Antibody Fragment and Affibody ImmunoPET Imaging Agents: Radiolabelling Strategies and Applications

Abstract Antibodies have long been recognised as potent vectors for carrying diagnostic medical radionuclides, contrast agents and optical probes to diseased tissue for imaging. The area of ImmunoPET combines the use of positron emission tomography (PET) imaging with antibodies to improve the diagnosis, staging and monitoring of diseases. Recent developments in antibody engineering and PET radiochemistry have led to a new wave of experimental ImmunoPET imaging agents that are based on a range of antibody fragments and affibodies. In contrast to full antibodies, engineered affibody proteins and antibody fragments such as minibodies, diabodies, single‐chain variable region fragments (scFvs), and nanobodies are much smaller but retain the essential specificities and affinities of full antibodies in addition to more desirable pharmacokinetics for imaging. Herein, recent key developments in the PET radiolabelling strategies of antibody fragments and related affibody molecules are highlighted, along with the main PET imaging applications of overexpressed antigen‐associated tumours and immune cells.


ImmunoPET:Antibodies for PET Imaging
ImmunoPET,i ni ts simplestt erms, is the combination of an antibody,orr elated molecule, with ad iagnostic positron-emitting radionuclide for the purposes of imaging an associated antigen in vivo. [1,2] ImmunoPET is playing an important role in the diagnosis, staginga nd monitoring of treatment response in cancer. Although the combination of radionuclides with antibodies for imaging or therapyi sn ot an ew concept, recently there has been ar esurgence in interest owing to advances in antibody engineering technology,t he greater availability of diagnostic PET radionuclides and the development of new site-specific chemicalc onjugation methods. The foundation of ImmunoPET is the matching of the antibody's ability to specifically engage at arget at sub-nanomolar affinities with the exquisitely high sensitivity of PET imaging. PET is ac linically-based non-invasive imaging technique that detects coincident gamma rays emit-ted from the positron decay/annihilation events from administered radiolabelled tracers. [3,4] Thea bility to detect very low levels of radioactivity via coincidence detection means that PET is incredibly sensitive, with only nanomolar amountso fa given PET tracer required for imaging. PET imaging is therefore ap owerful clinical techniqueu sed to map the biodistribution of tracers and to quantify their uptake in vivo. The combination of an exceptionally high-specificity/high-affinity antibody molecule with as ensitive imaging technique such as PET should, in principle, produce av ery powerful diagnostic tool that can complement other clinical imaging methods and interventions such as biopsied tissue and/ors urgery.I np ractice, however,f ull antibodyi maging agentsb ased on whole, intact antibodies suffer some significant drawbacks mainly as ad irect result of their large size ( % 150 kDa). They have sluggish pharmacokinetics and resultantly long circulation times (up to 3weeks);l onger-lived radionuclides (e.g., 89 Zr, 124 I) are therefore required for imaging. Such longer-lived radionuclides are less ideal for clinicali maging duet oh igher associated radiation doses and longer wait times for imaging.T he large size of intact antibodies typically resultsi nc learance via the liver which can preclude imagingo fl iver disease. Slow blood clearance times and nonspecific bindingo ft he tracer typically result in ah igherb ackground signal and therefore ad ecrease in the PET signal contrast; this ultimately leads to poorer image quality. [4] As many full antibodies are also therapeutic agents they could in principle stimulate unwanted biological responses due to the interaction of their Fc regions with cell surfacer eceptors, however,a tt he low concentrations typically used for PET imaging this is unlikely to happen. Ideally,t he tracer should not perturb the biological system under study, therefore having an understanding of the imaging agent dose Antibodies have long been recognised as potent vectors for carrying diagnostic medicalr adionuclides, contrast agentsa nd opticalp robest od iseased tissue for imaging. The area of Im-munoPET combines the use of positrone mission tomography (PET) imaging with antibodies to improve the diagnosis, staging and monitoring of diseases. Recent developments in antibody engineering and PET radiochemistry have led to an ew wave of experimental ImmunoPETi maging agents that are based on arange of antibodyf ragments and affibodies. In contrast to full antibodies, engineered affibody proteins and anti-body fragments such as minibodies, diabodies, single-chain variabler egion fragments (scFvs), and nanobodies are much smallerb ut retain the essential specificities and affinitieso ff ull antibodies in addition to more desirable pharmacokinetics for imaging. Herein, recent key developments in the PET radiolabelling strategies of antibody fragments and related affibody molecules are highlighted, along with the main PET imaging applicationso fo verexpressed antigen-associatedt umoursa nd immune cells.
