Tuning Supramolecular Selectivity for Hydrosulfide: Linear Free Energy Relationships Reveal Preferential C–H Hydrogen Bond Interactions

Abstract

Supramolecular anion receptors can be used to study the molecular recognition properties of the reactive yet biologically-critical hydrochalcogenide anions (HCh^–). Achieving selectivity for HCh^– over the halides is challenging but is necessary for not only developing future supramolecular probes for HCh^– binding and detection, but also for understanding the fundamental properties that govern these binding and recognition events. Here we demonstrate that linear free energy relationships (LFERs)—including Hammett and Swain-Lupton plots—reveal a clear difference in sensitivity to the polarity of an aryl C–H hydrogen bond (HB) donor for HS^– over other HCh^– and halides. Analysis using electrostatic potential maps highlights that this difference in sensitivity results from a preference of the aryl C–H HB donor for HS^– in this host scaffold. From this study, we demonstrate that LFERs are a powerful tool to gain interpretative insight into motif design for future anion-selective supramolecular receptors and highlight the importance of C–H HB donors for HS^– recognition. From our results, we suggest that aryl C–H HB donors should be investigated in the next generation of HS^– selective receptors based on the enhanced HS^– selectivity over other competing anions in this system.

Graphical Abstract

Introduction

The hydrochalcogenide anions (HCh^–, Ch = Group 16 element) hydrosulfide (HS^–) and hydroselenide (HSe^–) are highly reactive species that play crucial roles in biological systems.^1,2 At physiological pH, these anions are favored in solution over their diprotic counterparts hydrogen sulfide (H₂S) and hydrogen selenide (H₂Se), which are important as a biological gasotransmitter and in selenium metabolism, respectively.^1–4 The high nucleophilicity and redox activity of these anions, however, has stymied many investigations.^1,2 A better understanding of the molecular recognition properties of these anions could aid our understanding of the non-covalent forces used to stabilize these reactive species in biology and enable the development of future probes for HCh^– binding and detection.

The well-known ability of supramolecular receptors to reversibly bind anionic guests through noncovalent interactions^5–7 mimics strategies found in Nature to stabilize reactive species^8–10 and offers an attractive approach for studying the supramolecular chemistry of HCh^–. Our groups recently reported the first examples of organic receptors that can reversibly bind HS^– and HSe^– by employing urea or amide N–H hydrogen bond (HB) donors and aromatic C–H HB donors (1, 2, Figure 1).^11–13 More recently, an additional report of reversible supramolecular HS^– binding was reported in a system that employs bambusuril C–H HB donors (3, Figure 1).¹⁴

Figure 1.

Receptor scaffolds 1–3 all bind HS^– reversibly and all contain C–H HB donors. Only receptor classes 1 and 2 have been shown to reversibly bind HSe^–. The C–H HB donors that interact with the HCh^– are shown in red for clarity.

One advantage of using synthetic supramolecular receptors is the ability to engineer the receptor scaffold to improve specificity for one guest over other competing analytes. Supramolecular receptors with high selectivity could prove useful in applications such as sensing, extraction, controlled delivery, and cell-membrane transport.^15–17 Designing selective systems for HCh^–, however, is difficult given the similarity of size and shape to halides. This challenge in molecular recognition has not yet been solved and can be observed in all previously reported receptors 1-3 for HCh^–, all of which also bind Cl^– and Br^– with appreciable affinity to HCh^– of a similar size.^11–14

C–H HB donors serve as a unifying theme in receptors 1–3, which hints that such moieties could be important for HCh^– binding. Broadening these types of interactions, methionine C–H^…S HBs have been reported to be critical in the substrate specificity and catalytic activity of methionine aminopeptidase.¹⁸ These interactions are not limited to one protein, but rather have been observed in 20 other protein structures in the Protein Data Bank (PDB) prior to 2016.¹⁸ These findings prompted us to investigate the importance of C–H HB donor motifs in driving selective recognition of HCh^– further. We have previously investigated the effect of polarization of an aryl C–H HB donor on various anions through modifying the para-substituent with electron-donating and -withdrawing groups of 1.¹⁹ Analysis of the linear free energy relationships (LFERs), including Hammett plots, revealed a significant relationship between substituent effects on polarization of the aryl C–H HB and anion binding energy, at a time when non-traditional C–H HB donors were perhaps still considered controversial by some.^19,20 By conducting a similar LFER study with HCh^–, the effect of the C–H HB donor polarization on HCh^– binding could be compared directly to that on halides.

