1
Targeting Fibroblast Activation Protein:
Radiosynthesis and Preclinical Evaluation of an 18F‐labeled FAP Inhibitor
Johannes Toms*,1, Jürgen Kogler1, Simone Maschauer1, Christoph Daniel2, Christian Schmidkonz3, Torsten
Kuwert3, Olaf Prante1,3,§
1Friedrich‐Alexander University Erlangen‐Nürnberg (FAU), Department of Nuclear Medicine, Molecular
Imaging and Radiochemistry, Translational Research Center, Schwabachanlage 12, 91054 Erlangen,
Germany
2Friedrich‐Alexander University Erlangen‐Nürnberg (FAU), Department of Nephropathology,
Krankenhausstr. 8‐10, 91054 Erlangen, Germany
3Friedrich‐Alexander University Erlangen‐Nürnberg (FAU), Department of Nuclear Medicine, Ulmenweg
18, 91054 Erlangen, Germany
§ Corresponding author: Olaf Prante, Prof. Dr. Department of Nuclear Medicine, Molecular Imaging and Radiochemistry Friedrich‐Alexander University Erlangen‐Nürnberg (FAU) Schwabachanlage 12 91054 Erlangen Tel: +49‐9131‐85‐44440 Fax: +49‐9131‐85‐39288 Email: olaf.prante@uk‐erlangen.de * First author: Johannes Toms (PhD student) Department of Nuclear Medicine, Molecular Imaging and Radiochemistry Friedrich‐Alexander University Erlangen‐Nürnberg (FAU) Schwabachanlage 12 91054 Erlangen Tel: +49‐9131‐85‐47039 Email: johannes.toms@uk‐erlangen.de This work was financially supported by the Emerging Field Initiative (EFI) of the Friedrich‐Alexander
University Erlangen‐Nürnberg (FAU) and by the Deutsche Forschungsgemeinschaft (DFG, MA 4295/2‐1)
Short title: [18F]FGlc‐FAPI for imaging of FAP
Journal of Nuclear Medicine, published on April 24, 2020 as doi:10.2967/jnumed.120.242958
2
ABSTRACT
Fibroblast activation protein (FAP) has emerged as an interesting molecular target used in the imaging
and therapy of various types of cancers. Gallium‐68–labeled chelator‐linked FAP inhibitors (FAPIs) have
been successfully applied to positron emission tomography (PET) imaging of various tumor types. To
broaden the spectrum of applicable PET tracers for extended imaging studies of FAP‐dependent
diseases, we herein report the radiosynthesis and preclinical evaluation of an 18F–labeled glycosylated
FAP inhibitor ([18F]FGlc‐FAPI). Methods: An alkyne‐bearing precursor was synthesized and subjected to
click chemistry–based radiosynthesis of [18F]FGlc‐FAPI by two‐step 18F‐fluoroglycosylation. FAP‐
expressing HT1080hFAP cells were used to study competitive binding to FAP, cellular uptake,
internalization, and efflux of [18F]FGlc‐FAPI in vitro. Biodistribution studies and in vivo small animal PET
studies of [18F]FGlc‐FAPI compared to [68Ga]Ga‐FAPI‐04 were conducted in nude mice bearing
HT1080hFAP tumors or U87MG xenografts. Results: [18F]FGlc‐FAPI was synthesized with a 15%
radioactivity yield and a high radiochemical purity of >99%. In HT1080hFAP cells, [18F]FGlc‐FAPI showed
specific uptake, a high internalized fraction, and low cellular efflux. Compared to FAPI‐04 (IC50 = 32 nM),
the glycoconjugate, FGlc‐FAPI (IC50 = 167 nM), showed slightly lower affinity for FAP in vitro, while
plasma protein binding was higher for [18F]FGlc‐FAPI. Biodistribution studies revealed significant
hepatobiliary excretion of [18F]FGlc‐FAPI; however, small animal PET studies in HT1080hFAP xenografts
showed higher specific tumor uptake of [18F]FGlc‐FAPI (4.5 % injected dose per gram of tissue [ID/g])
compared to [68Ga]Ga‐FAPI‐04 (2 %ID/g). In U87MG tumor–bearing mice, both tracers showed similar
tumor uptake, but [18F]FGlc‐FAPI showed a higher tumor retention. Interestingly, [18F]FGlc‐FAPI
demonstrated high specific uptake in bone structures and joints. Conclusion: [18F]FGlc‐FAPI is an
interesting candidate for translation to the clinic, taking advantage of the longer half‐life and physical
imaging properties of F‐18. The availability of [18F]FGlc‐FAPI may allow extended PET studies of FAP‐
related diseases, such as cancer, but also arthritis, heart diseases, or pulmonary fibrosis.
3
Keywords: fibroblast activation protein, cancer‐associated fibroblasts, fluorine‐18, 18F‐
fluoroglycosylation, positron emission tomography
4
INTRODUCTION
The fibroblast activation protein (FAP) is a membrane‐bound serine protease with dipeptidyl peptidase
and endopeptidase activities (1). The regulation and role of FAP has been studied for almost three
decades and has gained special attention as a highly expressed marker in the vast majority of epithelial
tumors (2), as well as in fibrosis (3) and rheumatoid arthritis (4). More generally, FAP is associated with
the pathological and even non‐pathological remodeling of the extracellular matrix (5). The basal
expression of FAP in healthy human tissue is considered very low, whereas in mice detectable levels of
FAP expression were shown to be highest in the uterus, pancreas, submaxillary gland, and skin (6). In the
case of cancer, FAP is not expressed on malignant tumor cells themselves, but on cancer‐associated
fibroblasts (CAFs), a non‐malignant cell subtype of the stroma, which is a major part of a tumor’s
composition (7). However, even though recognized as non‐malignant cells, CAFs are heavily involved in
tumor growth, migration, and progression of the disease by communicating with several cell types
through the secretion of growth factors and chemokines (8). Not only is the mediating role of CAFs in
cancer manifold, but the cells originate from various sources. Endothelial cells, resident fibroblasts,
adipocytes and bone marrow–derived hematopoietic or mesenchymal stem cells have been described as
putative precursors that differentiate into CAFs (9). While low expression of FAP has been also shown on
some CAF precursors, FAP overexpression by CAFs is a key characteristic and is often connected to a bad
prognosis and outcomes for respective cancers (10). Several approaches targeting FAP have been made
using antibodies (11,12), peptide prodrugs (13), or small molecules (14). However, limited success has
been achieved in preclinical studies such that few candidates have entered clinical studies (15,16). The
protein structure of FAP was described in 2005 (17), together with the first generation of FAP inhibitors
that showed good affinity but lacked selectivity. Second‐generation FAP inhibitors were structurally
based on a quinoline amide core coupled to a 2‐cyanopyrrolidine moiety (18); these demonstrated
nanomolar affinity and selectivity for FAP with only a low affinity to other interfering dipeptidyl
5
peptidases (19,20). Recently, the development of PET ligands was successfully achieved by the
introduction of a dodecane tetraacetic acid (DOTA) chelator coupled to an alkyl‐piperazine spacer linked
to the quinoline core and labeled with 68Ga (21,22). A 68Ga–labeled FAPI tracer ([68Ga])Ga‐FAPI‐04; Fig. 1)
demonstrated high tumor‐to‐organ ratios and fast elimination in preclinical experiments. Moreover,
first‐in‐human PET studies with [68Ga]Ga‐FAPI‐04 revealed the excellent visualization of a broad range of
tumors, with high tumor‐to‐noise ratios and the successful imaging of tumor metastases (23,24). In
addition, derivatives of FAPI‐04 are under development that may allow radiolabeling with 177Lu for
application in endoradiotherapy, provided that the biological half‐life is appropriate for the therapeutic
demand (25).
We here report an alkyne‐bearing version of FAPI‐04 and its use as a precursor for the click
chemistry–based synthesis of an 18F‐labeled FAP inhibitor. In previous studies, we demonstrated that the
introduction of 18F‐labeled glycosyl moieties could positively influence the clearance behavior of a
radiotracer (26‐28), and the glycosylation of biomolecules has been frequently shown to improve the in
vivo stability in blood and allows for the optimization of active drug delivery (29,30). Therefore, herein,
we present the radiosynthesis and preclinical evaluation of an 18F–labeled glycosylated FAP inhibitor
([18F]FGlc‐FAPI; Fig. 1) as a new 18F‐fluoroglycosylated FAP ligand. In direct comparison with [68Ga]Ga‐
FAPI‐04, we report biodistribution studies and small animal PET studies of [18F]FGlc‐FAPI in nude mice
xenografts of FAP‐positive tumors.
6
MATERIALS AND METHODS
General
The stably FAP‐transfected fibrosarcoma cell line HT1080hFAP and the precursor FAPI‐04 were
kindly provided by Prof. Uwe Haberkorn (University Hospital Heidelberg, Germany). The procedures for
in vitro and animal experiments, including tumor transplantation, were given in detail in the
Supplemental Information File.
Chemistry
The syntheses of all non‐radioactive compounds, including the reference compound FGlc‐FAPI
are described in detail in the Supplemental Information.
