Mechanistic approach of the difference in hydrolysis rate between the 2- and 4-isomers of no-carrier-added [18F]fluoromethyl-L-phenylalanine

Article abstract of DOI:10.1002/jlcr.1852

No-carrier-added (n.c.a.) 2-[¹⁸F]fluoromethyl-l-phenylalanine (2-[¹⁸F]FMP) was found to be very sensitive to hydrolysis in aqueous solutions. In this paper, the defluorination reaction was studied in detail to elucidate its mechanism

Full text of DOI:10.1002/jlcr.1852

Research Article
Received 29 October 2010,
Accepted 5 November 2010
Published online 30 December 2010 in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/jlcr.1852
Mechanistic approach of the difference
in hydrolysis rate between the 2- and
4-isomers of no-carrier-added
[18F]fluoromethyl-L-phenylalanine
Ã
Ken Kersemans,a John Mertens,a Frank De Proft,b and Paul Geerlingsb
No-carrier-added (n.c.a.) 2-[18F]fluoromethyl-l-phenylalanine (2-[18F]FMP) was found to be very sensitive to hydrolysis in
aqueous solutions. In this paper, the defluorination reaction was studied in detail to elucidate its mechanism. Therefore,
besides 2-[18F]FMP and 4-[18F]FMP, 2-[18F]fluoromethyl-phenethylamine (2-[18F]FMPAM) and 4-[18F]FMPAM were
synthesized, both ‘mimetic’ molecules of the decarboxylated amino acid analogues. Radiosynthesis, using a customized
Scintomics automatic synthesis hotboxthree module, resulted in a high overall yield and a radiochemical purity of 499%.
The defluorination rates of all compounds were studied by HPLC. The defluorination rate of 2-[18F]FMLP at 501C was
approximately 300 times faster than that of n.c.a. 4-[18F]FMLP. The defluorination rate of 2-[18F]FMPAM is somewhat lower
than of 2-[18F]FMP but still very high in comparison with 4-[18F]FMPAM, which is virtually stable. It allowed to elucidate the
reaction mechanism ruled by two distinct intramolecular interactions. First, the hydrogen bond interaction between the
amine and the benzylic fluorine weakening the carbon–fluorine bond. Secondly, the formation of a second hydrogen bond
between the carboxyl oxygen atom and one of the benzylic hydrogen atoms rendering the benzyl fluoride group even more
susceptible to hydrolysis.
Keywords: 2 and 4-[18F]fluoromethyl-l-phenylalanine; hydrolysis; different; rates; mechanism; hydrogen bonds
AVANCE DRX 250 instrument and MS data were acquired on
a Fisons VG II Quattro Mass Spectrometer.
Introduction
Our group has previously shown that 2-amino-(S)-3-(2-[18F]fluoro-
methyl-phenyl)-propionic acid (2-[18F]fluoromethyl-L-phenyl-
Synthetic procedures
alanine; 2-[18F]FMLP) showed a high uptake in cancer cells both
in vitro and in vivo. However, in its no-carrier-added (n.c.a.) form it
Fluorinated amino acids
suffered from considerable non-radiolytic radiodefluorination in
water at room temperature.1,2 No-carrier added 2-amino-(S)-3-
(4-[18F]fluoromethyl-phenyl)-propionic acid (4-[18F]fluoromethyl-L-
The synthesis of the precursor molecules for the radiosynthesis
as well as of non-radioactive fluorinated analogues was
phenylalanine; 4-[18F]FMLP) was stable in the same conditions.
