EP-2104R: A fibrin-specific gadolinium-based MRI contrast agent for detection of thrombus

Published on Web 04/08/2008

EP-2104R: A Fibrin-Speciﬁc Gadolinium-Based MRI Contrast

Agent for Detection of Thrombus

Kirsten Overoye-Chan,^†Stefﬁ Koerner,^†Richard J. Looby,^†Andrew F. Kolodziej,^†

Stephan G. Zech,^†Qing Deng,^†Jaclyn M. Chasse,^†Thomas J. McMurry,^†and

Peter Caravan^‡,*

EPIX Pharmaceuticals, 4 Maguire Road, Lexington, Massachusetts 02421, and Athinoula A.

Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General

Hospital, Charlestown, Massachusetts 02429

Received February 1, 2008; E-mail: caravan@nmr.mgh.harvard.edu

Abstract: Thrombus (blood clot) is implicated in a number of life threatening diseases, e.g., heart attack,

stroke, pulmonary embolism. EP-2104R is an MRI contrast agent designed to detect thrombus by binding

to the protein ﬁbrin, present in all thrombi. EP-2104R comprises an 11 amino acid peptide derivatized with

2 GdDOTA-like moieties at both the C- and N-terminus of the peptide (4 Gd in total). EP-2104R was

synthesized by a mixture of solid phase and solution techniques. The La(III) analogue was characterized

by and 1D and 2D NMR spectroscopy and was found to have the expected structure. EP-2104R was

found to be signiﬁcantly more inert to Gd(III) loss than commercial contrast agents. At the most extreme

conditions tested (pH 3, 60 °C, 96 hrs), less than 10% of Gd was removed from EP-2104R by a challenge

with a DTPA based ligand, while the commercial contrast agents equilibrated within minutes to hours.

EP-2104R binds equally to two sites on human ﬁbrin (K_d) 1.7 ( 0.5 µM) and has a similar afﬁnity to

mouse, rat, rabbit, pig, and dog ﬁbrin. EP-2104R has excellent speciﬁcity for ﬁbrin over ﬁbrinogen (over

100-fold) and for ﬁbrin over serum albumin (over 1000-fold). The relaxivity of EP-2104R bound to ﬁbrin at

37 °C and 1.4 T was 71.4 mM^-1s^-1per molecule of EP-2104R (17.4 per Gd), about 25 times higher than

that of GdDOTA measured under the same conditions. Strong ﬁbrin binding, ﬁbrin selectivity, and high

molecular relaxivity enable EP-2104R to detect blood clots in vivo.

process.¹The coagulation enzymes thrombin and Factor XIII,

the glycoprotein GPIIB/IIIA receptor on activated platelets, and

the ﬁbrin polymer have all been exploited as imaging targets.

In the clotting process,¹platelets become activated and adhere

to the vessel wall and each other to form a plug. The enzyme

thrombin cleaves the ﬁbrinopeptides on circulating ﬁbrinogen

protein enabling end-to-end polymerization of the ﬁbrinogen

moieties to produce ﬁbrin. The ﬁbrin mesh is stabilized by cross-

links introduced by the enzyme Factor XIII. Activated platelets,

ﬁbrin, thrombin, and Factor XIII have all served as targets for

imaging agents. In the 1990s several groups^2–7used technetium-

99m labeled peptides or peptidomimetic agonists for the GPIIb/

IIIa receptor on activated platelets as a marker of thrombus.

1. Introduction

Thrombosis is the underlying pathology in a number of

cardiovascular diseases which affect millions worldwide. Heart

attack, ischemic stroke, pulmonary embolism, and deep vein

thrombosis all have culprit blood clots (thrombi). Thrombosis

can also occur in the cardiac chambers. For example, atrial

ﬁbrillation patients are at high risk for atrial thrombosis and

the presence of thrombus impacts their treatment. Another area

of intense research is vulnerable atherosclerotic plaque, where

unstable lesions can rupture leading to thrombosis and possibly

heart attack or stroke. In all of these diseases, there is clearly a

need to detect thrombus and to monitor its presence or absence

during treatment.

Many imaging techniques are used to identify thrombus, and

the preferred technique often depends on the anatomical region.

For instance ultrasound is often used to identify deep vein

thrombi, while spiral computed tomography (CT) is increasingly

used to detect pulmonary emboli. The underlying physics of

the different techniques are such that one technique is not ideally

suited for thrombus detection in all vascular territories. As a

result, there has been an effort to develop imaging probes that

speciﬁcally delineate thrombus pathology. Different approaches

have been taken to design thrombus-speciﬁc probes, and this is

perhaps not surprising given the complexity of the clotting

(1) Monroe, D. M.; Hoffman, M. Arterioscler. Thromb. Vasc. Biol. 2006,

26, 41–8.

(2) Barrett, J. A.; Crocker, A. C.; Damphousse, D. J.; Heminway, S. J.;

Liu, S.; Edwards, D. S.; Lazewatsky, J. L.; Kagan, M.; Mazaika, T. J.;

Carroll, T. R. Bioconjugate Chem. 1997, 8, 155–60.

(3) Ben-Haim, S.; Kahn, D.; Weiner, G. J.; Madsen, M. T.; Waxman,

A. D.; Williams, C. M.; Clarke-Pearson, D. L.; Coleman, R. E.;

Maguire, R. T. Nucl. Med. Biol. 1994, 21, 131–42.

(4) Francesconi, L. C.; Zheng, Y.; Bartis, J.; Blumenstein, M.; Costello,

C.; De Rosch, M. A. Inorg. Chem. 2004, 43, 2867–75.

(5) Lister-James, J.; Knight, L. C.; Maurer, A. H.; Bush, L. R.; Moyer,

B. R.; Dean, R. T. J. Nucl. Med. 1996, 37, 775–81.

(6) Lister-James, J.; Vallabhajosula, S.; Moyer, B. R.; Pearson, D. A.;

McBride, B. J.; De Rosch, M. A.; Bush, L. R.; Machac, J.; Dean,

R. T. J. Nucl. Med. 1997, 38, 105–11.

^†EPIX Pharmaceuticals.

(7) Mitchel, J.; Waters, D.; Lai, T.; White, M.; Alberghini, T.; Salloum,

A.; Knibbs, D.; Li, D.; Heller, G. V. Circulation 2000, 101, 1643–6.

^‡Massachusetts General Hospital and Harvard Medical School.

9

10.1021/ja800834y CCC: $40.75

2008 American Chemical Society

J. AM. CHEM. SOC. 2008, 130, 6025–6039 6025

A R T I C L E S

Overoye-Chan et al.

One Tc-based agent was eventually approved for imaging deep

vein thrombosis. Jaffer, Tung and co-workers reported near-

infrared probes that had speciﬁcity for thrombin,^8,9Factor

XIII,^10–12or activated platelets.¹³Recently, a phage display

study against clotted plasma identiﬁed peptides that target the

ity, and contained four gadolinium chelates to enhance the MRI

signal. EP-2104R was found to be effective at providing positive

contrast enhancement in preclinical models of carotid artery,²⁰

coronary artery,^21,22atrial,²³and cerebral venous sinus²⁴throm-

bosis as well as models of pulmonary embolism.^22,25,26The

compound proved efﬁcacious in acute (fresh clot) as well as

chronic (aged clot) models of thrombosis. A related, analogous

compound, EP-1873 was demonstrated to selectively enhance

ruptured atherosclerotic plaques in an animal model.²⁷EP-2104R

differs primarily from EP-1873 in the type of Gd chelator used:

a more kinetically inert DOTA-like chelator is used in EP-2104R

compared to a DTPA derivative used in EP-1873. Based on

these and other studies, EP-2104R was advanced into clinical

trials where preliminary results have been promising.^28,29

14

ﬁbrin-ﬁbronectin complex in clots.

Both nuclear and optical probes have very good detection

sensitivity. A key limitation of nuclear imaging, however, is

its relatively poor spatial resolution and the need to have a high

target/background ratio. For near-IR imaging, current technology

is limited by tissue penetration and scatter. Magnetic resonance

imaging (MRI) has excellent spatial resolution and no ionizing

radiation and can image deep tissues. MRI has an added

advantage of providing anatomical information to put the

location of the thrombus in context, e.g., is it in an artery or

vein and in which vessel precisely? A positive contrast (bright

spot) thrombus imaging agent for use with MRI would be a

powerful tool to faciliate a deﬁnitive diagnosis and could offer

the potential for whole body exams to pinpoint the source of

thrombus. However the challenge for MRI agents is sensitivity:

for gadolinium-based agents, micromolar concentrations are

required in order to generate robust image contrast.¹⁵

Particle-based technologies provide one means of delivering

large numbers of gadolinium chelates to facilitate thrombus

imaging. Wickline, Lanza and co-workers developed a platform

technology based on a perﬂuorocarbon emulsion for both MRI^16–18

and ultrasound¹⁹contrast agents. They could incorporate gado-

linium chelates into the emulsion via lipophilic chains on the

chelates resulting in thousands of gadolinium ions per particle. They

then linked a ﬁbrin speciﬁc antibody to the particle and used this

to image thrombus in canine models of disease.¹⁶

Existing clinical MRI contrast agents are markers of extra-

cellular space or provide tissue (blood vessels, liver) speciﬁc-

ity.³⁰EP-2104R represents a new class of MRI contrast agents,

one in which a speciﬁc protein is imaged. This advancement

beyond anatomical imaging down to the molecular level has

been termed molecular imaging.³¹In this report, the synthesis

and characterization of the ﬁrst MRI molecular imaging agent

to advance into human trials, EP-2104R, is described. The

paramagnetic compound is characterized by a number of

analytical methods; in addition NMR studies on the diamagnetic

La(III) analogue conﬁrm the expected structure. The kinetic

inertness with respect to gadolinium loss, a critical parameter

for patient safety, is addressed in a series of competition

experiments. The afﬁnity of EP-2104R for human ﬁbrin is

described as well as its speciﬁcity for ﬁbrin compared to

ﬁbrinogen or other plasma proteins. Finally, the NMR relaxation

enhancing properties of EP-2104R are also described in different

media and over a range of magnetic ﬁelds.

We took an alternate approach to designing a ﬁbrin-speciﬁc

MRI contrast agent. We reasoned that ﬁbrin could be a useful

target for a relatively small MRI probe because of its micromolar

concentration. In addition, ﬁbrin is present in all thrombi:

arterial, venous, and cardiac, but is not found in plasma, thus

providing speciﬁcity for the thrombus pathology. Moreover,

unlike platelets and enzymes, ﬁbrin is accessible in all active

stages of clot development. We developed a peptide-based agent

called EP-2104R that was targeted to ﬁbrin, providing speciﬁc-

(20) Sirol, M.; Fuster, V.; Badimon, J. J.; Fallon, J. T.; Moreno, P. R.;

Toussaint, J. F.; Fayad, Z. A. Circulation 2005, 112, 1594–600.

(21) Botnar, R. M.; Buecker, A.; Wiethoff, A. J.; Parsons, E. C., Jr.; Katoh,

M.; Katsimaglis, G.; Weisskoff, R. M.; Lauffer, R. B.; Graham, P. B.;

Gunther, R. W.; Manning, W. J.; Spuentrup, E. Circulation 2004, 110,

1463–6.

(22) Spuentrup, E.; Buecker, A.; Katoh, M.; Wiethoff, A. J.; Parsons, E. C.,

Jr.; Botnar, R. M.; Weisskoff, R. M.; Graham, P. B.; Manning, W. J.;

Gunther, R. W. Circulation 2005, 111, 1377–82.

(8) Jaffer, F. A.; Tung, C. H.; Gerszten, R. E.; Weissleder, R. Arterioscler.

Thromb. Vasc. Biol. 2002, 22, 1929–35.

(23) Spuentrup, E.; Fausten, B.; Kinzel, S.; Wiethoff, A. J.; Botnar, R. M.;

Graham, P. B.; Haller, S.; Katoh, M.; Parsons, E. C., Jr.; Manning,

W. J.; Busch, T.; Gunther, R. W.; Buecker, A. Circulation 2005, 112,

396–9.

(9) Tung, C. H.; Gerszten, R. E.; Jaffer, F. A.; Weissleder, R. ChemBio-

Chem 2002, 3, 207–11.

