Cite This: J. Med. Chem. 2017, 60, 8385-8393
`
`Article
`
`pubs.acs.org/jmc
`
`Inhibitor-Decorated Polymer Conjugates Targeting Fibroblast
`Activation Protein
`Petra Dvor ̌ákova ́,†,‡ Petr Bušek,§ Tomáš Knedlík,†,∥ Jiří Schimer,†,∥ Toma ́š Etrych,⊥ Libor Kostka,⊥
`Lucie Stollinova ́ Šromova ́,§ Vladimír Šubr,⊥ Pavel Šácha,*,†,∥ Aleksi Šedo,*,§ and Jan Konvalinka*,†,∥
`†Institute of Organic Chemistry and Biochemistry of The Czech Academy of Sciences, Flemingovo nám 2, 16610 Prague 6, Czech
`Republic
`‡Department of Cell Biology, Faculty of Science, Charles University, Viničná 7, 12843 Prague 2, Czech Republic
`§Institute of Biochemistry and Experimental Oncology, First Faculty of Medicine, Charles University, U Nemocnice 5, 12853 Prague
`2, Czech Republic
`∥Department of Biochemistry, Faculty of Science, Charles University, Hlavova 8, 12843 Prague 2, Czech Republic
`⊥Institute of Macromolecular Chemistry, The Czech Academy of Sciences, Heyrovského nám 2, 16206 Prague 6, Czech Republic
`
`*S Supporting Information
`
`ABSTRACT: Proteases are directly involved in cancer pathogenesis. Expression of fibroblast activation protein (FAP) is
`upregulated in stromal fibroblasts in more than 90% of epithelial cancers and is associated with tumor progression. FAP
`expression is minimal or absent in most normal adult tissues, suggesting its promise as a target for the diagnosis or treatment of
`various cancers. Here, we report preparation of a polymer conjugate (an iBody) containing a FAP-specific inhibitor as the
`targeting ligand. The iBody inhibits both human and mouse FAP with low nanomolar inhibition constants but does not inhibit
`close FAP homologues dipeptidyl peptidase IV, dipeptidyl peptidase 9, and prolyl oligopeptidase. We demonstrate the
`applicability of this iBody for the isolation of FAP from cell lysates and blood serum as well as for its detection by ELISA,
`Western blot, flow cytometry, and confocal microscopy. Our results show the iBody is a useful tool for FAP targeting in vitro and
`potentially also for specific anticancer drug delivery.
`
`■ INTRODUCTION
`Proteases in tumor and stromal cells play an important role in
`cancer progression by promoting tumor cell
`invasion and
`metastasis as well as facilitating neovascularization.1−3 Because
`of their pathogenic role and differential expression in tumor
`tissue, some proteases, such as matrix metalloproteinases, hold
`promise as therapeutic and diagnostic targets. However, several
`large-scale clinical trials testing low-molecular-weight matrix
`metalloproteinase inhibitors failed to show improvement in
`clinical outcomes. This was most likely due to the imperfect
`specificity of
`the tested compounds, on-target side effects
`caused by interference with the physiological functions of the
`proteases, and the incompletely understood involvement of the
`targeted proteolytic enzymes as well as their substitutability by
`other proteases in disease progression.4,5
`Since its discovery in the late 1980s, fibroblast activation
`protein (FAP; seprase, surface expressed protease) has been
`
`considered an interesting potential target for cancer therapeu-
`tics and diagnostics.6 FAP is expressed in stromal fibroblasts in
`more than 90% of epithelial cancers,6 and its expression is also
`in multiple myeloma7 and
`increased in stromal cells
`glioblastoma.8 In addition, FAP is expressed in malignant
`cells in glioblastoma8 and pancreatic,9,10 breast,11 colorectal,12
`cervical,13 and oral squamous cell14 carcinomas. Although the
`effects of FAP are tumor specific and in certain cancers FAP
`may even act as a tumor suppressor,15 it has been established in
`several cases that high FAP expression contributes to the
`invasiveness and increased proliferation of
`the tumor
`cells.14,16,17 Moreover, FAP in the bioptic material may be a
`prognostic marker of aggressive tumor progression, especially
`when expressed by cancer cells (recently reviewed in ref 18).
`
`Received: May 30, 2017
`Published: September 27, 2017
`
`© 2017 American Chemical Society
`
`8385
`
`DOI: 10.1021/acs.jmedchem.7b00767
`J. Med. Chem. 2017, 60, 8385−8393
`
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`Downloaded via UNIV OF MARYLAND COLG PARK on December 10, 2024 at 22:29:37 (UTC).