Moreover, these studies can begin to unravel the similarities and differences between selective recognition of halides and HCh^–. At first glance, these anions should behave drastically differently due to the differences in their polarizabilities, pK_bs,^21–23 solvation energies,^24,25 and reactivities; yet they tend to behave surprisingly similarly in their molecular recognition behavior at first glance. Differences in the observed effects could indicate that C–H polarization influences the selectivity between the two classes of anions, which is supported by prior results with 1–3 and in biological systems. Additionally, the systematic investigation of a series of receptors with physical organic methods is a rigorous way to uncover other important details in anion binding mechanisms, binding selectivity, and other anion-dependent effects.^26,27 Motivated by these needs, here we demonstrate that LFERs are a powerful tool that allow for not only anion-dependent solution binding energies (ΔG_binding) to be measured, but also for observing anion-dependent substituent effects and estimating of the difference in aryl C–H^…Anion (A^–) HB strengths between HCh^– and halides. These insights can be used as a first step to understanding the supramolecular chemistry of these anions as well as provide design elements for developing selective receptors for these reactive, yet biologically relevant anions.

RESULTS AND DISCUSSION

Synthesis and Characterization

A series of six arylethynyl bisurea receptors (1^R, Figure 2), differing only by the substituent in the position para to the participating aryl C–H HB donor (–R, Figure 2), was prepared for LFER studies with HCh^– (HS^– and HSe^–) and halides (Cl^– and Br^–).¹⁹ This host system, which can bind a guest molecule through HBs from one aryl C–H HB donor and four urea N–H HB donors, was chosen for its functional group tolerance of and compatibility with HCh^–. Hosts in this series had been previously shown to resist irreversible nucleophilic attack from the HCh^– guests on the titration timescale by preferentially binding HS^– and HSe^– through primarily noncovalent HBs.^11,13 The series of receptors featured one previously unreported host (1^CF3), which was prepared by similar synthetic methods and characterized by ¹H, ¹³C{¹H}, and ¹⁹F NMR spectroscopy (Figures S4–S6) as well as by mass spectrometry.

Figure 2.

The series of arylethynyl bisurea receptors used in this study.

To ensure that modulating the electron-withdrawing or -donating character of the –R substituent only affected the polarity of the C–H_a HB donor and not the NH_b/c HB donors as well, we compared the ¹H NMR spectra of the six 1^R free hosts (Figure 3). The ¹H NMR spectra of the most electron-donating (1^NMe2) and -withdrawing receptors (1^CF3) showed that the largest difference in chemical shift (Δδ) occurred in the aryl CH_a proton (Δδ = 0.927 ppm), followed by much smaller shifts in the proximal urea NH_b protons (Δδ = 0.058 ppm) and distal urea NH_c protons (Δδ = 0.024 ppm). We observed a general trend of downfield shifting of CH_a resonance with increasing electron-withdrawing nature of the –R substituent, with the exception of 1^Cl and 1^F, perhaps revealing the importance of resonance effects in the electron-withdrawing ability of these two substituents.

Figure 3.

¹H NMR spectra of six receptors 1^R in 10% DMSO-d₆/CD₃CN. The largest change in δ of possible HB donors occurs for the CH_a proton peak.

