Coupling fast water exchange to slow molecular tumbling in Gd3+ chelates: Why faster is not always better

Article

pubs.acs.org/IC

Coupling Fast Water Exchange to Slow Molecular Tumbling in Gd³⁺

Chelates: Why Faster Is Not Always Better

Stefano Avedano,^†Mauro Botta,*^,†Julian S. Haigh,^‡Dario L. Longo,^§and Mark Woods*^,‡,∥

^†Dipartimento di Scienze e Innovazione Tecnologica, Universita

I-15121 Alessandria, Italy

̀

del Piemonte Orientale “Amedeo Avogadro”, Viale T. Michel 11,

^‡Department of Chemistry, Portland State University, 1719 SW 10th Ave, Portland, Oregon 97201, United States

^§Dipartimento di Biotecnologie Molecolari e Scienze per la Salute and Molecular Imaging Center, Universita

I-10126 Torino, Italy

^∥Advanced Imaging Research Center, Oregon Health & Science University, 3181 S.W. Sam Jackson Park Road, Portland,

Oregon 97239, United States

̀

di Torino, Via Nizza 52,

S

* Supporting Information

ABSTRACT: The inﬂuence of dynamics on solution state

structure is a widely overlooked consideration in chemistry.

Variations in Gd³⁺chelate hydration with changing coordina-

tion geometry and dissociative water exchange kinetics substan-

tially impact the eﬀectiveness (or relaxivity) of monohydrated

Gd³⁺chelates as T₁-shortening contrast agents for MRI. Theory

shows that relaxivity is highly dependent upon the Gd³⁺−water

proton distance (r_GdH), and yet this distance is almost never

considered as a variable in assessing the relaxivity of a Gd³⁺

chelate as a potential contrast agent. The consequence of this

omission can be seen when considering the relaxivity of

isomeric Gd³⁺chelates that exhibit diﬀerent dissociative water

exchange kinetics. The results described herein show that the

relaxivity of a chelate with “optimal” dissociative water exchange kinetics is actually lower than that of an isomeric chelate with

“suboptimal” dissociative water exchange. When the rate of molecular tumbling of these chelates is slowed, an approach that has

long been understood to increase relaxivity, the observed diﬀerence in relaxivity is increased with the more rapidly exchanging

(“optimal”) chelate exhibiting lower relaxivity than the “suboptimally” exchanging isomer. The diﬀerence between the chelates

arises from a non-ﬁeld-dependent parameter: either the hydration number (q) or r_GdH. For solution state Gd³⁺chelates, changes

in the values of q and r_GdHare indistinguishable. These parametric expressions simply describe the hydration state of the

chelatei.e., the number and position of closely associating water molecules. The hydration state (q/r_GdH⁶) of a chelate is

intrinsically linked to its dissociative water exchange rate k_ex, and the interrelation of these parameters must be considered

when examining the relaxivity of Gd³⁺chelates. The data presented herein indicate that the changes in the hydration parameter

(q/r_GdH⁶) associated with changing dissociative water exchange kinetics has a profound eﬀect on relaxivity and suggest that

achieving the highest relaxivities in monohydrated Gd³⁺chelates is more complicated than simply “optimizing” dissociative water

exchange kinetics.

criteria by which these agents discriminate between tissue types

limit the diagnostic information that can be obtained by admi-

nistering these contrast agents. These limitations have provoked

the idea of a new class of contrast agent: so-called “targeted

agents” would possess a structural component that is designed

to bind to a biomarker associated with a disease of interest.¹

Binding of the agent will increase the localization of the agent

in regions where that biomarker is more abundant, thereby

increasing MR signal intensity of regions associated with the

disease. Binding has one further eﬀect: it will slow the agent’s

INTRODUCTION

■

Chelates of Gd³⁺are now routinely administered as contrast

agents for MRI. After intravenous injection, these chelates

extravasate through the pores at the endothelial junctions of the

vasculature into interstitial space. Time dependent modulation

of the dipolar interactions between the seven unpaired elec-

trons of Gd³⁺and proximate water protons leads to a shorten-

ing of the water proton longitudinal relaxation time constant

(T₁). In T₁-weighted MR images, this results in enhanced signal

intensity in those regions to which the agent is distributed.

For agents that are currently in clinical use, distribution is a

function of a number of vasculature characteristics, such as

blood ﬂow and pore size, and the size of the agent. The limited

Received: February 5, 2013

A

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

rate of molecular tumbling, characterized by the correlation

time τ_R.

agents, the Gd³⁺ion is chelated by an octadentate polyamino-

carboxylate ligand (DTPA or a derivative, or DOTA (1,4,7,10-

tetraazacyclododecane-1,4,7,10-tetraacetate) or a derivative),

and this leaves one coordination site available for occupation by

water. Opening additional coordination sites to water, and

therefore decreasing the sites coordinated by the chelating

ligand, is generally found to cause the stability constant of a

chelate to drop by as much as 3 orders of magnitude,¹³

suggesting that in practice, with perhaps a few exceptions,¹⁴⁻¹⁶

only q = 1 chelates are suitable for in vivo use.

The theory of paramagnetic relaxation given by the Solomon−

Bloembergen−Morgan (SBM) equations (eqs 1−3) describes

the relaxation of water protons that occurs through exchange of

water molecules coordinated directly to the paramagnetic metal

center.²⁻⁶These equations tell us that reducing the rate of molec-

ular tumbling (making τ_Rlonger) will aﬀord a more eﬀective

contrast agent.⁷⁻⁹This is critical because the Gd³⁺chelates that

are currently used clinically have high detection limits; a typical

dose (0.1 mmol kg⁻¹) equates to about 4.5 g of agent in a single

bolus for a 70 kg human. Clearly, if biomarkers of disease,

which are present only in very low abundance, are to be

detected by MRI, then it is absolutely critical that the detection

limit of any “targeted agent” be much lower than those of

agents currently employed. To this end the SBM equations

have been used to guide research into improving the function

of MRI contrast agents. These equations reveal several param-

eters that may be manipulated by the chemist to control the

eﬀectiveness of an agentdeﬁned as the longitudinal relaxivity, r₁.

From the SBM equations, τ_Ris found to limit the relaxivity of

low molecular weight clinical agents. τ_Ris readily made longer

either by increasing the size of the agent or by coupling its

molecular motion to that of a large (or stationary) structure

such as a protein or cell. However, it is commonly found that

when τ_Ris made longer, the relaxivity of a Gd³⁺chelate is

subsequently limited by the kinetics of inner-sphere water

exchange.¹⁰If τ_M, the residence lifetime of a water molecule on

Gd³⁺, is too long, then the coordination site on Gd³⁺is

needlessly occupied by a “relaxed” water molecule preventing

relaxation of other water molecules, slowing the catalysis of

bulk relaxation by the relaxation agent. This is the situation that

prevails in all clinically approved Gd³⁺chelates; if the limiting

eﬀect of τ_Ris lifted, then slow water exchange kinetics become

limiting and the highest relaxivities are not realized.¹⁰This

realization has prompted considerable research eﬀort into the

development of more rapidly exchanging Gd³⁺chelates that

would, in principle, aﬀord the highest relaxivities if the limiting

eﬀect of τ_Rwere lifted.

The recent observation of nephrogenic systemic ﬁbrosis

(NSF) in some renally compromised patients after admin-

istration of DPTA-based contrast agents further highlights the

importance of chelate stability as a consideration in contrast

agent development.^17,18Even though Gd³⁺chelates derived

from DTPA are q = 1 chelates, it is now generally accepted that

they are not suﬃciently robust to survive prolonged in vivo

residence completely intact and are only safe if the entire dose

is excreted rapidly.^12,19−21Targeted imaging applications also

envisage prolonged in vivo residence lifetimes, and as such Gd³⁺

chelates derived from DTPA should not be considered suited

for the purpose. The Gd³⁺chelates of DOTA and its derivatives

are more thermodynamically stable and more kinetically robust

than their DTPA counterparts.²²They are more suitable for

applications that require prolonged in vivo lifetimes, and

controversy surrounds the isolated cases in which GdDOTA

has been associated with cases of NSF.²³In general, DOTA

derivatives are considered suﬃciently stable for prolonged

in vivo residence lifetimes, and therefore for targeted applica-

tions. A further advantage of DOTA-derived chelates is the

superior electron spin relaxation characteristics (T_1eand T_2e) of

these chelates. Electron spin relaxation is sometimes cited as

another modiﬁable parameter that regulates relaxivity. How-

ever, even though considerable eﬀort is being made to better

understand this parameter,²⁴⁻²⁷at present its relationship with

coordination chemistry is vague at best, and at present the

chemist is really faced with accepting what a given chelate

provides. However, in general DOTA derivatives are on the

whole found to exhibit somewhat more favorable electron spin

relaxation characteristics than their DTPA-based counter-

parts.²⁸

⎛

⎜

⎞

q

1

r₁^is=

⎟

⎠

55.5 T + τ_M

⎝

(1)

Finally, one further parameter that also relates to hydration

of the chelate can potentially have a profound eﬀect upon

relaxivity. The distance from the metal to the inner sphere

water protons, r_GdH, is most widely viewed as a ﬁxed value. In

spite of this, the value used in the calculations and ﬁtting of

relaxation data of Gd³⁺chelates varies quite a bit, with values

anywhere from 2.9 to 3.1 Å commonly used.²⁹Given that

relaxivity scales to the negative sixth power of r_GdH, even small

changes in this value would be expected to have profound

eﬀects on relaxivity. Parker and co-workers have demonstrated

that faster dissociative water exchange kinetics are associated

with weaker (i.e., longer) Gd³⁺−OH₂bonds.^30,31Our own

results suggest that diﬀerences in r_GdHassociated with diﬀering

dissociative water exchange rates may not be conﬁned solely to

consideration of “bond” distances but that the diﬀerences in

1M

2

⎡

⎢

⎣

μ

⎛

⎞

⎟

7τ

1

2

1

0

2

c2

⎜

=

γ_Hγ_Sℏ S(S + 1)

1 + (ω τ )²

r_G⁶_dH

⎝

⎠

T

15 4π

1M

S c2

⎤

⎥

⎦

3τ

c1

+

1 + (ω_Hτ_c1)²

(2)

(3)

1

τ_Ci

1

τ_M

1

τ_R

1

=

+

i = 1, 2

T

ie

Several other parameters expressed in the SBM equations are

also potentially under the control of the chemist. Relaxivity scales

proportionally with the number of water molecules coordinated

directly to the Gd³⁺ion, q, so increasing the number of water

molecules directly coordinated to Gd³⁺will aﬀord higher

relaxivities. However, increasing the number of vacant

coordination sites on Gd³⁺is also found to reduce the stability

of the chelate. Unchelated, the Gd³⁺ion is quite toxic to

mammals (LD₅₀∼ 0.35 mmol kg⁻¹, i.v. in mice)¹¹and must

therefore be administered in the form of a kinetically and

thermodynamically robust chelate.^10,12In all clinically approved

r

_GdHmay be even greater in solution.³²⁻³⁴These diﬀerences in

r_GdHare comparatively smallcertainly less than the range of

values commonly employed in the published data analyses.

Such small variations in r_GdHare very diﬃcult to quantifythey

are less than the reported uncertainty in ENDOR measure-

ments for instance,³⁵and yet they could be large enough that

they signiﬁcantly impact relaxivity.

B

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Chart 1. The Structures of Conformationally Rigid Ln³⁺

Chelates

Several research groups around the world, including ourselves,

have developed q = 1 Gd³⁺chelates that possess very fast

dissociative water exchange kinetics.³⁶⁻⁴¹However, our chosen

approach is unique in that in addition to a chelate with very fast

dissociative water exchange kinetics the same methods can be

applied to generate an isomeric chelate with slower dissociative

water exchange kinetics.^34,42This allows us, for the ﬁrst time, to

compare the eﬀects of accelerating inner-sphere dissociative

water exchange on relaxivity. To achieve this aim in slowly

tumbling chelates, we have prepared prototypical targeted

contrast agents that employ a simple hydrophobic group that

will cause the agent to bind to various slowly tumbling systems,

such as human serum albumin (HSA). The advantages of this

approach are that it is easy to achieve and binding to HSA is

comparatively well understood; furthermore it also allows our

results to be compared to those obtained for related systems

that also bind HSA.

