��
`
`������
`
`M _A_Y_1�986
`� � N_AT_U _R_E_V_O_L_._32_1_29
`�
`
`�
`� � �����
`S=22"-� ��������
`ERSTONATURE
`LETT
`between NMDA and KA (Fig. 3c, d), and in individual
`13. Ault, 8., Evans, R. H., Francis, A. S., Oakes, D. J. & Watkins, J. C. J. Physiol., Lond. 307,
`differed
`413-428 (1980).
`
`spinal cord neurones KA-evoked increases in [Ca2+]; were
`V. & Mayer, M. L. Brain Res. 311, 392-396 (1984).
`14. Crunelli,
`
`15. Hamill, 0. P., Marty, A., Neher, E., Sakmann, B. & Sigworth, F. Pjliigers Arch. ges. Physiol.
`always much smaller than those evoked by NMDA. These
`391, 85-100 (1981).
`
`
`
`
`experiments suggest that Na+ is a poor trigger for inducing an
`S. G. & Ogden, D. C. Proc. R Soc. B224, 367-373 (1985).
`16. Cull.Candy,
`
`
`
`increase in [Caz+L since in several neurones the inward (Na+)
`L. A Rev. Neurosci. 4, 69-125 (1981).
`17. Hagiwara, S. & Bylerly,
`18. Smith, S. J., MacDermott, A. B. & Weight, F. F. Nature 304, 350-352 ( 1983).
`
`
`
`
`
`current activated by KA produced no detectable arsenazo III
`19. Gorman, A. L. F. & Thomas, M. V. J. PhysioL, Lond. 308, 259-285 (1980).
`
`
`
`
`signal. However, our results do not exclude the possibility that
`
`20. Berridge, M. J. & Irvine, R. F. Nature 312, 315-321 (1984).
`& Weiss, S. Nature 317, 717-719 (1985).
`
`21. Sladeczek, F., Pin, J.P., Recasens, M., Bockaert,J.
`
`
`Caz+ influx through ion channels activated by NMDA triggers
`
`22. Schoffelmeer, A. M. N. & Mulder, A.H. J. Neurochem. 40, 615-621 (1983).
`
`
`
`
`release of Ca2+ from intracellular stores27, contributing further
`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).
`
`
`
`to the NMDA-evoked arsenazo III signals reported here.
`25. Mayer, M. L. & Westbrook, G. L. Soc. Neurosci. Abstr. 11, 785 (1985).
`
`
`
`
`Although the present results suggest a high Ca2+ permeability
`26. Ascher, P. & Nowak, L. J. Physiol., Lond. Proc. (in the press).
`
`
`of NMDA-receptor-activated channels (Fig. 3), the net flux of
`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).
`
`
`
`
`monovalent cations (that is, conductance) decreases in the pres­
`
`
`F. Nature 305, 719-721 (1983). 29. Lynch, G., Larson, J., Kelso, S., Barrinuevo, G. & Schottler,
`
`
`
`ence of Ca2+. This reflects interactions between permeant ions
`
`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,
`
`within the channel with Ca2+ acting as both a permeant ion and
`1978).
`flux 25·26·28•
`
`
`as a blocker of monovalent cation
`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).
`The experiments reported here provide evidence for an
`
`
`
`34. Edwards, C. Neuroscience 7, 1335-1366 (1982).
`
`
`
`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
`
`9. The latter experi­
`
`excitatory amino acids in frog motoneurones
`
`
`
`ments suggested an increase in both [Na+]; and [Ca2+]; during
`
`
`
`perfusion with L-glutamate but the results were difficult to
`as (1) neurones
`
`interpret clearly
`
`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
`Peter T. Jones, Paul H. Dear, Jefferson Foote,
`and (2) L-glutamate
`
`is a mixed agonist that acts
`vated channels,
`Michael S. Neuberger & Greg Winter
`
`
`
`
`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
`
`Hills Road, Cambridge CB2 2QH, UK
`
`
`
`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
`The variable domains of an antibody consist of a fl-sheet
`
`
`more permeable to Caz+ than those linked to KA receptors, has
`framework with hypervariable regions (or complementarity-deter­
`
`
`
`
`implications for the role of excitatory amino-acid receptors in
`mining regions--CDRs) which fashion the antigen-binding site.
