Cation-π Binding Ability of BN Indole

Abstract

A BN indole-containing aromatic scaffold has been synthesized and the cation-π binding ability characterized by nuclear magnetic resonance (NMR) monitored titrations. The resulting chemical shifts were analyzed using a non-linear curve fitting procedure and the extracted association constants (K_a’s) compared with the natural indole scaffold. Computations were also performed to support our findings. This work shows that incorporation of a B–N bond in place of a C–C bond in an aromatic system slightly lowers the cation-π binding ability of the arene’s π-system with simple cations.

Graphical Abstract

Noncovalent interactions play a key role in many biological processes, such as protein-protein binding and drug-receptor interactions.¹ They also affect the structure and properties of biomacromolecules, such as the stability and secondary structure of proteins.² One such intermolecular force is the cation-π interaction, i.e. the attraction between a cationic acceptor species and the π-system of a donor molecule.^3,4 While any compound with π-electrons can act as the π-donor, aromatic molecules are the most common π-donors.^5,6

Despite its rarity in proteins, studies have shown that tryptophan, with its π-electron-rich indole side-chain, is one of the main contributors to cation-π interactions in biological settings.⁷ Tryptophan is prevalent in neurobiology and is found in neurotransmitter receptors, e.g., in acetylcholine receptors in acetylcholine binding proteins.⁸ In other cases, tryptophan is shown to play a crucial role in the stability of proteins; for instance, it stabilizes the secondary structure of β-hairpin peptides.^7d

One key factor in determining the strength of a cation-π interaction is the magnitude of the quadrupole moment of the arene, a feature determined by the arene’s electronic structure.⁹ BN/CC isosterism¹⁰ (the substitution of a C–C bond unit with a B–N bond unit) has emerged as a versatile strategy to introduce new physical and chemical properties into biologically relevant molecules while largely maintaining their steric profile.¹¹ In the context of indole, the key side chain motif of tryptophan, we have demonstrated for example that its BN isostere 2 exhibits a higher pK_a of the indole N-H proton, higher reactivity toward electrophilic aromatic substitution at the C3 position, and bathochromically shifted absorbance and emission spectra (Table 1) relative to the natural indole 1.¹² We also described the electronic structure of BN-indole through a combination of UV-photoelectron spectroscopy experiments, and computational studies and revealed that the BN indole has a smaller ground state dipole moment, is less aromatic, and has a similar first ionization energy (HOMO) compared to natural indole (Table 1).¹³

Table 1.

Selected properties of indole 1 and BN indole 2.

	pKa	λabsa	λema	Dipole Moment	NICS(1)	HOMO IEd
Indole 1	21	268 nm	315 nm	2.2 D	−11.7,b −10.9c	7.9 eV
BN-Indole 2	30	293 nm	360 nm	1.5 D	−7.8,b −8.1c	8.05 eV

^aMeasured in MeCN.

^bCalculated for the six-membered ring.

^cCalculated for the five-membered ring.

^dIonization energies (IE) measured by UV-PES.

Previously, Korona and coworkers studied the interaction of Na⁺ and Mg²⁺ with the three isomers of azaborine and benzene and their monosubstituted variants through computational methods.¹⁴ Here they found that the position of the cation over the arene was dependent on the boron and nitrogen positions in the ring and that the cation-π interaction is weaker in the azaborine cases versus benzene.^14a They also showed that the σ-polarization plays a primary role and the π-polarization a secondary role in the cation-π interaction.^14b

Recently, we were able to show that the endogenous tRNA synthetase in a tryptophan auxotrophic strain of E. coli can charge a BN-analogue of tryptophan derived from BN indole 2 in place of the natural substrate and further incorporate it into proteins.¹⁵ With this new ability to incorporate an unnatural BN-containing amino acid into proteins in our toolbox, learning about BN-indole’s π-donating abilities could help chemical biologists strategically apply BN-tryptophan to other proteins. However, despite the significant advances made in the fundamental understanding of the consequences of BN/CC isosterism on the electronic structure of arenes, an experimental investigation into the effect of BN/CC isosterism on cation-π interactions has remained elusive to date. In this communication, we evaluate the cation-π binding ability of BN indole 2 in the context of the C₃-symmetric supramolecular BN indole scaffold 3 (Figure 1), which was designed previously¹⁶ as a model for aromatic cages in biological systems, and we demonstrate that BN/CC isosterism of the indole framework leads to a slightly diminished cation binding ability. Our experimentally observed trend is corroborated by DFT calculations.

