Monovalent and bivalent fibrin-specific MRI contrast agents for detection of thrombus

Article abstract of DOI:10.1002/anie.200800563

Probing in contrast: Four gadolinium-DTPA moieties (DTPA = diethylenetriaminepentaacetic acid) and two fibrin-specific cyclic peptides are linked by a compact triethylenetetraamine core (see scheme) to create a highly sensitive probe for molecular MR imaging of thrombosis. The contrast agent has a high molecular relaxivity, and the dual peptide construct provides five-fold higher fibrin affinity than the monovalent analogue. This bivalent probe showed significant specific thrombus uptake in an in vivo model of thrombosis. (Figure Presented).

Full text of DOI:10.1002/anie.200800563

Communications
DOI: 10.1002/anie.200800563
Medical Imaging
Monovalent and Bivalent Fibrin-specific MRI Contrast Agents for
Detection of Thrombus
Shrikumar A. Nair, Andrew F. Kolodziej, Gandhali Bhole, Matthew T. Greenfield,
Thomas J. McMurry, and Peter Caravan*
Thromboembolic diseases such as stroke, heart attack, and
pulmonary embolism contribute significantly to the cost of
healthcare in the developed world. Many imaging techniques
(ultrasound, magnetic resonance imaging (MRI), gamma
scintigraphy, computed tomography) are used to identify
thrombus, and the preferred technique often depends on the
anatomical region. However these methods typically cannot
distinguish thrombus from other pathologies (e.g. athero-
sclerotic plaque, tumor). It would be useful to have a single,
three-dimensional imaging modality that can provide a
specific and sensitive diagnosis of
EP-2104R was identified by phage display. The phage-display
study also identified an 11 amino acid peptide with a seven
amino acid cyclic core (CDYYGTC) that had affinity for
fibrin. To convert this peptide lead (CDYYGTC) into a fibrin-
specific MR contrast agent, we initially derivatized the
peptide with a GdDTPA moiety (DTPA = diethylenetriami-
nepentaacetic acid) at the N-terminus by an amide bond
linkage to give EP-782. A control was prepared, denoted EP-
821, in which the amino acid sequence (Table 1) was
scrambled to eliminate fibrin binding. EP-782 binds with
thromboembolic disease over the
whole body.
Fibrin is a useful MR target
because it is the major protein
constituent of arterial and venous
clots and is present at 10–100 mm in
Table 1: Compounds with peptide sequence, their relaxivities (r1) in Tris buffered saline, 30 mm
fibrinogen, or 30 mm fibrin, and their stepwise association constants (Ka1, Ka2).[a]
Compound and peptide sequence
r1 per Gd [mmꢀ1 sꢀ1 [b]
]
Affinity [”10ꢀ5 mꢀ1
]
TBS
fgn
Fibrin
Ka1
Ka2
EP-782: GdDTPA-G·L·P·C·D·Y·Y·G·T·C·L·D
EP-821: GdDTPA-G·Y·L·C·G·D·Y·T·L·C·P·D
EP-1084: (GdDTPA)4-(L·P·C·D·Y·Y·G·T·C·Bip·d)2 20.6
EP-1086: (GdDTPA)4-L·P·C·D·Y·Y·G·T·C·Bip·d 16.9
10.1
14.6
10.5
13.3
ND
ND
14.9
13.0
27.7
26.0
2.6ꢁ0.1
<0.01
0.46ꢁ0.06
<0.01
the thrombus,
a concentration
10.3ꢁ2.1 2.1ꢁ0.3
range compatible with detection
by gadolinium-based contrast
agents.[1,2] Fibrin-specific antibod-
ies linked to gadolinium-contain-
ing nanoparticles have been suc-
1.7ꢁ0.3 0.40ꢁ0.07
[a] Amino acids given by one-letter abbreviation. Bip=l-biphenylalanine; d=d-aspartate. Error in r1
estimated at ꢁ10%. ND=not detectable. [b] 378C, 20 MHz; ND=not determined.
cessfully used to detect thrombi using MRI in animal
models.[3,4] EP-2104R, a fibrin-specific peptide derivatized
with four Gd chelates, has shown MR imaging efficacy in a
variety of animal thrombus models.[5–9] Herein we report a
new peptide sequence with specificity for fibrin and a
comparison of the fibrin-binding and relaxivity properties of
monovalent and bivalent fibrin-specific multimeric MRI
contrast agents. In vivo data are presented that demonstrate
localization of the bivalent agent in a model of venous
thrombosis.
low micromolar affinity to two sites on fibrin, while the
scrambled peptide isomer showed no detectable fibrin bind-
ing (Table 1). The relaxivity (r1 = D(1/T1)/[Gd]) of each
compound was determined in Tris (50 mm) buffered saline
(TBS), in TBS plus 30 mm human fibrinogen (fgn), or in TBS
plus 30 mm fgn that had been converted into fibrin (Table 1).
There was no significant difference in EP-782 relaxivity when
measured in TBS or TBS + fgn suggesting weak/no interac-
tion with fgn. But the relaxivity in fibrin was > 40% higher
consistent with fibrin binding and the receptor induced
magnetization enhancement (RIME) effect.[1,10,11] EP-821
showed the same relaxivity, within error, in all three media.
The lack of relaxivity enhancement of EP-821 in fibrin
demonstrates that the enhancement observed with EP-782 is
due to fibrin binding and not a viscosity effect.
Previous work utilized a cyclic six amino acid peptide[8] for
fibrin targeting. The peptide lead that ultimately resulted in
[*] Dr. P. Caravan
Athinoula A. Martinos Center for Biomedical Imaging
Massachusetts GeneralHospital
149 Thirteenth St, Suite 2301, Charlestown, MA 02129 (USA)
Fax : (+1)617-726-7422
Prior work[5–9] indicated that multiple Gd moieties were
required for robust MRI signal enhancement of thrombus
in vivo. We developed a bifunctional tetrameric GdDTPA-
based scaffold, 1, based on triethylenetetraamine (trien). The
trien scaffold is easily differentiated using the Dde protecting
group to selectively protect the primary amines (see Support-
ing Information). Two DTPA moieties were attached to each
secondary nitrogen by a short diaminopropionate linker to
minimize rotational flexibility and enhance relaxivity, while
the primary amines were used to introduce oxime groups for
further peptide conjugation. By controlling the stoichiometry
E-mail: caravan@nmr.mgh.harvard.edu
Dr. S. A. Nair
Affinergy, Inc.
617 Davis Drive, Suite 100, RTP, NC 27713 (USA)
Dr. A. F. Kolodziej, G. Bhole, M. T. Greenfield, Dr. T. J. McMurry
EPIX Pharmaceuticals, Inc.
4 Maguire Road, Lexington, MA 02421 (USA)
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
4918
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 4918 –4921

