Вы находитесь на странице: 1из 8

395

Recombinant antibody fragments


Peter J Hudson
Recombinant antibodies and their fragments now represent culture rather than from a live mouse. One significant
over 30% of all biological proteins undergoing clinical trials, advantage of thin new technology is the isolation of anti-
which recently culminated in FDA approval for the first bodies with new binding spccificities against hitherto
engineered cancer therapeutic antibody. Other successful rcfractorv antigens, thereby avoiding the limitations inher-
applications include both the effective enhancement of the ent in the mammalian immune response. Antibodies are
human immune response for cancer therapy and the reduction still the paradigm for the design of high-affinit> protein-
of unwanted immune rejections following transplantation and based binding rcagents [4"',9",10,11]. In the past year,
antibody therapy. Important advances have been made in the antibodies have bccn rcduccd in size, dissected into min-
methods for design, selection and production of these new imal binding fragments, and rebuilt into multivalcnt
types of engineered antibodies. Innovative selection methods high-axiditx/ reagents [9",12,13"]. Antibody fragments
have enabled the isolation of catalytic antibodies and high- have also bccn flised with a range of molecules limited
affinity viral-specific antibodies, the latter capable of only by the imagination, including enzymes for prodrug
redirecting viruses for gene therapy applications. In numerous therapy, toxins for cancer targeting, viruses for gene ther-
practical applications, recombinant antibody fragments have apy, lipids for improved systemic delixery and biosensors
been fused to radioisotopes, drugs, toxins, enzymes and for rcal-timc detection of target nmlecules. Clinical diag-
biosensor surfaces. nostic applications of antibody fragments includc thc full
range of i# vitro immunoassavs (30% of thc 10 billion pa
Addresses diagnostic industry) through to iH vivo tt, mour and clot
CRC for Diagnostic Technologies and CSIRO Molecular Science, imaging reagents. This revicw dcscribcs many of the
343 Royal Parade, Parkville,Victoria, Australia 3052; publications in the past 12 months and highlights devel-
e-mail: peter.hudson@molsci.csiro.au opments in the design, production and clinical applica-
URL: http://www.crc.sci.qut.edu.au and http:\\www.molsci.csiro.au
tions of rccombinant antibody fragments
CurrentOpinionin Bioteehnology1998, 9:395-402
http://biomednet.com/elecref/0958166900900395 Design of ' a n t i b o d y f r a g m e n t s ' and
i m p r o v e m e n t s in t h e 'antigen binding surface'
:c..Current Biology Publications ISSN 0958-1669 T h e Fv module comprises aligned antibody \:tt and VI.
Abbreviations domains and in the smallest fragment that retains the full
CDR complementarity-determining region monovatent binding affinity of the intact parent antibody
Fc constantregion fragment (Figure 1). Fvs cannot be easily produced by proteolysis
Fv variable region fragment of intact immunoglobulinn (Igs) and so recombinant sin-
Ig immunoglobulin
scFv single chain Fv gle-chain Fv (scFv) molecules, in which the Vll and VI.
V variable region domains arc tcthcred together with a polypeptide linker,
VH heavy chain variable region arc now uscd routinely to mimic Fv fragn)ents. T h e
VL light chain variable region polypeptide linker must span at least 35 A (3.5 nm)
between the carbox\ terminal of one variable region (\7)
Introduction domain to the amino terminal of the other V domain
A new wave of antibody fragments arc abot, t to enter the ~ithout compromising the fidclity of the V I t--\7l~ pairing
clinic, both for diagnosis and for therapy, and the term and antigcn-binding site. T h e 15 residue hydrophilic
"cnginccrcd antibodies' will be embraced by the general s c q u c n c c (Glv4Ser).~ is usuall\ chosen for the linker [13"1,
public for the first time [l°-3°]. Because it takes a decade but the length can xar\' between 10 and 25 amino acid
to dcvclop a new thcrapeutic reagent into a viable com- residues or be optimised b \ phagc-display [14]. T h e
mcrcial product, many of these new antibodies are con- amino-termini of both VII and VI. tire near to, but not part
sidcrcd primitive designs that were conceived 10 ycars oil the protein-binding interface, so there in little to
ago. Soinc are based on fusion of murine variable region choose between the t~vo possible orientations, V li-Vl, or
fragments (Fvs; complcx of hcaxv chain variable region ViTVtl. ScFxn gcncrally rctain the same affinity an the
(Vii) and light chain variable region (VI) domains) onto parent Fv module and express to higher lcxcls than Fabs
human frameworks [2"]; others are cleverly designed in bacteria. Both ncFxn and Fabs can be expressed to over
rcsurFacing of murinc antibodies, for example, by grafting 1 g per litre using fcrmcntcrn [1°], and usually incorpo-
murine C I ) R loops onto a human antibody framcwork rate a carboxv-terminal polypeptide tail for affinity
[3",4"',5]. Technology for antibody design has taken purification [13"]. ScFv and Fab fragments provide cffec-
cnorn~o/is strides forward and new library display and tixc and highly spccific iJl vivo targeting reagents [10]
selection procedures have made classic hvbridoma tech- and, unless conjugated to constant rcgion fragment (Fc)
nology obsolete [6"°,7,8]. It in now possible to select domains, will not recruit the c o m p l e m e n t cascade and
hi~h-afTinitv antibody fraemcnts directly from a viral related cvtotoxic functions.
396 Protein engineering

