WO 94/25591
`
`PCT/EP94/01442
`
`7
`
`Figure 2 Repertoire complexity and antigen binding capacity of camel IgG15 IgG2
`and IgG3 analysed by radioimmunoprecipitation (A) or Western blotting
`(B & C).
`Serum or purified IgG fractions from healthy or Trypanoma evansi
`(A)
`infected C. dromedarius (CATT titer 1/160 (7)) were incubated with labelled
`trypanosome lysate, recovered with Protein A Sepharose and analysed by SDS-
`PAGE. The relative counts recovered are inscribed below each lane. No
`trypanosome proteins bind to the Protein A or to the healthy camel
`immunoglobulins.
`20 pg of IgGj, IgG, and IgG3 from healthy and trypanosome
`(B)
`infected animals were separated by SDS-PAGE without prior reduction or heating.
`The electroblotted proteins were incubated with the labelled trypanosome lysate.
`The IgG2 shows a single antigen binding component corresponding to the heavy
`chain immunoglobulin whereas the IgG3 fraction appears to contain in addition two
`larger antigen binding components barely detectable by Ponceau Red staining (C).
`These are possibly Ig classes copurified as immunocomplexes present in the serum
`of the infected animals.
`
`METHODS. (35S)-methionine labelled Trypanosoma evansi lysate (500,000 counts)
`(22) was incubated (4°C, 1 hour) with 10 pl of serum or, 20 pg of IgG15 IgG2 or IgG3
`in 200 pl of 0.4 M NaCl, 10 mM EDTA, 10 mM Tris (pH 8.3), containing 0.1 M
`TLCK. 10 mg of Protein A SeDharose suspended in 200 pl of the same buffer was
`added (4°C, 1 hour). After washing and centrifugation, each pellet was resuspended
`in 75 pl SDS PAGE sample solution containing DTT, and heated for 3 min. at
`100°C. After centrifugation, 5 pl of the supernatant was saved for radioactivity
`counting and the remainder analysed by SDS PAGE and fluorography.
`The nitrocellullose filter of the Western blot of purified fractions IgG)5 IgG, apd
`IgG3 was stained with Ponceau Red (C) or incubated with 1% ovalbumin in TST
`buffer (Tris 10 mM, NaCl 150 mM, Tween 0,05%) (B). The membrane was
`extensively washed with TST buffer and incubated for 2 hours with (35S)-labelled
`trypanosome antigen. To avoid unspecific binding, the labelled trypanosome antigen
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 663 of 1663
`
`KELONIA EXHIBIT 1004 (Part 2 of 3)
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`8
`
`lysate was filtered (45 p) and incubated with healthy camel immunoglobulin and
`ovalbumin adsorbed on a nitrocellulose membrane.
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Figure 3 Amino acid sequences of the Vn framework, and hinge/Cn2 of Camelus
`dromedarius heavy chain immunoglobulins, compared to human (italic) VH
`framework (subgroup III) and hinges of human IgG (14).
`METHODS. Total RNA was isolated from a dromedary spleen (23). mRNA was
`purified with oligo T-paramagnetic beads (PolyATract-Promega). 1 pg mRNA was
`used for preparing double-strand cDNA (23) after an oligo-dT priming using
`enzymes provided by Boehringer Mannheim. 5 pg of cDNA was amplified by PCR
`in a 100 pl reaction mixture (WmM Tris-HCl pH 8.3, 50 mM KC1,15 mM MgCl2,
`0.01% (w/v) gelatine, 200 pM of each dNTP). 25 pmoles of each oligonucleotide of
`the mouse VH (24), containing a Xhol site, and 5'-CGCCATCAAGGTACCAGT-
`TGA-3' (see SEQ. ID. NO: 3) were used as primers. The 3' end primer was deduced
`from partial sequences corresponding to y chain amino acid 296 to 288
`(T.Atarhouch, C..Hamers-Casterman, G. Robinson, private communication) in which
`one mismatch was introduced to create a KDnl restriction site. After a round of
`denaturing annealing (94°C for 5 min. and 54°C for 5 min.), 2 U of Taq DNA
`polymerase were added, to the reaction mixture before subjecting it to 35 cycles of
`amplification (5). The PCR products were purified by phenol-chloroform extraction
`followed by HPLC (Genpak-fax column, Waters) and finally by MERMAID (BIO
`101, Inc.). After these purification steps, the amplified cDNA was digested with
`Xhol and Kpnl, and ligated into pBluescript.