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`Recent studies also demonstrated that FAP expression is
`increased in various nonmalignant disease states accompanied
`by extracellular matrix remodeling such as in idiopathic
`pulmonary fibrosis,19 liver cirrhosis,20 rheumatoid arthritis,21
`myocardial infarction,22 and advanced atherosclerotic plaques.23
`With the exception of pancreatic alpha cells,24 mesenchymal
`bone marrow cells,25 and endometrial stroma during the
`proliferative phase,26 FAP expression is minimal or absent in
`the majority of normal adult tissues.6,27,28 Its soluble form
`devoid of
`the cytoplasmic and transmembrane regions is
`physiologically present in blood plasma (known as antiplasmin-
`cleaving enzyme, APCE). The origin and function of blood
`plasma FAP are largely unknown.29 Given the limited
`expression of FAP in human tissues under physiological
`conditions, FAP seems to be a promising molecule for targeting
`cancer stroma as well as some types of transformed cancer cells.
`FAP is a type II transmembrane protein belonging to the S9B
`oligopeptidase subfamily of serine proteases. It consists of 760
`amino acids: 6 form the N-terminal cytoplasmic tail, 20 the
`transmembrane part, and the remaining 734 amino acids are
`part of a large extracellular C-terminal domain.30 FAP requires
`dimerization of two 97 kDa subunits for its catalytic activity31
`and cleaves off dipeptides from the N-terminus of its substrates
`after a proline residue (N-Xxx-Pro-). In addition, FAP exhibits
`postproline endopeptidase activity,32 which is thought
`to
`the extracellular matrix.33
`contribute to the remodeling of
`Nevertheless, it is likely that at least some of the complex effects
`of FAP in cancer-associated fibroblasts and cancer cells are
`mediated by nonhydrolytic protein−protein interactions. For
`example, introduction of FAP endowed normal fibroblasts with
`an inflammatory phenotype, which was mediated by the
`activation of FAK−Src−JAK2 signaling pathway by the
`urokinase-type plasminogen activator
`receptor (uPAR), a
`known FAP-interacting membrane protein.34 Similarly,
`the
`suppression of FAP in oral squamous cell carcinoma cells
`inactivated the PTEN/PI3K/AKT and Ras-ERK pathways and
`repressed the expression of genes regulating the epithelial−
`mesenchymal transition, thereby reducing the proliferation and
`invasiveness of these cells.14 Thus, the involvement of FAP in
`the pathogenesis of human malignancies is complex and seems
`to be cancer-type specific, which may have contributed to the
`failure of early clinical trials assessing FAP targeting with the
`rather nonspecific low-molecular-weight inhibitor talabostat35
`or the humanized antibody sibrotuzumab.36
`Recently, we have described novel biochemical tools called
`iBodies for the targeting of proteins with known ligands.37 The
`iBodies are based on a water-soluble and biocompatible N-(2-
`hydroxypropyl)methacrylamide (HPMA) copolymer carrier
`decorated with low-molecular-weight compounds such as
`enzyme inhibitors used as targeting ligands. The use of iBodies
`offers
`several advantages over classical approaches with
`antibodies. The iBodies are highly modular and versatile;
`conjugates containing virtually any desired compound can be
`easily prepared. Synthetic HPMA conjugates are well-
`characterized compounds for biomedical applications and
`have long been used as carriers for drug delivery to solid
`tumors, often making use of the enhanced permeability and
`retention (EPR) effect.38−40 Importantly, the molecular weight
`of the HPMA backbone can be easily adjusted to specifically
`tailor
`the pharmacokinetic properties. To prepare a new
`platform that would allow FAP targeting in cancer, we set
`out to develop FAP-targeting iBodies.
`
`Dipeptidyl peptidase IV (DPP-IV), the closest homologue of
`FAP, sharing 52% amino acid sequence identity, is a broadly
`expressed cell-surface serine protease involved in several
`physiological processes,
`including regulation of glucose
`metabolism41 and T-cell activation.42 Besides its physiological
`roles, DPP-IV was implicated in the pathogenesis of several
`cancers, acting as a tumor suppressor or promotor depending
`on the tumor type (recently reviewed in ref 43). The high
`selectivity of anti-FAP iBodies is critical to avoid interference
`with DPP-IV and to decrease the risk of undesired side effects.
