. Physiol. 396, 154-162 (1983).
`10. Zanotto, L. & Heinemann, U. Neuroscl Lett. 35, 79-84 (1983).
`11. Pumain, R. & Heinemann, U. J. NeurophysioL 53, 1-16 (1985).
`12. Lansman, J. B., Hess, P. & Tsien, R. W. J. gen. PhysioL (in the press).
`
`13. Ault, 8., Evans, R. H., Francis, A. S., Oakes, D. J. & Watkins, J. C. J. Physiol., Lond. 307,
`413-428 (1980).
`14. Crunelli, V. & Mayer, M. L. Brain Res. 311, 392-396 (1984).
`15. Hamill, 0. P., Marty, A., Neher, E., Sakmann, B. & Sigworth, F. Pjliigers Arch. ges. Physiol.
`391, 85-100 (1981).
`16. Cull.Candy, S. G. & Ogden, D. C. Proc. R Soc. B224, 367-373 (1985).
`17. Hagiwara, S. & Bylerly, L. A Rev. Neurosci. 4, 69-125 (1981).
`18. Smith, S. J., MacDermott, A. B. & Weight, F. F. Nature 304, 350-352 ( 1983).
`19. Gorman, A. L. F. & Thomas, M. V. J. PhysioL, Lond. 308, 259-285 (1980).
`20. Berridge, M. J. & Irvine, R. F. Nature 312, 315-321 (1984).
`21. Sladeczek, F., Pin, J.P., Recasens, M., Bockaert,J. & Weiss, S. Nature 317, 717-719 (1985).
`22. Schoffelmeer, A. M. N. & Mulder, A.H. J. Neurochem. 40, 615-621 (1983).
`23. Evans, R.H. & Watkins, J.C. J. Physiol., Lond. 277, 57P (1977).
`24. Nowak, L. M. & Ascher, P. Soc. Neurosci. Abstr. 10, 23 (1984).
`25. Mayer, M. L. & Westbrook, G. L. Soc. Neurosci. Abstr. 11, 785 (1985).
`26. Ascher, P. & Nowak, L. J. Physiol., Lond. Proc. (in the press).
`27. Fabiato, A. & Fabiato, F. Ann. N. Y. Acad. Sci. 307, 491-522 (1978).
`28. Nowak, L. M. & Ascher, P. Soc. Neurosci. Abstr. 11, 953 (1985).
`29. Lynch, G., Larson, J., Kelso, S., Barrinuevo, G. & Schottler, F. Nature 305, 719-721 (1983).
`30. Collingridge, G. L., Kehl, S. J. & McLennan, H.J. Physiol., Lond. 334, 33-46 (1983).
`31. Parsons, R. L. in Calcium in Drug Action (ed. Weiss, G. B.) 289-314 (Plenum, New York,
`1978).
`32. Miledi, R. Proc. R Soc. 8209, 447-452 (1980).
`33. Adams, D. J., Dwyer, T. M. & Hille, B. J. gen. Physiol. 75, 493-510 (1980).
`34. Edwards, C. Neuroscience 7, 1335-1366 (1982).
`
`Replacing the complementarity(cid:173)
`determining regions in a human
`antibody with those from a mouse
`
`Peter T. Jones, Paul H. Dear, Jefferson Foote,
`Michael S. Neuberger & Greg Winter
`
`Laboratory of Molecular Biology, Medical Research Council,
`Hills Road, Cambridge CB2 2QH, UK
`
`The variable domains of an antibody consist of a fl-sheet
`framework with hypervariable regions (or complementarity-deter(cid:173)
`mining regions--CDRs) which fashion the antigen-binding site.
`Here we attempted to determine whether the antigen-binding site
`could be transplanted from one framework to another by grafting
`the CD Rs. We substituted the CD Rs from the heavy-chain variable
`region of mouse antibody Bl--8, which binds the hapten NP-cap
`(4-hydroxy-3-nitrophenacetyl caproic acid; KNP-cap = 1.2 µM), for
`the corresponding CDRs of a human myeloma protein. We report
`that in combination with the Bl--8 mouse light chain, the new
`antibody has acquired the hapten affinity of the Bl--8 antibody
`(KNr-cap = 1.9 µM). Such 'CDR replacement' may offer a means
`of constructing human monoclonal antibodies from the correspond(cid:173)
`ing mouse monoclonal antibodies.
