Prediction of q-Values and Conformations of Gadolinium Chelates for

For gadolinium chelates, we determined that there is a linear correlation between calculated solvent-accessible surface area and q-value, the number o...
0 downloads 0 Views 303KB Size
Download PDF
958

Bioconjugate Chem. 1999, 10, 958−964

Prediction of q-Values and Conformations of Gadolinium Chelates for Magnetic Resonance Imaging L. A. Castonguay,*,† A. M. Treasurywala,†,‡ T. J. Caulfield,§ E. P. Jaeger,†,| and K. E. Kellar⊥ Departments of Biophysics and Computational Chemistry, Medicinal, and Analytical Chemistry, Sterling Winthrop Pharmaceutical Research Division, 1250 S. Collegeville Road, P.O. Box 5000, Collegeville, Pennsylvania 19426-0900. Received March 8, 1999; Revised Manuscript Received August 11, 1999

For gadolinium chelates, we determined that there is a linear correlation between calculated solventaccessible surface area and q-value, the number of rapidly exchanging water molecules directly bound to the gadolinium ion. A calibration curve was developed to predict q-value based on the solventaccessible surface area of gadolinium. This predictive method was validated with the following gadolinium crystal structures: (ethylenediaminetetraacetic acid)-gadolinium(III) [Gd(EDTA)] [Templeton, L. K., Templeton, D. H., Zalkin, A., and Ruben, H. W. (1982) Anomalous Scattering by Praseodymium, Samarium, and Gadolinium and Structures of their Thylenediaminetetraacetate (EDTA) Salts. Acta Crystallogr., Sect. B 38, 2155], (1,4,7,10-tetraazacyclododecane-N,N′,N′′,N′′′tetraacetic acid)-gadolinium(III) [Gd(DOTA)] [Dubost, J.-P., Leger, J.-M., Langlois, M.-H., Meyer, D., and Schaefer, M. (1991) Structure of a Magnetic Resonance Imaging Agent - The Gadolinium-DOTA Complex C16H24N4O8NaGd, 5H2O. C. R. Acad. Sci., Ser. 2 312, 349], (diethylenetriaminepentaacetic acid)-gadolinium(III) [Gd(DTPA)] [Stezowski, J. J., and Hoard, J. L. (1984) Heavy Metal Ionophores - Correlations Among Structural Parameters of Complexed Nonpeptide Polyamino Acids. Isr. J. Chem. 24, 323], (diethylenepenta-acetato)-gadolinium(III) [Gd(DTPA-BEA)] [Smith, P. H., Brainard, J. R., Morris, D. E., Jarvinen, G. D., and Ryan, R. R. (1989) Solution and Solid-State Characterization of Europium and Gadolinium Schiff-Base Complexes and Assessment of their Potential as Contrast Agents in Magnetic Resonance Imaging. J. Am. Chem. Soc. 111, 7437], and (1,7,13-triaza-4,10,16trioxacyclo-octadecane-N,N′,N′′-triacetato)-gadolinium(III) [Gd(TTTA)] [Chen, D., Squattrito, P. J., Martell, A. E., and Clearfield, A. (1990) Synthesis and Crystal Structure of a 9-Coordinate Gadolinium(III) Complex of 1,7,13-Triaza-4,10,16-Trioxacyclooctadecane-N,N′,N′′-Tri-Acetic Acid. Inorg. Chem. 29, 4366]. Predicted q-values were in complete agreement with experimentally determined q-values. A genetic algorithm-based conformational search method was developed to generate valid 3D models for gadolinium chelates. The method was successfully tested on the following gadolinium chelates: Gd(EDTA) (Templeton et al., 1982), Gd(DOTA) (Dubost et al., 1991), Gd(DTPA-BEA) (Smith et al., 1989), Gd(TTTA) (Chen et al., 1990), Gd(triethylene glycol) [Rogers, R. D., Voss, E. J., and Etzenhouser, R. D. (1988) F-Element Crown Ether Complexes. 17. Synthetic and Structural Survey of Lanthanide Chloride Tiethylene Glycol Complexes. Inorg. Chem. 27, 533], and Gd(tetraethylene glycol) [Rogers, R. D., Etzenhouser, R. D., Murdoch, J. S., and Reyes, E. (1991) Macrocycle Complexation Chemistry. 35. Survey of the Complexation of the Open-Chain 15-Crown-5 Analogue tetraethylene Glycol with the Lanthanide Chlorides. Inorg. Chem. 30, 1445].

