Reusable HNO Sensors Derived from Cu Cyclam: A DFT Study on the Mechanistic Origin of High Reactivity and Favorable Conformation Changes and Potential Improvements

Abstract

Nitroxyl (HNO) exhibits unique favorable properties in regulating biological and pharmacological activities. However, currently, there is only one Cu-based HNO sensor that can be recycled for reusable detection, which is a Cu cyclam derivative with a mixed thia/aza ligand. To elucidate the missing mechanistic origin of its high HNO reactivity and subsequent favorable conformation change toward a stable Cu^I product that is critical to be oxidized back by the physiological O₂ level for HNO detection again, a density functional theory (DFT) computational study was performed. It not only reproduced experimental structural and reaction properties but also, more importantly, revealed an unknown role of the coordination atom in high reactivity. Its conformation change mechanism was found to not follow the previously proposed one but involve a novel favorable rotation pathway. Several newly designed complexes incorporating beneficial effects of coordination atoms and substituents to further enhance HNO reactivity while maintaining or even improving favorable conformation changes for reusable HNO detection were computationally validated. These novel results will facilitate the future development of reusable HNO sensors for true spatiotemporal resolution and repeated detection.

Short abstract

HNO possesses significant biomedical functions. Unknown roles of coordination atoms and substituents on excellent reusable HNO sensors were discovered with several new designs to aid the future development.

Introduction

Nitric oxide (NO) is a well-known signaling molecule in many physiological processes.¹ Among biological nitrogen oxide species, nitroxyl (HNO), the one-electron-reduced derivative of NO, has gained much attention for its diverse physiological and pharmacological effects.²⁻⁵ For instance, HNO is more effective than NO in protecting against ischemia-reperfusion injury.⁶ Another remarkable activity is its potent inhibition of platelet aggregation in patients with sickle cell disease.^7,8 HNO was also used to prevent alcoholism by inhibiting aldehyde dehydrogenase⁹ and to treat cancer through reactions with thiols and thiol proteins.¹⁰ The extensive therapeutic effects of HNO ignite considerable scientific interest in studying its chemical and biological profiles.

However, experimental HNO investigation is difficult due to its reactive nature. It is highly desirable to develop efficient detection methods for recurrent in vivo use to better understand and examine its physiological and pharmacological functions. Conventional HNO detection methods use analytic instruments to quantify products of HNO dimerization/dehydration or reactions with thiols, phosphine, and heme proteins.¹¹⁻¹⁷ However, most of them are indirect, time-consuming, and relatively unreliable. To overcome the limitations, direct HNO detection methods based on mass spectrometry and electrochemistry were developed.¹⁸⁻²¹ For detections inside living cells, invivo HNO sensors are accessible with bioimaging functions using metal complexes,²²⁻³⁴ phosphines,³⁴⁻⁵² thiols,^53,54 or esters.⁵⁵

Despite these developments, there are no reusable fluorescent sensors for biological HNO. Recently, a Cu complex with a mixed thia/aza cyclam derivative 14-N₂S₂ (Scheme 1) was reported as the first Cu-based reusable HNO sensor.⁵⁶ Reusability is a strongly desired feature in HNO sensor development since only reusable sensors can provide true spatiotemporal resolution⁵⁷ and repeated detection⁵⁸ of sudden bursts of in vivo analyte production. It is also cost-effective compared to one-time sensors. Using soft sulfur atoms is intended to stabilize the reduced Cu center upon reaction with HNO to prevent demetalation in prior one-time-use Cu-based HNO sensors.⁵⁶ The stable Cu^I complex can then be oxidized back to the starting Cu^II system by the physiological level of O₂ for the next HNO detection. The 14-N₂S₂ ligand also offers excellent HNO reactivity compared to the inactive aza-only cyclam (14-N₄, Scheme 1).

Scheme 1. (A) Schematic Pathways of HNO Reactions with Cu Macrocyclic Complexes Studied in This Work (a Suffix of N, S, Se, Ts, or Bn Is Added, Respectively, for a Species of a Specific Ligand Listed Above; Coordination Atom Numberings Are Shown in I-1) and (B) trans-I, trans-III, and Tetrahedral Conformations of [Cu(14-N₂S₂)] (Bn = Benzyl; Ts = Tosyl; Atom Color Scheme (H Omitted for Clarity): C, Green; N, Blue; S, Yellow; Cu, Brown).

However, the mechanistic origins of its greater HNO reactivity and favorable conformation changes to retain Cu are unknown. So, we report a density functional theory (DFT) study of these missing mechanistic details, which were based on the previously found proton-coupled electron transfer (PCET) mechanism and some favorable substituent effects in other metal complexes and metalloproteins.⁵⁹⁻⁶¹

To help the future development of reusable HNO sensors, we examined new ligand designs (Scheme 1): 14-N₂Se₂ with modifications on coordination atoms and substituted ligands with both an electron-donating group (EDG) of benzyl (14-N₂S₂-Bn) and an electron-withdrawing group (EWG) of tosyl (14-N₂S₂-Ts). Bn/Ts groups were installed on similar Cu cyclams and thus are feasible for future experimental synthesis.^30,56 These new studies provide interesting aspects to further enhance HNO reactivities and/or conformation change efficiencies to facilitate the future development of potential reusable fluorescent sensors for biological HNO.

Computational Details

Geometry optimization and subsequent frequency calculations were performed using the mPW1PW91⁶² method with the LanL2DZ⁶³⁻⁶⁵ basis set for the Cu, the 6-311++G(2d,2p) basis set for HNO and Cu-coordinated N/S/Se atoms in the ligand, and the 6-31G(d) basis set for the rest of the atoms. The experimental aqueous solution effect was included using the PCM approach.⁶⁶ All calculations were performed using Gaussian 09⁶⁷ and Gaussian 16.⁶⁸ In each case, the frequency analysis was used to verify the nature of the stationary points on respective potential energy surfaces and to provide zero-point energy-corrected electronic energies (E_ZPE’s), enthalpies (H’s), and Gibbs free energies (G′s) at 298.15 K and 1 atm. This approach was selected based on a methodological and basis set study⁶⁹ of several DFT methods, including the more recently (compared to mPW1PW91) developed M06⁷⁰ and dispersion-corrected ωB97XD,⁷¹ as well as comparisons with all-electron basis sets 6-31+G(2d) and 6-311++G(2d,2p), on HNO reactions. This approach has also been recently shown to give accurate descriptions of HNO reactivities in diverse metalloproteins and model systems including quantitative errors well below 1 kcal/mol as compared to experimental reaction barriers.^59,60,72,73 The atomic charges and spin densities are from the natural population analysis (NPA) and Mulliken schemes implemented in Gaussian 09/16. Intrinsic reaction coordinate calculations in Gaussian 09/16 were performed to prove the connection between each transition state and its corresponding reactant and product.

