Manganese(II)-Based Responsive Contrast Agent Detects Glucose-Stimulated Zinc Secretion from the Mouse Pancreas and Prostate by MRI

Abstract

A Mn(II)-based zinc-sensitive MRI contrast agent, MnPyC3A–BPEN, was prepared, characterized, and applied in imaging experiments to detect glucose-stimulated zinc secretion (GSZS) from the mouse pancreas and prostate in vivo. Thermodynamic and kinetic stability tests showed that MnPy-C3A–BPEN has superior kinetic inertness compared to GdDTPA, is less susceptible to transmetalation in the presence of excess Zn²⁺ ions, and less susceptible to transchelation by albumin. In comparison with other gadolinium-based zinc sensors bearing a single zinc binding moiety, MnPyC3A–BPEN appears to be a reliable alternative for imaging β-cell function in the pancreas and glucose-stimulated zinc secretion from the prostate.

Graphical Abstract

INTRODUCTION

Magnetic resonance imaging (MRI) has become arguably the most powerful imaging modality because of its outstanding spatial and temporal resolution, its versatility, and its ability to detect functional and molecular events in tissue.^1,2 Although MRI is less sensitive than PET, SPECT, and optical methods,^3,4 interest in developing new molecular probes that report on specific biological events continues to grow.⁵ The implementation of more advanced techniques such as CEST and MR fingerprinting highlights the versatility of magnetic resonance.^6–9 Despite these major physics advances, interest in newer types of exogenous molecular contrast agents (CAs) remains strong. To date, the most widely used MRI CAs have been the gadolinium-based T₁ agents.¹⁰ Although gadolinium-based CAs (GBCA) have been widely used since the introduction of Magnevist in 1988, the appearance of Nephrogenic Systemic Fibrosis (NSF) in 2006^11–13 and, more recently, reports of Gd³⁺ deposition in the brain^14,15 have raised concerns about continuing the use of GBCA.

These issues have been largely attributed to the poor kinetic inertness of Gd-complexes formed with acyclic ligands such as those in Gadodiamide and Gadoversetamide.^10,16,17 Nonetheless, the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) raised concerns for all forms of GBCA including those derived from macrocyclic ligands. From this history, the scientific community has learned two principles; first, there is a need to develop safe alternatives to acyclic GBCA and, second, the kinetic inertness and thermodynamic stability should be thoroughly investigated for every new metal-based agent developed for medical imaging purposes.

Manganese-based MRI agents are beginning to emerge as alternatives to gadolinium-based agents because of their favorable spin state, (S = 5/2 for most Mn²⁺ complexes), long longitudinal electronic relaxation times, and fast water exchange rates.¹⁸ Moreover, manganese is generally considered to be less toxic because it is an endogenous metal ion and is quickly cleared via hepatobiliary excretion.^19,20 However, like any metal ion complexes injected in relatively high doses, Mn-complexes also have limitations. For example, if Mn²⁺ dissociates from a chelating ligand, free Mn²⁺ can catalyze the formation of reactive oxygen species (ROS) and reactive nitrogen species (RNS),^21–23 and some Mn²⁺ complexes have been shown to mimic mitochondrial manganese superoxide dismutase (MnSOD). The first and only Mn²⁺ complex approved for human injection, Mn(DPDP)³⁻ (Teslascan), is no longer commercially available because of unfavorable side effects and subsequent lack of use.^24–26 Historically, Mn²⁺ was one of the first paramagnetic ions considered for use as T₁-based CA for MRI but insufficient ligand field stabilization provided by most ligands makes the development of suitable manganese complexes for medical imaging quite challenging.²⁷ However, a renewed interest in Mn²⁺ has led to the development of newer types of ligands for optimal chelation, some derived from macrocyclic ligands and others from acyclic ligands. Both types vary in (i) thermodynamic stability, (ii) kinetic inertness, (iii) number of inner-sphere water molecules (q), (iv) water exchange rates (k_ex), (v) binding interactions with plasma proteins, (vi) oxidation (Mn²⁺/Mn³⁺), and (vii) general versatility.^20,28–36 Among the most promising Mn²⁺ chelates reported so far are those bearing picolyl coordinating groups attached to either a macrocyclic or acyclic amine.^37,38 This is the case for Mn(N-picolyl-N,N′,N′-trans-1,2-cyclo-hexylenediaminetriacetate hydrate, [MnPyC3A·(H₂O)⁻], a recently reported complex having a r₁ relaxivity comparable to commercially available GBCA that also displays rapid hepatobiliary/renal clearance in vivo and low toxicity.²⁰ A peptide-conjugated version of this agent has also been used to target fibrin filaments in cardiac thrombus.^20,39

