A gadolinium chelate for detection of β-glucuronidase: A self-immolative approach

Article abstract of DOI:10.1021/ja042162r

New classes of physiologically responsive magnetic resonance (MR) contrast agents are being developed that are activated by enzymes, secondary messengers, pH, and temperature. To this end, we have prepared a new class of enzyme-activated MR contrast agent

Full text of DOI:10.1021/ja042162r

Published on Web 08/23/2005
A Gadolinium Chelate for Detection of â-Glucuronidase: A
Self-Immolative Approach
Joseph A. Duimstra,1 Frank J. Femia,2 and Thomas J. Meade*
Contribution from the Departments of Chemistry, Biochemistry and Molecular and Cell Biology,
Neurobiology and Physiology, and Radiology, Northwestern UniVersity, EVanston, Illinois 60201
Received December 29, 2004; E-mail: tmeade@northwestern.edu
Abstract: New classes of physiologically responsive magnetic resonance (MR) contrast agents are being
developed that are activated by enzymes, secondary messengers, pH, and temperature. To this end, we
have prepared a new class of enzyme-activated MR contrast agents using a self-immolative mechanism
and investigated the properties of these agents using novel in vitro assays. We have synthesized in nine
steps a Gd(III) agent 1 that is activated by the oncologically significant â-glucuronidase. 1 consists of
Gd(III)DO3A (DO3A ) 1,4,7-tricarboxymethylene-1,4,7,10-tetraazacyclododecane) bearing a pendant
â-glucuronic acid moiety connected by a self-immolative linker to the macrocycle. LC-MS analysis reveals
that 1 is enzymatically processed as predicted by bovine liver â-glucuronidase, generating 2-aminoethyl-
GdDO3A, 2. Compound 2 was prepared independently in a four-step synthetic procedure. Complex 1
displays a decrease in relaxivity upon titration with bicarbonate anion. The relaxivity increases when 1 is
converted to 2 in a buffer mimicking in vivo anion concentrations (Parker, D. In Crown Compounds: Towards
Future Applications; Cooper, S. R., Ed.; VCH: New York, 1992; pp 51-67) by 17%, while the relaxivity
decreases by 27% for the same experiment in human blood serum. Hydrolytic kinetics catalyzed by bovine
liver â-glucuronidase at interstitial pH ) 7.4 fit the Michaelis-Menten model with kcat/Km ) 74.9 ( 10.9
M-1 s-1. Monitoring of bulk water proton T1 during incubation with enzyme shows an increase in T1 that
mirrors results obtained through the relaxivity measurements of compounds 1 and 2.
Introduction
The application of â-glucuronide prodrugs in prodrug mono-
therapy (PMT) has yielded mixed results.14,15 In PMT, â-glu-
curonic acid is liberated from the relatively nontoxic prodrug
via endogenous extracellular â-glucuronidase, yielding the more
potent chemotherapeutic. The drawbacks to this approach are
that high enzyme levels are found only near necrotic tumor
masses that have low perfusion and hence receive less prodrug,
and that enzyme concentration is variable between individu-
als.9,10 Further complicating matters is the short half-life of
glucuronide-conjugated prodrugs, necessitating fast enzymatic
conversion of the prodrug to its active form.15 Antibody-directed
enzyme prodrug therapy (ADEPT; a two-step approach) intro-
duces exogenous enzyme via an antibody-targeting moiety and,
in principle, should overcome the problems associated with
PMT. ADEPT progress, initially curtailed by host immune
response to the antibody-enzyme conjugate, has shown promise
with the advent of antibodies engineered via phage display16
and human-based fusion proteins.17 Another potential avenue
Cancer is a leading cause of death in developed nations,
prompting extensive research into the myriad aspects of this
heterogeneous disease. The ability to noninvasively detect a
small number of cells at the onset of metastatic growth presents
a major challenge not only to clinicians but also to the
experimental community. Potential exists for cancer detection
through the in vivo, noninvasive visualization of cellular and
biochemical events that has been gaining rapid momentum in
the burgeoning field of molecular imaging.4-8 Imaging con-
structs that function within the chemotherapeutic cancer prodrug
paradigm are particularly attractive given the extensive research
in the latter field.9-13
(1) Division of Chemistry and Chemical Engineering, California Institute of
Technology, Pasadena, CA 91125.
(2) Present address: Molecular Insight Pharmaceuticals Inc., Cambridge, MA
02142.
(3) Deleted in proof.
(4) J. Magn. Reson. Imaging, Special Issue: Molecular Imaging 2002, 4, 333-
483.
(5) Li, K.; Pandit, S.; Guccione, S.; Bednarski, M. Biomed. MicrodeV. 2004,
6, 113-116.
(13) Tietze, L. F.; Feuerstein, T. Aust. J. Chem. 2003, 56, 841-854.
(14) Bosslet, K.; Straub, R.; Blumrich, M.; Czech, J.; Gerken, M.; Sperker, B.;
Kroemer, H. K.; Gesson, J.-P.; Koch, M.; Monneret, C. Cancer Res. 1998,
58, 1195-1201.
(6) J. Cell. Biochem., Supplement: Molecular Imaging 2002, 87, supplement
39, 1-248.
(7) Nordberg, A. Lancet Neurol. 2004, 3, 519-527.
(8) Choy, G.; Choyke, P.; Libutti, S. K. Mol. Imaging 2003, 2, 303-312.
(9) Rooseboom, M.; Commandeur, J. N. M.; Vermeulen, N. P. E. Pharmacol.
ReV. 2004, 56, 53-102.
(10) Brusselbach, S. Methods Mol. Med. 2004, 90, 303-330.
(11) Papot, S.; Tranoy, I.; Tillequin, F.; Florent, J. C.; Gesson, J. P. Curr. Med.
Chem.: Anti-Cancer Agents 2002, 2, 155-185.
(12) Xu, G.; McLeod, H. L. Clin. Cancer Res. 2001, 7, 3314-3324.
(15) Guerquin-Kern, J.-L.; Volk, A.; Chenu, E.; Lougerstay-Madec, R.; Mon-
neret, C.; Florent, J.-C.; Carrez, D.; Croisy, A. NMR Biomed. 2000, 13,
306-310.
(16) Pedley, R. B.; Sharma, S. K.; Hawkins, R. E.; Chester, K. A. Methods
Mol. Med. 2004, 90, 491-514.
(17) Biela, B. H.; Khawli, L. A.; Hu, P.; Epstein, A. L. Can. Biother.
Radiopharm. 2003, 18, 339-353.
9
10.1021/ja042162r CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 12847-12855
12847

A R T I C L E S
Duimstra et al.
Scheme 1
involves gene-directed enzyme prodrug therapy (GDEPT). Here,
the diseased cells are transfected with DNA coding for the
enzyme that is then produced by the cell and effects the prodrug
cleavage. The cell surface display of â-glucuronidase has
recently been reported as a candidate for GDEPT.18
The proven clinical capacity of magnetic resonance imaging
(MRI) combined with the lack of ionizing radiation affords a
valuable tool for in vivo, whole organism molecular imaging.
Since the intrinsic contrast of MRI results from the physical
properties and local environment of endogenous water, this
technique has predominantly focused on anatomical structure.