To further evaluate the above method of choice, additional methodological studies of both methods and basis sets were conducted. Among several common basis set schemes for the first-row transition metal Cu, namely, 6-311++G(2d,2p), triple-ζ Def2TZVP,^74,75 and Wachters’ basis sets,⁷⁶ the all-electron 6-311++G(2d,2p) basis set has the greatest number of basis functions, so it was used for comparative analysis with the LanL2DZ basis set below. The latest hybrid DFT method MN15⁷⁷ in Gaussian 16 has the best combined performance compared to 12 selected functionals including the previously studied M06⁷⁰ and ωB97XD.^71,77 Therefore, we compared MN15 with mPW1PW91⁶² using 6-311++G(2d,2p) and LanL2DZ⁶³⁻⁶⁵ basis sets for Cu and the same basis set for other atoms as described above. Geometric comparison of optimized Cu(II) and Cu(I) mixed thia/aza complexes with the X-ray structures in aqueous solution showed no significant differences between the three computational approaches: mPW1PW91/LanL2DZ, MN15/LanL2DZ, and MN15/6-311++G(2d,2p). Compared to the mPW1PW91/LanL2DZ data, results from MN15/6-311++G(2d,2p) only have a slightly better average mean percentage deviation (MPD) of 0.4% and average mean absolute deviation (MAD) of 0.01 Å, while MN15/LanL2DZ results have a mildly worse MPD of 0.8% and MAD of 0.02 Å; see Table S2. The computed trends of the most favorable conformations for Cu(II) and Cu(I) mixed thia/aza complexes also exhibited identical results to experimental data for all studied methods (Table S3). Regarding reaction barriers of the HNO to NO conversion pathways for the mixed thia/aza and pure aza complexes [Cu(14-N₂S₂)] and [Cu(14-N₄)], compared to those from using the mPW1PW91/LanL2DZ method, ΔG data from MN15/LanL2DZ and MN15/6-311++G(2d,2p) are only with MADs of 0.44 and 0.67 kcal/mol; see Table S11. Additional results in Tables S12–S14 show that compared to the current method of mPW1PW91/LanL2DZ, the combination of MN15 with the largest basis studied here, i.e., 6-311++G(2d,2p), has average MADs of only ∼0.03 Å, ∼0.02 e, and ∼0.02 e, for bond lengths, atomic charges, and spin densities respectively, which are basically within computational errors. These methodological results show that the current method is more efficient with similar accuracy to the more advanced and time-consuming method, so the following discussion is based on data from the current approach.

Results and Discussion

Favorable Conformations of Cu^II and Cu^I Complexes

Due to the importance of conformation changes here, we first studied the conformation effects of experimental Cu^II/Cu^I complexes. The used computational method enabled accurate descriptions of HNO reactivities in various metalloproteins and model systems^{59−61,69,73,78} and accurate structural predictions of related Cu cyclams.⁵⁹ The optimized gas phase structures of [Cu^II(14-N₂S₂)]²⁺ and [Cu^I(14-N₂S₂)]⁺ showed excellent agreement with the crystal structures, with an average mean absolute deviation (MAD) of 0.045 Å for all coordination bond lengths and mean percentage deviation (MPD) of only 2.1% (Table S1). As experimental reactions occurred in aqueous solution, optimizations using water as solvent were done, which shows similar structures with a better MAD of 0.040 Å and MPD of 1.8% (Table S2). These data show excellent prediction accuracy of the current computational method.

Because Cu^II cyclams have two common square planar conformations,⁷⁹⁻⁸⁴trans-III and trans-I (Scheme 1B, with coordination atom’s angles of C–N–C or C–X–C featuring two pointing above/below the plane and all four pointing above the plane; here, C represents neighboring carbons), and the Cu^I complex of this reusable HNO sensor has a tetrahedral conformation, these three conformations were studied for [Cu^II(14-N₂S₂)]²⁺ and [Cu^I(14-N₂S₂)]⁺ as well as [Cu^II(14-N₄)]²⁺ and [Cu^I(14-N₄)]⁺. As shown in Table S3, the experimental trans-III form of [Cu^II(14-N₂S₂)]²⁺ and the tetrahedral conformation of [Cu^I(14-N₂S₂)]²⁺(56) were all found to be most stable, which further validates our calculations. In contrast, for the aza-only complexes, trans-III conformations are preferred for both oxidized and reduced Cu states, with the tetrahedral conformations being of >10 kcal/mol higher energies. Clearly, these results support the experimental use of sulfurs to selectively stabilize the tetrahedral conformation of [Cu^I(14-N₂S₂)]²⁺, critical to prevent demetalation and thus enable recycling it back for reuse.

Cu reduction results in coordination bond length increase; see Table S4. The average Cu bond length increases are 0.119 and 0.118 Å in the trans-III conformation, respectively, for the 14-N₂S₂ and 14-N₄ ligands, which are basically the same. However, the use of a tetrahedral conformation for the reduced complex minimizes structural changes since the average Cu coordination bond length variations are now significantly decreased to 0.024 and 0.063 Å, respectively, for the mixed and aza-only systems. The latter system has 2.6-fold larger geometric changes, indicating that the mixed thia/aza ligand dramatically decreases structural change and its associated energy cost. Larger sulfurs in 14-N₂S₂ vs nitrogens leads to an enlarged coordination sphere with an average Cu coordination bond length of 2.198 vs 2.039 Å in 14-N₄, which leaves a larger space to accommodate the Cu size increase after reduction to an average Cu bond length of 2.222 Å (only 0.024 Å expansion for 14-N₂S₂). These data provide the first theoretical insights into the preferential stabilization of the tetrahedral conformation for [Cu^I(14-N₂S₂)]⁺.

