A Review of Approaches to 18F Radiolabelling Affinity Peptide & Proteins

O. Morris1,2, M. Fairclough1,2, J. Grigg3, C. Prenant 1,2 & A. McMahon1,2

1. Wolfson Molecular Imaging Centre, The University of Manchester, UK2. CRUK/EPSRC Imaging Centre in Cambridge & Manchester, The University of Manchester, UK

3. GE Healthcare, Amersham Place, Little Chalfont, Bucks, UK

Abstract

Affinity peptide and protein- (APP) based radiotracers are an increasingly popular class of

radiotracer in positron emission tomography (PET), which was once dominated by the use of

small molecule radiotracers. Radiolabelled monoclonal antibodies (mAbs) are important

examples of APPs yet a preference for smaller APPs, which exhibit fast pharmacokinetics

and permit rapid PET aided diagnosis, has become apparent. 18F exhibits favourable physical

characteristics for APP radiolabelling and has been described as an ideal PET radionuclide.

Notwithstanding, 18F radiolabelling of APP is challenging and this is echoed in the literature

where a number of diverse approaches have been adopted. This review seeks to assess and

compare the approaches taken to 18F APP radiolabelling with the intention of highlighting

trends within this expanding field. Generic themes have emerged in the literature, namely the

use of mild radiolabelling conditions, a preference of site-specific methodologies with an

impetus for short, automated procedures which produce high-yielding [18F]APPs.

Introduction

The field of PET was dominated by the use of radiolabelled small molecules. Over recent

decades, however, radiolabelled affinity peptides & proteins (APPs) have become an

important class of radiotracer including radiolabelled somatostatin analogues and octreotides

[1]. The field of 18F APP radiochemistry is diverse and dynamic yet, despite the extensive

literature reporting 18F APP radiolabelling methods, very few have progressed to clinical-

trials or gained widespread use. A review by Stephanopoulos et al. [2] iterates the sentiments

in this review regarding the usefulness of APPs as molecular vehicles in PET and contains a

valuable flow-diagram that highlights the decision process for choosing an appropriate APP

radiolabelling strategy. The current work seeks to build on this topic and evaluate the various

approaches to APP radiolabelling with 18F whilst highlighting trends in the radiochemistry.

Separated into non-site-specific and site-specific methods, the review will comment on the

cost/benefit analysis of the approaches.

Affinity peptides & proteins in PET

The high selectivity and affinity of biological protein-protein interactions initially motivated

their exploitation in biomedical fields. Key developments in the knowledge of APP antigen-

binding, speed of generation using solid-phase and recombinant techniques, as well as

progressions in affinity selection techniques have strengthened their utility in a number of

areas [1]. APP-based pharmaceuticals represent a rapidly growing market in pharmaceutical

industry, despite some challenging absorption, distribution, metabolism and excretion

(ADME) properties of APPs, such as low cell membrane permeability, short biological half-

life and limited stability [3]. The advantages of their use including exquisitely high target

binding affinity and specificity alongside low toxicity and immunogenicity have

overwhelmed the challenges associated with their use [3].

Naturally, their utility generated attention in the field of PET, where APPs could be

radiolabelled and their high target selectivity and affinity could be exploited in vivo. The

protein-nature of many disease targets and their natural binding partners, including tumour

cell surface receptors and hallmarks of dementia such as tau and amyloid plaque, also

reinforces the use of radiolabelled APPs in the disease diagnoses and highlights their utility in

both peripheral and neurological pathologies [1].

Monoclonal antibodies (mAbs) (~150 kDa) are an important subclass of immunoglobulin; the

extensive use of mAbs is evident in the number of therapeutic antibodies that are available

and used in a clinical setting, with many more in the latter stages of clinical development [4].

The popularity of mAbs is credited to their naturally-evolved target selectivity and specificity

[4]; their ubiquity in other areas of medicine led to mAbs filtering across into the field of

nuclear medicine and literature as early as the 1950s can be found, documenting the use of

antibodies in oncology radiolabelled with iodine-131, a therapeutic radionuclide [5]. Koehler

et al. in 1975 [6] reported an in vitro method for the generation of high volumes of target-

specific antibodies and described the importance of this method in both medical and

industrial fields. The method is recognised as a key development in the use of radiolabelled

mAbs and more recent achievements in recombinant techniques, permitting humanisation and

chimerisation, have strengthened their clinical utility [7]. Encouraging results using

radiolabelled mAbs in clinical oncology trials have been reported, where their use has

permitted diagnosis, patient stratification and identification of the most appropriate treatment

[8].

The rate of uptake from the blood to the biological target is approximately inversely

proportional to the molecular mass of the APP, consequently target accumulation of mAbs is

relatively slow [9]. For this reason, longer-lived radioisotopes such as 89Zr and 64Cu have been

conventionally used to radiolabel mAbs [10, 11].

The slow biodistribution of radiolabelled mAbs can lead to poor target tissue to background

contrast unless the mAb is given sufficient time to accumulate at its biological target [12]. This

is time and labour intensive and the use of longer lived radioisotopes contribute to a high

patient radiation dose. Alternatively, pre-targeting approaches to mAb radiolabelling permit

the use of shorter-lived radioisotopes, however these methods are still relatively new [13].

The slow biodistribution and requirement for longer-lived radioisotopes are considered to be

significant limitations to the use of radiolabelled mAbs. In addition to this, large proteins

have complex tertiary structures with multi-domain features, formed by glycosylation

patterns and disulphide bridges, resulting in thermal and chemical vulnerability that tightly

constrains the radiolabelling method [14, 15]. The issue of fragility is underlined in the literature,

where the use of physiological temperatures and aqueous buffers for radiolabelling

procedures is consistently described [16-18]. These limitations have complicated the translation

of radiolabelled mAbs to routine clinical practice and, as a result, a change in the landscape

of APP-based radiotracers is evolving. This corresponds to a preference for smaller APPs that

exhibit rapid pharmacokinetics and enhanced robustness, whilst retaining high target affinity

and selectivity [4, 19]. Importantly, the rapid pharmacokinetics of small APPs permits the use of

short lived radioisotopes. The diagnostic value of small APPs has not yet been validated; yet

their exploitation in diagnostic PET has the potential to be a powerful tool helping to guide

patient stratification and identify those who would benefit from mAb-based therapy.

Radionuclide Nuclear equation Max. Positron energy (MeV)

Branching ratio

Half-life

Short lived

68Ga 70Zn(p,n)68Ga

Germanium-68 generator:68Ge/68Ga

1.8 0.81 68.3 mins

18F 18O(p,n)18F 0.6 0.97 109.80 mins

44Sc Titanium-44 generator:44Ti/44Sc

0.6 0.94 3.9 hrs

Long lived

64Cu 64Ni(p,n)64Cu 0.6 0.18 12.7 hrs89Zr 89Y(p,n)89Zr 0.8 0.23 3.30 days

124I 124Te(p,n)124I 1.5 0.23 4.2 daysTable 1) Commonly used radionuclides in APP radiolabelling [20, 21]

Table 1 documents the radioisotopes that have been utilised in APP radiolabelling; the Table

has been separated into short and long lived radionuclides. The majority of the radionuclides

listed are most commonly produced using a cyclotron however 68Ga is normally generator-

produced, which is considered a significant advantage. The wide range of radionuclides

available is highly advantageous in order to match the diverse biological half-lives of

different APPs.

Somatostatin analogues were amongst the first small APPs to be radiolabelled and approved

for clinical use in oncology, in the 1980s [1]. Early APP-based radiotracers used in PET,

including somatostatin analogues, typically made use of radioisotopes such as 68Ga [1]. Use of

18F radiolabelled APP in clinical trials would not be documented for another 20 years and this

has been ascribed to challenges associated with the radiochemistry [1].

Given the prevalence of mAbs, unsurprisingly, the utility of antibody fragments, such as Fab

(antigen-binding fragment, Mr ~50 kDa) and scFv (single-chain variable fragment, Mr ~25

kDa), as APP-based radiotracers have been investigated. However the target affinity of the

parent antibodies was rarely retained and issues of poor solubility and aggregation were

apparent [7]. The loss of affinity observed in mAb fragments was also observed in a anti-

HER2 Fab fragment, derived from trastuzumab, possibly related to the contribution of mAb

bivalency to antigen affinity [22].

Nanobodies (Mr ~15 kDa) are fragments of camelid heavy chain only antibodies and are

alternative immunoglobulin-based APPs. They exhibit high target affinity and selectivity

together with rapid biodistribution, whilst avoiding the issues of poor solubility and

aggregation observed in non-camelid derived antibody fragments [23]. Since the nanobody is

camelid-derived, it is necessary to humanise the APP for use in therapeutic applications [24].

Vaneycken et al.[25] described the use of an anti-HER2 (human epidermal growth factor

receptor 2) nanobody; the ease of nanobody production as well as improved affinity selection

techniques permitted efficient deduction of lead nanobodies that were able to identify HER2

expressing lesions and also regions of metastases [25]. This is indicative of the work that has

underlined the potential of nanobodies and they have thus been labelled the ‘magic bullet’ of

molecular imaging and are predicted to have an important future role in the field [23].

Advances in APP engineering have facilitated the arrival of non-immunoglobulin based APP

[4]. Designed using combinatorial protein libraries, non-immunoglobulin based APPs have

been engineered to possess characteristics, such as enhanced solubility or stability that

challenge some of the limitations of immunoglobulin-based APPs [4, 14]. The affibody

molecule is a promising example of a non-immunoglobulin APP and originated from

staphylococcal protein A (SPA). The affibody is derived from the B-domain: one of the five

domains that make up SPA. Its potential as a radiotracer was recognised based on its high

stability and aqueous solubility. Key mutations were subsequently made that afforded

enhanced stability of the newly named Z-domain; it is upon this protein-scaffold backbone

that more than 60 target-specific affibodies have been constructed [14]. Literature cited in this

review, describing the use of affibodies in oncology, unanimously reported successful and

rapid tumour targeting alongside fast clearance from non-target tissue [26-28]. Indeed, Kramer-

Marek et al. [27] reported tumour uptake as early as 20 minutes post-injection, thus stressing

the potential of the APP in rapid PET-guided diagnosis. It should, however, be noted that

tumour accumulation at tumour points exhibits lower image contrast due to accumulation of

the radiotracer in excretory organs.