Current dogma states that the antigen-binding surface is expect [18]. T h e s e reports shiny that u c arc still primitive
comprised of six C D R loops and, therefore, most of the in our ability to design specific interface mutations that
designed mutations for improving; affinity ha~e been tar- will restllt in affinity cnhancenlent, even using relatively
gctcd to these (]DR loops or the adjacent underlying high-resohition structural analysis of both the free and
rcsidues [10,15,16,17"]. Some of these mutants ha~e spec- antigen-bound antibody. Ftirthernlore, despite the surprise
tacular affinity enhancements, cvcn cnabling the produc- finding that camelids produce antibodies that display high-
tion of an efficient catalytic antibody capable of affinity Vlt domains rather than the F'x module, and which
prcforming new aldol reactions [17°]. Perhaps the only dis- therefore bind to tbeir target antigens using just three
appointmcnt has been that desig;ned interface mutations ( ; D R loops [19], most attempts to produce a similar mini-
[11] have been less effective than random mutation and rnal binding domain have been unsuccessfiil [12]. Perhaps
sclcction using; phage display [10,15,16,17"]. Part of thc the large loop size in camelid V domains, stabilised b \ an
problem is that we do not yet know how to design com- intra-loop distllfide bond, is a critical component in pro-
plementary surfaces precisel> T h e effect of solvating viding a sufficiently large antig;cn-binding surface [191.
buried water molecules is still in debate and it is uncertain
whether solvation of buried water molecules increases or Stabilised Fv molecules and improved
decreases entropy A rigorous analysis of antibody surfaces expression levels
shows that interfaces are complementary in electrostatic T h e crystal structure of a higb-affinity somatically n~aturcd
potential, but not in charge complementarit% as one would murinc Ig (solved as a crystallised Fab fragment), whcn

Figure 1

VH L VL
I I

Fab Fv scFv

VH VL
VH L VL I

Triabody = scFv trimer


Diabody = scFv dimer

TBR T8R

LI )L3

V H Lt V L L2 VH L3 VL
VH L1 VL L2
I I I

Two scFvs joined by linker L2 Two scFvs joined by adhesive L2 domain

Current Opinion in Biotechnology

Schematic representation of an intact Ig together with Fab and Fv fragments and the recombinant forms for scFv monomers, dimers (diabodies),
trimers (triabodies) and two other types of scFv dimers. Shown are both the gene construct and a hypothetical V-domain association for each
arrangement of V H and V L domains joined by a polypeptide linker (L). The target binding region (TBR) represents the antigen or hapten binding
sites in the Fv modules. V-domains are represented by ovals (V H is shaded grey with an amino-terminal dot) and linkers are represented as a black
line. The linker L2 can be either an adhesive polypeptide or protein domain.
RecombinantantibodyfragmentsHudson 397

compared to the low-affinity parent germline Ig, showed currently sitting idly in lab freezers, are predicted to have
that the key mutations responsible for the affinity enhance- significantly improved therapeutic value when engineered
ment were in the VIt-\:[, interface [20"]. Significant into flexible structures capable of cross-linking receptor
changes in the conformation of the binding site occur upon molecules [34]; however, cross-linked Igs are probably too
binding of hapten to the germline antibod>; whereas hap- large (300 kDa) for effective turnout penetration. For these
ten binds to the mature antibody by a lock-and-key fit reasons (high a~iditv and effective target cross-linking),
mechanism. T h e effect of stabilised and optimally aligned there have bccn many attempts to conjugate Fab or scFv
V domains was to reduce entropy loss upon antigen binding molecules into dimers or higher multimers (Figure 1) to
and significantly enhance affinity 30,000-fold [20°]. This produce high-avidity reagents of t'~0-120 kDa in size, capa-
observation extends the concept of the original 'interf:ace ble of rapid tumour penetration without fast kidney clear-
adaptor hypothesis' [21], that is, V domains are designed to ante. T h e s e designs have included chemical cross-linking
function as semi-flexible adaptors, allowing antigen to strategies and a variety of recombinant fl, sions using adhe-
l-aould slightly different C D R surfaces by realignment of sive protein domains or peptides [9°]. Perhaps the simplest
the two domains. Somatic mutation iH ~i~,o can then sta- design is a diabody [13"], in which a short 5-residue linker
bilise the conformation with optimal hapten complemen- b e t w e e n V domains prevents formation of aligned
tarity. T h e effect of these observations is to expand the V domains into a functional Fv module and instead results
problem of designing targeted mutations for affinity in combination of the V domains of two scFv molecules
enhancement because the sequence search now inch]des into a bivalent dimer. Recently, a surprising resuh showed
mutations in the V-domain interfaces. Indeed, a number of that reduction of linkers to less than three residues pre-
recent studies have identified key residues in both the vents the formation of a diabodv and instead directs three
interface and framework regions that improve folding and scFv moleculcs to associate into a trimer (triabody; see
stability and, thcrcb> enhance expression levels [22-26]. Figure 1) with three flmctional antigen binding sites
For example, Carter and co-workers [4"°,25] designed com- [13",35,36]. T h e gain in functional affinity through multi-
plementary surfaces to enhance \'-domain association and valent binding (often termed avidity; [9"]) makes trimeric
described these as 'knobs-into-holes' engineering. Fv frag- scFvs particularly attractive for iH ~,i~,o tumour imaging as
ments can also be stabilised by chemical cross-linking or an altemative reagent to diabodies [37 °] and multivalent
engineering of disulfide bonds between chains [26,27]. "Ib chemical conjugates [38].
select for optimal V-domain aligmnents and Fv stabilit>
novel phage selection me-tffods were developed [8,25], Of course, multiple binding to surface-bound antigens is
including a strategy of Fv fragment stabilisation in the pres- dependent on correct alignment and orientation in the Fv
ence of antigen [28]. In this latter method, VII and VI. modules of diabodies and triabodies, otherwise apparent
domains are displayed separately followed by an antigen gains in functional affinity are likely to be small. Diabodies
selection strategx/ that favours high-affinity Fv molecules and triabodies have been shown to be relatively flexible
with strong \:tt-VI, pairing. molecules from the orientation of antigen-binding sites
revealed in single-molecule EM images [39]. T h e effect of
Intracellular antibodies fl)r gene or protein-targeted thera- manipulating linker length, sequence and structure on dia-
pies have now been designed without intradomain disul- body and ttiabody stability and flexibility can now be
fides to improve the folding efficiency in the reducing analysed using single-molecule imaging. T h e ability of tri-
environment of the cytoplasm [29-31]. abodies to cross-link surface receptors is unknown, and
will obviously depend on flexibility between the F~ mod-
Engineered valency and flexibility in antibody ules and the orientation of the antigen-binding sites, as
fragments well as the structure of the receptor.
Intact antibodies are polyvalent molecules (Figure 1) with
generally enhanced functional affinity (avidity) relative to Cancer imaging studies using multivalent
mono~ alent Fab fragments, particularly in the case of pen- antibody fragments
tamcric IgM molecules. T h e r e is, therefore, a similar I;; ~'i~,o cancer imaging data has shown that radiolabelled
increase in functional affinity when Fab or scFv fragments diabodies (60 kDa), minibodies (dimers of s c F v - C H 3
arc complcxed into dimers, trimers or larger aggregates fusions: 90 kDa) and dimerised Fabs (120 kDa) all exhibit
[9"]. Flexibility between antigen-binding sites is impor- improved tun3our targeting, as measured by tumour-to-
tant; in intact lgs the 'elbow' angle within Fab arms and blood ratios, compared to monomeric scFv (30 kDa), Fab
'hingc' angles between arms allow significant flexibility, (60 kDa) and parcnt bivalent Ig (150 kI)a) [10,27,37°,38].
and numerous cross-linking geometries [32]. Despite this Monovalent antibody fragments (Fab and scFv) clear
flcxibilitx; intact Igs are often not able to cross-link adja- rapidly from the blood due to their small size and have rcl-
cent receptors on the same viral or cell surface [33,34]. ativcly high off rates due to the single binding site [10,37°].
Recently, cross-linked homodimers of whole lgs have been On the other hand, whole Igs appear to be too large for
sho~ n to cross-link adjacent receptors on the same cell sur- effective tumour penetration [37°]. Two recent studies
face, and thereby activate intracellular signalling and apop- [10,37 °] concur that the higher flmctional affinity (avidity)
tosis [34]. Manv cancer cell specific Igs, which are of multimers combined ~ith the reduced kidney clearance
398 Proteinengineering