HNO Reactivity

We then studied HNO reaction pathways for Cu complexes with the aza-only cyclam and its mixed thia/aza derivative to understand their significant experimental reactivity difference. For the starting Cu^II complexes with both ligands, the trans-III conformations are dominant with ∼98–99% based on their energy differences from other conformers (see Table S3), which was used in their reaction pathway calculations.

As shown in Scheme 1A, the basic pathway is like those with other Cu complexes and metalloproteins:^59−61,69 it starts with the HNO-bound intermediate I-1-S (its formation is experimentally facilitated by large excess of the HNO donor⁵⁶), which undergoes a PCET transition state (TS-S) to generate another intermediate I-2-S, where a proton transfers from HNO to one coordinated nitrogen and an electron from HNO transmits to Cu. It can be seen by the spin density data of the NO fragment (ρ_NO^αβ) for the 14-N₂S₂ systems increasing from ∼0 e in I-1-S to 0.122 e in TS-S and then to ∼1.0 e in I-2-S and accordingly Cu (ρ_Cu^αβ) decreasing from 0.395 e in I-1-S to 0.381 e in TS-S and then to ∼0 e in I-2-S for the electron transfer part. The proton transfer process is illustrated by the H–N₃ distance shortening from 3.661 Å in I-1-S to 2.027 Å in TS-S and then to 1.020 Å in I-2-S; see Table 1. The TS-S structure is shown in Figure 1B, and the structures of all species are shown in Figure S5. Another significant change is the Cu–N₃ bond cleavage due to proton transfer to N₃; see the Cu–N₃ bond lengths of 2.075, 3.153, and 3.416 Å in I-1-S, TS-S, and I-2-S, respectively. So, its elongation of ∼1 Å in TS-S is much more significant than the 0.001 Å elongation of the H–N bond (Table S23, also see key changes in Figure 1B). This suggests that breaking the Cu–N₃ bond is the most energy-demanding part. Then, NO dissociates to yield the reduced Cu complex I-3-S. This reaction’s Gibbs free energy of activation (ΔG^‡) of ∼21 kcal/mol is close to some room-temperature reactions,⁸⁵⁻⁸⁷ and the final reaction energy from I-1 to I-3+NO (ΔG°) of −18.53 kcal/mol is very negative, which shows a kinetically feasible and thermodynamically favorable reaction to support its experimentally observed facile HNO reactivity.⁵⁶

Table 1. Relative Energies, Key Geometric Parameters, Spin Densities, and Charges in the HNO to NO Conversion Pathways.

ligand	species	ΔE (kcal/mol)	ΔEzpe (kcal/mol)	ΔH (kcal/mol)	ΔG (kcal/mol)	RCuN (Å)	RHN3 (Å)	RCuN3 (Å)	QCu (e)	QN3 (e)	ραβCu (e)	ραβNO (e)
14-N2S2	I-1-S	0.00	0.00	0.00	0.00	2.552	3.661	2.075	0.604	–0.745	0.395	0.003
	TS-S	20.61	19.94	19.31	21.10	2.012	2.027	3.153	0.584	–0.764	0.381	0.122
	I-2-S	–7.74	–7.24	–6.85	–8.40	2.019	1.020	3.416	0.244	–0.604	0.017	0.987
	I-3-S + NO	–8.67	–8.94	–8.84	–18.53		1.030	3.080	0.354	–0.635
14-N4	I-1-N	0.00	0.00	0.00	0.00	2.692	4.104	2.048	0.966	–0.753	0.501	0.001
	TS-N	27.62	26.06	25.58	26.86	2.052	1.853	3.085	0.921	–0.771	0.484	0.106
	I-2-N	4.96	5.16	5.28	4.96	2.027	1.028	3.205	0.698	–0.625	–0.204	1.328
	I-3-N + NO	0.27	0.22	0.24	–0.22		1.031	2.895	0.621	–0.614
14-N2Se2	I-1-Se	0.00	0.00	0.00	0.00	2.576	4.197	2.090	0.531	–0.744	0.369	0.009
	TS-Se	18.26	17.88	17.24	19.14	1.990	2.059	3.291	0.533	–0.762	0.370	0.111
	I-2-Se	–10.38	–9.85	–9.53	–10.64	2.028	1.030	3.228	0.137	–0.637	0.015	0.962
	I-3-Se + NO	–12.26	–12.36	–12.31	–21.62		1.036	3.138	0.286	–0.646
14-N2S2-Ts	I-1-Ts	0.00	0.00	0.00	0.00	2.334	4.533	2.099	0.570	–0.737	0.391	0.020
	TS-Ts	20.81	19.70	19.10	20.92	2.021	1.815	3.094	0.546	–0.775	0.324	0.253
	I-2-Ts	–19.48	–18.54	–18.05	–19.66	2.181	1.027	3.280	0.226	–0.631	0.038	0.943
	I-3-Ts + NO	–16.06	–16.11	–15.83	–26.14		1.030	3.157	0.367	–0.625
14-N2S2-Bn	I-1-Bn	0.00	0.00	0.00	0.00	2.581	4.318	2.106	0.613	–0.579	0.408	0.003
	TS-Bn	16.94	15.93	15.42	16.57	2.026	1.952	3.208	0.628	–0.601	0.349	0.145
	I-2-Bn	–17.68	–16.91	–16.50	–18.78	2.229	1.029	3.293	0.236	–0.505	0.032	0.950
	I-3-Bn + NO	–15.37	–18.22	–15.33	–25.20		1.032	3.126	0.379	–0.506

Figure 1.

(A) Gibbs free energies of the HNO to NO conversion pathways with all studied Cu macrocyclic complexes. (B) Optimized transition-state structures with key geometric changes of R_CuN, R_CuN3, and R_HN from I-1 to TS. Atom color scheme: C, cyan; N, blue; O, red; S, yellow; Se, orange; H, gray; Cu, coral.