RESULTS AND DISCUSSION

■

Gd³⁺chelates of DOTA-type ligands can exist in two coordina-

tion geometries: a square antiprism (SAP) and a twisted square

antiprism (TSAP). It has long been appreciated that the water

exchange kinetics of the TSAP isomer are considerably faster

than those of the SAP isomer.^33,43To attain the highest

relaxivities, a TSAP isomer should therefore be the preferred

coordination geometry. In solution, these two coordination

isomers can interconvert, and a mixture of both is usually

observedfor GdDOTA itself, the SAP isomer is found to be

the predominant structure (∼85%).⁴⁴However, by appropri-

ately substituting the ligand framework, the conformational

changes by which the two isomers exchange can be “frozen

out,” producing a chelate that is “locked” into a single

coordination isomer (Chart 1).^45,46Careful consideration of

the stereochemistry enables one coordination isomer to be

selected over the other.⁴²The two chelates S-RRR-Ln1 and S-

SSS-Ln1 are locked into the SAP and TSAP coordination

geometries, respectively (Figure 1). Previous studies have

demonstrated that the water exchange kinetics of S-SSS-Gd1

(τ_M²⁹⁸≈ 6 ns) are more or less optimal for attaining the highest

relaxivities at current imaging ﬁelds (1.5 T).³⁴In contrast, the

water exchange kinetics of S-RRR-Gd1 are considerably slower,

τ_M²⁹⁸≈ 70 ns and yet close to the optimal value (∼ 20 − 40 ns)

predicted for lower magnetic ﬁeld strengths (0.5 T).³⁴

Prototypical “targeted” contrast agents derived from these

conformationally rigid isomeric chelates would allow the eﬀect

of varying dissociative water exchange kinetics to be probed in

the context of a targeted imaging approach. The two chelates

were modiﬁed to incorporate a hydrophobic biphenyl group

(Gd4) that could then be used to slow the rate of molecular

tumbling of each chelate through a binding interaction with

either poly-β-cyclodextrin (poly-β-CD) or human serum

albumin (HSA).

Synthesis. The preparation of the isomeric chelates S-RRR-4

and S-SSS-4 is shown in Scheme 1. The preparation of S-RRR-1

and S-SSS-1 has been reported previously,³⁴and functionaliza-

tion of these ligands, by conversion to the corresponding isothio-

cyanate, is facile.⁴⁸Catalytic hydrogenation using H₂over 10%

palladium on carbon aﬀorded the corresponding primary amines, 2,

in near quantitative conversions. The primary amines 2 were

then converted into the isothiocyanates 3 by reaction with

thiophosgene. Biphasic reaction conditions were employed with

the thiophosgene dissolved in chloroform and the amine 2

dissolved in water at pH 2. The pH of the reaction is crucial to

Figure 1. The lowest energy conformations of S-RRR-Gd4 (left,

orange) and S-SSS-Gd4 (right, green) obtained by simulated annealing

molecular dynamics conformational analysis. These structures high-

light the diﬀerent coordination geometries of the two chelates char-

acterized by diﬀerent torsion angles (φ) and the position of the

hydrophobic substituentin each case located on the corner of the

macrocyclic ring in an approximately similar position to that suggested

by 2D NMR experiments.⁴⁷Hydrogen atoms have been omitted for

clarity.

its success since the stability of isothiocyanates decreases as pH

rises. Isothiocyanates react readily with primary amines under

mildly basic conditions. Accordingly, the isothiocyanates 4 were

dissolved in water, the pH raised to 8 with sodium hydroxide,

and 4-phenylbenzylamine added with dioxane as a cosolvent to

facilitate dissolution. After stirring for 24 h, a solution of the

C

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

a

Scheme 1. The Synthesis of the Lanthanide Chelates of S-SSS-4

a

The chelates of S-RRR-4 were synthesized according to the same synthetic scheme. Reagents and conditions: (i) H₂and 10% Pd on C; (ii) SCCl₂/

CHCl₃/H₂O, pH 2; (iii) 4-phenylbenzylamine/dioxane/H₂O, pH 8; (iv) LnCl₃·6H₂O, pH 6.

appropriate lanthanide chloride in water was added and the

reaction stirred at room temperature for a further 48 h. At

periodic intervals, the pH of the reaction was monitored and

maintained above 6 by the addition of sodium hydroxide. The

Ln³⁺chelates of the two biphenyl conjugates S-RRR-4 and

S-SSS-4 were isolated and puriﬁed by preparative HPLC. Each

chelate was isolated as the more favored “corner” isomer^47,49

aﬀording the structures shown in Figure 1. The chelates

obtained thusly are isolated as the conjugate acid and are poorly

soluble in aqueous solution; however, neutralization, by

stoichiometric addition of NaOH, aﬀords the chelates as their

sodium salts, in which form the chelates are freely soluble in

water.

Relaxometric Studies of Biphenyl Conjugates in

Solution. The relaxivity of a discrete Gd³⁺chelate in solution

can be determined by measuring the proton relaxation rate con-

stant, R₁(= 1/T₁) over a range of Gd³⁺concentrations. Because

R₁and [Gd³⁺] are linearly related in this case, regression

analysis of these data aﬀords the relaxivity, r₁, of the chelate

from the slope of the line. This standard method of relaxivity

determination was applied to the chelates S-RRR-Gd4 and

S-SSS-Gd4. Unexpectedly, however, neither chelate exhibited a

linear dependence of R₁on [Gd³⁺] (Figure 2). In the con-

1

Figure 2. Plots of the paramagnetic H relaxation rate (R_1p) versus

centration range 5 to 1 mM, linear relationships were observed

Gd³⁺chelate concentration for S-RRR-Gd4 (top) and S-SSS-Gd4

aﬀording unusually high relaxivity values for each chelate

(Table 1 and Supporting Information). An abrupt change in the

(bottom) at 20 MHz and 25 °C, highlighting the change in relaxivity

with changing chelate concentration.

slope of the line is then observed followed by a second linear

region over the concentration range 300 to 10 μM. In this

Table 1. Physicochemical Properties of S-RRR-Gd4 and

region, the slope of the line aﬀorded relaxivity values much

S-SSS-Gd4 Obtained from the 1/T₁versus Concentration

closer to those normally expected for low molecular weight

Plots in Figure 2

chelates at 20 MHz and 25 °C (Table 1). This behavior is very

r₁/mM⁻¹s⁻¹

similar to that observed for other Gd³⁺chelates incorporating

long hydrocarbon groups, such as C17-AAZTA⁵⁰or a GdDOTA-

a

chelate

cmc /mM

below cmc

above cmc

calix[4]arene derivative.⁵¹As in that case, the change in relaxivity

can be attributed to the formation of micelles, in which the

rotation of Gd³⁺is slowed, at higher concentrations. Gd³⁺

chelates with large aromatic substituents, such as calix[4]arenes,

have previously been shown to form micelles.⁵¹The critical

micelle concentrations (cmc) of the Gd4 chelates are somewhat

below 0.5 mM, in closer agreement with the value determined

for C17-AAZTA than calix[4]arene substituted DOTA derivatives

S-RRR-Gd4

S-SSS-Gd4

0.42 0.02

0.29 0.02

10.7

9.0

40.8

28.2

a

cmc = critical micelle concentration.

(Table 1).^50,51The relaxivity values of both isomers of Gd4,

both above and below the cmc, are somewhat higher than

found for either the calix[4]arene⁴⁹derivatives or C17-AAZTA.⁵⁰

D

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

In the former case, this is most likely a reﬂection of the slow

exchange kinetics exhibited in this system.⁵¹The latter case is

noteworthy because C17-AAZTA is a q = 2 chelate and would

therefore be expected to have a somewhat higher relaxivity than

either Gd4 isomer: both q = 1 chelates. Given that the water

exchange kinetics of C17-AAZTA and S-RRR-Gd4 are

comparable, this suggests that the primary diﬀerence between

the chelates must arise from more eﬀective slowing of rotation

with the biphenyl substituent relative to that aﬀorded by the

hydrocarbon chain, perhaps through continued, weak inter-

molecular interactions. Strictly analogous behavior has also

been reported and discussed in the cases of functionalized Gd³⁺

DTPA,⁵²DOTA,⁵³and EGTA⁵⁴chelates.

The formation of micelles can unambiguously be demon-

strated by recording the nuclear magnetic relaxation dispersion

(NMRD) proﬁles of each chelate at concentrations above and

below the cmc. The NMRD proﬁles at low concentrations

(0.2 mM) of S-RRR-Gd4 and S-SSS-Gd4 are characteristic of

small molecule Gd³⁺chelates in which relaxivity is limited by τ_R

(Figure 3).¹⁰The relaxivity is higher at very low ﬁelds, and

diﬃculty the scientist faces in describing hydration. Both q and

r

_GdHare parametric descriptors of hydration, and the problem is

how to deﬁne each for a chelate in aqueous solution and

therefore in exchange. It has been possible to account for some

of the observed coordination chemistry of the later lanthanides

(Ho³⁺→ Lu³⁺) by invoking a TSAP coordination isomer that is

entirely dehydrated (q = 0) in solution.^44,55−57However, there

is no published evidence for the existence of a discrete, dehy-

drated TSAP isomer of DOTA-type chelates of lanthanides

from the middle of the series (Eu³⁺and Gd³⁺). In systems

where changes in hydration lead to changes in the coordination

5

geometry, the nondegenerate D₀↔⁷F₀transition of the Eu³⁺

ion can be used for probe hydration equilibria.⁵⁸The Eu³⁺

coordination geometry in DOTA-type chelates changes only

subtly with changing hydration,⁵⁶and the high resolution

emission spectra (Supporting Information, S1) of both S-RRR-

Eu4 and S-SSS-Eu4 each exhibit a single line for this transition

at about 578 nm. There are three possible explanations for this

observation: each chelate has a single hydration species (even

though the S-SSS-Eu1 is q = 0.74 as determined by Horrocks’

method);³¹any change in hydration does not aﬀord suﬃcient

change in the energy of this transition to resolve the diﬀerent

hydration species; or the spectrum aﬀords a time-average of all

hydration species in solution.⁵⁹As discussed previously, there

exists no good method for accurately determining r_GdH

in solution. Although Caravan and co-workers have proposed

ENDOR as a means of probing this parameter,³⁵it is important

to note that those experiments were performed at very low

temperatures on static chelates in frozen glasses, a condition

that does not in any way resemble the situation of a chelate

undergoing exchange in solution. Herein lies the quandary; the

hydration states of S-RRR-Eu1 and S-SSS-Eu1 are very dif-

ferent when determined by Horrocks’ methodq = 0.97 and

q = 0.74, respectivelyhow should this diﬀerence be

interpreted? Insight can be gained from two very diﬀerent

pioneering studies. The work of Parker and co-workers³¹has

unambiguously shown that Horrocks’ method does not provide

simply the hydration number. In fact, it provides both the

hydration number and the position of those water molecules in

a single parameter; i.e., Horrocks’ hydration state describes both

q and r_EuHin one parameter. Signiﬁcantly, the hydration state

also accounts for molecules in the second coordination sphere

as well. In his pioneering relaxometric studies on paramagnetic

metal ions, Bertini et al. encountered the same problem, and

recognizing what Parker would later demonstrate, he used a

single parameter to describe the hydration state of the metal

1

Figure 3. The H 1/T₁NMRD proﬁles of S-RRR-Gd4 (SAP, open

diamonds) and S-SSS-Gd4 (TSAP, closed circles) recorded at 0.2 mM

and 25 °C. Dashed lines represent the calculated outer-sphere

contribution to relaxivities.

there is a dispersion around 1 MHz followed by a more gradual

decrease in relaxivity at higher ﬁelds. However, the proﬁles have

two points of interest. First, the relaxivity of the agents is quite

high for discrete chelates, and this can be attributed to either

the increase in hydrodynamic volume associated with the

inclusion of a bulky biphenyl substituent in the structure, or

weak intermolecular associationsreducing the rate of

molecular tumbling (longer τ_R). Second, and perhaps

more signiﬁcantly, the two curves are almost identical across

the entire frequency range of the proﬁle but oﬀset from one

another by about 1.5 mM⁻¹s⁻¹. This same observation can be

made retrospectively for the previously reported NMRD

proﬁles of S-RRR-Gd1 and S-SSS-Gd1.³⁴The fact that these

proﬁles are almost superimposable but oﬀset in this way tells us

that the primary diﬀerence between the two chelates cannot

arise from a parameter whose eﬀect on relaxivity is modulated

by changes in B₀such as τ_M, τ_R, T_1e, or T_2e. There is a very

slight additional diﬀerence in the dispersions (0.5 to 5 MHz) of

the two proﬁles that suggests a small diﬀerence in one of these

ﬁeld-dependent parameters. The primary diﬀerence between

60,61

6

ion: q/r_MH

.

