Effect of Bridging Anions on the Structure and Stability of Phenoxide Bridged Zinc Dipicolylamine Coordination Complexes

Abstract

A series of four related phenol derivatives, with 2,2'-dipicolylamine substituents at the ortho positions, were prepared and their Zn²⁺ coordination complexes studied by spectroscopic methods. X-ray crystal diffraction analysis of a dinuclear zinc complex with two bridging acetate anions showed a ternary structure with highly charged interior and lipophilic exterior, which helps explain why this class of water-soluble complexes can effectively diffuse through cell membranes. The stability of the dinuclear zinc complexes in aqueous solution was found to be strongly anion dependent; that is, bridging oxyanions, such as acetate and pyrophosphate, lock the two Zn²⁺ cations to the surrounding ligand and greatly enhance ligand/zinc affinity. Overall, the results provide new insight into the structural and mechanistic factors that control the recognition and chemosensing performance of phenoxide bridged dipicolylamine molecular probes.

Introduction

Phenol derivatives with 2,2'-dipicolylamine (DPA) substituents at the ortho positions are known to readily form coordination complexes with Zn²⁺, Cu²⁺, Ni²⁺, Co³⁺, Fe²⁺, and Fe³⁺ cations, and quite a few X-ray crystal structures of dinuclear complexes have been described.¹ This report focuses on the Zn²⁺ complexes with generic molecular structure, BDPA•Zn₂ (Scheme 1). These compounds are part of a larger family of Zn²⁺-DPA complexes that have been developed for various supramolecular applications such as sensing of anionic biomolecules in aqueous solution, protein labeling, DNA targeting, RNA hydrolysis, biomembrane targeting, and cell penetration.^2,3,4 The long term goal of our research is to invent new classes of molecular probes for biological imaging, and we have discovered that Zn²⁺-DPA complexes have a remarkable ability to recognize anionic cell membrane surfaces selectively in complex biological environments such as cell culture and living animals.⁵ During these studies we observed an unusual phase transfer feature with water-soluble BDPA•Zn₂ complexes, namely, an ability to penetrate cell membranes.⁶ We exploited this finding by developing oligomers of MDPA•Zn as effective cell permeating peptides.⁷ However, non-selective cell penetration is not a desirable attribute for most bioimaging applications. In order to rationally improve the performance of BDPA•Zn₂ complexes as bioimaging probes and chemical sensors, a better understanding is needed of the fundamental coordination chemistry that controls the molecular recognition.

Scheme 1.

Here, we report a study of the solid-state structure and solution dynamics of some MDPA•Zn and BDPA•Zn₂ complexes using a combination of spectrometric methods. Specifically, we have examined the Zn²⁺ complexation properties of four ligands, L1–L4 (Scheme 1). We find that Zn²⁺ affinity of these ligands in water is strongly anion dependent, that is, Zn²⁺ affinity is greatly enhanced by the presence of bridging oxyanions, such as acetate (OAc⁻) and pyrophosphate (PPi⁴⁻). Furthermore, oxyanion bridging of the two Zn²⁺ cations in BDPA•Zn₂ produces a lipophilic ternary complex, which helps explain why water-soluble BDPA•Zn₂ complexes can effectively diffuse through cell membranes. The article concludes with a discussion of how this study can be used to improve the performance of DPA coordination complexes as molecular imaging agents and optical chemosensors for Zn²⁺ and oxyanions.

Results and Discussion

Synthesis and Solid-State Structures

The ligands, L1–L4, were prepared in straightforward fashion using established procedures that produced a Mannich reaction of the parent phenol with one or two molar equivalents of paraformaldehyde and DPA.^3c An X-ray crystal structure was obtained for the zinc acetate complex of L1. Single crystals were obtained by recrystallizing a methanol solution containing L1 mixed with two molar equivalents each of Zn(OAc)₂, triethylamine, and NaBF₄.^1a X-ray diffraction analysis revealed a molecular formula of Zn₂L1(OAc)₂(BF₄)•2MeOH. The C₂ symmetric structure (Figure 1) contains two, six-coordinate Zn²⁺ centers with distorted octahedral geometries and an N₃O₃ donor set. Each Zn²⁺ is coordinated by a DPA tertiary amine and two pyridyl nitrogen atoms, as well as the central phenoxy anion and two bridging OAc⁻ anions. The average Zn-N distances are 2.23 Å for the Zn-tertiary amine interactions and 2.17 Å for the Zn-pyridyl nitrogen interactions. The average Zn–O distances are 2.04 Å for the Zn-phenoxy bonds and 2.05 Å for the Zn-acetoxy bonds. The trans-axial angles through the distorted octahedral Zn atoms are all in the range of 162–171°, and the Zn1⋯Zn2 distance is 3.36 Å. These bond lengths and angles are consistent with similar Zn²⁺-DPA structures found in the literature.¹ The molecular structure has a highly charged interior and lipophilic exterior, and suggests a structural process that enables partitioning of water-soluble BDPA•Zn₂ complexes into vesicle and cell membranes.⁷ The process involves reversible association of BDPA•Zn₂ complexes with one or two fatty acids or phospholipids at the membrane surface to produce membrane-soluble coordination complexes that are analogous to the structure in Figure 1.

