Journal of
`
`Article
`- - - - - - - - - - - - - - - ' -
`pubs.acs.org/jmc
`
`Medicinal
`Chemistry
`Exploring Structural Parameters for Pretargeting Radioligand
`Optimization
`†
`†
`†
`Jan-Philip Meyer,
`Kristen M. Cunanan,
`James Jackson,
`Paul Kozlowski,
`Brian M. Zeglis,*,†,§,∥,#
`and Jason S. Lewis*,†,⊥,#
`Thomas R. Dilling,
`†
`Department of Radiology, Memorial Sloan Kettering Cancer Center, New York, New York 10065, United States
`‡
`Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, New York 10065, United States
`§Department of Chemistry, Hunter College of the City University of New York, New York, New York 10065, United States
`∥
`Ph.D. Program in Chemistry, Graduate Center of the City University of New York, New York, New York 10016, United States
`⊥
`Program in Molecular Pharmacology, Memorial Sloan Kettering Cancer Center, New York, New York 10065, United States
`#Departments of Radiology and Pharmacology, Weill Cornell Medical College, New York, New York 10065, United States
`*S Supporting Information
`
`†
`
`‡
`
`Pierre Adumeau,
`
`§
`
`Exploring Structural Parameters for Pretargeting Radioligand Optimization
`
`ABSTRACT: Pretargeting offers a way to enhance target spec-
`ificity while reducing off-target radiation dose to healthy tissue
`during payload delivery. We recently reported the development
`of an 18F-based pretargeting strategy predicated on the inverse
`electron demand Diels−Alder reaction as well as the use of this
`approach to visualize pancreatic tumor tissue in vivo as early as
`1 h postinjection. Herein, we report a comprehensive structure:
`pharmacokinetic relationship study of a library of 25 novel
`radioligands that aims to identify radiotracers with optimal
`pharmacokinetic and dosimetric properties. This investigation
`revealed key relationships between molecular structure and
`in vivo behavior and produced two lead candidates exhibiting
`rapid tumor targeting with high target-to-background activity
`concentration ratios at early time points. We believe this
`knowledge to be of high value for the design and clinical translation of next-generation pretargeting agents for the diagnosis and
`treatment of disease.
`
`Structure-Activity(cid:173)
`Relationship
`
`■ INTRODUCTION
`In vivo pretargeting for diagnostic1−6 and therapeutic7 appli-
`cations has emerged over the last three decades as a powerful
`technology to enhance target specificity and reduce off-target
`effects.2,8 Generally speaking, pretargeting strategies strive to
`combine the inherent advantages of macromolecular targeting
`vectors and small molecules, specifically high target specificity
`and short organ and tissue residence times, respectively.1,4
`To achieve this, the targeting vector is administered first and
`allowed to accumulate at the target site and clear from off-target
`organs prior to the injection of a small effector molecule carrying
`the payload of interest (e.g., radionuclide; Figure 1).4,5,9 To
`enable their in vivo recombination, both entities are equipped
`with complementary functionalities that enable an in vivo liga-
`tion reaction.10,11 Appropriate pairs of reaction partners that
`have been employed in in vivo pretargeting approaches include
`streptavidin−biotin,12 complementary oligonucleotide strands,13
`and click chemistry-based reaction pairs.10,14,15 While strategies
`based on streptavidin−biotin have shown somewhat deflating
`outcomes in the clinic,16 the use of a bispecific, CEA-targeting
`antibody in combination with a radiolabeled hapten peptide has
`shown very promising clinical results.17
`
`One of the newer members of the click chemistry toolbox,
`the inverse electron-demand Diels−Alder (IEDDA) reaction
`between trans-cyclooctene (TCO) and tetrazine (Tz) has proven
`particularly well suited for in vivo pretargeting.2−5,7,10,11 The
`IEDDA ligation is selective and bioorthogonal, but its principal
`advantage for pretargeting compared to other click reactions,
`such as the Staudinger ligation18 or the strain-promoted alkyne−
`azide cycloaddition14 (SPAAC), lies in its speed. To wit, the
`first-order rate constants of the Tz/TCO ligation lie in the realm
`of 104−105 M−1 s−1, orders of magnitude faster than the rates
`of the Staudinger and SPAAC reactions (0.002 M−1 s−1 and
`0.07 M−1 s−1, respectively).11 The in vivo feasibility of the
`IEDDA reaction between a TCO-modified monoclonal antibody
`(mAb) and a radiolabeled Tz has been demonstrated by various
`groups using a wide range of antigen-targeting immunoconju-
`gates and tetrazines labeled with an array of radionuclides for
`imaging (including 111In, 64Cu, 99mTc, 18F, 68Ga, and 11C)1,4,5,19−21
`and therapy (177Lu).7
`
`Received: August 1, 2017
`Published: August 31, 2017
`
`'V ACS Publications
`
`© 2017 American Chemical Society
`
`8201
`
`DOI: 10.1021/acs.jmedchem.7b01108
`J. Med. Chem. 2017, 60, 8201−8217
`
`Petitioner GE Healthcare – Ex. 1040, p. 8201
`
`

`

`Journal of Medicinal Chemistry
`
`a
`
`b
`
`Tetrazine
`
`O~OH 2NH 2
`
`I
`
`c:;,-
`:::...