Radiosynthesis of [18F]FGlc‐FAPI
Starting from the tosylate precursor 1, 2,3,4‐tri‐O‐acetyl‐6‐deoxy‐6‐[18F]fluoroglucosyl azide
([18F]2) was prepared, isolated and deacetylated with NaOH (270 µL, 60 mM) at 60 °C for 5 min as
described before (27). To the crude product 6‐deoxy‐6‐[18F]fluoroglucosyl azide ([18F]3), a mixture of
phosphate buffer (270 µL, 0.5 M, pH 8), Cu(OAc)2 (20 µL, 4 mM), tris(3‐
hydroxypropyltriazolylmethyl)amine (THPTA; 20 µL, 20 mM), sodium ascorbate (20 µL, 0.1 M) and alkyne
11 (1 mM, 400 µL) was added and the reaction was stirred for 15 min at 60 °C. The RCY of [18F]FGlc‐FAPI
was 52‐85% as determined by analytical HPLC (method 2) from a sample withdrawn from the reaction
mixture. Purification of [18F]FGlc‐FAPI was performed by semi‐preparative HPLC (method 1, tR = 10.3
min). The product fraction was diluted with water and passed through a RP‐18 cartridge (Sep‐Pak light
C18, Waters). [18F]FGlc‐FAPI was eluted with ethanol (2 mL) and after evaporation of the solvent,
[18F]FGlc‐FAPI was formulated with 0.9% saline. Starting from [18F]fluoride (500‐1000 MBq), [18F]FGlc‐
7
FAPI was obtained in a radioactivity yield (RAY) of 30‐130 MBq (6‐15%, referred to [18F]fluoride) in a total
synthesis time of 90‐110 min in specific activities of 30‐200 GBq/µmol (n = 11).
Biodistribution
The tumor‐bearing mice were anesthetized with O2/isofluran (2‐3% isofluran, 1.2 L/min O2) and
laid on a heating pad (37 °C). About 1‐2 MBq (100 µL) of [18F]FGlc‐FAPI (and 30 nmol/mouse of FAPI
alkyne 11 as blocking substance) were injected via the tail vein. The mice were then transferred to a cage
and allowed to recover from anesthesia. The mice were euthanized by cervical dislocation under deep
isoflurane anesthesia at 30 min (n = 2), 60 min (n = 2), 60 min (coinjection of 11, n = 2) and 90 min (n = 2)
post‐injection (p.i.) and samples from tissues and blood were removed, weighed and radioactivity
counted in the γ‐counter. The results were presented as percentage injected dose per gram tissue
(%ID/g) and tumor‐to‐organ ratios were calculated thereof. The values were given as mean values ±
standard deviation.
Small animal PET Imaging
Mice bearing xenografted HT1080hFAP or U87MG tumors were laid on a heating pad (37 °C) and
were anesthetized using an O2/isofluran (2‐3% isofluran, 1.2 L/min O2). A venous access was laid into the
tail vein of the animals and the cannula was fixed by an instant adhesive on the tail. The mice were
transferred to an Inveon™ microPET scanner (Siemens Healthcare, Erlangen), tempered by a heating pad
(37 °C). A dynamic PET scan was started from 0 to 60 min after injection of 2‐5 MBq of [18F]FGlc‐FAPI (n =
2‐3) or 3‐5 MBq [68Ga]Ga‐FAPI‐04 (n = 2‐5) in 100 µL of isotonic saline solution. For three mice a 15 min‐
static scan was acquired at 120 min after tracer injection. The blocking experiments were performed by
coinjection of [18F]FGlc‐FAPI and alkyne 11 (30 or 100 nmol/animal, n = 3). After iterative maximum a
posteriori image reconstruction of the decay and attenuation corrected images, regions of interest (ROIs)
were defined using the software PMOD (PMOD Technologies LLC, Switzerland). The mean radioactivity
8
concentration within the ROIs was converted to percentage injected dose per gram organ (%ID/g) and
provided as mean values ± standard deviation.
RESULTS
Chemistry and Radiochemistry
Inspired by recently published results on [68Ga]Ga‐FAPI‐04, we synthesized the alkyne‐bearing
derivative 11 as a derivative of FAPI‐04. This can lead to chemoselective copper(I)‐catalyzed alkyne azide
cycloaddition (CuAAC) reactions, such as a wide variety of 18F‐click chemistry–based labeling methods
(31). The alkyne precursor 11 is composed of a dipeptide analog synthesized according to Jansen et al.
(20), a quinoline unit, and a piperazine‐alkyl linker, which was synthesized by a modified procedure
following Lindner et al (22). The alkyne 11 was obtained with an overall yield of 13% over six reaction
steps (Supplemental Fig. 1). Applying chemoselective CuAAC reaction with 11 and 6‐deoxy‐6‐fluoro‐‐
glucosyl azide (3 (27)), we obtained glycoconjugate FGlc‐FAPI with a yield of 66% and a purity of 95% as a
FAP inhibitor test compound and reference compound of the 18F‐labeled analog.
Based on our previous experience in the development of 18F‐labeled glycoconjugates as PET
imaging agents (27,28,32), the radiosynthesis of [18F]FGlc‐FAPI was started from precursor 1 (27) for 18F‐
substitution, followed by 18F‐fluoroglycosylation of 11 (400 nmol) under optimized reaction conditions
(Fig. 2). The complete two‐step radiosynthesis was performed in a maximum of 90–110 min, providing a
[18F]FGlc‐FAPI formulated tracer with a radioactivity yield of 15%, a radiochemical purity of >99%, and a
molar activity of 30–200 GBq/µmol (Supplemental Figs. 2‐4).
9
In vitro Evaluation of [18F]FGlc‐FAPI
While [68Ga]Ga‐FAPI‐04 showed high hydrophilicity (logD7.4 = ‐2.41 ± 0.28), [18F]FGlc‐FAPI was less
polar (logD7.4 = ‐0.99 ± 0.03). Both tracers showed high stability in human serum after incubation for 55
min of >99% for [18F]FGlc‐FAPI and 93% for [68Ga]Ga‐FAPI‐04. Binding to human blood plasma proteins
was determined for [18F]FGlc‐FAPI (45 ± 6%, n = 6) and was significantly higher than for [68Ga]Ga‐FAPI‐04
(10 ± 3%, n = 6).
To determine whether the introduction of the alkyne moiety and glycosylation of [18F]FGlc‐FAPI
had any influence on FAP binding, we performed competition binding experiments with [177Lu]Lu‐FAPI‐04
as a radioligand in HT1080hFAP cells (Fig. 3A). The test compounds FGlc‐FAPI, alkyne 11, FAPI‐04, and
Ga‐FAPI‐04 showed IC50 values in a range from double‐ to triple‐digit nanomolar concentrations (Table
1). Due to differences in the assay setup, the IC50 values were in general higher than values reported in
the literature (22). All four FAPI derivatives lead to the displacement of [177Lu]Lu‐FAPI‐04 of about 75–
95%, while DOTA derivatives showed a 3–5‐fold higher affinity for FAP compared to alkyne 11 and FGlc‐
FAPI. We also determined the binding to the structurally‐related peptidase dipeptidyl peptidase 4 (DPP4,
CD26), which is highly expressed on blood cells and memory T cells and, in its soluble form, is present in
blood plasma. In direct comparison to FAPI‐04, the glycoconjugate FGlc‐FAPI and FAPI‐alkyne showed
comparable IC50 values in the micromolar range, confirming the binding selectivity of FGlc‐FAPI for FAP
(Table 1, Supplemental Fig. 5).
[18F]FGlc‐FAPI showed specific uptake in HT1080hFAP cells of 6.6 ± 0.8% after 60 min of
incubation, which could be significantly blocked by the addition of alkyne 11 to between 3 and 7% (Fig.
3B). Additionally, FAP‐negative MCF‐7 cells did not show any significant uptake of [18F]FGlc‐FAPI,
suggesting specific FAP‐mediated uptake of [18F]FGlc‐FAPI in HT1080hFAP cells (Supplemental Fig. 6).
[18F]FGlc‐FAPI showed a fast cellular uptake (>4% after 5 min, Fig. 3B) with a rapid and high
10
internalization rate in HT1080hFAP cells (95% after 5–120 min; Fig. 3C), confirming results for 177Lu‐
labeled FAPI derivatives reported by Lindner et al. (22). After 60 min of incubation with [18F]FGlc‐FAPI,
75% of the tracer was still internalized, demonstrating the low cellular efflux rate of [18F]FGlc‐FAPI in
vitro (Fig. 3C).
Biodistribution of [18F]FGlc‐FAPI
Biodistribution studies of [18F]FGlc‐FAPI in HT1080hFAP tumor‐bearing mice revealed tumor
uptake of 1.6–2.0 %ID/g (n = 2, Table 2, Supplemental Fig. 7A) with high retention from 30 to 90 min p.i.
The tracer uptake was shown to be specific by the coinjection of [18F]FGlc‐FAPI together with 11 (30
nmol/mouse), demonstrating diminished tumor uptake by 56% at 60 min p.i. (0.9 ± 0.2 %ID/g, n = 2,
Table 2, Supplemental Fig. 7B). Notably, partial necrosis of the tumors was observed, such that the
tumor uptake of [18F]FGlc‐FAPI was heterogeneous and moderate. Biodistribution studies of [18F]FGlc‐
FAPI also demonstrated pronounced uptake in the liver and intestine, suggesting a hepatobiliary
excretion pathway for [18F]FGlc‐FAPI. The observed uptake in the bones (femur) was constant over time,
similar to the uptake in tumors, and could be blocked by approximately 80% (3.4 %ID/g vs. 0.6 %ID/g, n =
2, Table 2, Supplemental Fig. 7B), suggesting specific binding to bone marrow. Similarly, the
concentration of [18F]FGlc‐FAPI in the blood was diminished under blocking conditions (Table 2,
Supplemental Fig. 7B).
