The acyclic ligands H4C3octapa and p-SCN-Bn-H4C3octapa were synthesized for the first time, using nosyl protection chemistry. These new ligands were compared to the previously studied ligands H4octapa and p-SCN-Bn-H4octapa to determine the extent to which the addition of a single carbon atom to the backbone of the ligand would affect metal coordination, complex stability, and, ultimately, utility for in vivo radiopharmaceutical applications. Although only a single carbon atom was added to H4C3octapa and the metal donor atoms and denticity were not changed, the solution chemistry and radiochemistry properties were drastically altered, highlighting the importance of careful ligand design and radiometal-ligand matching. It was found that [In(C3octapa)](-) and [Lu(C3octapa)](-) were substantially different from the analogous H4octapa complexes, exhibiting fluxional isomerization and a higher number of isomers, as observed by (1)H NMR, VT-NMR, and 2D COSY/HSQC-NMR experiments. Past evaluation of the DFT structures of [In(octapa)](-) and [Lu(octapa)](-) revealed very symmetric complexes; in contrast, the [In(C3octapa)](-) and [Lu(C3octapa)](-) complexes were much less symmetric, suggesting lower symmetry and less rigidity than that of the analogous H4octapa complexes. Potentiometric titrations revealed the formation constants (log K(ML), pM) were ~2 units lower for the In(3+) and Lu(3+) complexes of H4C3octapa when compared to that of the more favorable H4octapa ligand (~2 orders of magnitude less thermodynamically stable). The bifunctional ligands p-SCN-Bn-H4C3octapa and p-SCN-Bn-H4octapa were conjugated to the antibody trastuzumab and radiolabeled with (111)In and (177)Lu. Over a 5 day stability challenge experiment in blood serum, (111)In-octapa- and (111)In-C3octapa-trastuzumab immunoconjugates were determined to be ~91 and ~24% stable, respectively, and (177)Lu-octapa- and (177)Lu-C3octapa-trastuzumab, ~89% and ~4% stable, respectively. This work suggests that 5-membered chelate rings are superior to 6-membered chelate rings for large metal ions like In(3+) and Lu(3+), which is a crucial consideration for the design of bifunctional chelates for bioconjugation to targeting vectors for in vivo work.
The acyclic ligands H4C3octapa and p-SCN-Bn-H4C3octapa were synthesized for the first time, using nosyl protection chemistry. These new ligands were compared to the previously studied ligands H4octapa and p-SCN-Bn-H4octapa to determine the extent to which the addition of a single carbon atom to the backbone of the ligand would affect metal coordination, complex stability, and, ultimately, utility for in vivo radiopharmaceutical applications. Although only a single carbon atom was added to H4C3octapa and the metaldonor atoms and denticity were not changed, the solution chemistry and radiochemistry properties were drastically altered, highlighting the importance of careful ligand design and radiometal-ligand matching. It was found that [In(C3octapa)](-) and [Lu(C3octapa)](-) were substantially different from the analogous H4octapa complexes, exhibiting fluxional isomerization and a higher number of isomers, as observed by (1)H NMR, VT-NMR, and 2D COSY/HSQC-NMR experiments. Past evaluation of the DFT structures of [In(octapa)](-) and [Lu(octapa)](-) revealed very symmetric complexes; in contrast, the [In(C3octapa)](-) and [Lu(C3octapa)](-) complexes were much less symmetric, suggesting lower symmetry and less rigidity than that of the analogous H4octapa complexes. Potentiometric titrations revealed the formation constants (log K(ML), pM) were ~2 units lower for the In(3+) and Lu(3+) complexes of H4C3octapa when compared to that of the more favorable H4octapa ligand (~2 orders of magnitude less thermodynamically stable). The bifunctional ligands p-SCN-Bn-H4C3octapa and p-SCN-Bn-H4octapa were conjugated to the antibody trastuzumab and radiolabeled with (111)In and (177)Lu. Over a 5 day stability challenge experiment in blood serum, (111)In-octapa- and (111)In-C3octapa-trastuzumab immunoconjugates were determined to be ~91 and ~24% stable, respectively, and (177)Lu-octapa- and (177)Lu-C3octapa-trastuzumab, ~89% and ~4% stable, respectively. This work suggests that 5-membered chelate rings are superior to 6-membered chelate rings for large metal ions like In(3+) and Lu(3+), which is a crucial consideration for the design of bifunctional chelates for bioconjugation to targeting vectors for in vivo work.
In recent years, radiometal-based
radiopharmaceuticals have become
increasingly appreciated for their potential in diagnostic and therapeutic
medicine.[1−10] The modular and versatile nature of these systems allows for a continually
increasing number of routinely produced radiometals to be harnessed
for single photon emission computed tomography (SPECT), positron emission
tomography (PET), and therapy (e.g., Auger electron, β–, α).[1−10] Radiometal-based radiopharmaceuticals are typically broken down
into several modules and can be assembled in many permutations for
a high degree of customization: the targeting vector (e.g., peptide,
antibody, nanoparticle) is selected for site-specific delivery to
a chosen biological target or compartment; the bifunctional chelator
(BFC) is selected for optimal stability and radiolabeling properties
with the chosen radiometal; the bioconjugation method is selected
based on the reactivity and functional groups available on the bifunctional
chelator and targeting vector; and, finally, the radiometal is selected
for its decay characteristics.[1−9] Even subtle changes to the ligand structure, site of conjugation,
and donor arms of ligands can cause drastic changes to their radiolabeling
properties and the stability of their radiometal complexes. These
ligand alterations can include the addition of functional groups to
allow for conjugation with targeting vectors (e.g., p-benzyl-isothiocyanate, maleimide, activated esters), changes to
ligand donor arms and ligand denticity (e.g., peptide coupling to
a carboxylate donor arm), and changing ligand donor types (e.g., changing
primary amines to secondary amines, or methylenephosphonates for carboxylates),
and the effects of these changes on radiolabeling and stability properties
can be significant.[1−10] The difference in stability and radiolabeling kinetics between acyclic
and macrocyclic ligands is often significant, as macrocycles provide
a degree of preorganization of the metal binding cavity, which provides
a strong entropic incentive for metal complexation, although typically
at the cost of slow reaction kinetics.[11−13] There are a number of
competent and well-studied bifunctional chelators available for use;
however, none are without shortcomings, and for purposes ranging from
enhancing radiolabeling kinetics to enhancing kinetic inertness and
in vivo stability, development of new bifunctional chelators is an
important field of study.[3,7,8,14,15]The literature is rich with examples of how small changes
in cavity
size, chelate-ring size, bifunctionalization site, and coordination
environment can make drastic changes in the stability of radiometal
complexes. This article discusses differences in binding of two structurally
similar acyclic ligands to the specific radiometals 111In and 177Lu. Although each radiometal ion has unique
aqueous chemistry properties, and the difference between macrocyclic
and acyclic ligands is substantial, examples using different metal
ions and both acyclic and macrocyclic ligands will be discussed. As
an example of cavity size effects, the chemistry of the positron-emitting
radiometal ion 64Cu2+ with the widely used,
macrocyclic chelators DOTA and NOTA provides an excellent illustration
of the importance of the careful consideration of ligand properties
(Chart 1). The denticity and cavity size of
the macrocycle NOTA is smaller than that of DOTA (6 vs 8 maximum denticity)
and therefore is a better fit for radiometals such as 64Cu and 68Ga, resulting in increased stability and inertness
of these metal complex.[16−21] A common site of bifunctional derivatization of DOTA is one of its
4 carboxylic acid arms, which effectively masks one of the carboxylic
acid chelating arms, effectively dropping the denticity from 8 to
7 (although the carbonyl can still coordinate to metal ions, albeit
weakly).[22] This is a good example of how
changing ligand donor groups can effect the stability of radiometal
complexes, as this functionalization of DOTA causes a noted decrease
in the in vitro and in vivo stability of its radiometal complexes
with isotopes like 111In (DO3A, Chart 1).[22] An example of how ligand isomers
can affect stability is the acyclic ligand CHX-DTPA, where 4 possible
ligand isomers can be synthesized, with CHX-A″-DTPA being the
most stable isomer by a substantial margin (Chart 1).[23,24] Another example of how subtle
changes in ligand donor groups can effect stability are macrocyclic
TE2A derivatives (cyclam-based, Chart 1), which
undergo remarkable changes in radiolabeling and stability properties
as a result of simply alkylating the two secondary amine groups of
TE2A to form tertiary amines, either with methyl groups (MM/DM-TE2A)
or an ethylene bridge (CB-TE2A).[25,26] The result
of these TE2A modifications are that a large improvement in the in
vitro and in vivo stability of the 64Cu complexes of MM/DM-TE2A
is observed, relative to that of TE2A.[25,26]
Chart 1
Structures
of Selected Chelatorsa
Two pyridinecarboxylate-based
ligands that differ by only 1 n class="Chemical">carbon, H4C3octapa (propylene-bridged)
and H4octapa (ethylene-bridged); their bifunctional derivatives p-SCN-Bn-H4C3octapa and p-SCN-Bn-H4octapa; the popular 111In/177Lu/86/90Y bifunctional ligands C-DOTA, DO3A-NHS, DO3A-SCN, and p-SCN-Bn-CHX-A″-DTPA; and other ligand examples Cyclam,
Cyclen, NOTA, TETA, and CB-TE2A.
An example
of the predilection certain radiometals have for 5-
or 6-membered chelate rings can be found when comparing the stability
of cyclam- and cyclen-based ligands with the Cu2+ isotope 64Cu.[25,27] The inclusion of some 6-membered
chelate rings (cyclam-based) in ligands such as TETA and CB-TE2A yields
some improvements in stability in vitro and in vivo when compared
to that for ligands that contain only 5-membered chelate rings (cyclen-based)
such as DOTA and CB-DO2A.[25] These findings
suggest that 6-membered chelate rings may work well with Cu2+, at least when comparing the specific ligand scaffolds of cyclen
and cyclam.[27] In this specific example,
it appears that copper prefers some 6-membered chelate rings (TETA)
to strictly 5-membered chelate rings (DOTA); however, the smaller
macrocycle NOTA forms more stable complexes with Cu2+ than
TETA while forming only 5-membered chelate rings.[16,17] Conversely, previous work in our laboratory on the chelation of
Ga3+ with the ligand H2dedpa suggested a better
fit for Ga3+, which is even smaller than Cu2+ (62 vs 73 pm for CN = 6, respectively), with 5-membered chelate
rings over 6-membered chelate rings.[15] The
discussion of which chelate ring-size is the best fit and the most
stable for large metal ions such as In3+, Lu3+, and Y3+ is less opaque. DOTA has been shown to be superior
to TETA for large metal ions like In3+, Lu3+, and Y3+, suggesting that the optimal stability of large
metal ions is achieved when using 5-membered chelate rings, which
supports this work comparing H4octapa to H4C3octapa.[2,3,7,8] These
examples demonstrate the lack of clarity for the preference of some
metal ions between 5- and 6-membered chelate rings, providing a compelling
reason to perform further studies on the subject.We have previously
published work on the acyclic ligand H4octapa and the bifunctional
derivative p-SCN-Bn-H4octapa, which showed
very promising ambient temperature radiolabeling
performance and in vitro and in vivo stability with 111In and 177Lu. This study was performed by conjugating p-SCN-Bn-H4octapa to an antibody (trastuzumab)
as a proof-of-principle targeting vector for HER2/neu-expressing ovarian cancer cells (SKOV3).[14,28] Largely due to their exceptional specificity and affinity, antibodies
are extremely useful targeting vectors for the delivery of radioactive
isotopes to cancerous tissues.[29,13] Antibodies also possess
long biological half-lives (∼2 to 3 weeks in vivo), which makes
them excellent vehicles for proof-of-principle studies of new bifunctional
ligands in which stability in vitro and in vivo must be studied for
several days/weeks.[13] The isotopes 111In and 177Lu are widely used in both research
and clinical settings, with 111In being a cyclotron-produced
radiometal (111Cd(p,n)111In) for SPECT imaging (t1/2 ∼ 2.8 days) and Auger electron therapy and 177Lu being a reactor-produced therapeutic radiometal (176Lu(n,γ)177Lu) that emits β– particles as well as γ-rays (t1/2 ∼ 6.6 days).[30]To further evaluate the impact of ligand structure changes on radiolabeling
efficiency and stability, we have synthesized the novel derivative
of H4octapa, H4C3octapa, modified with only
a single additional carbon atom (1,3-propylenediamine vs 1,2-ethylenediamine
backbone) (Chart 1). Herein, we report the
synthesis, characterization, In3+/Lu3+ metal
complexation, density functional theory (DFT) structure prediction,
potentiometric titrations to determine thermodynamic stability constants,
and 111In/177Lu radiolabeling and in vitro blood
serum stability of the new ligand H4C3octapa and its bifunctional
derivative p-SCN-Bn-H4C3octapa (Chart 1). In this study, we have conjugated both the new p-SCN-Bn-H4C3octapa and the previously studied p-SCN-Bn-H4octapa to the HER2/neu-targeted antibody trastuzumab and directly compared their properties.
