Amphiphilic DTPA Multimer Assembled on Icosahedral Closo-Borane Motif as High-Performance MRI Blood Pool Contrast Agent

Abstract

A multimeric MRI blood pool contrast agent based on the closo-borane motif is reported. Twelve copies of an amphiphilic DTPA chelate with amine end groups are appended on carbonate-functionalized closo-borane motif using carbamate linkages. The presence of peripheral phenyl groups on the modified DTPA chelates results in high human serum albumin binding, high relaxivity, and excellent contrast enhancement in vitro and in vivo.

Graphical Abstract

Blood pool contrast agents (BPCAs) that can provide robust contrast enhancement (CE) of the vasculature and prolonged circulation half-life are required for efficient imaging of vascular abnormalities (angiography), for diagnosis of a variety of diseases such as atherosclerosis, aneurysm and stroke.^1–4 High molecular weight (HMW) macromolecules and nanomaterials are highly desirable, as they do not infuse into the interstitial space and are eliminated slowly by glomerular filtration. ^5,6 Recently, atomically precise nanomolecular structures such as organometallic nanoclusters, dendrimers and crown ethers have generated considerable interest for the rational design of nanoscale agents for imaging and therapy, as their chemical structure can be tuned for the topology, size and surface functions for probing biological interactions.^7–10

Most of the clinically approved contrast agents (CAs) exhibit non-specific biodistribution and minimal vascular retention, with limited promise for magnetic resonance angiography (MRA); although modern clinical MRI scanners with faster gradients can address this shortcoming.¹¹ The effectiveness of a CA is defined by its relaxivity, (r₁ or r₂). An ideal CA for MRA should display prolonged vascular retention and signal enhancement.^12,13 To achieve this goal, one approach is the conjugation of low molecular weight (LMW) Gd³⁺ chelates to macromolecules (liposomes, nanoparticles and polymers)¹¹ that can effectively curtail their rotational diffusion (τ_R).¹⁴ Further, by avoiding glomerular filtration due to their larger size, such macromolecule-bound chelates display long blood residence time and high relaxivity at 20–60 MHz. However, these CAs display undesirably long half-life, and at higher magnetic fields, they are hardly superior to their LMW counterparts.

An alternative to this strategy is receptor induced magnetization enhancement (RIME) wherein a LMW CA binds noncovalently to plasma proteins. This concept was utilized by Caravan and co-workers for the development of MS-325 as a BPCA, that was approved by the FDA in 2008 (Figure 1A).¹⁵ MS-325 binds strongly and reversibly to human serum albumin (HSA), leading to significant reduction of τ_R and a consequent increase in r₁ in serum as compared to in phosphate buffered saline (PBS). Secondly, the binding of MS-325 with HSA reduces its glomerular filtration, thus increasing its plasma half-life compared to its parent Gd³⁺-DTPA chelate. The lipophilic, aromatic moiety in MS-325 is responsible for the generation of noncovalent interactions with HSA.¹⁶ Conceptually, the presence of multiple such units can result in higher binding affinity to HSA and thereby lead to longer blood retention.¹⁷ It must be noted that the production of MS-325 has since been discontinued due to poor sales.¹⁸ Thus, our aim was to combine the RIME effect with a multimeric CA, using [closo-B₁₂]²⁻ scaffold as a core structural entity.

Figure 1.

Schematic representation of the concept. (A) Single point binding mode and structure of MS-325. (B) Multipoint binding mode and Closo-MRA, discussed in this paper.