is important to ensure that it has minimal pharmacological or toxicological effects on the model, and also to ensure the highest possible contrast PET images. Ar ecent study using an 89 Zr-labelled Cys-diabody fragment for the preclinical imaging of CD4 + T-cells demonstrated the importance of dose in obtaining high contrasti mages, and showed that when using their imaginga gent al ower dose resulted in higherq uality images. [5] Antibody fragments are specifically engineered parts of antibodies that retain the desirable high affinitiesa nd specificities of full-intact antibodies, but with more compatible pharmacokinetics for imaging. [6] They essentiallyc ontain only the basic targetinga nd binding components. Despite their relative lack of complexity,the number of antibody fragments in clinical development is much smaller than that of intact antibodies. [7,8] To enable imaging they also need to contain sufficient functionality to attach ar adionuclide. Typically,t hey range in size from 7 to 100 kDa, depending on the specific type of fragment ( Figure 1). Antibody fragments have much shorter circulation times (hours rather than weeks), deeper penetration into tissue, are therefore better matched with the more clinically applieds horter-lived PET radionuclides (e.g., 18 F, 68 Ga) and can enable same-dayi maging. [9] This should, in theory,l ead to improvedi mages as blood clearancew ill be faster giving ab etter signal-to-noise ratio for the specifically bound radiolabelled fragment. [10,11] It may also be expected that fragments will be better tolerated by the subject, as they have been stripped of their variable domains, the part of the antibody that typically provoke much stronger immune responses. Absence of the Fc region also decreases the nonspecific binding of the fragment to other cells andc an therefore improvei mage quality.I nc ontrast to intact antibodies, antibody fragments, due to their much smaller sizes, are excreted via the renal system and are therefore not significantly metabolised or retained by the liver (Figure 1). Although antibody fragments do have the distinct advantages of better tumour penetration and faster blood clearance (Figure 1), they can have lower affinities and typically display lower overall tumour uptake than full antibodies. [12] Affibody molecules are engineered proteinst hat have a5 8 amino acid sequence folded into three alpha helices. They mimic monoclonal antibodies and antibody fragments with their high affinities and selectivities, however,t hey are much smaller ( % 6-7 kDa, Figure 1) and chemically robustp roteins that can tolerate highert emperatures and more extreme pH. Their small size, chemical robustness and high affinity (nanomolar range) make them excellent candidates to act as imaging probes. [13] Their high affinities and short circulation times can result in high contrast PET imaging within hours of their administration. [14] They can be efficientlys elected for by phage display and produced on scale by recombinantt echniques or by chemical methods using solid-phase peptidesynthesis.
Despite all the potential advantages of antibody fragments and affibodies, they do not in anyw ay spell the end of fullintact antibodies for imaging. The applicationo fl ongerl ived isotopes such as 89 Zr (t1 = 2 % 78 h) and 124 I( t1 = 2 % 100 h) forl abelling intact antibodies can providei mportant continuous imaging information over longert ime periods. The use of pre-targeting strategies, wherebyf unctionalised antibodies are given enough time to engage their targetf ollowed by administration of ap ositrone mitting labelledm olecule that will conjugate in vivo, enables the use of shorter-lived PET radionuclides. [15,16] In this mini-review,w eh ighlight developments in the fragmentbased ImmunoPET area, discuss currents trategies for radiolabelling antibodyf ragments and affibodies, and comment on the applicantsfor targeted imaging.

Antibody Fragments: AD esign of the Times
The monoclonala ntibody therapeuticsm arket is set to be worth US$246 billion by 2024 with 74 approved antibodybased drugs as of mid-2017. [7] This success is fuelling the development of the next generation of antibody-based therapeu- tics and diagnostics. One area that has seen explosive growth has been in the development of alternative formats to intact antibodies especially in imaging and diagnostics driven by the inherenti ssues of both size and long serum residence time. Antibody fragments can be produced through either chemical/ enzymatic digestionorg enetic engineering. [17,18] The enzymatic digestion of intact antibodies with papain or pepsin is used to produce Fab'' and (Fab') 2 fragments,t hese have one or two antigen-binding regions, respectively but lack the Fc region that contains the heavy chain constant domains. Although Fab'' fragments can exhibit rapid renal clearance and improved tumour penetration theyc annotbep repared from all subclasses of antibodies through biochemical methods.S uch production methods are laborious and require large quantities of startingi ntact antibody.T he rapid development of antibody engineering technology has enabled the relativelye asy productiona nd isolation of the variable regions of antibodies, such as single-chain variable region fragments (scFvs), anda range of engineered diabodies, minibodies and single-domain antibodyv ariants. Each of these variantsh as unique binding and functional characteristics. The various commonf ragments are shown in Figure 1. One of the most popularf ormats is the scFv. [19] ScFv'sh ave their V H and V L domains linked by as hort flexible peptidec hain between the C-terminus of one Fv region to the N-terminus of another.These fragments are small (26 kDa) and when the linker is at least 12 residues long are both monomeric and monovalent. In terms of designa nd engineering, as ingle gene sequence codes for the entire scFv. Shortening the peptidel inker (< 11 residues) between the V domainsf orces the scFv to self-assemble,c reating ab ivalent diabody( 55 kDa) [20] and shortening the link even further( < 3 residues)f orces assembly into trivalent tribodies (80 kDa) and tetravalent tetrabodies (110 kDa). Genetically engineering an interchaind isulfide bond between the two variable domains createsadisulfide stabilised Fv fragment (dsFv)w ithout and with ap eptide linker (scdsFv). These can address some of the stabilitya nd aggregation issues associatedw ith scFvs. Minibodies are scFv-CH3 fusion proteins where two scFvs are linked by ac omponent of the heavy chain;t hese assemble into bivalent dimers (75 kDa). Bivalent single-chain variable fragments (bi-scFvs, 55 kDa) are engineered by linking two scFvs andh ave two different domains allowingt hem to bind to two different epitopes. The smallest known fragments that are stillc apable of selectively binding an antigen like an intact antibodya re single domains, and are derived from either V H or V L regionso ri solated from camelids. These single-domain antibodies (sdAbs, 12-15kDa), or nanobodies, [21] have an umber of distincta dvantages over scFvs in terms of stability,ease of productiona nd size. They are particularly amenable to applications such as radionuclide-based imaging, which requires a combination of enhanced tissuep enetration and rapid clearance. On an even smaller scale, affibodies( 6-7 kDa) are engineered affinity proteins derived from the B-domain in the immunoglobulin-binding region of staphylococcal protein A throughp hage display. [13] Affibodies seem ideally suited for Im-munoPET due to their small size, stability (extreme pH and temperature) and the presence of au nique C-terminalc ysteine for conjugation. [22] However,t heir rapid clearance and decreased avidity for the target remainchallenging.