Figure 1.

Indole scaffolds studied in this work

The synthesis of the BN indole scaffold 3 was accomplished by adapting the method to synthesize the CC indole scaffold 4 to the BN-system. Compound 4 was previously developed and explored by Hof and coworkers in order to study the cation-π binding ability of indole in aqueous and organic media.¹⁶ Deprotonation of the BN indole 2 with potassium hydride in dimethylformamide followed by addition of the 1,3,5-tribromomethylbenzene electrophile 5 led to the formation of the BN scaffold 3 (Scheme 1).

Scheme 1.

Synthesis of BN indole scaffold 3

With the scaffolds 3 and 4 in hand, NMR titrations were performed to determine the association constants for the binding of lithium and sodium cations to the scaffolds (Figures 2 and 3). In both cases, a 0.9-1 mM solution of scaffold was made in acetonitrile-D3 and titrated with a 2.38-2.56 M solution of lithium iodide or a 771-820 mM solution of sodium iodide.¹⁷ The scaffold concentration was kept consistent between the scaffold solution and the LiI/NaI solution in order to eliminate any dilution effects that may arise. The resulting changes in chemical shifts were fitted to a non-linear regression model using the online resource provided by Thordarson, and the association constants (K_a) were extracted.¹⁸ The chemical shift for scaffolds 3 and 4 were first unambiguously assigned via 2D NMR techniques (NOESY, HSQC, HMBC, COSY) prior to the titration experiments (See ESI for details). We chose to look specifically at the direct comparison of the chemical shift changes associated with the C5 proton of the scaffolds 3 and 4 for the following reasons: 1) It is known that cations preferentially bind to the six-membered portion of the bicyclic framework (vide infra),¹⁹ thus a proton position within the six-membered ring is appropriate, 2) the C5 proton signal shape is appropriate with sharp, assignable chemical shifts for both scaffolds²⁰ (on the other hand, the C6 proton shift for 3 and the C4 and C7 proton shifts for 4 are relatively broad in nature) (See Figures 2 and 3), 3) the electrostatic potential maps (vide infra) reveal a consistent negatively charged area located near the C5 and C6 positions for both scaffolds 3 and 4.

Figure 2.

Representative NMR spectra and titration curve for the titration of LiI to BN-indole scaffold 3

Figure 3.

Representative NMR spectra and titration curve for the titration of LiI to CC-indole scaffold 4

Our titration experiments (Figures 2 and 3) indicate that binding with lithium iodide gives a slightly lower association constant for the BN scaffold 3, with a K_a of 0.055 ± 0.003 M^–1, compared to a K_a of 0.060 ± 0.004 M⁻¹ for the CC scaffold 4 (Table 2). As expected from a solution phase cation-π binding system,²¹ the K_a with sodium iodide (vs. lithium iodide) was higher at 0.20 ± 0.01 M⁻¹ for the BN scaffold 3 and also for the CC scaffold 4 at 0.23 ± 0.04 M⁻¹. These results show that overall the incorporation of the B–N bond into the indole results in only a slightly lower cation-π binding ability.

Table 2.

Experimental K_a values of indole scaffolds 3 and 4 with LiI and NaI

	Li+	Na+
BN-Scaffold 31	0.055 ± 0.003 M−1	0.20 ± 0.01 M−1
CC-Scaffold 42	0.060 ± 0.004 M−1	0.23 ± 0.04 M−1

All values are an average of 3 titrations.