Great variety in the characteristics of APPs used in PET exists: from large immunoglobulin-

based mAbs to short-polypeptide chains such as anti-αvβ3 RGD motif-containing APPs used

as angiogenesis radiotracers. Further tailoring of APP characteristics is reported and

conjugation of biomolecules, such as poly-histidine and albumin, [3, 28] or inert polymers, such

polyethylene-glycol (PEG), used to manipulate the in vivo behaviour of the APP. Kalia and

Raines [29] reported the enhanced water-solubility, reduced immunogenicity and enhanced

biological half-life of PEGylated APPs.

The characteristics of smaller APPs described above are profitable attributes in PET; a key

advantage is their compatibility with short lived radioisotopes, in particular 18F which has

been described as an ideal PET radionuclide [1, 30]. However, as previously discussed, APP

radiolabelling with 18F can be challenging and this was outlined in a review article by Richter

et al. [1] where the incompatibility of direct fluorination and lack of automation of complex

multi-step methods was said to have impeded the translation of 18F radiolabelled APP to

clinical practice [1]. The review stresses the importance of process automation in aiding

conformance with GMP but at the same time highlighted the difficulties associated with

automation of multi-step approaches required for APP radiolabelling with 18F [1]. The

importance of process automation is reflected in the successes of 68Ga radiolabelled APPs,

where all-in-one, self-shielded, fully-automated radiosynthesisers are used to prepare the

radiotracers, stimulating an exponential growth in their use [31, 32]. Gallium-68 radiochemistry

lends itself more favourably to process automation due to one-step chelation methods, unlike

the current multi-step approaches to 18F radiolabelling. However, the advantageous physical

characteristics of 18F (Table 1) make the automation of 18F radiolabelling approaches, a

profitable endeavour [20, 33].

[18F]Fluoride

The biological half-life of a small or intermediately sized APP is complemented by a t1/2

=109.8 min, 18F radiolabel (≤ 60 kDa).

The introduction of fluorine into a molecule, at the outset, initiated much debate given the

limited natural abundance of fluorine in biological molecules [30]. It was argued that its

addition may alter the physical characteristics of any tracer in view of the electronegative

character of fluoride. It has, however, been found that its electronic properties can lead to

improved pharmacological characteristics and the strength of the C-F bond (112 kcal/mol), in

comparison to the C-H bond (98 kcal/mol), enhances its metabolic stability [30]. The small

size of fluorine (van der Waals radius 1.47 Å) bears similarity to hydrogen (1.2 Å) and

electronic nature is comparable to a hydroxyl group [30]. This, therefore, provides scope for

their replacement in a whole host of molecules, where fluoride can be described as a

bioisostere [34].

The highly favourable decay characteristics of 18F permit more complex radiochemistry and

longer imaging studies [1, 30]. This is of particular importance when radiolabelling APPs,

which frequently necessitates multi-step radiosynthetic strategies as direct fluorination of the

APP is often not possible due to the necessary harsher radiolabelling conditions and multiple

fluorination sites of the APP. The half-life also enables transportation to different clinical

locations; this is important as it permits off-site radiotracer synthesis.

Non-site-specific radiolabelling

Prosthetic groups are small molecules that conjugate to the APP under mild conditions, they

are 18F-fluorinated before conjugation to the APP thereby avoiding issues arising from the

harsher reaction conditions often required for direct fluorinations [35, 36].

Non-site-specific radiolabelling is commonly achieved using carbonyl-containing 18F labelled

prosthetic groups which can target the N-terminal or ε-amine belonging to lysine residues and

thus making use of the inherent nucleophilic sites of the APP. These nucleophilic sites have

been identified in Figure 1, which provides an example APP and where the N-terminus and

lysine ε-amines have been marked with a dashed ring.

Table 2 provides a summary of commonly used 18F prosthetic groups, the Table summarises

key features of prosthetic groups radiosynthesis allowing comparisons to be drawn. The

authors refer readers to a review by Schirrmacher et al. [37] for a more detailed review of small

prosthetic groups used in 18F radiochemistry.

Reductive alkylation, using amine-reactive prosthetic groups, is a commonly employed

strategy. It involves the formation of an imine (or Schiff base) which is subsequently reduced

with an appropriate reducing agent. [18F]fluorobenzaldehyde (FBA) is a popular example of

amine-reactive 18F prosthetic groups used in reductive alkylation [38]. More recently

[18F]fluoroacetaldehyde ([18F]FA) and [18F]fluoropyridinecarboxaldehyde ([18F]FPCA) were

reported, as water-soluble prosthetic group alternatives [39-42].

Figure 2 gives the mechanism of reductive alkylation in both acidic and basic conditions.

Reductive alkylation proceeds in both pH ranges but acidic conditions appear to be superior

Figure 1) Example of an APP where N-terminal or lysine ε-amines have been identified with dashed rings

[38, 39, 41]. The instability of the imine and production of H2O or hydroxyl group, which is

thermodynamically disfavoured in aqueous media, is overcome upon addition of an

appropriate reducing agent. NaCNBH3 is often used as a reducing agent on account of its mild

reducing properties, stability in water and compatibility with the pH range required for imine

formation [43].

Non-site-specific radiolabelling: [18F]Fluorobenzaldehyde ([18F]FBA)

[18F]FBA is a popular 18F prosthetic group and is generally synthesised by direct fluorination

of the commercially-available 4-formyl-N-N-N-trimethylammonium-triflate, this can be seen

in Figure 3.

The presence of an ortho/para (o/p) electron withdrawing group and powerful leaving group

permits nucleophilic aromatic substitution of the precursor, which without these elements

A

B

Figure 2) Reaction mechanism showing reductive alkylation in A) acidic and B) basic media

would not be susceptible to nucleophilic attack. Many of the arene-based 18F prosthetic

groups make use of a trimethylammonium precursor owing to the solubility of ammonium

salts, ease of separation of the 18F-labelled prosthetic group from the ammonium salt

precursor and stability of the trimethylamine leaving group. [18F]Fluoromethane has been

reported as a possible side-product, however its generation is disfavoured when substituents

with large Hammett constants are present on the arene in o/p positions [44].

Speranza et al. [45] report the automation of [18F]FBA radiosynthesis using a GE TRACERlab

FX-FN. Automated radiosynthesis of [18F]FBA was completed in 45 minutes and

radiochemical yields (RCY) of 90 % decay-corrected (d.c.) were reported [45]. The method

made use of SPE purification, rather than HPLC, for a number of reasons including ease of

automation, time-saving aspects and reproducibility [45]. Importantly the group report

comparable [18F]FBA recovery between the two methods and [18F]FBA radiochemical purity

(RCP) measurements of > 99 % were described [45]. The group insist on the importance of

automation in order to provide a reliable tool to produce large quantities of [18F]FBA to

radiolabel APPs for use in routine clinic [45].

Apana et al. [38] report the radiolabelling of an APP via reductive alkylation with [18F]FBA.

The APP, anginex, is a 33-residue peptide with anti-angiogenic properties and has been

reported to sensitise tumours to radiotherapy and inhibit tumour growth; [18F]anginex was

Figure 3) Reaction scheme to show [18F]FBA radiosynthesis

evaluated as an anti-angiogenic radiotracer in pre-clinical models [38]. [18F]Anginex yields of

76% were reported alongside molar activity measurements of 14.4 GBq/μmol [38].

[18F]Anginex was evaluated in tumour bearing mice and, although high liver uptake was

observed, tumour uptake of [18F]anginex was 16 % higher than the maximal liver uptake and

accounted for 30 % of the total body distribution. The authors concluded that [18F]anginex

had retained target selectivity, achieved high target to background ratios and is thus a

potentially useful antiangiogenic diagnostic radiotracer [38]. The yield and molar activity of

[18F]anginex were significantly improved by removal of polytetrafluoroethylene (PTFE)

which is a material used in tubing and as a surface coating for various components of

radiochemistry synthesisers. PFTE derived stable fluoride has a carrier effect and its removal

yielded ~25 % improvement in [18F]anginex yields [38, 46]. PTFE removal from the

radiosynthesis of [18F]FBA proved to be a profitable modification; the average mass of carrier

fluoride was calculated to be between 400-800 nmol, which was reduced to 25 nmol after its

elimination [38].

Non-site-specific radiolabelling: [18F]Fluoroacetaldehyde ([18F]FA)

[18F]FA is a small, water-soluble prosthetic group reported by Prenant et al. [42] and used to

radiolabel a recombinant human interleukin-1 receptor antagonist (rhIL1RA, 17.5 kDa) via

reductive alkylation [41]. A remotely-controlled semi-automated configuration was used for

[18F]FA radiosynthesis and subsequent rhIL1RA radiolabelling; the method achieved [18F]FA

yields of 34 ± 3 % (d.c., n = 3) and overall [18F]rhIL1RA RCY of 11.4 ± 4 % (d.c., n = 17) in

100 minutes [41, 42]. The non-site-specificity of reductive alkylation is a drawback of this

radiolabelling approach [41]. Nonetheless, ex-vivo binding assays revealed retention of

[18F]rhIL1RA binding specificity and it was thought that the small size of [18F]FA was, in

part, responsible [41].