described in detail in earlier publications.2,3
This suggested that the fast hydrolysis was due to the ortho-
position of the fluoromethyl group on the aromatic ring and
more specifically to an intra-molecular interaction between the
Synthesis of 2/4-fluoromethyl-phenethylamine
The amine group of commercially available 2/4-methylphe-
nethylamine (Sigma-Aldrich) was protected with N-Boc by
reaction with Boc2O in the presence of triethylamine in
tetrahydrofurane, as described by Leftheris et al.4 After silica
gel column chromatography (50 g Si-gel (Merck), id 2 ꢀ 25 cm,
negatively charged benzyl fluorine atom and the positively
charged NH13 of the amino acid group.2,3 The problem of
radiodefluorination in the radiopharmaceutical formulation was
solved for the larger part by the addition of calcium ions (0.04 M)
ensuring a shelf-life of at least 6 h. This stabilizing effect was
assumed to be related to the formation of a complex between
the Ca21 ions and the negatively charged fluorine and oxygen
atom of the dissociated carboxylic acid.2 In this paper, we study
the mechanism of the fast defluorination of n.c.a. 2-[18F]FMLP.
aRadiopharmaceutical Chemistry, BEFY, Vrije Universiteit Brussel, Laarbeeklaan
103, 1090 Brussels, Belgium
bDepartment of General Chemistry, ALGC, Vrije Universiteit Brussel, Pleinlaan 2,
1050 Brussels, Belgium
Materials and methods
*Correspondence to: Ken Kersemans, Radiopharmaceutical Chemistry, BEFY,
Laarbeeklaan 103, 1090 Brussels, Belgium.
E-mail: kkersema@vub.ac.be
All products were at least of analytical grade. All solvents were
of HPLC quality or better. NMR data were obtained using an
J. Label Compd. Radiopharm 2011, 54 220–223
Copyright r 2010 John Wiley & Sons, Ltd.

K. Kersemans et al.
petroleum ether/diethyl ether: 2/8 (v/v)) (2-o-Tolyl-ethyl)-carba- (for the new compounds 2-[18F]FMPAM and 4-[18F]FMPAM this
mic acid tert-butyl ester was obtained as a colourless oil was 8.2 min).
1
(88% yield). H NMR (CDCl3, 500 MHz):d 7.16–7.12 (m, 4 H), 2.91
(t, J = 8 Hz, 2 H), 2.51 (t, J = 7.5 Hz, 2 H), 2.34 (s, 3 H) and 1.46 Quality control
(s, 9 H). MS (EI) m/z 236 (MH1). (4-o-Tolyl-ethyl)-carbamic acid
Quality control of the radiofluorinated amino acid analogues
was performed as described earlier.2,3
1
tert-butyl ester was obtained as a colourless oil (88% yield). H
NMR (CDCl3, 500 MHz):d 7.20–7.10 (m, 4 H), 3.33 (m, 2H), 2.73
Quality control of the n.c.a. 2-(2/4-[18F]fluoromethyl-phenyl)-
ethylamine was achieved by HPLC analysis, performed on a
(t, J = 7 Hz, 2H), 2.35 (s, 3H) and 1.43 (s, 9H). MS (EI) m/z 236 (MH1).
For the radical bromination of the methyl side chain, (2 or 4-o-
Polaris C8 125 ꢀ 4 mm, 5 m column (Varian) using an EtOH/
aqueous solution of 1 mM CH3COONH4 and 1 mM NaF: 5/95 (v/v)
tolyl-ethyl)-carbamic acid tert-butyl ester (270 mg, 1.49 mmol)
was reacted with N-bromosuccinimide (245 mg, 1.38 mmol),
of pH 6.5 as mobile phase with a flow rate of 1 mL/min while
using azobisisobutyronitrile (28 mg, 0.17 mmol) in dichloro-
monitoring both radioactivity (NaI(Tl), Harshaw Chemie) and UV
absorption at 254 nm (Shimadzu). The k0 values were 0.6 and 8.3
for [18F]fluoride and 2-(2/4-[18F]fluoromethyl-phenyl)-ethyla-
mine, respectively.