Angewandte
Chemie
of peptide conjugation, this scaffold afforded the possibility to
test the effect of bivalent versus monovalent fibrin targeting.
The EP-782 result indicated that the peptide could be safely
modified at the N-terminus without affinity loss. Concur-
rently, we performed structure–activity studies on the peptide
to improve fibrin affinity. The full structure–activity results
will be reported separately, but one positive finding was a Leu
to biphenylalanine (Bip) substitution to improve fibrin
affinity. The C-terminus l-Asp (D) could also be converted
into d-Asp (d) without loss of affinity. Nonnatural amino
acids can often impart metabolic stability, and by substituting
the peptide at the C-terminus with d-Asp and the N-terminus
with gadolinium chelates it was anticipated that the resultant
complex would be stable in vivo. Chemoselective ligation was
used to introduce one or two equivalents of peptide to the
DTPA tetramer giving EP-1086 and EP-1084 (Scheme 1). A
Ser residue was incorporated at the N-terminus of the peptide
and mild periodate oxidation gave an N-terminal a-keto
aldehyde, 2, that reacts selectively with the oxime-containing
DTPA-tetramer scaffold at pH4.6 in water. Complexation
with GdCl3 gave the desired product.
Fibrin binding data of EP-1084 and EP-1086 shown in
Figure 1 demonstrates that both compounds bind to two sites
on fibrin and that fibrin affinity is higher for bivalent EP-1084.
A stoichiometric binding model [Eq. (1)] with stepwise
association constants Ka1, Ka2 (reported in Table 1) were
fitted to the data.
2
n
½boundŠ
½fibrinŠ
Ka1½freeŠ þ 2 Ka1 Ka2 ½freeŠ þ . . . þ n Ka1
K
K
a2 . . . Kan ½freeŠ
¼
2
n
1 þ Ka1 ½freeŠ þ Ka1 Ka2 ½freeŠ þ . . . þ Ka1
a2 . . . Kan ½freeŠ
ð1Þ
The binding curve for EP-1086 was well described by two
association constants. For EP-1084, the binding curve did not
saturate at [bound]/[fibrin] = 2 and a third, much weaker
*
*
Figure 1. Binding isotherms for EP-1084 ( ) and EP-1086 ( ) at
pH 7.4, 378C, TBS.
Scheme 1.
Angew. Chem. Int. Ed. 2008, 47, 4918 –4921
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4919