Labelling Strategies and Applications
Ar ange of common PET radionuclides have been used to label antibody fragments and affibodies (Table 1). Their physical properties-half-life, decay characteristics and labelling chemistry-dictate both the types of fragments that can be labelled and how they are labelled.K ey to the labelling of any tracer,i ncludinga ntibody fragments, is the appropriate matching of the radionuclide half-life with the biological process under study to ensure that there is sufficient tracer accumulation to observe as ignal. Because fragments have longerb iological half-lives than smallm olecule tracers ( % 200-800 Da), tissue penetration is slower and clearance times are longer, thus radionuclides with t1 = 2 on the order hours are required. The positron yield and positrone nergy of the radionuclide affects the sensitivity and the image resolution, respectively. High positron yields as observed for 11 C, 19 Fa nd 68 Ga result in higher sensitivity,a st he major decay pathway is via positron emission. Radionuclides with alternative decay pathways, such as 64 Cu or 89 Zr,m ay requirel arger doses to be administered to the subject in order to obtain the required level of signal. A low positrone nergy is desirable because this determines the distance the positronw ill travel in the body after decay,p rior to annihilation. This is one of the physical limitations that determines the resolution of aPET image.
The longer livedP ET radionuclides 89 Zr and 124 Ih ave been used to radiolabel intact antibodies, [23,24] whilst the faster pharmacokinetics of antibodyf ragments have enabled the use of the shorter-lived nuclides 18 F, 64 Cu, 44 Sc and 68 Ga for imaging. The fragment labelling strategy depends on both the particular isotope chosen and the availablefunctionality on the particular fragment for conjugation.B ecause antibody fragments are large biomolecules and their tertiarys tructures are determined by many complex noncovalent bonds, extreme temperatures, pH and reducing conditions during the radiolabelling process can affect their structurali ntegrity.S electing an appropriate radiolabelling strategy is therefore key to ensuring that the affin- ity and specificity of the antibody fragment is retained. In addition, the radionuclide needs to be covalently bound to the fragment, and remain kinetically and thermodynamically stable over the time-course of imaging. Degradation of the fragment in vivow ith the formation of variousr adio-metabolites can confound the signal giving poor quality images. As ar esult, intact antibodies and variousa ntibody fragments are almost exclusively radiolabelled via indirect labellingr outes that first involvet he preparation of ar adiolabelled" prosthetic" group or ab ifunctional chelate complex (in the case of radiometals), that can then be conjugated to the fragment under much milder reaction conditions. Additionally,t he position of the chelate on the fragment, type of chelate, number of chelators attached,a nd relative size of chelate compared to the fragment may affect the targeting of the antibody fragment. The conjugation of such chelators can affect the local chargeo rl ipophilicity of the fragment and therefore its physicochemical properties. This is known to have ag reater influence on the binding of smallerf ragments and affibodies. [25,26] The options for radiolabelling antibody fragments and affibodies are more limited than for intact antibodies, as they are smaller and have fewer potential sites for conjugation. For example,s ite-specific conjugation with modified glycans, commonly used for intact antibodies, is not possible for antibody fragments because the C H 2d omain region is no longerp resent. [27] The majority of routes to generating labelledf ragments therefore exploit either the exposed and reactive primary amine groups on lysine residues or the thiol groups of cysteines. For example, activated esters such as N-hydroxysuccinimide (NHS) esters will rapidly react with primary amines at room temperature under am ild pH 7-9 range. Functionalised isothiocyanate (SCN) groups can also be used to react with free amine groups generating stable thiourea conjugates.T he abundance of lysine residues on ag iven fragment can, however,l ead to nonspecific conjugation and ah eterogeneously labelled product. [11] To achieve site-specific conjugation,a nd a more homogenously labelled product, disulfide bonds and cysteine residues can be targeted. [28] Cysteine residues commonly form disulfide bonds within proteins. These disulfide bonds act as an inter-chain linkage forming ab ridge between the heavy and light chains of intact antibodies. The disulfide bridges within larger fragments such as F(ab') 2 or Fabs can be exploited to form free thiol groups for conjugation.F ollowing am ild reduction,apair of cysteine thiols are formed whichc an rapidly react with am aleimide containing label at pH 6.5-7.5 generating at hioether.O ns maller antibodyf ragments, such as nanobodies, diabodies and scFv,t here are no such inter-chain disulfideb ridges, however,f ree cysteiner esidues can be engineered into the fragment. For example, Olafsene tal. developed