`
`I
`
`:::...
`
`Slow
`clearanca
`
`Rapid
`clearance
`
`~
`
`,c
`
`So
`
`1,
`
`~ 1/ ~
`N~ N
`
`N " N
`
`NI N
`
`2
`
`2NH 2
`
`N
`
`N " N
`
`No" ✓. N
`
`" N
`,,:;
`
`I
`
`2NH2
`
`"'
`
`.0
`
`~ " ~
`N---,:> N
`
`-
`
`Article
`
`Linker
`
`Chelator
`
`Polyethylene glycol
`
`HzN y
`
`Boe
`
`O~NH
`n
`
`n = 7 (5)
`n = 11 (6)
`
`Amino acids
`
`0
`
`HO~R
`NH 2
`
`I .o
`R° '
`
`OH ( N)
`0;_:N
`~N
`
`Ho-4.O
`
`R = NCS (11 ), NH 2 (12)
`
`R=~NH 2
`
`(7)
`
`R2~
`
`:::...
`
`R = ~
`
`NH
`N'=:/
`
`(8)
`
`R = ~
`
`OH
`
`(9)
`
`0
`
`NH
`
`1 CO zH
`H0 2 ' -N NJ
`
`(._,, N
`)
`HO2C
`
`R2 = NCS (13), NH 2(14)
`
`Inject
`antibody
`
`•
`•••
`
`Inject
`radiollgand
`
`In vivo
`click react1011
`
`·1
`
`~·
`
`2
`
`••
`
`3
`
`4
`
`R = c'::;~ N)lNH
`H
`2
`
`(10)
`
`Figure 1. (a) Schematic illustration of the pretargeting approach: a macromolecular targeting vector (in our case an antibody−TCO conjugate) is
`injected first and allowed to reach the target site while clearing slowly from systemic circulation. After a specific accumulation time, the small molecule
`effector probe (in this case a radiolabeled tetrazine probe) is administered systemically and undergoes bioorthogonal click reaction with the TCO groups
`of the immunoconjugate at the target site. (b) Modular chemistry approach for radioligand design: radioligands consisted of a Tz moiety for in vivo click
`chemistry, a linker for altering the biodistribution, and a chelator for the attachment of the positron-emitting metal ions 68Ga3+ and [18F]-AlF2+.
`
`In terms of pretargeted positron emission tomography (PET)
`imaging, Zeglis et al. presented promising results in 2013 in
`pretargeting experiments using the gpA33-targeting mAb
`huA33-TCO and a 64Cu-labeled tetrazine radioligand.4 Shortly
`thereafter, our laboratories developed a second-generation Tz for
`64Cu-based pretargeted PET imaging applications by integrating
`the sarcophagine chelator system into the radioligand structure.3
`At
`the same time, our laboratories demonstrated that an
`18F-labeled Tz-based radioligand in combination with the
`carbohydrate antigen 19.9 (CA19.9)-targeting fully human
`mAb 5B1-TCO22 allowed for the successful PET imaging of
`subcutaneous (sc) pancreatic cancer xenografts as early as 1 h
`postinjection (p.i.).1 Critically, this new pretargeting approach
`utilized the short-lived radionuclide 18F (t1/2 = 109 min),
`resulting in only a fraction of off-target radiation doses to healthy
`tissues compared to directly labeled immunoconjugates with
`long-lived isotopes (124I or 89Zr, t1/2 > 3 days). Despite clear
`delineation of tumor tissue at early imaging time points demon-
`strating the general feasibility of this approach, the relatively low
`tumor-to-background activity concentration ratios at even 4 h
`p.i., such as tumor-to-intestines (1.6 ± 0.1) and tumor-to-kidney
`(1.8 ± 0.4) ratios, inspired us to undertake a thorough investi-
`gation into the fundamental relationships between molec-
`ular structure, pharmacokinetics, and pretargeting performance.
`This first-of-its-kind structure−pharmacokinetics relationship
`(SPR) study was further fueled by the increased popularity of
`IEDDA pretargeting strategy. However, current approaches lack
`fundamental insight into the relationship between physicochemical
`
`properties and pretargeting performance. To address those
`issues, the study at hand was designed with two main objectives:
`(1) to identify a radiopharmaceutical lead candidate suitable for
`clinical development and (2) to generate experimental evidence
`for a rational understanding of how molecular parameters such
`as overall molecular net charge, distribution coefficient, plasma
`half-life (PHL), and stability influence the in vivo performance of
`small-molecule radioligands in pretargeting systems.