The alternative pathway with a proton transferred to another coordinated nitrogen N₁ was also studied. It has a much higher ΔG^‡ by 6.96 kcal/mol. This is because N₃ points toward HNO, while N₁ points away from HNO (see Figures S1 and S5), and thus HNO’s H is much closer to N₃ in I-1-S than to N₁ in I-1′-S by 0.233 Å to shorten the proton transfer distance for the N₃ site (Table S16). So, only the favorable N₃ pathway was used in the subsequent study.

Experimentally, the use of the 14-N₄ ligand does not exhibit any reactivity toward HNO.⁵⁶ As seen from Table 1 and Figure 1A, this aza-only ligated system has a much higher barrier of ∼27 kcal/mol, a positive ΔG for I-2-N formation, and a basically net-zero reaction energy (thus higher than that for the 14-N₂S₂ system by 18.31 kcal/mol). So, its reaction has no kinetic feasibility and thermodynamic favorability. These data also point out interesting reaction feature differences because of coordination atom changes not reported before:^59,61,69 (1) HNO binds more favorably to the Cu center in I-1-S than in I-1-N, as indicated by the 0.14 Å shorter Cu–N distance in Table 1. This could be a result of less steric hindrance for HNO binding due to the enlarged coordinated sphere for the 14-N₂S₂ ligand (S is significantly larger than N). This makes HNO’s H closer to N₃ by 0.443 Å in I-1-S than in I-1-N to facilitate the proton transfer (Table 1); (2) the enlarged coordination sphere makes the Cu–N₃ bond longer by ∼0.03 Å in I-1-S and thus weaker for lower cleavage cost; (3) the less electronegative sulfur vs nitrogen renders a significantly less positively charged Cu (0.604 e vs 0.966 e for Cu in I-1-S and I-1-N, respectively, Table 1), which reduces the attraction between Cu and the negatively charged N₃ to reduce the energy cost to break the Cu–N₃ bond (the largest energy consumption part in TS).

Based on the effects of S vs N, we hypothesized that using the even larger and softer Se atom to replace S could further enhance the HNO reactivity and evaluated it. Similar to 14-N₂S₂, [Cu^II(14-N₂Se₂)]²⁺ also exists predominantly in the trans-III conformation (95% based on the energy difference with other common conformations, Table S6), which was used in the pathway calculations. As shown in Table 1 and Figure 1A (blue energy levels), this system indeed reduces ΔG^‡ and ΔG° further by 2–3 kcal/mol compared to 14-N₂S₂. The even larger and softer Se atom leads to an even longer Cu–N₃ bond (by 0.015 Å) and less positively charged Cu (by 0.073 e) to further weaken the Cu–N₃ bond to cut energy cost. These data provide additional evidence of the coordination atom effect on HNO reactivity found here to facilitate future HNO sensor design.

Since a recent computational work⁵⁹ revealed beneficial effects of either EDG or EWG substituents on the aza-only 14-N₄ ligand in experimental Cu cyclams,^30,56 we examined such effects on the mixed thia/aza ligand, i.e., 14-N₂S₂-Ts and 14-N₂S₂-Bn (Scheme 1). Bn/Ts atoms were installed on similar Cu cyclams and thus are feasible for future experimental synthesis.^30,56 The reason to have only one substituent is to minimize the steric hindrance, the primary effect on HNO reactivity found recently.⁵⁹

We first optimized their starting Cu^II complexes in the commonly found trans-III and trans-I conformations to determine the more favorable one for subsequent pathway studies, which show <1 kcal/mol energy differences (Table S6), i.e., within typical computational errors. This suggests coexistence. We then calculated the pathways for both conformations and found that the trans-I conformation has more favorable ΔG^‡ and ΔG° by 3–4 kcal/mol (Table S10), which was used in the following discussion. Among the two nitrogen sites on these two ligands, the nitrogen with the Bn substitution for 14-N₂S₂-Bn and that opposite to the Ts substituent for 14-N₂S₂-Ts were used as proton acceptors based on their more favorable protonation energies from a prior work on similarly substituted cyclams.⁵⁹

As shown in Table 1 and Figure 1A (green energy levels), the Ts group on the mixed thia/aza framework further decreases ΔG° by 7.61 kcal/mol to make it much more thermodynamically favorable, although its effect on ΔG^‡ is mild (0.18 kcal/mol). This EWG helps attract the electron-rich HNO to facilitate its binding, leading to the shortest Cu–N distance for all I-1 species (Table 1). Its Cu–N₃ bond is longer than that in the parent I-1-S system by 0.024 Å, indicating a more weakened Cu–N₃ bond to facilitate this reaction. The electron-withdrawing nature of Ts may help enhance Cu’s redox potential to facilitate the electron transfer part of this PCET reaction. Indeed, ρ_Cu^αβ change from I-1-Ts to TS-Ts due to electron transfer is the largest here; see Table 1.

In contrast, Bn further reduces ΔG^‡ by ∼4 kcal/mol while having a similar ΔG° to Ts; see Table 1 and violet energy levels in Figure 1A. This EDG on N₃ not only helps improve this nitrogen’s proton affinity to promote the proton transfer in this reaction but also contributes to a steric effect on the Cu–N₃ bond to elongate it, which results in the longest Cu–N₃ bond among all I-1 species. Therefore, it has the weakest Cu–N₃ bond to save energy cost most in TS, which makes its ΔG^‡ the lowest here.

Overall, all three designed complexes show promising improved HNO reactivities with favorable coordination atom and substituent effects. Results also highlight the importance of reducing Cu–N₃ bond cleavage energy cost on HNO reactivity. In fact, for all five complexes here, the linear correlation coefficients for barriers vs R_CuN3’s and barriers vs |Q_Cu × Q_N3| data at I-1 are 0.92 and 0.96, respectively (see Figure 2A,B). These excellent correlations show that a weak Cu–N₃ bond with a long Cu–N₃ bond length and a small electrostatic attraction between Cu and N₃, as indicated by a small value of |Q_Cu × Q_N3|, is associated with a small barrier. These excellent quantitative relationships may help future evaluation of new HNO sensors.

