Modular Synthesis of Peptide-Based Single- and Multimodal Targeted Molecular Imaging Agents

Abstract

A practical, modular synthesis of targeted molecular imaging agents (TMIAs) containing near-infrared dyes for optical molecular imaging (OMI) or chelated metals for magnetic resonance imaging (MRI) and single-photon emission correlation tomography or positron emission tomography (PET) has been developed. In the method, imaging modules are formed early in the synthesis by attaching imaging agents to the side chain of protected lysines. These modules may be assembled to provide a given set of single- or dual-modal imaging agents, which may be conjugated in the last steps of the synthesis under mild conditions to linkers and targeting groups. A key discovery was the ability of a metal such as gadolinium, useful in MRI, to serve as a protecting group for the chelator, 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA). It was further discovered that two lanthanide metals, La and Ce, can double as protecting groups and placeholder metals, which may be transmetalated under mild conditions by metals used for PET in the final step. The modular method enabled the synthesis of discrete targeted probes with two of the same or different dyes, two same or different metals, or mixtures of dyes and metals. The approach was exemplified by the synthesis of single- or dual-modal imaging modules for MRI–OMI, PET–OMI, and PET–MRI, followed by conjugation to the integrin-seeking peptide, c(RGDyK). For Gd modules, their efficacy for MRI was verified by measuring the NMR spin–lattice relaxivity. To validate functional imaging of TMIAs, dual-modal agents containing Cy5.5 were shown to target A549 cancer cells by confocal fluorescence microscopy.

Graphical Abstract

INTRODUCTION

As a result of advances in molecular imaging, there has been an acceleration of research that compels the need for safe and effective targeted molecular imaging agents (TMIAs) in a variety of modalities. These include TMIAs for magnetic resonance imaging (MRI), positron emission tomography (PET), single-photon emission correlation tomography (SPECT), optical molecular imaging (OMI), photoacoustic imaging (PAI), multispectral optoacoustic imaging (MSOT), confocal fluorescence microscopy (CFM), and other imaging methods.^1–12 By targeting distinct biomarkers, TMIAs can enhance early diagnosis of cancer and other diseases and be used to guide surgery or biopsies for monitoring progression and active surveillance during and after therapy.

In addition to the use of targeting agents to seek and bind to biomarkers, current trends include an eagerness to combine these into dual-modal TMIAs for PET–MRI, PET–OMI, and MRI–OMI. Images from these complementary features of each modality are merged to enhance the interpretation and ensure accurate diagnoses.^4,13–21 This trend is evidenced by a growing number of preclinical instruments now available or in development for dual imaging including PET–OMI, MRI–OMI, PET–CT, and PET–MRI.^22–27 Despite advances in instrumentation, the number of methods for combining two dyes or two metals or mixing dyes with metals in a practical synthesis of dual imaging agents has diminished, particularly methods which include conjugation to targeting agents.

The increased interest in dual agents for PET–MRI stems from advantages in fusing the high sensitivity of PET with the high spatial resolution in MRI.^28–31 Synthetic approaches to dual PET-MRI agents have focused on non-targeted small molecules^32–34 or nanoparticle systems which combine metals such as Gd with radiotracers such as Ga, In, Y, or Cu.^35–37 Two reported methods for targeted probes for PET–MRI include a dendritic platform for 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA)–Gd complexes centered around a single 2,2′,2″-(l,4,7-triazacyclononane-1,4,7-triyl)triacetic acid (NOTA)–Ga construct, using the well-known cancer-targeting peptide c(RGDyK).³⁸ In a second report, 5% of Gd is transmetalated to a radioactive Cu analogue, resulting in a defined ratio of metals for MRI–PET in a fibrin-targeted probe EP-2104R but not a discrete TMIA.³⁹

The quest for dual-modal TMIAs for PET–OMI seeks to combine the advantages of whole-body nuclear imaging and highly sensitive near-infrared (NIR) fluorescence imaging. This has led to dual PET–OMI probes that target a variety of cancer types.^13,40–47 Advantages of combined modalities are described as ensuring simultaneous bioavailability, pharmacokinetic distribution, and ensuring signals from each reporter originate from the same biomarkers at the same time points.⁴⁴

Dual-modal agents for MRI–OMI offer a high spatial resolution and whole-body imaging in MRI combined with highly sensitive fluorescence or acoustic signals by OMI or PAI (also known as MSOT), which could be used in complementary detection and therapeutic methods.^48–55

Another approach to improve contrast agents for MRI is to combine two or more Gd metals in the same probe. As described in peptide-based dendrimeric approaches, this results in an increase in proton relaxivity, dramatically increasing the signal in MRI.^38,56,57

An additional desirable feature in probe design is the ability to methodically place two different NIR dyes on the same agent for multispectral imaging in OMI, CFM, and MSOT.^58–60

The use of small peptides as a platform to construct diverse TMIAs has been beneficial due to their inherent features in bioavailability, biodistribution, and rapid clearance.^44,61 Peptides are well-defined structures with which site-specific conjugations of targeting agents can be made. They also have advantages in tumor penetration and can provide high target-to-background ratios at early time points. Peptides may be tuned for proteolytic stability in blood using d-amino acids and other means.^62–64 In addition, they can be readily purified by HPLC, characterized by MS, and readily assayed in blood and other fluids. These features augment the safety and efficacy studies in preclinical studies, enhancing clinical translation.