ad iabodyb earing ad isulfide bond linkage in order to provide free thiol groups for further conjugation. [29] Further functionalisation of free cysteine thiol or lysine amine groups is possible to enable even greaterc ontrol of site-specific conjugation. [30] The conjugation of ar ange of reactive groups such as trans-cyclooctenes, alkynes, azides and tetrazinest of unctionalised amine or thiol residues to facilitate click reactions of complementary radiolabelled precursors can furtheri mprove the rate, efficiency and versatility of the labelling. [31] Similar conjugation methods have been used to radiolabelling of affibodies. Because affibody molecules can be prepared via wellknown chemical peptides ynthesis chelators, fluorescent dyes or radiolabelled groups can be introduced site-specifically on the protein sequence. [13] Such site-specific labellingi st ypically achieved either via chelation of ar adiometal,w here the chelator has been introduced at as pecific site in the sequence, or via reaction of an incorporatedc ysteine residue witha na ppropriately radiolabelled maleimide group, for example. The sitespecific labellingo fs uch affibodies andf ragments is important to ensure that the site of the radiolabel or chelator does not impact the binding, and also to ensure that the fragment is homogenously labelledi nasingle position and well-characterised to better enable clinicaltranslation.
ImmunoPET is playing an important role in cancer diagnosis, staginga nd monitoring of treatmentr esponse. Conventional biopsyd etectionm ethods are invasive and can cause significant discomfort to the patient. Biopsies can also suffer from diagnostic inaccuracies due to heterogeneous natureo fr eceptors within atumour mass. PET imaging by comparison is minimally invasive and can provide more accurate quantitative information about the primary tumour mass and secondary lesions. Relative to [ 18 F]FDG PET,w hich monitors metabolic uptake, ImmunoPET provides biomarker information on ad isease by directly targeting specific receptors. An umber of recents tudiesh ave shown the potential of antibody fragmentbased ImmunoPET for detecting ab road range of diseases. The mostc ommona pplication to date has been the imaging of overexpressed antigen-associated tumours; the human epidermal growth factor receptor (HER1, HER2 and HER3) being the most widely investigated. The detection of T-cells have also been investigated by specifically targeting T-cell receptors for the detection of inflammatory relatedc onditions such graft-versus-host disease (MHC class II), inflammatory bowel disease (CD4) and inflammatory responses to hematopoietic stem cell transplant (CD4, CD8). Cardiovascular diseases, such as atherosclerosis, have also been targeted and evaluated by detecting the vascular cell adhesion molecule (VCAM-1). An emerging area of ImmunoPET imaging is the direct targeting of pathogen specific antigens on viruses, bacteria and fungi. The growing area of antibody drugs also makes ImmunoPET an excellent diagnostic companion for monitoring the efficacy of these antibody-based treatments.
3.1. Non-metals: labelling with 18 F, 124 Iand 11 C Fluorine-18 is the most widely used PET radionuclide owing to its high positron yield, low positrone nergy,a pproximate two hour half-life and routine cyclotron production via proton bombardment of [ 18  directly radiolabel antibody fragments using various 18 F-prosthetic groups, thus avoiding the harsh direct labelling conditions. The caveat of prostheticg roup labelling is the additional time and complexity required to preparea nd purify these groups for further conjugation.S everal preliminary studies of 18 F-labelling of F(ab') 2 fragments were achieved via conjugation reactions of lysines with para-[ 18 F]fluorophenacyl bromide ([ 18 F]FPB, Scheme 1a nd Figure 2) [32] and the NHS activated ester N-succinimidyl-4-[ 18 F]fluorobenzoate ([ 18 F]SFB, Scheme 2 and Figure 2). [33] More recently diabodies [9] ands cFvs [34] have been labelledu sing [ 18 F]SFB. N-2-(4-[ 18 F]fluoro-benzamido)ethyl]maleimide ([ 18 F]FBEM,S cheme2)i sa lso an effective reagent for taggingt hiol groups that has been used to develop aH ER2-bindingc ysteine rich affibody molecule. [35] The presence of cysteine in the C-terminus of the HER2 affibody molecule Z HER2:2891 has been exploited for labelling by three methods:s ilicon-fluoride acceptor approach [ 18 [36,37] More recent labellings trategies have focused on conjugation reactions based on click chemistry that occur more efficiently and at much faster rates. This has been more widely adopted for affibody labelling where reactive cysteine residues can be incorporated into the sequence that can enable sitespecificl abelling. Aw ide range of click reactions have been investigated for the bioconjugation of radiolabelled prosthetic groups to biomolecules, the most commona pproaches being: 1,3-dipolar azide-alkyne cycloaddition catalysed with copper, strain-promoted azide-alkynec ycloaddition( SPAAC), Staudinger ligation, and the inverse electron demandD iels-Alder (IEDDA). [31] The IEDDA reaction, [38] using a1 ,2,4,5-tetrazine (Tz)/ trans-cyclooctene( TCO) pair,h