`Tracer Library and SPR Study Design. The synthesis of
`the radioligands library as the first step of this study was based
`on previously reported protocols.1,2,4 New reaction routes
`and radiolabeling procedures developed within this study are
`described in the Supporting Information (sections 2 and 3).
`Overall, the radioligands were designed to display structural
`variation in order to cover a broad spectrum of physicochemical
`properties,
`thereby enabling the study of
`the relationship
`between structure and in vivo behavior (Table 1). Each radio-
`ligand is composed of three different structural building blocks
`(Table 1; Figure 2a). First, a Tz component (1−4) (Scheme 1)
`was selected for in vivo click chemistry. Second, a linker moi-
`ety consisting of polyethylene glycol [PEG7 (5) or PEG11 (6)],
`amino acids (AA) [AA = lysine (K, 7), histidine (H, 8), aspartate
`(R, 9), and arginine (D, 10)], or a combination of both is
`attached. Finally, a bifunctional chelator, either 1,4,7-triazacy-
`clononane-1,4-diacetic acid (NODA, 11,12) or 1,4,7-triazacy-
`clononane-1,4,7-triacetic acid (NOTA, 13,14), was introduced,
`allowing for the installation of 18F and 68Ga radionuclides. Tz
`moieties 1−4 were selected based on their previously reported
`
`8202
`
`DOI: 10.1021/acs.jmedchem.7b01108
`J. Med. Chem. 2017, 60, 8201−8217
`
`Petitioner GE Healthcare – Ex. 1040, p. 8202
`
`

`

`Journal of Medicinal Chemistry
`Article
`- - - - - - - - - - - - - - - - - - - - - - - ' -
`Table 1. All 25 Radioligands Were Employed in the First Characterization Processa
`
`Tz
`
`Chelator Precursor Tracer
`
`Linker
`PEGx (x = 7,11) AA (D, H, K, R)
`-
`-
`K
`-
`7
`-
`-
`11
`K
`-
`
`r1 sFl15
`[1 sF]l6
`[1SF] 17
`[1SF] 18
`[1 sF] 19
`r6sGa l 19
`r1sF120
`r6sGal20
`r1sF121
`[6sGa]21
`r6sGa]22
`r1sF123
`r1sF124
`[6sGa]24
`[1sF]25
`[6sGa]26
`[1sF]27
`[6sGal27
`[1sF]28
`r6sGa]28
`r1sF129
`r1sF13o
`r6sGa 130
`[1sF]31
`4
`31
`NODA
`R
`11
`[1sF]32
`11
`4
`NOTA
`32
`D
`aOn the basis of those results, 15 radioligands were selected for the next step of testing to investigate their performance in pretargeting experiments
`(blue). Finally, radioligands [18F]27 and [68Ga]27 were identified as lead compounds based on their overall tumoral uptake and tumor-to-NT activity
`concentration ratios (red).
`
`1
`1
`1
`1
`
`2
`
`2
`
`2
`
`2
`2
`
`3
`
`4
`4
`
`4
`
`4
`
`4
`
`4
`
`NODA
`NOTA
`NOTA
`NOTA
`
`NOTA
`
`NOTA
`
`NOTA
`
`NODA
`NOTA
`
`NOTA
`
`NODA
`NOTA
`
`NODA
`
`NOTA
`
`NODA
`
`NODA
`
`15
`16
`17
`18
`
`19
`
`20
`
`21
`
`22
`23
`
`24
`
`25
`26
`
`27
`
`28
`
`29
`
`30
`
`-
`
`-
`-
`11
`
`11
`
`7
`7
`
`11
`
`11
`
`7
`
`11
`
`K-K
`
`K-K-K
`K-K-K
`-
`-
`
`-
`-
`
`-
`
`-
`K
`
`H
`
`stability and reaction kinetics with TCO to ensure a wide range
`of properties.4,5,21,23 The use of PEG linkers to modulate in vivo
`PK of small molecules has previously been reported, including
`accelerated nontarget organ clearance as well as increased renal
`clearance.1,2,24 We included them into our study in order to
`investigate their impact on radiotracer PK alone or in combi-
`nation with AA linkers (which, to the best of our knowledge,
`have not yet been systematically reported in any SPR study).
`The amino acids lysine, arginine, histidine, and aspartate were
`regarded as useful structural components to significantly influ-
`ence in vivo behavior and PK parameters. It was reasoned that
`their charged side chains should have a measurable effect on
`tracer PK and would further allow us to establish a correlation
`between molecular net charge and PK parameters. Both bifunc-
`tional chelator moieties NODA and NOTA currently find broad
`application in preclinical25,26 and clinical27 research for the
`radiolabeling of biological macromolecules and small molecule
`targeting probes. Precursors 15−32 were synthesized in good
`overall chemical yields (18−37%) via 3−8-step syntheses,
`depending on the starting materials (Figure 2b).