Synthetic approaches to peptide-based dual-modal probes for PET, MRI, and NIR have been described earlier.^{13,44,61,65–68} They most often begin with the targeting agent followed by the coupling of imaging groups to it. As targeting agents are often a precious or sensitive sample such as a peptide, an antibody, an aptamer, or an inhibitor and are often available in minute quantities, they may be troublesome as starting materials and it would be better to conjugate them in the last step.

The unprotected form of a chelating agent such as DOTA is often added in the penultimate step, followed by introduction of Gd or a radioactive metal in the last step in synthetic approaches to TMIAs containing metals. This is because DOTA, when conjugated, contains four amines and three carboxylic acids, which can cause side reactions in peptide-coupling reactions and hamper purification. In addition, the half-life of radioactive metals generally necessitates their late arrival into the syntheses.

When the unprotected form of DOTA is not amenable to the synthesis, the use of the tri-(tert-butyl) ester form is most often utilized and carried through the synthesis.⁶¹ Removal of tert-butyl groups, however, requires strong acids such as trifluoroacetic acid in which many peptides, proteins, antibodies, and NIR dyes are unstable. These restrictions severely limit the way in which DOTA and metal–DOTA complexes can be incorporated into dual TMIAs for MRI, PET, or SPECT.

In our initial approach to dual-modal TMIAs, we employed a traditional orthogonal protection scheme to create a dual-modal probe for MRI–OMI. The approach suffered drawbacks arising from incompatibilities between deprotection conditions and imaging moieties. For example, NIR dyes and chelated metals were not stable in strongly acidic conditions or hydrogenation. Our options to create dual agents were likewise constrained to attaching imaging groups onto targeting agents.

It became clear that conjugating the targeting group in the final steps of synthesis under mild conditions to a pre-constructed single or dual imaging system would be ideal. To facilitate this, a practical solution was needed to assemble metals and NIR dyes onto a scaffold in the initial steps. This would further enable the coupling of a given imaging system to diverse sets of given targeting systems in the final steps.

MODULAR METHOD

Our strategy was to utilize peptides as scaffolds but to reassemble these through a modular method. To enhance the bioavailability, pharmacokinetics, and tissue/cell penetration, desirable features in the design were being compact in size (less than 3000 amu), available from inexpensive materials, easily purified and characterized, water soluble, and stable in ambient atmosphere, and the ability to introduce the targeting group in the last step.

In a key finding, we discovered that, when introduced early in the synthesis, the MRI contrast metal Gd could double as a protecting group for DOTA, thus avoiding the use of the tri-(tert-butyl) ester form. Using this approach, imaging modules could be assembled by first coupling Gd–DOTA to the side chain of lysine. This enabled a modular strategy shown for single-modal TMIAs in Scheme 1 as listed in Table 1.

Scheme 1. Modular Synthesis of Single-Modal TMIAs for OMI, MRI, or PET Utilizing Standard Peptide Chemistry: (i) Fmoc Deprotection with DEA, (ii) Addition of Linker, DIPEA, and DMF, and (iii) Conjugation of Targeting Group, DIPEA, and DMF or NMP.

Table 1.

Single-Modal Imaging Agents: Intermediates, Metal or NIR Dye Imaging Modules, Linkers Utilized, and Final Single-Modal TMIAs

	compound number	compound name	linker	targeting agent	imaging module
precursors	1	Fmoc-Lys(Boc)-NH2
	2	Fmoc-Lys-NH2
	3	c(RGDyK)-DSS	DSS	c(RGDyK)
MRI	4	Fmoc-Lys(Gd-DOTA)-NH2a			Gd–DOTA
	5	H-Lys(Gd-DOTA)-NH2			Gd–DOTA
	6	SMCC-Lys(Gd-DOTA)-NH2	SMCC		Gd–DOTA
	7	c(RGDyK)-SMCC-Lys(Gd-DOTA)-NH2a	SMCC	c(RGDyK)	Gd–DOTA
	8	Fmoc-d-Lys(Gd-DOTA)-OHa			Gd–DOTA
PET	9	Fmoc-Lys(La-DOTA)-NH2			La–DOTA
	10	H-Lys(La-DOTA)-NH2			La–DOTA
	11	Fmoc-d-Lys(La-DOTA)-OH			La–DOTA
	12	SMCC-Lys(La-DOTA)-NH2	SMCC		La–DOTA
	13	c(RGDyK)-SMCC-Lys(La-DOTA)-NH2	SMCC	c(RGDyK)	La–DOTA
	14	c(RGDyK)-SMCC-Lys(Cu-DOTA)-NH2	SMCC	c(RGDyK)	Cu–DOTA
OMI	15	Fmoc-Lys(Cy5.5–3S)-NH2			Cy5.5–3S
	16	H-Lys(Cy5.5–3S)-NH2			Cy5.5–3S
	17	Fmoc-d-Lys(Cy5.5–3S)-OH			Cy5.5–3S
	18	Fmoc-d-Lys(Cy7)-OH			Cy7

^aCompounds tested in relaxivity studies by NMR with data shown in Figure 8 and Table 3.

The same strategy was next applied to form modules based on NIR dyes. Fortuitously, dye modules were also found to be stable to Fmoc-based peptide synthesis and coupling conditions, enabling the synthesis of dual-modal imaging modules as illustrated in Scheme 2 and listed in Table 2. These included combinations of two same or different dyes, two same or different metals, or combinations of dyes and metals.