`The three-dimensional structures of several immunoglobulins
`show that the variable domains consist of two /3-sheets pinned
`together by a disulphide bridge, with their hydrophobic faces
`packed together1-3. The individual /3-strands are linked by loops
`which at one tip of the /3-sheet may fashion a binding pocket
`for small haptens1·2. Sequence comparisons among heavy- and
`light-chain variable domains (V8 and Y1. respectively) reveal
`that each domain has three CDRs flanked by four relatively
`conserved regions (framework regions-FRs) 4
`. As seen in the
`structure of the human myeloma protein NEWM (Fig. 1), the
`CDRs include each of the three main loops. Often the CDRs
`also include the ends of the /3-strands, suggesting that side
`chains at the ends of the fl-strands may help to fix the conforma(cid:173)
`tion or orientation of the loops. The framework regions form
`the bulk of the /3-sheet, although for example in the V8 domain
`of NEWM, FRI includes part of the loop between the two
`/3-sheets and CDR2 not only forms a loop but a complete
`/3-strand (Fig. 1). The structure of the /3-sheet framework is
`similar in different antibodies, as the packing of different side
`chains is accommodated by slight shifts between the two /3-
`strands5. Furthermore, the packing together of Y1. and V8 FRs
`is conserved6
`, therefore the orientation of Y1. with respect to V8
`is fixed. We wondered whether the FRs represent a simple
`/3-sheet scaffold on which new binding sites may be built, and
`
`
`
`© Nature Publishing Group1986
`
`1 of 4
`
`Celltrion, Inc., Exhibit 1033
`
`

`

`Fig. 1 Stereo pairs of the VH (right)
`and VL (left) domains of the human
`myeloma protein NEWM 1
`8 gener(cid:173)
`•
`ated using the computer graphics
`program FROD025
`. The tracings
`indicate the backbone of C" atoms
`for the framework regions. a, The C"
`atoms of the CDRs (e) cluster at the
`tip of the variable domain. b, A view
`into the hapten binding pocket with
`the CDRs in the order (clockwise
`from noon): VH CDR3, CDRl and
`CDR2, and VL CDR3, CDRl and
`CDR2. The side chains lining the
`binding pocket (VL A 28, N 30, Y 90,
`S93, R95; VH W47, Y50, F52,
`I 100, A 101) lie almost entirely in
`the CDRs. c, The C" atoms in the
`NEWM VH domain are marked (e)
`where side chains in the mouse Bl-8
`V H domain are different. The side
`chains (VL Y35, Q37, A42, P43,
`y 86, F 99; VH v 37, Q 39, L 45, y 94,
`W 107) involved in packing VH and
`VL framework regions are traced. In
`the VH domain all these side chains
`are conserved in mouse Bl-8.
`
`whether the structure of the CDRs (and antigen binding) is
`therefore independent of the FR context. To answer these ques(cid:173)
`tions experimentally, we have grafted the CDRs from one anti(cid:173)
`body to another, to determine whether antigen binding transfers
`with the CDRs.
`We grafted the CDRs from the VH domain of the mouse
`monoclonal antibody Bl-8 (ref. 7) into the VH domain of the
`human myeloma protein NEWM, whose crystallographic struc(cid:173)
`8
`ture is known 1
`. The VH domain of the CDR donor (Bl-8) is
`•
`attached to a µ, constant region and associated with a mouse
`A 1 light chain, and the antibody is directed against the hapten
`NP-cap. Both the VH and VL domains seem to have a role in
`determining the affinity of the antibody for NP-cap as the
`substitution of either domain by other, often highly related
`variable domains can destroy hapten binding (refs 7, 9 and
`M.S.N., unpublished results). In the VH domain, each of the
`CDRs has been implicated in NP-cap binding10
`, but the class
`of constant domains attached to VH does not seem to affect
`12
`binding of hapten 11
`. The CD Rs from the VH domain of anti(cid:173)
`•
`body Bl-8 (ref. 13) are longer than the CDRs which they replace
`in NEWM4 and this may give rise to a deeper binding pocket.