`The clones were sequenced by the dideoxy chain termination method (25). The
`sequences were translated into amino acids which allowed their assignment to well
`defined domains of the Ig molecule (14); see SEO. ID. NO: 4-12
`
`Figure 4 Schematic representation of the structural organisation of the camel
`immunoglobulins (adapted from 26).
`On the basis of size consideration, the IgG! fraction possess probably the normal
`antibody assembly of two light and two heavy chains. IgG3 would have a hinge
`comparable in size to the human IgG|t IgG, and IgG^. The two antigen binding sites
`
`Page 664 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`9
`
`are much closer to each other as this camel IgG lacks the CB1 domain. In the camel
`IgG, the long hinge, being formed of Pro-X repeats (X = Glu, Gin or Lys), most
`likely adopt a rigid structure (19,20). This long hinge could therefore substitute the
`CH1 domain and bring the two antigen binding sites of IgG, to normal positions.
`— End of Draft publication —
`
`Background of the invention
`Already at a very early stage during evolution antibodies have been developed to
`protect the host organisms against invading molecules or organisms. Most likely one
`of the earliest forms of antibodies must have been developed in Agnatha. In these
`primitive fishes antibodies of the IgM type consisting of heavy and lights chains have
`been detected. Also in many other forms of life ranging from amphibians to
`mammals antibodies are characterized by the feature that they consist of two heavy
`and two light chains, although the heavy chains of the various classes of
`immunoglobulins are quite different. These heavy and light chains interact with each
`other by a number of different physical forces, but interactions between hydrophobic
`patches present on both the heavy and light chain are always important. The interac¬
`tion between heavy and light chains exposes the complementarity determining
`regions (CDRs) of both chains in such a way that the immunoglobulin can bind the
`antigen optimally. Although individual heavy or light chains have also the capability
`to bind antigens (Ward et al., Nature 341 (1989) 544-546 = ref. 5 of the above given
`draft publication) this binding is in general much less strong than that of combined
`heavy and light chains.
`Heavy and light chains are composed of constant and variable domains. In the
`organisms producing immunoglobulins in their natural state the constant domains
`are very important for a number of functions, but for many applications of
`antibodies in industrial processes and products their variable domains are sufficient.
`Consequently many methods have been described to produce antibody fragments.
`
`One of these methods is characterized by cleavage of the antibodies with proteolytic
`enzymes like papain and pepsin resulting in (a) antibody fragment comprising a light
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 665 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`10
`
`chain bound via an S-S bridge to part of a corresponding heavy chain formed by
`proteolytic cleavage of the heavy chain (Fab), or (b) a larger fragment of the
`antibody comprising two of these Fabs still connected to each other via an S-S
`bridge in enlargements of the heavy chain parts, indicated with F(ab)„ respectively
`(see patent applications EP-A-0125023 (GENENTECH / Cabilly et al., 1984) and
`WO-A-93/02198 (TECH. RES. CENT. FINLAND / Teeri et al., 1993) for defini¬
`tions of these abbreviations). The disadvantage of the enzymatic route is that the
`production of whole antibodies is expensive and the enzymatic processing increases
`the costs of these fragments even more. The high costs of antibody fragments block
`the application of these fragments in processes and products outside the
`pharmaceutical industry'.
`
`Another method is based on linkage on DNA level of the genes encoding (parts of)
`the heavy chain and the light chain. This linkage and the subsequent production of
`these chimeric immunoglobulins in microorganisms have been described (for Fab
`fragments see e.g, Better et al., Science 240 (1988) 1041-1043, for Fv fragments
`(combination of variable fragments of the heavy chain (VH) and light chain (VjJ still
`connected to each other by non-covalent binding interactions) see e.g. Skerra et aL,
`Science 240 (1988) 1938, and for single chain Fv fragments (ScFv; an Fv fragment in
`which the two variable fragments are linked to each other by a linker peptide) see
`e.g. Bird et al., Science 242 (1988) 423-426. Provided that an appropriate signal
`sequence has been placed in front of the single chain VH and VL antibody fragment
`(ScFv), these products are translocated in E. coli into the periplasmic space and can
`be isolated and activated using quite elaborate and costly procedures. Moreover the
`application of antibody fragments produced by E. coli in consumer products requires
`extensive purification processes to remove pyrogenic factors originating from E. coli.