The purpose of this work was to determine the extent to which a small
structural change of the ligand, which, importantly, does not change
the ligand denticity or metal coordinating groups, may affect radiolabeling
performance and most importantly the stability of radiometal complexes.
Our laboratory has been studying new acyclic chelators based on the
pyridinecarboxylate scaffold for several years now, and this work
aims to expand upon our understanding of ligand design and radiometal–ligand
matching (Chart 1).[14,15,31−37]
Experimental Section
Materials and Methods
All solvents and reagents were
purchased from commercial suppliers (Sigma-Aldrich, St. Louis, MO;
TCI America, Portland, OR; Fisher Scientific, Waltham, MA) and were
used as received unless otherwise indicated. The bifunctional chelator p-SCN-Bn-H4octapa was synthesized as previously
described.[28] DMSO used for chelator stock
solutions was of molecular biology grade (>99.9%: Sigma, D8418).
1,4,7,10-Tetraazacyclododecane-1,4,7,10-tetraacetic
acid para-benzylisothiocyanate (p-SCN-Bn-DOTA) was purchased from Macrocyclics, Inc. (Dallas, TX).
Methyl-6-bromomethylpicolinate was synthesized according to a literature
protocol.[14] Water used was ultrapure (18.2
MΩ cm–1 at 25 °C, Milli-Q, Millipore,
Billerica, MA). The analytical thin-layer chromatography (TLC) plates
were aluminum-backed ultrapure silica gel (Siliaplate, 60 Å pore
size, 250 μM plate thickness, Silicycle, Quebec, QC). Flash
column silica gel was provided by Silicycle (Siliaflash Irregular
Silica Gels F60, 60 Å pore size, 40–63 mm particle size,
Silicycle, Quebec, QC). Automated column chromatography was performed
using a Teledyne Isco (Lincoln, NE) CombiFlash R automated system with solid load cartridges
packed with flash column silica gel and RediSep R Gold reusable normal-phase silica columns
and neutral alumina columns (Teledyne Isco, Lincoln, NE). 1H and 13CNMR spectra were recorded on Bruker AV300, AV400,
or AV600 instruments; all spectra were internally referenced to residual
solvent peaks except for 13CNMR spectra in D2O, which were externally referenced to a sample of CH3OH/D2O. Low-resolution mass spectrometry was performed
using a Waters liquid chromatography–mass spectrometer (LC–MS)
consisting of a Waters ZQ quadrupole spectrometer equipped with an
ESCI electrospray/chemical ionization ion source and a Waters 2695
HPLC system (Waters, Milford, MA). High-resolution electrospray–ionization
mass spectrometry (ESI–MS) was performed on a Waters Micromass
LCT time-of-flight instrument. Microanalyses for C, H, and N were
performed on a Carlo Erba EA 1108 elemental analyzer. The HPLC system
used for purification of nonradioactive compounds consisted of a semipreparative
reverse-phase C18 Phenomenex synergi hydro-RP (80 Å pore size,
250 × 21.2 mm, Phenomenex, Torrance, CA) column connected to
a Waters 600 controller, a Waters 2487 dual-wavelength absorbance
detector, and a Waters delta 600 pump. 177Lu(chelate) analysis
was performed using an HPLC system comprised of a Shimadzu SPD-20A
prominence UV/vis, LC-20AB prominence LC, a Bioscan flow-count radiation
detector, and a C18 reverse-phase column (Phenomenex Luna
Analytical 250 × 4.6 mm). UV/vis measurements for determining
antibody stock solution concentrations were taken on a Thermo Scientific
Nanodrop 2000 spectrophotometer (Wilmington, DE).111In was cyclotron produced (Advanced Cyclotron Systems, Model TR30)
by proton bombardment through the reactions 111Cd(p,n)111In and was provided by Nordion as 111InCl3 in 0.05 M HCl. 177Lu was procured from PerkinElmer
(PerkinElmer Life and Analytical Sciences, Wellesley, MA, effective
specific activity of 29.27 Ci/mg) as 177LuCl3 in 0.05 M HCl. Labeling reactions were monitored using silica-gel
impregnated glass-microfiber instant thin-layer chromatography paper
(iTLC-SG, Varian, Lake Forest, CA) and analyzed on a Bioscan AR-2000
radio-TLC plate reader using Winscan Radio-TLC software (Bioscan Inc.,
Washington, DC). All radiolabeling chemistry was performed with ultrapure
water (>18.2 MΩ cm–1 at 25 °C, Milli-Q,
Millipore, Billerica, MA) that had been passed through a 10 cm column
of Chelex resin (BioRad Laboratories, Hercules, CA). Human blood serum
(Sigma, Sera, human, aseptically filled, S7023-100mL) competition
solutions were agitated at 550 rpm and held at 37 °C using an
Eppendorf thermomixer, and then177Lu(chelate) mixtures
were analyzed using GE Healthcare Life Sciences PD-10 desalting columns
(GE Healthcare, United Kingdom, MW < 5000 Da filter) that were
conditioned by elution of 25 mLphosphate-buffered saline (PBS) before
use. 177Lu/111In-immunoconugates were analyzed
using iTLC as described above and purified using PD-10 desalting columns
and Corning 50k MW Amicon Ultra centrifugation filters. Radioactivity
in samples was measured using a Capintec CRC-15R dose calibrator (Capintec,
Ramsey, NJ).
Propylenediamine (173 μL, 2.05 mmol)
was dissolved in n class="Chemical">THF (10 mL) and placed in an ice bath; then, sodium
bicarbonate (∼1 g) was added, followed by slow addition of
2-nitrobenzenesulfonyl chloride (0.954 g, 4.31 mmol). The reaction
mixture was allowed to warm to ambient temperature and stirred overnight.
The yellow mixture was filtered to remove sodium bicarbonate, rotary
evaporated to a red oil, purified by silica chromatography (CombiFlash R automated column system;
120 g HP silica; A: dichloromethane, isocratic elution), and dried
in vacuo to yield the product (1) as yellow solid (75%,
∼0.684 g). 1HNMR (300 MHz, CDCl3, 25
°C) δ 8.13–8.06 (m, 2H), 7.97–7.84 (m, 6H),
6.65 (br s, 2H, -NH-), 3.19 (m, 4H), 1.81 (quin, J = 6.9 Hz, 2H). 13CNMR (75 MHz, DMSO-d6, 25 °C) δ 149.5, 135.2, 134.6, 133.9, 131.7,
126.2, 41.9, 31.2. HR-ESI-MS calcd for [C15H16N4O8S2+H]+, 445.0488;
found, 445.0481 [M + H]+, PPM = −1.5.
To a solution of 2 (0.685 g,
0.922 mmol) in tetrahydrofuran (10 mL) was added thiophenol (223 μL,
2.03 mmol) and potassium carbonate (excess, ∼0.4 g). The reaction
mixture was stirred at 60 °C for 48 h, during which time the
color changed from light yellow to dark yellow. The reaction mixture
was filtered with a large fritted glass filter to remove K2CO3, rinsed liberally with THF and CH3CN, and
then concentrated to dryness in vacuo. The resulting crude yellow
oil was purified by alumina chromatography (CombiFlash R automated column system; 24 g neutral
alumina; A: dichloromethane, B: methanol, 100% A to 30% B gradient)
to yield 3 as yellow oil (89%, ∼0.306 g). Compound 3 was purified using neutral alumina, as it demonstrates an
abnormally high affinity for silica and requires the use of ammonium
hydroxide and >20% methanol to be eluted, resulting in partial
methyl-ester
deprotection and dissolving of some silica. 1HNMR (300
MHz, CDCl3, 25 °C) δ 7.95–7.93 (m, 2H),
7.77–7.74 (m, 2H), 7.54–7.52 (m, 2H), 4.01 (s, 4H),
3.92 (s, 6H), 3.61 (br s, -NH-, 2H), 2.81 (s, 4H), 1.78 (s, 2H). 13CNMR (75 MHz, CDCl3, 25 °C) δ 165.5,
159.5, 147.3, 137.4, 125.7, 123.5, 54.3, 52.7, 48.0, 28.6. HR-ESI-MS
calcd for [C19H24N4O4+H]+, 373.1876; found, 373.1881, [M + H]+, PPM = 1.3.
To a solution of 4 (248 mg,
0.413 mmol) in a mixture
of tetrahydrofuran/deionized water (3:1, 5 mL) was added LiOH (150
mg). The reaction mixture was stirred at ambient temperature for 4
h. A portion of HCl was added (5 mL, 6 M), and then the mixture was
reduced to dryness in vacuo. The mixture was dissolved in deionized
water (4 mL) and purified via semipreparative reverse-phase HPLC (10
mL/min, gradient: A: 0.1% TFA (trifluoroacetic acid) in deionized
water, B: CH3CN. 5 to 100% B linear gradient in 30 min. tR = 11.5–13.3 min, broad). Product fractions
were pooled, concentrated in vacuo, dissolved in HCl (5 mL, 6 M),
and then concentrated in vacuo again to remove trifluoroacetic acid.
This processes was repeated three times to remove traces of TFA. The
HCl saltH4C3octapa·4HCl·2H2O (5) was obtained as a yellow solid (74% yield, ∼0.195
g, using the molecular weight of the HCl salt as determined by elemental
analysis). 1HNMR (600 MHz, D2O) δ 7.79
(br s, 4H, pyr-H), 7.43 (br s, 2H, pyr-H), 4.56 (s, 4H, Pyr-CH2-N), 4.09 (s,
4H, HOOC-CH2-N), 3.30 (br s, 4H, propylene-1,3-H), 1.95 (m, 2H, propylene-2-H). 13CNMR (150 MHz, D2O) δ 167.2, 166.1, 148.9, 145.2,
139.5, 126.7, 126.5, 125.1, 57.4, 54.6, 52.7, 19.4. IR (neat, ATR-IR):
υ = 1732 cm–1 (C=O), 1629/1614 cm–1 (C=C py). HR-ESI-MS calcd for [C21H24N4O8 + H]+, 461.1672;
found, 461.1665, [M + H]+, PPM = −1.5. Anal. Calcd
for H4C3octapa·4HCl·2H2O (C21H24N4O8·4HCl·2H2O = 642.312 g/mol): C, 40.40; H, 4.84; N, 8.97. Found: C, 40.14 (Δ
= 0.26); H, 4.74 (Δ = 0.10); N, 8.84 (Δ = 0.13).
Na[In(C3octapa)]
(6)
H4C3octapa·4HCl·2H2O (5) (15.54 mg, 0.024 mmol) was suspended in
0.1 M HCl (1.5 mL), and In(NO3)3·6H2O (12.9 mg, 0.031 mmol) was added. The pH was adjusted to
4–5 using 0.1 M NaOH, and then the solution was stirred at
room temperature. After 1 h, the product was confirmed via mass spectrometry,
and the solvent was removed in vacuo to yield Na[In(C3octapa)] (6). 1HNMR (400 MHz, D2O, 25 °C)
δ 8.42–7.88 (m, 4H), 7.76–7.47 (m, 2H), 4.45–3.85
(m, 4H), 3.64–3.33 (m, 2H), 3.15–2.73 (m, 2H), 2.53–2.19
(m, 4H), 1.30 (m, 1H), 1.04 (m, 1H). HR-ESI-MS calcd for [C21H18115InN4O8 + 2H]+, 571.0320; found, 571.0317, [M + 2H]+, PPM = −0.5.
Na[Lu(C3octapa)] (7)
H4C3octapa·4HCl·2H2O (5) (15.9 mg, 0.025 mmol) was suspended in
0.1 M HCl (1.5 mL), and Lu(NO3)3·6H2O (15.1 mg, 0.032 mmol) was added. The pH was adjusted to
4–5 using 0.1 M NaOH, and then the solution was stirred at
room temperature. After 1 h, the product was confirmed via mass spectrometry,
and the solvent was removed in vacuo to yield Na[Lu(C3octapa)] (6). 1HNMR (300 MHz, D2O, 25 °C)
δ 8.22–7.66 (m, 5H), 7.36–7.22 (m, 1H), 4.54–4.38
(m, 1H), 4.22–3.85 (m, 3H), 3.58–3.21 (m, 4H), 2.81–2.32
(m, 4H), 1.87–1.64 (m, 2H). HR-ESI-MS calcd for [C21H18175LuN4O8 + 2H]+, 631.0689; found, 631.0680, [M + 2H]+, PPM = −1.4.