`
`
`
`
`CNS function. It is possible that Caz+ influx activated by NMDA
`Here we attempted to determine whether the antigen-binding site
`
`
`
`
`
`receptors underlies the synaptic plasticity generating long-term
`could be transplanted from one framework to another by grafting
`
`
`
`
`potentiation, as the latter is prevented by intracellular injection
`the CD Rs. We substituted the CD Rs from the heavy-chain variable
`Ca2+ (ref. 29), or by blocking
`of EGTA to chelate
`NMDA
`region of mouse antibody Bl--8, which binds the hapten NP-cap
`(4-hydroxy-3-nitrophenacetyl caproic acid; KNP-cap = 1.2 µM), for
`
`
`
`receptors with selective antagonists30. For example, Ca2+ influx
`
`
`localized at transmitter-operated ion channels could have a role
`
`the corresponding CDRs of a human myeloma protein. We report
`
`
`
`
`in organizing and regulating postsynaptic structures in an
`that in combination with the Bl--8 mouse light chain, the new
`
`
`
`
`appropriate spatial relation to transmitter-releasing presynaptic
`antibody has acquired the hapten affinity of the Bl--8 antibody
`
`
`
`
`terminal boutons, and it is important to consider that Ca2+ influx
`
`
`
`
`occurring at NMDA receptors located on dendritic spines might
`of constructing human monoclonal antibodies from the correspond­
`
`
`
`produce an especially large but localized elevation in intracel­
`ing mouse monoclonal antibodies.
`
`
`lular Ca2+ concentration, due to restriction of Ca2+ diffusion
`The three-dimensional structures of several immunoglobulins
`
`
`
`
`along the narrow shaft of the spine. In addition, our results
`
`
`show that the variable domains consist of two /3-sheets pinned
`faces
`
`
`have some bearing on the mechanisms of desensitization of
`
`
`
`together by a disulphide bridge, with their hydrophobic
`
`NMDA receptors, as the link that has been demonstrated
`
`
`
`packed together1-3. The individual /3-strands are linked by loops
`
`which at one tip of the /3-sheet may fashion a binding pocket
`
`
`
`
`between [Caz+]; and desensitization ofnicotinic receptors at the
`
`neuromuscular junction31·32 may occur also for other receptor­
`
`
`
`for small haptens1·2. Sequence comparisons among heavy-and
`domains (V8 and Y1. respectively)
`
`
`
`ionophore complexes. Thus our results may help to explain the
`reveal
`
`light-chain variable
`
`
`similar desensitization evoked by either large doses of NMDA
`that each domain has three CDRs flanked by four relatively
`
`
`
`or depolarizing voltage jumps 7, which trigger Caz+ entry through
`
`
`
`
`conserved regions (framework regions-FRs)4• As seen in the
`
`NMDA channels and voltage-dependent calcium channels,
`
`
`structure of the human myeloma protein NEWM (Fig. 1), the
`respectively.
`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 /3-strands may help to fix the conforma­
`
`
`tion or orientation of the loops. The framework regions form
`for example in the V8 domain
`the bulk of the /3-sheet, although
`
`of NEWM, FRI includes part of the loop between the two
`/3-sheets and CDR2 not only forms a loop but a complete
`
`(Fig. 1). The structure
`of the /3-sheet framework is
`/3-strand
`
`
`
`
`similar in different antibodies, as the packing of different side
`
`chains is accommodated by slight shifts between the two /3-
`of Y1. and V8 FRs
`
`
`strands5. Furthermore, the packing together
`of Y1. with respect
`to V8
`
`is conserved6, therefore the orientation
`
`is fixed. We wondered whether the FRs represent a simple
`
`/3-sheet scaffold on which new binding sites may be built, and
`
`
`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).