`For this and other reasons the production of ScFv in microorganisms that are
`normally used in the fermentation industry, like prokaryotes as Streptomyces or
`Bacillus (see e.g. Wu et al. Bio/Technology 11 (1993) 71) or yeasts belonging to the
`genera Saccharomyces (Teeri et al., 1993, supra), Kluyveromyces, Hansenula, or Pichia
`or moulds belonging to the genera Aspergillus or Trichodenna is preferred. However
`with a very few exceptions the production of ScFv antibodies using these systems
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 666 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`11
`
`proved to be impossible or quite poor. Although the exact reasons for the poor
`production are not well known, the use of linkers between the Vn and VL chains not
`designed for secretion (Teeri et al., 1993, supra) may be a reason.
`
`Another reason may be incorrect folding of ScFv. The frameworks and to a limited
`extend the CDRs of variable domains of light and heavy chains interact with each
`other. It has been described by Chothia et al. (J. Mol. Biol. 186 (1985) 651-663 =
`ref. 13 of the above given draft publication) that this interaction involves amino
`acids at the following positions of the variable region of the heavy chain: 35, 37, 39,
`44-45, 47, 100-103 and 105 (numbering according to Kabat et al, In "Sequences of
`Proteins of Immunological Interest, Public Health Service, NIH, Washington DC,
`1983 = ref. 14 of the above given draft publication). Especially leucine at position
`45 is strongly conserved and the whole apolar side chain of this amino acid seems to
`be involved in the interaction with the light chain. These strong interactions may
`fold the ScFv into a structure that can not be translocated in certain types of lower
`eukaryotes.
`
`Thus the use of a linker in the production of ScFv for connecting a VH chain to a VL
`chain, might negatively influence either the translocation, or the folding of such ScFv
`or both.
`
`Not prior-published European patent application 92402326.0 filed 21.08.92 (C.
`Casterman & R. Hamers) discloses the isolation of new animal-derived immuno¬
`globulins devoid of light chains (also indicated as heavy chain immunoglobulins),
`which can especially originate from animals of the camelid family (Camelidae). This
`European patent specification, now publicly available as EP-A1-0 584 421, is incor¬
`porated herein by reference. These heavy chain immunoglobulins are characterized
`in that they comprise two heavy polypeptide chains sufficient for the formation of
`one or more complete antigen binding sites, whereby a complete antigen binding
`site means a site which will alone allow the recognition and complete binding of an
`antigen, which can be verified by any known method regarding the testing of the
`binding affinity. The European patent specification further discloses methods for
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 667 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`12
`
`isolating these heavy chain immunoglobulins from the serum of Camelidae and
`details of the chemical structure of these heavy chain immunoglobulins. It also indi¬
`cates that these heavy chain immunoglobulins and derivatives thereof can be made
`by using recombinant DNA technology in both prokaryotes and eukaryotes. The
`present invention relates to a further development of the work disclosed in that
`prior-filed but not prior-published European specification.
`
`Due to the absence of light chains in most of the immunoglobulins of Camelidae
`such linkers are not necessary, thereby avoiding the above-mentioned potential
`problems.
`As described above in the draft publication for Nature, now publicly available as
`Nature 363 (3 June 1993) 446-448, and in the not prior-published European patent
`application 92402326.0 (supra) it was surprisingly found that the majority of the
`protein A-binding immunoglobulins of Camelidae consists just of two heavy chains
`and that these heavy chains are quite different from common forms of heavy chains,
`as the CH1 domain is replaced by a long or short hinge (indicated for IgG2 and IgG3,
`respectively, in Figure 4 of the above given draft publication for Nature).
`Moreover these heavy chains have a number of other features that make them
`remarkably different from the heavy chains of common immunoglobulins.
`One of the most significant features is that they contain quite different amino acid
`residues at those positions involved in binding to the light chain, which amino acids
`are highly conserved in common immunoglobulins consisting of two heavy and two
`light chains (see Table 1 and SEO. ID. NO: 13-31).
`
`5
`
`10
`
`15
`
`20
`
`Page 668 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`Table 1
`Comparison af ainino acid sequences of various immunoglobulins
`Alignment of a number of V,, regions of Camel heavy chain antibodies compared
`with those of mouse (M, top line) and human (H, second line). Framework
`fragments are indicated in capitals, CDR fragments in small print; see SEQ. ID.