as risen in prominence to become ap owerful and convenient route for both modifying and labelling proteins. This [4 + 2] cycloaddition, which eliminates am olecule of dinitrogen and generates an ew six-membered ring (Scheme3), fulfills an umber of key criteria in terms of rapid reactionr ates,s electivity and biocompatibility which are essential for radiolabelling applications.S ince the first ap-  plication of Tz/TCO for 18 F-labelling, [39] ar ange of 18 F-Tz-and TCO-labelled molecules have been reported to facilitate protein andp eptide labelling. [38] Ar ecent elegant example of this Tz/TCO approach was used to radiolabel an anobody.A na minooxy-tetrazine was first reactedw ith [ 18 F]2-deoxy-2-fluoroglucose ([ 18 F]FDG) to generate the 18 F-labelled tetrazine (Scheme 4). [40] Reaction with [ 18 F]FDG is ac onvenientw ay to introduce 18 F, as it is produced daily at many PET centers and thus is both readily available and avoidsm ore complicatedl abelling methods. The nanobody (V H H) wasm odified at its Cterminus with as ortase-recognition motif that was site-specifically modified with a( Gly) 3 [41] involves the reactiono fabifunctional chelator with [ 18 F]AlF (generated in situ from 18 F À and AlCl 3 ), followed by ac onjugation step. Alternatively, the chelate can be conjugated to the vector molecule first, followed by direct[ 18 F]AlF labelling. Typically,[ 18 F]AlF is chelated to functionalised 1,4,7-triazacyclononane-1,4,diacetic acid (NODA) derivate at 100 8Cf or 15 min at pH 4( Scheme 5, Figure 3). The key advantages of the [ 18 F]AlF methoda re the improved efficiencies (RCY and molar activities) and simplification of the labellingp rocess. NODA route has been used to successfully labelarange of F(ab') 2 ,F ab, diabody and affibody fragments using both maleimide and tetrazine functionalised chelates (Figure 4). [36,42,43] The elevated temperatures required to effect this chelation can affect the integrity of proteinsi fadirect labelling protocoli su sed. Recently,B ormans and co-workers have overcome this issue with their (AE)-H3RESCA chelator (Scheme 6). [44] They were ablet os ynthesise [ 18 F]AlF labelled nanobodies and affibodies at room temperature that had been prefunctionalised with their chelator,a chieving RCYs similart ot hose of previously reported NOTAvariations.
Scheme4.Reaction[ 18 F]FDG with an aminooxy-modified tetrazine to generate a 18 F-labelled Tz. Reactiono ft he 18 F-labelled Tz with aT CO-modified singledomain antibody fragment( V H H) for 20 min generates an 18 F-radolabelling nanobodyf ollowing size-exclusion chromatography. [40] Scheme5.Preparation of [ 18 F]AlF-NODA-Tzv ia reaction of the NODA-functionalised tetrazine with [ 18 F]{AlF} 2 + ,g enerated in situ from AlCl 3 and 18 F À . [36,42,43]  124 Iodine (t1 = 2 % 4days) is al ongerl ived radionuclide that is well matched for the labelling of antibodies andl arger fragments.L abelling with 124 Ii st ypically achieved via electrophilic substitution reactions with suitablel eaving groups or activated phenylr ings to generate ac ovalent 124 I-carbon bond. [24] Iodination reagents, such as Iodogen and chloroamine-T, are able to oxidise [ 124 I]NaI to generate 124 I + in situ whichc an then iodinate as uitability activated aromatic rings. The method is sufficiently mild to directly radiolabel proteins at low temperatures, and is frequently used to label the phenolr ing of tyrosine. Such radioiodinations cannot however be used to label specific phenolic or tyrosine groups within the protein. Several minibodies for the imagingo fP CSA and CD20 have been developed by Wu and co-workers, [46][47][48][49] wherein they compared the performance of 124 Ii odinated fragments to that of 89 Zr and 64 Cu versions.I odine labelled fragments are more susceptible to degradation and deiodinationi nv ivo, releasing free [ 124 I]iodideo r[ 124 I]iodotyrosine. This can lead to loss of signal in the region of interesto vert ime. However,b ecause iodine is able to diffuse out of the tissue, al ower background signal can lead to improved images relative to radiometal-based tracers. Alternatively,anumber of 124 I-prosthetic groups have also been developed that are suitable for mild and selectiveprotein labelling, one example is N-succinimidyl-3-(4-hydroxy-5-[ 124 I]iodophenyl)propionate ([ 124 I]SHPP), which can react with the amineso fl ysine groups (Scheme 7). This route has been used to radiolabel aH ER2 targeting diabody;h owever,t he [ 124 I]SHPP labelling method wass hown to decrease the immunreactivity of this fragment relative to the 124 Il abelledv ersion using the Iodogen method. [50] This likely depends on the specificm acromolecule as other examples exist in the intact antibody literature wherein the [ 124 I]SHPP labelling method is preferred. [51] The anti-HER2 affibody Z HER2:342 has been labelled with 124 I using as imilart ypeo fp rosthetic group, N-succinimidyl-para-[ 124 I]iodobenzoate( [ 124 I]SPIB). [52] In ac omparative study with [ 124 I]trastuzumab, the total uptake radioactivity of [ 124 I]trastuzumabi nt umours was found to be highert han [ 124 I]SPIB-Z HER2:342 ,h owever,t umour-to-organ ratios were lower.