Acyclic chelates such as in MnPyC3A·(H₂O)⁻ undergo transmetalation when challenged with excess ZnCl₂ more easily than macrocyclic chelates but less easily when compared to linear GBCAs.²⁰ Transmetalation by Zn²⁺ is thought to be one of the main mechanisms for the release of Gd³⁺ from linear amine-based GBCA. Given the widespread interest in responsive MR CAs for the detection of local changes in freely available Zn²⁺ in the brain,^40,41 pancreas,^42–45 and prostate,^46,47 it is important to design zinc-sensitive agents in which Zn²⁺ does not displace the paramagnetic ion from the agent itself.^48–51 Our first zinc-sensitive agent, GdDOTA—diBPEN, was a macrocyclic 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) derivative with two bispyridine (BPEN)-extended side chains for zinc recognition (Figure 1a).⁴⁵ When exposed to Zn²⁺, the two BPEN moieties each bind a single Zn²⁺ ion and the resulting complex then forms a ternary complex with albumin. This protein interaction results in the reduced molecular rotation of the Gd³⁺ complex and a resulting increase in r₁ relaxivity and an increase in MR signal intensity in T₁-weighted images (Figure 1b). It was also shown that excess Zn²⁺ added to GdDOTA—diBPEN does not displace the Gd³⁺ ion from the macrocyclic ligand. Given this prior information, we hypothesized that by conjugating a zinc recognition unit such as BPEN onto MnPyC3A, one might actually protect the Mn center against transmetalation by zinc while retaining the zinc-responsiveness of the agent.

Figure 1.

Chemical structures and mechanisms for sensitive detection of Zn²⁺ in tissues. (a) GdDOTA–diBPEN, GdDO3A–BPEN, the parent MnPyC3A compound, and MnPyC3A–BPEN. (b) Mechanism of contrast enhancement involves binding of Zn²⁺ to the agent, followed by the agent-Zn²⁺ complex forming a ternary complex with albumin.

We report here a comprehensive chemical–biophysical study of MnPyC3A–BPEN and demonstrate its potential as a zinc-sensitive MRI CA. Our overarching goal was to create an alternative to GdDOTA–diBPEN^42,52 or GdDO3A–BPEN⁴⁵ for the in vivo detection of glucose stimulated zinc secretion (GSZS) in the prostate and the pancreas by MRI. In vivo imaging comparisons of MnPyC3A–BPEN with GdDO3A–BPEN, a derivative bearing a single zinc binding side chain (Figure 1a), show that MnPyC3A–BPEN may indeed be a viable alternative for functional imaging of zinc secretion in both the mouse pancreas and prostate.

RESULTS

Synthesis.

Encouraged by the promising report of MnPyC3A·(H₂O)⁻ as an imaging probe, we sought to use this same synthon as the basis of a new zinc-sensitive MRI agent. The synthetic route to this new derivative is outlined in Scheme 1. Compound 1 was prepared using reported protocols.²¹ We took advantage of the pyridyl moiety to provide an easy and achiral means to incorporate the zinc binding moiety. The 5-position in the pyridyl moiety was functionalized for this purpose. N′,N′-bis(pyridine-2-ylmethyl)ethane-1,2-diamine (BPEN) was coupled to 1 followed by the de-protection of 2 which yielded 3 in reasonable yields. The zinc sensor MnPyC3A–BPEN was prepared by stirring 3 with MnCl₂ at pH 6.5.

Scheme 1.

Preparation of MnPyC3A–BPEN (a) (7-Azabenzotriazol-1-yloxy)tripyrrolidinophosphonium Hexafluorophosphate (PyAOP), DMF, DIPEA, BPEN, (b) 3 M HCl, and (c) MnCl₂, pH 6.5

Relaxometry and Binding Characteristics.

The T₁ relaxation efficiency of paramagnetic agents such as this are typically compared by their longitudinal relaxivity r₁ values as defined by eq 1.

$$r_{1^p}=\frac{1}{[M]}\left(\frac{1}{T_{1^{\mathrm{obs}}}}-\frac{1}{T_{1^{\mathrm{dia}}}}\right)$$ (1)

For zinc-responsive agents like MnPyC3A–BPEN, one must consider the r₁ values of several species including the agent itself, the binary MnPyC3A–BPEN·Zn²⁺ complex, and the ternary MnPyC3A–BPEN·Zn²⁺•albumin complex. The r₁ values of MnPyC3A–BPEN ± Zn²⁺ and ±0.6 mM HSA are listed in Table 1 and compared with r₁ values previously reported for GdDO3A–BPEN. The data show that the r₁ of MnPyC3A–BPEN is slightly lower than GdDO3A–BPEN in the absence of Zn²⁺, increases only slightly in the presence of one equivalent of Zn²⁺ but increases by 4-fold in the presence of both Zn²⁺ and HSA. These results parallel the r₁ changes previously reported for GdDO3A–BPEN. Although the relaxivity data reported in Table 1 are measured at 0.5T, these values are magnetic field dependent, especially for those agents that bind to larger macromolecules. For comparison, the relaxivity values measured at 9.4T are also reported in Table S1. Here, the differences between the binary (5.0 mM⁻¹ s⁻¹) and ternary complexes (5.4 mM⁻¹ s⁻¹) are less dramatic but this small difference appears to be sufficient to detect release of Zn²⁺ in vivo (see below).