methane (50 mL) at 501C for 2 h. The crude product was purified
via silica gel column chromatography (50 g Si-gel (Merck), id
2 ꢀ 25 cm, petroleum ether/diethyl ether: 1/9 (v/v)) to yield [2-(2-
bromomethyl-phenyl)-ethyl]-carbamic acid tert-butyl ester as a
yellow oil in 48% yield. 1H NMR (CDCl3, 500 MHz):d 7.21–7.15 (m,
4 H), 4.58 (s, 2 H), 2.96 (t, J = 7.5, 2 H), 2.54 (t, J = 7.0 Hz, 2 H) and
Shelf-life study experiments
1.42 (s, 9 H). MS (EI) m/z 314 (M1). [2-(4-bromomethyl-phenyl)-
Quantities of 2.86 GBq of each radiofluorinated compound were
ethyl]-carbamic acid tert-butyl ester was obtained as a yellow oil
in (yield 46%). 1H NMR (CDCl3, 250 MHz):d 7.14–7.01 (m, 4 H),
4.58 (s, 2 H), 3.27 (m, 2 H), 2.76 (t, J = 6.5 Hz, 2 H) and 1.40(s, 9 H).
MS (EI) m/z 314 (M1).
synthesized as described earlier. Immediately after the radio-
synthesis CaCl2 was added to a final concentration of 40 mM in a
volume of 5 mL. Solutions were stored at room temperature.
Follow-up of the stock solution showed that the defluorination
rate was limited to 0.5% per hour.
In the next step, [2-(2 or 4-bromomethyl-phenyl)-ethyl]-
carbamic acid tert-butyl ester (106 mg, 0.339 mmol) was treated
with an excess of AgF (180 mg, 1.42 mmol) in dry acetonitrile
(5 mL) for 3 h at 651C. The reaction product was purified via silica
gel column chromatography (50 g Si-gel (Merck), id 2 ꢀ 25 cm)
using a gradient of ethyl acetate 1–10% (v/v) in petroleum ether
(40–601C). [2-(2-fluoromethyl-phenyl)-ethyl]-carbamic acid tert-
butyl ester was obtained as a white amorphous solid (yield 54%).
1H NMR (CDCl3, 250 MHz):d 7.18–7.16 (m, 4 H), 5.52 (dd, J1 =
10.0 Hz, J2 = 15.0 Hz, J[19F–1H] = 50.0 Hz, 1 H), 5.42 (1 H, dd,
J1 = 10.0 Hz, J2 = 15.0 Hz, J[19F–1H] = 50 Hz), 3.37 (m, 2H), 2.88
(t, J = 7.5 Hz, 2 H) and 1.41 (s, 9 H). MS (EI) m/z 254 (MH1). [2-(4-
fluoromethyl-phenyl)-ethyl]-carbamic acid tert-butyl ester was
obtained as a white amorphous solid (yield 58%). 1H NMR
(CDCl3, 250 MHz): d 7.23–7.21 (m, 4 H), 5.52 (dd, J1 = 10.0 Hz,
J2 = 15.0 Hz, J[19F–1H] = 50.0 Hz, 1 H), 5.42 (1 H, dd, J1 = 10.0 Hz,
J2 = 15.0 Hz, J[19F–1H] = 50 Hz), 3.38 (m, 2 H), 2.86 (t, J = 7.0 Hz,
2H) and 1.40 (s, 9H). MS (EI) m/z 254 (MH1).
From the stock solution 10mL aliquots were added to a 20 mL
vial containing 9.990 mL of water, resulting in a solution contain-
ing 5.72 MBq of radiofluorinated compound and 0.04 mM CaCl2.
At this concentration of CaCl2, the stabilizing effect is negligible,
as was demonstrated by comparison with a sample without CaCl2.
Samples of 10 mL, containing about 5.72kBq at time zero, were
directly taken from this solution and injected on the HPLC system
at regular time points. The HPLC conditions applied were the
same as described in the Quality Control subsection, but using
only the radioactivity detector. For each series a t = 0 analysis was
performed and the results corrected for the free 18Fꢁ if present.
The results were expressed as a fraction of the activity at t = 0,
using the surface under the peaks of 18Fꢁ and the radio-
fluorinated compound. Collection of the radioactive peak of a
calibrated 18Fꢁ solution showed that in presence of 1 mM NaF,
the amount of 18Fꢁ absorbed in the HPLC system was negligible.