`All 18F- and 68Ga-labeled [t1/2 (68Ga) = 68 min] Tz-derived
`radioligands were furnished in high radiochemical yields [RCYs,
`
`>55% (18F); >83% (68Ga)], with specific activities (SAs) of
`>19 MBq/nmol and high radiochemical and radionuclidic purity
`(Supporting Information, section 3). For in vitro analysis,
`purified tracers were incubated in human serum at 37 °C to
`analyze their stability under physiological conditions. Further,
`the distribution coefficient (log D, n = 3) of all radioligands in a
`1:1 mixture of PBS:1-octanol was determined using the shake-
`flask method. Subsequently, the tracers (4−8 MBq, 0.5−1 nmol)
`were injected into healthy athymic nude mice via the lateral tail
`vein. At various time points (between 2−120 min p.i.), blood
`was drawn via the lateral tail vein or saphenous vein (n = 4) to
`calculate PHLs (all radioligands, n = 1) and plasma stabilities
`(n = 3). Tracers (11−14 MBq, 0.8−1.3 nmol) were injected into
`healthy athymic nude mice, and general biodistribution experi-
`ments for all tracers were performed using serial PET imaging
`hourly between 1 and 4 h p.i., unless stated otherwise (n = 4).
`Decay-corrected PET imaging data and reconstructed 3D images
`of the tracer distribution were then used to determine radio-
`activity concentrations (given as percent injected dose per gram,
`%ID/g) in the kidney and large intestine (quantitative ROI
`analysis). Additional ex vivo organ uptake values were deter-
`mined for selected compounds and were found to be in line with
`
`8203
`
`DOI: 10.1021/acs.jmedchem.7b01108
`J. Med. Chem. 2017, 60, 8201−8217
`
`Petitioner GE Healthcare – Ex. 1040, p. 8203
`
`

`

`Journal of Medicinal Chemistry
`a 100
`
`...
`
`f
`e ~ 6
`
`6
`
`t· t
`
`Yy 1
`....
`
`.........
`
`b
`
`-0.5
`
`en -1 .0
`
`CD
`Cl g; -1.5
`Cl 0
`_3 ~ -2.0
`u
`0
`::::, -2.5
`
`-3.0
`
`e
`
`-
`
`Article
`
`• Lysine (18F)
`♦ Lysine (68Ga)
`.ii. PEG (18F)
`Y PEG (68Ga)
`■ PEG + AA (18F)
`• PEG + AA (68Ga)
`* No linker
`
`♦
`
`4
`
`5
`
`C
`
`-0.5
`
`_-1.0
`Cf)
`CD
`C g; -1 .5
`Cl 0
`0 C:
`-' .!l! -2.0
`u
`9
`::::. -2.5
`
`-3.0
`
`t
`
`■
`
`•
`...
`i
`y
`•
`•
`• • •
`
`■
`
`♦
`
`♦
`
`-3
`
`-2
`
`3
`2
`0
`-1
`Molecular charge
`
`f
`
`20
`
`c I 15
`
`~
`;r
`iii 10
`.s:::.
`"' E
`"' a:
`
`1/1
`
`5
`
`0
`-3.5
`
`♦
`
`4
`
`5
`
`•••
`•
`
`•
`
`~·
`
`••
`
`-3.0
`
`-2.5
`
`-2.0
`LogD
`
`-1.5
`
`-1.0
`
`-0.5
`
`80
`
`.t: -
`st (.) 90
`@ ;:._
`Q) ~
`u E
`... ~
`"'
`::,
`-
`Q)
`<J) 70
`u C:
`-
`~ E
`. , :::,
`60
`-;f!.E,
`
`d
`
`20
`c
`§.15
`~
`~ 10
`.s::
`"' E
`1/1 5
`"'
`a:
`
`0
`
`if
`• ¼
`
`..
`
`♦
`
`y
`
`-!-
`•
`
`......
`
`*
`
`20
`
`c
`I 15
`~
`;r
`iii 10
`.s:::.
`"' E
`"' a:
`
`1/1
`
`5
`
`•
`
`0
`-3
`
`-2
`
`I
`y
`
`y
`
`...
`...
`t ...
`y
`* ■
`♦ •
`• •
`• • ' ♦
`
`3
`2
`0
`-1
`Molecular charge
`
`Figure 2. Correlation diagrams showing how structural and physicochemical parameters (e.g., stability, linker, molecular charge, log D, PHL) influence
`each other. (a) Tracer stabilities after a 3 h incubation time in human serum are shown after tracers were divided into groups with the same linker and
`radionuclide. Serum stabilities were generally >60%, except for 68Ga-labeled compound 46. (b) Log D values (median shown, n = 3) summarized as a
`group diagram. (c) Distribution coefficients in relation to molecular charge (under physiological conditions) showed a significant correlation. (d−f)
`in vivo PHLs plotted as a group diagram (d), in dependence of molecular net charge (e), as well as plotted against log D (f). Lysine-containing
`compounds showed overall fast clearance from circulation, with PEGylated compounds having three times longer PHLs. Interestingly, a positive
`correlation between a tracer’s PHL and its log D value was found.