¹Measured with C5 ¹H NMR shifts

²Measured with C5/6 ¹H NMR shifts

We performed a gas-phase computational prediction of the cation-π binding energy of Li⁺ and Na⁺ to the aromatic CC and BN indole rings. All theoretical results were obtained with Gaussian 09 software.²² The range-separated hybrid (RSH) functional CAM-B3LYP²³ is used, with a triple-ζ basis set and polarization functions 6-311G(d,p)²⁴, in order to keep homogeneous with according to our previous computational studies, for the description of electronic energies, which are in good agreement with experimental values of ionization energies obtained by UV-PES (UV photoelectron spectroscopy).²⁵ Binding energies were calculated rigorously taking into account fragment relaxation energies²⁶ and basis set superposition error (BSSE).²⁷

Three trends can be derived from the gas-phase computational results (Table 3): 1) The cations (Li⁺ and Na⁺) bind more strongly to the π6 ring than to the π5 ring for both CC indole¹⁹ and BN indole. The interaction of both the CC indole (π5) and BN indole (π5) rings with K⁺ is predicted to be unfavorable, in contrast to the cation-π interaction of the corresponding π6 rings. 2) Consistent with the experimentally observed trend, the cations bind to CC indole stronger than to BN indole by 2.5 kcal/mol for Li⁺, 1.4 kcal/mol for Na⁺, and 1.3 kcal/mol for K⁺, with respect to the π6 ring. 3) As is expected in the gas phase where solvation effects are absent, the smaller Li⁺ binds the strongest and the larger K⁺ the weakest.²¹

Table 3.

Calculated binding energies of indole 1 and BN indole 2 with various cations

	ΔE(Li+)	ΔE(Na+)	ΔE(K+)
BN indole 2 (π6)	−45.8	−30.1	−21.7
CC indole 1 (π6)	−48.3	−31.5	−23.0
BN indole 2 (π5)	−41.7	−26.8	N/A
CC indole 1 (π5)	−42.9	−28.1	N/A

Energy values in kcal/mol. ΔE corresponds to the binding energy taking into account the basis set superposition error (BSSE).

The calculated electrostatic potential (ESP) map (Figure 4) are also consistent with our titration results (Figures 2 and 3). The ESP maps show that in both arenes there is a significant negative charge over the six-membered ring in both indoles, which is consistent with predicted stronger cation binding affinity for the π6 ring system over the π5 system for both CC and BN indole. There appears to be a correlation between the location of the negative charge and the magnitude of the chemical shift change Δδ upon cation binding. For CC indole 1, the C2 position appears to be less electron rich according to the ESP map, resulting in a small observed Δδ upon Li⁺ binding (Figure 3). For BN indole 2, the C2, C3, and C4 positions are relatively electron deficient, and consequently, the observed Δδ for these positions are smaller than the Δδ for the C5-C7 positions (Figure 2). The additional nitrogen atom in BN indole 2 appears to exert its electronegative influence on the C3 and C4 positions giving rise to a more non-uniform and position-dependent Δδ as a function of the titrated cation concentration relative to the CC indole 1 (Figure 2 vs Figure 3). Furthermore, we note that the C7 position in BN indole 2 appears to be the most electron rich position (partially due to the inductively donating effect of the adjacent boron atom, see ESP map). As a consequence, the C7 proton experiences the largest Δδ upon titration with cation (Figure 2).

Figure 4.

Electrostatic potential (ESP) map of indole 1 and BN indole 2

In conclusion, we demonstrate that BN/CC isosterism of the indole motif slightly decreases the cation-π binding ability of the resulting BN indole through comparative NMR titration experiments of the supramolecular scaffolds 3 and 4. This work represents the first experimental investigation probing the consequence of BN/CC isosterism on the quadrupole behavior of a BN arene in the context of the biologically relevant indole motif. The experimental findings are corroborated by computational predictions of gas-phase binding energies of indole and BN indole with Li⁺ and Na⁺. The fact that BN/CC isosterism of indole does not significantly perturb the fundamental quadrupole character bodes well for future chemical biology applications involving the unnatural BN tryptophan.