`Jansen et al. recently developed low nanomolar FAP inhibitors
`based on a (4-quinolinoyl)glycyl-2-cyanopyrrolidine scaffold,
`which showed high selectivity against
`related proteases,
`including DPP-IV and prolyl-specific proteases prolyl oligo-
`peptidase (PREP) and dipeptidyl peptidase 9 (DPP9).44−46
`In this work, we prepared anti-FAP iBody containing a highly
`specific FAP inhibitor as a targeting ligand and tested its utility
`using FAP-expressing malignant glioblastoma cells. We
`demonstrate that
`this iBody can be used for the specific
`detection of FAP in various biological matrices by a number of
`biochemical methods, and we show that it is suitable for the
`specific targeting and visualization of FAP as well as the
`inhibition of its enzymatic activity.
`
`■ RESULTS
`Expression of Recombinant Human and Mouse FAP
`and DPP-IV. To test
`the selectivity of
`the compounds
`described in this study, we prepared recombinant human and
`mouse FAP and DPP-IV bearing cleavable N-terminal
`purification tags (SF-tag or Avi-tag) (Figure 1a). All proteins
`were expressed in Drosophila S2 cells and purified via affinity
`chromatography according to previously published protocols
`(using Streptavidin Mutein matrix for Avi-tagged constructs48
`or Strep-Tactin resin for SF-tagged versions49). Originally, all
`four proteins were prepared with an Avi-tag; however, the Avi-
`hFAP and Avi-mFAP expression yields were not sufficient for
`biochemical characterization and subsequent experiments.
`Therefore, we replaced the Avi-tag with the recently described
`SF-tag, which comprises two Strep-tags and a Flag-tag.49 The
`resulting constructs, SF-hFAP and SF-mFAP, were expressed in
`larger quantities compared to their Avi-tagged counterparts and
`could be obtained in quantity and purity sufficient for their
`biochemical characterization, with overall yields of 0.7 and 0.2
`mg, respectively (per 1 L of conditioned medium) (Figure 1b).
`Similarly, we obtained 0.1 and 0.2 mg of Avi-hDPP-IV and Avi-
`mDPP-IV, respectively. The kinetic properties of the SF-tagged
`and Avi-tagged proteins were virtually identical (data not
`shown).
`Design, Synthesis, and Characterization of Anti-FAP
`iBody 1. To select the most suitable targeting ligand for FAP
`in terms of potency and selectivity, we prepared a small panel of
`FAP inhibitors (compounds 1−4; Scheme 1 and Figure 2a) and
`assessed their structure−activity relationships. Most
`impor-
`tantly, we investigated the appropriate linkage of the targeting
`ligand to the polymer backbone. The compounds are based on
`the previously published structure of a FAP inhibitor with a (4-
`quinolinoyl)glycyl-2-cyanopyrrolidine scaffold44,45 and contain
`a PEG linker for attachment to the HPMA copolymer. We then
`determined IC50 values of compounds 1−4 for SF-hFAP using
`a FAP activity inhibition assay. Compounds 1−3 exhibited
`comparable inhibition constants (0.23, 0.28, and 0.37 nM),
`whereas compound 4 was substantially less potent (4.8 nM)
`(Figure 2a). For further experiments, we chose compound 1 to
`
`8386
`
`DOI: 10.1021/acs.jmedchem.7b00767
`J. Med. Chem. 2017, 60, 8385−8393
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`comparable with the IC50 value obtained from the FAP activity
`assay. Concordantly, we confirmed that inhibition of FAP
`enzymatic activity persists even when FAP-containing cell
`lysates preincubated with effective concentrations of anti-FAP
`iBody were diluted to decrease the iBody concentration prior to
`the enzymatic assay. In contrast, the inhibition by the low-
`molecular-weight (4-quinolinoyl)glycyl-2-cyanopyrrolidine-
`based FAP inhibitors was reversible in the same experimental
`setup (data not shown).
`We tested the anti-FAP conjugate in several biochemical
`applications. Using iBody 1 immobilized to streptavidin agarose
`via biotin, we pulled down FAP from a cell lysate of U251 cells
`stably transfected with human FAP (U251_FAP+; see the
`“Biochemical Methods” section in the Supporting Information
`for the discussion of the cell line denomination) (Figure 3b).
`As negative controls, we used iBody 2, which lacks the FAP
`inhibitor, and blank streptavidin agarose (NC-SA) to show
`potential nonspecific binding to HPMA copolymer backbone
`and/or the streptavidin agarose resin (Figure 3b). The presence
`of isolated FAP was verified by LC-MS/MS, which detected
`FAP protein in the iBody 1 elution sample. Using the same
`setup, we also successfully isolated FAP protein from human
`blood plasma (verified by LC-MS/MS; data not shown),
`further confirming the functionality of anti-FAP conjugates in
`complex biological samples.
