fluorescent proteins with ties that bind
TRANSCRIPT
N E W S A N D V I E W S
Harnessing the immune system’s ability togenerate high-affinity binding pockets hasproven to be an enormously important tech-nology for biomedical science. The ability to generate protein-specific affinity reagentshas formed the basis of thousands of types of assays for purifying, detecting and probingthe molecular components encoded bygenomes—assays that have touched virtuallyevery facet of modern molecular biology.In this issue, Bradbury and colleagues1
demonstrate a powerful new hybrid reagent:intrinsically fluorescent proteins that sport the complementarity-determining regions(CDRs) normally found in antibody variabledomains. This new class of reagent, termed a‘fluorobody’, thus marries one of the greatestrecent boons to cell biology and physiologyresearch, fluorescent proteins (such as greenfluorescent proteins or GFPs), with the abilityto select for high-affinity binding to a target ofchoice. Although antibodies have proven pow-erful and versatile, they require a secondarydetection scheme to be useful, which typicallyrequires the use of at least two other reagents.With fluorobodies the detection scheme isbuilt in.
The structure of an intact antibody is com-plex, and high-affinity binding to antigensarises from the detailed crevices, surfaces andpockets formed by the apposition of six CDRs,three from each of two different gene prod-ucts, the heavy- and light-chain variable dom-ains, respectively. Considerable efforts havegone into creating reagents that maintain high-affinity binding and large diversity but aremore amenable to high-throughput expres-sion systems. Single-chain antibodies (scFvs)
maintain the diversity and high-affinitycapacity of intact antibodies in a simpler pack-age, whereby the variable domains of the lightand heavy chains are expressed as a single geneproduct connected by a flexible linker2.Recently, this technology was simplified evenfurther with the creation of single-domainantibody (IDab) libraries, which, althoughcombinatorially less complex than scFvs,maintain excellent antigen-binding proper-ties3. These schemes have allowed the creationof high-diversity libraries that can be com-bined with efficient expression and selectionschemes to identify high-affinity reagents4,5
that work well both inside and outside cells.Although in principle the gene for GFP
could be fused directly into either scFv orIDab libraries to provide a built-in detectionscheme, this would significantly increase thesize of the reagent and would be likely todecrease the efficiency of production because
GFPs normally fold in the intracellular space,whereas antibodies normally fold withinlumenal domains of intracellular organelles.Instead, Bradbury and colleagues take advan-tage of the fact that GFPs arise from a novelthree-dimensional fold called a β-can formedby 11 β-strands wrapping into a cylinder (Fig. 1a). In intact antibodies, the CDRs arebrought together at the top of a cylindricalsurface defined by two β-sheets, one fromeach of the heavy- and light-chain variabledomains (Fig. 1b). The similarity in thesestructures suggested that GFPs themselvesmight provide an appropriate scaffold to allowCDRs to define a high-affinity binding arr-angement. As the various β-strands in GFPalso have connecting loops at the top of thecylinder, Bradbury and colleagues replacedfour of the GFP loops (Fig. 1c) with a libraryof the CDR with the greatest diversity(CDR3); and thus fluorobodies were born.
Expression of only a modestly diverselibrary (107 elements) with a phage-displaysystem led to the isolation of numerous fluo-robodies against various antigens with affini-ties as high as the nanomolar range. Theintrinsic fluorescence itself is very useful dur-ing the selection and purification procedures,as the binding properties are inherently linkedto the stability of the protein and therefore its
Timothy A. Ryan is in the Department ofBiochemistry, Joan and Sanford I. Weill MedicalCollege, Cornell University, 525 E. 68th Street,New York, New York 10021, USA.e-mail: [email protected]
Fluorescent proteins with ties that bindTimothy A Ryan
Antigen-binding fluorescent proteins provide a powerful tool for immunodetection-based assays.