`Most of the residues conserved between the VH domains of
`B 1-8 and NEWM are located in FR2, FR4 and the carboxy(cid:173)
`terminal third of FR3 (Fig. 2a) and largely form the region of
`{:!-sheet which is packed against the light chain. Therefore, it
`might be expected that the VH domain of Bl-8 (hereafter
`abbreviated to MVNP) and the hybrid Bl-8/NEWM domain
`(HuVNP) would dock in a similar manner with the mouse VL
`domain to form the antigen-combining site6
`• The more variable
`
`FRI and N-terminal two-thirds of FR3 form the other {:!-sheet
`which is exposed to solvent (Fig. le).
`The gene encoding the HuVNP domain was constructed by
`gene synthesis (Fig. 2b ). We then constructed a plasmid, pSV(cid:173)
`Hu VNPHe, in which the HuVNP domain is linked to a humane
`constant region, and cloned into a pSV2gpt-derived vector14
`.
`The plasmid DNA was introduced into cells of the J558L mouse
`myeloma by spheroplast fusion. J558L secretes A 1 light chains
`which have been shown to associate with heavy chains contain(cid:173)
`ing a MVNP variable domain, to create a binding site for NP-cap
`or the related hapten NIP-cap (3-iodo-4-hydroxy-5-nitrophenyl(cid:173)
`acetyl caproic acid) 7
`• As the plasmid pSV-HuVNpHe contains
`the gpt marker (encoding guanine phosphoribosyltransferase),
`stably transfected myeloma cells could be selected in medium
`containing mycophenolic acid 14
`; transfectants would be ex(cid:173)
`pected to secrete an antibody (HuVNp-IgE) with a heavy chain
`composed of a HuVNP variable domain and human e constant
`regions, and the A 1 light chain of the J558L myeloma. The
`culture supematants of several gpt+ clones were assayed by
`radioimmunoassay and found to contain NIP-cap-binding anti(cid:173)
`body. The antibody secreted by one such clone was purified
`from the culture supernatant by affinity chromatography on
`NIP-cap-Sepharose, and by SDS-polyacrylamide gel electro(cid:173)
`phoresis the protein was indistinguishable from the mouse
`chimaeric MVNp-IgE (ref. 12) (results not shown). The HuVNP(cid:173)
`IgE antibody competes effectively with MVNp-IgE for binding
`to both anti-human e (Fig. 3a) and NIP-cap coupled to bovine
`serum albumin (NIP-BSA) (Fig. 3b).
`The affinities of HuVNp-IgE for NP-cap and NIP-cap were
`
`
`
`© Nature Publishing Group1986
`
`2 of 4
`
`Celltrion, Inc., Exhibit 1033
`
`

`

`-5~~~~~~~~~~~~~~~~~~~LE1TERSTQNATLJR£~~~~~~~~~--'-N~A~TU=R=E~v~o=L~·~32~1~2~9~M~A~Y~l~98~6
`
`a
`t£un
`
`··-·
`
`FAI
`1 - - - - -
`- - - -30
`KUQLQESGPGLURPSQTLSL TCTUSGSTFS
`~QP_!!RE~K!GRSUK~S£KR~Y~T
`
`31
`
`35
`
`CORI
`HDYYT
`SYMMH
`
`FA2
`36- - - - - -4 ;
`t£un
`MURQPPGRGLEM I G
`11-8 ~KgR~
`
`l50
`
`CDA2
`65
`YUFYHGTSDDTTPLRS
`RI DPHSGGTKYHEKFKS
`
`t£un
`
`··-·
`.....
`
`81-8
`
`FA3
`- - ---94 ~
`66 - -
`-
`CllA3
`102
`RUTMLUOTSKHQFSLRLSSUTRRDTRUYYCRR
`HLIRGCIOU
`KR!l T!!!!KPSSTRYMQ~l,!.SE,!!S~
`YDYYGSSYFOY
`
`FA4
`100--
`--113
`MGOGSLUTUSS
`!!E2ETTl~
`
`b
`
`-48
`I H;ndl 11
`-23
`-7
`. ATiiCiWilccrcT-TCTACATOGTfWITATllOOTTTOTC~
`5 . .• . .
`. . . .