`NO: 13-31 for sequences indicated by M, H, 1, 2, 3, 7, 9, 11, 13, 16, 17, 18, 19, 20,
`21, 24, 25, 27, 29, respectively.
`
`m
`EVKLVESGGG LVQPGGSLRL SCATSGFTFS dfyme..WVR QPPGRRLEWI
`EVQLVESGGG LVQPGGSLRL SCAASGFTFS syams..WVR QAPGRGLEWV
`h
`GG SVQAGGSLRL SCAASGYSNC pltWS..WYR QFPGTEREFV
`caml
`cam2 DVQLVASGGG SVQAGGSLRL SCTASGDSFS rfams..WFR QAPGKECELV
`GG SVQTGGSLRL SCAVSGFSFS tscma..WFR QASGRQREGV
`cam3
`GG SVQGGGSLRL SCAISGYTYG sfcmg..WFR EGPGREREGI
`cam7
`GG SVQAGGSLTL SCVYTNDTGT ...mg..WFR QAPGRECERV
`cam9
`GG SVQAGGSLRL SCNVSGSPSS tyclg..WFR QAPGREREGV
`camll
`GG SVEAGGSLRL SCTASGYVSS ...ma..WFR QVPGQEREGV
`caml3
`GG SAQAGGSLRL SCAAHGIPLN gyyia..WFR QAPGRGREGV
`caml6
`GG SVQPGGSLTL SCTVSGATYS dysig..WIR QAPGRDREW
`caml7
`GG SVQAGGSLRL SCTGSGFPYS tfcig..WFR QAPGREREGV
`caml8
`GG SVQAGGSLRL SCAASDYTIT dycma..WFR QAPGRERELV
`caml9
`..GG SVQVGGSLRL SCVASTHTDS stcig..WFR QAPGREREGV
`cam20
`..GG SVQVGGSLRL SCRISGGTPD rvpkslaWFR QAPEREREGI
`cam21
`..GG SVQAGGSLRL SCNVSGSPSS tyclg..WFR QAPGREREGV
`cam24
`.-GG SVQTGGSLRL SCEISGLTFD dsdvg..WYR QAPGDECRLV
`cam25
`..GG SVQAGGSLRL SCASSSRYMP ctydmt.WYR QAPGREREFV
`cam27
`exxGG SVQAGGSLRL SCVASGFNFE tsrma..WYR QTPGNVCELV
`cam29
`
`100
`m
`A..asrnkan dytteysasv kgRFIVSRDT SQSILYLQMN ALRAEDTAIY
`h S..xisxktd ggxtyyadsv kgRFTISRDN SRNTLYLQMN SLRAEDTAVY
`caml S..smd...p dgntkytysv kgRFTMSRGS TEYTVFLQMD NLRPEDTAMY
`cam2 S..siq
`s ngrtteadsv qgRFTISRDN SRNTVYLQMN SLRPEDTAVY
`Aainsgggrt yyntyvaesv kgRFAISQDN ARTTVYLDMN NLTPEDTATY
`cam3
`A..tiln..g gtntyyadsv kgRFTISQDS TLRTMYLLMN NLRPEDTGTY
`cam7
`A..hit...p dgmtfidepv kgRFTISRDN AQRTLSLRMN SLRPEDTAVY
`cam9
`camll T..aint..d gsiiyaadsv kgRFTISQDT ARETVHLQMN NLQPEDTATY
`A..fvqt..a dnsalygdsv kgRFTISRDN ARNTLYLQMR NLQPDDTGVY
`caml3
`A..ting..g rdvtyyadsv tgRFTISRDS PKNTVYLQMN SLRPEDTAIY
`caml6
`A..aant..g atskfyvdfv kgRFTISQDN AKNTVYLQMS FLKPED.TAIY
`caml7
`A..gins..a ggntyyadav kgRFTISQGN AKNTVFLQMD NLKPEDTAIY
`caml8
`A.aiqvvrsd trltdyadsv kgRFTISQGN TRNTVNLQMN SLTPEDTAIY
`caml9
`cam20 A..siyf..g dggtnyrdsv kgRFTISQLN AQNTVYLQMN SLRPEDSAMY
`cam21 A..vlst..k dgktfyadsv kgRFTIFLDN DRTTFSLQLD RLNPEDTADY
`cam24 T..aint..d gsviyaadsv kgRFTISQDT ARKTVYLQMN NLQPEDTATY
`cam25 Sgilsdgtpy tksgdyaesv rgRVTISRDN ARNMIYLQMN DLRPEDTAMY
`cam27 S..sin...i dgkttyadsv kgRFTISQDS ARNTVYLQMN SLRPEDTAMY
`cam29 S..siy...s dgktyyvdrm kgRFTISREN AKNTLYLQLS GLKPEDTAMY
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`Page 669 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`14
`
`Table 1 (Cont.) Comparison al' amino acid sequences of various immunoglobulins
`Alignment of a number of VB regions of Camel heavy chain antibodies compared
`with those of mouse (M, top line) and human (H, second line). Framework
`fragments are indicated in capitals, CDR fragments in small print; see SEQ. ID.