Diethyl-2-(4-nitrobenzyl)malonate (8)
This
synthesis was adapted from a literature preparation.[38] Sodium ethoxide (3.47 g, 50.9 mmol) was added
to ethanol (100 mL), followed by slow addition of diethyl malonate
(14.1 mL, 14.8 g, 92.6 mmol) and then4-nitrobenzyl bromide (10.0
g, 46.3 mmol). The dark orange reaction mixture was heated to reflux
overnight, and it was observed that a fine white precipitate formed.
The volume of the reaction mixture was reduced to 20 mL, it was placed
in the freezer for several hours to encourage maximum precipitation,
and then the shiny white solid was isolated via suction filtration.
The procedure was repeated, reducing the volume to ∼5–10
mL and then placing in the freezer to recover more product. The white
solid was filtered and rinsed with cold ethanol to isolate pure 8 (3.76 g, 12.7 mmol, ∼27%). 1HNMR (300
MHz, CDCl3) δ 8.16 (m, 2H, NO2-Ph-H),
7.40 (m, 2H, NO2-Ph-H), 4.18 (m, 4H, -O-CH2-),
2.91 (t, 1H, -CO-CH-CO, 3J = 7.8 Hz),
3.32 (d, 2H, -CH2-Ph-NO2, 3J = 7.8 Hz), 1.23 (t, 6H, -O-CH2-CH3-, 3J = 7.1 Hz). 13CNMR (100
MHz, CDCl3) δ 168.2, 146.9, 145.6, 129.8, 123.7, 61.8, 53.1,
34.3, 13.9. HR-ESI-MS calcd for [C14H17NO6+H]+, 296.1134; found, 296.1133, [M + H+]+, PPM = −0.4.
2-(4-Nitrobenzyl)malondiamide
(9)
Compound 8 (3.52 g, 11.9 mmol)
was dissolved in methanol (125 mL) to
form a clear and colorless mixture and was then placed in an ice/salt
bath. After cooling, ammonia gas was purged through the reaction vessel
for ∼10 min with vigorous stirring to allow for NH3(g) saturation. After NH3(g) saturation, the reaction
mixture turned clear yellow. The ammonia gas flow was ceased, and
the reaction vessel was sealed with a rubber septum and then allowed
to stir in the ice bath overnight. While stirring overnight, the ice
bath melted, and the reaction mixture warmed to ambient temperature.
After stirring overnight, a precipitate was observed to have formed,
and the reaction mixture was a yellow/orange suspension. The reaction
mixture was allowed to stir at ambient temperature for a second night
(∼36 h total). The precipitate was filtered out, washed with
methanol followed by boiling acetonitrile, and then dried in vacuo.
The product (9) was isolated as yellow solid (2.34 g,
9.86 mmol, ∼83%). 1HNMR (300 MHz, DMSO-d6) δ 8.15–8.12 (m, 2H, NO2-Ph-H), 7.49–7.46 (m, 2H, NO2-Ph-H), 7.30 (s, 2H,
-NH2), 7.08 (s, 2H, -NH2), 3.39 (t, 1H, 3J = 7.8 Hz, -CO-CH-CO-), 3.10 (d, 2H, 3J = 7.6 Hz, -CH2-Ph-NO2-). 13CNMR (75 MHz, DMSO-d6) δ 170.1, 148.0, 146.0, 130.1, 123.2, 54.0, 34.5. HR-ESI-MS
calcd for [C10H11N3O4+Na]+, 260.0647; found, 260.0654, [M + Na+]+, PPM = 2.6. Anal. Calcd for C10H11N3O4·0.1CH3OH = 240.424 g/mol: C, 50.46;
H, 4.78; N, 17.71. Found: C, 50.61 (Δ = 0.15); H, 4.60 (Δ
= 0.18); N, 17.26 (Δ = 0.22).
Compound 9 (1.99 g, 8.40 mmol)
was added to a two-necked
round-bottomed flask under argon gas, and an addition funnel was attached.
Borane stabilized in THF (BH3·THF, 1 M, 15 mL) was
added, and the reaction mixture was heated to reflux under argon gas
for 20 h. The reaction mixture was clear yellow with solid precipitate
on the sides of the flask. The reaction mixture was removed from heat,
∼30 mL of concentrated HCl (12 M) was added slowly, and the
mixture heated to reflux for ∼1 h. The reaction mixture was
reduced to dryness in vacuo, and then ∼40 mL of NaOH (6 M)
was added. The aqueous phase was extracted with dichloromethane (5
× 30 mL), and the light yellow organic extractions were pooled
and dried over magnesium sulfate, filtered, and reduced to dryness
in vacuo. The product was then suspended in ∼20 mL of ethanol,
concentrated HCl was added (∼3 mL, 12 M), and the flask placed
in the freezer. A light yellow precipitate formed, which was filtered
out, washed with dichloromethane, and dried in vacuo to isolate pure 10 (0.544 g, 2.39 mmol, ∼28%). 1HNMR (400
MHz, D2O) δ 8.18 (m, 2H, NO2-Ph-H (meta
to NO2)), 7.51 (m, 2H, NO2-Ph-H (ortho to NO2)), 3.19 (dd, 2H, H2N-CH2-), 3J = 6.7 Hz, 2J = 13.6
Hz), 3.06 (dd, 2H, H2N-CH2-), 3J = 6.7 Hz, 2J = 13.5 Hz), 2.97
(2, 2H, NO2-Ph-CH2-, 3J = 7.6 Hz), 2.56 (septet, 1H, -CH-, 3J = 6.9 Hz). 13CNMR (100 MHz, D2O) δ
146.6, 145.6, 130.4, 124.2, 40.22, 36.7, 34.9. HR-ESI-MS calcd for
[C10H15N3O2+H]+, 210.1214; found, 210.1240, [M + H+]+, PPM
= −1.2.
To a solution of 12 (224.4 mg, 0.278 mmol) in tetrahydrofuran
(5 mL) were added thiophenol (58.2 μL, 0.569 mmol) and potassium
carbonate (excess, ∼0.4 g). The reaction mixture was stirred
at ambient temperature for 72 h as a slow color change from light
yellow to dark yellow occurred. The crude reaction mixture was filtered
out with a large fritted glass filter, rinsed liberally with THF and
CH3CN, and then concentrated to dryness in vacuo. The resulting
crude yellow oil was purified by silica chromatography (CombiFlash R automated column system;
24 g HP silica; A: dichloromethane, B: methanol, 100% A to 30% B gradient)
to yield 13 as clear yellow oil (90%, ∼109 mg). 1HNMR (300 MHz, CDCl3, 25 °C) δ 8.08
(d, J = 9.0 Hz, 2H), 7.32 (d, J =
9.0 Hz, 2H), 3.20 (s, 4H), 2.74 (d, J = 8.0 Hz, 2H),
2.53 (d, J = 8.0 Hz, 4H), 1.93 (m, 1H), 1.85 (s,
2H), 1.39 (s, 18H). 13CNMR (75 MHz, CDCl3,
25 °C) δ 171.6, 148.6, 146.2, 129.8, 123.3, 80.9, 51.8,
51.5, 40.8, 36.8, 27.9. HR-ESI-MS calcd for [C22H35N3O6+H]+, 438.2604; found, 438.2593,
[M + H]+, PPM = −2.5.
Compound 14 (92.4 mg, 0.126 mmol) was dissolved
in glacial acetic acid (2.5 mL) with hydrochloric acid (2.5 mL, 3
M). Palladium on carbon (10 wt %) was added, and hydrogen gas (balloon)
was purged through the reaction vessel that was then sealed with a
rubber septum. The reaction mixture was stirred vigorously at ambient
temperature for 1 h and then filtered to remove Pd/C and washed ad
libitum with methanol and hydrochloric acid (3 M). The crude reaction
mixture was concentrated in vacuo, dissolved in a mixture of tetrahydrofuran/deionized
water (3:1, 5 mL), LiOH (150 mg) was added, and the mixture was stirred
at ambient temperature for 1 h. The crude reaction mixture was dissolved
in hydrochloric acid (∼2 mL, 3 M), heated to near boiling with
a heat gun to affect full tert-butyl ester deprotection,
and then allowed to cool to ambient temperature. The crude reaction
mixture was then mixed with thiophosgene (purchased suspended in chloroform)
in ∼0.2 mL of additional chloroform (15 equiv, 144 μL,
1.88 mmol) to react overnight at ambient temperature with vigorous
stirring. The reaction mixture was washed with chloroform (5 ×
1 mL) by vigorous biphasic stirring, followed by removal of the organic
phase using a pipet to extract excess thiophosgene. The aqueous phase
was then diluted to a volume of 4.5 mL with deionized water and injected
directly onto a semiprep HPLC column for purification (A: 0.1% TFA
in deionized water, B: 0.1% TFA in CH3CN, 100% A to 60%
B gradient over 40 min). p-SCN-Bn-H4C3octapa
(15) was found in the largest peak at tR = 35 min (broad), lyophilized overnight, and was isolated
as a fluffy off-white solid (53% over three steps from 15, ∼40 mg). 1HNMR (400 MHz, MeOD, 25 °C) δ
8.06–8.03 (m, 2H), 7.91 (t, J = 7.9 Hz, 2H),
7.55–7.53 (m, 2H), 7.27 (d, J = 8.6 Hz, 2H),
7.16 (d, J = 8.2 Hz, 2H), 4.70 (d, J = 7.3 Hz, 2H), 4.48 (d, J = 14.3 Hz, 2H), 4.19
(d, J = 17.7 Hz, 2H), 3.84 (d, J = 17.7 Hz, 2H), 3.44 (d, J = 12.0 Hz, 2H), 3.28
(d, J = 12.6 Hz, 2H), 2.94 (m, 1H), 2.61 (d, J = 7.2 Hz, 2H). 13CNMR (100 MHz, MeOD, 25 °C)
δ 171.7, 167.4, 155.0, 149.7, 139.9, 139.5, 137.1, 131.8, 139.5,
129.4, 126.9, 126.1, 61.2, 58.7, 55.3, 37.2, 33.3. IR (neat, ATR-IR):
υ = 2097 cm–1 (S=C=N-), 1718/1661
cm–1 (C=O), 1594 cm–1 (C=C
py). HR-ESI-MS calcd for [C29H29N5O8S+H]+, 608.1815; found, 608.1822, [M + H]+, PPM = 1.2.
Molecular Modeling
Calculations
were performed using
the Gaussian 09[39] and GaussView packages.
Molecular geometries and electron densities were obtained from density
functional theory calculations, with the B3LYP functional employing
the 6-31+G(d,p) basis set for first- and second-row elements, and
the Stuttgart/Dresden and associated ECP’s basis set was employed
for the metals, lutetium and indium.[40,41] Solvent (water)
effects were described through a continuum approach by means of the
IEF PCM as implemented in G09. The electrostatic potential was mapped
onto the calculated electron density surface. The corresponding harmonic
vibration frequencies were computed at the same level to characterize
the geometry as a minimum.