Figure 1.

X-ray crystal structure of Zn₂L1(OAc)₂(BF₄)•2MeOH. The structure omits the BF₄⁻ anion and two MeOH molecules for clarity. Thermal ellipsoids are indicated for all non-hydrogen atoms at the 50% probability level.

Solution-State UV Titrations

The two phenolic ligands, L3 and L4, have appended naphthalene chromophores and they are known to exhibit ratiometric spectral changes when they form zinc complexes.^3c Thus, L3 and L4 were chosen as sensing ligands for a study of the effects of bridging oxyanions on relative zinc complex stoichiometry and stability. Shown in Figure 2 are the results of an absorbance titration experiment that added Zn(NO₃)₂ to control mono-DPA ligand L4 in methanol:water (4:1). There is a large bathochromic shift in absorption maxima as the ligand forms the expected L4:Zn²⁺ salt structure (see MDPA•Zn in Scheme 1) with 1:1 stoichiometry. The next set of absorbance titration experiments added Zn(NO₃)₂ to bis-DPA ligand L3 in methanol:water (4:1) and yielded intriguing results. As shown in Figure 3 the stoichiometry at signal saturation depended on the identity of the counter anion. Addition of Zn(NO₃)₂ to a solution of L3 reached saturation when the L3:Zn²⁺ stoichiometry was 1:1 (Figure 3a). In contrast, the saturated L3:Zn²⁺ stoichiometry was clearly 1:2 when the titration was conducted in the presence of Na₄PPi (Figure 3b) or NaOAc (Figure 3c). The data is rationalized by considering the two stepwise, zinc association equilibria in Scheme 2. When the free L3 ligand (represented as generic ligand BDPA in Scheme 2) is titrated with Zn(NO₃)₂, the first apparent Zn²⁺ association constant, K₁, is larger than the second stepwise constant, K₂.⁸ Thus, the absorbance titration isotherm in Figure 3a reflects the first Zn²⁺ association step in Scheme 2; that is, the conversion of free BDPA ligand into mononuclear zinc complex BDPA•Zn.⁹ When high-affinity, bridging oxyanions, such as PPi⁴⁻ or OAc⁻, are present in the solution, the second Zn²⁺ association constant, K₂, is greatly enhanced such that it is larger than K₁ (Scheme 2). Thus, the absorbance changes in Figure 3b and 3c reflect the conversion of generic ligand BDPA directly into dinuclear zinc complex BDPA•Zn₂ with no measurable accumulation of the intermediate mononuclear zinc complex BDPA•Zn.

Figure 2.

(left) Absorbance spectra of L4 (20 μM) upon titration of Zn(NO₃)₂ in methanol/water (4:1 volume ratio); (right) absorbance at 306 nm.

Figure 3.

(a) Absorbance spectra of L3 (20 μM) upon titration of Zn(NO₃)₂ in methanol/water (4:1 volume ratio); (b) same titration repeated in the presence of Na₄PPi (40 μM); (c) same titration repeated in the presence of NaOAc (40 μM). In each case, the graph on the right shows absorbance intensity change for the spectral wavelength maxima band.

Scheme 2.

Stepwise association of BDPA with Zn²⁺ to form BDPA•Zn₂ is pushed to the right by the presence of bridging oxyanions, A⁻ = OAc⁻ or PPi⁴⁻.