Table 1.

r₁ Relaxivities and K_D Values for Several Zinc-Sensitive MRI Agents in 100 mM Tris Buffer, 37 °C

	r10.5T (mM−1 s−1)
CA	No Zn2+	(+) Zn2+	(+) 0.6 mM HSA	(+) Zn2+ (+) 0.6 mM HSA	KD(Zn) (nM)	KD(HSA) (μM)		r1max (mM−1 s−1)	kex310 (×108 s−1)
MnPyC3A-BPEN	3.7 ± 0.1	4.1 ± 0.1a	10.8 ± 0.1b	17.4 ± 0.5b	93 ± 4c	28 ± 1d	66 ± 7e	21 ± 3d	0.7 ± 0.2
GdDO3A-BPEN	4.8 ± 0.1	5.1 ± 0.1a	5.8 ± 0.1b	17.8 ± 0.5b	118 ± 3c	36 ± 1d	48 ± 6e	20 ± 1d	0.029 ± 0.001
MnPyC3A-FBP*	8.5f		11.4f	13.5f		0.11i			1.0

^aCA/Zn = 1:1 (1 molar equivalent of Zn²⁺).

^bObtained by measuring T₁ values of four different samples of sensors (0.1, 0.2, 0.3, and 0.5 mM) in 0.6 mM human serum albumin (HSA) (±0.6 mM Zn²⁺). The r₁ values observed are only representative of these experimental conditions.

^cDetermined by competitive binding experiments with the Zn-sensitive fluorescence ligand, ZnAF–2F (see Supporting Information).

^dObtained by fitting proton relaxation enhancement data to equation 2 in Supporting Information.

^eObtained by one site competition fitting with dansylglycine.

^fB = 1.4 T.

^g4.5% w/v BSA.

^hFibrin gel.

ⁱFluorescein-labeled fibrin binding peptide titrations.²⁰

A titration of MnPyC3A–BPEN with Zn²⁺ showed that r₁ increases with incremental addition of Zn²⁺ ions until a 1:1 complex is formed then levels off with further addition of Zn²⁺. The binding affinity of the BPEN unit on MnPyC3A–BPEN with Zn²⁺ was determined by competitive binding experiments with the Zn-sensitive fluorescence ligand, ZnAF–2F.^42,45 These titrations yielded a dissociation constant of K_D = 93 ± 4 nM for Zn²⁺ binding with MnPyC3A–BPEN (Figure S1). These data show that the BPEN moiety retains a high affinity for Zn²⁺ when conjugated to MnPyC3A.

The number of inner-sphere water molecules (q) and the water exchange rate (k_ex) in MnPyC3A–BPEN were determined by simultaneous analysis of ¹⁷O reduced T₂ data.^53–56 These data, summarized in Table 2 and Table S2, indicate that MnPyC3A–BPEN has a single inner-sphere water coordination site with an exchange rate (k_ex = 0.7 ± 0.1 × 10⁸ s⁻¹) similar to that reported for MnPyC3A.²⁰

Table 2.

Number of Coordination Sites for Water Molecules (q), Enthalpy of Activation, Mn–¹⁷O Hyperfine Coupling Constants (A_O/ℏ), and Water Exchange Rates at T = 310 K (k_ex³¹⁰)

Cas	q calc	ΔH‡ [kJ/mol]	A/h [x107 rad/s]	kex310 (×108 s−1)
MnPyC3A-BPEN	0.9 ± 0.2	33 ± 0.50	2.8 ± 0.10	0.7 ± 0.1
MnPyC3A*	1	37.2	2.9	1.0

Albumin Binding Studies.