INTRODUCTION

The importance of intravenously administered contrast agents for magnetic resonance imaging (MRI) is evidenced by the fact that approximately one-half of all routine clinical MRI procedures involve the use of contrast agents (8). Particularly, small water-soluble * To whom correspondence should be addressed. Current address: Merck Research Laboratories, P.O. Box 2000, RY50SW100, Rahway, NJ 07065. E-mail: [email protected]. † Department of Biophysics and Computational Chemistry. ‡ Current address: Pfizer Central Research, Groton, CT 06340. § Department of Medicinal Chemistry. Current address: Rhone-Poulenc Rorer, 500 Arcola Rd., P.O. Box 1200 Collegeville, PA 19426. | Current address: 3D Pharmaceuticals, 665 Stockton Dr., Suite 104, Exton, PA 19341. ⊥ Department of Analytical Cheimisty. Current address: Nycomed Amersham Imaging, 466 Devon Park Dr., Wayne, PA 19087.

chelates of Gd(III) such as [Gd(DTPA)]2- and [Gd(DOTA)]have found widespread clinical application as MRI contrast agents due to their favorable characteristics of high chemical stability, low toxicity, and rapid clearance. The effectiveness of a contrast agent is determined by the ability of the complex to increase the relaxation rate of bulk water protons, or relaxivity in vivo. For small water soluble polyaminocarboxylate gadolinium chelates at field strengths pertinent to MRI, the magnitude of the relaxivity is determined mainly by four parameters: (1) tR, the rotational correlation time of the chelate; (2) r, the distance between the centers of the gadolinium ion and the protons of the coordinated water molecules; (3) q, the number of water molecules bound in the inner coordination sphere of the gadolinium ion; and (4) τM, the residence lifetime of water molecules in the inner coordination sphere of the gadolinium ion. For most gadolinium chelates reported in the literature (911), the values of tR, r, and τM affect the relaxivity to a minor extent as compared to the value of q. Indeed,

10.1021/bc990027b CCC: $18.00 © 1999 American Chemical Society Published on Web 10/09/1999

q-Values and Conformations of Gadolinium Chelates

previous work has established a linear relationship between integer q-value and relaxivity at 1.2 T, a fieldstrength typical of clinical MRI (10). Due to the toxicity of free lanthanide ions, the biological tolerance of contrast agents is largely determined by the ability of the gadolinium complex to remain intact in vivo. The acute toxicity of a number of polyaminocarboxylate gadolinium complexes has been demonstrated to be inversely proportional to the conditional binding constants of the ligands. Despite a 50-fold range of LD50 values for four Gd(III) complexes, all become lethally toxic to mice at dosages where 13-15 mM of Gd(III) is released in vivo (12). The authors propose that release of Gd(III) from the complex is responsible for the toxicity and that this appears to be a consequence of Zn(II), Cu(II), and Ca(II) transmetalation in vivo. In general, the stability of metal-ligand complexes increases with the number of potential coordination sites which are occupied. The goal in designing improved MRI contrast agents is to maximize both the biological tolerance (conditional stability) and efficacy (q-value) of the gadolinium chelate. Therefore, predictive methods for determining both conditional stability constants and q-value for gadolinium chelates would help facilitate the discovery of improved contrast agents. Previously, molecular mechanics calculations and dynamics simulations, based on the AMBER force field (13), were used to examine the structure and stabilities of gadolinium complexes (14). However, no predictive methods for the determination the q-values has been reported. Herein, we would like to report a computational method to predict the q-values and structures of gadolinium chelates for magnetic resonance imaging. COMPUTATIONAL DETAILS

q-Value Calculation and Validation. The Connolly surface is the area traced out by rolling a defined probe (a sphere of 1.4 Å which represents a water molecule) over the gadolinium ion. A calibration curve for translating calculated Connolly surface area to q-value was developed using the crystal structure of Gd(H2O)9 (15). If one water is removed, then there is one vacancy to which a water molecule can bind; in other words, eight of the nine coordination sites are already occupied. This is analogous to a q ) 1 system where there is space for one solvent water molecule to coordinate to the gadolinium ion, and the rest of the “solvent sites” are involved in complexation to the chelate. Therefore, for q ) 1, one water is removed from the structure and the Connolly surface area is calculated for the gadolinium using SYBYL software (16). There are nine possible ways to remove one water molecule, each of the nine possible Connolly surface areas are calculated, and the average, maximum, and minimum values are recorded. This same procedure was repeated for q ) 0 to q ) 9 and the resultant data plotted as a calibration curve, Figure 1. For q ) 2, 3, 4, 5, 6, 7, and 8, only contiguous waters were removed. This restriction was applied to prevent “partial sites” from being mistakenly counted. For example, if one and a half sites were available on the top of the molecule and half a site was available on the bottom, there would not actually be room for two solvent water molecules to bind since the sites are not contiguous. Since the structures generated during conformational analysis will be minimized structures, a new calibration curve was developed using the same procedure for the Gd(H2O)9 crystal structure after minimization. Gado-