rates when muhimers wcrc 60-90 kDa in size wcrc partic- the anti-bispecific response in the host is minimised duc to
ularly suited to tumour targeting. the small diabody size and, again, lack of Fc domains.

Clinical trials arc in progress to evaluate the in c'i~,o stabil- Therapeutic and diagnostic applications of
it3. and radiotherapeutic potential of these diabodies and antibody fragment fusions
minibodies. Folh)wing this trend, we would expect that rl'hc original 'Magic Bullet' concept is still alive: a number
radiolabellcd triabndies (90 kI)a) would outperform dia- of cell-specific antibodies are being cvaluatcd as delivery
bodies due to higher avidity and reduced kidney clearancc. agents to dircct a cvtotoxic con]ponent to a ttlnlotlr site.
Improvements in tumour labelling and therapy \sere T h e original principals ha~c now been significantly
reported for chemically conjugated Fabs (120 kI)a) com- refined; exquisitely sensitive laumaniscd cancer antibod\
pared to the parent Ig and were attributed to the increase fragments can be fused to Fc domains to initiate cvtotoxic
in stability of thc maleimide cross-linkers and improved effector functions, prmiding thc connecting hinge to l"c
flexibility at the hinge region, alhming better cross-linking retains thc disufide linkage and flcxibilit\ requircd for
of adjacent receptors on the same cell surface [38]. T h e effcctixc clustering and presentation of heax\ r chain con-
i m p r m e d ftmctional affinit% tumour penetration and stant ((]H2) domains [46,47]. New types of immunotoxins
biodistribution of these novel cross-linked antibody frag- haxe been produccd 13\ recombinant or chemical fusion of
ments is promising fi)r the development of a n c w genera- antibody fragments to cytotoxic drugs, pcptidcs and pro-
tion of antibody fragments for imaging and therapx. tcins [48,49",50-57]. I n n m a t i v e improvements to thc
imnaunotoxin concept [48,49",50] include A 1 ) E P T - - thc
Engineering multiple specificity in antibody concept of antibody-enzyme fusions that specifically acti-
fragments vate a prodrug into a cytotoxic agent at the cell target site
Cross-linking reagents have bccn a major fi)cus in clinical [51,52]. Refined designs include fusions to intcrlcukin 2
therapy, particularly for rccrnitmcnt of cytotoxic T cells for for cvtokinc stimulation and T cell proliferation at the tar-
canccr trcatment or ancrgy of T -cclls fi)r transplant thera- get turnout site [53], fusions to T-cell costimulator\ molc-
py [1",4"']. T h e r e are also numcrous diagnostic applica- cults for "Fccll modulation (either actixation or ancrgy),
tions, as many immunoassays require cross-linking fusions to transmembrane cell-signalling domains as 'T-
rcagcnts [40,41]. Bispecific antibodies can bc produccd by bodies' [54] and fusions to a viral surface markcrs for viral-
fusion of hvbridomas into quadroma cell lines but this targctcd genc then~py [55-57]. In gcnc thcrap.x, the
tcchniquc is complex, time consuming and often incffcc- dclivered g e n t may encode a cytotoxic protcin or a surfacc
tixe. Far simpler strategies incorporate chemical conjuga- antibody (fusion) capablc of rccruiting complen~cnt or
tion or recombinant expression systems to couplc two T cells [54]. Antibody fragments ha~c been fused to lipids
different Fab or F~ modules together (Figure 1) [9"]. [58] and alkaline phosphatase [59] for diagnostic applica-
Rcccntly, superb mammalian cell expression systems ha~c tions. T h e antibody-fusion proteins provide vcrx high sen-
been dcveh)ped for the production of bispecific 'minibod- sitivity in these diagnostic assays by enabling signal
ics', using designed heavv chain CH3-constant domain amplification at'tcr the initial antibody binding interaction.
intcrfaccs that forcc efficient hctcrodimcrisation of two T h c ultimate diagnostic device will be a recombinant anti-
different Fab-CH3 molcculcs [4",42]. b e d \ fused directly to a bioscnsor platform capablc of real-
time detection of target antigens. T h e current research
Bispccific diabodics arc simple recombinant constructs biosensor (BIAcoreTM: Pharmacia) dctccts molccular mass
comprising tx~o scFv molecules that are forced to combine increase on a surl'ace and, although somewhat limited in
together because each contains a short (five rcsiduc) link- sensitixit3, has bccn cspccially useful for rcal-timc moni-
cr joining \:ll and VI. domains that prevents monomeric toring of a n t i g e n - a n t i b o d y interactions [9,13°].
scFx fi)nnation [43,44]. In one example [43], bispecific dia- Recombinant designs for antibody l'usions arc rcall\ onl\
bodies wcrc designed to cross-link colon cancer cells to limited by thc imagination and an expression svstcna capa-
serum lg, thereby inducing the complement cascade, ble of producing c()rrectl\ folded protein products.
including mononuclear phagocyte respirator\ burst and
phagocytosis, and directing synergistic T-cell cytotoxicity. Antibody fragment libraries: construction,
Furthcrmorc, b \ virtue of binding to serum lg their half- display and selection
life (beta-phase) was incrcascd fivefold compared to a con- Bacteriophage libraries
trol diabody of the same molecular weight. In a second Phage display technology is currentl\ the most popular in
example [441, bispccific diabodics with specificity to C lq c'/tro method for the selection of hi~h-affinity antibody
recruited thc complement cascade directly. In a third fragments (JK Scott and co-~orkcrs, see this issuc
example [45], a bispecific scFv dimcr was produccd that pp 427-435: also [6°',7,8]). T h e current fd phagc and
can recruit T cells (via CI)3) to cancer cells bearing the phagcmid ~cctors cnablc coupling of affinit\ selection
I tER2 antigcn. Bispccific diabodics and scFv dimers haxe (based on the displayed rcpcrtoires of antibody Fab and
two advantages over bispecific antibodies: firstly: by lack- scFv fragmcnts) to thc recovery of the packaged g e n t
ing Fc domains, bispecific diabodies will only activate encoding that antibody. This system is highly cffcctixc
T cells ~ h e n cr()ss-linkcd to tarect (cancer) cells: secondly, and has been used cxtcnsixclv to isolate human Fab and
RecombinantantibodyfragmentsHudson 399