Scheme 2. Modular Synthesis of Dual-Modal TMIAs for MRI−OMI Utilizing Standard Peptide Chemistry: (i) Coupling with HATU or TBTU, and DIPEA; (ii) Fmoc Deprotection with DEA; (iii) Linker, DIPEA, and DMF; and (iv) Targeting Group, DIPEA, DMF, or NMP.

Table 2.

Dual-Modal, Dual Dye, and Di-Gd Imaging Agents: Intermediates, Metal or NIR Dye Modules, Linkers, and Final TMIAs in Scheme 4

	compound number	compound name	linker	targeting agent	imaging module I	imaging module II
MRI-OMI	19	Fmoc-d-Lys(Cy5.5–3S)-Lys(Gd–DOTA)-NH2a			Cy5.5–3S	Gd–DOTA
	20	H-d-Lys(Cy5.5–3S)-Lys(Gd–DOTA)-NH2			Cy5.5–3S	Gd–DOTA
	21	SMCC-d-Lys(Cy5.5–3S)-Lys(Gd-DOTA)-NH2	SMCC		Cy5.5–3S	Gd–DOTA
	22	c(RGDyK)-SMCC-d-Lys(Cy5.5–3S)-Lys(Gd–DOTA)-NH2	SMCC	c(RGDyK)	Cy5.5–3S	Gd–DOTA
MRI-OMI	23	Fmoc-d-Lys(Cy5.5–3S)-Lys(La–DOTA)-NH2			Cy5.5–3S	La–DOTA
	24	H-d-Lys(Cy5.5–3S)-Lys(La–DOTA)-NH2			Cy5.5–3S	La–DOTA
	25	c(RGDyK)-DSS-d-Lys(Cy5.5–3S)-Lys(La–DOTA)-NH2	DSS	c(RGDyK)	Cy5.5–3S	La–DOTA
	26	c(RGDyK)-DSS-d-Lys(Cy5.5–3S)-Lys(Cu–DOTA)-NH2	DSS	c(RGDyK)	Cy5.5–3S	Cu–DOTA
PET-MRI	27	Fmoc-d-Lys(La-DOTA)-Lys(Gd–DOTA)-NH2			La–DOTA	Gd–DOTA
	28	H-d-Lys(La-DOTA)-Lys(Gd–DOTA)-NH2			La–DOTA	Gd–DOTA
	29	SMCC-d-Lys(La-DOTA)-Lys(Gd–DOTA)-NH2	SMCC		La–DOTA	Gd–DOTA
	30	c(RGDyK)-SMCC-d-Lys(La-DOTA)-Lys(Gd–DOTA)-NH2	SMCC	c(RGDyK)	La–DOTA	Gd–DOTA
	31	c(RGDyK)-SMCC-d-Lys(Cu-DOTA)-Lys(Gd–DOTA)-NH2	SMCC	c(RGDyK)	Cu–DOTA	Gd–DOTA
	32	c(RGDyK)-DSS-d-Lys(La-DOTA)-Lys(Gd–DOTA)-NH2	DSS	c(RGDyK)	La–DOTA	Gd–DOTA
	33	c(RGDyK)-DSS-d-Lys(Cu–DOTA)-Lys(Gd–DOTA)-NH2	DSS	c(RGDyK)	Cu–DOTA	Gd–DOTA
	34	c(RGDyK)-DSS-d-Lys(Ga–DOTA)-Lys(Gd–DOTA)-NH2	DSS	c(RGDyK)	Ga–DOTA	Gd–DOTA
Di-Gd MRI	35	Fmoc-d-Lys(Gd–DOTA)-Lys(Gd–DOTA)-NH2a			Gd–DOTA	Gd–DOTA
	36	H-d-Lys(Gd–DOTA)-Lys(Gd–DOTA)-NH2			Gd–DOTA	Gd–DOTA
di-OMI	37	c(RGDyK)-DSS-d-Lys(Gd–DOTA)-Lys(Gd–DOTA)-NH2	DSS	c(RGDyK)	Gd–DOTA	Gd–DOTA
	38	Fmoc-d-Lys(Cy7)-Lys(Cy5.5–3S)-NH2			Cy7	Cy5.5–3S

^aCompounds tested in relaxivity studies by NMR with data shown in Figure 8 and Table 3.

A spacer between the imaging agent and the targeting agent is often employed in the design of TMIAs. Linkers such as SMCC and DSS may position the imaging agent outside the binding pocket to provide optimal receptor binding. Additionally, these may impart enhanced bioavailability and tissue permeation. Linkers can be attached to the modules in the second to last step and targeting group in the last step as shown in Schemes 1 and 2. The result was a truly modular approach to single- and dual-modal TMIAs.⁶⁹ The peptide-based approach is simple, practical, and highly versatile.

RESULTS: SINGLE- AND DUAL-MODAL TMIAS FOR OMI AND MRI

All imaging peptides and TMIAs were prepared by solution-phase synthesis. To avoid the potential reactivity of a C-terminal acid, the first l-lysine residue in each case was prepared as the carboxamide. For increasing the proteolytic stability in vivo, d-lysine was utilized for the second module.^70,71 The nomenclature employed, with imaging agents on peptide side chains in parentheses, is described in the Supporting Information.