Additionally, the LFER between the ¹H NMR chemical shift (δ) of H_a-c in 10%-DMSO-d₆/CD₃CN and the Hammett parameter σ_p was used to quantitatively evaluate the influence of the electron-withdrawing and -donating nature of the –R substituent on δ of the free-host. Figure S7 and Table S1 show that the plot for the aryl C–H_a proton has a slope of 0.64 ± 0.11, with 90% of the change in δ stemming from the electron-donating or -withdrawing nature of the substituent (R² = 0.90). Conversely, the fits for the urea NH_b/c protons are poor (0.13 ≤ R² ≤ 0.49), and the slopes of the plots for these protons are close to 0, indicating that these protons are not significantly affected by the nature of the functional group. These data show that electronic communication between the urea protons and the –R substituent diminishes with increasing distance, which is consistent with observations in other systems.^27,28 Furthermore, the aryl C–H bond is most affected by the para-substituent modifications, which is consistent with previous work from our group.¹⁹

¹H NMR Spectroscopy Titrations

Previous work from our lab tested the hypothesis that softer HCh^– could interact more favorably with aryl C–H HB donors,¹¹ suggesting that substituent effects that polarize this motif may affect the binding affinity of HCh^– more than the presumptively harder halides. As a result, selectivity between these similar anions could potentially be achieved in this system by exploiting suitably polarized C–H HB donor motifs. To test this hypothesis, we measured binding affinities (K_as) using ¹H NMR spectroscopy titration experiments between the six 1^R hosts and HS^–, HSe^–, Cl^–, and Br^– guests as the tetrabutylammonium (NBu₄⁺) salts^13,29 in 10%-DMSO-d₆/CD₃CN at 25 °C, as shown for host 1^CF3 and HS^– (Figure 4). All experiments were performed under anaerobic and anhydrous conditions since the presence of oxygen or water resulted in brightly colored guest solutions, noisy NMR spectra, poor data fitting, and accelerated irreversible reactivity between hosts and HCh^–. Titrations were performed in triplicate (Method 1 and 2, SI), and K_as and energy of binding in solution (ΔG_binding) (Table 1) were obtained by the Thordarson method.³⁰ Note that some K_a values for 1^H and 1^tBu were previously reported by our groups^11,13 and were reused in this study after replication under the exact conditions reported in this paper.

Figure 4.

(a) Representation of the host–guest equilibrium between 1^CF3 and HS^–. (b) ¹H NMR titration of 2.2 mM 1^CF3 with NBu₄SH in 10% DMSO-d₆/CD₃CN.

Table 1.

Association Constants and Binding Free Energies for Receptors 1^R at 1–3 mM with HS^–, HSe^–, Cl^–, and Br^– in 10% DMSO-d₆/CD₃CN at 25 °C^a

	HS–		HSe–		Cl–		Br–
Host	Ka (M–1)	ΔGbinding (kcal mol–1)	Ka (M–1)	ΔGbinding (kcal mol–1)	Ka (M–1)	ΔGbinding (kcal mol–1)	Ka (M–1)	ΔGbinding (kcal mol–1)
1CF3	15000 ± 1800	−5.69 ± 0.07	940 ± 80	−4.05 ± 0.05	2420 ± 120	−4.61 ±0.03	173 ± 9	−3.05 ± 0.03
1Cl	8480 ± 1170	−5.35 ± 0.08	810 ± 60	−3.96 ± 0.04	2300 ± 180	−4.58 ± 0.05	133 ± 7	−2.89 ± 0.03
1F	8330 ± 940	−5.34 ± 0.07	610 ± 40	−3.79 ± 0.04	1890 ± 90	−4.47 ± 0.03	134 ± 7	−2.90 ± 0.03
1Hb	5010 ± 810b	−5.04 ± 0.10b	530 ± 60	−3.71 ± 0.06	1780 ± 120b	−4.43 ± 0.04b	120 ± 7	−2.84 ± 0.03
1tBub	3600 ± 500b	−4.85 ± 0.08b	460 ± 50b	−3.63 ± 0.06b	1700 ± 200b	−4.40 ± 0.07b	110 ± 20b	−2.78 ± 0.11b
1NMe2	1660 ± 100	−4.39 ± 0.04	360 ± 40	−3.48 ± 0.06	1120 ± 150	−4.15 ± 0.08	85 ± 8	−2.63 ± 0.06

^aAll values were obtained by fitting ¹H NMR titration data to 1:1 binding isotherm model with the error as the standard deviation of three titrations.³⁰ Minimum error is assumed to be 5% of K_a value.