[ 124 I]SPIB-Z HER2:342 showed ar apid clearance of radioactivity from blooda nd organs,a nd therefore gives better contrast than the intact antibody.T he inherentlimitations of 124 Ii nclude low positron yield which affects sensitivity,h igh energy positron emission giving lower resolution and higher gammae missions which increase patient doses.E ven with these limitations longer-lived tracers,s uch as 124 I, continue to be important for PET imaging. The pairing of imaging isotopes such as 124 Iw ith am atched therapeutic partners uch as 131 I( for radiotherapy) will also become more important for combined imaging and therapy.

Zr and 44 Sc
Radiometalsw ith longerh alf-lives, such as 64 Cu (t1 = 2 = 12.7 h) and 111 In (t1 = 2 = 2.8 days), have been used form any years for the labellinga nd imagingo fi ntact antibodies. [55] The half-life of 64 Cu also makes it suitable fori maging largerf ragments such as minibodies,( Fab') 2, Fab and diabodies that have slower clearance rates.K ey pioneering studies of antibody fragments for PET imaging have been conducted using 64 Cu. [29,56] 1,4,7,10tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) and 1,4,7-triazacyclononane-1,4,7-triacetic acid (NOTA) chelation methods have been the mainstay for 64 Cu labelling of antibody fragments. [57][58][59][60] The 12.7 hh alf-life relieves the time pressures for labelling, permitting extended reaction times of % 60 min to ensurec omplete chelation of DOTAd erivatives. 64 Cu-DOTA labelleda ntibodies and antibody fragments have shown high circulation radioactivity and nonspecific uptake in kidney,l iver, and spleen. [29,61] This is partially due to competition for 64 Cu from endogenousp roteins that are able to strongly chelate copper. [60,62] NOTAc helators have been found to largely circumventt his issue, formingm uch more stable copperc omplexes within shorter labelling times, displaying decreased nonspecific uptake in the liver and spleen. [63,64] For these reasons there has been as hift toward using NOTAa st he chelator of choice for 64 Cu-labelling of fragments. Cai and co-workers [65] recently reported the 64 Cu-labelling of ab ispecific fragment using aN ODA chelator and used it for imaging CD105 and tissue factor (TF) in mice bearing pancreatic cancerx enographs. The same fragment was also labelled with af luorescent tag, to generate ad ual labelledp robe, that also enabled near-infrared fluorescence( NIRF) imaging. The labelledh eterodimer probe showed increased tumour uptake relativet ot he homodimer and the fluorescence imaging was used to validated the PET imaging and to show delineation of the tumours. Such dual-modalityo ptical-PET imaging is advantageous over single modality imaging, as the much highers patial resolution of opticali magingm ethods relative to PET can aid in more accuratelyl ocalising the probe. Dual optical-PET and optical fragment-based probes may further helpinthe diagnosis and staging of cancer,a sw ell as in providing greater delineationo f healthyand cancerous tissue during surgery. [66][67][68] Recently,s horter-lived radiometalss uch as 68 Ga (t1 = 2 = 67.7 min) and 44 Sc (t1 = 2 = 3.9 h), with half-lives more closely matching the pharmacokinetics of antibodyf ragments, have garnered closer attention for fragment labelling. 68 Ga is particularly convenient for radiolabelling because it is produced via a 68 Ge/ 68 Ga generatora nd thusc ircumventst he need (and associated costs) of ac yclotron facility.Ac ommon approach for 68 Ga-labelled is reaction with the widely used DOTAa nd NOTA, [69] however,arange of other bifunctional chelators have been investigated. [70] The FDA approved 68 GaDOTA-TATEt racer for somatostatin receptor positive neuroendocrine tumours is an example of its growing importance for clinicalu se. [71] 68 Ga-DOTAl abelling, however,t ypicallyr equires elevated temperatures to ensure high yielding and rapid incorporation;t herefore direct labelling of heat sensitive proteins with 68 Ga is not typicallya chieved with DOTAc helators.N OTAc helators are more suitable for direct 68 Ga-labelling of antibody fragments. Both the thermodynamic stabilityG a III -NOTAc omplexes (logK = 31.0) and the kinetics of chelation are superior to that of Ga III -DOTA( logK = 21.3). [70] 68 Ga-NOTAl abelling can therefore be achieved at room temperature within much shorter reaction times to generate complexes that are typicallymuch more stable in vivo. Ar ange of Fabs, nanobodies and affibodies have been successfully labelledw ith 68 Ga using the cyclic DOTA [14,[72][73][74] and NOTA [75][76][77] chelators ( Figure 5). Recently,a HER2 specific 68 Ga-NOTA-Bn-SCN-Nanobody underwent phase I clinicaltrials and displayed high uptake in metastases. [76] Desferrioxamine B( DFO), an acyclicc helator and naturally occurring siderophore, hasb een widely used to radiolabelled antibodies. The acyclic and multidentate nature of this ligand enables the rapid and stable formation of Ga-chelates under mild reactionc onditions, typically within 5min at room temperature. Several 68 Ga-DFO nanobody and scFv fragments have been reported for imagingE GFR andH ER2 (Figure6). [78,79] There