The quest for higher resolution and shorter acquisition times
has spurred the development of agents that increase the observed
contrast in MR images.19,20 Much work has focused on
increasing the effectiveness of small-molecule contrast agents
based on Gd(III) chelates with particularly promising results
derived from interactions with macromolecules.21-23 While these
compounds delineate anatomical structure, new agents that are
biochemical and physiological reporters are being developed.24-29
Barriers to the development of new generations of biologically
responsive agents include the lack of adequate in vitro assays
to determine efficacy, enzyme kinetics, delivery, and inherent
contrast.
Here, we report a new class of MR contrast agents that uses
a self-immolative mechanism for activation in the presence of
an enzyme. The prototype of this class is a Gd(III) agent
designed to function within the â-glucuronidase prodrug family.
This agent’s effect on water proton T1 relaxation is modulated
by hydrolysis of â-glucuronic acid (Scheme 1). Previous studies
using â-galactosidase-sensitive EGad and EGadMe revealed that
the in vitro behavior of these agents does not reflect their in
vivo activity.24,25 Therefore, we have developed and exploited
in vitro assays using compounds 1 and 2 to gain insight into
factors affecting in vivo activity.
from Aldrich as anhydrous Sure-Seal bottles. Water was purified using
a Millipore Milli-Q Synthesis purifier. Sugar-containing compounds
were visualized on silica TLC plates with CAM stain (1 g of
(NH4)4Ce(SO4)4, 2.5 g of (NH4)4Mo2O7, 6 mL of concentrated H2SO4,
94 mL of water), while compounds containing unmetalated cyclen could
be easily detected using a platinum stain (150 mg of K2PtCl6, 10 mL
of 1 N HCl, 90 mL of water, 3 g of KI). NMR spectra were recorded
on either a Varian Mercury 400 MHz or Varian Inova 500 MHz
instrument. Peaks were referenced to an internal TMS standard. Infrared
spectra were measured using a KBr plate on a Biorad FTS-60 FTIR
spectrometer. Electrospray mass spectra were obtained via direct
infusion of a methanolic solution of the compound of interest on a
Varian 1200L single quadrupole mass spectrometer. Elemental analysis
was performed by Desert Analytics (Tucson, AZ). ICP-MS were
recorded on a VG Elemental PQ Excell spectrometer standardized with
eight concentrations spanning the range of 0-50 ppb Gd(III). One part
per billion In(III) was used as the internal standard for all runs. UV-
visible spectroscopy was performed on a HP 8452A diode array
spectrometer thermostated to 37 °C.
Experimental Section
General Methods. All reagents were used as purchased. 1,4,7,10-
Tetraazacyclododecane (cyclen) was obtained from Strem. Prohance
was purified from the clinically available sample from Bracco Inc. using
HPLC. Bovine liver â-glucuronidase [EC 3.2.1.31] Sigma cat G 0251
(Type B-1), BSA fraction V Sigma cat A 3059, p-nitrophenyl-â-D-
glucuronide, and male human blood serum Sigma cat H 1388 were
procured from Sigma. Dry solvents, where indicated, were obtained
(18) Heine, D.; Muller, R.; Brusselbach, S. Gene Therapy 2001, 8, 1005-1010.
(19) Merbach, A. E.; Toth, E. The Chemistry of Contrast Agents in Medical
Magnetic Resonance Imaging; John Wiley and Sons: West Sussex, New
York, 2001.
LC-MS. Analytical LC-MS was performed on a computer-
controlled Varian Prostar system consisting of a 410 autosampler
equipped with a 100 µL sample loop, two 210 pumps with 5 mL/min
heads, a 363 fluorescence detector, a 330 photodiode array (PDA)
detector, and a 1200L single quadrupole ESI-MS. All runs were
executed with a 0.8 mL/min flow rate using a ThermoElectron 4.6 ×
150 mm 5 µm Aquasil C18 column, with a 3:1 split directing one part
to the MS and three parts to the series-connected light detectors. Mobile
phase consisted of water (solvent A) and HPLC-grade acetonitrile
(solvent B) except where noted. All injections were full-loop.
Preparative LC. The preparative system is a Varian Prostar. Two
210 pumps with 25 mL/min heads fed a 5 mL manual inject sample
loop. Detection was performed after a 20:1 split by a two-channel 325
UV-visible detector and, on the low-flow leg, a HP 1046A fluorescence
detector. The mobile phases were the same as in the LC-MS
instrument. Preparative runs were typically 50-100 mg dissolved in
water and run at 15 mL/min on a ThermoElectron 20 × 250 mm 5 µm
Aquasil C18 column.
(20) Caravan, P.; Ellison, J. J.; McMurry, T. J.; Lauffer, R. B. Chem. ReV. 1999,
99, 2293-2352.
(21) Caravan, P. et al. J. Am. Chem. Soc. 2002, 124, 3152-3162.
(22) Aime, S.; Botta, M.; Fasano, M.; Crich, S. G.; Terreno, E. J. Biol. Inorg.
Chem. 1996, 1, 312-319.
(23) Doble, D. M. J.; Botta, M.; Wang, J.; Aime, S.; Barge, A.; Raymond, K.
N. J. Am. Chem. Soc. 2001, 123, 10758-10759.
(24) Moats, R. A.; Fraser, S. E.; Meade, T. J. Angew. Chem., Int. Ed. Engl.
1997, 36, 726-728.
(25) Louie, A. Y.; Huber, M. M.; Ahrens, E. T.; Rothbacher, U.; Moats, R.;
Jacobs, R. E.; Fraser, S. E.; Meade, T. J. Nat. Biotechnol. 2000, 18, 321-
325.
(26) Alauddin, M. M.; Louie, A. Y.; Shahinian, A.; Meade, T. J.; Conti, P. S.
Nucl. Med. Biol. 2003, 30, 261-265.
(27) Allen, M. J.; Meade, T. J. Metal Ions Biol. Syst. 2004, 42, 1-38.
(28) Morawski, A. M.; Winter, P. M.; Crowder, K. C.; Caruthers, S. D.; Fuhrhop,
R. W.; Scott, M. J.; Robertson, J. D.; Abendschein, D. R.; Lanza, G. M.;
Wickline, S. A. Magn. Reson. Med. 2004, 51, 480-486.
(29) Nivorozhkin, A. L.; Kolodziej, A. F.; Caravan, P.; Greenfield, M. T.;
Lauffer, R. B.; McMurry, T. J. Angew. Chem., Int. Ed. 2001, 40, 2903-
2906.
9
12848 J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005

Gadolinium Chelate for Detection of
â
-Glucuronidase
A R T I C L E S
Methyl-1-(4-(2-(1-(1,4,7,10-tetraazacyclododecyl))ethylcarbam-
oyloxymethyl)-2-nitrophenyl)-â-D-glucopyronuronate (12). Cyclen
(541 mg, 3.14 mmol) and 10 (640 mg, 1.26 mmol) were combined in
19 mL of DMSO, and the reaction was allowed to stir overnight. TLC
analysis at this time (10% MeOH/CH2Cl2) indicated no unreacted sugar.
The solvent was removed in vacuo, yielding a viscous yellow oil. The
oil was dissolved in 7 mL of MeOH, and a pale yellow solid precipitated
upon addition of 50 mL of Et2O. Upon storage at -20 °C for 1 h, the
hygroscopic solid was collected on a glass frit, washed with Et2O (3 ×
3 mL), and dried under vacuum, yielding 908 mg of solid. TLC (silica;
1:9:90 sat. KNO3 (aq.):water:MeCN; Pt stain visualization) and ESI-
MS showed very low intensity di- and trisubstituted side product peaks.