Full process automation of [18F]FA radiosynthesis and subsequent APP radiolabelling was

later reported, where rhIL1RA was radiolabelled via reductive alkylation, in order to make

direct comparisons between the semi- and fully- automated procedures [39]. The automation

permitted larger starting quantities of [18F]fluoride which contributed to larger quantities of

[18F]FA, despite some observed loss in [18F]FA yields (~25 % loss) [39]. Overall [18F]rhIL1RA

yields of 5 ± 3 % (d.c., n = 5) and apparent molar activity measurements of 13.5 GBq/μmol

were reported. [18F]rhIL1RA apparent molar activity measurements reported using the fully-

automated radiosynthesis [39] are significantly higher than measurements described by Prenant

et al. [41] (0.9 – 5.0 GBq/μmol) despite the inability of both methods to separate the product

from unlabelled rhIL1RA. This was largely attributed to the ability to use higher starting

levels (15-fold higher) of radioactivity [39, 41, 42].

Non-site-specific radiolabelling: [18F]Fluoropyridinecarboxaldehyde ([18F]FPCA)

The ease with which pyridine can be fluorinated at both the C2 and C4 positions, attributable

to the presence of the nitrogen ‘electron-sink’, has led to an increasing number of pyridine-

based prosthetic groups [40, 47, 48]. One such example is [18F]FPCA, a water-soluble 18F

prosthetic group developed by Morris et al. [40]. Its radiosynthesis has been fully-automated

using a customised GE Tracerlab FX-FN with [18F]FPCA RCY of 28 ± 2 % (d.c., n = 10)

reported [40]. Morris et al. [40] describe the replaced purification of [18F]FPCA with SPE for

time-saving reasons and to simplify automation and improve reproducibility [40], as was

reported by Speranza et al. [45] for the purification [18F]FBA. The automation of [18F]FPCA

was thereby extended to include the subsequent radiolabelling of an APP.

Non-site-specific radiolabelling: N-Succinimidyl-4-[18F]fluorobenzoate ([18F]SFB)

[18F]SFB, similarly to [18F]FPCA, radiolabels APPs through conjugation with the inherent

nucleophilic sites of the APP. Unlike the other amine-reactive prosthetic groups described,

[18F]SFB has the advantage that it does not require a reducing agent. Alkylation of APPs with

[18F]SFB is commonly carried out between pH 8 and 9 to deprotonate APP amine groups [49].

Automation of [18F]SFB radiosynthesis has been described by Zijlstra et al. [49]; in brief:

[18F]SFB was produced by first synthesising ethyl 4-[18F]fluorobenzoate through fluorination

of a trimethylammonium-precursor [49]. 4-[18F]Fluorobenzoic acid was then produced by

base hydrolysis of the fluorinated ester [49]. In the final step, a succinimide-containing synthon

(O-[N-succinimidyl]-N,N,N,N-tetramethyluronium tetrafluoroborate) (TSTU) was incubated

with 4-[18F]fluorobenzoic acid to yield [18F]SFB in yields of ≤ 35 % (d.c.) and RCP of 98 %

[49].

High RCP values were reported despite having to reuse SPE cartridges owing to limitations

of the automated radiochemistry module [49]. [18F]SFB was used to radiolabel the apoptosis

radiotracer, annexin-V (30 kDa), and RCY of ≤ 20 % (d.c.) were reported [49]. In vitro

experiments using apoptotic cells verified the selectivity of [18F]annexin-V and cell uptake of

the radiotracer was dose dependent in experiments where varying concentrations of

unradiolabelled annexin-V were used, thereby verifying retention of the biological activity of

[18F]annexin-V [49]. Conjugation of [18F]SFB to an APP, such as annexin-V, can be seen in

Figure 4.

Figure 4) Reaction scheme to show APP radiolabelling with [18F]SFB

Non-site-specific radiolabelling: Evaluation of non-site-specific methods

Thus far, this review has described the use of 18F prosthetic groups to radiolabel APPs via a

non-site-specific pathway. Apana et al. [38] sought to establish the preferential site of non-site-

specific radiolabelling and their results indicated that the singly-substituted labelled APP,

also the major product, was N-terminally labelled suggesting this as the preferred site and the

likely site of [18F]FBA radiolabelling [38]. This was a useful undertaking in order to understand

the impact of labelling on the interaction of the APP with its binding site.

Table 2 summarises the radiosyntheses of the 18F-prosthetic groups discussed above. The

literature shows widespread use of [18F]FBA and [18F]SFB: they are well-established methods

and can achieve attractive RCY (≤ 90% d.c.) [45]. The commercial availability of the

[18F]FBA precursor augments its popularity and the additional advantage of not having to use

a reducing agents adds to the appeal of [18F]SFB. The popularity of the [18F]FBA and

[18F]SFB prosthetic reagents is reflected in the commercial availability of bespoke

molecularly imprinted polymer (MIP) SPE for their purification following nucleophilic

aromatic substitution of ammonium-containing precursors [40, 50].

Process automation has been reported for all the above prosthetic groups with radiosyntheses

not exceeding 45 minutes. This conforms to the generally held view that radiotracer

synthesis, purification, formulation and analysis should not exceed two half-lives of the PET

radionuclide [1]. The automated radiosynthesis of all prosthetic groups uses SPE purification

[40, 45], with the exception of [18F]fluoroacetaldehyde which uniquely uses distillation

purification [39, 41, 42]. There are, however, significant challenges and considerations to process

automation. Losses in yield are reported [39], as are considerations relating to the component

materials of imposed by automated radiosynthesisers: limitations in the number of

compatible valves and ports that are available can place restrictions on process automation

[49]. Additionally, the use of radiochemistry synthesisers can increase the exposure of 18F to

stable fluoride owing to the multiple reactors and tubing used to transfer a radiolabelling

reaction mixture in a multi-step radiosynthesis [38, 46].

The ability to use larger starting levels of 18F and the consistency and reliability that process

automation can bring means that its challenges are outweighed by its advantages. This

importance of process automation is reflected in the literature, where there is a trend towards

automation of prosthetic group synthesis.

Alongside the encouraging movement towards automated radiosynthesis is the advent of

water-soluble 18F prosthetic groups, such as [18F]FA and [18F]FPCA. The water-solubility of

these prosthetic groups allows milder radiolabelling conditions; radiolabelling of APP with

[18F]FBA, for example, is often carried out in an organic solvent, such as methanol [38].

Non-site-specific radiolabelling: A summary

Radiosynthesis simplification, process automation and compatibility of a radiolabelling

strategy with mild reaction conditions are key characteristics that have emerged regarding

methods of non-site-specific radiolabelling with 18F. The developments are indicative of the

measures being taken to align with the generic features that are required of a versatile

platform for 18F radiolabelling and would facilitate the translation of [18F]APPs to GMP.

The ability to radiolabel the native APP using its naturally-occurring nucleophilic sites,

without pre-modification, is the foremost advantage of this approach [38, 41, 49]. Despite this,

there has been significant interest in the development of site-specific APP radiolabelling

methods. The appeal of these methods is not only characterised by site-selectivity but also the

faster rates of reaction that often result [19, 29, 51].

18F prosthetic

group

RCY d.c.(approx.

%)

Purification No. of steps

Automation Radiosynthesis time (min)

APP radiolabelling

temp (°C)

Comments Refs

SFB 35 SPEMIP SPE

3 Yes 45 (estimated from data in the paper)

30 1. No reducing agent2. Availability of bespoke MIP SPE cartridge

[49, 50]

FBA 90 C18 SPEMIP SPE

1 Yes 45 60 1. High RCY2. Availability of bespoke MIP SPE cartridge3. Commercial availability of precursor

[38, 45]

FA 26 Distillation 2 Yes 45 40 1. No SPE or HPLC purification2. Water-soluble3. Full process automation3. Commercial availability of precursor

[39, 41, 42]

FPCA 28 MIP SPE 1 Yes 45 50 1. Water-soluble2. Full process automation3. Facile synthesis of precursor

[40, 52]

Table 2) Summary of selected carbonyl-containing 18F prosthetic groups used in non-site-specific APP radiolabelling

18F site-specific radiolabelling

Site-specific APP radiolabelling addresses the concerns of possible radiolabel conjugation

at or near the antigen binding site [53]. By virtue of their chemistry, these methods can offer

faster rates of reaction and thus are more high yielding approaches to APP radiolabelling,

given that synthesis is time-limited.

By way of comparison, Table 3 documents the rates that have been reported in the literature

for non-site-specific and site-specific methods of bioconjugation. Consequently site-specific

approaches have become preferable means by which to radiolabel APPs and safeguard

biological activity [1, 19].

Approach Method Rate (M-1s-1) Refs

Non-site-specific Reductive alkylation

(imine formation)

k1 = 10-1 [54]

Site-specific Oxime & Hydrazone

Huisgen 1,3-dipolar cycloaddition

Inverse electron demand Diels-Alder

k = 10-3 – 1.0

k = 2.3

k = 104 - 105

[51]

[55]

[56]

Table 3: A comparison of rates between non-site-specific and site-specific bioconjugation methods

Rate constants for site-specific methods have been included, however overall rate constants

of reductive alkylation are not readily available in the literature, for this reason, k1 values of

imine formation have been included [54].

Site-specific approaches to APP radiolabelling include 18F acceptor chemistries and click-

chemistry methods; the methods and their functional groups have been summarised in Table

4.

Site specific labelling method

Functional group on APP 18F prosthetic group Radiolabelled APP Refs

([18F]Al)2+

[18F]AlCl3

[57-59]

Isotopic exchange

[58, 60,

61]

Oxime bond formation

[15, 40,

52, 62]

Hydrazone formation

[51, 55]

Thiol-Michael addition

[53]

CuAAC [63, 64]

SPAAC [65]

IEDDA [66]

Table 4) Summary of site-specific methods and the functional groups required for the chemistry

Site-specific radiolabelling: 18F acceptor chemistry

Richter et al.[1] described 18F acceptor synthons as attractive alternatives to the use of

prosthetic groups. Two types of 18F acceptor chemistry used to radiolabel APPs will be

discussed: ([18F]AlF)2+ complexation and isotopic exchange. The authors refer readers to a

recent review of recent 18F radiochemistry advances, which includes 18F acceptor approaches

[58] and also a review of ([18F]AlF)2+ coordination chemistries by Kumar et al. [59].