`
`Replacing the complementarity­
`determining regions in a human
`antibody with those from a mouse
`
`Laboratory of Molecular Biology, Medical Research Council,
`
`(KNr-cap = 1.9 µM). Such 'CDR replacement' may offer a means
`
`
`
`© Nature Publishing Group1986
`
`1 of 4
`
`BI Exhibit 1033
`
`

`

`�NA�TU�R�E_v�o=L_.�32�1�29�M�A_Y_1_98_6
`
`���������
`
`- LETT ERS TQNATLJRE � ���� � � � �� � ������ �- s2_3
`
`Fig. 1 Stereo pairs of the VH (right)
`and VL (left) domains of the human
`myeloma protein NEWM1•8 gener­
`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,
`S 93, R 95; 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, P 43,
`y 86, F 99; VH v 37, Q 39, L 45, y 94,
`
`W 107) involved in packing VH and
`
`
`
`V L framework regions are traced. In
`the VH domain all these side chains
`
`are conserved in mouse Bl-8.
`
`FRI and N-terminal two-thirds of FR3 form the other {:!-sheet
`
`
`
`
`whether the structure of the CDRs (and antigen binding) is
`(Fig. le).
`which is exposed to solvent
`
`
`
`therefore independent of the FR context. To answer these ques­
`The gene encoding the HuVNP domain was constructed by
`
`
`
`
`tions experimentally, we have grafted the CDRs from one anti­
`
`
`
`gene synthesis (Fig. 2b ). We then constructed a plasmid, pSV­
`
`
`
`body to another, to determine whether antigen binding transfers
`Hu VNPHe, in which the HuVNP domain is linked to a humane
`with the CDRs.
`
`
`
`constant region, and cloned into a pSV2gpt-derived vector14.
`
`We grafted the CDRs from the VH domain of the mouse
`Bl-8 (ref. 7) into the VH domain of the
`
`The plasmid DNA was introduced into cells of the J558L mouse
`monoclonal antibody
`A 1 light chains
`
`
`myeloma by spheroplast fusion. J558L secretes
`
`
`human myeloma protein NEWM, whose crystallographic struc­
`
`which have been shown to associate with heavy chains contain­
`
`
`ture is known1•8• The VH domain of the CDR donor (Bl-8) is
`
`attached to a µ, constant
`
`
`ing a MVNP variable domain, to create a binding site for NP-cap
`
`region and associated with a mouse
`A 1 light chain, and the antibody
`
`
`or the related hapten NIP-cap (3-iodo-4-hydroxy-5-nitrophenyl­
`
`
`is directed against the hapten
`
`
`
`acetyl caproic acid)7• As the plasmid pSV-HuVNpHe contains
`NP-cap. Both the VH and VL domains seem to have a role in
`the gpt marker (encoding
`guanine phosphoribosyltransferase),
`
`
`determining the affinity of the antibody for NP-cap as the
`
`
`stably transfected myeloma cells could be selected in medium
`
`
`substitution of either domain by other, often highly related
`domains can destroy hapten binding (refs 7, 9 and
`
`
`
`
`containing mycophenolic acid14; transfectants would be ex­
`variable
`
`
`
`pected to secrete an antibody (HuVNp-IgE) with a heavy chain
`
`
`M.S.N., unpublished results). In the VH domain, each of the
`
`composed of a HuVNP variable domain and human e constant
`
`
`CDRs has been implicated in NP-cap binding10, but the class
`and the A 1 light chain of the J558L myeloma. The
`regions,
`
`
`of constant domains attached to VH does not seem to affect
`gpt+ clones were assayed by
`
`
`culture supematants of several
`
`binding of hapten11•12. The CD Rs from the VH domain of anti­
`
`
`
`radioimmunoassay and found to contain NIP-cap-binding anti­
`body Bl-8 (ref. 13) are longer than the CDRs which they replace
`
`
`body. The antibody secreted by one such clone was purified
`in NEWM4 and this may give rise to a deeper binding pocket.