A meaningful quantitative analysis of the above absorption titration curves was not possible, primarily for two reasons — the aqueous methanol solvent was not buffered (K₁ is pH dependent) and the ligand/Zn²⁺ association was too strong for the absorption titration method. However, semi-quantitative confirmation of the ligand/Zn²⁺ stabilization provided by the bridging oxyanions was gained by conducting competition experiments using ethylenediaminetetraacetic acid (EDTA) as a competitive Zn²⁺ binder. The top of Figure 4 shows the change in absorption spectra when a solution of L3:2•Zn(NO₃)₂ in water/methanol (4:1) was titrated with two molar equivalents of EDTA. The spectra clearly show that the EDTA stripped the two zinc cations and produced the free L3 ligand. In contrast, the set of spectra at the bottom of Figure 4 show that addition of two molar equivalents of Na₄PPi to the solution of L3:2•Zn(NO₃)₂ greatly stabilized the dinuclear L3•Zn₂ structure and prevented EDTA stripping of the zinc cations. Even the addition of one hundred molar equivalents of EDTA was unable to remove any measurable amount of Zn²⁺. The data show clearly that PPi⁴⁻, a strongly binding, bridging oxyanion, can stabilize the dinuclear L3•Zn₂ complex by several orders of magnitude compared to a weakly binding anion such as NO₃⁻.

Figure 4.

(top) Absorption spectra of L3:2•Zn(NO₃)₂ complex (20 μM) showing complete conversion to L3 upon addition of two molar equivalents of EDTA in water/methanol (4:1 volume ratio). (a) L3:2•Zn(NO₃)₂ with no EDTA, (b) L3:2•Zn(NO₃)₂ plus 0.5 equiv. EDTA, (c) L3:2•Zn(NO₃)₂ plus 1.0 equiv. EDTA, (d) L3:2•Zn(NO₃)₂ plus 2.0 equiv. EDTA, (e) L3. (bottom) Absorption spectra of L3:2•Zn(NO₃)₂ complex (20 μM) in the presence of Na₄PPi (40 μM) is unchanged by the addition of one hundred molar equivalents of EDTA in water/methanol (4:1 volume ratio). (a) L3:2•Zn(NO₃)₂ plus Na₄PPi, (b) L3:2•Zn(NO₃)₂ plus Na₄PPi and 100 equiv. EDTA, (c) L3.

Solution-State NMR Titrations

Additional structural evidence for the bridging anion stabilization effect was obtained by monitoring analogous Zn(NO₃)₂ titration experiments using ¹H NMR spectroscopy. Shown in Figure 5 are partial ¹H NMR spectra from the titration of mono-DPA ligand L2 with Zn(NO₃)₂ in CD₃OD:D₂O (4:1). As expected, there was a smooth transition from free L2 to L2:Zn complex with very strong affinity and 1:1 ligand/zinc stoichiometry. The two pyridyl rings in the DPA unit exhibit an equivalent set of four proton chemical shifts. Figures 6–8 show ¹H NMR spectra obtained during the titration of bis-DPA ligand L1 with Zn(NO₃)₂ in CD₃OD:D₂O (4:1) under three different conditions respectively: (i) no additional salt present, (ii) presence of NaOAc, and (iii) presence of Na₄PPi. The peak assignments on the spectra refer to the atom labels in Scheme 2 as elucidated by analyzing COSY spectra. The partial NMR spectra in Figure 6 show the titration of L1 with Zn(NO₃)₂ with no other salt present. The first species to appear is the asymmetric L1:Zn complex with 1:1 stoichiometry. The most diagnostic peaks are the two aryl peaks on the central phenoxy ring which are non-equivalent at 6.98 and 7.25 ppm (assigned as f and f' on the generic mononuclear BDPA·Zn structure in Scheme 2). These peaks disappear as more Zn(NO₃)₂ is added, and they are replaced by the symmetric dinuclear L1:2•Zn complex. However, the pyridyl proton peaks are very broad broad, indicating a fluxional structure. The only sharp NMR peak is at 6.86 ppm and corresponds to the two equivalent protons on the central phenoxy ring (labeled as k on the generic BDPA·Zn₂ structure in Scheme 2). The sodium salts of several anions were investigated to see if they influenced the zinc association equilibria. The presence of NaCl had no effect, with the titration spectra clearly showing stepwise production of the mononuclear species followed by the dinuclear complex (data not shown). In contrast, the presence of NaOAc promoted formation of the dinuclear zinc species with no evidence for any intermediate mononuclear structure. The spectra in Figure 7 show direct conversion of free L1 into the dinuclear L1:2•Zn complex with eight, broadened inequivalent pyridyl proton signals. This spectral feature is consistent with the X-ray structure in Figure 1, which is a C₂ symmetric bis-acetoxy bridged complex with one zinc-coordinated pyridyl ring in pseudo axial position and the partner pyridyl ring in pseudo equatorial position. The broadness of the pyridyl peaks indicates that the pyridyl rings undergo exchange of pseudo axial/pseudo equatorial Zn²⁺-coordination positions. In Figure 8 is the titration of L1 with Zn(NO₃)₂ in the presence of Na₄PPi. Again, the spectra show direct conversion of free L1 into the dinuclear L1:2•Zn complex. In this case, the eight inequivalent pyridyl peaks of the dinuclear L1:2•Zn complex are quite sharp indicating that the PPi strongly bridges the two zinc cations and produces a rigid, C₂ symmetric structure. Indeed, it is extremely likely that the structure is analogous to that shown in Figure 1 with a single tetradentate PPi⁴⁻ replacing the two bidentate OAc⁻ anions.^3b,3d The increased rigidity of the PPi⁴⁻ bridged structure is due to the higher anionic charge and stronger tetradentate zinc-coordination by the PPi⁴⁻ which prevents the zinc-oxygen bond dissociation events that allow the pyridyl rings to exchange between pseudo axial and pseudo equatorial coordination positions.