^bValues were previously reported by our groups in references 11 (1^H) and 13 (1^tBu)

LFERs Reveal Anion-Dependent ΔG_binding Trends

With K_as and ΔG_binding values determined for each host/guest combination, we endeavored to use LFERs to visualize binding energy trends within a host/guest series with one anion and across a range of anions. Plotting ΔG_binding of each host/guest complex against σ_p of the –R substituent (Figure 5) revealed a strong linear response of the HCh^– and halide anion binding energies to the electron-withdrawing or -donating nature of the –R substituent. Table 2 summarizes the parameters of the linear fit for each anion, determined through linear regression. For each anion, more than 90% of the change in ΔG_binding can be attributed to the electronics of the –R substituent (0.91 ≤ R² ≤ 0.97). The LFER for HSe^– has the lowest R² value, which we attribute to slight reactivity of HSe^– with the receptors that is not detectable by ¹H NMR spectroscopy over the titration timescale but has been observed previously over several hours.¹³ Competing receptor reactivity has been shown in other systems to adversely affect fits.²⁶ Importantly, the p-values for the contribution of the slopes and intercepts to all the regressions are statistically significant, meaning that both σ_p and ΔG_binding at the intercept are meaningful predictors of anion binding in our systems. The p-values for overall models also reveal that all the LFERs are statistically significant.

Figure 5.

LFER between ΔG_binding and σ_p values for 1^R with HS^–, HSe^–, Cl^–, and Br^–. Dashed lines represent the 95% confidence interval for each linear trend.

Table 2.

Fitting Statistics for the LFER Between ΔG_binding and σ_p for all Four Anions

Guest	Slope	Intercept	p-value			R2	R2adj.
			Slope	Intercept	Model
HS–	−0.97 ± 0.10	−5.14 ± 0.04	< 0.01	< 0.01	< 0.01	0.96	0.95
HSe–	−0.43 ± 0.07	−3.78 ± 0.03	< 0.01	< 0.01	< 0.01	0.91	0.88
Cl–	−0.35 ± 0.03	−4.45 ± 0.01	< 0.01	< 0.01	< 0.01	0.97	0.96
Br–	−0.30 ± 0.03	−2.86 ± 0.01	< 0.01	< 0.01	< 0.01	0.96	0.96

Several trends can immediately be extracted from this LFER study. First, each LFER has a negative slope, indicating that more electron-withdrawing para substituents favor guest binding for all four anions as expected. This trend is likely due to the increasing polarization, and by extension acidity, of the C–H_a HB donor, as noted in previous LFERs of HB receptors with anionic guests.^26,31 Additionally, we previously observed that 1^tBu binds smaller, more basic anions with a higher affinity.¹³ These trends were again observed in this study: for similarly sized anions (e.g., HS^– vs. Cl^– and HSe^– vs. Br^–), the more basic anion is more strongly bound by each receptor, whereas between the anions in the same Group (e.g., HS^– vs. HSe^– and Cl^– vs. Br^–), the smaller anion is more strongly bound. Stronger bases would clearly form stronger HBs with the host, and the more diffuse nature of the larger anions may weaken their HB affinity.

As a result of these trends, the LFERs reveal that all six receptors bind HS^– (a small, basic guest) the strongest, with a preference for this anion over the other anions investigated. Surprisingly, we also saw that when the polarization of the aryl C–H HB donor increases with more electron-withdrawing substituents, the preference of our receptors for HS^– over the other anions increases (Figure 5). The most electron-donating receptor 1^NMe2 shows little preference for HS^– over Cl^– (ΔΔG_binding = −0.24 ± 0.09 kcal mol^–1), whereas the most electron withdrawing receptor 1^CF3 has the largest difference in binding energy between the two anions (ΔΔG_binding = −1.30 ± 0.08 kcal mol^–1), which corresponds to an approximately nine-fold increase in selectivity (Table 2). This unexpected result is significant because polar C–H HB donors may provide a route to future, more selective supramolecular receptors for HS^–.³²