`
`
`
`from the culture supernatant by affinity chromatography on
`
`
`Most of the residues conserved between the VH domains of
`
`
`NIP-cap-Sepharose, and by SDS-polyacrylamide gel electro­
`
`B 1-8 and NEWM are located in FR2, FR4 and the carboxy­
`
`
`phoresis the protein was indistinguishable from the mouse
`
`
`terminal third of FR3 (Fig. 2a) and largely form the region of
`
`
`chimaeric MVNp-IgE (ref. 12) (results not shown). The HuVNP­
`
`
`
`{:!-sheet which is packed against the light chain. Therefore, it
`
`
`
`IgE antibody competes effectively with MVNp-IgE for binding
`
`might be expected that the VH domain of Bl-8 (hereafter
`
`to both anti-human e (Fig. 3a) and NIP-cap coupled to bovine
`
`abbreviated to MVNP) and the hybrid Bl-8/NEWM domain
`serum albumin (NIP-BSA) (Fig. 3b).
`
`(HuVNP) would dock in a similar manner with the mouse VL
`
`
`domain to form the antigen-combining site6• The more variable
`
`
`The affinities of HuVNp-IgE for NP-cap and NIP-cap were
`
`
`
`© Nature Publishing Group1986
`
`2 of 4
`
`BI Exhibit 1033
`
`

`

`-5�� � � ��
`
`��
`
`� � � ���
`
`� � ��
`
`� LflTE
`
`� ���
`
`��
`
`N_A_TU_R_E� vo_L_. _J2_1_2_9 _M_A_Y�l�98�6
`
`a
`t£un
`8t-8
`
`FAI
`CllR'
`, __ - --
`- - -- 30
`35
`31HDYYT
`
`KUQLQESGPGLURPSQTLSL TCTUSGSTFS
`� QP_!!RE�K!GRSUK�S£KR�Y�T
`SYMnH
`FA2
`4;
`36- - ----
`t£un
`
`MURQPPGRGLEM I G
`81-8
`�KgR�
`
`FA3
`� CllA3 102
`66 - -
`- - - ---94
`t£un
`
`RUTMLUOTSKHQFSLRLSSUTRRDTRUYYCRR
`HLIAGCIOU
`··-·
`KR!L T!!!!KPSSTRYMQ� L,!.SEES�
`
`YDYYGSSYFOY
`FA4
`100-- --113
`.....
`MGQGSLUTUSS
`81-8
`�TTL�
`
`-7
`
`RS TQNATLJR£ � ���
`b
`I H;ndl 11 -48
`-23
`'. .•. . . . .
`ATiiCiWilccrcT-TCTACATOGTfWITATNJOTTTOTCTimi:
`............. Pltft 5lcrts ......._. NtA •tar ts
`CACAAACl\G
`C.ATGAGATCACAGTTCTCTCTACAOTTRCTGAGCRCRCAOGACCTC
`"" l!Ndr
`CllA2
`Spl ic:.
`l50
`6S
`'" 0 W S C I I L F L V ft
`YUFYHGTSDDTTPLRS
`T A rJ!
`RI DPHSGGTKYHEKFKS
`..::CRTGGGA TGGAGCTGTATCATCCTCTTCTTGGTAGCMCAGCT
`ACAGGTAAGGGGC
`TC
`ACllG TROC AOOC rTOIWJGTCTOOACATATATATOOGT GACAA TGACATCCACTTTGCCTT
`Spl ic•
`� P9tl
`10
`1
`
`lo U H S Q U 0 L Q ! E S G P G l U R
`,.