`............. Piil 5lcrts
`............. NtA starts
`CACAMCftOll Z U CATGAGATCACfllGTTCTCTCTACAOTTRCTGAGCRCRCAOGACCTC
`
`,. leodr
`Spl ic:.
`ft T A rJ!
`'" 0 W S C
`I L F L V
`I
`..::CRTGGGATGGAGCTGTATCATCCTCTTCTTGGTAGCMCAGCTACAGGTAAGGGGCTC
`
`ACllGTROCAOOCTTOAOGTCTOOACATATATATOOOTGACAATGACATCCACTTTGCCTT
`~ P9 tl
`lo U H S Q U 0 L 0 ! E S G P G
`Spl ic•
`1
`10
`
`3'
`
`l U R
`TCTCTCCACAGGTGTCCACTCCCllOOTCCAACTGCA GGAGAGCGGTCCllOOTCTTOTGAG
`5'
`1 - - -
`2~,.---
`30
`CORI
`~
`~
`~
`P SQ T L SL T C TU~ GS T F sEL:Li)
`ACCTAOCCllOACCCTOAGCCTOACCTOCACCOTOTCTOOCAOCACCTTCAOCAOCTACTO
`-''-------3
`5
`L-7-
`6------i r - - - -8 - - -
`4' ------,
`-
`50 ~ 52R
`3'
`40
`45
`c:::n::::Jj]w u. Q pp G . G L E w I ol• 1 0 •I
`GATGCACTGGGTGAGACAGCCACCTGGACGAGGTCTTGAGTGGATTGGAAGGATTGATCC
`- -7 .___J ~-----' '---9b---J L---11-
`-----, r----1oa---., ,....--- 10b----, . - - 12114-
`COA2
`60
`65
`70
`~5
`Ins o o T KY n EK F K slR v T,, Luo
`TMTRGTGGTGGTRC'fMGTACfllllTOAGAAGTTCAAOAOCAGAGTGRCARTGCTGGTRGA
`--11-----' L--t3Q----' L---13b~ ~1s-
`12114
`r - - -16 - - -
`80
`82ABC
`85
`~
`T S K H Q F S L R L S S V T R A D T A V
`CACCAGCAAGAACCAOTTCAOCCTGAGllCTCAGCAGCGTGACAGCCGCCGACACCGCGGT
`- tS - . . .J
`17
`L - -10 -
`ro - - -
`'lO
`IOOfl B C
`~ COA3
`lo:!
`y y c A Riv 0 y y G s
`s y F 0 Ylw 0 Q 0
`CTRTTRTTOTOCfWQUftCO'tTTACTACOOTAOTAOCTACTTTOACTACTOOOOTCflllllJO
`- - IQ - - - ' ' - -21 - -J L - - -23 - ' L--
`r--20a-
`22/24
`
`11
`
`Fi1. 2 a, Amino-acid sequence of the VH domain of the human myeloma
`protein NEWM compared with that of the mouse Bl-8 (anti-NP) antibody,
`and divided into FRs and CDRs according to Kabat et al.4. Conserved
`residues are marked with a line above and below the residue. b, Amino-acid
`and nucleotide sequence of the HuVNP gene, in which the CDRs from the
`VH domain of the mouse Bl-8 antibody alternate with the FRs of human
`NEWM protein. The gene was constructed by replacing a section of the
`MVNP gene in the vector pSV-VNr (ref. 12) with a synthetic fragment
`encoding the HuVNP domain. Thus the 5' and 3' noncoding sequences, the
`leader sequence, the leader-variable region intron, five N-terminal and four
`C-terminal amino acids are derived from the MVNP gene; the rest of the
`coding sequence is derived from the synthetic HuVNr fragment. The oligonu(cid:173)
`cleotides are aligned with the corresponding portions of the HuVNr gene.
`For convenience in cloning, the ends of oligonucleotides 25 and 26b form
`a Hindll site followed by a Hindlll site, and the sequences of the 25/ 26b
`oligonucleotides therefore differ from the HuVNP gene.