`NO: 13-31 for sequences indicated by M, H, 1, 2, 3, 7, 9, 11, 13, 16, 17, 18, 19, 20,
`21, 24, 25, 27, 29, respectively.
`
`101
`139
`m
`y.. f
`dvWG AGTTVTVSS
`YCARdyygss
`h YCARxxxxxx xxxxxyyyyh x....fdyWG QGTLVTVSS
`caml YCKTalqpgg ycgygx
`clWG QGTQVTVSS
`cam2 YCGAvslmdr isqh
`gcRG QGTQVTVSL
`cam3 YCAAvpahlg pgaildlkky
`kyWG QGTQVTVSS
`dyWG QGTQVTVSS
`cam7 YCAAelsggs celpllf
`gqWG QGAQVTVSS
`cam9 YCAAdwkywt cgaqtggyf
`camll YCAArltemg acdarwatla trtfaynyWG QGTQVTVSS
`caml3 YCAAqkkdrt rwaeprew
`nnWG QGTQVTASS
`caml6 FCAAgsrfss pvgstsrles .sdy..nyWG QGIQVTASS
`caml7 YCAAadpsiy ysilxiey
`kyWG QGTQVTVSS
`caml8 YCAAdspcym ptmpappird sfgw..ddFG QGTQVTVSS
`nvWG QGTQVTVSS
`caml9 SCAAtssfyw ycttapy
`cam20 YCAIteiewy gcnlrttf
`trWG QGTQVTVSS
`cam21 YCAAnqlagg wyldpnywls vgay..aiWG QGTHVTVSS
`cam24 YCAArltemg acdarwatla trtfaynyWG RGTQVTVSS
`cam25 YCAVdgwtrk eggiglpwsv qcedgynyWG QGTQVTVSS
`dvWG QGTQVTVSS
`cam27 YCKIdsypch 11
`cam29 YCAPveypia dmcs
`ryGD PGTQVTVSS
`
`For example, according to Pessi et al. (1993) a subdomain portion of a VH region of
`common antibodies (containing both heavy chains and light chains) is sufficient to
`direct its folding, provided that a cognate VL moiety is present. Thus it might be
`expected from literature on the common antibodies that without VL chains proper
`folding of heavy chains cannot be achieved. A striking difference between the
`common antibodies and the Camelidae-derived heavy chain antibodies is, that the
`highly conserved apolar amino acid leucine (L) at place 45 present in commoh .
`antibodies is replaced in most of the Camelidae-dQmed heavy chain antibodies by
`the charged amino acid arginine (R), thereby preventing binding of the variable
`region of the heavy chain to that of the light chains.
`Another remarkable feature is that one of the CDRs of the heavy chains of this type
`of immunoglobulins from Canudidae, CDR3, is often much longer than the
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`Page 670 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`15
`
`corresponding CDR3 of common heavy chains. Besides the two conserved cysteines
`forming a disulphide bridge in common VH fragments, the Camelidae VH fragments
`often contain two additional cysteine residues, one of which often is present in
`CDR3.
`According to the present inventors these features indicate that CDR3 may play an
`important role in the binding of antigens by these heavy chain antibodies and can
`compensate for the absence of light chains (also containing CDRs) in binding of
`antigens by immunoglobulins in Camelidae.
`Thus, as the heavy chains of Camelidae do not have special features for interacting
`with corresponding light chains (which are absent), these heavy chains are very dif¬
`ferent from common heavy chains of immunoglobulins and seem intrinsically more
`suitable for secretion by prokaryotic and lower eukaryotic cells.