µPET Imaging
For a proof‐of‐concept study and an assessment of the tracer performance of [18F]FGlc‐FAPI in
comparison to [68Ga]Ga‐FAPI‐04, small animal PET studies using HT1080hFAP xenografts were
conducted. Dynamic measurements over the course of 60 min p.i. revealed fast biodistribution and
specific uptake in HT1080hFAP tumors in vivo, both for [18F]FGlc‐FAPI and [68Ga]Ga‐FAPI‐04 (Fig. 4).
11
Tumor uptake of [18F]FGlc‐FAPI was limited to intact tumor tissue, as necrotic areas inside the tumor did
not show any significant tracer uptake (Fig. 4A). At an early time point after tracer injection (10 min p.i.),
[18F]FGlc‐FAPI reached a tumor uptake of 4.6 %ID/g and showed retention in the tumor over the time
period of the PET scan (Fig. 4C). In comparison, [68Ga]Ga‐FAPI‐04 showed a lower tumor uptake of 2.1
%ID/g (Fig. 4B, D), though homogenous tracer distribution was noted in the tumors.
As an alternative tumor model for FAP‐transfected HT1080 cancer cell xenografts, we generated
U87MG xenografts for PET imaging studies. In vitro, U87MG cells lack FAP expression, however, U87MG
xenograft tumors are FAP‐positive in vivo, which is ascribed to the recruitment and activation of mouse
fibroblasts (CAF) (33). In these more homogenous growing tumors, a similarly high uptake of [18F]FGlc‐
FAPI and [68Ga]Ga‐FAPI‐04 was observed (approx. 7 %ID/g, Fig. 5A, B). Interestingly, [68Ga]Ga‐FAPI‐04
reached maximum uptake at 15 min p.i., which subsequently slightly decreased. However, the tumor
uptake of [18F]FGlc‐FAPI slightly but constantly increased, beginning from an initial rapid uptake phase at
10 min p.i. to the end of the total scan time of 60 min (Fig. 5B). The high tumor retention of [18F]FGlc‐
FAPI has been additionally proven by PET scans at 120‐130 min p.i. (Supplemental Fig. 8).
Unexpectedly, a high and specific uptake of [18F]FGlc‐FAPI was also detected in bone structures,
for example in the joints, spine, skull, and shoulders (Fig. 5C, Supplemental Fig. 9), as demonstrated by
uptake of around 6 %ID/g in the knee joints (Fig. 5C, D). However, [68Ga]Ga‐FAPI‐04 accumulated in the
knee joints to a lower extent (2.8 ± 1.4 %ID/g). Blocking studies confirmed the specific uptake [18F]FGlc‐
FAPI in the knee joint region ([18F]FGlc‐FAPI: 6.0 ± 2.5 %ID/g vs. blocking: 0.36 ± 0.06 %ID/g; n = 4–6, all
values for 55 min p.i., Fig. 5D). Both tracers showed fast blood clearance (Fig. 5E). However, contrary to
the low concentration of [68Ga]Ga‐FAPI‐04 in the blood at 55 min p.i., [18F]FGlc‐FAPI showed a higher
residual blood concentration than [68Ga]Ga‐FAPI‐04 (Fig. 5E, insert).
12
DISCUSSION
To take advantage of the favorable properties of fluorine‐18, we envisaged the design and
synthesis of the alkyne‐bearing FAPI derivative 11, rendering click chemistry–based 18F‐
fluoroglycosylation feasible for the radiosynthesis of a new 18F‐labeled FAPI glycoconjugate, [18F]FGlc‐
FAPI. 18F‐Fluoroglycosylation is well established in our lab and has proven its potential in the
development of new PET radiopharmaceuticals (26,34). We have optimized an automated synthesis
module for 18F‐glycosylation, successfully running the radiosynthesis of four different 18F‐glycoconjugates
in a reliable manner, with a robust radioactivity yield of 15–20% (35). Currently, the module is used in a
clean room laboratory for the good manufacturing practice‐compliant manufacture of 18F‐
glycoconjugates, such as [18F]FGlc‐FAPI, to facilitate first‐in‐human PET imaging studies.
In vitro, cellular uptake in FAP‐transfected human fibrosarcoma cells, ligand binding,
internalization, and efflux of [18F]FGlc‐FAPI were similar to those of [68Ga]Ga‐FAPI‐04, as expected.
However, FAP binding affinity was five times lower for [18F]FGlc‐FAPI, but still within the three‐digit
nanomolar range. Also, plasma protein binding and lipophilicity were distinctly higher for [18F]FGlc‐FAPI
compared to [68Ga]Ga‐FAPI‐04. The replacement of the hydrophilic gallium complex with a triazolyl
glucosyl moiety (Fig. 1) induced lipophilicity in the molecule under physiological conditions (pH 7.4) and
increased the logD7.4 value from ‐2.4 to ‐1.0. This increase in lipophilicity is in agreement with higher
binding to plasma proteins. In fact, the difference in blood‐plasma binding between [18F]FGlc‐FAPI (45%)
and [68Ga]Ga‐FAPI‐04 (10%) was the most striking difference of all measured in vitro results. The most
abundant plasma protein, serum albumin, binds a large diversity of small organic molecules to shield
their lipophilic character, thereby increasing their solubility in plasma and prolonging the circulating half‐
life, as shown by our in vivo results for [18F]FGlc‐FAPI compared to [68Ga]Ga‐FAPI‐04.
13
In agreement to increased plasma protein binding and lipophilicity, biodistribution studies of
[18F]FGlc‐FAPI in mice revealed that the excretion of [18F]FGlc‐FAPI was through the kidneys as well as the
hepatobiliary pathway, which was comparable to the recently reported biodistribution of [64Cu]Cu‐FAPI‐
04 (36). For comparison, [68Ga]Ga‐FAPI‐04 is almost exclusively excreted by the kidneys and bladder (22).
Clearly, a pronounced nonspecific uptake of [18F]FGlc‐FAPI in the liver and intestine may be a problem for
the detection of abdominal FAP‐positive lesions in tumor patients. Therefore, first‐in‐human studies are
necessary to clarify whether the excretion of [18F]FGlc‐FAPI in human is similar to that in mice or whether
predominant renal elimination occurs in human.
Our results of small animal PET studies using HT1080hFAP xenografts demonstrated that
[18F]FGlc‐FAPI specifically bound to the tumor, revealing equal or even higher tumor uptake values
compared to [68Ga]Ga‐FAPI‐04. However, PET images with higher tumor‐to‐background ratios were
achieved with [68Ga]Ga‐FAPI‐04 due to the lower plasma protein binding and, thereby, faster blood
clearance of [68Ga]Ga‐FAPI‐04. In addition, we performed small animal PET studies in U87MG tumor–
bearing mice, confirming previous studies that, in vivo, U87MG cells recruit activated FAP‐positive
fibroblasts and/or, by communicating with cancer‐associated fibroblasts, express FAP (33); such cells
were previously used for preclinical imaging including PET with [68Ga]Ga‐FAPI‐04 (33,37).
The most eye‐catching observation using [18F]FGlc‐FAPI for PET imaging was its high uptake in
bone structures in U87MG tumor–bearing mice. In comparison with [68Ga]Ga‐FAPI‐04, the accumulation
of [18F]FGlc‐FAPI in the spine, femur, skull, knees, and shoulders was unexpectedly high. Blocking
experiments revealed that [18F]FGlc‐FAPI uptake was significantly diminished by coinjection of 11,
suggesting specific binding of [18F]FGlc‐FAPI in these regions. In vivo defluorination of [18F]FGlc‐FAPI was
excluded by the blocking experiments, and was, to date, never observed for 18F‐fluoroglycosylated
compounds.
14
Our in vitro and in vivo results demonstrated higher plasma protein binding of [18F]FGlc‐FAPI
compared to [68Ga]Ga‐FAPI‐04, negligible binding to DPP4‐positive blood cells, and a prolonged
circulation time in blood for [18F]FGlc‐FAPI. These may be the main reasons for the continuously slightly
increasing tumor uptake over time (Fig. 5B), but also for the higher uptake of [18F]FGlc‐FAPI in the knee
joints and in bone structures compared to [68Ga]Ga‐FAPI‐04.
Tran et al. have reported FAP expression of freshly isolated osteogenic cells of nonhematopoietic
origin, including multipotent bone marrow stromal cells (BMSCs) (38). The authors found that FAP‐
reactive T cells induced severe cachexia and dose‐limiting bone toxicity in mice, which appeared to be
the result of targeting of FAP‐expressing multipotent BMSCs. Therefore, the in vivo accumulation pattern
of [18F]FGlc‐FAPI may be ascribed to binding to the bone marrow in vivo. Moreover, it has been shown
that BMSCs are capable of directed migration towards various tumor types (39), and BMSCs are a source
of circulating stem cells that are recruited from the blood into peripheral solid organs in times of tissue
remodeling (40). In the U87MG tumor‐bearing nude mice model, such circulating murine BMSCs are
potential precursor cells of cancer‐associated fibroblasts (CAFs) that are recruited by stroma‐building
U87MG xenografts (9). Therefore, it is tempting to speculate that [18F]FGlc‐FAPI is capable of binding to
FAP‐positive BMSCs and circulating BMSCs, besides binding to FAP‐positive CAFs in the tumor tissue.