Table 5 summarises the key features of site-specific 18F APP radiolabelling approaches.

Site-specific radiolabelling: ([18F]AlF)2+

([18F]AlF)2+ has been used to radiolabel a number of APPs in a one-pot, high yielding

reaction characterised by Al chelation followed by the formation of a strong Al-F bond, first

reported by McBride et al. [58, 67, 68].

Da Pieve et al.[57] used the ([18F]AlF)2+ method to radiolabel a HER3-specific affibody with a

view to assessing its usefulness as radiotracer in oncology; the method mirrored conditions

reported by Glaser et al. [69]. The affibody was derivatised with 1,4,7-triazacyclononane-1,4,7-

triacetic acid (NOTA) and incubated with AlCl3. Final incubation of the complex with 18F

gave rise to yields of 38.8 ± 5.8 % non-decay-corrected (n.d.c) and molar activity

measurements of 6.0 – 11.9 GBq/μmol were reported [57]. A reaction by-product was

observed, presumed to be an APP degradation product as a result of the radiolabelling

conditions [57, 69], as a result Da Pieve et al. [57] report the requirement for both RP-HPLC and

SPE purification to ensure a final RCP of > 98 %. In vitro analysis using HER3 expressing

cells confirmed retention of [18F]Al-F-NOTA-affibody target specificity and affinity,

indicated by preservation of the affibody’s low nanomolar binding constant [57]. These results

were echoed by the in vivo analysis, using HER3 tumour-bearing mice, where successful

tumour targeting and fast blood clearance of Al[18F]NOTA-affibody was observed [57].

A recent publication from the group describes full process automation of the approach,

having been encouraged by the results of the method and the recognised requirement for

automated processes in GMP production [70]. High temperatures and a water-miscible solvent

were, again, used in the fully-automated procedure which resulted in efficient radiolabelling

of two model APPs which required single SPE-purification of the final radiolabelled product

[70]. A clear advantage of the approach is the simplicity of the single-vial method and ability to

use commercially available aqueous [18F]fluoride/H2O. Application of the fully-automated

approach to a NOTA-derivatised octreotide, chosen for its commercial availability, achieved

RCY of >45 % and RCP measurements of >98 % within 26 minutes [70].

In a similar vein, McBride et al. [67] report the development of single-vial lyophilized kits as a

method to simply and rapidly label APPs with ([18F]AlF)2+ and although not automated, the

group demonstrate the feasibility of translating the simplified single-vial method into a

clinical setting, owing to the ease with which freeze-dried APPs can be radiolabelled and

purified using SPE [67].

The potential of the ([18F]AlF)2+ radiolabelling method is clear and encouraging achievements

in process automation will surely stimulate the appeal of the approach and an additional

advantage to the technique is the ability to exploit the myriad NOTA-derivatised APPs that

are commercially available and used as precursors for 68Ga radiolabelling. However, the

electronic and/or steric influences of a chelate and metal ion complex on the in vivo

performance of an APP should be considered. Strand et al.[26] report the difference in tumour

accumulation of chelate-derivatised affibodies, which was in agreement with previous work

reported by the group that demonstrated a significant influence of the radiolabelling method

on the in vivo behaviour of affibodies. It is also important to consider the toxicity of Al and

that exposure limits to aluminium, ranging from 5-15 mg/m3, have been established in Europe

and the US [71].

Site-specific radiolabelling: Isotopic exchange

Silicon-based isotopic exchange agents are attractive alternatives to 18F radiolabelled

prosthetic groups owing to the higher rate of fluorination at silicon relative to that at carbon

and greater stability of the Si-F bond, compared to that of C-F (142.3 and 116.4 kcal/mol

respectively) [72, 73]

Rosenthal et al. [72] report the development of a silicon-fluoride acceptor (SiFA),

[18F]fluorotrimethylsilane ([18F]FTMS), by nucleophilic substitution of a

chlorotrimethylsilane with 18F, achieving RCY of 80 % (d.c.). However, the in vivo instability

of the Si-F limited its applicability but encouraged developmental of isotopic exchange

agents with bulky alkoxy substituents which exhibit enhanced in vivo stability, whilst

retaining high RCY [60, 72]. The stabilising effect of bulky alkoxy substituents is in agreement

with the work of Brinker et al. [74] where steric factors were shown to have the greatest impact

on the hydrolytic stability of organosilanes.

Schirrmacher et al. [60] reported the development of an isotopic exchange agent, [18F]fluorodi-

tert-butylphenylfluorosilane, described as a highly efficient and hydrolytically stable SiFA.

The group reported a RCY of 80-95 % and molar activity between 194-230 GBq/μmol; the

competitive RCY and molar activity measurements were ascribed to the rapid rates of isotope

exchange [60].

The group identified the requirement for a simplified and, if possible, single-vial method to

yield [18F]APPs to act as an attractive alternative to multi-step radiosyntheses [60]. In order to

achieve this, they bonded the SiFA to a benzaldehyde-derivative for site-specific conjugation

with an aminooxy-functionalised APP; RCY of 60 ± 5 % were attained alongside RCP

measurements of > 98 % [60]. The group describe the APP radiolabelling conditions as the

mildest documented to date. This is true of the pH and room temperature conditions used, yet

the use of 100 % acetonitrile as the reaction medium could denature some APPs. The

approach was also applied to non-azeotropically dried [18F]fluoride; comparable yields to

those attained using azeotropically-dried 18F were only attained at higher temperature (95

°C), yet highlights the translation of the method to a single-vial approach.

The method was further developed by Niedermoser et al. [75] where somatostatin analogues

bearing a SiFA motif were radiolabelled with dried 18F achieving RCY of 50 ± 6 % (d.c.)

within 25 minutes. The radiolabelled APPs was SPE purified and did not require HPLC

purification which contributed significantly to the short overall radiosynthesis time.

Encouragingly, the [18F]APPs exhibited higher tumour accumulation and, although, signal to

noise ratios were lower than those achieved using a 68Ga analogue the single-vial approach

was described as being advantageous for routine clinical use [75].

Site-specific radiolabelling: Click-chemistry

Click-chemistry is characterised by high-yielding and chemoselective reactions that are

compatible with benign solvents and do not generate offensive by-products; see to the

comprehensive list of criteria provided by Sharpless et al. [76]. Click-chemistry reactions were

originally divided into two types of reactions: ring-opening and cycloaddition reactions [76],

however a number of other reactions are now included in the definition, which will be

discussed.

The hallmarks of click-chemistry reactions make clear its appeal in radiochemistry and in

APP radiolabelling terms allows the site-specific conjugation of a radiolabelled synthon,

containing a reactive group, to a complementary reactive group on the APP.

A number of click-chemistry approaches have been reported for APP radiolabelling with 18F;

including oxime bond and hydrazone formation, the products of the reaction between amine-

reactive prosthetic groups and aminooxy- and hydrazide- functional groups respectively [15, 51,

62]. The efficiency of hydrazone and oxime bond formation can be explained by the alpha

effect, where an adjacent lone-pair enhances the nucleophilicity of the amine group by raising

the energy of the highest occupied molecular orbital (HOMO) and stabilisation of the

transition state [77].

Another click-chemistry approach is the Michael-addition of thiol-containing APPs with

maleimide containing prosthetic groups [53]; thiol-Michael additions are the only click-

chemistry approaches for APP radiolabelling using a naturally-occurring residue: thiol-

containing cysteine.

An additional, and very important, category of click-chemistry approaches are cycloadditions

in which new σ-bonds are formed at the expense of π-bonds to form a cyclic compound [78];

examples include copper catalysed and strain-promoted azide-alkyne cycloadditions and

inverse electron demand Diels-Alder cycloadditions [13, 64, 65].

Site-specific radiolabelling: Oxime bond formation

Aminooxyacetyl groups, or aminooxy (Aoa), are important functionalities in site-specific

conjugation to APPs and have been used in a number of fields, including the synthesis of

complex proteins from peptide fragments [62]. As such, oxime bond formation for the purpose

of APP radiolabelling has become an increasingly popular approach; moreover the oxime

bond has been described as hydrolytically stable thereby highlighting its suitability for in vivo

studies [51, 55]. Figure 6 shows the conjugation of an Aoa-functionalised APP with a carbonyl-

bearing prosthetic group, such as [18F]FBA.

Poethko et al. [62] sought to verify the site-specificity of the conjugation and for this [18F]FBA

was incubated with a range of amino-acids both in the presence and absence of aminooxy-

acetic acid. Results indicated that, in the presence of aminooxy-acetic acid, oxime bond

formation accounted for 93 % of the radiolabelled product; [18F]FBA showed some reactivity

with cysteine but, nevertheless, site-specificity was high [62]. [18F]FBA was subsequently used

to radiolabel Aoa-derivatised model APPs: an RGD peptide and a somatostatin analogue,

with overall RCY of ≤ 40 [62].

The highest yields of radiolabelled APPs (60-80 %), were reported at a pH of 2-3 [62]. The

radiolabelled APPs were evaluated in vivo to confirm hydrolytic stability of the oxime bond

but results also showed good tumour-to-background ratios [62].