Figure 2.

Linear correlation plots of (A) reaction barriers vs R_CuN3 at I-1 and (B) reaction barriers vs |Q_Cu × Q_N3| at I-1.

Deprotonation

The previously unknown mechanism of spontaneous changes from trans-III I-3-S to the experimental tetrahedral Cu^I product P–S, essential to prevent the demetalation problem of prior Cu-based HNO sensors,⁵⁶ was then investigated together with the three new ligands.

Because I-3-S has an extra proton compared to P–S, deprotonation on N₃ is needed via a base (A^–) in the solution; see Scheme 2. We first performed calculations to see whether the trans-III to tetrahedral conformation change should occur before or after deprotonation. As shown in Table S19, I-3-S is preferred in trans-III by 2 kcal/mol, which corresponds to 97% population. The deprotonated I-5-S, however, has a thermodynamic trend to undergo the conformation change to be tetrahedral, which is more stable by ∼4 kcal/mol. These results suggest that I-3-S is to be deprotonated first to trigger the conformation change.

Scheme 2. Deprotonation Step.

Color sphere represents the transferred proton.

According to pK_a data of the potential proton abstractors in the experimental buffered pH 7 solution,⁵⁶ such as the buffer PIPES (piperazine-N,N′-bis(2-ethanesulfonic acid); pK_a, 6.76), the counterion OTf^– (pK_a, −14), water (pK_a, −1.7), and OH^– (pK_a, 15.74),⁸⁸ PIPES is the most possible one for use in calculations.

As shown in Scheme 2 with optimized structures in Figure S6, PIPES first forms an adduct I-4-S. In TS-1-S, the proton that is more outward is abstracted by PIPES (the inward one is inaccessible due to steric hindrance of PIPES) to form I-5-S. The increase of the distance between this proton and N₃ and the concomitant decrease of the distance between this proton and N(PIPES), the nitrogen atom in PIPES that accepts it, can be clearly seen in Figure 3C.

Figure 3.

(A) Gibbs free energies of deprotonation pathways for 14-N₂S₂, 14-N₂Se₂, and 14-N₂S₂-Ts. (B) Gibbs free energies of the first substituent rotation and the deprotonation pathway for 14-N₂S₂-Bn. (C) Optimized transition-state structures with key geometric changes of R_HN3 and R_HN(PIPES) from I-4 to TS-1 (TS-2 for 14-N₂S₂-Bn). Atom color scheme: C, cyan; N, blue; O, red; S, yellow; Se, orange; H, gray; Cu, coral.

Energy results from I-3-S to I-5-S in Figure 3A show that deprotonation is kinetically and thermodynamically favorable with ΔG^‡ and ΔG° of 4.38 and −7.30 kcal/mol, respectively. These species all have negative overall energies with respect to I-1-S. So, deprotonation is spontaneous in this experimental condition.

14-N₂Se₂ and 14-N₂S₂-Ts follow the same mechanism and have similarly favorable reaction barriers and energies: ΔG^‡ of ∼5–6 kcal/mol and ΔG° of −5 to −8 kcal/mol; see Figure 3A. With the more favorable I-3-Ts and I-3-Se over I-3-S, their deprotonated I-5 species are still more favorable.

However, due to steric hindrance, it is not possible for PIPES to reach the inward proton (transferred from HNO) in 14-N₂S₂-Bn for deprotonation at the Bn-substituted N₃ site. So, Bn first needs to be rotated away in I-3-Bn via TS-1-Bn to expose the N₃ proton (see the optimized structures in Figure S21), which forms I-4-Bn with a low energy cost of 2.79 kcal/mol. This rotation can be seen by a large bond angle increase for ∠CuN₃C_Bn (benzyl carbon bonded to N₃) from 118.2° in I-3-Bn to 154.3° in I-4-Bn (Table S66). It then binds with PIPES to form adduct I-5-Bn with subsequent TS-2-Bn to take the proton away to yield I-6-Bn; see Figure 3B for energies and Figure S22 for optimized structures. Compared with other ligands, the deprotonation ΔG^‡ of 13.98 kcal/mol from I-3-Bn to TS-2-Bn (Figure 3B) is the highest due to the extra Bn rotation and the more difficult proton abstraction at an EDG-substituted site with an enhanced proton affinity. However, it is still lower than the HNO to NO conversion barrier. The energy of TS-2-Bn from I-1-Bn (−10.91 kcal/mol) is still negative. This step is thermodynamically favorable; see Figure 3B. These results show that all four mixed ligands have spontaneous deprotonations.

Proton or Substituent Rotation

The Cu–N₃ bond in I-5-S is still broken (∼3 Å, Table S27) because the remaining proton points toward Cu to intrude the Cu–N₃ bonding. Since it is reformed (2.1 Å, Tables S1 and S2) in the final product, a proton rotation is needed to position N₃ in a closer distance and have its lone pair in the right orientation to rebond with Cu; see Scheme 3 and the optimized structures in Figure S7. After rotation, I-6-S has Cu–N₃ rebonded (2.131 Å, Table S30), which restores the symmetry to the Cu–N₁ bond like in a typical trans-III structure. The proton rotation can be seen by the increased ∠CuN₃H from 51.7° in I-5-S via 80.5° in TS-2-S (see the change in Figure 4C) to 102.9° in I-6-S and the change of ∠S2CN₃H from −53.5° in I-5-S to −110.5° in I-6-S. As shown in Table S29, from I-5-S, the proton rotation barrier is 1.16 kcal/mol and its ΔG° toward I-6-S is −4.97 kcal/mol, indicating a fast step to make it ready for following conformational changes.⁸⁹⁻⁹²

Scheme 3. Proton Rotation Step.

Green box represents the rotated proton.

Figure 4.