To demonstrate the utility of the modular approach, single- and dual-modal TMIAs were synthesized using a variety of imaging modalities prepared in our labs including DOTA chelates of gadolinium (Gd³⁺) for MRI and non-radioactive metals such as copper (Cu²⁺) and gallium (Ga³⁺) as models for PET. In order to reduce costs, a tri-sulfonated analogue (Cy5.5–3S) of commercial Cy5.5 was synthesized,⁷² along with Cy7.^69,73

To demonstrate targeting, we chose the established cyclic peptide, c(RGDyK), which binds to a_vβ₃ integrin receptors that are over-expressed in many types of cancer including A549 cancer cells, which we utilized for testing in CFM.^74–76

For modules containing chelated metals, a three-step procedure described earlier⁶⁹ was replaced by a one-pot procedure shown in Figure 1.⁷⁷ The resulting modules may be coupled using standard Fmoc deprotection followed by coupling to a targeting system or to the C-terminal acid of a second Fmoc-protected imaging module as shown in Scheme 2.^69,77

Figure 1.

One-pot synthesis of modules containing DOTA-chelated metals. M₁ = Gd or La and R=OH or NH₂. (i) DOTA, TSTU, and NMM in DMF; (ii) Fmoc-Lys-OH and NMM; and (iii) Gd(OAc)₃. Lysines are d-Lys when R=OH and l-Lys when R=NH₂.

Modules containing NIR dyes were synthesized as shown in Figure 2. Single-modal TMIAs for the OMI and PAI of prostate cancer (PCa) based on dye modules were described earlier by the Pomper group.⁷⁸ We synthesized analogues of these by the modular method, coupled the linker DSS and then urea DCL to target PSMA, an overexpressed receptor in PCa, and report further testing of these along with our prior reports.^78–80

Figure 2.

Synthesis of imaging modules containing NIR dyes where R=OH or NH₂. (i) DIPEA, DMF, or NMP. Lysine residues are d-Lys when R=OH and l-Lys when R=NH₂.

Single-Modal TMIAs for MRI.

Of the TMIAs shown in Table 1, an example of an easily constructed single-modal TMIA (7) prepared by the method in Scheme 1, containing the chelated Gd–DOTA with SMCC linker, is shown in Figure 3. In an earlier publication, we described a related application of the modular method to single-modal TMIAs for MRI and PET of PCa, similarly targeting PSMA by the use of DSS-DCL.^{67,77,81–83}

Figure 3.

Compound (7), c(RGDyK)-SMCC-Lys(Gd-DOTA)-NH₂, a single-modal TMIA containing Gd-DOTA for MRI.

Dual-Modal TMIA for MRI-OMI.

An illustrative example of a dual-modal agent for MRI-OMI is compound (22), depicted in Scheme 2 and in Figure 4, which contains Cy5.5–3S and Gd-DOTA, with the SMCC linker.⁸⁴ The efficacy of the c(RGDyK) targeted dual-modal TMIA was tested and shown to bind effectively to A549 cancer cells as described below.

Figure 4.

Compound (22), c(RGDyK)-SMCC-d-Lys(Cy5.5–3S)-Lys(Gd−DOTA)-NH₂, a dual-modal TMIA for MRI−OMI containing Gd−DOTA for MRI and Cy5.5–3S for OMI.

Placeholder Method for PET/SPECT Agents.

For the construction of TMIAs for PET agents, it was clearly not feasible to carry a radioactive metal through a multi-step synthesis. However, by considering literature studies involving demetalation by Sherry and Chang,^85,86 a transmetalation method was envisioned that could take advantage of the unique fragility of certain lanthanides when chelated to DOTA. In a mild acid, the stability of chelation decreases across the row of lanthanides from the right to left due to the lanthanide contraction.⁸⁷ Employing this concept, we postulated that La³⁺ or Ce³⁺ could be used as acid-labile “placeholders” while also doubling as protecting groups. When tested, these two metals were readily transmetalated by Cu²⁺, Ga³⁺, Y³⁺, and In³⁺, thus providing a route to TMIAs for PET, as illustrated in Scheme 3.

Scheme 3. Modular Synthesis of Single-Modal TMIAs for PET and SPECT from Imaging Modules: (i) Linker, DIPEA, and DMF; (ii) Targeting Group, DIPEA, and DMF; (iii) Transmetalation in Two Steps: 0.2M TFA in H₂O; and (iv) Addition of Ga, In, Y, or Cu in NaOAc Buffer.

Transmetalation Methods.

The use of La³⁺ over Ce³⁺ was found to be preferable as La³⁺ was displaced more rapidly than Ce³⁺. However, Ce³⁺ could be used in cases where a more stable placeholder is desired.⁶⁹ The conditions for the demetalation by Sherry⁸⁵ were optimized for synthesis in further kinetic studies and applied to the modular method for TMIAs. The mildest conditions for transmetalation while still allowing complete removal of the placeholder were found to be between 0.05 M TFA for La³⁺ and 0.2 M TFA for Ce³⁺.

While the transmetalation could be carried out in one step with the metal present (method A, Figure 5), the long reaction times necessary would not be compatible with the half-lives of the radiolabels. After verifying that the rate-limiting step was de-metalation, a two-step method was developed (method B, Figure 5), with treatment of placeholder-TMIA with dilute acid (0.1–0.2 M TFA), for 16–24 h to first de-metalate and then add the metal for PET to rapidly chelate the DOTA.

Figure 5.

Single-modal TMIAs with c(RGDyK)-SMCC for PET in placeholder M₁ = La are transmetalated to M₂ = Cu.