(10) Jaffer, F. A.; Tung, C. H.; Wykrzykowska, J. J.; Ho, N. H.; Houng,

A. K.; Reed, G. L.; Weissleder, R. Circulation 2004, 110, 170–6.

(11) Kim, D. E.; Schellingerhout, D.; Jaffer, F. A.; Weissleder, R.; Tung,

C. H. J. Cereb. Blood Flow Metab. 2005, 25, 226–33.

(12) Tung, C. H.; Ho, N. H.; Zeng, Q.; Tang, Y.; Jaffer, F. A.; Reed, G. L.;

Weissleder, R. ChemBioChem 2003, 4, 897–9.

(24) Stracke, C. P.; Katoh, M.; Wiethoff, A. J.; Parsons, E. C.; Spangenberg,

P.; Spuntrup, E. Stroke 2007, 38, 1476–81.

(25) Spuentrup, E.; Katoh, M.; Buecker, A.; Fausten, B.; Wiethoff, A. J.;

Wildberger, J. E.; Haage, P.; Parsons, E. C., Jr.; Botnar, R. M.;

Graham, P. B.; Vettelschoss, M.; Gunther, R. W. InVest. Radiol. 2007,

42, 586–595.

(13) Tung, C. H.; Quinti, L.; Jaffer, F. A.; Weissleder, R. Mol. Pharm.

2005, 2, 92–5.

(26) Spuentrup, E.; Katoh, M.; Wiethoff, A. J.; Parsons, E. C., Jr.; Botnar,

R. M.; Mahnken, A. H.; Gunther, R. W.; Buecker, A Am. J. Respir.

Crit. Care Med. 2005, 172, 494–500.

(14) Pilch, J.; Brown, D. M.; Komatsu, M.; Jarvinen, T. A.; Yang, M.;

Peters, D.; Hoffman, R. M.; Ruoslahti, E. Proc. Natl. Acad. Sci. U.S.A.

2006, 103, 2800–4.

(27) Botnar, R. M.; Perez, A. S.; Witte, S.; Wiethoff, A. J.; Laredo, J.;

Hamilton, J.; Quist, W.; Parsons, E. C., Jr.; Vaidya, A.; Kolodziej,

A.; Barrett, J. A.; Graham, P. B.; Weisskoff, R. M.; Manning, W. J.;

Johnstone, M. T. Circulation 2004, 109, 2023–9.

(15) Caravan, P. Chem. Soc. ReV. 2006, 35, 512–23.

(16) Flacke, S.; Fischer, S.; Scott, M. J.; Fuhrhop, R. J.; Allen, J. S.;

McLean, M.; Winter, P.; Sicard, G. A.; Gaffney, P. J.; Wickline, S. A.;

Lanza, G. M. Circulation 2001, 104, 1280–5.

(28) Katoh, M.; Wiethoff, A. J.; Sparing, R.; Gunther, R. W.; Spuentrup,

E. In Proceedings of the International Society for Magnetic Resonance

in Medicine; Berlin, Germany, 2007; p 424.

(17) Winter, P. M.; Caruthers, S. D.; Yu, X.; Song, S. K.; Chen, J.; Miller,

B.; Bulte, J. W.; Robertson, J. D.; Gaffney, P. J.; Wickline, S. A.;

Lanza, G. M. Magn. Reson. Med. 2003, 50, 411–6.

(29) Vymazal, J.; Spuentrup, E.; Cardenas, G.; Stefancik, R.; Hartmann,

M.; Uprichard, A. In Proceedings of the Radiological Society of North

America; Chicago, 2006; p SSK06–07.

(18) Yu, X.; Song, S. K.; Chen, J.; Scott, M. J.; Fuhrhop, R. J.; Hall, C. S.;

Gaffney, P. J.; Wickline, S. A.; Lanza, G. M. Magn. Reson. Med.

2000, 44, 867–72.

(30) Caravan, P.; Lauffer, R. B. In Clinical Magnetic Resonance Imaging,

3rd ed.; Edelman, R. R., Hesselink, J. R., Zlatkin, M. B., Crues, J. V.,

Eds.; Saunders: Philadelphia, 2005; Vol. 1, pp 357–75..

(31) Weissleder, R.; Mahmood, U. Radiology 2001, 219, 316–33.

(19) Lanza, G. M.; Wallace, K. D.; Scott, M. J.; Cacheris, W. P.;

Abendschein, D. R.; Christy, D. H.; Sharkey, A. M.; Miller, J. G.;

Gaffney, P. J.; Wickline, S. A. Circulation 1996, 94, 3334–40.

9

6026 J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008

EP-2104R: Thrombus Speciﬁc MRI Contrast Agent

A R T I C L E S

Scheme 1. Synthesis of Peptide Tetraamine 9

2. Results and Discussion

romethane/5 water). The peptide was cyclized in a DMSO

solution. Once cyclization was complete, the peptide was

puriﬁed by reversed phase chromatography.

2.1. Synthesis of EP-2104R. The peptide portion of EP-2104R

was synthesized in 18 steps (when each amino acid addition is

counted) as outlined in Schemes 1 and 2. The peptide used in

EP-2104R was based on a similar, all natural amino acid

sequence identiﬁed using a phage display screen against human

ﬁbrin. The unnatural amino acids D-glutamate, L-3-chloroty-

rosine, and hydroxyproline were found to improve ﬁbrin binding.

EP-2104R consists of an 11 amino acid peptide for targeting

ﬁbrin with 4 Gd-DOTAGA chelate moieties for MRI signal

enhancement. The gadolinium chelates are linked at both the

N- and C-termini. In order to facilitate coupling of the chelates

to both termini, the C-terminus was ﬁrst converted to an amino

terminus. This was achieved by derivatizing the Rink amide

MBHA resin with an amino acid-linker-amino acid moiety.

Standard Fmoc-peptide chemistry was then employed to elon-

gate the peptide on the resin. After addition of the ﬁnal amino

acid, the peptide was fully deprotected using an acidic cocktail

(80 TFA/5 triisopropylsilane/5 1-dodecanethiol/5 dichlo-

We used the DOTAGA chelating moiety³²to introduce

gadolinium(III), Chart 1. This synthon has been used to label

peptides that target the somatostatin receptor with ⁹⁰Y(III).³³

We chose the DOTAGA synthon for two reasons. First, the

GdDOTA-like complex was expected to be highly stable and

extremely inert to gadolinium loss. Second, use of this synthon

results in the same donor atom set as in [Gd(DOTA)].³³DOTA-

like ligands are often conjugated to peptides using one of the

carboxylates to form an amide linkage (DOTA-monoamide).

However the amide oxygen donor is known to decrease the

water exchange rate at the gadolinium, and this can result in

decreased relaxivity.

The tert-butyl ester protected DOTAGA intermediate was

prepared in an analogous manner to that of Eisenwiener et al.,³²

except that the stereochemistry was controlled to give the

R-DOTAGA enantiomer.³⁴The pentaﬂuorophenyl ester was

9

J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008 6027

A R T I C L E S

Overoye-Chan et al.

Scheme 2. Conversion of Peptide Tetraamine 6 to EP-2104R

prepared, and this was used to couple to the peptide tetraamine,

Scheme 2. The coupling of the four protected DOTAGA

molecules to the peptide was carried out in DMF, and the

reaction was monitored by HPLC. It is important to control the

stoichiometry at this stage to ensure full conversion to the

tetramer but at the same time minimize overacylation which

results in a pentamer that is very difﬁcult to separate from the

desired product. The tert-butyl groups are then hydrolyzed, and

the crude peptide-(DOTAGA)₄ligand is chelated with GdCl₃

at pH 6.5. A slight excess of Gd³⁺is used to ensure complete

chelation. The excess Gd³⁺is scavenged with EDTA, and the

crude product is then puriﬁed by preparative reversed phase

HPLC to give EP-2104R as the sodium salt, Scheme 2.

Direct infusion of EP-2104R for high resolution ESI-TOF

MS showed a loss of the coordinated water molecules but gave

the correct mass for the sodium adducts [M + Na + H]²⁺and

[M + 2Na]²⁺. LC-MS analysis of EP-2104R showed a single

peak with the expected average mass minus the coordinated

water molecules and the sodium ions which are likely dissoci-

ated during the chromatography and replaced by ammonium

counterions. Thus the molecular weight seen in the LC-MS

method reﬂects EP-2104R, minus four coordinated water

molecules, and four ammonium ions. Reduction of EP-2104R

with tris(2-carboxyethyl)-phosphine (TCEP) resulted in a reten-

tion time shift of 1.3 min and mass change of +2, indicating

the presence of the disulﬁde bond (Supporting Information

Figure S1). IR spectroscopy was consistent with the structure

with absorptions characteristic of peptide bonds. The UV

spectrum and extinction coefﬁcients were consistent with the

presence of 3 tyrosine residues and the xylene moiety. Enan-

tiomeric purity was estimated based on chiral GC-MS,³⁵and

these results indicated no signiﬁcant epimerization or racem-

ization (<0.5%).

2.2. NMR Characterization of Lanthanum Analogue

EP-2104R-La. Since EP-2104R is a potent NMR relaxation

agent, structural characterization by NMR was impossible.

Instead, the diamagnetic lanthanum(III) analogue was synthe-

sized. EP-2104R-La shows similar LC-MS behavior to that of

EP-2104R but with the expected 64 lower mass unit shift. Figure

1 shows the 1D proton NMR spectra of EP-2104R (A) and EP-

2104R-La (B). While the Gd containing compound shows a

broad unresolved spectrum due to enhanced relaxation, the

(32) Eisenwiener, K. P.; Powell, P.; Macke, H. R. Bioorg. Med. Chem.

Lett. 2000, 10, 2133–5.

(33) Laznicek, M.; Laznickova, A.; Macke, H. R.; Eisenwiener, K.; Reubi,

J. C.; Wenger, S. Cancer Biother. Radiopharm. 2002, 17, 527–33.

(34) Amedio, J., C.; Caravan, P.; Jacques, V.; Zhou, K. L.; Levy, S.;

Kalageropoulos, S.; Greenﬁeld, M. Patent Application: WO 2005/

001415, 2005.

(35) Ermer, J.; Gerhardt, J.; Siewert, M. Arch. Pharm. (Weinheim, Ger)

1995, 328, 635–639.

9

6028 J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008

EP-2104R: Thrombus Speciﬁc MRI Contrast Agent

A R T I C L E S

Chart 1. Compounds Discussed in This Article

diamagnetic La analogue yields a well resolved spectrum

amenable to further characterization by 2D NMR spectroscopy.

Assignment of the proton resonances was obtained from 2D

¹H-¹H correlation spectra measured at a 400 MHz proton

frequency and are listed in Supporting Information Table S2.

Figure 2 shows the H_Nregion of the COSY (A) and TOCSY

(B) spectra obtained in H₂O/D₂O (5:1) phosphate buffer

solution.³⁶The individual amino acids comprising the peptide

part of EP-2104R-La are readily identiﬁed by their characteristic

spin coupling pattern observed in COSY and TOCSY spectra.³⁷

The p-xylenediamine (PXD) moiety and the two linker mol-

ecules, 2-diaminobutyric acid (DAB), could also be identiﬁed

by their H_N-side chain couplings. Only the metal chelating

DOTAGA (DTG) moieties could not be site speciﬁcally

assigned due to spectral overlap. Carbon atoms directly bound

presence of various NOE cross peaks between the two cysteine

residues further corroborates the cyclic nature of the peptide.

The observation of long-range NOE peaks is only possible

for systems with a well-deﬁned tertiary structure. The presence

of a large number of long-range connectivities (Figure 2C)

therefore indicates that EP-2104R-La exhibits only one dominant

conformation in solution. This enables further evaluation by

NMR and molecular modeling revealing the 3D structure of

the peptide in solution which will be reported elsewhere.

2.3. Kinetic Inertness and Thermodynamic Stability. For MRI

contrast agents to be used clinically, it is imperative that the

gadolinium remains chelated and be completely eliminated from

the body. Dissociated gadolinium has been linked to a condition

known as nephrogenic systemic ﬁbrosis³⁸(NSF), and this has

recently prompted the U.S. Food and Drug Administration to

issue a public health advisory regarding gadolinium-containing

contrast agents and a possible link to the development of NSF.

While it is necessary to employ a chelate with high thermody-

namic stability, kinetic inertness is probably more important.

For this reason the substituted DOTA chelator termed DOTAGA

was used. In this way EP-2104R is an improvement over an

earlier analogue, EP-1873,²⁷which contains the more labile

DTPA chelator (Vide infra). The [Gd(DOTA)]^-complex has a

very large formation constant³⁹(log K_ML) 25.3) and is quite

inert to gadolinium loss⁴⁰or transmetalation.^41,42It was expected

1

to protons were assigned on the basis of 2D H-¹³C HSQC

experiments (Table S1). The nomenclature used for the assign-

ment is outlined in Figure 1C.