scFv fragments against a wide rangc of proteins, D N A , T h e two domain structure of the Fv module enables chain
cell-surfacc markers, viruses and parasites. H u m a n anti- shuffling to be used both for sequential i m p r o v e m e n t s in
body gencs can be extracted from p r e - i m m u n i s e d donors affinity and for hunlanisation [631. In chain shuffling, just
and have alreadx' b e e n somatically m a t u r e d to high affin- one V domain is altcrcd at a time, relying on promiscuous
ity (Kd >10 -7 M). An alternative strategy is to sclcct anti- V domain paring to crcatc ncx~ flmctional Fv modules. For
b o d y fragments from very largc 'nafve" repertoires that cxample, starting with an antibody of known target affini-
have b e e n e i t h e r c o n s t r u c t e d from germline V domains or ty, only onc V domain is displayed as a library while the
s y n t h e s i s e d with large n u m b e r s of mutations, typically in other V domain is hcld constant to provide a defined target
C I ) R loops. D c s p i t c these successes, f\l vectors impose a affinity. R e p e a t e d cycles of chain shuffling and affinity
series of major limitations, including gene deletion, plas- selection will produce high-affinity antibodies. Using this
mid instability, gross phagc cross-contaminations and lim- "guidcd selection" or ' e p i t o p e imprinting' strateg 5 it has
itcd library size. bccn possible to sequentially replace the \; domains of a
parcnt mousc antibody to form a fully human Fv with
Polysome libraries high-affinity to thc samc c p i t o p c [63].
Polvsomes arc stable p r o t e i n - r i b o s o m e - m R N A c o m p l e x -
es that can bc uscd to replace live b a c t e r i o p h a g e as the New affinity maturation technologies
d i s p l a y v e h i c l e for r e c o m b i n a n t a n t i b o d y f r a g m e n t s T h e f l m d a m c n t a l problcm in library mutagenesis strate-
[60"]. T h e p o l y s o m e s arc formed by p r e v e n t i n g release gies is the librar\: size required to search the c o m p l e t e set
of newly sx'nthesised and corrcctly fnldcd protein from of residue mutations (for example, a c o m p l e t e m u t a g e n e -
thc ribosome. As for phage display, thc selection is based sis of just six residues requires 20 ~' = 6.4 × 107 molecules).
on protein affinity ~ i t h c o n c o m i t a n t r e c m c r y of the gene A further limitation is that the target antigen is not itself
(as m R N A ) that e n c o d e s the protein. Repertoire diversi- part of the maturation process, but is a d d e d as a static mol-
ty is d i r e c t e d by thc m R N A population. Polysome display ecule in soluble or i m m o b i l i s e d form. Recently, an
has many advantages over phage-displa% including easy improved strategy has been d e x c l o p c d , t c r m e d Selection
synthesis of very large libraries t)3' avoiding the need for and Amplification of Phage (SAP: [64]) or Selective
cell transfections. Infection (SIP; [65,66°]). T h i s sccond-gencration format
uncouplcs the display system (currently as s c F v / F a b on
Library selection phage heads) so that a n t i g e n - a n t i b o d y recognition is an
A n t i b o d y f r a g m e n t s d i s p l a v c d on c i t h e r p h a g e or intcgral c o m p o n c n t of the viral transfection process
polysomcs can be selected by binding to covalcntlv immo- [64,65,66"]. In this way, continuous evolution (affinity
bilised antigen. The choice of affinity matrix can often maturation) of the a n t i g e n - a n t i b o d y interaction can be
determine the kinetics of the selected antibody gone flag- theoretically achieved as affinity is now coupled to the
m e n t [6"',7]. Protocols have bcen established for selection efficiency of viral transfection. T h c antigen is p r e s e n t e d
on live cells, as well as establishing competitive selection as a soluble fusion molecule with the amino-terminal
procedures in solution, to discriminate b e t ~ e e n mutants domain of f d - g p l I I and thc antibody fragment repertoircs
with as little as a txvofold affinity difference [6"']. arc displaycd on phage as fusions to the carboxv-terminal
f d - g p l l l domain. Phage propagation iv ~,i~,o during con-
Affinity enhancement of antibody fragments tinuous cell culture provides a simple selection process for
using libraries the highest affinity l i g a n d - r c c c p t o r interaction. Similar to
(]urrcnt strategies all e m u l a t e the natural somatic hypcr- phage selection, the usc of c o m p e t i n g concentrations ot
mutation and selection process that occurs /,v F/co and are antigen allows selcction of the higher affinity scFv/Fabs.
based on the principal of introducing gcnctic mutations T h e ~ectors and host cclls are still under d e v e l o p m e n t
follCmed by selection of the e n c o d e d antibodies that and it would be unwise to suggcst that S I P - S A P m e t h o d s
exhibit e n h a n c e d affinity. One of the most effcctixe ran- will replace the establishcd fd-phage affinit\ maturation
dora mutagenesis methods incorporates replication iJl c/co procedures in the short tcrm. A major hurdle is the pre-
using F:,~c~eriJzia Fo/i mutator cells [61]. Mutations can bc sentation of antigen as a soluble, correctly folded fusion
targeted to tile V-domain genes by homologous g e n t rcas- molecule [66"].
sortments [62"] or error-prone P e R [6"]. Specific window
mutagenesis targets the introduced sequence diversity to Transgenic mice with human humoral immune
just short regions of p o l x p c p t i d c sequence (e.g. 5-10 systems
residues) and is the most effective m e t h o d in generating a Injection of live animals is still tile simplest route to pro-
c o m p l e t e set of m u t a t e d residues. It is important to make ducc antibody fragments, dcspite the fact that phage
the correct choice for the short regions of sequencc and display' libraries can supcrsede natural i n m n m e repertoires
tlsuallv thc mutations have heen directed to the (',DR in size and afl'inity maturation strategies can s u p e r s e d e
loops [ 6 - ] . Mutations in thc framework regions, however, sonmtic mutation in the specd of selecting high affinity
provide an alternative target for mutagenesis since thcv mutants. Transgenic micc have been produccd that lack
often exhibit a pronouccd cffcct on affinity, stability and the native murine antibody repertoire and instead harbour
expression levels [8,22-25]. most of thc human antibody \:-gencs in the germline [67].
400 Protein e n g i n e e r i n g