Namavari et al. [15] adopted the same approach to radiolabel an Aoa-derivatised dimeric

HER2 affibody with [18F]FBA to assess its use as a radiotracer for HER2 detection: HER2 is

another member of the human epidermal growth factor family and is also associated with a

more aggressive tumour phenotype and poorer patient prognosis [22]. [18F]HER2 RCY of 27 ±

3 % (d.c.) were reported [15]; the lower yields of Aoa-functionalised APP, compared to those

described by Poethko et al. [62], were ascribed to the use of a higher pH (pH 4). Namavari et

al. [15] had wanted to assess the applicability of the method to APPs with more complex

proteins that might be susceptible to the use of low pH, especially in combination with

elevated temperatures, such as 60 °C [15]. The group report the success with which the APP

was radiolabelled via oxime bond formation and the high in vivo stability of the oxime bond

[15].

Low pH conditions are widely used in conjunction with oxime-bond formation [40, 62], this

reflects results reported by Dirksen et al. [79] where slow kinetics of formation are observed at

neutral pH. Further work by Dirksen et al. [79] demonstrated the utility of aniline to catalyse

oxime-bond formation. Aniline catalysis was used in the work by Glaser et al. [69], who

radiolabelled an Aoa-functionalised APP with yields of 13 ± 3 % (n.d.c, n=5) and where

aniline catalysed oxime-formation even at neutral pH. Figure 5 shows the mechanism of

aniline catalysis on oxime bond formation.

Full process automation of APP radiolabelling via oxime bond formation has recently been

reported [40] with a 4-fold reduction in overall yields of radiolabelled APP ascribed to the

challenges of translating the radiosynthesis from a semi-manual to an automated approach [40,

52]. This was recompensed by the enhanced consistency, reliability and process robustness

afforded by the automated radiosynthesisers, alongside the ability to use larger starting levels

of 18F [40].

Site-specific radiolabelling: Hydrazone formation

Hydrazone formation is an additional chemoselective approach to site-specific APP

radiolabelling [51]. Hydrazide-derivatised APPs can be radiolabelled using well-established

amine-reaction 18F prosthetic groups. Hydrazone formation has, arguably, had limited utility

in APP radiolabelling and this can be ascribed to the limited stability of the hydrazone

product. Oxime bonds possess greater hydrolytic stability compared to hydrazones [51]; this is

supported by the work of Kalia et al. [55] which documents the hydrolytic half-lives of an

oxime (25 d) and a corresponding hydrazone (2 h).

The rates of hydrazone and oxime bond formation have been compared [51]; Keq of oxime

formation were reported to be two orders of magnitude larger than those for hydrazone

formation despite a 20-fold smaller k1, attributable to the instability of the hydrazone product

Figure 5) Reaction scheme to show the catalysis of oxime bond formation with aniline

[51]. The work provides evidence for the limited utility of hydrazone formation in APP

radiolabelling; despite encouraging rates of hydrazone formation, the stability of the

radiolabelled product would translate to low radiolabelled APP yields and does not align with

some of the generic features of 18F APP radiolabelling, where fast rates of reaction coupled

with high yields as well as in vivo stability are important in order to motivate the translation

of 18F radiolabelled APP into the clinic. Figure 6 shows the conjugation of an hydrazide-

functionalised APP with a carbonyl-bearing prosthetic group, such as [18F]FBA

Site-specific radiolabelling: Thiol-Michael addition

N-[2-(4-18fluorobenzamido)ethyl]maleimide ([18F]FBEM) is a maleimide-containing 18F

prosthetic group, which site-specifically radiolabels thiol-containing APPs. The relative ease

with which thiol-groups can be incorporated in APPs, in comparison to Aoa-derivatives,

drove the developments in the area [15]. As seen in Figure 7, [18F]FBEM can be synthesised

through addition of N-(2-aminoethyl)maleimide to [18F]SFB in the presence of a non-

nucleophilic base, typically N,N-diisopropylethylamine (DIPEA).

B

A

Figure 6) Reaction scheme to show oxime bond (A) and hydrazone formation (B)

Conjugation of [18F]FBEM to an N-terminal cysteine proceeds via a Michael addition, the

mechanism of which can be seen in Figure 8. Radiolabelling mixtures containing thiol-

bearing APPs require a reducing agent (such as tris(2-carboxyethyl)phosphine (TCEP) ) to

prevent disulphide bridge formation between two cysteine residues.

Cai et al. [53] reported using [18F]FBEM to site-specifically radiolabel a thiolated RGD peptide

with yields of 80 % (n.d.c.) and molar activity measurements of 100-150 GBq/μmol, yet the

overall RCY was impacted by the low [18F]FBEM RCY of 5 ± 2 % (n.d.c.). In vitro assays in

cells verified the binding specificity of [18F]RGD and were able to inhibit binding of the avβ3

antagonist echistatin. In vivo assessment showed specific accumulation of the radiotracer

alongside good metabolic stability [53].

Cai et al. [53] described the method as a high-yielding approach to radiolabel thiol-containing

APPs with 18F with high molar activity and the mild reaction conditions highlight the

versatility of the technique [53]. The relevance of this approach for routine production was

unclear, due to the time-consuming multistep radiosynthesis (150 ± 20 min) and low yields,

Figure 7) Reaction scheme to show [18F]FBEM radiosynthesis from [18F]SFB

until the radiosynthesis was later automated by Kiesewetter et al. [80]. The automated approach

reported increased RCY of 17 ± 6 % and reduced synthesis time of 115 minutes (n=21).

Site-specific radiolabelling: CuI catalysed azide alkyne cycloaddition (CuAAC)

CuAAC is a prevalent click-chemistry technique, characterised by the catalysis of azide and

alkyne conjugation by copper (CuI) and the formation of a stable triazole [81]. The chemistry

of CuAAC, a type of Huisgen 1,3-dipolar cycloaddition, originally necessitated the use of

high reaction temperatures or pressures. Important methodical improvements reported by

Rostovtsev et al. [81] involved the use of a copper catalysts which eliminated the requirement

for such harsh reaction conditions [81]. CuI is often generated in situ by reduction of Cu(II)SO4,

often with sodium ascorbate. Understanding of the mechanism of copper catalysis has

*

Figure 8) Reaction scheme to show [18F]FBEM Michael addition to N-terminal cysteine bearing APP

evolved, from originally hypothesising the formation of copper(I)acetylide, via coordination

to the alkyne, to, more recently, the supposition that two copper atoms are involved. This is

discussed in more detail in a review by Pretze et al. [82].

Glaser et al. [64] compared the CuAAC approach with oxime bond formation to radiolabel an

RGD peptide. For the investigation, the azide-containing prosthetic group

[18F]fluoroethylazide ([18F]FEA) was conjugated to an alkyne-derivatised APP, in the

presence of a copper catalyst [76]. [18F]RGD RCY of 47 ± 8 %, (d.c., n=3) were reported;

comparable to those attained using oxime bond formation (40 ± 12 %, d.c., n=5). Optimised

CuAAC radiolabelling was reported in slightly basic conditions; the use of pH 8, however,

led to the formation of a side-product, which was thought to be a copper catalyst adduct.

Generation of the by-product was minimised by reducing the amounts of Cu(II)SO4 and

ascorbic acid [64], an adaptation that would also be of benefit to APPs that are vulnerable to

reduction and susceptible to structural rearrangements to coordinate CuII [56].

Further optimisation of CuAAC approaches would promote its use in APP radiolabelling; of

particular importance is the necessity to ensure the formulated radiotracer is purified from

copper. Pretze et al. [83] reported the formation of copper complexes with amino acid residues:

serine, histidine and arginine and the difficulties in separating these copper species from the

radiolabelled APP. Additional concerns surround the use of Cu(I) stabilising ligands, such as

tris(benzyltriazolylmethyl)amine (TBTA), which are also cytotoxic and must be purified

from the formulated radiotracer.

Site-specific radiolabelling: Bioorthogonality & pre-targeting

More recently, click-chemistry criteria have matured to include a preference for

bioorthogonal methods, which are highly site-selective and compatible with physiological

conditions and inert to biological reactions [56]. Strain promoted azide-alkyne cycloaddition

(SPAAC) and inverse electron demand Diels-Alder (IEDDA) are click-chemistry techniques

that have seen significant attention for their bioorthogonality [56].

Bioorthogonality has enabled the development of a pre-targeting approach: a two-step

methodology involving the labelling of a prosthetic group followed by reaction with an APP

functionalised with a click reagent, which can selectively react in vivo. Firstly, the APP is

administered, and allowed to accumulate at its biological target in accordance with its

relatively slow pharmacokinetics, followed by administration of the more rapid, radiolabelled

prosthetic group. Pre-targeting has been applied in clinical practice, most particularly in

therapeutic oncology applications for localised radiotherapy or chemotherapy rather than

imaging [84].

Site-specific radiolabelling: Strain promoted cycloaddition

A highly strained alkyne, characterised by constrained bond angles flanking the alkyne

moiety (< 180°, linearity of sp-hybridisation), can be used to drive the cycloaddition and

lower the activation barrier and thus eliminate the requirement for a catalyst. A number of

highly strained alkyne reagents are commercially available; those bearing electron

withdrawing groups have reportedly faster kinetics due to alkyne polarisation - although a

trade-off exists between alkyne stability and reactivity [85].

Site-specific radiolabelling: Strain promoted azide-alkyne cycloaddition (SPAAC)

Like CuAAC, strain-promoted azide-alkyne cycloadditions (SPAAC) involve the formation

of a triazole product.

Kim et al. [65] radiolabelled an RGD peptide using the SPAAC approach in order to evaluate

its potential as a PET radiotracer in oncology. The APP was functionalised with a

dibenzocyclooctyne (DBCO) moiety for strain-promoted click-labelling with an azide-

bearing 18F prosthetic group [65]; the radiosynthesis of the prosthetic group had previously

been reported on and achieved a RCY of 63 % (d.c.) [86].