However, a significant role of the binding of [18F]FGlc‐FAPI to plasma proteins or most abundant proteins
in the murine synovial fluid of the knee joints cannot be ruled out. Although preliminary in vitro
experiments did not show significant binding to serum albumin, other shared proteins in plasma or
synovial fluid, such as several immunoglobulins (41), could bind and trap [18F]FGlc‐FAPI. [68Ga]Ga‐FAPI‐04
also accumulated in bone structures, however, to a much lower extent than [18F]FGlc‐FAPI. A reason for
this may be the extremely fast and not‐adhesive clearance properties of [68Ga]Ga‐FAPI‐04. The higher
specific uptake of [18F]FGlc‐FAPI by bone structures compared to [68Ga]Ga‐FAPI‐04 may be beneficial in
monitoring FAP‐dependent bone tissue remodeling in diseases such as rheumatoid arthritis by PET.
15
CONCLUSION
The glycoconjugate FAP inhibitor [18F]FGlc‐FAPI showed suitable in vitro and in vivo properties for
the tumor imaging of FAP expression in mice, but slower in vivo clearance properties than [68Ga]Ga‐FAPI‐
04. While hepatobillary clearance of [18F]FGlc‐FAPI in mice could be a disadvantage for successful clinical
translation, the high uptake of [18F]FGlc‐FAPI in murine bone structures may be an interesting property
of this tracer. The glycoconjugate, [18F]FGlc‐FAPI, is a viable alternative 18F‐labeled FAP inhibitor and a
promising candidate for translation to the clinic for first‐in‐human studies on FAP‐dependent bone tissue
remodeling in diseases such as rheumatoid arthritis by PET.
Disclosure
There are no conflicts of interest. This work was supported by the Emerging Fields Initiative (EFI)
of the Friedrich–Alexander University Erlangen–Nürnberg (grant 3_Nat_01, “Chemistry in live cells”) and
by the German Research Foundation (DFG, MA 4295/2‐1).
ACKNOWLEDGMENTS
The authors gratefully thank Dr. Annette Altmann and Prof. Uwe Haberkorn (University Hospital
Heidelberg and DKFZ, Heidelberg, Germany) for kindly providing the FAP‐transfected HT1080 cell line
and FAPI‐04. Additional acknowledgements go to Mr. Stefan Söllner, Mr. Manuel Geisthoff and Mrs.
Ulrike Ittstein for excellent technical assistance as well as to Mr. Jan Hellmann, Mrs. Anke Seitz, and Prof.
Peter Gmeiner (Chair of Pharmaceutical Chemistry, FAU) for excellent collaboration and help in chemical
analyses.
16
Key Points
Question:
Is an 18F‐labeled glycoconjugate inhibitor targeting fibroblast activation protein (FAP) an effective PET
tracer for the imaging of FAP expression?
Pertinent Findings:
This preclinical study showed that 18F‐FGlc‐FAPI had suitable in vitro and in vivo properties for the
imaging of FAP‐expressing tumors and is therefore a viable alternative 18F‐labeled FAP inhibitor.
Implications for Patient Care:
The glycoconjugate, [18F]FGlc‐FAPI, is a viable alternative 18F‐labeled FAP inhibitor and a promising
candidate for translation to the clinic for the first‐in‐human imaging of FAP expression in FAP‐related
diseases such as rheumatoid arthritis by PET.
17
REFERENCES
1. Park JE, Lenter MC, Zimmermann RN, Garin‐Chesa P, Old LJ, Rettig WJ. Fibroblast activation
protein, a dual specificity serine protease expressed in reactive human tumor stromal fibroblasts. J
Biol Chem. 1999;274:36505‐36512.
2. Garin‐Chesa P, Old LJ, Rettig WJ. Cell surface glycoprotein of reactive stromal fibroblasts as a
potential antibody target in human epithelial cancers. Proc Natl Acad Sci U S A. 1990;87:7235‐
7239.
3. Acharya PS, Zukas A, Chandan V, Katzenstein A‐LA, Puré E. Fibroblast activation protein: a serine
protease expressed at the remodeling interface in idiopathic pulmonary fibrosis. Hum Pathol.
2006;37:352‐360.
4. Bauer S, Jendro MC, Wadle A, et al. Fibroblast activation protein is expressed by rheumatoid
myofibroblast‐like synoviocytes. Arthrit Res Ther. 2006;8:R171.
5. Jacob M, Chang L, Pure E. Fibroblast Activation Protein in Remodeling Tissues. Curr Mol Med.
2012;12:1220‐1243.
6. Keane FM, Yao T‐W, Seelk S, et al. Quantitation of fibroblast activation protein (FAP)‐specific
protease activity in mouse, baboon and human fluids and organs. FEBS oopen bio. 2014;4:43‐54.
7. Kalluri R. The biology and function of fibroblasts in cancer. Nat Rev Cancer. 2016;16:582.
8. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer. 2006;6:392.
9. Shiga K, Hara M, Nagasaki T, Sato T, Takahashi H, Takeyama H. Cancer‐associated fibroblasts: their
characteristics and their roles in tumor growth. Cancers (Basel). 2015;7:2443‐2458.
10. Cohen SJ, Alpaugh RK, Palazzo I, et al. Fibroblast activation protein and its relationship to clinical
outcome in pancreatic adenocarcinoma. Pancreas. 2008;37:154‐158.
18
11. Welt S, Divgi CR, Scott AM, et al. Antibody targeting in metastatic colon cancer: a phase I study of
monoclonal antibody F19 against a cell‐surface protein of reactive tumor stromal fibroblasts. J Clin
Oncol. 1994;12:1193‐1203.
12. Mersmann M, Schmidt A, Rippmann JF, et al. Human antibody derivatives against the fibroblast
activation protein for tumor stroma targeting of carcinomas. Int J Cancer. 2001;92:240‐248.
13. LeBeau AM, Brennen WN, Aggarwal S, Denmeade SR. Targeting the cancer stroma with a
fibroblast activation protein‐activated promelittin protoxin. Mol Cancer Ther. 2009;8:1378‐1386.
14. Ryabtsova O, Jansen K, Van Goethem S, et al. Acylated Gly‐(2‐cyano) pyrrolidines as inhibitors of
fibroblast activation protein (FAP) and the issue of FAP/prolyl oligopeptidase (PREP)‐selectivity.
Bioorg Med Chem Lett. 2012;22:3412‐3417.
15. Eager RM, Cunningham CC, Senzer NN, et al. Phase II assessment of talabostat and cisplatin in
second‐line stage IV melanoma. BMC Cancer. 2009;9:263.
16. Scott AM, Wiseman G, Welt S, et al. A Phase I dose‐escalation study of sibrotuzumab in patients
with advanced or metastatic fibroblast activation protein‐positive cancer. Clin Cancer Res.
2003;9:1639‐1647.
17. Aertgeerts K, Levin I, Shi L, et al. Structural and kinetic analysis of the substrate specificity of
human fibroblast activation protein α. J Biol Chem. 2005;280:19441‐19444.
18. Tsai T‐Y, Yeh T‐K, Chen X, et al. Substituted 4‐carboxymethylpyroglutamic acid diamides as potent
and selective inhibitors of fibroblast activation protein. J Med Chem. 2010;53:6572‐6583.
19. Jansen K, Heirbaut L, Cheng JD, et al. Selective inhibitors of fibroblast activation protein (FAP) with
a (4‐quinolinoyl)‐glycyl‐2‐cyanopyrrolidine scaffold. ACS Med Chem Lett. 2013;4:491‐496.
20. Jansen K, Heirbaut L, Verkerk R, et al. Extended structure–activity relationship and
pharmacokinetic investigation of (4‐quinolinoyl) glycyl‐2‐cyanopyrrolidine inhibitors of fibroblast
activation protein (FAP). J Med Chem. 2014;57:3053‐3074.
19
21. Loktev A, Lindner T, Mier W, et al. A tumor‐imaging method targeting cancer‐associated
fibroblasts. J Nucl Med. 2018;59:1423‐1429.
22. Lindner T, Loktev A, Altmann A, et al. Development of quinoline‐based theranostic ligands for the
targeting of fibroblast activation protein. J Nucl Med. 2018;59:1415‐1422.
23. Giesel FL, Kratochwil C, Lindner T, et al. 68Ga‐FAPI PET/CT: biodistribution and preliminary
dosimetry estimate of 2 DOTA‐containing FAP‐targeting agents in patients with various cancers. J
Nucl Med. 2019;60:386‐392.
24. Kratochwil C, Flechsig P, Lindner T, et al. 68Ga‐FAPI PET/CT: Tracer Uptake in 28 Different Kinds of
Cancer. J Nucl Med. 2019;60:801‐805.
25. Loktev A, Lindner T, Burger E‐M, et al. Development of novel FAP‐targeted radiotracers with
improved tumor retention. J Nucl Med. DOI:jnumed. 118.224469.
26. Maschauer S, Einsiedel J, Haubner R, et al. Labeling and Glycosylation of Peptides Using Click
Chemistry: A General Approach to 18F‐Glycopeptides as Effective Imaging Probes for Positron
Emission Tomography. Angew Chem Int Ed. 2010;49:976‐979.
27. Maschauer S, Haubner R, Kuwert T, Prante O. 18F‐Glyco‐RGD peptides for PET imaging of integrin
expression: efficient radiosynthesis by click chemistry and modulation of biodistribution by
glycosylation. Mol Pharm. 2014;11:505‐515.
28. Potemkin R, Strauch B, Kuwert T, Prante O, Maschauer S. Development of 18F‐Fluoroglycosylated
PSMA‐Ligands with Improved Renal Clearance Behavior. Mol Pharm. 2020;17:933‐943.
29. Davis BG, Robinson MA. Drug delivery systems based on sugar‐macromolecule conjugates. Curr
Opin Drug Discov Devel. 2002;5:279‐288.