`
`The present inventors realized that these features make both intact heavy chain
`immunoglobulins of Camelidae and fragments thereof very attractive for their
`production by microorganisms. The same holds for derivatives thereof including
`functionalized fragments. In this specification the term "functionalized fragment" is
`used for indicating an antibody or fragment thereof to which one or more functional
`groups, including enzymes and other binding polypeptides, are attached resulting in
`fusion products of such antibody fragment with another biofunctional molecule.
`
`Summary of the invention
`In a broad sense the invention provides a process for the production of an antibody
`or a fragment or functionalized fragment thereof using a transformed lower
`eukaryotic host containing an expressible DNA sequence encoding the antibody or
`(functionalized) fragment thereof, wherein the antibody or (functionalized) fragment
`thereof is derived from a heavy chain immunoglobulin of Camelidae and is devoid of
`light chains, and wherein the lower eukaryotic host is a mould or a yeast. Thus the
`lower eukaryotic host can be a mould, e.g. belonging to the genera Aspergillus or
`Trichoderma, or a yeast, preferably belonging to the yeast genera Saccharomyces,
`Kluyveromcyes, Hansentda, or Pichia. Preferably the fragments still contain the whole
`variable domain of these heavy chains.
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 671 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`16
`
`The invention also provides methods to produce such heavy chain immunoglobulins
`or (functionalized) fragments thereof in which methods the framework or the CDRs
`of these heavy chains are modified by random or directed mutagenesis in such a way
`that the mutated heavy chain is optimized for secretion by the host microorganism
`into the fermentation medium.
`Another embodiment of the invention is that CDRs can be grafted on these
`optimized frameworks (compare grafting of CDRs on human immunoglobulins as
`described by e.g. Jones et al., Nature 321 (1986) 522). These CDRs can be obtained
`from common antibodies or they may originate from heavy chain immunoglobulins
`of Camelidae. The binding properties may be optimized by random or directed
`mutagenesis. Thus in a process according to the invention an antibody or
`(functionalized) fragment thereof derived from a heavy chain immunoglobulin of
`Camelidae can be produced which comprises a CDR different from the CDR
`belonging to the natural antibody ex Camelidae which is grafted on the framework
`of the variable domain of the heavy chain immunoglobulin ex Camelidae.
`The invention also provides a method for the microbiological production of catalytic
`antibodies. These antibodies are preferably raised in Camelidae against transition
`state molecules following procedures similar to the one described by Lerner et al..
`Science 252 (1991) 659-667. Using random or site-directed mutagenesis such
`catalytic antibodies or fragments thereof can be modified in such a way that the
`catalytic activity of these (functionalized) antibodies or fragments can be further
`improved.
`For preparing modified heavy chain antibodies a process according to the invention
`is provided, in which the DNA sequence encodes a modified heavy chain immuno¬
`globulin or a (functionalized) fragment thereof derived from Camelidae and being
`devoid of light chains, and is made by random or directed mutagenesis or both.
`Thus the resulting immunoglobulin or (functionalized) fragment thereof is modified
`such that
`-
`it is better adapted for production by the host cell, or
`it is optimized for secretion by the lower eukaryotic host into the fermentation
`-
`medium, or
`its binding properties (kon and kofr) are optimized, or
`
`-
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 672 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`17
`
`-
`-
`-
`
`its catalytic activity is improved, or
`it has acquired a metal chelating activity, or
`its physical stability is improved.
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Another particular embodiment of the present invention relates to genes encoding
`fusion proteins consisting of both a heavy chain immunoglobulin from Camelidae or
`part thereof and a second protein or another polypeptide, e.g. an enzyme, in parti¬
`cular an oxido-reductase, and to expression products of such genes. By means of the
`heavy chain immunoglobulin (fragment) the protein or enzyme can be guided to a
`target thereby increasing the local efficiency of the protein or enzyme significantly.
`Thus according to this embodiment of the invention a process is provided, in which
`the functionalized antibody or fragment thereof comprises a fusion protein of both a
`heavy chain immunoglobulin from Camelidae or a fragment thereof and another
`polypeptide, e.g. an enzyme, preferably an oxido-reductase.
`
`As a result of a process according to the invention known products may be
`produced, e.g. antibodies also produced by Camelidae, but many of the possible
`products will be new products, thus the invention also provides new products
`obtainable by a process according to the invention.
`The products so produced can be used in compositions for various applications.