The precipitation procedure removed excess free base cyclen; however,
Calcd for C30H40N6O17Gd‚2.5H2O (87% Na+ salt): C 37.85, H 4.46,
N 8.83, Gd 16.52, Na 2.10. Found: C 37.69, H 4.28, N 9.12, Gd 16.88,
Na 2.10.
Europium(III)-1-(4-(2-(1-(4,7,10-triscarboxymethyl-(1,4,7,10-tetra-
azacyclododecyl)))ethylcarbamoyloxymethyl)-2-nitrophenyl)-â-D-
glucopyronuronate (4). This compound was synthesized and purified
in the same manner as 1 using 168 mg of 13 and substituting EuCl3
for GdCl3. Yield: 70 mg of 4 (18.1% from 10). The compound was
stored at -20 °C. Analysis of this material by analytic LC-MS (using
the same method as in the preparative runs) gave a single peak in the
PDA at 12.9 min with a m/z ) 907.2 (M - H+ ESI-MS) of appropriate
isotope pattern.
1-(2-tBoc-aminoethyl)-(1,4,7,10-tetraazacyclododecane) (17). A
quantity of 1.0 g (4.46 mmol) of 2-tBoc-aminoethylbromide30 was added
to a stirring solution of 1.92 g (11.1 mmol) of cyclen in 60 mL of dry
toluene. The solution was refluxed overnight under N2 and extracted
with 3 × 100 mL of water. The aqueous layer was extracted with 3 ×
75 mL of CH2Cl2, and the combined CH2Cl2 extracts were dried over
MgSO4. Removal of solvent gave a white solid that was washed with
1
MS and H NMR showed that the desired product was contaminated
with cyclen hydrobromide salt. This mixture was used in the subsequent
reaction without further purification.
Methyl-1-(4-(2-(1-(4,7,10-trisethylcarboxymethyl-(1,4,7,10-tetra-
azacyclododecyl)))ethylcarbamoyloxymethyl)-2-nitrophenyl)-â-D-
glucopyronuronate (13). A quantity of 887 mg of the mixture
containing 12 and 1.23 g of K2CO3 were suspended in 30 mL of
acetone; 820 µL of R-bromoethylacetate was added, and the solution
was allowed to stir at room temperature overnight. An additional 164
µL of R-bromoethylacetate and 210 mg K2CO3 were added after 24 h.
At 48 h, the reaction mixture was filtered to remove solids and purified
by flash chromatography (silica, 0-13.3% MeOH in CH2Cl2). The
resulting solid was dissolved in acetone and filtered through a 0.2 µm
PTFE filter to remove excess silica. This yielded 510 mg of 13.
Elemental bromine analysis indicated the presence of a mixture of free
base and hydrobromide salt. ESI-MS and 1H NMR showed no evidence
of lactamization. 1H NMR (500 MHz, CD3OD): δ 1.25 (m, 9H,
COOCH2CH3), 2.0-3.4 (br, 24H, cyclen H’s, 2H-sugar, acetate CH2),
3.52 (m, 4H), 3.64 (m, 1H), 3.76 (s, 3H, COOCH3), 4.10 (d, 1H, J )
10 Hz), 4.12-4.24 (m, 6H, COOCH2CH3), 5.07 (s, 2H, benzylic CH2),
5.21 (d, 1H, H-1, J ) 7 Hz), 7.39 (d, ArH, J ) 8 Hz), 7.60 (d, 1H,
ArH, J ) 8 Hz), 7.83 (s, 1H, ArH); 13C NMR (126 MHz, CD3OD) δ
14.64, 14.67, 38.73, 53.10, 55.99, 56.44, 56.88 (br), 62.61, 62.83, 66.01,
72.79, 74.46, 76.90, 77.25, 102.30, 118.81, 125.33, 133.50, 134.31,
142.19, 150.54, 159.14, 170.76, 175.57, 175.66 (br). ESI-MS m/z (M
+ H)+ 859.2 (40%), (M + Na)+ 881.2 (100%). Anal. Calcd for
C37H58N6O17‚acetone‚2.5H2O‚0.75HBr: C 46.98, H 6.87, N 8.22, Br
5.86. Found: C 47.05, H 6.55, N 8.23, Br 6.07.
1
cold Et2O and dried in vacuo. This yielded 890 mg (63%) of 15. H
NMR (500 MHz, CDCl3): δ 1.44 (s, 9H), 2.59 (br s, 10H), 2.63 (br s,
4H), 2.82 (br s, 4H), 3.22 (br s, 2H); 13C NMR (126 MHz, CDCl3) δ
28.58, 38.65, 46.21, 47.28, 47.92, 52.20, 54.28, 78.96, 156.14; ESI-
MS m/z (M + H)+ 316.3. Anal. Calcd for C15H33N5O2: C 57.11, H
10.54, N 22.20. Found: C 57.43, H 10.47, N 22.51.
1-(2-tBoc-aminoethyl)-4,7,10-(tris-tbutylcarboxymethyl)-(1,4,7,10-
tetraazacyclododecane) (18). To a solution of 950 mg of 17 (3.01
mmol) and 3.29 g (31.0 mmol) of Na2CO3 in dry MeCN under N2 was
added 2.4 mL (15.1 mmol) of R-bromo-tbutyl acetate. The suspension
was refluxed for 24 h, filtered, washed with 3 × 250 mL of hexanes,
and concentrated in vacuo to give a yellow oil. The resulting oil was
purified by chromatography (silica, 0-10% MeOH in CH2Cl2) to give
1.70 g (76%) of 18 as a white solid. Spectral and analytic data indicate
1
a mixture of free base and HBr salt. The H NMR was very broad
between 2 and 3.8 ppm and therefore unasssignable.13C NMR (126
MHz, CDCl3): δ (major product) 27.73, 27.92, 28.03, 28.32, 37.69,
48.01, 49.88, 50.09, 52.50, 53.01, 53.82, 55.60, 56.40 (br), 56.91, 79.17,
81.70, 82.74, 156.40, 169.99, 172.51; (minor peaks): 79.30, 81.80,
82.36, 170.33, 173.28 (br); ESI-MS m/z (M + H)+ 658.4 (60%), (M
+ Na)+ 680.3 (100%). Anal. Calcd for C33H63N5O8‚0.9HBr‚H2O: C
52.94, H 8.87, N 9.35, Br 9.60. Found: C 52.81, H 8.99, N 9.02, Br
9.78.
1-(2-Aminoethyl)-4,7,10-(triscarboxymethyl)-(1,4,7,10-tetraaza-
cyclododecane) TFA Salt (19). Deprotection of 192 mg of 18 was
achieved by stirring at room temperature in 4.75 mL of trifluoroacetic
acid with 125 µL of both triisopropylsilane and water. After 17 h, the
volatiles were removed in vacuo, and 40 mL of Et2O was added to
precipitate the ligand. The suspension was centrifuged and the white
pellet washed with 3 × 50 mL of Et2O. The resulting solid was dried
under vacuum and yielded 135 mg of the TFA salt, 19. 1H NMR showed
Gadolinium(III)-1-(4-(2-(1-(4,7,10-triscarboxymethyl-(1,4,7,10-
tetraazacyclododecyl)))ethylcarbamoyloxymethyl)-2-nitrophenyl)-
â-D-glucopyronuronate (1). A quantity of 455 mg of 13 in 10 mL of
water was cooled to 0 °C; 2.12 mL of 1 N NaOH was added over 1
min, and the solution was allowed to stir for 75 min. The pH was
brought to 6.5 with 0.1 N HCl, and 216 mg of GdCl3 (dissolved in 5
mL of water and brought to pH ) 6.5 with NaOH) was added dropwise.