Solution Thermodynamics
The experimental
procedures
and details of the apparatus closely followed our reported studies
of H2dedpa/Ga3+, and H4octapa with
In3+, Lu3+, and Y3+.[14,15,28,42] As a result of the strength of the binding of the In3+ and Lu3+ complexes [In(C3octapa)]− and
[Lu(C3octapa)]−, the complex formation constant
with this ligand could not be determined directly, and ligand–ligand
competition using the known competitor, Na2H2EDTA, was employed. Potentiometric titrations were performed using
a Metrohm Titrando 809 equipped with a Ross combination pH electrode
and a Metrohm Dosino 800. Data were collected in triplicate using
PC Control (version 6.0.91, Metrohm). The titration apparatus consisted
of a water-jacketed glass vessel maintained at 25.0 (±0.1 °C,
Julabo water bath). Prior to and during the course of the titration,
a blanket of nitrogen, passed through 10% NaOH to exclude any CO2, was maintained over the sample solution. Indium and lutetium
ion solutions were prepared by dilution of the appropriate atomic
absorption standard (AAS) solution. The exact amount of acid present
in the indium and lutetium standard was determined by separate titrations
of equimolar solutions of In3+ or Lu3+ and Na2H2EDTA. The amount of acid present was determined
by Gran’s method.[43] Calibration
of the electrode was performed prior to each measurement by titrating
a known amount of HCl with 0.1 M NaOH. Calibration data were analyzed
by standard computer treatment provided within the program MacCalib[44] to obtain the calibration parameters E0. Ligand solutions were prepared 24 h in advance
of titrations to allow for equilibration. Electrode equilibration
times for titrations were up to 10 min for pKa titrations and up to 3 h for metal complex titrations (<0.2
mV/min drift allowed). Ligand and metal concentrations were 0.75–1.0
mM for potentiometric titrations. The data were treated with Hyperquad2008.[45] The proton dissociation constants corresponding
to hydrolysis of In3+(aq) and Lu3+(aq) ions
included in the calculations were taken from Baes and Mesmer.[46] The KML value for
the indium–EDTA and lutetium–EDTA complexes was taken
from Martell. Values of pM were calculated under physiologically relevant
conditions of pH 7.4, 10 μM ligand, and 1 μM metal. All
values and errors represent the average of at least three independent
experiments.
Trastuzumab (purchased commercially as Herceptin,
Genentech, San
Francisco, CA) was purified using centrifugal filter units with a
50 000 molecular weight cutoff (Amicon ultra centrifuge filters,
Ultracel-50: regenerated cellulose, Millipore Corp., Billerica, MA)
and phosphate buffered saline (PBS, pH 7.4) to remove α-α-trehalose
dihydrate, l-histidine, and polysorbate 20 additives. After
purification, the antibody was taken up in PBS, pH 7.4. Subsequently,
300 μL of antibody solution (150–250 μM) was combined
with 100 μL of PBS (pH 8.0); the pH of the resulting solution
was adjusted to 8.8–9.0 with 0.1 M Na2CO3, and 4 equiv of the p-SCN-Bn-H4C3octapa
or p-SCN-Bn-H4octapa was added in 10 μL
DMSO. The reactions were incubated at 37 °C for 1 h, followed
by centrifugal filtration to purify the resultant antibody conjugate.
The final modified antibody stock solutions were stored in PBS (pH
7.4) at 4 °C.
111In- and 177Lu-C3octapa/octapa–Trastuzumab
Radiolabeling
Aliquots of H4C3octapa/H4octapa–trastuzumab immunoconjugates (∼400 μg)
were transferred to 2 mL microcentrifuge tubes and made up to 1 mL
of ammonium acetate buffer (pH 5.5, 200 mM), and then aliquots of 177Lu or 111In were added (∼2–3 mCi).
The H4octapa–trastuzumab mixtures were allowed to
radiolabel at room temperature for 30 min and then analyzed via iTLC
with an eluent of 50 mM EDTA (pH 5) and confirmed reproducible values
>95% RCY, with 111In being ∼94–95% (∼2.23
mCi/mg) and 177Lu being ∼97–98% (∼2.52
mCi/mg) (Figures S21–S24). The H4C3octapa–trastuzumab mixtures were allowed to radiolabel
at room temperature for 60 min and then analyzed via iTLC with RCY
for 111In of ∼95–96% (∼2.25 mCi/mg)
and RCY of 177Lu of ∼8–9% (∼0.3 mCi/mg).
The low radiolabeling yield for 177Lu appears to be due
to immediate transchelation of 177Lu to the iTLC EDTA mobile
phase, suggesting poor stability of the H4C3octapa complex.
EDTA solution (30 μL, 50 mM, pH 5) was then added to the reaction
mixture; the resultant radiolabeled immunoconjugates were then purified
using size-exclusion chromatography (Sephadex G-25 M, PD-10 column,
30 kDa, GE Healthcare; dead volume = 2.5 mL, eluted with 1 mL fractions
of PBS, pH 7.4) and centrifugal column filtration (Amicon ultra 50k
MWCO). The radiochemical purity of the final radiolabeled bioconjugate
was assayed by radio-iTLC and was found to be >99% for both H4octapa–trastuzumab samples (Figures
S21–24), but for 111In-C3octapa–trastuzumab
a small amount of 111In was transchelated by the EDTA mobile
phase after purification (Figures S21–24), and for 177Lu–C3octapa–trastuzumab, a
very substantial amount of 177Lu was leached out of the
ligand and transchelated by the EDTA mobile phase (Figures S21–24). In the iTLC experiments, 111In- and 177Lu-octapa/DOTA–trastuzumab remained
at the baseline, whereas 177Lu3+/111In3+ ions complexed as [111In/177Lu]–EDTA eluted with or near the solvent front.
111In- and 177Lu-C3octapa/octapa–Trastuzumab
Blood Serum Competition Experiments
Frozenhuman blood serum
was thawed for 30 min, and 300 μL aliquots were transferred
to 2.0 mL Corning centrifuge vials. A portion of radiolabeled immunoconjugate
(∼50 μg, ∼400 μCi) was transferred to the
blood serum. Serum competition samples were then incubated at 37 ±
0.1 °C with gentle agitation (300 rpm) and analyzed via iTLC
with an EDTA eluent (50 mM, pH 5.0) (Bioscan AR-2000) at time points
of 0, 24, 48, 72, 96, and 120 h.
[177Lu(chelate)]
Radiolabeling and Serum Stability
For 177Lu experiments
with nonbifunctional ligands,
the ligands H4octapa, H4C3octapa, DTPA, and
DOTA were used. Aliquots of each chelator stock solution (1 mg/mL)
were transferred to Corning 2.0 mL self-standing microcentrifuge tubes
containing ∼1.2 mCi of 177Lu to a ligand concentration
of ∼180 μM and made up to 1 mL with NaOAc buffer (10
mM, pH 5.0). H4octapa, H4C3octapa, and DTPA
were allowed to radiolabel at ambient temperature for 10 min, and
DOTA was radiolabeled for 1 h at 90 °C. Radiometal complexes
were then evaluated using radio-HPLC to confirm quantitative radiometal
coordination (linear gradient A: 0.1% TFA in H2O, B: CH3CN, 0–80% B over 30 min). Radiolabeled ligands were
then used for blood serum stability assays. Frozenhuman blood serum
was thawed for 30 min, and 750 μL aliquots were transferred
to 2.0 mL Corning centrifuge vials. Three hundred microliters of each 177Lu(chelate) was transferred to 750 μL of blood serum
along with 450 μL of PBS to a total volume of 1.5 mL (n = 3 for each ligand). The final 177Lu(chelate)
concentration present in serum was ∼36 μM. Serum competition
samples were then incubated at 37 ± 0.1 °C with constant
agitation (550 rpm) and analyzed via PD-10 size-exclusion column elution
(filters MW < 5000 Da) at 1.5 and 24 h time points and counted
using a Capintec CRC-15R dose calibrator. Aliquots of 500 μL
of each serum/177Lu(chelate) competition solution (n = 3) were removed from the competition vial, diluted to
2.5 mL with PBS, and counted. The diluted aliquot of serum competition
mixture was loaded onto a conditioned PD-10 column. The loading volume
(2.5 mL) was eluted into radioactive waste, and then an additional
3.5 mL of PBS was loaded, collected, and counted in the dose calibrator
as the serum-bound 177Lu (nonchelate bound). Percent stability
was reported as a percentage of 177Lu still chelate-bound
and not associated with serum proteins (MW < 5000 Da). Data is
shown in Table 3. In previous work, free radiometal
(111In) was loaded onto a PD10 column in the same manner
as described above, and it was found that no activity was eluted.[14] Additionally, 111In was added directly
to a mixture of mouse serum, loaded onto a PD10 column, and it was
found that no activity was retained on the column after elution.[14]
Table 3
Relevant Bond Lengths (Å) and
Angles (deg) Comparing the DFT-Calculated In3+ and Lu3+ Complexes of H4octapa and H4C3octapa
bond lengths
(Å)
[In(octapa)]–
[In(C3octapa)]–
[Lu(octapa)]–
[(Lu(C3octapa)]–
(Ac-COO) O1–M
2.200
2.213
2.218
2.296
(Ac-COO) O2–M
2.201
2.250
2.218
2.280
(pyr-COO) O3–M
2.295
2.263
2.315
2.326
(pyr-COO) O4–M
2.294
2.280
2.315
2.325
(en/pn-N) N1–M
2.538
2.494
2.756
2.593
(en/pn-N) N2–M
2.538
2.486
2.756
2.576
(pyr-N) N3–M
2.241
2.256
2.501
2.451
(pyr-N) N4–M
2.241
2.280
2.501
2.399
N1–M–N2 angle (deg)
74.8
93.0
67.8
90.8
Results
Synthesis and Characterization
The ligand H4C3octapa and its bifunctional derivative p-SCN-Bn-H4C3octapa were synthesized for the
first time using a synthetic
route similar to a recently developed method for H4octapa
(Scheme 1).[28] Although
the synthesis of p-SCN-Bn-H4octapa utilizes
the enantiopure starting material l-4-nitrophenylalanine,
for p-SCN-Bn-H4C3octapa a different synthesis
was used, starting from diethylmalonate and 4-nitrobenzyl bromide,
so that a single additional carbon atom could be inserted to obtain
a symmetric propylene bridged backbone (Scheme 2).[15,47,14,38] The 2-nitrobenzenesulfonamide (nosyl) amine-protecting
group has made the synthesis of picolinic acid-based ligands higher
yielding and more efficient.[28,42,48−53] Synthesis of H4C3octapa was completed in five steps with
a cumulative yield of ∼44%, and p-SCN-Bn-H4C3octapa, in seven steps with a cumulative yield of ∼29%
(not including synthesis of the backbone diamine fragment).
Scheme 1
Synthesis
of H4C3octapa (5) Utilizing Nosyl
Protection Chemistry
(i) THF, NaHCO3 (excess),
2-nitrobenzenesulfonyl chloride, 0 °C → RT, 20 h (1); (ii) DMF, Na2CO3 (excess), methyl-6-bromomethylpicolinate,
50 °C, 20 h (2); (iii) THF, thiophenol, K2CO3 (excess), RT, 72 h (3); (iv) CH3CN, Na2CO3 (excess), tert-butylbromoacetate,
60 °C, 20 h (4); (v) THF/H2O (3:1), LiOH,
RT, 4 h (H4C3octapa, 5). Cumulative yield
of ∼44% in five steps.
Scheme 2
Synthesis of p-SCN-Bn-H4C3octapa (15) Utilizing
Nosyl Protection Chemistry
(i) EtOH, NaOEt, Δ,
20
h (8); (ii) MeOH, NH3(g), 0 °C →
RT, 36 h ; (iii) BH3·THF, Δ, Ar(g), 20 h (10); (iv) THF, NaHCO3 (excess), 0 °C →
RT, 20 h (11); (v) DMF, Na2CO3 (excess),
50 °C, 20 h (12); (vi) THF, K2CO3 (excess), RT, 72 h (13); (vii) CH3CN, Na2CO3 (excess), 60 °C, 20 h (14); (viii) 5 mL of (1:1) AcOH (glacial)/HCl (3 M), Pd/C (10
wt %), H2(g), RT, 1 h; (ix) THF/H2O (3:1), LiOH,
RT, 4 h; (x) thiophosgene in CHCl3 (15 equiv), HCl (3 M),
RT, 20 h (p-SCN-Bn-H4C3octapa, 15). Cumulative yield of ∼29% in seven steps.
Synthesis
of H4C3octapa (5) Utilizing Nosyl
Protection Chemistry
(i) THF, n class="Chemical">NaHCO3 (excess),
2-nitrobenzenesulfonyl chloride, 0 °C → RT, 20 h (1); (ii) DMF, Na2CO3 (excess), methyl-6-bromomethylpicolinate,
50 °C, 20 h (2); (iii) THF, thiophenol, K2CO3 (excess), RT, 72 h (3); (iv) CH3CN, Na2CO3 (excess), tert-butylbromoacetate,
60 °C, 20 h (4); (v) THF/H2O (3:1), LiOH,
RT, 4 h (H4C3octapa, 5). Cumulative yield
of ∼44% in five steps.