30. Moradi SV, Hussein WM, Varamini P, Simerska P, Toth I. Glycosylation, an effective synthetic
strategy to improve the bioavailability of therapeutic peptides. Chemical Science. 2016;7:2492‐
2500.
20
31. Mirfeizi L, Campbell‐Verduyn L, Dierckx RA, Feringa BL, Elsinga PH. Application of Click Chemistry
for PET. Curr Org Chem. 2013;17:2108‐2118.
32. Maschauer S, Ott JJ, Bernhardt G, Kuwert T, Keller M, Prante O. (18)F‐labelled triazolyl‐linked
argininamides targeting the neuropeptide Y Y1R for PET imaging of mammary carcinoma. Sci Rep.
2019;9:12990.
33. Röhrich M, Loktev A, Wefers AK, et al. IDH‐wildtype glioblastomas and grade III/IV IDH‐mutant
gliomas show elevated tracer uptake in fibroblast activation protein‐specific PET/CT. Eur J Nucl
Med Mol Imaging. 2019;46:2569‐2580.
34. Maschauer S, Prante O. Sweetening pharmaceutical radiochemistry by 18F‐fluoroglycosylation: a
short review. BioMed Res Int. 2014;2014:16.
35. Ott JJ. Selective neuropeptide and opioid receptor radioligands for imaging studies in vivo by
positron emission tomography (PET), PhD thesis, FAU Erlangen‐Nürnberg; 2018.
36. Watabe T, Liu Y, Kaneda‐Nakashima K, et al. Theranostics targeting fibroblast activation protein in
the tumor stroma: (64)Cu and (225)Ac labelled FAPI‐04 in pancreatic cancer xenograft mouse
models. J Nucl Med. 2019:doi:10.2967/jnumed.2119.233122.
37. Li J, Chen K, Liu H, et al. Activatable near‐infrared fluorescent probe for in vivo imaging of
fibroblast activation protein‐alpha. Bioconj Chem. 2012;23:1704‐1711.
38. Tran E, Chinnasamy D, Yu Z, et al. Immune targeting of fibroblast activation protein triggers
recognition of multipotent bone marrow stromal cells and cachexia. J Exp Med. 2013;210:1125‐
1135.
39. Hamada H, Kobune M, Nakamura K, et al. Mesenchymal stem cells (MSC) as therapeutic
cytoreagents for gene therapy. Cancer Sci. 2005;96:149‐156.
40. Kaplan RN, Psaila B, Lyden D. Niche‐to‐niche migration of bone‐marrow‐derived cells. Trends Mol
Med. 2007;13:72‐81.
21
41. Bennike T, Ayturk U, Haslauer CM, et al. A normative study of the synovial fluid proteome from
healthy porcine knee joints. J Proteome Res. 2014;13:4377‐4387.
22
Tables and Figures
TABLE 1
IC50 values as determined by radioligand displacement studies for FAP (using 177Lu‐FAPI‐04) and by
inhibition of DPP4 enzyme activity
Compound IC50 (FAP)
MW ± SD, (n = 3) IC50 (DPP4)
MW ± SD, (n = 2)
FAPI‐alkyne (11) 109 ± 60 nM 9.5 µM
FGlc‐FAPI 167 ± 110 nM 26.6 µM
FAPI‐04 49 ± 22 nM 7.9 µM
Ga‐FAPI‐04 32 ± 7 nM* n.d.**
* n.d. not determined; ** n = 2
23
TABLE 2
Biodistribution data of [18F]FGlc‐FAPI in HT1080hFAP xenografts for 30, 60, and 90 min p.i. and of
[18F]FGlc‐FAPI coinjected with competitor 11 (30 nmol/mouse, blocking) for 60 min p.i. Data are given as
mean values ± standard deviation (n = 2).
30 min 60 min 90 min 60 min blocking
Blood 2.3 ± 0.9 1.1 ± 0.6 0.7 ± 0.2 0.2 ± 0.1
Lung 1.5 ± 0.2 0.8 ± 0.3 0.5 ± 0.2 0.3 ± 0.1
Liver 4.0 ± 1.4 2.2 ± 0.1 1.7 ± 0.2 2.9 ± 1.2
Kidney 1.7 ± 0.1 1.5 ± 0.3 1.0 ± 0.2 1.4 ± 0.3
Heart 1.3 ± 0.5 0.7 ± 0.5 0.4 ± 0.2 0.1 ± 0.1
Spleen 1.0 ± 0.5 0.5 ± 0.1 0.3 ± 0.1 0.4 ± 0.3
Brain 0.13 ± 0.02 0.05 ± 0.02 0.07 ± 0.03 0.02 ± 0.01
Muscle 1.5 ± 0.3 1.2 ± 0.5 0.9 ± 0.2 0.2 ± 0.1
Femur 3.2 ± 0.5 3.4 ± 0.6 2.6 ± 0.5 0.6 ± 0.8
HT1080hFAP Tumor 1.6 ± 0.4 2.0 ± 0.5 1.6 ± 0.2 0.9 ± 0.2
Intestine 6.1 ± 4.4 6.4 ± 6.0 5.2 ± 6.1 8.0 ± 9.9
24
FIGURE 1. Molecular structure of [68Ga]Ga‐FAPI‐04 (22) and [18F]FGlc‐FAPI (this work).
25
FIGURE 2. 18F‐fluoroglycosylation of FAPI alkyne 11: a) [18F]fluoride, K222, K2CO3, KH2PO4, acetonitrile, 85°C, 10 min;
b) NaOH, 60°C, 5 min; (27) c) Cu(OAc)2, THPTA, sodium ascorbate, phosphate buffer pH 8, 11, 60°C, 15 min.
26
FIGURE 3. A) Competitive binding in HT1080hFAP cells of four reference compounds against [177Lu]Lu‐FAPI‐04 (n =
6–9). B) Tracer uptake of [18F]FGlc‐FAPI in HT1080hFAP cells, with and without blocking using FAPI alkyne 11 as a
competitor (n = 4). C) Internalization of [18F]FGlc‐FAPI in HT1080hFAP cells and efflux of [18F]FGlc‐FAPI out of
HT1080FAP cells after 1 hour incubation and exchange by fresh medium. Each data point represents the mean ±
standard deviation (n = 8).
27
FIGURE 4. Representative µPET images and time‐activity‐curves of HT1080hFAP xenografts using [18F]FGlc‐FAPI (A,
C) and [68Ga]Ga‐FAPI‐04 (B, D) and respective blocking experiments with competitor 11 (30 nmol/mouse).
28
FIGURE 5. Representative µPET images and time‐activity curves of U87MG xenografts using [18F]FGlc‐FAPI (A, C left
side), [18F]FGlc‐FAPI together with competitor 11 (100 nmol/mouse) (A, C middle), and [68Ga]Ga‐FAPI‐04 (A, C right
side). A, B) Projection of the U87MG tumor (red arrows) and time activity curves of U87MG tumor uptake in %ID/g.
C, D, E) Projection of the knee joints (dashed circles) and time activity curves for blood represented by the
integration of radioactivity in the heart.
1
Chemistry
General
All chemicals were purchased in the highest available quality and used without further purification. NMR
spectra were acquired on a Bruker Avance Nanobay V3-I 400 MHz or a Bruker Avance III HD 600 MHz
spectrometer. ESI mass spectra were recorded on Bruker amaZon SL or Bruker ESI timsTOF mass
spectrometers. The FPLC system (FlashPure, Büchi Labortechnik AG) was equipped with a UV-detector
(254 nm) and C-18 columns.
Synthesis of FGlc-FAPI
Precursor 1 as well as compounds 2 and 3 were synthesized according to Maschauer et al (1). Compound
10 was synthesized as described by Jansen et al (2). The analytical data was in agreement with the
literature. For the synthesis of compounds 4 to 11 the route was adapted form Lindner et al (3) and is
described below (Supplemental Figure 1). Since the reaction was up-scaled compared to the previous
procedure the carboxylic acid moiety was methylated to provide simplified workup without the use of a
preparative HPLC system. Finally, FGlc-FAPI was obtained by copper-catalyzed alkyne azide cycloaddition
with compounds 3 and 11 according to Maschauer et al (1).
NN
MMeeOO
OOHHOO
NN
HHOO
OOOO
NN
OO
OOOO
CCll
NN
OONNNN
BBBBrr33
DDCCMM,, 00 °°CC ---->> rr tt
CCll BBrr
HH22OO//MMeeOOHH,, 6600 °°CC,, 3300 mmiinn
KKIIDDMMFF,, 6600 °°CC,, 1188 hh
NNHHNN
NN
MMeeOO
OOOO
OOOO
NNaaOOHH
MMeeOOHH,, 6600 °°CC,, 1188 hh
HH22SSOO44 CCss22CCOO33
DDMMFF,, 6600 °°CC,, 1188 hh
NN
OONNNN
OOHHOO
HHBBTTUU,, HHOOBBtt,, DDIIPPEEAA
DDMMFF,, rrtt,, 2244 hh
NNFF
FFCCNN
OO NNHH22NN
NN OO
NN
HHNN NN
FFFF
NNCC HHOO
OO
44 55 66
77 88
99
1100
1111
NNNN
NNNN
NN OO
NN
HHNN NN
FFFF
NNCC HHOO
OOOOHHOO
HHOO OOHH
FF
CCuuSSOO44,, NNaaAAsscc
HH22OO,, 6600 °°CC,, 22 hh
OOHHOOHHOO OOHH
FF
NN33
33
FFGGllcc--FFAAPPII Supplemental Figure 1. Synthesis route of alkyne 11 and reference substance FGlc-FAPI.