The pH was kept above 5.5 during metal addition with 1 N NaOH.
The solution was allowed to warm to room temperature while stirring,
and the pH was adjusted periodically to keep it between 6 and 6.5.
After 3 days at room temperature, the pH showed no change and the
reaction was considered complete. The pH was brought to 8.2 and the
solution centrifuged to remove excess gadolinium as Gd(OH)3. Trace
solids were removed by filtration through a 0.2 µm nylon filter, and
the solution was lyophilized. The solid was brought up in 3 mL of
water and purified on preparative HPLC using the following method:
0-10% B over 10 min, hold for 15 min at 10% B, then wash to 98%
B before returning to 0% B. Two runs using this method were sufficient
to give material that was pure by microanalysis. Yield: 185 mg 1
(17.7% from 10). The compound was stored at -20 °C. Analysis of
this material by analytic LC-MS (using the same method as in the
preparative runs) gave a single peak in the PDA at 12.9 min with a
m/z ) 914.4 (M - H+ ESI-MS) of appropriate isotope pattern.
t
no remaining butyl resonances, while 19F NMR showed a signal for
TFA. ESI-MS m/z (M + H)+ 390.2.
Gadolinium(III)-1-(2-aminoethyl)-4,7,10-(triscarboxymethyl)-
(1,4,7,10-tetraazacyclododecane) (2). A quantity of 128 mg (0.61
mmol) of Gd(OH)3‚H2O and 239 mg of 19 were combined in 10 mL
of water, and the suspension was refluxed for 48 h. The solution was
brought to pH ) 10 with concentrated NH4OH and centrifuged to
remove excess Gd(OH)3. The pellet was washed, and the combined
washings and supernatant were lyophilized. The resulting solid was
dissolved in water and purified by successive runs on preparative HPLC
using the following method: 0-20% B over 10 min, hold at 20% B
for 15 min, then wash to 98% B before returning to 0% B. Due to the
lack of chromophores, the compound displays little UV absorption;
(30) Sakai, N.; Gerard, D.; Matile, S. J. Am. Chem. Soc. 2001, 123, 2517-
2524.
9
J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005 12849

A R T I C L E S
Duimstra et al.
fluorescence, however, can be detected by exciting at 271 nm and
monitoring the emission at 314 nm. Due to peak tailing, fractions were
analyzed by analytic LC-MS, and those containing 2 were pooled and
lyophilized; 101 mg of 2 was obtained analytically pure by this approach
(40% from 18). ESI-MS m/z (M + Na)+ 567.0 with Gd isotope pattern.
Anal. Calcd for C16H28N5O6Gd‚H2O: C 34.21, H 5.38, N 12.47.
Found: C 34.16, H 5.31, N 12.08.
showed negligible absorption over the time period of the experiment.
Fitting of the resulting data was performed using Origin 7.0. Analysis
of the reaction mixture by LC-MS using the method detailed in
purification of 1 after 10 min confirmed the presence of 4-hydroxy-
3-nitrobenzyl alcohol and compound 2. The ꢀ of p-nitrophenol at 422
nm was determined to be 7660 ( 180 M-1 cm-1 in the same fashion
as 4-hydroxy-3-nitrobenzyl alcohol. The determination of Km and Vmax
for p-nitrophenyl-â-D-glucuronide was performed in the same manner
as for 1 except the substrate concentrations were varied from 0.5 to
30.0 mM.
Magnetic Resonance. Samples consisting of 0.2 mM 1 and 1.0 mg/
mL of enzyme in 500 µL of phosphate buffer were monitored by UV-
visible and magnetic resonance. T1 was determined using a saturation
recovery (90-τ-90) pulse sequence on the Bruker mq60 operating as
detailed in the relaxivity section. The runs were performed in triplicate.
The substrate-only control was also examined in this manner. The
enzyme-only controls showed T1’s that were identical to neat buffer.
In all instances, with the exception of the magnetic resonance substrate
control in acetate buffer, the controls showed very little change over
the 1 h duration of each experiment. The substrate in acetate buffer
showed a slight (2%) decrease in T1 over the course of an hour. This
trend is in the opposite direction of the kinetics runs that show an
increase in T1.
Determination of q. Europium(III) compound 4 was dissolved in
H2O and D2O. The emission was monitored at 614 nm with excitation
at 394 nm on a Hitachi F4500 fluorometer operating in phosphorescence
lifetime (short) mode. The shortest lifetime measurable with this
instrument is about 0.3 ms. Fifteen scans were averaged and fit to a
monoexponential decay to give phosphorescent lifetimes: τ (D2O) )
1.274 ms; τ (H2O) ) 0.527 ms at room temperature. Using the equation
of Supkowski and Horrocks,32a q ) 0.89 (correction for a single N-H
amide-like oscillator gives q ) 0.81), while the equation from Beeby
et al.32b generates q ) 1.0 (N-H correction gives q ) 0.95). At 37 °C,
τ (D2O) ) 1.197 ms; τ (H2O) ) 0.458 ms. Supkowski and Horrocks
give q ) 1.2 (correction for a single N-H amide-like oscillator gives
q ) 1.1). Using the equation from Beeby et al., q ) 1.3 (N-H
correction gives q ) 1.2).
Europium(III)-1-(2-aminoethyl)-4,7,10-(triscarboxymethyl)-
(1,4,7,10-tetraazacyclododecane) (3). A quantity of 128 mg of 19 and
132 mg (0.36 mmol) of EuCl3‚6H2O were combined in water, and the
pH was adjusted to 6 with 1 N NaOH. The reaction was stirred for 3
days at room temperature, filtered, and lyophilized. The freeze-dried
solid was purified in the same manner (and exhibited similar peak
tailing) as 2 except fluorescence detection used λex ) 395 nm and λem
) 615 nm; 45 mg of 3 was obtained analytically pure in this fashion
(33% from 18). ESI-MS m/z (M + Na)+ 559.8, 561.7, Eu isotope
pattern. Anal. Calcd for C16H28N5O6Eu‚0.5H2O: C 35.11, H 5.34, N
12.79, Eu 27.76. Found: C 35.03, H 5.41, N 12.54, Eu 27.89.
Relaxivity. A 4 mM stock solution of either 1 or 2 in the appropriate
buffer was diluted to give 500 µL of each of seven approximate
concentrations for each run: 0, 0.05, 0.15, 0.3, 0.5, 1.0, and 2.0 mM.