`
`image-derived uptake data, justifying the use of image-derived
`organ uptake data for the majority of compounds in order to
`reduce the number of animals euthanized. Data analysis led to the
`selection of 15 radioligands (Table 1, highlighted in blue) that
`were further tested in in vivo pretargeting experiments using
`athymic nude mice bearing subcutaneous pancreatic ductal
`adenocarcinoma (PDAC) xenografts (n = 4). A TCO-modified
`immunoconjugate of
`the CA19.9-targeting antibody 5B1
`(5B1−TCO; 1.3 nmol, 200 μg per mouse) was injected
`72 h prior to the injection of the small molecule radioligands
`(1.3−1.6 nmol, 0.6−1.4 μg, 1−1.2 equiv). PET images were
`acquired hourly between 1−4 h p.i. unless stated otherwise.
`In addition, ex vivo biodistribution experiments were conducted
`for selected tracers in order to obtain quantitative information on
`tracer distribution in up to 14 organs. Ultimately, this investi-
`gation identified two radioligands ([18F]27 and [68Ga]27) as the
`most promising lead compounds (Table 1, highlighted in red).
`Physicochemical and Pharmacokinetic Properties of
`Tz-Derived PET Tracer. Structural, in vitro, and pharmacoki-
`netic (PK) data obtained for all 25 tracers were used to
`investigate potential correlations between physicochemical
`parameters. Tracer stabilities (given as % intact, n = 3) in
`human serum (incubation for 4 h at 37 °C, Figure 3c) ranged
`from >90% ([18F]23) to 54 ± 7% ([68Ga]24) (Figure 3a,
`Supporting Information, section 4). Considering all of the structural
`
`elements of the radioligands, we reasoned that both the tetrazine
`moiety and the metal complex would have an impact on in vitro
`(and presumably in vivo) stability. However, we found that
`instability was due primarily to the decomposition of
`the
`tetrazine moiety and that the radioactive metal complexes were
`stable over the course of our experiments. The majority of the
`radioligands did not show any elevated protein binding (<2% of
`total radioactivity, Supporting Information, section 4), with the
`exception of [18F]20, [18F]21, [68Ga]21, [68Ga]22, and [18F]29,
`each of which exhibited up to 16% protein-bound radioactivity.
`This result may be explained by the high lysine content residues
`and positive charge of these radioligands, prompting electrostatic
`interactions between the tracers and plasma proteins.24
`Not surprisingly, a significant negative correlation between the
`log D value of a tracer and its overall molecular charge under
`physiological conditions was found (p value <0.0001, Supporting
`Information, sections 4 and 9). Generally speaking, the closer the
`net charge of a tracer is to 0, the higher its partition coefficient.
`However, it is not quite that simple. For instance, radioligands
`possessing the same overall molecular charge but a dif ferent
`number of formally charged functional groups could still show
`significant differences in their distribution coefficients. One
`explanation could be enhanced solvation by water molecules
`through charge−dipole interactions: if the molecule exhibits a
`greater number of formal charges, it would lead to a higher
`
`8204
`
`DOI: 10.1021/acs.jmedchem.7b01108
`J. Med. Chem. 2017, 60, 8201−8217
`
`Petitioner GE Healthcare – Ex. 1040, p. 8204
`
`

`

`Journal of Medicinal Chemistry
`Article
`_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ J _
`
`Scheme 1. Reaction Scheme Exemplifying the Synthesis of Precursor Molecules Using Tz 1 and Tz 4 via Two Different Synthetic
`Routes, Both of Which Furnish Final Compounds in Good Overall Yields and High Chemical Puritiesa
`In-situ carboxylic acid activation
`and amide bond formation
`
`Non-activated carboxylic acids
`
`020H
`
`1""
`,,,;
`
`~ "~
`N~N
`
`HzNV'o~~~c
`
`(5,6)
`
`n=7,11
`(1-1 .3 eq.)
`AND/OR
`
`HzNXR
`
`(7-10)
`
`OH
`0
`(0.8-4 eq .)
`
`►
`
`1. DMF, EDC (2.0 eq.)
`HOB! (2 .0 eq.), RT, 4-8 h
`>75%.
`2. Boc-deprotection (if applicable)
`
`X'LI _L_i_nk_e_r _ _,f-R 2
`y R2 = NH 2, COOH
`
`N" N
`'
`" NvN
`
`Simple amide coupling
`
`Boe
`H2NV '0~NH
`(5,6)
`n=7, 11
`(1-1.3 eq.)
`
`AND/OR
`
`H2NXR
`
`(7-10)
`
`OH
`0
`(0.8-4 eq.)
`
`NHS-activated ester
`
`2gg:J
`
`~1/ ~
`NvN
`
`4
`
`1. DMSO, TEA (1.5 eq.), RT,
`30-45 min , >90%.