Figure 5.

Partial ¹H NMR spectra of L2 (10 mM) with increasing amounts of Zn(NO₃)₂ in CD₃OD/D₂O (4:1 volume ratio). The asterisk indicates a solvent peak.

Figure 6.

Partial ¹H NMR spectra of L1 (10 mM) with increasing amounts Zn(NO₃)₂ in CD₃OD/D₂O (4:1 volume ratio).

Figure 8.

Partial ¹H NMR spectra of L1 (10 mM) in the presence of Na₄PPi (20 mM) with increasing amounts Zn(NO₃)₂ in CD₃OD/D₂O (4:1 volume ratio).

Figure 7.

Partial ¹H NMR spectra of L1 (10 mM) in the presence of NaOAc (20 mM) with increasing amounts Zn(NO₃)₂ in CD₃OD/D₂O (4:1 volume ratio).

Conclusions

The crystallographic and spectroscopic results demonstrate that strongly bridging oxyanions such as OAc⁻ and PPi⁴⁻ associate with water soluble, dinuclear BDPA·Zn₂ complexes in aqueous solution and form relatively rigid, C₂ symmetric coordination structures with octahedral geometries around each Zn²⁺ center (Figure 1).^1,3b,3d It is reasonable to conclude that if the bridging anions were fatty acids or phospholipids then the resulting coordination complexes would be lipophilic enough to partition into a biomembrane. The relatively high lipophilicity of DPA ligands has been reported by nuclear imaging researchers to produce poor pharmacokinetics and they have mitigated this problem by developing DPA analogues with increased hydrophilicity.¹⁰ It should be possible to refine these hydrophilic molecular designs and produce next-generation, membrane impermeable dinuclear zinc bis-DPA complexes with improved cell recognition and bioimaging performance.

The solution-state absorption and NMR titration data show clearly that bridging oxyanions lock the two Zn²⁺ cations to the surrounding ligand and greatly enhance ligand/zinc affinity (Scheme 2). Thus, at sufficiently low zinc concentrations the molecular recognition process can be viewed as a three-component assembly of mononuclear zinc-ligand complex, second Zn²⁺ cation, and bridging oxyanion. This stepwise process can be employed to create optical anion sensor designs that use the anion-mediated association of the second zinc cation to induce a change in ligand optical properties.^2c,5b At higher zinc concentrations the bis-DPA ligand is saturated with both Zn²⁺ cations (i.e., it exists entirely as a dinuclear BDPA•Zn₂ complex) but the two sets of zinc-coordinated pyridyl rings are in a fluxional state. Association of a bridging oxyanion rigidifies the zinc-coordinated pyridyl rings, a process that may decrease non-radiative decay of excited state energy and be the basis of a “turn on” fluorescent sensor design.^3c,11 In addition to anion sensing, many DPA-containing receptors have been investigated as Zn²⁺ cation sensors and while it has been reported that the response of a bis-DPA Zn²⁺ sensor can be anion dependent,¹² the impact of this anion dependence on analytical performance for this class of zinc sensors has not been widely discussed. Finally, it is worth noting that the three-component molecular assembly describe here mimics a common biological strategy for targeting the surface of biomembranes.¹³ Metal cations often act as bridging cofactors that enable membrane association of peripheral membrane proteins (e.g., membrane targeting by annexin V mediated by Ca²⁺ ions).¹⁴

Experimental

Synthesis

The known ligands L3 and L4 were prepared by a literature method,^3c (an improved procedure is described in reference 11) and ligands L1 and L2 were prepared by the following related procedure.