Hammett Plots Reveal Anion-Dependent Substituent Effects

$$\text{log}\frac{{\text{K}}_{\text{a}}^{\text{R}}}{{\text{K}}_{\text{a}}^{\text{H}}}=\rho {\sigma }_{\text{p}}+\epsilon$$ (1)

To better understand and visualize differences in anion binding sensitivity, we used Hammett relationships (log(K_a^R/K_a^H) vs. σ_p) generated by fitting K_a data to Equation 1. Table 3 summarizes the parameters of each linear fit, determined through linear regression. These were fit to a modified Hammett equation that includes an origin offset value (ε). The reported p-values for each slope, ρ, indicate that σ_p is a significant predictor of anion sensitivity to the polarity of the C–H HB donor in our systems, and the p-values of the regression models show the Hammett plots for each anion are statistically significant. Non-significant p-values for the contribution of ε to the overall model indicates that factors beyond the electronics of the –R substituent described by σ_p do not have a meaningful effect on the change in binding energies for the individual anion guests. Indeed, forcing the Hammett plot through the origin of the graph does not result in appreciably different slopes (Table S20).

Table 3.

Fitting Statistics for Hammett Plots for HS^–, HSe^–, Cl^–, and Br^–

Guest	ρ	ε	p-value			R2	R2adj.
			ρ	ε	Model
HS–	0.71 ± 0.07	0.07 ± 0.03	< 0.01	0.08	< 0.01	0.96	0.95
HSe–	0.32 ± 0.05	0.06 ± 0.02	< 0.01	0.06	< 0.01	0.90	0.88
Cl–	0.25 ± 0.02	0.02 ± 0.01	< 0.01	0.16	< 0.01	0.97	0.96
Br–	0.22 ± 0.02	0.02 ± 0.01	< 0.01	0.16	< 0.01	0.96	0.96

The Hammett plots for HSe^–, Cl^–, and Br^– (Figure 6) show a smaller substituent effect than the benchmark deprotonation of benzoic acid in water at 25 °C,³³ with slopes ranging from 0.22 to 0.32. These slopes are similar to our previously reported receptors¹⁹ and other HB donor and acceptor systems.³⁴ In our system these values represent a description of how sensitive anion binding is to the polarization and strength of the aryl C–H HB donor. Using analysis of covariance (ANCOVA) to compare the linear regression models of HSe^–, Cl^–, and Br^–, we found that the slopes of the three Hammett plots are not statistically different (Table S21). These anions may have very different binding energies in our host system (Table 1, Figure 5), but the substituent effect on binding is independent of the identity of the three anions. Stated another way, HSe^–, Cl^–, and Br^– have experimentally identical effects from C–H HB donor modulation in this receptor class.

Figure 6.

Hammett plot between 1^R and HS^–, HSe^–, Cl^–, and Br^–. The slope of HS^– is significantly different from HSe^–, Cl^–, and Br^– illustrating the increased sensitivity of HS^– to substituent effects.

Intriguingly, the larger magnitude of the slope for the Hammett plot of HS^– binding (ρ = 0.71 ± 0.07, Table 3) is statistically significantly larger than that of the other anions, confirmed by ANCOVA (Table S21). These results show that HS^– is more sensitive to the polarization of the C–H HB donor than the other anions. To investigate this results further, we used Swain-Lupton parameters to investigate the relative inductive/field and resonance contributions to the observed substituent effect in each anion binding event, and electrostatic potential (ESP) surface maps to interrogate the strength of each C–H^…A^– HB.