`Additionally, we developed sandwich ELISA for FAP
`quantification employing iBody 1 as a substitute for the
`detection antibody (Figure 3c). Recombinant SF-hFAP was
`first captured by the FAP-specific monoclonal antibody F-19
`and then detected with iBody 1, followed by incubation with
`neutravidin−HRP conjugate. The detection limit of this newly
`developed ELISA was as low as 0.4 ng/mL of FAP.
`iBody 1 (followed by IRDye 800CW streptavidin conjugate)
`could also be used to visualize FAP on a “semi-native” Western
`blot (Figure 3d). Both recombinant SF-tagged FAP (SF-hFAP)
`and endogenous full-length FAP migrated at around 130 kDa,
`corresponding to FAP dimers; the detection limit was about 50
`ng of FAP (Figure 3d).
`Application of Anti-FAP iBody 1 for Imaging of FAP-
`Expressing Cells. We tested the suitability of iBody 1 as a tool
`for the specific imaging of FAP-positive cells using confocal
`microscopy (Figure 4a,d) and flow cytometry (Figure 4b,c).
`Live cells expressing (U251_FAP+) or not expressing
`(U251_FAP−) FAP were incubated with anti-FAP iBody 1;
`iBody 2, which lacks the FAP inhibitor, was used as a negative
`control. Confocal microscopy imaging showed that iBody 1
`binds only to FAP-expressing cells and not to cells lacking FAP,
`whereas control
`iBody 2 did not bind to any of the cells
`analyzed (Figure 4a). Upon binding to FAP on the cell surface,
`iBody 1 underwent slow internalization, as evidenced by the
`accumulation of
`the signal
`inside cells after prolonged
`incubation. Similar results were obtained with cells transfected
`with mouse FAP (data not shown).
`The binding of anti-FAP iBody 1 to mouse FAP was also
`confirmed by flow cytometry. Anti-FAP iBody 1 strongly
`stained mouse GL261 glioma cells transfected with mouse FAP,
`whereas neither the FAP-negative parental cell line GL261 nor
`Gl261 cells transfected with mouse DPP-IV were stained by the
`conjugate (Figure 4b). We further analyzed the utility of the
`compounds in detecting endogenous levels of FAP expression.
`Using flow cytometry, FAP expression was visualized in
`cultured human fibroblasts and human glioblastoma U87
`cells, which are known to express the protein,50 by anti-FAP
`
`8387
`
`DOI: 10.1021/acs.jmedchem.7b00767
`J. Med. Chem. 2017, 60, 8385−8393
`
`Figure 1. Design and purification of recombinant FAP and DPP-IV.
`(a) Schematic structures of the recombinant human and mouse FAP
`and DPP-IV proteins expressed in Drosophila S2 cells. The
`extracellular parts of human and mouse FAP containing Strep-tag II
`and Flag-tag were purified via Strep-Tactin affinity chromatography.
`The extracellular parts of human and mouse DPP-IV containing Avi-
`tag were purified via streptavidin mutein affinity purification. (b) A
`silver-stained SDS-PAGE gel showing a typical two-round affinity
`purification of recombinant SF-hFAP protein expressed in Drosophila
`S2 cells. Load, concentrated medium; FT, flow-through; W1−W2,
`wash fractions; E1−E3, elution fractions. Ten microliters were loaded
`onto the gel, except for Load, FT-1, and FT-2 (0.5 μL).
`
`avoid steric problems after conjugation (long linker) and ensure
`the best inhibitory properties (difluoro substitution at C4 of the
`proline derivative).
`Compound 1, together with an ATTO488 fluorophore and
`the affinity anchor biotin, were conjugated to the HPMA
`copolymer carrier, yielding an HPMA copolymer conjugate
`targeting FAP (iBody 1: Mn = 110600 g/mol, Mw = 149900 g/
`mol, Đ = 1.36; Figure 2b). As a negative control, a
`corresponding conjugate lacking the FAP inhibitor was
`prepared (iBody 2: Mn = 80900 g/mol, Mw = 131000 g/mol,
`Đ = 1.62). Attachment of compound 1 to the copolymer chain
`led to an increase in the IC50 value [IC50(iBody 1) = 1.1 nM]
`(Figure 2c). Importantly, using a DPP-IV activity assay, we
`determined that iBody 1 is highly selective for FAP with IC50
`for the FAP homologue DPP-IV more than 4 orders of
`magnitude higher (IC50 > 10 μM) (Figure 2c). We also
`determined the IC50 values for iBody 1 toward mouse FAP and
`mouse DPP-IV and observed similar selectivity for FAP (IC50 =
`3.0 nM and IC50 > 10 μM, respectively). In addition, iBody 1
`did not inhibit recombinant prolyl oligopeptidase (PREP) and
`dipeptidyl peptidase 9 (DPP9) (Figure 2c).