The hydration states of the two Eu1 chelates

determined by Horrocks’ method unambiguously demonstrate

a diﬀerence in the hydration states of the two chelates.³⁴This

diﬀerence can be represented either as a diﬀerence in q, as a

diﬀerence in r_GdH, or more properly as a diﬀerence in q/r⁶.

For the purposes of the data analyses performed herein, we

have employed values of q/r_GdH⁶= 1372 nm⁻⁶and 1127 nm⁻⁶

for S-RRR-Gd4 and S-SSS-Gd4, respectively. For reasons of

simplicity and familiarity, these values can be considered to

represent two q = 1 chelates in which r_GdHincreases by 0.1 Å or

3.3% (from 3.0 Å to 3.1 Å) on passing from S-RRR-Gd4 to

S-SSS-Gd4. Throughout this article, we will employ this descrip-

tion of hydration; however, the reader must recognize two

important points: (1) Although we will describe q as ﬁxed with

only r_GdHin variance, in truth either or both hydration

parameters may be diﬀerent between the two chelates. (2) The

the two chelates must be a parameter with which relaxivity

6

scales directly: in other words, either q or r_GdH

.

Hydration in Ln³⁺chelates has been something of a

contentious matter for over a decade now. The disagreements

have largely arisen from diﬀering opinions of how to describe

diﬀerences in hydration between chelates. The debate may

appear largely one of semantics, but it arises because of the

6

values of q/r_GdHfor each chelate are not known; the values,

E

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

and the diﬀerence between them, that are employed herein are

more illustrative than precise representations of the hydration

in these chelates. However, given the previously noted diﬀerences

of r_LnH(2.8% in the crystal, 4.5% in solution)³²⁻³⁴between

SAP and TSAP isomers, the diﬀerence seems reasonable, and

the values of r_GdHlie within the range of commonly used values

for this type of analysis. Any errors in the values employed will

be reﬂected in deviations in the values of τ_Rfrom the real

values. Muller and co-workers have shown that when ﬁtting

NMRD proﬁles, variance in the value of r_GdH, usually ﬁxed,

causes a variance in the obtained value of τ_R.²⁹Thus, unless τ_R

is independently determined and ﬁxed during ﬁtting, there is

some ﬂexibility to the value of r_GdHemployed provided that the

eﬀect on the value of τ_Ris appreciated. This ﬂexibility may

explain why it has not previously been found necessary to

consider the eﬀect of variation in hydration (q/r_GdH⁶) between

chelates when undertaking these ﬁttings.

Taking this approach to hydration, both NMRD proﬁles ﬁt

well to theory. In all NMRD ﬁttings herein, the water exchange

parameter τ_Mhas been taken from studies on the corresponding

isomer of Gd1; it is not possible to measure the water exchange

kinetics of Gd4 chelates directly by ¹⁷O NMR methods due to

solubility constraints. The assumption was made that the

peripheral incorporation of a biphenyl group has no signiﬁcant

eﬀect upon the water exchange kinetics of the chelate. The τ_M

value for S-RRR-Gd4 was ﬁxed at the value determined for the

corresponding isomer of Gd1.³⁴A variation of this τ_Mvalue by

30 ns resulted in a relaxivity change of less than 0.3 mM⁻¹s⁻¹

at all ﬁelds. In the case of S-SSS-Gd4, ﬁtting was found to be

largely insensitive to small changes in τ_Mover a range 5−10 ns,

and data were ﬁtted using a τ_Mvalue, 8 ns, that best ﬁts the

data. The values of τ_Robtained from ﬁtting the low

concentration NMRD proﬁles of both chelates are somewhat

longer than expected for relatively small chelates (Table 2).

usually obtained in this type of exercise for these types of chelate.

Crucially, both isomers have very similar electron spin

relaxation characteristics, a feature that has been previously

established through EPR analysis of analogous chelates to the

two isomers of the Gd1.²⁶Given that the relationship between

electron spin relaxation and coordination chemistry is poorly

understood, it is diﬃcult to point to a direct physical reason as

to why τ_Vis longer than might be expected. However, it is

worth noting that a marked lengthening of this parameter is

very often observed in macromolecular systems.^10,64That the

diﬀerence between the two NMRD proﬁles arises primarily

from the diﬀerence in hydration state between the two chelates

is shown by simulating an NMRD proﬁle using the same

parameters obtained from the ﬁtting of the proﬁle of S-RRR-

Gd4 but extending r_GdHfrom 3.0 to 3.1 Å. This aﬀords a curve

that almost exactly ﬁts the experimental data obtained for

S-SSS-Gd4 (Supporting Information Figure S2), indicating that

almost the entire diﬀerence between the two proﬁles is

accounted for in the change in r_GdH. Furthermore, constraining

the hydration states of the two chelates to the same value during

ﬁtting aﬀords unrealistic values for other ﬁtting parameters

(supplementary Figure S8 and Tables S1 and S2).

The NMRD proﬁles of the two isomers of Gd4 recorded at

3.98 mM, above the cmc (Figure 4), are characteristic of more

a b

,

Table 2. Best-Fit Parameters of the NMRD Proﬁles of

S-RRR-Gd4 and S-SSS-Gd4 in Figures 3 and 4

1

Figure 4. The high ﬁeld region of the H 1/T₁NMRD proﬁles of

[Gd³⁺] = 0.20 mM

S-RRR-Gd4 S-SSS-Gd4

3.0 3.1

70

0.62

[Gd³⁺] = 3.98 mM

S-RRR-Gd4 S-SSS-Gd4

3.0 3.1

70

2.4

29

S-RRR-Gd4 (SAP, open diamonds) and S-SSS-Gd4 (TSAP, closed

circles) recorded at 3.98 mM and 25 °C. Solid lines represent ﬁts to

the data; the dashed line is a simulated proﬁle taking the ﬁtting

parameters from the proﬁle of S-RRR-Gd4 but applying the water

exchange parameters τ_Mand r_GdHfrom the S-SSS-Gd4 proﬁle. Proﬁles

including the low ﬁeld region are provided in the Supporting

Information (S3, S5, and S7).

b

r_Gd−H/Å

b

τ_M/ns

8

Δ²/10¹⁹s⁻¹

τ_V/ps

τ_R/ps

τ_g/ns

0.61

2.6

42

45

220

20

229

3.84

3.38

0.43

0.33

slowly tumbling chelates: a classical high ﬁeld “hump” is

observed in each case indicating the formation of a slowly

rotating species, micelles, at higher concentrations. It should be

noted that NMRD proﬁles for slowly rotating systems are

shown and ﬁtted herein only in the high ﬁeld region because of

the known limitations of SBM theory in the slowly rotating

regime that render it unable to completely account for the

behavior of more slowly rotating chelates at very low magnetic

ﬁeld strengths.⁶⁵These proﬁles are notable because the high

ﬁeld relaxivity enhancement arising when molecular tumbling is

slow (longer τ_R) is greater for S-RRR-Gd4, the isomer (SAP)

with the more slowly exchanging water molecule. This is in

direct contrast to the general expectation that more rapid water

exchange kinetics (up to an optimal value of about 6 ns at

1.5 T)^7,34will aﬀord higher relaxivities. Fitting these proﬁles to

SBM theory, incorporating the Lipari−Szabo model,^66,67

aﬀords valuable information about the tumbling dynamics of

each chelate in the micelle. The advantage of the Lipari−Szabo

τ_l/ns

S²

0.45

0.34

a

Fitting used a = 3.8 Å, ²⁹⁸D = 2.24 × 10⁻⁵cm²s⁻¹, and q = 1.

Parameter ﬁxed during ﬁtting.

b

This may reﬂect an underestimation of r_GdHon our part.

Alternatively, these longer than expected values could reﬂect an

increased level of intermolecular π−π interaction (without

forming micelles) that will tend to slow rotation and increase

relaxivity as discussed earlier. Eﬀects of this type have been

observed by Merbach and Helm for other chelates including

aromatic groups.^62,63The similarity in the values of τ_Ris

expected since the two chelates are isomeric. The parameters

Δ²and τ_Vare reﬂective of the zero-ﬁeld splitting and its

transient modulation, respectively, which govern the electron

spin relaxation time constants of the chelate. In both isomers of

Gd4, the value of τ_Vis somewhat diﬀerent (larger) than is

F

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

model is that it separates the local and global tumbling motions

of the chelate, allowing the eﬀect of local molecular motion on

relaxivity to be considered. Fitting was undertaken with the

aforementioned diﬀerence of 0.1 Å in the values of r_GdHused

for the two isomeric chelates. The ﬁts aﬀorded comparable

values of τ_gand τ_l, the global and local molecular tumbling

correlation times, respectively. These results indicate that the

micelles formed by each isomer of Gd4 are of comparable size

and that within each the Gd³⁺chelate has broadly similar

freedom of rotation. The diﬀerence in relaxivity of the isomers

of Gd4 in micelles can only partially be the result of diﬀerences

in molecular rotation between the two systems. The major

diﬀerence between the two proﬁles again stems from the longer

r_GdHvalue found for S-SSS-Gd4. This is demonstrated by a

simulated NMRD proﬁle using the ﬁtting parameters from the

proﬁle of S-RRR-Gd4 but employing the hydration parameters

from the S-SSS-Gd4 proﬁleτ_M= 8 ns and r_GdH= 3.1 Å

(dashed line, Figure 4). This simulation accounts for the

majority of the relaxivity enhancement observed for S-RRR-

Gd4 and indicates that the primary limitation to enhancing

relaxivity at high ﬁelds for S-SSS-Gd4 is a combination of the

longer r_GdHvalue observed for this chelate and a value of τ_M

that is rather too short to optimize relaxivity at 20 MHz.

As previously noted the τ_Mvalue of S-SSS-Gd4, at 8 ns, is

somewhat shorter than optimal at 20 MHz, under the tra-

ditional SBM paradigm.^10,34This is reﬂected in a shift of the

maxima of the two relaxivity “humps” in the NMRD proﬁles

shown in Figure 4. The more rapidly exchanging S-SSS-Gd4

exhibits a “hump” maximum that is at 10 to 20 MHz higher

ﬁeld than the more slowly exchanging S-RRR-Gd4, consistent

with expectation based on the diﬀerent water exchange kinetics

of the two isomers. However, it is important to note that even

at the higher ﬁelds where the exchange rate of S-SSS-Gd4

would be considered optimal, the relaxivity of S-RRR-Gd4

Figure 5. The eﬀect of poly-β-CD on the longitudinal relaxation rate

(20 MHz, 25 °C) of dilute solutions (0.17 mM) of S-RRR-Gd4 (SAP,

open diamonds) and S-SSS-Gd4 (TSAP, closed circles).