`TCTCTCCACAGG TGTCCllCTCCCAOO TCCAACTGCA GGAGAGCGGTCC
`AGG TCTTGTGAG
`, __ _
`3'
`2��-
`�
`30 CORI
`�
`�
`U � G S T F sCL:Li)
`P S Q T l S l T C r
`ACCTAOCCllOAC
`CCTOAGCCTGACCTQCACCOTOTCT
`OOCAOCACC
`TTCRQCAOCTACTO
`L-7-
`�
`-''-----3
`&------, r----8---
`- 4 ------,
`3'
`45
`40
`50 � 52R
`c:::c:::::Jj]w u . Q p p G . G l. E w I ol• I 0 Pl
`GATGCACT
`GGG TGAGACAGCCACCTGGAC
`GAGG TCTTGAGTGGATT GGAAGGA
`TTGATCC
`--7
`.___J �___...J l--9t>,---J L---11-
`12/ 14-
`-----, r---- 1oa---, ,...----- ICl!t----, r--
`�S COA2
`60
`&�
`70
`I rt s o o T K Y rt E K F K siR v T ,., L u o
`TMTRGTGGTGGTAC1'RAGT
`ACfllllTOAGAAG
`TTCAAOAOCRGAG
`TGRCARTGCTGGTRGA
`--11
`-----' '--t3G---' L.--
`13b� � 15-
`12/14
`16---
`r---
`�
`80
`82A 8 C
`85
`
`T S K H Q F S L R L S S V T R R D T A V
`CACCAGCAAGAAC
`CAOTTCAOCCTGAGACTCAGCAGCGT
`GACAGCC
`GCC GACACCOCGGT
`-tS-..J
`L--
`1Q-
`17
`..-----
`11-----�.-----ro---
`IDS
`� COA3
`IOOfl I C
`'lO
`y y c A Riv 0 y y G s s y F D vlw 0 Q 0
`CTRTTRTTOTOCfWOUftCO'tTTACT
`RCOO TAOTAOCTACTTTORCTACT
`OOOO TCfWJO
`--1Q---'
`'--
`21---J L---
`23-'
`L..-
`r-- 20a -
`..--------:22124
`!Splice
`11� u s s
`s L u
`CAGCCTCGTCACAOTCTCCTCllOGT. . .... IQ:Jbp.
`-25--GACA
`3'
`-, ,-211D -CTGTTCGA s·
`
`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 CORs 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 CORs 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-VNP (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 HuVNP fragment. The oligonu­
`
`
`cleotides are aligned with the corresponding portions of the HuYNr 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 HuYNr gene.
`as a Pstl-Hindlll
`Methods. The synthetic gene for HuVNP was constructed
`
`
`fragment. The nucleotide sequence was derived from the protein sequence
`using the computer program ANAL YSEQ26 with optimal codon usage taken
`made from the sequences of mouse constant-region genes. The oligonucleotides used in synthesis (l-26b28 in total) vary in size from 14 to 59 bases and were
`
`
`
`
`
`i
`
`
`
`
`
`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
`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
`
`blocks (A-0 and A'-0') containing oligonucleotides
`B'), 15 and 17 (block C), 16 and 18 (block C'), 19, 21, 23 and 25 (block 0), and 20, 22/24, 26a and 26b (block D'). In a typical assembly, for example of block
`
`
`kinase and mixed with 5 pmol of the terminal oligonucleotide A, 50 pmol of oligonucleotides 1, 3, 5 and 7 were phosphorylated at the 5' end with T4 pol�nucleotide
`
`with 5 µ.Ci of [ y-32P]ATP (Amersham; 3,000 Ci mmol- ). These oligonucleotides
`were annealed by heating to 80 °C and cooling
`which had been phosphorylated
`
`
`
`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, 10 mM MgC12• For the ligation, ATP
`(I 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 1 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-D and A'-0' using splint oligonucleotides; thus,
`
`
`A-D 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'-D' were constructed
`
`
`
`
`with oligonucleotides 7, 13b and 17 as splints). After phenol/ether extraction blocks A-D were annealed with blocks A'-0', small amounts were cloned in the
`vector Ml3 mpl8 (ref. 28) then cut with Psti and Hindlll,
`as a Hindlll-BamHI
`
`and the gene sequenced by the dideoxy technique29• The MVNP gene was transferred
`
`
`
`fragment from the vector pSV-YNr (ref. 12) to the vector Ml3mp8 (ref. 30). To facilitate the replacement of MVNP coding sequences by the synthetic HuVNP
`and a new Hindll site subsequently
`
`introduced near the
`
`fragment, three Hindll sites were removed from the 5' noncoding sequence by site.directed mutagenesis,
`end of FR4. By cutting the vector with Pstl and Hindll, most of the VNP coding sequence falls out and the synthetic
`as a Pstl-Hindll
`fragment could be introduced
`fragment. The sequence at the Hindll site was corrected to give NEWM FR4 by site-directed
`The Hindlll-BamHI
`mutagenesis.