The final step was the deprotection of [2-(2-fluoromethyl-
phenyl)-ethyl]-carbamic acid tert-butyl ester (137 mg, 0.541 mmol)
in 20 mL dichloromethane/trifluoroactic acid: 1/1 (v/v) (30 min at
room temperature). After the final reaction, solvents were removed
by rotatory evaporation and 2-(2-fluoromethyl-phenyl)-ethylamine
was obtained as a pale yellow oil (82%).1H NMR (CDCl3,
250 MHz):d 7.10–6.99 (m, 4 H), 5.49 (s, J[19F–1H] = 50.0 Hz, 1 H),
5.39 (s, J[19F–1H] = 50 Hz, 1 H), 3.16 (t, J= 8 Hz, 2 H), 2.73 (t, J=8.0Hz,
2 H), 1.20 (s, 2 H). MS (EI) m/z 154 (MH1). 2-(4-Fluoromethyl-phenyl)-
ethylamine was obtained as a pale yellow oil (84%). 1H NMR (CDCl3,
250 MHz):d 7.20–7.01 (m, 4 H), 5.50 (s, J[19F–1H] = 47.0 Hz, 1 H), 5.37
(s, J[19F–1H] = 47 Hz, 1 H), 3.16 (t, J= 8 Hz, 2 H), 2.73 (t, J=8.0Hz, 2
H), 1.22 (s, 2 H). MS (EI) m/z 154 (MH1).
Results and discussion
Radiosynthesis
No-carrier-added 2-[18F]fluoromethyl-L-phenylalanine (2-[18F]FMLP)
and 4-[18F]fluoromethyl-L-phenylalanine (4-[18F]FMLP) were synthe-
sized as described earlier3 with a 30% yield and a radio-
pharmaceutical purity of at least 99%. The ‘mimetic molecules’
n.c.a. 2-[18F]fluoromethyl-phenethylamine (2-[18F]FMPAM) and
n.c.a. 4-[18F]fluoromethyl-phenethylamine (4-[18F]FMPAM) were
synthesized with a 40% yield and 99% radiopharmaceutical purity
using the same automated modular system and similar synthesis
strategies. All radiopharmaceuticals were stabilized by addition of
CaCl2. The samples used in the stability experiments were diluted
to obtain a final CaCl2 concentration of 40 mM. The defluorination
in these solutions was proven to be the same as in pure water.
Radiosynthesis
All radioactive compounds were synthesized using the synthetic
pathways and the customized modular Scintomics hotboxthree
Hydrolysis study
system (Scintomics, Fu¨rstenfeldbruck, Germany) as described Defluorination of n.c.a. 2-[18F]FMLP and n.c.a. 4-[18F]FMLP in
earlier3 using the appropriate time for recovery of the n.c.a. water at neutral pH at 501C or 801C is shown in Figure 1.
[18F]-labeled fully protected compounds after HPLC separation At equal activities and concentrations (ꢂ5.8 ꢀ 10ꢁ10 M), the
J. Label Compd. Radiopharm 2011, 54 220–223
Copyright r 2010 John Wiley & Sons, Ltd.
www.jlcr.org