Field/Inductive vs. Resonance Substituent Effects on Anion Binding

$$\text{log}\frac{{\text{K}}_{\text{a}}^{\text{R}}}{{\text{K}}_{\text{a}}^{\text{H}}}={\rho }_{\text{f}}\text{F}+{\rho }_{\text{r}}\text{R}+\text{I}$$ (2)

We first hypothesized the increased sensitivity of HS^– to changing the –R substituent could be attributed to differences in the relative resonance contribution from the –R substituent to the aryl C–H HB in binding with each anion. To test this hypothesis we fit experimentally determined K_as to the Swain-Lupton equation (Equation 2), which splits the ρσ_p term in the Hammett equation (Equation 1) into contributions from field/inductive effects (denoted by ρ_fF) and contributions from resonance effects (denoted by ρ_rR).³⁵ Although our systems should not exhibit any important resonance contributors involving the aryl C–H HB donors (Figure S37), resonance effects have been shown to play a significant role in anion binding within our scaffolds.¹⁹ Table 4 summarizes the parameters of each linear fit, determined through multivariable linear regression. The linear regression of each anion has an excellent fit to Swain-Lupton parameters F and R (0.96 ≤ R² ≤ 0.99). The reported p-values for the contribution of ρ_f and ρ_r to the regressions for all anions indicate that inductive/field effects and resonance are both meaningful contributors to anion binding. In addition, the overall model for the Swain-Lupton plots are statistically significant. Non-significant p-values for the contribution of the intercept I to the regressions (Table 4) indicate that factors beyond field/inductive and resonance substituent effects, such as polarizability and steric interactions, do not make meaningful contributions to the change in binding energies. Again, forcing the plot through the origin of the graph does not significantly affect results (Table S22).

Table 4.

Fitting Statistics from the Multivariable Linear Fit to the Swain-Lupton Equation for the K_a Values for HS^–, HSe^–, Cl^–, and Br^–

Guest	ρf	ρr	I	p-value				R2	R2adj.	%R
				ρf	ρr	I	Model
HS–	1.00 ± 0.08	0.62 ± 0.04	−0.02 ± 0.03	< 0.01	< 0.01	0.50	< 0.01	0.99	0.99	38 ± 3
HSe–	0.48 ± 0.09	0.27 ± 0.05	0.00 ± 0.03	0.01	0.01	0.94	0.01	0.96	0.93	36 ± 8
Cl–	0.28 ± 0.06	0.25 ± 0.03	0.01 ± 0.02	0.02	< 0.01	0.71	0.01	0.97	0.94	47 ± 9
Br–	0.26 ± 0.05	0.20 ± 0.03	0.00 ± 0.02	0.01	< 0.01	0.96	0.01	0.97	0.95	43 ± 8

To better compare the regression results in Table 4, we calculated the percent resonance contribution (%R) to anion binding with Equation 3.¹⁹ Previous computational studies have shown no resonance contribution to anion binding in other C–H HB donor systems in the gas phase.^36,37 This does not take into account the role of solvent, however, which may be crucial in allowing resonance effects to participate in anion binding; in fact, our system shows a high %R contribution to anion binding, ranging from 36 to 47%. We were also able to directly compare the %R contribution for the halides in 10% DMSO-d₆/CD₃CN (solvent dielectric constant ε ~ 42)³⁸ with previously published %R for the halides in a similar receptor series in H₂O_sat./CHCl₃ (ε ~ 4.9).^19,38 We found that despite moving from a comparatively non-competitive solvent system (H₂O_sat./CHCl₃) to a competitive solvent system (10% DMSO-d₆/CD₃CN), all the %R contributions are identical within error.¹⁹

$$\%\text{R}=\frac{{\rho }_r}{{\rho }_f+{\rho }_r}\times 100$$ (3)

We also found that the %R contributions for all the anions in 10% DMSO-d₆/CD₃CN are identical within error, despite HS^– having much larger coefficients for ρ_f and ρ_r. Therefore, we cannot ascribe such a big difference in sensitivity seen for HS^– to a change in the relative contributions from resonances substituent effects. However, analysis of the linear fits of the anions with alternative Hammett parameters σ_m and σ_p⁺, which give more weight to field/inductive effects and resonance effects, respectively,³⁵ indicates that resonance contributions from the substituent may still play a larger role in binding the halides than HCh^– (Section S8).

