The synthesis scheme is shown in Scheme 2. The initial common synthetic steps [steps 11→16, Scheme 2] were carried out in bulk using standard glassware while the remaining diverse regions [steps 17→18, Scheme 2] were carried out using a 96-well synthesis filter plate [Scheme 2 (insert)]. First, a spacer between the bead and the DOTA scaffold was introduced. Methionine was added as the first residue of the spacer arm so the final product could be cleaved from the resin with cyanogen bromide for subsequent mass spectroscopic confirmation. To obtain reasonably symmetrical DOTA-tetraamide products and to limit possible steric hindrance between the EuDOTA complex and the bead surface, two neutral methoxyethylamine moieties were added next. Here, both the bromoacetic acid addition and amine coupling reactions were assisted by microwave irradiation.^17,19 Then, β-alanine was added to mimic the ethylendiamine spacer planned for the other three DOTA arms and finally DO3A tert-butylester was coupled. The three remaining carboxyl groups of DOTA were then reacted with mono-protected 1,2-diaminoethane [step 15→16, Scheme 2] to yield 16, the starting material for building the peptoid library.

A critical test was carried out at this point to verify that resin-bound DOTA-tetraamide forms a complex with Eu(III) and to test whether a CEST signal could be detected with the EuDOTA-tetraamide complex bound to the beads. Eu(III) complexation was performed by addition of ~15-fold excess EuCl₃ at pH=6.3 overnight, followed by removal of excess Eu(III) by washing the resin several times with water. A few beads were then removed and placed in a glass capillary for collection of a CEST spectrum using a vertical bore 400 MHz NMR spectrometer. Importantly, a large Eu(III)-bound water-exchange peak was observed in the CEST spectrum near 50 ppm (Fig. 2, squares). No CEST was observed for ligand 16 alone [no Eu(III)] or empty beads treated with EuCl₃, confirming that the CEST signal reflects Eu*16 attached to the beads (Fig. 2). The CEST signal of a solution of Eu*16 after it was cleaved from beads (see supplementary Fig. S2 online) displayed a similar chemical shift as the bead-bound complex but the exchange peak was slightly broader (Fig. 2, crosses). This implies that water exchange may be somewhat slower for the complex attached to beads, although absolute comparisons of exchange rates were not possible because the concentrations of the two samples were unknown. Most importantly, these data show that the CEST signal from a covalently-attached complex (Eu*16) is not strongly influenced by the presence of the Tentagel beads. This observation allowed us to proceed with the synthesis of a diverse chemical library of CEST ligands, including hydrophobic side-arms that might not ordinarily be soluble when prepared in bulk.

The parallel combinatorial synthesis approach was used to couple two peptoid residues onto 16. Bromoacetic acid was coupled (of the first peptoid residue) to 16 in bulk before the beads were distributed equally into a 96-well synthesis plate [Scheme 2 (insert)]. To avoid possible inhomogeneous microwave irradiation in the different wells of the plate, the remaining peptoid coupling steps were performed for an extended period of time at room temperature.¹⁷ The first eight rows and ten columns were used for synthesis of the library; column 11 was used for controls and column 12 was used for synthesis of additional eight random combinations of amines (see supplementary Table S3 online) for cleavage and mass spectroscopy analyses.

Eight different amines [1-8, Chart 1] were coupled to remaining three arms of DOTA to complete the first residue in each row [i.e., R₁ in row 1 was methoxyethylamine (1), R₁ in row 2 was piperonylamine (2), R₁ in row 3 was 4-methoxybenzylamine (3), etc]. This was followed by a second bromoacetic acid coupling step to prepare for the second amine. Amines 1-10 (Chart 1) were chosen as residue R₂; 1 was added to each of eight wells in column 1, 2 was added to each of eight wells in column 2, etc. to yield an array of compounds having the same first peptoid along each row (R₁ = 1-8) and the same second residue along each column (R₂ = 1-10). This simple approach allowed rapid access to 8 × 10 = 80 different compounds, all built on the same parental DOTA-tetraamide scaffold and differing only in the chemical identity of R₁ and R₂. Finally, Eu(III) complexation was performed as described previously. Before proceeding with the MRI measurements, the quality of the library was checked by analyzing the representative compounds synthesized in the 12^th column using MALDI-TOF MS to confirm the complete synthesis of all 8 compounds (see supplementary Table S3 online). Random samples were not withdrawn from the actual 80 compound library for this purpose in order to have same number of beads in each well for imaging.

To register a single image for all 80 compounds plus controls, the beads were first transferred from the synthesis/filtration plate to the central 80 wells of a 145 well-plate to maximize the height of the packed beads in 4 mm diameter well. All 145 wells were filled with water, including those on the perimeter that did not contain beads, to make the sample as homogeneous as possible. The plate was then positioned in the center of a 63 mm quadrature volume coil and CEST imaging was performed by applying a 10 μT frequency-selective presaturation pulse for 5 s followed by a spin-echo sequence using a 9.4 T Varian scanner. A 2 mm-thick coronal slice was selected near the bottom of each well to insure that each image would reflect an equal number of beads (the height of packed beads in each well was > 3 mm). The resulting CEST image (Fig. 3) is defined as the difference in water intensities in an image collected with the presaturation pulse set to -50 ppm (control) minus a second image collected with the presaturation pulse set to 50 ppm. This experiment was repeated on another day and the same image was obtained thereby confirming the reproducibility of this library imaging experiment.