Albumin, the most abundant protein in plasma, plays an important role in the transport of drugs and the delivery of essential poorly soluble molecules to cells. Free Zn²⁺ ions also have a high affinity (29.5 nM) binding site on albumin,⁵⁷ an affinity about 3-fold stronger than the binding affinity between MnPyC3A–BPEN and Zn²⁺.^42,58 This means that albumin must play a key role in the formation of the ternary complex involving Zn²⁺.⁴⁵ At low-to-medium magnetic fields, r₁ is dominated by the rotational correlation time, τ_R, of the protein. This is clearly the case in our studies because r₁ relaxivity of MnPyC3A–BPEN is amplified upon the addition of both Zn²⁺ and 600 μM human serum albumin (HSA) (Table 1). In the presence of 600 μM HSA but no zinc, a modest increase in r₁ was observed. This demonstrates that MnPyC3A–BPEN alone, unlike GdDOTA–diBPEN or GdDO3A–BPEN, interacts weakly with HSA even in the absence of Zn²⁺ ions. An alternative explanation might be that some Mn²⁺ is released from the chelate (transchelation) and bound to a metal ion binding site on HSA. The r₁ data in general show that MnPyC3A–BPEN does respond to the presence of Zn²⁺ and HSA by showing an increase in r₁ similar to that reported for GdDO3A–BPEN.⁵²

The binding of MnPyC3A–BPEN·Zn²⁺ with HSA was evaluated using two different methods: (1) a proton relaxation enhancement (PRE) titration and fluorescence titrations using dansylglycine, a drug site 2 binding molecule (Figure 2a,b). The K_D values obtained from these experiments yielded comparable binding affinities (Table 1). In a complex mixture containing a Mn-based zinc sensor, HSA, and Zn²⁺ ions, several species are present in the solution. The combined data suggest that HSA heavily mediates the amount of MnL_x–Zn²⁺–HSA present in this mixture. This is due to the fact that HSA is normally present at a higher concentration in the plasma compared to Zn²⁺ (<20 μM) and also has a higher affinity for Zn²⁺ in comparison to BPEN-based sensors such as these.^42,57,59

Figure 2.

Relaxometry and binding characteristics of MnPyC3A-BPEN. (a) PRE titrations of MnPyC3A–BPEN (0.1 mM) as a function of increasing [HSA]. [Zn²⁺] was held constant (0.6 mM, equal to the highest concentration of HSA) in each titration. All measurements were performed at 20 MHz, 310 K in 100 mM Tris buffer at pH 7. (b) Competition binding curve for the determination of the MnPyC3A–BPEN·Zn²⁺ binding dissociation constants with HSA with dansylglycine (drug site 2) and warfarin (drug site 1). (c) Transmetalation studies. MnPyC3A–BPEN and GdDTPA were separately incubated with 25 mol excess Zn²⁺ at pH = 6 and the T₁ of water protons was measured using a mq60 relaxometer (B₀ = 1.5 T) over 210 min. The fitted lines reflect pseudo-first order rate constants for the dissociation of free Gd³⁺ from GdDTPA and free Mn²⁺ from MnPyC3A–BPEN. (d) Evolution of the relative water proton paramagnetic relaxation rate of 0.1 mM aqueous solutions of MnPyC3A–BPEN (■) or MnPyC3A (●) in the presence of 0.6 mM of HSA. The plots show changes in R₁ (at 23 MHz) for 1 mM samples of each agent over time, pH 7.2 in TRIS buffer, 310 K. The concentration labels reflect the calculated [Mn²⁺] transchelated from each complex to HSA.

Kinetic Inertness.

PyC3A forms a complex with Mn²⁺ with moderate thermodynamic stability (log K_MnL = 14.14, p_Mn = 98.17).²⁰ and one would predict that PyC3A would form even more stable complexes with Zn²⁺ and Cu²⁺ as predicted by Irving–Williams theory.⁶⁰ Hence, the kinetic inertness of MnPyC3A–BPEN is quite important if one intends to use this agent as a reporter of Zn²⁺ release from tissues. To test this, the complex was first challenged by the addition of 25-fold excess Zn²⁺ to MnPyC3A–BPEN at pH 6.0, 37 °C while monitoring changes in water proton r₁ (Figure 2c). The data show that MnPyC3A–BPEN is quite inert to transmetalation by Zn²⁺ in comparison to GdDTPA which dissociates very quickly. A fit of these data to a pseudo-first-order kinetic model showed that MnPyC3A–BPEN was about 2-fold more inert toward transmetalation by Zn²⁺ (k = 3.2 × 10⁻⁴ s⁻¹) compared to the parent compound, MnPyC3A (k = 6.7 × 10⁻⁴ s⁻¹). This suggests that the BPEN moiety provides some protection against transmetalation of Mn²⁺ by excess Zn²⁺ even though it binds only one equivalent of Zn²⁺ ions.