Estimation of Aryl C–H^…HB A^– Strength through LFERs

Our next step was to analyze the LFER between ΔG_binding of each anion and the electrostatic potential (ESP) surface of the aryl hydrogen atom participating in anion binding (Figure 7), similar to previous work.¹⁹ ESP maps at the 0.02 Å isoelectronic surface for the six receptors were computed at the PBE/6–31G(d) level of theory.³⁹ The ESP value of the C–H HB donor was used to ascertain the electronic effect that the –R substituent has on the C–H HB donor, and by extension the binding strength of the four anions. The statistically significant linear fit (Table 5) agrees with previous work that anion binding in our systems and others is often strongly governed by electrostatics.^19,36,37,40 The fact that HS^– and Cl^– seem to have a better fit (0.97 ≤ R² ≤ 0.98) than HSe^– and Br^– (0.93 ≤ R² ≤ 0.95) could point to attractive binding forces other than electrostatics, such as induction and dispersion,⁴¹ playing a more significant role in binding the larger anions in our systems. The ChelpG charges, which give atomic charges that map to the ESP,⁴² for the key C–H HB donor did not reveal statistically significant linear fits. This showed that it is not the average charge of the C–H atoms that are meaningful to explain the binding; instead, the anisotropy of the ESP map is crucial to explain the chemistry of binding in these receptors.

Figure 7.

(a) Free receptors 1^R can twist into the “W” conformation.⁴⁴ (b) Representative ESP maps of 1^CF3, 1^H, and 1^NMe2, calculated at the PBE level of theory. The values describe the energy at the 0.02 Å isoelectric surface of the C–H HB donor para to the –R substituent.

Table 5.

Fitting Statistics for the LFER Between ΔG_binding and ESP Surfaces of the Model Compounds for HS^–, HSe^–, Cl^–, and Br^–

Guest	Slope	Intercept	p-value			R2	R2adj.
			Slope	Intercept	Model
HS–	−105 ± 9	−3.17 ± 0.17	< 0.01	< 0.01	< 0.01	0.97	0.97
HSe–	−48 ± 5	−2.89 ± 0.10	< 0.01	< 0.01	< 0.01	0.95	0.94
Cl–	−37 ± 3	−3.75 ± 0.06	< 0.01	< 0.01	< 0.01	0.98	0.97
Br–	−31 ± 4	−2.27 ± 0.08	< 0.01	< 0.01	< 0.01	0.93	0.92

This LFER was used in conjunction with previously published Equation 4 to estimate the strength of the aryl C–H^…A^– HB.¹⁹ The equation assumes that at the intercept (when the ESP at the C–H HB is equal to 0), all of the remaining binding energy is due to attractive interactions other than the C–H HB. Subtracting out all other attractive interactions from the experimental ΔG_binding should provide an estimate for the C–H^…A^– HB strength. Note that since the binding of anions in our hosts is a cooperative event between the N–H_b/c urea and aryl C–H_a HBs,⁴³ and other forces such as contribution from solvent are not accounted for, the following data are simply estimates.

$$\text{Aryl CH}\cdots {\text{A}}^{–}\text{HB Strength}=\Delta {\text{G}}_{\text{binding}}–\text{intercept}$$ (4)

The estimated aryl C–H^…A^– HB bond strengths for each of the six receptors with each of the four anions are shown in Table 6. Unlike the Swain-Lupton analysis, the plot of binding strengths vs. ESP shows a clear difference between HS^– relative to the other three anions (Figure 8), reflecting the differences in Hammett plots of the anion guests. The strength of the aryl C–H^…A^– HBs are very similar for each of the six receptors with HSe^–, Cl^–, and Br^–. Conversely, HS^– shows considerably stronger C–H HB strengths for all the receptors in this system. From these data, it is clear that the aryl C–H donor has a preference for HS^– binding over the other anions, emphasizing the importance of using such moieties in the design of HS^– sensitive and selective probes and materials, and indicating that the nature of the C–H^…A^– interaction is the cause of the difference in slopes in the Hammett plot.

Table 6.