The bioorthogonality of SPAAC click-chemistry enabled use of ambient temperature and

neutral pH; quantitative conjugation of the azide-bearing prosthetic group was reported and

[18F]RGD RCY of 92 % (d.c.) were attained [65]. In vivo analyses of [18F]RGD, in tumour-

bearing mice, showed good tumour accumulation and tumour to background ratios and thus

highlighted the preservation of the APP’s biological activity [65].

Figure 9 shows the reaction pathway for SPAAC click-radiolabelling with

[18F]fluoroethylazide, a commonly used azide-bearing prosthetic group, and a DBCO

functionalised APP. Owing to the robustness and adaptability of the approach, both 18F azides

and DBCO prosthetic groups are reported for use in APP radiolabelling via SPAAC [56].

Site-specific radiolabelling: Inverse electron demand Diels-Alder cycloadditions (IEDDA)

The [4+2] Diels-Alder cycloaddition is the most common pericyclic reaction and can be

reversible or irreversible depending on the diene and dienophile substituents. Figure 10

shows the differences between reversible and irreversible Diels-Alder reactions and, as can

be seen, the inverse electron demand Diels-Alder (IEDDA) occurs between an electron

deficient diene and a dienophile [87].

Figure 9) Reaction scheme to SPAAC between DBCO-functionalised APP and radiolabelled azide

The absence of charged intermediates permits a solely aqueous reaction medium and

proceeds well, via a single-step, on heating due the stability of the transition state which has

six delocalised π-electrons.

Tetrazine and transcyclooctyne (TCO) have been reported as useful IEDDA synthons as a

result of their fast reaction rates [66, 82].

Selvaraj et al.[66] adopted the approach to successfully radiolabel a tetrazine-derivatised RGD

with a fluorinated TCO synthon. The mechanism of IEDDA between tetrazine and TCO can

be seen in Figure 11; the endo product is formed (the kinetic product). [ 18F]TCO RCY were

not provided, but [18F]RGD yields of 90 % (n.d.c.) were achieved within 5 minutes and at

room temperature.

Keliher et al. [88] adopted the IEDDA approach to radiolabel an exendin derivative, targeting

glucagon-liked peptide 1 (GLP-1). For this, a radiolabelled TCO synthon was produced,

achieving RCY of 46 ± 12 % (d.c.), and conjugated to a tetrazine-modified exendin

derivative which yielded d.c. RCY of 47 ± 17 % [88]. Previous attempts to modify exendin

analogues had been reported to have a significant impact on its pharmacokinetics;

importantly, the selectivity of the exendin derivative was retained when radiolabelled using

the IEDDA approach [88].

Figure 10) Irreversible inverse electron demand Diels-Alder and B) reversible normal electron demand Diels-Alder

The mild reaction conditions and compatibility of IEDDA to a range of solvents including

cell media and buffer highlight the bioorthogonality of the approach [66]. These favourable

characteristics promoted an investigation into the application of IEDDA in pre-targeting

methods [12].

Meyer et al. [13] notably used the IEDDA reaction between a tetrazine and TCO to radiolabel a

mAb with 18F using a pre-targeting approach [13]. For the investigation, an [18F]Al-F-NOTA-

radiolabelled tetrazine synthon was synthesised achieving RCY of 60 ± 5 % (d.c.) [13]. The

mAb, which targeted the tumour-associated antigen CA19.9, was derivatised with TCO and

administered to mice bearing CA19.9-expressing tumours before administration of the

radiolabelled tetrazine. In vivo evaluation of the pre-targeting method confirmed that tumour

delineation was possible after 1 hour post administration of Al[18F]NOTA-tetrazine and good

tumour to background ratios were reported between 2 and 4 hours [13]. The in vivo results are

encouraging, yet a real merit of the technique is the ability to radiolabel this influential class

of APP with 18F, an ideal PET radionuclide [1, 30].

Figure 11) Reaction scheme to show IEDDA cycloaddition between tetrazine and TCO

Site-specific radiolabelling: Evaluation of methods

A review by Meyer et al. [56] describes the importance of site-specific APP radiolabelling and

the superiority of the approaches over non-site-specific methods by providing a ‘homogenous

product’ with better in vivo behaviour [56]. A number of site-specific approaches to APP

radiolabelling with 18F have been adopted, Tables 4 and 5 provide a summary of these

methods.

Encouraging features of fluoride acceptor chemistries have emphasised their appeal and

highlighted their potential as attractive alternatives to click-chemistry based approaches [1].

Full process automation of an ([18F]AlF)2+ radiolabelling method [70] and the development of

single-vial methods using lyophilized APPs [67] are key developments in the approach that

enhance the ease with which the methods can be translated to GMP. The ability to use

commercially available aqueous [18F]fluoride and removal of azeotropic drying steps

simplifies the overall procedure and increases the ease at which the methods can be

implemented in a clinical setting. The versatility of the fluoride acceptor approaches may,

however, be limited to APPs where the use of organic solvents and/or high temperatures does

not lead to loss of function [15].

Oxime bond formation and thiol-Michael additions are additional approaches to APP

radiolabelling that have yielded good results. The flexibility of the methods to a variety of

media, including aqueous buffers, and numerous APPs, including those with more complex

structures, highlights the versatility of the approaches [15, 27, 53]. Alongside this are the

developments towards automated radiosyntheses that have improved the consistency and

reliability of multi-step methods [40] and, in the case of [18F]FBEM, resulted in higher yields

[80]. The compatibility of the methods with mild labelling conditions also underlines their

appeal and although neutral pH had been reported to significantly impact yields attained

using oxime bond formation, the use of an aniline catalyst circumvents this issue [69, 79].

Approaches to APP radiolabelling via cycloaddition are very encouraging; the faster rates of

reaction that are afforded using cycloaddition chemistries have given rise to a number of

advantages. Glaser et al. [64] report near complete consumption of crude radiolabelled

synthon using CuAAC, thereby underlining the robustness and high-yielding nature of the

method and highlights the feasibility of a single-vial method of APP radiolabelling.

Comparable RCY using oxime bond formation required a 3-fold increase in reaction time [64].

The ability to use mild reaction conditions opened the door to bioorthogonality and the

applicability of an APP radiolabelling method to physiological conditions and in vivo

labelling. SPAAC and IEDDA are very promising methods, their use of mild reaction

conditions highlights their potential in pre-targeting approaches, which promises to be a very

powerful technique. However, the size and hydrophobic nature of the

dibenzocyclooctatriazole SPAAC and bicyclic IEDDA products and their impact on the

pharmacokinetics of the APP should be considered [56]. An additional disadvantage of SPAAC

is isomerisation in the final radiolabelled APP product, which raises issues of product

consistency and homogeneity.

The use of pre-targeting methods overthrows the convention of matching a PET radioisotope

to the pharmacokinetics of an APP and enhances the versatility of the approach, removing the

need for longer-lived radioisotopes, where radioactive burden to non-target tissues is a

considerable limitation of the technique. Meyer et al. [13] report effective dose measurements

of an 18F pre-targeting approach that were 60-fold lower than directly labelled mAb using

89Zr. Yet the translation of a pre-targeting method to clinical practice may be complicated by

the inconvenience of the technique, where a lag time of 24-72 hours between administration

of the APP and radiolabelled prosthetic group is not uncommon [13, 56], relative to a method

that involves a single injection and rapid scanning protocol which would be more suitable.

Site-specific radiolabelling: A summary

Alongside process automation and radiosynthesis simplification, a trend towards click-

chemistry methods of APP radiolabelling is apparent in the literature with some momentum

towards bioorthogonal approaches. The notion of a generic radiolabelling platform by which

APPs can be radiolabelled with 18F appeared unfeasible due to the striking diversity of APPs

yet the encouraging developments in these bioorthogonal methods, most particularly pre-

targeting methods, and the flexibility and versatility that these methods can afford relieve

some of the challenges of APP radiolabelling with 18F.

Labelling method

Reaction conditions

Overall yields (%)

Comments Refs

Click-chemistry Oxime pH 3-4,50-60°C

≤ 40 (d.c.u.) 1. Good stability of oxime-product2. Aniline catalyst enhances rate, even at neutral pH3. Full process automation

[40, 52,

62, 79]

Hydrazone - - 1. Instability of hydrazone product has limited the appeal of the approach, despite rapid rates of hydrazone formation- k1 of hydrazone and oxime (aniline catalysed), respectively: 170 ± 10 and 8.2

± 1.0 M-1

[51]

Thiol-Michael

pH 7-7.5, room temp

~4 (n.d.c, calculated)

1. Physiological reaction conditions2. Low overall yield as a result of low [18F]FBEM RCY3. Requirement for reducing agent for disulphide bridging (TCEP.HCl)4. Process automation of [18F]FBEM synthesis

[53, 80]

CuAAC slightly basic,80°C

≤ 55 (d.c.) 1. Issues of Cu(I) toxicity2. Higher temperatures often used3. Slightly basic pH enhanced yield

[64]

SPAAC Neutral, room temp

~58 (d.c., calculated)

1. Bioorthogonalitya. No Cu(I) toxicityb. Mild reaction conditions

[65]

IEDDA Neutral, room temp

~60 (d.c.) 1. Bioorthogonalitya. No Cu(I) toxicityb. Mild reaction conditions

2. Pre-targeting application

[13]

Fluoride acceptor chemistry

([18F]AlF)2+ pH 4,100°C

≤ 45 (n.d.c.) 1. Water miscible solvent enhances rate2. Single-vial/fully-automated methods3. No azeotropic distillation4. Electronic/steric impact of macrocyclic on APP

[57-59,

67, 68,

70]

Isotopic exchange

pH 9, room temp

≤ 65 (d.c.u.) 1. Mild APP radiolabelling conditions2. No azeotropic distillation3. Competition with boron exchange (BF3) side reactions

[58, 60]

Table 5) Summary of site-specific methods using 18F (non-decay corrected (n.d.c), decay-corrected (d.c.), decay-correction status unknown (d.c.u))

18F radiolabelling methods: final remarks

A number of approaches have been adopted to radiolabel small and intermediately sized

APPs with 18F. Non-site-specific methods, generally achieved using amine targeting

prosthetic groups, offer a rapid approach to radiolabel native APPs allowing the potential of

the radiolabelled APP to be rapidly validated. The importance of this is reiterated in a

publication by Wester et al. [89], where it was argued that the ability to make a ‘go’ or ‘no go’

decision as early as possible is vital in order to help focus efforts towards the most promising

radiotracers. Non-site-specific radiolabelling methods can, arguably, be used to identify these

strong APP candidates, yet it should be noted that different radiolabelling methods can alter a

radiotracer’s pharmacokinetics, metabolism and excretion [26, 28, 90].