Our approach, encompassing the multipoint binding mode with the borane closomer CA, Closo-MRA, is shown in Figure 1B. The three dimensional icosahedral [closo-B₁₂]²⁻ and its derivatives provide a unique multi-centric core scaffold with well-established chemistry.¹⁹ In recent years, there has been a lot of interest in using a bottom-up approach for the design of nanoscale materials using well-defined building blocks.^8,10 These novel architectures include dendrimers, polymers and various nanomaterials that can be surface engineered for multivalent binding.^20–22 Our group has been actively pursuing the development of polyhedral boranes as a scaffold for the targeted high payload delivery of drugs and imaging agents (termed as closomers).²³ Closomers are quite similar to recently reported gold-nanoparticle based super assemblies that are either rigid or monodisperse;^24,25 but are unique in that they combine these two features in addition to being highly symmetrical as compared to dendrimers and polymeric systems of similar size. This can considerably reduce the rotational motion of the functional payload (in this case Gd³⁺ chelates) attached to the closo-B₁₂²⁻ core via suitable linkers. This in turn contributes to high r₁ and greater CE. Herein, we report the synthesis, relaxivity and in vivo MRA assessment of a new class of multimeric BPCA based on a closo-B₁₂²⁻ motif (Figure 1B) represented by twelve radial arms, each carrying an amphiphilic Gd³⁺-DTPA moiety. It was anticipated that the high Gd³⁺ payload as well as the presence of multiple HSA-binding moieties within a single molecule would result in enhanced relaxivity values together with prolonged intravascular retention time and improved CE during in vivo MRA.

It has been well established that the lipophilic component represented by the aromatic moiety in BPCAs is responsible for the generation of strong noncovalent interactions between the corresponding Gd³⁺ chelate and the plasma protein, HSA. Therefore, we designed a novel amphiphilic ligand by attaching a hydrophobic substituent comprised of a 3, 3-diphenylpropan-1-amine moiety to the Gd³⁺ chelating ligand, DTPA.

The synthesis and characterization details of MRA Lig-and-I (S6) and its Gd³⁺ complex (Gd-S6) are provided in Supporting Information (SI). The Gd-complexation was carried out by reacting MRA ligand-I (S6) with Gd₂O₃ under reflux to yield Gd-S6 (Scheme S2). Gd-S6 was characterized via HRMS and inductively coupled plasma optical emission spectroscopy (ICP-OES).

A prerequisite for CA design for in vivo use is their ready solubility in aqueous media. The MRA ligand-I displayed less than optimal solubility in water, raising concerns about the solubility of the final closomer molecule appended with twelve copies of the MRA ligand on it. To address this shortcoming, we incorporated a small oligoethylene glycol (OEG) linker in the MRA Ligand-I. Although the incorporation of OEG would increase the degree of flexibility along the chain and thus compromise the overall rigidity of the final multimeric CA, this compromise was necessary. The addition of an OEG linker has two potential advantages: assistance in achieving an efficient and robust synthesis of the 12-fold functionalization of closo-B12²⁻ core and it would enhance aqueous solubility of final CA, Closo-MRA. Thus, to address these concerns, we synthesized MRA ligand-II with a short OEG chain.

The consequent steps for the synthesis of MRA Ligand-II (Scheme 1) were strategically similar to that of MRA Ligand-I (Scheme S2). Thus, the first step involved installing an aromatic moiety within the commercially available Boc-Lys(Z)-OH, 1, via the reaction of the free –COOH function in 1 with 3,3-diphenylpropan-1-amine to obtain 2 in 93% yield. Standard Boc deprotection using Trifluoroacetic acid (TFA) in DCM (20%) with 2 was used to obtain the free amine 3 in 96% yield. This pendant free –NH₂ on 3 was next alkylated with the bromo chain S2 to afford protected DTPA analogue 4 in 77% yield. This was a major improvement from the meager 33% yield obtained for the DTPA analogue, S5; corresponding to MRA-Ligand-I (S6, Scheme S2). Next, the Cbz-group in 4 was removed using Pd catalyzed hydrogenolysis to obtain 5 in 96% yield. The –NH₂ functionality in 5 can be used for attachment to various targets/chelates for biomedical applications.

Scheme 1.

Synthesis of MRA Ligand-II.