Bioconjugate Chem., Vol. 10, No. 6, 1999 959

Figure 1. Graph of q-value vs calculated Connolly surface area for the crystal structure of Gd(H2O)9. The dots denote the average value of Connolly surface area for each q-value. The shaded bar denotes the data spread between highest and lowest calculated Connolly surface area for each q-value. Note no bands overlap.

linium nonahydrate was minimized in SYBYL using the parameters and conditions listed below. The data were plotted as before and the same linear calibration curve was observed. Gadolinium Parameter Development. The SYBYL force field was extended in this work to include an atom type for Gd(III). The parameters for the remaining atoms are taken directly from the published SYBYL force field (17). Since the gadolinium is not covalently bonded to the chelates, only nonbonded parameters needed to be developed. The van der Waals energy term used in the SYBYL force field is as follows: Natoms

Evdw )

∑∑

Eij (1/aij12 - 2/aij6)

i)1 j>1 nonbonded

where Eij is the van der Waals constant, aij equals rij/(Ri + Rj), rij is the distance between atoms i and j, Ri is the van der Waals radius of the ith atom, and Rj is the van der Waals radius of the jth atom. These parameters were developed empirically by varying Eij and Ri for a database of gadolinium complex crystal structures. The optimal parameters being the ones that fit the structural data obtained from the crystal structures. Crystal structures from the Cambridge Structural Database (CSD) (18) were used to construct this database. A search of the CSD resulted in 76 structures containing gadolinium and carbon. Structures were eliminated if they were duplicates, contained more than one gadolinium, were diagnostic imaging agents that would be used later to test these methods, contained uncomplexed gadolinium with the ligand, contained just gadolinium and solvent, or contained other metals as part of the ligand. A total of 59 structures were eliminated, leaving 17 structures in the database. The database contained the following structures: (18crown-6)-ethanol-dichloro-gadolinium(III) chloride (19),