Synthesis of p-SCN-Bn-H4C3octapa (15) Utilizing
Nosyl Protection Chemistry
(i) EtOH, NaOEt, Δ,
20
h (8); (ii) n class="Chemical">MeOH, NH3(g), 0 °C →
RT, 36 h ; (iii) BH3·THF, Δ, Ar(g), 20 h (10); (iv) THF, NaHCO3 (excess), 0 °C →
RT, 20 h (11); (v) DMF, Na2CO3 (excess),
50 °C, 20 h (12); (vi) THF, K2CO3 (excess), RT, 72 h (13); (vii) CH3CN, Na2CO3 (excess), 60 °C, 20 h (14); (viii) 5 mL of (1:1) AcOH (glacial)/HCl (3 M), Pd/C (10
wt %), H2(g), RT, 1 h; (ix) THF/H2O (3:1), LiOH,
RT, 4 h; (x) thiophosgene in CHCl3 (15 equiv), HCl (3 M),
RT, 20 h (p-SCN-Bn-H4C3octapa, 15). Cumulative yield of ∼29% in seven steps.
Following ligand synthesis and purification, portions
of the pure
H4C3octapa·4HCl·2H2O salt (formula
determined by elemental analysis) were mixed separately with both
indium nitrate [In(NO3)3·6H2O] and lutetium nitrate [Lu(NO3)3·6H2O] to form the coordination complexes. These metal ion complexes
were characterized using standard 1HNMR and HR-ESI-MS
techniques, 2D 1H–1H COSY and 1H–13C HSQC NMR, and variable temperature (VT) NMR
spectroscopy. Due to the presence of multiple isomers formed under
aqueous conditions for [In(C3octapa)]− and [Lu(C3octapa)]−, and fluxional isomerization observed for [Lu(C3octapa)]− at ambient temperature, complete 13CNMR
spectra could not be obtained in a reasonable time (e.g., no suitable
spectrum for 13CNMR, 150 MHz, 20 h acquisition). 2D-HSQC
heteronuclear single bond correlation experiments provide more sensitive
acquisition of 13C signals than do standard 13CNMR experiments, so because of the difficulty obtaining complete 13CNMR spectra, HSQC spectra were obtained in their place.
It is important to note that the different isomers of these metal
complexes could not be separated during purification by semipreparative
reverse-phase high-performance liquid chromatography, as the complexes
eluted as single broad peaks, as did the previously studied [In(octapa)]−, [Lu(octapa)]−, and [Y(octapa)]− complexes.[14,28,42]
1H NMR Spectroscopy of [In(C3octapa)]− and [Lu(C3octapa)]− Compared to that of [In(octapa)]− and [Lu(octapa)]−
The previously
studied compound, p-SCN-Bn-H4octapa, is
asymmetric and contains a single stereocenter, which results in weak
nuclear magnetic resonance (NMR) spectra with complicated coupling
patterns, making NMR characterization challenging.[28] In contrast, p-SCN-Bn-H4C3octapa
contains no stereocenters and retains ∼C2 symmetry, which results in simpler coupling patterns, easier
interpretation of NMR spectra, and faster spectral acquisition (Figures S8–S10). The nonbifunctional ligands
H4octapa and H4C3octapa both possess ∼C2 symmetry, but when they
are complexed to metal ions such as In3+ and Lu3+, they show substantial changes in coupling patterns and chemical
shifts.The [C3octapa]4– metal ion complexes
with In3+ and Lu3+ were compared to the analogous
[octapa]4– complexes to compare the differences
between the coordination chemistry of these two structurally similar
ligands. A comparison of Figures 1 and 2 highlights a large difference in the NMR spectra
of the In3+ and Lu3+ metal complexes of [C3octapa]4– and [octapa]4–, despite these ligands
only differing by a single backbone carbon. H4C3octapa
forms a 6-membered chelate ring when coordinated to metal ions, whereas
H4octapa forms only 5-membered chelate rings. It is established
that 5-membered chelate rings are generally more thermodynamically
favorable than 6-membered rings for large metal ions, so this small
structure difference between ligands (ethylene vs propylene bridge)
is expected to impact stability to an extent.[11] The metal ion complexes of H4octapa (Figure 2) display less isomerization and fluxionality than
do those of H4C3octapa (Figure 1), with the [In(octapa)]− complex showing only
1 static isomer at ambient temperature, where the 1HNMR
spectrum of [In(C3octapa)]− shows a complicated
set of sharp peaks, suggesting the presence multiple static isomers
at ambient temperature. The contrast betweenLu3+ complexes
is similar, with the 1HNMR spectrum of [Lu(C3octapa)]− showing broad signals suggesting fluxional isomerization
between multiple isomers at ambient temperature (slow on the NMR time
scale), whereas [Lu(octapa)]− shows sharp and well-resolved 1HNMR peaks, suggesting the presence of multiple static isomers
at ambient temperature.
Figure 1
Stacked NMR spectra of H4C3octapa
(D2O, 25
°C, 600 MHz), metal complex [In(C3octapa)]− (D2O, 25 °C, 400 MHz) showing complicated but sharp
signals suggesting multiple static isomers, and [Lu(C3octapa)]− (D2O, 25 °C, 300 MHz) showing complicated
and broad signals suggesting fluxional isomerization.
Figure 2
Stacked 1H NMR spectra of H4octapa
(D2O, 25 °C, 300 MHz), metal complex [In(octapa)]− (D2O, 25 °C, 600 MHz) showing simple
and sharp diastereotopic
splitting suggesting the presence of one static isomer, and [Lu(octapa)]− (D2O, 25 °C, 400 MHz) showing complicated
and sharp signals suggesting multiple static isomers.
Stacked NMR spectra of n class="Chemical">H4C3octapa
(D2O, 25
°C, 600 MHz), metal complex [In(C3octapa)]− (D2O, 25 °C, 400 MHz) showing complicated but sharp
signals suggesting multiple static isomers, and [Lu(C3octapa)]− (D2O, 25 °C, 300 MHz) showing complicated
and broad signals suggesting fluxional isomerization.
Stacked 1Hn class="Chemical">NMR spectra of H4octapa
(D2O, 25 °C, 300 MHz), metal complex [In(octapa)]− (D2O, 25 °C, 600 MHz) showing simple
and sharp diastereotopic
splitting suggesting the presence of one static isomer, and [Lu(octapa)]− (D2O, 25 °C, 400 MHz) showing complicated
and sharp signals suggesting multiple static isomers.
The ethylene protons of H4octapa originally
appear as
a single multiplet at ∼3.0 ppm (Figure 2), but these signals in the [Lu(octapa)]− complex
appear as two broad multiplets at ∼2.0–2.5 ppm (∼50:50
integration, Figure 2), suggesting the presence
of at least two isomers. On the basis of these observations, H4octapa appears to form In3+ and Lu3+ complexes with less fluxionality and fewer isomers than the analogous
H4C3octapa complexes, suggesting that the extra carbon
present in H4C3octapa results in a more flexible backbone
and a lower energetic barrier to interconversion between isomers.
Because these ligands are studied with the long-term goal of being
used in radiopharmaceutical agents, it would be logical that more
inert and static complexes such as those formed by H4octapa
would be favorable in order to maximize stability and inertness in
vivo.
2D COSY and HSQC NMR Spectroscopy of [In(C3octapa)]− and [Lu(C3octapa)]−
The 1HNMR spectra of [In(C3octapa)]− and [Lu(C3octapa)]− revealed substantially more complicated splitting
patterns than did the spectra of the analogous H4octapa
complexes and thus required the 2D NMR techniques 1H–1H COSY and 1H–13C HSQC to provide
information on the number of isomers present. Due to the solution
behavior of these metal complexes showing fluxional isomerization
or the presence of multiple isomers, 13CNMR spectra were
not obtained, as the sensitivity was very poor (e.g., no usable 13C spectrum after 20 h at 150 MHz). In order to gain more
insight into the type of isomerization occurring with these metal
complexes, 2D-COSY NMR experiments were performed to assess the 1H–1H correlations.It can be observed
in the 1HNMR spectrum of [In(C3octapa)]− (Figure 1) that the signal arising from the
central propylene-bridge −CH2– changed from
a singlet in the free ligand at ∼1.9 ppm (integration of 2H)
to two broad signals at ∼0.9 and ∼1.3 ppm (total integration
of 2H, ∼50% for each peak), which show no correlation to each
other in the COSY spectra (see red arrows, Figures 3 and S4). The splitting of this 1HNMR singlet into two broad peaks upon In3+ coordination
suggest that at least two major isomers are formed (∼50:50),
and the lack of correlation between these two peaks in the COSY NMR
suggests that they arise from chemically distinct isomers in solution
(Figure 3). The 1HNMR signal arising
from the central propylene-bridge CH2 of [Lu(C3octapa)]− was observed as a single broad signal (Figure 1) rather than two broad signals, as observed for
[In(C3octapa)]− (Figures 1 and 3), most likely due to signal averaging
from a faster rate of fluxional isomerization at ambient temperature.
Figure 3
1H–COSY NMR (400 mHz, D2O, 25 °C)
spectrum of [In(C3octapa)]− showing an expansion
of the alkyl-region, highlighting two broad signals with red arrows
arising from the central −CH2– of the propylene
bridge, showing no 1H–1H correlations
to each other (∼50:50 integration between both peaks), suggesting
that they arise from chemically distinct static isomers in solution
(for full COSY spectrum and expansions, see Figures
S13 and S14).
1H–COSY NMR (400 mHz, D2O, 25 °C)
spectrum of [In(C3octapa)]− showing an expansion
of the alkyl-region, highlighting two broad signals with red arrows
arising from the central −CH2– of the propylene
bridge, showing no 1H–1H correlations
to each other (∼50:50 integration between both peaks), suggesting
that they arise from chemically distinct static isomers in solution
(for full COSY spectrum and expansions, see Figures
S13 and S14).In addition to COSY experiments, 2D-HSQC heteronuclear single
bond
correlation (1H–13C) NMR spectra were
obtained, which utilized an interesting diagnostic handle to assess
isomerization via the 1J C–H correlations
of the pyridine rings (Figure 4). As previously
mentioned, the 2D-HSQC experiments were more sensitive than standard 13C experiments, allowing for detection of 13C signals
via 1H–13C cross-peaks. The 2D-HSQC 1H–13C experiments revealed a complicated
set of cross-peaks arising from multiple static isomers, with the
aromatic C–H pyridine signals being the most diagnostic and
simple to evaluate. In the case of the free ligand or a single metal–ligand
isomer, 6 carbon signals (3 if C2 symmetric) should be observed in the HSQC spectra arising
from the 1J C–H pyridine ring correlations
of the 3 aromatic C–H groups on each pyridine ring. It was
observed in the HSQC spectra of [In(C3octapa)]− that
13 unique 13C signals were detected, suggesting that at
least 2 major isomers were present and possibly 1 minor isomer (Figure 4). If minor isomers are present, then the additional
signals may have been too weak to observe from this experiment. The
number of alkyl-region C–H correlations in the HSQC spectrum
of [In(C3octapa)]− reveals correlations to at least
17 unique carbon atoms, with each single molecule possessing only
7 alkyl-region carbon atoms (Figure S17). Because 14 unique carbon atoms would be expected to arise from
two unique isomers, this observation suggests that there are more
than 2 static isomers present in aqueous solution, with the most likely
explanation being two major isomers (∼50:50 ratio) and one
or more minor isomers.
Figure 4
1H–13C HSQC NMR (400/100
mHz, D2O, 25 °C) expansion of aromatic signals in
the spectrum
of [In(C3octapa)]−, showing correlations to 12 unique 13C signals, with an additional aromatic signal at 125.32 ppm
not shown (13 aromatic carbon atoms total) (13C NMR spectra
externally referenced to MeOH in D2O) (for full HSQC spectrum
and expansions, see Figures S17 and S18).
1H–13C HSQC NMR (400/100
mHz, D2O, 25 °C) expansion of aromatic signals in
the spectrum
of [In(C3octapa)]−, showing correlations to 12 unique 13C signals, with an additional aromatic signal at 125.32 ppm
not shown (13 aromatic carbon atoms total) (13CNMR spectra
externally referenced to MeOH in D2O) (for full HSQC spectrum
and expansions, see Figures S17 and S18).The COSY (Figures S15 and S16) and HSQC
spectral signals of [Lu(C3octapa)]− were substantially
weaker than those of [In(C3octapa)]−, as would be
expected due to the increased rate of fluxional behavior of the Lu3+ complex at ambient temperature, as demonstrated by the broad 1H spectrum at 25 °C (Figures 1 and 5). The HSQC spectrum of [Lu(C3octapa)]− was poorly resolved, with 10 unique carbons being
observed in the aromatic region, which corresponded to pyridine C–H
carbons, suggesting the presence of at least 2 fluxional isomers in
solution at ambient temperature (Figure 5).