In our case, e.g., 5 is reacted with a 4-nitrophenyl carbonate derivative, S8 (Scheme S3), to afford the OEG linked DTPA derivative 6 in 87% yield. Finally, hydrazine assisted deprotection of the phthalimide group in 6 led to the formation of 7, MRA Ligand-II, in 94% yield. All the compounds were characterized using various spectroscopic techniques such as NMR, FTIR and HRMS.

We next focused on the attachment of the –NH₂ appended MRA Ligand-II to the closomer core. Closomer-I (Scheme 2),²⁶ a B₁₂-analogue with pendant meta-chloro phenylcarbonate arms, was reacted with MRA Ligand-II (6 equiv per vertex) to obtain complete 12-fold functionalized carbamate closomers, Closomer-II in 80% yield (Scheme 2). Closomer-II was purified by size-exclusion column chromatography on Lipophilic Sephadex LH-20 using MeOH as the eluent and was characterized by NMR. A highly symmetrical singlet peak centered at –17.7 ppm in the ¹¹B NMR spectrum of Closomer-II was indicative of fully substituted closomer.

Scheme 2.

Synthesis of Closo-MRA.

The generation of the final closomer MRA contrast agent, Closo-MRA, was straightforward. First, complete removal of the tert-butyl ester groups from Closomer-II was achieved by treatment with 80% TFA in DCM and confirmed by the absence of the large singlet at δ ~1.4 ppm from its ¹H NMR (Figure S1). The presence of a sharp singlet in the ¹¹B NMR spectra confirmed that 12-fold integrity was maintained at the closomer cage after the TFA deprotection step (Figure S2). Subsequently, the product was reacted with GdCl₃·6H₂O (10 equiv per vertex) in citrate buffer at pH 7 to obtain Closo-MRA in 79% yield. Closo-MRA was purified by extensive dialysis using 1000 MWCO dialysis bags in ultrapure water and was characterized by FTIR and ICP-OES. FTIR spectral analysis showed a characteristic shift of the carbonyl stretch from 1723 cm⁻¹ to 1613 cm⁻¹, confirming the complexation of Gd³⁺ with carbonyl groups of the ligands (Figure S3). The purity of Closo-MRA was assessed using HPLC, a single peak confirming the presence of a singular species (Figure S4). Unfortunately, the HRMS analysis attempts for Closomer II and Closo-MRA were unsuccessful. It is worthwhile to mention that mass spectral data of a similar boron cluster compound was previously reported by our group,²⁷ however, failure in acquiring clean HRMS data for Closomer II and closo-MRA could be attributed to high molecular weight (>10 kDa) of both compounds and the possibility of generation of multiple charged species during ionization. Nevertheless, the ICP-OES of Closo-MRA, specifically the Boron: Gd ratio, showed an average of 12 Gd³⁺ ions per molecule, indicating the formation of fully loaded complex. Dynamic light scattering (DLS) analysis of Closo-MRA in 2% Tween-80/PBS and 4.5% HSA/PBS solution rendered an average particle size of 8–10 nm, while the DLS analysis of Closo-MRA in PBS only showed larger particle size, possibly due to aggregation (Figure S5).

We next evaluated the percentage binding of Closo-MRA and Gd-S6 to HSA (4.5% w/v = 0.67 mM) in PBS at pH 7.4 and 37 °C using an ICP-OES protocol.²⁸ As anticipated, Closo-MRA had higher binding (97%) to HSA than its parent ligand Gd-S6 (44%) and also slightly higher than that of MS-325 (88%)²⁸ (Figure 2A). The significantly higher binding of Closo-MRA to HSA, compared to Gd-S6, is indicative of the advantages of using closo-B₁₂²⁻ core to carry 12 copies of protein binding ligands as potential BPCAs, and further highlights the concept of “multilocus binding” demonstrated earlier by Caravan and colleagues.¹⁷

Figure 2.

(A) Percentage binding of Gd-S6, Closo-MRA and MS-325 with HSA (4.5% w/v = 0.67 mM) in PBS at pH 7.4 and 37 °C. (B) In vivo dynamic contrast enhanced MRA of Closo-MRA in CF1 mice at various time points post injection at 0.1 mmol/kg Gd³⁺ dose and 7 T. (C) Blood vessel contrast enhancement ratio (CER) of Closo-MRA (solid dot) significantly higher than that of Gd-DTPA-BMA (open circle) at all time points up to 3 h (p < 0.01).