Similarly, relaxometric data on samples containing 0.1 mM MnPyC3A–BPEN or MnPyC3A plus 0.6 mM HSA show that the former complex was somewhat less susceptible to transchelation by albumin (Figure 2d). Using the relaxivity values in Table 1 and the reported relaxivity of Mn²⁺ bound to albumin (97 mM⁻¹ s⁻¹),⁶¹ the amount of Mn²⁺ transchelated from MnPyC3A to HSA was estimated at ~2% over 1 h and ~12% over 13 h. Similar experiments with MnPyC3A–BPEN showed that slightly less Mn²⁺ moves from the chelate to HSA over this same time period (~2% over 1 h and ~8% over 13 h). This effect was also observed by ¹⁷O NMR experiments which showed an increase in q = 2.5 ± 0.2 in the presence of 1 equiv of Zn²⁺ and excess HSA (Figure S2, Table S2). Data also suggest that the presence of zinc and HSA favorably impacts q and k_ex, and the different species present in the solution contribute to the overall observed r₁ enhancement. Thus, it appears that having a BPEN moiety attached to the chelate protects against both transmetalation by Zn²⁺ and transchelation by HSA. The exact mechanism of this protection is yet to be investigated.

In Vivo MRI.

Several mouse imaging experiments were performed to evaluate the potential use of MnPyC3A–BPEN for detecting Zn²⁺ secretion from tissues in vivo by MRI. As shown previously, the pancreas co-releases insulin and Zn²⁺ after the bolus injection of glucose and that the increase in Zn²⁺ in the extracellular space of β-cells can be detected by MRI using a Gd-based zinc sensor. To date, this is the only reported method for imaging β-cell function in vivo.^43,52 More recently, GSZS was also observed in the prostate of fasted mice by MRI.^46,62 Although the molecular mechanism of GSZS from healthy prostate cells remains to be fully elucidated, this response has been shown to be useful for distinguishing healthy versus malignant prostate cells.^46,62,63 Figure 3a shows typical in vivo T₁-weighted MRI images of mice before and after a bolus injection of MnPyC3A–BPEN plus glucose. Contrast enhancement was quite evident in the pancreas and prostate after a bolus of glucose and no significant increase in signal intensity was seen in these organs in control mice receiving saline instead of glucose (Figure S3). Furthermore, the administration of MnPyC3A plus glucose instead of MnPyC3A–BPEN plus glucose showed little to no contrast enhancement in either tissue (Figure S4). Together these experiments highlight the specific interactions between secreted Zn²⁺, plasma albumin, and MnPyC3A–BPEN leading to the formation of the ternary complex. Conversely, MnPyC3A interacts only weakly with plasma proteins.²⁰ The % signal intensity gain after the injection of each agent in ROIs in the pancreas, prostate, kidneys, and liver normalized to muscle are shown in Figure 3b. These data show that significantly higher signal enhancement is observed after the injection of MnPyC3A–BPEN versus MnPyC3A not only in those tissues known to release Zn²⁺ (pancreas and prostate) but also in the liver and kidneys. For comparison purposes, the area-under-the curve (AUC) over the first 16 min post CA injection (AUC^{0–16 min}) for each tissue are compared in Figure 3b. These results indicate that after glucose stimulation, MnPyC3A–BPEN induces a larger MR signal enhancement over the secretory period (0–16 min) in comparison to parent compound MnPyC3A. Figure 3b (inset) shows that the liver-to-kidney ratio for the two agents do not differ, consistent with equivalent excretion mechanisms for MnPyC3A–BPEN and MnPyC3A. Nevertheless, the observation that MnPyC3A–BPEN induces greater signal enhancement in all tissues compared to MnPyC3A delivered at the same dose indicates that MnPyC3A–BPEN circulates in all tissues as the higher relaxivity ternary MnPyC3A–BPEN·Zn²⁺·albumin species.

Figure 3.

Imaging Zn²⁺ secretion from tissues in vivo by MRI at 9.4 T. (a) T₁-weighted images (3D gradient echo TE/TR = 1.69/3.35 ms, averages = 4, θ = 20°) of fasted C57Bl6 male mice after receiving 0.07 mmol/kg MnPyC3A–BPEN i.v. and 2.2 mmol/kg glucose i.p. to stimulate the release of zinc from secretory organs. (b) Quantitative changes in signal intensity in each organ normalized to muscle after administration of 0.07 mmol/kg MnPyC3A–BPEN or non-zinc sensitive control MnPyC3A i.v. and 2.2 mmol/kg glucose i.p. (N = 3). Integrated normalized signal intensity profiles as AUC^{0–16 min} for each agent and each organ. Bars represent standard error of the mean; *p value < 0.05, **p value < 0.01.

Tissue Bio-Distribution.