Through space contacts between protons of adjacent amino

1

acids were observed in 2D H-¹H NOESY spectra measured

at various mixing times between 80 and 300 ms. Figure 2C

shows the NOESY spectrum at the 300 ms mixing time.

Sequential neighboring amino acids are identiﬁed from observa-

tion of sequential NOE connectivities within H_N-H_N,

H_N-H_R,ꢀ,γ, or possibly side chain-side chain correlations.³⁷

Along those lines, more than 110 sequential inter-residue

contacts were identiﬁed which conﬁrm the proposed sequence

of EP-2104R-La as that shown in Figure 1. In addition, the

(38) Grobner, T.; Prischl, C. Kidney Int. 2007, 72, 260–264.

(39) Kumar, K.; Chang, C. A.; Francesconi, L. C.; Dischino, D. D.; Malley,

M. F.; Gougoutas, J. Z.; Tweedle, M. F. Inorg. Chem. 1994, 33, 3567–

75.

(36) Cavanagh, J.; Fairbrother, W. J.; Palmer, A. G.; Skelton, N. J. Protein

NMR Spectroscopy: Principles and Practice; Academic Press: San

Diego, 1996.

(37) Wüthrich, K. NMR of Proteins and Nucleic Acids; John Wiley & Sons:

New York, 1986.

(40) Pulukkody, K. P.; Norman, T. J.; Parker, D.; Royle, L.; Broan, C. J.

J. Chem. Soc., Perkin Trans. 2 1993, 605, 20.

(41) Laurent, S.; Elst, L. V.; Muller, R. N. Contrast Media Mol. Imaging

2006, 1, 128–37.

9

J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008 6029

A R T I C L E S

Overoye-Chan et al.

Figure 1. (A) Proton NMR spectrum of EP-2104R (ca. 2 mM in D₂O) and (B) EP-2104R-La (ca. 12 mM in H₂O/D₂O (5:1)) at 400 MHz and 300 K. Both

spectra have been measured using presaturation for water suppression. The “/” denotes the residual water signals. The nomenclature and numbering scheme

used for the assignment are shown in (C).

that EP-2104R would have a similar stability and inertness with

respect to Gd loss.

compound were pH 3 at 60 °C for 4 days in a citric acid buffer.

EP-2104R was challenged with 4 equiv of MS-325-L (1:1 in

metal binder) at pH 3, pH 4, and pH 5. Figure 3 shows that

transmetalation is acid catalyzed; however very little Gd is lost

from EP-2104R. After 4 days, 9.6% of Gd is removed at pH 3,

1.7%, at pH 4, and 0.27%, at pH 5. It is clear from Figure 3

that the system has not equilibrated after 4 days at 60 °C. If

one assumes that the protonation constants and stability constant

for the EP-2104R system are the same for [Gd(DOTA)]^-, then

one predicts that 60% of the Gd should be in the form of MS-

325 at pH 3 at equilibrium.

Similar studies were performed with the macrocyclic complex

[Gd(HP-DO3A)] (Prohance) and the acyclic complexes [Gd-

(DTPA)]^2-(Magnevist) and [Gd(DTPA-BMA)] (Omniscan) at

pH 3 and pH 5 at 60 °C. The structures of these commercial

agents are shown in Chart 1. For the acyclic ligands, the

transmetalation reaction was complete within 10 min at both

pH values. The equilibrium result was very close to that

predicted based on reported stability constants measured at 25

°C. For example at pH 5, [Gd(DTPA)]^2-lost 71.7% of its Gd

compared to the 71.1% predicted,^43,44and [Gd(DTPA-BMA)]

lost 88.4% of its Gd compared to the 86.6% predicted.^43,45The

macrocyclic complex [Gd(HP-DO3A)] was more inert. Figure

The presence of the peptide obviates the ability to measure

thermodynamic stability and kinetic inertness as in other reports.

To determine the stability constant, one needs to know the

protonation constants of the ligand. However, the peptide is

unstable at the high pH required to measure the protonation

constants of the macrocyclic amine nitrogens. For inertness,

some investigators measured gadolinium dissociation in strongly

acidic media.⁴⁰Again, at this extreme pH, the peptide is unstable

to degradation. In order to get a measure of the relative stability

and inertness of EP-2104R, we instead subjected EP-2104R to

a challenge with a substituted DTPA ligand and compared the

results to those obtained with some commercial MRI contrast

agents. The ligand of MS-325 (Chart 1, denoted here as MS-

325-L) was used as a challenge. This molecule has a high

afﬁnity⁴³for Gd (log K_ML) 22.06), and the biphenylcyclohexyl

moiety is a useful chromophore and provides retention on a

reversed phase HPLC column. Transmetalation was measured

by using HPLC to monitor the conversion of MS-325-L to MS-

325 over time. Both MS-325 and MS-325-L were found to be

stable with respect to degradation over all the pH/temperature

combinations employed.

EP-2104R was extremely resistant to transmetalation. The

most forcing conditions identiﬁed that did not degrade the

(44) Martell, A. E.; Smith, R. M.; Motekaitis, R. J. NIST Critically Selected

Stability Constants of Metal Complexes: NIST Standard Reference

Data 46, 3.0 ed.; National Institute of Standards and Technology:

Washington, DC, 1997.

(45) Cacheris, W. P.; Quay, S. C.; Rocklage, S. M. Magn. Reson. Imaging

1990, 8, 467–81.

(42) Laurent, S.; Elst, L. V.; Copoix, F.; Muller, R. N. InVest. Radiol. 2001,

36, 115–22.

(43) Caravan, P.; Comuzzi, C.; Crooks, W.; McMurry, T. J.; Choppin, G. R.;

Woulfe, S. R. Inorg. Chem. 2001, 40, 2170–6.

9

6030 J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008

EP-2104R: Thrombus Speciﬁc MRI Contrast Agent

A R T I C L E S

1

Figure 2. Amide proton region of the 2D H-¹H correlation spectra of EP-2104R-La (ca. 12 mM) at 400 MHz and 300 K in H₂O/D₂O (5:1) buffer. (A)

Double-quantum ﬁltered COSY spectrum, (B) TOCSY spectrum with 80 ms MLEV-17 spin lock, (C) NOESY spectrum at 300 ms mixing time. All spectra

are acquired with a watergate pulse sequence for water suppression.³⁶The assignments in (C) have been omitted for clarity.

Figure 4. Time course of reaction of MS-325-L with either EP-2104R

(b) or [Gd(HP-DO3A)] (2) and conversion to MS-325 at pH 3, citrate

buffer, 60 °C.

Figure 3. Time course of reaction of EP-2104R with 4 equiv MS-325-L

(1:1 in chelator) showing conversion to product MS-325 at 60 °C in citrate

buffer at pH 3 (b), pH 4 (9), or pH 5 (2).

9

J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008 6031

A R T I C L E S

Overoye-Chan et al.

4 shows gadolinium loss as a function of time for [Gd(HP-

DO3A)] and EP-2104R at pH 3. The [Gd(HP-DO3A)] system

has reached equilibrium by 44 h (99.8% Gd loss compared to

96.2% predicted^39,43). At pH 5, [Gd(HP-DO3A)] is more inert

and has not reached equilibrium at 48 h (39.2% loss vs 92.6%

predicted).

These studies demonstrate that EP-2104R is much more inert

to gadolinium transmetalation than commercially available

contrast agents. Figure 4 shows that EP-2104R is signiﬁcantly

more inert than [Gd(HP-DO3A)], which itself has been shown

to be very robust with respect to gadolinium loss under a number

of challenges.^41,42,46Under physiological conditions where the

temperature is lower (37 °C), pH is higher (7.4), and endogenous

chelators are less potent than MS-325-L, it is highly unlikely

that EP-2104R will release gadolinium in vivo.

Figure 5. (Left) EP-2104R binding to human ﬁbrin in the presence of

human plasma in two assay formats: a plasma derived clot (open circles)

and a plate-based assay using puriﬁed ﬁbrin (ﬁlled symbols). Lines are ﬁts

to the data as described in the text. (Right) Data from plate assay plotted

on a linear scale to highlight saturation of binding sites at ∼2 sites per

ﬁbrin monomer.

2.4. Binding of EP-2104R to Human Fibrin. Fibrin binding

localizes EP-2104R to the site of clotting and thus is critical to

the EP-2104R mechanism of action. Fibrin is an insoluble

polymer formed during the clotting cascade by thrombin-induced

polymerization of a soluble ﬁbrinogen monomer, followed by

Factor XIII mediated cross-linking. In these in Vitro studies,

ﬁbrin is generated by adding thrombin and calcium to ﬁbrinogen

and allowing the solution to polymerize to a gel state. Here,

ﬁbrin concentrations refer to that of the polymerized monomeric

ﬁbrinogen unit.

Binding of EP-2104R to human ﬁbrin in plasma was

measured by equilibrium binding at 37 °C in two assay formats:

with a plasma-derived clot or with puriﬁed ﬁbrin. In the plasma

clot assay, known concentrations of EP-2104R were incubated

in human plasma, and thrombin was added to induce clot

formation. The clot was separated from the serum, and the

concentration of EP-2104R in the serum ([EP-2104R]_free) was

determined by ICP-MS analysis; the concentration of ﬁbrin-

bound EP-2104R in the clot was calculated by subtraction ([EP-

2104R]_bound) [EP-2104R]_total- [EP-2104R]_free). The plasma

used contained a functional ﬁbrinogen concentration of 2.3 mg/

mL, within the range of normal plasma, 2.0–3.0 mg/mL.⁴⁷This

value was used to calculate the moles of EP-2104R bound per

mole of ﬁbrin for each data point.

Table 1. Comparison of EP-2104R Binding to Different Animal

Fibrins Measured Either in the Species’ Plasma, in TBS, or in

4.5% HSA^a

ﬁbrin source

medium

K_d(µM)

N binding sites

human

pig

dog

rabbit

rat

TBS

1.7 ( 0.5

1.6 ( 0.2

4.9 ( 0.9

2.2 ( 0.3

2.4 ( 0.3

1.8 ( 0.1

1.5 ( 0.3

2.2 ( 0.1

2.2 ( 0.2

1.7 ( 0.4

1.7 ( 0.1

1.6 ( 0.05

2.0 ( 0.1

1.7 ( 0.1

4.5% HSA in TBS

human plasma

pig plasma

dog plasma

rabbit plasma

rat plasma

mouse

mouse plasma

^aRelaxivity has units of mM^–1s^–1. Binding parameters derived from

best-ﬁts to data ([EP-2104R]_boundvs [EP-2104R]_free) using eq 1.

electron microscopy and that ﬁbrin speciﬁc antibodies, but not

ﬁbrinogen speciﬁc antibodies, react with dried ﬁbrin.⁴⁸

For both assays, the data were ﬁt to a model of N equivalent

binding sites with the dissociation constant, K_d, with units of

µM, given by eq 1:⁵¹

[EP-2104R]_bound

[fbn]_total

N × [EP-2104R]_free

)

(1)

([EP-2104R]_free+ K_d)

In the second assay, puriﬁed ﬁbrinogen (2.5 mg/mL; 7 µM

ﬁbrin) was clotted with thrombin and dried to a thin ﬁlm in

wells of a microtiter plate. The resulting clots bind to the plate

without loss of protein. The ﬁbrin is condensed into a thin disk

that is rapidly penetrated by EP-2104R and other molecules of

similar molecular weight (full equilibration occurs within 30

min) and is without the diffusion limitations posed by other

binding assays. The clots were rehydrated with plasma contain-

ing known concentrations of EP-2104R dissolved in human

citrated plasma. After incubation at 37 °C, the concentration of

EP-2104R in the supernatant ([EP-2104R]_free) was determined

by ICP-MS. The contents of the media or the ﬁbrin structure

can be further modiﬁed to test the effects of speciﬁc proteins,

buffer conditions, or ﬁbrin organization on ﬁbrin binding. Dried

ﬁbrin prepared by this procedure is comparable in structure to

freshly polymerized ﬁbrin. SDS-PAGE of ﬁbrin prepared by

this procedure indicated that the ﬁbrin is fully cross-linked by

factor XIII at the γ subunits. Studies from other laboratories

have shown that the dried and rehydrated ﬁbrils bear a similar

morphology to freshly polymerized ﬁbrin when examined by

where the total ﬁbrin (fbn) concentration is the concentration

of polymerized ﬁbrinogen monomer in µM. The data with ﬁtted

curves are shown in Figure 5. These two different methods

(plasma clot vs dried ﬁbrin assay) yield comparable results, in

both the estimated number of binding sites (1.9 ( 0.1 vs 1.7 (

0.1) and the K_d(0.9 ( 0.3 vs 1.6 ( 0.2), respectively.

