A Janus Chelator Enables Biochemically Responsive MRI Contrast with Exceptional Dynamic Range

Abstract

We introduce a new biochemically responsive Mn-based MRI contrast agent that provides a 9-fold change in relaxivity via switching between the Mn³⁺ and Mn²⁺ oxidation states. Interchange between oxidation states is promoted by a “Janus” ligand that isomerizes between binding modes that favor Mn³⁺ or Mn²⁺. It is the only ligand that supports stable complexes of Mn³⁺ and Mn²⁺ in biological milieu. Rapid interconversion between oxidation states is mediated by peroxidase activity (oxidation) and L-cysteine (reduction). This Janus system provides a new paradigm for the design of biochemically responsive MRI contrast agents.

Molecular magnetic resonance imaging (MRI) adds a dimension of biochemical specificity to the rich anatomic and physiological data attainable through MRI.^1,2 Biochemical specificity is achieved either by conjugating an MRI contrast agent to a targeting vector, or engineering the contrast agent to undergo change in magnetic relaxation, or “activation,” in the presence of a biochemical stimulus. Innovative chemistry has generated molecular MRI contrast agents capable of detecting protein targets,³ pH change,⁴ redox activity,⁵ hypoxia,⁶ ion flux,⁷ neurotransmitters,⁸ necrosis,⁹ and enzyme activity.¹⁰

Sensitivity, dynamic range, and rate of response are the three main challenges to developing molecular MRI contrast agents. T₁-relaxation agents are detected at ≥10 μM (metal ion), thus only a handful of cellular and protein targets are candidates for imaging with targeted agents.¹ Activatable agents suffer from poor dynamic range. With a few notable exceptions,^3,11,12 activatable Gd-based agents rarely achieve >2-fold r₁ change in the presence of physiologically relevant levels of biochemical stimuli. Often, activatable contrast agents require prolonged incubation times before measurable relaxation change is observed.

Elegant examples of activatable agents detected via changes in the chemical exchange saturation transfer (CEST) effect have been reported,⁴ but these agents are typically detected with far lower sensitivity than T₁-agents.¹

Although comparatively underexplored, coordination complexes that undergo a biochemically mediated change in paramagnetism offer a promising strategy to expand the dynamic range of activatable probes.^13–15 We, and others, have pursued biochemically activated MR contrast agents that utilize the Mn^3+/2+ couple as the activation mechanism.^16–20 The Mn^3+/2+ couple is physiologically tenable and can be tuned through ligand modifications.²¹ High spin Mn²⁺ is a potent relaxation agent, whereas Mn³⁺ is a much less effective relaxation agent.²²

Rational design of redox activated Mn-based contrast agents is challenging. Most ligand systems support a single oxidation state. Polyaminocarboxylate chelators like BPED, Chart 1, and EDTA bind Mn²⁺ with high affinity, Table S1,²³ but the redox potential of the corresponding Mn³⁺ complex is very high and the Mn³⁺ chelates decompose within seconds in aqueous media.²⁴ Mn³⁺ is stabilized by strongly electron releasing ligands like HBED, Chart 1, and TPPS, Fig S1. Coordination of Mn²⁺ by these ligands triggers spontaneous oxidation to Mn^3+.18,25 The HBET ligand, Fig S1, stabilizes both oxidation states but the Mn^3+/2+ redox potential is >0.50 V more positive than that of tissues and cells.^26,27 Although Mn³⁺-HBET can be prepared and is stable in aqueous solution, it does not persist in blood plasma, Figs S2–3.

Chart 1.

(A) Mn³⁺ and Mn²⁺ selective chelators, HBED and BPED, respectively, and the Janus chelate, JED, designed to support both Mn³⁺- and Mn²⁺. (B) Redox triggered isomerization of Mn³⁺- and Mn²⁺-JED.