K. Kersemans et al.
initial defluorination rate Vt= 0 of 2-[18F]FMLP at 501C was The subsequent nucleophilic hydroxylation is related to the
approximately 300 times faster than that of n.c.a. 4-[18F]FMLP. solvation of [18F]fluoride ions and not to the formation of HF (pKa
At room temperature the defluorination of 4-[18F]FMLP is too slow of 3.2). These findings can only be explained by the earlier
to allow kinetic measurements (o0.5% after 3 h). These results proposed intramolecular interaction followed by ꢁOH substitu-
show that the fluorobenzyl moiety was more activated for tion. This intramolecular interaction must then occur between the
nucleophilic hydroxylation when the benzylic carbon atom was negatively charged benzyl fluorine atom (natural charge 0.45) and
present in the ortho position compared with the para position of a proton of the positively charged ammonium group (the pKa of
the aromatic ring. Remarkably, a difference in the order of the amino group in phenylalanine is 9.5), in which a hydrogen
reaction kinetics between the ortho- and para-substituted bond is formed resulting in an intermediate H2N1–H–Fdꢁ–CBzl
analogues was observed. The defluorination of 4-[18F]FMLP (Bzl= benzyl). This model can explain the zero-order kinetics
followed a pseudo-first-order reaction (dotted line showing ln % whereby the role of the intramolecular hydrogen bond formation
4-[18F]FMLP a.f.o. time), indicating that the concentration of the in the reaction kinetics can be compared with the common
second reactant remained constant during the reaction. This known keto–enol tautomerism in the halogenation of ketones,
reactant could be water ([H2O]= 55M) or the hydroxyl ions, since the typical example of a zero-order reaction. Our hypothesis is
the ratio [OH]/[4-[18F]FMLP] at pH 7.0 amounts to at least 5 ꢀ 102 supported by literature through analogy of the role of the
and [OHꢁ] can assumed to be constant.
positively charged tropaneN atom (R3NH1) in the non-enzymatic
Less evident was the fact that the hydrolysis of 2-[18F]FMLP acid catalyzed hydrolysis of the methyl ester group of cocaine in
showed zero-order kinetics. Therefore, defluorination of water, proven experimentally and by quantum chemical first-
2-[18F]FMLP was studied at room temperature as a function of principle electronic structure calculations.6,7
pH, ranging from pH 3.5 to pH 8.0. The reaction rate constant Kobs
In order to further our hypothesis we also studied the
(% 18Fꢁ/min.) was calculated for each pH value and the results hydrolytic properties of n.c.a 2-[18F]FMPAM as a ‘mimetic’
presented in Figure 2. Within the recorded pH range the compound, since it is in fact decarboxylated 2-fluoromethyl-
hydrolysis rate constant of 2-[18F]FMLP increases exponentially phenylalanine. This means that both the position of the NH31
with the [OHꢁ] concentration, ruling out H2O as the nucleophile. group vis a vis the fluorobenzyl group and the probability of
`
On the other hand, the fact that at pH 3.5 Kobs is approximately 25 AA–NH13 –Fdꢁ–CBzL (benzyl) interaction are comparable to these
times lower than at pH 8.0 eliminates acid hydrolysis by the bulk in 2-[18F]FMP. In water at pH 6.5 n.c.a. 2-[18F]FMPAM showed a
protons as the initial step in the overall hydrolysis reaction. defluorination rate of 8.5 and 24% per hour at room temper-
ature and 371C, respectively (Figure 3). N.c.a. 4-[18F]FMPAM was
stable under these conditions. This proves our hypothesis that a
protonated amine in close proximity of the benzylic fluoro-
methyl group activates defluorination by an intramolecular
interaction.
Moreover, at room temperature and neutral pH, the defluor-
ination rate of 2-[18F]FMPAM, lacking the acid group, is lower than
2-[18F]FMLP (0.14%/min relative to 0.95%/min, respectively). This
finding suggests an additional hydrogen bond involving the acid
group. It let us to assume that besides the AA-NH13 –Fdꢁ-CBzl
internal bridge, a second hydrogen bond type interaction can
occur between the O of the carboxylic group and a proton of the
benzylic carbon atom. According to the Gutman rules8 on
electron donor–acceptor complexes, confirmed by ab initio
quantum chemical calculations on H-bonding, this interaction
transfers the negative charge to the F-atom (the spill-over effect),
Figure 1. Defluorination of 2-[18F]FMLP and 4-[18F]FMLP. Y-axis left: percentage of
original compound a.f.o. time at 50 and 801C. Y-axis right: ln (% 4-[18F]FMLP) a.f.o.
time and represented by the dotted line.
5
4.5
4
3.5
y = 1.49E-02e
3
R
= 9.94E-01
2.5
2
1.5
1
0.5
0
3
3.5
4
4.5
5
5.5
6
6.5
7
7.5
8
8.5
9
pH
Figure 2. Defluorination of N.C.A. 2-[18F]FMLP at room temperature at different
pH values.