Figure 5 shows binding of EP-2104R in the presence of

human plasma. The ﬁbrin plate assay was also run with EP-

2104R in buffer or EP-2104R in 4.5% HSA solution, and very

similar dissociation constants were obtained, Table 1. The

congruence of results obtained in the presence and absence of

plasma proteins indicates that EP-2104R has a high degree of

binding speciﬁcity for ﬁbrin over proteins in plasma, particularly

serum albumin. This property is especially important, as binding

to plasma proteins would reduce binding to the target, increase

background MRI enhancement by the agent, and slow clearance

of nonclot associated compounds from the vasculature (which

also contributes to signal background).

The ∼15% difference between an exact 2.0 sites per ﬁbrin

monomer and the observed data in the ﬁbrin plate assay may

(46) Kumar, K.; Chang, C. A.; Tweedle, M. F. Inorg. Chem. 1993, 32,

587–93.

(48) Tymkewycz, P. M.; Creighton-Kempsford, L. J.; Hockley, D.; Gaffney,

(47) Redman, C. M.; Xia, H. Ann. N.Y. Acad. Sci. 2000, 936, 480–95.

P. J. Thromb. Haemost. 1992, 68, 48–53.

9

6032 J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008

EP-2104R: Thrombus Speciﬁc MRI Contrast Agent

A R T I C L E S

Table 2. Relaxivities for EP-2104R in Various Media Measured at 0 < [EP-2104R] < 40 µM, 37 °C, pH 7.4 and Expressed on a per

Gadolinium and per Molecule Basis^a

r₁(20 MHz)

r₁(60 MHz)

r₂(60 MHz)

medium

per Gd

per molecule

per Gd

per molecule

per Gd

per molecule

TBS

Fgn/TBS

Fgn/plasma

fbn/TBS

11.1 (0.7)

12.4 (0.8)

11.2 (0.7)

24.8 (0.7)

24.9 (1.6)

3.4

44.4 (2.8)

49.4 (3.2)

44.8 (2.9)

99.3 (2.8)

99.7 (6.5)

3.4

10.1 (0.1)

11.1 (0.7)

10.6 (0.7)

17.9 (0.3)

17.8 (1.2)

2.9^b

40.3 (0.3)

44.2 (2.9)

42.2 (2.7)

71.4 (1.2)

71.2 (4.6)

2.9^b

12.8 (0.2)

14.4 (0.9)

12.4 (0.8)

32.1 (2.1)

27.7 (1.8)

3.2^b

51.1 (0.6)

57.4 (3.7)

49.6 (3.2)

128.2 (8.3)

110.7 (7.2)

3.2^b

clotted plasma

[Gd(DOTA)]^-

^aRelaxivity has units of mM^-1s^-1. The number in parentheses corresponds to one standard deviation. Relaxivities for [Gd(DOTA)]^-in water at

37 °C are provided for reference.^{56 b}measured at 64 MHz.

be attributed to several sources. There is some uncertainty as

to the functional protein concentration. An extinction coefﬁcient

determined from the literature for pure ﬁbrin monomer was used

which may not correspond exactly to the commercial prepara-

tions used in this study. Reduction in site accessibility may have

occurred at the protein-plate interface. A small percentage of

the ﬁbrin molecule in the preparation may also be misfolded

and not competent to bind. Nonetheless, the availability of two

sites per ﬁbrin monomer is consistent with the 2-fold symmetry

of the ﬁbrinogen molecule, an R₂ꢀ₂γ₂oligomer of three protein

chains. The ability to bind to two sites on each ﬁbrin molecule

effectively doubles the potential concentration of EP-2104R that

may localize in a clot, greatly enhancing the signal that may be

obtained with this agent.

In order to address the question of whether the binding sites

were equivalent and independent, the data were also ﬁt to a

more general stoichiometric model of binding^49,50where the

binding sites are not assumed to be equivalent. This ﬁt returned

binding constants that were statistically equivalent indicating

that there is no cooperativity in the binding of EP-2104R binding

to ﬁbrin and that the two sites do in fact behave independently.⁵¹

The ﬁbrin concentration in thrombi is expected to be rather

high since the circulating ﬁbrinogen concentration is ∼7 µM.

During the clotting process, the ﬁbrin concentration at the clot

increases further by ongoing ﬁbrin formation as well as clot

retraction, resulting in 10s to 100s µM ﬁbrin. The low

micromolar afﬁnity of EP-2104R coupled with the high micro-

molar concentration of binding sites in thrombi assures ap-

preciable binding of EP-2104R to clots as has been reported in

various animal models.

The peptide portion of EP-2104R (without the 4 gadolinium

chelates) was also assessed for ﬁbrin afﬁnity. The afﬁnity of

EP-2104R was 2.5 times lower than that of the parent peptide.

This reduction in afﬁnity may be expected given the bulk of

the 4 GdDOTAGA moieties (∼75% of the total EP-2104R

molecular weight). However the afﬁnity loss is not dramatic

suggesting that C- and N-terminal substitution is well tolerated.

2.5. Species Dependence on Fibrin Binding. The dried ﬁbrin

assay was also utilized to determine the afﬁnity of EP-2104R

for ﬁbrin from other species. In a similar manner, puriﬁed

ﬁbrinogen from pig, dog, rabbit, rat, or mouse plasma was used

to create ﬁbrin plates. The binding of EP-2104R to these ﬁbrins

was then assessed in the presence of that species’ blood plasma.

The data are reported in Table 1. There was no strong species

dependence on afﬁnity. All species showed low micromolar

afﬁnity to two equivalent binding sites, indicating that critical

residues in the EP-2104R binding site are largely conserved

between species.

2.6. Relaxivity of EP-2104R in Different Media. The efﬁciency

of contrast agents in terms of their ability to enhance the

solvent–water relaxation rate is deﬁned as relaxivity, r₁, which

is the degree to which a given concentration can change 1/T₁:

∆(1/T₁)

r₁)

(2)

[Gd]

A higher relaxivity will enable detection of the agent at lower

concentrations. Relaxivity is dependent on the applied magnetic

ﬁeld and on many molecular parameters. Because of this,

relaxivity measurements were carried out at several ﬁeld

strengths in different media. The proton relaxivity of EP-2104R

was measured at 20 and 60 MHz in Tris buffered saline (TBS),

in the presence of ﬁbrinogen, in a ﬁbrin gel, in human plasma,

and in clotted human plasma at 37 °C. The results are

summarized in Table 2. Data were collected at 20 MHz because

there is a large body of relaxivity data at 20 MHz with which

to compare. However 60 MHz is closer to the common clinical

imaging frequency of 64 MHz (1.5 T) used on the majority of

clinical imagers. The data in Table 2 are reported on a per

gadolinium basis to better compare with other compounds and

on a per molecule basis to better relate T₁change associated

with ﬁbrin binding.

The per Gd relaxivity in TBS is about 3-fold higher for EP-

2104R than for [Gd(DOTA)]^-. This is attributed to the larger

size of EP-2104R which results in a longer rotational correlation

time. Increasing the correlation time is an established method

to increase relaxivity.^15,52,53The correlation time can also be

increased by protein binding. Lauffer⁵⁴proposed noncovalent

binding of contrast agents to target proteins as a means of

increasing relaxivity many years ago, and this strategy was

successfully employed with MS-325 targeting serum albumin^49,55

and with numerous compounds since. In the case of ﬁbrin

binding, the relaxivity of EP-2104R more than doubles at 20

MHz and sees an 80% increase at 60 MHz. The relaxivity is

similar when measured in either ﬁbrin gel or in a plasma clot.

(56) Rohrer, M.; Bauer, H.; Mintorovitch, J.; Requardt, M.; Weinmann,

H. J. InVest. Radiol. 2005, 40, 715–24.

(49) Caravan, P.; Cloutier, N. J.; Greenﬁeld, M. T.; McDermid, S. A.;

Dunham, S. U.; Bulte, J. W.; Amedio, J. C., Jr.; Looby, R. J.;

Supkowski, R. M.; Horrocks, W. D., Jr.; McMurry, T. J.; Lauffer,

R. B. J. Am. Chem. Soc. 2002, 124, 3152–62.

(52) Caravan, P.; Ellison, J. J.; McMurry, T. J.; Lauffer, R. B. Chem. ReV.

1999, 99, 2293–352.

(53) Lauffer, R. B. Chem. ReV. 1987, 87, 901–927.

(54) Lauffer, R. B. Magn. Reson. Med. 1991, 22, 339.

(55) Lauffer, R. B.; Parmelee, D. J.; Dunham, S. U.; Ouellet, H. S.; Dolan,

R. P.; Witte, S.; McMurry, T. J.; Walovitch, R. C. Radiology 1998,

207, 529–538.

(50) Klotz, I. M. Ligand Receptor Energetics - A Guide for the Perplexed;

John Wiley & Sons: New York, 1987.

(51) Koshland, D. E., Jr.; Nemethy, G.; Filmer, D. Biochemistry 1966, 5,

365–85.

9

J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008 6033

A R T I C L E S

Overoye-Chan et al.

[Gd(DTPA)]^-was unaffected by the presence of the 1% ﬁbrin

gel, suggesting that the increased relaxivity observed is due to

ﬁbrin binding.

2.7. Speciﬁcity of EP-2104R for Fibrin. For effective imaging

of thrombi, which by deﬁnition reside in blood vessels, it is

important that the imaging agent displays speciﬁcity for ﬁbrin

over blood plasma proteins. Of particular concern is ﬁbrinogen

(Fgn) which obviously shares a great deal of structural homology

with ﬁbrin and is present at a circulating concentration of ∼7.5

µM. The peptide portion of EP-2104R is based on a sequence

identiﬁed by phage display against ﬁbrin,⁶⁰and in that selection

the phage library was ﬁrst prescreened against ﬁbrinogen to

exclude ﬁbrinogen binders. The NMRD data shown in Figure

6 suggests that there is not a large fraction of EP-2104R bound

to ﬁbrinogen, otherwise one would have expected the relaxivity

to be enhanced as in the ﬁbrin gel.

The binding of EP-2104R to ﬁbrinogen was assessed with

two techniques. We used isothermal titration calorimetry to

measure the heat change that occurs when a ligand binds to a

protein target. No signiﬁcant heat change was detected for the

titration of 20 mM EP-2104R with 59.7 µM Fgn, indicating

that the interaction between EP-2104R and ﬁbronogen is weak.

From the calorimetry studies, a lower limit of a dissociation

constant could be estimated, K_d> 100 µM for the binding of

EP-2104R to ﬁbrinogen.

Figure 6. Nuclear magnetic relaxation dispersions (NMRD) displayed as

observed relaxivity per gadolinium of 50 µM EP-2104R in Tris buffered

saline (2), human blood plasma (4), 30 µM ﬁbrinogen solution (O), or 30

µM gelled ﬁbrin (b) at 35 °C.

On the other hand, if relaxivity is measured in ﬁbrinogen

solution or in unclotted plasma, the values are closer to those

obtained in buffer.

At 60 MHz, the relaxivity of EP-2104R in ﬁbrin is about 6

times higher than that of[Gd(DOTA)] on a per gadolinium basis

and 25 times higher on a per molecule basis. It is this increased

sensitivity that allows for robust detection of thrombi in a variety

of in vivo models.

The proton relaxation enhancement (PRE) effect was also

used to estimate EP-2104R binding to Fgn. In the case of EP-

2104R binding to a single site on ﬁbrinogen, there should be

two species in solution: free (unbound) and ﬁbrinogen bound.

These are related by a dissociation constant, K_d, eq 3.