`Methods. The synthetic gene for HuVNP was constructed as a Pstl-Hindlll
`fragment. The nucleotide sequence was derived from the protein sequence
`using the computer program ANAL YSEQ26 with optimal codon usage taken
`from the sequences of mouse constant-region genes. The oligonucleotides used in synthesis (l-26bi 28 in total) vary in size from 14 to 59 bases and were made
`on a Biosearch SAM or an Applied Biosystems machine, and purified on 8 M urea/ polyacrylamide gels 7
`• The oligonucleotides were assembled in eight single-stranded
`blocks (A-0 and A'-D') containing oligonucleotides 1, 3, 5 and 7 (block A), 2, 4, 6 and 8 (block A'), 9, 11, 13a and 13b (block B), !Oa, !Ob and 12/14 (block
`B'), 15 and 17 (block C), 16 and 18 (block C'), 19, 21, 23 and 25 (block D), and 20, 22/24, 26a and 26b (block D'). In a typical assembly, for example of block
`A, 50 pmol of oligonucleotides 1, 3, 5 and 7 were phosphorylated at the 5' end with T4 pol~nucleotide kinase and mixed with 5 pmol of the terminal oligonucleotide
`which had been phosphorylated with 5 µ.Ci of [ y- 32P]ATP (Amersham; 3,000 Ci mmol- ). These oligonucleotides were annealed by heating to 80 °C and cooling
`to room temperature over 30 min with unkinased oligonucleotides 2, 4 and 6 as splints in 150 µ.l of 50 mM Tris-HCI pH 7.5, to mM MgC1 2 • For the ligation, ATP
`(1 mM) and dithiothreitol (10 mM) were added, together with 50 units of T4 DNA ligase (Anglian Biotechnology Ltd ), and the mixture was incubated for 30 min
`at room temperature. EDTA was added to 10 mM, the sample extracted with phenol, precipitated from ethanol, dissolved in 20 µ.I of water and boiled for I min
`with an equal volume of formamide dyes. Then the sample was loaded onto a thin (0.3 mm) 8 M urea/ 10% polyacrylamide gel27 and a band of the expected size
`detected by autoradiography and eluted by soaking. The two full-length single strands were assembled from A-0 and A'-0' using splint oligonucleotides; thus,
`A-0 were annealed and ligated in 30 µ.las above with 100 pmol each of oligonucleotides !Oa, 16 and 20 as splints, then incubated overnight (A'-0' were constructed
`with oligonucleotides 7, 13b and 17 as splints). After phenol/ ether extraction blocks A-0 were annealed with blocks A'-0', small amounts were cloned in the
`vector Ml3 mpl8 (ref. 28) then cut with Psti and Hindlll, and the gene sequenced by the dideoxy technique29
`• The MVNP gene was transferred as a Hindlll-BamHI
`fragment from the vector pSV-VNP (ref. 12) to the vector Ml3mp8 (ref. 30). To facilitate the replacement of MVNP coding sequences by the synthetic HuVNP
`fragment, three Hindll sites were removed from the 5' noncoding sequence by site-directed mutagenesis, and a new Hindll site subsequently introduced near the
`end of FR4. By cutting the vector with Pstl and Hindll, most of the VNP coding sequence falls out and the synthetic fragment could be introduced as a Pstl-Hindll
`fragment. The sequence at the Hindll site was corrected to give NEWM FR4 by site-directed mutagenesis. The Hindlll-BamHI fragment, now carrying the
`HuVNP gene, was excised from M13 and cloned back into pSV-VNP to replace the MVNP gene (and yield the vector pSV-HuVNp). Finally, the heavy-chain constant
`domains of human lgE (ref. 31) were introduced as a BamHI fragment to yield the vector pSV-HuVNPH., which was transfected into the myeloma line J558L by
`spheroplast fusion. The sequence of the HuVNP gene in pSV-HuVNpHE was checked by re-cloning the Hindlll-BamHl fragment back into Ml3mp8 (ref. 30).
`
`s L u11~ u s s 15plice
`CAGCCTCGTCACAOTCTCCTCllOGT .
`. .. .. IQ:Jbp .