A listing of all intermediates, metal and dye imaging modules, linkers, and final single-modal c(RGDyK) TMIAs prepared are listed in Table 1. All structures are shown in the Supporting Information.

Dual-Modal TMIAs for PET-OMI.

The placeholder method was then applied to the facile synthesis of c(RGDyK)-targeted TMIA (26) from (25) as shown in Figure 6 after first combining modules (10) and (17) containing La–DOTA and Cy5.5 by the modular method in Schemes 2 and 4.

Figure 6.

Compound (25), c(RGDyK)-DSS-d-Lys(Cy5.5–3S)-Lys(La−DOTA)-NH₂, is transmetalated to replace La with Cu to produce (26), a dual-modal TMIA for PET−OMI.

Scheme 4. Modular Synthesis of Dual TMIAs for PET, OMI, and SPECT: (i) Deprotection with DEA; (ii) DIPEA, DMF, and Convergent Addition of TG-DSS; and (iii) Transmetalation.

To confirm complete removal of the placeholder metal ion, the demetalation was monitored to the detection limit of the La³⁺ module by HPLC-MS. While the chelation by Cu²⁺ was complete in method B, La³⁺ was found to complete with Ga³⁺ for re-metalation under these conditions. To solve this and ensure complete transmetalation, a method was devised to adhere the demetalated TMIA in mild acid to a C-18 solid phase extraction (SPE) column, washing away the La³⁺, then treating the column with a 5 mM aqueous solution of Ga³⁺, and eluting the pure product with an acetonitrile-water gradient.^77,88

Dual-Modal TMIAs for PET–MRI.

In addition to the advantages of serving as a protecting group for DOTA during peptide synthesis and serving as a means of storing a penultimate form of the non-radioactive TMIA, a third advantage of the placeholder metal arose when pursuing dual metal TMIAs for PET–MRI as follows.

Through detailed kinetic studies in which various DOTA–metal chelates as their Fmoc modules (compounds 4 and 5) were subjected to a dilution series of TFA (not yet reported), it was discovered that by using a molarity to 0.02–0.10 M (pH 2–3), it was possible to keep the Gd–DOTA chelate entirely intact while de-metalating either La or Ce. This differential stability of lanthanide metals and robustness of Gd–DOTA was consistent with earlier studies by Sherry⁸⁵ and ultimately enabled the selective synthesis of metal pairs, Cu–Gd and Ga–Gd, from La–Gd or Ce–Gd via the modular method.

In control experiments, it was further determined that indocyanine dyes such as Cy5.5–3S, Cy7, and the widely used dye IRDye800CW (used in separate studies in our lab) were also stable in the mildly acidic transmetalation conditions (0.02–0.10 M TFA) to similarly enable transmetalation of La³⁺ and Ce³⁺ in the presence of these NIR dyes.

These results paved the way for a practical route to dual-modal agents for both PET–MRI and PET–OMI with examples of each shown in Scheme 4.⁸⁹ The dual-modal agents for PET–MRI were assembled by first preparing the C-terminus imaging module Fmoc-Lys(Gd–DOTA)-NH₂, then deprotection of the Fmoc, and coupling to the second imaging module Fmoc-d-Lys(La–DOTA)-OH to produce the dual imaging module Fmoc-d-Lys(La–DOTA)-Lys(Gd–DOTA)-NH₂. This in turn was followed by deprotection of the Fmoc and convergent coupling to a targeting module comprising c(RGDyK) and the DSS linker.

The dual-modal agents for PET–OMI as shown in Scheme 4 were assembled by preparing the C-terminus imaging module Fmoc-Lys(La–DOTA)-NH₂ followed by deprotection of the Fmoc. This was coupled to the placeholder module as described above to provide the dual imaging module Fmoc-d-Lys(Cy5.5–3S)-Lys(La–DOTA)-NH₂. Deprotection of the Fmoc and coupling similarly in a convergent step to the targeting module c(RGDyK)-DSS yielded the penultimate TMIA which was, in turn, transmetalated to the Cu analogue (26).

Summary of Intermediates and TMIAs.

A listing of intermediates for dual-modal modules and dual-modal TMIAs is shown in Table 2. Detailed structures for each of these, along with experimental methods and structural characterization, are shown in the Supporting Information.

Convergent versus Linear Conjugation of the Linker and Targeting Groups.

Two approaches were developed for the linker and targeting groups. In a first linear approach, the linker SMCC or DSS was added in a sequential manner. An advantage of this route is to be able to store and transport a reactive linker form of the dual module (SMCC or DSS) for later conjugation to a targeting group, which could be chosen from a diverse set at a distant clinic or research lab. It was found that these reactive linker conjugates are stable when stored at −20 °C.

A second option is the convergent approach shown in Figure 7. Here, the targeting moiety was pre-attached to the linker to form a targeting-linker module. The advantage of this option is one less step in the synthesis.

Figure 7.

Convergent method for the attachment of a dual-imaging module to a targeting-linker module to form (32) followed by transmetalation to TMIAs for PET−MRI, (33) and (34).

Di–Gd–DOTA TMIAs.

It is well established that an increase in relaxivity per Gd may be achieved in agents with a multiplicity of Gd atoms.^32,38,56,57 We therefore extended the method as a route to bis-gadolinium agents. The di-peptide module Fmoc-d-Lys(Gd–DOTA)-Lys(Gd–DOTA)-NH₂ (35) was prepared and then coupled via the convergent approach to the module, resulting in the c(RGDyK)-targeted di-Gd TMIA (37).