`2. Boc-deprotection (if applicable)
`
`21
`
`~''-I _L_in_k_e_r_-'f-R2
`R2 = NH 2, COOH
`
`N" N
`
`' " NvN
`
`Amine-isothiocyanide addition
`or
`Amide coupling
`
`NODNNOTA-NCS/NH 2
`(11-14)
`f - - - - - - - - - - - - (Precursors 15-32)
`if NH2 + NCSchelator:
`DMSO, RT, TEA(1.5 eq.),
`1 h, >80%
`
`if C02H + NH2, chelator:
`DMF, EDC (2.0 eq.)
`HOB! (2.0 eq.), RT, 4-8 h
`>68%.
`
`aFor Tz starting materials such as Tz 1, in situ activation of the carboxylic acid group under mild conditions was performed using 1-ethyl-3-(3-
`dimethylamino-propyl)carbodiimide (EDC) and 1-hydroxybenzotriazole (HOBt), enabling subsequent amide coupling with the terminal amine of
`an appropriate linker moiety. NHS-activated, commercially available starting materials such as Tz 4 could be coupled to the linker moiety under
`weakly basic conditions with high conversion rates and without the need for further activation reagents. After TFA-mediated deprotection of the
`terminal Boc-group (if applicable), terminal amines (or carboxylic acids) were reacted with a bifunctional NODA or NOTA chelator construct that
`possessed a complementary functional group for chemical ligation.
`
`degree of solvation.28 For example, radioligands with a net charge
`of 0 containing a PEG linker (or no linker) exhibited log D values
`between −1.0 and −1.5, while radioligands with the same overall
`charge but with lysine residues in the linker displayed log D
`values between −1.8 and −3.1. PHLs for the radioligands corre-
`lated (p value <0.0001, Supporting Information, sections 5 and 9)
`with their net charge. Tracers with lower (more negative) log D
`values and higher net charges possessed shorter PHLs (Figure 3d,e).
`The replacement of a PEG linker by AAs or the incorporation of
`AA(s) to an already existing PEG linker resulted in reduced
`PHLs. PEG-containing tracers [18F]21−[68Ga]28, even though
`ranging from −1 to +1 in terms of net charge, possessed calcu-
`lated PHLs of >10 min. For instance, lead compounds [18F]27
`and [68Ga]27 exhibited PHL of 17.1 and 15.1 min, respecti-
`vely. In contrast, AA-containing radioligands exhibited PHLs <
`10 min. Tracers that solely contained lysine moieties as linkers
`displayed the shortest PHLs, values that decreased further as the
`number of lysine residues increased. For example, [68Ga]22
`(with a charge of +4 and 3 lysine residues) possessed a PHL of
`1.9 min, the shortest of all tracers. Longitudinal PET imaging
`experiments conducted in healthy animals revealed a significant
`difference in biodistribution patterns and clearance pathways
`between tracers (Supporting Information, section 7). In general,
`lysine-containing radioligands (regardless of the presence of a
`PEG group) showed fast clearance from circulation and relatively
`high kidney radioactivity concentration of >5%ID/g as early as
`1 h p.i. Radioactivity uptake and retention increased in a nearly
`linear fashion with the number of lysine residues per radioligand
`
`(Figure 4a−c). High retention of the tracers containing lysine
`residues was likely due to reabsorption of those tracers in the
`proximal tubules of the kidney. In fact, recent studies have shown
`that peptides high in lysine residues are powerful kidney targeting
`agents, facilitating the uptake and retention of those constructs in
`the renal clearing organs.29,30 All other radioligands that did not
`contain lysine residues exhibited significantly lower kidney activity
`concentrations (<3.5%ID/g, Figure 4). Compound [68Ga]30, for
`instance, exhibited the lowest uptake values of 1.2 ± 0.3%ID/g at
`2 h p.i., although it also displayed higher liver (>1.8%ID/g) and
`intestinal uptake (>3%ID/g). Generally, tracers with overall net
`charges were cleared faster from circulation through globular
`filtration, whereas compounds with low or no net charge exhibited
`elongated circulation times and were predominantly cleared
`hepatically and excreted via the intestines.
`Evaluation of Pretargeting Performance in a PDAC
`Xenograft Model. On the basis of this in vitro and PK data, 15
`radioligands were selected for in vivo pretargeting experiments to
`probe correlations between PK parameters and pretargeting per-
`formance. Compounds were selected in order to cover a broad
`range of structural and physicochemical diversity. Generally, we
`reasoned that PHL and the primary clearance pathway of a tracer
`should have significant impact on its pretargeting performance.
`PHL would determine how much tracer molecules reach the
`target site before clearance, and the clearance pathway should
`influence the target-to-background ratio, as renal clearance
`should reduce background noise more quickly than hepatic
`clearance.