Site-specific methods of APP radiolabelling have become the preferred approach for APP

radiolabelling [19]. Site-specific methods help safeguard the biological properties of the APP

antigen-binding site, provide consistency in the final radiolabel product and also permit

exploitation of the faster reaction rates [15, 19]. However, site-specific approaches often require

APP-functionalisation and whilst functionalisation of smaller APPs is often possible during

solid phase peptide synthesis, functionalisation of larger, more complex, multi-domain APPs

often requires a different approach. Theile et al. [91] report an enzyme-mediated method by

which APPs can be functionalised. The enzyme, sortase, is expressed by gram-positive

bacteria and involved in cell-wall synthesis. An APP and a derivatised small peptide, bearing

the sortase recognition motif, are conjugated by the enzyme [91]. It is clear that APP-

functionalisation has time and financial implications, most particularly using multi-step,

enzyme-mediated derivatisation methods. Such investments may well be worthwhile, in order

to exploit the advantage of site-specific methods, if the potential of an APP as a radiotracer

has already been established.

It may, therefore, be concluded that non-site-specific radiolabelling approaches serve as rapid

tool to first, determine the feasibility of APP radiolabelling and second, the usefulness of the

labelled APP as a PET radiotracer before adapting the approach to permit site-specific APP

radiolabelling.

Although mAbs are likely to remain the prevalent APP PET tracers, there are a number of

very promising 18F radiolabelled APPs. The potential of [18F]APPs hasn’t yet met its full

potential due to the inherent challenges of fluoride radiolabelling and this is plain from the

dearth of [18F]APPs in routine clinical use [1]. In response to the recognised need to harness

the unmet potential of [18F]APPs in the clinic, momentum in the field has unfolded and a

diverse range of approaches have been reported. Despite the diversity of the field, a number

of generic features of 18F APP radiolabelling methods have manifested, including process

automation and site-specificity of methods that are high-yielding and mild. The propensity

for mild APP radiolabelling conditions has given rise to an appetite for bioorthogonality. The

coming together of these generic features has afforded the arrival of very promising APP

radiolabelling methods that are high-yielding, site-specific and bioorthogonal. Whilst these

developments are very promising, the definition of high-yielding is disputable when

comparing against enzyme-mediated conjugations, where rate constants of 2.7x106 M-1s-1 and

yields of > 90 % are typical [29]. Nevertheless, the impetus in the field of APP radiolabelling

with 18F is encouraging and the concerted efforts alongside the afforded results provide

assurance for the long awaited appearance of [18F]APPs in routine clinical use.

Disclosure statement

The authors declare no potential sources of conflict of interest or competing financial interest.

Acknowledgements

OM gratefully acknowledges the CRUK/EPSRC Imaging Centre in Cambridge &

Manchester (CRUK/EPSRC [C8742/A18097]) and would like to thanks AM and JG for

critically reviewing this article.

References

[1] S. Richter, F. Wuest, Molecules 2014, 19, 20536.[2] N. Stephanopoulos, M. B. Francis, Nat Chem Biol 2011, 7, 876.[3] L. Di, The AAPS Journal 2015, 17, 134.[4] C. Gronwall, S. Stahl, J Biotechnol 2009, 140, 254.[5] D. Pressman, Annals of the New York Academy of Sciences 1957, 69, 644.[6] G. Köhler, C. Milstein, Nature 1975, 256, 495.[7] P. Holliger, P. J. Hudson, Nature Biotechnology 2005, 23, 1126.[8] S. M. Larson, H. A. Macapinlac, A. M. Scott, C. R. Divgi, Acta Oncologica 1993, 32, 709.[9] I. Vaneycken, C. Xavier, A. Blykers, N. Devoogdt, V. Caveliers, T. Lahoutte, Journal of

Nuclear Medicine 2011, 52, 664.[10] E. C. F. Dijkers, J. G. W. Kosterink, A. P. Rademaker, L. R. Perk, G. A. M. S. van Dongen, J.

Bart, J. R. de Jong, E. G. E. de Vries, M. N. Lub-de Hooge, Journal of Nuclear Medicine 2009, 50, 974.

[11] H. Kurihara, A. Hamada, M. Yoshida, S. Shimma, J. Hashimoto, K. Yonemori, H. Tani, Y. Miyakita, Y. Kanayama, Y. Wada, M. Kodaira, M. Yunokawa, H. Yamamoto, C. Shimizu, K. Takahashi, Y. Watanabe, Y. Fujiwara, K. Tamura, EJNMMI Res 2015, 5, 8.

[12] I. Vaneycken, M. D'Huyvetter, S. Hernot, J. De Vos, C. Xavier, N. Devoogdt, V. Caveliers, T. Lahoutte, Current Opinion in Biotechnology 2011, 22, 877.

[13] J.-P. Meyer, J. L. Houghton, P. Kozlowski, D. Abdel-Atti, T. Reiner, N. V. K. Pillarsetty, W. W. Scholz, B. M. Zeglis, J. S. Lewis, Bioconjugate Chemistry 2016, 27, 298.

[14] J. Löfblom, J. Feldwisch, V. Tolmachev, J. Carlsson, S. Ståhl, F. Y. Frejd, FEBS Letters 2010, 584, 2670.

[15] M. Namavari, O. P. De Jesus, Z. Cheng, A. De, E. Kovacs, J. Levi, R. Zhang, J. K. Hoerner, H. Grade, F. A. Syud, S. S. Gambhir, Molecular Imaging and Biology 2008, 10, 177.

[16] I. Verel, G. W. M. Visser, R. Boellaard, M. Stigter-van Walsum, G. B. Snow, G. van Dongen, Journal of Nuclear Medicine 2003, 44, 1271.

[17] M. Vosjan, L. R. Perk, G. W. M. Visser, M. Budde, P. Jurek, G. E. Kiefer, G. van Dongen, Nature Protocols 2010, 5, 739.

[18] D. J. Vugts, C. Klaver, C. Sewing, A. J. Poot, K. Adamzek, S. Huegli, C. Mari, G. W. M. Visser, I. E. Valverde, G. Gasser, T. L. Mindt, G. A. M. S. van Dongen, European Journal of Nuclear Medicine and Molecular Imaging 2017, 44, 286.

[19] S. M. Okarvi, European Journal of Nuclear Medicine 2001, 28, 929.[20] R. C. Welch MJ, Handbook of Radiopharmaceuticals. Radiochemistry and Applications,

Wiley, 2003.[21] F. Roesch, Curr Radiopharm 2012, 5, 187.[22] H. S. Cho, K. Mason, K. X. Ramyar, A. M. Stanley, S. B. Gabelli, D. W. Denney, Jr., D. J.

Leahy, Nature 2003, 421, 756.[23] R. Chakravarty, S. Goel, W. B. Cai, Theranostics 2014, 4, 386.[24] C. Vincke, R. Loris, D. Saerens, S. Martinez-Rodriguez, S. Muyldermans, K. Conrath,

Journal of Biological Chemistry 2009, 284, 3273.[25] I. Vaneycken, N. Devoogdt, N. Van Gassen, C. Vincke, C. Xavier, U. Wernery, S.

Muyldermans, T. Lahoutte, V. Caveliers, Faseb Journal 2011, 25, 2433.

[26] J. Strand, H. Honarvar, A. Perols, A. Orlova, R. K. Selvaraju, A. E. Karlström, V. Tolmachev, PLoS ONE 2013, 8, e70028.

[27] G. Kramer-Marek, D. O. Kiesewetter, L. Martiniova, E. Jagoda, S. B. Lee, J. Capala, Eur J Nucl Med Mol Imaging 2008, 35, 1008.

[28] S. Trousil, S. Hoppmann, Q. D. Nguyen, M. Kaliszczak, G. Tomasi, P. Iveson, D. Hiscock, E. O. Aboagye, Clin Cancer Res 2014, 20, 1632.

[29] J. Kalia, R. T. Raines, Current organic chemistry 2010, 14, 138.[30] O. Jacobson, D. O. Kiesewetter, X. Chen, Bioconjug Chem 2015, 26, 1.[31] M. Kroselj, A. Socan, K. Zaletel, T. Dreger, R. Knopp, T. Gmeiner, P. Kolenc Peitl, Nucl

Med Commun 2015, 37, 207.[32] I. Velikyan, Journal of Labelled Compounds and Radiopharmaceuticals 2015, 58, 99.[33] M. B. Haskali, P. D. Roselt, R. J. Hicks, C. A. Hutton, Journal of Labelled Compounds and

Radiopharmaceuticals 2018, 61, 61.[34] M. Alauddin, Am J Nucl Med Mol Imaging 2012, 2, 55.[35] M. Piel, F. Rösch, in Molecular Imaging in the Clinical Neurosciences, Vol. 71 (Ed.: G.

Gründer), Humana Press, 2012, pp. 41.[36] H. H. Coenen, Ernst Schering Research Foundation workshop 2007, 15.[37] R. Schirrmacher, B. Wängler, J. Bailey, V. Bernard-Gauthier, E. Schirrmacher, C. Wängler,

Seminars in Nuclear Medicine 2017, 47, 474.[38] S. M. Apana, R. J. Griffin, N. A. Koonce, J. S. Webber, R. P. M. Dings, K. H. Mayo, M. S.