`Use of Anti-FAP iBody 1 for Detection and Visual-
`ization of FAP. We used surface plasmon resonance (SPR) to
`evaluate the interaction between iBody 1 and FAP (Figure 3a).
`iBody 1 was immobilized to a neutravidin layer via biotin, and
`four concentrations of recombinant SF-hFAP were loaded. The
`SPR analysis indicated a relatively high association rate (kon =
`3860 M−1 s−1) and remarkably low dissociation rate (koff < 2 ×
`10−5 s−1), which was under the detection limit of our SPR
`instrument. The resulting dissociation constant (KD < 6 nM) is
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`Scheme 1. Synthesis of Compounds 1−4, Specific Inhibitors of FAP Modified with PEG Linkersa
`
`aReagents and conditions: (a*) described in ref 47; (a) SOCl2, MeOH, reflux; (b) NaH, t-butyl 2-bromoacetate, DMF, −80 °C to RT; (c) TFA; (d)
`TBTU, DIEA, NH2-PEGn-NH-BOC (n = 5, or n = 15), DMF; (e) 5 M NaOH/H2O, THF/MeOH; (f) (1) TSTU, DIEA, DMF, (2) (S)-2-(2-cyano-
`4,4-difluoropyrrolidin-1-yl)-2-oxoethanaminium chloride or (S)-2-(2-cyanopyrrolidin-1-yl)-2-oxoethanaminium chloride; (g) Ts-OH, ACN.
`
`iBody 1 followed by an amplification step with a streptavidin−
`phycoerythrin conjugate (Figure 4c). Finally, anti-FAP iBody 1
`was used to visualize FAP-transfected tumor cells in frozen
`tissue sections of glioma tumor xenografts (Figure 4d).
`Collectively, these data suggest the applicability of anti-FAP
`iBodies in a broad spectrum of methodologies traditionally
`utilizing antibodies.
`
`■ DISCUSSION
`Multiple proteases are involved in oncogenesis. FAP, along with
`other proteases, is proposed to participate in the processes of
`cell adhesion, invasion, migration, and tumor neovasculariza-
`tion.51 However, in contrast to most other cancer-associated
`proteases, FAP is expressed very sparsely in healthy adult
`tissues. In cancerous tissues, FAP is characteristically present in
`stromal cells as well as in the transformed elements of several
`malignancies.51 This makes FAP a promising potential target to
`exploit for cancer therapeutics and/or diagnostics. Recently, we
`described the iBody concept for specific targeting of enzymes.37
`In this work, we aimed to prepare anti-FAP iBodies and
`demonstrate their potential to specifically bind FAP and FAP-
`expressing cells.
`To identify the most potent and specific FAP-targeting
`ligand, we synthesized and characterized four FAP inhibitors
`based on the (4-quinolinoyl)glycyl-2-cyanopyrrolidine scaffold,
`which has high potency and selectivity for FAP.44 The
`
`inhibitors were synthesized with PEG linkers of two different
`lengths, as we had previously observed that short
`linkers
`impaired binding of the protein target to the inhibitor molecule
`“immobilized” on the polymer backbone. We prepared
`inhibitors with or without a 4,4-difluoro substitution of the 2-
`cyanopyrrolidine moiety, as Jansen et al. showed that this
`substitution leads to more potent FAP binding and improved
`selectivity with respect to the close FAP homologue prolyl
`oligopeptidase.45 Conjugation of compound 1 to the HPMA
`copolymer resulted in a 5-fold increase in the IC50 value. This
`was somewhat surprising, as we observed a significant drop in
`IC50 value for the iBody targeting GCPII.37 Nevertheless, anti-
`FAP iBody 1 is a low nanomolar binder of FAP, which is still
`more than sufficient for effective in vitro and in vivo targeting.