We hypothesized that we could stabilize both oxidation states using a ligand that isomerizes between Mn³⁺ or Mn²⁺ selective chelators. Our prototype ligand to test this strategy, JED (“Janus HBED/BPED”), Chart 1, is designed to present a HBED-type donor set to Mn³⁺ and a BPED-type donor set to Mn²⁺. Upon oxidation or reduction, the opposite Janus face will capture the otherwise unstable oxidation state. By analogy with previously characterized Mn²⁺ complexes with acyclic hexadentate chelators, we anticipated the Mn²⁺ complex would form a 7-coordinate, ternary complex with a rapidly exchanging water co-ligand, a requisite to high relaxivity.^28,29 The smaller Mn³⁺ ion is expected to be 6-coordinate with no inner-sphere water ligand and low relaxivity, analogous to the related [Mn(EHPG)]⁻ complex, Fig S1.³⁰

Diasteromerically pure JED (R,R/S,S) was isolated following a 7 step synthesis, Scheme 1. Aldol reaction of phenol and 2-pyrdinecarboxaldyde, 1, followed by SeO₂ oxidation yielded the synthon for the Janus pyridyl-N/phenlato-O donor, 2. Double condensation of 2 with ethylenediamine yielded 3, which was reduced with NaBH₄ to a diastereomeric mixture of diamine 4. Preparative separation of the diastereomers of 4 proved difficult, but facile separation was achieved after Zn chelation, Figs S4–5. Zn was stripped from the diastereomerically pure 4 with excess DTPA at pH 5.0. Introduction of the acetate-O donors via reductive amination of (R,R/S,S) 4 with glyoxylic acid yielded JED. The Mn²⁺ and Mn³⁺ complexes were independently synthesized via reaction of JED with MnCl₂ or MnF₃, respectively.

Scheme 1. Synthesis of JED^a,b.

^a(i) MgCl₂, Me₂NEt, CH₂Cl₂, RT; (ii) SeO₂, dioxane, 100 °C; (iii) 0.5 mol. equiv. ethylenediamine, MeOH, RT; (iv) NaBH₄, (v) Zn(OTf)₂, 1:1 MeCN:H₂O, RT – diastereomers of Zn•4 separated by RP-HPLC; (vi) excess DTPA, pH 5.0; (vii) glyoxylic acid, NaBH₃CN, NaHCO₃, MeOH. ^bJED was prepared from the (R,R/S,S) diastereomer of 4; (S,S) 4 and JED are depicted here.

“BPED-type” binding of the Mn²⁺ was confirmed by UV-vis spectroscopy. Prior work with HBET and Mn²⁺-HBET complexes showed that phenol ionization is accompanied by a significant red-shift that can be used to track phenolato-O coordination.¹⁷ JED exhibits a similar red shift (268 to 308 nm) at pH > 10, Fig S9. The absorbance profile of Mn²⁺-JED shows no evidence of phenol ionization out to pH 8, Fig S10. The Mn³⁺ complex can only exist if the Janus switch to HBED-type binding occurs. HBED-type binding to Mn³⁺ is further evidenced by UV-vis spectroscopy. The Mn³⁺-JED LMCT transitions, Fig S12, reflect those of known Mn³⁺ complexes of ligands similar to HBED.³⁰

Cyclic voltammetry measurements performed on Mn-JED reveal irreversible oxidation and reduction events, Fig S11. From this data we estimate the oxidation and reduction potentials occur at 0.91 V and 0.01 V vs. NHE, respectively. The Mn-JED oxidation and reduction events mirror the Mn²⁺-BPED oxidation event, 1.11 V, and Mn³⁺-HBED reduction event, −0.17 V, Table S1.

The thermodynamic stability of Mn²⁺-JED was evaluated by monitoring the direct competition reaction with BPED (log K_{pH 7.4} = 11.2, Table S1) using HPLC (0.1 M KNO₃, RT) and yielded a log K_{pH 7.4} = 10.8±0.2 for Mn²⁺-JED. Mn²⁺-JED is among the most stable Mn²⁺ complexes reported, Table S1. Direct measurement of Mn³⁺ formation constants is challenging, given the fact that the Mn³⁺ aqua ion spontaneously disproportionates to Mn²⁺ and Mn⁴⁺ in water. Instead, the Mn³⁺-JED formation constant, log K_{pH 7.4} = 28.6, was estimated from the Mn²⁺ formation constant and redox potentials,³¹ Eq S3. A log K_{pH 7.4} = 29.4 is estimated for Mn³⁺-HBED. The congruence between the Mn²⁺- and Mn³⁺-JED formation constants with those of Mn²⁺-BPED and Mn³⁺-HBED, respectively, provides additional evidence of BPED- and HBED-type JED coordination.