NATURE BIOTECHNOLOGY VOLUME 21 NUMBER 12 DECEMBER 2003 1447
HCDR1
HCDR3
HCDR2LCDR1
LCDR3
LCDR2
a
c
bFigure 1 The making of a fluorobody. (a) β-canstructure of GFP made up of 11 β-strands into acylinder. At the top of the can, loop regions jointhe strands. (b) The antigen-binding site ofantibodies is made by two sets of β-sheets fromboth heavy and light chains organized in acylindrical surface. The hypervariable regions, theloops at the top of this cylindrical arrangement(HCDR1, HCDR2 and HCDR3 and LCDR1,LCDR2 and LCDR3), define the antigen-bindingsite. (c) In a fluorobody, four of the loops of GFPare replaced by a library of HDCR3. These loopsdefine a new binding site using the β-can as ascaffold rather than the light chain– and heavychain–defined β-cylinder.
Bob
Crim
i
©20
03 N
atu
re P
ub
lish
ing
Gro
up
h
ttp
://w
ww
.nat
ure
.co
m/n
atu
reb
iote
chn
olo
gy
N E W S A N D V I E W S
fluorescent properties. Fluorobodies haveproven useful for detection of proteins in gels,enzyme-linked immunosorbent assays, flowcytometry and immunocytochemistry oncells, indicating that they are essentially capa-ble of replacing conventional antibodies formost uses with the benefit that secondaryreagents are not needed to detect the antibody.
In cell biological applications, the ability tolabel proteins in their cellular context withexogenous reagents is very valuable. As intactantibodies are bivalent, labeling often res-ults in crosslinking of the target molecule.Although less of a concern in fixed tissue, thiscan be a significant perturbation in live cells.Furthermore, although the production of use-ful Fab fragments that have been covalentlycoupled to a fluorophore is possible, they frequently lose antigen-binding capacity atthis step. Fluorobodies will almost certainlybind monovalently and are immediately use-ful for fluorescence detection without theneed for secondary antibodies or fluorophorecoupling.
One exciting avenue for development is thepotential to use the ever-expanding repertoireof GFPs in conjunction with this technology.Since their original discovery, GFPs have bec-ome an invaluable tool in many aspects of bio-medical research and have been coaxed intorevealing not only the dynamics and subcellu-lar locations of given proteins but also infor-mation such as small-molecule-messenger
dynamics, enzyme activation, protein-proteininteractions and intracellular as well asintralumenal pH6. The possibility of couplingthese types of changes in fluorescence proper-ties with antigen binding would almost cer-tainly expand the number of useful app-lications of intrinsically fluorescent proteins.Finally, this technology may be a boon simplyin the production of protein-specific affinityreagents. In contrast to antibody productionfrom immune cells, which is expensive in largequantities, fluorobodies should be easily gen-erated reproducibly and in large quantities inbacterial expression systems.
Although it remains to be seen whether fluorobodies can replace antibodies for alluses, particularly because secondary reagentsused for detection often increase sensitivity,this technological breakthrough will likelycompel many researchers to change how theynormally use protein-specific affinityreagents, for the simple reason that where onepreviously needed three reagents one nowonly needs one.
1. Zeytun, A., et al. Nat. Biotechnol. 21, 1473–1479(2003).
2. Huston, J.S. et al. Proc. Natl. Acad. Sci. USA 85,5879–5883 (1988).
3. Tanaka, T., Natividad, L. & Rabbitts, T.H. J. Mol. Biol.331, 1109–1120 (2003).
4. Sheets, M.D. et al. Proc. Natl. Acad. Sci. USA 95,6157–6162 (1998).
5. Feldhaus et al. Nat. Biotechnol. 21, 163–170 (2003).6. Zhang, J. Campbell, R.E., Ting, A.Y. & Tsien, R.Y. Nat.
Rev. Mol. Cell Biol. 3, 906–918 (2002).