To gain a better understanding of the ﬁeld dependence on

relaxivity, nuclear magnetic relaxation dispersion (NMRD)

proﬁles were measured at 35 °C. Here relaxation rates are

recorded at different frequencies for TBS, ﬁbrinogen solution,

ﬁbrin gel, or human plasma in the presence or absence of EP-

2104R, and relaxivity at each frequency in each medium is

calculated according to eq 22. The proﬁles are shown in Figure

6. The NMRD proﬁle of EP-2104R in ﬁbrin gel is typical of a

protein bound gadolinium complex. The increasing relaxivity

between 4 and 30 MHz followed by dispersion is indicative of

a slow motion component inducing relaxation, i.e., a slow

[EP-2104R]_f([Fgn]_tot- [EP-2104R]_b)

K_d)

(3)

[EP-2104R]_b

Here, [EP-2104R]_frefers to the concentration of free (un-

bound) EP-2104R, and the bound concentration is denoted [EP-

2104R]_b. The total ﬁbrinogen concentration is denoted [Fgn]_tot.

The bound and free EP-2104R will have different relaxivies,

r_1fand r_1brespectively, and the measured relaxation rate (1/T₁)

will depend on these relaxivities and the concentrations of the

two species, eq 44.

motion arising from docking to the protein.¹⁵Solomon-

52,57–59

Bloembergen-Morgan

theory predicts that as the

correlation time gets longer, the relaxivity should increase in

this high frequency region. The curve should be rather ﬂat at

short correlation times and then take on this peaked appearance

in this 4–60 MHz frequency range as the correlation time gets

longer.¹⁵In principle these proﬁles can be modeled. However

given the structural complexity of EP-2104R it is obvious that

rotational motion cannot be attributed to a single isotropic

correlation time. Moreover it is not clear whether the correlation

time for each gadolinium center would be the same. This leads

to too many assumptions and more parameters than can be

meaningfully ﬁt.

1

T₁

1

) r_1f[EP-2104R]_f+ r_1b[EP-2104R]_b+

(4)

T₁⁰

Here, 1/T₁⁰is the relaxation rate of the solution in the absence

of EP-2104R. Alternately one can rearrange eq 4 to calculate

the observed relaxivity as deﬁned by eq 5.

1

T₁

1

T₁⁰

-

r^o₁^bs

)

) f_fr_1f+ f_br_1b

(5)

[EP-2104R]_tot

For the ﬁbrinogen and plasma solutions, relaxivity is enhanced

slightly at all frequencies. There is also a hint of this peaked

effect at higher frequencies suggesting that there may be a small

fraction of the compound bound to ﬁbrinogen or plasma

proteins.

Here f refers to the fraction of EP-2104R that is either free

(f_f) or bound to ﬁbrinogen (f_b). The concentration of EP-2104R

bound can be calculated from the initial concentrations (totals)

of EP-2104R and ﬁbrinogen if the K_dis known, eq 6:

To verify that the relaxivity increase in the ﬁbrin gel was

not due to some microviscosity effect, the relaxivity of [Gd-

(DTPA)]^2-was also measured in ﬁbrin gel. The relaxivity of

a - a²- 4[Fgn] [EP-2104]

√

t

[EP-2104]_b)

;

2

a ) [Fgn]_t+ [EP-2104]_t+ K_d(6)

(57) Bloembergen, N. J. Chem. Phys. 1957, 27, 572–573.

(58) Bloembergen, N.; Morgan, L. O. J. Chem. Phys. 1961, 34, 842–850.

(59) Solomon, I. Phys. ReV. 1955, 99, 559.

(60) Wescott, C. R.; Nair, S. A.; Kolodziej, A.; Beltzer, J. P. Patent

Application WO2001009188, 2001.

9

6034 J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008

EP-2104R: Thrombus Speciﬁc MRI Contrast Agent

A R T I C L E S

assess plasma protein binding, the interaction of EP-2104R with

the most abundant plasma protein, human serum albumin (HSA),

was investigated. In the dried ﬁbrin binding assay, there was

no discernible difference in EP-2104R binding to ﬁbrin if the

assay was carried out in Tris buffered saline, in 4.5% HSA

solution, or in human plasma indicating that HSA or other

plasma proteins do not compete with ﬁbrin from EP-2104R

binding. Isothermal calorimetry using a 560 µM HSA solution

to titrate a 20 mM EP-2104R solution showed no signiﬁcant

heat changes indicating weak protein binding. Using the PRE

method and a 670 µM HSA solution, a K_dof 4600 ( 460 µM

was estimated for the binding of EP-2104R to serum albumin,

over a 1000-fold lower afﬁnity than that for binding to ﬁbrin.

Figure 7. Effect of increasing amounts of EP-2104R on observed relaxivity

in 30 µM ﬁbrin (b), 30 µM ﬁbrinogen (O), or buffer (2) measured at 20

MHz (left panel). Right panel: same data, plotted as relaxation rate (1/T₁)

at low EP-2104R concentrations to demonstrate the small but signiﬁcant

relaxation enhancement due to weak ﬁbrinogen binding. Solid lines are ﬁts

to the data as described in the text.

3. Conclusions

EP-2104R consists of an 11 amino acid peptide functionalized

with two GdDOTAGA moieties at each of the C- and N-termini.

The peptide imparts high afﬁnity for ﬁbrin, the most concen-

trated component of blood clots, and 100- to 1000-fold higher

selectivity for ﬁbrin compared to blood plasma proteins such

as ﬁbrinogen or serum albumin. The ﬁbrin afﬁnity and selectivity

explain the localization of EP-2104R to thrombi observed in

various animal models^20–26and clinical studies.^28,29The four

GdDOTAGA moieties provide a signiﬁcant relaxation enhance-

ment effect that allows EP-2104R to enhance thrombi in MR

images. The relaxivity of EP-2104R bound to ﬁbrin at 37 °C is

about 25 times higher than that of [Gd(DOTA)]^-at the clinical

ﬁeld strength of 1.5 T. This combination of strong ﬁbrin binding,

ﬁbrin selectivity, and high molecular relaxivity is why EP-2104R

appears effective at detecting blood clots in vivo.

The relaxivity of the unbound fraction is known from

measurements in the absence of protein, Table 2. It is assumed

that the relaxivity of the ﬁbrinogen-bound fraction is the same

as the relaxivity for EP-2104R bound to ﬁbrin, Table 2. One

can iteratively vary K_duntil the calculated and observed

relaxation rates or relaxivities agree. In principle, the relaxivity

of ﬁbrinogen-bound EP-2104R can also be obtained from ﬁtting

the data; however for weak binding K_dand r_1bare correlated,

so we make the reasonable assumption that r_1bis the same for

ﬁbrinogen and ﬁbrin binding.

A series of solutions of EP-2104R (0–300 µM) were prepared

at a ﬁxed 30 µM ﬁbrinogen concentration. T₁was measured at

both 20 and 60 MHz before and after polymerizing the

ﬁbrinogen to ﬁbrin via addition of thrombin and Ca²⁺. The left

panel of Figure 7 shows the results of the experiments at 20

MHz plotted as observed relaxivity (r₁^obs) versus total EP-2104R

concentration, along with measurements made in the absence

of protein. The solid line through the ﬁbrin relaxivity data in

Figure 7 is calculated based on the binding data in Table 1.

The relaxivity data are in accord with the equilibrium binding

studies. Note that after the two ﬁbrin binding sites are saturated,

the relaxivity decreases and begins to approach r_1fat very high

concentrations.

4. Experimental Section

4.1. Materials. Human ﬁbrinogen, grade L (Lots 18 and 21)

was purchased from American Diagnostica. Citrated human plasma

(Product Code 19201) was received from Red Cross, Danvers, MA.

The buffer used in these studies was 50 mM tris(hydroxy-

methyl)aminomethane (Tris) buffered saline (TBS). The thrombin

was obtained from Haematologic Technologies, diluted 1:100 into

TBS, and frozen as aliquots. These aliquots had activities of 0.3–0.4

NIH Units/µL. Calcium chloride solution (2 M) was prepared in

water from reagent grade material. The commercial contrast agents

were obtained as 0.5 M formulated solutions: [Gd(DTPA)]^2-(with

N-methylglucamine as the counterion, Magnevist, Berlex, Wayne,

NJ), [Gd(HP-DO3A)] (Prohance, Bracco, Princeton, NJ), and

[Gd(DTPA-BMA)] (Omniscan, GE Healthcare, Princeton, NJ).

Citric acid (Sigma) and sodium citrate (JT Baker) were USP grade.

The derivatized DTPA ligand for MS-325, denoted here as MS-

325-L, was synthesized as described previously.⁶¹High resolution

ESI-TOF mass spectra were recorded at Tandem Laboratories

(Woburn, MA).

4.2. Synthesis of EP-2104R. 4.2.1. tert-Butyl 4-(amino-

methyl)benzylcarbamate, 1. p-Xylylenediamine (75.0 g, 550.6

mmol) was dissolved in dioxane (1 L) at room temperature. Di-

tert-butyldicarbonate (60.0 g, 275.3 mmol) was dissolved in dioxane

(0.5 L) and added dropwise to the solution over 5 h at room

temperature. A solid precipitated, was ﬁltered, and washed with

dioxane (200 mL). The ﬁltrate was concentrated to a volume of

400 mL under reduced pressure. The solution was then poured into

water (1400 mL). A precipitate formed, was ﬁltered, and washed

with water (200 mL). The ﬁltrate was then extracted with ethyl

acetate (4 × 500 mL). The combined extracts were dried over

There is a small but signiﬁcant increase in r₁^obsfor EP-2104R

in ﬁbrinogen relative to buffer-only. In the left panel of Figure

7, the uncertainties in each individual relaxivity data point are

large because at low EP-2104R concentrations the difference

between 1/T₁and 1/T₁⁰is small compared to the uncertainty in

T₁(5%). This is compounded by dividing this difference by

the small EP-2104R concentration (estimated at (3%) to obtain

r₁^obs. However the measured 1/T₁data are more compelling and

are shown in the right panel of Figure 7 with the same error

bars ((5% in T₁, (3% in concn). The ﬁbrinogen 1/T₁data

consistently lie above the dashed line denoting the slope

(relaxivity) of the 1/T₁data in the absence of protein. Fitting

the ﬁbrinogen data (solid line) as described above resulted in

K_d) 240 ( 30 µM for EP-2104R binding to ﬁbrinogen. EP-

2104R displays a 100-fold higher selectivity for ﬁbrin than for

ﬁbrinogen. Although there are two sites on ﬁbrin for EP-2104R,

we assumed the simplest binding model for the ﬁbrinogen

interaction (one site) since binding to ﬁbrinogen is so weak

(e11% bound with [Fgn] at 4 times physiological levels).

Protein solubility limits the degree of site occupancy that can

be achieved.

(61) McMurry, T. J.; Parmelee, D. J.; Sajiki, H.; Scott, D. M.; Ouellet,

H. S.; Walovitch, R. C.; Tyeklar, Z.; Dumas, S.; Bernard, P.; Nadler,

S.; Midelfort, K.; Greenﬁeld, M.; Troughton, J.; Lauffer, R. B. J. Med.

Chem. 2002, 45, 3465–74.

The NMRD data in human plasma, Figure 6, were not

consistent with strong binding to plasma proteins. To further

9

J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008 6035

A R T I C L E S

Overoye-Chan et al.

sodium sulfate and concentrated under reduced pressure to give 1

(32.7 g, 50% yield) as a pale beige solid. ¹H NMR (300 MHz,

DMSO-d₆): 7.28 (d, 2H, ³J ) 8 Hz), 7.18 (d, 2H, ³J ) 8 Hz), 4.09

with water (1 L), and dried under reduced pressure to give 5 (103.9

g). This compound was not puriﬁed further and used crude in the

1

3

next step. H NMR (300 MHz, DMSO-d₆): 8.36 (t, 1H, J ) 6

3

(d, 2H, J ) 6 Hz), 3.75 (s, 2H), 1.38 (s, 9H). ¹³C NMR: 156.2,

Hz), 8.19 (t, 1H, J ) 5 Hz), 7.89 (d, 2H, J ) 7.5 Hz), 7.74 (d,

3

141.0, 139.0 127.6, 127.2, 78.0, 45.0, 43.5, 28.6. High resolution

ESI-TOF-MS: m/z expected [C₁₃H₂₀N₂O₂+ H]⁺237.1604, found

237.1392.

2H, J ) 7 Hz), 7.59–7.27 (m, 5H), 7.19 (s, 4H), 6.88 (d, 1H, J

) 8 Hz), 6.68 (m, 1H), 4.26 (m, 5H), 4.06 (m, 2H), 3.95 (m, 1H),

2.96 (m, 2H), 2.27 (m, 2H), 1.95 (m, 1H), 1.77 (m, 2H), 1.65 (m,

1H), 1.37 (s, 18H). ¹³C NMR: 174.2, 172.2, 171.8, 156.3, 155.7,

144.2, 141.1, 138.2, 128.0, 127.4, 125.7, 120.5, 78.5, 77.9, 66.1,

54.5, 47.1, 42.2, 37.74, 30.7, 28.6, 28.5, 27.7. High resolution ESI-

TOF-MS: m/z expected [C₄₂H₅₃N₅O₁₀+ H]⁺788.3871, found

788.3641.