960 Bioconjugate Chem., Vol. 10, No. 6, 1999

nona-aqua-gadolinium(III) tris(trifluoromethanesulfonate) (14), tris(acetylacetonato)-diaqua-gadolinium(III) monohydrate (20), tris(acetylacetonato-O,O′)-bis(acetylacetoniminato)-gadolinium(III) acetylacetonimine solvate (21), chloro-diaqua-(18-crown-6)-gadolinium(III) dichloride dihydrate (22), bis(η-1,3,5-tri-tert-butylbenzene)-gadolinium(0) (23), penta-aqua-(triethylene glycol-O,O′,O′′,O′′′′)gadolinium(III) trichloride (6), cis-trichloro-tris(2, 6-dimethyl-4-pyrone)-gadolinium(III) (24), aqua-tris(nitratoO,O′)-bis[di-isopropyl(1, 2-bis(diethylcarbamoyl)ethyl) phosphonato-O]gadolinium(III) (25), tris[(3-aza-4-methylhept-4-en-6-on-1-yl)amine-O,O′,O′′]-tris(nitrato-O,O′)gadolinium(III) (26), triaqua-chloro-(tetraethylene glycol)gadolinium(III) dichloride monohydrate (7), [10-(2-((2hydroxy-1-(hydroxymethyl)ethyl) amino]-1-[(phenylmethoxy)methyl]-2-oxo-ethyl)-1,4,7,10-tetra-azacyclododecane-1,4,7-triacetic acid)-gadolinium(III) trihydrate (27), pentasodium tris(oxy-diacetato-O,O′,O′′)gadolinium(III) diperchlorate hexahydrate (28), trinitrato-1, 2-bis(pyridine-2-aldimino)ethane-gadolinium(III) (29), acidic gadolinium tetrakis(benzoylacetonate) piperidine (30), bis[diaqua-(2-formyl-4-methyl-6-(4-(2-pyridyl)-2-azoniabut-1-enyl)phenolato-O,O′)-gadolinium(III)] diperchlorate chloride hydrate (31), and trichloro-bis(dimethoxyethaneO,O′)-gadolinium(III) (32). All structures in the database were subjected to the following protocol after extraction from the CSD. The crystal structures were imported into SYBYL, and counterions and water molecules were removed. Atoms were atom typed, and bonding was fixed if necessary. Hydrogens were added and minimized, holding the rest of the molecule fixed (default conditions: simplex method (33), 20 steps for any atom with a force above 1000 kcal/mol Å2 followed by the Powell method of conjugate gradient minimization (34) to a convergence 0.05 kcal/molÅ2 RMS (root-mean-square gradient). Gasteiger charges were calculated setting the charge on gadolinium to +3.0. The Ri of Gd was varied from 1.8 to 2.3 Å in increments of 0.1 Å, the database was minimized (default conditions) using the default Eij (van der Waals constant or well depth) value of 0.1. The resulting minimized structures were saved in a new database, and the total RMS deviations (no hydrogen atoms) from original crystal structures were used as a measure of “goodness” of parameter fit. The following four structures were eliminated since the RMS deviation (no hydrogen atoms) was very poor (>1) for all values of Ri: cis-trichloro-tris(2,6dimethyl-4-pyrone)-gadolinium(III) (24), bis(diethylcarbamoyl)ethyl-phosphonato-O-gadolinium(III) (25), acidic gadolinium tetrakis(benzoylacetonate)piperidine (30), and bis[diaqua-(2-formyl-4-methyl-6-(4-(2-pyridyl)-2-azoniabut-1-enyl)phenolato-O,O′)-gadolinium(III)] diperchlorate chloride hydrate (31). A narrower range of suitable Ri was obtained by graphing the sum of the total RMS deviations (no hydrogen atoms) for each run vs Ri. An approximately parabolic curve was obtained with the best radius range in the well being 1.9-2.1. Using a similar procedure as before, both Ri (1.9-2.1 with increments of 0.05) and Eij (0.001, 0.01, 0.05, 0.1, 0.2, 1.0) were varied. The best parameters found were Ri ) 2.0 and Eij ) 0.2. Genetic Algorithm Conformational Search Method. A flow chart representation of this method is shown in Scheme 1. Initially in SYBYL, the gadolinium chelate is built, then atom typed and charged using Gasteiger charges, setting the charge on Gd to +3.0. To go from a 2D drawing to a reasonable starting gadolinium chelate complex, initial constraints are applied. Distance constraints of 2.6 Å are applied to all gadolinium-

Castonguay et al. Scheme 1. Flowchart of Gd-Chelate Conformation Generation and q-Value Prediction

nitrogen pairs, and the structure is minimized [Powell method of conjugate gradient minimization (34) until a convergence of 0.01 energy gradient tolerance is reached]. Then additional distance constraints of 2.4 Å are added for the gadolinium-oxygen pairs (one constraint for each carboxylate group, arbitrarily picking just one of the oxygens in the COO- group), and the chelate is minimized. Then all constraints are removed, and the structure is minimized. The structure is output in SYBYL’s mol2 format. This conformation is used as the starting structure for the genetic algorithm conformational searching technique described below. A genetic algorithm (GA) search method was used to identify favorable binding conformation for the gadolinium ligand system. This search was performed using a hybrid system known as GASY which linked the genetic algorithm package GENESIS (35) with the molecularmodeling package SYBYL, hence the name GASY (GA linked through the SYBYL software). The GENESIS package was configured to generate lists of floating point numbers whose values had a possible range of -180 to 180. These lists were passed to SYBYL via named pipes. SYBYL was configured using SYBYL programming language (SPL) procedures to retrieve the sets of floating point numbers from the incoming pipe and interpret them as strings of torsional values. GASY then used the values to set a conformation of the molecular system, minimize the conformation [simplex method (33), 20 steps for any atom with a force above 1000 kcal/molÅ2 followed by the Powell method of conjugate gradient minimization (34), 200 steps or a minimum energy change of 0.05kcal/ mol], and measure the energy of this binding conformation. The energy value was then passed to the GENESIS program via a second pipe where it was treated as a measure of the fitness of the set of floating point numbers. This completed the evaluation loop and allowed

q-Values and Conformations of Gadolinium Chelates

Bioconjugate Chem., Vol. 10, No. 6, 1999 961

Table 1. Gd Chelates Used to Test the Predictive Capability of This Methoda

a Connolly surface area was calculated for each of these known crystal structures, and the calibration curve shown in Figure 1 was used to predict the q-value. The predicted and experimental q-values are in agreement.