`-25--GllCA 3 '
`- , ,--:zet.-CTGTTCGA 5 '
`
`Table 1 Affinity of HuVNp-lgE and MVNp-lgE for the haptens
`NP-cap and NIP-cap
`
`MVNp-lgE
`HuVNp-lgE
`
`KNP-cap (µ.M)
`1.2±0.1
`1.9±0.2
`
`KNIP-<0ap (µ.M)
`0.02±0.01
`O.o?±0.02
`
`The affinity of HuVNp-lgE and MVNp-lgE for NP-cap was determined
`by fluorescence quenching with excitation at 295 nm and emission
`observed at 340 nm (ref. 22). Antibody solutions were diluted to 100 nM
`in phosphate-buffered saline, filtered (0.45 µ.m-pore cellulose acetate)
`and titrated with NP-cap in the range 0.2-20 µ.M. As a control, the
`mouse Dl-3 antibody23
`, which does not bind hapten, was titrated in
`parallel. A decrease in the ratio of the fluorescence of HuVNp-lgE or
`MVNp-lgE (as appropriate) to that of the Dl-3 antibody was taken as
`being proportional to NP-cap occupancy of the antigen-binding sites.
`The maximum quench was -40% for both HuVNp-lgE and MVNp-lgE,
`and hapten dissociation constants were determined from least-squares
`fits of triplicate data sets to a hyperbola. The concentration of NIP-cap
`was varied from 10 to 300 nM, and -50% quenching of fluorescence
`was observed at saturation. As the antibody concentrations were compar(cid:173)
`able to the values of the dissociation constants, data were fitted by
`least-squares to an equation24 describing tight binding inhibition.
`
`then measured directly using the fluorescence quench technique,
`and compared with those ofMVNp-IgE (Table 1). The antibodies
`HuVNp-lgE and MVNp-lgE have similar affinities for either
`hapten (NP-cap or NIP-cap), and although the affinity of
`HuVNp-IgE for both haptens is slightly lower than that of MVNP(cid:173)
`lgE (2-3-fold, 0.3-0.6 kcal mol- 1
`), the difference in affinity is
`less than expected for loss of either a hydrogen bond or van
`der Waals' contact from the active site of an enzyme 15
`16
`• Thus,
`•
`it seems that binding affinity and specificity for hapten can be
`conferred on a human antibody by grafting in the CDRs from
`an appropriate mouse antibody.
`Is this result likely to be general? This would assume (1) that
`antigen usually binds to the CD Rs, and any contacts to the FRs
`are made to the polypeptide backbone or to conserved side
`chains, and (2) that substitutions in the FRs do not usually
`affect the conformation of the CDR loops. These assumptions
`seem reasonable: thus, in the structure of a complex of the Dl-3
`antibody with lysozyme (R. A. Mariuzza, S. Phillips and R. J.
`Poljak, personal communication) most contacts to the lysozyme
`are made by the CDRs, but there is also a hydrogen bond in
`FRI of the V H domain from the /3-0H ofThr 30 (often conserved
`or replaced by Ser). Similarly, the conformation of CDR loops
`
`
`
`© Nature Publishing Group1986
`
`3 of 4
`
`Celltrion, Inc., Exhibit 1033
`
`

`

`Fig. 3 Comparison ofHuVNP and MVNP lgEs in binding
`inhibition assays. Various concentrations of HuVNp-lgE
`(e) and MVNp-lgE (0) were used to compete the binding
`of radiolabelled MVNP-lgE to polyvinyl microtitre plates
`that had been coated with a, sheep anti-humane antiserum
`(Seward Laboratory); b, (NIP-cap)i4-BSA; c, Ac38 anti(cid:173)
`idiotypic antibody; d, Ac146 anti-idiotypic antibody; e,
`rabbit anti-MVNP antiserum. Binding was also carried out
`in the presence of MVNp-lgM antibody JWl/2/2 (ref. 32)
`(•)as well as in the presence of JW5/1/2 (D), which is
`an lgM antibody that differs from JWl/2/2 at 13 residues
`mainly located in VH CDR2 (M.S.N., unpublished
`results). Values of binding are relative to the binding in
`the absence of inhibitor.