The Mn²⁺ complex is a much stronger relaxation agent than its Mn³⁺ sister complex. Relaxivity values were recorded in water and human blood plasma at 1.4 T, 4.7 T and 11.7 T, 37 °C, Table 1. The Mn²⁺ vs. Mn³⁺ r₁ ratios are amongst the largest measured for any reported activatable Gd- or Mn-based relaxation agent, with a 9-fold change measured in human plasma at 1.4T. For Gd-based agents, the largest dynamic ranges reported are achieved via reactions that result in products that constrain Gd rotational dynamics.^11,12 Altering rotational dynamics provides a profound effect at field strengths ≤1.5 T, but is nearly obsolete by 3.0T.³² Mn-JED maintains a 3.3 to 5.0-fold r₁ change at 4.7T and 11.7T. The high relaxivity of the Mn²⁺ complex is consistent with a tertiary complex with a rapidly exchanging water co-ligand. The higher relaxivity of the Mn²⁺ complex is consistent with a tertiary complex with a rapidly exchanging water co-ligand, while the lower relaxivity of the Mn³⁺ complex is consistent with a coordinatively saturated complex that precludes water coordination. The increase in relaxivity observed in blood plasma at 1.4T suggests some degree of protein binding.

Table 1.

Relaxivity values, r₁, of Mn³⁺- and Mn²⁺-JED recorded in water or human plasma (in parenthesis) at 1.41T, 4.7T^a or 11.7T, 37 °C. Relaxivity of Mn²⁺-JED is much higher regardless of applied field strength.

	Mn3+	Mn2+	r1 Mn2+/ r1 Mn3+
1.4T	0.5±0.01 (0.9±0.01)	3.3±0.06 (8.0±0.39)	6.6 (8.9)
4.7Ta	0.9±0.02 (1.1±0.05)	4.3±0.32 (3.6±0.21)	4.8 (3.3)
11.7T	0.5±0.01 (0.5±0.02)	2.5±0.08 (1.9±0.08)	5.0 (4.8)

^a4.7T measurements performed at RT.

Independently isolated Mn³⁺- and Mn²⁺-JED persist in human blood plasma at 37 °C with little inter-conversion over the course of 24 h. Mn speciation was evaluated by HPLC interfaced to an ICP for Mn detection, Fig 1. Some inter-conversion between oxidation states was observed, but the reaction proceeds slowly. Comparison of the resultant 24 h HPLC traces indicates that equilibrium had not yet been achieved. Both complexes are very stable in plasma; >95% of Mn is present as JED complexes at 24 h.

Figure 1.

HPLC-ICP-MS traces showing Mn speciation of solutions of Mn²⁺-JED (top trace) and Mn³⁺-JED (bottom trace) incubated for 24 h, 37 °C, in human blood plasma. Mn³⁺- and Mn²⁺-JED elute at 8.0 and 10.9 min. Inter-conversion between the Mn³⁺- and Mn²⁺- complexes is slow, not reaching equilibrium within 24h.

Peroxidase enzymes amplify the reactivity of reactive oxygen species. High peroxidase activity is a salient feature of the acute inflammatory response and molecular imaging of peroxidase has been pursued in animal models of vasculitis, stroke, aneurism, and myocardial infarction.^33–36 Up to 250 U/mg (~250,000 U/mL) peroxidase activity has been measured in atherosclerotic lesions,³⁷ while here we show that Mn²⁺-JED to Mn³⁺-JED conversion is rapidly mediated by peroxidase activity 4 orders of magnitude lower than what is observed in vivo. Fig 2A shows HPLC traces recorded pre- and post- H₂O₂/ peroxidase incubation and confirms that conversion to Mn³⁺-JED occurs without byproducts. Fig 2B depicts rapid peroxidase mediated oxidation of Mn²⁺-JED, k_obs = 19.1±4.75 s⁻¹, in human blood plasma supplied with a steady state of H₂O₂ via the glucose/glucose oxidase reaction. Peroxidase mediated oxidation to Mn³⁺-JED occurs on the order of seconds while in blood plasma inter-conversion between oxidation states occurs on the order of days. H₂O₂ alone triggers negligible conversion, further underscoring a selectively for peroxidase mediated oxidation. The rate measured using r₁ change tracks with the rate measured by optical absorbance at 450 nm (ε = 1180 M⁻¹cm⁻¹ for Mn³⁺-JED), Figs S12–13.

Figure 2.