`
`8205
`
`DOI: 10.1021/acs.jmedchem.7b01108
`J. Med. Chem. 2017, 60, 8201−8217
`
`Petitioner GE Healthcare – Ex. 1040, p. 8205
`
`

`

`- - - - - - - - - - - - - - - - - - - - - - - - - - -
`Journal of Medicinal Chemistry
`
`Article
`
`10
`
`ci.
`<::
`N
`@) 6
`~
`Q
`~
`0
`E
`t=
`
`8
`
`4
`
`2
`
`"'"'
`
`o+----.----.----.---~
`10
`20
`15
`0
`5
`Plasma half-life [min]
`
`10
`
`8
`
`4
`
`ci.
`.s::.
`N
`@) 6
`~ 'cf.
`0
`E 2
`t=
`
`*
`
`o,.i....-.---.---.---.---.---..--
`
`Figure 3. (a) Chemical structures and maximum intensity projections (MIPs) acquired 2 and 4 h p.i. for five selected radioligands ([18F]19, [18F]20,
`[18F]21, [18F]27, and [18F]30) in healthy athymic nude mice (B = bladder, K = kidney, I = intestine). Tracers (11−14 MBq, 0.8−1.3 nmol) were
`administered via the lateral tail vein (t = 0). Notably, lysine-containing radioligands [18F]19, [18F]20, and [18F]21 showed fast renal clearance, whereas
`[18F]30 showed more predominantly uptake in the intestines. Lead compound [18F]27 showed clearance via both excretion routes. The increasing
`uptake and retention in the kidneys facilitated by an increasing number of lysine residues in the radioligand structure is well visible for [18F]19−21.
`(b) Kidney uptake of radioligands at 2 h p.i. is shown as a function of molecular net charge. (c) Kidney uptake values at 2 h p.i. dependent on the linker
`and radionuclide. Solely lysine containing radioligands exhibited elevated kidney uptake values of up to 16.0 ± 5.8 (2 h p.i.) and 21.1 ± 5.6%ID/g
`(4 h p.i.) for [68Ga]22. Even one lysine residue had a dramatic effect on kidney uptake as shown for [18F]29, where a lysine residue was integrated
`between the PEG7 linker and the chelator moiety, increasing kidney uptake from 2.6 ± 0.3 ([18F]25, Tz-PEG7-NODA) to 8.3 ± 1.5%ID/g ([18F]29,
`Tz-PEG7-Lysine-NODA). Substitution of lysine by other amino acids with either positively or negatively charged side chains reduced kidney uptake to
`below 4%ID/g.
`
`All initial pretargeting experiments were carried out in mice
`bearing sc BxPC3 xenografts. Approximately 3−4 weeks after
`inoculation (2.5 × 106 cells), TCO-modified anti-CA19.9 mAb
`5B1−TCO (1.33 nmol, 200 μg in 150 μL of 0.9% saline) was
`injected via the tail vein. Then 72 h later, Tz-derived radioligands
`(1.3−1.6 nmol, 1−1.2 equiv, in 150 μLof 0.9% saline, containing
`<5% v/v ethanol) were injected into the opposite lateral tail vein
`(t = 0).1,2
`PET imaging and ex vivo biodistribution experiments were
`conducted at 2 and 4 h p.i. Tumor uptake values of all the radio-
`ligands were determined via PET image analysis (ROI analysis)
`as well as via ex vivo biodistribution analysis for selected com-
`pounds (Supporting Information, sections 6 and 7). The median
`tumoral activity concentration values at 2 h p.i. ranged from 7.6 ±
`1.8%ID/g for 18F-labeled lead compound [18F]27, to 1.7 ± 0.8%
`ID/g for the fast-clearing tracer [18F]20 (Figures 5 and 6).
`Lysine-containing radioligands generally possessed tumor uptake
`values of <3%ID/g at all time points, with tumor uptake values
`of <2%ID/g for compounds solely containing a lysine linker. The
`highest tumor uptake values were observed for radioligands
`containing PEG linkers only. The introduction of AA linker led
`to accelerated plasma clearance and thus reduced the tumoral
`uptake due to fewer tracer molecules capable of reaching the
`
`tumor site. As expected, the PHL of a radiotracer correlated
`(p-value <0.0001, Supporting Information, section 9) with the
`tracer’s overall tumoral uptake, as illustrated for the 2 and 4 h
`time points (Figure 5c,d). Additionally, tracer stabilities in vivo
`were determined to be >50% at 3 h p.i. and did not correlate with
`tumoral uptake (Figure 5b).
`In case of tracer [68Ga]24, possessing low stabilities in both
`human serum (66.9 ± 7.9%) and in vivo (31 ± 6.3%), stability
`may indeed explain the relatively poor tumoral uptake of 3.1 ±
`0.7%ID/g at 2 h p.i., despite exhibiting a PHL (15.5 min) that
`would allow for higher accumulation at the target site over time.