After the plate was imaged, 10 beads were collected from each of eleven wells selected from those that showed large variations in CEST signal intensities that includes empty beads and beads from the well containing Eu*16. These beads were treated with HCl to displace all Eu(III) from each chelate and analyzed for total Eu(III) using a commercially available fluorescence enhancement assay.^20-21 These data (Fig. 4) indicated that there was an identical amount of total Eu(III) in each 10-bead sample and that empty beads treated with Eu(III) did not retain a significant amount of ion. Thus, any intensity differences in the CEST images shown in Fig. 3 could be attributed solely to consequences of the chemical differences of the side-arms in the 80 different EuDOTA-tetraamide complexes and not differences in agent concentrations.

The resulting CEST image (Fig. 3) illustrates the power of the technique; the CEST signal in some wells was quite intense (up to a 30% reduction in water intensity) while little to no CEST was detected in other wells. To check whether this may be due to frequency differences in the exchanging Eu(III)-bound water resonance in each complex, complete CEST spectra were collected on several individual wells (Fig. 5) and to show that the spectra were consistent with the imaging data. For those wells showing a CEST signal, the Eu(III)-bound water exchange peak appeared between 50-51 ppm (Fig. 5) in all cases. The quantitative % CEST signal (1-(M₍₊₅₀₎/M_(-50)*100) from each well varied from zero to 25% (Table 1, Supplementary Information) and was consistent with the qualitative image intensities shown in Fig. 3. The most intriguing part of the study came from a comparison of the combined effects of various physico-chemical properties of R₁ and R₂ on water exchange at each Eu(III) center.

The most intense CEST signals came from wells containing Eu(III) complexes with small ether, polar, or charged residues like methoxyethyl (1), carboxyethyl (7), or the furans (6 and 8) [compounds in wells (1,1), (1,6), (1,7), (7,1), (7,6), (7,7) and (7,8); where the designation (R₁,R₂) refers to (row, column), Fig. 3A]. Interestingly, the position of each residue also had consequences on the CEST signal intensity. A more intense CEST signal was seen when a carboxyethyl group (7) was the first residue, (almost independent of which residue was present at the second position; see row 7) compared to when it was the second residue (column 7). For example, equally intense CEST signals were seen for compounds containing the carboxyethyl group (7) in the first peptoid position and either 2-methylfuran (7,6) or 2-methylthiophene (7,8) in the second position but when the sequence was inverted [(6,7) and (8,7)], both resulted in a weak CEST signal. Thus, one can conclude that the presence of a negatively charged group in the peptoid positioned closest to the Eu(III) center has a positive influence on the CEST signal. This is consistent with previous data reporting that EuDOTA-tetraglycinate has one of the slowest water exchange rates of any CEST agent reported to date.²² The opposite trend was seen for the positively-charged aminoethyl group (4). Here, the CEST intensities did depend on the identity of the second group but, in general, a larger CEST signal was seen for those compounds where the positively charged group (4) occupied the second position of the peptoid sequence, not the first (compare the three bright signals in column 4 with the virtually all dark signals in row 4). This was highlighted even further for those compounds having one positively and one negatively charged group together in one sequence; in this case, as usual the best CEST signal was found for those compounds with the negatively charged carboxyethyl group (7) closest to the metal center (7,4). When the positively charged aminoethyl group (4) occupied the first position, this cancelled most of the advantage of having the negatively charged carboxyethyl group in the second position [compare the signal intensity of (4,7) with that of (3,7) or (5,7)]. This suggests that having a non-polar group close to the metal center is more advantageous than having a positively charged group near the metal center, as long as the second position is occupied by the negatively charged, carboxyethyl group [(1,7) (3,7), (5,7), (6,7)].

In all cases, the bulkier aromatic groups, 5-benzyl-1,3-dioxole (2), 4-methoxybenzyl (3), or 4-chlorobenzyl (9) resulted in the weakest CEST signals, regardless of their position in the peptoid sequence (compare the number of dark wells in columns 2, 3, 8 and 9 with any of the other columns). This is in agreement with our prior observations that bulkier aromatic residues such as these increase the rate water exchange in EuDOTA-tetraamide complexes.²³ Interestingly, the smaller aliphatic, hydrophobic isobutyl group (5) did not show a similar trend. When this group appeared in combination with a carboxylethyl group (in either order), the CEST intensities were similar [compare wells (5,7) with (7,5)]. A similar but less obvious trend was observed with 2-methylfuran (6) or 2-methylthiophene (8) acting as the small non-polar residue in combination with a charged group (either positive or negative) as the second component [(6,4)(8,4)(6,7)(8,7)]. This suggests that having a small non-polar group closest to the Eu(III)-water exchange site may act to slow dissociation of the inner-sphere water molecule from the Eu(III) coordination site while the charged group further away acts to organize second-sphere water molecules to slow their entrance into a position where they can be in a position to exchange with the single, inner-sphere water molecule.

It has been reported that the resonance frequency of the bound-water is quite sensitive to the identity of the first amide side-chain group in a series of tetra-substituted amino acids, for example.²⁴ This did not hold true for the 80 complexes prepared in this library because all four side-chains were positioned well away from the central EuDOTA structure by the four ethylenediamine arms. In this case, the frequency of the water exchange peak did not vary substantially (all were near 50-51 ppm) because the ethylendiamine spacer placed the chemically diverse substituents far enough away from the metal to minimize any changes in ligand field. Nevertheless, the most surprising finding of this study was that the water exchange rate and hence the CEST intensity can vary substantially even with these chemically diverse groups are positioned well-away from the central Eu(III). Thus, the differences in CEST signal intensities observed here can only reflect differences in water accessibility to the inner-sphere coordination sphere.

1_si_001