Injcction of these animals with a foreign antigcn o r hapten using recombinant antibodies and will be followed by many other
publications in 1998 describing more detailed clinical trials with a number
effectively evokes an immune response and a human anti- of therapeutic recombinant antibodies.
body is produced in B -cells [67]. T h e antibody genes can
3. Parren PW, Burton DR: Antibodies against HIV-1 from phage
bc recovcrcd from B -cells either bv P e R and library sclec- • display libraries: mapping of an immune response and progress
tion o r bv fusion into a monoclonal cell line bv classic towards antiviral immunotherapy. Chem Immunol 1997, 65:18-56.
h vhridtmm teclmology. An excellent review describing the antiviral applications of recombinant
antibodies.
4. Datl'Acqua W, Carter P: Antibody engineering. Curr Opin Struct Biol
Conclusions •. 1998, in press.
This review provides a detailed description of the most recent structural
During the past 12 months there has been a rapid increase in designs and modifications that are possible with engineered antibody
both scientific and commercial interest in recombinant anti- fragments. It summarises the latest methods and strategies and should be
bodies, leading to the large number of publications covered read in conjuction with this review (above).

in this reviex~: (~linical interest is due to encouraging results 5. Banfield M J, King DJ, Mountain A, Brady RL: V-domain rotations
in engineered antibodies-crystal structures of the Fab
from phase 3 trials, which has led to several recent Food and fragments from two murine antitumour antibodies and their
1)rug Adrninistration approvals for therapeutic antibodies. engineered human constructs. Proteins - Struct Func Genet
Scientific interest has stemmed frona the ehtcidation of the 1997, 29:161-171.
key elements required for antibody fragment design and effi- 6. Hoogenboom HR, De Bruine AP, Hufton SE, Hoet RM, Arends J-W,
cient expression. Many of the antibody fragments and fusion •. Roovers Re: Antibody Phage Display technology and its
applications. Immunotechnology 1998, in press.
proteins discussed in this review are now tmdergoing scale- This review provides the most recent detailed description of the strategies
ttp production, some in transgenic animal or plant expression for phage display and selection (see also Scott et al. for phage display of
peptides; this issue, Curr Opin Biotechnol pp 427-435). The review also
systems, and wc cagerly a'wait the results of clinical trials. In provides detailed sections for the currnet in-vogue methods and a section
the next 12 months, it is probablc that wc will havc gcncric on troubleshooting.
solutions for improved Fv stabilit> leading to improved 7. Rader C, Barbas CF II1: Phage display of combinatorial antibody
expression yields, and generic strategies for the design of libraries, Curr Opin Biotechnol 1997, 8:503-508.
mttltimcrs, immunotoxins and other innmative fltsion pro- 8. Baca M, Presta LG, O'Connor S J, Wells JA: Antibody humanization
teins. I.ibrar3: display and selection strategies have been sig- using monovalent phage display. J Biol Chem 1997,
272:10678-10684.
nificantly improved to the point that these i~] ~'i:,o systems
are nov, more effective than conventional hvbridonaa tech- 9. PIL~ckthun A, Pack P: New protein engineering approaches to
• multivalent and bispecific antibody fragments. Immunotechnol
nology. By providing a highly stablc, protcasc-rcsistant scaf- 1997, 3:83-105.
fold, recombinant antibody fragmcnts will continuc to be the A review of multivalent and multispecific antibody fragments with an
excellent explanation of the functional affinity (avidity) gains via apparently
paradigm for selection of high-affinity protein-based reduced off-rates, which result from both multiple binding and rebinding to
rcagents, including the deveh)pmcnt of catalytic antibodies. the target antigens.
10. Adams GP, Schier R, Marshall K, Wolf EJ, McCall AM, Marks JD,
Useful Internet sites for antibodies and fragments: Weiner LM: Increased affinity leads to improved selective tumor
htrp:llx~x~.mfacn.uni-hcidclbcrg.dclSl)lSl)scF~Sirc.htnfl (1"', structure,,) delivery of single-chain Fv antibodies. Cancer Res 1998,
58:485-490.
http://',~ w~.mrc-cpc.caln.ac.uk/m~t-dou (Vbasc: h u m a n anribud~, s e q u e n c e s )
11. Dougan DA, Malby RL, Gruen LC, Kortt AA, Hudson PJ: Effects of
http://x~ ~ ~.antibodvrcsourcc.com {t 5;A links)
substitutions in an antibody binding surface on antigen affinity.
Protein Eng 1998, 11:65-74.
12. Cat XH, Garen A: Comparison of fusion phage libraries displaying
Acknowledgements V H or single chain Fv antibody fragments derived from the
I thank John :\t~cl]. {';rcA (]oia and Alex Kot'tt for their t o t n m c n r s on this antibody repertoire of a vaccinated melanoma patient as a source
rcxicx~ and particularly, John Atwcll fi~r t:igurc 1. of melanoma-specific targeting molecules. Proc Natl Acad Sci
USA 1997, 94:9261-9266.