2
Methyl 6-methoxyquinoline-4-carboxylate (5)
N
O
OO
6-Methoxyquinoline-4-methyl carboxylic acid 4 (0.291 g, 1.43 mmol) was suspended in MeOH (2 mL).
H2SO4 (0.382 mL, 7.16 mmol) was added to the reaction while cooling to 0°C. The reaction was stirred at
60 °C for 18 hours. The solvent was evaporated under reduced pressure. Saturated sodium bicarbonate
solution (10 mL) was added and extracted three times with dichloromethane. The combined organic
phases were washed three times with brine and dried over sodium sulfate. The solvent was evaporated
and the residue was dried under vacuum. The product 5 was obtained as light brown solid (0.303 g, 1.40
mmol, 97%). 1H NMR (600 MHz, DMSO-d6) δ 8.89 (d, J = 4.4 Hz, 1H), 8.08 (d, J = 2.8 Hz, 1H), 8.05 (d, J = 9.2 Hz, 1H),
7.94 (d, J = 4.4 Hz, 1H), 7.52 (dd, J = 9.2, 2.8 Hz, 1H), 3.99 (s, 3H), 3.92 (s, 3H).
LC-MS (ESI): m/z 217.77 [M+H]+ , calculated: 218.07 [M+H]+
Methyl 6-hydroxyquinoline-4-carboxylate (6)
N
HO
OO
Methyl 6-methoxyquinoline-4-carboxylate (0.155 g, 0.71 mmol) was dissolved in dry dichloromethane (4
mL) under argon atmosphere. BBr3 (1 M solution, 1.43 mL, 1.43 mmol) was added dropwise to the
reaction while cooling to 0 °C. The reaction was allowed to warm to room temperature and was stirred
overnight. Saturated sodium bicarbonate solution (10 mL) was added and the solution was extracted
with ethyl acetate three times. The combined organic phases were dried over sodium sulfate. The
solvent was evaporated and the residue was dried under vacuum. The product 6 was obtained as yellow
solid (0.120 g, 0.59 mmol, 82%). 1H NMR (600 MHz, DMSO-d6) δ 8.89 (d, J = 4.6 Hz, 1H), 8.03 (d, J = 9.1 Hz, 1H), 7.99 (d, J = 2.6 Hz, 1H),
7.96 (d, J = 4.6 Hz, 1H), 7.48 (dd, J = 9.1, 2.7 Hz, 1H), 3.99 (s, 3H).
LC-MS (ESI): m/z 203.72 [M+H]+ , calculated: 204.06 [M+H]+
3
Methyl 6-(4-chloropropoxy)quinoline-4-carboxylate (7)
N
O
OO
Cl
Methyl 6-hydroxyquinoline-4-carboxylate (0.105 g, 0.56 mmol) and Cs2CO3 (0.906 g, 2.78 mmol) were
dissolved in DMF (3 mL) and 1-bromo-3-chloropropane (0.19 mL, 1.95 mmol) was added. The reaction
was stirred at 60 °C overnight. Then the solvent was removed under reduced pressure. The residue was
partitioned between ethyl acetate and brine and the aqueous phase was extracted with ethyl acetate
two times. The combined organic phases were dried over sodium sulfate and the solvent was evaporated
under reduced pressure. Finally, the crude product was purified by column chromatography (n-
hexane/ethyl acetate, 2:1) using silica. The product 7 was obtained as yellow oil (0.055 g, 0.20 mmol,
35%). 1H NMR (600 MHz, DMSO-d6) δ 8.90 (dd, J = 4.4, 3.2 Hz, 1H), 8.05 (d, J = 9.2 Hz, 1H), 8.01 (d, J = 4.4 Hz,
1H), 7.95 (d, J = 4.4 Hz, 1H), 7.57 – 7.51 (m, 1H), 4.26 (t, J = 6.0, 2H), 3.99 (s, 3H), 3.91 – 3.81 (m, 2H), 2.32
– 2.23 (m, 2H).
LC-MS (ESI): m/z 279.82 [M+H]+ , calculated: 280.07 [M+H]+
Methyl 6-(3-(1-piperazinyl-4-prop-2-ynyl)propoxy)quinoline-4-carboxylate (8)
N
ONN
O O
Methyl 6-(4-chloropropoxy)quinoline-4-carboxylate 7 (0.050 g, 0.18 mmol) and 1-prop-2-ynyl-piperazine
(0.077 g, 0.62 mmol) were dissolved in DMF (2.5 mL) together with potassium iodide (0.206 g,
1.24 mmol). The reaction was stirred under argon atmosphere at 60 °C overnight. The solvent was
evaporated under reduced pressure. Saturated sodium bicarbonate solution (10 mL) was added and
extracted three times with ethyl acetate. The combined organic phases were dried over sodium sulfate
and the solvent was evaporated under reduced pressure. The crude product was purified by column
chromatography (dichloromethane/methanol, 15:1 + 1% ammonia solution) using silica. The product 8
was obtained as yellow oil (0.033 g, 0.09 mmol, 50%).
4
1H NMR (600 MHz, DMSO-d6) δ 8.90 (dd, J = 4.5, 3.2 Hz, 1H), 8.04 (d, J = 9.1 Hz, 1H), 8.00 (d, J = 4.4 Hz,
1H), 7.94 (d, J = 4.4 Hz, 1H), 7.56 – 7.49 (m, 1H), 4.27 (t, J = 6.2, 2H), 3.99 (s, 3H), 3.52 (s, 2H), 3.35 (s, 1H),
3.32 – 3.24 (m, 2H), 3.23 – 2.82 (m, 8H), 2.30 – 2.24 (m, 2H).
LC-MS (ESI): m/z 368.20 [M+H]+ , calculated: 368.19 [M+H]+
6-(3-(1-Piperazinyl-4-prop-2-ynyl)propoxy)quinoline-4-carboxylic acid (9)
N
ONN
O OH
Methyl 6-(3-(1-piperazinyl-4-prop-2-ynyl)propoxy)quinoline-4-carboxylate 8 (0.033 g, 0.09 mmol) was
dissolved in a mixture of methanol (1.5 mL) and water (0.5 mL) and sodium hydroxide solution (2 M, 0.2
mL) was added. The reaction was stirred at 60 °C for 30 min. The reaction was neutralized with HCl and
the solvent was evaporated under reduced pressure. The residue was purified by column
chromatography using a FPLC system with a C18-column (25% acetonitrile in H2O + 0.1% TFA). The
product 9 was obtained as a light yellow solid which could be a hydrochloride or sodium salt (0.049 g,
0.13 mmol, quantitative). 1H NMR (400 MHz, D2O) δ 8.95 (d, J = 5.5 Hz, 1H), 8.18 (d, J = 9.4 Hz, 1H), 7.99 (d, J = 5.5 Hz, 1H), 7.80 (dd,
J = 9.4, 2.6 Hz, 1H), 7.70 (d, J = 2.7 Hz, 1H), 4.37 (t, J = 5.7 Hz, 2H), 3.85 – 3.60 (m, 8H), 3.85 – 3.60 (m,
2H), 3.85 – 3.60 (m, 1H), 3.58 (t, J = 7.8 Hz, 2H), 2.39 (p, J = 7.6, 5.8 Hz, 2H).
LC-MS (ESI): m/z 354.13 [M+H]+ , calculated: 354.17 [M+H]+
(S)-N-(2-(2-Cyano-4,4-difluoropyrrolidin-1-yl)-2-oxoethyl)-6-(3-(1-piperazinyl-4-prop-2-
ynyl)propoxy)quinoline-4-carboxamide (11)
6-(3-(1-Piperazinyl-4-prop-2-ynyl)propoxy)quinoline-4-carboxylic acid 9 (0.065 g, 184 µmol) and HOBt
(0.037 g, 276 µmol) were dissolved in DMF (1.2 mL). Then DIPEA (78 µL, 460 µmol) and HBTU (0.084 g,
221 µmol) were added to the reaction and stirred for 10 min before addition of compound 10 (0.083 g,
NN O
N
HN
N
F F
NC HO
O
5
230 µmol) dissolved in DMF (0.6 mL) and DIPEA (78 µL, 460 µmol). The reaction was stirred for 24 hours
at room temperature before quenching it with H2O. The solvent was evaporated and the residue purified
by column chromatography (dichloromethane:methanol 9:1) using silica. This gave a light colored solid
which was again purified using a FPLC system with a C18-column (10% acetonitrile in H2O + 0.1% TFA).
The product 11 was obtained as slightly yellow oil (0.094 g, 179 µmol, 97%). 1H NMR (400 MHz, DMSO-d6) δ 9.13 (t, J = 6.0 Hz, 1H), 8.87 (d, J = 4.4 Hz, 1H), 8.04 (d, J = 9.2 Hz, 1H),
7.92 (d, J = 2.8 Hz, 1H), 7.59 (d, J = 4.4 Hz, 1H), 7.50 (dd, J = 9.2, 2.8 Hz, 1H), 5.15 (dd, J = 9.4, 2.9 Hz, 2H),
4.38-4.20 (m, 2H), 4.30-4.11 (m, 2H), 4.25 (t, J = 6.0 Hz, 2H), 3.50 (s, 2H), 3.36 (s, 1H), 3.30 (t, J = 8.1 Hz,
2H), 3.18-2.80 (m, 8H), 2.22 (p, J = 14.9, 7.0 Hz, 2H).