(A) Gibbs free energies of proton rotation pathways for 14-N₂S₂, 14-N₂Se₂, and 14-N₂S₂-Ts. (B) Gibbs free energies of the second substituent rotation pathway for 14-N₂S₂-Bn. (C) Optimized transition-state structures with the geometric change of ∠CuN₃H for proton rotation from I-5 to TS-2 and ∠CuN₃C_Bn for substituent rotation from I-6 to TS-3 with 14-N₂S₂-Bn. Atom color scheme: C, cyan; N, blue; O, red; S, yellow; Se, orange; H, gray; Cu, coral.

Additional calculations were carried out to see whether the conformation change could proceed without proton rotation. However, as shown in Table S20 for the first conformation change transition state TS-3-S (which is sulfur inversion; see more discussion of the conformation change pathway later), without this proton rotation step, the barrier becomes much higher by 5 kcal/mol, supporting the necessity of proton rotation.

We then examined this step for 14-N₂Se₂ and 14-N₂S₂-Ts (optimized structures in Figures S12 and S17, respectively). As shown in Tables S42 and S55, they have similar barriers (∼1 kcal/mol) and similar or more favorable reaction energies (∼− 4 or −8 kcal/mol). Overall, these two new systems have more favorable proton rotation than the 14-N₂S₂ ligand; see lower energy levels in Figure 4A.

Due to Bn substitution on N₃, there is no proton attached to N₃ at I-6-Bn and thus no proton rotation step. However, the Bn substituent is still bent away. So, I-6-Bn then undergoes a Bn rotation via TS-3-Bn to restore its normal stable conformation in I-7-Bn with a significant ΔG° drop of ∼11 kcal/mol and relatively small ΔG^‡ of 2.25 kcal/mol (Table S71), which also reforms the Cu–N₃ bond with a length of 2.237 Å (Table S72). This is the second substituent rotation, which can be seen by the decreased ∠CuN₃C_Bn (C_Bn is the benzylic C) from 152.3° in I-6-Bn via 147.8° in TS-3-Bn (see the change in Figure 4C) to 106.9° in I-7-Bn.

Overall, this proton or substituent rotation step for all of these ligands is more favorable than the previous proton abstraction step and new ligands have more negative energies than 14-N₂S₂.

Conformation Change via Deprotonation

The conformation change mechanisms toward tetrahedral for Cu^I cyclams and S-containing derivatives^89,91,92 were reported before, involving sequential inversion of certain coordinated atoms depending on the starting conformation. While S inversion can occur directly, the inversion of a nitrogen center requires deprotonation and then reprotonation before/after the inversion. However, there are no computational studies of this mechanism. So, we first investigated it for 14-N₂S₂.

As shown in Scheme 1B, in trans-III conformation, one pair of N and S points up and the other pair points down, while both nitrogens point up and both sulfurs point down in the tetrahedral conformation. Thus, one N and one S inversions are needed for this conformation change. For 14-N₂S₂, specifically, N₁ and S₄ are to be inverted; see Scheme 4.

Scheme 4. Nitrogen Inversion via Deprotonation.

Color spheres represent the involved proton.

Based on the proposed nitrogen inversion mechanism for metal cyclams and derivatives,^89,91,92 N₁’s proton in I-6-S is deprotonated by a base via a transition state TS-3*-S to form I-7*-S. Afterward, N₁ inversion occurs when a proton is added to the opposite side to generate I-8*-S to complete the needed nitrogen site structural change, which then undergoes a sulfur inversion to form the final P–S. Since the experimental reaction occurred at pH 7, PIPES was again used as a deprotonation agent. However, the reaction energy to form I-7*-S from I-6-S is 33 kcal/mol (Table S21), which is too high to be thermodynamically favorable. Because a reaction barrier is always higher than the reaction energy, it suggests that this deprotonation is also kinetically not feasible. In fact, all efforts to locate TS-3*-S failed. So, these results suggest that this mechanism does not operate here.

Conformation Change via Rotation

We then proposed a rotation pathway for the needed N₁ and S₄ inversions. To determine the inversion sequence, we calculated I-7-S and I-7′-S with sulfur and nitrogen inverted first, respectively. The S first-inverted intermediate I-7-S has a lower energy by −10.85 kcal/mol (Table S21) than the N first-inverted I-7′-S and is the only thermodynamically favorable one. So, our proposed pathway in Scheme 5 has S inversion first. For S₄ inversion (the cyan box part in TS-3-S, Scheme 5), the ΔG^‡ for TS-3-S from I-6-S is 13.81 kcal/mol (Figure 5A and Table S32), which is below barriers of some room-temperature reactions⁸⁵⁻⁸⁷ and thus supports its kinetical feasibility. This S₄ inversion can be seen by the dihedral angle of ∠N₁CuS₄C₁ (see atom labels in I-6-SFigure 5B) from positive 25.6° (C₁ below the N₁–Cu–S₄ plane) to 6.9° (C₁ almost in that plane) to negative −6.8° (C₁ above that plane); see the data in Table S33.

Scheme 5. Conformation Change Pathway via Rotation.

Color boxes represent the rotation site.

Figure 5.

(A) Gibbs free energies of the rotational conformation change pathway with Cu macrocyclic complexes: 14-N₂S₂ and 14-N₂Se₂. (B) Optimized structures of the species involved in the conformation change pathway in the 14-N₂S₂ system (the 14-N₂Se₂ system follows the same reaction pathway; see Figure S13). Atom color scheme: C, cyan; N, blue; O, red; S, yellow; H, gray; Cu, coral. Key atoms’ labels (here, H is the N₁’s proton that points down) are shown in the optimized structure of I-6-S.

I-7-S then undergoes a small cost (2.56 kcal/mol) carbon bridge rotation to form I-8-S, where the carbon chain between N₁ and S₄ (the green box part in I-7-S, Scheme 5; carbons C₁ and C₂ labeled in I-6-SFigure 5B) has changed to the same conformation in P–S; see the optimized structures in Figure 5B. This C₁–C₂ orientation change is shown by the dihedral angle of ∠N₁CuC₁C₂ from a negative value of −27.8° in I-7-S to a positive number of 12.5° in I-8-S; see Table S33 and Figure 5B.