4.2.2. 1-Fmoc-Glu(tBu)-tert-butyl-4-(aminomethyl)benzyl-

carbamate, 2. 1 (32.0 g, 135.4 mmol) was dissolved in dimeth-

ylformamide (300 mL). Fmoc-Glu(tBu)-OH (54.7 g, 128.6 mmol),

EDC (24.6 g, 128.6 mmol), and 1-hydroxybenzotriazole (17.4 g

128.6 mmol) were dissolved acetonitrile (1.2 L) and stirred for 15

min at room temperature. The amino acid solution was added to

the amine solution, and the reaction was stirred for 2 h at room

temperature. The reaction mixture was then concentrated to a

volume of 300 mL under reduced pressure and poured into water

(1.7 L). The resulting solid was ﬁltered, washed with water (600

mL), and dried under reduced pressure to give 2 (96.59 g). This

compound was not puriﬁed further and used crude in the next step.

¹H NMR (300 MHz, DMSO-d₆): 8.34 (t, 1H, ³J ) 6 Hz), 7.89 (d,

4.2.6. Fmoc Glu(Rink Amide MBHA resin)-1-{4-

[Boc-Dab(Boc)-methyl]benzylamide, 6. Rink Amide MBHA resin

(31.25 g, 20 mmol, 0.64 mmol/g) was washed with dimethylfor-

mamide (2 × 300 mL). The resin was suspended in 20% piperidine

in dimethylformamide (300 mL) and shaken for 1 h. The resin was

then ﬁltered, washed with dimethylformamide (2 × 300 mL),

dichloromethane (2 × 300 mL), and dimethylformamide (2 × 300

mL). 5 (31.5 g, 40 mmol) and 1-hydroxybenzotriazole (5.4 g, 40

mmol) were dissolved in dimethylformamide (300 mL) and added

to the resin. 1,3-Diisopropylcarbodiimide (6.19 mL, 40 mmol) was

added, and the reaction was shaken for 2 days at room temperature.

The reaction was then ﬁltered and washed with dimethylformamide

(1 × 300 mL), dichloromethane (2 × 300 mL), and dimethylfor-

mamide (3 × 300 mL). A solution (300 mL) of 5% acetic anhydride

and 6% diisopropylethylamine in dimethylformamide was added,

and the reaction was agitated for 1 h. The resin was then ﬁltered

and washed with dimethylformamide (1 × 300 mL), dichlo-

romethane (2 × 300 mL), and dimethylformamide (3 × 300 mL)

to give 6.

3

2H, J ) 8 Hz), 7.72 (m, 2H), 7.58–7.24 (m, 6H), 7.18 (s, 4H),

3

4.27 (m, 4H), 4.07 (m, 4H), 2.24 (t, 2H, J ) 7.5 Hz), 1.93 (m,

1H), 1.70 (m, 1H), 1.40 (s, 9H), 1.38 (s, 9H). ¹³C NMR: 172.0,

156.7, 144.3, 141.1, 139.2, 138.2, 138.1, 128.0, 127.4, 127.3, 125.7,

120.5, 80.1, 78.1, 66.0, 54.6, 47.0, 43.5, 42.2, 31.9, 28.6, 28.2, 27.7.

High resolution ESI-TOF-MS: m/z expected [C₃₇H₄₅N₃O₇+ H]⁺

644.3336, found 644.3326.

4.2.3. Fmoc-Glu-1,4-bisaminomethylbenzene, 3. 2 (96.0 g) was

dissolved in a solution (1.5 L) of triﬂuoroacetic acid/dichlo-

romethane/triisopropylsilane/water (47.5/50/1.25/1.25). The reaction

was stirred for 1.5 h at room temperature. The reaction was then

concentrated to a volume of 400 mL under reduced pressure and

poured into isopropyl ether (1 L). The resulting solid was ﬁltered,

washed with isopropyl ether (1 L), and dried under reduced pressure

to give 3 (81.9 g). This compound was not puriﬁed further and

used crude in the next step. ¹H NMR (300 MHz, DMSO-d₆): 8.46

4.2.7. Boc-Dab(Boc)-Tyr(tBu)-dGlu(OtBu)-Cys(Trt)-Hyp-

(tBu)-3ClTyr-Gly-Leu-Cys(Trt)-Tyr(tBu)-Ile-Glu(Rink Amide

MBHA resin)-1-{4-[Boc-Dab(Boc)-methyl]-benzylamide, 7. The

following general procedure was used to attach each subsequent

amino acid. The resin was washed with dimethylformamide (2 ×

300 mL). 20% Piperidine in dimethylformamide (300 mL) was

added, and the resin was agitated for 1 h. The resin was then washed

with dimethylformamide (2 × 300 mL), dichloromethane (2 × 300

mL), and dimethylformamide (2 × 300 mL). The amino acid (40

mmol) and 1-hydroxybenzotriazole (5.4 g, 40 mmol) were dissolved

in dimethylformamide (300 mL) and added to the resin. 1,3-

Diisopropylcarbodiimide (6.19 mL) was added, and the reaction

was agitated for 18 h. The resin was then ﬁltered and washed with

dimethylformamide (1 × 300 mL), dichloromethane (2 × 300 mL),

and dimethylformamide (3 × 300 mL). This procedure was repeated

for each amino acid in the following order: Fmoc-Ile-OH, Fmoc-

Tyr(tBu)-OH, Fmoc-Cys(Trt)-OH, Fmoc-Leu-OH, Fmoc-Gly OH,

Fmoc-Tyr(3-Cl)-OH, Fmoc-Hyp(tBu)-OH, Fmoc-Cys(Trt)-OH, Fmoc-

D-Glu(OtBu)-OH, Fmoc-Tyr(tBu)-OH, and Boc-Dab(boc)-OH. The

elongated peptide on the resin was dried under reduced pressure to

give compound 7.

3

(t, 1H, J ) 6 Hz), 8.22 (2H, m), 7.90 (d, 2H, J ) 7.5 Hz), 7.57

3

(d, 1H, J ) 8.0 Hz), 7.48–7.13 (m, 7H), 4.26 (s, 4H), 3.99 (m,

3H), 2.27 (m, 2H), 1.93 (m, 1H), 1.79 (m, 1H). ¹³C NMR: 174.3,

172.0, 156.2, 144.3, 141.6, 140.5, 132.9, 129.2, 128.1, 127.7, 127.4,

125.8, 120.6, 66.1, 54.7, 47.1, 42.4, 42.2, 30.7, 27.6. High resolution

ESI-TOF-MS: m/z expected [C₂₈H₂₉N₃O₅+ H]⁺488.2186, found

488.2170.

4.2.4. Boc-Dab(Boc)-OPfp, 4. Boc-Dab(Boc)-OH ·DCHA (100

g, 200 mmol) was dissolved in dichloromethane (2 L) at room

temperature. Solid pentaﬂuorophenol (36.8 g, 200 mmol) and EDC

(38.2 g, 200 mmol) were added. The reaction was stirred at room

temperature for 2 h. The resulting solid was ﬁltered and washed

with dichloromethane (200 mL). The ﬁltrate was concentrated to a

volume of 1 L under reduced pressure, washed with 0.1 N

hydrochloric acid (2 × 500 mL), washed with water (500 mL),

and washed with brine (500 mL). The organic layer was passed

through a plug of silica gel (230 g) and rinsed with dichloromethane

(1 L). The ﬁltrate was concentrated to a white glassy solid under

reduced pressure to give 4 (86.9 g, 90% yield). ¹H NMR (300 MHz,

4.2.8.

L-2,4-Diamino-N-butyramide-[Tyr-dGlu-Cys-Hyp-

Tyr(3-Cl)-Gly-Leu-Cys-Tyr-Ile-Gln-1-{4-[(L-2,4-diamino-butyr-

ylamino)-methyl]-benzylamide}, 8. 7 (83.18 g) was washed with

dichloromethane (3 × 300 mL). The resin was suspended in a

solution (1.25 L, 15 mL/g of resin) of triﬂuoroacetic acid,

triisopropylsilane, 1-dodecanethiol, dichloromethane, and water (80:

5:5:5:5). The mixture was agitated for 1.25 h. The peptide resin

was ﬁltered and washed with triﬂuoroacetic acid (2 × 100 mL).

The ﬁltrate was poured into isopropyl ether (4 L). The resulting

solid was washed with isopropyl ether (2 × 600 mL) and acetonitrile

(1 × 600 mL). The solid was dried under reduced pressure. The

solid was washed with acetonitrile (2 × 200 mL) and dried under

reduced pressure to give crude 8 (40.03 g, 92% yield). ESI-MS:

m/z 1719.5 [M + H]⁺.

3

DMSO-d₆): 7.45 (d, 1H, J ) 7 Hz), 6.82 (m, 1H), 4.11 (s, 1H),

2.89 (d, 2H, ³J ) 4.5 Hz), 1.70 (m, 2H), 1.17 (s, 9H), 1.14 (s, 9H).

¹³C NMR: 169.7, 155.9, 142.2, 139.3, 136.2, 125.0, 79.1, 78.0,

51.6, 36.9, 30.7, 28.5. ¹⁹F NMR: –149.4 (d, 2F, ³J ) 23 Hz), –154.0

(m, 1F), –158.7 (d, 2F, ³J ) 23 Hz). ESI-MS: m/z 330.2 [M-2Boc

+ 2Na]⁺.

4.2.5. Fmoc-Glu(OH)-1-{4-[Boc-Dab(Boc)-methyl]benzyl-

amide, 5. 3 (81 g, 135 mmol) was dissolved in dimethylformamide

(1 L). The pH was adjusted to 7 with diisopropylethylamine (30

mL, 173 mmol). 4 (62 g, 128 mmol) was added to the reaction,

and the pH was adjusted to 7 with diisopropylethylamine (10 mL,

57.5 mmol). The reaction was stirred for 2 h. The reaction was

monitored by HPLC. The reaction was then poured into water (3.5

L) over 20 min to form a solid. The solid was ﬁltered, washed

4.2.9. L-2,4-Diamino-N-butyramide-[Tyr-dGlu-Cys-Hyp--

Tyr(3-Cl)-Gly-Leu-Cys-Tyr-Ile-Gln cyclic (2 f 8) disulﬁde]-

9

6036 J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008

EP-2104R: Thrombus Speciﬁc MRI Contrast Agent

A R T I C L E S

1-{4-[(L-2,4-diamino-butyrylamino)-methyl]-benzylamide}, 9. 8

(40.03 g) was dissolved in a solution (3150 mL) of dimethylsul-

foxide, acetonitrile, and water (15:10:75). The reaction was stirred

for 24 h at room temperature. The reaction was then diluted with

water (3150 mL). The solution was divided in equal portions of

500 mL. 500 mL of solution were injected directly onto a reversed

prep HPLC column (Kromasil C18, 100 Å, 10 µm, 5 cm × 25 cm)

equilibrated with 95:5 of the following mobile phase (A: 0.1%

triﬂuoroacetic acid in water, B: 0.1% triﬂuoroacetic acid in

acetonitrile). Starting conditions were held for 5 min before

initiating the gradient which was 5% to 20% B over 5 min, then

20% to 35% B over 25 min followed by a 3 min ramp to 95% B

and wash for 10 min followed by re-equilibration at 5% B for 10

min. Flow rate was 100 mL/min and peaks were detected at 220

nm. Fractions with a purity greater than 98.5% purity were

combined and lyophilized to give compound 9 (16.83 g, 97.1%

purity, 34% yield (based on peptide content)). High resolution ESI-

TOF-MS: m/z expected [C₇₈H₁₀₉ClN₁₈O₂₀S₂+ H]⁺1717.7274,

found 1717.6541. ¹H, ¹³C, and ¹⁵N chemical shifts of all residues

were assigned by means of 2D COSY and HSQC spectra at 9.4T

(see Supporting Information Tables S3 and S4 for list of shifts).

NOESY spectra yielded more than 90 contacts between adjacent

amino acids and conﬁrm the peptide sequence as shown in Figure

1C.