`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
`into the myeloma line J558L by
`
`of human lgE (ref. 31) were introduced as a BamHI fragment to yield the vector pSV-HuYNrH .. which was transfected
`the Hindlll-BamHI
`fragment back into MJ3mp8 (ref. 30).
`
`spheroplast fusion. The sequence of the HuVNP gene in pSV-HuVNpHE was checked by re-cloning
`
`MVNp-lgE
`HuVNp-lgE
`
`Table 1 Affinity of HuVNp-lgE and MVNp-lgE for the haptens
`NP-cap and NIP-cap
`
`then measured directly using the fluorescence quench technique,
`
`
`
`
`and compared with those ofMVNp-lgE (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
`KNP-<:ap (µ.M)
`KN IP-<:ap (µ.M)
`
`
`
`HuVNp-lgE for both haptens is slightly lower than that of MVNP­
`0.02±0.01
`1.2±0.1
`
`
`
`lgE (2-3-fold, 0.3-0.6 kcal mol-1 ), the difference in affinity is
`O.o?±0.02
`1.9±0.2
`
`
`less than expected for loss of either a hydrogen bond or van
`
`
`der Waals' contact from the active site of an enzyme15•16• Thus,
`The affinity of HuVNp-lgE and MVNp-lgE for NP-cap was determined
`
`it seems that binding affinity and specificity for hapten can be
`at 295 nm and emission
`
`
`by ftuorescence quenching with excitation
`
`
`
`conferred on a human antibody by grafting in the CDRs from
`at 340 nm (ref. 22). Antibody solutions
`to 100 nM
`observed
`were diluted
`
`an appropriate mouse antibody.
`(0.45
`in phosphate-buffered saline, filtered
`
`
`µ.m-pore cellulose acetate)
`
`
`
`
`Is this result likely to be general? This would assume (1) that
`with NP-cap in the range 0.2-20
`and titrated
`
`µ.M. As a control, the
`
`
`mouse Dl-3 antibody23, which does not bind hapten, was titrated in
`
`
`
`antigen usually binds to the CD Rs, and any contacts to the FRs
`
`are made to the polypeptide backbone or to conserved side
`
`
`
`
`parallel. A decrease in the ratio of the ftuorescence of HuVNp-lgE or
`MVNp-lgE (as appropriate)
`
`to that of the Dl-3 antibody was taken as
`
`
`chains, and (2) that substitutions in the FRs do not usually
`
`
`
`being proportional to NP-cap occupancy of the antigen-binding sites.
`
`affect the conformation of the CDR loops. These assumptions
`The maximum quench was -40% for both HuVNp-lgE and MVNp-lgE,
`
`
`seem reasonable: thus, in the structure of a complex of the Dl-3
`
`
`
`and hapten dissociation constants were determined from least-squares
`
`
`
`
`antibody with lysozyme (R. A. Mariuzza, S. Phillips and R. J.