NMR Relaxivity Measurements.

To assess the efficacy of the Gd-based contrast agents in MRI, the spin lattice relaxivity (r₁) values of key modules marked (*) in Tables 1 and 2 were measured and compared to a sample of Gd-DOTA (Dotarem). These included two single-modal, one dual-modal, and one di-Gd imaging modules.

Aqueous solutions which varied in concentrations from 0.5 to 5 mM were placed in 1 mm I.D. capillary tubes and relaxivity measurements were carried out in a magnetic field strength of40 MHz T. The r₁ values and data plots presented in Figure 8 and Table 3 show that the single Gd-containing modules have r₁ values comparable to Dotarem. The di-Gd modules exhibited a value of 11.33 mM⁻¹ s⁻¹ T, which, as expected, doubled the value of the mono-Gd agents.

Figure 8.

Plot of T₁ relaxation times vs concentration (mM) of compounds 4, 7, 8, 19, and 35 compared to Dotarem. Measurements were performed at 20 °C in H₂O at 40 MHz.

Table 3.

Relaxivity Values (r₁) for Gd-Containing Modules

contrast agent	r1 (mM−1s−1) 1T
Fmoc-Lys(Gd–DOTA)-NH2 (4)	4.95
c(RGDyK)-SMCC-Lys(Gd–DOTA)-NH2 (7)	4.40
Fmoc-d-Lys(Gd–DOTA)-OH (8)	5.26
Fmoc-d-Lys(Cy5.5–3S)-Lys(Gd–DOTA)-NH2 (19)	4.72
Fmoc-d-Lys(Gd–DOTA)-Lys(Gd–DOTA)-NH2 (35)	11.33
Gd–DOTA (Dotarem)	4.09

Gd-DOTA conjugates of (RGDyK) have been shown to be effective tumor-targeting MRI contrast agents in vivo.⁹⁰ The concentrations of the a_v and β3 receptor per cell are in a range from 3 × 10³ to 1.4 × 10⁴. For a direct conjugate, an injection concentration of 1.42 mmol kg⁻¹ appears suitable.⁹¹ In a second study, an RGD-targeted poly(l-glutamic acid)-cystamine-(Gd-DO3A) conjugate was shown to be an effective probe for MR imaging of cancer via T₁ mapping.⁹² With T₁ relaxivities in a, comparable range as TMIAs (7) and (22), these would be expected to provide similar efficacy with enhancement of contrast in the tumor due to targeting. A third c(RGDyK) targeted system involving a dendritic construct of five Gd atoms per single Ga atom for dual OMI-PET imaging reports a large increase in relaxivity and excellent targeting.³⁸ The increase is due to both multiplicity of Gd atoms and the large molecular size of the TMIA.

Confocal Fluorescence Microscopy.

To test the efficacy of a representative dual-modal probe, compound (22), c(RGDyK)-SMCC-d-Lys(Cy5.5–3S)-Lys(Gd–DOTA)-NH₂ was chosen and its binding to human lung adenocarcinoma epithelia A549 cancer cells was examined. The affinity of c(RGDyK) to the a_vβ₃ integrin receptor is a well-established system set forth by the Gambhir and Achilefu groups. In these pioneering studies, the direct conjugate of commercial tetra-sulfonated Cy5.5 to c(RGDyK) was shown to effectively target A549 lung cancer cells and the U87MG glioma brain cancer cells.^74,93

As a control to test this system, an analogous direct conjugate, of c(RGDyK) and our tri-sulfonated Cy5.5 analogue Cy5.5–3S-c(RGDyK) was synthesized as described in our earlier report.⁷² The probe is similar to the commercial IntegriSense agent, which employs the related NIR dye Alexafluor 680, (Abs 677, Em 701 nm).⁷²

A549 cells were stained with the control, Cy5.5–3S-c(RGDyK), and scanned under a bright field and fluorescent light Figure 9. The differential interference contrast (DIC) images showed healthy cells with a morphology similar to untreated cells. Fluorescence imaging revealed punctate, cytoplasmic staining by the control, Cy5.5–3S-c(RGDyK), surrounding the nucleus region. A 30 min time lapse scan was carried out on the live, stained cells. Various regions of the plate were scanned and imaged to display what appeared to be the uptake and accumulation of TMIA within the cells by endocytosis. Staining of the conjugate was observed in cells located on all areas of the plate, indicating uptake of the TMIA occurred in most cells by 30 min. The observed staining was consistent with the findings reported previously by Gambhir and Achilefu groups.^66,74,93

Figure 9.

DIC image of confluent A549 cell lines grown on a glass slide (left): Live A549 cells stained with 5 μM Cy5.5–3S-c(RGDyK) (emission 674 nm, right). Punctate staining is observed in 1 h (not shown) and 4 h (shown).