Communications
association constant was required to fit the data, Ka3 = (6 ꢁ
EP-1084 was investigated in a guinea pig venous thrombus
model. Prior to thrombus formation 125I-fibrinogen was
administered to the animal. A thrombus was formed by
isolating a section of the inferior vena cava and injecting
thrombin. The clot was aged for 30 min, and then 111In-labeled
EP-1084 (2 mmolkgꢀ1) was given as a bolus injection in the
jugular vein. Immediately after the EP-1084 was added, 99mTc-
DTPA mixed with 2 mmolkgꢀ1 GdDTPA was injected. After
another 30 min, blood was drawn and the clot removed. The
blood and clot samples were weighed and the amount of each
radiolabeled species was determined. The three isotopes used
emit at sufficiently different energies that it is possible to
quantify each in the presence of the others.
1) ” 103 mꢀ1. This likely represents non-specific binding.
The stepwise thermodynamic binding constants Ka1 and
Ka2 are related to site-specific binding constants, K’a1 and K’a2,
as Ka1 = 2K’a1 and Ka2 = K’a2/2, and association constants are
related to dissociation constants as Kd = 1/Ka.[12] For EP-1086,
K’d1 = 11.2 ꢁ 1.6
ꢂ K’d2 = 12.6 ꢁ 2.4 mm and for EP-1084,
K’d1 = 1.9 ꢁ 0.4 ꢂ K’d2 = 2.4 ꢁ 0.4 mm. The fact that K’d1 and
K’d2 are approximately constant for a given compound
strongly suggests that both binding sites are equivalent. This
result is consistent with the dimeric subunit structure of this
protein. The addition of the second peptide on EP-1084
improves the fibrin affinity five-fold relative to the monop-
eptide EP-1086.
The clot:blood ratio was 6.6 ꢁ 0.4 for 125I-fibrinogen.
Guinea pig plasma fgn concentration is 9.7 mm and its
hematocrit is 0.44.[17] Assuming conversion of fibrinogen
into fibrin in the clot, the mean fibrin concentration in this
model is 36 ꢁ 2 mm. EP-1084 showed a 4.4 ꢁ 1.2 clot:blood
ratio demonstrating positive uptake in the clot. To ensure that
this positive result was not due to a trapping phenomenon, the
extracellular tracer GdDTPA was co-administered and gave a
clot:blood ratio of 0.64 ꢁ 0.17; the less than unity ratio is
consistent with distribution in the clot but no specific uptake.
Figure 3 shows the concentration of EP-1084 and GdDTPA in
the clot and in blood, highlighting the fact that while the blood
levels for both compounds were very similar, there was seven-
times more EP-1084 in the clot than GdDTPA.
Nuclear magnetic relaxation dispersion (NMRD) profiles
of these compounds in the presence and absence of fibrin are
shown in Figure 2. In buffer, r1 is relatively high for both EP-
1084 and EP-1086 because of the increased size of these
Figure 2. Relaxivity per Gd ion for 60 mm EP-1084 (solid symbols) and
33 mm EP-1086 (open symbols) in TBS (triangles) or in the presence of
30 mm fibrin (circles) at pH 7.4, 358C.
Figure 3. Distribution of GdDTPA and EP-1084 in a venous clot and in
the blood pool, 30 min after injection of 2 mmolkgꢀ1 of each probe.
contrast agents. In the presence of fibrin, both compounds
show increased relaxivity owing to the RIME effect, and the
relaxivity maximum shifts to lower frequency, approximately
30 MHz, consistent with slower rotational dynamics. How-
ever, the NMRD curves in fibrin are very similar for both
compounds suggesting similar rotational dynamics at the
GdDTPA moiety. Previously we found that a multilocus
binding approach could increase relaxivity in the protein-
bound state by rigidifying the molecule upon protein bind-
ing.[13] The absence of such an effect here may suggest that
only one peptide unit from EP-1084 is binding to fibrin.
The per Gd relaxivity of fibrin-bound EP-1084 is similar
to the commercial albumin-binding agent gadofosveset (MS-
325)[14,15] and about six-fold higher than extracellular contrast
agents such as GdDTPA (24-fold higher per molecule)[16] at
the usual MRI field of 1.5 T.
The in vivo thrombus uptake and high relaxivity of EP-
1084 are compatible with this compound being employed as a
thrombus-specific imaging agent. A low dose of EP-1084
(2 mmolkgꢀ1, cf. 100–200 mmolkgꢀ1 for most MRI agents)
delivered 64 ꢁ 28 mm Gd to the thrombus. At 1.5 T, D(1/T1) =
1.5 sꢀ1 is calculated at the thrombus. The inherent thrombus
T1 depends on the clot age, but in many instances the
thrombus 1/T1 is very close to that of blood,[18] approximately
0.75 sꢀ1. Under these conditions, a 200% increase is expected
in relaxation rate at the thrombus and ready detection of clot.
In conclusion, we have described new fibrin-specific, high
relaxivity contrast agents for thrombus MR imaging, and
demonstrated thrombus localization in an in vivo model.
4920 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 4918 –4921