Table 3. Fitting Parameters for the Binding of the Two

Isomers of Gd4 to poly-β-CD

a

K_a/M⁻¹

r₁^bound/mM⁻¹s⁻¹

S-RRR-Gd4

S-SSS-Gd4

(0.51 0.05) × 10³

(0.87 0.07) × 10³

46 0.7

30 0.5

a

Measured at 20 MHz and 25 °C.

The association constants determined in this way (Table 3)

are probably a good reﬂection of the strength of the interaction

between each isomer of Gd4 and a β-CD unit. Although these

values are an average over all β-CD of the polymer, there is no

reason to suppose that the binding of one chelate by poly-β-CD

will substantially change the binding of any of the others, and

each binding site must be very similar, if subtly diﬀerent. The

binding of S-SSS-Gd4 to β-CD is somewhat stronger than that

of S-RRR-Gd4 even though the same group is responsible for

binding in each case. Since these are interactions between the

chiral host and chiral guest in each case, diﬀerences in the

binding of the two isomers were expected. The results of

molecular modeling studies (below) further highlight how these

diﬀerences in interaction are likely to occur. Unlike K_a, the

relaxivity of Gd4 bound to poly-β-CD is expected to vary

somewhat depending on the location of the β-CD unit in the

polymerchelates bound closer to the middle of the molecular

assembly could reasonably be assumed to have less (or at least

remains higher, and this must be a direct result of the increase

10,34

in r_GdH

.

Interactions of Biphenyl Conjugates with Poly-β-

cyclodextrin. Hydrophobic groups such as the aromatic

substituent of Gd4 are known to form inclusion compounds

with β-cyclodextrin (β-CD). Cyclodextrins themselves are not

suﬃciently large to slow rotation to the extent that substantial

gains in relaxivity can be realized, but polymers of cyclodextrins

are large enough and have previously been used to slow

molecular tumbling and increase relaxivity.⁶⁸⁻⁷⁰Poly-β-cyclo-

dextrin (poly-β-CD) contains an average of 10−11 β-CD units,

each of which can bind and slow the tumbling of Gd4. The

advantage of this approach is that, unlike the serum albumin

binding described below, poly-β-CD aﬀords only approximately

equivalent binding sites and will slow the rotation of each

chelate it binds approximately equivalently. As a consequence,

the binding of Gd4 to poly-β-CD conforms, to a ﬁrst

approximation, to a simple 1:1 binding model. Titrating a

dilute solution of Gd4 with poly-β-CD and measuring the

change in water proton relaxation rate aﬀords typical binding

curves for both isomers of Gd4 (Figure 5). The increase in

water proton relaxation rate as Gd4 is added to poly-β-CD is a

clear indication of a reduction in the rate of molecular tumbling

of the Gd³⁺chelates (longer τ_R) as the chelates bind to the

polymer. The data were ﬁtted to a simple model that considers

slower) motion of rotation than those closer to the ends. In this

bound

light, it is important to treat r₁

values as averaged values

rather than absolute values for a discrete chelate. Nonetheless,

this exercise is highly instructive; the bound relaxivity (Table 3)

of the more rapidly exchanging TSAP isomer again has lower

value than the more slowly exchanging SAP isomer when

molecular tumbling is slowed.

NMRD proﬁles of the two isomers of Gd4 were recorded

under conditions that ensured >96% of the chelate was bound

to poly-β-CD (Figure 6). Again, only the high ﬁeld regions of

the proﬁles are shown and ﬁtted. Notably, the two proﬁles

closely resemble those obtained for the chelates in micelles.

Fitting the proﬁles to SBM theory (including the Lipari−Szabo

model) aﬀords similar rotational correlation times, τ_gand τ_l, for

the two isomers of Gd4 bound to poly-β-CD (Table 4),

indicating that diﬀerences in molecular tumbling between the

two isomers are not the primary cause of the diﬀerence in

relaxivity. Yet again, it is found that the critical impact of the

longer r_GdHvalue of S-SSS-Gd4 is the primary cause of the

lower relaxivity observed for the more rapidly exchanging TSAP

isomer. A proﬁle simulated with ﬁtting parameters for S-SSS-

Gd4, but using the water exchange parameters for S-RRR-Gd4,

the presence of 11 equivalent and independent binding sites,

bound

aﬀording association constants and bound relaxivities (r₁

)

for the two isomeric chelates (Table 3).

G

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Figure 7. Relaxometric titrations of HSA into 100 μM solutions of

S-RRR-Gd4 (SAP, open diamonds) and S-SSS-Gd4 (TSAP, closed

circles) at 20 MHz and 25 °C. A qualitative data ﬁt is provided (Table 5)

using a binding model with three equivalent binding sites.

Figure 6. The ¹H 1/T₁NMRD proﬁles of S-RRR-Gd4 (red diamonds)

and S-SSS-Gd4 (blue circles) recorded at [Gd³⁺] = 0.17 mM in the

presence of 6 mM poly-β-CD at 25 °C.

Table 4. Best-Fit Parameters of the NMRD Proﬁles (25 °C)

Table 5. Binding Parameters Obtained from the Qualitative

Fit of the Relaxometric Titration of HSA with S-RRR-Gd4

and S-SSS-Gd4 Using a 3:1 Binding Model with Equivalent

Binding Sites

of the Inclusion Complexes of the Two Isomers of Gd4 with

a

poly-β-CD

S-RRR-Gd4

S-SSS-Gd4

b

a

K_HSA‑Gd4( × 10³M⁻¹

)

r₁

(mM⁻¹s⁻¹

)

bound

r_Gd−H/Å

3.0

70

3.1

8

b

τ_M/ns

S-RRR-Gd4

S-SSS-Gd4

71.3 12.4

49.7 7.9

46.8 0.9

37.6 0.6

Δ²/10¹⁸s⁻¹

τ_V/ps

τ_g/ns

2.8

3.7

12

18

a

Apparent values (see text).

4.2

3.6

0.45

0.42

τ_l/ns

S²

0.40

aﬃnity of S-RRR-Gd4 appears to be higher than that of S-SSS-

Gd4. Furthermore, the eﬀect of the longer r_GdHvalue for S-SSS-

Gd4 is again evident as its relaxivity is evidently lower than that

of S-RRR-Gd4.

0.45

a

Fitting used a = 3.8 Å, ²⁹⁸D = 2.2 × 10⁻⁵cm²s⁻¹, and q = 1.

Parameter ﬁxed during ﬁtting.

b

Relaxometric titrations cannot discriminate between chelates

bound to diﬀerent sites on the protein, and it is highly unlikely

that either chelate is able to ﬁnd three equivalent sites at which

to bind. To probe the binding interactions in more depth, site

speciﬁc binding assays are required. The work of Sudlow and

co-workers^71,72has provided a great deal of information about

the binding interactions of HSA as well as methods for probing

these interactions. Caravan and co-workers employed those

methods to probe the interactions of the clinical blood pool

agent MS-325.⁷³Given the well-developed nature of this

experimental protocol, identical techniques were used to probe

the binding of each isomer of Gd4 to HSA. A ﬂuorescent probe

speciﬁc for a particular binding site on HSA was added to a

solution of defatted HSA and its displacement by Gd4 followed

by ﬂuorescence. Warfarin is a probe that is known to selectively

bind in what Sudlow designated drug binding site I.⁷¹When

either isomer of Gd4 was titrated into the solution, no

displacement of warfarin from HSA was observed (Supporting

Information, S4). This appears to indicate that there is no

binding of either isomer at this site. However, drug binding site

I is a very large binding domain, and three distinct subdomains,

a, b, and c, have been noted within drug binding site I.⁷⁴

Warfarin binds in subdomain Ia, which is the subdomain

located closest to the mouth of the binding pocket and between

subdomains Ib and Ic. When bound, warfarin is known to

extend partially into both subdomains Ib and Ic, and therefore

binding of Gd4 in either subdomain Ib or Ic was expected to

lead to at least partial displacement of warfarin from drug

binding site I. In light of the results from the relaxometric

titrations, the possibility that drug binding site I is able to

accommodate Gd4 and warfarin simultaneously, without

displacement of warfarin, cannot be excluded. Indeed, the

results of molecular modeling studies (below) suggest that,

from an energetic perspective, binding of the chelates in this

accounts for all of the diﬀerence between the proﬁles at the

high and low ﬁeld regions of the relaxivity “hump” and most

(about 80%) of the “hump” around 20 MHz (dashed line,

Figure 6). The remaining diﬀerences in relaxivity may be

attributed to the small diﬀerences that exist between the local

rotations (τ_l) of the two chelates when bound to poly-β-CD.

The relaxivity maximum observed for each chelate exhibits the

same ﬁeld dependence as observed in the micelle systemthe

more rapidly exchanging isomer peaking at higher ﬁeld

consistent with expectation.

Interactions of Biphenyl Conjugates with Human

Serum Albumin. Titrating human serum albumin (HSA) into

dilute solutions of Gd4 below the cmc clearly demonstrates the

relaxivity enhancement aﬀorded by the interaction of the Gd³⁺

chelate with the protein (Figure 7). As the amount of HSA

present increases, the relaxivity of each chelate increases as the

chelate binds to the protein and molecular tumbling is slowed.

However, unlike poly-β-CD with its approximately equivalent

binding sites, HSA has multiple, very diﬀerent hydrophobic

binding sites and is capable of binding a large number of

hydrophobic molecules simultaneously. Because chelate bind-

ing in proteins such as HSA is allosteric, the binding of Gd4 at

any given site alters the chelate−protein interaction at all other

binding sites. As a consequence, ﬁtting this type of titration data

does not provide information with true physical meaning about

the agent, its binding, or relaxivity. A binding model

incorporating three equivalent binding sites on the protein

describes the data for each chelate quite well and allows a

qualitative assessment of the titration data in Table 5. From the

inﬂection of the binding curve which occurs signiﬁcantly before

a 1:1 stoichiometry, it is clear that both chelates bind

reasonably avidly to more than one site on the protein,

justifying the use a 3:1 binding model. The overall binding

H

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Table 6. The Association Constants of the Two Isomeric Biphenyl Conjugates Gd4 in the Two Drug Binding Sites of HSA

Determined at 25 °C

K_HSA‑Gd4/M⁻¹

a

ﬂuorescent probe

binding site

S-RRR-Gd4

n.d.

9.2 0.5 × 10⁻³

S-SSS-Gd4

MS-325

n.d.

11.8 × 10⁻³

warfarin

I

n.d.

22.1 1.9 × 10⁻³

dansyl sarcosine

II

a

Taken from ref 73 and determined at 37 °C; n.d. indicates that no displacement of the ﬂuorescent probe was detected.

site remains a possibility. In contrast, dansyl sarcosine binds

selectively in what Sudlow designated drug binding site II.⁷¹

Gd³⁺chelates, including the clinical HSA binders MS-325 and

GdBOPTA, are often found to bind in drug binding site II.

Perhaps not unexpectedly then, addition of either isomer of

Gd4 resulted in displacement of the dansyl sarcosine, causing a

decrease in emission intensity (Supporting Information, S4).

Following the methods of Caravan and co-workers, inhibition

constants for the ﬂuorescent probe at drug binding site II can

be determined for each isomer from which the strength of each

association can be calculated (Table 6). The association

constants determined in this way are of a similar magnitude

to that observed for MS-325. Notably the two isomers have

quite diﬀerent association constants for binding at drug binding

site II but these results are in contradiction to those obtained

from the relaxometric titration (Figure 7). The binding of

S-RRR-Gd4, globally stronger from the relaxometric titration, is

the weaker of the two isomers at drug binding site II. Two

factors could contribute to this diﬀerence. First, S-RRR-Gd4

could either be binding more strongly to other sites or binding

to more sites on HSAthe results of molecular modeling

experiments). Given the distribution of environments in which

Gd4 could be found in these systems, any NMRD ﬁtting is

without meaning, and thus none was attempted. The same

overall pattern is observed as for the other slowly tumbling

systems studied herein. High ﬁeld relaxivity “humps” are again

observed for both chelates, but the “hump” maximum observed

for S-RRR-Gd4 is signiﬁcantly higher than that of S-SSS-Gd4.