Then, N₁ inversion starts. Because N₁’s H points down in I-6-S and should point up like N₃′s in P–S, N₁ inversion requires simultaneous H rotation, which takes two steps. First, this H (the pink box part in TS-4-S, Scheme 5) rotates from pointing away from Cu in I-8-S to toward Cu in I-9-S via TS-4-S. This is indicated by the gradual decrease of ∠HN₁Cu from 103.0° via 71.7 to 61.3° in this step; see Table S33 and Figure 5B. It has a small barrier of 6.46 kcal/mol (Figure 5A). This proton rotation disrupts the Cu–N₁ bond: the bond length changes from 2.177 via 2.778 to 3.295 Å from I-8-S via TS-4-S to I-9-S (Table S33), which allows more flexibility for the following concerted nitrogen and hydrogen inversion from pointing down to up via TS-5-S to I-10-S. The N₁ inversion can be seen by a ∠C₂C₃CuN₁ change from +31.8° in I-9-S to −31.7° in I-10-S, corresponding to the N₁ change from below the C₂–C₃–Cu plane to above; see Table S33 and Figure 5B. Accordingly, its proton inversion is shown by a positive ∠CuC₂C₃H value of 67.2° in I-9-S to a negative value of −68.6° in I-10-S, reflecting the H position from below the Cu–C₂–C₃ plane to above. The Cu–N₁ bond reforms (2.114 Å) in I-10-S. Now, the only difference between I-10-S and P–S is the carbon bridge between N₁ and S₂: the relative heights for the three carbons from N₁ to S₂ are high–low–high (see the blue box in I-10-S in Scheme 5), which are changed to low–high–low in P–S. This step is energetically downhill with −2.76 kcal/mol.

As shown in Figure 5A, in this rotation pathway from I-6-S to P–S, the rate-determining step (RDS) is the first S rotation (TS-3-S) with a small barrier of ∼14 kcal/mol. This conformation change pathway is also thermodynamically favorable, with a reaction energy of −7.37 kcal/mol. With the starting HNO to NO conversion reaction, the overall reaction energy is −38.17 kcal/mol (Figure 5A). So, this mechanism is the first plausible one to explain the experimentally observed facile conformation change for 14-N₂S₂.

As 14-N₂Se₂ has the same starting conformation as 14-N₂S₂, it follows the same pathway; see the blue energy levels in Figure 5A and optimized structures in Figure S13. Compared to 14-N₂S₂, the ΔG^‡ from I-6-Se (19.18 kcal/mol) is higher but still within the range of room-temperature reactions,⁸⁵⁻⁸⁷ while the reaction energy (−9.16 kcal/mol) is lower. This may be a result of the larger size of Se, which increases the strain (Cu–N₁/N₃ bond lengths elongated by ca. 0.05 Å, Tables S33 and S46) in TS-3-Se for Se inversion but allows more space for subsequent N inversion to make them more favorable; see the lower (blue vs red) energy levels in Figure 5A.

Pathways for 14-N₂S₂-Ts and 14-N₂S₂-Bn are different because their Cu complexes are in trans-I conformation, not trans-III for 14-N₂S₂ and 14-N₂Se₂. Structural differences between trans-I and tetrahedral lie in both sulfurs, which point up and down, respectively (Scheme 1B). Hence, the conformation change involves two S inversions here, which take fewer steps (a favorable reaction feature) than the nonsubstituted systems; see Figures 6 and 7. Since both S sites need to be inverted and they are symmetrically distributed on two sides of the substituent, we chose the sulfur with a relatively longer Cu–S bond length (slightly more flexible for structural changes) to be first inverted. Barriers are similar (<1–2 kcal/mol differences) for two S sites due to symmetry; see Figures 6 and 7.

Figure 6.

(A) Gibbs free energies of the rotational conformation change pathway in 14-N₂S₂-Ts. (B) Optimized structures of species involved in the conformation change pathway in the 14-N₂S₂-Ts system. Atom color scheme: C, cyan; N, blue; O, red; S, yellow; H, gray; Cu, coral.

Figure 7.

(A) Gibbs free energies of the rotational conformation change pathway in 14-N₂S₂-Bn. (B) Optimized structures of species involved in the conformation change pathway in the 14-N₂S₂-Bn system. Atom color scheme: C, cyan; N, blue; O, red; S, yellow; H, gray; Cu, coral.

For 14-N₂S₂-Ts, the RDS has a ΔG^‡ of 15.56 kcal/mol from I-6-Ts; see Table S58. The EWG substituent on N₁ destabilized the Cu–N₁ bond, which is longer than the Cu–N₃ bond by ca. 0.5 Å in I-6-Ts. The S inversion further weakens it to be broken in TS-3-Ts and restored in P-Ts; see the optimized structures in Figure S18 and geometric data in Table S59. The two S inversions can be seen by the sign changes of corresponding dihedral angles of ∠C₅C₆CuS₂ and ∠C₁₀C₁CuS₄ (see atom labels in Figure 6), respectively, to reflect that S₂ and S₄ change from above their respective planes to below; see Table S59. The conformation change alone has a negative ΔG° of −1.39 kcal/mol. The 14-N₂S₂-Ts system overall has a reaction energy of −43.39 kcal/mol, which is ca. 5 kcal/mol lower than that of 14-N₂S₂.

For 14-N₂S₂-Bn, the starting trans-I structure for conformation change is I-7-Bn with a normal Bn conformation. The two S inversions can be seen by the sign changes of corresponding dihedral angles of ∠C₅C₆CuS₂ and ∠C₁₀C₁CuS₄ (see atom labels in Figure 7), respectively, to reflect that S₂ and S₄ change from above their respective planes to below; see Table S75. As seen from Table S74 (optimized structures in Figure S24), the RDS has a ΔG^‡ of 19.62 kcal/mol, which is higher than other ligands but still in the range for room-temperature reactions. Its conformation change ΔG° of −1.77 kcal/mol is like that for 14-N₂S₂-Ts. Overall, the final tetrahedral product formation energy of −38.82 kcal/mol is slightly lower than that of 14-N₂S₂. So, the new mechanism is favorable for all four ligands.