`
`
`
`fits of triplicate data sets to a hyperbola. The concentration of NIP-cap
`
`
`
`
`Poljak, personal communication) most contacts to the lysozyme
`was varied from 10 to 300 nM, and -50% quenching
`of ftuorescence
`
`are made by the CDRs, but there is also a hydrogen bond in
`
`
`
`
`was observed at saturation. As the antibody concentrations were compar­
`FRl of the V H domain from the /3-0H ofThr 30 (often conserved
`
`
`able to the values of the dissociation constants, data were fitted by
`
`
`least-squares to an equation24 describing tight binding inhibition.
`
`
`
`or replaced by Ser). Similarly, the conformation of CDR loops
`
`
`
`© Nature Publishing Group1986
`
`3 of 4
`
`BI Exhibit 1033
`
`

`

`100
`
`Anti-human £
`
`� \ \ \ \ \
`\,,...,,,_ ............
`
`\
`
`"u.. __
`
`"- --
`
`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
`50
`
`
`
`of radiolabelled MVNP-lgE to polyvinyl microtitre plates
`that had been coated with a, sheep anti-human
`e antiserum
`b, (NIP-cap)i4-BSA;
`c, Ac38 anti­
`(Seward Laboratory);
`d, Ac146 anti-idiotypic
`e,
`
`idiotypic antibody;
`antibody;
`
`
`
`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.
`
`50
`
`a
`
`c
`
`[lnhibitor]lµg
`ml-1 l
`
`(ed. Sayre, D.) 303-310 (Clarendon, Oxford,
`
`Regulation of human insulin
`gene expression in transgenic mice
`
`10. Reth, M., Bothwell, A. L. M. & Rajewsky, K. in Immunoglobulin
`between ,B-strands depends on loop size and specific interactions
`
`
`Jdiotypes and Their
`(eds Janeway, C., Wigzell, H. & Fox, C. F.) 169-178 (Academic, New York,
`Expression
`
`
`of the loop back to the ,B-sheet. However, in the same class of
`1981).
`11. Neuberger, M. S. & Rajewsky, K. Proc. natn. Acad. Sci. U.S.A. 78, 1138-1142 (1981).
`
`
`are usually variable domains (V H• V K or VA) these interactions
`12. Neuberger, M. S. et al. Nature 314, 268-270 (1985).
`
`conserved (ref. 5 and A. M. Lesk and C. Chothia,
`personal
`13. Bothwell, A. L. M. et al. Ce/124, 625-637 (1981).
`communication).
`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).
`
`
`
`While human monoclonal antibodies have therapeutic poten­
`A. J., Carter, P. & Winter, G. Biochemistry
`24, 5858-5861 (1985).
`16. Fersht, A. R., Wilkinson,
`
`
`tial in human disease, they can be difficult to prepare17 and
`17. Boyd, J.E., James, K. & McClelland, D. B. L. Trends Biotechnol.
`2, 70-77 (1984).
`R. M., Smith, L. M. & Dilman, R. 0. J. Immun. 135, 1530-1535
`18. Shawler, D. L., Bartholomew,
`
`
`
`
`treatment of patients with mouse monoclonal antibodies often
`(1985).
`
`
`
`
`increases the titre of circulating antibody against the mouse
`
`19. Morrison, S. L., Johnson, M. J., Herzenberg, L.A. & Oi, V. T. Proc. natn. Acad. Sci. U.S.A.
`81, 6851-6855 (1984).
`
`
`
`
`immunoglobulin18. As chimaeric antibodies containing human
`20. Boulianne, G. L., Hozumi, N. & Shulman, M. J. Nature 312, 643-646 (1984).
`
`
`constant domains12•19•20 and variable domains made by grafting
`
`T. & Rajewsky, K. Eur. J. lmmun. 9, 1004-1013 (1979).
`21. Reth, M., Imanishi-Kari,
`
`mouse CD Rs into human FRs, could have therapeutic potential,
`22. Eisen, H. N. Meth. med. Res. 10, 115-121 (1964).