`Therefore, the invention also relates to compositions containing a product produced
`by a process according to the invention. This holds for both old products and new
`products.
`
`Brief Description of the Figures
`Figures 1-4 were already described above in the draft publication.
`Characterisation and purification of camel IgG classes on Protein
`Figure 1
`A, Protein G and gel filtration.
`Repertoire complexity and antigen binding capacity of camel IgGj,
`IgG, and IgG, analysed by radioimmunoprecipitation (A) or
`Western blotting (B & C).
`
`Figure 2
`
`Page 673 of 1663
`
`

`

`WO 94/25591
`
`Figure 3
`
`5
`
`Figure 4
`
`Figure 5
`
`10
`
`Figure 6
`
`Figure 7
`Figure 8
`Figure 9
`Figure 10
`Figure 11
`Figure 12
`Figure 13
`Figure 14
`Figure 15
`Figure 16
`Figure 17
`Figure 18
`Figure 19
`
`Figure 20
`Figure 21
`
`15
`
`20
`
`25
`
`30
`
`PCT/EP94/01442
`
`18
`
`Amino acid sequences of the VH framework, and hinge/CqZ of
`Camelus divniedarius heavy chain immunoglobulins, compared to
`human (italic) Vj, framework (subgroup III) and hinges of human
`IgG (14); see SEQ. ID. NO: 4-12.
`Schematic representation of the structural organisation of the camel
`immunoglobulins (adapted from 26).
`DNA and amino acid sequences of the Camel VH fragments fol¬
`lowed by the Flag sequence as present in pB03 (Figure 5A), pB09
`(Figure 5B) and pB24 (Figure 5C); see SEQ. ID. NO: 32-37.
`Nucleotide sequence of synthetic DNA fragment cloned into
`pEMBL9 (Example 1); see SEQ. ID. NO: 38-41.
`Schematic drawing of plasmid pUR4423
`Schematic drawing of plasmid pUR4426
`Schematic drawing of plasmid pUR2778
`Schematic drawing of plasmid pUR4429
`Schematic drawing of plasmid pUR4430
`Schematic drawing of plasmid pUR4445
`Schematic drawing of plasmid pUR4446
`Schematic drawing of plasmid pUR4447
`Schematic drawing of plasmid pUR4451
`Schematic drawing of plasmid pUR4453
`Schematic drawings of plasmids pUR4437 and pUR4438
`Schematic drawings of plasmids pUR4439 and pUR4440
`Nucleotide sequence of synthetic DNA fragment cloned into
`pEMBL9 (Example 6); see SEQ. ID. NO: 42-45.
`Schematic drawing of plasmid pAW14B.
`Western blot analysis of culture medium of S. cerevisiae trans;
`formants containing pUR4423M (see A) or pUR4425M (see B).
`Samples were taken after 24 (see 1) or 48 hours (see 2). For
`pUR4425M two bands were found due to glycosylation of the
`antibody fragment.
`
`Page 674 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`19
`
`Detailed description of the invention
`The present invention relates io the production of antibodies or (functionalized)
`fragments thereof derived from heavy chain immunoglobulins of Camelidae by
`eukaryotes, more in particular by lower eukaryotes such as yeasts and fungi.
`Therefore, mRNA encoding immunoglobulins of Camelidae was isolated and
`transcribed into cDNA according to the procedures described in the above given
`draft publication and not prior-published European patent application 92402326.0.
`In each case primers for the PCR reaction directed to the N-terminus of the VH
`domain and PCR primers that either hybridize with the C-terminal regions of the
`Vn domain or with the short or large hinge regions as described in the above given
`draft publication, or with the C-terminal region of the Crl2 or CH3 domains can be
`used. In this way structural genes can be obtained encoding the following fragments
`of heavy chain immunoglobulins of Camelidae (Table 2).
`
`Table 2. The various forms of immunoglobulins of Camelidae that can be
`expressed in microorganisms.
`
`a.
`b.
`c.
`d.
`
`e.
`
`the variable domain of a heavy chain;
`the variable domain and the short hinge of a heavy chain;
`the variable domain and the long hinge of a heavy chain;
`the variable domain, the CH2 domain, and either the short or long hinge
`of a heavy chain;
`a complete heavy chain, including either the short or long hinge.
`
`According to procedures described in detail in the Examples these cDNAs can be
`integrated into expression vectors.