are, however,s ome concerns over the stabilityo f 68 Ga-DFO complexes and in theirl abellinge fficiencies relative to 68 Ga-NOTAc omplexes. [79] N,N'-Bis(2-hydroxybenzyl)ethylenediamine-N,N'-diacetic acid (HBED) is another acyclic chelator that enablesr apid 68 Ga-labellinga tr oom temperature. The HBED-CC derivative has been used for rapid and room temperature labellingo fr ecombinant scFvs and diabodies ( Figure 6). [80,81] A tris(hydroxypyridinone) (THP) bifunctional chelator has recently been used to label scFv of J591, (Figure 6) for prostate-specific membrane antigen (PSMA)i maging. [82] The THP derivative (THP-mal) was coupled the C-terminus of ac ysteiner esidue of the scFv via am aleimide linker and labelled 68 Ga at room temperaturea nd neutralp Ha chieving RCY > 95 %w ithout further purification. The THP ligand systemi sp roving to be ah ighly effective chelator for 68 Ga that shows excellent stability. [83] Zirconium-89( t1 = 2 % 78 h) is al ongerl ived PET radionuclide that has found applications for the labelling of intact antibodies and forl arger antibody fragments such as minibodies and diabodies. The DFO ligand is commonly used to chelate 89 Zr IV , however,s tudies have demonstrated 89 Zr-DFO complexes to be unstablei nv ivo, resulting in 89 Zr releasea nd accumulation of 89 Zr in bone. [23] Despite this, DFO has been mostly used for fragment labelling due to its high chelation yields, mild reac- Figure 5. Cyclic chelators DOTA [14,[72][73][74] and NOTA [75][76][77] that have been conjugated to antibody fragments or affibodies and radiolabelled with gallium-68.  89 Zr labelling of PMSA selective minibody and diabody fragments. [84] Labelling was achieved in both cases with good RCY (> 70 %) via chelation with DFO modified fragments. In comparison with the intact antibody,a lso labelled with 89 Zr-DFO, both fragments displayed faster tumour delineation and background clearance in tumour-bearing mice, thus demonstrating their potentiala sp robesf or the detection and staging of PSMA-positivet umours. A[ 89 Zr]ZrDFO-Cys-diabody has recently been developedf or trackinge ndogenous CD8 + T-cells which couldb eu sed to evaluate the tumour immune responseo fnovel immunotherapies. [85] The Cys-diabody was conjugated to am aleimide-DFO via the engineered C-terminal cysteine of the fragment and labelled with 89 Zr in high RCY.T he labelled diabody showeds pecific targeting to CD8 in tumour-bearing mice. An interesting nanobody construct, composed of three individual nanobodies, that is capable of binding albumina nd two different epitopes of HER3 has also been labelledv ia 89 Zr via DFO chelation. [86] It has been previously found that two anti-HER3 antibodies inhibited tumourc ell growth better than each antibodyi ndependently. The labelledc onstruct showed an uptake correlation with HER3 expression in tumour-bearing mice, and enabled tumour visualisation for up 96 hp ost injection. A 89 Zr-DFO labelled minibody fragment has recently undergone ap hase Ic linical trial for imaging metastatic prostatec ancer. [87] The labelled minibody proved to be safe and showedf avourableb iodistribution for imaging metastatic prostate cancer.T he limitations of 89 Zr-labelling, include the low positrona bundance (23 %) and long half-life which mean that patients are potentially exposed higherd oses of radioactivity.R ecent studies have demonstrated that 89 Zr-DOTAd erivatives show greaters tability, however,c helation requires heating at > 90 8Cw hich could not be applied to the direct labellingw ith fragments. [88] Scandium-44i sr eceiving current interest owing to its intermediate 3.97 hh alf-life, high positron yield, route of production andc oordination chemistry. [89] It is typically produced from 44 Ca in ac yclotron, [90] but hast he potential to be more conveniently produced via a 44 Ti/ 44 Sc generators ystem. [91] 44 Sc, which is considered to be ap otential alternative to 68 Ga,d isplays similar coordination chemistry to 68 Ga and could be used to extend the PET scanning window which may resulti nb etter images for certain probes. [92] The longer 3.97 hh alf-life is particularly well matched for antibody fragment imaging studies. The vast majority of 44 Sc labelling has been focused on DOTA functionalised peptides, however,c helation of Sc III to DOTAr equires high temperature duet oi ts slow reaction kinetics which is unfavourable for protein labelling. An affibody molecule, Z HER2:2891 ,h as been recently labelledw ith 44 Sc using an Nterminal conjugatedD OTAc helator. [93] High RCY were achieved via reaction with 44 ScCl 3 at 95 8Cf or 30 min. The [ 44 Sc]Sc-DOTA-Z HER2:2891 conjugate is promising and displays specific binding to HER2-expressing cells, and high-contrast for the imaging of tumour-bearing mice. Cai and co-workers [94] reported 44 Sc labelling of ac etuximab Fab fragment modified with ad iaminepentaacetic acid (DTPA) chelator ([ 44 Sc]CHX-A''-DTPA-cetuximab-Fab, Scheme 9) which enabled radiolabelling within 30 min, under optimisedc onditions. The labelledf ragment displayedg ood stability in mouse serum, specific uptake in tumour-bearing mice and rapid renal clearance.