13. Kortt AA, Lah M, Oddie GW, Gruen LC, Burns JE, Pearce LA,
References and r e c o m m e n d e d reading • Atwell JA, McCoy A J, Howlett G J, Metzger DW et aL: Single chain Fv
Papers of particular interest, published within the annual period of review, Fragments of anti-neuraminidase antibody NCl 0 containing five
have been highlighted as: and ten residue linkers form dimers and with zero residue linker a
trimer. Protein Eng 1997, 10:423-433
° of special interest This publication provides the first detailed description of a triabody and
°* of outstanding interest explains why scFv trimers are the preferred conformation when very short
1. Carter P, Merchant AM: Engineering antibodies for imaging and polypeptide linkers between V domains prevent diabody formation.
• therapy. Curr Opin Biotechnol 1997, 8:449-454.
This review is an excellent summary of the clinical applications using 14. Turner DJ, Ritter MA, George AJT: Importance of the linker in
recombinant antibodies from January 1996 to April 1992. A comparison to expression of single-chain Fv antibody fragments - optirnisation
this current review (above) demonstrates the very rapid progress that has of peptide sequences using phage display technology. J Immunol
been made in the design and engineering of recombinant antibody Methods 1997, 205:43-54.
fragments in 1997 and 1998. 15. De Wildt RM, Ruytenbeek R, van Venrooij WJ, Hoet RM: Heavy chain
2. Maloney DG, Grillo-Lopez A J, White CA, Bodkin D, Schilder RJ, CDR3 optimization of a germline encoded recombinant antibody
• Neidhart JA, Janakiraman N, Foon KA, Liles TM, Dallaire BK et al.: fragment predisposed to bind the U1A protein. Protein Eng 1997,
IDEC-C2B8 (Rituximab) anti-CD20 monoclonal antibody therapy 10:835-841.
in patients with relapsed low-grade non-Hodgkin's lymphoma.
Blood 1997, 90:2188-2195. 16. Komissarov AA, Marchbank MT, Calcutt M J, Quinn TP, Deutscher SL:
This publication describes the pharmocokinetics and Phase I, II trials for the Site-specific mutagenesis of a recombinant anti-single-stranded
first FDA approved recombinant cancer therapeutic antibody. The antibody DNA Fab. Role of heavy chain complementarity-determining
was designed as a simple chimaera of mouse V domains fused to human region 3 residues in antigen interaction. J Biol Chem 1997,
constant region C domains. It is a useful reference for initial clinical data 272:26864-26870.
Recombinant antibody fragments Hudson 401

17. Barbas CF III, Heine A, Zhong G, Hoffmann T, Gramatikova S 35. Iliades P, Kortt AA, Hudson PJ: Triabodies: single chain Fv
• Bjornestedt R, Anderson J, Stura EA, Wilson IA, Lerner RA: Immune fragments without a linker form trivalent trimers. FEBS Lett 1997,
versus natural selection: antibody aldolases with enzymic rates 409:437-441.
but broader scope. Science 1997, 278:2085-2092.
Catalytic antibodies provide a new range of commercial opportunities for 36. Pei XY, Holliger P, Murzin AG, Williams RL: The crystal structure of a
engineered antibodies. This publication is the best demonstration to date trimeric antibody fragment with non-cognate MH-VL domain pairs
that describes the selection of new antibodies with hitherto unknown shows a rarrangement of VH CDR3. Proc Nat/Acad Sci USA 1997,
enzyme (catalytic) activity and with specific catalysis of new chemical 94:963?-9642.
reactions (aldol condensations). 3?. Adams GP, Schier R, McCall AM, Crawford RS, Wolf EJ, Weiner LM,
18. McCoy A J, Epa VC, Colman PM: Electrostatic complementarity at • Marks .ID: Prolonged in vivo tumor retention of a human diabody
protein/protein interfaces. J Mol Bio/1997, 268:5?0-584. targeting the extracellular domain of human HER2/neu. Brit E
Cancer 1998, 77:1405-1412.
19. Arbabi Ghahroudi M, Desmyter A, Wyns L, Hamers R, The second study on tumour imaging with bivalent diabodies, with results
Muyldermans S: Selection and identification of single domain that concur with the previous anti-CEA results of Anna Wu et aL
antibody fragments from camel heavy-chain antibodies. FEBS (Immunotechnol 1996, 2:21-36).
Lett 1997, 414:521-526.
38. Casey JL, King DJ, Chaplin LC, Haines AM, Pedley RB, Mountain A,
20. Wedemayer G J, Patten PA, Wang LH, Schultz PG, Stevens RC: Yarranton GT, Begent RH: Preparation, characterisation and
• Structural insights into the evolution of an antibody combining tumour targeting of cross-linked divalent and trivalent anti-
site. Science 1997, 276:1665-1669. tumour Fab' fragments. Brit J Cancer 1996, 74:1397-1405.
This important publication demonstrates for the first time that affinity
maturation in vivo can result from mutations exclusively in the V domain 39. Lawrence U, Kortt AA, Iliades P, Tulloch PA, Hudson PJ: Orientation
interface that stabilise V domain pairing rather than resulting from mutations of antigen binding sites in dimeric and trimeric single chain Fv
in the CDR loops that form the antigen binding surface. antibody fragments. FEBS Lett 1998,425:479-484.