`23. Mariuzza, R. A. et al. J. molec. Biol. 170, 1055-1058 (1983).
`
`
`we wondered whether the HuVNp-lgE antibody loses antigenic
`
`24. Segal, I. H. in Enzyme Kinetics, 73-74 (Wiley, New York, 1975).
`
`
`
`determinants associated with the MVNP variable region
`25 Jones, T. A. in Computational
`
`Crystallography
`
`
`(idiotopes). The binding of HuVNP-lgE and MVNp-lgE to both
`1982).
`26. Staden, R. Nucleic Acids Res. 12, 521-538 (1984).
`
`
`
`
`monoclonal and polyclonal anti-idiotypic antibodies directed
`27. Sanger, F. & Coulson, A. FEBS Lett. 87, 107-110 (1978).
`C., Vieira, J. & Messing, J. Gene 33, 103-119 (1985).
`
`against the MV NP domain was examined by using inhibition
`28. Yanisch-Perron,
`29. Sanger, F., Nicklen, S. & Coulson, A. R. Proc. natn. Acad. Sci. U.S.A. 74, 5463-5467 (1977).
`
`
`
`assays. As shown in Fig. 3d, the HuVNp-lgE antibody has lost
`30. Messing, J. & Vieira, J. Gene 19, 269-276 (1982).
`
`
`
`
`the MVNP idiotypic determinant recognized by antibody Ac146
`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).
`
`
`
`(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
`Richard F Selden*, Marek J. Skoskiewiczt,
`
`HuVNp-lgE antibody retains hapten binding but has lost
`Kathleen Burke Howie*, Paul S. Russellt
`
`
`
`
`idiotypic determinants, indicating that the immunoglobulin uses
`& Howard M. Goodman*
`
`
`
`different sites to bind hapten and anti-idiotypic antibodies. It
`
`
`appears, therefore, that both FR and CDR side chains form the
`Departments of *Molecular Biology and tSurgery, Massachusetts
`
`
`
`
`
`binding site for these anti-idiotopes, but mainly CDR side chains
`
`
`
`General Hospital, and Departments of *Genetics and tSurgery,
`
`interact with hapten.
`
`
`Harvard Medical School, Massachusetts General Hospital, Boston,
`
`
`
`We thank C. Milstein for suggesting this project, K. Rajewsky
`
`Massachusetts 02114, USA
`
`
`and M. Reth for the anti-idiotypic antibodies Ac38 and Ac146,
`and A. M. Lesk, C.
`
`
`Chothia, R. J. Leatherbarrow and C. Milstein
`Insulin is a polypeptide hormone of major physiological import­
`
`
`for helpful discussions. J.F. is a Fellow of the Jane Coffin Childs
`ance in the regulation of fuel homeostasis in animals (reviewed in
`Memorial Fund for Medical Research.
`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
`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
`most vertebrates, contain one proinsulin gene3•4, although several
`species, including mice5 and rats6•7, have two highly homologous
`insulin genes. We have studied the regulation of serum insulin
`
`Received 17 February; accepted 17 March 1986.
`
`1. Poljak, R. J. et al. Proc. natn. 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).
`J., Huber, R. & Palm, W. J. molec. Biol. 141, 369-391 (1980).
`3. Marquart, M., Oeisenhofer,
`4 Kabat, E. A, Wu, T. T., Bilofsky,
`M. & Perry, H. in Sequences of Proteins
`H., Reid-Miller,
`
`Interest (U.S. Department of Health and Human Services, 1983).
`of Immunological
`5. Lesk, A. M. & Chothia, C. J. molec. Biol. 160, 325-342 (1982).
`R. & Karplus, M. J. molec. Biol. 186, 651-663 (1985).
`6. Chothia, C., Novotny, J., Bruccoleri,
`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).
`
`
`
`© Nature Publishing Group1986
`
`4 of 4
`
`BI Exhibit 1033
`
`