The T1 of each concentration was determined using an inversion
recovery pulse sequence with appropriate recycle delays on a Bruker
mq60 Minispec. This instrument has a proton Larmor frequency of 60
MHz and operates at 37 °C. The resulting curves were fit to a
monoexponential function to obtain T1. Ten microliters of each sample
was digested in concentrated nitric acid, diluted with water, and
analyzed for exact Gd(III) concentration using ICP-MS. The reciprocal
of the longitudinal relaxation time was plotted against the concentration
obtained from ICP-MS and fit to a straight line. All lines were fit with
R2 > 0.998. This was performed for each buffer in duplicate. The
buffers were all made to have the appropriate pH at 37 °C and remade
if the pH had drifted more than 0.05 pH units upon storage. Anion
mimic and carbonate-containing buffers were made fresh daily.
Enzyme Kinetics. Bovine liver â-glucuronidase (Type B-1, Sigma
G 0251) stability in several buffers was assayed by the Sigma quality
control assay for â-glucuronidase from bovine liver. It was determined
that MOPS and anion mimic31 engendered the enzyme with poor
stability, while 100 mM phosphate with 0.01% (w/v) bovine serum
albumin (BSA), pH ) 7.4 at 37 °C, gave suitable stability (>2 h at 37
°C without loss of activity) provided the enzyme concentration was
greater than 0.5 mg/mL. The enzyme was stable for at least 24 h at 37
°C at its native pH of 5.0 in 100 mM acetate buffer.
Synthesis
The synthesis of 1 begins with methyl 1-bromo-2,3,4-tri-
O-acetyl-R-D-glucopyronuronate33 (Scheme 2). Coupling of
4-hydroxy-3-nitrobenzaldehyde to the pyranose via a Koenigs-
Knorr reaction followed by sodium borohydride reduction
produced 7 in 92% yield without recourse to chromatography.
Initial attempts at formation of the carbamate using p-nitrophenyl
chloroformate34,35 and triphosgene36,37 gave intermediates that
showed insufficient reactivity toward 2-bromoethylamine and
2-hydroxyethylamine. The reaction occurred smoothly, however,
when carbonyldiimidazole (CDI) was used as the carbonyl
equivalent. Methylation of the monoimidazolyl intermediate
gave an increased yield through precipitation of the cationic
intermediate.
LC-MS. After incubating 1 (0.2 mM) with bovine liver â-glu-
curonidase (1 mg/mL) in the above phosphate buffer at 37 °C, the
reaction mixture was analyzed by LC-MS and showed the presence
of 4-hydroxy-3-nitrobenzyl alcohol at 4.0 min, substrate 1 at 7.5 min,
and 2 at 11.8 min. The alcohol and 1 are readily distinguished by their
absorption spectra (PDA), their appropriate negative mode ESI-MS
patterns, and through spiking with authentic compound. 2 could be
observed using fluorescence. The HPLC method was as follows:
0-10% B over 10 min, hold at 10% B for 15 min, with fluorescence
using λex ) 271 nm and λem ) 314 nm.
UV-Visible. The enzymatic hydrolysis of the pyranose from 1 and
subsequent linker decomposition generates 4-hydroxy-3-nitrobenzyl
alcohol. The molar absorptivity, ꢀ, of this alcohol at 422 nm was
determined in triplicate to be 2840 ( 40 M-1 cm-1 in 100 mM
phosphate, 0.01% (w/v) BSA, pH ) 7.4 at 37 °C. At pH ) 7.4, the
substrate, 1, does not absorb at this wavelength. The kinetics were
sampled on the HP 8452A at 37 °C every 5 s for 10 min. The initial
velocities for pH ) 7.4 buffers were determined through a linear fit of
the first 10% change in absorbance. The Michaelis-Menten parameters
for 1 were determined by plotting the initial velocity versus substrate
concentration for concentrations ranging from 0.1 to 4.0 mM. The
measurements were made in triplicate with 1.0 mg/mL enzyme in a
500 µL sample. Control experiments without enzyme or substrate
Alkylation reactions involving 9 with either cyclen or
DO3A(tris-tbutyl ester)38 generated large amounts of R,â-
(31) Parker, D. In Crown Compounds: Towards Future Applications; Cooper,
S. R., Ed.; VCH: New York, 1992; pp 51-67.
(32) (a) Supkowski, R. M.; Horrocks, W. D., Jr. Inorg. Chim. Acta 2002, 340,
44-48. (b) Beeby, A.; Clarkson, I. M.; Dickins, R. S.; Faulkner, S.; Parker,
D.; Royle, L.; de Sousa, A. S.; Williams, J. A. G.; Woods, M. J. Chem.
Soc., Perkin Trans. 2 1999, 493-504.
(33) Bollenback, G. N.; Long, J. W.; Benjamin, D. G.; Lindquist, J. A. J. Am.
Chem. Soc. 1955, 77, 3310-3315.
(34) Florent, J.-C. et al. J. Med. Chem. 1998, 41, 3572-3581.
(35) Leu, Y. L.; Roffler, S. R.; Chern, J. W. J. Med. Chem. 1999, 42, 3623-
3628.
(36) Eckert, H.; Forster, B. Angew. Chem. 1987, 99, 922-923.
(37) Majer, P.; Randad, R. S. J. Org. Chem. 1994, 59, 1937-1938.
9
12850 J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005

Gadolinium Chelate for Detection of
â
-Glucuronidase
A R T I C L E S
Scheme 2
Scheme 4 a
a Coupling of macrocycle to 10, via DO3A route (A) or cyclen route
(B).
with cyclen by 1H NMR confirmed the elimination.41 The
formation of the R,â-unsaturated byproducts was circumvented
by selective removal, in good yield, of the acetyl groups from
9 using sodium methoxide, with 11 as the byproduct (Scheme
3). Introduction of the macrocycle was then attempted through
two approaches.
Scheme 3
In the first approach, DO3A(tris-tbutyl ester) was reacted with
10 (Scheme 4, A). After several days at room temperature, the
reaction was not complete and byproducts had begun to develop.
Heating the reaction induced sugar decomposition. The cyclen
route, in which the sugar-containing arm is added prior to the
acetate arms (Scheme 4, B), was successful, presumably because
the macrocycle free base is more nucleophilic toward the
unactivated alkyl bromide electrophile than is the DO3A
compound. Increasing the electrophilicity of 9 (Scheme 2) by
making the mesylated compound, 14 from 8 (Scheme S1), was
not successful due to the propensity of 14 to cyclize via
elimination of the mesyl group in the presence of cyclen.
Purification of 12 proved difficult, so the compound was
alkylated with ethyl bromoacetate. Excess cyclen, as the perethyl
DOTA ester, could then be removed by chromatography on
silica. Final deprotection and metalation were performed in a
one-pot procedure. Compound 1 was obtained analytically pure,
showing one peak by LC-MS with the appropriate isotope
pattern after purification by preparative HPLC. The overall yield
for the nine-step procedure was 8.0%. Compound 4 was
obtained in 8.2% overall yield by substituting EuCl3 for
GdCl3.
unsaturated byproducts due to an acid-base reaction between
cyclen and the acidic proton R to the methyl ester (Scheme 3).
It is known that this type of elimination can happen with acetyl-
protected glucuronic acids,39,40 but it is rarely mentioned in the
literature. Synthesis of methyl-1-ethoxy-2,3,4-tri-O-acetyl-â-D-
glucopyronuronate and subsequent monitoring of the reaction
(39) Schmidt, H. W. H.; Neukom, H. Tetrahedron Lett. 1969, 2011-2012.