21. Colman PM: Structure of antibody-antigen complexes: implications 40. Kontermann RE, Martineau P, Cummings CE, Karpas A, Allen D,
for immune recognition. Adv Immunol 1988, 43:99-132. Derbyshire E, Winter G: Enzyme immunoassays using bispecific
diabodies. Immunotechno/1997, 3:137-144.
22. Kipriyanov SM, Moldenhauer G, Martin ACR, Kupriyanova OA,
Little M: Two amino acid mutations in an anti-human CD3 single 41. Libyh MT, Goossens D, Oudin S, Gupta N, Dervillez X, Juszczak G,
chain Fv antibody fragment that affect the yield on bacterial Cornillet P, Bougy F, Reveil B, Philbert F et aL: A recombinant human
secretien but not the affinity. Protein Eng 1997, 10:445-453. scFv anti-Rh(D) antibody with multiple valences using a
C-terminal fragment of C4-binding protein. Blood 1997,
23. Nieba L, Honegger A, Krebber C, Pl~ckthun A: Distrupting the 90:3978-3983.
hydrophobic patches at the antibody variable/constant domain
interface improved in vivo folding and physical characterisation
-
42. Li E, Pedraza A, Bestagno M, Mancardi S, Sanchez R, Burrone O:
of an engineered scFv fragment. Protein Eng 1997, 10:435-444. Mammalian cell expression of dimeric small immune proteins
(SIP). Protein Eng 1997, 10:731-736.
24. Forsberg G, Forsgren M, Jaki M, Norin M, Sterky C, Enhorning A,
Larsson K, Ericsson M, Bjork P: Identification of framework 43. Holliger P, Wing M, Pound JD, Bohlen H, Winter G: Retargeting
residues in a secreted recombinant antibody fragment that serum immunoglobulin with bispecific diabodies. Nat Biotechnol
control production level and Iocalisation in E. coil J Bio/Chem 1997, 15:632-636.
1997, 272:12430-12436.
44. Kontermann RE, Wing MG, Winter G: Complement recruitment
25. Atwell S, Ridgway JB, Wells JA, Carter P: Stable heterodimers from using bispecific diabodies. Nat Biotechnol 199?, 15:629-631.
remodeling the domain interface of a homodimer using a phage
45. Kufer P, Mack M, Gruber R, Lutterbuse R, Zettl F, Riethmuller G:
display library. J Mol Biol 1997, 270:26-35.
Construction and biological activity of a recombinant bispecific
26. Brinkmann U, Dicarlo A, Vasmatzis G, Kurochkina N, Beers R, Lee B, single-chain antibody designed for therapy of minimal residual
Pastan h Stabilisation of a recombinant Fv fragment by base- colorectal cancer. Cancer Immunol 199?, 45:193-197.
loop interconnection and VH-V L permutation. J Mol Biol 1997,
268:107-117. 46. Brekke OH, Michaelsen TE, Sandlin R, Sandlie I: Activation of
complement by an IgG molecule without a genetic hinge. Nature
27. Fitzgerald K, Holliger P, Winter G: Improved turnout targeting by 1996, 383:103.
disulphide stabilised diabodies expressed in Pichia pastoris.
Protein Eng 1997, 10:1221-1225.
47. Coloma MJ, Trinh KR, Wims LA, Morrison SL: The hinge as a spacer
contributes to covalent assembly and is required for function of
28. Tsumoto K, Nishimiya Y, Kasai N, Ueda H, Nagamune T, Ogasahara K, IgG. J Immunol 1997, 158:733-740.
Yutani K, Tokushisa K, Matsushima M, Kumagai I: Novel selection
method for engineered antibodies using the mechanism of Fv 48. Ghetie MA, Ghetie V, Vitetta ES: Immunotoxins for the treatment of
fragment stabilization in the presence of antigen. Protein Eng B-cell lymphomas. Mol Med 1997, 3:420-427.
1997, 10:1311-1318. 49. Zewe M, Rybak SM, Dubel S, Coy JF, Welschof M, Newton DL, Little M:
29. Uson I, Bes MT, Sheldrick GM, Schneider TR, Hartsch T, Fritz H-J: • Cloning and cytotoxicity of a human pancreatic RNase
X-ray crystallography reveals stringent conservation of protein immunofusion. Immunotechnol 1997, 3:127-136.
fold after removal of the only disulfide bridge from a stabilised Clinical trials with immunotoxins have been compromised by the
immunoglobulin variable domain. Fold Des 1997, 2:357-362. immunogenicity of the conventional antibody fusion proteins to ricin or
bacterial toxins. This publication demonstates that conjugation of antibodies
30. Rondon IJ, Marasco WA: Intracellular antibodies (intrabodies) for to human cytotoxic proteins, RNase or onconase, is an effective and
gene therapy of infectious diseases. Ann Rev Microbiol 1997, presumably less immunogenic alternative therapeutic reagent.
51:257-283.
50. Pastan I: Targeted therapy of cancer with recombinant
31.. Gargano N, Cattaneo A: Inhibition of routine leukaemia virus immunotoxins. Biochim Biophys Acta 1997, 1333:C1-6.
retrotranscription by the intracellular expression of a phage-
derived anti-reverse transcriptase antibody fragment. J Gen Virol 51. Martin J, Stribbling SM, Poon GK, Begent RH, Napier M, Sharma SK,
1997, 78:2591-2599. Springer CJ: Antibody-directed enzyme prodrug therapy:
pharmacokinetics and plasma levels of prodrug and drug in a
32. Roux KH. Strelets L, Michaelsen TE: Flexibility of human IgG phase I clinical trial. Cancer Chemother Pharmacol 1997,
subclasses. J Immunol 1997, 159:3372-3382. 40:189-201.
33. Thouvenin E, Laurent S, Madelaine MF, Rasschaert D, Vautherot JF, 52. Siemers NO, Kerr DE, Yarnold S, Stebbins MR, Vrudhula VM,
Hewat EA: Bivalent binding of a neutralising antibody to a Hellstrom I, Hellstrom KE, Senter PD: Construction, expresssion
calicivirus involves the torsional flexibility of the antibody hinge. and activities of a single-chain antibody-beta-lactamase fusion
J Mol Biol 1997, 270:238-246. protein for anticancer prodrug activation. Bioconjug Chem 1997,
8:510-519.
34. Ghetie MA, Podar EM, Ilgen A, Gordon BE, Uhr JW, Vitetta ES:
Homodimerization of tumor-reactive monoclonal antibodies 53. Penichet ML, Harvill ET, Morrison SL: Antibody-lL-2 fusion proteins:
markedly increases their ability to induce growth arrest or apoptosis a novel strategy for immune protection. Hum Antibodies 1997,
of tumor cells. Proc Nat/Acad Sci USA 1997. 94:7509-7514. 8:106-118
402 Protein engineering