repair is the nascent primary transcript, whichon the average spans ∼ 27,000 nucleotides andis interrupted by seven introns (see ref. 2 andreferences therein). These primary transcriptsare frequently alternatively spliced to producemany different mRNA isoforms, and thisprocess can be redirected to preferentiallyexpress some isoforms over others in a type ofreprogramming where no new information is provided to the revised transcript4. The nascent transcript is also the target for spliceosome-mediated RNA trans-splicing2,5.Endogenous spliceosomes catalyze splicing of an exon in the target transcript (exon 2 inFig. 1) with an exon in an engineered pre-trans-splicing molecule (PTM)2,5. The PTMexon replaces the mutant exon 3 (m) for awild-type exon (Fig. 1, reaction 1). PTM tar-geting is accomplished through base-pairinginteractions, which in the figure are schemati-cally drawn within the second intron of thetranscript. After successful trans-splicing, thereprogrammed mRNA is transported to thecytoplasm and translated to produce func-tional protein.
Spliceosome-mediated trans-splicing hasbeen used to repair endogenous mRNAs andrestore function not only in cell culture butalso in animals5–7. Although these studiesmerit cautious optimism, several issues limitthe potential of this methodology—mostimportantly, this type of trans-splicing re-quires the presence of nuclear introns and tar-gets trans-splicing only to available splice sites.The specificity of spliceosome-mediatedtrans-splicing has not yet been completelyascertained, and competing reactions, such asnonspecific trans-splicing, may lead to expres-sion of undesirable products.
After nuclear RNA processing and nucleo-cytoplasmic transport, mutant mRNAs havebeen reprogrammed by trans-splicing medi-ated by group I ribozymes (Fig. 1, reaction 2).This reaction, which provided the first exam-ple of RNA reprogramming8, requires theexpression of a targeting transcript thatencodes for both the trans-splicing group Ienzyme and the coding sequence to beincluded in the revised RNA. Group Iribozyme–mediated trans-splicing has beenused to effect mRNA repair in cells in cul-ture9,10, but has yet to restore function in ani-mals. An advantage of this technique is that itis not limited to naturally occurring splicesites, thus extending the potential range ofRNA reprogramming to mRNAs fromintronless genes and also to noncoding RNAs.As with spliceosome-mediated trans-splicing,targeting by group I introns is achieved bybase pairing of the ribozyme to the targetmRNA, and therefore similar questions about
1448 VOLUME 21 NUMBER 12 DECEMBER 2003 NATURE BIOTECHNOLOGY
Gene therapy, just like the current economicrecovery, is buoyed by hopeful long-term fore-casts but encumbered by important everydayhurdles. To overcome these problems, we relyon the evolution of novel approaches andtechniques. One such general approach isknown as mRNA reprogramming—a form ofgene therapy that modifies transcripts directly.Messenger RNA reprogramming in mam-
malian cells has been achieved primarily byrecombining mRNA molecules via trans-splicing using introduced group I ribozymes1
or endogenous spliceosomes2. In this issue,Deidda et al.3 report on a third and intriguingvariation of this reprogramming theme. Theauthors use a combination of a splicing endo-nuclease from an archaeal bacterium and atargeting RNA to alter mRNAs in cells in cul-ture. Because the splicing endonuclease medi-ates trans-splicing only if it encounters a veryspecific structure, this reprogramming meth-od could offer significant advantages in casesrequiring exquisite specificity.
RNA reprogramming can be achieved atmultiple sites during the process of geneexpression (Fig. 1). The first target for RNA
Mending the messageMariano A Garcia-Blanco
An ancient, archaeal splicing endonuclease provides a promising alternativefor gene therapy.
Mariano A. Garcia-Blanco is in the Departmentof Molecular Genetics and Microbiology,Department of Medicine, Box 3053 (424 CARL),Duke University Medical Center, ResearchDrive, Durham, North Carolina 27710, USA.e-mail: [email protected]
©20
03 N
atu
re P
ub
lish
ing
Gro
up
h
ttp
://w
ww
.nat
ure
.co
m/n
atu
reb
iote
chn
olo
gy