This can again be attributed to the diﬀerence in metal−water

distance, r_GdH, between the two isomeric chelates. On the

higher ﬁeld edge of the “hump,” the relaxivity of both agents

falls oﬀ quickly, consistent with theory. It is evident that the

relaxivity of S-RRR-Gd4 falls oﬀ more quickly than that of

S-SSS-Gd4 until, at around 40 MHz, it falls below that of

S-SSS-Gd4. This observation is expected on the basis of the

previously determined dissociative water exchange rates of each

chelate, that determined for S-SSS-Gd4 being more suitable for

achieving higher relaxivities at higher ﬁelds, according to the

traditional SBM paradigm.^10,34

Variable Temperature Relaxometry and Molecular

Modeling Studies. One key assumption has been made in the

analysis of all the relaxivity measurements herein: the water

exchange kinetics of both chelates remains virtually unaﬀected

by the interactions that slow the global rate of molecular tum-

bling. This is a point of particular relevance when considering

the agents bound to HSA as it has previously been reported

that the interaction of a chelate with the protein can aﬀect its

water exchange kinetics.^73,75−77The water exchange kinetics of

Gd³⁺chelates are accurately determined by measuring the eﬀect

of changing temperature on the ¹⁷O transverse relaxation rate

of water in a solution of the chelate.^78,79The water exchange

kinetics of the two isomeric Gd1 chelates were determined

using this method.³⁴The limitation of this technique is that it

requires relatively high concentrations of Gd³⁺(typically 10⁻²

M), much higher than the limits of solubility of HSA or poly-β-

CD. Samples cannot then be prepared in which suﬃcient

chelate is present in solution under conditions where the

majority of the chelate is bound to the macromolecule. The

water exchange kinetics of the bound chelate cannot therefore

be determined in this manner. Caravan and co-workers

proposed a method by which absolute quantiﬁcation of water

exchange kinetics could be achieved from variable temperature

¹H relaxation measurements at much lower chelate concen-

trations using the corresponding Dy³⁺chelate.⁷⁵It is generally

considered that Ln³⁺ions may be interchanged to aﬀord

diﬀering types of information that build up a picture of the

overall coordination chemistryas noted earlier where the

Eu³⁺chelate was studied to probe hydration equilibria.

However, in our case, there are concerns that switching to a

heavier Ln³⁺ion (rather than the lighter Eu³⁺ion) will bring the

TSAP isomer closer to a tipping point at which the hydration

state of the chelates in the TSAP coordination geometry is

thought to abruptly begin to decrease. Dy³⁺is adjacent to Ho³⁺

in the lanthanide series, and from recent crystallographic data

on a TSAP Ho³⁺chelate, it is evident that the hydration

studies (below) suggest that even drug binding site I could be

bound

occupied by this chelate. Second, the apparent r₁

value

determined from the relaxometric titration (Table 5), as noted

previously, has no physical relation to the behavior of any single

chelate molecule. This value is a weighted average over chelate

bound to all sites, none of which are expected to have identical

relaxivities, and the model allows for only three bound chelates

when in fact there could be many more. As a result, this

parameter may easily be an underestimate, and if this were

indeed the case, then the value of the association constants

determined from the ﬁtting would tend to be overestimated.

NMRD proﬁles were collected for both isomers under

conditions designed to maximize the amount of Gd4 bound to

HSA (Figure 8). Conditions were chosen under which >99% of

Figure 8. The ¹H 1/T₁NMRD proﬁles of solutions of the two isomers

of Gd4 (0.15 mM) and HSA (1.8 mM): S-RRR-Gd4 (open symbols)

and S-SSS-Gd4 (closed symbols) recorded at 25 °C.

each isomer of Gd4 is estimated to be globally bound to HSA

(80 and 85% bound to site II, on the basis of the displacement

I

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

equilibrium of a TSAP Ho³⁺chelate in solution is complex

indeed.⁸⁰The extent to which the hydration and exchange

kinetics of a TSAP Dy³⁺chelate can be said to reﬂect those of a

TSAP Gd³⁺, with its much simpler hydration behavior, is highly

questionable. For this reason, we took two diﬀerent approaches

to assess the likelihood of changes in water exchange kinetics as

the rate of molecular tumbling is slowed.

First, molecular models were used to examine the orientation

of the chelates when bound to the macromolecules HSA and

poly-β-CD. The results of modeling the interactions between

Gd4 and β-CD suggest very similar interactions for both

isomeric chelates (interaction energies of −24.4 1.4 kcal/mol

and −25.7 0.8 kcal/mol for the S-RRR-Gd4 and S-SSS-Gd4,

respectively), consistent with the experimentally determined

association constants. The orientation of each isomeric chelate

when bound to a β-CD unit (Figure 9) may help explain the

process of both isomers of Gd4 when bound to poly-β-CD. We

may reasonably assume that binding of Gd4 to β-CD does not

signiﬁcantly inﬂuence the rate of water exchange in either

chelateany inﬂuence from this binding can reasonably be

assumed to be very similar for each isomeric chelate.

Although experimentally no displacement of warfarin was

observed, molecular modeling studies suggest that both isomers

of Gd4 are capable of binding in drug binding site I (Figure 10).

Figure 9. Results of the calculated docking procedure applied to β-CD

and S-RRR-Gd4 (orange) and β-CD and S-SSS-Gd4 (green). The

β-CD surface is depicted with a red semitransparent Gauss−Connolly

surface. Both isomers place the para-phenyl group inside the hydro-

phobic cavity, with the thiourea moiety forming hydrogen bonding

interactions with the primary hydroxyl groups of the narrow rim.

Figure 10. Calculated docking of S-RRR-Gd4 (orange) and S-SSS-Gd4

(green) to HSA (a) to drug binding site I and (b) to drug binding site

II. The hydrogen bonding interaction between S-SSS-Gd4 with Ser489

(drug binding site II) is highlighted in purple. Hydrogen atoms have

been omitted for clarity.

observed diﬀerences in the K_afor the two chelates. From the

molecular models, it is clear that in each case the biphenyl

group extends right through the cyclodextrin binding pocket,

and it is the para- substituted phenyl and thiourea group that

are held in the binding cavity. Binding is thus the result of

hydrophobic interactions between the phenyl group and the

inner surface of the cavity and hydrogen bonding interactions

between the thiourea group and the primary hydroxyl groups

on the narrow rim. This binding mode brings the chiral chelate

in close proximity to the wider rim of the β-CD cavity. β-CD

interacts stereoselectively with chiral compounds,⁸¹and here

the secondary hydroxyl groups of the wider rim will interact

diﬀerently with the chelates’ pendant arms depending upon

their orientation (either Λ or Δ). In each case, the hydrophobic

substituent is located on the corner of the macrocycle

(Figure 1) which orients the water coordination site such

that it will rotate pointing up and away from the β-CD unit,

into the bulk water. It would be expected that such an

orientation would minimize interference in the water exchange

They would need to be capable of doing so without displacing

warfarin, and it is possible to model the docking of both

isomers of Gd4 in such a way as to accommodate both Gd4

and warfarin in the binding pocket. This possibility arises only

because the binding pocket of drug binding site I is so large that

accommodation of the Gd³⁺chelate simultaneously with the

ﬂuorescent probe is possible.⁵⁴Despite the size of this binding

pocket, docking calculations show the biphenyl groups held in

the binding pocket with the chelates held at some distance away

from the mouth of the pocket by the para-phenyl linker. This

suggests a considerable degree of freedom of motion on the

part of the chelate end of the molecule. Although the calculated

docking modes orient the water binding face of the isomeric

chelates in diﬀerent directions relative to the protein surface,

they are both pointing essentially outward toward the bulk

water. It is important to bear in mind that these are single low

energy minima of the chelate positions and are not a reﬂection

of the time-averaged orientation of the water binding face.

J

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

Given the apparent freedom of motion of each chelate when

bound to HSA, it seems unlikely that the exchangeable water

molecule of either chelate is constrained to face toward the

protein surface which could result in an apparent deceleration

of water exchange.

Both isomers of Gd4 are experimentally found to bind to

drug binding site II of HSA (Table 6). Docking calculations

position S-SSS-Gd4 more deeply in the pocket than S-RRR-

Gd4, this permits the thiourea to engage the side chain of

Ser489 in hydrogen bonding interactions (Figure 10). Such a

diﬀerence in binding mode may have a signiﬁcant impact on the

freedom of local rotation of the two chelates. The chelate of

S-SSS-Gd4 will be held more closely to the surface of the

protein, which may reasonably be presumed to reduce local

rotation of the chelate. This would tend to increase relaxivity.

However, the diﬀerence in binding mode provides no reason to

suppose that the water exchange rate of either chelate would be

aﬀected by this binding. Both chelates will rotate locally around

a point of attachment (the corner of the macrocycle) which

would seem to maintain the water coordination site of each

isomer pointing away from the protein surface and into the

bulk. These results may account for the stronger binding

interactions observed for the S-SSS-Gd4.

In a qualitative sense, the accuracy of the binding predictions

that come out of these modeling exercises can be tested

through variable temperature ¹H relaxation measurements.

Since a number of parameters that are temperature dependent

control relaxivity, the temperature proﬁle of relaxivity can be

very informative. This is of particular relevance here since τ_M

has a non-negligible eﬀect upon the characteristic correlation

time τ_C(eq 3) when τ_Ris long, i.e. in a slowly rotating system.

Furthermore, τ_Mis a primary determinant of the eﬀectiveness of

the transfer of the paramagnetic eﬀect to the bulk (eq 1).

From the Solomon−Bloembergen−Morgan equations

1

Figure 11. Variation in the paramagnetic H relaxivity of S-RRR-Gd4

(open symbols) and S-SSS-Gd4 (closed circles) at 20 MHz: (a) as

discrete chelates in solution below the cmc, (b) bound to poly β-CD,

and (c) bound to HSA at the same concentrations as in Figure 8. Solid

lines are (a) ﬁts to the data and (b and c) a guide to the eyes only.

is

(eqs 1−3), two limiting cases for inner-sphere relaxivity (r₁)

may be deﬁned:

• The Fast Exchange Regime, τ_M< T_1M. For low molecular

weight Gd³⁺chelates, this condition typically occurs

temperature. As temperature dips below about 15 °C, the

behavior of the two chelates begins to deviate: while S-SSS-Gd4

remains in the fast exchange regime, for S-RRR-Gd4 the value

of τ_Mbecomes so long as the temperature continues to drop

that the chelate drops out of the fast exchange regime and

relaxivity begins to ﬂatten out, eventually dropping below that

of S-SSS-Gd4. These temperature proﬁles can be ﬁtted in terms

of the parameters ΔH_i^#(i = M, V, R, D) using the best-ﬁt

parameters from Table 2 and assuming an Eyring-type behavior

for the parameters τ_R, τ_M, τ_V, and D (the relative diﬀusion

coeﬃcient of solvent and Gd³⁺chelate). This procedure aﬀords

excellent ﬁts that strongly support the previously determined

values of the τ_Mvalues for the two Gd4 isomers.³⁴

In the two slowly tumbling systems (Figure 11b and c), the

relaxivity of S-SSS-Gd4 exhibits an increase in relaxivity with

decreasing temperature similar to that observed for the discrete

chelates, indicating that the chelate remains in the fast exchange

regime throughout. In contrast, the relaxivity of S-RRR-Gd4

decreases with decreasing temperatures, indicating that the

water exchange is slower. The decrease in T_1Marising from

slower molecular tumbling drives the chelate into a slow/

intermediate exchange regime. Notably, in each proﬁle there is

a single temperature at which the relaxivities of the two isomers

are equal, and this temperature increases as the system rotates

more slowly, reﬂecting the eﬀect of decreasing T_1Mon S-RRR-

Gd4 in the slow/intermediate exchange regime. What these

298

when τ_M< 100−200 ns.

