Multilocus binding increases the relaxivity of protein-bound MRI contrast agents

Article abstract of DOI:10.1002/anie.200502245

Spinning around: The sensitivity of MRI contrast agents can be increased through multilocus binding. This restricts the internal motion and therefore increases the relaxivity of a bound protein multimer (see picture; A) relative to a single binding group

Full text of DOI:10.1002/anie.200502245

Communications
Protein Relaxivity
DOI: 10.1002/anie.200502245
Multilocus Binding Increases the Relaxivity of
Protein-Bound MRI Contrast Agents
Zhaoda Zhang, Matthew T. Greenfield, Marga Spiller,
Thomas J. McMurry, Randall B. Lauffer, and
Peter Caravan*
Dedicated to Professor Iwao Ojima
on the occasion of his 60th birthday
The utility of clinical magnetic resonance imaging (MRI) has
been greatly expanded by the use of contrast agents—
gadolinium complexes that serve to increase the longitudinal
nuclear relaxation rate (1/T1) of water protons.Commercial
contrast agents are only effective at high concentrations
(> 0.1 mm) and, as a result, there has been considerable effort
to increase their sensitivity.One approach is to increase the
number of gadolinium ions in the molecule by making
polymeric or dendrimeric contrast agents.[1]
A second
approach is to optimize the parameters that influence
relaxation enhancement, such as rotational diffusion and
water exchange.[2]
An effective mechanism for increasing relaxivity (r1 =
(D1/T1)/[Gd]) is the receptor-induced magnetization-
enhancement (RIME) effect.[3] An example is the GdDTPA
derivative MS-325, which targets serum albumin.When MS-
325 binds to albumin, the rotational correlation time (tR)
increases from that of a small molecule to that of the protein,
[*] Dr. Z. Zhang, M. T. Greenfield, Dr. T. J. McMurry, Dr. R. B. Lauffer,
Dr. P. Caravan
EPIX Pharmaceuticals, Inc.
67 Rogers Street, Cambridge, MA 02142 (USA)
Fax : (+1)617-250-6127
E-mail: pcaravan@epixpharma.com
M. Spiller
Department of Radiology
New York Medical College
Valhalla, NY 10595 (USA)
Supporting information (synthesis of 1, 2, and 3) for this article is
available on the WWW under http://www.angewandte.org or from
the author.
6766
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 6766 –6769