LC-MS (ESI): m/z 525.18 [M+H]+ , calculated: 525.23 [M+H]+
N-(2-((R)-2-Cyano-4,4-difluoropyrrolidin-1-yl)-2-oxoethyl)-6-(3-(4-((1-((2R,3R,4S,5S,6S)-6-
(fluoromethyl)-3,4,5-trihydroxytetrahydro-2H-pyran-2-yl)-1H-1,2,3-triazol-4-yl)methyl)piperazin-1-
yl)propoxy)quinoline-4-carboxamide (FGlc-FAPI)
NN
NN
N O
N
HN
N
F F
NC HO
OOHO
HOOH
F
Compound 11 (3.15 mg, 6.01 µmol) was dissolved in H2O (300 µL) and 6-deoxy-6-fluoro-β-D-
glucopyranosyl-1-azide 3 (1.57 mg, 7.58 µmol) dissolved in H2O (150 µL) was added. A CuSO4
pentahydrate solution (75 µL, 0.2M) was added and the reaction was started by addition of a sodium
ascorbate solution (75 µL, 0.6M) and heating to 60 °C. After stirring for 2 hours the reaction was diluted
with H2O and purified by HPLC (method 1) followed by lyophilization to obtain FGlc-FAPI as a pale white
solid (2.9 mg, 3.96 µmol, 66%). 1H NMR (600 MHz, D2O) δ 8.87 (d, J = 4.8 Hz, 1H), 8.27 (s, 1H), 8.11 (d, J = 9.3 Hz, 1H), 7.79 (d, J = 4.8 Hz,
1H), 7.70 (d, J = 2.7 Hz, 1H), 7.62 (dd, J = 9.3, 2.7 Hz, 1H), 5.82 (d, J = 9.3 Hz, 1H), 5.16 (dd, J = 9.0, 3.9 Hz,
1H), 4.68 (d, J = 3.7 Hz, 1H), 4.43 – 4.35 (d, 2H), 4.33 (t, J = 6.2, 2H), 4.35 – 4.13 (ddt, J = 20.3, 10.5 Hz,
1H), 4.06 (t, J = 9.2 Hz, 1H), 3.98 – 3.88 (m, 1H), 3.94 (s, 2H), 3.75 (t, J = 9.2 Hz, 1H), 3.70 (t, J = 9.6 Hz,
1H), 3.41 (t, J = 7.8 Hz, 2H), 3.20 – 2.80 (m, 8H), 3.05 – 2.89 (m, 2H), 2.32 (p, J = 6.1 Hz, 2H).
LC-MS (ESI): m/z 732.23 [M+H]+ , calculated: 732.30 [M+H]+
HRMS (ESI): m/z 732.3076 [M+H]+ , calculated: 732.3071 [M+H]+
6
Radiochemistry
General
The FAPI-04 precursor was obtained from the Department of Nuclear Medicine, University of Heidelberg,
Germany.
The HPLC-System (Series 1100, Agilent) was equipped with a VWD UV-Lamp (detection at 214 or 254 nm)
and was for radio-HPLC additionally connected to a radio-detector (500 TR Series, Packard). Computer
analysis of the HPLC data was performed using FLO-One software (Canberra Packard). The conditions
used for preparative and analytical HPLC are described in methods 1-5.
Method 1: Kromasil C8, 125 × 8 mm, 0-30% acetonitrile (0.1% trifluoroacetic acid (TFA)) in water (0.1%
TFA) in a linear gradient over 20 min, 4 mL/min.
Method 2: Chromolith RP-18e, 100 × 4.6 mm, 0-30% acetonitrile (0.1% TFA) in water (0.1% TFA) in a
linear gradient over 5 min, 4 mL/min.
Method 3: Kromasil C8, 250 × 4.6 mm, 0-30% acetonitrile (0.1% TFA) in water (0.1% TFA) in a linear
gradient over 20 min, 1.5 mL/min.
Method 4: Chromolith RP-18e, 100 × 4.6 mm, 10-50% acetonitrile (0.1% TFA) in water (0.1% TFA) in a
linear gradient over 5 min, 4 mL/min.
Method 5: Chromolith RP-18e, 100 × 4.6 mm, 0-70% acetonitrile (0.1% TFA) in water (0.1% TFA) in a
linear gradient over 5 min, 4 mL/min.
The time difference between UV detector and radio-detector was 0.06 min for method 2 (flow: 4
mL/min) and 0.15 min for method 3 (flow: 1.5 mL/min).
A γ-counter (Wallac Wizard, Perkin Elmer, Waltham, MA, USA) was used to measure the radioactivity of
samples generated in cell-based experiments, in vitro experiments and the determination of the
biodistribution.
No-carrier-added [18F]fluoride was produced through the 18O(p,n)18F reaction on a PETtrace 800
cyclotron using H2[18O]O as a target at the University Hospital Würzburg (Center of Radiopharmacy,
Würzburg, Germany). [177Lu]LuCl3 was obtained from Isotope Technologies Garching (ITG, Garching).
[68Ga]Ga3+ was obtained by elution from a 68Ge/68Ga-generator (Eckert & Ziegler, Berlin).
Characterization of [18F]FGlc-FAPI
For characterization of [18F]FGlc-FAPI, a coinjection with non-radioactive FGlc-FAPI was performed using
method 2 on the HPLC system (tR= 2.90 min, Supplemental Figure 2). The time delay of 0.06 min between
7
the radio and the UV detector was corrected by the processing program. A coinjection using method 3
was additionally performed which showed the same retention time for radio and UV peak (tR= 9.9 min,
data not shown).
Supplemental Figure 2. Coinjection of [18F]FGlc-FAPI (above) and non-radioactive reference FGlc-FAPI
(below).
Supplemental Figure 3. Chromatogram of a representative isolation of the fraction containing [18F]FGlc-
FAPI (dashed lines) by semi-preparative HPLC (method 1).
n (FGlc-FAPI) [nmol]
UV p
eak
area
[mAU
]
0 5 10 15 200
1000
2000
3000
4000
Supplemental Figure 4. Calibration curve of the HPLC UV detector for the quantification of the amount
of FGlc-FAPI (linear regression of the amount of FGlc-FAPI vs. peak area (UV, 214 nm)).
8
Radiosynthesis of [177Lu]Lu-FAPI-04
For the radiosynthesis of 177Lu-labeled FAPI-04 2-12 µL [177Lu]LuCl3-solution (2-12 µL, 20-120 MBq) were
diluted with HEPES-solution (0.5M, 200 µL, pH 5) and FAPI-04 precursor (4 µL, 4.6 nmol) was added. The
reaction was heated to 99 °C in an Eppendorf tube and the reaction progress was monitored on HPLC (tR
= 1.19 min, method 4). After 15-20 min, the radiochemical yield was 92 ± 5% (n = 5), as determined by
integration of the radio-HPLC peaks and [177Lu]Lu-FAPI-04 was used without further purification or was
isolated on a C18-cartridge and eluted with ethanol, followed by evaporation and formulation with cell
culture media for the use of [177Lu]Lu-FAPI-04 as a radioligand in the competitive FAP binding assay.
Radiosynthesis of [68Ga]Ga-FAPI-04
[68Ga]Ga3+ was eluted with 0.1M HCl (10 mL) and trapped on a PS-H+-cartridge (Chromafix, Machery-
Nagel) which was then eluted with a NaCl solution (5M, 1 mL). The FAPI-04 precursor (3 µL, 3.44 nmol)
was diluted with HEPES solution (2.5M, 300 µL, pH 5) and the 68Ga-eluate was added (500 µL, 100 - 250
MBq). The reaction was incubated in an Eppendorf tube for 10 min at 99 °C to obtain a radiochemical
yield of 86 ± 11% (n=3) as determined by HPLC (tR = 1.10 min, method 4). The reaction mixture was fixed
on a C18-cartridge and eluted with ethanol, which was finally evaporated. The radiotracer was obtained
in a radioactivity yield of 53 ± 2% in a total synthesis time of 30 min, with a radiochemical purity of >97%
and an apparent molar activity of approximately 15-40 GBq/µmol.
In-vitro studies
Determination of the partition coefficient (logD7.4)
An aliquot (50 kBq) of [18F]FGlc-FAPI or [68Ga]Ga-FAPI-04 was added to an Eppendorf tube with 500 µL
PBS (pH 7.4) and 500 µL n-octanol and was vortexed for 1 min. After centrifugation (20000 rpm, 1 min),
three samples (each 100 µL) were taken from each phase and measured in a γ-counter. The cpm values
were calculated as logarithm of the octanol/PBS ratio. The determination of logD7.4 was performed in
triplicate in two independent experiments. The values of logD7.4 were given as mean values ± standard
deviation.
Tracer stability in human serum
An aliquot of [18F]FGlc-FAPI or [68Ga]Ga-FAPI-04 (20 µL) was added to human serum (400 µL) and
incubated at 37 °C. At defined time points (5, 15, 25, 35, 45, 55 min), samples (50 µL) were taken and
added to a mixture of 10% TFA (50 µL) and acetonitrile (50 µL). The mixture was centrifuged (2 min,
9
20,000 g) and a sample of the supernatant (10 µL) was analyzed by radio HPLC using method 5 ([18F]FGlc-
FAPI) or method 4 ([68Ga]Ga-FAPI-04).
Plasma protein binding
The binding of tracers to plasma proteins was determined using a micro column separation method. The
radiotracer (1 µL, 100-500 kBq) was added to human plasma (100 µL) and incubated for 10 min at 37 °C.