(40) Stachulski, A. V.; Jenkins, G. N. Nat. Prod. Rep. 1998, 15, 173-186.
(41) Duimstra, J. A. Carbohydrate-Based Gadolinium(III) Magnetic Resonance
Imaging Contrast Agents: A Synthetic RetrospectiVe. M.S. Thesis, Cali-
fornia Institute of Technology, Pasadena, 2002.
(38) Dadabhoy, A.; Faulkner, S.; Sammes, P. G. J. Chem. Soc., Perkin Trans.
2 2002, 348-357.
9
J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005 12851

A R T I C L E S
Scheme 5
Duimstra et al.
Figure 1. T1 relaxivity of GdHPDO3A, 1, and 2 at 60 MHz, 37 °C, pH )
7.4. Error is (1 standard deviation of duplicate measurements: a10 mM
MOPS, 100 mM NaCl; b100 mM sodium phosphate; c100 mM sodium
phosphate, 0.01% (w/v) bovine serum albumin (BSA).
of water molecules coordinated to the metal center), Gd(III)-
2-hydroxypropyl-DO3A (GdHP-DO3A) was measured in pH
) 7.4 MOPS buffer and gave an r1 ) 2.99 ( 0.44 mM-1 s-1
under these conditions. It has been reported that MOPS does
not coordinate to the metal center. Bretonniere and co-
workers42,43 performed experiments in MOPS without interfer-
ence with carbonate binding to Eu(III) trisamide cyclen com-
plexes, while Bruce et al.44 indicate that experiments with related
compounds may be run in the structurally similar HEPES. Initial
relaxivity measurements of 1 and 2 made in the same MOPS
buffer showed a higher value for 1 and a value comparable to
GdHP-DO3A for 2 (Figure 1, MOPS columns). The magnitude
of r1 for 1 is somewhat low for a q ) 2 complex,45 indicating
intramolecular coordination of the sugar-containing arm to the
metal center. Determination of q at room temperature via Eu-
(III) fluorescence32 for 4 (the Eu(III) analogue of 1) in the
absence of buffer salts gave a q ) 1, supporting the intramo-
lecular coordination postulate. Presumably, this coordination
occurs through the carbamate carbonyl oxygen. A similar seven-
membered intramolecular ring has been invoked to explain low
q values for EuDO3A-type complexes containing tethered
carboxylates.46,47 The donating ability of a carbamate carbonyl
oxygen is expected to be lower than that of a carboxylate due
to increased electron density delocalization and charge neutrality.
This does not, however, preclude weak binding of the carbamate
moiety to the metal center in 1, giving a species that has one
inner-sphere water molecule. Measurement of q at 37 °C
supports this postulate. Here, the q is slightly higher with a value
of 1.1-1.2, a result which indicates a weakening of the
interaction between carbamate and metal center.48
The synthesis of compounds 2 and 3 proved to be more
straightforward (Scheme 5). Utilizing the monoalkylation of
excess cyclen, intermediate 17 was obtained analytically pure
following extraction and an ether wash. Subsequent alkylation
t
with 3 equiv of butyl bromoacetate gave 18 as a 9:1 mixture
of hydrobromide salt to free base. Deprotection was achieved
through the use of trifluoroacetic acid. Metalation of the free
ligand, 19, with Gd(OH)3 gave crude 2. Purification of 2 via
HPLC was difficult for two reasons. The lack of a chromophore
on 2 limited detection to fluorescence, and the presence of the
primary amine ligand gave a peak with a long tail. These
difficulties lowered the overall yield to 19% for four steps. The
Eu(III) compound 3 was obtained in 16% overall yield by
substituting EuCl3 for Gd(OH)3. Both compounds were deter-
mined to be authentic by elemental analysis.
Relaxivity
The defining parameter of contrast agent efficacy is relaxivity.
In this context, relaxivity, r1, is a measure of the extent to which
the agent, per unit, catalyzes the shortening of the longitudinal
relaxation time, T1, of protons on the hydrogen atoms in bulk
water. The presence of other species in solution, be they salts,
proteins, or small molecules, can have a marked effect on an
agent’s relaxivity. Relaxivity measurements made in solutions
of varying composition not only describe how the agent responds
to that composition but also provide insight into the microscopic
processes occurring at or near the Gd(III) center. Attributing
relaxivity effects to solution composition can only be made when
the contrast agent under study is of a known purity. The use of
analytically pure contrast agents allows for facile and accurate
determination of agent concentration through the use of Gd(III)
ICP-MS. This, in tandem with measurements made in duplicate,
reduces the systematic error in the relaxivity measurements.
The measurements shown in Figure 1 reveal trends in
relaxivity related to buffer composition. To provide a reference,
the relaxivity of the known q ) 1 contrast agent (q is the number
Due to enzyme instability in MOPS buffer, the relaxivities
of 1 and 2 were determined in the enzyme kinetics buffer
composed of phosphate and BSA (Figure 1). Here, the relax-
(42) Bretonniere, Y.; Cann, M. J.; Parker, D.; Slater, R. Chem. Commun. 2002,
1930-1931.
(43) Bretonniere, Y.; Cann, M. J.; Parker, D.; Slater, R. Org. Biomol. Chem.
2004, 2, 1624-1632.
(44) Bruce, J. I. et al. J. Am. Chem. Soc. 2000, 122, 9674-9684.
(45) Since r1 is directly proportional to q (see ref 13), a q ) 2 complex should
show an r1 approximately twice that of a q ) 1 complex, such as GdHP-
DO3A
(46) Messeri, D.; Lowe, M. P.; Parker, D.; Botta, M. Chem. Commun. 2001,
2742-2743.
(47) Lowe, M. P.; Parker, D.; Reany, O.; Aime, S.; Botta, M.; Castellano, G.;
Gianolio, E.; Pagliarin, R. J. Am. Chem. Soc. 2001, 123, 7601-7609.
9
12852 J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005

Gadolinium Chelate for Detection of
â
-Glucuronidase
A R T I C L E S
ivities are higher (4.73 vs 3.68 mM-1 s-1) but show the same
trend as observed in MOPS buffer, namely, a 20% drop in
relaxivity upon going from 1 to 2. In this case, however, it is
unclear if the species in solution is an intramolecularly
coordinated 1 or a ternary adduct in which phosphate has
displaced carbamate binding. Phosphate is known to coordinate
to DO3A analogues in a monodentate fashion44,49-52 and could,
therefore, compete with intramolecular coordination, especially
at the relatively high concentrations employed here.
The native lysosomal environment of â-glucuronidase is
acidic, with maximum activity observed between pH ) 4-5.53
The relaxivity of compounds 1 and 2 was thus measured at pH
) 5.0. The results from two different buffers at the same pH
are dramatically different. In 10 mM pyridine, 100 mM sodium
chloride buffer, compound 1 has a relaxivity of 3.89 ( 0.05
mM-1 s-1, while 2 displays an r1 of 4.16 ( 0.09 mM-1 s-1. In
100 mM acetate buffer, those values are 2.53 ( 0.01 and 2.16
( 0.04 mM-1 s-1, respectively. The difference between buffers
may be due to the coordinating ability of the buffer constituents.