Angewandte
Chemie
and the relaxivity increases ninefold to 51 mmꢀ1 sꢀ1
(378C, 20 MHz).[4] This is shown schematically in
Figure 1a and b.Despite this increased relaxivity,
there is still a drive to increase relaxivity further to
allow imaging of lower concentration targets.One
approach for increased molecular relaxivity would be
to combine the RIME effect with a multimeric gadoli-
nium agent (Figure 1c).However, polymeric contrast
agents such as GdDTPA-monoamide tethered to poly-
lysine[5–8] or dextran[9,10] have much lower relaxivities
(r1 < 13 mmꢀ1 sꢀ1 per ion at 378C) than one might expect
based on their molecular weight because of internal
motion along the polymer.[11] Furthermore, targeting
multimeric contrast agents to a specific protein may also
result in lower relaxivity than expected because of
internal flexibility within the multimer.
It was hypothesized that a large RIME effect could
be maintained if the multimer exhibited a certain
rigidity when bound to the protein target.One approach
would be to prepare a rigid multimer[12] by limiting the
degrees of rotational freedom along the chain and then
attaching a targeting vector to the multimer.A second
approach in which a second targeting vector is used to
anchor the terminus of the chain (multilocus binding) is
outlined in Figure 1d.The second vector can be
designed to bind nonspecifically to a hydrophobic
patch on the target.In the absence of a target, there is
rotational flexibility and low relaxivity; however, upon
binding of the protein with both binding groups, the
bound complex becomes more rigid and the relaxivity
increases.In the absence of the second targeting group,
the relaxivity of the bound species is lower because of internal
flexibility.It is possible that the second targeting group could
target a second protein which would also induce rigidity and
enhance relaxivity.
Figure 2. Compounds described in this study: GdDTPA tetramer containing zero (1),
one (2), or two (3) HSA-binding groups (BG), and benchmark GdDTPA monomer MS-
325.
two protein-binding moieties.The DTPA moiety [13] shown in
Figure 2 was used instead of the monoamide, because the
GdDTPA complex has a more favorable water-exchange rate
and the thermodynamically stable GdDTPA core is already
present in several clinically approved contrast agents.[14]
Human serum albumin (HSA) was chosen as the protein
target for proof of principle and MS-325 (Figure 2) was used
as a benchmark compound.Although MS-325 does not
contain the same biphenyl albumin-binding group, it exhibits
both high relaxivity and albumin binding and represents the
state of the art.
To test this hypothesis, three tetrameric GdDTPA-based
compounds were prepared (Figure 2) containing zero, one, or
The binding of each compound 1–3 (0.1 mm) to HSA
(4.5% w/v = 0.67 mm) was assessed by ultrafiltration[4]
through a 5-kDa-molecular-weight cut-off membrane at
378C and at pH 7.4; the gadolinium concentration in both
the initial solution (total) and the filtrate was measured by
inductively coupled plasma mass spectrometry (ICP-MS).
Not surprisingly, compound 3 (with two binding groups)
exhibits the highest affinity for HSA (Table 1).The relaxiv-
ities of 1–3 (0.1 mm) were measured at 20 MHz, pH 7.4, 378C
in phosphate-buffered saline (PBS), in the presence or
absence of HSA (0.67 mm).As a result of their increased
molecular size, the relaxivities per Gd of the tetramers in
buffer solution are higher than that of MS-325.When the
measurement is made in an excess of HSA, the observed
relaxivities follow the order 3 > 2 > 1.To determine if this is
simply a result of increasing albumin affinity, the relaxivity of
the bound fraction can also be calculated (rb1d = {r1obsꢀffree r1free}/
fbd).Table 1 shows that compound 3 has the highest albumin-
Figure 1. Strategies for increasing relaxivity: a) Discrete Gd chelate and
targeting moiety, for example, the fast tumbling of MS-325 limits the
relaxivity. b) Binding to the target slows tumbling, and relaxivity is
increased. c) Gd chelate multimer with targeting; relaxivity may be
limited by the fast internal motion. d) Gd chelate multimer with multi-
locus binding rigidifies the bound complex and increases the relaxivity.
Angew. Chem. Int. Ed. 2005, 44, 6766 –6769
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6767