Berridge, Journal of Labelled Compounds and Radiopharmaceuticals 2011, 54, 708.[39] O. Morris, A. McMahon, H. Boutin, J. Grigg, C. Prenant, Journal of Labelled Compounds &

Radiopharmaceuticals 2016, 59, 277.[40] O. Morris, J. Gregory, M. Kadirvel, F. Henderson, A. Blykers, A. McMahon, M. Taylor, D.

Allsop, S. Allan, J. Grigg, H. Boutin, C. Prenant, Applied Radiation and Isotopes 2016, 116, 120.

[41] C. Prenant, C. Cawthorne, M. Fairclough, N. Rothwell, H. Boutin, Applied Radiation and Isotopes 2010, 68, 1721.

[42] C. Prenant, J. Gillies, J. Bailey, G. Chimon, N. Smith, G. C. Jayson, J. Zweit, Journal of Labelled Compounds & Radiopharmaceuticals 2008, 51, 262.

[43] N. Jentoft, D. G. Dearborn, Journal of Biological Chemistry 1979, 254, 4359.[44] S. Preshlock, M. Tredwell, V. Gouverneur, Chemical Reviews 2016, 116, 719.[45] A. Speranza, G. Ortosecco, E. Castaldi, A. Nardelli, L. Pace, M. Salvatore, Applied Radiation

and Isotopes 2009, 67, 1664.[46] M. S. Berridge, S. M. Apana, J. M. Hersh, Journal of Labelled Compounds &

Radiopharmaceuticals 2009, 52, 543.[47] B. de Bruin, B. Kuhnast, F. Hinnen, L. Yaouancq, M. Amessou, L. Johannes, A. Samson, R.

Boisgard, B. Tavitian, F. Dolle, Bioconjug Chem 2005, 16, 406.[48] B. Kuhnast, F. Hinnen, B. Tavitian, F. Dollé, Journal of Labelled Compounds and

Radiopharmaceuticals 2008, 51, 336.[49] S. Zijlstra, J. Gunawan, W. Burchert, Appl Radiat Isot 2003, 58, 201.[50] A. Blykers, C. Xavier, I. Vaneycken, D. Laoui, N. Devoogdt, M. D'Huyvetter, T. Lahoutte, J.

A. Van Ginderachter, V. Caveliers, European Molecular Imaging Meeting (EMIM), Turin Italy, 2013.

[51] A. Dirksen, P. E. Dawson, Bioconjug Chem 2008, 19, 2543.[52] O. Morris, M. A. Elsawy, M. Fairclough, K. J. Williams, A. McMahon, J. Grigg, D. Forster,

A. F. Miller, A. Saiani, C. Prenant, Journal of Labelled Compounds and Radiopharmaceuticals 2017, 60, 481.

[53] W. Cai, X. Zhang, Y. Wu, X. Chen, Journal of nuclear medicine : official publication, Society of Nuclear Medicine 2006, 47, 1172.

[54] T. C. French, T. C. Bruice, Biochemical and Biophysical Research Communications 1964, 15, 403.

[55] J. Kalia, R. T. Raines, Angewandte Chemie International Edition 2008, 47, 7523.[56] J. P. Meyer, P. Adumeau, J. S. Lewis, B. M. Zeglis, Bioconjugate Chemistry 2016, 27, 2791.

[57] C. Da Pieve, L. Allott, C. D. Martins, A. Vardon, D. M. Ciobota, G. Kramer-Marek, G. Smith, Bioconjug Chem 2016, 27, 1839.

[58] V. Bernard-Gauthier, M. L. Lepage, B. Waengler, J. J. Bailey, S. H. Liang, D. M. Perrin, N. Vasdev, R. Schirrmacher, J Nucl Med 2018, 59, 568.

[59] K. Kumar, A. Ghosh, Bioconjugate Chemistry 2018, 29, 953.[60] R. Schirrmacher, G. Bradtmöller, E. Schirrmacher, O. Thews, J. Tillmanns, T. Siessmeier, H.

G. Buchholz, P. Bartenstein, B. Wängler, C. M. Niemeyer, K. Jurkschat, Angewandte Chemie International Edition 2006, 45, 6047.

[61] E. Schirrmacher, B. Wängler, M. Cypryk, G. Bradtmöller, M. Schäfer, M. Eisenhut, K. Jurkschat, R. Schirrmacher, Bioconjugate Chemistry 2007, 18, 2085.

[62] T. Poethko, M. Schottelius, G. Thumshirn, M. Herz, R. Haubner, G. Henriksen, H. Kessler, M. Schwaiger, H. J. Wester, Radiochimica Acta 2004, 92, 317.

[63] M. Glaser, E. Årstad, Bioconjugate Chemistry 2007, 18, 989.[64] M. Glaser, M. Solbakken, D. R. Turton, R. Pettitt, J. Barnett, J. Arukwe, H. Karlsen, A.

Cuthbertson, S. K. Luthra, E. Arstad, Amino Acids 2009, 37, 717.[65] H. L. Kim, K. Sachin, H. J. Jeong, W. Choi, H. S. Lee, D. W. Kim, ACS Med Chem Lett

2015, 6, 402.[66] R. Selvaraj, S. Liu, M. Hassink, C.-w. Huang, L.-p. Yap, R. Park, J. M. Fox, Z. Li, P. S.

Conti, Bioorganic & Medicinal Chemistry Letters 2011, 21, 5011.[67] W. J. McBride, C. A. D'Souza, H. Karacay, R. M. Sharkey, D. M. Goldenberg, Bioconjug

Chem 2012, 23, 538.[68] W. J. McBride, R. M. Sharkey, H. Karacay, C. A. D'Souza, E. A. Rossi, P. Laverman, C. H.

Chang, O. C. Boerman, D. M. Goldenberg, J Nucl Med 2009, 50, 991.[69] M. Glaser, P. Iveson, S. Hoppmann, B. Indrevoll, A. Wilson, J. Arukwe, A. Danikas, R.

Bhalla, D. Hiscock, Journal of Nuclear Medicine 2013, 54, 1981.[70] L. Allott, C. Da Pieve, D. R. Turton, G. Smith, Reaction Chemistry & Engineering 2017, 2,

68.[71] D. Krewski, R. A. Yokel, E. Nieboer, D. Borchelt, J. Cohen, J. Harry, S. Kacew, J. Lindsay,

A. M. Mahfouz, V. Rondeau, Journal of toxicology and environmental health. Part B, Critical reviews 2007, 10, 1.

[72] M. S. Rosenthal, A. L. Bosch, R. J. Nickles, S. J. Gatley, International Journal of Applied Radiation and Isotopes 1985, 36, 318.

[73] R. S. Grev, in Advances in Organometallic Chemistry, Vol. Volume 33 (Eds.: F. G. A. Stone, W. Robert), Academic Press, 1991, pp. 125.

[74] C. J. Brinker, Journal of Non-Crystalline Solids 1988, 100, 31.[75] S. Niedermoser, J. Chin, C. Wangler, A. Kostikov, V. Bernard-Gauthier, N. Vogler, J. P.

Soucy, A. J. McEwan, R. Schirrmacher, B. Wangler, J Nucl Med 2015, 56, 1100.[76] H. C. Kolb, M. G. Finn, K. B. Sharpless, Angewandte Chemie-International Edition 2001, 40,

2004.[77] N. J. Fina, J. O. Edwards, International Journal of Chemical Kinetics 1973, 5, 1.[78] R. Huisgen, Angewandte Chemie International Edition in English 1963, 2, 565.[79] A. Dirksen, T. M. Hackeng, P. E. Dawson, Angewandte Chemie International Edition 2006,

45, 7581.[80] D. O. Kiesewetter, O. Jacobson, L. Lang, X. Chen, Appl Radiat Isot 2011, 69, 410.[81] V. V. Rostovtsev, L. G. Green, V. V. Fokin, K. B. Sharpless, Angewandte Chemie

International Edition 2002, 41, 2596.[82] M. Pretze, D. Pietzsch, C. Mamat, Molecules 2013, 18, 8618.[83] M. Pretze, M. Kuchar, R. Bergmann, J. Steinbach, J. Pietzsch, C. Mamat, Chemmedchem

2013, 8, 935.[84] R. M. Sharkey, H. Karacay, T. M. Cardillo, C.-H. Chang, W. J. McBride, E. A. Rossi, I. D.

Horak, D. M. Goldenberg, Clinical Cancer Research 2005, 11, 7109s.[85] L. Carroll, H. L. Evans, E. O. Aboagye, A. C. Spivey, Org Biomol Chem 2013, 11, 5772.[86] K. Sachin, V. H. Jadhav, E.-M. Kim, H. L. Kim, S. B. Lee, H.-J. Jeong, S. T. Lim, M.-H.

Sohn, D. W. Kim, Bioconjugate Chemistry 2012, 23, 1680.[87] J. Y. Choi, B. C. Lee, Nucl Med Mol Imaging 2015, 49, 258.

[88] E. J. Keliher, T. Reiner, G. M. Thurber, R. Upadhyay, R. Weissleder, ChemistryOpen 2012, 1, 177.

[89] H. J. Wester, M. Schottelius, in PET Chemistry: The Driving Force in Molecular Imaging, Vol. 62 (Eds.: P. A. Schubiger, L. Lehmann, M. Friebe), 2007, pp. 79.

[90] S. Richter, M. Wuest, C. N. Bergman, J. D. Way, S. Krieger, B. E. Rogers, F. Wuest, Bioconjug Chem 2015, 26, 201.

[91] C. S. Theile, M. D. Witte, A. E. M. Blom, L. Kundrat, H. L. Ploegh, C. P. Guimaraes, Nat. Protocols 2013, 8, 1800.