Additional tissue biodistribution studies were performed in mice after the injection of MnPyC3A–BPEN. In these studies, either 0.07 mmol/kg MnPyC3A–BPEN, 0.04 mmol/kg MnCl₂, or saline were injected followed by an immediate injection of glucose. The kidney, brain, liver, heart, spleen, muscle, pancreas, and prostate were resected at either 15 or 90 min postinjection to monitor short-versus long-term accumulation and excretion. After tissue digestion, total Mn was measured by ICP–MS (Figure 4). In the MnCl₂ group, significant Mn was found in the kidney, liver, heart, and pancreas at 15 min postinjection but little at 90 min (Figure 4b), consistent with a previous report.²⁰ In the MnPyC3A–BPEN group, about 2- to 3-fold less Mn was found in any of these same tissues at 15 min postinjection, with the most found in the kidney and liver. At 90 min, the amount of Mn in the kidney and liver was reduced by ~60%. No significant Mn was found in the heart tissue in the MnPyC3A–BPEN group, consistent with stable chelation of Mn throughout. Given that the tissue biodistribution data showed comparable amounts of Mn in the kidney and liver after the injection of MnPyC3A–BPEN, this indicates that the excretion pathway is about 50% biliary and 50% renal. To evaluate this further, a separate cohort of mice were imaged serially after receiving 0.07 mmol/kg MnPyC3A–BPEN plus glucose. Figure 5a (left) shows coronal images of a mouse prior to and at 90 min postinjection. These images show once again that the agent is largely cleared from all tissues, including the kidney and liver, at 90 min but the gallbladder remained hyperintense consistent with hepatobiliary excretion of MnPyC3A–BPEN. Dynamic quantitative excretion information was obtained by analyzing the change in signal intensity in the respective excretory organs (Figure 5b). The signal change in the kidneys showed a maximum at 27 min postinjection, and only 25% of that signal was lost due to excretion after 90 min. On the other hand, the change in the liver signal postinjection also showed a maximum at 27 min, but at 90 min–89% of the signal was reduced because of the excretion of the compound and accumulation in the gallbladder. The gallbladder signal gradually increased and reached a plateau starting at 60 min and ending with a signal change of 242 ± 68% after 90 min. To evaluate the integrity of MnPyC3A–BPEN after excretion via the two pathways, we collected bile from the gallbladder or duodenum and urine from the bladder 90 min postinjection. The LC–MS elution profiles of samples of bile and urine collected 90 min postinjection are shown in Figure 5c. The gallbladder showed an intense MR signal in T₁-weighted scans, and the bile LC–MS trace shows a peak at an elution time of ~11 min where the mass coincides with that of intact MnPyC3A–BPEN. Similarly, the LC–MS profile of urine showed a peak at ~14 min consistent with intact MnPyC3A–BPEN. To validate that the peaks we were observing were indeed the intact compound, we measured the Mn concentration collected from urine and bile 90 min post i.v. injection of MnPyC3A–BPEN by inductively coupled plasma mass spectrometry. Naïve urine and bile (see Figure S5 for chromatogram) were then spiked with an authentic sample of the original compound. The LC–MS chromatograms of spiked urine and bile both are consistent with the intact compound seen in the samples collected from urine and bile illustrated in Figure 5c. These results suggest that MnPyC3A–BPEN is largely excreted via renal filtration and hepatobiliary pathways as the intact complex.

Figure 4.

Manganese content as measured by ICP–MS in organs of mice after receiving either (a) MnPyC3A–BPEN plus glucose or (b) MnCl₂ plus glucose. Animals were sacrificed 15 or 90 min postinjection (N = 3 each group). *p < 0.05.

Figure 5.

MnPyC3A–BPEN excretion profiles. (a) Coronal T₁-weighted MR images of a mouse pre- and 90 min postinjection of MnPyC3A–BPEN plus glucose. (b) MR signal intensity normalized to muscle for the kidney, liver, and gallbladder as a function of time (N = 4). (c) LC–MS chromatogram of urine and bile collected 90 min postinjection. (d) LC–MS chromatogram of urine and bile before and after spiking with MnPyC3A–BPEN at concentrations measured in urine and bile samples collected 90 min post i.v. injection.