Estimated Aryl C–H⋯A^– HB Strengths Between 1^R and all Four Anions

		HS–	HSe–	Cl–	Br–
Host	ESP (kcal mol−1)	Aryl C–H…A– HB Strength (kcal mol−1)
1CF3	0.0234	−2.52 ± 0.18	−1.17 ± 0.11	−0.86 ± 0.24	−0.78 ± 0.08
1Cl	0.0219	−2.18 ± 0.19	−1.08 ± 0.11	−0.83 ± 0.24	−0.63 ± 0.08
1F	0.0200	−2.17 ± 0.18	−0.91 ± 0.11	−0.72 ± 0.24	−0.63 ± 0.08
1H	0.0176	−1.87 ± 0.19	−0.82 ± 0.12	−0.68 ± 0.24	−0.57 ± 0.08
1tBu	0.0166	−1.68 ± 0.19	0.74 ± 0.12	−0.65 ± 0.25	−0.52 ± 0.13
1NMe2	0.0114	−1.22 ± 0.17	−0.60 ± 0.12	0.40 ± 0.25	−0.36 ± 0.10

Figure 8.

Estimated aryl C–H^…A^– HB strengths for all six receptors, with all four anions. While HSe^–, Cl^–, and Br^– show similar HB strengths, HS^– shows a stronger binding with the C–H HB donor motif.

CONCLUSIONS

Efforts by many others in the field,^{20,36,45–55} as well as our studies on equilibrium isotope effects⁴³ and LFERs on aryl C–H^…A^– HBs,¹⁹ have helped establish C–H HBs as an important anion recognition motif in supramolecular chemistry. We endeavored to apply this study to HCh^–, due to previous speculation that aryl C–H^…A^– HBs could be important to binding these anions.^11,13,56

We therefore used our model aryl ethynyl bisurea anion-binding systems to report the first example of a LFER with HCh^– receptors in this manuscript. During this process, we also expanded the family of receptors shown to reversibly bind the reactive hydrochalcogenide anions. Importantly, we observed significant differences in HS^– binding in comparison to Cl^–, HSe^–, and Br^– anions. The LFERs of ΔG_binding vs. σ_p show that our receptors prefer to bind smaller, more basic anions. Additionally, selectivity for HS^– over the other anions increased with increasing C–H HB donor acidity. Furthermore, Hammett plots illustrated a significantly greater substituent effect on HS^– binding when compared to that of the other three anionic guests. Finally, the estimate of the aryl C–H^…A^– HB bond strength again revealed unique behavior for HS^– binding, where significantly higher C–H HB strengths were observed.

Taken together, the insights from our investigation highlight the design principles needed for the next generation of selective hosts, materials, and probes for HCh^– anion binding. These studies demonstrate the importance of polarization of aryl C–H HB donors for HS^–, which may be utilized to bind this anion more tightly and selectively for a variety of applications in biomedical and environmental research. In addition, our studies on a fairly simple model systems may shed light on more complicated systems where C–H^…Ch HBs and HCh^–recognition are gaining attention in molecular and structural biology. It has been well established that C–H^…O HBs are important in organocatalysis^57,58 and in defining the structure and function of biomolecules such as DNA, RNA, and proteins.^59,60 Moving down the periodic table, C–H HBs and other non-covalent binding interactions with chalcogens are less well studied. Methionine C–H^…S HBs have been found to be important in the catalytic activity and substrate specificity of methionine aminopeptidase.¹⁸ Furthermore, a bacterial cell ion channel for HS^– was found to use only non-covalent molecular recognition in transport of the anion,⁶¹ and HS^– has been found in the turnover state of a nitrogenase enzyme, held in place by HB interactions.⁶² Perhaps our studies on fairly simple model systems lead to a hypothesis that C–H HBs are over-represented in biological examples of sulfur compound recognition.

ABBREVIATIONS

Ch

chalcogen

HCh^–

hydrochalcogenide anion

HS^–

hydrosulfide

HSe^–

hydroselenide

H₂S

hydrogen sulfide

H₂Se

hydrogen selenide

HB

hydrogen bond

PDB

Protein Data Base

LFER

linear free energy relationship

ΔG_binding

solution binding energy

A^–

anion

NMR

nuclear magnetic resonance

K_a

binding affinity

ANCOVA

analysis of covariance

ESP

electrostatic potential

NBu4+

tetrabutylammonium