The higher rate of fluxional isomerization and the coalescence of
NMR signals observed for [Lu(C3octapa)]− explains
why fewer 13CNMR signals were detected relative to those
of [In(C3octapa)]− (Figure 4 vs 5). 2D-HMBC heteronuclear multiple bond
correlation NMR experiments were also performed; however, after 10–12
h of acquisition time, the spectral signals obtained were too weak
to evaluate (∼15 mg of compound in ∼300 μL of
D2O, 600 mHz).
Figure 5
1H–13C HSQC NMR
(600/150 mHz, D2O, 25 °C) expansion of aromatic signals
in spectrum of
[Lu(C3octapa)]−, showing correlations to 10 unique 13C signals, with spectral resolution being worse than that
of [In(C3octapa)]− in Figure 4 due to the broad signals from fluxional isomerization (13C NMR spectra externally referenced to MeOH in D2O) (for
full HSQC spectrum and expansions, see Figures
S19 and S20).
1H–13C HSQC NMR
(600/150 mHz, D2O, 25 °C) expansion of aromatic signals
in spectrum of
[Lu(C3octapa)]−, showing correlations to 10 unique 13C signals, with spectral resolution being worse than that
of [In(C3octapa)]− in Figure 4 due to the broad signals from fluxional isomerization (13CNMR spectra externally referenced to MeOH in D2O) (for
full HSQC spectrum and expansions, see Figures
S19 and S20).
Variable Temperature (VT) 1H NMR of [In(C3octapa)]− and [Lu(C3octapa)]−
In
order to gain more insight into the solution chemistry of these metal
complexes, variable temperature (VT) 1HNMR experiments
were performed with [In(C3octapa)]− and [Lu(C3octapa)]−. At 85 °C (maximum temperature for D2O), full coalescence could not be achieved with either complex, but
significant broadening from fluxional isomerization could be clearly
observed in both samples (Figures 6 and 7). The 1HNMR of [Lu(C3octapa)]− demonstrated broad signals at ambient temperature (Figure 6), suggesting fluxional isomerization (slow on the
NMR time scale). As the temperature was increased to 85 °C in
20 °C increments, the peaks were observed to coalesce, merging
toward 4 broad alkyl-region peaks (Figure 6) as the rate of fluxional interconversion increased, correlating
to the 4 unique 1HNMR signals observed as singlets in
the spectra of the free ligand H4C3octapa (Figure 1). [In(C3octapa)]− showed sharp
but complicated 1HNMR splitting patterns at ambient temperature,
suggesting the presence of multiple static isomers, with peaks broadening
and beginning to coalesce as the temperature was increased to 85 °C
(Figure 7). At 85 °C, the 1HNMR signals of [In(C3octapa)]− also appeared
to be coalescing toward 4 broad signals in a manner similar to that
observed with [Lu(C3octapa)]−; however, the signals
from [Lu(C3octapa)]− remained substantially broader
(Figure 6) than those observed for the In3+ complex (Figure 7), suggesting a
higher energetic barrier to fluxional interconversion for [In(C3octapa)]−. It can also be observed that the signals arising
from the central propylene bridge −CH2– group
of [In(C3octapa)]− (red arrow, Figure 3) began to merge at elevated temperatures (Figure 7), further supporting the assignment of these two
signals as arising from different isomers, where coalescence from
signal averaging began to occur at faster rates of fluxional interconversion
at higher temperatures. At higher temperatures (e.g., 135 °C
in DMSO-d6), a further increase in coalescence
would be expected for both samples; however, D2O was chosen
as solvent for its biological relevance.
Figure 6
Variable temperature
(VT) NMR experiments with [Lu(C3octapa)]− (D2O, 400 MHz), showing broad signals at
25 °C most likely arising from fluxional isomers and/or aqua
ligand exchange at 25 °C, with further broadening and coalescing
being observed as the temperature was increased to 85 °C in 20
°C increments, suggesting fluxional isomerization between multiple
isomers was increased elevated temperatures.
Figure 7
Variable temperature (VT) NMR experiments with [In(C3octapa)]− (D2O, 400 MHz), showing sharp signals and
coupling patterns most likely arising from multiple static isomers
at 25 °C, broadening and coalescing as the temperature was increased
to 85 °C in 20 °C increments, suggesting fluxional isomerization
between multiple isomers at elevated temperatures.
Variable temperature
(VT) n class="Chemical">NMR experiments with [Lu(C3octapa)]− (D2O, 400 MHz), showing broad signals at
25 °C most likely arising from fluxional isomers and/or aqua
ligand exchange at 25 °C, with further broadening and coalescing
being observed as the temperature was increased to 85 °C in 20
°C increments, suggesting fluxional isomerization between multiple
isomers was increased elevated temperatures.
Variable temperature (VT) n class="Chemical">NMR experiments with [In(C3octapa)]− (D2O, 400 MHz), showing sharp signals and
coupling patterns most likely arising from multiple static isomers
at 25 °C, broadening and coalescing as the temperature was increased
to 85 °C in 20 °C increments, suggesting fluxional isomerization
between multiple isomers at elevated temperatures.
Fluxional complexes of radiometals may ultimately
be unfavorable.
The rapid process of metal ion deligation–ligation by ligand
donor arms, which can occur as the fluxional complex constantly rearranges
and shifts between isomers in solution, could open temporary holes
in the coordination sphere. When these holes in the coordination sphere
occur in vivo, they may provide competing ligands such as aqua, phosphate,
and serum proteins (e.g., transferrin or serum albumin) opportunities
to bind to the metal ion and facilitate transchelation/demetalation.
Furthermore, the presence of a well-defined, single static metal–ligand
isomer (as in [In(octapa)]−) of an acyclic metal
complex would be ideal. Even having multiple static isomers of a metal–ligand
complex in solution is not ideal because, due to their different structures,
they may exhibit different physical properties and susceptibilities
to external ligand attack. If this were the case, then these different
static isomers may effectively exhibit different kinetic off-rates
and therefore different in vivo stability and kinetic inertness. It
is important to note that this discussion pertains to isomers formed
during metal coordination by a ligand and not isomers of the ligand
itself. The case of ligand isomerization is demonstrated by CHX-DTPA
(Chart 1), which has 4 isomers (two pairs of
diastereomers, CHX-A′-DTPA, CHX-A″-DTPA, CHX-B′-DTPA,
CHX-B″-DTPA), and it was observed that the single CHX-A″-DTPA
isomer was substantially more stable than the other 3 isomers.[23,24,54,55] The idea of correlating the degree of metal–ligand fluxional
isomerization to in vivo stability is not established. However, the
increased fluxionality and number of isomers observed here for H4C3octapa complexes compared to the analogous H4octapa complexes does appear to correlate to substantially inferior
radiolabeling and serum stability properties for H4C3octapa
(vide infra). This concept may be valid only for acyclic ligands and
therefore may not translate to macrocyclic complexes. For example,
[In(DOTA)]− has been observed to be fluxional at
ambient temperature through 1HNMR experiments but is remarkably
stable and kinetically inert in vivo.[56,57] The rapid
interconversion between isomers of a macrocyclic metal ion complex
like [In(DOTA)]− may not provide the same holes
in the coordination sphere for competing ligands to access the metal
center, as the metal ion remains enshrouded and protected within the
macrocyclic framework during this process, making transchelation slower
and less likely to occur. A thorough investigation comparing the level
of fluxionality of acyclic metal complexes to their radiolabeling
and stability properties would be required, which, to our knowledge,
has not been adequately performed.
Acid Dissociation Constants
The pKa values of ligand donor groups
can effect metal coordination
and radiolabeling properties, and to further study this, we experimentally
determined the acid dissociation constants (pKa) for H4C3octapa by potentiometric titrations.
For example, a lower pKa value for functional
groups such as phenols and carboxylic acids means that those acidic
protons are more easily removed (at lower pH values), which allows
for metal coordination and radiolabeling at lower pH values. A difference
in pKa values can be observed when comparing
H4C3octapa to H4octapa. Table 1 shows the acid dissociation constants for both H4C3octapa and H4octapa, with the HL and H2L
equilibrium quotients corresponding to the protonation of backbone
nitrogen atoms (1,3-propylenediamine and 1,2-ethylenediamine, respectively).
Although chemically identical, these two nitrogen atoms have significantly
different pKa values, as a result of charge
repulsion when the two nitrogen atoms are protonated in the H2L species. For H4octapa, the difference in pKa is 3 units between these two protonated nitrogen
atoms, with the deprotonation of the first backbone-ennitrogen occurring
with a pKa of 5.59(6) (H2L),
and the second, with a pKa of 8.59(4)
(HL). The explanation for this is that the two positively charged
protonated nitrogen atoms (H2L) of the 1,2-ethylenediamine
backbone of H4octapa are held physically close together
through the ethylene bridge, creating substantial charge repulsion
between them, therefore making the first deprotonation event from
H2L to HL more favorable and more acidic with a lower pKa (5.59(6) vs 8.59(4)). The effect of extending
the two-carbon ethylene (en) bridge of H4octapa to a three-carbonpropylene (pn) bridge for the H4C3octapa derivative is
that this charge repulsion is decreased, therefore reducing the thermodynamic
driving force for removal of the first H2L proton. The
effect of this is that the pKa for removal
of the first H2L1,3-propylenediaminenitrogen proton in
H4C3octapa is over 1 unit higher (more basic) than that
for H4octapa, with values of 6.95(6) for H2L
and 8.86(3) for HL (Table 1). Table 1 also contains pKa values
for the picolinic acid groups (H4L and H3L)
and the carboxylic acid arms (H6L and H5L).
The pKa value for the first deprotonation
event of the H6Lcarboxylic acid could not be determined,
as the value was below the threshold of the electrode (pH < 2).
Table 1
Acid Dissociation Constants (pKa), Formation Constants (log KML), and pMa Values for In3+ and
Lu3+ Complexes of H4octapa and
H4C3octapa
log K
equilibrium
quotient
H4C3octapa, C3octapa4– = L
H4octapa, octapa4– = L
[H6L]/[H5L][H]
N/D
N/D
[H5L]/[H4L][H]
2.0(1)
2.79(4)
[H4L]/[H3L][H]
2.65(8)
2.77(4)
[H3L]/[H2L][H]
3.54(7)
3.77(2)
[H2L]/[HL][H]
6.95(6)
5.59(6)
[HL]/[H][L]
8.86(3)
8.59(4)
[InL]/[In][L]
24.6(3)
26.76(14)
[InHL]/[InL][H]
N/D
2.89(23)
pMa
24.0
26.5
[LuL]/[Lu][L]
18.8(3)
20.08(9)
pMa
18.1
19.8
Calculated for 10 μM total
ligand and 1 μM total metal at pH 7.4 and 25 °C.
Calculated for 10 μM total
ligand and 1 μM total metal at pH 7.4 and 25 °C.
Thermodynamic Stability
In order
to indirectly test
the hypothesis that decreased fluxional isomerization may correlate
to increased stability (with acyclic ligands), we have evaluated the
thermodynamic stability of the In3+ and Lu3+ metal complexes of H4C3octapa using potentiometric titrations
to determine their formation constants. The molecular formula of H4C3octapa was determined by elemental analysis (H4C3octapa·4HCl·2H2O), and thermodynamic formation
constants for H4octapa were previously studied.[14,28] The thermodynamic stability constants (log KML) for H4C3octapa with In3+ and Lu3+ were determined by competitive potentiometric titration
experiments with EDTA, under the same experimental conditions as those
previously used for H4octapa.[14,28] The thermodynamic stability constants (log KML) for In3+ and Lu3+ with H4C3octapa were experimentally determined to be 24.6 ± 0.3 (pM
= 24.0) and 18.8 ± 0.3 (pM = 18.1), respectively (Table 2). The thermodynamic stability constants of the
In3+ complexes of H4octapa, DOTA, and DTPA have
been previously determined to be 26.6 (pM = 26.5), 23.9 (pM = 18.8),
and 29.0 (pM = 25.7), respectively (Table 2).[14] The thermodynamic stability constants
of the Lu3+ complexes of H4octapa, DOTA, and
DTPA have been previously determined to be 20.08 (pM = 19.8), 21.6–29.2
(pM = 17.1), and 22.6 (pM = 19.1), respectively (Table 2).[28] The reason for the large variance
in DOTA-Lu3+ stability constants is that many different
methods have been used due to the very slow kinetics of DOTA for forming
metal ion complexes.[28,58−60] The pM value
is a condition-dependent value that is calculated from the standard
thermodynamic stability constant (log KML), accounting for variables and conditions such as ligand basicity
(pKa), metal ion hydrolysis, pH (physiological,
7.4), and a set ligand/metal ratio (10:1). The pM (= −log[M])
is essentially the metal scavenging ability of the ligand under biologically
relevant conditions; the higher the pM value, the lower the concentration
of free metal ion.[61−64] These experimentally determined values show that H4C3octapa
forms less thermodynamically stable complexes with In3+ and Lu3+ than does the ligand H4octapa. The
log KML and pM values were determined
to be ∼2 units lower (∼2 orders of magnitude difference
in thermodynamic stability) for H4C3octapa than for H4octapa with both In3+ and Lu3+. These
lower log KML and pM values are consistent
with H4C3octapa being a worse fit than H4octapa
for these large metal ions, with DFT calculations predicting lower
symmetry and less ideal bond lengths/angles (see DFT section), and
NMR studies showing higher levels of isomerization and fluxional interconversion.