the GA to evolve better conformations. For cyclic chelates, a bond must be broken before the GA run and then reestablished before the SYBYL minimization and energy evaluation. The messaging scheme used by GASY was necessary because both programs are designed to be in control of the sequence of operations and there is no easy way to make either program a subprogram of the other. GASY is a general modeling optimization tool. It allows the easy optimization of any number of problems where a string of numbers can be encoded into a molecular system and a fitness evaluation of the resulting system can be performed. Analysis of Genetic Algorithm Searches. The analysis of the conformations that were generated by the method outlined above was performed within SYBYL using a macro that had been created in SYBYL Programming Language (SPL) (see Scheme 1.). This macro reads a file that was generated by the Genetic Algorithm that contains an ordered list of the values of the torsions of each of the rotatable bonds. All energetically acceptable conformations were defined as those that were less than or equal to the minimum energy found over the entire run, plus 4 kcals. The program creates a table in SYBYL whose rows represent the conformations and whose columns represent a rotatable bond. Atom numbers of the four atoms which define each of the torsions from the GA initialization file are determined. Lastly, the program generates each of the conformations in the 4 kcal set and reports their energies as calculated in the SYBYL package. The resulting conformations are minimized with another SPL and deposited into a SYBYL database along with their corresponding calculated energies. All lowenergy structures (within 3 kcal of lowest) are considered valid models, and their q-values are calculated using the above-described method. The Connolly surface of the

gadolinium for each model is calculated in SYBYL, and the calibration curve is used to convert Connolly surface area to corresponding q-value. RESULTS AND DISCUSSION

Our hypothesis was that q-value would be proportional to the solvent-accessible surface area of the gadolinium ion. Typically, calculated Connolly surface area has been used to estimate the solvent accessible surface area of proteins. Using the crystal structure of gadolinium nonahydrate (15), it was shown that a linear correlation exists between calculated Connolly surface area of the gadolinium ion and q-value (see Figure 1). This calibration curve was developed by iteratively removing waters from the crystal structure of gadolinium nonahydrate and calculating Connolly surface area. To test the predictive capability of this method, this calibration curve was used to determine the q-values of the following gadolinium crystal structures: Gd(EDTA) (1), Gd(DOTA) (2), Gd(DTPA) (3), Gd(DTPA-BEA) (4), and Gd(TTTA) (5) (see Table 1). The predicted q-values were in total agreement with the experimentally determined values (36). To develop this as a tool for novel agent design, where no prior structural information exists, we needed to develop a method to generate valid 3D structures for the gadolinium chelates. Since any computationally generated model would be a minimized structure, the force field used had to be augmented to include gadolinium. Nonbonded force field parameters for Gd(III) were developed empirically within SYBYL. A new calibration curve was developed for minimized structures using the minimized gadolinium nonahydrate crystal structure (vide infra). The predictive capability was again tested using the minimized crystal structures listed above. Again, the predicted q-values were in complete agreement with the experimentally determined values (36).

962 Bioconjugate Chem., Vol. 10, No. 6, 1999

Castonguay et al.

Figure 2. Superposition of minimized crystal structures (green) and the GA generated conformations (magenta) for (A) Gd(DOTA), second unique GA generated conformation shown in blue; (B) Gd(EDTA); (C) Gd(tetraethylene glycol), two crystal structures are reported for this molecule and the second minimized crystal strucure is shown in yellow; (D) Gd(triethylene glycol), the second unique GA generated conformation is shown in blue; (E) Gd(TTTA).

A genetic algorithm based conformational search method (vide infra) was used to generate 3D models for the following chelates: Gd(EDTA) (1), Gd(DOTA) (2), Gd(DTPA-BEA) (4), Gd(TTTA) (5), Gd(tetraethylene glycol) (7), and Gd(triethylene glycol) (6). For Gd(EDTA) and Gd(tetraethylene glycol), one unique low-energy structure was generated (see Figure 2, panels B and C). Two unique equivalent energy structures were generated for Gd(DOTA) and for Gd(triethylene glycol). For Gd(DOTA), one structure is a diastereomer of the minimized crystal structure differing only in the rotation about the N-CH2 bond and the other is the enantiomer of the crystal structure (see Figure 2A). For Gd(triethylene glycol), one structure is the same as the minimized crystal structure and the other is a slightly lower energy structure which is similar in conformation to the minimized crystal structure (see Figure 2D). Two families of structures are generated for Gd(TTTA); one is the same as the minimized crystal structure and the other is the enantiomer of the minimized crystal structure (see Figure 2E). All the GA-generated structures are similar to the minimized crystal structures and have the same q-value and number of metal coordinating ligands as the minimized crystal structures. For Gd(DTPA-BEA), two families of structures were generated. Within the families, the structures