Comparisons with Other Systems

The HNO reaction with the oxidized metal center to form a reduced metal and a proton transfer to the system is a common feature of the metal-based HNO reaction sensors or probes.²²⁻³⁴ We have studied a number of such reactions including both Cu-based systems and heme proteins,⁵⁹⁻⁶¹ which all exhibit the PCET mechanism. Table 2 presents the values of ΔG^‡’s, ΔG°’s, and ΔR_M···N’s (change of the distance between the metal and the coordination nitrogen that accepts HNO’s proton from the transition state to its preceding intermediate, i.e., TS vs I-1) of this reaction from the HNO-bound complex to the NO-released stage, i.e., from I-1 to I-3 and NO.

Table 2. Comparisons of Key Energies and Geometric Parameters in the HNO to NO Conversion Pathways.

system	ΔG‡ (kcal/mol)	ΔG° (kcal/mol)	ΔRM···Na (Å)	refsb
[Cu(14-N4)]	26.86	–0.22	1.037	t.w.
[Cu(14-N2S2)]	21.10	–18.53	1.078	t.w.
[Cu(14-N2Se2)]	19.14	–21.62	1.201	t.w.
[Cu(14-N2S2-Ts)]	20.92	–26.14	0.995	t.w.
[Cu(14-N2S2-Bn)]	16.57	–25.20	1.102	t.w.
CuDHX1	20.59	–17.22	1.025	(59)
Cu-Fl	18.84	–11.21	1.050	(59)
Cu-Ts	23.53	–11.76	0.932	(59)
Cu-Me4	19.02	–14.29	0.971	(59)
CuZnSOD	10.98	–42.66	0.374	(61)
CuZnSOD model	12.12	–12.80	1.076	(69)
myoglobin	–1.92	–10.07c	N.A.	(60)
catalase	–1.18	–11.13c	N.A.	(60)

^aN.A. means not applicable.

^bReference number. t.w. means this work.

^cThe reaction energy is up to the NO-bound reduced metal, not the NO-released system.

It can be seen that the Cu cyclam derivative systems with coordination S/Se atoms studied in this work have a relatively lower reaction barrier range (17–21 kcal/mol) and lower reaction energy range of −19 to −26 kcal/mol than Cu cyclams (19–27 kcal/mol and ∼0 to −17 kcal/mol, respectively) reported recently.⁵⁹ These data clearly show that the mixed S/Se–N coordination systems have better HNO reactivities, which were found to originate from at least three favorable factors described in detail in the HNO Reactivity Section under Results and Discussion Section: (1) less steric hindrance due to the enlarged coordinated sphere facilitates HNO binding; (2) the enlarged coordination sphere weakens the Cu–N₃ bond for lower cleavage cost; and (3) the less electronegative S/Se vs N renders a significantly less positively charged Cu to reduce its attraction with the negatively charged N₃ to decrease the bond cleavage cost.

However, the CuZnSOD (SOD, superoxide dismutase) protein has the lowest barrier and reaction energy among all Cu systems (see Table 2), which was found to be a result of a more flexible nonplanar coordination environment in the protein to lower the Cu–N bond-breaking cost and a stronger proton acceptor due to a negatively charged coordination histidine ligand.⁶¹ The previously developed CuZnSOD model for HNO detection with a similar coordination environment but with a neutral ligand has higher ΔG^‡ and ΔG°,⁶⁹ see Table 2.

In contrast with these Cu systems, the heme proteins (myoglobin and catalase) have basically barrierless (ΔG^‡’s are in Table 2; ΔE^‡’s are 0.25 and 0.83 kcal/mol, respectively) HNO to NO conversion because HNO’s proton is transferred to a protein residue that has strong hydrogen bonding with it before the reaction.⁶⁰ Thus, the heme systems have no metal-coordination-N bond breaking to save the energy cost involved in the Cu systems. However, the strong binding of NO to ferrous heme centers makes the final systems remain nitrosylated and thus difficult to recycle for reuse.

Overall, the introduction of sulfur (and selenium) as coordination atoms in Cu cyclam derivatives leads to stable tetrahedral structures of the reduced species, a unique feature the all-nitrogen coordination system does not have (e.g., [Cu(14-N₄)] still prefers the trans-III conformation in the reduced form, Table S3). This stable reduced species is also free of anything from HNO after the reaction, which makes it clean to be recycled back for repeated uses. Therefore, only the Cu systems with the mixed S/Se coordination atoms have subsequent conformation change pathways, which were found to be kinetically feasible and thermodynamically favorable from the above-mentioned studies.

Conclusions

Quantum chemical calculations accurately (∼2% error) predicted experimental structures of Cu^II/Cu^I complexes of the mixed thia/aza system and offered the first theoretical insights into the selective stabilization of the tetrahedral conformation for the reduced system. Results also reproduced the experimental facile HNO reactivity of 14-N₂S₂ vs inactivity for 14-N₄ and the spontaneous conformation change to the final product for 14-N₂S₂. The complete pathway of HNO to NO conversion, deprotonation, proton or substituent rotation, and conformation change via rotation was uncovered for this reusable HNO sensor. A novel conformation change mechanism was discovered, which is more favorable than the previously proposed deprotonation–inversion–protonation pathway for the room-temperature, neutral-pH condition of the experimental or physiological systems. The whole reaction pathway was then examined for three new designs, which were found to have improved reactivities due to the introduction of softer and larger coordinate atoms and electron-withdrawing or electron-donating substituents. The favorable results with substituted mixed thia/aza systems are particularly interesting, which paves new venues to incorporate fluorophores into reusable HNO sensors that have not been discovered before. Overall, this first and systematic mechanistic study of a reusable HNO sensor uncovered numerous insightful results to facilitate HNO sensor development for broad biochemical and biomedical uses.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.3c04506.

• Details of all the studied structures and pathways with various geometric and electronic parameters and additional figures, together with absolute energies and coordinates of favorable species (PDF)

Author Contributions

Y.Z. conceived the idea and designed the research. J.M.C., D.B., and V.N. conducted the computational studies. All authors participated in the data analyses and preparation of data tables and figures. J.M.C. and Y.Z. wrote the manuscript together with input from all authors. Y.Z. supervised the research.

The authors declare no competing financial interest.