The lower thermodynamic stability of H4C3octapa with these
metal ions also correlates to inferior radiolabeling and serum stability
properties with 111In and 177Lu (vide infra).
Table 2
Formation Constants (log KML) and pMa Values for In3+, Lu3+, and Y3+ Complexes of Relevant
Ligands
ligand
metal ion
log KML
pMa
ref
dedpa2–
In3+
26.60(4)
25.9
(14)
octapa4–
In3+
26.8(1)
26.5
(14)
Lu3+
20.08(9)
19.8
(28)
C3octapa4–
In3+
24.6(3)
24.0
Lu3+
18.8(3)
18.1
DTPA4–
In3+
29.0
25.7
(65 and 66)
Lu3+
22.6
19.1
(28 and 66)
DOTA4–
In3+
23.9(1)
18.8
(61 and 66)
Lu3+
21.6(1), 23.6, 25, 29.2
17.1
(28 and 58−60)
transferrin
In3+
18.3
18.7
(67)
Lu3+
11.08
(68)
Calculated for 10 μM total
ligand and 1 μM total metal at pH 7.4 and 25 °C.
Calculated for 10 μM total
ligand and 1 μM total metal at pH 7.4 and 25 °C.Although thermodynamic stability
is generally regarded as an unreliable
predictor of in vivo stability on its own[69] and the kinetic inertness of a radiometal complex is a much more
relevant factor, we hope to observe trends by comparing NMR, DFT,
thermodynamic, and in vitro/in vivo stability data. A common example
of this is the case of [111In(DOTA)]−, where it is widely accepted as being significantly more stable
than [111In(DTPA)]2– both in vivo and
in vitro; however, the thermodynamic stability constant for DTPA with
In3+ is ∼5 units higher (∼5 orders of magnitude
difference in thermodynamic stability) than that for DOTA (29.0 and
23.9, respectively) (Table 1). Comparing the
thermodynamic stability and the in vivo stability of DTPA and DOTA
like this may appear to be a compelling example of why thermodynamic
stability values are not reliable predictors of in vivo stability;
however, as mentioned above, it may be that comparing acyclic ligands
to macrocyclic ligands is not possible by these methods, and perhaps
trends within each class of ligand, acyclic and macrocyclic, may emerge
in the future as more data are collected. It appears that in this
work the lower thermodynamic stability of H4C3octapa (∼2
orders of magnitude) compared to that of H4octapa does
indeed correlate to lower in vitro stability in radiolabeling experiments
(vide infra) and presumably lower in vivo stability.
Density Functional
Theory Structure Calculations
Growing
crystals of the metal complexes discussed herein for solid-state X-ray
analysis proved to be difficult, so density functional theory (DFT)
calculations were performed. The coordination geometries of 8-coordinate
[In(C3octapa)]− and [Lu(C3octapa)]− were calculated in silico using DFT methods to compare their geometries,
polar surface areas, and bond lengths/angles (Figure 8). The DFT structures of the 8-coordinate [In(octapa)]− and [Lu(octapa)]− complexes were
calculated previously, and MEP polar surface area maps were superimposed
onto the structures.[14,28] A visual inspection of In3+ and Lu3+ complexes of both ligands suggests they
are very similar to one another, with similar charge distributions
determined by the overlaid molecular electrostatic potential (MEP)
polar surface area maps (Figure 8).[14,28]
Figure 8
In
silico DFT structure predictions: (a) 8-coordinate structure
of [In(C3octapa)]− (top) from two perspectives;
(b) 8-coordinate structure of [Lu(C3octapa)]− (bottom)
from two perspectives, with both structures showing overlaid MEP polar-surface
area maps predicting the charge distribution over the solvent-exposed
surface of the metal complexes (red = negative, blue = positive, representing
a maximum potential of 0.254 au and a minimum of −0.254 au,
mapped onto electron density isosurfaces of 0.002 Å–3). Performed using the B3LYP functional employing the 6-31+G(d,p)
basis set for first- and second-row elements, and the Stuttgart/Dresden
and associated ECP’s basis set was employed for the metals,
lutetium and indium.[40,41] Solvent (water) effects were
described through a continuum approach by means of the IEF PCM as
implemented in Gaussian 09.
In
silico DFT structure predictions: (a) 8-coordinate structure
of [In(C3octapa)]− (top) from two perspectives;
(b) 8-coordinate structure of [Lu(C3octapa)]− (bottom)
from two perspectives, with both structures showing overlaid MEP polar-surface
area maps predicting the charge distribution over the solvent-exposed
surface of the metal complexes (red = negative, blue = positive, representing
a maximum potential of 0.254 au and a minimum of −0.254 au,
mapped onto electron density isosurfaces of 0.002 Å–3). Performed using the B3LYP functional employing the 6-31+G(d,p)
basis set for first- and second-row elements, and the Stuttgart/Dresden
and associated ECP’s basis set was employed for the metals,
lutetium and indium.[40,41] Solvent (water) effects were
described through a continuum approach by means of the IEF PCM as
implemented in Gaussian 09.Upon qualitative visual comparison, the structures and charge
distributions
are very similar between these [C3octapa]− complexes
and the [In(octapa)]− and [Lu(octapa)]− DFT structures, but a deeper analysis of bond lengths and angles
reveals a more interesting comparison.[14,28] It has been
proposed by Hancock that the ideal L–M–L bond angle
and ideal M–L bond length for a 6-membered L–M–L
chelate ring (e.g., 1,3-propylenediamine) is 109.5° and 1.6 Å,
respectively, and for a 5-membered chelate ring (e.g., 1,2-ethylenediamine),
69° and 2.5 Å, respectively, with 5-membered chelate rings
being ideal for larger metal ions (e.g., In3+ and Lu3+) and 6-membered chelate rings being ideal for a small metal
(the approximate size of a sp3-hybridized carbon).[11] These predictions are not concrete but will
be compared to the DFT calculated values for H4octapa and
H4C3octapametal complexes. A selection of bond lengths
and angles from DFT calculated ML complexes are shown in Table 3.The bond lengths for
the H4C3octapa complexes diverge
from the optimum M–L values proposed by Hancock of 1.6 Å
for a 6-membered chelate ring, with bond lengths calculated of 2.494/2.486
Å for the In3+ complex and 2.593/2.576 Å for
the Lu3+ complex (Table 3). The
1,2-ethylenediamine M–L bond lengths of the H4octapa
complexes were close to the reported ideal lengths of 2.5 Å,
with values being calculated to be 2.538/2.538 Å for the In3+ complex and 2.756/2.756 Å for the Lu3+ complex
(Table 3). For the H4C3octapa complexes,
the numbers obtained for the 1,3-propylenediamine L–M–L
bond angles deviated more significantly from the proposed optimal
L–M–L angle of 109.5°, with calculated values of
93.0 and 90.8° for In3+ and Lu3+, respectively
(Table 3). For the H4octapa complexes,
on the other hand, the ethyleneN1–M–N2 bond angle was
calculated to be closer to the proposed ideal angle of 69°, with
values of 74.8 and 67.8° for In3+ and Lu3+, respectively. An inspection of the bond lengths presented in Table 3 highlights the very symmetric nature of M–L
bonds in the H4octapa complexes (∼C2 symmetry) and the comparatively asymmetric bonds for
the H4C3octapa complexes, which supports the NMR data (vide
supra). The NMR spectra of the H4C3octapa complexes revealed
lower symmetry and more fluxional species than the H4octapa
complexes, mirroring the data from DFT-calculated structures (Table 3). These results suggest that the H4octapa
complexes should be more stable and inert than the analogous H4C3octapa complexes. Although it is interesting to compare
these DFT-calculated bond lengths and angles to the ideal values proposed
by Hancock, evaluating the stability of the radiometal complexes is
perhaps more relevant.
Trastuzumab Antibody Conjugation, 111In/177Lu Radiolabeling, and In Vitro Serum Stability
To gain further
insight into their differences in stability (thermodynamic and kinetic),
we conjugated the bifunctional ligand derivatives p-SCN-Bn-H4C3octapa and p-SCN-Bn-H4octapa to the antibody trastuzumab, radiolabeled the resulting
ligand–antibody immunoconjugate with 111In and 177Lu, and evaluated their stability via their inertness to
transchelation by serum proteins over a period of 5 days (vide infra).
As discussed in the introduction, the boundary
between large and small metal ions for the most stable fit between
5- and 6-membered chelate rings is unclear. These radiolabeling and
serum stability studies aim to make this boundary clear for the radiometals 111In and 177Lu. NMR studies, thermodynamic formation
constants, and DFT structure analysis have suggested that H4octapa is a superior ligand to H4C3octapa for the large
metal ions In3+ and Lu3+, and that the addition
of a single carbon to the structure of H4C3octapa was detrimental.
Ultimately, the stability and inertness of these metal ion complexes
must be evaluated in vitro and in vivo. Radiometric isotopic dilution
assays for determining the number of chelates bound to each antibody
were not performed, as previous experiments with p-SCN-Bn-H4octapa and p-SCN-Bn-DOTA found
that the addition of 4 equiv. of isothiocyanate-ligand to trastuzumab
would reliably conjugate ∼3 chelates per antibody.[28] Further to this point, the purpose of this study
is to determine the difference in stability between ligands, and a
difference of a few chelates per antibody would have a negligible,
if any, effect on serum stability and dissociation rates. For studies
trying to determine the maximum specific activity achievable for each
ligand–antibody conjugate, however, the number of chelates
per antibody would be a crucial piece of data.To evaluate and
compare the stability of these ligands in biologically relevant media,
the nonbifunctional ligands H4octapa and H4C3octapa,
along with gold standard ligands DOTA and DTPA were radiolabeled with 177Lu and incubated in human blood serum as a transchelation
challenge. The amount of 177Lu transchelated from ligands
to serum proteins was measured at 1.5 and 24 h using PD10 size-exclusion
columns. The results summarized in Table 2 reveal
that the stability of H4octapa, H4C3octapa,
and DOTA in human serum was nearly identical during a 24 h period,
with DTPA having slightly lower stability. These results are observed
to diverge from the serum stability values obtained for the same ligands
but as bifunctional derivatives conjugated to the antibody trastuzumab
(vide infra).Although the results summarized in Table 4 suggest that the 177Lu complexes of
H4C3octapa
and H4octapa possess identical stability against serum
transchelation, a much more relevant experiment is to evaluate their
stability when attached to a targeting vector like trastuzumab. The
practical difficulty with this approach is that bifunctional ligands
are typically much more difficult to synthesize, so for early screening
of new ligands, the nonbifunctional derivatives are often more expedient
to use. Despite convenience, a more biologically relevant model system
was evaluated by conjugating the bifunctional ligands to the HER2/neu-targeting antibody trastuzumab. In order to provide
a basis for comparison, immunoconjugates were synthesized bearing
both p-SCN-Bn-H4C3octapa and p-SCN-Bn-H4octapa bifunctional chelates, they were radiolabeled,
and serum stability experiments were performed. To begin, trastuzumab
was purified by centrifuge filtration to remove additives present
in the antibody kit (e.g., l-histidine HCl), mixed under
basic conditions (pH 8.5–9.0) with 4 equiv of p-SCN-Bn-H4C3octapa or p-SCN-Bn-H4octapa to react, and finally purified via size-exclusion chromatography
(PD-10, GE Healthcare, UK). H4C3octapa–trastuzumab
and H4octapa–trastuzumab were then radiolabeled
with either 111In or 177Lu in NH4OAc buffer (pH 5.5, 200 mM) for 60 min at room temperature to ensure
maximum radiolabel incorporation. H4octapa–trastuzumab
has been previously determined to radiolabel quantitatively with 111In and 177Lu in 15 min at ambient temperature;[28] results that were reproduced here with >95%
radiochemical yields for both 111In and 177Lu
with high radiochemical purity (>99% in each case) and specific
activity
(∼2.2 and ∼2.5 mCi/mg, respectively, iTLC traces in Figures S21 and S22). H4C3octapa–trastuzumab
radiolabeled with 111In and 177Lu in radiochemical
yields and specific activities of ∼95% (∼2.3 mCi/mg)
and ∼8% (∼0.3 mCi/mg), respectively (iTLC traces in Figures S23 and S24). It was observed that a
substantial amount of 177Lu was eluted by the EDTA-containing
iTLC mobile phase upon analysis of radiolabeling yields for 177Lu-C3octapa–trastuzumab, suggesting facile transchelation
of 177Lu by EDTA and therefore very low stability of the
H4C3octapa complex.