`Highly specific discrimination between FAP and its close
`homologue DPP-IV, an almost ubiquitously expressed multi-
`functional protease,52 is essential to prevent off-target effects of
`anti-FAP iBodies in vivo. We showed that iBody 1 is highly
`selective for FAP, exhibiting a more than four-order-of-
`magnitude lower IC50 for FAP than for DPP-IV. Therefore,
`even hundred nanomolar concentrations should lead to specific
`FAP targeting. We also verified that IC50 values of iBody 1
`toward other FAP homologues, prolyl oligopeptidase (PREP)
`and dipeptidyl peptidase 9 (DPP9), are more than three-orders-
`of-magnitude higher than for FAP itself, meaning that iBody 1
`
`8388
`
`DOI: 10.1021/acs.jmedchem.7b00767
`J. Med. Chem. 2017, 60, 8385−8393
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`Figure 2. Structures of FAP inhibitors and anti-FAP polymer conjugate iBody 1 and their kinetic characterization. (a) The effect of prolyl moiety
`modification of the (4-quinolinoyl)glycyl-2-cyanopyrrolidine scaffold and PEG linker length on the kinetic properties of the tested inhibitors. The
`IC50 values for modified inhibitors with or without fluorine substitution and PEG linkers of various lengths are presented as mean ± standard
`deviation. Measurements were performed in duplicate. (b) Schematic structure of anti-FAP polymer conjugate (iBody 1) containing an ATTO488
`fluorophore, an affinity anchor (biotin), and compound 1, the FAP-specific inhibitor. (c) Comparison of IC50 values of the FAP inhibitor and the
`anti-FAP iBody toward recombinant FAP, DPP-IV, DPP9, and PREP (IC50 values are presented as mean ± standard deviation).
`
`Figure 3. Application of anti-FAP iBody 1 in biochemical methods. (a) SPR analysis of SF-hFAP binding to immobilized iBody 1 (KD < 6 nM). (b)
`Affinity isolation of FAP from FAP-transfected U251 (U251_FAP+) cell lysate using iBody 1. iBody 2 (lacking FAP inhibitor) and blank streptavidin
`agarose (NC-SA) were used as negative controls. Load = U251_FAP+ cell lysate. (c) Sandwich ELISA for quantification of FAP using iBody 1 and
`anti-FAP F-19 antibody as a detection “agent” and a capture antibody, respectively. Each sample was measured in triplicate; values are presented as
`the mean ± standard deviation. (d) Western blot visualization of FAP using iBody 1 followed by an IRDye 800CW streptavidin conjugate. Purified
`recombinant human FAP (SF-hFAP) and a lysate of U251_FAP+ cells were used. The right section refers to the membrane probed with IRDye
`800CW streptavidin conjugate only.
`
`these predominantly
`keeps the selectivity for FAP over
`intracellularly localized homologues as well.
`In contrast to antibodies, which target surface epitopes, the
`binding of iBodies relies on a specific interaction between the
`
`inhibitor molecule and the active site of the enzyme, which is
`usually the most conserved part of a protein molecule.
`Therefore, we expected anti-FAP iBody to bind FAP
`orthologues with similar affinity. Indeed, we found that iBody
`
`8389
`
`DOI: 10.1021/acs.jmedchem.7b00767
`J. Med. Chem. 2017, 60, 8385−8393
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`Figure 4. Detection and visualization of FAP on cells using anti-FAP iBody 1. (a) Confocal microscopy analysis of cells expressing and not
`expressing FAP. Cells with inducible FAP expression were incubated with 200 nM iBody 1 (at 37 °C for 1 h), which was then visualized by confocal
`microscopy; the cell nuclei were stained with Hoechst H34580 dye. (b) Flow cytometry detection of FAP expression in GL261 cells transfected with
`mouse FAP but not in cells transfected with the closely related mouse DPP-IV. Nontransfected GL261 cells stained with iBody 1 and streptavidin−
`PE conjugate, and FAP-GL261 transfectants stained with streptavidin−PE conjugate only were used as negative controls. (c) Flow cytometry
`detection of endogenous FAP expression in human fibroblasts and U87 glioma cells using anti-FAP iBody 1. GL261 cells incubated with iBody 1 and
`streptavidin−PE conjugate and human fibroblasts incubated with streptavidin−PE conjugate only were used as negative controls. (d)
`Immunohistochemistry on frozen sections of a tumor (delineated by a dashed line) generated by xenotransplantation of FAP-transfected glioma cells
`into mouse brain.
`
`1 interacts with mouse FAP with an inhibition constant
`comparable to that of human FAP. This extends
`the
`applicability of anti-FAP iBodies, suggesting that they can be
`used as a versatile instrument for animal model experiments in
`preclinical translational studies.
`We used various biochemical methods to assess the potential
`iBody 1 for FAP targeting in vitro. Using SPR, we
`of
`characterized iBody 1 binding to FAP and determined its KD.