Next, the relaxivities r₁ and r₂ were determined for Closo-MRA, Gd-S6 and a clinically approved CA gadodiamide (Gd-DTPA bis-methylamide or Gd-DTPA-BMA), in PBS and 4.5% HSA/PBS (4.5% w/v = 0.67 mM), at pH 7.4 and 7 T (Table S1). In presence of HAS, the r₁ values (Closo-MRA per Gd³⁺ ion r₁ = 6.7 mM⁻¹s⁻¹ and Gd-S6 r₁ = 6.9 mM⁻¹s⁻¹) are higher than the r₁ in PBS (Closo-MRA r₁ = 5.5 mM⁻¹s⁻¹ and Gd-S6 r₁ = 4.6 mM⁻¹s⁻¹). The higher r₁ in presence of HAS is explained by the binding of Closo-MRA and Gd-S6 to HSA, leading to slower molecular tumbling of the resulting species, and a subsequently improved relaxivity. In contrast, Gd-DTPA-BMA does not display any such binding to HSA and showed no increase in relaxivity in presence of HSA. The higher r₁ values upon binding to HSA for Closo-MRA and Gd-S6 are similar to that reported for another BPCA MS-325 (r₁ = 7.1 mM⁻¹s⁻¹ at 9.4 T).²⁹ However, considering the strong binding of both Closo-MRA and MS-325 to HSA (67 kDa), the increase in r₁ is not as high as expected in comparison to Gd-S6 that displays lower binding. This can be attributed to the diminishing contribution of the slow tumbling τ_R towards the enhancement of r₁ at higher magnetic field strengths (≥ 4.7 T). Nevertheless, Closo-MRA has a significantly high molecular relaxivity (in presence of HSA: r₁ = 80.9 mM⁻¹s⁻¹ and r₂ = 683.5 mM⁻¹s⁻¹) (Table S1) and is a proof-of-concept demonstration of a multimeric BPCA in combining the RIME effect for protein-binding enhancement.

Finally, the in vivo potential of Closo-MRA as BPCA was examined by performing 3D dynamic CE MRA in CF1 mice (n = 4) via tail vein intravenous injection of Closo-MRA (Gd³⁺ dose: 0.1 mmol/kg body weight). Mice exhibited strong and prolonged CE of vascular system immediately post injection (p.i.) of Closo-MRA, which persisted for 1 h before diminishing at 3 h and cleared at 24 h (Figure 2B and Figure S8). Control CF1 mice (n = 4), injected with Gd-DTPA-BMA, did not show such persisted CE in the vasculature (Figure S9). The vascular half-life of Closo-MRA was ~30 min; significantly longer than that of Gd-DTPA-BMA (~ 2.5 min) (Figure 2C), and similar to MS-325 (28.5 min in rodents).³⁰ The contrast enhancement ratio (CER) (Figure S10) further confirms the significant CE in blood vessels as compared to the liver, kidney and muscle p.i. of Closo-MRA. The CE of the kidneys, liver and bladder on MRI at 3 h (Figure S11) indicates that Closo-MRA is cleared through both the renal and hepatic routes, as is commonly observed for compounds with a considerable amphiphilic component. ICP-OES analysis of organs 24 h p.i. revealed minimal Gd retention in the tissues, thus enhancing the safety profile of Closo-MRA (Figure S12).

In conclusion, Closo-MRA, a multimeric closo-borane appended with twelve copies of an amphiphilic variation of the DTPA ligand, demonstrates high binding to HSA, high relaxivity and remarkable in vivo CE profile of vasculature. Closo-MRA expands the library of previously reported MRI CAs based on the closo-borane motif and is the first example of such a blood pool CA with favorable in vivo characteristics.