Summary and Outlook
Ar ange of PET radionuclides have been used to generate an array of labelledf ragments and affibodies for imaging studies; key labelling methods, targets and imaging applicationsa re Figure 6. The acyclic chelatorsT HP,H BED and DFO that have been conjugated to antibody fragments and radiolabelled with gallium-68. [79][80][81][82] Scheme9. 44 Sc-labelling of aFab fragment using athiourea-conjugated DTPAc helator. [94] ChemMedChem  Tables 2a nd 3. The short-to-medium lived radionuclides 18 F, 68 Ga, 124 Ia nd 64 Cu are currently the most widely used for fragment labelling studies owing to their availability and half-lives that match the biological clearance rates of fragments.L abelling methods have evolvedc onsiderably over the past ten years. There is now ad iverser ange of prosthetic groups and conjugation methods that can be used to reliably label fragments, although there is room for improvement in terms of labelling timeframes, radiochemicaly ields and simplification of labellingp rocedures. Key considerations for fragment labellingi nclude matching the radionuclide to the pharmacokinetics of the fragment, labelling rapidly under mild conditions to ensuref ragment integrity and labelling at specific sites within the fragment to ensure that the immunoreactivity of the fragment with its target is not compromised. Lysine and cysteineb ioconjugation methods are now well-established for fragment and affibody labelling. Newerb ioconjugation strategies, such as the IEDDA reaction (using Tz/TCO), offer improvedr ates of reactionu nder milder conditions but require more complex synthetic steps to prepare appropriate fragment conjugates and labelling partners.I mproved chelator chemistry has also evolved for rapid and mild sequestering of radiometals. THP and NOTAd erivatives are two key examples of chelators that show the greatest kinetic and thermodynamics tabilities, in addition to mild and rapid chelation reaction conditions. The establishment of novel labellingr outes such as the [ 18 F]AlF labelling, greater access to existing PET radionuclides (e.g.,g eneratorp roduced 68 Ga) and access to new radionuclides with favourable physical characteristics for labelling (e.g., 44 Sc) have also enabled the development of fragmentbased imaging agents. As new fragments are developed for both therapeutic and imaging, new mild, rapid, site-specific and robust labelling methods will aid in the understanding of their biological behaviour and facilitate clinicalt ranslation. Inevitably,r esearch in fundamental chemistry andi ts translation can profoundly impact imaging chemistry.T he development of new bioconjugation and chelation methods are likely to be key to improvingf ragment labelling protocols, and ultimately impact clinical use.
The development of labelledf ragments and affibodies is withoutd oubt al ong, challenging and expensive task that requires iterative stages of protein engineering, selection of a fragment/affibody,a ppropriate radiolabellinga nd preclinical testing. For clinical translation there are regulatory hurdles and challenges in scaling-up of the fragment/affibody production process. The timeframes and cost of developing an ovel antibody fragment/affibody imaging agent may,h owever,b er educed compared with full antibody agents, due to their more rapid production,s election and characterisation methods. There is enormousp otential and many opportunities fori maging with antibody fragments and affibodies based on their high affinities, specific targeting and fast clearance rates. The non-invasive imaging of cell-surface antigens forc ancerd etection has been the main focus to date and there has been ajustified emphasis on targeting EGFRs (HER1-HER3) owing to their over-expression in aw ide range of cancers. Future challenges in this area involve improving the specificity of fragmentbased probest ot he HER family in order to better select patients for receptor-targeted therapy and monitoring of therapeutic response. [95] Encouragingly,s everal labelled fragments are now undergoing clinical trials as canceri maging agents. [76,87] An umber of other targets have been investigated (Tables 2a nd 3), and there is much currenti nterest in imaging inflammatoryr e- sponses via the direct targeting of T-cells. It is also expected that labelled fragments will be used to specifically target antigens of pathogenic bacteria and viruses, and thus enable the development of new pathogen-specific tracers to discriminate between infectious and sterile sites of inflammation. Future developmentsi nt his area will depend on the identification of new fragments and affibodies that have appropriate characteristics for imaging (high target affinity,s electivity,s tability,r apid clearance, ease of labelling etc.). The adoption of newer and more site-specific conjugation chemistries is currently underway,a si st he development of theranostic fragments that will make use of complementary radionuclide pairs for both imaging and therapy (e.g., 44 Sc/ 47 Sc, 124 I, 131 I). It is also anticipated that PET imagingw ill play ar ole in the assessment of the burgeoning field of fragment-based antibody drug conjugates (ADCs).