• The Slow/Intermediate Regime, τ_M≥ T_1M. In this

regime, the rate of water exchange is so slow that the

condition τ_M> T_1Moccurs over an extended temperature

range.

In the fast exchange regime, T_1Mis the primary determinant

of inner-sphere relaxivity. At low temperatures, τ_Rand T_1,2e, and

therefore τ_C, are long. The result is that T_1Mshortens with

decreasing temperature, causing relaxivity to rise with

decreasing temperature. In slowly tumbling macromolecular

systems, the values of T_1Mare also shorter, and the fast

exchange regime is not reached until τ_M²⁹⁸< 30 ns. In the slow/

intermediate exchange regime, inner-sphere relaxivity decreases

with decreasing temperature and eventually tends toward zero,

following the increase of the water exchange lifetime.

The temperature relaxivity proﬁles of each isomer of Gd4

were recorded under three of the molecular tumbling regimes

described herein: as discrete chelates below the cmc, when

bound to poly-β-CD, and when bound to HSA (Figure 11).

The temperature response of the relaxivity of the discrete

chelates (Figure 11a) reﬂects the known diﬀerence in water

exchange kinetics between the two isomeric chelates. At room

temperature and above, both chelates are in the fast-exchange

regime, and their relaxivity decreases exponentially with

K

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

electromagnet operating in the range 20−80 MHz. The synthesis of the

ligands S-SSS-1 and S-RRR-1 have been reported previously.³⁴

(4S,7S,10S)-α,α′,α″-Trimethyl-[(S)-2-(4-aminobenzyl)]-1,4,7,10-

tetraazacyclododecane-1,4,7,10-tetraacetic Acid (S-SSS-2). The

nitrobenzyl ligand S-SSS-1 (125 mg, 0.187 mmol) was dissolved in

water (10 mL), and 10% palladium on carbon (20 mg) was added.

The reaction mixture was shaken on a Parr Hydrogenator apparatus

for 12 h under H₂(25 psi). The catalyst was removed by ﬁltration and

the solvents lyophilized to aﬀord the title compound as a colorless

solid (111 mg, 94%).

data show is that the relative rates of water exchange of the two

isomers remain consistent across the systems studied herein:

S-RRR-Gd4 is a more slowly (but still rapidly) exchanging

chelate and S-SSS-Gd4 a more rapidly exchanging chelate in all

these systems, and binding of these agents to either poly-β-CD

and HSA does not aﬀect this situation.

CONCLUSIONS

■

Conventional wisdom in the ﬁeld of contrast agent develop-

ment has held that in order to maximize relaxivity it is necessary

to make two modiﬁcations to the low molecular weight Gd³⁺

chelate currently employed in this role. First, the chelate must

be prevented from tumbling rapidly in solution, such that τ_R>

1 ns or so. Second, water exchange must be accelerated to some

optimal value, such that τ_M< 20 ns. While it is undeniably the

case that slow molecular tumbling and fast water exchange hold

the key to the highest relaxivities, the results presented herein

show that conventional wisdom does not have it quite right.

Preparing two isomeric Gd³⁺chelates with very similar elec-

tronic relaxation properties but very diﬀerent dissociative water

exchange kinetics aﬀorded the opportunity to undertake a

direct side by side comparison of the eﬀect of changing water

exchange kinetics on relaxivity. Theory indicates that one

chelate should have had vastly superior relaxivity: that should

have been the one that exchanged water most rapidly (the

TSAP isomer). Instead, what we observe is that the more

slowly exchanging one actually has substantially higher

relaxivity. Analysis of the relaxometric data reveals that the

origin of this unique and unexpected result lies in a diﬀerence in

hydration state (q/r_GdH⁶) between the two isomers. The lower

hydration state of the rapidly exchanging TSAP isomer has a

profoundly limiting eﬀect on relaxivity. These results

demonstrate that, far from being a matter of little importance,

the hydration state (number and position of water molecules in

the inner coordination sphere) can have a very profound eﬀect

on relaxivity. This demonstrates that, contrary to the commonly

performed superﬁcial analyses of theory, simply “optimizing”

water exchange in Gd³⁺chelates is no guarantee that very high

relaxivities will be attained.

¹H NMR (300 MHz, D₂O pD 7), δ: 6.84 (2H, d, ³J_H−H= 7 Hz, Ar),

3

6.62 (2H, d, J_H−H= 7 Hz, Ar), 1.7−3.5 (22H, m br), 0.98 (9H, m,

CH₃). ¹³C NMR (75.5 MHz, D₂O pD 7), δ: 7.3 (CH₃), 7.5 (CH₃), 7.7

(CH₃), 43.3, 45.1, 45.2, 46.9, 47.2, 47.4, 54.8, 56.9, 57.9, 58.0, 58.2,

58.5, 59.8, 116.8 (Ar), 130.1 (Ar), 131.5 (Ar), 144.2 (Ar), 182.3

(CO₂), 182.6 (CO₂), 182.7 (CO₂), 182.8 (CO₂). m/z (ESMS ESI+):

590 (8%, [H₄L + K]⁺), 612 (55%, [NaH₃L + K]⁺), 634 (71%,

[Na₂H₂L + K]⁺), 656 (100%, [Na₃HL + K]⁺). ν_max/cm⁻¹(ATR/pH 7):

3338 (NH), 2968, 2829, 1573 (CO₂), 1462, 1408, 1258, 1227, 1166,

1126, 1032.

(4R,7R,10R)-α,α′,α″-Trimethyl-[(S)-2-(4-aminobenzyl)]-1,4,7,10-

tetraazacyclododecane-1,4,7,10-tetraacetic Acid (S-RRR-2). The

title compound was prepared from S-RRR-1 according to the pro-

cedure employed for S-SSS-2 and was isolated after removal of the

solvents by lyopholization to aﬀord a pale yellow solid (156 mg, 92%).

¹H NMR (300 MHz, D₂O pD 2), δ: 7.23 (4H, m, Ar), 2.6−4.1 (22

H, m br), 1.30 (9 H, m, CH₃). ¹³C NMR (75.5 MHz, D₂O pD 2), δ:

13.6 (2 × CH₃), 14.5 (CH₃), 31.7, 32.4, 46.7 (br), 49.3 (br), 51.8,

53.4, 57.8, 58.7, 59.8, 61.6, 62.4 123.6 (Ar), 128.9 (Ar), 131.2 (Ar),

138.1 (Ar), 172.1 (2 × CO₂), 175.1 (CO₂), 176.0 (CO₂). m/z

(ESMS ESI+): 552 (83%, [H₄L + H]⁺), 574 (45%, [H₄L + Na]⁺) 590

(100%, [H₄L + K]⁺). ν_max/cm⁻¹(ATR/pH 2): 3333 (NH), 2842,

2569, 1713 (CO₂H), 1620, 1556, 1540, 1506, 1473, 1455, 1207,

1163, 1099.

(4S,7S,10S)-α,α′,α″-Trimethyl-[(S)-2-(4-isothiocyanatobenzyl)]-

1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic Acid (S-SSS-3).

The amine S-SSS-2 (108 mg, 0.170 mmol) was dissolved in water

(4 mL) and the pH of the resulting solution adjusted to 2 by the

addition of a dilute HCl solution. Chloroform (6 mL) was added to

the reaction, which was then stirred vigorously at room temperature.

Thiophosgene (68 mg, 0.59 mmol) was added to the reaction, which

was then stoppered and stirred vigorously for 18 h at room tem-

perature. The reaction mixture was then transferred to a separatory

funnel, and the chloroform layer was allowed to run oﬀ. The aqueous

layer was then washed with chloroform (2 × 15 mL). The aqueous

layer was then collected and the solvents removed under reduced

pressure to aﬀord the title compound as a colorless solid (111 mg,

95%).

EXPERIMENTAL SECTION

■

General Remarks. All solvents and reagents were purchased from

commercial sources and used as received. HPLC puriﬁcations were

performed on a Waters δ-Prep 150 HPLC system using a Phenomenex

Luna C-18 reversed-phase (50 × 250 mm) column. In all cases,

absorbance was monitored at 205 and 254 nm. The solvent system

employed for the puriﬁcation of chelates was elution with water

(0.037% HCl) for 5 min and then with a linear gradient to 80% MeCN

and 20% water (0.037% HCl) after 40 min, at a ﬂow rate of 50 mL

¹H NMR (300 MHz, D₂O), δ: 7.35 (4H, m, Ar), 2.6−4.6 (22H, m

br), 1.61 (3H, s br, CHCH₃), 1.50 (3H, s br, CHCH₃), 1.33 (3H, s br,

CHCH₃). m/z (ESMS ESI+): 594 (39%, [M + H]⁺), 616 (100%, [M

+ Na]⁺), 532 (10%, [M+K]⁺). ν_max/cm⁻¹: 3345 (OH), 2986, 2102

(SCN), 1722 (CO), 1516, 1455, 1394 1222, 1160, 1102, 1027.

Anal. Found: C, 44.8%; H, 6.3%; N, 9.7%. C₂₇H₃₉N₅O₉S·3(H₂O)·2-

(HCl) requires C, 45.0%; H, 6.6%; N, 9.7%.

1

min⁻¹. H and ¹³C NMR spectra were recorded on a JEOL Eclipse

270 spectrometer at 270.17 and 67.93 MHz, respectively, or on a

Varian Mercury 300 spectrometer operating at 300.01 and 75.47 MHz,

respectively. Relaxometric HSA titrations were performed on a Bruker

MiniSpec operating at 20 MHz. The 1/T₁nuclear magnetic relaxation

dispersion proﬁles of water protons were measured over a continuum

of magnetic ﬁeld strength from 0.00024 to 0.5 T (corresponding to

0.01−20 MHz proton Larmor frequency) on the fast ﬁeld-cycling

Stelar Spinmaster FFC 2000 relaxometer equipped with a silver

magnet. The relaxometer operates under complete computer control

with an absolute uncertainty in the 1/T₁values of 1%. The typical

ﬁeld sequences used were the NP sequence between 40 and 8 MHz

and PP sequence between 8 and 0.01 MHz. The observation ﬁeld was

set at 13 MHz. Sixteen experiments of two scans were used for the T₁

determination for each ﬁeld. Additional data at higher ﬁelds (30−70 MHz)

were measured on a Stelar Spinmaster relaxometer equipped with a Bruker

(4R,7R,10R)-α,α′,α″-Trimethyl-[(S)-2-(4-isothiocyanatobenzyl)]-

1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic Acid (S-RRR-3).

The title compound was prepared from S-RRR-2 according to the

procedure employed for S-SSS-3 and was isolated after removal of the

solvents under reduced pressure to aﬀord a colorless solid (127 mg,

93%).

¹H NMR (300 MHz, D₂O), δ: 7.15 (4H, m, Ar), 2.5−4.4 (22H, m br),

1.34 (9H, m, CH₃). m/z (ESMS ESI-): 592 (100%, [M − H]⁻). ν_max/cm⁻¹:

2924, 2098 (SCN), 1716 (CO), 1558, 1520, 1506, 1456, 1394,

1204, 1097.