Pyridine is expected to be a poor ligand for the oxophilic Gd(III)
in water54 and thus does not restrict water access to the metal
center as much as acetate. At pH ) 7, propionate binding to
GdDO3A has been shown to be monodentate,55 and TbDO3A,
in the presence of a large excess of acetate, gives a q value of
1.44 To our knowledge, the coordination of acetate or pyridine
to LnDO3A and its analogues has not been examined at pH )
5.0.
Figure 2. T1 relaxivity of 1 and 2 at 60 MHz, 37 °C, pH ) 7.4. Error is
(1 standard deviation of duplicate measurements: a100 mM sodium
b
phosphate, 0.01% (w/v) BSA, 24 mM NaHCO3; 10 mM MOPS, 24 mM
NaHCO3; c100 mM NaCl, 0.9 mM Na2HPO4, 30 mM NaHCO3, 0.13 mM
sodium citrate, 2.3 mM sodium lactate;44 dmale human blood serum.
Our studies on EGad and EGadMe demonstrate that in vitro
measurements do not correlate with in vivo efficacy. While an
MR contrast agent may display efficacy in vitro, the addition
of the components found in blood plasma could result in a
significant change in the properties of the agent. To investigate
these effects and the interplay of coordinating bidentate anions
on the present â-glucuronidase-sensitive agent, we measured
the relaxivity of 1 and 2 in buffers of increasing compositional
complexity. These results are depicted in Figure 2. For the
MOPS and phosphate data, the effects are superimposed upon
the overall relaxivity differences observed between the two
buffers (Figure 1). Addition of physiologically relevant carbon-
ate concentrations (24 mM)31 to the buffers displayed in Figure
1 gave similar results independent of buffer; namely, the
relaxivity of 1 decreased by 20-30%, while that of 2 remained
the same within error. These data support our initial hypothesis
of stronger binding of carbonate to 1 versus 2 and would indicate
that carbonate binding can displace the seven-membered car-
bamate chelate. Results with the anion extracellular mimic31,44
continue the trend begun in the carbonate-containing buffers.
Here, the data exhibit the desired dark to bright (lower to higher
Figure 3. Change in T1 of 1 (green [) and 2 (orange 9) as a function of
added sodium bicarbonate: 0.5 mM complex, 100 mM MOPS, 60 MHz,
37 °C, pH ) 7.4. Error is (1 standard deviation of duplicate measurements.
Lines are best linear least-squares fit to data.
relaxivity) change. In this case, it is a 17% increase. Once again,
the decrease is more dramatic for 1 (22%) than for 2 (6%),
providing more evidence for the increased chelating anion
affinity of 1 compared to that of 2.
Carbonate is known to bind seven-coordinate cis-divacant
LnDO3A analogues more strongly than to eight-coordinate
LnDOTA analogues.44,46,50,51 The relaxivities of 1 and 2 (Figure
2) follow this trend. Figure 3 displays the change in T1 observed
upon titration of these agents with sodium bicarbonate in a
physiologically relevant range at pH ) 7.4. 2 has a small
dependence on added bicarbonate (0.14% increase in T1 per mM
bicarbonate) in this range, while 1 displays a stronger correlation
at 1.1% increase in T1 per mM bicarbonate,56 indicating
displacement of intramolecular carbamate coordination. The
results correlate well with the change in relaxivity of the two
compounds upon addition of 24 mM bicarbonate (Figures 1 and
2). The titration of 1 has not begun to saturate at 36 mM added
bicarbonate, indicative of a small binding constant for the ternary
complex. This result contrasts sharply with the results observed
for the tri-positive Ln(III) tris amide cyclen complexes. These
(48) The q values at elevated temperatures must be viewed with caution since
the empirically observed relationship between phosphorescence lifetime
and coordination number was evaluated at room temperature. It is unlikely
that the empirical parameters will be independent of temperature.
(49) Aime, S.; Gianolio, E.; Terreno, E.; Giovenzana, G. B.; Pagliarin, R.; Sisti,
M.; Palmisano, G.; Botta, M.; Lowe, M. P.; Parker, D. J. Biol. Inorg. Chem.
2000, 5, 488-497.
(50) Dickins, R. S. et al. J. Am. Chem. Soc. 2002, 124, 12697-12705.
(51) Supkowski, R. M.; Horrocks, W. D., Jr. Inorg. Chem. 1999, 38, 5616-
5619.
(52) Vander Elst, L.; Van Haverbeke, Y.; Goudemant, J. F.; Muller, R. N. Magn.
Reson. Med. 1994, 31, 437-444.
(53) Himeno, M.; Hashiguchi, Y.; Kato, K. J. Biochem. (Tokyo) 1974, 76, 1243-
1252.
(54) A search of the CSD returned no structures containing a lanthanide
coordinated to both a pyridyl and aquo ligand.
(55) Aime, S.; Terreno, E.; Botta, M.; Bruce, J. I.; Parker, D.; Mainero, V. Chem.
Commun. 2001, 115-116.
(56) The concentration of the various carbonate species is a function of pH, as
discussed in Bruce et al.44
9
J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005 12853

A R T I C L E S
Duimstra et al.
complexes display saturation with as little as 2 mM added
bicarbonate.44 The lower affinity for carbonate species of
complex 1 compared to that of the tris amide complexes can
be attributed to the lower charge (neutral versus 3+) and the
binding competition with the intramolecular carbamate chelation.
If the coordination of endogenous anions was the sole
intermolecular contributor to the relaxivities observed in these
contrast agents, then results from the anion extracellular mimic
would translate well to results obtained in human blood serum.
The present case shows the dramatic differences on going from
a competitive extracellular anion mimic to human serum. The
data are entirely different as compound 1 shows an increase in
relaxivity of 240%, while compound 2 displays a 150% increase.
Furthermore, the relaxivity differential has switched with 1 27%
brighter than 2 in serum. On the basis of our previous work,
this difference should be sufficient for in vivo imaging.25 For
these compounds in serum, the relaxivity indicates a species
containing inner-sphere water molecules, in contrast to the q )
0 species detected by Aime et al. for DO3A analogues in the
presence of albumin.49 The complex composition of human
serum makes it difficult to ascribe the results to any given
component, but the higher viscosity and possible macro-
molecular interactions could affect τR and hence the relaxivity.
Figure 4. Kinetics of hydrolysis of 1 catalyzed by bovine liver â-
glucuronidase. Each point is the average of three runs (1 standard deviation.
Line represents best fit to Michaelis-Menten model. Conditions are 1.0
mg/mL of enzyme in 100 mM sodium phosphate, 0.01% (w/v) bovine serum
albumin (BSA), pH ) 7.4 at 37 °C.
Table 1. Enzyme Kinetic Parameters for 1 and PNG
Km
Vmax
(pmol s-
kcat/Km
1
1
substrate
pHa
(mM)
)
(M-1 s-
)
Enzyme Kinetics
1b
7.4
7.4
5
0.906 (0.133
18.5 ( 5.6
0.33
98 ( 5.5
184 ( 30
74.0 ( 10.9
6.76 ( 2.05
PNGb
PNGc
PNGd
To study the enzymatic processing of 1, we chose to use
â-glucuronidase isolated from bovine liver. This commercially
available enzyme is more similar to the human variant than the
E. coli version.57,58 The bovine liver enzyme is localized in the
lysosome and hence has maximal activity at pH ) 5.53 At neutral
interstitial pH, the activity of both the human and bovine liver
enzymes is 10% that at pH ) 5 and falls off rapidly in more
alkaline solutions.53,59 The attenuation of activity is significant
since ADEPT and PMT function at interstitial pH. The lower
pH around tumor masses is advantageous in this respect as
enzyme activity will be higher. Although the bacterial enzyme
has much higher activity at extracellular pH,60 it is obviously
not endogenous to humans and its use in ADEPT has resulted
in host immune response16,18 severely restricting its potential.