The microcolumns (illustra MicroSpin G-50 columns, GE Healthcare) were preconditioned as described
by the manufacturer and a sample of the plasma (50 µL) was carefully applied to the resin. The columns
were placed in an Eppendorf tube and centrifuged (2 min, 2000 g). The eluate and the remaining resin
were both collected separately and measured using a γ-counter. The radioactivity which eluted from the
resin, representing the plasma protein bound tracer, was calculated as percentage of the total amount of
radioactivity in the sample. The experiment was performed twice each in triplicate.
Cell-based experiments
For the cell-based experiments the stably FAP-transfected fibrosarcoma cell line HT1080hFAP was used
which was obtained from the Department of Nuclear Medicine, University of Heidelberg, Germany. The
cells were cultivated in Dulbecco's modified Eagle's medium containing 10% fetal calf serum at 37 °C and
a CO2 content of 5%. If not stated otherwise, the experiments were performed in 48-well plates with a
final confluence of 80-100%.
Competitive FAP binding assay
Competitive binding experiments were performed in 96-well plates using [177Lu]Lu-FAPI-04 as
radioligand. For this HT1080hFAP cells (100,000 cells/well) were seeded 24 h before the experiment,
washed with media once before applying the non-radioactive test substances FGlc-FAPI, FAPI alkyne 11,
Ga-FAPI-04 and DOTA-FAPI-04 in concentrations ranging from 1 nM to 5 µM in 90 µL media. For FGlc-
FAPI, FAPI alkyne 11 and DOTA-FAPI-04 the concentrations were achieved by weighting and dissolving of
the substance, for Ga-FAPI-04 by HPLC measurement using a calibration curve. [177Lu]Lu-FAPI-04 was
subsequently added (in 10 µL media, approximately 0.1 MBq/well) and incubated for 60 min at 37 °C.
After setting the plate on ice, the cells were washed with media and cold PBS. Finally, the cells were
lysed with 1 M NaOH and counted. IC50 values were calculated using Prism 6.0 Software (GraphPad, One-
Site Fit logIC50) from the raw data (in cpm). The experiments were performed in triplicates and two to
three independent experiments were conducted.
10
DPP4 activity assay
For testing of inhibitory effects of FAPI-04, FAPI alkyne 11 and FGlc-FAPI on DPP4 inhibitors were
incubated in concentrations between 1 nM and 1 mM together with 125 mU purified DPP4 (Sigma
Aldrich, Germany) in 130 µL Hepes buffer (pH 7.6) and incubated for 10 minutes at 37°C in 96 well
microtiter plates. Purified DPP4 with activity between 3.9-125 mU was used as standards. After addition
of Gly-Pro-pNA (4-nitroaniline, Sigma Aldrich, Germany) as DPP4 substrate (100 µL of 1 mM solution)
kinetics of enzymatic pNA release (25 cycles in 20 min) was monitored at 405 nm in a microplate ELISA-
reader (Synergy II, BioTek Instruments Inc., Germany). IC50 values were calculated using Prism 6.0
software (GraphPad, One-Site Fit logIC50, Supplemental Fig. 5). All measurements were carried out in
duplicates.
-8 -6 -4 -20
50
100
150
Log c (M)
DPP4
act
ivity
(mU) FAPI-04
FAPI alkyneFGlc-FAPI
Supplemental Figure 5. Inhibition of DPP4 activity by FAPI-04, FAPI alkyne 11, and FGlc-FAPI (n = 2).
Cellular uptake
For determination of cellular uptake cells were seeded 2 or 3 days before the experiment and were
washed with medium once before incubation with the radiotracer. The cells were incubated for 5, 15,
30, 60 or 120 min at 37 °C in a total volume of 250 µL in binding buffer. Blocking of specific binding was
achieved by the addition of FAPI alkyne 11 in an end concentration of 1 µM. After the incubation time
the plates were put on ice and washed twice with cold PBS. After detachment and lysis of the cells by 2
min incubation in 0.1M NaOH the radioactivity was determined in a γ-counter. The counting results were
normalized to 1 million cells and calculated as the percentage of applied dose (%ID/1 million cells). The
experiment was performed in quadruplicate for each time point.
The uptake of [18F]FGlc-FAPI was determined in HT1080hFAP cells and compared to the uptake in FAP-
negative MCF-7 cells, demonstrating FAP-specific uptake of [18F]FGlc-FAPI (Supplemental Figure 6). The
11
incubation was performed for 60 min at 37 °C as described above for the uptake experiments. The
experiment was performed in quadruplicate for each condition.
upta
ke [%
/1 m
io c
ells
]
0
2
4
6
8
HT1080hFAP cellsblocked HT1080hFAP cellsMCF-7 cellsblocked MCF-7 cells
Supplemental Figure 6. Uptake of [18F]FGlc-FAPI in HT1080hFAP cells (with FAP expression) and MCF-7 cells (without FAP expression). Nonspecific uptake was defined in the presence of an excess of FAPI alkyne 11 in a concentration of 1 µM. Each bar represents the mean ± SD (n = 4). Internalization and efflux
The internalization of the radiotracer was determined after incubation times of 5, 15, 30, 60 and 120
minutes. After the respective incubation times the cells were washed with PBS (300 µL/well) twice and
were then incubated two times for 1 min each with an acid wash solution (0.02 M NaOAc, pH 5) to
detach extracellulary bound tracer. After washing one more time with PBS the cells were lysed by
addition of 0.1 M NaOH for 2 min. Radioactivity of the cell lysates and the acid wash solutions was
measured in the γ-counter. Values were calculated by the relation of counts in the acid wash solution
(extracellularly bound tracer) to counts in the corresponding cell lysate (internalized tracer). The
experiments were performed in quadruplicates and two independent experiments were conducted.
Efflux of the radiotracer out of the cells was measured after a 60 min incubation at 37 °C. After setting
the plate on ice and washing with cold PBS fresh media was applied on the cells. The media was
collected 5, 15, 30, 60 and 120 min after its addition to the cells. After washing with cold PBS the
remaining cells were lysed with 0.1 M NaOH. Radioactivity of the cell lysates and the media solutions was
measured in the γ-counter. The values are calculated by the relation of counts in the collected media
(effluxed tracer) to counts in the corresponding cell lysate (still internalized tracer). The experiments
were performed in quadruplicates and two independent experiments were conducted.
12
Animal experiments
All animal experiments were approved by the local animal protection authorities (Government of Central
Franconia, Germany, No. 55.2 2532-2-279) and performed in accordance with the relevant E.U.
guidelines and regulations.
Tumor xenotransplantation
Female athymic nude mice (CD1-Foxn1/nu, homozygous) were obtained from Charles River Laboratories
(Germany) at 4 weeks of age and were kept under standard conditions (12 h light/dark) with nude mice
food and water available ad libitum for at least 5 weeks. Viable cells were harvested, suspended in
PBS/Matrigel (1:1, 100 μL) and were injected subcutaneously in the left (5x106 HT1080hFAP cells/mouse
or 2x106 U87MG cells/mouse) back of the mice. The body weight and diameter of each tumor were
determined and documented every 2-3 days. Two to four weeks after inoculation, the mice, now
weighing about 30 g and bearing tumors of 6.5 - 15 mm (for HT1080hFAP) and 2.5 - 7 mm (for U87MG) in
diameter, were used for biodistribution and small-animal PET studies.
13
bloodlung
liver
kidney
heart
splee
nbrai
n
muscle
femur
HT1080
hFAP tumor
intestin
e0
5
10
15[18
F]FG
lc-F
API
(%ID
/g) 30 min
60 min90 min
bloodlung
liver
kidney
heart
splee
nbrai
n
muscle
femur
HT1080
hFAP tumor
intestin
e0
5
10
15
20
[18F]
FGlc
-FA
PI (%
ID/g
) 60 min control60 min blocking
A B
Supplemental Figure 7. A) Biodistribution of [18F]FGlc-FAPI in HT1080hFAP xenografts after 30, 60, and 90 min p.i. (n = 2). B) Comparison of biodistribution of [18F]FGlc-FAPI in HT1080hFAP xenografts at 60 min p.i., without (control) and with coinjection of 11 (blocking, 30 nmol/mouse, n = 2).
0
2
4
6
8
10
12
Tum
or u
ptak
e (%
ID/g
)12
0-13
0 m
in p
.i.
HT1080hFAP U87MG
ControlBlocking
Supplemental Figure 8. Tumor uptake values [18F]FGlc-FAPI at 120-130 min p.i. derived from PET scans
(additional data for Fig. 4 and Fig. 5 of the main manuscript).
14
Supplemental Figure 9. Representative µPET images of U87MG xenografted mice in sagittal view using [18F]FGlc-FAPI (left), [18F]FGlc-FAPI together with alkyne 11 (middle), and [68Ga]Ga-FAPI (right).
References
1. Maschauer S, Haubner R, Kuwert T, Prante O. 18F-Glyco-RGD peptides for PET imaging of integrin expression: efficient radiosynthesis by click chemistry and modulation of biodistribution by glycosylation. Mol Pharm. 2014;11:505-515. 2. Jansen K, Heirbaut L, Verkerk R, et al. Extended structure–activity relationship and pharmacokinetic investigation of (4-quinolinoyl) glycyl-2-cyanopyrrolidine inhibitors of fibroblast activation protein (FAP). J Med Chem. 2014;57:3053-3074. 3. Lindner T, Loktev A, Altmann A, et al. Development of quinoline-based theranostic ligands for the targeting of fibroblast activation protein. J Nucl Med. 2018;59:1415-1422.