Communications
Table 1: Binding of MS-325, 1, 2, and 3.[a]
the complexity of the system (multiple chelates, complex
rotational motion), it is informative to model these curves and
this was performed in an approach previously described for
the high-field part of the NMRD curve.[15] The NMRD of a
q = 0 triethylenetetraamine hexaacetic acid (TTHA) deriva-
tive reported earlier[4] was used to estimate the second- and
outer-sphere contributions to relaxivity.The water residency
time at 358C was fixed for all three compounds at 100 ns
based on an analogy with other substituted GdDTPA com-
pounds.[4,14] As the chelate part of the molecule is the same for
all three compounds, it was assumed that the water-exchange
rate would also be the same.The Gd—H(water) distance was
fixed at 3.1 Š.[16] The three bound relaxivity NMRD data sets
were fit simultaneously, the two parameters for electronic
relaxation (D2, tv) were treated as global parameters (owing
to the identical chelating portions), and the rotational
parameters were treated locally.
These data were fit in two ways: first, an isotropic model
of rotation was employed to analyze these data.This model
did not fit well—the peak shape was not well reproduced—
and the fitted curve gave a much flatter peak that under-
estimated the relaxivity maximum and overestimated the
relaxivity at higher fields.The correlation times from the
fitting are in the 1-ns range and do not reflect the physical
reality of an albumin tethered complex.The second method
was a Lipari–Szabo model-free approach that has been
described previously.[15] Equation (1) describes the dipolar
Compound Number
of Gd
%
ro1bs
bound (PBS)
ro1bs
HSA
+
rb1d
(HSA)
MS-325
1
4
4
4
88
30
78
94
5.8 (5.8) 46 (46)
11.3 (45.2) 15.0 (60)
10.1 (40.4) 26.2 (104.8) 30.8 (123.2)
10.3 (41.2) 39.1 (156.4) 41.3 (165.2)
50.8 (50.8)
23.9 (95.6)
1
2
3
[a] Percentage of 0.1 mm compound bound to 0.67 mm HAS, observed
per Gd relaxivities (mMꢀ1 sꢀ1) at 20 MHz in the absence and presence of
0.67 mm HSA, and calculated albumin-bound relaxivities. Numbers in
parentheses refer to relaxivity per molecule; all measurements were
performed at pH 7.4, 378C, with 0.1 mm compound in PBS.
bound relaxivity of the three multimeric chelates.Although
the GdDTPA moieties of 3 have a somewhat flexible linkage
to the albumin-binding groups, the relaxivity per Gd of 3 in
HSA approaches that of the more compact and rigid MS-325/
HSA.On a per molecule basis, the relaxivity of 3 is clearly the
highest.
To demonstrate that this result is indeed a rotational
phenomenon, nuclear magnetic relaxation dispersion
(NMRD) profiles were recorded for each of the three
compounds, both with and without excess HSA at 358C.
Figure 3 shows the relaxivity profiles for each compound
ꢀ
ꢁ
3 F tcg
1
T1M
C
3 ð1ꢀFÞtcl
¼
þ
;
r6GdH 1 þ w2Ht2cg 1 þ wH2 tc2l
ð1Þ
1
1
1
1
1
1
1
¼
þ
þ
;
¼
þ
tcg tg T1e tm tcl tcg tl
relaxation of the coordinated water.In this case, the spectral
density function has two terms that are related by a factor F
(sometimes called an order parameter and denoted S2), which
describes the degree of spatial restriction.If the motion of the
Gd chelate is independent of the protein, then F = 0; if the
chelate is completely immobilized on the protein, then F = 1.
There are two correlation times, tcg and tcl, which contain
contributions from global (tg) and local (tl) rotational
correlation times, as well as from the electronic relaxation
time T1e.
A global correlation time based on fluorescence aniso-
tropy decay was fixed at 41 ns,[17] whereas the local correlation
time for each compound was allowed to vary along with its
order parameter.In this case, the peak shape is much better
reproduced, and the resultant fitted parameters are given in
Table 2.As may be expected, the order parameter F increases
from compound 1 to 2 to 3 as the number of binding groups
Figure 3. 1H NMRD relaxivities expressed per Gd (358C, pH 7.4) of 1
(triangles), 2 (squares), and 3 (circles) (0.1 mm) in PBS only (open
symbols) or bound to 4.5% HSA (filled symbols).
when bound to albumin (solid symbols, rb1d) or in buffer
solution only (open symbols).The NMRD of the three
compounds in buffer solution are similar and rather feature-
less, with an estimated rotational correlation time of about
0.3 ns. In the presence of excess albumin, however, the
relaxivities of all three compounds are increased.Notably, the
appearance of a high-field peak at about 25 MHz is typical of
a slow-motion fraction.
Table 2: Fitted NMRD parameters for compounds 1, 2, and 3.[a]
tv
[ps] [1018 sꢀ2
D2
1
2
3
]
F
tl [ns]
F
tl [ns]
F
tl [ns]
13
7.1
0.05
0.68
0.09
0.92
0.16
1.40
(1)
(0.2)
(0.01) (0.03) (0.01) (0.03) (0.02) (0.04)
The relaxivity in buffer solution and the HSA-binding
data were used to calculate the bound relaxivity NMRD
profiles.The high-field region is shown in Figure 3.Despite
[a] Global (HSA) rotational correlation time fixed at 41 ns. Numbers in
parentheses represent one standard deviation.
6768 www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 6766 –6769