(4S,7S,10S)-α,α′,α″-Trimethyl-[(S)-2-(4-[3-(biphenyl-4-ylmethyl)-

thioureido]phenyl methyl)]-1,4,7,10-tetraazacyclododecane-

1,4,7,10-tetraacetate gadolinium(III) Chelate (H[Gd(S-SSS-4)]). The

isothiocyanate S-SSS-3 (102 mg, 0.150 mmol) was dissolved in water

(5 mL) and the pH of the solution adjusted to 8 (1 M NaOH

L

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

solution). The solution was stirred at room temperature, and a

solution of 4-phenylbenzylamine (38 mg, 0.21 mmol) in dioxane

(5 mL) was added. The reaction mixture was stirred at room

temperature for 18 h. A solution of gadolinium chloride hexahydrate

(64 mg, 0.17 mmol) in water (2 mL) was then added to the reaction,

the pH being maintained at 6 by periodic addition of a 1 M solution of

NaOH. The reaction was stirred at room temperature for 48 h. The

solvents were removed in vacuo and the residue dissolved in a mixture

of water and THF prior to HPLC puriﬁcation. After removal of the

HPLC eluent by lyophilization, the title compound was obtained as a

colorless solid (58 mg, 43%).

chelates was performed using a simulated annealing molecular

dynamics (SAMD) method. High-temperature MD calculations were

carried out at 1000 K with the starting velocities calculated from the

Boltzmann distribution. Each simulation ran for 2000 ps in steps of 0.1

fs with coordinates saved every 2 ps, resulting in 1000 conformations.

Each conformation was subject to an energy-minimization step until

0.01 convergence and then to a second molecular dynamic at 300 K

for 20 interations, followed by conjugate gradient energy minimization

until a convergence of 0.001. Clustering of conformations was perfor-

med by considering two identical conformers when their diﬀerence in

energy was below 1 kcal mol⁻¹and their RMSD less than 3.0 Å. The

structure of β-CD was taken from the CSD (entry code BCDEXD10).

The high-resolution three-dimensional coordinates of human serum

albumin (HSA) were obtained from the Protein Data Bank (PDB

code: 1E7H). Prior to docking calculations, the structures of HSA and

β-CD were prepared by adding hydrogen atoms and completing

missing atoms. The starting positions for the docking procedure were

obtained by modifying the ligand positions and orientations to

optimize binding geometry while ﬁlling the available space in the HSA

drug sites I and II and in the β-CD cavity. Minimization was achieved

by a multistep procedure, until convergence was less than 0.01 kcal

mol⁻¹Å⁻¹. For all calculations, a modiﬁcation of the Amber99 force

ﬁeld⁸²was used with in-house parametrization to treat Gd³⁺chelates

within the framework of the ionic method.⁵⁰The docking procedure

was performed using the Moe-Dock module with Tabu Search with

10 runs, 1000 steps per run. The ligand binding moiety was kept

ﬂexible during the docking calculations. For the β-CD docking, the

solvent was modeled by using a dielectric constant equal to 20,

whereas for the HSA docking an implicit solvation contribution

(continuum model) was included to model solvent eﬀects⁸³in the

docking calculations. The results of the docking calculations were

sorted by utilizing a force-ﬁeld-based scoring function, and for each

isomer the ﬁve best poses were chosen comparing the interaction

energies.

HPLC R_T= 32.77 min. m/z (ESMS ESI−): 930 (100%, [GdL]⁻).

The appropriate isotope pattern was observed. Anal. Found: C, 46.9%;

H, 6.0%; N, 8.0%. C₄₀H₄₉N₆O₈SGd·5(H₂O) requires C, 47.0%; H,

5.8%; N, 8.2%.

(4R,7R,10R)-α,α′,α″-Trimethyl-[(S)-2-(4-[3-(biphenyl-4-ylmethyl)-

thioureido]phenyl methyl)]-1,4,7,10-tetraazacyclododecane-

1,4,7,10-tetraacetate gadolinium(III) Chelate (H[Gd(S-RRR-4)]).

The title compound was prepared from S-RRR-3 according to the

procedure employed for GdS-SSS-4 and was isolated after removal of

the solvents by lyophilization to aﬀord a colorless solid (49 mg, 39%).

HPLC R_T= 31.55 min. m/z (ESMS ESI−): 930 (100%, [GdL]⁻).

The appropriate isotope pattern was observed.

(4S,7S,10S)-α,α′,α″-Trimethyl-[(S)-2-(4-[3-(biphenyl-4-ylmethyl)-

thioureido]phenyl methyl)]-1,4,7,10-tetraazacyclododecane-

1,4,7,10-tetraacetate europium(III) Chelate (H[Eu(S-SSS-4)]). The

title compound was prepared from S-SSS-3 and europium chloride

hexahydrate according to the procedure employed for GdS-SSS-4 and

was isolated after removal of the solvents by lyophilization to aﬀord a

colorless solid (52 mg, 40%).

HPLC R_T= 33.73 min. ¹H NMR (300 MHz, D₂O), δ: 18.97

(NCH₂−H_ax), 17.93 (2H, NCH₂−H_ax), 17.20 (NCH₂−H_ax), 8.35−6.0

(17H, m br, Ar and CH₂Ar), 0.77 (NCH₂−H_eq), 0.84 (2H, NCH₂−

H_eq), −0.56 (CH₃), −1.11 (NCH₂−H_eq), −2.06 (CH₃), −2.36

(NCH₂−H_ax), −2.74 (NCH₂−H_ax), −3.04 (CH₃), −3.99 (NCH₂−

H_ax), −4.76 (NCH₂−H_ax), −5.18 (2H, NCH₂−H_eq) −5.48 (NCH₂−

H_eq), −6.69 (H_ac), −7.14 (H_ac), −7.84 (H_ac), −9.73 (H_ac), −11.57

(H_ac). m/z (ESMS ESI−): 925 (100%, [EuL]⁻). The appropriate

isotope pattern was observed. Anal. Found: C, 41.5%; H, 5.5%; N,

7.1%. C₄₀H₄₈N₆O₈SEuNa·11(H₂O) requires C, 41.9%; H, 6.1%; N,

7.3%.

ASSOCIATED CONTENT

* Supporting Information

■

S

Details of the cmc determination, emission spectra of S-RRR-

Eu4 and S-SSS-Eu4, full NMRD proﬁles of all slowly rotating

systems, displacement and relaxometric titrations and to

determine HSA binding. This material is available free of

charge via the Internet at http://pubs.acs.org.

(4R,7R,10R)-α,α′,α″-Trimethyl-[(S)-2-(4-[3-(biphenyl-4-ylmethyl)-

thioureido]phenyl methyl)]-1,4,7,10-tetraazacyclododecane-

1,4,7,10-tetraacetate europium(III) Chelate (H[Eu(S-RRR-4)]). The

title compound was prepared from S-RRR-3 and europium chloride

hexahydrate according to the procedure employed for GdS-SSS-4 and

was isolated after removal of the solvents by lyophilization to aﬀord a

colorless solid (44 mg, 37%).

AUTHOR INFORMATION

Corresponding Author

■

*Tel.: +39-0131-360253 (M.B.); + 1-503-725-8238 or 1-503-

418-5530 (M.W.). E-mail: mauro.botta@unipmn.it (M.B.);

mark.woods@pdx.edu or woodsmar@ohsu.edu (M.W.).

HPLC R_T= 32.23 min. ¹H NMR (300 MHz, D₂O), δ: 37.99

(NCH₂−H_ax), 36.20 (NCH₂−H_ax), 35.55 (NCH₂−H_ax), 35.34

2

(NCH₂−H_ax), 11.59 (1H, d, J_H−H7 Hz, NCHCH₂Ar), 9.74 (1H, d,

Notes

2

²J_H−H7 Hz, NCHCH₂Ar), 8.45 (2H, aa′, J_H−H13 Hz, NHCH₂Ar),

The authors declare no competing ﬁnancial interests.

3

7.80 (3H, m, Ar), 7.76 (2H, d, J_H−H8 Hz, para-substituted Ar), 7.67

(2H, t, ³J_H−H7 Hz, Ar), 7.44 (2H, d, ³J_H−H8 Hz, para-substituted Ar),

7.24 (2H, d, ³J_H−H8 Hz, para-substituted Ar), 7.03 (2H, d, ³J_H−H8 Hz,

para-substituted Ar), 1.51 (NCH₂−H_eq), 0.78 (NCH₂−H_eq), 0.27

(NCH₂−H_eq), −1.02 (NCH₂−H_eq), −2.92 (CH₃), −3.73 (CH₃),

−4.05 (CH₃), −5.80 (NCH₂−H_ax), −6.06 (NCH₂−H_ax), −6.97

(NCH₂−H_ax), −7.66 (2H, NCH₂−H_axand NCH₂−H_eq), −9.90

(NCH₂−H_eq), −11.37 (NCH₂−H_eq), −12.70 (H_ac), −14.64 (H_ac),

−19.50 (H_ac), −20.17 (H_ac), −20.31 (H_ac). m/z (ESMS ESI−): 925

(100%, [EuL]⁻). The appropriate isotope pattern was observed.

Molecular Modeling. All modeling and docking procedures were

carried out using the MOE molecular modeling package (MOE,

version 2004.03, Chemical Computing Group Inc., Montreal,

Canada). The structures of the chelates of Gd4 were built from the

crystal structure of the DOTA-type chelates obtained from the

Cambridge Structural Database (entry code JOPJIH; www.ccdc.cam.

ac.uk/) and modeled using the Moe-Builder module keeping the Gd³⁺

coordination cage ﬁxed. Conformational analysis of the isomeric Gd4

ACKNOWLEDGMENTS

■

The authors thank the National Institutes of Health (EB-

11687) (MW), Oregon Nanoscience and Microtechnologies

Institute (N00014-11-1-0193) (MW), Regione Piemonte

(Italy) through the NANO-IGT and Prin 2009 Projects (MB

& DL), Portland State University and the Oregon Opportunity

for Biomedical Research for ﬁnancial support of this work, and

Drs. Silvio Aime and Enzo Terreno for insightful discussion.

REFERENCES

■

(1) Caravan, P.; Zhang, Z. Eur. J. Inorg. Chem. 2012, 2012, 1916.

(2) Bloembergen, N. J. Chem. Phys. 1957, 27, 572.

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

(4) Bloembergen, N.; Purcell, E. M.; Pound, R. V. Phys. Rev. 1948,

73, 679.

M

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

(5) Solomon, I. Phys. Rev. 1955, 99, 559.

(41) Rodriguez-Rodriguez, A.; Esteban-Gomez, D.; de Blas, A.;

Rodriguez-Blas, T.; Fekete, M.; Botta, M.; Tripier, R.; Platas-Iglesias,

C. Inorg. Chem. 2012, 51, 2509.

(42) Woods, M.; Kovacs, Z.; Zhang, S.; Sherry, A. D. Angew. Chem.

Int’l. Ed. 2003, 42, 5889.

(43) Aime, S.; Barge, A.; Bruce, J. I.; Botta, M.; Howard, J. A. K.;

Moloney, J. M.; Parker, D.; de Sousa, A. S.; Woods, M. J. Am. Chem.

Soc. 1999, 121, 5762.

(6) Solomon, I.; Bloembergen, N. J. Chem. Phys. 1956, 25, 261.

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

Rev. 1999, 99, 2293.

(8) Aime, S.; Botta, M.; Fasano, M.; Terreno, E. Chem. Soc. Rev.

1998, 27, 19.

(9) Botta, M.; Tei, L. Eur. J. Inorg. Chem. 2012, 2012, 1945.

(10) Lauffer, R. B. Chem. Rev. 1987, 87, 901.

(11) Bousquet, J. C.; Saini, S.; Stark, D. D.; Hahn, P. F.; Nigam, M.;

Wittenberg, J.; Ferrucci, J. T. Radiology 1988, 166, 693.

(12) Sherry, A. D.; Caravan, P.; Lenkinski Robert, E. J. Magn. Reson.

Imag. 2009, 30, 1240.

(13) Kumar, K.; Jin, T.; Wang, X.; Desreux, J. F.; Tweedle, M. F.

Inorg. Chem. 1994, 33, 3823.

(14) Aime, S.; Calabi, L.; Cavallotti, C.; Gianolio, E.; Giovenzana, G.