are very similar and only differ by end methyl rotations. Both structures are similar in energy to the minimized crystal structure, but are inverted at one nitrogen (see Figure 3). These two models still yield the correctly predicted q-value and number of metal coordinating ligands and, therefore, are reasonable models to use for design purposes. Water molecules that exchange rapidly between the inner coordination sphere of the gadolinium ion and the bulk solvent dominate the relaxivity. Despite the fact that our calculations do not take water exchange rates (1/τM) into account, the exchange rates for all gadolinium chelates reported thus far are sufficiently rapid to contribute to the relaxivity. In a recent study, the exchange rates [298 K, units of inverse seconds (s-1)] ranged from 8.3 × 108 to 4.3 × 105 for a series of gadolinium compounds (11). Although this range of exchange rates spans 4 orders of magnitude, the exchange rates are sufficiently rapid to make the value of the relaxivity insensitive to the precise value of the exchange rate for small gadolinium complexes (37). Consequently for small gadolinium complexes, the water molecules in the inner coordination sphere may be considered in rapid exchange with the bulk.

q-Values and Conformations of Gadolinium Chelates

Figure 3. Superposition of minimized crystal structure (green) and the two GA generated conformations (magenta and blue) for Gd(DTPA-BEA). SUMMARY AND CONCLUSIONS

A quick computational method for predicting integer q-value of small molecule gadolinium chelates has been developed. Using the crystal structure of gadolinium nonahydrate, a calibration curve to predict q-value from solvent accessible surface area of the gadolinium ion was developed. The calibration curve was tested using the crystal structures of gadolinium chelates obtained from the CSD, thus establishing a linear relationship between the calculated Connolly surface area of the gadolinium ion and experimental q-value. Furthermore, a genetic algorithm based protocol which generates valid threedimensional structures of known gadolinium chelates has been developed. The combination of the genetic algorithm conformational search method and q-value calibration curve produce literature consistent q-values and structures for selected MRI contrast agents. Use of the described method will help facilitate the discovery of novel MRI contrast agents. LITERATURE CITED (1) Templeton, L. K., Templeton, D. H., Zalkin, A., and Ruben, H. W. (1982) Anomalous Scattering by Praseodymium, Samarium, and Gadolinium and Structures of their Thylenediaminetetraacetate (EDTA) Salts. Acta Crystallogr., Sect. B 38, 2155. (2) Dubost, J.-P., Leger, J.-M., Langlois, M.-H., Meyer, D., and Schaefer, M. (1991) Structure of a Magnetic Resonance Imaging Agent - The Gadolinium-DOTA Complex C16H24N4O8NaGd, 5H2O. CR Acad. Sci., Ser. 2 312, 349. (3) Stezowski, J. J., and Hoard, J. L. (1984) Heavy Metal Ionophores - Correlations Among Structural Parameters of Complexed Nonpeptide Polyamino Acids. Isr. J. Chem. 24, 323. (4) Smith, P. H., Brainard, J. R., Morris, D. E., Jarvinen, G. D., and Ryan, R. R. (1989) Solution and Solid-State Characterization of Europium and Gadolinium Schiff-Base Complexes and Assessment of their Potential as Contrast Agents in Magnetic Resonance Imaging. J. Am. Chem. Soc. 111, 7437. (5) Chen, D., Squattrito, P. J., Martell, A. E., and Clearfield, A. (1990) Synthesis and Crystal Structure of a 9-Coordinate Gadolinium(III) Complex of 1,7,13-Triaza-4,10,16-Trioxacyclooctadecane-N,N′,N”-Tri-Acetic Acid. Inorg. Chem. 29, 4366. (6) Rogers, R. D., Voss, E. J., and Etzenhouser, R. D. (1988) F-Element Crown Ether Complexes. 17. Synthetic and Struc-