DISCUSSION

In this study, the well-described zinc binding unit, BPEN, was attached to a previously reported stable Mn²⁺ complex, MnPyC3A,²⁰ and the resulting complex was evaluated as a responsive MR imaging agent for the detection of Zn²⁺ released from the tissue in vivo. This responsive agent, like prior Gd-based zinc-responsive agents, showed only a modest increase in r₁ relaxivity in the presence of Zn²⁺ or HSA alone, but when both were present, a ternary complex is formed and r₁ is significantly increased. Although the primary protein contributing to the increase in relaxivity in this work is albumin, it is important to denote that other noncovalent interactions with proteins found in the plasma may also contribute to the relaxivity increase. The goal of this study was (1) to create a Mn-based Zn²⁺-responsive MRI CA, (2) to demonstrate its utility in vivo for detection of Zn²⁺ secretion from tissues in live animals, (3) to evaluate the stability of MnPyC3A–BPEN against transmetalation by Zn²⁺, and (4) to determine the tissue biodistribution and excretion pathways of this new agent. The results show that MnPyC3A–BPEN not only detects Zn²⁺ secretion from the pancreas and prostate in mice by MRI but the BPEN unit also increases the kinetic inertness of the complex toward transmetalation by Zn²⁺ and transchelation by HSA. T₁-weighted MR images of live animals showed contrast enhancement in the pancreas and prostate only after the injection of glucose to stimulate Zn²⁺ secretion from these tissues, similar to the contrast observed previously with the most effective Gd-based Zn²⁺-responsive agents.^45,46,52 Moreover, the tissue-biodistribution and excretory characterization studies both indicate that MnPyC3A–BPEN is excreted intact via both renal filtration and hepatobiliary clearance pathways. The later clearance pathway may reflect a combination of its slightly more lipophilic character⁶⁴ plus its negative charge for transport into hepatocytes by a family of organic anion transporting proteins expressed on the sinusoidal membrane of hepatocytes.⁶⁵

Possible limitations include: (1) the consideration that when using Zn²⁺ as a biomarker for malignant transformations, tissues may exhibit aberrant pH or oxygenation levels potentially altering the binding to the compound and thus the ternary complex and (2) the use of only male animals in the imaging studies. We observed that in pH aberrant environments (pH 6 or 8) the binding mechanism to Zn²⁺ may be slightly affected (Table S1) and should be considered when imaging zinc content and secretion in cancer tissues. Although there could potentially be gender differences in the secretory capacity of the pancreas, we used only male mice here so that both the pancreas and prostate could be studied in the same imaging setting. Given current concerns about the use of Gd-based CAs in humans,¹⁴ hopefully, this study will help advance discoveries of other Mn-based Zn²⁺-responsive MRI agents for imaging glucose-stimulated zinc secretion in different organs so that one can image the pathological effects associated with dysregulation in Zn²⁺ homeostasis.

CONCLUSIONS

In summary, MnPyC3A–BPEN offers an alternative to similar Gd-based zinc-sensitive MRI CAs for the in vivo detection of GSZS from pancreatic β-cells and from the prostate by MRI. This new sensor also offers superior kinetic inertness toward Zn²⁺ transmetalation compared to other GBCAs based upon linear amine ligand platforms. Given that MnPyC3A appears to be moving toward clinical trials as an alternative to GBCA,⁶⁶ MnPyC3A–BPEN may also have a translational value for the early detection of prostate cancer and for monitoring β-cell function during the development of type 2 diabetes.

EXPERIMENTAL SECTION

Synthesis.

Refer to Scheme S1 for structures. To a stirred solution of 1 (0.592 g, 1 mmol), (7-azabenzotriazolyl-1-yloxy)trispyrrilodino phosphonium hexaflurophosphate (1.042 g, 2 mmol) and N,N-diisopropylethylamine (1.29 g.10 mmol) in 5 mL of anhydrous N,N-dimethylformamide was added N′,N′-bis(pyridine-2-ylmethyl)ethane-1,2-diamine (0.482 g, 2 mmol). The mixture was stirred at room temperature for 2 h; 100 mL of dichloromethane was added and washed with water (50 mL × 3) followed by brine (100 mL). The organic layer was dried over Na₂SO₄ and concentrated to a brown oil. The crude product was purified by flash chromatography (alumina, 5% MeOH in dichloromethane) to yield 0.490 (58.7%) g of 2 as a pale brown oil. ¹H NMR CDCl₃ 400 MHz δ: 9.15 (1H, s), 8.71 (3H, m), 8.12 (3H, t), 7.84 (2H, d), 7.60 (2H, m), 4.24 (CH₂NCH₂C, 8H, s), 3.11 (NCH₂, 10H, m), 2.99 (CH₂CHN, 4H), 1.79 (OCCH₃, 21H, m), 1.35 (NCHCH₂, 8H, s); ¹³C NMR CDCl₃ 100 MHz δ: 168.8, 168.0 (CONH), 159.4, 159.0, 152.0, 143.0, 142.9, 129.8, 125.9, 124.3, 123.3 (CH–Py), 81.7 (CCH₃), 62.6, 60.9, 55.4, 52.8 (NCH₂), 26.9,25.3 (CCH₃), 24.2, 23.6 (CH₂). ESIMS positive mode m/z 816.1 [M + H]⁺ calculated for M + H⁺ C₄₅H₆₆N₇O₇ m/z 816.5.