54. EshharZ: Tumor-specific T-bodies: towards clinical application. 62. Zhao H, Giver L, Shao Z, Affholter .JA, Arnold FH: Molecular
Cancer Immune/Immunother 1997, 45:131-136 • evolution by staggered extension process (STEP) in vitro
recombination. Nat Biotechnol 1998, 16:258-261.
55. Wickham TJ, Lee GM, Titus JA, Sconocchia G, Bakacs T, Kovesdi I, An important new method that complements the DNA shuffling and
Segal DM: Targeted adenovirus-mediated gene delivery to T cells
via CD3. J Virol 1997, 71:7663-7669. homologous recombination strategies discussed by Willem Stemmer (Curt
Opin Biotechnol 1997, 8:724-733).
56. Watkins S.I, Mesyanzhinov VV, Kurochkina LP, Hawkins RE: The
'adenobody' approach to viral targeting: specific and enhanced 63. MahlerSM, Marquis CP, Brown G, Roberts A, Hoogenboom HR:
adenoviral gene delivery. Gene Ther 1997, 4:1004-1012. Cloning and expression of human V-genes derived from phage
display libraries as fully assembled human anti-TNF alpha
57. Chu THT, Dornburg R: Toward highly efficient cell-type specific monoclonal antibodies. Immunotechnology 1997, 3:31-43.
gene transfer with retroviral vectors displaying single-chain
antibodies. J Virol 1997, 71:720-725. 64. Malmborg AC, Soderlind E, Frost L, Borrebaeck CA: Selective phage
infection mediated by epitope expression on F pilus. J Mol Biol
58. Kobatake E, Sasakura H, Haruyama T, Laukkanen ML, Keinanen K, 1997, 273:544-551.
Aizawa MA: Fluoroimmunoassay based on immunoliposomes
containing genetically engineered lipid-tagged antibody. Anal 65. PedrazziG, Schwesinger F, Honegger A, Krebber C, Pl~JckthunA:
Chem 1997, 69:1295-1298. Affinity and folding properties both influence the selection of
antibodies with the selectively infective phage (Sl P)
59. Kerschbaumer RJ, Hirschl S, Kaufmann A, Ibl M, Koenig R, Himmler G: methodology. FEBS Lett 1997, 415:289-293.
Single chain Fv fusion proteins suitable as coating and detecting
reagents in a double antibody sandwich enzyme-linked 66. Krebber C, Spada S, Desplancq D, Krebber A, Ge L, PI0ckthun A:
immunosorbent assay. Anal Biochem 1997, 249:219-227. • Selectively infective phage (SIP): a mechanistic dissection of a
60. Hanes J, Pl(Jckthun A: In vitro selection and evolution of functional novel method to select for protein-ligand interactions. J Mol Biol
eo proteins by using ribosome display. Proc Nat/Acad Sci USA 1997, 1997, 268:619-630.
94:4937-4942. The best description of SIP technology which may provide the first
The first detailed description on the use of polysomes for the display of continuous molecular evolution strategy by phage display.
antibody fragment repertoires. 6?. Mendez MJ, Green LL, Corvalan JR, Jia XC, Maynard-Currie CE,
61. Cola G, Ayres A, Lilley GG, Hudson PJ, Irving RA: Use of mutater Yang XD, Gallo ML, Louie DM, Lee DV, Erickson KL et al.: Functional
cells as a means for increasing production levels of a transplant of megabase human immunoglobulin loci recapitulates
recombinant antibody directed against Hepatitis B. Gene 1997, human antibody response in mice. Nat Genet 1997, 15:146-156.
201:203-209. [Published erratum appears in Nat Genet 1997, 16:410.]

Вам также может понравиться