`Identification of Lead Candidates. Taking all of the data
`into account, our study identified two lead candidates: [18F]27
`and [68Ga]27. Both tracers possessed good stabilities in vitro
`as well as in vivo and exhibited PHLs (17.1 and 15.1 min,
`respectively) that allowed for a fast enough clearance from circu-
`lation without impairing tumor accumulation. [18F]27 showed
`tumor uptake values of 7.6 ± 1.8%ID/g at 2 h p.i. and 8.8 ± 1.7%
`ID/g after 4 h, as well as promising tumor-to-NT (nontarget)
`activity concentration ratios, although it did display some uptake
`in the intestines (Figure 6a,b). [68Ga]27 exhibited tumoral
`uptake values of 6.8 ± 1.4%ID/g at 2 h p.i. and 7.1 ± 1.8%ID/g at
`4 h p.i. and boasted superior tumor-to-NT activity concentration
`
`8206
`
`DOI: 10.1021/acs.jmedchem.7b01108
`J. Med. Chem. 2017, 60, 8201−8217
`
`Petitioner GE Healthcare – Ex. 1040, p. 8206
`
`

`

`Article
`
`• Lysine ('"F)
`• Lysine (68Ga)
`• PEG ('"F)
`• PEG (68Ga)
`■ PEG+ AA ('"F)
`• No linker ('" F)
`
`•
`
`80
`
`20
`
`..
`•
`....
`
`....
`• • ..
`•
`..
`'
`
`000 -----r----:,:----
`5
`10
`15
`Plasma half rt
`• 0 e[min]
`
`C
`
`d
`
`ci
`,:
`
`"' @
`
`Journal of Medicinal Chemistry
`
`10
`
`..
`
`..
`•
`..
`
`..
`
`♦
`
`o,..---
`20
`
`60
`40
`0
`.
`1/omtact tracer @ 2 h
`• p.1 .
`
`10
`
`..
`
`..
`
`.. . ..
`....
`
`•
`..
`
`•
`•• • •
`
`0
`0
`
`10
`5
`15
`Plasma halt-1·1
`•
`oe[mm]
`
`20
`
`Figure 4. (a) MIPs and transverse image slices acquired in pretargeted PET imaging experiments 2 h after radioligand administration. White arrows
`indicate tumor tissue. To this end, 5B1−TCO (200 μg, 1.33 nmol, in 150 μL of 0.9% saline) was injected into the lateral tail vein 72 h prior to the
`radioligand (1.3−1.6 nmol in 150 μL of 0.9% saline, containing <5% v/v ethanol). [18F]19 exhibited slightly higher tumor uptake (2.2 ± 0.2%ID/g, one
`lysine residue) than [18F]20 (1.7 ± 0.2%ID/g, two lysine residues), where the tumor was not visible due to the high background of kidney. However,
`those uptake values were significantly lower compared to the PEGylated lead compounds [18F]27 (7.6 ± 1.8%ID/g) and [68Ga]27 (6.6 ± 1.4%ID/g).
`Incorporation of a lysine residue as for [18F]29 (PEG + Lysine) increased kidney uptake, while decreasing PHL and tumoral uptake, indicating a positive
`correlation. (b−d) Plots showing tumoral uptake in dependence of in vivo stability (2 h p.i.) and PHL (2, 4 h p.i.). In vivo stability (shown as % intact
`tracer and determined at 2 h p.i.) of all radioligands employed in pretargeting experiments was >50%, except for [18F]24 with a measured stability of
`31.6 ± 6.3%. In vivo stability had no measurable effect on the pretargeting performance, most likely because the majority of intact tracer had undergone
`click reaction at the target site within the first 2 h of the experiment. Tumor uptake of radioligands did however correlate with their PHLs: the longer the
`PHL the higher the overall tumoral uptake.
`
`ratios compared to its 18F-labeled counterpart, especially with
`regard to the large intestines (>3 at 2 h p.i. compared to ∼1.2 for
`[18F]27; Supporting Information, section 6).
`However, [68Ga]27 showed elevated activity concentrations in
`the blood pool in both tumor models with tumor-to-blood ratios
`of 0.8 ± 0.2 and 1.7 ± 0.4 at 2 and 4 h p.i., respectively, in the
`PDAC model, suggesting that a significant amount of 5B1−TCO
`was still present in circulation when the tracer was administered.
`High tumor uptake as well as similar residual blood radioactivity
`concentrations were observed when [68Ga]27 was tested in
`a sc colorectal cancer (CRC, SW1222 cell line, Supporting Infor-
`mation, section 7) model, performed to demonstrate the versa-
`tility and modularity of this approach. To this end, the anti-A33
`mAb huA33−TCO was injected 48 h prior to the radioligand.31,32
`Despite a relatively low tumor-to-blood activity concentration ratio
`due to residual circulating mAb, [68Ga]27 showed excellent tumor
`uptake as early as 1 h p.i., with maximum uptake values of 7.7 ±
`1.0%ID/g at 4 h p.i. (Figure 6a,b). In accordance to our data
`