Angewandte
Chemie
increase.The local correlation time also increases in the order
1 < 2 < 3, which suggests that compound 3 is the most rigid
when bound to albumin.Although the order parameters are
relatively low, it was noted that they correlated with the global
correlation time—arbitrarily chosen shorter global correla-
tion times also gave good fits, but the order parameter was
increased.Although the Lipari–Szabo approach models the
data well, motion within this system is complex and care
should be taken in comparing these rotational parameters
with other unrelated systems.
There have been gains reported in increasing the relax-
ivity of single gadolinium complexes through the optimization
of water exchange, hydration number, etc.Multilocus binding
may prove to be a useful technique for maintaining this high
relaxivity when these complexes are incorporated into
targeted multimeric MRI contrast agents.
[1] H.Kobayashi, M.W.Brechbiel, Curr. Pharm. Biotechnol. 2004,
5, 539.
[2] M.P.Lowe, Aust. J. Chem. 2002, 55, 551.
[3] R.B.Lauffer, Magn. Reson. Med. 1991, 22, 339.
[4] 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,
Am. Chem. Soc. 2002, 124, 3152.
J.
[5] P.F.Sieving, A.D.Watson, S.M.Rocklage,
Bioconjugate Chem.
1990, 1, 65.
[6] M.Spanoghe, D.Lanens, R.Dommisse, A.van der Linden, F.
Alder Weireldt, Magn. Reson. Imaging 1992, 10, 913.
[7] G.Schuhmann-Giampieri, H.Schmitt-Willich, T.Frenzel, W.R.
Press, H.J.Weinmann, Invest. Radiol. 1991, 26, 969.
[8] A.A.Bogdanov, R.Weissleder, H.W.Frank, A.V.Bogdanova,
N.Nossif, BK. .Schaffer, E.Tsai, MI..Papisov, TJ..Brady,
Radiology 1993, 187, 701.
[9] R.Rebizak, M.Schaefer, E.Dellacherie,
Bioconjugate Chem.
1998, 9, 94.
[10] C.Casali, M.Janier, E.Canet, J.F.Obadia, S.Benderbous, C.
Carot, D.Revel, Acta Radiol. 1998, 5, S214.
[11] E.Toth, I.van Uffelen, L.Helm, A.E.Merbach, D.Ladd, K.
Briley-Saebo, K.E.Kellar, Magn. Reson. Chem. 1998, 36, S125.
[12] R.S.Ranganathan, M.E.Fernandez, S.I.Kang, A.D.Nunn,
Experimental Section
The compounds 1, 2, and 3 were synthesized in an analogous fashion
(see Supporting Information for full details).4-Mesitylbenzoic acid
was coupled to zero, one, or two of the primary amine groups of
triethylenetetraamine.After reduction of the amide, the resultant
tetraamine was coupled to four protected DTPA moieties.Depro-
tection, chelation, and purification by HPLC yielded the complexes
shown in Figure 2.All Gd concentrations were determined by ICP-
MS (Agilent 7500).NMRD profiles were recorded on a Koenig-
Brown field-cycling relaxometer.
P.C. Ratsep, K.M.R. Pillai, X. Zhang, M.F. Tweedle,
Invest.
Radiol. 1998, 33, 779.
[13] J.C.Amedio, Jr,. G.Van Wagenen, Jr,. G.Zavlin, A.Gyorkos,
S.A.Peterson, Synth. Commun. 2000, 30, 3755.
[14] P.Caravan, J.J.Ellison, T.J.McMurry, R.B.Lauffer,
1999, 99, 2293.
Chem. Rev.
[15] J.S.Troughton, M.T.Greenfield, J.M.Greenwood, S.Dumas,
A.J.Wiethoff, J.Wang, M.Spiller, T.J.McMurry, P.Caravan,
Inorg. Chem. 2004, 43, 6313.
Received: June 26, 2005
Published online: September 20, 2005
[16] P.Caravan, A.V.Astashkin, A.M.Raitsimring,
2003, 42, 3972.
[17] J.D. Dattelbaum, O.O. Abugo, J.R. Lakowicz,
Chem. 2000, 11, 533.
Inorg. Chem.
Keywords: gadolinium · imaging agents · MRI ·
NMR spectroscopy · relaxivity
Bioconjugate
.
Angew. Chem. Int. Ed. 2005, 44, 6766 –6769
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6769