`Known expression vectors for Sacchammyces, Khiyveromcyes, Hansenula, Pichia and
`Aspergillus can be used for incorporating a cDNA or a recombinant DNA according
`to the invention. The resulting vectors contain the following sequences that are re¬
`quired for expression: (a) a constitutive, or preferably an inducible, promoter; (b) a
`leader or signal sequence; (c) one of the structural genes as described in Table 2
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 675 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`20
`
`and (d) a terminator. If the vector is an episomal vector, it preferably comprises an
`origin of replication as well as a selection marker, preferably a food grade selection
`marker, (EP-A-487159, UNILEVER / Leenhouts et al.). If the vector is an
`integration vector, then it preferably comprises sequences that ensure integration
`and a selection marker in addition to the sequences required for expression of the
`structural gene encoding a form of the heavy chain immunoglobulin of Camelidae or
`derivatives thereof. The preferred sequences for integration are sequences encoding
`ribosomal DNA (WO 91/00920, 1991. UNILEVER / Giuseppin et al.) whereas the
`selection marker will be preferably a food grade marker.
`For Saccharomyces the preferred inducible promoter is the GAL7 promoter (EP-A-
`0255153, UNILEVER / Fellinger et al.); for Kluyveromyces the preferred inducible
`promoter is the inulinase promoter (not yet published EP application 92203932.6,
`UNILEVER / Toschka & Verbakel, which is incorporated herein by reference); for
`Hansenula or Pichia the preferred inducible promoter is the methanol-oxidase
`promoter (Sierkstra et al., Current Genetics 19 (1991) 81-87) and for Aspergillus the
`preferred inducible promoter is the endo-xylanase promoter (not prior-published
`PCT application PCT/EP 92/02896, UNILEVER / Gouka et al., now publicly
`available as WO-A-93/12237, which is incorporated herein by reference).
`To achieve efficient secretion of the heavy chain immunoglobulin or parts thereof
`the leader (secretion) sequences of the following proteins are preferred: invertase
`and a-factor for Saccharomyces, inulinase for Kluyveromyces, invertase for Hansenula
`or Pichia (Sierkstra et al., 1991 supra) and either glucoamylase or xylanase for
`Aspergillus (not prior-published PCT application WO-A-93/12237, supra). As food¬
`grade selection markers, genes encoding anabolic functions like the leucine2 and
`tryptophan3 are preferred (Giuseppin et al. 1991, supra). The present invention
`describes the heterologous production of (functionalized) derivatives or fragments of
`immunoglobulins in a microorganism, which immunoglobulins in nature occur .not as
`a composite of heavy chains and light chains, but only as a composite of heavy
`chains. Although the secretion mechanism of mammals and microorganisms is quite
`similar, in details there are differences that are important for developing industrial
`processes.
`
`5
`
`10
`
`15
`
`20
`
`25
`
`30
`
`Page 676 of 1663
`
`

`

`WO 94/25591
`
`PCT/EP94/01442
`
`21
`
`To obtain frameworks of the heavy chain immunoglobulins, that are optimally
`secreted by lower eukaryotes, genes encoding several different heavy chains can be
`cloned into the coat protein of bacteriophages and subsequently the frameworks of
`these heavy chain immunoglobulins can be mutated using known PCR technology,
`e.g. Zhou et al., (1991). Subsequently the mutated genes can be been cloned in
`Saccharomyces and Aspergillus and the secretion of the mutated genes can be
`compared with the wild type genes. In this way frameworks optimized for secretion
`may be selected.
`Alternatively these structural genes can be linked to the cell wall anchoring part of
`cell wall proteins, preferably GPl-linked cell wall proteins of lower eukaryotes,
`which result in the expression of a chimeric protein on the cell wall of these lower
`eukaryotes (not prior-published EP application 92202080.5, UNILEVER / Klis et
`al., now publicly available as International (PCT) patent application WO-A-
`94/01567, which is incorporated herein by reference).
`Both methods have the advantage that the binding parts of the immunoglobulins are
`well exposed to the surrounding of the cell, microorganism, or phage and therefore
`can bind antigens optimally. By changing the external conditions the binding rates
`and dissociation rates of this binding reaction can be influenced. Therefore, these
`systems are very suitable to select for mutated immunoglobulins that have different
`binding properties. The mutation of the immunoglobulins can either be obtained by
`random mutagenesis, or directed mutagenesis based on extensive molecular
`modelling and molecul