`
`Anti-human £
`
`100
`
`50
`
`CJ)
`c
`u
`c
`:c;
`
`a
`
`~ Ol---'-~~----'~~~'--_,4=-----'~~~'--~~-'-~.at-~~~~~~~~~-,
`Anti-V 81 _8
`
`~\ Ac38
`
`~
`100
`
`~
`
`\__,
`'',,,,, \
`
`50
`
`c
`
`\
`
`\,,',,
`
`d
`
`e
`
`\
`
`\ ~
`
`\
`\, ____ __
`
`'a.. __
`
`o~-o~.1~~~~~~,~o~~~~o_-,~~~~~-,~o~~~-0~.1~~~~~~1~0~-
`
`[lnhibitor]lµg ml- 1 l
`
`between ,B-strands depends on loop size and specific interactions
`of the loop back to the ,B-sheet. However, in the same class of
`variable domains (V H• V K or VA) these interactions are usually
`conserved (ref. 5 and A. M. Lesk and C. Chothia, personal
`communication).
`While human monoclonal antibodies have therapeutic poten(cid:173)
`tial in human disease, they can be difficult to prepare17 and
`treatment of patients with mouse monoclonal antibodies often
`increases the titre of circulating antibody against the mouse
`immunoglobulin 18
`. As chimaeric antibodies containing human
`constant domains 12
`19
`20 and variable domains made by grafting
`•
`•
`mouse CD Rs into human FRs, could have therapeutic potential,
`we wondered whether the HuVNp-lgE antibody loses antigenic
`determinants associated with
`the MVNP variable
`region
`(idiotopes). The binding of HuVNP-lgE and MVNp-lgE to both
`monoclonal and polyclonal anti-idiotypic antibodies directed
`against the MV NP domain was examined by using inhibition
`assays. As shown in Fig. 3d, the HuVNp-lgE antibody has lost
`the MVNP idiotypic determinant recognized by antibody Ac146
`(ref. 21). Furthermore, HuVNp-lgE also binds the antibody Ac38
`(ref. 21) less well (Fig. 3c), therefore it is not surprising that
`HuVNp-lgE has lost many of the determinants recognized by a
`polyclonal rabbit anti-idiotypic antiserum (Fig. 3e ). While the
`loss of idiotypic determinants that accompanies 'humanizing'
`of the VH region is reassuring in view of potential therapeutic
`applications, it does suggest that the recognition of the hapten
`and of anti-idiotypic antibodies is not equivalent. Thus the
`HuVNp-lgE antibody retains hapten binding but has lost
`idiotypic determinants, indicating that the immunoglobulin uses
`different sites to bind hapten and anti-idiotypic antibodies. It
`appears, therefore, that both FR and CDR side chains form the
`binding site for these anti-idiotopes, but mainly CDR side chains
`interact with hapten.
`We thank C. Milstein for suggesting this project, K. Rajewsky
`and M. Reth for the anti-idiotypic antibodies Ac38 and Ac146,
`and A. M. Lesk, C. Chothia, R. J. Leatherbarrow and C. Milstein
`for helpful discussions. J.F. is a Fellow of the Jane Coffin Childs
`Memorial Fund for Medical Research.
`
`Received 17 February; accepted 17 March 1986.
`
`1. Poljak, R. J. et al. Proc. narn. Acad. Sci. U.S.A. 70, 3305-3310 (1973).
`2. Segal, D. M. et al. Proc. natn. Acad. Sci U.S.A. 71, 4298-4302 (1974).
`3. Marquart, M., Oeisenhofer, J., Huber, R. & Palm, W. J. molec. Biol. 141, 369-391 (1980).
`4. Kabat, E. A, Wu, T. T., Bilofsky, H., Reid-Miller, M. & Perry, H. in Sequences of Proteins
`of Immunological Interest (U.S. Department of Health and Human Services, 1983).
`5. Lesk, A. M. & Chothia, C. J. molec. Biol. 160, 325-342 (1982).
`6. Chothia, C., Novotny, J., Bruccoleri, R. & Karplus, M. J. molec. Biol. 186, 651-663 (1985).
`7. Reth, M., Hiimmerling, G. J. & Rajewsky, K. Eur. J. Immun. 8, 393-400 (1978).
`8. Saul, F. A., Amzel, M. & Poljak, R. J. J. biol. Chem. 253, 585-597 (1978).
`9. Briiggemann, M., Radbruch, A. & Rajewsky, K. EMBO J. 1, 629-634 (1982).