Biochemically mediated interconversion of Mn-JED oxidation states by peroxidase and thiols. (A) Conversion of Mn²⁺-JED (top trace, 7.6 min) to Mn³⁺-JED (bottom trace, 6.1 min) triggered by H₂O₂/ peroxidase (25 U/mL) in PBS buffer monitored by HPLC with 254 nm detection. The difference in peak height between the pre- and post-peroxidase treated samples results from the 2-fold higher extinction co-efficient of Mn³⁺-JED. (B) Oxidation of Mn²⁺-JED in the presence of H₂O₂ (open circles) or H₂O₂/peroxidase (15 U/mL, closed circles) in human blood plasma monitored by NMR. H₂O₂ was generated in situ by the glucose/ glucose oxidase reaction. (C) Reduction of Mn³⁺-JED in human blood plasma without (open circles) and with 5 mol. equiv. L-Cys (filled circles) added. Relaxometry measurements were performed at 1.4T, 37 °C.

Proliferating tumors are characterized by regions of hypoxia,³⁸ and thiol-rich microenvironments.^39,40 Mn³⁺-JED is readily reduced to Mn²⁺-JED in the presence of thiols. Fig 2C depicts the cysteine mediated reduction of Mn³⁺-JED by incubation with 5 mol. equiv. L-Cys in blood plasma with k_obs = 3.60±0.50 s⁻¹. L-Cys reduction results in a remarkable 8.5-fold increase in r₁.

The MRI contrast between equimolar solutions of Mn²⁺-JED and Mn³⁺-JED is profound. Fig 3A shows a conventional T₁-weighted image at 4.7 T of phantoms containing 0.5 mM Mn²⁺-JED before (−) and after (+) peroxidase mediated oxidation. In this image the higher relaxivity Mn²⁺-JED solution is much brighter than the low relaxivity Mn³⁺-JED solution. Gd-bis-5HT-DTPA is a state of the art peroxidase sensing Gd-based agent whose relaxivity change is mediated by a change in rotational correlation time upon oxidation, Fig S14.⁴¹ However at 4.7 T oxidation only results in a 15% r₁ change and this small difference is highlighted in the image in Fig 3B. By using an inversion pre-pulse we can generate positive contrast in the oxidized Mn³⁺-JED complex. Fig 3C shows the same phantoms as in 3A imaged with a 325 ms inversion pre-pulse. The inversion time was chosen to null signal in the untreated sample, generating large positive contrast in the oxidized sample. Fig 3D shows the same inversion pre-pulse sequence applied to the Gd-bis-5HT-DTPA samples from Fig 3B. Regardless of the scanning protocol, the contrast generated following oxidation of the Mn-based agent is much larger than that possible with the Gd-based agent. Fig 3E compares percentage r₁ change observed afer peroxidase oxidation of Mn-JED and Gd-bis-5HT-DTPA at 4.7 T. The Mn agent undergoes 380% r₁ change whereas the activatable Gd-based agents experiences <15% r₁ change. At 1.4 T, the r₁ change is also much greater for the Mn-JED system (790%) than for Gd-bis-5HT-DTPA (60%). Images of phantoms prepared in human blood plasma exhibit comparable contrast, Fig S15.

Figure 3.

MR phantom images at 4.7 T, RT, of water with 0.5 mM Mn²⁺-JED (A and C) or Gd-bis-5HT-DTPA (B and D) before (−) and after (+) incubation with H₂O₂/ peroxidase (45 U/mL). Images (A and C) acquired using a T₁-weighted FLASH sequence, or (B and D) with a 325 ms inversion pre-pulse to generate large positive contrast in the oxidized sample; see SI for image acquisition details. Note the high contrast between the samples containing Mn²⁺- vs. Mn³⁺-JED, but low contrast for the Gd-based system. (E) Percentage r₁ change after peroxidase mediated oxidation of the Mn- and Gd- based agents. A detailed description of the scanning parameters is in the SI.

To our knowledge, JED is the only chelator capable of stabilizing both Mn³⁺ and Mn²⁺ in the physiological milieu, and enables rapid and reversible biochemically mediated interconversion between Mn³⁺ (low r₁) and Mn²⁺ (high r₁). The biochemically mediated r₁ change observed with Mn-JED is the largest of any activatable Gd- or Mn-based contrast agent. Peroxidase mediated oxidation of Mn²⁺- to Mn³⁺-JED provides over an order of magnitude greater r₁ change than the state of the art Gd-based peroxidase sensor, and is achieved in minutes at peroxidase concentrations 1000-fold below what is seen in vivo. Mn-JED offers a promising new paradigm for activatable MRI contrast agent development.