4.2.12. L-2,4-[[(1-(1-Carboxy-3-carboxypropyl)-4,7,10-(car-

boxymethyl)-1,4,7,10-tetraazacyclodecane)]-butyramide]-[Tyr-

dGlu-Cys-Hyp-Tyr(3-Cl)-Gly-Leu-Cys-Tyr-Ile-Gln cyclic (2 f

8) disulﬁde]-1-{4-[(L-2,4-)-[[(1-(1-carboxy-3-carboxypropyl)-

4,7,10-(carboxymethyl)-1,4,7,10-tetraazacyclodecane)]methyl-

]benzylamide}, 12. The tert-butyl esters were removed by dis-

solving 11 (30.6 g, 6.9 mmol) in a cocktail of triﬂuoromethanesulfonic

acid, methanesulfonic acid, 1-dodecanethiol, and water (91.6:2.8:

2.8:2.8, 500 mL) at room temperature. The reaction was stirred for

1 h and monitored by HPLC. Cold ethyl acetate (1.25 L) was added,

and a solid precipitated. The mixture was stirred for 15 min. The

solid was ﬁltered, washed with ethyl acetate (2 × 250 mL), and

dried under reduced pressure to give compound 12 (24.4 g, 100%

yield) as an off-white solid. ESI-TOF-MS: m/z expected

[C₁₅₄H₂₂₉ClN₃₄O₅₆S₂+ H + Na]²⁺1786.7589, found 1786.9338,

expected [C₁₅₄H₂₂₉ClN₃₄O₅₆S₂+2Na]²⁺1797.7521, found 1797.9050.

4.2.13. L-2,4-[Gadolinate(1-), aqua[[(1-(1-carboxy-3-carboxy-

propyl)-4,7,10-(carboxymethyl)-1,4,7,10-tetraazacyclodecane),

monosodium]]-butyramide]-[Tyr-dGlu-Cys-Hyp-Tyr(3-Cl)-Gly-

Leu-Cys-Tyr-Ile-Gln cyclic (2 f 8) disulﬁde]-1-{4-[(L-2,4-)-

[Gadolinate(1-), aqua[[(1-(1-carboxy-3-carboxypropyl)-4,7,10-

(carboxymethyl)-1,4,7,10-tetraazacyclodecane), monosodium]]methyl]-

benzylamide}, EP-2104R. The ligand 12 (24.4 g, 6.9 mmol) was

added in portions to a solution of water (500 mL) at room

temperature, while maintaining a pH of 6.5 using 1 N sodium

hydroxide (277 mL total). Gadolinium(III) chloride hexahydrate

(GdCl₃·6H₂O) (9.8 g, 26.4 mmol) was added to the solution, and

the pH was adjusted to 6.5 with 1 N sodium hydroxide (57 mL).

The reaction was stirred for 1 h at room temperature. The reaction

was monitored by HPLC. Residual gadolinium(III) chloride was

neutralized with ethylenediaminetetraacetic acid (EDTA) (10 mL,

1 mmol). The reaction was stirred for 15 min, and the pH was

adjusted to 6.5 with 1 N sodium hydroxide (3 mL). The neutraliza-

tion was monitored by testing with xylenol orange. The product

was isolated from solution and puriﬁed by reversed phase chro-

matography. The solution (850 mL) was divided into ﬁve equal

170 mL portions. Each portion was injected onto the column (C4

Kromasil, 100 Å, 10 µm, 5 cm × 25 cm) with acetonitrile/water

(2:98). Starting conditions were held for 5 min before the gradient

was started which was 2% to 18% acetonitrile (ACN) over 25 min,

then 18% to 50% ACN over 1 min followed by a 7 min ramp to

80% ACN and wash for 3 min followed by re-equilibration at 2%

ACN. Flow rate was 100 mL/min, and peaks were detected at 220

nm. The puriﬁed fractions were analyzed by HPLC. Fractions >

95% purity were combined and lyophilized to give EP-2104R (21.5

g, >98% purity by analytical HPLC (see below for method), 4.28

mmol, 62.1% yield from the peptide) as a white powder. ESI-TOF-

MS: m/z expected [C₁₅₄H₂₁₂ClGd₄Na₅N₃₄O₅₆S₂+ H + Na]²⁺

2151.5173, found 2151.5958, expected [C₁₅₄H₂₁₂ClNa₅N₃₄O₅₆S₂+

2Na]²⁺2162.5082, found 2162.5650. UV: λ (ꢀ) 192 nm (327 000

cm^-1M^-1), 219 nm (60 600 cm^-1M^-1), 277 nm (5700 cm^-1

M^-1). Chiral amino acid analysis³⁵showed all amino acids were

>99.5% in the L form except Glu which was analyzed as 49.7%

D-Glu. This is expected from the D-Glu in the molecule and the

L-Gln which is hydrolyzed to Glu in the analytical method.

4.2.10. 2-(R)-(4,7,10-Tris-tert-butoxycarbonylmethyl-1,4,7,10-

tetraazacyclododec-1-yl)-pentanedioic Acid, 1-tert-Butyl-5-pentaﬂuoro-

phenyl Ester, DOTAGA(OtBu)₄-OPfp, 10. The activated ester

of the protected DOTAGA ligand (10) was synthesized as described

previously.³⁴2-(R)-2-(4,7,10-Tris-tert-butylcarboxymethyl-1,4,7,10-

tetraazacyclododec-1-yl)-pentanedioic acid 1-tert-butyl ester (DOT-

AGA(OtBu)₄-OH) (29.3 g, 29.4 mmol)^32,34and pentaﬂuorophenol

(8.7 g, 47.3 mmol) were dissolved in dichloromethane (500 mL).

Polystyrene-carbodiimide (PS-DCC; 41.2 g, 47.3 mmol; 1.23

mmol/g loading capacity) was added, and the reaction was shaken

on an orbital shaker for 3 h. The reaction was monitored by HPLC.

The resin was ﬁltered and rinsed with methylene chloride (200 mL).

The ﬁltrate was evaporated under reduced pressure. The glassy solid

was dried under reduced pressure to give 10 (30.9 g, 85% yield;

excess pentaﬂuorophenol will not react in the next step). ESI-MS:

m/z 867.3 [M + H]⁺.

4.2.11. L-2,4-[[(1-(1-Carboxy-3-carboxypropyl)-4,7,10-(tris-tert-

butoxycarbonylmethyl)-1,4,7,10-tetraazacyclodecane)]-butyr-

amide]-[Tyr-dGlu-Cys-Hyp-Tyr(3-Cl)-Gly-Leu-Cys-Tyr-Ile-

Gln cyclic (2 f 8) disulﬁde]-1-{4-[(L-2,4-)-[[(1-(1-carboxy-3-

carboxypropyl)-4,7,10-(tris-tert-butoxycarbonylmethyl)-1,4,7,10-

tetraazacyclodecane)]methyl]benzylamide}, 11. Peptide 9 (11.83

g, 6.89 mmol) was dissolved in dimethylformamide (325 mL).

DOTAGA(OtBu)₄-OPfp (23.9 g, 27.6 mmol) was added as a solid

to the reaction. The pH was adjusted to 6.5 with diisopropylethyl-

amine (6.5 mL). The reaction was stirred at room temperature for

1 h. To the reaction mixture was added additional DOTAGA(OtBu)₄

(3.1 g, 3.6 mmol). The pH was adjusted to 6.5 with diisopropyl-

ethylamine (1.5 mL), and the reaction was stirred for 0.5 h at room

temperature. To the reaction was added additional DOTAGA(Ot-

Bu)₄(1.4 g, 1.6 mmol). The pH was adjusted to 6.5 with

diisopropylethylamine (1.0 mL), and the reaction was stirred for

0.5 h at room temperature. To the reaction was added additional

DOTAGA(OtBu)₄(0.4 g, 0.46 mmol). The pH was adjusted to 6.5

with diisopropylethylamine (0.5 mL), and the reaction was stirred

for 1 h. The reaction was monitored by HPLC. Saturated sodium

chloride (800 mL) was added to the reaction over 30 min at room

temperature, and a solid precipitated. The suspension was stirred

for an additional 30 min at room temperature. The solid was ﬁltered,

washed with water (500 mL), and dried under reduced pressure to

give compound 11 (30.6 g, 100% yield) as an off-white solid. ESI-

TOF-MS: m/z expected [C₂₁₈H₃₅₇ClN₃₄O₅₆S₂,+,2H]²⁺2224.8091,

found 2224.8091.

4.3. LC-MS of EP-2104R and its Disulﬁde Reduced

Product. EP-2104R was analyzed using a Waters Capillary HPLC

(30 °C, 15 µL/min, Thermo-Keystone HyPurity C18, 150 mm ×

0.5 mm, 3 µm, 0.2 µL injection) with UV (220 nm) detection

connected to a Waters (Micromass) Quattro LC mass spectrometer

(a triple quadrupole MS) using a gradient method. Mobile phase

A was 50 mM ammonium formate, pH 6.8. Mobile phase B was

90% acetonitrile and 10% mobile phase A. Starting from 5% B,

the fraction of B increased to 50% over 20 min and was held for

2 min. After a 1 min ramp down to 5% B, the system re-equilibrated

at 5% B over 7 min (total time ) 30 min). EP-2104R elutes at

14.5 min. The disulﬁde bond was reduced by treating 90 µL of a

1.2 mM EP-2104R solution with 10 µL of a 200 mM tris(carboxy-

9

J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008 6037

A R T I C L E S

Overoye-Chan et al.

ethyl)phosphine (TCEP) hydrochloride solution in a HPLC vial and

allowing the reduction to proceed for 30 min at room temperature.

LC-MS of the reduced EP-2104R elutes at 13.1 min with a +2

a.m.u mass shift (see Supporting Information Figure S1).

4.4. NMR Experiments. EP-2104R-La was synthesized in a

similar manner as that for EP-2104R, with the exception that LaCl₃

was used in place of GdCl₃(Scheme 2). LC/MS analysis indicated

that EP-2104R-La was 93.5% pure. About 30 mg of lyophilized

EP-2104R-La were dissolved in 600 µL of a solvent mixture

comprised of 100 µL of D₂O (Aldrich) and 500 µL of H₂O and

transferred into a 5 mm NMR sample tube (Wilmad).

Experiments were performed at 300 K on a Bruker Avance-II

400 MHz spectrometer using an inverse triple channel (¹H/¹³C/

¹⁵N) probe. Initial 1D ¹H spectra (Figure 1) were taken to optimize

experimental parameters such as the proton frequency for water

suppression by means of presaturation.⁶²The ¹H chemical shift of

all spectra has been referenced to DSS (2,2-dimethyl-2-silapentane-

5-sulfonic acid). Phase-sensitive gradient-enhanced ¹H-¹H COSY,

TOCSY (80 ms MLEV-17 spinlock), and NOESY spectra³⁶were

recorded with a spectral width of 12.25 ppm and 512 increments

in t₁with 16 averaged transients per increment using the watergate

pulse sequence for water suppression. NOESY experiments were

recorded at mixing times between 80 and 300 ms. All 2D data sets

were processed using NMRPipe⁶³and subsequently analyzed with

SPARKY version 3.106.⁶⁴

The other solutions were analyzed by HPLC (37 °C, 1.0 mL/

min, Zorbax C8, 150 mm × 4.6 mm, 2 µL injection) with UV

(220 nm) detection using an isocratic (20% acetonitrile, 80% mobile

phase A from above) method for 8 min. This method separates

MS-325 from MS-325-L while the other species elute in the void

volume.

4.6. Preparation of Fibrinogen. Fibrinogen was dissolved in

H₂O or TBS buffer, pH 7.4 and dialyzed in a Slyde-a-Lyzer (10 000

MWCO, Pierce Chemical Co.) vs TBS containing 5 mM citrate at

4 °C. After two changes of buffer, the ﬁbrinogen was centrifuged

(10 min, 2000 × g) to remove undissolved material. Fibrinogen

concentration was determined by measuring the absorbance at 280

nm (A₂₈₀), after dilution into TBS, pH 7.4. A 1 mg/mL ﬁbrinogen

solution has an absorbance of 1.512 OD units.⁶⁵

Stock ﬁbrinogen solutions ranged in concentration from 10 to

20 mg/mL. Based on the concentration determined postdialysis and

the initial mass of the ﬁbrinogen used, the ﬁbrinogen content of

the solid commercial material could be calculated. Typically this

was about 30% by weight.