`
`10. Reth, M., Bothwell, A. L. M. & Rajewsky, K. in Immunoglobulin Jdiotypes and Their
`Expression (eds Janeway, C., Wigzell, H. & Fox, C. F.) 169-178 (Academic, New York,
`1981).
`11. Neuberger, M. S. & Rajewsky, K. Proc. natn. Acad. Sci. U.S.A. 78, 1138-1142 (1981).
`12. Neuberger, M. S. et al. Nature 314. 268-270 (1985).
`13. Bothwell, A. L. M. et al. Cell 24. 625-637 (1981).
`14. Mulligan, R. C. & Berg, P. Proc. natn. Acad. Sci. U.S.A. 78, 2072-2076 (1983).
`15. Fersht, A. R. et al. Nature 314, 235-238 (1985).
`16. Fersht, A. R., Wilkinson, A. J., Carter, P. & Winter, G. Biochemistry 24, 5858-5861 (1985).
`17. Boyd, J.E., James, K. & McClelland, D. B. L. Trends Biotechnol. 2, 70-77 (1984).
`18. Shawler, D. L., Bartholomew, R. M., Smith, L. M. & Dilman, R. 0. J. Immun. 135, 1530-1535
`(1985).
`19. Morrison, S. L., Johnson, M. J., Herzenberg, L.A. & Oi, V. T. Proc. natn. Acad. Sci. U.S.A.
`81, 6851-6855 (1984).
`20. Boulianne, G. L., Hozumi, N. & Shulman, M. J. Nature 312, 643-646 (1984).
`21. Reth, M., Imanishi-Kari, T. & Rajewsky, K. Eur. J. lmmun. 9, 1004-1013 (1979).
`22. Eisen, H. N. Meth. med. Res. 10, 115-121 (1964).
`23. Mariuzza, R. A. et al. J. molec. Biol. 170, 1055-1058 (1983).
`24. Segal, I. H. in Enzyme Kinetics, 73-74 (Wiley, New York, 1975).
`25. Jones, T. A. in Computational Crystallography (ed. Sayre, D.) 303-310 (Clarendon, Oxford,
`1982).
`26. Staden, R. Nucleic Acids Res. 12, 521-538 (1984).
`27. Sanger, F. & Coulson, A. FEBS Lett. 87, 107-110 (1978).
`28. Yanisch-Perron, C., Vieira, J. & Messing, J. Gene 33, 103-119 (1985).
`29. Sanger, F., Nicklen, S. & Coulson, A. R. Proc. natn. Acad. Sci. U.S.A. 74, 5463-5467 (1977).
`30. Messing, J. & Vieira, J. Gene 19, 269-276 (1982).
`31. Flanagan, J. G. & Rabbitts, T. H. EMBO J. 1, 655-660 (1982).
`32. Neuberger, M. S., Williams, G. T. & Fox, R. 0. Nature 312, 604-608 (1984).
`
`Regulation of human insulin
`gene expression in transgenic mice
`
`Richard F Selden*, Marek J. Skoskiewiczt,
`Kathleen Burke Howie*, Paul S. Russellt
`& Howard M. Goodman*
`
`Departments of *Molecular Biology and tSurgery, Massachusetts
`General Hospital, and Departments of *Genetics and tSurgery,
`Harvard Medical School, Massachusetts General Hospital, Boston,
`Massachusetts 02114, USA
`
`Insulin is a polypeptide hormone of major physiological import(cid:173)
`ance in the regulation of fuel homeostasis in animals (reviewed in
`refs 1, 2). It is synthesized by the /J-cells of pancreatic islets, and
`circulating insulin levels are regulated by several small molecules,
`notably glucose, amino acids, fatty acids and certain pharmacologi(cid:173)
`cal agents. Insulin consists of two polypeptide chains (A and B,
`linked by disulphide bonds) that are derived from the proteolytic
`cleavage of proinsulin, generating equimolar amounts of the
`mature insulin and a connecting peptide (C-peptide). Humans, like
`4
`most vertebrates, contain one proinsulin gene3
`, although several
`•
`7
`species, including mice5 and rats6
`, have two highly homologous
`•
`insulin genes. We have studied the regulation of serum insulin
`
`
`
`© Nature Publishing Group1986
`
`4 of 4
`
`Celltrion, Inc., Exhibit 1033
`
`