L1: A solution of 4-bromophenol (0.88 g), 1 M HCl (0.5 mL), paraformaldehyde (0.35 g) and 2,2'-dipicolylamine (2.1 g) in ethanol (20 mL) and water (40 mL) was refluxed for 36 hours, then cooled to room temperature and neutralized with Na₂CO₃. The neutral suspension was extracted into chloroform, dried with sodium sulfate, and the solvent was evaporated to yield a yellow oil that was purified on a silica gel column using an eluent of 97:3 CHCl₃: MeOH give to give L1 as a pale yellow oil (2.5 g), 83% yield. ¹H NMR (300 MHz, CDCl₃, ppm): δ 8.53 (m, 4H), 7.61 (td, 4H, J₁=7.8 Hz, J₂=1.8 Hz), 7.44 (d, 4H, J= 8.1 Hz), 7.33 (s, 2H), 7.14 (m, 4H), 3.87 (s, 8H), 3.77 (s, 4H). ¹³C NMR (75 MHz, CDCl_3, ppm): δ 158.9, 155.1, 148.8, 136.6, 131.4, 126.5, 122.8, 122.0, 110.2, 59.7, 54.2. MS (FAB⁺): HRMS C₃₂H₃₂BrN₆O calcd 595.1821, found 595.1832.

L2: The above procedure was used with the following stoichiometry: 4-bromophenol (0.88 g), 1 M HCl (0.5 mL), paraformaldehyde (0.35 g), 2,2'-dipicolylamine (1.0 g). The same work up and purification gave L2 as pale yellow oil (1.7 g), 87 % yield. ¹H NMR (300 MHz, CDCl₃, ppm): δ 8.55 (m, 2H), 7.60 (td, 2H, J₁=7.8 Hz, J₂=1.8 Hz), 7.31 (d, 2H, J=7.8 Hz), 7.22 (dd, 1H, J₁=8.7 Hz, J₂=2.4 Hz), 7.14 (m, 3H), 6.80 (d, 1H, t=8.7 Hz), 3.87 (s, 4H), 3.74 (s, 2H). ¹³C NMR (75 MHz, CDCl_3, ppm): δ 157.8, 156.7, 148.6, 136.7, 132.4, 131.5, 125.0, 122.9, 122.1, 118.3, 110.1, 58.6, 56.1. MS (FAB⁺): HRMS C₁₉H₁₉BrN₃O calcd 384.0711, found 384.0699.

X-ray Crystallography

L1 Dinuclear Zinc Acetate Complex

A solution of L1 ligand and two molar equivalents of triethylamine, Zn(OAc)₂, and NaBF₄ in methanol was refluxed for 30 minutes and then allowed to cool to room temperature.^4a Colorless crystals were formed after 24 h and they were found to be suitable for X-ray diffraction analysis. C₃₈H₄₄BBrF₄N₆O₇Zn₂, M_r=994.25, triclinic, unit cell dimensions a=10.2479(5), b=12.5046(6), c=17.2421(10) Å, α=69.363(3), ß=81.066(3), γ=86.542(3)°, V=2042.60(18) ų, T=100(2) K, $\mathrm{P}\stackrel{‒}{1}$, Z=2. The structure was refined on F₂ to wR₂=0.0809, conventional R₁=0.0310 [11493 reflections with I >2σ(I)], and a goodness of fit=1.055 for 10254 refined parameters. The asymmetric unit contains the Zn₂L1 complex coordinated by two OAc anions, one BF₄ anion, and two molecules of CH₃OH. One of the CH₃OH molecules is disordered over two sites, and site occupancy for the principal component is 0.872(70).

Absorption Titrations

Appropriate aliquots of Zn(NO₃)₂ solution (20 mM in 4:1 CH₃OH:H₂O volume ratio) were added to 3.0 mL solutions of ligand L3 or L4 in the presence or absence of NaCl, NaOAc, or Na₄PPi in the same solvent and the absorption spectrum acquired at 25 °C.

¹H NMR Titrations

Appropriate aliquots of Zn(NO₃)₂ solution (100 mM in 4:1 CH₃OD:D₂O volume ratio) were added to 0.75 mL solutions of ligand L1 or L2 in the presence or absence of NaCl, NaOAc, or Na₄PPi in the same solvent and the ¹H NMR spectrum acquired at 25 °C.