4.7. Functional Fibrinogen Concentration in Plasma. The

concentration of ﬁbrinogen that is competent to form a ﬁbrin clot

was determined by a thrombin clotting time assay against a certiﬁed

standard using a ﬁbrinogen test kit (Cat. No. 886-A; Sigma

Diagnostics). The kit included a 2.52 mg/mL ﬁbrinogen standard,

Thrombin (75 U/mL), and imidazole (Im) buffer (30 mM imidazole,

130 mM NaCl, pH 7.35). Clotting times were recorded on an

Amelung KC1A Micro Coagulometer. To determine ﬁbrinogen

concentration, [Fgn], in the plasma, the standard was diluted in Im

to concentrations of 0.50, 0.252, 0.168, 0.126, and 0.084 mg/mL.

The sample (150 µL) was added to an Amelung KC1A/KC4A

microcuvette and incubated at 37 °C. A 50 µL aliquot of thrombin

was added, and the clotting time (TT) was recorded. Plasma samples

of unknown concentration were diluted into a linear range (10-

fold) in Im, and clotting times were recorded following thrombin

addition. The TT of the unknown was used to extrapolate the

ﬁbrinogen concentration of the unknown from a linear least-squares

ﬁt plot of the log(TT) vs log([Fgn]) data obtained for the standard.

4.8. EP-2104R Sample Preparation. EP-2104R stock solutions

(1–10 mM) were prepared by dissolving solid EP-2104R in water.

The concentration was determined by gadolinium analysis using

ICP-MS. All solutions were prepared by weight. Gadolinium

concentrations were calculated based on the concentration of the

stock solution and appropriate dilution factors.

Samples in TBS were prepared by mixing appropriate amounts

of EP-2104R stock solution and TBS buffer to a ﬁnal volume of

200 µL. Samples in ﬁbrinogen were prepared by mixing appropriate

amounts of EP-2104R stock solution, ﬁbrinogen stock solution, and

TBS buffer to a ﬁnal volume of 200 µL and ﬁnal ﬁbrinogen

concentration of 10 mg/mL. Samples in ﬁbrin gels were prepared

by mixing appropriate amounts of EP-2104R stock solution,

ﬁbrinogen stock solution, and TBS buffer to a ﬁnal volume of 200

µL and ﬁnal ﬁbrinogen concentration of 10 mg/mL. To these

solutions was ﬁrst added a 4 µL aliquot of 2 M CaCl₂followed by

2 µL aliquot of thrombin (0.6 U). The sample was vigorously mixed

for 3 s and then allowed to stand at room temperature for 20 min

as the ﬁbrinogen polymerized to ﬁbrin and the solution became a

gel. For the samples in human plasma, ﬁrst a stock solution of EP-

2104R in ﬁbrinogen enriched plasma was prepared. The relaxivity

samples were prepared by mixing appropriate amounts of the EP-

2104R/plasma stock with appropriate amounts of ﬁbrinogen

enriched plasma to a total volume of 200 µL. The plasma gels (clots)

were prepared by taking the plasma samples after measuring the

T₁and T₂data and then adding ﬁrst 4 µL of 2 M CaCl₂followed

by 2 µL of thrombin (0.6 U). The sample was vigorously mixed

for 3 s and then allowed to stand at room temperature for 20 min

as the ﬁbrinogen polymerized to ﬁbrin and the solution became a

gel.

4.5. Inertness of EP-2104R and Comparison to

Commercial Contrast Agents. MS-325-L solutions at 5 and 20

mM were prepared in 50 mM citrate buffer at pH 3.0, 4.0, and 5.0.

The MS-325-L concentration was determined by absorbance at 280

nm and comparison to a standard curve generated by a series of

MS-325 solutions of known concentration. A 5 mM solution of

EP-2104R was prepared in 50 mM citrate buffer at pH 3.0, 4.0,

and 5.0. For a given pH (3, 4, or 5), 1.0 mL of the 20 mM MS-

325-L solution was vigorously mixed with 1.0 mL of the 5 mM

EP-2104R solution. A 50 µL aliquot was removed for HPLC

analysis as described below, and the remainder of the solution was

incubated at 60 °C. A 50 µL aliquot was removed and analyzed at

24, 48, 72, and 96 h.

For each of the commercial agents, [Gd(DTPA)]^2-, [Gd(HP-

DO3A)], and [Gd(DTPA-BMA)], a 10 µL aliquot of the 500 mM

contrast agent solution was added to 1.0 mL of 5 mM MS-325-L

at pH 3.0 or 5.0. After vigorous mixing, a 50 µL aliquot was

removed for HPLC analysis as described below and the remainder

of the solution incubated at 60 °C. A 50 µL aliquot was removed

and analyzed at 10 min intervals out to 40 min after mixing. The

[Gd(HP-DO3A)] solution was further sampled at 50 min, 60 min,

and then at 4 to 12 h intervals out to 48 h of incubation. The total

gadolinium in all solutions was measured by ICP-MS. To control

for sample degradation, individual MS-325-L, EP-2104R, [Gd-

(DTPA)]^2-, [Gd(HP-DO3A)], and [Gd(DTPA-BMA)] solutions

were incubated at 60 °C and analyzed.

The EP-2104R/MS-325-L solutions were analyzed by HPLC (37

°C, 0.8 mL/min, Kromasil C4, 250 mm × 4.6 mm, 5 µL injection)

with UV (220 nm) detection using a gradient method. Mobile phase

A was 100 mM ammonium phosphate, pH 6.5 plus 200 µM

Na₂H₂EDTA. Mobile phase B was 60% acetonitrile and 40% mobile

phase A. Starting from 20% B, the fraction of B increased to 32%

over 10 min and then from 32 to 50% B over 25 min. After a 30 s

ramp to 80% B, the system re-equilibrated at 20% B over 6.5 min

(total time ) 42 min).

(62) Neuhaus, D.; Williamson, M. P. The Nuclear OVerhauser Effect in

Structural and Conformational Analysis; 2nd ed.; Wiley-VCH: New

York, 2000.

(63) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax,

A. J. Biomol. NMR 1995, 6, 277–93.

(64) Goddard, T. D.; Kneller, D. G.; SPARKY 3.106; University of

California, San Francisco, Available online at http://www.cgl.ucsf.edu/

home/sparky/manual/.

(65) Ugarova, T. P.; Budzynski, A. Z. J. Biol. Chem. 1992, 267, 13687–

93.

9

6038 J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008

EP-2104R: Thrombus Speciﬁc MRI Contrast Agent

A R T I C L E S

4.9. Binding of EP-2104R to Human Plasma Clots. A 10 mM

EP-2104R solution was diluted 100-fold with human plasma to give

a EP-2104R/plasma stock solution that was 100 µM. Diluting this

solution with citrated human plasma resulted in a series of solutions

where the total EP-2104R concentration varied between 0.6 and

36 µM. An aliquot of each of these solutions was removed to

determine [EP-2104R]_totalby ICP-MS. Calcium chloride (2 M) and

thrombin were added to yield ﬁnal concentrations of 10 mM CaCl₂

and 1 U/mL thrombin in order to induce clotting. To minimize

dilution, small volumes of these reagents (1.7% total volume) were

added. The plasma clots were incubated for 1 h at 37 °C in a

temperature controlled incubator. The samples were removed

individually, and the supernatant was quickly recovered by vigor-

ously stirring the clot with a pipet tip while aspirating the released

liquid. About half of the clot volume was recovered from each clot,

The plasma was treated with 10 µM PPACK, an irreversible

thrombin inhibitor, to prevent clotting from residual thrombin

present in the ﬁbrin plates. Assayed solutions ranged from 0.4 µM

e [EP-2104R]_totale 52 µM, and aliquots were taken for ICP

determination of [EP-2104R]_total. The EP-2104R solutions (100 µL)

were added to duplicate wells of the dried ﬁbrin microtiter plate,

and the plate was shaken for 2 h. After incubation, solution was

removed and the concentration of unbound EP-2104R, [EP-

2104R]_free, was determined by ICP-MS.

4.11. Relaxation Rate Measurements. Relaxation rates were

measured at 20 MHz (0.47 T) and 60 MHz (1.41 T) using a Bruker

NMS 120 minispec and a Bruker mq60 minispec, respectively.

Samples, 200 µL, were pipetted into glass vials. All samples were

equilibrated for at least 20 min at 37 °C and measured at 37 °C.

The ﬁbrin gels were prepared in the tube by addition of 4 µL of 2

M CaCl₂and 2 µL of 100 U/mL thrombin solution to a solution

already containing EP-2104R and ﬁbrinogen (30 µM), followed

by rapid mixing and then equilibration at room temperature for at

least 30 min. T₁was measured with an inversion–recovery pulse

sequence.⁶⁶Typically 10–20 spectra were acquired with differing

delay times. The preacquisition delay was always set to at least

5T₁. T₂was measured using a Carr-Purcell-Meiboom-Gill

sequence with 300 to 500 echoes collected.⁶⁶Error in T₁and T₂

was estimated to be (5%.⁶⁷

this was assayed for Gd content by ICP-MS to give [EP-2104R]_free

.

4.10. Dried Fibrin Plate Preparation. Plasminogen-depleted

ﬁbrinogen from human plasma (Calbiochem Cat No.341578) was

dialyzed against 50 mM Tris, pH 7.4, 150 mM sodium chloride, 5

mM sodium citrate (TBS ·citrate) prior to use. The ﬁbrinogen

concentration was adjusted to 5 mg/mL, and CaCl₂was added (7

mM). The ﬁbrinogen solution (50 µL) was dispensed into the wells

of a 96-well polystyrene microplate (Immulon-II). A solution (50

µL) of human thrombin (2 U/mL) in TBS was added to each well

to clot the ﬁbrinogen and to yield a ﬁnal ﬁbrin concentration close

to 2.5 mg/mL. In one of the assays, ﬁbrin plates were formulated

at 1.25 mg/mL. The plates were incubated at 37 °C and evaporated

to dryness overnight. The plates were sealed with tape and stored

at -20 °C until use. The extent of ﬁbrinogen polymerization

(clottability) of individual ﬁbrinogen batches was determined by

ﬁrst measuring the absorbance at 280 nm of the ﬁbrinogen solutions

(A^pre₂₈₀). An aliquot of the ﬁbrinogen solution was then clotted by

addition of 1 U/mL thrombin, incubated at 37 °C for 1 h, and then

centrifuged to separate the clot. The absorbance of the supernatant

(A^post₂₈₀) was measured. The percent clottability is given by

¹H nuclear magnetic relaxation dispersion (NMRD) measure-

ments were recorded on a ﬁeld cycling relaxometer⁶⁸over the

Larmor frequency range 0.01 to 50 MHz on 600 µL samples. A

total of 22 data points were collected for each dispersion. NMRD

were recorded at 5, 15, 25, and 35 °C.

Acknowledgement Acknowledgement is made to John

Amedio and Phil Graham for useful discussions and to Frank

Qi and Steve Witowski for analytical support.

Supporting Information Available: LC-MS data for EP-

1

2104R and its reduced form; H, ¹³C, and ¹⁵N chemical shifts

post

280

A

%C ) 1 -

× 100%

(7)

for peptide tetraamine 9 and EP-2104R-La. This material is

available free of charge via the Internet at http://pubs.acs.org.

pre

280

(

)

A

The ﬁbrin concentration, [fbn] (µM), in the plate was calculated

from [Fgn] using a molecular weight of 340 kD for ﬁbrin(ogen),

as follows:

JA800834Y

%C[fgn]

100 0.34

(66) Braun, S.; Kalinowski, H.-O.; Berger, S. 150 and More Basic NMR

Experiments; Wiley-VCH: Weinheim, 1998.

(67) Pitner, T. P.; Whidby, J. F. Anal. Chem. 1979, 51, 2203–6.

(68) Koenig, S. H.; Brown, R. D., III Prog. Nucl. Magn. Reson. Spectrosc.

1990, 22, 487–567.

[Fbn] )

(8)

Solutions of EP-2104R (10 µM and 100 µM) were prepared from

a 1 mM stock solution by serial 1:10 dilutions into human plasma.

9

J. AM. CHEM. SOC. VOL. 130, NO. 18, 2008 6039

Article Doi

EP-2104R: A fibrin-specific gadolinium-based MRI contrast agent for detection of thrombus

DOI: 10.1021/ja800834y

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja800834y

Full text of DOI:10.1021/ja800834y

Products guided by the article

R&D Labs maybe for 817562-96-2

Relevant to this article

Hot Product