Compound 2 (0.204 g. 0.25 mmol) was stirred in 5 mL 3N HCl for 48 h. Acid was removed in vacuo and the residue lyophilized to yield compound 3 as an off white solid in quantitative yield. (0.16 g). ¹H NMR CDCl₃ 400 MHz δ: 9.01 (1H, s), 8.65 (3H, d), 8.40 (3H, t), 7.98 (2H, d), 7.87 (2H, m), 4.25 (8H, s), 3.55 (6H, m), 2.91 (2H, m), 2.22 (2H, m), 1.80 (2H, m), 1.38 (2H, m), 1.23 (2H, m); ¹³C NMR CDCl₃ 100 MHz δ: 169.8, 165.9, 159.4, 159.0, 152.4, 147.2, 141.3, 127.1, 126.3, 117.7, 114.8, 62.6, 55.3, 53.3, 37.4, 26.9, 24.2, 23.6. ESIMS positive mode m/z 648.0 [M + H]⁺ calculated for M + H⁺ C₃₃H₄₂N₇O₇ m/z 648.3.

Free ligand 3 (0.194 g 0.3 mmol) was dissolved in 5 mL of water and pH adjusted to 6.5, and MnCl₂.4H₂O (0.059 g. 0.3 mmol) was added and pH re-adjusted to 6.5. The complex formation was monitored using LC–MS. The mixture was purified on a RP–HPLC C18 column using 50 mM ammonium acetate buffer at pH 6.5 and acetonitrile containing 5% 50 mM ammonium acetate buffer at pH 6.5 as the mobile phase to get 0.165 g (78.5%) of Mn(3). Mn²⁺content: 6.6% by ICP–OES; ESIMS positive mode m/z 701.0 [M + 2H]⁺ calculated for M + 2H⁺ C₃₃H₄₀N₇O₇Mn m/z 701.3.

In Vivo MRI.

All animal experiments were carried out following UT Southwestern guidelines for animal handling provided by the institutional animal care and use committee. Fasted male C57Bl6 mice were imaged at 4.7 T(Figure S3) and 9.4 T(Figure S4) using Varian/Agilent scanners. Mice were anesthetized with 2–5% isofluorane/oxygen mixture and their body temperature was maintained at 37 °C using a heated airflow. Two ge3d T₁-weighted scans were obtained as a baseline (TE/TR = 1.69/3.35 ms, average = 4, θ = 20°, matrix 128 × 128 × 128). Mice then received 0.07 mmol/kg of either (1) MnPyC3A–BPEN plus 2.2 mmol/kg d-glucose i.p, (2) MnPyC3A×BPEN plus saline, and (3) MnPyC3A i.v. plus 2.2 mmol/kg d-glucose i.p. Following the administration of CAs, sequential 3d T₁-weighted scans were obtained for 30 (N = 3) or 90 min (N = 4) until clearance of the agent was evident. Using ImageJ, the organs of interest were identified and ROIs were measured and normalized against ROIs drawn and measured from the back muscle found in same slice and time point. The change in the MR signal intensity is reported as a percentage compared to pre-injection scans. The area under the curve was measured over a period of 1–16 min postcontrast administration using GraphPad Prism 7 software, statistical significance was evaluated using unpaired two-tailed t-tests to compare between agents, p-values < 0.05 were considered significant.

Biodistribution by ICP–MS and LC–MS.

Biodistribution studies were performed in mice fasted for at least 12 h by injecting 0.07 mmol/kg of Mn(PyC3A–BPEN), 0.04 mmol/kg MnCl₂, or saline i.v. followed by an immediate injection of 2.2 mmol/kg d-glucose i.p. The kidney, brain, liver, heart, spleen, muscle, pancreas, and prostate were resected 15 and 90 min postinjection to monitor long-term accumulation/excretion. The tissue was digested by dissolving in 2 mL of freshly prepared aqua regia (1:3 mixture of nitric acid and hydrochloric acid) and lysing for 24 h. The lysed tissue samples were heated at 120 °C till the aqua regia evaporated. The residual digested tissue was dissolved in 0.5N HCl by sonicating for 30 min. The samples were centrifuged at 4000g for 5 min to eliminate any residues. The resultant sample solutions (10 μL) were diluted up to 5 mL with 4% HNO₃ and analyzed by ICP–MS for Mn²⁺ ion concentration. Along with collecting tissue, urine and bile were collected by carefully extracting at least 20 μL of fluid from both organs using a 30 G needle and a 1 mL syringe. The fluids were immediately inserted into a LC–MS and the traces were obtained. Additionally, Mn concentration was obtained by ICP–MS. The bile and urine of animals receiving only saline i.v. and 2.2 mmol/Kg d-glucose i.p. were collected for LC–MS trace composition analysis of background fluid. Naïve urine and bile were spiked with Mn(PyC3A–BPEN) at the concentrations obtained from ICP–MS of the injected animals. These spiked fluids were then inserted into a LC–MS and the traces were obtained and compared.