Furthermore, the 2-nitroquinonemethide that results from en-
zymatic processing of 1 is not an irreversible inhibitor of bovine
â-glucuronidase.57
LC-MS provided a facile means to detect and identify the
components of the enzyme-catalyzed hydrolysis of 1. After a 2
h reaction in phosphate buffer, the presence of 4-hydroxy-3-
nitrobenzyl alcohol could be easily detected in the LC data at
4.0 min; unreacted 1 was observed at 7.5 min, and 2 appeared
at 11.8 min. 1 had disappeared after 24 h. The verification of
the presence of 4-hydroxy-3-nitrobenzyl alcohol enabled the
enzyme kinetics in phosphate/BSA buffer to be quantified using
a continuous UV-visible assay.61 The alcohol has an absorption
maximum at 422 nm, while 1 does not absorb at this wavelength.
A plot of initial velocity versus substrate concentration fits well
to a single-site Michaelis-Menten model (Figure 4). The kinetic
7.2
1.40 ( 0.42
47
a At 37 °C. b With 1.0 mg/mL of bovine liver â-glucuronidase (type B-1),
100 mM sodium phosphate, 0.01% (w/v) bovine serum albumin (BSA).
Data are average of three runs ( 1 standard deviation. c Bovine liver
â-glucuronidase.53 d Human recombinant â-glucuronidase.62
parameters from the nonlinear fit are tabulated in Table 1. For
comparison, kcat/Km was determined for the standard assay
substrate, p-nitrophenyl-â-D-glucuronide (PNG). The data ob-
tained show that 1 is a better substrate by an order of magnitude
for bovine liver â-glucuronidase than is PNG (Table 1). These
kinetics demonstrate the marked improvement afforded by
incorporation of the self-immolative linker. The nonimmolative
â-galactosidase-sensitive agent, EGadMe, displayed enzyme
kinetics that were 2-3 orders of magnitude worse than that of
the standard â-galactosidase substrate, 2-nitrophenyl-â-D-
galactopyranoside.25
The kinetics can also be monitored using magnetic resonance.
The use of MR allowed observation of kinetics in human serum,
something that could not be done by visible light spectroscopy
due to light scatter by suspended particles. Figure 5 shows the
normalized change in T1 as a function of enzyme incubation
time. The results show excellent correlation with the relaxivities
of the substrate 1 and the product 2, measured in the absence
of enzyme (Figures 1 and 2). The largest change detected for
the pH ) 7.4 buffers comes from the data in human serum.
Here, a 14% increase is observed over the course of 1 h, and
the curve has not saturated at this time. The control without
enzyme maintains a constant T1 over this period. These serum
experiments demonstrate that the contrast agent functions well
in the complex biological milieu represented by human serum.
(57) Azoulay, M.; Chalard, F.; Gesson, J. P.; Florent, J. C.; Monneret, C.
Carbohydr. Res. 2001, 332, 151-156.
(58) Brot, F. E.; Bell, C. E., Jr.; Sly, W. S. Biochemistry 1978, 17, 385-391.
(59) Jain, S.; Drendel, W. B.; Chen, Z.-w.; Mathews, F. S.; Sly, W. S.; Grubb,
J. H. Nat. Struct. Biol. 1996, 3, 375-381.
(60) Jefferson, R. A.; Burgess, S. M.; Hirsh, D. Proc. Natl. Acad. Sci. U.S.A.
1986, 83, 8447-8451.
(62) O’Leary, K. A.; Day, A. J.; Needs, P. W.; Sly, W. S.; O’Brien, N. M.;
(61) Aich, S.; Delbaere, L. T. J.; Chen, R. BioTechniques 2001, 30, 846-850.
Williamson, G. FEBS Lett. 2001, 503, 103-106.
9
12854 J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005

Gadolinium Chelate for Detection of
â
-Glucuronidase
A R T I C L E S
absorption spectroscopy (solid line in Figure 5) and the 15%
decrease in relaxivity observed between the pure compounds
in acetate buffer.
The successful synthesis of an MRI contrast agent sensitive
to the oncologically relevant enzyme â-glucuronidase expands
the repertoire of agents responsive to biochemical reactions into
the realm of clinically important processes. The relaxivity and
kinetic assays developed in the present work elucidate the
marked dependence that buffer composition can have upon the
efficacy of an MR agent. Affinity for anions, such as carbonate,
may be improved by increasing the electrostatic charge at the
metal center through the replacement of anionic carboxylate
ligands with neutral amide coordinating groups. The viability
of the self-immolative linker conjugated to the macrocycle
provides a model for future agents in which water access to the
Gd(III) ion is restricted by a bridge connecting opposite sides
of the macrocycle. Cleavage of a pendant moiety results in a
cascade reaction that liberates the bridge, allowing water to
access the inner-sphere coordination site. This class of agent in
which inner-sphere coordination is blocked intramolecularly
should function independently of endogenous anions.
Figure 5. Kinetics of enzyme-catalyzed hydrolysis of 1 monitored by bulk
water T1 relaxation (60 MHz, 37 °C). Error bars on data (filled symbols)
are (1 standard deviation of three independent measurements. Open
symbols are control runs without enzyme: (blue 9) 0.2 mM 1, 1.0 mg/mL
of â-glucuronidase, 100 mM sodium phosphate, 0.01% (w/v) bovine serum
albumin (BSA), pH ) 7.4; (orange 2) 0.2 mM 1, 1.0 mg/mL of
â-glucuronidase, 100 mM sodium phosphate, 0.01% (w/v) bovine serum
albumin (BSA), 24 mM NaHCO3, pH ) 7.4. Green line is hydrolysis of 1
monitored by UV-visible at 354 nm, conditions as for: (green [) 0.2 mM
1, 0.1 mg/mL of â-glucuronidase, 100 mM sodium acetate, pH ) 5.0; (red
b) 0.2 mM 1, 1.0 mg/mL of â-glucuronidase, male human blood serum.
Acknowledgment. This paper is dedicated in memory of
Lorenzo J. Ponza. We thank Profs. Scott E. Fraser and Harry
B. Gray for helpful discussions. Assistance with ICP-MS was
provided by Keith MacRenaris. This work was supported by
DOD Grant DAMD 17-02-1-0693 and NIH Grant 5 RO1
A147003.
Enzyme instability precluded determination at long reaction
times, but the shapes of the curves match those obtained from
the UV-visible assays, indicating that compound 2 forms on a
time scale similar to the change in absorbance of the aromatic
linker. In particular, the longitudinal relaxation time observed
in acetate buffer levels off at +15% around 35 min, corroborat-
ing well with both the complete conversion detected by
Supporting Information Available: Synthesis of compounds
6-10 and 14-16 and complete refs 21, 34, 44, 50. This material
is available free of charge via the Internet at http://pubs.acs.org.
JA042162R
9
J. AM. CHEM. SOC. VOL. 127, NO. 37, 2005 12855