Bioconjugate Chem., Vol. 10, No. 6, 1999 963 tural Survey of Lanthanide Chloride Tiethylene Glycol Complexes. Inorg. Chem. 27, 533. (7) Rogers, R. D., Etzenhouser, R. D., Murdoch, J. S., and Reyes, E. (1991) Macrocycle Complexation Chemistry. 35. Survey of the Complexation of the Open-Chain 15-Crown-5 Analogue tetraethylene Glycol with the Lanthanide Chlorides. Inorg. Chem. 30, 1445. (8) Koenig, S. H., and Brown, R. D. (1990) Field-Cycling Relaxometry of Protein Solutions and Tissue-Implications for MRI. Progr. NMR Spectrosc. 22, 487. (9) Geraldes, C. F. G. C., Sherry A. D., Cacheris, W. P., Kuan, K.-T., Brown, R. D., Koenig, S. H., and Spiller, M. (1988) Number of Inner-Sphere Water Molecules in Gd3+ and Eu3+ Complexes of DTPA-Amide and DTPA-Ester Conjugates. Magn. Reson. Med. 8, 191. (10) Geraldes, C. F. G. C., Brown, R. D., Brucher, E., Koenig, S. H., Sherry A. D., and Spiller, M. (1992) Nuclear Magnetic Relaxation Dispersion Profiles of Aqueous Solutions of a Series of Gd(NOTA) Analogues. Magn. Reson. Med. 27, 284. (11) Gonzalez, G., Powell, D. H., Tissieres, V., and Merbach, A. E. (1994) Water Exchange, Electronic Relaxation, and Rotational Dynamics of the MRI Contrast Agent [Gd(DTPABMA)(H2O)} in Aqueous Solution - A Variable Pressure, Temperature, and Magnetic Field O17 NMR Study. J. Phys. Chem. 98, 53. (12) Cacheris, W. P., Quay, S. C., and Rocklage, S. M. (1990) The Relationship between Thermodynamics and the Toxicity of Gadolinium Complexes. Magn. Reson. Imaging 8, 467. (13) Weiner, S. J., Kollman, P. A., Nguyen, D. T., and Case, D. A. (1986) An All Atom Force Field for Simulations of Proteins and Nucleic Acids. J. Comput. Chem. 7, 230. (14) a) Fossheim, R., Dugstadt, H., and Dahl, S. G. (1991) Structure Stability Relationships of Gd(III) Ion Complexes for Magnetic Resonance Imaging. J. Med. Chem. 34, 819. (b) Fossheim, R., and Dahl, S. G. (1990) Molecular Structure and Dynamics of Aminopolycarboxylates and their Lanthanide Ion Complexes. Acta Chem. Scand. 44, 698. (15) Chatterjee, A., Maslen, E. N., and Watson, K. J. The Effect of the Lanthanoid (1988) Contraction on the Nonaaqualanthanoid(III) Tris(Trifluoromethanesulfonates). Acta Crystallogr., Sect. B 44, 381. (16) SYBYL, Version 6.0, TRIPOS Assoc. Inc., St. Louis, Mo. 63144. (17) Clark, M., Cramer, R. D., III, and Van Opdenbosch, N. (1989) Validation of the General Purpose Tripos 5.2 ForceField. J. Comput. Chem. 10, 982. (18) Cambridge Crystallographic Data Center, 12 Union Road, Cambridge CB2 1EZ, U.K. (19) Forsellini, E., Benetollo, F., Bombieri, G., Cassol, A., and De Paoli, G. (1985) Preparation, Crystal and Molecular Structure of GdCl3(18-Crown-6)EtOH. Inorg. Chim. Acta 109, 167. (20) Cheng, S., Yuguo, F., Guofa, L., Yutian, W., and Pinzhe, L. (1983) Gaodeng Xuexiao Huaxue Xuebao (Chem. J. Chin. Univ.) 4, 769. (21) Batsanov, A. S., Struchkov, Y. T., Trembovetskii, G. V., Martynenko, L. I., and Muraveva, I. A. (1986) Crystal Structure of Tris(Acetylacetonato)Bis(O-Acetylacetoneimino)Gadolinium(III) (1-2) Solvate with Acetylacetoneimine. Zh. Neorg. Khim. 31, 345. (22) Rogers, R. D., and Kurihara, L. K. (1987) F-Element Crown Ether Complexes. 4. Synthesis and Crystal and Molecular Structures of [SmCl(OH2)2(18-Crown-6)Cl2.2H2O, (GdCl(OH2)2(18-Crown-6))Cl2.2H20, (TbCl(OH2)2(18-Crown-6))Cl2.2H2O. Inorg. Chem. 26, 1498. (23) Brennan, J. G., Cloke, F. G. N., Sameh, A. A., and Zalkin, A. (1987) Synthesis of Bis(Eta-1,3,5-Tri-tert-Butylbenzene) Sandwich Complexes of Yttrium(0) and Gadolinium(0) - the X-ray Crystal Structure of the 1st Aunthentic Lanthanide(0) Complex, [Gd(ETA-But3C6H3)2]. J. Chem. Soc., Chem. Commun. 1668. (24) Castellani, C. B., and Tazzoli, V. (1984) Cis-Trichlorotris(2,6-Dimethyl-4-Pyrone)Gadolinoum(III), [GdCl3(C7H8O2)3] Acta Crystallogr., Sect. C 40, 183. (25) McCabe, D. J., Duesler, E. N., and Paine, R. T. (1988) Synthesis and Coordination Chemistry of Tripodal Dialkyl-