Endocytosis of a similar, Cy5-based, tetrameric c(RGDyK) targeting vector was reported in a similar study in HEK293(β3) tumors.⁹⁴ The same tetrameric vector is the basis of the targeted probe Angiostamp 800, which has been utilized for in vivo PAI/MSOT imaging in liver tumors⁹⁵ and fluorescence image-guided surgery in an orthotopic animal model of head and neck squamous cell carcinoma.⁹⁶

Using the same procedure and the same concentrations as for the control, Cy5.5–3S-c(RGDyK) (Figure 9), the A549 cells were stained with the dual TMIA (22), c(RGDyK)-SMCC-d-Lys(Cy5.5–3S)-Lys(Gd-DOTA)-NH₂. The results, shown in Figure 10, with incubation at 1 and 4 h, mirrored the imaging experiments using the direct c(RGDyK) conjugate. Similar punctate, cytoplasmic staining was observed along with visible endocytosis of the TMIA at both 1 and 4 h time points. This result is consistent with a report describing enhanced cell penetration of Gd-based MRI contrast agents by conjugation with hydrophobic NIR dyes.⁵⁴ It would be expected that an increase in concentration of Gd within each cell would result in higher-resolution MRI images as well.

Figure 10.

Imaging of A549 cells stained with 5 μM dual-modal TMIA (22). Confluent monolayers of live cells were stained for 1 (left) or 4 h (right). The nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; blue).

Both the direct conjugate Cy5.5–3S-c(RGDyK) and dual agent (22) were imaged using a triple-staining technique in which nuclei were stained with DAPI (emission: 461 nm), actin filaments were stained with phalloidin (emission: 518 nm), and the TMIAs conjugated to Cy5.5–3S (emission: 681 nm), as shown in Figure 11. Sequential scanning was carried out to eliminate bleed-through and to obtain better dye separation for DAPI and Phalloidin.

Figure 11.

Living A549 cells (left) stained with 5 μM Cy5.5–3S-c(RGDyK), endocytosed in cytoplasm (red), DAPI (nucleus, blue), and phalloidin (outer membrane, green). Experiments with TMIA (22) yielded similar results. Integrin-receptor-targeting mechanism was verified by the blocking experiment using the c(RGDyK) peptide (right).

To determine if the mechanism of binding was specific to the a_vβ₃ integrin receptors, a control experiment was carried out emulating the previous literature precedent.⁷⁴ Cells were pre-incubated with a blocking agent that contained 200 nM of the unconjugated c(RGDyK) for 30 min. The cells were then co-incubated with the blocking agents and Cy5.5–3S-c(RGDyK) for 1 h. The reduction in signal, as depicted in Figure 11, clearly showed that the c(RGDyK) peptide bound to integrin receptors and effectively competed with the Cy5.5–3S-c(RGDyK) conjugate for receptor-binding sites.

CONCLUSIONS

A versatile, modular method devised for the synthesis of TMIAs is poised to offer a broad impact in molecular imaging. In this method, NIR dyes for OMI, PAI, or MSOT and metals for MRI, PET, and SPECT can be used singly or in specific combinations early in the synthesis, followed by conjugation to linkers and targeting groups in the final steps. The method features mild conditions at each step to maintain the integrity of imaging dyes and chelated metals. Using this method, a single- or dual-imaging system containing an active linker can be prepared, stored, or transported and then conjugated to a variety of targeting motifs in the last step. The method further facilitates the synthesis of targeted probes for PET by the use of a lanthanide metal as both a placeholder and a protecting group for DOTA. This alleviates the concern for the loss of radioactivity during synthesis or transportation. A simple transmetalation in mild acid at the point of use can provide the final radiolabeled TMIAs when needed. The design further enables transmetalation of La by a radioactive metal while leaving the Gd-chelated module intact, thereby facilitating the synthesis of dual probes for MRI-PET.

The design of TMIAs using peptide modules is advantaged by the inherent features of small peptides in bioavailability, biodistribution, and cell and tissue penetration. In addition, the TMIAs can be readily characterized by LC–MS and purified by SPE or preparative-HPLC, which are features that enhance translation to human studies and clinical development. The application to TMIAs has been exemplified by use of a well-established cancer-targeting agent, c(RGDyK), with verification of the receptor binding to A59 cancer cells by CFM. Their efficacy for MRI has been validated by measurement of r₁ relaxivity.

Further on-going work in our lab is focused on synthesis of TMIAs for prostate cancer imaging based on the PSMA inhibitor DCL as a targeting motif.⁷⁷ Synthesized by the same modular method, these will provide targeted probes for OMI and MRI, which are analogous to clinically useful TMIAs such as ⁶⁸Ga-DKFZ-PSMA-11^65,97 and therapeutic agents such as 177Lu-PSMA-617. ⁹⁸

EXPERIMENTAL PROCEDURES

The detailed synthesis of compounds, Tables 1 and 2, are described in detail, along with structures, in the Supporting Information. This includes in all dyes, modules, and single- and dual-modal TMIAs. Data from LC–MS and high-resolution mass spectroscopy (HRMS), 1-D and 2-D spectra are also included in the Supporting Information. Important analytical methods are described here along with methods for evaluation by CFM.

Analytical and Purification Methods.

For LC–MS, a Waters 2695 Alliance HPLC with a Waters 2998 diode array detector and a Waters 3100 SQ mass spectrometer was used. For HPLC analysis of dye-containing modules, the instrument used was an Agilent 1100 with a diode array detector. HPLC columns used were an Agilent XDB C18 column, with dimensions of 3 mm × 100 mm or a Waters XBridge C18 column of 50 mm × 3 mm dimensions and 3μ particle size. Mass spectra from this instrument were recorded at unit resolution with positive and negative switching modes at 35 or 50 V cone voltages. The flow rate for HPLC-MS was 0.5 mL/min. All aqueous mobile phases for HPLC are 0.01 M ammonium acetate unless otherwise noted. Preparative HPLC (prep-HPLC) was carried out with a Waters 600E system controller and a Waters 600 multi-solvent delivery system using a 30 mL/min flow rate.