Figure 3. Interaction of the amino group with the benzylic fluorine in 2-[18F]FMLP.
www.jlcr.org
Copyright r 2010 John Wiley & Sons, Ltd.
J. Label Compd. Radiopharm 2011, 54 220–223

K. Kersemans et al.
increasing its leaving group capacity or nucleofugality.9,10
This mechanism also explains the stabilizing role of Ca21 ions.2
The Oꢁ atom of the dissociated carboxylic acid and the negatively
charged F-atom of the fluorobenzyl entity are involved as ligands
in a bidentate Ca21 complex, preventing the occurrence of
aforementioned interaction.
References
[1] K. Kersemans, M. Bauwens, T. Lahoutte, A. Bossuyt, J. Mertens,
Nucl. Instrum. Methods A 2007, 571, 151.
[2] K. Kersemans, M. Bauwens, J. Mertens, Nucl. Med. Biol. 2008,
35, 425.
[3] K. Kersemans, J. Mertens, V. Caveliers, J. Label. Compd. Radio-
pharm 2009, 53(2), 58–62.
Conclusion
[4] K. Leftheris, G. Ahmed, R. Chan, A. J. Dyckman, Z. Hussain, K. Ho,
J. Jr Hynes, J. Letourneau, W. Li, S. Lin, A. Metzger, K. J. Moriarty,
C. Riviello, Y. Shimshock, J. Wen, J. Wityak, S. T. Wrobleski, H. Wu,
J. Wu, M. Desai, K. M. Gillooly, T. H. Lin, D. Loo, K. W. McIntyre,
S. Pitt, D. Ren Shen, D. J. Shuster, R. Zhang, D. Diller, A. Doweyko,
J. Sack, J. Baldwin, J. Barrish, J. Dodd, I. Henderson, S. Kanner,
G. L. Schieven, M. Webb, J. Med. Chem. 2004, 47, 6283.
[5] The geometry of 2-FMLP was optimised and the atomic charges
were subsequently calculated using the natural population
analysis within the B3LYP/6-31G(d) level of theory, using the
Gaussian 03 software package.
The considerable difference in reaction rate for the hydrolysis of
n.c.a. 2- and 4-[18F]FMP was studied. The study of the reaction
kinetics lead to the elucidation of the reaction mechanism,
which is governed by two distinct intramolecular interactions.
First, the hydrogen bond interaction between the amine and the
benzylic fluorine weakens the carbon–fluorine bond. Secondly,
the formation of an additional hydrogen bond between the
carboxyl negatively charged oxygen atom and one of the
benzylic hydrogen atoms renders the fluorine atom even more
susceptible to hydrolysis.
[6] P. Li, K. Zhao, S. Deng, D. W. Landry, Helv. Chim. Acta. 1999,
82, 85.
[7] C. G. Zhan, S. X. Deng, J. G. Skiba, B. A. Hayes, S. M. Tschampel,
G. C. Shields, D. W. Landry, J. Comput. Chem. 2005, 26, 980.
[8] V. Gutman, The Donor– Acceptor Approach to Molecular Inter-
actions, Plenum Press, New York, 1978, Chapter 3.
[9] P. W. Ayers, J. S. M. Anderson, J. L. Rodriguez, Z. Jawed, Phys.
Chem. Chem. Phys. 2005, 7, 1918.
Acknowledgements
K. Kersemans wishes to thank the Research Unit High Resolution
NMR and the department of Organic Chemistry of the Vrije
Universiteit Brussel for performing the 1H NMR and ESI-MS analysis.
[10] L. Broeckaert, J. Moens, G. Roos, F. De Proft, P. Geerlings, J. Phys.
Chem. A 2008, 112, 12164.
J. Label Compd. Radiopharm 2011, 54 220–223
Copyright r 2010 John Wiley & Sons, Ltd.
www.jlcr.org