`Importantly, these results suggested the formation of a stable
`complex between FAP and its targeting iBody despite the fact
`that inhibition by the low-molecular-weight (4-quinolinoyl)-
`glycyl-2-cyanopyrrolidine-based FAP inhibitors is reversible.45
`When bound to streptavidin agarose resin, iBody 1 specifically
`pulled down FAP from complex protein matrices, including cell
`lysate and human blood plasma. As seen in Figure 3b, the
`negative control experiment with blank streptavidin agarose
`showed that the nonspecifically isolated proteins are adsorbed
`to the agarose resin and not to the HPMA conjugate backbone.
`
`For quantitative assays of FAP, we developed a sandwich
`ELISA using iBody 1 in place of the detection antibody (Figure
`3c). Even without optimizing the assay, the limit of detection of
`the antibody−iBody sandwich ELISA was comparable to that of
`the commercially available ELISA (Human FAP DuoSet
`ELISA, no. DY3715; R&D Systems, Inc.). FAP could also be
`detected on Western blot, which is somewhat surprising as
`iBody binding requires an intact active site, while SDS-PAGE
`and blotting generally lead to protein structure destabilization.
`However, FAP apparently preserved its native structure during
`these processes and was recognized by iBody 1. The observed
`FAP band migrating at about 130 kDa corresponds to the
`dimeric form of FAP, suggesting that the FAP dimer was not
`dissociated in the used experimental setup.
`Next, we tested iBody 1 in cell culture experiments. Live cell
`imaging using confocal microscopy confirmed the high
`selectivity of anti-FAP iBody for FAP and revealed negligible
`nonspecific binding to cells. Up to micromolar concentrations
`
`8390
`
`DOI: 10.1021/acs.jmedchem.7b00767
`J. Med. Chem. 2017, 60, 8385−8393
`
`

`

`Journal of Medicinal Chemistry
`
`Article
`
`of the anti-FAP polymer conjugate did not stain cells not
`expressing FAP. Our data showed that the FAP-iBody complex
`is internalized, which is beneficial
`for future drug delivery
`concepts exploiting FAP iBodies in cancer therapy. In addition
`to cell culture experiments with human FAP, we also showed
`that iBody 1 specifically stained GL261 cells transfected with
`mouse FAP, confirming its possible use in mouse models.
`Finally, we used iBody 1 to detect and visualize FAP on cells
`within complex tissue by performing immunohistochemistry
`experiments with frozen tissue sections.
`
`■ CONCLUSION
`We designed, synthesized, and characterized a novel type of a
`highly selective FAP targeting agent, an iBody based on an
`HPMA copolymer decorated with a FAP inhibitor. The
`specificity, modularity, and versatility of the anti-FAP iBody
`make it suitable for a broad spectrum of biochemical and
`biomedical applications. Thus, anti-FAP iBodies may represent
`an attractive theranostic tool for future in vivo imaging and
`selective drug delivery into the tumor microenvironment.
`
`■ EXPERIMENTAL SECTION
`Chemistry. All chemicals were purchased from Sigma-Aldrich,
`unless stated otherwise. All inhibitors tested in the biological assays
`were purified using preparative scale HPLC Jasco PU-975 (flow rate
`10 mL/min, water phase containing 0.1% of TFA; gradient shown for
`each compound, including Rt) equipped with UV detector UV-975
`and with column Waters YMC-PACK ODS-AM C18 prep column, 5
`μm, 20 mm × 250 mm. The purity of compounds was tested on
`analytical Jasco PU-1580 HPLC (flow rate 1 mL/min,
`invariable
`gradient 2−100% ACN in 30 min, Rt shown for each compound, water
`phase contained 0.1% of TFA) with column Watrex C18 analytical
`column, 5 μm, 250 mm × 5 mm. The final inhibitors were all at least
`of 99% purity. Structure was further confirmed by HRMS at LTQ
`Orbitrap XL (Thermo Fisher Scientific) and by NMR (Bruker Avance
`I 400 MHz).
`(S)-1-((4-((2-(2-Cyano-4,4-difluoropyrrolidin-1-yl)-2-oxoethyl)-
`c a r b a m o y l ) q u i n o l i n - 7 - y l ) o x y ) - 2 - o x o -
`6,9,12,15,18,21,24,27,30,33,36,39,42,45,48-pentadecaoxa-3-aza-
`pentacontan-50-aminium 2,2,2-Trifluoroacetate (1). The crude
`evaporated reaction mixture of compound 10a was dissolved in 1.5
`mL of ACN and 400 mg of p-toluenesulfonic acid were added in one
`portion. The reaction was left stirring overnight, after which it was
`evaporated and the product was isolated using preparative scale HPLC
`(grad, 10−40% ACN in 50 min; Rt = 36 min) 133 mg isolated upon
`dry freezing (isolated yield over two steps = 57%). Analytical HPLC:
`Rt = 16.5 min. HRMS: (ESI+) m/z for C51H83O19N6F2 [M + H]+
`calcd 1121.56756, found 1121.56767.