In many cases, preparative SPE purification was an adequate substitute for preparative HPLC, particularly on small scales. For SPE purification, a 10 or 20 g C-18 Sep-pack Varian Mega Bond Elut (20CC/5GRM) SPE cartridge was utilized, scales under 30 mg and the DOTA transmetalations utilized a Varian Bond Elut (C18, 12CC/2GRM) SPE cartridge. Further details on solvents and columns are presented in the Supporting Information.

High-resolution mass spectra were obtained on a Waters Synapt G2Si (School of Chemical Sciences, University of Illinois at Urbana-Champaign) using the following parameters: flow injection at a flow rate of 0.1 mL/min, H₂O/ACN/0.1% formic acid, positive and negative mode ESI, cone voltage = 25, capillary voltage = 3.0, ion source temperature = −100 °C, desolation temperature = 180 °C, nebulizing gas (N₂) flow = 200 L/h, and cone gas (N₂) flow = 5 L/h.

The NMR (¹H, ¹³C, 2D correlation) data were obtained using a Bruker AVANCEIII 500 MHz NMR spectrometer. Chemical shifts (δ) are reported in ppm relative to TMS.

Relaxivity Measurements.

Relaxivity measurements were performed at 40 MHz on a Magritek Spinsolve Benchtop 1 T NMR at 29 °C.

Aqueous solutions of the Gd-containing compounds were prepared at 0.0, 0.5, 1.0, 2.0, 3.0, and 4.0 mM concentrations using deionized water. 50 μL of each solution was transferred into an ~0.9 mm ID × 90 mm length borosilicate glass capillary tube (Kimble Chase). The tube was loaded into a standard 5 mm OD glass NMR tube (Wilmad-LabGlass) with a Teflon capillary tube adapter.

An inversion recovery sequence was used to measure the spin-lattice relaxation time (T₁) of the solutions. At least 20 inversion time (T_I) values were chosen for each sample with the maximum of ~5T₁ The maximum T_I ranged from 0.2 to 10 s, depending on the sample. Plots of the integrated water signal versus T_I were fit with an exponential growth to determine T₁ for a sample. The relaxivity (r₁) was calculated from the slope of a plot of 1/T₁ (s⁻¹) versus [Gd³⁺ (mM) using the T₁ values for the six concentrations of Gd(III) and eq 1.

$$1/T_1=r_1[\text{Gd}(\text{III})]+1/T_1\left({\text{H}}_2\text{O}\right)$$ (1)

Cell Culture, Staining, and CFM Materials and Methods: Cell Preparation.

An A549 culture obtained from the American Type Culture Collection (ATCC) was grown to 50–70% confluency in an In Vitro Scientific 35 mm culture dish with a 10 mm glass bottom well. The culture was incubated in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (Atlanta Biologicals) overnight at 37 °C and 5% CO₂. To obtain a suspension cell culture for transfer purpose, the cells were washed with phosphate-buffered saline (PBS) and then 0.25% Trypsin–EDTA (1×) (Invitrogen) was added to detach the cells.

Cell Staining.

The cells were washed 3× with pre-heated PBS. 5 μM of the control, Cy5.5–3S-c(RGDyK), and dual TMIA (22), diluted to 5 μM in PBS were added to the 10 mm glass bottom well and incubated at 37 °C and 5% CO₂. Cells were then washed 4× with chilled PBS. Coverslips were flipped over onto a sterilized glass slides and sealed with clear nail polish. The samples were then viewed immediately under a scanning laser confocal microscope (Leica Microsystems Inc.).

Confocal Microscopy Scanning.

A Leica TCS SP5 IIAOBS Filter Free Tunable Spectral Confocal Research Microscope with a Resonant Scanner and Hybrid Detectors attached to a Leica DMI6000 fully automated microscope was used, with a 40X water immersion objective. The 405 nm diode and He–Ne 633 nm lasers were employed for visualizing suspected auto-fluorescence and Cy5.5–3S-c(RGDyK), peptide conjugate, respectively. Images of stained cells were obtained using a sequential scan. The final image was captured at 1000 Hz with a resolution of 1024 × 1024 pixel and a frame average of 6. LAS AF software was used to analyze the data.

Maximum projection, created from Z-stack operation using Leica software, was utilized to confirm the endocytosis of the Cy5.5–3S-c(RGDyK) conjugate in living cells.

ABBREVIATIONS

DEA

diethylamine

dIPEA

N,N-diisopropylethylamine

DMF

dimethylformamide

DOTA

1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid

HPLC

high-performance liquid chromatography

MRI

magnetic resonance imaging

MS

mass spectrometry

MSOT

multispectral optoacoustic tomography

NMP

N-methyl-2-pyrrolidone

NMR

nuclear magnetic resonance

NOTA

2,2′,2″-(1,4,7-triazacyclononane-1,4,7-triyl)triacetic acid

OMI

optical molecular imaging

PAI

photoacoustic imaging

PET

positron emission tomography

SPE

solid phase extraction

TBTU

2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethylaminium tetrafluoroborate

TFA

trifluoroacetic acid

TSTU

O-(N-succimmidyl)-N,N,N′,N′-tetramethyluronium tetrafluoroborate