Angewandte
Chemie
[7] E. Spuentrup, B. Fausten, S. Kinzel, A. J. Wiethoff, R. M. Botnar,
Thrombus uptake is consistent with bright spot thrombus
MRI and future work will describe in vivo MRI with this class
of compounds.
P. B. Graham, S. Haller, M. Katoh, E. C. Parsons, Jr., W. J.
Manning, T. Busch, R. W. Gꢀnther, A. Buecker, Circulation
2005, 112, 396.
[8] E. Spuentrup, M. Katoh, A. Buecker, B. Fausten, A. J. Wiethoff,
J. E. Wildberger, P. Haage, E. C. Parsons, Jr., R. M. Botnar, P. B.
Graham, M. Vettelschoss, R. W. Gꢀnther, Invest. Radiol. 2007,
42, 586.
[9] C. P. Stracke, M. Katoh, A. J. Wiethoff, E. C. Parsons, P.
Spangenberg, E. Spꢀntrup, Stroke 2007, 38, 1476.
[10] P. Caravan, J. J. Ellison, T. J. McMurry, R. B. Lauffer, Chem. Rev.
1999, 99, 2293.
Experimental Section
Details of compound syntheses, fibrin binding and relaxivity assays,
and biodistribution are given in the Supporting Information.
Received: February 3, 2008
Published online: May 21, 2008
[11] R. B. Lauffer, Magn. Reson. Med. 1991, 22, 339.
[12] I. M. Klotz, Ligand Receptor Energetics—A Guide for the
Perplexed, Wiley, New York, 1987.
[13] Z. Zhang, M. T. Greenfield, M. Spiller, T. J. McMurry, R. B.
Lauffer, P. Caravan, Angew. Chem. 2005, 117, 6924; Angew.
Chem. Int. Ed. 2005, 44, 6766.
[14] P. Caravan, N. J. Cloutier, M. T. Greenfield, S. A. McDermid,
S. U. Dunham, J. W. Bulte, J. C. Amedio, Jr., R. J. Looby, R. M.
Supkowski, W. D. Horrocks, Jr., T. J. McMurry, R. B. Lauffer, J.
Am. Chem. Soc. 2002, 124, 3152.
[15] H. B. Eldredge, M. Spiller, J. M. Chasse, M. T. Greenwood, P.
Caravan, Invest. Radiol. 2006, 41, 229.
[16] M. Rohrer, H. Bauer, J. Mintorovitch, M. Requardt, H. J.
Weinmann, Invest. Radiol. 2005, 40, 715.
[17] W. C. Hou, H. S. Tsay, H. J. Liang, T. Y. Lee, G. J. Wang, D. Z.
Liu, J. Ethnopharmacol. 2007, 111, 483.
[18] W. D. Rooney, G. Johnson, X. Li, E. R. Cohen, S. G. Kim, K.
Ugurbil, C. S. Springer, Jr., Magn. Reson. Med. 2007, 57, 308.
Keywords: contrast agent · fibrin · gadolinium ·
magnetic resonance imaging · thrombus
.
[1] P. Caravan, Chem. Soc. Rev. 2006, 35, 512.
[2] A. D. Nunn, K. Linder, M. Tweedle, Q. J. Nuc. Med. 1997, 41,
155.
[3] S. Flacke, S. Fischer, M. J. Scott, R. J. Fuhrhop, J. S. Allen, M.
McLean, P. Winter, G. A. Sicard, P. J. Gaffney, S. A. Wickline,
G. M. Lanza, Circulation 2001, 104, 1280.
[4] P. M. Winter, S. D. Caruthers, X. Yu, S. K. Song, J. Chen, B.
Miller, J. W. Bulte, J. D. Robertson, P. J. Gaffney, S. A. Wickline,
G. M. Lanza, Magn. Reson. Med. 2003, 50, 411.
[5] R. M. Botnar, A. Buecker, A. J. Wiethoff, E. C. Parsons, Jr., M.
Katoh, G. Katsimaglis, R. M. Weisskoff, R. B. Lauffer, P. B.
Graham, R. W. Gunther, W. J. Manning, E. Spuentrup, Circu-
lation 2004, 110, 1463.
[6] M. Sirol, V. Fuster, J. J. Badimon, J. T. Fallon, P. R. Moreno, J. F.
Toussaint, Z. A. Fayad, Circulation 2005, 112, 1594.
Angew. Chem. Int. Ed. 2008, 47, 4918 –4921
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
4921