`(S)-1-((4-((2-(2-Cyano-4,4-difluoropyrrolidin-1-yl)-2-oxoethyl)-
`carbamoyl)quinolin-7-yl)oxy)-2-oxo-6,9,12,15,18-pentaoxa-3-azai-
`cosan-20-aminium 2,2,2-trifluoroacetate (2). The crude evaporated
`intermediate from previous reaction (compound 10b) was dissolved in
`1 mL of ACN and 100 mg of p-toluenesulfonic acid was added in one
`portion. The reaction mixture was left stirring overnight, after which it
`was evaporated and the product purified (gradient, 10−40 ACN in 50
`min; Rt = 29 min) 6 mg of oily yellowish substance obtained upon dry
`freezing (isolated yield = 30%). Analytical HPLC: Rt = 14.0 min.
`HRMS: (ESI+) m/z for C31H43O9N6F2 [M + H]+ calcd 681.30449,
`found 681.30541.
`(S)-1-((4-((2-(2-Cyanopyrrolidin-1-yl)-2-oxoethyl)carbamoyl)-
`quinolin-7-yl)oxy)-2-oxo-6,9,12,15,18,21,24,27,30,33,36,39,
`42,45,48-pentadecaoxa-3-azapentacontan-50-aminium 2,2,2-tri-
`fluoroacetate (3). Crude evaporated reaction mixture from the
`previous reaction (compound 10c; product theoretically 79 μmol, 1.0
`equiv) was dissolved in 2 mL of ACN and 150 mg of Tos-OH·H2O
`(790 μmol, 10.0 equiv) were added in one portion and the
`deprotection was left to proceed overnight. The reaction mixture
`was then evaporated, and the crude product was purified on a
`
`preparative scale HPLC (grad, 10−40% ACN in 50 min; Rt = 36 min).
`(Note: deprotection with TFA leads to a major side reaction where
`isobutylene is added on the nitrile group.) Analytical HPLC: Rt = 16.1
`min. HRMS: (ESI+) m/z for C51H85O19N6 [M + H]+ calcd
`1085.58640, found 1085.58646.
`(S)-1-((4-((2-(2-Cyanopyrrolidin-1-yl)-2-oxoethyl)carbamoyl)-
`quinolin-7-yl)oxy)-2-oxo-6,9,12,15,18-pentaoxa-3-azaicosan-20-
`aminium 2,2,2-trifluoroacetate (4). First, 30 mg of compound 10d
`(40 μmol, 1.0 equiv) were dissolved in 750 μL of ACN and 23 mg of
`Tos-OH·H2O were added in one portion. The reaction mixture was
`monitored by analytical HPLC. After 16 h, the reaction mixture was
`evaporated and the product was purified using preparative scale HPLC
`(gradient, 20−60% ACN in 50 min; Rt = 30 min) 17 mg of oily
`yellowish substance obtained upon dry freezing (isolated yield = 56%).
`Note: the reaction does undergo the deprotection using TFA as well,
`however, it is much less clean, probably due to addition of t-butyl
`cation on nitrile group. Analytical HPLC: Rt = 14.2 min. HRMS: (ESI
`+) m/z for C31H45O9N6 [M + H]+ calcd 645.32425, found 645.32420.
`iBody 1. Copolymer precursor poly(HPMA-co-Ma-β-Ala-TT) (15
`mg; Mn = 61700 g/mol, Mw = 66600 g/mol, Đ = 1.08; 11.7 mol % TT;
`the preparation is described in the Supporting Information),
`compound 1 (3.78 mg dissolved in 36 μL of DMSO), ATTO488-
`NH2 (0.75 mg), and N-(2-aminoethyl)biotinamid hydrobromide
`(biotin-NH2) (2 mg) were dissolved in 0.1 mL of N,N-
`dimethylacetamide (DMA). Then 4.9 μL of N,N-diisopropylethyl-
`amine (DIPEA) was added. Reaction was carried out for 4 h at room
`temperature, and then 2 μL of 1-aminopropan-2-ol was added and the
`reaction was stirred for 10 min. Copolymer conjugate poly(HPMA-co-
`Ma-β-Ala-compound1-co-Ma-β-Ala-ATTO488-co-Ma-β-Ala-NH-bio