964 Bioconjugate Chem., Vol. 10, No. 6, 1999 [1,2-Bis(Diethylcarbamoyl)Ethyl]Phosphonates with Lanthanide Nitrates. Inorg. Chim. Acta 147, 265. (26) Smith, A., Rettig, S. J., and Orvig, C. (1988) Lanthanide Complexes of Potentially Heptadentate Ligands Including the Structure of [Tris(3-Aza-4-Methylhept-4-Ene-6-On-1-Yl)Amine]Tris(Nitrato)Gadolinium(III). Inorg. Chem. 27, 3929. (27) Aime, S., Anelli, P. L., Botta, M., Fedeli, F., Grandi, M., Paoli, P., and Uggeri, F. (1992) Synthesis, Characterization, and 1/T(1) NMRD Profiles of Gadolinium(III) Complexes of Monoamide Derivatives of DOTA-like Ligands - X-ray Structure of the 10-[-[[2-Hydroxy-1-(Hydroxymethyl)Ethyl]Amino]1-[(Phenylmethoxy) Methyl]-2-Oxo-Ethyl]-1,4,7-Triacetic Acid Gadolinium(III) Complex. Inorg. Chem. 31, 2422. (28) Albertson, J. (1968) Structural Studies on the Rare Earth Carboxylates. I. Crystal and Molecular Structure of Na3(M(O2CCH2OCH2CO2)3).2NaClO4.6H20; M) Neodynium, Gadolinium, and Ytterbium. Acta Chem. Scand. 22, 1563. (29) Smith, G. D., Caughlan, C. N., Haque, M., and Hart, F. A. (1973) Crystal and Molecular Structure of Trinitrato-1,2-Bis(Pyridine-2-Aldimino)Ethanegadolinium(III). Inorg. Chem. 12, 2654. (30) Butman, L. A., Aslanov, L. A., and Porai-Koshits, M. A. (1970) Crystal and Molecular Structure of Acidic Gadolinium Tetrakis(benzoylacetonate). Zh. Strukt. Khim. 11, 46.

Castonguay et al. (31) Bakalbassis, E., Kahn, O., Sainton, J., Trombe, J. C., and Galy, J. (1991) Synthesis of Dissymmetric Schiff-Base Ligands Through Gadolinium(III) Coordination. J. Chem. Soc., Chem. Commun. 1991, 755. (32) Gecheng, W., Hanrong, G., Zhongsheng, J., and Qi, S. (1989) Jiegou Huaxue (J. Struct. Chem.) 8, 61. (33) Press, W. H., Teukolsky, S. A., Vetterling, W. T., and Flannery, B. P. (1995) Numerical Recipes in C, 2nd ed., Cambridge University Press. (34) Powell, M. J. D. (1981) Approximation Theory and Methods, Cambridge University Press. (35) Grefenstette, J. J., and Schraudolph, N. N. GENESIS 1.2ucsd, available via anonymous ftp from ftp.aic.nrl.navy.mil in /pub/galist/source-code/gasource/gaucsd12.tar. (36) Lauffer, R. B. (1987) Paramagnetic Metal Complexes as Water Proton Relaxation Agents for NMR Imaging: Theory and Design. Chem. Rev. 87, 901. (37) Koenig, S. H. (1994) The Need for Electron Paramagnetic Resonance and Water Exchange-Rate Data for Understanding Small Magnetic Imaging Contrast Agents and their Macromolecular Complexes. Invest. Radiol. 29 (Suppl. 2), S127.

BC990027B