Table 4
Human Serum Stability
Challengea
complex
1.5 h stability (%)
24 h stability (%)
[177Lu(C3octapa)]−
90.3 ± 1.8
86.2 ± 1.0
[177Lu(octapa)]−
88.1 ± 1.2
87.7 ± 0.7
[177Lu(DOTA)]−
87.7 ± 0.7
87.4 ± 2.1
[177Lu(DTPA)]2–
77.4 ± 1.2
81.6 ± 2.3
Performed at 37.5 °C (n = 3), with stability
shown as the percent intact 177Lu complex, determined by
PD10 size-exclusion column elution.
Performed at 37.5 °C (n = 3), with stability
shown as the percent intact n class="Chemical">177Lu complex, determined by
PD10 size-exclusion column elution.
In order to assay the stability of these radioimmunoconjugates
under biologically relevant conditions, all four radiolabeled constructs
were incubated in human serum and PBS (control) at 37 °C for
a period of 5 days (Figures 9 and 10). At the end of 5 days, the stability of the 111In-C3octapa– and 111In-octapa–trastuzumab
immunoconjugates was determined to be ∼24 and ∼91%,
respectively, revealing a significant decrease in stability of the
radiometal complexes of H4C3octapa compared to that of
H4octapa (Figure 9). The 177Lu-C3octapa– and 177Lu-octapa–trastuzumab
conjugates were found to be ∼4 and ∼89% stable after
5 days, respectively, showing a more drastic decrease in stability
than that observed for the analogous 111In complexes (Figure 10). Although it was known that 6-membered ligand–metal
chelate rings (as with H4C3octapa) are less favorable than
5-membered chelate rings for large metal ions (as with H4octapa),[11] these serum stability results
prove this point more dramatically than expected.
Figure 9
Stability of the immunoconjugates 111In(octapa)–trastuzumab
and 111In(C3octapa)–trastuzumab in both phosphate
buffered saline (PBS) and human blood serum, evaluated by spotting
∼1 μCi of serum competition mixture on silica-embedded
paper iTLC strips and eluting with an aqueous EDTA (50 mM, pH 5) mobile
phase.
Figure 10
Stability of the immunoconjugates 177Lu(octapa)–trastuzumab
and 177Lu(C3octapa)–trastuzumab in both phosphate
buffered saline (PBS) and human blood serum, evaluated by spotting
∼1 μCi of serum competition mixture on silica-embedded
paper iTLC strips and eluting with an aqueous EDTA (50 mM, pH 5) mobile
phase.
Stability of the immunoconjugates 111In(octapa)–trastuzumab
and 111In(C3octapa)–trastuzumab in both phosphate
buffered saline (PBS) and human blood serum, evaluated by spotting
∼1 μCi of serum competition mixture on silica-embedded
paper iTLC strips and eluting with an aqueous EDTA (50 mM, pH 5) mobile
phase.Stability of the immunoconjugates 177Lu(octapa)–n class="Chemical">trastuzumab
and 177Lu(C3octapa)–trastuzumab in both phosphate
buffered saline (PBS) and human blood serum, evaluated by spotting
∼1 μCi of serum competition mixture on silica-embedded
paper iTLC strips and eluting with an aqueous EDTA (50 mM, pH 5) mobile
phase.
During elution of the iTLC strips,
the ligand EDTA is present in
great excess (50 mM) over the small amount of radioimmunoconjugate
spotted onto the iTLC strips (0.5–1 μL of serum competition
mixture) and therefore transchelation during elution is possible (Figures S21–S24). The four serum stability
challenges were therefore run in both human serum and phosphate buffered
saline (PBS) as a control to determine if transchelation and radiometal
leaching was occurring as a result of serum proteins or because of
the excess of EDTA present in the iTLC mobile phase (Figures 9 and 10). For the 111In/177Lu-octapa–trastuzumab samples, there was
<1% difference in stability between serum and PBS solutions. The
H4C3octapa samples demonstrated more interesting results,
with the stability of 111In-C3octapa–trastuzumab
after 120 h in serum being ∼24%, but this was much higher,
at ∼85%, in PBS (Figure 9). This result
reveals that in a solution of PBS the 111In-C3octapa–trastuzumab
immunoconjugate remains stable and that very little 111In is transchelated by EDTA during iTLC elution, but when placed
in a solution of blood serum, a substantial amount of 111In was transchelated by serum proteins.The stability of 177Lu-C3octapa–trastuzumab after
5 days in PBS and serum was ∼8 and ∼4%, respectively,
suggesting that the [177Lu(C3octapa)]− complex is very unstable, being easily transchelated by PBS and/or
the EDTA present in the iTLC mobile phase (Figure 10). One of the reasons for the poor radiolabeling yield and
specific activity of H4C3octapa–trastuzumab with 177Lu was that the complex formed was very weak, with initial
iTLC analyses after radiolabeling showing substantial leaching of 177Lu from the immunoconjugate to the EDTA mobile phase (Figures S21–S24).The difference
of a single carbon in the backbone of this ligand
scaffold clearly results in a substantial decrease in stability with
these radiometals. The denticity and binding groups were not changed
betweenH4octapa and H4C3octapa, demonstrating
the crucial importance of chelate ring size and ligand–metal
bite size/angle, reinforcing the importance of carefully matching
bifunctional chelators with radiometals for imaging and therapeutic
applications.[8] These results also suggest
that evaluating the stability of nonbifunctional ligands with serum
transchelation experiments, using PD10 size-exclusion elution as a
method of analysis, may not be a reliable method. It was found that
the nonbifunctional [177Lu(C3octapa)]− complex was very stable to serum transchelation by PD10 size-exclusion
analysis but was very unstable as the bifunctional trastuzumab conjugate
by iTLC analysis. This mirrors previous results where the radiometal
complex [111In(decapa)]2– was found to
be stable by PD10 size-exclusion analysis, but in vivo biodistribution
results suggested that it was not stable. These results suggest that
in vitro serum stability analysis of nonbifunctional ligands should
be investigated with a more rigorous technique like size-exclusion
HPLC. It may also be the case that only experiments with bifunctional
ligand derivatives should be undertaken, as the nonbifunctional ligands
will generally not be used for radiopharmaceutical applications and
only bifunctional chelate conjugates are of utility. The difficulty
of synthesizing most bifunctional ligands could, however, make this
approach time consuming and inefficient.
Conclusions
The
ligands H4C3octapa and p-SCN-Bn-H4C3octapa were synthesized for the first time, using nosyl
protection chemistry, for comparison to the previously studied ligands
H4octapa and p-SCN-Bn-H4octapa.
The purpose of this work was to determine whether addition of a single
carbon atom to the backbone of these ligand scaffolds would effect
metal/radiometal chelation and stability. The In3+ and
Lu3+ complexes of H4C3octapa were synthesized,
studied by NMR spectroscopy, DFT structure analysis, and potentiometric
titrations, and compared to the analogous H4octapa complexes.
It was found that the 1HNMR spectra of [In(C3octapa)]− and [Lu(C3octapa)]− were substantially
different from the analogous H4octapa complexes, with fluxional
isomerization and a higher number of isomers being observed by VT-NMR
and 2D COSY/HSQC-NMR experiments. Evaluation of DFT structures revealed
very symmetric [In(octapa)]− and [Lu(octapa)]− complexes, with a 5-member chelate ring being formed
between the ,′-ethylene backbone and the metal centers yielding
very symmetric structures. In contrast, the [In(C3octapa)]− and [Lu(C3octapa)]− complexes were much less symmetric,
with a 6-membered chelate ring being formed between the ,′-propylene
backbone and the metal centers, supporting the NMR spectral analysis
that these complexes are less symmetric and rigid than the analogous
H4octapa complexes. Potentiometric titrations were performed
to determine the formation constants (log KML, pM), which found the thermodynamic stability to be ∼2 orders
of magnitude lower for the In3+ and Lu3+ complexes
of H4C3octapa when compared to that of the analogous complexes
of the H4octapa ligand. The bifunctional ligands p-SCN-Bn-H4C3octapa and p-SCN-Bn-H4octapa were conjugated to the HER2/neu targeting
antibody trastuzumab and radiolabeled with 111In and 177Lu, and their radiochemical yields and serum stability were
directly compared. It was observed that the H4C3octapa–trastuzumab
conjugates displayed inferior radiochemistry properties to that of
H4octapa–trastuzumab, with radiochemical yields
and serum stability being substantially worse. Over a 5 day stability
challenge experiment in blood serum, 111In-octapa–
and 111In-C3octapa–trastuzumab immunoconjugates
were determined to be ∼91 and ∼24% stable, respectively,
and 177Lu-octapa– and 177Lu-C3octapa–trastuzumab,
∼89 and ∼4% stable, respectively. Although only a single
carbon atom was added to H4C3octapa and the metaldonor
atoms and denticity were not changed, the solution chemistry and radiochemistry
properties were drastically altered, highlighting the importance of
careful ligand design and radiometal–ligand matching when selecting
a bifunctional chelator to conjugate to targeting vectors for radiopharmaceutical
applications. The exact boundary between what is considered to be
small and large metal ion radii for 5/6-membered chelate ring preference
is still unclear. This work further suggests that large metal ions
such as In3+ and Lu3+ form the most stable complexes
when chelated by 5-membered chelate rings.
Authors: Eszter Boros; Cara L Ferreira; Jacqueline F Cawthray; Eric W Price; Brian O Patrick; Dennis W Wester; Michael J Adam; Chris Orvig Journal: J Am Chem Soc Date: 2010-11-10 Impact factor: 15.419
Authors: Eric W Price; Jacqueline F Cawthray; Gwendolyn A Bailey; Cara L Ferreira; Eszter Boros; Michael J Adam; Chris Orvig Journal: J Am Chem Soc Date: 2012-05-11 Impact factor: 15.419
Authors: Jason P Holland; Vadim Divilov; Neil H Bander; Peter M Smith-Jones; Steven M Larson; Jason S Lewis Journal: J Nucl Med Date: 2010-07-21 Impact factor: 10.057
Authors: Raquel Ferreirós-Martínez; David Esteban-Gómez; Carlos Platas-Iglesias; Andrés de Blas; Teresa Rodríguez-Blas Journal: Dalton Trans Date: 2008-09-09 Impact factor: 4.390
Authors: Eric W Price; Kimberly J Edwards; Kathryn E Carnazza; Sean D Carlin; Brian M Zeglis; Michael J Adam; Chris Orvig; Jason S Lewis Journal: Nucl Med Biol Date: 2016-06-28 Impact factor: 2.408
Authors: Elaheh Khozeimeh Sarbisheh; Akam K Salih; Shvan J Raheem; Jason S Lewis; Eric W Price Journal: Inorg Chem Date: 2020-07-28 Impact factor: 5.165
Authors: Aohan Hu; Ivan Keresztes; Samantha N MacMillan; Yang Yang; Erdong Ding; Warren R Zipfel; Robert A DiStasio; John W Babich; Justin J Wilson Journal: Inorg Chem Date: 2020-03-27 Impact factor: 5.165
Authors: Jia Wang; Alejandra Rios; Ksenia Lisova; Roger Slavik; Arion F Chatziioannou; R Michael van Dam Journal: Nucl Med Biol Date: 2019-12-17 Impact factor: 2.408
Authors: Fátima Lucio-Martínez; Zoltán Garda; Balázs Váradi; Ferenc Krisztián Kálmán; David Esteban-Gómez; Éva Tóth; Gyula Tircsó; Carlos Platas-Iglesias Journal: Inorg Chem Date: 2022-03-11 Impact factor: 5.165
Chart 1
Scheme 1
Scheme 2
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10