B.; Losi, P.; Maiocchi, A.; Palmisano, G.; Sisti, M. Inorg. Chem. 2004,

43, 7588.

(15) Werner, E. J.; Datta, A.; Jocher, C. J.; Raymond, K. N. Angew.

Chem., Int. Ed. 2008, 47, 8568.

(16) Baranyai, Z.; Botta, M.; Fekete, M.; Giovenzana, G. B.; Negri,

R.; Tei, L.; Platas-Iglesias, C. Chem.Eur. J. 2012, 18, 7680.

(17) Cowper, S. E.; Robin, H. S.; Steinberg, S. M.; Su, L. D.; Gupta,

S.; LeBoit, P. E. Lancet 2000, 356, 1000.

(18) Grobner, T. Nephrol. Dial. Transplant. 2006, 21, 1104.

(19) White Gregory, W.; Gibby Wendell, A.; Tweedle Michael, F.

Invest. Radiol. 2006, 41, 272.

(20) Sarka, L.; Burai, L.; Brucher, E. Chem.Eur. J. 2000, 6, 719.

(21) Aime, S.; Caravan, P. J Magn Reson Imaging 2009, 30, 1259.

(22) Baranyai, Z.; Palinkas, Z.; Uggeri, F.; Maiocchi, A.; Aime, S.;

(44) Aime, S.; Botta, M.; Fasano, M.; Marques, M. P. M.; Geraldes,

C. F. G. C.; Pubanz, D.; Merbach, A. E. Inorg. Chem. 1997, 36, 2059.

(45) Howard, J. A. K.; Kenwright, A. M.; Moloney, J. M.; Parker, D.;

Woods, M.; Port, M.; Navet, M.; Rousseau, O. Chem. Commun. 1998,

1381.

(46) Woods, M.; Kovacs, Z.; Kiraly, R.; Brucher, E.; Zhang, S.;

̈

Sherry, A. D. Inorg. Chem. 2004, 43, 2845.

(47) Webber, B. C.; Woods, M. Inorg. Chem. 2012, 51, 8576.

(48) Ruegg, C. L.; Anderson-Berg, W. T.; Brechbiel, M. W.;

Mirzadeh, S.; Gansow, O. A.; Strand, M. Cancer Res. 1990, 50, 4221.

(49) Tircso, G.; Webber, B. C.; Kucera, B. E.; Young, V. G.; Woods,

M. Inorg. Chem. 2011, 50, 7966.

(50) Gianolio, E.; Giovenzana, G. B.; Longo, D.; Longo, I.;

Menegotto, I.; Aime, S. Chem.Eur. J. 2007, 13, 5785.

(51) Schuehle, D. T.; Schatz, J.; Laurent, S.; Vander Elst, L.; Mueller,

R.; Stuart, M. C. A.; Peters, J. A. Chem.Eur. J. 2009, 15, 3290.

(52) Torres, S.; Martins, J. A.; Andre, J. P.; Geraldes, C. F. G. C.;

Merbach, A. E.; Toth, E. Chem.Eur. J. 2006, 12, 940.

(53) Nicolle, G. M.; Toth, E.; Eisenwiener, K.-P.; Macke, H. R.;

Merbach, A. E. J. Biol. Inorg. Chem. 2002, 7, 757.

(54) Botta, M.; Avedano, S.; Giovenzana, G. B.; Lombardi, A.; Longo,

D.; Cassino, C.; Tei, L.; Aime, S. Eur. J. Inorg. Chem. 2011, 802.

(55) Benetollo, F.; Bombieri, G.; Calabi, L.; Aime, S.; Botta, M. Inorg.

Chem. 2003, 42, 148.

Brucher, E. Chem.Eur. J. 2012.

̈

(23) Morcos, S. K.; Dawson, P. Nephrol. Dial. Transpl. Plus 2010, 3,

501.

(24) Rudovsky, J.; Botta, M.; Hermann, P.; Hardcastle, K. I.; Lukes,

̌

(56) Kotek, J.; Rudovsky, J.; Hermann, P.; Lukes, I. Inorg. Chem.

2006, 45, 3097.

(57) Lebduskova, P.; Hermann, P.; Helm, L.; Toth, E.; Kotek, J.;

Binnemans, K.; Rudovsky, J.; Lukes, I.; Merbach, A. E. Dalton Trans.

2007, 493.

I.; Aime, S. Bioconjugate Chem. 2006, 17, 975.

(25) Benmelouka, M.; Borel, A.; Moriggi, L.; Helm, L.; Merbach, A.

E. J. Phys. Chem. B 2007, 111, 832.

(26) Borel, A.; Bean, J. F.; Clarkson, R. B.; Helm, L.; Moriggi, L.;

Sherry, A. D.; Woods, M. Chem.Eur. J. 2008, 14, 2658.

(27) Senn, F.; Helm, L.; Borel, A.; Daul, C. A. C. R. Chim. 2012, 15,

250.

(28) Sherry, A. D.; Brown, R. D., III; Geraldes, C. F. G. C.; Koenig, S.

H.; Kuan, K. T.; Spiller, M. Inorg. Chem. 1989, 28, 620.

(29) Muller, R. N.; Raduchel, B.; Laurent, S.; Platzek, J.; Pierart, C.;

Mareski, P.; Vander Elst, L. Eur. J. Inorg. Chem. 1999, 1949.

(30) Aime, S.; Botta, M.; Parker, D.; Williams, J. A. G. J. Chem. Soc.,

Dalton Trans. 1996, 17.

(31) Beeby, A.; Clarkson, I. M.; Dickins, R. S.; Faulkner, S.; Parker,

D.; Royle, L.; de Sousa, A. S.; Williams, J. A. G.; Woods, M. J. Chem.

Soc., Perkin Trans. 2 1999, 493.

(32) Aime, S.; Botta, M.; Garda, Z.; Kucera, B. E.; Tircso, G.; Young,

V. G.; Woods, M. Inorg. Chem. 2011, 50, 7955.

(58) Horrocks, W. D., Jr.; Arkle, V. K.; Liotta, F. J.; Sudnick, D. R. J.

Am. Chem. Soc. 1983, 105, 3455.

(59) Batsanov, A. S.; Beeby, A.; Bruce, J. I.; Howard, J. A. K.;

Kenwright, A. M.; Parker, D. Chem. Commun. 1999, 1011.

(60) Banci, L.; Bertini, I.; Luchinat, C. Inorg. Chim. Acta 1985, 100,

173.

(61) Aime, S.; Botta, M.; Ermondi, G. J. Magn. Reson. 1991, 92, 572.

(62) Costa, J.; Toth, E.; Helm, L.; Merbach, A. E. Inorg. Chem. 2005,

44, 4747.

(63) Costa, J.; Balogh, E.; Turcryl, V.; Tripier, R.; Le Baccon, M.;

Chuburu, F.; Handel, H.; Helm, L.; Toth, E.; Merbach, A. E. Chem.

Eur. J. 2006, 12, 6841.

(64) Koenig, S. H.; Brown, R. D., III Prog. Nucl. Magn. Reson.

Spectrosc. 1990, 22, 487.

(65) Ferreira, M. F.; Mousavi, B.; Ferreira, P. M.; Martins, C. I. O.;

Helm, L.; Martins, J. A.; Geraldes, C. F. G. C. Dalton Trans. 2012, 41,

5472.

(33) Woods, M.; Aime, S.; Botta, M.; Howard, J. A. K.; Moloney, J.

M.; Navet, M.; Parker, D.; Port, M.; Rousseaux, O. J. Am. Chem. Soc.

2000, 122, 9781.

(34) Woods, M.; Botta, M.; Avedano, S.; Wang, J.; Sherry, A. D.

(66) Lipari, G.; Szabo, A. J. Am. Chem. Soc. 1982, 104, 4546.

(67) Lipari, G.; Szabo, A. J. Am. Chem. Soc. 1982, 104, 4559.

(68) Aime, S.; Botta, M.; Fedeli, F.; Gianolio, E.; Terreno, E.; Anelli,

P. Chem.Eur. J. 2001, 7, 5261.

Dalton Trans. 2005, 3829.

(35) Caravan, P.; Astashkin, A. V.; Raitsimring, A. M. Inorg. Chem.

2003, 42, 3972.

(36) Doble, D. M. J.; Botta, M.; Wang, J.; Aime, S.; Barge, A.;

Raymond, K. N. J. Am. Chem. Soc. 2001, 123, 10758.

(37) Ruloff, R.; Toth, E.; Scopelliti, R.; Tripier, R.; Handel, H.;

Merbach, A. E. Chem. Commun. 2002, 2630.

(69) Aime, S.; Botta, M.; Frullano, L.; Crich, S. G.; Giovenzana, G.

B.; Pagliarin, R.; Palmisano, G.; Sisti, M. Chem.Eur. J. 1999, 5, 1253.

(70) Gianolio, E.; Napolitano, R.; Fedeli, F.; Arena, F.; Aime, S.

Chem. Commun. 2009, 6044.

(38) Laus, S.; Ruloff, R.; Toth, E.; Merbach, A. E. Chem.Eur. J.

(71) Sudlow, G.; Birkett, D. J.; Wade, D. N. Mol. Pharmacol. 1975,

2003, 9, 3555.

11, 824.

(39) Rudovsky, J.; Cigler, P.; Kotek, J.; Hermann, P.; Vojtisek, P.;

(72) Sudlow, G.; Birkett, D. J.; Wade, D. N. Mol. Pharmacol. 1976,

12, 1052.

Lukes,

2005, 11, 2373.

̌

I.; Peters, J. A.; Vander Elst, L.; Muller, R. N. Chem.Eur. J.

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

(40) Polasek, M.; Hermann, P.; Peters, J. A.; Geraldes, C. F. G. C.;

̌

Lukes, I. Bioconjugate Chem. 2009, 20, 2142.

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

N

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry

Article

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

J. Am. Chem. Soc. 2002, 124, 3152.

(74) Yamasaki, K.; Maruyama, T.; Kragh-Hansen, U.; Otagiri, M.

Biochim. Biophys. Acta 1996, 1295, 147.

(75) Zech, S. G.; Eldredge, H. B.; Lowe, M. P.; Caravan, P. Inorg.

Chem. 2007, 46, 3576.

(76) Aime, S.; Gianolio, E.; Longo, D.; Pagliarin, R.; Lovazzano, C.;

Sisti, M. ChemBioChem 2005, 6, 818.

(77) Aime, S.; Botta, M.; Fasano, M.; Crich, S. G.; Terreno, E. J. Biol.

Inorg. Chem. 1996, 1, 312.

(78) Powell, D. H.; Ni Dhubhghaill, O. M.; Pubanz, D.; Helm, L.;

Lebedev, Y. S.; Schlaepfer, W.; Merbach, A. E. J. Am. Chem. Soc. 1996,

118, 9333.

(79) Toth, E.; Connac, F.; Helm, L.; Adzamli, K.; Merbach, A. E. J.

Biol. Inorg. Chem. 1998, 3, 606.

(80) Payne, K. M.; Aime, S.; Botta, M.; Valente, E.; Woods, M. Chem.

Commun. 2013, 2320.

(81) Hamilton, J. A.; Chen, L. J. Am. Chem. Soc. 1988, 110, 5833.

(82) Ponder, J. W.; Case, D. A. Adv. Protein Chem. 2003, 66, 27.

(83) Qiu, D.; Shenkin, P. S.; Hollinger, F. P.; Still, W. C. J. Phys.

Chem. A 1997, 101, 3005.

O

dx.doi.org/10.1021/ic400308a | Inorg. Chem. XXXX, XXX, XXX−XXX

Article Doi

Coupling fast water exchange to slow molecular tumbling in Gd3+ chelates: Why faster is not always better

DOI: 10.1021/ic400308a

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ic400308a

Full text of DOI:10.1021/ic400308a

Products guided by the article

R&D Labs maybe for 1447145-70-1

Relevant to this article

Hot Product